diff --git "a/model2/trainer_state.json" "b/model2/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/model2/trainer_state.json"
@@ -0,0 +1,87290 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0001283579352709,
+  "eval_steps": 500,
+  "global_step": 109070,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 9.168423947923352e-05,
+      "learning_rate": 0.00019998349683689374,
+      "loss": 1.923,
+      "mean_token_accuracy": 0.5671112179756165,
+      "num_tokens": 23240.0,
+      "step": 10
+    },
+    {
+      "epoch": 0.00018336847895846704,
+      "learning_rate": 0.0001999651599889979,
+      "loss": 1.3823,
+      "mean_token_accuracy": 0.6590937197208404,
+      "num_tokens": 46441.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.0002750527184377006,
+      "learning_rate": 0.00019994682314110205,
+      "loss": 1.2459,
+      "mean_token_accuracy": 0.6675321936607361,
+      "num_tokens": 69296.0,
+      "step": 30
+    },
+    {
+      "epoch": 0.0003667369579169341,
+      "learning_rate": 0.00019992848629320618,
+      "loss": 1.1608,
+      "mean_token_accuracy": 0.6786929786205291,
+      "num_tokens": 92571.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.0004584211973961676,
+      "learning_rate": 0.00019991014944531038,
+      "loss": 1.1545,
+      "mean_token_accuracy": 0.6873579978942871,
+      "num_tokens": 115866.0,
+      "step": 50
+    },
+    {
+      "epoch": 0.0005501054368754011,
+      "learning_rate": 0.00019989181259741452,
+      "loss": 1.0964,
+      "mean_token_accuracy": 0.6990690946578979,
+      "num_tokens": 138611.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.0006417896763546347,
+      "learning_rate": 0.00019987347574951868,
+      "loss": 1.081,
+      "mean_token_accuracy": 0.7047873675823212,
+      "num_tokens": 162354.0,
+      "step": 70
+    },
+    {
+      "epoch": 0.0007334739158338682,
+      "learning_rate": 0.00019985513890162282,
+      "loss": 1.1199,
+      "mean_token_accuracy": 0.6927294492721557,
+      "num_tokens": 185436.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.0008251581553131017,
+      "learning_rate": 0.00019983680205372696,
+      "loss": 1.0239,
+      "mean_token_accuracy": 0.7098340153694153,
+      "num_tokens": 209182.0,
+      "step": 90
+    },
+    {
+      "epoch": 0.0009168423947923352,
+      "learning_rate": 0.00019981846520583112,
+      "loss": 1.0153,
+      "mean_token_accuracy": 0.7134263336658477,
+      "num_tokens": 232622.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.0010085266342715687,
+      "learning_rate": 0.0001998001283579353,
+      "loss": 1.0558,
+      "mean_token_accuracy": 0.7057909607887268,
+      "num_tokens": 255986.0,
+      "step": 110
+    },
+    {
+      "epoch": 0.0011002108737508023,
+      "learning_rate": 0.00019978179151003945,
+      "loss": 1.0556,
+      "mean_token_accuracy": 0.7068513870239258,
+      "num_tokens": 278812.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.0011918951132300357,
+      "learning_rate": 0.0001997634546621436,
+      "loss": 1.1141,
+      "mean_token_accuracy": 0.6901987969875336,
+      "num_tokens": 301968.0,
+      "step": 130
+    },
+    {
+      "epoch": 0.0012835793527092693,
+      "learning_rate": 0.00019974511781424773,
+      "loss": 1.0658,
+      "mean_token_accuracy": 0.709293419122696,
+      "num_tokens": 325897.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.0013752635921885027,
+      "learning_rate": 0.0001997267809663519,
+      "loss": 1.0533,
+      "mean_token_accuracy": 0.7050972282886505,
+      "num_tokens": 348899.0,
+      "step": 150
+    },
+    {
+      "epoch": 0.0014669478316677363,
+      "learning_rate": 0.00019970844411845603,
+      "loss": 1.0351,
+      "mean_token_accuracy": 0.7042410492897033,
+      "num_tokens": 372353.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.00155863207114697,
+      "learning_rate": 0.0001996901072705602,
+      "loss": 1.0595,
+      "mean_token_accuracy": 0.7090585947036743,
+      "num_tokens": 395742.0,
+      "step": 170
+    },
+    {
+      "epoch": 0.0016503163106262033,
+      "learning_rate": 0.00019967177042266436,
+      "loss": 1.0508,
+      "mean_token_accuracy": 0.7078003644943237,
+      "num_tokens": 418213.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.001742000550105437,
+      "learning_rate": 0.0001996534335747685,
+      "loss": 1.0229,
+      "mean_token_accuracy": 0.7085508227348327,
+      "num_tokens": 441971.0,
+      "step": 190
+    },
+    {
+      "epoch": 0.0018336847895846704,
+      "learning_rate": 0.00019963509672687267,
+      "loss": 1.0318,
+      "mean_token_accuracy": 0.7104375839233399,
+      "num_tokens": 465422.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.001925369029063904,
+      "learning_rate": 0.0001996167598789768,
+      "loss": 1.0309,
+      "mean_token_accuracy": 0.714245331287384,
+      "num_tokens": 487928.0,
+      "step": 210
+    },
+    {
+      "epoch": 0.0020170532685431374,
+      "learning_rate": 0.00019959842303108097,
+      "loss": 1.0399,
+      "mean_token_accuracy": 0.7037406146526337,
+      "num_tokens": 512219.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.002108737508022371,
+      "learning_rate": 0.0001995800861831851,
+      "loss": 1.0346,
+      "mean_token_accuracy": 0.7041298568248748,
+      "num_tokens": 535130.0,
+      "step": 230
+    },
+    {
+      "epoch": 0.0022004217475016046,
+      "learning_rate": 0.00019956174933528928,
+      "loss": 1.0245,
+      "mean_token_accuracy": 0.7015379905700684,
+      "num_tokens": 558131.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.0022921059869808378,
+      "learning_rate": 0.00019954341248739344,
+      "loss": 1.0207,
+      "mean_token_accuracy": 0.7124298512935638,
+      "num_tokens": 580761.0,
+      "step": 250
+    },
+    {
+      "epoch": 0.0023837902264600714,
+      "learning_rate": 0.00019952507563949758,
+      "loss": 1.0441,
+      "mean_token_accuracy": 0.7096422195434571,
+      "num_tokens": 604278.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.002475474465939305,
+      "learning_rate": 0.00019950673879160174,
+      "loss": 1.047,
+      "mean_token_accuracy": 0.7109373688697815,
+      "num_tokens": 627516.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.0025671587054185386,
+      "learning_rate": 0.00019948840194370588,
+      "loss": 0.9743,
+      "mean_token_accuracy": 0.7240226328372955,
+      "num_tokens": 651591.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.0026588429448977722,
+      "learning_rate": 0.00019947006509581002,
+      "loss": 1.0425,
+      "mean_token_accuracy": 0.7046778917312622,
+      "num_tokens": 674320.0,
+      "step": 290
+    },
+    {
+      "epoch": 0.0027505271843770054,
+      "learning_rate": 0.0001994517282479142,
+      "loss": 0.9836,
+      "mean_token_accuracy": 0.7111404120922089,
+      "num_tokens": 697451.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.002842211423856239,
+      "learning_rate": 0.00019943339140001835,
+      "loss": 1.0388,
+      "mean_token_accuracy": 0.708145147562027,
+      "num_tokens": 721193.0,
+      "step": 310
+    },
+    {
+      "epoch": 0.0029338956633354727,
+      "learning_rate": 0.00019941505455212252,
+      "loss": 1.0164,
+      "mean_token_accuracy": 0.7146294176578522,
+      "num_tokens": 744067.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.0030255799028147063,
+      "learning_rate": 0.00019939671770422666,
+      "loss": 1.032,
+      "mean_token_accuracy": 0.7135140478610993,
+      "num_tokens": 767476.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.00311726414229394,
+      "learning_rate": 0.0001993783808563308,
+      "loss": 1.0081,
+      "mean_token_accuracy": 0.7130200088024139,
+      "num_tokens": 790545.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.003208948381773173,
+      "learning_rate": 0.00019936004400843496,
+      "loss": 0.9808,
+      "mean_token_accuracy": 0.721145749092102,
+      "num_tokens": 813522.0,
+      "step": 350
+    },
+    {
+      "epoch": 0.0033006326212524067,
+      "learning_rate": 0.0001993417071605391,
+      "loss": 0.9681,
+      "mean_token_accuracy": 0.7215366899967194,
+      "num_tokens": 836771.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.0033923168607316403,
+      "learning_rate": 0.00019932337031264326,
+      "loss": 1.0226,
+      "mean_token_accuracy": 0.7119088530540466,
+      "num_tokens": 859901.0,
+      "step": 370
+    },
+    {
+      "epoch": 0.003484001100210874,
+      "learning_rate": 0.00019930503346474743,
+      "loss": 0.9835,
+      "mean_token_accuracy": 0.7127217948436737,
+      "num_tokens": 883177.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.003575685339690107,
+      "learning_rate": 0.00019928669661685157,
+      "loss": 0.999,
+      "mean_token_accuracy": 0.7090746223926544,
+      "num_tokens": 907431.0,
+      "step": 390
+    },
+    {
+      "epoch": 0.0036673695791693407,
+      "learning_rate": 0.00019926835976895573,
+      "loss": 1.0289,
+      "mean_token_accuracy": 0.7040601015090943,
+      "num_tokens": 930277.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.0037590538186485743,
+      "learning_rate": 0.00019925002292105987,
+      "loss": 1.0032,
+      "mean_token_accuracy": 0.7104796707630158,
+      "num_tokens": 952904.0,
+      "step": 410
+    },
+    {
+      "epoch": 0.003850738058127808,
+      "learning_rate": 0.00019923168607316404,
+      "loss": 1.028,
+      "mean_token_accuracy": 0.7142015218734741,
+      "num_tokens": 976112.0,
+      "step": 420
+    },
+    {
+      "epoch": 0.0039424222976070416,
+      "learning_rate": 0.00019921334922526817,
+      "loss": 1.0175,
+      "mean_token_accuracy": 0.7111793398857117,
+      "num_tokens": 1000129.0,
+      "step": 430
+    },
+    {
+      "epoch": 0.004034106537086275,
+      "learning_rate": 0.00019919501237737234,
+      "loss": 1.0108,
+      "mean_token_accuracy": 0.7194611966609955,
+      "num_tokens": 1024034.0,
+      "step": 440
+    },
+    {
+      "epoch": 0.004125790776565509,
+      "learning_rate": 0.0001991766755294765,
+      "loss": 1.001,
+      "mean_token_accuracy": 0.7095412254333496,
+      "num_tokens": 1047274.0,
+      "step": 450
+    },
+    {
+      "epoch": 0.004217475016044742,
+      "learning_rate": 0.00019915833868158064,
+      "loss": 1.0316,
+      "mean_token_accuracy": 0.7048232853412628,
+      "num_tokens": 1070814.0,
+      "step": 460
+    },
+    {
+      "epoch": 0.004309159255523975,
+      "learning_rate": 0.0001991400018336848,
+      "loss": 1.0021,
+      "mean_token_accuracy": 0.7127786338329315,
+      "num_tokens": 1093506.0,
+      "step": 470
+    },
+    {
+      "epoch": 0.004400843495003209,
+      "learning_rate": 0.00019912166498578895,
+      "loss": 1.0864,
+      "mean_token_accuracy": 0.6990128040313721,
+      "num_tokens": 1116273.0,
+      "step": 480
+    },
+    {
+      "epoch": 0.004492527734482442,
+      "learning_rate": 0.00019910332813789309,
+      "loss": 0.9868,
+      "mean_token_accuracy": 0.7214479863643646,
+      "num_tokens": 1139494.0,
+      "step": 490
+    },
+    {
+      "epoch": 0.0045842119739616756,
+      "learning_rate": 0.00019908499128999728,
+      "loss": 1.0224,
+      "mean_token_accuracy": 0.7078819572925568,
+      "num_tokens": 1162185.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.00467589621344091,
+      "learning_rate": 0.00019906665444210142,
+      "loss": 0.9564,
+      "mean_token_accuracy": 0.7251663327217102,
+      "num_tokens": 1185649.0,
+      "step": 510
+    },
+    {
+      "epoch": 0.004767580452920143,
+      "learning_rate": 0.00019904831759420558,
+      "loss": 1.0394,
+      "mean_token_accuracy": 0.7065913915634155,
+      "num_tokens": 1208749.0,
+      "step": 520
+    },
+    {
+      "epoch": 0.004859264692399377,
+      "learning_rate": 0.00019902998074630972,
+      "loss": 0.9923,
+      "mean_token_accuracy": 0.7146309971809387,
+      "num_tokens": 1232124.0,
+      "step": 530
+    },
+    {
+      "epoch": 0.00495094893187861,
+      "learning_rate": 0.00019901164389841386,
+      "loss": 1.0036,
+      "mean_token_accuracy": 0.717809009552002,
+      "num_tokens": 1255100.0,
+      "step": 540
+    },
+    {
+      "epoch": 0.005042633171357843,
+      "learning_rate": 0.00019899330705051802,
+      "loss": 1.0539,
+      "mean_token_accuracy": 0.7106794893741608,
+      "num_tokens": 1278052.0,
+      "step": 550
+    },
+    {
+      "epoch": 0.005134317410837077,
+      "learning_rate": 0.00019897497020262216,
+      "loss": 1.0064,
+      "mean_token_accuracy": 0.7144601583480835,
+      "num_tokens": 1301376.0,
+      "step": 560
+    },
+    {
+      "epoch": 0.0052260016503163104,
+      "learning_rate": 0.00019895663335472633,
+      "loss": 1.0455,
+      "mean_token_accuracy": 0.7059086263179779,
+      "num_tokens": 1324700.0,
+      "step": 570
+    },
+    {
+      "epoch": 0.0053176858897955445,
+      "learning_rate": 0.0001989382965068305,
+      "loss": 0.9906,
+      "mean_token_accuracy": 0.7112154304981232,
+      "num_tokens": 1347260.0,
+      "step": 580
+    },
+    {
+      "epoch": 0.005409370129274778,
+      "learning_rate": 0.00019891995965893463,
+      "loss": 0.9914,
+      "mean_token_accuracy": 0.7184651911258697,
+      "num_tokens": 1370130.0,
+      "step": 590
+    },
+    {
+      "epoch": 0.005501054368754011,
+      "learning_rate": 0.0001989016228110388,
+      "loss": 0.977,
+      "mean_token_accuracy": 0.7227175831794739,
+      "num_tokens": 1393754.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.005592738608233245,
+      "learning_rate": 0.00019888328596314294,
+      "loss": 0.9863,
+      "mean_token_accuracy": 0.7218797028064727,
+      "num_tokens": 1417095.0,
+      "step": 610
+    },
+    {
+      "epoch": 0.005684422847712478,
+      "learning_rate": 0.0001988649491152471,
+      "loss": 1.0015,
+      "mean_token_accuracy": 0.7211240470409394,
+      "num_tokens": 1440173.0,
+      "step": 620
+    },
+    {
+      "epoch": 0.005776107087191712,
+      "learning_rate": 0.00019884661226735127,
+      "loss": 0.9524,
+      "mean_token_accuracy": 0.7226812183856964,
+      "num_tokens": 1462615.0,
+      "step": 630
+    },
+    {
+      "epoch": 0.005867791326670945,
+      "learning_rate": 0.0001988282754194554,
+      "loss": 0.9427,
+      "mean_token_accuracy": 0.7317568361759186,
+      "num_tokens": 1485642.0,
+      "step": 640
+    },
+    {
+      "epoch": 0.0059594755661501785,
+      "learning_rate": 0.00019880993857155957,
+      "loss": 0.9702,
+      "mean_token_accuracy": 0.7181661605834961,
+      "num_tokens": 1508865.0,
+      "step": 650
+    },
+    {
+      "epoch": 0.0060511598056294125,
+      "learning_rate": 0.0001987916017236637,
+      "loss": 0.9822,
+      "mean_token_accuracy": 0.7227430582046509,
+      "num_tokens": 1531240.0,
+      "step": 660
+    },
+    {
+      "epoch": 0.006142844045108646,
+      "learning_rate": 0.00019877326487576787,
+      "loss": 0.9823,
+      "mean_token_accuracy": 0.7115278661251068,
+      "num_tokens": 1554582.0,
+      "step": 670
+    },
+    {
+      "epoch": 0.00623452828458788,
+      "learning_rate": 0.000198754928027872,
+      "loss": 1.0114,
+      "mean_token_accuracy": 0.71690434217453,
+      "num_tokens": 1577156.0,
+      "step": 680
+    },
+    {
+      "epoch": 0.006326212524067113,
+      "learning_rate": 0.00019873659117997615,
+      "loss": 0.9203,
+      "mean_token_accuracy": 0.7258795380592347,
+      "num_tokens": 1600592.0,
+      "step": 690
+    },
+    {
+      "epoch": 0.006417896763546346,
+      "learning_rate": 0.00019871825433208034,
+      "loss": 0.9611,
+      "mean_token_accuracy": 0.7181583404541015,
+      "num_tokens": 1622497.0,
+      "step": 700
+    },
+    {
+      "epoch": 0.00650958100302558,
+      "learning_rate": 0.00019869991748418448,
+      "loss": 0.995,
+      "mean_token_accuracy": 0.7175450980663299,
+      "num_tokens": 1645055.0,
+      "step": 710
+    },
+    {
+      "epoch": 0.006601265242504813,
+      "learning_rate": 0.00019868158063628865,
+      "loss": 0.9704,
+      "mean_token_accuracy": 0.7163600087165832,
+      "num_tokens": 1668230.0,
+      "step": 720
+    },
+    {
+      "epoch": 0.0066929494819840465,
+      "learning_rate": 0.00019866324378839279,
+      "loss": 0.9659,
+      "mean_token_accuracy": 0.719957035779953,
+      "num_tokens": 1690963.0,
+      "step": 730
+    },
+    {
+      "epoch": 0.006784633721463281,
+      "learning_rate": 0.00019864490694049692,
+      "loss": 0.9757,
+      "mean_token_accuracy": 0.7256332635879517,
+      "num_tokens": 1714141.0,
+      "step": 740
+    },
+    {
+      "epoch": 0.006876317960942514,
+      "learning_rate": 0.0001986265700926011,
+      "loss": 0.9343,
+      "mean_token_accuracy": 0.7251800775527955,
+      "num_tokens": 1737597.0,
+      "step": 750
+    },
+    {
+      "epoch": 0.006968002200421748,
+      "learning_rate": 0.00019860823324470525,
+      "loss": 0.9828,
+      "mean_token_accuracy": 0.7137405216693878,
+      "num_tokens": 1760673.0,
+      "step": 760
+    },
+    {
+      "epoch": 0.007059686439900981,
+      "learning_rate": 0.0001985898963968094,
+      "loss": 1.0008,
+      "mean_token_accuracy": 0.716589343547821,
+      "num_tokens": 1783485.0,
+      "step": 770
+    },
+    {
+      "epoch": 0.007151370679380214,
+      "learning_rate": 0.00019857155954891356,
+      "loss": 0.9541,
+      "mean_token_accuracy": 0.727026391029358,
+      "num_tokens": 1806277.0,
+      "step": 780
+    },
+    {
+      "epoch": 0.007243054918859448,
+      "learning_rate": 0.0001985532227010177,
+      "loss": 0.9587,
+      "mean_token_accuracy": 0.7290895402431488,
+      "num_tokens": 1829597.0,
+      "step": 790
+    },
+    {
+      "epoch": 0.007334739158338681,
+      "learning_rate": 0.00019853488585312186,
+      "loss": 0.9833,
+      "mean_token_accuracy": 0.7171423137187958,
+      "num_tokens": 1852391.0,
+      "step": 800
+    },
+    {
+      "epoch": 0.0074264233978179155,
+      "learning_rate": 0.000198516549005226,
+      "loss": 0.968,
+      "mean_token_accuracy": 0.7270346641540527,
+      "num_tokens": 1875683.0,
+      "step": 810
+    },
+    {
+      "epoch": 0.007518107637297149,
+      "learning_rate": 0.00019849821215733017,
+      "loss": 1.0124,
+      "mean_token_accuracy": 0.7099732458591461,
+      "num_tokens": 1899044.0,
+      "step": 820
+    },
+    {
+      "epoch": 0.007609791876776382,
+      "learning_rate": 0.00019847987530943433,
+      "loss": 0.9879,
+      "mean_token_accuracy": 0.7223415613174439,
+      "num_tokens": 1922578.0,
+      "step": 830
+    },
+    {
+      "epoch": 0.007701476116255616,
+      "learning_rate": 0.00019846153846153847,
+      "loss": 1.0117,
+      "mean_token_accuracy": 0.710689491033554,
+      "num_tokens": 1945702.0,
+      "step": 840
+    },
+    {
+      "epoch": 0.007793160355734849,
+      "learning_rate": 0.00019844320161364263,
+      "loss": 1.0074,
+      "mean_token_accuracy": 0.7130803108215332,
+      "num_tokens": 1968971.0,
+      "step": 850
+    },
+    {
+      "epoch": 0.007884844595214083,
+      "learning_rate": 0.00019842486476574677,
+      "loss": 0.9564,
+      "mean_token_accuracy": 0.7260822772979736,
+      "num_tokens": 1992042.0,
+      "step": 860
+    },
+    {
+      "epoch": 0.007976528834693316,
+      "learning_rate": 0.00019840652791785094,
+      "loss": 0.9605,
+      "mean_token_accuracy": 0.7239135682582856,
+      "num_tokens": 2014709.0,
+      "step": 870
+    },
+    {
+      "epoch": 0.00806821307417255,
+      "learning_rate": 0.00019838819106995508,
+      "loss": 0.9672,
+      "mean_token_accuracy": 0.7277372121810913,
+      "num_tokens": 2037689.0,
+      "step": 880
+    },
+    {
+      "epoch": 0.008159897313651783,
+      "learning_rate": 0.00019836985422205924,
+      "loss": 0.9373,
+      "mean_token_accuracy": 0.724119883775711,
+      "num_tokens": 2060452.0,
+      "step": 890
+    },
+    {
+      "epoch": 0.008251581553131018,
+      "learning_rate": 0.0001983515173741634,
+      "loss": 0.9671,
+      "mean_token_accuracy": 0.7217941105365753,
+      "num_tokens": 2083439.0,
+      "step": 900
+    },
+    {
+      "epoch": 0.00834326579261025,
+      "learning_rate": 0.00019833318052626755,
+      "loss": 0.9593,
+      "mean_token_accuracy": 0.726180350780487,
+      "num_tokens": 2106422.0,
+      "step": 910
+    },
+    {
+      "epoch": 0.008434950032089484,
+      "learning_rate": 0.0001983148436783717,
+      "loss": 0.9969,
+      "mean_token_accuracy": 0.7121199488639831,
+      "num_tokens": 2129419.0,
+      "step": 920
+    },
+    {
+      "epoch": 0.008526634271568717,
+      "learning_rate": 0.00019829650683047585,
+      "loss": 0.988,
+      "mean_token_accuracy": 0.7151882767677307,
+      "num_tokens": 2153284.0,
+      "step": 930
+    },
+    {
+      "epoch": 0.00861831851104795,
+      "learning_rate": 0.00019827816998258,
+      "loss": 0.9936,
+      "mean_token_accuracy": 0.7165880739688874,
+      "num_tokens": 2177157.0,
+      "step": 940
+    },
+    {
+      "epoch": 0.008710002750527185,
+      "learning_rate": 0.00019825983313468415,
+      "loss": 0.9236,
+      "mean_token_accuracy": 0.7331863820552826,
+      "num_tokens": 2199698.0,
+      "step": 950
+    },
+    {
+      "epoch": 0.008801686990006418,
+      "learning_rate": 0.00019824149628678832,
+      "loss": 0.9382,
+      "mean_token_accuracy": 0.7276931762695312,
+      "num_tokens": 2222343.0,
+      "step": 960
+    },
+    {
+      "epoch": 0.008893371229485652,
+      "learning_rate": 0.00019822315943889246,
+      "loss": 1.0353,
+      "mean_token_accuracy": 0.706400340795517,
+      "num_tokens": 2245100.0,
+      "step": 970
+    },
+    {
+      "epoch": 0.008985055468964885,
+      "learning_rate": 0.00019820482259099662,
+      "loss": 0.943,
+      "mean_token_accuracy": 0.7241316556930542,
+      "num_tokens": 2267882.0,
+      "step": 980
+    },
+    {
+      "epoch": 0.009076739708444118,
+      "learning_rate": 0.00019818648574310076,
+      "loss": 0.9818,
+      "mean_token_accuracy": 0.715369588136673,
+      "num_tokens": 2290572.0,
+      "step": 990
+    },
+    {
+      "epoch": 0.009168423947923351,
+      "learning_rate": 0.00019816814889520493,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.7224092602729797,
+      "num_tokens": 2312974.0,
+      "step": 1000
+    },
+    {
+      "epoch": 0.009260108187402586,
+      "learning_rate": 0.00019814981204730906,
+      "loss": 0.9275,
+      "mean_token_accuracy": 0.7284131586551666,
+      "num_tokens": 2336739.0,
+      "step": 1010
+    },
+    {
+      "epoch": 0.00935179242688182,
+      "learning_rate": 0.00019813147519941323,
+      "loss": 0.9704,
+      "mean_token_accuracy": 0.7230646967887878,
+      "num_tokens": 2360652.0,
+      "step": 1020
+    },
+    {
+      "epoch": 0.009443476666361052,
+      "learning_rate": 0.0001981131383515174,
+      "loss": 0.9735,
+      "mean_token_accuracy": 0.713595587015152,
+      "num_tokens": 2383611.0,
+      "step": 1030
+    },
+    {
+      "epoch": 0.009535160905840286,
+      "learning_rate": 0.00019809480150362153,
+      "loss": 0.9426,
+      "mean_token_accuracy": 0.7269194841384887,
+      "num_tokens": 2406931.0,
+      "step": 1040
+    },
+    {
+      "epoch": 0.009626845145319519,
+      "learning_rate": 0.0001980764646557257,
+      "loss": 0.9544,
+      "mean_token_accuracy": 0.7190650165081024,
+      "num_tokens": 2430341.0,
+      "step": 1050
+    },
+    {
+      "epoch": 0.009718529384798754,
+      "learning_rate": 0.00019805812780782984,
+      "loss": 0.9765,
+      "mean_token_accuracy": 0.7188243806362152,
+      "num_tokens": 2453112.0,
+      "step": 1060
+    },
+    {
+      "epoch": 0.009810213624277987,
+      "learning_rate": 0.000198039790959934,
+      "loss": 0.9685,
+      "mean_token_accuracy": 0.7198343932628631,
+      "num_tokens": 2476467.0,
+      "step": 1070
+    },
+    {
+      "epoch": 0.00990189786375722,
+      "learning_rate": 0.00019802145411203814,
+      "loss": 0.9433,
+      "mean_token_accuracy": 0.7255321025848389,
+      "num_tokens": 2499409.0,
+      "step": 1080
+    },
+    {
+      "epoch": 0.009993582103236453,
+      "learning_rate": 0.0001980031172641423,
+      "loss": 0.9668,
+      "mean_token_accuracy": 0.7166699886322021,
+      "num_tokens": 2522250.0,
+      "step": 1090
+    },
+    {
+      "epoch": 0.010085266342715686,
+      "learning_rate": 0.00019798478041624647,
+      "loss": 0.9958,
+      "mean_token_accuracy": 0.7167232811450959,
+      "num_tokens": 2545370.0,
+      "step": 1100
+    },
+    {
+      "epoch": 0.010176950582194921,
+      "learning_rate": 0.0001979664435683506,
+      "loss": 0.9621,
+      "mean_token_accuracy": 0.7260641932487488,
+      "num_tokens": 2568971.0,
+      "step": 1110
+    },
+    {
+      "epoch": 0.010268634821674155,
+      "learning_rate": 0.00019794810672045478,
+      "loss": 0.9318,
+      "mean_token_accuracy": 0.7268799662590026,
+      "num_tokens": 2591958.0,
+      "step": 1120
+    },
+    {
+      "epoch": 0.010360319061153388,
+      "learning_rate": 0.00019792976987255891,
+      "loss": 0.9242,
+      "mean_token_accuracy": 0.7272241413593292,
+      "num_tokens": 2615239.0,
+      "step": 1130
+    },
+    {
+      "epoch": 0.010452003300632621,
+      "learning_rate": 0.00019791143302466305,
+      "loss": 0.9699,
+      "mean_token_accuracy": 0.7197751224040985,
+      "num_tokens": 2638063.0,
+      "step": 1140
+    },
+    {
+      "epoch": 0.010543687540111854,
+      "learning_rate": 0.00019789309617676724,
+      "loss": 0.9832,
+      "mean_token_accuracy": 0.7191377222537995,
+      "num_tokens": 2660601.0,
+      "step": 1150
+    },
+    {
+      "epoch": 0.010635371779591089,
+      "learning_rate": 0.00019787475932887138,
+      "loss": 0.9897,
+      "mean_token_accuracy": 0.7226161777973175,
+      "num_tokens": 2682907.0,
+      "step": 1160
+    },
+    {
+      "epoch": 0.010727056019070322,
+      "learning_rate": 0.00019785642248097552,
+      "loss": 1.0066,
+      "mean_token_accuracy": 0.7087887823581696,
+      "num_tokens": 2706737.0,
+      "step": 1170
+    },
+    {
+      "epoch": 0.010818740258549555,
+      "learning_rate": 0.0001978380856330797,
+      "loss": 0.9805,
+      "mean_token_accuracy": 0.7189956307411194,
+      "num_tokens": 2729568.0,
+      "step": 1180
+    },
+    {
+      "epoch": 0.010910424498028789,
+      "learning_rate": 0.00019781974878518383,
+      "loss": 0.9767,
+      "mean_token_accuracy": 0.7154843866825104,
+      "num_tokens": 2753432.0,
+      "step": 1190
+    },
+    {
+      "epoch": 0.011002108737508022,
+      "learning_rate": 0.000197801411937288,
+      "loss": 0.9598,
+      "mean_token_accuracy": 0.7268693804740906,
+      "num_tokens": 2775300.0,
+      "step": 1200
+    },
+    {
+      "epoch": 0.011093792976987257,
+      "learning_rate": 0.00019778307508939213,
+      "loss": 0.9813,
+      "mean_token_accuracy": 0.7163785696029663,
+      "num_tokens": 2798396.0,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01118547721646649,
+      "learning_rate": 0.0001977647382414963,
+      "loss": 0.9622,
+      "mean_token_accuracy": 0.7209876716136933,
+      "num_tokens": 2820421.0,
+      "step": 1220
+    },
+    {
+      "epoch": 0.011277161455945723,
+      "learning_rate": 0.00019774640139360046,
+      "loss": 0.9366,
+      "mean_token_accuracy": 0.7272763848304749,
+      "num_tokens": 2842838.0,
+      "step": 1230
+    },
+    {
+      "epoch": 0.011368845695424956,
+      "learning_rate": 0.0001977280645457046,
+      "loss": 0.9896,
+      "mean_token_accuracy": 0.7131923735141754,
+      "num_tokens": 2865965.0,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01146052993490419,
+      "learning_rate": 0.00019770972769780876,
+      "loss": 0.9337,
+      "mean_token_accuracy": 0.7314384281635284,
+      "num_tokens": 2888744.0,
+      "step": 1250
+    },
+    {
+      "epoch": 0.011552214174383424,
+      "learning_rate": 0.0001976913908499129,
+      "loss": 0.9389,
+      "mean_token_accuracy": 0.7257712483406067,
+      "num_tokens": 2911838.0,
+      "step": 1260
+    },
+    {
+      "epoch": 0.011643898413862657,
+      "learning_rate": 0.00019767305400201707,
+      "loss": 0.9817,
+      "mean_token_accuracy": 0.7177240908145904,
+      "num_tokens": 2934685.0,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01173558265334189,
+      "learning_rate": 0.0001976547171541212,
+      "loss": 0.9847,
+      "mean_token_accuracy": 0.7176860392093658,
+      "num_tokens": 2957712.0,
+      "step": 1280
+    },
+    {
+      "epoch": 0.011827266892821124,
+      "learning_rate": 0.00019763638030622537,
+      "loss": 0.9527,
+      "mean_token_accuracy": 0.7200366497039795,
+      "num_tokens": 2981205.0,
+      "step": 1290
+    },
+    {
+      "epoch": 0.011918951132300357,
+      "learning_rate": 0.00019761804345832954,
+      "loss": 0.9404,
+      "mean_token_accuracy": 0.7285177111625671,
+      "num_tokens": 3005064.0,
+      "step": 1300
+    },
+    {
+      "epoch": 0.012010635371779592,
+      "learning_rate": 0.00019759970661043367,
+      "loss": 0.9749,
+      "mean_token_accuracy": 0.7237740933895112,
+      "num_tokens": 3028358.0,
+      "step": 1310
+    },
+    {
+      "epoch": 0.012102319611258825,
+      "learning_rate": 0.00019758136976253784,
+      "loss": 0.9407,
+      "mean_token_accuracy": 0.7212411761283875,
+      "num_tokens": 3051640.0,
+      "step": 1320
+    },
+    {
+      "epoch": 0.012194003850738058,
+      "learning_rate": 0.00019756303291464198,
+      "loss": 0.9969,
+      "mean_token_accuracy": 0.7143770158290863,
+      "num_tokens": 3074093.0,
+      "step": 1330
+    },
+    {
+      "epoch": 0.012285688090217291,
+      "learning_rate": 0.00019754469606674612,
+      "loss": 0.9663,
+      "mean_token_accuracy": 0.721057939529419,
+      "num_tokens": 3097195.0,
+      "step": 1340
+    },
+    {
+      "epoch": 0.012377372329696525,
+      "learning_rate": 0.0001975263592188503,
+      "loss": 0.9828,
+      "mean_token_accuracy": 0.7231638550758361,
+      "num_tokens": 3120301.0,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01246905656917576,
+      "learning_rate": 0.00019750802237095445,
+      "loss": 1.0089,
+      "mean_token_accuracy": 0.7083804249763489,
+      "num_tokens": 3143453.0,
+      "step": 1360
+    },
+    {
+      "epoch": 0.012560740808654993,
+      "learning_rate": 0.00019748968552305859,
+      "loss": 0.9372,
+      "mean_token_accuracy": 0.7232567846775055,
+      "num_tokens": 3166434.0,
+      "step": 1370
+    },
+    {
+      "epoch": 0.012652425048134226,
+      "learning_rate": 0.00019747134867516275,
+      "loss": 0.9585,
+      "mean_token_accuracy": 0.7252507746219635,
+      "num_tokens": 3189720.0,
+      "step": 1380
+    },
+    {
+      "epoch": 0.012744109287613459,
+      "learning_rate": 0.0001974530118272669,
+      "loss": 0.9243,
+      "mean_token_accuracy": 0.7273086667060852,
+      "num_tokens": 3213296.0,
+      "step": 1390
+    },
+    {
+      "epoch": 0.012835793527092692,
+      "learning_rate": 0.00019743467497937106,
+      "loss": 1.0068,
+      "mean_token_accuracy": 0.7107485592365265,
+      "num_tokens": 3236429.0,
+      "step": 1400
+    },
+    {
+      "epoch": 0.012927477766571925,
+      "learning_rate": 0.0001974163381314752,
+      "loss": 1.0114,
+      "mean_token_accuracy": 0.7163017809391021,
+      "num_tokens": 3259605.0,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01301916200605116,
+      "learning_rate": 0.00019739800128357936,
+      "loss": 0.9551,
+      "mean_token_accuracy": 0.7215744376182556,
+      "num_tokens": 3282382.0,
+      "step": 1420
+    },
+    {
+      "epoch": 0.013110846245530394,
+      "learning_rate": 0.00019737966443568352,
+      "loss": 0.9525,
+      "mean_token_accuracy": 0.7207176864147187,
+      "num_tokens": 3305393.0,
+      "step": 1430
+    },
+    {
+      "epoch": 0.013202530485009627,
+      "learning_rate": 0.00019736132758778766,
+      "loss": 0.9652,
+      "mean_token_accuracy": 0.7264041602611542,
+      "num_tokens": 3328431.0,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01329421472448886,
+      "learning_rate": 0.00019734299073989183,
+      "loss": 0.9875,
+      "mean_token_accuracy": 0.7188103854656219,
+      "num_tokens": 3351574.0,
+      "step": 1450
+    },
+    {
+      "epoch": 0.013385898963968093,
+      "learning_rate": 0.00019732465389199597,
+      "loss": 0.9556,
+      "mean_token_accuracy": 0.7261079728603363,
+      "num_tokens": 3375174.0,
+      "step": 1460
+    },
+    {
+      "epoch": 0.013477583203447328,
+      "learning_rate": 0.00019730631704410013,
+      "loss": 0.9486,
+      "mean_token_accuracy": 0.7252309262752533,
+      "num_tokens": 3399179.0,
+      "step": 1470
+    },
+    {
+      "epoch": 0.013569267442926561,
+      "learning_rate": 0.0001972879801962043,
+      "loss": 0.9087,
+      "mean_token_accuracy": 0.7351897895336151,
+      "num_tokens": 3423090.0,
+      "step": 1480
+    },
+    {
+      "epoch": 0.013660951682405794,
+      "learning_rate": 0.00019726964334830844,
+      "loss": 0.9608,
+      "mean_token_accuracy": 0.7285245358943939,
+      "num_tokens": 3446736.0,
+      "step": 1490
+    },
+    {
+      "epoch": 0.013752635921885028,
+      "learning_rate": 0.0001972513065004126,
+      "loss": 0.9423,
+      "mean_token_accuracy": 0.7259980082511902,
+      "num_tokens": 3470538.0,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01384432016136426,
+      "learning_rate": 0.00019723296965251674,
+      "loss": 0.9585,
+      "mean_token_accuracy": 0.7208814442157745,
+      "num_tokens": 3493164.0,
+      "step": 1510
+    },
+    {
+      "epoch": 0.013936004400843496,
+      "learning_rate": 0.0001972146328046209,
+      "loss": 0.9534,
+      "mean_token_accuracy": 0.7244306862354278,
+      "num_tokens": 3515939.0,
+      "step": 1520
+    },
+    {
+      "epoch": 0.014027688640322729,
+      "learning_rate": 0.00019719629595672504,
+      "loss": 0.9566,
+      "mean_token_accuracy": 0.7232561886310578,
+      "num_tokens": 3539096.0,
+      "step": 1530
+    },
+    {
+      "epoch": 0.014119372879801962,
+      "learning_rate": 0.00019717795910882918,
+      "loss": 0.941,
+      "mean_token_accuracy": 0.7308221578598022,
+      "num_tokens": 3562532.0,
+      "step": 1540
+    },
+    {
+      "epoch": 0.014211057119281195,
+      "learning_rate": 0.00019715962226093337,
+      "loss": 0.9695,
+      "mean_token_accuracy": 0.7229514122009277,
+      "num_tokens": 3585952.0,
+      "step": 1550
+    },
+    {
+      "epoch": 0.014302741358760428,
+      "learning_rate": 0.0001971412854130375,
+      "loss": 0.9583,
+      "mean_token_accuracy": 0.7235263407230377,
+      "num_tokens": 3608803.0,
+      "step": 1560
+    },
+    {
+      "epoch": 0.014394425598239663,
+      "learning_rate": 0.00019712294856514165,
+      "loss": 0.9381,
+      "mean_token_accuracy": 0.7337869882583619,
+      "num_tokens": 3631898.0,
+      "step": 1570
+    },
+    {
+      "epoch": 0.014486109837718896,
+      "learning_rate": 0.00019710461171724582,
+      "loss": 0.9422,
+      "mean_token_accuracy": 0.7355702638626098,
+      "num_tokens": 3654752.0,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01457779407719813,
+      "learning_rate": 0.00019708627486934995,
+      "loss": 0.9378,
+      "mean_token_accuracy": 0.7221675395965577,
+      "num_tokens": 3678326.0,
+      "step": 1590
+    },
+    {
+      "epoch": 0.014669478316677363,
+      "learning_rate": 0.00019706793802145412,
+      "loss": 0.9546,
+      "mean_token_accuracy": 0.7216339230537414,
+      "num_tokens": 3701736.0,
+      "step": 1600
+    },
+    {
+      "epoch": 0.014761162556156596,
+      "learning_rate": 0.00019704960117355828,
+      "loss": 0.9792,
+      "mean_token_accuracy": 0.7194464564323425,
+      "num_tokens": 3724626.0,
+      "step": 1610
+    },
+    {
+      "epoch": 0.014852846795635831,
+      "learning_rate": 0.00019703126432566242,
+      "loss": 0.9423,
+      "mean_token_accuracy": 0.7257243514060974,
+      "num_tokens": 3747992.0,
+      "step": 1620
+    },
+    {
+      "epoch": 0.014944531035115064,
+      "learning_rate": 0.0001970129274777666,
+      "loss": 0.9693,
+      "mean_token_accuracy": 0.7192770183086395,
+      "num_tokens": 3770426.0,
+      "step": 1630
+    },
+    {
+      "epoch": 0.015036215274594297,
+      "learning_rate": 0.00019699459062987073,
+      "loss": 0.9852,
+      "mean_token_accuracy": 0.7204219222068786,
+      "num_tokens": 3793564.0,
+      "step": 1640
+    },
+    {
+      "epoch": 0.01512789951407353,
+      "learning_rate": 0.0001969762537819749,
+      "loss": 0.9204,
+      "mean_token_accuracy": 0.7410035967826843,
+      "num_tokens": 3815840.0,
+      "step": 1650
+    },
+    {
+      "epoch": 0.015219583753552764,
+      "learning_rate": 0.00019695791693407903,
+      "loss": 0.9559,
+      "mean_token_accuracy": 0.7224495232105255,
+      "num_tokens": 3838591.0,
+      "step": 1660
+    },
+    {
+      "epoch": 0.015311267993031999,
+      "learning_rate": 0.0001969395800861832,
+      "loss": 0.8758,
+      "mean_token_accuracy": 0.7431340098381043,
+      "num_tokens": 3861813.0,
+      "step": 1670
+    },
+    {
+      "epoch": 0.015402952232511232,
+      "learning_rate": 0.00019692124323828736,
+      "loss": 0.9994,
+      "mean_token_accuracy": 0.7104192674160004,
+      "num_tokens": 3885929.0,
+      "step": 1680
+    },
+    {
+      "epoch": 0.015494636471990465,
+      "learning_rate": 0.0001969029063903915,
+      "loss": 0.9614,
+      "mean_token_accuracy": 0.7201364398002624,
+      "num_tokens": 3908600.0,
+      "step": 1690
+    },
+    {
+      "epoch": 0.015586320711469698,
+      "learning_rate": 0.00019688456954249567,
+      "loss": 0.9499,
+      "mean_token_accuracy": 0.7221241772174836,
+      "num_tokens": 3931576.0,
+      "step": 1700
+    },
+    {
+      "epoch": 0.015678004950948933,
+      "learning_rate": 0.0001968662326945998,
+      "loss": 0.9617,
+      "mean_token_accuracy": 0.7238831579685211,
+      "num_tokens": 3953883.0,
+      "step": 1710
+    },
+    {
+      "epoch": 0.015769689190428166,
+      "learning_rate": 0.00019684789584670397,
+      "loss": 0.9744,
+      "mean_token_accuracy": 0.7155179023742676,
+      "num_tokens": 3976360.0,
+      "step": 1720
+    },
+    {
+      "epoch": 0.0158613734299074,
+      "learning_rate": 0.0001968295589988081,
+      "loss": 0.9624,
+      "mean_token_accuracy": 0.7200638294219971,
+      "num_tokens": 3998956.0,
+      "step": 1730
+    },
+    {
+      "epoch": 0.015953057669386633,
+      "learning_rate": 0.00019681122215091227,
+      "loss": 0.9652,
+      "mean_token_accuracy": 0.7259124040603637,
+      "num_tokens": 4022236.0,
+      "step": 1740
+    },
+    {
+      "epoch": 0.016044741908865866,
+      "learning_rate": 0.00019679288530301644,
+      "loss": 0.9405,
+      "mean_token_accuracy": 0.7326333880424499,
+      "num_tokens": 4044454.0,
+      "step": 1750
+    },
+    {
+      "epoch": 0.0161364261483451,
+      "learning_rate": 0.00019677454845512058,
+      "loss": 0.9286,
+      "mean_token_accuracy": 0.7249843358993531,
+      "num_tokens": 4068244.0,
+      "step": 1760
+    },
+    {
+      "epoch": 0.016228110387824332,
+      "learning_rate": 0.00019675621160722471,
+      "loss": 0.9753,
+      "mean_token_accuracy": 0.7207641959190368,
+      "num_tokens": 4091048.0,
+      "step": 1770
+    },
+    {
+      "epoch": 0.016319794627303565,
+      "learning_rate": 0.00019673787475932888,
+      "loss": 0.9255,
+      "mean_token_accuracy": 0.7323376059532165,
+      "num_tokens": 4113954.0,
+      "step": 1780
+    },
+    {
+      "epoch": 0.0164114788667828,
+      "learning_rate": 0.00019671953791143302,
+      "loss": 0.9517,
+      "mean_token_accuracy": 0.7274047136306763,
+      "num_tokens": 4136647.0,
+      "step": 1790
+    },
+    {
+      "epoch": 0.016503163106262035,
+      "learning_rate": 0.00019670120106353718,
+      "loss": 0.9591,
+      "mean_token_accuracy": 0.7191266596317292,
+      "num_tokens": 4159354.0,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01659484734574127,
+      "learning_rate": 0.00019668286421564135,
+      "loss": 0.9155,
+      "mean_token_accuracy": 0.7271833419799805,
+      "num_tokens": 4182599.0,
+      "step": 1810
+    },
+    {
+      "epoch": 0.0166865315852205,
+      "learning_rate": 0.0001966645273677455,
+      "loss": 0.9437,
+      "mean_token_accuracy": 0.7300412118434906,
+      "num_tokens": 4205609.0,
+      "step": 1820
+    },
+    {
+      "epoch": 0.016778215824699735,
+      "learning_rate": 0.00019664619051984965,
+      "loss": 0.9627,
+      "mean_token_accuracy": 0.7181828320026398,
+      "num_tokens": 4228312.0,
+      "step": 1830
+    },
+    {
+      "epoch": 0.016869900064178968,
+      "learning_rate": 0.0001966278536719538,
+      "loss": 0.9653,
+      "mean_token_accuracy": 0.7201115906238555,
+      "num_tokens": 4252258.0,
+      "step": 1840
+    },
+    {
+      "epoch": 0.0169615843036582,
+      "learning_rate": 0.00019660951682405796,
+      "loss": 0.9111,
+      "mean_token_accuracy": 0.7319241642951966,
+      "num_tokens": 4274681.0,
+      "step": 1850
+    },
+    {
+      "epoch": 0.017053268543137434,
+      "learning_rate": 0.0001965911799761621,
+      "loss": 0.932,
+      "mean_token_accuracy": 0.7312391996383667,
+      "num_tokens": 4298157.0,
+      "step": 1860
+    },
+    {
+      "epoch": 0.017144952782616667,
+      "learning_rate": 0.00019657284312826626,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.734565258026123,
+      "num_tokens": 4321214.0,
+      "step": 1870
+    },
+    {
+      "epoch": 0.0172366370220959,
+      "learning_rate": 0.00019655450628037043,
+      "loss": 0.9306,
+      "mean_token_accuracy": 0.7280169785022735,
+      "num_tokens": 4345197.0,
+      "step": 1880
+    },
+    {
+      "epoch": 0.017328321261575134,
+      "learning_rate": 0.00019653616943247456,
+      "loss": 0.9564,
+      "mean_token_accuracy": 0.7204728484153747,
+      "num_tokens": 4368106.0,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01742000550105437,
+      "learning_rate": 0.00019651783258457873,
+      "loss": 0.9601,
+      "mean_token_accuracy": 0.712797486782074,
+      "num_tokens": 4391291.0,
+      "step": 1900
+    },
+    {
+      "epoch": 0.017511689740533604,
+      "learning_rate": 0.00019649949573668287,
+      "loss": 0.9795,
+      "mean_token_accuracy": 0.7219728231430054,
+      "num_tokens": 4414395.0,
+      "step": 1910
+    },
+    {
+      "epoch": 0.017603373980012837,
+      "learning_rate": 0.00019648115888878703,
+      "loss": 0.9308,
+      "mean_token_accuracy": 0.7315888285636902,
+      "num_tokens": 4438227.0,
+      "step": 1920
+    },
+    {
+      "epoch": 0.01769505821949207,
+      "learning_rate": 0.00019646282204089117,
+      "loss": 0.9861,
+      "mean_token_accuracy": 0.7164831399917603,
+      "num_tokens": 4460466.0,
+      "step": 1930
+    },
+    {
+      "epoch": 0.017786742458971303,
+      "learning_rate": 0.00019644448519299534,
+      "loss": 0.9795,
+      "mean_token_accuracy": 0.7084175288677216,
+      "num_tokens": 4483128.0,
+      "step": 1940
+    },
+    {
+      "epoch": 0.017878426698450536,
+      "learning_rate": 0.0001964261483450995,
+      "loss": 0.9667,
+      "mean_token_accuracy": 0.7209431111812592,
+      "num_tokens": 4505460.0,
+      "step": 1950
+    },
+    {
+      "epoch": 0.01797011093792977,
+      "learning_rate": 0.00019640781149720364,
+      "loss": 0.9912,
+      "mean_token_accuracy": 0.7063813924789428,
+      "num_tokens": 4528513.0,
+      "step": 1960
+    },
+    {
+      "epoch": 0.018061795177409003,
+      "learning_rate": 0.00019638947464930778,
+      "loss": 0.9687,
+      "mean_token_accuracy": 0.7258190512657166,
+      "num_tokens": 4552384.0,
+      "step": 1970
+    },
+    {
+      "epoch": 0.018153479416888236,
+      "learning_rate": 0.00019637113780141194,
+      "loss": 0.9719,
+      "mean_token_accuracy": 0.7236562371253967,
+      "num_tokens": 4575577.0,
+      "step": 1980
+    },
+    {
+      "epoch": 0.01824516365636747,
+      "learning_rate": 0.00019635280095351608,
+      "loss": 0.9615,
+      "mean_token_accuracy": 0.7232940793037415,
+      "num_tokens": 4598952.0,
+      "step": 1990
+    },
+    {
+      "epoch": 0.018336847895846702,
+      "learning_rate": 0.00019633446410562028,
+      "loss": 0.9049,
+      "mean_token_accuracy": 0.7341329097747803,
+      "num_tokens": 4623559.0,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01842853213532594,
+      "learning_rate": 0.00019631612725772441,
+      "loss": 0.9213,
+      "mean_token_accuracy": 0.7316263079643249,
+      "num_tokens": 4646930.0,
+      "step": 2010
+    },
+    {
+      "epoch": 0.018520216374805172,
+      "learning_rate": 0.00019629779040982855,
+      "loss": 0.9193,
+      "mean_token_accuracy": 0.7297214806079865,
+      "num_tokens": 4669607.0,
+      "step": 2020
+    },
+    {
+      "epoch": 0.018611900614284405,
+      "learning_rate": 0.00019627945356193272,
+      "loss": 0.9295,
+      "mean_token_accuracy": 0.7318191885948181,
+      "num_tokens": 4692450.0,
+      "step": 2030
+    },
+    {
+      "epoch": 0.01870358485376364,
+      "learning_rate": 0.00019626111671403686,
+      "loss": 0.9164,
+      "mean_token_accuracy": 0.7271653711795807,
+      "num_tokens": 4715980.0,
+      "step": 2040
+    },
+    {
+      "epoch": 0.01879526909324287,
+      "learning_rate": 0.00019624277986614102,
+      "loss": 0.9373,
+      "mean_token_accuracy": 0.7283840656280518,
+      "num_tokens": 4739203.0,
+      "step": 2050
+    },
+    {
+      "epoch": 0.018886953332722105,
+      "learning_rate": 0.00019622444301824516,
+      "loss": 0.9827,
+      "mean_token_accuracy": 0.7198963284492492,
+      "num_tokens": 4762605.0,
+      "step": 2060
+    },
+    {
+      "epoch": 0.018978637572201338,
+      "learning_rate": 0.00019620610617034933,
+      "loss": 0.9484,
+      "mean_token_accuracy": 0.7245118319988251,
+      "num_tokens": 4786308.0,
+      "step": 2070
+    },
+    {
+      "epoch": 0.01907032181168057,
+      "learning_rate": 0.0001961877693224535,
+      "loss": 0.9124,
+      "mean_token_accuracy": 0.7333156585693359,
+      "num_tokens": 4810147.0,
+      "step": 2080
+    },
+    {
+      "epoch": 0.019162006051159804,
+      "learning_rate": 0.00019616943247455763,
+      "loss": 0.9018,
+      "mean_token_accuracy": 0.7366366147994995,
+      "num_tokens": 4832350.0,
+      "step": 2090
+    },
+    {
+      "epoch": 0.019253690290639038,
+      "learning_rate": 0.0001961510956266618,
+      "loss": 0.9334,
+      "mean_token_accuracy": 0.7254868447780609,
+      "num_tokens": 4855821.0,
+      "step": 2100
+    },
+    {
+      "epoch": 0.019345374530118274,
+      "learning_rate": 0.00019613275877876593,
+      "loss": 0.9588,
+      "mean_token_accuracy": 0.7224485158920289,
+      "num_tokens": 4878509.0,
+      "step": 2110
+    },
+    {
+      "epoch": 0.019437058769597507,
+      "learning_rate": 0.0001961144219308701,
+      "loss": 0.927,
+      "mean_token_accuracy": 0.7318493902683259,
+      "num_tokens": 4902071.0,
+      "step": 2120
+    },
+    {
+      "epoch": 0.01952874300907674,
+      "learning_rate": 0.00019609608508297426,
+      "loss": 0.9281,
+      "mean_token_accuracy": 0.736470353603363,
+      "num_tokens": 4923762.0,
+      "step": 2130
+    },
+    {
+      "epoch": 0.019620427248555974,
+      "learning_rate": 0.0001960777482350784,
+      "loss": 0.93,
+      "mean_token_accuracy": 0.7328227162361145,
+      "num_tokens": 4946563.0,
+      "step": 2140
+    },
+    {
+      "epoch": 0.019712111488035207,
+      "learning_rate": 0.00019605941138718257,
+      "loss": 0.9769,
+      "mean_token_accuracy": 0.7141354322433472,
+      "num_tokens": 4969913.0,
+      "step": 2150
+    },
+    {
+      "epoch": 0.01980379572751444,
+      "learning_rate": 0.0001960410745392867,
+      "loss": 0.9012,
+      "mean_token_accuracy": 0.7336618602275848,
+      "num_tokens": 4993848.0,
+      "step": 2160
+    },
+    {
+      "epoch": 0.019895479966993673,
+      "learning_rate": 0.00019602273769139084,
+      "loss": 0.9385,
+      "mean_token_accuracy": 0.7301294863224029,
+      "num_tokens": 5016536.0,
+      "step": 2170
+    },
+    {
+      "epoch": 0.019987164206472906,
+      "learning_rate": 0.000196004400843495,
+      "loss": 0.9444,
+      "mean_token_accuracy": 0.720094096660614,
+      "num_tokens": 5039552.0,
+      "step": 2180
+    },
+    {
+      "epoch": 0.02007884844595214,
+      "learning_rate": 0.00019598606399559915,
+      "loss": 0.9669,
+      "mean_token_accuracy": 0.720612359046936,
+      "num_tokens": 5062464.0,
+      "step": 2190
+    },
+    {
+      "epoch": 0.020170532685431373,
+      "learning_rate": 0.00019596772714770334,
+      "loss": 0.9375,
+      "mean_token_accuracy": 0.7292058229446411,
+      "num_tokens": 5084932.0,
+      "step": 2200
+    },
+    {
+      "epoch": 0.02026221692491061,
+      "learning_rate": 0.00019594939029980748,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.7395495593547821,
+      "num_tokens": 5108191.0,
+      "step": 2210
+    },
+    {
+      "epoch": 0.020353901164389843,
+      "learning_rate": 0.00019593105345191162,
+      "loss": 0.9611,
+      "mean_token_accuracy": 0.7236228704452514,
+      "num_tokens": 5131864.0,
+      "step": 2220
+    },
+    {
+      "epoch": 0.020445585403869076,
+      "learning_rate": 0.00019591271660401578,
+      "loss": 1.0113,
+      "mean_token_accuracy": 0.714591783285141,
+      "num_tokens": 5154579.0,
+      "step": 2230
+    },
+    {
+      "epoch": 0.02053726964334831,
+      "learning_rate": 0.00019589437975611992,
+      "loss": 0.949,
+      "mean_token_accuracy": 0.7267784833908081,
+      "num_tokens": 5177731.0,
+      "step": 2240
+    },
+    {
+      "epoch": 0.020628953882827542,
+      "learning_rate": 0.00019587604290822409,
+      "loss": 0.9255,
+      "mean_token_accuracy": 0.7317762017250061,
+      "num_tokens": 5201252.0,
+      "step": 2250
+    },
+    {
+      "epoch": 0.020720638122306775,
+      "learning_rate": 0.00019585770606032825,
+      "loss": 0.9554,
+      "mean_token_accuracy": 0.7194037437438965,
+      "num_tokens": 5224340.0,
+      "step": 2260
+    },
+    {
+      "epoch": 0.02081232236178601,
+      "learning_rate": 0.0001958393692124324,
+      "loss": 0.9506,
+      "mean_token_accuracy": 0.7309067130088807,
+      "num_tokens": 5247853.0,
+      "step": 2270
+    },
+    {
+      "epoch": 0.020904006601265242,
+      "learning_rate": 0.00019582103236453655,
+      "loss": 0.968,
+      "mean_token_accuracy": 0.7199053347110749,
+      "num_tokens": 5270539.0,
+      "step": 2280
+    },
+    {
+      "epoch": 0.020995690840744475,
+      "learning_rate": 0.0001958026955166407,
+      "loss": 0.9228,
+      "mean_token_accuracy": 0.7256809651851654,
+      "num_tokens": 5293157.0,
+      "step": 2290
+    },
+    {
+      "epoch": 0.021087375080223708,
+      "learning_rate": 0.00019578435866874486,
+      "loss": 0.9045,
+      "mean_token_accuracy": 0.7299234926700592,
+      "num_tokens": 5316419.0,
+      "step": 2300
+    },
+    {
+      "epoch": 0.021179059319702945,
+      "learning_rate": 0.000195766021820849,
+      "loss": 0.9354,
+      "mean_token_accuracy": 0.7320650398731232,
+      "num_tokens": 5340590.0,
+      "step": 2310
+    },
+    {
+      "epoch": 0.021270743559182178,
+      "learning_rate": 0.00019574768497295316,
+      "loss": 0.9831,
+      "mean_token_accuracy": 0.7165019750595093,
+      "num_tokens": 5363922.0,
+      "step": 2320
+    },
+    {
+      "epoch": 0.02136242779866141,
+      "learning_rate": 0.00019572934812505733,
+      "loss": 0.9671,
+      "mean_token_accuracy": 0.717933064699173,
+      "num_tokens": 5387419.0,
+      "step": 2330
+    },
+    {
+      "epoch": 0.021454112038140644,
+      "learning_rate": 0.00019571101127716147,
+      "loss": 0.9082,
+      "mean_token_accuracy": 0.7313336968421936,
+      "num_tokens": 5410015.0,
+      "step": 2340
+    },
+    {
+      "epoch": 0.021545796277619877,
+      "learning_rate": 0.00019569267442926563,
+      "loss": 0.9328,
+      "mean_token_accuracy": 0.7289097845554352,
+      "num_tokens": 5432736.0,
+      "step": 2350
+    },
+    {
+      "epoch": 0.02163748051709911,
+      "learning_rate": 0.00019567433758136977,
+      "loss": 0.9688,
+      "mean_token_accuracy": 0.7147701561450959,
+      "num_tokens": 5455943.0,
+      "step": 2360
+    },
+    {
+      "epoch": 0.021729164756578344,
+      "learning_rate": 0.0001956560007334739,
+      "loss": 0.9308,
+      "mean_token_accuracy": 0.7298171997070313,
+      "num_tokens": 5478813.0,
+      "step": 2370
+    },
+    {
+      "epoch": 0.021820848996057577,
+      "learning_rate": 0.00019563766388557807,
+      "loss": 0.9489,
+      "mean_token_accuracy": 0.7274266302585601,
+      "num_tokens": 5502569.0,
+      "step": 2380
+    },
+    {
+      "epoch": 0.02191253323553681,
+      "learning_rate": 0.00019561932703768224,
+      "loss": 0.9286,
+      "mean_token_accuracy": 0.7251901865005493,
+      "num_tokens": 5525763.0,
+      "step": 2390
+    },
+    {
+      "epoch": 0.022004217475016043,
+      "learning_rate": 0.0001956009901897864,
+      "loss": 0.9473,
+      "mean_token_accuracy": 0.727522361278534,
+      "num_tokens": 5548671.0,
+      "step": 2400
+    },
+    {
+      "epoch": 0.022095901714495277,
+      "learning_rate": 0.00019558265334189054,
+      "loss": 0.9852,
+      "mean_token_accuracy": 0.7218357384204864,
+      "num_tokens": 5571555.0,
+      "step": 2410
+    },
+    {
+      "epoch": 0.022187585953974513,
+      "learning_rate": 0.00019556431649399468,
+      "loss": 0.9594,
+      "mean_token_accuracy": 0.7173354685306549,
+      "num_tokens": 5594684.0,
+      "step": 2420
+    },
+    {
+      "epoch": 0.022279270193453746,
+      "learning_rate": 0.00019554597964609885,
+      "loss": 0.9623,
+      "mean_token_accuracy": 0.7192810893058776,
+      "num_tokens": 5617557.0,
+      "step": 2430
+    },
+    {
+      "epoch": 0.02237095443293298,
+      "learning_rate": 0.00019552764279820298,
+      "loss": 0.9095,
+      "mean_token_accuracy": 0.7417798578739166,
+      "num_tokens": 5640621.0,
+      "step": 2440
+    },
+    {
+      "epoch": 0.022462638672412213,
+      "learning_rate": 0.00019550930595030715,
+      "loss": 0.9789,
+      "mean_token_accuracy": 0.718708848953247,
+      "num_tokens": 5663739.0,
+      "step": 2450
+    },
+    {
+      "epoch": 0.022554322911891446,
+      "learning_rate": 0.00019549096910241132,
+      "loss": 0.969,
+      "mean_token_accuracy": 0.7202953457832336,
+      "num_tokens": 5686800.0,
+      "step": 2460
+    },
+    {
+      "epoch": 0.02264600715137068,
+      "learning_rate": 0.00019547263225451545,
+      "loss": 0.9173,
+      "mean_token_accuracy": 0.7371094405651093,
+      "num_tokens": 5710180.0,
+      "step": 2470
+    },
+    {
+      "epoch": 0.022737691390849912,
+      "learning_rate": 0.00019545429540661962,
+      "loss": 0.9098,
+      "mean_token_accuracy": 0.7396210551261901,
+      "num_tokens": 5733142.0,
+      "step": 2480
+    },
+    {
+      "epoch": 0.022829375630329145,
+      "learning_rate": 0.00019543595855872376,
+      "loss": 0.9503,
+      "mean_token_accuracy": 0.723417603969574,
+      "num_tokens": 5755600.0,
+      "step": 2490
+    },
+    {
+      "epoch": 0.02292105986980838,
+      "learning_rate": 0.00019541762171082792,
+      "loss": 0.945,
+      "mean_token_accuracy": 0.7276733875274658,
+      "num_tokens": 5778864.0,
+      "step": 2500
+    },
+    {
+      "epoch": 0.023012744109287612,
+      "learning_rate": 0.00019539928486293206,
+      "loss": 0.942,
+      "mean_token_accuracy": 0.7287339627742767,
+      "num_tokens": 5801942.0,
+      "step": 2510
+    },
+    {
+      "epoch": 0.02310442834876685,
+      "learning_rate": 0.00019538094801503623,
+      "loss": 0.9009,
+      "mean_token_accuracy": 0.7320193171501159,
+      "num_tokens": 5825136.0,
+      "step": 2520
+    },
+    {
+      "epoch": 0.02319611258824608,
+      "learning_rate": 0.0001953626111671404,
+      "loss": 0.9748,
+      "mean_token_accuracy": 0.719618844985962,
+      "num_tokens": 5848051.0,
+      "step": 2530
+    },
+    {
+      "epoch": 0.023287796827725315,
+      "learning_rate": 0.00019534427431924453,
+      "loss": 0.934,
+      "mean_token_accuracy": 0.7233562409877777,
+      "num_tokens": 5872042.0,
+      "step": 2540
+    },
+    {
+      "epoch": 0.023379481067204548,
+      "learning_rate": 0.0001953259374713487,
+      "loss": 0.9327,
+      "mean_token_accuracy": 0.7312821507453918,
+      "num_tokens": 5894234.0,
+      "step": 2550
+    },
+    {
+      "epoch": 0.02347116530668378,
+      "learning_rate": 0.00019530760062345283,
+      "loss": 0.9526,
+      "mean_token_accuracy": 0.7317610204219818,
+      "num_tokens": 5917184.0,
+      "step": 2560
+    },
+    {
+      "epoch": 0.023562849546163014,
+      "learning_rate": 0.00019528926377555697,
+      "loss": 0.9128,
+      "mean_token_accuracy": 0.7336754083633423,
+      "num_tokens": 5941067.0,
+      "step": 2570
+    },
+    {
+      "epoch": 0.023654533785642248,
+      "learning_rate": 0.00019527092692766114,
+      "loss": 0.8943,
+      "mean_token_accuracy": 0.7358475089073181,
+      "num_tokens": 5963855.0,
+      "step": 2580
+    },
+    {
+      "epoch": 0.02374621802512148,
+      "learning_rate": 0.0001952525900797653,
+      "loss": 0.9766,
+      "mean_token_accuracy": 0.7097460091114044,
+      "num_tokens": 5987158.0,
+      "step": 2590
+    },
+    {
+      "epoch": 0.023837902264600714,
+      "learning_rate": 0.00019523425323186947,
+      "loss": 0.9264,
+      "mean_token_accuracy": 0.729276442527771,
+      "num_tokens": 6010071.0,
+      "step": 2600
+    },
+    {
+      "epoch": 0.023929586504079947,
+      "learning_rate": 0.0001952159163839736,
+      "loss": 0.9145,
+      "mean_token_accuracy": 0.7364008963108063,
+      "num_tokens": 6032716.0,
+      "step": 2610
+    },
+    {
+      "epoch": 0.024021270743559184,
+      "learning_rate": 0.00019519757953607775,
+      "loss": 0.9328,
+      "mean_token_accuracy": 0.732559758424759,
+      "num_tokens": 6055494.0,
+      "step": 2620
+    },
+    {
+      "epoch": 0.024112954983038417,
+      "learning_rate": 0.0001951792426881819,
+      "loss": 0.9641,
+      "mean_token_accuracy": 0.7260844290256501,
+      "num_tokens": 6078113.0,
+      "step": 2630
+    },
+    {
+      "epoch": 0.02420463922251765,
+      "learning_rate": 0.00019516090584028605,
+      "loss": 0.9092,
+      "mean_token_accuracy": 0.7366998136043549,
+      "num_tokens": 6101402.0,
+      "step": 2640
+    },
+    {
+      "epoch": 0.024296323461996883,
+      "learning_rate": 0.00019514256899239021,
+      "loss": 0.9108,
+      "mean_token_accuracy": 0.7387801766395569,
+      "num_tokens": 6124010.0,
+      "step": 2650
+    },
+    {
+      "epoch": 0.024388007701476117,
+      "learning_rate": 0.00019512423214449438,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.740670245885849,
+      "num_tokens": 6146406.0,
+      "step": 2660
+    },
+    {
+      "epoch": 0.02447969194095535,
+      "learning_rate": 0.00019510589529659852,
+      "loss": 0.9599,
+      "mean_token_accuracy": 0.725754851102829,
+      "num_tokens": 6170514.0,
+      "step": 2670
+    },
+    {
+      "epoch": 0.024571376180434583,
+      "learning_rate": 0.00019508755844870268,
+      "loss": 0.9334,
+      "mean_token_accuracy": 0.7298601746559144,
+      "num_tokens": 6193378.0,
+      "step": 2680
+    },
+    {
+      "epoch": 0.024663060419913816,
+      "learning_rate": 0.00019506922160080682,
+      "loss": 0.9265,
+      "mean_token_accuracy": 0.7361769676208496,
+      "num_tokens": 6215842.0,
+      "step": 2690
+    },
+    {
+      "epoch": 0.02475474465939305,
+      "learning_rate": 0.000195050884752911,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.7371805787086487,
+      "num_tokens": 6238984.0,
+      "step": 2700
+    },
+    {
+      "epoch": 0.024846428898872282,
+      "learning_rate": 0.00019503254790501513,
+      "loss": 0.9446,
+      "mean_token_accuracy": 0.7247693419456482,
+      "num_tokens": 6262221.0,
+      "step": 2710
+    },
+    {
+      "epoch": 0.02493811313835152,
+      "learning_rate": 0.0001950142110571193,
+      "loss": 0.9376,
+      "mean_token_accuracy": 0.7261334121227264,
+      "num_tokens": 6285174.0,
+      "step": 2720
+    },
+    {
+      "epoch": 0.025029797377830752,
+      "learning_rate": 0.00019499587420922346,
+      "loss": 0.9415,
+      "mean_token_accuracy": 0.7219414234161377,
+      "num_tokens": 6307994.0,
+      "step": 2730
+    },
+    {
+      "epoch": 0.025121481617309985,
+      "learning_rate": 0.0001949775373613276,
+      "loss": 0.9269,
+      "mean_token_accuracy": 0.7352584719657898,
+      "num_tokens": 6331386.0,
+      "step": 2740
+    },
+    {
+      "epoch": 0.02521316585678922,
+      "learning_rate": 0.00019495920051343176,
+      "loss": 0.9408,
+      "mean_token_accuracy": 0.7218868672847748,
+      "num_tokens": 6354377.0,
+      "step": 2750
+    },
+    {
+      "epoch": 0.025304850096268452,
+      "learning_rate": 0.0001949408636655359,
+      "loss": 0.9418,
+      "mean_token_accuracy": 0.7339106857776642,
+      "num_tokens": 6377356.0,
+      "step": 2760
+    },
+    {
+      "epoch": 0.025396534335747685,
+      "learning_rate": 0.00019492252681764004,
+      "loss": 0.9771,
+      "mean_token_accuracy": 0.7172683537006378,
+      "num_tokens": 6400570.0,
+      "step": 2770
+    },
+    {
+      "epoch": 0.025488218575226918,
+      "learning_rate": 0.0001949041899697442,
+      "loss": 0.913,
+      "mean_token_accuracy": 0.7312417685985565,
+      "num_tokens": 6423983.0,
+      "step": 2780
+    },
+    {
+      "epoch": 0.02557990281470615,
+      "learning_rate": 0.00019488585312184837,
+      "loss": 0.9636,
+      "mean_token_accuracy": 0.7262275397777558,
+      "num_tokens": 6446226.0,
+      "step": 2790
+    },
+    {
+      "epoch": 0.025671587054185385,
+      "learning_rate": 0.00019486751627395253,
+      "loss": 0.9491,
+      "mean_token_accuracy": 0.7296705484390259,
+      "num_tokens": 6468335.0,
+      "step": 2800
+    },
+    {
+      "epoch": 0.025763271293664618,
+      "learning_rate": 0.00019484917942605667,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7369650721549987,
+      "num_tokens": 6491446.0,
+      "step": 2810
+    },
+    {
+      "epoch": 0.02585495553314385,
+      "learning_rate": 0.0001948308425781608,
+      "loss": 0.9175,
+      "mean_token_accuracy": 0.7304332315921783,
+      "num_tokens": 6514073.0,
+      "step": 2820
+    },
+    {
+      "epoch": 0.025946639772623088,
+      "learning_rate": 0.00019481250573026498,
+      "loss": 0.9796,
+      "mean_token_accuracy": 0.719984632730484,
+      "num_tokens": 6535888.0,
+      "step": 2830
+    },
+    {
+      "epoch": 0.02603832401210232,
+      "learning_rate": 0.00019479416888236911,
+      "loss": 0.916,
+      "mean_token_accuracy": 0.729135948419571,
+      "num_tokens": 6559120.0,
+      "step": 2840
+    },
+    {
+      "epoch": 0.026130008251581554,
+      "learning_rate": 0.0001947758320344733,
+      "loss": 0.9709,
+      "mean_token_accuracy": 0.7150510847568512,
+      "num_tokens": 6582047.0,
+      "step": 2850
+    },
+    {
+      "epoch": 0.026221692491060787,
+      "learning_rate": 0.00019475749518657744,
+      "loss": 0.9787,
+      "mean_token_accuracy": 0.7159995019435883,
+      "num_tokens": 6605452.0,
+      "step": 2860
+    },
+    {
+      "epoch": 0.02631337673054002,
+      "learning_rate": 0.00019473915833868158,
+      "loss": 0.8902,
+      "mean_token_accuracy": 0.7358322203159332,
+      "num_tokens": 6628820.0,
+      "step": 2870
+    },
+    {
+      "epoch": 0.026405060970019253,
+      "learning_rate": 0.00019472082149078575,
+      "loss": 0.9268,
+      "mean_token_accuracy": 0.7268377065658569,
+      "num_tokens": 6651668.0,
+      "step": 2880
+    },
+    {
+      "epoch": 0.026496745209498487,
+      "learning_rate": 0.0001947024846428899,
+      "loss": 0.8938,
+      "mean_token_accuracy": 0.739132821559906,
+      "num_tokens": 6674632.0,
+      "step": 2890
+    },
+    {
+      "epoch": 0.02658842944897772,
+      "learning_rate": 0.00019468414779499405,
+      "loss": 0.9266,
+      "mean_token_accuracy": 0.7303313732147216,
+      "num_tokens": 6698564.0,
+      "step": 2900
+    },
+    {
+      "epoch": 0.026680113688456953,
+      "learning_rate": 0.0001946658109470982,
+      "loss": 0.9477,
+      "mean_token_accuracy": 0.7183525085449218,
+      "num_tokens": 6721992.0,
+      "step": 2910
+    },
+    {
+      "epoch": 0.026771797927936186,
+      "learning_rate": 0.00019464747409920236,
+      "loss": 0.9278,
+      "mean_token_accuracy": 0.7257976174354553,
+      "num_tokens": 6745472.0,
+      "step": 2920
+    },
+    {
+      "epoch": 0.026863482167415423,
+      "learning_rate": 0.00019462913725130652,
+      "loss": 0.9158,
+      "mean_token_accuracy": 0.7347129762172699,
+      "num_tokens": 6769153.0,
+      "step": 2930
+    },
+    {
+      "epoch": 0.026955166406894656,
+      "learning_rate": 0.00019461080040341066,
+      "loss": 0.9305,
+      "mean_token_accuracy": 0.728382408618927,
+      "num_tokens": 6792791.0,
+      "step": 2940
+    },
+    {
+      "epoch": 0.02704685064637389,
+      "learning_rate": 0.00019459246355551482,
+      "loss": 0.9253,
+      "mean_token_accuracy": 0.7284856557846069,
+      "num_tokens": 6816008.0,
+      "step": 2950
+    },
+    {
+      "epoch": 0.027138534885853122,
+      "learning_rate": 0.00019457412670761896,
+      "loss": 0.942,
+      "mean_token_accuracy": 0.7243313372135163,
+      "num_tokens": 6839554.0,
+      "step": 2960
+    },
+    {
+      "epoch": 0.027230219125332356,
+      "learning_rate": 0.0001945557898597231,
+      "loss": 0.9324,
+      "mean_token_accuracy": 0.7303346991539001,
+      "num_tokens": 6862086.0,
+      "step": 2970
+    },
+    {
+      "epoch": 0.02732190336481159,
+      "learning_rate": 0.0001945374530118273,
+      "loss": 0.9493,
+      "mean_token_accuracy": 0.7256803274154663,
+      "num_tokens": 6884659.0,
+      "step": 2980
+    },
+    {
+      "epoch": 0.027413587604290822,
+      "learning_rate": 0.00019451911616393143,
+      "loss": 0.9152,
+      "mean_token_accuracy": 0.7365429759025574,
+      "num_tokens": 6908688.0,
+      "step": 2990
+    },
+    {
+      "epoch": 0.027505271843770055,
+      "learning_rate": 0.0001945007793160356,
+      "loss": 0.9504,
+      "mean_token_accuracy": 0.7272844731807708,
+      "num_tokens": 6931862.0,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02759695608324929,
+      "learning_rate": 0.00019448244246813974,
+      "loss": 0.9153,
+      "mean_token_accuracy": 0.7273892223834991,
+      "num_tokens": 6955720.0,
+      "step": 3010
+    },
+    {
+      "epoch": 0.02768864032272852,
+      "learning_rate": 0.00019446410562024387,
+      "loss": 0.9306,
+      "mean_token_accuracy": 0.7319213390350342,
+      "num_tokens": 6978669.0,
+      "step": 3020
+    },
+    {
+      "epoch": 0.027780324562207758,
+      "learning_rate": 0.00019444576877234804,
+      "loss": 0.8965,
+      "mean_token_accuracy": 0.732559609413147,
+      "num_tokens": 7001564.0,
+      "step": 3030
+    },
+    {
+      "epoch": 0.02787200880168699,
+      "learning_rate": 0.00019442743192445218,
+      "loss": 0.9654,
+      "mean_token_accuracy": 0.7271079003810883,
+      "num_tokens": 7025252.0,
+      "step": 3040
+    },
+    {
+      "epoch": 0.027963693041166224,
+      "learning_rate": 0.00019440909507655637,
+      "loss": 0.9704,
+      "mean_token_accuracy": 0.7184332847595215,
+      "num_tokens": 7048790.0,
+      "step": 3050
+    },
+    {
+      "epoch": 0.028055377280645458,
+      "learning_rate": 0.0001943907582286605,
+      "loss": 0.9123,
+      "mean_token_accuracy": 0.7326594293117523,
+      "num_tokens": 7072185.0,
+      "step": 3060
+    },
+    {
+      "epoch": 0.02814706152012469,
+      "learning_rate": 0.00019437242138076465,
+      "loss": 0.9122,
+      "mean_token_accuracy": 0.7348964989185334,
+      "num_tokens": 7095313.0,
+      "step": 3070
+    },
+    {
+      "epoch": 0.028238745759603924,
+      "learning_rate": 0.0001943540845328688,
+      "loss": 0.9319,
+      "mean_token_accuracy": 0.7342008352279663,
+      "num_tokens": 7118411.0,
+      "step": 3080
+    },
+    {
+      "epoch": 0.028330429999083157,
+      "learning_rate": 0.00019433574768497295,
+      "loss": 0.9054,
+      "mean_token_accuracy": 0.7365982592105865,
+      "num_tokens": 7140666.0,
+      "step": 3090
+    },
+    {
+      "epoch": 0.02842211423856239,
+      "learning_rate": 0.00019431741083707712,
+      "loss": 0.9356,
+      "mean_token_accuracy": 0.7332864880561829,
+      "num_tokens": 7164803.0,
+      "step": 3100
+    },
+    {
+      "epoch": 0.028513798478041624,
+      "learning_rate": 0.00019429907398918128,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7311400473117828,
+      "num_tokens": 7188525.0,
+      "step": 3110
+    },
+    {
+      "epoch": 0.028605482717520857,
+      "learning_rate": 0.00019428073714128542,
+      "loss": 0.9382,
+      "mean_token_accuracy": 0.7273888885974884,
+      "num_tokens": 7212173.0,
+      "step": 3120
+    },
+    {
+      "epoch": 0.028697166957000093,
+      "learning_rate": 0.00019426240029338959,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.7442355751991272,
+      "num_tokens": 7235299.0,
+      "step": 3130
+    },
+    {
+      "epoch": 0.028788851196479327,
+      "learning_rate": 0.00019424406344549372,
+      "loss": 0.9424,
+      "mean_token_accuracy": 0.7223499357700348,
+      "num_tokens": 7258654.0,
+      "step": 3140
+    },
+    {
+      "epoch": 0.02888053543595856,
+      "learning_rate": 0.0001942257265975979,
+      "loss": 0.9632,
+      "mean_token_accuracy": 0.7214014410972596,
+      "num_tokens": 7282501.0,
+      "step": 3150
+    },
+    {
+      "epoch": 0.028972219675437793,
+      "learning_rate": 0.00019420738974970203,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.7162065744400025,
+      "num_tokens": 7304978.0,
+      "step": 3160
+    },
+    {
+      "epoch": 0.029063903914917026,
+      "learning_rate": 0.00019418905290180617,
+      "loss": 0.932,
+      "mean_token_accuracy": 0.7355128765106201,
+      "num_tokens": 7327854.0,
+      "step": 3170
+    },
+    {
+      "epoch": 0.02915558815439626,
+      "learning_rate": 0.00019417071605391036,
+      "loss": 0.9462,
+      "mean_token_accuracy": 0.7246469616889953,
+      "num_tokens": 7350491.0,
+      "step": 3180
+    },
+    {
+      "epoch": 0.029247272393875492,
+      "learning_rate": 0.0001941523792060145,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.7225529968738555,
+      "num_tokens": 7374164.0,
+      "step": 3190
+    },
+    {
+      "epoch": 0.029338956633354726,
+      "learning_rate": 0.00019413404235811866,
+      "loss": 0.9212,
+      "mean_token_accuracy": 0.7264076113700867,
+      "num_tokens": 7396796.0,
+      "step": 3200
+    },
+    {
+      "epoch": 0.02943064087283396,
+      "learning_rate": 0.0001941157055102228,
+      "loss": 0.9442,
+      "mean_token_accuracy": 0.721256959438324,
+      "num_tokens": 7420750.0,
+      "step": 3210
+    },
+    {
+      "epoch": 0.029522325112313192,
+      "learning_rate": 0.00019409736866232694,
+      "loss": 0.9102,
+      "mean_token_accuracy": 0.7344326734542846,
+      "num_tokens": 7443504.0,
+      "step": 3220
+    },
+    {
+      "epoch": 0.029614009351792425,
+      "learning_rate": 0.0001940790318144311,
+      "loss": 0.9198,
+      "mean_token_accuracy": 0.7297407686710358,
+      "num_tokens": 7466958.0,
+      "step": 3230
+    },
+    {
+      "epoch": 0.029705693591271662,
+      "learning_rate": 0.00019406069496653527,
+      "loss": 0.921,
+      "mean_token_accuracy": 0.7294909179210662,
+      "num_tokens": 7490438.0,
+      "step": 3240
+    },
+    {
+      "epoch": 0.029797377830750895,
+      "learning_rate": 0.00019404235811863944,
+      "loss": 0.9084,
+      "mean_token_accuracy": 0.7329320967197418,
+      "num_tokens": 7512827.0,
+      "step": 3250
+    },
+    {
+      "epoch": 0.029889062070230128,
+      "learning_rate": 0.00019402402127074357,
+      "loss": 0.9329,
+      "mean_token_accuracy": 0.7312842011451721,
+      "num_tokens": 7536334.0,
+      "step": 3260
+    },
+    {
+      "epoch": 0.02998074630970936,
+      "learning_rate": 0.0001940056844228477,
+      "loss": 0.9365,
+      "mean_token_accuracy": 0.7257433235645294,
+      "num_tokens": 7559931.0,
+      "step": 3270
+    },
+    {
+      "epoch": 0.030072430549188595,
+      "learning_rate": 0.00019398734757495188,
+      "loss": 1.0004,
+      "mean_token_accuracy": 0.7128322720527649,
+      "num_tokens": 7584030.0,
+      "step": 3280
+    },
+    {
+      "epoch": 0.030164114788667828,
+      "learning_rate": 0.00019396901072705602,
+      "loss": 0.9296,
+      "mean_token_accuracy": 0.7273230791091919,
+      "num_tokens": 7606954.0,
+      "step": 3290
+    },
+    {
+      "epoch": 0.03025579902814706,
+      "learning_rate": 0.00019395067387916018,
+      "loss": 0.9096,
+      "mean_token_accuracy": 0.7309522807598114,
+      "num_tokens": 7630471.0,
+      "step": 3300
+    },
+    {
+      "epoch": 0.030347483267626294,
+      "learning_rate": 0.00019393233703126435,
+      "loss": 0.9365,
+      "mean_token_accuracy": 0.7274603307247162,
+      "num_tokens": 7654156.0,
+      "step": 3310
+    },
+    {
+      "epoch": 0.030439167507105527,
+      "learning_rate": 0.00019391400018336848,
+      "loss": 0.9204,
+      "mean_token_accuracy": 0.7304625988006592,
+      "num_tokens": 7676993.0,
+      "step": 3320
+    },
+    {
+      "epoch": 0.03053085174658476,
+      "learning_rate": 0.00019389566333547265,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7398621261119842,
+      "num_tokens": 7700705.0,
+      "step": 3330
+    },
+    {
+      "epoch": 0.030622535986063997,
+      "learning_rate": 0.0001938773264875768,
+      "loss": 0.9062,
+      "mean_token_accuracy": 0.7303354918956757,
+      "num_tokens": 7723897.0,
+      "step": 3340
+    },
+    {
+      "epoch": 0.03071422022554323,
+      "learning_rate": 0.00019385898963968095,
+      "loss": 0.9335,
+      "mean_token_accuracy": 0.7265379667282105,
+      "num_tokens": 7747534.0,
+      "step": 3350
+    },
+    {
+      "epoch": 0.030805904465022464,
+      "learning_rate": 0.0001938406527917851,
+      "loss": 0.9404,
+      "mean_token_accuracy": 0.7280780255794526,
+      "num_tokens": 7770833.0,
+      "step": 3360
+    },
+    {
+      "epoch": 0.030897588704501697,
+      "learning_rate": 0.00019382231594388926,
+      "loss": 0.9224,
+      "mean_token_accuracy": 0.7308550953865052,
+      "num_tokens": 7794236.0,
+      "step": 3370
+    },
+    {
+      "epoch": 0.03098927294398093,
+      "learning_rate": 0.00019380397909599342,
+      "loss": 0.8752,
+      "mean_token_accuracy": 0.7406655371189117,
+      "num_tokens": 7817781.0,
+      "step": 3380
+    },
+    {
+      "epoch": 0.031080957183460163,
+      "learning_rate": 0.00019378564224809756,
+      "loss": 0.9681,
+      "mean_token_accuracy": 0.7218480706214905,
+      "num_tokens": 7840338.0,
+      "step": 3390
+    },
+    {
+      "epoch": 0.031172641422939396,
+      "learning_rate": 0.00019376730540020173,
+      "loss": 0.9164,
+      "mean_token_accuracy": 0.7277542173862457,
+      "num_tokens": 7862721.0,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03126432566241863,
+      "learning_rate": 0.00019374896855230587,
+      "loss": 0.9116,
+      "mean_token_accuracy": 0.7325112044811248,
+      "num_tokens": 7885451.0,
+      "step": 3410
+    },
+    {
+      "epoch": 0.031356009901897866,
+      "learning_rate": 0.00019373063170441,
+      "loss": 0.9064,
+      "mean_token_accuracy": 0.7332186162471771,
+      "num_tokens": 7907812.0,
+      "step": 3420
+    },
+    {
+      "epoch": 0.031447694141377096,
+      "learning_rate": 0.00019371229485651417,
+      "loss": 0.9386,
+      "mean_token_accuracy": 0.7319149374961853,
+      "num_tokens": 7931090.0,
+      "step": 3430
+    },
+    {
+      "epoch": 0.03153937838085633,
+      "learning_rate": 0.00019369395800861833,
+      "loss": 0.962,
+      "mean_token_accuracy": 0.7256182253360748,
+      "num_tokens": 7953775.0,
+      "step": 3440
+    },
+    {
+      "epoch": 0.03163106262033556,
+      "learning_rate": 0.0001936756211607225,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.747320145368576,
+      "num_tokens": 7976398.0,
+      "step": 3450
+    },
+    {
+      "epoch": 0.0317227468598148,
+      "learning_rate": 0.00019365728431282664,
+      "loss": 0.9331,
+      "mean_token_accuracy": 0.7238705515861511,
+      "num_tokens": 8000784.0,
+      "step": 3460
+    },
+    {
+      "epoch": 0.03181443109929403,
+      "learning_rate": 0.00019363894746493078,
+      "loss": 0.9115,
+      "mean_token_accuracy": 0.7366018474102021,
+      "num_tokens": 8023750.0,
+      "step": 3470
+    },
+    {
+      "epoch": 0.031906115338773265,
+      "learning_rate": 0.00019362061061703494,
+      "loss": 0.9215,
+      "mean_token_accuracy": 0.7342450380325317,
+      "num_tokens": 8047137.0,
+      "step": 3480
+    },
+    {
+      "epoch": 0.0319977995782525,
+      "learning_rate": 0.00019360227376913908,
+      "loss": 0.9106,
+      "mean_token_accuracy": 0.728040623664856,
+      "num_tokens": 8069307.0,
+      "step": 3490
+    },
+    {
+      "epoch": 0.03208948381773173,
+      "learning_rate": 0.00019358393692124325,
+      "loss": 0.9269,
+      "mean_token_accuracy": 0.7291705071926117,
+      "num_tokens": 8092285.0,
+      "step": 3500
+    },
+    {
+      "epoch": 0.03218116805721097,
+      "learning_rate": 0.0001935656000733474,
+      "loss": 0.9263,
+      "mean_token_accuracy": 0.73028724193573,
+      "num_tokens": 8115420.0,
+      "step": 3510
+    },
+    {
+      "epoch": 0.0322728522966902,
+      "learning_rate": 0.00019354726322545155,
+      "loss": 0.9064,
+      "mean_token_accuracy": 0.7332595527172089,
+      "num_tokens": 8137777.0,
+      "step": 3520
+    },
+    {
+      "epoch": 0.032364536536169435,
+      "learning_rate": 0.00019352892637755571,
+      "loss": 0.8974,
+      "mean_token_accuracy": 0.7386172652244568,
+      "num_tokens": 8161065.0,
+      "step": 3530
+    },
+    {
+      "epoch": 0.032456220775648664,
+      "learning_rate": 0.00019351058952965985,
+      "loss": 0.8913,
+      "mean_token_accuracy": 0.7372443437576294,
+      "num_tokens": 8184175.0,
+      "step": 3540
+    },
+    {
+      "epoch": 0.0325479050151279,
+      "learning_rate": 0.00019349225268176402,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.7191249907016755,
+      "num_tokens": 8206748.0,
+      "step": 3550
+    },
+    {
+      "epoch": 0.03263958925460713,
+      "learning_rate": 0.00019347391583386816,
+      "loss": 0.9263,
+      "mean_token_accuracy": 0.7323255717754364,
+      "num_tokens": 8230621.0,
+      "step": 3560
+    },
+    {
+      "epoch": 0.03273127349408637,
+      "learning_rate": 0.00019345557898597232,
+      "loss": 0.9426,
+      "mean_token_accuracy": 0.7250443637371063,
+      "num_tokens": 8254704.0,
+      "step": 3570
+    },
+    {
+      "epoch": 0.0328229577335656,
+      "learning_rate": 0.0001934372421380765,
+      "loss": 0.8757,
+      "mean_token_accuracy": 0.7412980794906616,
+      "num_tokens": 8278789.0,
+      "step": 3580
+    },
+    {
+      "epoch": 0.032914641973044834,
+      "learning_rate": 0.00019341890529018063,
+      "loss": 0.9462,
+      "mean_token_accuracy": 0.7306847631931305,
+      "num_tokens": 8302665.0,
+      "step": 3590
+    },
+    {
+      "epoch": 0.03300632621252407,
+      "learning_rate": 0.0001934005684422848,
+      "loss": 0.9305,
+      "mean_token_accuracy": 0.7272827684879303,
+      "num_tokens": 8326571.0,
+      "step": 3600
+    },
+    {
+      "epoch": 0.0330980104520033,
+      "learning_rate": 0.00019338223159438893,
+      "loss": 0.9385,
+      "mean_token_accuracy": 0.728124612569809,
+      "num_tokens": 8349752.0,
+      "step": 3610
+    },
+    {
+      "epoch": 0.03318969469148254,
+      "learning_rate": 0.00019336389474649307,
+      "loss": 0.9288,
+      "mean_token_accuracy": 0.7261069536209106,
+      "num_tokens": 8373054.0,
+      "step": 3620
+    },
+    {
+      "epoch": 0.033281378930961766,
+      "learning_rate": 0.00019334555789859723,
+      "loss": 0.9452,
+      "mean_token_accuracy": 0.726563709974289,
+      "num_tokens": 8396409.0,
+      "step": 3630
+    },
+    {
+      "epoch": 0.033373063170441,
+      "learning_rate": 0.0001933272210507014,
+      "loss": 0.9343,
+      "mean_token_accuracy": 0.7256794333457947,
+      "num_tokens": 8419902.0,
+      "step": 3640
+    },
+    {
+      "epoch": 0.03346474740992023,
+      "learning_rate": 0.00019330888420280556,
+      "loss": 0.892,
+      "mean_token_accuracy": 0.7357638061046601,
+      "num_tokens": 8442209.0,
+      "step": 3650
+    },
+    {
+      "epoch": 0.03355643164939947,
+      "learning_rate": 0.0001932905473549097,
+      "loss": 0.9634,
+      "mean_token_accuracy": 0.7277958691120148,
+      "num_tokens": 8465645.0,
+      "step": 3660
+    },
+    {
+      "epoch": 0.0336481158888787,
+      "learning_rate": 0.00019327221050701384,
+      "loss": 0.9424,
+      "mean_token_accuracy": 0.7206253051757813,
+      "num_tokens": 8488580.0,
+      "step": 3670
+    },
+    {
+      "epoch": 0.033739800128357936,
+      "learning_rate": 0.000193253873659118,
+      "loss": 0.9341,
+      "mean_token_accuracy": 0.7325787842273712,
+      "num_tokens": 8512041.0,
+      "step": 3680
+    },
+    {
+      "epoch": 0.033831484367837165,
+      "learning_rate": 0.00019323553681122214,
+      "loss": 0.964,
+      "mean_token_accuracy": 0.7234341681003571,
+      "num_tokens": 8535048.0,
+      "step": 3690
+    },
+    {
+      "epoch": 0.0339231686073164,
+      "learning_rate": 0.0001932171999633263,
+      "loss": 0.9443,
+      "mean_token_accuracy": 0.7267716944217681,
+      "num_tokens": 8558636.0,
+      "step": 3700
+    },
+    {
+      "epoch": 0.03401485284679564,
+      "learning_rate": 0.00019319886311543048,
+      "loss": 0.9254,
+      "mean_token_accuracy": 0.7347033023834229,
+      "num_tokens": 8581147.0,
+      "step": 3710
+    },
+    {
+      "epoch": 0.03410653708627487,
+      "learning_rate": 0.00019318052626753461,
+      "loss": 0.8836,
+      "mean_token_accuracy": 0.7345532178878784,
+      "num_tokens": 8604458.0,
+      "step": 3720
+    },
+    {
+      "epoch": 0.034198221325754105,
+      "learning_rate": 0.00019316218941963878,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.7402276337146759,
+      "num_tokens": 8628067.0,
+      "step": 3730
+    },
+    {
+      "epoch": 0.034289905565233335,
+      "learning_rate": 0.00019314385257174292,
+      "loss": 0.9612,
+      "mean_token_accuracy": 0.7229425013065338,
+      "num_tokens": 8651721.0,
+      "step": 3740
+    },
+    {
+      "epoch": 0.03438158980471257,
+      "learning_rate": 0.00019312551572384708,
+      "loss": 0.9006,
+      "mean_token_accuracy": 0.7286960244178772,
+      "num_tokens": 8674637.0,
+      "step": 3750
+    },
+    {
+      "epoch": 0.0344732740441918,
+      "learning_rate": 0.00019310717887595122,
+      "loss": 0.9431,
+      "mean_token_accuracy": 0.7249775469303131,
+      "num_tokens": 8698077.0,
+      "step": 3760
+    },
+    {
+      "epoch": 0.03456495828367104,
+      "learning_rate": 0.0001930888420280554,
+      "loss": 0.9275,
+      "mean_token_accuracy": 0.729182767868042,
+      "num_tokens": 8720532.0,
+      "step": 3770
+    },
+    {
+      "epoch": 0.03465664252315027,
+      "learning_rate": 0.00019307050518015955,
+      "loss": 0.9447,
+      "mean_token_accuracy": 0.7250973403453826,
+      "num_tokens": 8744014.0,
+      "step": 3780
+    },
+    {
+      "epoch": 0.034748326762629504,
+      "learning_rate": 0.0001930521683322637,
+      "loss": 0.9265,
+      "mean_token_accuracy": 0.7238422811031342,
+      "num_tokens": 8766595.0,
+      "step": 3790
+    },
+    {
+      "epoch": 0.03484001100210874,
+      "learning_rate": 0.00019303383148436786,
+      "loss": 0.9248,
+      "mean_token_accuracy": 0.7350191116333008,
+      "num_tokens": 8790233.0,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03493169524158797,
+      "learning_rate": 0.000193015494636472,
+      "loss": 0.95,
+      "mean_token_accuracy": 0.7251366436481476,
+      "num_tokens": 8813193.0,
+      "step": 3810
+    },
+    {
+      "epoch": 0.03502337948106721,
+      "learning_rate": 0.00019299715778857613,
+      "loss": 0.9388,
+      "mean_token_accuracy": 0.7282186210155487,
+      "num_tokens": 8835927.0,
+      "step": 3820
+    },
+    {
+      "epoch": 0.03511506372054644,
+      "learning_rate": 0.00019297882094068032,
+      "loss": 0.9124,
+      "mean_token_accuracy": 0.7330652952194214,
+      "num_tokens": 8859275.0,
+      "step": 3830
+    },
+    {
+      "epoch": 0.035206747960025674,
+      "learning_rate": 0.00019296048409278446,
+      "loss": 0.9275,
+      "mean_token_accuracy": 0.7347700655460357,
+      "num_tokens": 8883352.0,
+      "step": 3840
+    },
+    {
+      "epoch": 0.0352984321995049,
+      "learning_rate": 0.00019294214724488863,
+      "loss": 0.9137,
+      "mean_token_accuracy": 0.7327137053012848,
+      "num_tokens": 8906655.0,
+      "step": 3850
+    },
+    {
+      "epoch": 0.03539011643898414,
+      "learning_rate": 0.00019292381039699277,
+      "loss": 0.9246,
+      "mean_token_accuracy": 0.7234176814556121,
+      "num_tokens": 8929167.0,
+      "step": 3860
+    },
+    {
+      "epoch": 0.03548180067846337,
+      "learning_rate": 0.0001929054735490969,
+      "loss": 0.9326,
+      "mean_token_accuracy": 0.7283679842948914,
+      "num_tokens": 8952692.0,
+      "step": 3870
+    },
+    {
+      "epoch": 0.035573484917942606,
+      "learning_rate": 0.00019288713670120107,
+      "loss": 0.9383,
+      "mean_token_accuracy": 0.7266233503818512,
+      "num_tokens": 8976287.0,
+      "step": 3880
+    },
+    {
+      "epoch": 0.035665169157421836,
+      "learning_rate": 0.0001928687998533052,
+      "loss": 0.888,
+      "mean_token_accuracy": 0.7429258465766907,
+      "num_tokens": 9001139.0,
+      "step": 3890
+    },
+    {
+      "epoch": 0.03575685339690107,
+      "learning_rate": 0.00019285046300540937,
+      "loss": 0.9509,
+      "mean_token_accuracy": 0.7248917520046234,
+      "num_tokens": 9024293.0,
+      "step": 3900
+    },
+    {
+      "epoch": 0.03584853763638031,
+      "learning_rate": 0.00019283212615751354,
+      "loss": 0.8998,
+      "mean_token_accuracy": 0.7370175182819366,
+      "num_tokens": 9048108.0,
+      "step": 3910
+    },
+    {
+      "epoch": 0.03594022187585954,
+      "learning_rate": 0.00019281378930961768,
+      "loss": 0.9311,
+      "mean_token_accuracy": 0.7286861360073089,
+      "num_tokens": 9071283.0,
+      "step": 3920
+    },
+    {
+      "epoch": 0.036031906115338776,
+      "learning_rate": 0.00019279545246172184,
+      "loss": 0.8983,
+      "mean_token_accuracy": 0.7360428929328918,
+      "num_tokens": 9094154.0,
+      "step": 3930
+    },
+    {
+      "epoch": 0.036123590354818005,
+      "learning_rate": 0.00019277711561382598,
+      "loss": 0.9033,
+      "mean_token_accuracy": 0.7331046104431153,
+      "num_tokens": 9116850.0,
+      "step": 3940
+    },
+    {
+      "epoch": 0.03621527459429724,
+      "learning_rate": 0.00019275877876593015,
+      "loss": 0.9409,
+      "mean_token_accuracy": 0.7291643500328064,
+      "num_tokens": 9140874.0,
+      "step": 3950
+    },
+    {
+      "epoch": 0.03630695883377647,
+      "learning_rate": 0.0001927404419180343,
+      "loss": 0.9208,
+      "mean_token_accuracy": 0.7249537229537963,
+      "num_tokens": 9164480.0,
+      "step": 3960
+    },
+    {
+      "epoch": 0.03639864307325571,
+      "learning_rate": 0.00019272210507013845,
+      "loss": 0.9405,
+      "mean_token_accuracy": 0.7252800822257995,
+      "num_tokens": 9187966.0,
+      "step": 3970
+    },
+    {
+      "epoch": 0.03649032731273494,
+      "learning_rate": 0.00019270376822224262,
+      "loss": 0.9088,
+      "mean_token_accuracy": 0.7344141483306885,
+      "num_tokens": 9210424.0,
+      "step": 3980
+    },
+    {
+      "epoch": 0.036582011552214175,
+      "learning_rate": 0.00019268543137434675,
+      "loss": 0.9366,
+      "mean_token_accuracy": 0.7268620252609252,
+      "num_tokens": 9233443.0,
+      "step": 3990
+    },
+    {
+      "epoch": 0.036673695791693404,
+      "learning_rate": 0.00019266709452645092,
+      "loss": 0.9438,
+      "mean_token_accuracy": 0.7300260365009308,
+      "num_tokens": 9255951.0,
+      "step": 4000
+    },
+    {
+      "epoch": 0.03676538003117264,
+      "learning_rate": 0.00019264875767855506,
+      "loss": 0.906,
+      "mean_token_accuracy": 0.7370599687099457,
+      "num_tokens": 9279251.0,
+      "step": 4010
+    },
+    {
+      "epoch": 0.03685706427065188,
+      "learning_rate": 0.0001926304208306592,
+      "loss": 0.9123,
+      "mean_token_accuracy": 0.7281017899513245,
+      "num_tokens": 9303004.0,
+      "step": 4020
+    },
+    {
+      "epoch": 0.03694874851013111,
+      "learning_rate": 0.0001926120839827634,
+      "loss": 0.9501,
+      "mean_token_accuracy": 0.7276595234870911,
+      "num_tokens": 9325422.0,
+      "step": 4030
+    },
+    {
+      "epoch": 0.037040432749610344,
+      "learning_rate": 0.00019259374713486753,
+      "loss": 0.9224,
+      "mean_token_accuracy": 0.7300042688846589,
+      "num_tokens": 9348725.0,
+      "step": 4040
+    },
+    {
+      "epoch": 0.037132116989089574,
+      "learning_rate": 0.0001925754102869717,
+      "loss": 0.9135,
+      "mean_token_accuracy": 0.7325215935707092,
+      "num_tokens": 9372069.0,
+      "step": 4050
+    },
+    {
+      "epoch": 0.03722380122856881,
+      "learning_rate": 0.00019255707343907583,
+      "loss": 0.9213,
+      "mean_token_accuracy": 0.7293917417526246,
+      "num_tokens": 9394823.0,
+      "step": 4060
+    },
+    {
+      "epoch": 0.03731548546804804,
+      "learning_rate": 0.00019253873659117997,
+      "loss": 0.8938,
+      "mean_token_accuracy": 0.7401670753955841,
+      "num_tokens": 9417990.0,
+      "step": 4070
+    },
+    {
+      "epoch": 0.03740716970752728,
+      "learning_rate": 0.00019252039974328414,
+      "loss": 0.9218,
+      "mean_token_accuracy": 0.7370520889759063,
+      "num_tokens": 9441655.0,
+      "step": 4080
+    },
+    {
+      "epoch": 0.03749885394700651,
+      "learning_rate": 0.0001925020628953883,
+      "loss": 0.9203,
+      "mean_token_accuracy": 0.7329012930393219,
+      "num_tokens": 9464852.0,
+      "step": 4090
+    },
+    {
+      "epoch": 0.03759053818648574,
+      "learning_rate": 0.00019248372604749244,
+      "loss": 0.9245,
+      "mean_token_accuracy": 0.7307345628738403,
+      "num_tokens": 9488579.0,
+      "step": 4100
+    },
+    {
+      "epoch": 0.03768222242596498,
+      "learning_rate": 0.0001924653891995966,
+      "loss": 0.9244,
+      "mean_token_accuracy": 0.7316778898239136,
+      "num_tokens": 9512651.0,
+      "step": 4110
+    },
+    {
+      "epoch": 0.03777390666544421,
+      "learning_rate": 0.00019244705235170074,
+      "loss": 0.9031,
+      "mean_token_accuracy": 0.7334048569202423,
+      "num_tokens": 9535376.0,
+      "step": 4120
+    },
+    {
+      "epoch": 0.037865590904923446,
+      "learning_rate": 0.0001924287155038049,
+      "loss": 0.9133,
+      "mean_token_accuracy": 0.7282979667186738,
+      "num_tokens": 9558779.0,
+      "step": 4130
+    },
+    {
+      "epoch": 0.037957275144402676,
+      "learning_rate": 0.00019241037865590905,
+      "loss": 0.9576,
+      "mean_token_accuracy": 0.7185169041156769,
+      "num_tokens": 9581924.0,
+      "step": 4140
+    },
+    {
+      "epoch": 0.03804895938388191,
+      "learning_rate": 0.0001923920418080132,
+      "loss": 0.8996,
+      "mean_token_accuracy": 0.7348777174949646,
+      "num_tokens": 9604941.0,
+      "step": 4150
+    },
+    {
+      "epoch": 0.03814064362336114,
+      "learning_rate": 0.00019237370496011738,
+      "loss": 0.9546,
+      "mean_token_accuracy": 0.7209603607654571,
+      "num_tokens": 9628525.0,
+      "step": 4160
+    },
+    {
+      "epoch": 0.03823232786284038,
+      "learning_rate": 0.00019235536811222152,
+      "loss": 0.9247,
+      "mean_token_accuracy": 0.7281001746654511,
+      "num_tokens": 9652060.0,
+      "step": 4170
+    },
+    {
+      "epoch": 0.03832401210231961,
+      "learning_rate": 0.00019233703126432568,
+      "loss": 0.9508,
+      "mean_token_accuracy": 0.7243199825286866,
+      "num_tokens": 9674742.0,
+      "step": 4180
+    },
+    {
+      "epoch": 0.038415696341798845,
+      "learning_rate": 0.00019231869441642982,
+      "loss": 0.903,
+      "mean_token_accuracy": 0.7305576920509338,
+      "num_tokens": 9698353.0,
+      "step": 4190
+    },
+    {
+      "epoch": 0.038507380581278075,
+      "learning_rate": 0.00019230035756853398,
+      "loss": 0.8963,
+      "mean_token_accuracy": 0.7371667623519897,
+      "num_tokens": 9720928.0,
+      "step": 4200
+    },
+    {
+      "epoch": 0.03859906482075731,
+      "learning_rate": 0.00019228202072063812,
+      "loss": 0.9727,
+      "mean_token_accuracy": 0.7168154358863831,
+      "num_tokens": 9743613.0,
+      "step": 4210
+    },
+    {
+      "epoch": 0.03869074906023655,
+      "learning_rate": 0.0001922636838727423,
+      "loss": 0.8997,
+      "mean_token_accuracy": 0.7314011812210083,
+      "num_tokens": 9766340.0,
+      "step": 4220
+    },
+    {
+      "epoch": 0.03878243329971578,
+      "learning_rate": 0.00019224534702484645,
+      "loss": 0.9387,
+      "mean_token_accuracy": 0.7212896049022675,
+      "num_tokens": 9789134.0,
+      "step": 4230
+    },
+    {
+      "epoch": 0.038874117539195015,
+      "learning_rate": 0.0001922270101769506,
+      "loss": 0.9679,
+      "mean_token_accuracy": 0.7234770894050598,
+      "num_tokens": 9812019.0,
+      "step": 4240
+    },
+    {
+      "epoch": 0.038965801778674244,
+      "learning_rate": 0.00019220867332905476,
+      "loss": 0.9112,
+      "mean_token_accuracy": 0.7349667072296142,
+      "num_tokens": 9835484.0,
+      "step": 4250
+    },
+    {
+      "epoch": 0.03905748601815348,
+      "learning_rate": 0.0001921903364811589,
+      "loss": 0.906,
+      "mean_token_accuracy": 0.7317733287811279,
+      "num_tokens": 9858660.0,
+      "step": 4260
+    },
+    {
+      "epoch": 0.03914917025763271,
+      "learning_rate": 0.00019217199963326303,
+      "loss": 0.9464,
+      "mean_token_accuracy": 0.7197532892227173,
+      "num_tokens": 9881340.0,
+      "step": 4270
+    },
+    {
+      "epoch": 0.03924085449711195,
+      "learning_rate": 0.0001921536627853672,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.7415769517421722,
+      "num_tokens": 9904520.0,
+      "step": 4280
+    },
+    {
+      "epoch": 0.03933253873659118,
+      "learning_rate": 0.00019213532593747136,
+      "loss": 0.9393,
+      "mean_token_accuracy": 0.7288309633731842,
+      "num_tokens": 9927905.0,
+      "step": 4290
+    },
+    {
+      "epoch": 0.039424222976070414,
+      "learning_rate": 0.0001921169890895755,
+      "loss": 0.9101,
+      "mean_token_accuracy": 0.7318473041057587,
+      "num_tokens": 9951267.0,
+      "step": 4300
+    },
+    {
+      "epoch": 0.03951590721554965,
+      "learning_rate": 0.00019209865224167967,
+      "loss": 0.9621,
+      "mean_token_accuracy": 0.724247795343399,
+      "num_tokens": 9974399.0,
+      "step": 4310
+    },
+    {
+      "epoch": 0.03960759145502888,
+      "learning_rate": 0.0001920803153937838,
+      "loss": 0.9643,
+      "mean_token_accuracy": 0.7237609922885895,
+      "num_tokens": 9998102.0,
+      "step": 4320
+    },
+    {
+      "epoch": 0.03969927569450812,
+      "learning_rate": 0.00019206197854588797,
+      "loss": 0.9104,
+      "mean_token_accuracy": 0.7343532621860505,
+      "num_tokens": 10020895.0,
+      "step": 4330
+    },
+    {
+      "epoch": 0.03979095993398735,
+      "learning_rate": 0.0001920436416979921,
+      "loss": 0.9486,
+      "mean_token_accuracy": 0.7267881989479065,
+      "num_tokens": 10044334.0,
+      "step": 4340
+    },
+    {
+      "epoch": 0.03988264417346658,
+      "learning_rate": 0.00019202530485009628,
+      "loss": 0.9248,
+      "mean_token_accuracy": 0.732563316822052,
+      "num_tokens": 10067261.0,
+      "step": 4350
+    },
+    {
+      "epoch": 0.03997432841294581,
+      "learning_rate": 0.00019200696800220044,
+      "loss": 0.9232,
+      "mean_token_accuracy": 0.729229062795639,
+      "num_tokens": 10091244.0,
+      "step": 4360
+    },
+    {
+      "epoch": 0.04006601265242505,
+      "learning_rate": 0.00019198863115430458,
+      "loss": 0.9276,
+      "mean_token_accuracy": 0.7269666433334351,
+      "num_tokens": 10114219.0,
+      "step": 4370
+    },
+    {
+      "epoch": 0.04015769689190428,
+      "learning_rate": 0.00019197029430640875,
+      "loss": 0.9053,
+      "mean_token_accuracy": 0.7333428621292114,
+      "num_tokens": 10136903.0,
+      "step": 4380
+    },
+    {
+      "epoch": 0.040249381131383516,
+      "learning_rate": 0.00019195195745851288,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7340048730373383,
+      "num_tokens": 10159936.0,
+      "step": 4390
+    },
+    {
+      "epoch": 0.040341065370862746,
+      "learning_rate": 0.00019193362061061705,
+      "loss": 0.9274,
+      "mean_token_accuracy": 0.7230561375617981,
+      "num_tokens": 10182391.0,
+      "step": 4400
+    },
+    {
+      "epoch": 0.04043274961034198,
+      "learning_rate": 0.0001919152837627212,
+      "loss": 0.9443,
+      "mean_token_accuracy": 0.7281116962432861,
+      "num_tokens": 10204517.0,
+      "step": 4410
+    },
+    {
+      "epoch": 0.04052443384982122,
+      "learning_rate": 0.00019189694691482535,
+      "loss": 0.9196,
+      "mean_token_accuracy": 0.7343146979808808,
+      "num_tokens": 10228254.0,
+      "step": 4420
+    },
+    {
+      "epoch": 0.04061611808930045,
+      "learning_rate": 0.00019187861006692952,
+      "loss": 0.925,
+      "mean_token_accuracy": 0.7307032346725464,
+      "num_tokens": 10251253.0,
+      "step": 4430
+    },
+    {
+      "epoch": 0.040707802328779685,
+      "learning_rate": 0.00019186027321903366,
+      "loss": 0.9239,
+      "mean_token_accuracy": 0.7301364064216613,
+      "num_tokens": 10274680.0,
+      "step": 4440
+    },
+    {
+      "epoch": 0.040799486568258915,
+      "learning_rate": 0.00019184193637113782,
+      "loss": 0.9009,
+      "mean_token_accuracy": 0.7370919942855835,
+      "num_tokens": 10297380.0,
+      "step": 4450
+    },
+    {
+      "epoch": 0.04089117080773815,
+      "learning_rate": 0.00019182359952324196,
+      "loss": 0.9127,
+      "mean_token_accuracy": 0.7353456616401672,
+      "num_tokens": 10320125.0,
+      "step": 4460
+    },
+    {
+      "epoch": 0.04098285504721738,
+      "learning_rate": 0.0001918052626753461,
+      "loss": 0.9246,
+      "mean_token_accuracy": 0.727199786901474,
+      "num_tokens": 10342990.0,
+      "step": 4470
+    },
+    {
+      "epoch": 0.04107453928669662,
+      "learning_rate": 0.0001917869258274503,
+      "loss": 0.9192,
+      "mean_token_accuracy": 0.7290555357933044,
+      "num_tokens": 10367321.0,
+      "step": 4480
+    },
+    {
+      "epoch": 0.04116622352617585,
+      "learning_rate": 0.00019176858897955443,
+      "loss": 0.9051,
+      "mean_token_accuracy": 0.7330770313739776,
+      "num_tokens": 10390203.0,
+      "step": 4490
+    },
+    {
+      "epoch": 0.041257907765655084,
+      "learning_rate": 0.00019175025213165857,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.7316236853599548,
+      "num_tokens": 10413697.0,
+      "step": 4500
+    },
+    {
+      "epoch": 0.041349592005134314,
+      "learning_rate": 0.00019173191528376273,
+      "loss": 0.9472,
+      "mean_token_accuracy": 0.7224329710006714,
+      "num_tokens": 10437138.0,
+      "step": 4510
+    },
+    {
+      "epoch": 0.04144127624461355,
+      "learning_rate": 0.00019171357843586687,
+      "loss": 0.9147,
+      "mean_token_accuracy": 0.7303707957267761,
+      "num_tokens": 10460102.0,
+      "step": 4520
+    },
+    {
+      "epoch": 0.04153296048409279,
+      "learning_rate": 0.00019169524158797104,
+      "loss": 0.9283,
+      "mean_token_accuracy": 0.7304834306240082,
+      "num_tokens": 10483933.0,
+      "step": 4530
+    },
+    {
+      "epoch": 0.04162464472357202,
+      "learning_rate": 0.00019167690474007518,
+      "loss": 0.9127,
+      "mean_token_accuracy": 0.7311757445335388,
+      "num_tokens": 10507204.0,
+      "step": 4540
+    },
+    {
+      "epoch": 0.041716328963051254,
+      "learning_rate": 0.00019165856789217934,
+      "loss": 0.9085,
+      "mean_token_accuracy": 0.7350265562534333,
+      "num_tokens": 10530801.0,
+      "step": 4550
+    },
+    {
+      "epoch": 0.041808013202530483,
+      "learning_rate": 0.0001916402310442835,
+      "loss": 0.8878,
+      "mean_token_accuracy": 0.7378389596939087,
+      "num_tokens": 10553989.0,
+      "step": 4560
+    },
+    {
+      "epoch": 0.04189969744200972,
+      "learning_rate": 0.00019162189419638764,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.734668105840683,
+      "num_tokens": 10577138.0,
+      "step": 4570
+    },
+    {
+      "epoch": 0.04199138168148895,
+      "learning_rate": 0.0001916035573484918,
+      "loss": 0.8842,
+      "mean_token_accuracy": 0.7381771743297577,
+      "num_tokens": 10601297.0,
+      "step": 4580
+    },
+    {
+      "epoch": 0.042083065920968186,
+      "learning_rate": 0.00019158522050059595,
+      "loss": 0.9061,
+      "mean_token_accuracy": 0.7319147646427154,
+      "num_tokens": 10624206.0,
+      "step": 4590
+    },
+    {
+      "epoch": 0.042174750160447416,
+      "learning_rate": 0.0001915668836527001,
+      "loss": 0.8852,
+      "mean_token_accuracy": 0.7353412926197052,
+      "num_tokens": 10648182.0,
+      "step": 4600
+    },
+    {
+      "epoch": 0.04226643439992665,
+      "learning_rate": 0.00019154854680480428,
+      "loss": 0.9555,
+      "mean_token_accuracy": 0.7207437753677368,
+      "num_tokens": 10670133.0,
+      "step": 4610
+    },
+    {
+      "epoch": 0.04235811863940589,
+      "learning_rate": 0.00019153020995690842,
+      "loss": 0.9053,
+      "mean_token_accuracy": 0.7397615969181061,
+      "num_tokens": 10693264.0,
+      "step": 4620
+    },
+    {
+      "epoch": 0.04244980287888512,
+      "learning_rate": 0.00019151187310901258,
+      "loss": 0.9227,
+      "mean_token_accuracy": 0.7346453845500946,
+      "num_tokens": 10715885.0,
+      "step": 4630
+    },
+    {
+      "epoch": 0.042541487118364356,
+      "learning_rate": 0.00019149353626111672,
+      "loss": 0.8899,
+      "mean_token_accuracy": 0.7341054856777192,
+      "num_tokens": 10739445.0,
+      "step": 4640
+    },
+    {
+      "epoch": 0.042633171357843586,
+      "learning_rate": 0.00019147519941322089,
+      "loss": 0.9314,
+      "mean_token_accuracy": 0.7296093761920929,
+      "num_tokens": 10761786.0,
+      "step": 4650
+    },
+    {
+      "epoch": 0.04272485559732282,
+      "learning_rate": 0.00019145686256532502,
+      "loss": 0.9078,
+      "mean_token_accuracy": 0.7359691679477691,
+      "num_tokens": 10784647.0,
+      "step": 4660
+    },
+    {
+      "epoch": 0.04281653983680205,
+      "learning_rate": 0.00019143852571742916,
+      "loss": 0.9195,
+      "mean_token_accuracy": 0.7348001301288605,
+      "num_tokens": 10808385.0,
+      "step": 4670
+    },
+    {
+      "epoch": 0.04290822407628129,
+      "learning_rate": 0.00019142018886953336,
+      "loss": 0.8954,
+      "mean_token_accuracy": 0.7353730797767639,
+      "num_tokens": 10831964.0,
+      "step": 4680
+    },
+    {
+      "epoch": 0.04299990831576052,
+      "learning_rate": 0.0001914018520216375,
+      "loss": 0.9343,
+      "mean_token_accuracy": 0.7262511432170868,
+      "num_tokens": 10855313.0,
+      "step": 4690
+    },
+    {
+      "epoch": 0.043091592555239755,
+      "learning_rate": 0.00019138351517374163,
+      "loss": 0.9107,
+      "mean_token_accuracy": 0.7381054639816285,
+      "num_tokens": 10878614.0,
+      "step": 4700
+    },
+    {
+      "epoch": 0.043183276794718985,
+      "learning_rate": 0.0001913651783258458,
+      "loss": 0.9033,
+      "mean_token_accuracy": 0.7366665303707123,
+      "num_tokens": 10901215.0,
+      "step": 4710
+    },
+    {
+      "epoch": 0.04327496103419822,
+      "learning_rate": 0.00019134684147794994,
+      "loss": 0.9308,
+      "mean_token_accuracy": 0.7277913689613342,
+      "num_tokens": 10923953.0,
+      "step": 4720
+    },
+    {
+      "epoch": 0.04336664527367746,
+      "learning_rate": 0.0001913285046300541,
+      "loss": 0.9264,
+      "mean_token_accuracy": 0.7296841561794281,
+      "num_tokens": 10947865.0,
+      "step": 4730
+    },
+    {
+      "epoch": 0.04345832951315669,
+      "learning_rate": 0.00019131016778215827,
+      "loss": 0.8988,
+      "mean_token_accuracy": 0.7380661249160767,
+      "num_tokens": 10970854.0,
+      "step": 4740
+    },
+    {
+      "epoch": 0.043550013752635924,
+      "learning_rate": 0.0001912918309342624,
+      "loss": 0.9156,
+      "mean_token_accuracy": 0.7335442006587982,
+      "num_tokens": 10994578.0,
+      "step": 4750
+    },
+    {
+      "epoch": 0.043641697992115154,
+      "learning_rate": 0.00019127349408636657,
+      "loss": 0.9366,
+      "mean_token_accuracy": 0.7311666131019592,
+      "num_tokens": 11018190.0,
+      "step": 4760
+    },
+    {
+      "epoch": 0.04373338223159439,
+      "learning_rate": 0.0001912551572384707,
+      "loss": 0.8864,
+      "mean_token_accuracy": 0.7349504888057709,
+      "num_tokens": 11040756.0,
+      "step": 4770
+    },
+    {
+      "epoch": 0.04382506647107362,
+      "learning_rate": 0.00019123682039057487,
+      "loss": 0.9226,
+      "mean_token_accuracy": 0.7343326091766358,
+      "num_tokens": 11064224.0,
+      "step": 4780
+    },
+    {
+      "epoch": 0.04391675071055286,
+      "learning_rate": 0.000191218483542679,
+      "loss": 0.9116,
+      "mean_token_accuracy": 0.7310377061367035,
+      "num_tokens": 11088507.0,
+      "step": 4790
+    },
+    {
+      "epoch": 0.04400843495003209,
+      "learning_rate": 0.00019120014669478318,
+      "loss": 0.9202,
+      "mean_token_accuracy": 0.7256883978843689,
+      "num_tokens": 11111192.0,
+      "step": 4800
+    },
+    {
+      "epoch": 0.04410011918951132,
+      "learning_rate": 0.00019118180984688734,
+      "loss": 0.9189,
+      "mean_token_accuracy": 0.737100613117218,
+      "num_tokens": 11134725.0,
+      "step": 4810
+    },
+    {
+      "epoch": 0.04419180342899055,
+      "learning_rate": 0.00019116347299899148,
+      "loss": 0.9303,
+      "mean_token_accuracy": 0.7256133377552032,
+      "num_tokens": 11158174.0,
+      "step": 4820
+    },
+    {
+      "epoch": 0.04428348766846979,
+      "learning_rate": 0.00019114513615109565,
+      "loss": 0.8911,
+      "mean_token_accuracy": 0.7342537999153137,
+      "num_tokens": 11181872.0,
+      "step": 4830
+    },
+    {
+      "epoch": 0.044375171907949026,
+      "learning_rate": 0.00019112679930319979,
+      "loss": 0.8922,
+      "mean_token_accuracy": 0.7380224943161011,
+      "num_tokens": 11205167.0,
+      "step": 4840
+    },
+    {
+      "epoch": 0.044466856147428256,
+      "learning_rate": 0.00019110846245530395,
+      "loss": 0.9255,
+      "mean_token_accuracy": 0.7311283588409424,
+      "num_tokens": 11228416.0,
+      "step": 4850
+    },
+    {
+      "epoch": 0.04455854038690749,
+      "learning_rate": 0.0001910901256074081,
+      "loss": 0.8655,
+      "mean_token_accuracy": 0.7448768377304077,
+      "num_tokens": 11252013.0,
+      "step": 4860
+    },
+    {
+      "epoch": 0.04465022462638672,
+      "learning_rate": 0.00019107178875951223,
+      "loss": 0.9079,
+      "mean_token_accuracy": 0.7323541164398193,
+      "num_tokens": 11275455.0,
+      "step": 4870
+    },
+    {
+      "epoch": 0.04474190886586596,
+      "learning_rate": 0.00019105345191161642,
+      "loss": 0.8912,
+      "mean_token_accuracy": 0.7325319647789001,
+      "num_tokens": 11298923.0,
+      "step": 4880
+    },
+    {
+      "epoch": 0.04483359310534519,
+      "learning_rate": 0.00019103511506372056,
+      "loss": 0.8714,
+      "mean_token_accuracy": 0.7456724584102631,
+      "num_tokens": 11321675.0,
+      "step": 4890
+    },
+    {
+      "epoch": 0.044925277344824426,
+      "learning_rate": 0.0001910167782158247,
+      "loss": 0.8966,
+      "mean_token_accuracy": 0.732540625333786,
+      "num_tokens": 11344222.0,
+      "step": 4900
+    },
+    {
+      "epoch": 0.045016961584303655,
+      "learning_rate": 0.00019099844136792886,
+      "loss": 0.9664,
+      "mean_token_accuracy": 0.7202228009700775,
+      "num_tokens": 11367101.0,
+      "step": 4910
+    },
+    {
+      "epoch": 0.04510864582378289,
+      "learning_rate": 0.000190980104520033,
+      "loss": 0.9181,
+      "mean_token_accuracy": 0.7350785553455352,
+      "num_tokens": 11390164.0,
+      "step": 4920
+    },
+    {
+      "epoch": 0.04520033006326213,
+      "learning_rate": 0.00019096176767213717,
+      "loss": 0.8982,
+      "mean_token_accuracy": 0.7460702061653137,
+      "num_tokens": 11413866.0,
+      "step": 4930
+    },
+    {
+      "epoch": 0.04529201430274136,
+      "learning_rate": 0.00019094343082424133,
+      "loss": 0.9133,
+      "mean_token_accuracy": 0.7296039283275604,
+      "num_tokens": 11437073.0,
+      "step": 4940
+    },
+    {
+      "epoch": 0.045383698542220595,
+      "learning_rate": 0.00019092509397634547,
+      "loss": 0.9185,
+      "mean_token_accuracy": 0.7308907270431518,
+      "num_tokens": 11460639.0,
+      "step": 4950
+    },
+    {
+      "epoch": 0.045475382781699825,
+      "learning_rate": 0.00019090675712844963,
+      "loss": 0.9383,
+      "mean_token_accuracy": 0.7259656071662903,
+      "num_tokens": 11484530.0,
+      "step": 4960
+    },
+    {
+      "epoch": 0.04556706702117906,
+      "learning_rate": 0.00019088842028055377,
+      "loss": 0.909,
+      "mean_token_accuracy": 0.7335390865802764,
+      "num_tokens": 11508137.0,
+      "step": 4970
+    },
+    {
+      "epoch": 0.04565875126065829,
+      "learning_rate": 0.00019087008343265794,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.740622091293335,
+      "num_tokens": 11531716.0,
+      "step": 4980
+    },
+    {
+      "epoch": 0.04575043550013753,
+      "learning_rate": 0.00019085174658476208,
+      "loss": 0.8933,
+      "mean_token_accuracy": 0.7343130946159363,
+      "num_tokens": 11554888.0,
+      "step": 4990
+    },
+    {
+      "epoch": 0.04584211973961676,
+      "learning_rate": 0.00019083340973686624,
+      "loss": 0.8836,
+      "mean_token_accuracy": 0.7344820499420166,
+      "num_tokens": 11578097.0,
+      "step": 5000
+    },
+    {
+      "epoch": 0.045933803979095994,
+      "learning_rate": 0.0001908150728889704,
+      "loss": 0.9201,
+      "mean_token_accuracy": 0.7351269721984863,
+      "num_tokens": 11601339.0,
+      "step": 5010
+    },
+    {
+      "epoch": 0.046025488218575224,
+      "learning_rate": 0.00019079673604107455,
+      "loss": 0.8898,
+      "mean_token_accuracy": 0.7382785975933075,
+      "num_tokens": 11623472.0,
+      "step": 5020
+    },
+    {
+      "epoch": 0.04611717245805446,
+      "learning_rate": 0.0001907783991931787,
+      "loss": 0.9112,
+      "mean_token_accuracy": 0.7351314783096313,
+      "num_tokens": 11645914.0,
+      "step": 5030
+    },
+    {
+      "epoch": 0.0462088566975337,
+      "learning_rate": 0.00019076006234528285,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7480172634124755,
+      "num_tokens": 11669016.0,
+      "step": 5040
+    },
+    {
+      "epoch": 0.04630054093701293,
+      "learning_rate": 0.00019074172549738702,
+      "loss": 0.9295,
+      "mean_token_accuracy": 0.7314546644687653,
+      "num_tokens": 11692104.0,
+      "step": 5050
+    },
+    {
+      "epoch": 0.04639222517649216,
+      "learning_rate": 0.00019072338864949115,
+      "loss": 0.9463,
+      "mean_token_accuracy": 0.7246203124523163,
+      "num_tokens": 11715266.0,
+      "step": 5060
+    },
+    {
+      "epoch": 0.04648390941597139,
+      "learning_rate": 0.00019070505180159532,
+      "loss": 0.941,
+      "mean_token_accuracy": 0.7293759286403656,
+      "num_tokens": 11738485.0,
+      "step": 5070
+    },
+    {
+      "epoch": 0.04657559365545063,
+      "learning_rate": 0.00019068671495369948,
+      "loss": 0.9232,
+      "mean_token_accuracy": 0.7281923830509186,
+      "num_tokens": 11762529.0,
+      "step": 5080
+    },
+    {
+      "epoch": 0.04666727789492986,
+      "learning_rate": 0.00019066837810580362,
+      "loss": 0.8803,
+      "mean_token_accuracy": 0.735435402393341,
+      "num_tokens": 11784620.0,
+      "step": 5090
+    },
+    {
+      "epoch": 0.046758962134409096,
+      "learning_rate": 0.00019065004125790776,
+      "loss": 0.9113,
+      "mean_token_accuracy": 0.7347549855709076,
+      "num_tokens": 11807447.0,
+      "step": 5100
+    },
+    {
+      "epoch": 0.046850646373888326,
+      "learning_rate": 0.00019063170441001193,
+      "loss": 0.8987,
+      "mean_token_accuracy": 0.7351148843765258,
+      "num_tokens": 11830174.0,
+      "step": 5110
+    },
+    {
+      "epoch": 0.04694233061336756,
+      "learning_rate": 0.00019061336756211606,
+      "loss": 0.9323,
+      "mean_token_accuracy": 0.7337264120578766,
+      "num_tokens": 11852919.0,
+      "step": 5120
+    },
+    {
+      "epoch": 0.04703401485284679,
+      "learning_rate": 0.00019059503071422023,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7345613598823547,
+      "num_tokens": 11875515.0,
+      "step": 5130
+    },
+    {
+      "epoch": 0.04712569909232603,
+      "learning_rate": 0.0001905766938663244,
+      "loss": 0.9532,
+      "mean_token_accuracy": 0.7227387726306915,
+      "num_tokens": 11898333.0,
+      "step": 5140
+    },
+    {
+      "epoch": 0.047217383331805265,
+      "learning_rate": 0.00019055835701842853,
+      "loss": 0.9127,
+      "mean_token_accuracy": 0.7374009609222412,
+      "num_tokens": 11921640.0,
+      "step": 5150
+    },
+    {
+      "epoch": 0.047309067571284495,
+      "learning_rate": 0.0001905400201705327,
+      "loss": 0.9097,
+      "mean_token_accuracy": 0.7349705874919892,
+      "num_tokens": 11945069.0,
+      "step": 5160
+    },
+    {
+      "epoch": 0.04740075181076373,
+      "learning_rate": 0.00019052168332263684,
+      "loss": 0.904,
+      "mean_token_accuracy": 0.7365436136722565,
+      "num_tokens": 11968865.0,
+      "step": 5170
+    },
+    {
+      "epoch": 0.04749243605024296,
+      "learning_rate": 0.000190503346474741,
+      "loss": 0.9122,
+      "mean_token_accuracy": 0.7329197347164154,
+      "num_tokens": 11991891.0,
+      "step": 5180
+    },
+    {
+      "epoch": 0.0475841202897222,
+      "learning_rate": 0.00019048500962684514,
+      "loss": 0.8986,
+      "mean_token_accuracy": 0.7397706091403962,
+      "num_tokens": 12014467.0,
+      "step": 5190
+    },
+    {
+      "epoch": 0.04767580452920143,
+      "learning_rate": 0.0001904666727789493,
+      "loss": 0.8919,
+      "mean_token_accuracy": 0.7435442924499511,
+      "num_tokens": 12037783.0,
+      "step": 5200
+    },
+    {
+      "epoch": 0.047767488768680665,
+      "learning_rate": 0.00019044833593105347,
+      "loss": 0.8918,
+      "mean_token_accuracy": 0.7346071600914001,
+      "num_tokens": 12060304.0,
+      "step": 5210
+    },
+    {
+      "epoch": 0.047859173008159894,
+      "learning_rate": 0.0001904299990831576,
+      "loss": 0.9051,
+      "mean_token_accuracy": 0.7371912002563477,
+      "num_tokens": 12083174.0,
+      "step": 5220
+    },
+    {
+      "epoch": 0.04795085724763913,
+      "learning_rate": 0.00019041166223526178,
+      "loss": 0.8998,
+      "mean_token_accuracy": 0.7349052250385284,
+      "num_tokens": 12106308.0,
+      "step": 5230
+    },
+    {
+      "epoch": 0.04804254148711837,
+      "learning_rate": 0.00019039332538736591,
+      "loss": 0.9604,
+      "mean_token_accuracy": 0.7219788312911988,
+      "num_tokens": 12129160.0,
+      "step": 5240
+    },
+    {
+      "epoch": 0.0481342257265976,
+      "learning_rate": 0.00019037498853947008,
+      "loss": 0.9377,
+      "mean_token_accuracy": 0.7283720731735229,
+      "num_tokens": 12152124.0,
+      "step": 5250
+    },
+    {
+      "epoch": 0.048225909966076834,
+      "learning_rate": 0.00019035665169157422,
+      "loss": 0.945,
+      "mean_token_accuracy": 0.7313692808151245,
+      "num_tokens": 12175087.0,
+      "step": 5260
+    },
+    {
+      "epoch": 0.048317594205556064,
+      "learning_rate": 0.00019033831484367838,
+      "loss": 0.909,
+      "mean_token_accuracy": 0.7319585144519806,
+      "num_tokens": 12197643.0,
+      "step": 5270
+    },
+    {
+      "epoch": 0.0484092784450353,
+      "learning_rate": 0.00019031997799578255,
+      "loss": 0.9346,
+      "mean_token_accuracy": 0.7274428248405457,
+      "num_tokens": 12220518.0,
+      "step": 5280
+    },
+    {
+      "epoch": 0.04850096268451453,
+      "learning_rate": 0.0001903016411478867,
+      "loss": 0.8683,
+      "mean_token_accuracy": 0.745925348997116,
+      "num_tokens": 12244221.0,
+      "step": 5290
+    },
+    {
+      "epoch": 0.04859264692399377,
+      "learning_rate": 0.00019028330429999083,
+      "loss": 0.9096,
+      "mean_token_accuracy": 0.7328312158584595,
+      "num_tokens": 12267036.0,
+      "step": 5300
+    },
+    {
+      "epoch": 0.048684331163472996,
+      "learning_rate": 0.000190264967452095,
+      "loss": 0.9198,
+      "mean_token_accuracy": 0.7322596430778503,
+      "num_tokens": 12290136.0,
+      "step": 5310
+    },
+    {
+      "epoch": 0.04877601540295223,
+      "learning_rate": 0.00019024663060419913,
+      "loss": 0.8922,
+      "mean_token_accuracy": 0.7314849436283112,
+      "num_tokens": 12312714.0,
+      "step": 5320
+    },
+    {
+      "epoch": 0.04886769964243146,
+      "learning_rate": 0.00019022829375630332,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7430943608283996,
+      "num_tokens": 12336333.0,
+      "step": 5330
+    },
+    {
+      "epoch": 0.0489593838819107,
+      "learning_rate": 0.00019020995690840746,
+      "loss": 0.8905,
+      "mean_token_accuracy": 0.7302204668521881,
+      "num_tokens": 12358377.0,
+      "step": 5340
+    },
+    {
+      "epoch": 0.049051068121389936,
+      "learning_rate": 0.0001901916200605116,
+      "loss": 0.918,
+      "mean_token_accuracy": 0.7341718196868896,
+      "num_tokens": 12381783.0,
+      "step": 5350
+    },
+    {
+      "epoch": 0.049142752360869166,
+      "learning_rate": 0.00019017328321261576,
+      "loss": 0.8849,
+      "mean_token_accuracy": 0.7394325435161591,
+      "num_tokens": 12404907.0,
+      "step": 5360
+    },
+    {
+      "epoch": 0.0492344366003484,
+      "learning_rate": 0.0001901549463647199,
+      "loss": 0.9342,
+      "mean_token_accuracy": 0.7280310750007629,
+      "num_tokens": 12428138.0,
+      "step": 5370
+    },
+    {
+      "epoch": 0.04932612083982763,
+      "learning_rate": 0.00019013660951682407,
+      "loss": 0.9071,
+      "mean_token_accuracy": 0.7335152506828309,
+      "num_tokens": 12452551.0,
+      "step": 5380
+    },
+    {
+      "epoch": 0.04941780507930687,
+      "learning_rate": 0.0001901182726689282,
+      "loss": 0.8968,
+      "mean_token_accuracy": 0.7261455655097961,
+      "num_tokens": 12476343.0,
+      "step": 5390
+    },
+    {
+      "epoch": 0.0495094893187861,
+      "learning_rate": 0.00019009993582103237,
+      "loss": 0.9005,
+      "mean_token_accuracy": 0.7367361962795258,
+      "num_tokens": 12499385.0,
+      "step": 5400
+    },
+    {
+      "epoch": 0.049601173558265335,
+      "learning_rate": 0.00019008159897313654,
+      "loss": 0.9273,
+      "mean_token_accuracy": 0.7294711172580719,
+      "num_tokens": 12523284.0,
+      "step": 5410
+    },
+    {
+      "epoch": 0.049692857797744565,
+      "learning_rate": 0.00019006326212524067,
+      "loss": 0.9013,
+      "mean_token_accuracy": 0.7319343149662018,
+      "num_tokens": 12547085.0,
+      "step": 5420
+    },
+    {
+      "epoch": 0.0497845420372238,
+      "learning_rate": 0.00019004492527734484,
+      "loss": 0.8736,
+      "mean_token_accuracy": 0.7341129004955291,
+      "num_tokens": 12569829.0,
+      "step": 5430
+    },
+    {
+      "epoch": 0.04987622627670304,
+      "learning_rate": 0.00019002658842944898,
+      "loss": 0.9254,
+      "mean_token_accuracy": 0.7253397226333618,
+      "num_tokens": 12592456.0,
+      "step": 5440
+    },
+    {
+      "epoch": 0.04996791051618227,
+      "learning_rate": 0.00019000825158155314,
+      "loss": 0.9164,
+      "mean_token_accuracy": 0.733900386095047,
+      "num_tokens": 12614992.0,
+      "step": 5450
+    },
+    {
+      "epoch": 0.050059594755661505,
+      "learning_rate": 0.0001899899147336573,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.7450060486793518,
+      "num_tokens": 12638048.0,
+      "step": 5460
+    },
+    {
+      "epoch": 0.050151278995140734,
+      "learning_rate": 0.00018997157788576145,
+      "loss": 0.9016,
+      "mean_token_accuracy": 0.7367283046245575,
+      "num_tokens": 12661835.0,
+      "step": 5470
+    },
+    {
+      "epoch": 0.05024296323461997,
+      "learning_rate": 0.0001899532410378656,
+      "loss": 0.9191,
+      "mean_token_accuracy": 0.7345572352409363,
+      "num_tokens": 12685048.0,
+      "step": 5480
+    },
+    {
+      "epoch": 0.0503346474740992,
+      "learning_rate": 0.00018993490418996975,
+      "loss": 0.8806,
+      "mean_token_accuracy": 0.7392894685268402,
+      "num_tokens": 12708402.0,
+      "step": 5490
+    },
+    {
+      "epoch": 0.05042633171357844,
+      "learning_rate": 0.0001899165673420739,
+      "loss": 0.9381,
+      "mean_token_accuracy": 0.7295340657234192,
+      "num_tokens": 12731798.0,
+      "step": 5500
+    },
+    {
+      "epoch": 0.05051801595305767,
+      "learning_rate": 0.00018989823049417806,
+      "loss": 0.9145,
+      "mean_token_accuracy": 0.7341712176799774,
+      "num_tokens": 12755263.0,
+      "step": 5510
+    },
+    {
+      "epoch": 0.050609700192536904,
+      "learning_rate": 0.0001898798936462822,
+      "loss": 0.9494,
+      "mean_token_accuracy": 0.7211965560913086,
+      "num_tokens": 12777560.0,
+      "step": 5520
+    },
+    {
+      "epoch": 0.05070138443201613,
+      "learning_rate": 0.00018986155679838639,
+      "loss": 0.9024,
+      "mean_token_accuracy": 0.7308406412601471,
+      "num_tokens": 12800675.0,
+      "step": 5530
+    },
+    {
+      "epoch": 0.05079306867149537,
+      "learning_rate": 0.00018984321995049052,
+      "loss": 0.9153,
+      "mean_token_accuracy": 0.7297637462615967,
+      "num_tokens": 12823598.0,
+      "step": 5540
+    },
+    {
+      "epoch": 0.05088475291097461,
+      "learning_rate": 0.00018982488310259466,
+      "loss": 0.9173,
+      "mean_token_accuracy": 0.7339243113994598,
+      "num_tokens": 12847125.0,
+      "step": 5550
+    },
+    {
+      "epoch": 0.050976437150453836,
+      "learning_rate": 0.00018980654625469883,
+      "loss": 0.8923,
+      "mean_token_accuracy": 0.7382059693336487,
+      "num_tokens": 12869988.0,
+      "step": 5560
+    },
+    {
+      "epoch": 0.05106812138993307,
+      "learning_rate": 0.00018978820940680297,
+      "loss": 0.8706,
+      "mean_token_accuracy": 0.7482873499393463,
+      "num_tokens": 12892199.0,
+      "step": 5570
+    },
+    {
+      "epoch": 0.0511598056294123,
+      "learning_rate": 0.00018976987255890713,
+      "loss": 0.9112,
+      "mean_token_accuracy": 0.7290290057659149,
+      "num_tokens": 12915562.0,
+      "step": 5580
+    },
+    {
+      "epoch": 0.05125148986889154,
+      "learning_rate": 0.0001897515357110113,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7361009776592254,
+      "num_tokens": 12937961.0,
+      "step": 5590
+    },
+    {
+      "epoch": 0.05134317410837077,
+      "learning_rate": 0.00018973319886311544,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7427524626255035,
+      "num_tokens": 12961230.0,
+      "step": 5600
+    },
+    {
+      "epoch": 0.051434858347850006,
+      "learning_rate": 0.0001897148620152196,
+      "loss": 0.8949,
+      "mean_token_accuracy": 0.7323137044906616,
+      "num_tokens": 12983738.0,
+      "step": 5610
+    },
+    {
+      "epoch": 0.051526542587329235,
+      "learning_rate": 0.00018969652516732374,
+      "loss": 0.9205,
+      "mean_token_accuracy": 0.7301993727684021,
+      "num_tokens": 13006233.0,
+      "step": 5620
+    },
+    {
+      "epoch": 0.05161822682680847,
+      "learning_rate": 0.0001896781883194279,
+      "loss": 0.9052,
+      "mean_token_accuracy": 0.7336772859096528,
+      "num_tokens": 13029753.0,
+      "step": 5630
+    },
+    {
+      "epoch": 0.0517099110662877,
+      "learning_rate": 0.00018965985147153204,
+      "loss": 0.8929,
+      "mean_token_accuracy": 0.7403875470161438,
+      "num_tokens": 13053550.0,
+      "step": 5640
+    },
+    {
+      "epoch": 0.05180159530576694,
+      "learning_rate": 0.0001896415146236362,
+      "loss": 0.935,
+      "mean_token_accuracy": 0.725757896900177,
+      "num_tokens": 13075579.0,
+      "step": 5650
+    },
+    {
+      "epoch": 0.051893279545246175,
+      "learning_rate": 0.00018962317777574037,
+      "loss": 0.9334,
+      "mean_token_accuracy": 0.7271854400634765,
+      "num_tokens": 13099007.0,
+      "step": 5660
+    },
+    {
+      "epoch": 0.051984963784725405,
+      "learning_rate": 0.0001896048409278445,
+      "loss": 0.9228,
+      "mean_token_accuracy": 0.7263003051280975,
+      "num_tokens": 13122262.0,
+      "step": 5670
+    },
+    {
+      "epoch": 0.05207664802420464,
+      "learning_rate": 0.00018958650407994868,
+      "loss": 0.9106,
+      "mean_token_accuracy": 0.7385031998157501,
+      "num_tokens": 13144882.0,
+      "step": 5680
+    },
+    {
+      "epoch": 0.05216833226368387,
+      "learning_rate": 0.00018956816723205282,
+      "loss": 0.9616,
+      "mean_token_accuracy": 0.723107373714447,
+      "num_tokens": 13168316.0,
+      "step": 5690
+    },
+    {
+      "epoch": 0.05226001650316311,
+      "learning_rate": 0.00018954983038415695,
+      "loss": 0.9367,
+      "mean_token_accuracy": 0.7253693878650666,
+      "num_tokens": 13191038.0,
+      "step": 5700
+    },
+    {
+      "epoch": 0.05235170074264234,
+      "learning_rate": 0.00018953149353626112,
+      "loss": 0.9034,
+      "mean_token_accuracy": 0.7317043125629425,
+      "num_tokens": 13213894.0,
+      "step": 5710
+    },
+    {
+      "epoch": 0.052443384982121574,
+      "learning_rate": 0.00018951315668836529,
+      "loss": 0.9056,
+      "mean_token_accuracy": 0.7340876698493958,
+      "num_tokens": 13237425.0,
+      "step": 5720
+    },
+    {
+      "epoch": 0.052535069221600804,
+      "learning_rate": 0.00018949481984046945,
+      "loss": 0.9071,
+      "mean_token_accuracy": 0.7308786451816559,
+      "num_tokens": 13261397.0,
+      "step": 5730
+    },
+    {
+      "epoch": 0.05262675346108004,
+      "learning_rate": 0.0001894764829925736,
+      "loss": 0.9176,
+      "mean_token_accuracy": 0.7290667653083801,
+      "num_tokens": 13283673.0,
+      "step": 5740
+    },
+    {
+      "epoch": 0.05271843770055928,
+      "learning_rate": 0.00018945814614467773,
+      "loss": 0.9148,
+      "mean_token_accuracy": 0.7335974693298339,
+      "num_tokens": 13306127.0,
+      "step": 5750
+    },
+    {
+      "epoch": 0.05281012194003851,
+      "learning_rate": 0.0001894398092967819,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7394903182983399,
+      "num_tokens": 13328718.0,
+      "step": 5760
+    },
+    {
+      "epoch": 0.052901806179517744,
+      "learning_rate": 0.00018942147244888603,
+      "loss": 0.8896,
+      "mean_token_accuracy": 0.7327750205993653,
+      "num_tokens": 13351679.0,
+      "step": 5770
+    },
+    {
+      "epoch": 0.05299349041899697,
+      "learning_rate": 0.0001894031356009902,
+      "loss": 0.9118,
+      "mean_token_accuracy": 0.731440144777298,
+      "num_tokens": 13374935.0,
+      "step": 5780
+    },
+    {
+      "epoch": 0.05308517465847621,
+      "learning_rate": 0.00018938479875309436,
+      "loss": 0.8995,
+      "mean_token_accuracy": 0.7352271199226379,
+      "num_tokens": 13398259.0,
+      "step": 5790
+    },
+    {
+      "epoch": 0.05317685889795544,
+      "learning_rate": 0.0001893664619051985,
+      "loss": 0.8953,
+      "mean_token_accuracy": 0.737835818529129,
+      "num_tokens": 13421088.0,
+      "step": 5800
+    },
+    {
+      "epoch": 0.053268543137434676,
+      "learning_rate": 0.00018934812505730267,
+      "loss": 0.8999,
+      "mean_token_accuracy": 0.7353157579898835,
+      "num_tokens": 13443602.0,
+      "step": 5810
+    },
+    {
+      "epoch": 0.053360227376913906,
+      "learning_rate": 0.0001893297882094068,
+      "loss": 0.9085,
+      "mean_token_accuracy": 0.7335394084453583,
+      "num_tokens": 13466446.0,
+      "step": 5820
+    },
+    {
+      "epoch": 0.05345191161639314,
+      "learning_rate": 0.00018931145136151097,
+      "loss": 0.9248,
+      "mean_token_accuracy": 0.7322831153869629,
+      "num_tokens": 13489637.0,
+      "step": 5830
+    },
+    {
+      "epoch": 0.05354359585587237,
+      "learning_rate": 0.0001892931145136151,
+      "loss": 0.9242,
+      "mean_token_accuracy": 0.7337723314762116,
+      "num_tokens": 13512273.0,
+      "step": 5840
+    },
+    {
+      "epoch": 0.05363528009535161,
+      "learning_rate": 0.00018927477766571927,
+      "loss": 0.8608,
+      "mean_token_accuracy": 0.7417926669120789,
+      "num_tokens": 13536002.0,
+      "step": 5850
+    },
+    {
+      "epoch": 0.053726964334830846,
+      "learning_rate": 0.00018925644081782344,
+      "loss": 0.8768,
+      "mean_token_accuracy": 0.7452898561954499,
+      "num_tokens": 13559219.0,
+      "step": 5860
+    },
+    {
+      "epoch": 0.053818648574310075,
+      "learning_rate": 0.00018923810396992758,
+      "loss": 0.9125,
+      "mean_token_accuracy": 0.732683789730072,
+      "num_tokens": 13582383.0,
+      "step": 5870
+    },
+    {
+      "epoch": 0.05391033281378931,
+      "learning_rate": 0.00018921976712203174,
+      "loss": 0.8871,
+      "mean_token_accuracy": 0.7352430641651153,
+      "num_tokens": 13605476.0,
+      "step": 5880
+    },
+    {
+      "epoch": 0.05400201705326854,
+      "learning_rate": 0.00018920143027413588,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.7374521434307099,
+      "num_tokens": 13628328.0,
+      "step": 5890
+    },
+    {
+      "epoch": 0.05409370129274778,
+      "learning_rate": 0.00018918309342624002,
+      "loss": 0.8935,
+      "mean_token_accuracy": 0.7319090247154236,
+      "num_tokens": 13651322.0,
+      "step": 5900
+    },
+    {
+      "epoch": 0.05418538553222701,
+      "learning_rate": 0.00018916475657834418,
+      "loss": 0.9031,
+      "mean_token_accuracy": 0.7305229842662812,
+      "num_tokens": 13674324.0,
+      "step": 5910
+    },
+    {
+      "epoch": 0.054277069771706245,
+      "learning_rate": 0.00018914641973044835,
+      "loss": 0.9023,
+      "mean_token_accuracy": 0.733366072177887,
+      "num_tokens": 13696523.0,
+      "step": 5920
+    },
+    {
+      "epoch": 0.054368754011185474,
+      "learning_rate": 0.00018912808288255252,
+      "loss": 0.9223,
+      "mean_token_accuracy": 0.7279128074645996,
+      "num_tokens": 13719611.0,
+      "step": 5930
+    },
+    {
+      "epoch": 0.05446043825066471,
+      "learning_rate": 0.00018910974603465665,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7450281620025635,
+      "num_tokens": 13743060.0,
+      "step": 5940
+    },
+    {
+      "epoch": 0.05455212249014394,
+      "learning_rate": 0.0001890914091867608,
+      "loss": 0.9195,
+      "mean_token_accuracy": 0.7355048298835755,
+      "num_tokens": 13766559.0,
+      "step": 5950
+    },
+    {
+      "epoch": 0.05464380672962318,
+      "learning_rate": 0.00018907307233886496,
+      "loss": 0.8898,
+      "mean_token_accuracy": 0.7425531446933746,
+      "num_tokens": 13789749.0,
+      "step": 5960
+    },
+    {
+      "epoch": 0.054735490969102414,
+      "learning_rate": 0.0001890547354909691,
+      "loss": 0.8948,
+      "mean_token_accuracy": 0.7356940090656281,
+      "num_tokens": 13813212.0,
+      "step": 5970
+    },
+    {
+      "epoch": 0.054827175208581644,
+      "learning_rate": 0.0001890363986430733,
+      "loss": 0.8955,
+      "mean_token_accuracy": 0.7427564442157746,
+      "num_tokens": 13836278.0,
+      "step": 5980
+    },
+    {
+      "epoch": 0.05491885944806088,
+      "learning_rate": 0.00018901806179517743,
+      "loss": 0.8781,
+      "mean_token_accuracy": 0.7427315890789032,
+      "num_tokens": 13858837.0,
+      "step": 5990
+    },
+    {
+      "epoch": 0.05501054368754011,
+      "learning_rate": 0.00018899972494728156,
+      "loss": 0.9124,
+      "mean_token_accuracy": 0.7382710576057434,
+      "num_tokens": 13882252.0,
+      "step": 6000
+    },
+    {
+      "epoch": 0.05510222792701935,
+      "learning_rate": 0.00018898138809938573,
+      "loss": 0.912,
+      "mean_token_accuracy": 0.7302774846553802,
+      "num_tokens": 13905670.0,
+      "step": 6010
+    },
+    {
+      "epoch": 0.05519391216649858,
+      "learning_rate": 0.00018896305125148987,
+      "loss": 0.8934,
+      "mean_token_accuracy": 0.7333192527294159,
+      "num_tokens": 13928573.0,
+      "step": 6020
+    },
+    {
+      "epoch": 0.05528559640597781,
+      "learning_rate": 0.00018894471440359403,
+      "loss": 0.9205,
+      "mean_token_accuracy": 0.7285143196582794,
+      "num_tokens": 13951207.0,
+      "step": 6030
+    },
+    {
+      "epoch": 0.05537728064545704,
+      "learning_rate": 0.00018892637755569817,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7364872217178344,
+      "num_tokens": 13974493.0,
+      "step": 6040
+    },
+    {
+      "epoch": 0.05546896488493628,
+      "learning_rate": 0.00018890804070780234,
+      "loss": 0.8833,
+      "mean_token_accuracy": 0.7420256316661835,
+      "num_tokens": 13996879.0,
+      "step": 6050
+    },
+    {
+      "epoch": 0.055560649124415516,
+      "learning_rate": 0.0001888897038599065,
+      "loss": 0.9403,
+      "mean_token_accuracy": 0.7265816628932953,
+      "num_tokens": 14019520.0,
+      "step": 6060
+    },
+    {
+      "epoch": 0.055652333363894746,
+      "learning_rate": 0.00018887136701201064,
+      "loss": 0.9022,
+      "mean_token_accuracy": 0.7378275334835053,
+      "num_tokens": 14042872.0,
+      "step": 6070
+    },
+    {
+      "epoch": 0.05574401760337398,
+      "learning_rate": 0.0001888530301641148,
+      "loss": 0.896,
+      "mean_token_accuracy": 0.7368328273296356,
+      "num_tokens": 14066443.0,
+      "step": 6080
+    },
+    {
+      "epoch": 0.05583570184285321,
+      "learning_rate": 0.00018883469331621894,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.7459207534790039,
+      "num_tokens": 14089901.0,
+      "step": 6090
+    },
+    {
+      "epoch": 0.05592738608233245,
+      "learning_rate": 0.00018881635646832308,
+      "loss": 0.9238,
+      "mean_token_accuracy": 0.7314928591251373,
+      "num_tokens": 14112575.0,
+      "step": 6100
+    },
+    {
+      "epoch": 0.05601907032181168,
+      "learning_rate": 0.00018879801962042725,
+      "loss": 0.906,
+      "mean_token_accuracy": 0.7364032506942749,
+      "num_tokens": 14135704.0,
+      "step": 6110
+    },
+    {
+      "epoch": 0.056110754561290915,
+      "learning_rate": 0.00018877968277253141,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7461249709129334,
+      "num_tokens": 14158714.0,
+      "step": 6120
+    },
+    {
+      "epoch": 0.056202438800770145,
+      "learning_rate": 0.00018876134592463558,
+      "loss": 0.9115,
+      "mean_token_accuracy": 0.7310883283615113,
+      "num_tokens": 14181618.0,
+      "step": 6130
+    },
+    {
+      "epoch": 0.05629412304024938,
+      "learning_rate": 0.00018874300907673972,
+      "loss": 0.9306,
+      "mean_token_accuracy": 0.7311136484146118,
+      "num_tokens": 14204124.0,
+      "step": 6140
+    },
+    {
+      "epoch": 0.05638580727972861,
+      "learning_rate": 0.00018872467222884386,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.7353039979934692,
+      "num_tokens": 14226147.0,
+      "step": 6150
+    },
+    {
+      "epoch": 0.05647749151920785,
+      "learning_rate": 0.00018870633538094802,
+      "loss": 0.9053,
+      "mean_token_accuracy": 0.7393592476844788,
+      "num_tokens": 14249515.0,
+      "step": 6160
+    },
+    {
+      "epoch": 0.056569175758687085,
+      "learning_rate": 0.00018868799853305216,
+      "loss": 0.8776,
+      "mean_token_accuracy": 0.7368683636188507,
+      "num_tokens": 14272137.0,
+      "step": 6170
+    },
+    {
+      "epoch": 0.056660859998166314,
+      "learning_rate": 0.00018866966168515635,
+      "loss": 0.9018,
+      "mean_token_accuracy": 0.733753764629364,
+      "num_tokens": 14295267.0,
+      "step": 6180
+    },
+    {
+      "epoch": 0.05675254423764555,
+      "learning_rate": 0.0001886513248372605,
+      "loss": 0.9312,
+      "mean_token_accuracy": 0.731525057554245,
+      "num_tokens": 14318197.0,
+      "step": 6190
+    },
+    {
+      "epoch": 0.05684422847712478,
+      "learning_rate": 0.00018863298798936463,
+      "loss": 0.9351,
+      "mean_token_accuracy": 0.7326232314109802,
+      "num_tokens": 14341827.0,
+      "step": 6200
+    },
+    {
+      "epoch": 0.05693591271660402,
+      "learning_rate": 0.0001886146511414688,
+      "loss": 0.9383,
+      "mean_token_accuracy": 0.7253555476665496,
+      "num_tokens": 14364640.0,
+      "step": 6210
+    },
+    {
+      "epoch": 0.05702759695608325,
+      "learning_rate": 0.00018859631429357293,
+      "loss": 0.9435,
+      "mean_token_accuracy": 0.7301494657993317,
+      "num_tokens": 14388414.0,
+      "step": 6220
+    },
+    {
+      "epoch": 0.057119281195562484,
+      "learning_rate": 0.0001885779774456771,
+      "loss": 0.907,
+      "mean_token_accuracy": 0.7335768043994904,
+      "num_tokens": 14411936.0,
+      "step": 6230
+    },
+    {
+      "epoch": 0.057210965435041714,
+      "learning_rate": 0.00018855964059778124,
+      "loss": 0.9083,
+      "mean_token_accuracy": 0.7336957097053528,
+      "num_tokens": 14434648.0,
+      "step": 6240
+    },
+    {
+      "epoch": 0.05730264967452095,
+      "learning_rate": 0.0001885413037498854,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.744956886768341,
+      "num_tokens": 14456764.0,
+      "step": 6250
+    },
+    {
+      "epoch": 0.05739433391400019,
+      "learning_rate": 0.00018852296690198957,
+      "loss": 0.8716,
+      "mean_token_accuracy": 0.7397426307201386,
+      "num_tokens": 14479853.0,
+      "step": 6260
+    },
+    {
+      "epoch": 0.057486018153479417,
+      "learning_rate": 0.0001885046300540937,
+      "loss": 0.8937,
+      "mean_token_accuracy": 0.7402728736400604,
+      "num_tokens": 14503349.0,
+      "step": 6270
+    },
+    {
+      "epoch": 0.05757770239295865,
+      "learning_rate": 0.00018848629320619787,
+      "loss": 0.9142,
+      "mean_token_accuracy": 0.7332817733287811,
+      "num_tokens": 14526076.0,
+      "step": 6280
+    },
+    {
+      "epoch": 0.05766938663243788,
+      "learning_rate": 0.000188467956358302,
+      "loss": 0.9258,
+      "mean_token_accuracy": 0.7274915874004364,
+      "num_tokens": 14549465.0,
+      "step": 6290
+    },
+    {
+      "epoch": 0.05776107087191712,
+      "learning_rate": 0.00018844961951040615,
+      "loss": 0.9064,
+      "mean_token_accuracy": 0.7345609068870544,
+      "num_tokens": 14573191.0,
+      "step": 6300
+    },
+    {
+      "epoch": 0.05785275511139635,
+      "learning_rate": 0.00018843128266251034,
+      "loss": 0.9096,
+      "mean_token_accuracy": 0.734123432636261,
+      "num_tokens": 14596590.0,
+      "step": 6310
+    },
+    {
+      "epoch": 0.057944439350875586,
+      "learning_rate": 0.00018841294581461448,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.7332545578479767,
+      "num_tokens": 14619528.0,
+      "step": 6320
+    },
+    {
+      "epoch": 0.058036123590354816,
+      "learning_rate": 0.00018839460896671864,
+      "loss": 0.8747,
+      "mean_token_accuracy": 0.7394288897514343,
+      "num_tokens": 14643136.0,
+      "step": 6330
+    },
+    {
+      "epoch": 0.05812780782983405,
+      "learning_rate": 0.00018837627211882278,
+      "loss": 0.9034,
+      "mean_token_accuracy": 0.739733350276947,
+      "num_tokens": 14665759.0,
+      "step": 6340
+    },
+    {
+      "epoch": 0.05821949206931328,
+      "learning_rate": 0.00018835793527092692,
+      "loss": 0.8784,
+      "mean_token_accuracy": 0.7444514751434326,
+      "num_tokens": 14688597.0,
+      "step": 6350
+    },
+    {
+      "epoch": 0.05831117630879252,
+      "learning_rate": 0.00018833959842303109,
+      "loss": 0.9192,
+      "mean_token_accuracy": 0.7293457150459289,
+      "num_tokens": 14712072.0,
+      "step": 6360
+    },
+    {
+      "epoch": 0.058402860548271755,
+      "learning_rate": 0.00018832126157513522,
+      "loss": 0.9162,
+      "mean_token_accuracy": 0.7286231815814972,
+      "num_tokens": 14735369.0,
+      "step": 6370
+    },
+    {
+      "epoch": 0.058494544787750985,
+      "learning_rate": 0.00018830292472723942,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7397731244564056,
+      "num_tokens": 14757974.0,
+      "step": 6380
+    },
+    {
+      "epoch": 0.05858622902723022,
+      "learning_rate": 0.00018828458787934356,
+      "loss": 0.9018,
+      "mean_token_accuracy": 0.7344457864761352,
+      "num_tokens": 14780663.0,
+      "step": 6390
+    },
+    {
+      "epoch": 0.05867791326670945,
+      "learning_rate": 0.0001882662510314477,
+      "loss": 0.9278,
+      "mean_token_accuracy": 0.73030526638031,
+      "num_tokens": 14803159.0,
+      "step": 6400
+    },
+    {
+      "epoch": 0.05876959750618869,
+      "learning_rate": 0.00018824791418355186,
+      "loss": 0.8956,
+      "mean_token_accuracy": 0.7391948223114013,
+      "num_tokens": 14826264.0,
+      "step": 6410
+    },
+    {
+      "epoch": 0.05886128174566792,
+      "learning_rate": 0.000188229577335656,
+      "loss": 0.9161,
+      "mean_token_accuracy": 0.7264982461929321,
+      "num_tokens": 14849010.0,
+      "step": 6420
+    },
+    {
+      "epoch": 0.058952965985147154,
+      "learning_rate": 0.00018821124048776016,
+      "loss": 0.893,
+      "mean_token_accuracy": 0.7374953508377076,
+      "num_tokens": 14872077.0,
+      "step": 6430
+    },
+    {
+      "epoch": 0.059044650224626384,
+      "learning_rate": 0.00018819290363986433,
+      "loss": 0.9364,
+      "mean_token_accuracy": 0.7247084021568299,
+      "num_tokens": 14895186.0,
+      "step": 6440
+    },
+    {
+      "epoch": 0.05913633446410562,
+      "learning_rate": 0.00018817456679196847,
+      "loss": 0.8867,
+      "mean_token_accuracy": 0.7435551762580872,
+      "num_tokens": 14917517.0,
+      "step": 6450
+    },
+    {
+      "epoch": 0.05922801870358485,
+      "learning_rate": 0.00018815622994407263,
+      "loss": 0.872,
+      "mean_token_accuracy": 0.7501089930534363,
+      "num_tokens": 14940031.0,
+      "step": 6460
+    },
+    {
+      "epoch": 0.05931970294306409,
+      "learning_rate": 0.00018813789309617677,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7309714913368225,
+      "num_tokens": 14963179.0,
+      "step": 6470
+    },
+    {
+      "epoch": 0.059411387182543324,
+      "learning_rate": 0.00018811955624828094,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7429340958595276,
+      "num_tokens": 14985751.0,
+      "step": 6480
+    },
+    {
+      "epoch": 0.05950307142202255,
+      "learning_rate": 0.00018810121940038507,
+      "loss": 0.8904,
+      "mean_token_accuracy": 0.7367467224597931,
+      "num_tokens": 15008446.0,
+      "step": 6490
+    },
+    {
+      "epoch": 0.05959475566150179,
+      "learning_rate": 0.0001880828825524892,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7398522615432739,
+      "num_tokens": 15031349.0,
+      "step": 6500
+    },
+    {
+      "epoch": 0.05968643990098102,
+      "learning_rate": 0.0001880645457045934,
+      "loss": 0.8951,
+      "mean_token_accuracy": 0.7324127376079559,
+      "num_tokens": 15054551.0,
+      "step": 6510
+    },
+    {
+      "epoch": 0.059778124140460256,
+      "learning_rate": 0.00018804620885669754,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.7435194730758667,
+      "num_tokens": 15078018.0,
+      "step": 6520
+    },
+    {
+      "epoch": 0.059869808379939486,
+      "learning_rate": 0.0001880278720088017,
+      "loss": 0.8842,
+      "mean_token_accuracy": 0.743254816532135,
+      "num_tokens": 15101025.0,
+      "step": 6530
+    },
+    {
+      "epoch": 0.05996149261941872,
+      "learning_rate": 0.00018800953516090585,
+      "loss": 0.8871,
+      "mean_token_accuracy": 0.7371959745883941,
+      "num_tokens": 15124153.0,
+      "step": 6540
+    },
+    {
+      "epoch": 0.06005317685889795,
+      "learning_rate": 0.00018799119831300999,
+      "loss": 0.8699,
+      "mean_token_accuracy": 0.7455730557441711,
+      "num_tokens": 15147110.0,
+      "step": 6550
+    },
+    {
+      "epoch": 0.06014486109837719,
+      "learning_rate": 0.00018797286146511415,
+      "loss": 0.9192,
+      "mean_token_accuracy": 0.7300807714462281,
+      "num_tokens": 15170377.0,
+      "step": 6560
+    },
+    {
+      "epoch": 0.060236545337856426,
+      "learning_rate": 0.00018795452461721832,
+      "loss": 0.9205,
+      "mean_token_accuracy": 0.7324627101421356,
+      "num_tokens": 15194174.0,
+      "step": 6570
+    },
+    {
+      "epoch": 0.060328229577335656,
+      "learning_rate": 0.00018793618776932248,
+      "loss": 0.875,
+      "mean_token_accuracy": 0.7377755105495453,
+      "num_tokens": 15217608.0,
+      "step": 6580
+    },
+    {
+      "epoch": 0.06041991381681489,
+      "learning_rate": 0.00018791785092142662,
+      "loss": 0.8978,
+      "mean_token_accuracy": 0.7367534399032593,
+      "num_tokens": 15240404.0,
+      "step": 6590
+    },
+    {
+      "epoch": 0.06051159805629412,
+      "learning_rate": 0.00018789951407353076,
+      "loss": 0.9103,
+      "mean_token_accuracy": 0.7309245347976685,
+      "num_tokens": 15262817.0,
+      "step": 6600
+    },
+    {
+      "epoch": 0.06060328229577336,
+      "learning_rate": 0.00018788117722563492,
+      "loss": 0.8727,
+      "mean_token_accuracy": 0.7372923493385315,
+      "num_tokens": 15286113.0,
+      "step": 6610
+    },
+    {
+      "epoch": 0.06069496653525259,
+      "learning_rate": 0.00018786284037773906,
+      "loss": 0.8703,
+      "mean_token_accuracy": 0.7400227010250091,
+      "num_tokens": 15309220.0,
+      "step": 6620
+    },
+    {
+      "epoch": 0.060786650774731825,
+      "learning_rate": 0.00018784450352984323,
+      "loss": 0.9371,
+      "mean_token_accuracy": 0.7229818522930145,
+      "num_tokens": 15332332.0,
+      "step": 6630
+    },
+    {
+      "epoch": 0.060878335014211055,
+      "learning_rate": 0.0001878261666819474,
+      "loss": 0.8735,
+      "mean_token_accuracy": 0.7370070874691009,
+      "num_tokens": 15355692.0,
+      "step": 6640
+    },
+    {
+      "epoch": 0.06097001925369029,
+      "learning_rate": 0.00018780782983405153,
+      "loss": 0.9187,
+      "mean_token_accuracy": 0.7303988337516785,
+      "num_tokens": 15378602.0,
+      "step": 6650
+    },
+    {
+      "epoch": 0.06106170349316952,
+      "learning_rate": 0.0001877894929861557,
+      "loss": 0.8962,
+      "mean_token_accuracy": 0.7402011632919312,
+      "num_tokens": 15403457.0,
+      "step": 6660
+    },
+    {
+      "epoch": 0.06115338773264876,
+      "learning_rate": 0.00018777115613825983,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.7377009212970733,
+      "num_tokens": 15426229.0,
+      "step": 6670
+    },
+    {
+      "epoch": 0.061245071972127994,
+      "learning_rate": 0.000187752819290364,
+      "loss": 0.9219,
+      "mean_token_accuracy": 0.7281396806240081,
+      "num_tokens": 15449562.0,
+      "step": 6680
+    },
+    {
+      "epoch": 0.061336756211607224,
+      "learning_rate": 0.00018773448244246814,
+      "loss": 0.8789,
+      "mean_token_accuracy": 0.7357682764530182,
+      "num_tokens": 15473220.0,
+      "step": 6690
+    },
+    {
+      "epoch": 0.06142844045108646,
+      "learning_rate": 0.0001877161455945723,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7406415164470672,
+      "num_tokens": 15496530.0,
+      "step": 6700
+    },
+    {
+      "epoch": 0.06152012469056569,
+      "learning_rate": 0.00018769780874667647,
+      "loss": 0.8978,
+      "mean_token_accuracy": 0.7423011541366578,
+      "num_tokens": 15519869.0,
+      "step": 6710
+    },
+    {
+      "epoch": 0.06161180893004493,
+      "learning_rate": 0.0001876794718987806,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7424899756908416,
+      "num_tokens": 15542720.0,
+      "step": 6720
+    },
+    {
+      "epoch": 0.06170349316952416,
+      "learning_rate": 0.00018766113505088477,
+      "loss": 0.8759,
+      "mean_token_accuracy": 0.7321468591690063,
+      "num_tokens": 15565691.0,
+      "step": 6730
+    },
+    {
+      "epoch": 0.06179517740900339,
+      "learning_rate": 0.0001876427982029889,
+      "loss": 0.9021,
+      "mean_token_accuracy": 0.7332452654838562,
+      "num_tokens": 15587986.0,
+      "step": 6740
+    },
+    {
+      "epoch": 0.06188686164848262,
+      "learning_rate": 0.00018762446135509305,
+      "loss": 0.9058,
+      "mean_token_accuracy": 0.7323048353195191,
+      "num_tokens": 15611069.0,
+      "step": 6750
+    },
+    {
+      "epoch": 0.06197854588796186,
+      "learning_rate": 0.00018760612450719721,
+      "loss": 0.9216,
+      "mean_token_accuracy": 0.7371990263462067,
+      "num_tokens": 15634139.0,
+      "step": 6760
+    },
+    {
+      "epoch": 0.06207023012744109,
+      "learning_rate": 0.00018758778765930138,
+      "loss": 0.8878,
+      "mean_token_accuracy": 0.7320496499538421,
+      "num_tokens": 15657348.0,
+      "step": 6770
+    },
+    {
+      "epoch": 0.062161914366920326,
+      "learning_rate": 0.00018756945081140555,
+      "loss": 0.9143,
+      "mean_token_accuracy": 0.7285496592521667,
+      "num_tokens": 15680247.0,
+      "step": 6780
+    },
+    {
+      "epoch": 0.06225359860639956,
+      "learning_rate": 0.00018755111396350968,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.7439673006534576,
+      "num_tokens": 15703156.0,
+      "step": 6790
+    },
+    {
+      "epoch": 0.06234528284587879,
+      "learning_rate": 0.00018753277711561382,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7361136078834534,
+      "num_tokens": 15725451.0,
+      "step": 6800
+    },
+    {
+      "epoch": 0.06243696708535803,
+      "learning_rate": 0.000187514440267718,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7394465565681457,
+      "num_tokens": 15748133.0,
+      "step": 6810
+    },
+    {
+      "epoch": 0.06252865132483726,
+      "learning_rate": 0.00018749610341982213,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.7421841084957123,
+      "num_tokens": 15771816.0,
+      "step": 6820
+    },
+    {
+      "epoch": 0.0626203355643165,
+      "learning_rate": 0.00018747776657192632,
+      "loss": 0.9246,
+      "mean_token_accuracy": 0.7252350389957428,
+      "num_tokens": 15795042.0,
+      "step": 6830
+    },
+    {
+      "epoch": 0.06271201980379573,
+      "learning_rate": 0.00018745942972403046,
+      "loss": 0.9047,
+      "mean_token_accuracy": 0.7368955314159393,
+      "num_tokens": 15818614.0,
+      "step": 6840
+    },
+    {
+      "epoch": 0.06280370404327495,
+      "learning_rate": 0.0001874410928761346,
+      "loss": 0.8984,
+      "mean_token_accuracy": 0.7376978814601898,
+      "num_tokens": 15841511.0,
+      "step": 6850
+    },
+    {
+      "epoch": 0.06289538828275419,
+      "learning_rate": 0.00018742275602823876,
+      "loss": 0.9005,
+      "mean_token_accuracy": 0.7304648399353028,
+      "num_tokens": 15864316.0,
+      "step": 6860
+    },
+    {
+      "epoch": 0.06298707252223343,
+      "learning_rate": 0.0001874044191803429,
+      "loss": 0.8921,
+      "mean_token_accuracy": 0.7313158929347991,
+      "num_tokens": 15887153.0,
+      "step": 6870
+    },
+    {
+      "epoch": 0.06307875676171266,
+      "learning_rate": 0.00018738608233244706,
+      "loss": 0.893,
+      "mean_token_accuracy": 0.7359376966953277,
+      "num_tokens": 15910352.0,
+      "step": 6880
+    },
+    {
+      "epoch": 0.0631704410011919,
+      "learning_rate": 0.0001873677454845512,
+      "loss": 0.8977,
+      "mean_token_accuracy": 0.7376785218715668,
+      "num_tokens": 15933915.0,
+      "step": 6890
+    },
+    {
+      "epoch": 0.06326212524067112,
+      "learning_rate": 0.00018734940863665537,
+      "loss": 0.9317,
+      "mean_token_accuracy": 0.7326904177665711,
+      "num_tokens": 15957172.0,
+      "step": 6900
+    },
+    {
+      "epoch": 0.06335380948015036,
+      "learning_rate": 0.00018733107178875953,
+      "loss": 0.8774,
+      "mean_token_accuracy": 0.7328968286514282,
+      "num_tokens": 15980904.0,
+      "step": 6910
+    },
+    {
+      "epoch": 0.0634454937196296,
+      "learning_rate": 0.00018731273494086367,
+      "loss": 0.898,
+      "mean_token_accuracy": 0.7409002304077148,
+      "num_tokens": 16004324.0,
+      "step": 6920
+    },
+    {
+      "epoch": 0.06353717795910883,
+      "learning_rate": 0.00018729439809296784,
+      "loss": 0.8682,
+      "mean_token_accuracy": 0.7439389050006866,
+      "num_tokens": 16026946.0,
+      "step": 6930
+    },
+    {
+      "epoch": 0.06362886219858806,
+      "learning_rate": 0.00018727606124507198,
+      "loss": 0.9173,
+      "mean_token_accuracy": 0.7388446688652038,
+      "num_tokens": 16050303.0,
+      "step": 6940
+    },
+    {
+      "epoch": 0.0637205464380673,
+      "learning_rate": 0.00018725772439717611,
+      "loss": 0.9191,
+      "mean_token_accuracy": 0.7310415506362915,
+      "num_tokens": 16072316.0,
+      "step": 6950
+    },
+    {
+      "epoch": 0.06381223067754653,
+      "learning_rate": 0.0001872393875492803,
+      "loss": 0.8931,
+      "mean_token_accuracy": 0.7374844074249267,
+      "num_tokens": 16095978.0,
+      "step": 6960
+    },
+    {
+      "epoch": 0.06390391491702577,
+      "learning_rate": 0.00018722105070138444,
+      "loss": 0.8632,
+      "mean_token_accuracy": 0.7394248306751251,
+      "num_tokens": 16119124.0,
+      "step": 6970
+    },
+    {
+      "epoch": 0.063995599156505,
+      "learning_rate": 0.0001872027138534886,
+      "loss": 0.9068,
+      "mean_token_accuracy": 0.732264769077301,
+      "num_tokens": 16142278.0,
+      "step": 6980
+    },
+    {
+      "epoch": 0.06408728339598423,
+      "learning_rate": 0.00018718437700559275,
+      "loss": 0.8759,
+      "mean_token_accuracy": 0.7376603662967682,
+      "num_tokens": 16165905.0,
+      "step": 6990
+    },
+    {
+      "epoch": 0.06417896763546346,
+      "learning_rate": 0.0001871660401576969,
+      "loss": 0.9255,
+      "mean_token_accuracy": 0.7317251801490784,
+      "num_tokens": 16189684.0,
+      "step": 7000
+    },
+    {
+      "epoch": 0.0642706518749427,
+      "learning_rate": 0.00018714770330980105,
+      "loss": 0.9138,
+      "mean_token_accuracy": 0.7325927197933197,
+      "num_tokens": 16212901.0,
+      "step": 7010
+    },
+    {
+      "epoch": 0.06436233611442194,
+      "learning_rate": 0.0001871293664619052,
+      "loss": 0.9256,
+      "mean_token_accuracy": 0.7301843285560607,
+      "num_tokens": 16236792.0,
+      "step": 7020
+    },
+    {
+      "epoch": 0.06445402035390116,
+      "learning_rate": 0.00018711102961400938,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7444778919219971,
+      "num_tokens": 16260056.0,
+      "step": 7030
+    },
+    {
+      "epoch": 0.0645457045933804,
+      "learning_rate": 0.00018709269276611352,
+      "loss": 0.9333,
+      "mean_token_accuracy": 0.7293393611907959,
+      "num_tokens": 16282761.0,
+      "step": 7040
+    },
+    {
+      "epoch": 0.06463738883285963,
+      "learning_rate": 0.00018707435591821766,
+      "loss": 0.8976,
+      "mean_token_accuracy": 0.7393335163593292,
+      "num_tokens": 16305315.0,
+      "step": 7050
+    },
+    {
+      "epoch": 0.06472907307233887,
+      "learning_rate": 0.00018705601907032183,
+      "loss": 0.9162,
+      "mean_token_accuracy": 0.7291788518428802,
+      "num_tokens": 16328631.0,
+      "step": 7060
+    },
+    {
+      "epoch": 0.06482075731181809,
+      "learning_rate": 0.00018703768222242596,
+      "loss": 0.8899,
+      "mean_token_accuracy": 0.7410358488559723,
+      "num_tokens": 16351733.0,
+      "step": 7070
+    },
+    {
+      "epoch": 0.06491244155129733,
+      "learning_rate": 0.00018701934537453013,
+      "loss": 0.8874,
+      "mean_token_accuracy": 0.7387078881263733,
+      "num_tokens": 16374963.0,
+      "step": 7080
+    },
+    {
+      "epoch": 0.06500412579077657,
+      "learning_rate": 0.0001870010085266343,
+      "loss": 0.9475,
+      "mean_token_accuracy": 0.7295987010002136,
+      "num_tokens": 16397461.0,
+      "step": 7090
+    },
+    {
+      "epoch": 0.0650958100302558,
+      "learning_rate": 0.00018698267167873843,
+      "loss": 0.8814,
+      "mean_token_accuracy": 0.7374467432498932,
+      "num_tokens": 16420852.0,
+      "step": 7100
+    },
+    {
+      "epoch": 0.06518749426973504,
+      "learning_rate": 0.0001869643348308426,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.739473843574524,
+      "num_tokens": 16444586.0,
+      "step": 7110
+    },
+    {
+      "epoch": 0.06527917850921426,
+      "learning_rate": 0.00018694599798294674,
+      "loss": 0.8925,
+      "mean_token_accuracy": 0.731457096338272,
+      "num_tokens": 16467033.0,
+      "step": 7120
+    },
+    {
+      "epoch": 0.0653708627486935,
+      "learning_rate": 0.0001869276611350509,
+      "loss": 0.9007,
+      "mean_token_accuracy": 0.7407608687877655,
+      "num_tokens": 16489861.0,
+      "step": 7130
+    },
+    {
+      "epoch": 0.06546254698817273,
+      "learning_rate": 0.00018690932428715504,
+      "loss": 0.8853,
+      "mean_token_accuracy": 0.7345590412616729,
+      "num_tokens": 16513864.0,
+      "step": 7140
+    },
+    {
+      "epoch": 0.06555423122765197,
+      "learning_rate": 0.00018689098743925918,
+      "loss": 0.8828,
+      "mean_token_accuracy": 0.7381375849246978,
+      "num_tokens": 16537198.0,
+      "step": 7150
+    },
+    {
+      "epoch": 0.0656459154671312,
+      "learning_rate": 0.00018687265059136337,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.7393272697925568,
+      "num_tokens": 16560335.0,
+      "step": 7160
+    },
+    {
+      "epoch": 0.06573759970661043,
+      "learning_rate": 0.0001868543137434675,
+      "loss": 0.8972,
+      "mean_token_accuracy": 0.736215102672577,
+      "num_tokens": 16583397.0,
+      "step": 7170
+    },
+    {
+      "epoch": 0.06582928394608967,
+      "learning_rate": 0.00018683597689557167,
+      "loss": 0.9127,
+      "mean_token_accuracy": 0.7304023504257202,
+      "num_tokens": 16605837.0,
+      "step": 7180
+    },
+    {
+      "epoch": 0.0659209681855689,
+      "learning_rate": 0.0001868176400476758,
+      "loss": 0.8992,
+      "mean_token_accuracy": 0.735389119386673,
+      "num_tokens": 16629642.0,
+      "step": 7190
+    },
+    {
+      "epoch": 0.06601265242504814,
+      "learning_rate": 0.00018679930319977995,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7340928137302398,
+      "num_tokens": 16652890.0,
+      "step": 7200
+    },
+    {
+      "epoch": 0.06610433666452736,
+      "learning_rate": 0.00018678096635188412,
+      "loss": 0.9363,
+      "mean_token_accuracy": 0.7297744393348694,
+      "num_tokens": 16676579.0,
+      "step": 7210
+    },
+    {
+      "epoch": 0.0661960209040066,
+      "learning_rate": 0.00018676262950398826,
+      "loss": 0.8672,
+      "mean_token_accuracy": 0.7379568219184875,
+      "num_tokens": 16700076.0,
+      "step": 7220
+    },
+    {
+      "epoch": 0.06628770514348584,
+      "learning_rate": 0.00018674429265609245,
+      "loss": 0.9195,
+      "mean_token_accuracy": 0.7359227120876313,
+      "num_tokens": 16724267.0,
+      "step": 7230
+    },
+    {
+      "epoch": 0.06637938938296507,
+      "learning_rate": 0.00018672595580819659,
+      "loss": 0.9032,
+      "mean_token_accuracy": 0.7368239104747772,
+      "num_tokens": 16747543.0,
+      "step": 7240
+    },
+    {
+      "epoch": 0.0664710736224443,
+      "learning_rate": 0.00018670761896030072,
+      "loss": 0.9144,
+      "mean_token_accuracy": 0.7297830283641815,
+      "num_tokens": 16771820.0,
+      "step": 7250
+    },
+    {
+      "epoch": 0.06656275786192353,
+      "learning_rate": 0.0001866892821124049,
+      "loss": 0.8988,
+      "mean_token_accuracy": 0.7404946625232697,
+      "num_tokens": 16795665.0,
+      "step": 7260
+    },
+    {
+      "epoch": 0.06665444210140277,
+      "learning_rate": 0.00018667094526450903,
+      "loss": 0.8852,
+      "mean_token_accuracy": 0.7353927493095398,
+      "num_tokens": 16819389.0,
+      "step": 7270
+    },
+    {
+      "epoch": 0.066746126340882,
+      "learning_rate": 0.0001866526084166132,
+      "loss": 0.8787,
+      "mean_token_accuracy": 0.7369317412376404,
+      "num_tokens": 16842826.0,
+      "step": 7280
+    },
+    {
+      "epoch": 0.06683781058036124,
+      "learning_rate": 0.00018663427156871736,
+      "loss": 0.8929,
+      "mean_token_accuracy": 0.7297127783298493,
+      "num_tokens": 16866166.0,
+      "step": 7290
+    },
+    {
+      "epoch": 0.06692949481984047,
+      "learning_rate": 0.0001866159347208215,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7410301506519318,
+      "num_tokens": 16888677.0,
+      "step": 7300
+    },
+    {
+      "epoch": 0.0670211790593197,
+      "learning_rate": 0.00018659759787292566,
+      "loss": 0.889,
+      "mean_token_accuracy": 0.7322735846042633,
+      "num_tokens": 16912050.0,
+      "step": 7310
+    },
+    {
+      "epoch": 0.06711286329879894,
+      "learning_rate": 0.0001865792610250298,
+      "loss": 0.8856,
+      "mean_token_accuracy": 0.7414264559745789,
+      "num_tokens": 16935508.0,
+      "step": 7320
+    },
+    {
+      "epoch": 0.06720454753827818,
+      "learning_rate": 0.00018656092417713397,
+      "loss": 0.9075,
+      "mean_token_accuracy": 0.7362886309623718,
+      "num_tokens": 16958563.0,
+      "step": 7330
+    },
+    {
+      "epoch": 0.0672962317777574,
+      "learning_rate": 0.0001865425873292381,
+      "loss": 0.915,
+      "mean_token_accuracy": 0.7296898543834687,
+      "num_tokens": 16982051.0,
+      "step": 7340
+    },
+    {
+      "epoch": 0.06738791601723663,
+      "learning_rate": 0.00018652425048134224,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.7311189293861389,
+      "num_tokens": 17004712.0,
+      "step": 7350
+    },
+    {
+      "epoch": 0.06747960025671587,
+      "learning_rate": 0.00018650591363344644,
+      "loss": 0.9058,
+      "mean_token_accuracy": 0.7329726874828338,
+      "num_tokens": 17028050.0,
+      "step": 7360
+    },
+    {
+      "epoch": 0.06757128449619511,
+      "learning_rate": 0.00018648757678555057,
+      "loss": 0.9303,
+      "mean_token_accuracy": 0.7248231410980225,
+      "num_tokens": 17051053.0,
+      "step": 7370
+    },
+    {
+      "epoch": 0.06766296873567433,
+      "learning_rate": 0.00018646923993765474,
+      "loss": 0.8843,
+      "mean_token_accuracy": 0.7385094702243805,
+      "num_tokens": 17073754.0,
+      "step": 7380
+    },
+    {
+      "epoch": 0.06775465297515357,
+      "learning_rate": 0.00018645090308975888,
+      "loss": 0.9123,
+      "mean_token_accuracy": 0.7319236874580384,
+      "num_tokens": 17097349.0,
+      "step": 7390
+    },
+    {
+      "epoch": 0.0678463372146328,
+      "learning_rate": 0.00018643256624186302,
+      "loss": 0.8846,
+      "mean_token_accuracy": 0.7347530663013458,
+      "num_tokens": 17120271.0,
+      "step": 7400
+    },
+    {
+      "epoch": 0.06793802145411204,
+      "learning_rate": 0.00018641422939396718,
+      "loss": 0.9098,
+      "mean_token_accuracy": 0.7334261775016785,
+      "num_tokens": 17143657.0,
+      "step": 7410
+    },
+    {
+      "epoch": 0.06802970569359128,
+      "learning_rate": 0.00018639589254607135,
+      "loss": 0.9247,
+      "mean_token_accuracy": 0.7289936065673828,
+      "num_tokens": 17166438.0,
+      "step": 7420
+    },
+    {
+      "epoch": 0.0681213899330705,
+      "learning_rate": 0.0001863775556981755,
+      "loss": 0.9361,
+      "mean_token_accuracy": 0.7277748346328735,
+      "num_tokens": 17189881.0,
+      "step": 7430
+    },
+    {
+      "epoch": 0.06821307417254974,
+      "learning_rate": 0.00018635921885027965,
+      "loss": 0.8601,
+      "mean_token_accuracy": 0.7468758761882782,
+      "num_tokens": 17213078.0,
+      "step": 7440
+    },
+    {
+      "epoch": 0.06830475841202897,
+      "learning_rate": 0.0001863408820023838,
+      "loss": 0.9185,
+      "mean_token_accuracy": 0.7321870148181915,
+      "num_tokens": 17236890.0,
+      "step": 7450
+    },
+    {
+      "epoch": 0.06839644265150821,
+      "learning_rate": 0.00018632254515448795,
+      "loss": 0.882,
+      "mean_token_accuracy": 0.7418410241603851,
+      "num_tokens": 17259974.0,
+      "step": 7460
+    },
+    {
+      "epoch": 0.06848812689098743,
+      "learning_rate": 0.0001863042083065921,
+      "loss": 0.8628,
+      "mean_token_accuracy": 0.7472536444664002,
+      "num_tokens": 17283370.0,
+      "step": 7470
+    },
+    {
+      "epoch": 0.06857981113046667,
+      "learning_rate": 0.00018628587145869626,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7388462424278259,
+      "num_tokens": 17306312.0,
+      "step": 7480
+    },
+    {
+      "epoch": 0.0686714953699459,
+      "learning_rate": 0.00018626753461080042,
+      "loss": 0.9212,
+      "mean_token_accuracy": 0.7299582719802856,
+      "num_tokens": 17328831.0,
+      "step": 7490
+    },
+    {
+      "epoch": 0.06876317960942514,
+      "learning_rate": 0.00018624919776290456,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7480060636997223,
+      "num_tokens": 17352124.0,
+      "step": 7500
+    },
+    {
+      "epoch": 0.06885486384890438,
+      "learning_rate": 0.00018623086091500873,
+      "loss": 0.919,
+      "mean_token_accuracy": 0.735031658411026,
+      "num_tokens": 17376609.0,
+      "step": 7510
+    },
+    {
+      "epoch": 0.0689465480883836,
+      "learning_rate": 0.00018621252406711287,
+      "loss": 0.8857,
+      "mean_token_accuracy": 0.7383214592933655,
+      "num_tokens": 17399810.0,
+      "step": 7520
+    },
+    {
+      "epoch": 0.06903823232786284,
+      "learning_rate": 0.00018619418721921703,
+      "loss": 0.8489,
+      "mean_token_accuracy": 0.744429212808609,
+      "num_tokens": 17422162.0,
+      "step": 7530
+    },
+    {
+      "epoch": 0.06912991656734208,
+      "learning_rate": 0.00018617585037132117,
+      "loss": 0.8776,
+      "mean_token_accuracy": 0.7415482640266419,
+      "num_tokens": 17445079.0,
+      "step": 7540
+    },
+    {
+      "epoch": 0.06922160080682131,
+      "learning_rate": 0.00018615751352342533,
+      "loss": 0.8773,
+      "mean_token_accuracy": 0.7432332038879395,
+      "num_tokens": 17467641.0,
+      "step": 7550
+    },
+    {
+      "epoch": 0.06931328504630054,
+      "learning_rate": 0.0001861391766755295,
+      "loss": 0.8946,
+      "mean_token_accuracy": 0.7381612658500671,
+      "num_tokens": 17491272.0,
+      "step": 7560
+    },
+    {
+      "epoch": 0.06940496928577977,
+      "learning_rate": 0.00018612083982763364,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.742399537563324,
+      "num_tokens": 17514326.0,
+      "step": 7570
+    },
+    {
+      "epoch": 0.06949665352525901,
+      "learning_rate": 0.0001861025029797378,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.7429358124732971,
+      "num_tokens": 17538380.0,
+      "step": 7580
+    },
+    {
+      "epoch": 0.06958833776473825,
+      "learning_rate": 0.00018608416613184194,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7393517673015595,
+      "num_tokens": 17562048.0,
+      "step": 7590
+    },
+    {
+      "epoch": 0.06968002200421748,
+      "learning_rate": 0.00018606582928394608,
+      "loss": 0.9625,
+      "mean_token_accuracy": 0.7224134266376495,
+      "num_tokens": 17584969.0,
+      "step": 7600
+    },
+    {
+      "epoch": 0.0697717062436967,
+      "learning_rate": 0.00018604749243605025,
+      "loss": 0.9035,
+      "mean_token_accuracy": 0.7296406388282776,
+      "num_tokens": 17608461.0,
+      "step": 7610
+    },
+    {
+      "epoch": 0.06986339048317594,
+      "learning_rate": 0.0001860291555881544,
+      "loss": 0.8754,
+      "mean_token_accuracy": 0.7468557894229889,
+      "num_tokens": 17631323.0,
+      "step": 7620
+    },
+    {
+      "epoch": 0.06995507472265518,
+      "learning_rate": 0.00018601081874025858,
+      "loss": 0.9026,
+      "mean_token_accuracy": 0.7325952231884003,
+      "num_tokens": 17655593.0,
+      "step": 7630
+    },
+    {
+      "epoch": 0.07004675896213441,
+      "learning_rate": 0.00018599248189236271,
+      "loss": 0.8996,
+      "mean_token_accuracy": 0.7371292054653168,
+      "num_tokens": 17678836.0,
+      "step": 7640
+    },
+    {
+      "epoch": 0.07013844320161364,
+      "learning_rate": 0.00018597414504446685,
+      "loss": 0.8808,
+      "mean_token_accuracy": 0.7433316528797149,
+      "num_tokens": 17702229.0,
+      "step": 7650
+    },
+    {
+      "epoch": 0.07023012744109287,
+      "learning_rate": 0.00018595580819657102,
+      "loss": 0.8986,
+      "mean_token_accuracy": 0.7327707171440124,
+      "num_tokens": 17725528.0,
+      "step": 7660
+    },
+    {
+      "epoch": 0.07032181168057211,
+      "learning_rate": 0.00018593747134867516,
+      "loss": 0.885,
+      "mean_token_accuracy": 0.7369406819343567,
+      "num_tokens": 17747838.0,
+      "step": 7670
+    },
+    {
+      "epoch": 0.07041349592005135,
+      "learning_rate": 0.00018591913450077932,
+      "loss": 0.9538,
+      "mean_token_accuracy": 0.7233248591423035,
+      "num_tokens": 17771786.0,
+      "step": 7680
+    },
+    {
+      "epoch": 0.07050518015953057,
+      "learning_rate": 0.0001859007976528835,
+      "loss": 0.9146,
+      "mean_token_accuracy": 0.7280868113040924,
+      "num_tokens": 17794416.0,
+      "step": 7690
+    },
+    {
+      "epoch": 0.0705968643990098,
+      "learning_rate": 0.00018588246080498763,
+      "loss": 0.9179,
+      "mean_token_accuracy": 0.7320180058479309,
+      "num_tokens": 17818764.0,
+      "step": 7700
+    },
+    {
+      "epoch": 0.07068854863848904,
+      "learning_rate": 0.0001858641239570918,
+      "loss": 0.8764,
+      "mean_token_accuracy": 0.7389637172222138,
+      "num_tokens": 17841747.0,
+      "step": 7710
+    },
+    {
+      "epoch": 0.07078023287796828,
+      "learning_rate": 0.00018584578710919593,
+      "loss": 0.9236,
+      "mean_token_accuracy": 0.7238744258880615,
+      "num_tokens": 17864897.0,
+      "step": 7720
+    },
+    {
+      "epoch": 0.07087191711744752,
+      "learning_rate": 0.0001858274502613001,
+      "loss": 0.9008,
+      "mean_token_accuracy": 0.7376737713813781,
+      "num_tokens": 17888390.0,
+      "step": 7730
+    },
+    {
+      "epoch": 0.07096360135692674,
+      "learning_rate": 0.00018580911341340423,
+      "loss": 0.9364,
+      "mean_token_accuracy": 0.7339943230152131,
+      "num_tokens": 17911341.0,
+      "step": 7740
+    },
+    {
+      "epoch": 0.07105528559640598,
+      "learning_rate": 0.0001857907765655084,
+      "loss": 0.8868,
+      "mean_token_accuracy": 0.7426157474517823,
+      "num_tokens": 17933928.0,
+      "step": 7750
+    },
+    {
+      "epoch": 0.07114696983588521,
+      "learning_rate": 0.00018577243971761256,
+      "loss": 0.9084,
+      "mean_token_accuracy": 0.7337997078895568,
+      "num_tokens": 17956464.0,
+      "step": 7760
+    },
+    {
+      "epoch": 0.07123865407536445,
+      "learning_rate": 0.0001857541028697167,
+      "loss": 0.8883,
+      "mean_token_accuracy": 0.7383899211883544,
+      "num_tokens": 17980429.0,
+      "step": 7770
+    },
+    {
+      "epoch": 0.07133033831484367,
+      "learning_rate": 0.00018573576602182087,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7452596187591553,
+      "num_tokens": 18003751.0,
+      "step": 7780
+    },
+    {
+      "epoch": 0.07142202255432291,
+      "learning_rate": 0.000185717429173925,
+      "loss": 0.9186,
+      "mean_token_accuracy": 0.7293454945087433,
+      "num_tokens": 18027433.0,
+      "step": 7790
+    },
+    {
+      "epoch": 0.07151370679380215,
+      "learning_rate": 0.00018569909232602914,
+      "loss": 0.8958,
+      "mean_token_accuracy": 0.7356851279735566,
+      "num_tokens": 18050922.0,
+      "step": 7800
+    },
+    {
+      "epoch": 0.07160539103328138,
+      "learning_rate": 0.00018568075547813334,
+      "loss": 0.8912,
+      "mean_token_accuracy": 0.7396374762058258,
+      "num_tokens": 18074551.0,
+      "step": 7810
+    },
+    {
+      "epoch": 0.07169707527276062,
+      "learning_rate": 0.00018566241863023748,
+      "loss": 0.886,
+      "mean_token_accuracy": 0.7380398035049438,
+      "num_tokens": 18097734.0,
+      "step": 7820
+    },
+    {
+      "epoch": 0.07178875951223984,
+      "learning_rate": 0.00018564408178234164,
+      "loss": 0.9089,
+      "mean_token_accuracy": 0.7365286886692047,
+      "num_tokens": 18120093.0,
+      "step": 7830
+    },
+    {
+      "epoch": 0.07188044375171908,
+      "learning_rate": 0.00018562574493444578,
+      "loss": 0.9603,
+      "mean_token_accuracy": 0.718100905418396,
+      "num_tokens": 18144102.0,
+      "step": 7840
+    },
+    {
+      "epoch": 0.07197212799119831,
+      "learning_rate": 0.00018560740808654992,
+      "loss": 0.888,
+      "mean_token_accuracy": 0.7385085821151733,
+      "num_tokens": 18167554.0,
+      "step": 7850
+    },
+    {
+      "epoch": 0.07206381223067755,
+      "learning_rate": 0.00018558907123865408,
+      "loss": 0.8856,
+      "mean_token_accuracy": 0.7322841823101044,
+      "num_tokens": 18190684.0,
+      "step": 7860
+    },
+    {
+      "epoch": 0.07215549647015677,
+      "learning_rate": 0.00018557073439075822,
+      "loss": 0.9147,
+      "mean_token_accuracy": 0.7345363199710846,
+      "num_tokens": 18212828.0,
+      "step": 7870
+    },
+    {
+      "epoch": 0.07224718070963601,
+      "learning_rate": 0.0001855523975428624,
+      "loss": 0.9369,
+      "mean_token_accuracy": 0.7276594460010528,
+      "num_tokens": 18236208.0,
+      "step": 7880
+    },
+    {
+      "epoch": 0.07233886494911525,
+      "learning_rate": 0.00018553406069496655,
+      "loss": 0.886,
+      "mean_token_accuracy": 0.7395935773849487,
+      "num_tokens": 18259184.0,
+      "step": 7890
+    },
+    {
+      "epoch": 0.07243054918859448,
+      "learning_rate": 0.0001855157238470707,
+      "loss": 0.9038,
+      "mean_token_accuracy": 0.7328357338905335,
+      "num_tokens": 18282023.0,
+      "step": 7900
+    },
+    {
+      "epoch": 0.07252223342807372,
+      "learning_rate": 0.00018549738699917486,
+      "loss": 0.916,
+      "mean_token_accuracy": 0.7275796115398407,
+      "num_tokens": 18304842.0,
+      "step": 7910
+    },
+    {
+      "epoch": 0.07261391766755294,
+      "learning_rate": 0.000185479050151279,
+      "loss": 0.9065,
+      "mean_token_accuracy": 0.7362442374229431,
+      "num_tokens": 18328316.0,
+      "step": 7920
+    },
+    {
+      "epoch": 0.07270560190703218,
+      "learning_rate": 0.00018546071330338316,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7417624533176422,
+      "num_tokens": 18350784.0,
+      "step": 7930
+    },
+    {
+      "epoch": 0.07279728614651142,
+      "learning_rate": 0.00018544237645548733,
+      "loss": 0.9288,
+      "mean_token_accuracy": 0.7282447874546051,
+      "num_tokens": 18373584.0,
+      "step": 7940
+    },
+    {
+      "epoch": 0.07288897038599065,
+      "learning_rate": 0.00018542403960759146,
+      "loss": 0.9354,
+      "mean_token_accuracy": 0.7260125935077667,
+      "num_tokens": 18396471.0,
+      "step": 7950
+    },
+    {
+      "epoch": 0.07298065462546988,
+      "learning_rate": 0.00018540570275969563,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7401220858097076,
+      "num_tokens": 18419692.0,
+      "step": 7960
+    },
+    {
+      "epoch": 0.07307233886494911,
+      "learning_rate": 0.00018538736591179977,
+      "loss": 0.9266,
+      "mean_token_accuracy": 0.7221021354198456,
+      "num_tokens": 18443411.0,
+      "step": 7970
+    },
+    {
+      "epoch": 0.07316402310442835,
+      "learning_rate": 0.00018536902906390393,
+      "loss": 0.8414,
+      "mean_token_accuracy": 0.746176129579544,
+      "num_tokens": 18466740.0,
+      "step": 7980
+    },
+    {
+      "epoch": 0.07325570734390759,
+      "learning_rate": 0.00018535069221600807,
+      "loss": 0.9206,
+      "mean_token_accuracy": 0.7348521947860718,
+      "num_tokens": 18489677.0,
+      "step": 7990
+    },
+    {
+      "epoch": 0.07334739158338681,
+      "learning_rate": 0.0001853323553681122,
+      "loss": 0.8849,
+      "mean_token_accuracy": 0.7342815041542053,
+      "num_tokens": 18512294.0,
+      "step": 8000
+    },
+    {
+      "epoch": 0.07343907582286605,
+      "learning_rate": 0.0001853140185202164,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7435256361961364,
+      "num_tokens": 18535164.0,
+      "step": 8010
+    },
+    {
+      "epoch": 0.07353076006234528,
+      "learning_rate": 0.00018529568167232054,
+      "loss": 0.872,
+      "mean_token_accuracy": 0.7408172130584717,
+      "num_tokens": 18558999.0,
+      "step": 8020
+    },
+    {
+      "epoch": 0.07362244430182452,
+      "learning_rate": 0.0001852773448244247,
+      "loss": 0.868,
+      "mean_token_accuracy": 0.7401656746864319,
+      "num_tokens": 18582912.0,
+      "step": 8030
+    },
+    {
+      "epoch": 0.07371412854130376,
+      "learning_rate": 0.00018525900797652884,
+      "loss": 0.8189,
+      "mean_token_accuracy": 0.7549288332462311,
+      "num_tokens": 18605803.0,
+      "step": 8040
+    },
+    {
+      "epoch": 0.07380581278078298,
+      "learning_rate": 0.00018524067112863298,
+      "loss": 0.9304,
+      "mean_token_accuracy": 0.7264647424221039,
+      "num_tokens": 18628917.0,
+      "step": 8050
+    },
+    {
+      "epoch": 0.07389749702026222,
+      "learning_rate": 0.00018522233428073715,
+      "loss": 0.8898,
+      "mean_token_accuracy": 0.7439450800418854,
+      "num_tokens": 18652810.0,
+      "step": 8060
+    },
+    {
+      "epoch": 0.07398918125974145,
+      "learning_rate": 0.0001852039974328413,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7408682942390442,
+      "num_tokens": 18675968.0,
+      "step": 8070
+    },
+    {
+      "epoch": 0.07408086549922069,
+      "learning_rate": 0.00018518566058494545,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.733596795797348,
+      "num_tokens": 18698822.0,
+      "step": 8080
+    },
+    {
+      "epoch": 0.07417254973869991,
+      "learning_rate": 0.00018516732373704962,
+      "loss": 0.9037,
+      "mean_token_accuracy": 0.7288919508457183,
+      "num_tokens": 18721128.0,
+      "step": 8090
+    },
+    {
+      "epoch": 0.07426423397817915,
+      "learning_rate": 0.00018514898688915375,
+      "loss": 0.8833,
+      "mean_token_accuracy": 0.7353435218334198,
+      "num_tokens": 18744285.0,
+      "step": 8100
+    },
+    {
+      "epoch": 0.07435591821765838,
+      "learning_rate": 0.00018513065004125792,
+      "loss": 0.8756,
+      "mean_token_accuracy": 0.739633196592331,
+      "num_tokens": 18767682.0,
+      "step": 8110
+    },
+    {
+      "epoch": 0.07444760245713762,
+      "learning_rate": 0.00018511231319336206,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7327218413352966,
+      "num_tokens": 18791529.0,
+      "step": 8120
+    },
+    {
+      "epoch": 0.07453928669661686,
+      "learning_rate": 0.00018509397634546622,
+      "loss": 0.8942,
+      "mean_token_accuracy": 0.7419546723365784,
+      "num_tokens": 18814752.0,
+      "step": 8130
+    },
+    {
+      "epoch": 0.07463097093609608,
+      "learning_rate": 0.0001850756394975704,
+      "loss": 0.8967,
+      "mean_token_accuracy": 0.7323192059993744,
+      "num_tokens": 18837598.0,
+      "step": 8140
+    },
+    {
+      "epoch": 0.07472265517557532,
+      "learning_rate": 0.00018505730264967453,
+      "loss": 0.8914,
+      "mean_token_accuracy": 0.7411729276180268,
+      "num_tokens": 18860319.0,
+      "step": 8150
+    },
+    {
+      "epoch": 0.07481433941505455,
+      "learning_rate": 0.0001850389658017787,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7420401394367218,
+      "num_tokens": 18883411.0,
+      "step": 8160
+    },
+    {
+      "epoch": 0.07490602365453379,
+      "learning_rate": 0.00018502062895388283,
+      "loss": 0.9161,
+      "mean_token_accuracy": 0.7286843657493591,
+      "num_tokens": 18905727.0,
+      "step": 8170
+    },
+    {
+      "epoch": 0.07499770789401301,
+      "learning_rate": 0.000185002292105987,
+      "loss": 0.8704,
+      "mean_token_accuracy": 0.7459618389606476,
+      "num_tokens": 18929436.0,
+      "step": 8180
+    },
+    {
+      "epoch": 0.07508939213349225,
+      "learning_rate": 0.00018498395525809114,
+      "loss": 0.9291,
+      "mean_token_accuracy": 0.7256647884845734,
+      "num_tokens": 18952552.0,
+      "step": 8190
+    },
+    {
+      "epoch": 0.07518107637297149,
+      "learning_rate": 0.0001849656184101953,
+      "loss": 0.9214,
+      "mean_token_accuracy": 0.7314865827560425,
+      "num_tokens": 18975090.0,
+      "step": 8200
+    },
+    {
+      "epoch": 0.07527276061245072,
+      "learning_rate": 0.00018494728156229947,
+      "loss": 0.8783,
+      "mean_token_accuracy": 0.7414880692958832,
+      "num_tokens": 18998071.0,
+      "step": 8210
+    },
+    {
+      "epoch": 0.07536444485192996,
+      "learning_rate": 0.0001849289447144036,
+      "loss": 0.8938,
+      "mean_token_accuracy": 0.741072142124176,
+      "num_tokens": 19020958.0,
+      "step": 8220
+    },
+    {
+      "epoch": 0.07545612909140918,
+      "learning_rate": 0.00018491060786650777,
+      "loss": 0.9333,
+      "mean_token_accuracy": 0.7330272257328033,
+      "num_tokens": 19044328.0,
+      "step": 8230
+    },
+    {
+      "epoch": 0.07554781333088842,
+      "learning_rate": 0.0001848922710186119,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7314849197864532,
+      "num_tokens": 19066770.0,
+      "step": 8240
+    },
+    {
+      "epoch": 0.07563949757036766,
+      "learning_rate": 0.00018487393417071605,
+      "loss": 0.9039,
+      "mean_token_accuracy": 0.7377885401248931,
+      "num_tokens": 19090650.0,
+      "step": 8250
+    },
+    {
+      "epoch": 0.07573118180984689,
+      "learning_rate": 0.0001848555973228202,
+      "loss": 0.8775,
+      "mean_token_accuracy": 0.7388021290302277,
+      "num_tokens": 19113998.0,
+      "step": 8260
+    },
+    {
+      "epoch": 0.07582286604932612,
+      "learning_rate": 0.00018483726047492438,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.7420752584934235,
+      "num_tokens": 19137048.0,
+      "step": 8270
+    },
+    {
+      "epoch": 0.07591455028880535,
+      "learning_rate": 0.00018481892362702852,
+      "loss": 0.8743,
+      "mean_token_accuracy": 0.7413016796112061,
+      "num_tokens": 19160849.0,
+      "step": 8280
+    },
+    {
+      "epoch": 0.07600623452828459,
+      "learning_rate": 0.00018480058677913268,
+      "loss": 0.9035,
+      "mean_token_accuracy": 0.7368860065937042,
+      "num_tokens": 19183747.0,
+      "step": 8290
+    },
+    {
+      "epoch": 0.07609791876776383,
+      "learning_rate": 0.00018478224993123682,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.741813462972641,
+      "num_tokens": 19207895.0,
+      "step": 8300
+    },
+    {
+      "epoch": 0.07618960300724305,
+      "learning_rate": 0.00018476391308334098,
+      "loss": 0.8937,
+      "mean_token_accuracy": 0.7364883244037628,
+      "num_tokens": 19231139.0,
+      "step": 8310
+    },
+    {
+      "epoch": 0.07628128724672228,
+      "learning_rate": 0.00018474557623544512,
+      "loss": 0.9097,
+      "mean_token_accuracy": 0.7353035271167755,
+      "num_tokens": 19254845.0,
+      "step": 8320
+    },
+    {
+      "epoch": 0.07637297148620152,
+      "learning_rate": 0.0001847272393875493,
+      "loss": 0.9021,
+      "mean_token_accuracy": 0.7364081740379333,
+      "num_tokens": 19278248.0,
+      "step": 8330
+    },
+    {
+      "epoch": 0.07646465572568076,
+      "learning_rate": 0.00018470890253965345,
+      "loss": 0.9109,
+      "mean_token_accuracy": 0.7344849526882171,
+      "num_tokens": 19300732.0,
+      "step": 8340
+    },
+    {
+      "epoch": 0.07655633996516,
+      "learning_rate": 0.0001846905656917576,
+      "loss": 0.9152,
+      "mean_token_accuracy": 0.7305308163166047,
+      "num_tokens": 19324322.0,
+      "step": 8350
+    },
+    {
+      "epoch": 0.07664802420463922,
+      "learning_rate": 0.00018467222884386176,
+      "loss": 0.9046,
+      "mean_token_accuracy": 0.7361734569072723,
+      "num_tokens": 19348400.0,
+      "step": 8360
+    },
+    {
+      "epoch": 0.07673970844411845,
+      "learning_rate": 0.0001846538919959659,
+      "loss": 0.9312,
+      "mean_token_accuracy": 0.7255568861961365,
+      "num_tokens": 19371307.0,
+      "step": 8370
+    },
+    {
+      "epoch": 0.07683139268359769,
+      "learning_rate": 0.00018463555514807006,
+      "loss": 0.8451,
+      "mean_token_accuracy": 0.7498118042945862,
+      "num_tokens": 19393832.0,
+      "step": 8380
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "learning_rate": 0.0001846172183001742,
+      "loss": 0.9102,
+      "mean_token_accuracy": 0.7367885828018188,
+      "num_tokens": 19416290.0,
+      "step": 8390
+    },
+    {
+      "epoch": 0.07701476116255615,
+      "learning_rate": 0.00018459888145227837,
+      "loss": 0.8905,
+      "mean_token_accuracy": 0.7381061017513275,
+      "num_tokens": 19438403.0,
+      "step": 8400
+    },
+    {
+      "epoch": 0.07710644540203539,
+      "learning_rate": 0.00018458054460438253,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7411992430686951,
+      "num_tokens": 19461862.0,
+      "step": 8410
+    },
+    {
+      "epoch": 0.07719812964151462,
+      "learning_rate": 0.00018456220775648667,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7488736093044281,
+      "num_tokens": 19484569.0,
+      "step": 8420
+    },
+    {
+      "epoch": 0.07728981388099386,
+      "learning_rate": 0.00018454387090859083,
+      "loss": 0.8825,
+      "mean_token_accuracy": 0.7367926299571991,
+      "num_tokens": 19507935.0,
+      "step": 8430
+    },
+    {
+      "epoch": 0.0773814981204731,
+      "learning_rate": 0.00018452553406069497,
+      "loss": 0.8891,
+      "mean_token_accuracy": 0.728267252445221,
+      "num_tokens": 19531556.0,
+      "step": 8440
+    },
+    {
+      "epoch": 0.07747318235995232,
+      "learning_rate": 0.0001845071972127991,
+      "loss": 0.8769,
+      "mean_token_accuracy": 0.737507826089859,
+      "num_tokens": 19554621.0,
+      "step": 8450
+    },
+    {
+      "epoch": 0.07756486659943156,
+      "learning_rate": 0.00018448886036490328,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.7386962890625,
+      "num_tokens": 19578394.0,
+      "step": 8460
+    },
+    {
+      "epoch": 0.07765655083891079,
+      "learning_rate": 0.00018447052351700744,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7423336207866669,
+      "num_tokens": 19601913.0,
+      "step": 8470
+    },
+    {
+      "epoch": 0.07774823507839003,
+      "learning_rate": 0.00018445218666911158,
+      "loss": 0.9025,
+      "mean_token_accuracy": 0.7393895745277405,
+      "num_tokens": 19626183.0,
+      "step": 8480
+    },
+    {
+      "epoch": 0.07783991931786925,
+      "learning_rate": 0.00018443384982121575,
+      "loss": 0.8913,
+      "mean_token_accuracy": 0.738505357503891,
+      "num_tokens": 19648997.0,
+      "step": 8490
+    },
+    {
+      "epoch": 0.07793160355734849,
+      "learning_rate": 0.00018441551297331988,
+      "loss": 0.8809,
+      "mean_token_accuracy": 0.7413359463214875,
+      "num_tokens": 19671810.0,
+      "step": 8500
+    },
+    {
+      "epoch": 0.07802328779682773,
+      "learning_rate": 0.00018439717612542405,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7465264201164246,
+      "num_tokens": 19695597.0,
+      "step": 8510
+    },
+    {
+      "epoch": 0.07811497203630696,
+      "learning_rate": 0.0001843788392775282,
+      "loss": 0.8899,
+      "mean_token_accuracy": 0.7404637336730957,
+      "num_tokens": 19719712.0,
+      "step": 8520
+    },
+    {
+      "epoch": 0.0782066562757862,
+      "learning_rate": 0.00018436050242963235,
+      "loss": 0.9133,
+      "mean_token_accuracy": 0.7294314622879028,
+      "num_tokens": 19742476.0,
+      "step": 8530
+    },
+    {
+      "epoch": 0.07829834051526542,
+      "learning_rate": 0.00018434216558173652,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7467393636703491,
+      "num_tokens": 19766150.0,
+      "step": 8540
+    },
+    {
+      "epoch": 0.07839002475474466,
+      "learning_rate": 0.00018432382873384066,
+      "loss": 0.8754,
+      "mean_token_accuracy": 0.741338175535202,
+      "num_tokens": 19788595.0,
+      "step": 8550
+    },
+    {
+      "epoch": 0.0784817089942239,
+      "learning_rate": 0.00018430549188594482,
+      "loss": 0.9201,
+      "mean_token_accuracy": 0.7328977167606354,
+      "num_tokens": 19811650.0,
+      "step": 8560
+    },
+    {
+      "epoch": 0.07857339323370313,
+      "learning_rate": 0.00018428715503804896,
+      "loss": 0.9023,
+      "mean_token_accuracy": 0.7369591057300567,
+      "num_tokens": 19835189.0,
+      "step": 8570
+    },
+    {
+      "epoch": 0.07866507747318235,
+      "learning_rate": 0.00018426881819015313,
+      "loss": 0.8898,
+      "mean_token_accuracy": 0.7325446903705597,
+      "num_tokens": 19857559.0,
+      "step": 8580
+    },
+    {
+      "epoch": 0.07875676171266159,
+      "learning_rate": 0.00018425048134225726,
+      "loss": 0.8951,
+      "mean_token_accuracy": 0.7413085997104645,
+      "num_tokens": 19880426.0,
+      "step": 8590
+    },
+    {
+      "epoch": 0.07884844595214083,
+      "learning_rate": 0.00018423214449436143,
+      "loss": 0.8995,
+      "mean_token_accuracy": 0.735448557138443,
+      "num_tokens": 19903141.0,
+      "step": 8600
+    },
+    {
+      "epoch": 0.07894013019162006,
+      "learning_rate": 0.0001842138076464656,
+      "loss": 0.8479,
+      "mean_token_accuracy": 0.7454135417938232,
+      "num_tokens": 19926504.0,
+      "step": 8610
+    },
+    {
+      "epoch": 0.0790318144310993,
+      "learning_rate": 0.00018419547079856973,
+      "loss": 0.8897,
+      "mean_token_accuracy": 0.7353550672531128,
+      "num_tokens": 19950113.0,
+      "step": 8620
+    },
+    {
+      "epoch": 0.07912349867057852,
+      "learning_rate": 0.0001841771339506739,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7345813035964965,
+      "num_tokens": 19974333.0,
+      "step": 8630
+    },
+    {
+      "epoch": 0.07921518291005776,
+      "learning_rate": 0.00018415879710277804,
+      "loss": 0.8743,
+      "mean_token_accuracy": 0.7420864105224609,
+      "num_tokens": 19996898.0,
+      "step": 8640
+    },
+    {
+      "epoch": 0.079306867149537,
+      "learning_rate": 0.00018414046025488218,
+      "loss": 0.8682,
+      "mean_token_accuracy": 0.7440168380737304,
+      "num_tokens": 20019741.0,
+      "step": 8650
+    },
+    {
+      "epoch": 0.07939855138901623,
+      "learning_rate": 0.00018412212340698637,
+      "loss": 0.9167,
+      "mean_token_accuracy": 0.7251489818096161,
+      "num_tokens": 20042555.0,
+      "step": 8660
+    },
+    {
+      "epoch": 0.07949023562849546,
+      "learning_rate": 0.0001841037865590905,
+      "loss": 0.8747,
+      "mean_token_accuracy": 0.744568943977356,
+      "num_tokens": 20066376.0,
+      "step": 8670
+    },
+    {
+      "epoch": 0.0795819198679747,
+      "learning_rate": 0.00018408544971119464,
+      "loss": 0.9263,
+      "mean_token_accuracy": 0.7316726863384246,
+      "num_tokens": 20089115.0,
+      "step": 8680
+    },
+    {
+      "epoch": 0.07967360410745393,
+      "learning_rate": 0.0001840671128632988,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.735486525297165,
+      "num_tokens": 20112260.0,
+      "step": 8690
+    },
+    {
+      "epoch": 0.07976528834693317,
+      "learning_rate": 0.00018404877601540295,
+      "loss": 0.8894,
+      "mean_token_accuracy": 0.7351312339305878,
+      "num_tokens": 20135132.0,
+      "step": 8700
+    },
+    {
+      "epoch": 0.07985697258641239,
+      "learning_rate": 0.00018403043916750711,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.7413839042186737,
+      "num_tokens": 20157358.0,
+      "step": 8710
+    },
+    {
+      "epoch": 0.07994865682589163,
+      "learning_rate": 0.00018401210231961125,
+      "loss": 0.9046,
+      "mean_token_accuracy": 0.7350801885128021,
+      "num_tokens": 20181209.0,
+      "step": 8720
+    },
+    {
+      "epoch": 0.08004034106537086,
+      "learning_rate": 0.00018399376547171542,
+      "loss": 0.8739,
+      "mean_token_accuracy": 0.7423152804374695,
+      "num_tokens": 20204229.0,
+      "step": 8730
+    },
+    {
+      "epoch": 0.0801320253048501,
+      "learning_rate": 0.00018397542862381958,
+      "loss": 0.8953,
+      "mean_token_accuracy": 0.7356089353561401,
+      "num_tokens": 20227928.0,
+      "step": 8740
+    },
+    {
+      "epoch": 0.08022370954432934,
+      "learning_rate": 0.00018395709177592372,
+      "loss": 0.9026,
+      "mean_token_accuracy": 0.7324627876281739,
+      "num_tokens": 20250216.0,
+      "step": 8750
+    },
+    {
+      "epoch": 0.08031539378380856,
+      "learning_rate": 0.0001839387549280279,
+      "loss": 0.9058,
+      "mean_token_accuracy": 0.7393997013568878,
+      "num_tokens": 20272232.0,
+      "step": 8760
+    },
+    {
+      "epoch": 0.0804070780232878,
+      "learning_rate": 0.00018392041808013202,
+      "loss": 0.8813,
+      "mean_token_accuracy": 0.7367434084415436,
+      "num_tokens": 20295550.0,
+      "step": 8770
+    },
+    {
+      "epoch": 0.08049876226276703,
+      "learning_rate": 0.0001839020812322362,
+      "loss": 0.9454,
+      "mean_token_accuracy": 0.7237827122211457,
+      "num_tokens": 20318767.0,
+      "step": 8780
+    },
+    {
+      "epoch": 0.08059044650224627,
+      "learning_rate": 0.00018388374438434036,
+      "loss": 0.9218,
+      "mean_token_accuracy": 0.7363382518291474,
+      "num_tokens": 20341763.0,
+      "step": 8790
+    },
+    {
+      "epoch": 0.08068213074172549,
+      "learning_rate": 0.0001838654075364445,
+      "loss": 0.9096,
+      "mean_token_accuracy": 0.7324922680854797,
+      "num_tokens": 20364896.0,
+      "step": 8800
+    },
+    {
+      "epoch": 0.08077381498120473,
+      "learning_rate": 0.00018384707068854866,
+      "loss": 0.896,
+      "mean_token_accuracy": 0.7368214428424835,
+      "num_tokens": 20387231.0,
+      "step": 8810
+    },
+    {
+      "epoch": 0.08086549922068396,
+      "learning_rate": 0.0001838287338406528,
+      "loss": 0.8796,
+      "mean_token_accuracy": 0.7440318286418914,
+      "num_tokens": 20411114.0,
+      "step": 8820
+    },
+    {
+      "epoch": 0.0809571834601632,
+      "learning_rate": 0.00018381039699275696,
+      "loss": 0.9121,
+      "mean_token_accuracy": 0.7343628704547882,
+      "num_tokens": 20434243.0,
+      "step": 8830
+    },
+    {
+      "epoch": 0.08104886769964244,
+      "learning_rate": 0.0001837920601448611,
+      "loss": 0.9109,
+      "mean_token_accuracy": 0.7294663310050964,
+      "num_tokens": 20457325.0,
+      "step": 8840
+    },
+    {
+      "epoch": 0.08114055193912166,
+      "learning_rate": 0.00018377372329696524,
+      "loss": 0.885,
+      "mean_token_accuracy": 0.7388871312141418,
+      "num_tokens": 20480852.0,
+      "step": 8850
+    },
+    {
+      "epoch": 0.0812322361786009,
+      "learning_rate": 0.00018375538644906943,
+      "loss": 0.859,
+      "mean_token_accuracy": 0.7362000644207001,
+      "num_tokens": 20504421.0,
+      "step": 8860
+    },
+    {
+      "epoch": 0.08132392041808013,
+      "learning_rate": 0.00018373704960117357,
+      "loss": 0.8667,
+      "mean_token_accuracy": 0.7418944597244262,
+      "num_tokens": 20527108.0,
+      "step": 8870
+    },
+    {
+      "epoch": 0.08141560465755937,
+      "learning_rate": 0.0001837187127532777,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7480157673358917,
+      "num_tokens": 20550605.0,
+      "step": 8880
+    },
+    {
+      "epoch": 0.0815072888970386,
+      "learning_rate": 0.00018370037590538187,
+      "loss": 0.9114,
+      "mean_token_accuracy": 0.7304875373840332,
+      "num_tokens": 20573090.0,
+      "step": 8890
+    },
+    {
+      "epoch": 0.08159897313651783,
+      "learning_rate": 0.000183682039057486,
+      "loss": 0.8376,
+      "mean_token_accuracy": 0.745467746257782,
+      "num_tokens": 20596228.0,
+      "step": 8900
+    },
+    {
+      "epoch": 0.08169065737599707,
+      "learning_rate": 0.00018366370220959018,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.7430700659751892,
+      "num_tokens": 20618943.0,
+      "step": 8910
+    },
+    {
+      "epoch": 0.0817823416154763,
+      "learning_rate": 0.00018364536536169434,
+      "loss": 0.9316,
+      "mean_token_accuracy": 0.7364117920398712,
+      "num_tokens": 20641169.0,
+      "step": 8920
+    },
+    {
+      "epoch": 0.08187402585495554,
+      "learning_rate": 0.00018362702851379848,
+      "loss": 0.8253,
+      "mean_token_accuracy": 0.7549974143505096,
+      "num_tokens": 20664607.0,
+      "step": 8930
+    },
+    {
+      "epoch": 0.08196571009443476,
+      "learning_rate": 0.00018360869166590265,
+      "loss": 0.8802,
+      "mean_token_accuracy": 0.7424738585948945,
+      "num_tokens": 20687667.0,
+      "step": 8940
+    },
+    {
+      "epoch": 0.082057394333914,
+      "learning_rate": 0.00018359035481800679,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.738900226354599,
+      "num_tokens": 20710902.0,
+      "step": 8950
+    },
+    {
+      "epoch": 0.08214907857339324,
+      "learning_rate": 0.00018357201797011095,
+      "loss": 0.8374,
+      "mean_token_accuracy": 0.7451739192008973,
+      "num_tokens": 20734252.0,
+      "step": 8960
+    },
+    {
+      "epoch": 0.08224076281287247,
+      "learning_rate": 0.0001835536811222151,
+      "loss": 0.8889,
+      "mean_token_accuracy": 0.7412276923656463,
+      "num_tokens": 20756954.0,
+      "step": 8970
+    },
+    {
+      "epoch": 0.0823324470523517,
+      "learning_rate": 0.00018353534427431925,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7402545034885406,
+      "num_tokens": 20779877.0,
+      "step": 8980
+    },
+    {
+      "epoch": 0.08242413129183093,
+      "learning_rate": 0.00018351700742642342,
+      "loss": 0.8855,
+      "mean_token_accuracy": 0.7334759831428528,
+      "num_tokens": 20802755.0,
+      "step": 8990
+    },
+    {
+      "epoch": 0.08251581553131017,
+      "learning_rate": 0.00018349867057852756,
+      "loss": 0.8862,
+      "mean_token_accuracy": 0.7427118003368378,
+      "num_tokens": 20826197.0,
+      "step": 9000
+    },
+    {
+      "epoch": 0.0826074997707894,
+      "learning_rate": 0.00018348033373063172,
+      "loss": 0.9096,
+      "mean_token_accuracy": 0.7359176874160767,
+      "num_tokens": 20848927.0,
+      "step": 9010
+    },
+    {
+      "epoch": 0.08269918401026863,
+      "learning_rate": 0.00018346199688273586,
+      "loss": 0.8515,
+      "mean_token_accuracy": 0.7440576672554016,
+      "num_tokens": 20871663.0,
+      "step": 9020
+    },
+    {
+      "epoch": 0.08279086824974786,
+      "learning_rate": 0.00018344366003484003,
+      "loss": 0.9031,
+      "mean_token_accuracy": 0.7336151599884033,
+      "num_tokens": 20894820.0,
+      "step": 9030
+    },
+    {
+      "epoch": 0.0828825524892271,
+      "learning_rate": 0.00018342532318694417,
+      "loss": 0.8684,
+      "mean_token_accuracy": 0.7430043578147888,
+      "num_tokens": 20917421.0,
+      "step": 9040
+    },
+    {
+      "epoch": 0.08297423672870634,
+      "learning_rate": 0.00018340698633904833,
+      "loss": 0.875,
+      "mean_token_accuracy": 0.7408009827136993,
+      "num_tokens": 20940502.0,
+      "step": 9050
+    },
+    {
+      "epoch": 0.08306592096818557,
+      "learning_rate": 0.0001833886494911525,
+      "loss": 0.8969,
+      "mean_token_accuracy": 0.7269556581974029,
+      "num_tokens": 20963538.0,
+      "step": 9060
+    },
+    {
+      "epoch": 0.0831576052076648,
+      "learning_rate": 0.00018337031264325664,
+      "loss": 0.9083,
+      "mean_token_accuracy": 0.7323137760162354,
+      "num_tokens": 20986338.0,
+      "step": 9070
+    },
+    {
+      "epoch": 0.08324928944714403,
+      "learning_rate": 0.00018335197579536077,
+      "loss": 0.8996,
+      "mean_token_accuracy": 0.7331121683120727,
+      "num_tokens": 21010087.0,
+      "step": 9080
+    },
+    {
+      "epoch": 0.08334097368662327,
+      "learning_rate": 0.00018333363894746494,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7416777074337005,
+      "num_tokens": 21033354.0,
+      "step": 9090
+    },
+    {
+      "epoch": 0.08343265792610251,
+      "learning_rate": 0.00018331530209956908,
+      "loss": 0.8994,
+      "mean_token_accuracy": 0.730178689956665,
+      "num_tokens": 21056357.0,
+      "step": 9100
+    },
+    {
+      "epoch": 0.08352434216558173,
+      "learning_rate": 0.00018329696525167324,
+      "loss": 0.8907,
+      "mean_token_accuracy": 0.7331460475921631,
+      "num_tokens": 21079081.0,
+      "step": 9110
+    },
+    {
+      "epoch": 0.08361602640506097,
+      "learning_rate": 0.0001832786284037774,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7385563969612121,
+      "num_tokens": 21102943.0,
+      "step": 9120
+    },
+    {
+      "epoch": 0.0837077106445402,
+      "learning_rate": 0.00018326029155588155,
+      "loss": 0.8868,
+      "mean_token_accuracy": 0.7314861476421356,
+      "num_tokens": 21126669.0,
+      "step": 9130
+    },
+    {
+      "epoch": 0.08379939488401944,
+      "learning_rate": 0.0001832419547079857,
+      "loss": 0.8772,
+      "mean_token_accuracy": 0.7403957903385162,
+      "num_tokens": 21149283.0,
+      "step": 9140
+    },
+    {
+      "epoch": 0.08389107912349868,
+      "learning_rate": 0.00018322361786008985,
+      "loss": 0.916,
+      "mean_token_accuracy": 0.7340789377689362,
+      "num_tokens": 21171927.0,
+      "step": 9150
+    },
+    {
+      "epoch": 0.0839827633629779,
+      "learning_rate": 0.00018320528101219402,
+      "loss": 0.8674,
+      "mean_token_accuracy": 0.7459503710269928,
+      "num_tokens": 21194580.0,
+      "step": 9160
+    },
+    {
+      "epoch": 0.08407444760245714,
+      "learning_rate": 0.00018318694416429815,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7379239141941071,
+      "num_tokens": 21217462.0,
+      "step": 9170
+    },
+    {
+      "epoch": 0.08416613184193637,
+      "learning_rate": 0.00018316860731640232,
+      "loss": 0.9298,
+      "mean_token_accuracy": 0.7346334517002105,
+      "num_tokens": 21240716.0,
+      "step": 9180
+    },
+    {
+      "epoch": 0.08425781608141561,
+      "learning_rate": 0.00018315027046850648,
+      "loss": 0.8791,
+      "mean_token_accuracy": 0.735994029045105,
+      "num_tokens": 21264109.0,
+      "step": 9190
+    },
+    {
+      "epoch": 0.08434950032089483,
+      "learning_rate": 0.00018313193362061062,
+      "loss": 0.9068,
+      "mean_token_accuracy": 0.734699672460556,
+      "num_tokens": 21287737.0,
+      "step": 9200
+    },
+    {
+      "epoch": 0.08444118456037407,
+      "learning_rate": 0.0001831135967727148,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.7357621252536773,
+      "num_tokens": 21311051.0,
+      "step": 9210
+    },
+    {
+      "epoch": 0.0845328687998533,
+      "learning_rate": 0.00018309525992481893,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7430299639701843,
+      "num_tokens": 21334874.0,
+      "step": 9220
+    },
+    {
+      "epoch": 0.08462455303933254,
+      "learning_rate": 0.0001830769230769231,
+      "loss": 0.9054,
+      "mean_token_accuracy": 0.7372971832752228,
+      "num_tokens": 21358174.0,
+      "step": 9230
+    },
+    {
+      "epoch": 0.08471623727881178,
+      "learning_rate": 0.00018305858622902723,
+      "loss": 0.8813,
+      "mean_token_accuracy": 0.7368301928043366,
+      "num_tokens": 21382025.0,
+      "step": 9240
+    },
+    {
+      "epoch": 0.084807921518291,
+      "learning_rate": 0.0001830402493811314,
+      "loss": 0.8823,
+      "mean_token_accuracy": 0.7336306691169738,
+      "num_tokens": 21404632.0,
+      "step": 9250
+    },
+    {
+      "epoch": 0.08489960575777024,
+      "learning_rate": 0.00018302191253323556,
+      "loss": 0.8547,
+      "mean_token_accuracy": 0.7515554547309875,
+      "num_tokens": 21428131.0,
+      "step": 9260
+    },
+    {
+      "epoch": 0.08499128999724948,
+      "learning_rate": 0.0001830035756853397,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7392058372497559,
+      "num_tokens": 21451432.0,
+      "step": 9270
+    },
+    {
+      "epoch": 0.08508297423672871,
+      "learning_rate": 0.00018298523883744384,
+      "loss": 0.8871,
+      "mean_token_accuracy": 0.7410492539405823,
+      "num_tokens": 21474230.0,
+      "step": 9280
+    },
+    {
+      "epoch": 0.08517465847620793,
+      "learning_rate": 0.000182966901989548,
+      "loss": 0.9153,
+      "mean_token_accuracy": 0.7351513147354126,
+      "num_tokens": 21497153.0,
+      "step": 9290
+    },
+    {
+      "epoch": 0.08526634271568717,
+      "learning_rate": 0.00018294856514165214,
+      "loss": 0.8844,
+      "mean_token_accuracy": 0.7379489421844483,
+      "num_tokens": 21520721.0,
+      "step": 9300
+    },
+    {
+      "epoch": 0.08535802695516641,
+      "learning_rate": 0.00018293022829375633,
+      "loss": 0.8737,
+      "mean_token_accuracy": 0.7409141719341278,
+      "num_tokens": 21543904.0,
+      "step": 9310
+    },
+    {
+      "epoch": 0.08544971119464564,
+      "learning_rate": 0.00018291189144586047,
+      "loss": 0.8947,
+      "mean_token_accuracy": 0.7339003920555115,
+      "num_tokens": 21566753.0,
+      "step": 9320
+    },
+    {
+      "epoch": 0.08554139543412487,
+      "learning_rate": 0.0001828935545979646,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7357079029083252,
+      "num_tokens": 21590527.0,
+      "step": 9330
+    },
+    {
+      "epoch": 0.0856330796736041,
+      "learning_rate": 0.00018287521775006878,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7461161673069,
+      "num_tokens": 21613767.0,
+      "step": 9340
+    },
+    {
+      "epoch": 0.08572476391308334,
+      "learning_rate": 0.00018285688090217291,
+      "loss": 0.8909,
+      "mean_token_accuracy": 0.7372100174427032,
+      "num_tokens": 21637037.0,
+      "step": 9350
+    },
+    {
+      "epoch": 0.08581644815256258,
+      "learning_rate": 0.00018283854405427708,
+      "loss": 0.9097,
+      "mean_token_accuracy": 0.7302749633789063,
+      "num_tokens": 21660356.0,
+      "step": 9360
+    },
+    {
+      "epoch": 0.08590813239204181,
+      "learning_rate": 0.00018282020720638122,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.7322764992713928,
+      "num_tokens": 21683097.0,
+      "step": 9370
+    },
+    {
+      "epoch": 0.08599981663152104,
+      "learning_rate": 0.00018280187035848538,
+      "loss": 0.9033,
+      "mean_token_accuracy": 0.7378346085548401,
+      "num_tokens": 21705375.0,
+      "step": 9380
+    },
+    {
+      "epoch": 0.08609150087100027,
+      "learning_rate": 0.00018278353351058955,
+      "loss": 0.8736,
+      "mean_token_accuracy": 0.7481322288513184,
+      "num_tokens": 21728893.0,
+      "step": 9390
+    },
+    {
+      "epoch": 0.08618318511047951,
+      "learning_rate": 0.0001827651966626937,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7349271535873413,
+      "num_tokens": 21751697.0,
+      "step": 9400
+    },
+    {
+      "epoch": 0.08627486934995875,
+      "learning_rate": 0.00018274685981479785,
+      "loss": 0.86,
+      "mean_token_accuracy": 0.7382111191749573,
+      "num_tokens": 21774770.0,
+      "step": 9410
+    },
+    {
+      "epoch": 0.08636655358943797,
+      "learning_rate": 0.000182728522966902,
+      "loss": 0.9093,
+      "mean_token_accuracy": 0.733877283334732,
+      "num_tokens": 21796772.0,
+      "step": 9420
+    },
+    {
+      "epoch": 0.0864582378289172,
+      "learning_rate": 0.00018271018611900616,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.742811793088913,
+      "num_tokens": 21820725.0,
+      "step": 9430
+    },
+    {
+      "epoch": 0.08654992206839644,
+      "learning_rate": 0.00018269184927111032,
+      "loss": 0.896,
+      "mean_token_accuracy": 0.7337358832359314,
+      "num_tokens": 21843697.0,
+      "step": 9440
+    },
+    {
+      "epoch": 0.08664160630787568,
+      "learning_rate": 0.00018267351242321446,
+      "loss": 0.8616,
+      "mean_token_accuracy": 0.7448089361190796,
+      "num_tokens": 21866861.0,
+      "step": 9450
+    },
+    {
+      "epoch": 0.08673329054735492,
+      "learning_rate": 0.00018265517557531863,
+      "loss": 0.8978,
+      "mean_token_accuracy": 0.7339885711669922,
+      "num_tokens": 21889922.0,
+      "step": 9460
+    },
+    {
+      "epoch": 0.08682497478683414,
+      "learning_rate": 0.00018263683872742276,
+      "loss": 0.9258,
+      "mean_token_accuracy": 0.7256360352039337,
+      "num_tokens": 21912965.0,
+      "step": 9470
+    },
+    {
+      "epoch": 0.08691665902631338,
+      "learning_rate": 0.0001826185018795269,
+      "loss": 0.8548,
+      "mean_token_accuracy": 0.7446251630783081,
+      "num_tokens": 21936199.0,
+      "step": 9480
+    },
+    {
+      "epoch": 0.08700834326579261,
+      "learning_rate": 0.00018260016503163107,
+      "loss": 0.9191,
+      "mean_token_accuracy": 0.732340258359909,
+      "num_tokens": 21958595.0,
+      "step": 9490
+    },
+    {
+      "epoch": 0.08710002750527185,
+      "learning_rate": 0.0001825818281837352,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7465512156486511,
+      "num_tokens": 21981051.0,
+      "step": 9500
+    },
+    {
+      "epoch": 0.08719171174475107,
+      "learning_rate": 0.0001825634913358394,
+      "loss": 0.8792,
+      "mean_token_accuracy": 0.733728039264679,
+      "num_tokens": 22004657.0,
+      "step": 9510
+    },
+    {
+      "epoch": 0.08728339598423031,
+      "learning_rate": 0.00018254515448794354,
+      "loss": 0.9001,
+      "mean_token_accuracy": 0.7300061106681823,
+      "num_tokens": 22027822.0,
+      "step": 9520
+    },
+    {
+      "epoch": 0.08737508022370954,
+      "learning_rate": 0.00018252681764004768,
+      "loss": 0.8689,
+      "mean_token_accuracy": 0.745732969045639,
+      "num_tokens": 22051033.0,
+      "step": 9530
+    },
+    {
+      "epoch": 0.08746676446318878,
+      "learning_rate": 0.00018250848079215184,
+      "loss": 0.88,
+      "mean_token_accuracy": 0.7341371715068817,
+      "num_tokens": 22074337.0,
+      "step": 9540
+    },
+    {
+      "epoch": 0.08755844870266802,
+      "learning_rate": 0.00018249014394425598,
+      "loss": 0.9054,
+      "mean_token_accuracy": 0.734076589345932,
+      "num_tokens": 22096998.0,
+      "step": 9550
+    },
+    {
+      "epoch": 0.08765013294214724,
+      "learning_rate": 0.00018247180709636014,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7426896154880523,
+      "num_tokens": 22119842.0,
+      "step": 9560
+    },
+    {
+      "epoch": 0.08774181718162648,
+      "learning_rate": 0.0001824534702484643,
+      "loss": 0.8793,
+      "mean_token_accuracy": 0.7316954851150512,
+      "num_tokens": 22142707.0,
+      "step": 9570
+    },
+    {
+      "epoch": 0.08783350142110571,
+      "learning_rate": 0.00018243513340056845,
+      "loss": 0.8868,
+      "mean_token_accuracy": 0.7348066747188569,
+      "num_tokens": 22166344.0,
+      "step": 9580
+    },
+    {
+      "epoch": 0.08792518566058495,
+      "learning_rate": 0.00018241679655267261,
+      "loss": 0.8568,
+      "mean_token_accuracy": 0.744666600227356,
+      "num_tokens": 22188550.0,
+      "step": 9590
+    },
+    {
+      "epoch": 0.08801686990006417,
+      "learning_rate": 0.00018239845970477675,
+      "loss": 0.8764,
+      "mean_token_accuracy": 0.7450824737548828,
+      "num_tokens": 22211255.0,
+      "step": 9600
+    },
+    {
+      "epoch": 0.08810855413954341,
+      "learning_rate": 0.00018238012285688092,
+      "loss": 0.9055,
+      "mean_token_accuracy": 0.7375835120677948,
+      "num_tokens": 22234451.0,
+      "step": 9610
+    },
+    {
+      "epoch": 0.08820023837902265,
+      "learning_rate": 0.00018236178600898506,
+      "loss": 0.8869,
+      "mean_token_accuracy": 0.7426702976226807,
+      "num_tokens": 22257234.0,
+      "step": 9620
+    },
+    {
+      "epoch": 0.08829192261850188,
+      "learning_rate": 0.00018234344916108922,
+      "loss": 0.8714,
+      "mean_token_accuracy": 0.746830689907074,
+      "num_tokens": 22280204.0,
+      "step": 9630
+    },
+    {
+      "epoch": 0.0883836068579811,
+      "learning_rate": 0.0001823251123131934,
+      "loss": 0.8792,
+      "mean_token_accuracy": 0.7361852407455445,
+      "num_tokens": 22304714.0,
+      "step": 9640
+    },
+    {
+      "epoch": 0.08847529109746034,
+      "learning_rate": 0.00018230677546529752,
+      "loss": 0.9159,
+      "mean_token_accuracy": 0.732985931634903,
+      "num_tokens": 22327734.0,
+      "step": 9650
+    },
+    {
+      "epoch": 0.08856697533693958,
+      "learning_rate": 0.0001822884386174017,
+      "loss": 0.8742,
+      "mean_token_accuracy": 0.7432840228080749,
+      "num_tokens": 22350796.0,
+      "step": 9660
+    },
+    {
+      "epoch": 0.08865865957641882,
+      "learning_rate": 0.00018227010176950583,
+      "loss": 0.8694,
+      "mean_token_accuracy": 0.7500789582729339,
+      "num_tokens": 22374105.0,
+      "step": 9670
+    },
+    {
+      "epoch": 0.08875034381589805,
+      "learning_rate": 0.00018225176492160997,
+      "loss": 0.9417,
+      "mean_token_accuracy": 0.7292441070079804,
+      "num_tokens": 22397321.0,
+      "step": 9680
+    },
+    {
+      "epoch": 0.08884202805537728,
+      "learning_rate": 0.00018223342807371413,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.7461090803146362,
+      "num_tokens": 22420772.0,
+      "step": 9690
+    },
+    {
+      "epoch": 0.08893371229485651,
+      "learning_rate": 0.00018221509122581827,
+      "loss": 0.8963,
+      "mean_token_accuracy": 0.7368881940841675,
+      "num_tokens": 22443852.0,
+      "step": 9700
+    },
+    {
+      "epoch": 0.08902539653433575,
+      "learning_rate": 0.00018219675437792246,
+      "loss": 0.9074,
+      "mean_token_accuracy": 0.7292484223842621,
+      "num_tokens": 22467109.0,
+      "step": 9710
+    },
+    {
+      "epoch": 0.08911708077381499,
+      "learning_rate": 0.0001821784175300266,
+      "loss": 0.8251,
+      "mean_token_accuracy": 0.7499877512454987,
+      "num_tokens": 22490388.0,
+      "step": 9720
+    },
+    {
+      "epoch": 0.08920876501329421,
+      "learning_rate": 0.00018216008068213074,
+      "loss": 0.8924,
+      "mean_token_accuracy": 0.7306248307228088,
+      "num_tokens": 22513923.0,
+      "step": 9730
+    },
+    {
+      "epoch": 0.08930044925277345,
+      "learning_rate": 0.0001821417438342349,
+      "loss": 0.9283,
+      "mean_token_accuracy": 0.7331124186515808,
+      "num_tokens": 22537418.0,
+      "step": 9740
+    },
+    {
+      "epoch": 0.08939213349225268,
+      "learning_rate": 0.00018212340698633904,
+      "loss": 0.8983,
+      "mean_token_accuracy": 0.7407335758209228,
+      "num_tokens": 22560079.0,
+      "step": 9750
+    },
+    {
+      "epoch": 0.08948381773173192,
+      "learning_rate": 0.0001821050701384432,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7470311105251313,
+      "num_tokens": 22582433.0,
+      "step": 9760
+    },
+    {
+      "epoch": 0.08957550197121115,
+      "learning_rate": 0.00018208673329054737,
+      "loss": 0.8609,
+      "mean_token_accuracy": 0.7422303915023803,
+      "num_tokens": 22605173.0,
+      "step": 9770
+    },
+    {
+      "epoch": 0.08966718621069038,
+      "learning_rate": 0.0001820683964426515,
+      "loss": 0.8944,
+      "mean_token_accuracy": 0.7427703142166138,
+      "num_tokens": 22627989.0,
+      "step": 9780
+    },
+    {
+      "epoch": 0.08975887045016961,
+      "learning_rate": 0.00018205005959475568,
+      "loss": 0.8471,
+      "mean_token_accuracy": 0.7471030592918396,
+      "num_tokens": 22650742.0,
+      "step": 9790
+    },
+    {
+      "epoch": 0.08985055468964885,
+      "learning_rate": 0.00018203172274685982,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7479526579380036,
+      "num_tokens": 22673623.0,
+      "step": 9800
+    },
+    {
+      "epoch": 0.08994223892912809,
+      "learning_rate": 0.00018201338589896398,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.7459315598011017,
+      "num_tokens": 22696704.0,
+      "step": 9810
+    },
+    {
+      "epoch": 0.09003392316860731,
+      "learning_rate": 0.00018199504905106812,
+      "loss": 0.9011,
+      "mean_token_accuracy": 0.7355086147785187,
+      "num_tokens": 22720122.0,
+      "step": 9820
+    },
+    {
+      "epoch": 0.09012560740808655,
+      "learning_rate": 0.00018197671220317229,
+      "loss": 0.9328,
+      "mean_token_accuracy": 0.7215277791023255,
+      "num_tokens": 22742374.0,
+      "step": 9830
+    },
+    {
+      "epoch": 0.09021729164756578,
+      "learning_rate": 0.00018195837535527645,
+      "loss": 0.9456,
+      "mean_token_accuracy": 0.7291106104850769,
+      "num_tokens": 22765871.0,
+      "step": 9840
+    },
+    {
+      "epoch": 0.09030897588704502,
+      "learning_rate": 0.0001819400385073806,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.7470474421977997,
+      "num_tokens": 22788967.0,
+      "step": 9850
+    },
+    {
+      "epoch": 0.09040066012652426,
+      "learning_rate": 0.00018192170165948475,
+      "loss": 0.8991,
+      "mean_token_accuracy": 0.7358954310417175,
+      "num_tokens": 22811479.0,
+      "step": 9860
+    },
+    {
+      "epoch": 0.09049234436600348,
+      "learning_rate": 0.0001819033648115889,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.7358669996261596,
+      "num_tokens": 22834308.0,
+      "step": 9870
+    },
+    {
+      "epoch": 0.09058402860548272,
+      "learning_rate": 0.00018188502796369303,
+      "loss": 0.9043,
+      "mean_token_accuracy": 0.7353496193885803,
+      "num_tokens": 22857743.0,
+      "step": 9880
+    },
+    {
+      "epoch": 0.09067571284496195,
+      "learning_rate": 0.0001818666911157972,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7453127682209015,
+      "num_tokens": 22880218.0,
+      "step": 9890
+    },
+    {
+      "epoch": 0.09076739708444119,
+      "learning_rate": 0.00018184835426790136,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7360309422016144,
+      "num_tokens": 22903463.0,
+      "step": 9900
+    },
+    {
+      "epoch": 0.09085908132392041,
+      "learning_rate": 0.00018183001742000553,
+      "loss": 0.8704,
+      "mean_token_accuracy": 0.7368826031684875,
+      "num_tokens": 22926125.0,
+      "step": 9910
+    },
+    {
+      "epoch": 0.09095076556339965,
+      "learning_rate": 0.00018181168057210967,
+      "loss": 0.8731,
+      "mean_token_accuracy": 0.7351884782314301,
+      "num_tokens": 22949921.0,
+      "step": 9920
+    },
+    {
+      "epoch": 0.09104244980287889,
+      "learning_rate": 0.0001817933437242138,
+      "loss": 0.9423,
+      "mean_token_accuracy": 0.7227811276912689,
+      "num_tokens": 22973397.0,
+      "step": 9930
+    },
+    {
+      "epoch": 0.09113413404235812,
+      "learning_rate": 0.00018177500687631797,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7420787274837494,
+      "num_tokens": 22996340.0,
+      "step": 9940
+    },
+    {
+      "epoch": 0.09122581828183735,
+      "learning_rate": 0.0001817566700284221,
+      "loss": 0.8551,
+      "mean_token_accuracy": 0.746318656206131,
+      "num_tokens": 23018876.0,
+      "step": 9950
+    },
+    {
+      "epoch": 0.09131750252131658,
+      "learning_rate": 0.00018173833318052627,
+      "loss": 0.9181,
+      "mean_token_accuracy": 0.7355397641658783,
+      "num_tokens": 23041450.0,
+      "step": 9960
+    },
+    {
+      "epoch": 0.09140918676079582,
+      "learning_rate": 0.00018171999633263044,
+      "loss": 0.8854,
+      "mean_token_accuracy": 0.7342116951942443,
+      "num_tokens": 23064800.0,
+      "step": 9970
+    },
+    {
+      "epoch": 0.09150087100027506,
+      "learning_rate": 0.00018170165948473458,
+      "loss": 0.8047,
+      "mean_token_accuracy": 0.7571393847465515,
+      "num_tokens": 23087381.0,
+      "step": 9980
+    },
+    {
+      "epoch": 0.09159255523975429,
+      "learning_rate": 0.00018168332263683874,
+      "loss": 0.9063,
+      "mean_token_accuracy": 0.7324288666248322,
+      "num_tokens": 23110734.0,
+      "step": 9990
+    },
+    {
+      "epoch": 0.09168423947923351,
+      "learning_rate": 0.00018166498578894288,
+      "loss": 0.9023,
+      "mean_token_accuracy": 0.7366482257843018,
+      "num_tokens": 23133633.0,
+      "step": 10000
+    },
+    {
+      "epoch": 0.09177592371871275,
+      "learning_rate": 0.00018164664894104705,
+      "loss": 0.901,
+      "mean_token_accuracy": 0.7339980900287628,
+      "num_tokens": 23156919.0,
+      "step": 10010
+    },
+    {
+      "epoch": 0.09186760795819199,
+      "learning_rate": 0.00018162831209315118,
+      "loss": 0.909,
+      "mean_token_accuracy": 0.7299727261066437,
+      "num_tokens": 23180293.0,
+      "step": 10020
+    },
+    {
+      "epoch": 0.09195929219767122,
+      "learning_rate": 0.00018160997524525535,
+      "loss": 0.8927,
+      "mean_token_accuracy": 0.740869790315628,
+      "num_tokens": 23203594.0,
+      "step": 10030
+    },
+    {
+      "epoch": 0.09205097643715045,
+      "learning_rate": 0.00018159163839735952,
+      "loss": 0.8973,
+      "mean_token_accuracy": 0.7331617951393128,
+      "num_tokens": 23225983.0,
+      "step": 10040
+    },
+    {
+      "epoch": 0.09214266067662968,
+      "learning_rate": 0.00018157330154946365,
+      "loss": 0.9039,
+      "mean_token_accuracy": 0.7249781310558319,
+      "num_tokens": 23248764.0,
+      "step": 10050
+    },
+    {
+      "epoch": 0.09223434491610892,
+      "learning_rate": 0.00018155496470156782,
+      "loss": 0.9008,
+      "mean_token_accuracy": 0.7372535169124603,
+      "num_tokens": 23271310.0,
+      "step": 10060
+    },
+    {
+      "epoch": 0.09232602915558816,
+      "learning_rate": 0.00018153662785367196,
+      "loss": 0.9034,
+      "mean_token_accuracy": 0.7302026450634003,
+      "num_tokens": 23293634.0,
+      "step": 10070
+    },
+    {
+      "epoch": 0.0924177133950674,
+      "learning_rate": 0.0001815182910057761,
+      "loss": 0.875,
+      "mean_token_accuracy": 0.7396530449390412,
+      "num_tokens": 23316858.0,
+      "step": 10080
+    },
+    {
+      "epoch": 0.09250939763454662,
+      "learning_rate": 0.00018149995415788026,
+      "loss": 0.8783,
+      "mean_token_accuracy": 0.7374988317489624,
+      "num_tokens": 23339146.0,
+      "step": 10090
+    },
+    {
+      "epoch": 0.09260108187402585,
+      "learning_rate": 0.00018148161730998443,
+      "loss": 0.8557,
+      "mean_token_accuracy": 0.7460222184658051,
+      "num_tokens": 23363154.0,
+      "step": 10100
+    },
+    {
+      "epoch": 0.09269276611350509,
+      "learning_rate": 0.0001814632804620886,
+      "loss": 0.9,
+      "mean_token_accuracy": 0.7356420278549194,
+      "num_tokens": 23386881.0,
+      "step": 10110
+    },
+    {
+      "epoch": 0.09278445035298433,
+      "learning_rate": 0.00018144494361419273,
+      "loss": 0.8941,
+      "mean_token_accuracy": 0.7426242411136628,
+      "num_tokens": 23409505.0,
+      "step": 10120
+    },
+    {
+      "epoch": 0.09287613459246355,
+      "learning_rate": 0.00018142660676629687,
+      "loss": 0.8851,
+      "mean_token_accuracy": 0.7426791250705719,
+      "num_tokens": 23432008.0,
+      "step": 10130
+    },
+    {
+      "epoch": 0.09296781883194279,
+      "learning_rate": 0.00018140826991840103,
+      "loss": 0.8507,
+      "mean_token_accuracy": 0.7449314594268799,
+      "num_tokens": 23455531.0,
+      "step": 10140
+    },
+    {
+      "epoch": 0.09305950307142202,
+      "learning_rate": 0.00018138993307050517,
+      "loss": 0.8933,
+      "mean_token_accuracy": 0.7333428859710693,
+      "num_tokens": 23480250.0,
+      "step": 10150
+    },
+    {
+      "epoch": 0.09315118731090126,
+      "learning_rate": 0.00018137159622260936,
+      "loss": 0.8911,
+      "mean_token_accuracy": 0.7399576127529144,
+      "num_tokens": 23503635.0,
+      "step": 10160
+    },
+    {
+      "epoch": 0.0932428715503805,
+      "learning_rate": 0.0001813532593747135,
+      "loss": 0.8613,
+      "mean_token_accuracy": 0.7430021822452545,
+      "num_tokens": 23527143.0,
+      "step": 10170
+    },
+    {
+      "epoch": 0.09333455578985972,
+      "learning_rate": 0.00018133492252681764,
+      "loss": 0.8669,
+      "mean_token_accuracy": 0.7376038372516632,
+      "num_tokens": 23550196.0,
+      "step": 10180
+    },
+    {
+      "epoch": 0.09342624002933896,
+      "learning_rate": 0.0001813165856789218,
+      "loss": 0.9163,
+      "mean_token_accuracy": 0.7347811579704284,
+      "num_tokens": 23573328.0,
+      "step": 10190
+    },
+    {
+      "epoch": 0.09351792426881819,
+      "learning_rate": 0.00018129824883102595,
+      "loss": 0.8665,
+      "mean_token_accuracy": 0.7413029789924621,
+      "num_tokens": 23596881.0,
+      "step": 10200
+    },
+    {
+      "epoch": 0.09360960850829743,
+      "learning_rate": 0.0001812799119831301,
+      "loss": 0.9292,
+      "mean_token_accuracy": 0.7272100985050202,
+      "num_tokens": 23620234.0,
+      "step": 10210
+    },
+    {
+      "epoch": 0.09370129274777665,
+      "learning_rate": 0.00018126157513523425,
+      "loss": 0.8538,
+      "mean_token_accuracy": 0.7449434697628021,
+      "num_tokens": 23643870.0,
+      "step": 10220
+    },
+    {
+      "epoch": 0.09379297698725589,
+      "learning_rate": 0.00018124323828733841,
+      "loss": 0.9175,
+      "mean_token_accuracy": 0.7336058020591736,
+      "num_tokens": 23666228.0,
+      "step": 10230
+    },
+    {
+      "epoch": 0.09388466122673512,
+      "learning_rate": 0.00018122490143944258,
+      "loss": 0.8563,
+      "mean_token_accuracy": 0.7398800075054168,
+      "num_tokens": 23690317.0,
+      "step": 10240
+    },
+    {
+      "epoch": 0.09397634546621436,
+      "learning_rate": 0.00018120656459154672,
+      "loss": 0.8536,
+      "mean_token_accuracy": 0.7451589703559875,
+      "num_tokens": 23713023.0,
+      "step": 10250
+    },
+    {
+      "epoch": 0.09406802970569358,
+      "learning_rate": 0.00018118822774365088,
+      "loss": 0.9027,
+      "mean_token_accuracy": 0.7337402164936065,
+      "num_tokens": 23736779.0,
+      "step": 10260
+    },
+    {
+      "epoch": 0.09415971394517282,
+      "learning_rate": 0.00018116989089575502,
+      "loss": 0.8761,
+      "mean_token_accuracy": 0.7383345663547516,
+      "num_tokens": 23759880.0,
+      "step": 10270
+    },
+    {
+      "epoch": 0.09425139818465206,
+      "learning_rate": 0.00018115155404785916,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.7440394878387451,
+      "num_tokens": 23782863.0,
+      "step": 10280
+    },
+    {
+      "epoch": 0.0943430824241313,
+      "learning_rate": 0.00018113321719996335,
+      "loss": 0.9068,
+      "mean_token_accuracy": 0.7336950719356536,
+      "num_tokens": 23806069.0,
+      "step": 10290
+    },
+    {
+      "epoch": 0.09443476666361053,
+      "learning_rate": 0.0001811148803520675,
+      "loss": 0.8937,
+      "mean_token_accuracy": 0.7380026817321778,
+      "num_tokens": 23829547.0,
+      "step": 10300
+    },
+    {
+      "epoch": 0.09452645090308975,
+      "learning_rate": 0.00018109654350417166,
+      "loss": 0.9029,
+      "mean_token_accuracy": 0.7320657312870026,
+      "num_tokens": 23852311.0,
+      "step": 10310
+    },
+    {
+      "epoch": 0.09461813514256899,
+      "learning_rate": 0.0001810782066562758,
+      "loss": 0.8758,
+      "mean_token_accuracy": 0.7432155132293701,
+      "num_tokens": 23875025.0,
+      "step": 10320
+    },
+    {
+      "epoch": 0.09470981938204823,
+      "learning_rate": 0.00018105986980837993,
+      "loss": 0.8583,
+      "mean_token_accuracy": 0.7418897330760956,
+      "num_tokens": 23898321.0,
+      "step": 10330
+    },
+    {
+      "epoch": 0.09480150362152746,
+      "learning_rate": 0.0001810415329604841,
+      "loss": 0.8863,
+      "mean_token_accuracy": 0.7378986597061157,
+      "num_tokens": 23922891.0,
+      "step": 10340
+    },
+    {
+      "epoch": 0.09489318786100669,
+      "learning_rate": 0.00018102319611258824,
+      "loss": 0.8774,
+      "mean_token_accuracy": 0.7400460600852966,
+      "num_tokens": 23945787.0,
+      "step": 10350
+    },
+    {
+      "epoch": 0.09498487210048592,
+      "learning_rate": 0.00018100485926469243,
+      "loss": 0.9182,
+      "mean_token_accuracy": 0.7314947068691253,
+      "num_tokens": 23969040.0,
+      "step": 10360
+    },
+    {
+      "epoch": 0.09507655633996516,
+      "learning_rate": 0.00018098652241679657,
+      "loss": 0.9068,
+      "mean_token_accuracy": 0.7313754498958588,
+      "num_tokens": 23992210.0,
+      "step": 10370
+    },
+    {
+      "epoch": 0.0951682405794444,
+      "learning_rate": 0.0001809681855689007,
+      "loss": 0.8971,
+      "mean_token_accuracy": 0.7337064266204834,
+      "num_tokens": 24014546.0,
+      "step": 10380
+    },
+    {
+      "epoch": 0.09525992481892363,
+      "learning_rate": 0.00018094984872100487,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.7389392912387848,
+      "num_tokens": 24037577.0,
+      "step": 10390
+    },
+    {
+      "epoch": 0.09535160905840286,
+      "learning_rate": 0.000180931511873109,
+      "loss": 0.8682,
+      "mean_token_accuracy": 0.7403826177120209,
+      "num_tokens": 24060043.0,
+      "step": 10400
+    },
+    {
+      "epoch": 0.09544329329788209,
+      "learning_rate": 0.00018091317502521318,
+      "loss": 0.8946,
+      "mean_token_accuracy": 0.737509262561798,
+      "num_tokens": 24083426.0,
+      "step": 10410
+    },
+    {
+      "epoch": 0.09553497753736133,
+      "learning_rate": 0.00018089483817731734,
+      "loss": 0.8635,
+      "mean_token_accuracy": 0.74267076253891,
+      "num_tokens": 24106035.0,
+      "step": 10420
+    },
+    {
+      "epoch": 0.09562666177684057,
+      "learning_rate": 0.00018087650132942148,
+      "loss": 0.8746,
+      "mean_token_accuracy": 0.741468733549118,
+      "num_tokens": 24128673.0,
+      "step": 10430
+    },
+    {
+      "epoch": 0.09571834601631979,
+      "learning_rate": 0.00018085816448152564,
+      "loss": 0.8833,
+      "mean_token_accuracy": 0.7361687660217285,
+      "num_tokens": 24151675.0,
+      "step": 10440
+    },
+    {
+      "epoch": 0.09581003025579903,
+      "learning_rate": 0.00018083982763362978,
+      "loss": 0.9131,
+      "mean_token_accuracy": 0.734002023935318,
+      "num_tokens": 24174869.0,
+      "step": 10450
+    },
+    {
+      "epoch": 0.09590171449527826,
+      "learning_rate": 0.00018082149078573395,
+      "loss": 0.881,
+      "mean_token_accuracy": 0.7427867949008942,
+      "num_tokens": 24198932.0,
+      "step": 10460
+    },
+    {
+      "epoch": 0.0959933987347575,
+      "learning_rate": 0.00018080315393783809,
+      "loss": 0.9253,
+      "mean_token_accuracy": 0.7226717889308929,
+      "num_tokens": 24220961.0,
+      "step": 10470
+    },
+    {
+      "epoch": 0.09608508297423674,
+      "learning_rate": 0.00018078481708994222,
+      "loss": 0.9018,
+      "mean_token_accuracy": 0.7350903332233429,
+      "num_tokens": 24243768.0,
+      "step": 10480
+    },
+    {
+      "epoch": 0.09617676721371596,
+      "learning_rate": 0.00018076648024204642,
+      "loss": 0.9096,
+      "mean_token_accuracy": 0.7298749208450317,
+      "num_tokens": 24267112.0,
+      "step": 10490
+    },
+    {
+      "epoch": 0.0962684514531952,
+      "learning_rate": 0.00018074814339415056,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7433853507041931,
+      "num_tokens": 24290280.0,
+      "step": 10500
+    },
+    {
+      "epoch": 0.09636013569267443,
+      "learning_rate": 0.00018072980654625472,
+      "loss": 0.876,
+      "mean_token_accuracy": 0.7451959550380707,
+      "num_tokens": 24313419.0,
+      "step": 10510
+    },
+    {
+      "epoch": 0.09645181993215367,
+      "learning_rate": 0.00018071146969835886,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.7386884212493896,
+      "num_tokens": 24336293.0,
+      "step": 10520
+    },
+    {
+      "epoch": 0.09654350417163289,
+      "learning_rate": 0.000180693132850463,
+      "loss": 0.8308,
+      "mean_token_accuracy": 0.7562291979789734,
+      "num_tokens": 24359607.0,
+      "step": 10530
+    },
+    {
+      "epoch": 0.09663518841111213,
+      "learning_rate": 0.00018067479600256716,
+      "loss": 0.9069,
+      "mean_token_accuracy": 0.7358807027339935,
+      "num_tokens": 24383134.0,
+      "step": 10540
+    },
+    {
+      "epoch": 0.09672687265059136,
+      "learning_rate": 0.00018065645915467133,
+      "loss": 0.8859,
+      "mean_token_accuracy": 0.7362659752368927,
+      "num_tokens": 24405313.0,
+      "step": 10550
+    },
+    {
+      "epoch": 0.0968185568900706,
+      "learning_rate": 0.0001806381223067755,
+      "loss": 0.9059,
+      "mean_token_accuracy": 0.7316841602325439,
+      "num_tokens": 24428495.0,
+      "step": 10560
+    },
+    {
+      "epoch": 0.09691024112954984,
+      "learning_rate": 0.00018061978545887963,
+      "loss": 0.8603,
+      "mean_token_accuracy": 0.7400094747543335,
+      "num_tokens": 24451005.0,
+      "step": 10570
+    },
+    {
+      "epoch": 0.09700192536902906,
+      "learning_rate": 0.00018060144861098377,
+      "loss": 0.8448,
+      "mean_token_accuracy": 0.7427947461605072,
+      "num_tokens": 24474038.0,
+      "step": 10580
+    },
+    {
+      "epoch": 0.0970936096085083,
+      "learning_rate": 0.00018058311176308794,
+      "loss": 0.8342,
+      "mean_token_accuracy": 0.7481784880161285,
+      "num_tokens": 24497775.0,
+      "step": 10590
+    },
+    {
+      "epoch": 0.09718529384798753,
+      "learning_rate": 0.00018056477491519207,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7447938144207,
+      "num_tokens": 24520891.0,
+      "step": 10600
+    },
+    {
+      "epoch": 0.09727697808746677,
+      "learning_rate": 0.00018054643806729624,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.7382242679595947,
+      "num_tokens": 24544005.0,
+      "step": 10610
+    },
+    {
+      "epoch": 0.09736866232694599,
+      "learning_rate": 0.0001805281012194004,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.750395393371582,
+      "num_tokens": 24566293.0,
+      "step": 10620
+    },
+    {
+      "epoch": 0.09746034656642523,
+      "learning_rate": 0.00018050976437150454,
+      "loss": 0.8766,
+      "mean_token_accuracy": 0.7411053121089936,
+      "num_tokens": 24589497.0,
+      "step": 10630
+    },
+    {
+      "epoch": 0.09755203080590447,
+      "learning_rate": 0.0001804914275236087,
+      "loss": 0.859,
+      "mean_token_accuracy": 0.7437268733978272,
+      "num_tokens": 24611898.0,
+      "step": 10640
+    },
+    {
+      "epoch": 0.0976437150453837,
+      "learning_rate": 0.00018047309067571285,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7459022343158722,
+      "num_tokens": 24636191.0,
+      "step": 10650
+    },
+    {
+      "epoch": 0.09773539928486293,
+      "learning_rate": 0.000180454753827817,
+      "loss": 0.8518,
+      "mean_token_accuracy": 0.7471199095249176,
+      "num_tokens": 24659648.0,
+      "step": 10660
+    },
+    {
+      "epoch": 0.09782708352434216,
+      "learning_rate": 0.00018043641697992115,
+      "loss": 0.8632,
+      "mean_token_accuracy": 0.7440553903579712,
+      "num_tokens": 24682546.0,
+      "step": 10670
+    },
+    {
+      "epoch": 0.0979187677638214,
+      "learning_rate": 0.00018041808013202532,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.738535338640213,
+      "num_tokens": 24705496.0,
+      "step": 10680
+    },
+    {
+      "epoch": 0.09801045200330064,
+      "learning_rate": 0.00018039974328412948,
+      "loss": 0.8723,
+      "mean_token_accuracy": 0.7417737483978272,
+      "num_tokens": 24729135.0,
+      "step": 10690
+    },
+    {
+      "epoch": 0.09810213624277987,
+      "learning_rate": 0.00018038140643623362,
+      "loss": 0.8556,
+      "mean_token_accuracy": 0.7444552779197693,
+      "num_tokens": 24752266.0,
+      "step": 10700
+    },
+    {
+      "epoch": 0.0981938204822591,
+      "learning_rate": 0.00018036306958833779,
+      "loss": 0.8946,
+      "mean_token_accuracy": 0.7364124059677124,
+      "num_tokens": 24774873.0,
+      "step": 10710
+    },
+    {
+      "epoch": 0.09828550472173833,
+      "learning_rate": 0.00018034473274044192,
+      "loss": 0.8723,
+      "mean_token_accuracy": 0.7340504467487335,
+      "num_tokens": 24797617.0,
+      "step": 10720
+    },
+    {
+      "epoch": 0.09837718896121757,
+      "learning_rate": 0.00018032639589254606,
+      "loss": 0.8674,
+      "mean_token_accuracy": 0.7438235223293305,
+      "num_tokens": 24821454.0,
+      "step": 10730
+    },
+    {
+      "epoch": 0.0984688732006968,
+      "learning_rate": 0.00018030805904465023,
+      "loss": 0.9345,
+      "mean_token_accuracy": 0.7265106618404389,
+      "num_tokens": 24843857.0,
+      "step": 10740
+    },
+    {
+      "epoch": 0.09856055744017603,
+      "learning_rate": 0.0001802897221967544,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.7420856058597565,
+      "num_tokens": 24867256.0,
+      "step": 10750
+    },
+    {
+      "epoch": 0.09865224167965526,
+      "learning_rate": 0.00018027138534885856,
+      "loss": 0.9076,
+      "mean_token_accuracy": 0.7344007968902588,
+      "num_tokens": 24890614.0,
+      "step": 10760
+    },
+    {
+      "epoch": 0.0987439259191345,
+      "learning_rate": 0.0001802530485009627,
+      "loss": 0.8276,
+      "mean_token_accuracy": 0.7482860684394836,
+      "num_tokens": 24913653.0,
+      "step": 10770
+    },
+    {
+      "epoch": 0.09883561015861374,
+      "learning_rate": 0.00018023471165306683,
+      "loss": 0.9213,
+      "mean_token_accuracy": 0.7255630433559418,
+      "num_tokens": 24936755.0,
+      "step": 10780
+    },
+    {
+      "epoch": 0.09892729439809297,
+      "learning_rate": 0.000180216374805171,
+      "loss": 0.8781,
+      "mean_token_accuracy": 0.7428742527961731,
+      "num_tokens": 24960279.0,
+      "step": 10790
+    },
+    {
+      "epoch": 0.0990189786375722,
+      "learning_rate": 0.00018019803795727514,
+      "loss": 0.8811,
+      "mean_token_accuracy": 0.7340105235576629,
+      "num_tokens": 24982434.0,
+      "step": 10800
+    },
+    {
+      "epoch": 0.09911066287705143,
+      "learning_rate": 0.0001801797011093793,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7397896409034729,
+      "num_tokens": 25005674.0,
+      "step": 10810
+    },
+    {
+      "epoch": 0.09920234711653067,
+      "learning_rate": 0.00018016136426148347,
+      "loss": 0.8681,
+      "mean_token_accuracy": 0.7424762189388275,
+      "num_tokens": 25029219.0,
+      "step": 10820
+    },
+    {
+      "epoch": 0.0992940313560099,
+      "learning_rate": 0.0001801430274135876,
+      "loss": 0.8851,
+      "mean_token_accuracy": 0.7344321787357331,
+      "num_tokens": 25051932.0,
+      "step": 10830
+    },
+    {
+      "epoch": 0.09938571559548913,
+      "learning_rate": 0.00018012469056569177,
+      "loss": 0.9175,
+      "mean_token_accuracy": 0.7275677442550659,
+      "num_tokens": 25075651.0,
+      "step": 10840
+    },
+    {
+      "epoch": 0.09947739983496837,
+      "learning_rate": 0.0001801063537177959,
+      "loss": 0.8489,
+      "mean_token_accuracy": 0.7498330891132354,
+      "num_tokens": 25098936.0,
+      "step": 10850
+    },
+    {
+      "epoch": 0.0995690840744476,
+      "learning_rate": 0.00018008801686990008,
+      "loss": 0.9123,
+      "mean_token_accuracy": 0.7343634068965912,
+      "num_tokens": 25121710.0,
+      "step": 10860
+    },
+    {
+      "epoch": 0.09966076831392684,
+      "learning_rate": 0.00018006968002200422,
+      "loss": 0.8283,
+      "mean_token_accuracy": 0.7467993855476379,
+      "num_tokens": 25144107.0,
+      "step": 10870
+    },
+    {
+      "epoch": 0.09975245255340608,
+      "learning_rate": 0.00018005134317410838,
+      "loss": 0.8872,
+      "mean_token_accuracy": 0.7347580850124359,
+      "num_tokens": 25166860.0,
+      "step": 10880
+    },
+    {
+      "epoch": 0.0998441367928853,
+      "learning_rate": 0.00018003300632621255,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.7424611270427703,
+      "num_tokens": 25190298.0,
+      "step": 10890
+    },
+    {
+      "epoch": 0.09993582103236454,
+      "learning_rate": 0.00018001466947831668,
+      "loss": 0.8873,
+      "mean_token_accuracy": 0.7385188579559326,
+      "num_tokens": 25213595.0,
+      "step": 10900
+    },
+    {
+      "epoch": 0.10002750527184377,
+      "learning_rate": 0.00017999633263042085,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7345197677612305,
+      "num_tokens": 25236233.0,
+      "step": 10910
+    },
+    {
+      "epoch": 0.10011918951132301,
+      "learning_rate": 0.000179977995782525,
+      "loss": 0.9006,
+      "mean_token_accuracy": 0.7335421979427338,
+      "num_tokens": 25259579.0,
+      "step": 10920
+    },
+    {
+      "epoch": 0.10021087375080223,
+      "learning_rate": 0.00017995965893462913,
+      "loss": 0.891,
+      "mean_token_accuracy": 0.7389984250068664,
+      "num_tokens": 25282412.0,
+      "step": 10930
+    },
+    {
+      "epoch": 0.10030255799028147,
+      "learning_rate": 0.0001799413220867333,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7452229619026184,
+      "num_tokens": 25305507.0,
+      "step": 10940
+    },
+    {
+      "epoch": 0.1003942422297607,
+      "learning_rate": 0.00017992298523883746,
+      "loss": 0.8811,
+      "mean_token_accuracy": 0.7357382118701935,
+      "num_tokens": 25329647.0,
+      "step": 10950
+    },
+    {
+      "epoch": 0.10048592646923994,
+      "learning_rate": 0.00017990464839094162,
+      "loss": 0.8681,
+      "mean_token_accuracy": 0.7397084474563599,
+      "num_tokens": 25353597.0,
+      "step": 10960
+    },
+    {
+      "epoch": 0.10057761070871916,
+      "learning_rate": 0.00017988631154304576,
+      "loss": 0.8788,
+      "mean_token_accuracy": 0.744835251569748,
+      "num_tokens": 25377101.0,
+      "step": 10970
+    },
+    {
+      "epoch": 0.1006692949481984,
+      "learning_rate": 0.0001798679746951499,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.739315527677536,
+      "num_tokens": 25400047.0,
+      "step": 10980
+    },
+    {
+      "epoch": 0.10076097918767764,
+      "learning_rate": 0.00017984963784725406,
+      "loss": 0.8854,
+      "mean_token_accuracy": 0.733559387922287,
+      "num_tokens": 25423305.0,
+      "step": 10990
+    },
+    {
+      "epoch": 0.10085266342715687,
+      "learning_rate": 0.0001798313009993582,
+      "loss": 0.8619,
+      "mean_token_accuracy": 0.7448963761329651,
+      "num_tokens": 25446291.0,
+      "step": 11000
+    },
+    {
+      "epoch": 0.10094434766663611,
+      "learning_rate": 0.00017981296415146237,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.7425753057003022,
+      "num_tokens": 25469812.0,
+      "step": 11010
+    },
+    {
+      "epoch": 0.10103603190611533,
+      "learning_rate": 0.00017979462730356653,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7363907277584076,
+      "num_tokens": 25492493.0,
+      "step": 11020
+    },
+    {
+      "epoch": 0.10112771614559457,
+      "learning_rate": 0.00017977629045567067,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7488499164581299,
+      "num_tokens": 25516184.0,
+      "step": 11030
+    },
+    {
+      "epoch": 0.10121940038507381,
+      "learning_rate": 0.00017975795360777484,
+      "loss": 0.8408,
+      "mean_token_accuracy": 0.7440124809741974,
+      "num_tokens": 25539516.0,
+      "step": 11040
+    },
+    {
+      "epoch": 0.10131108462455304,
+      "learning_rate": 0.00017973961675987898,
+      "loss": 0.9001,
+      "mean_token_accuracy": 0.7308463513851166,
+      "num_tokens": 25563494.0,
+      "step": 11050
+    },
+    {
+      "epoch": 0.10140276886403227,
+      "learning_rate": 0.00017972127991198314,
+      "loss": 0.8787,
+      "mean_token_accuracy": 0.7380222618579865,
+      "num_tokens": 25587273.0,
+      "step": 11060
+    },
+    {
+      "epoch": 0.1014944531035115,
+      "learning_rate": 0.00017970294306408728,
+      "loss": 0.8705,
+      "mean_token_accuracy": 0.739801687002182,
+      "num_tokens": 25610363.0,
+      "step": 11070
+    },
+    {
+      "epoch": 0.10158613734299074,
+      "learning_rate": 0.00017968460621619145,
+      "loss": 0.8621,
+      "mean_token_accuracy": 0.7454406261444092,
+      "num_tokens": 25632975.0,
+      "step": 11080
+    },
+    {
+      "epoch": 0.10167782158246998,
+      "learning_rate": 0.0001796662693682956,
+      "loss": 0.8685,
+      "mean_token_accuracy": 0.7407660961151123,
+      "num_tokens": 25656036.0,
+      "step": 11090
+    },
+    {
+      "epoch": 0.10176950582194921,
+      "learning_rate": 0.00017964793252039975,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.739461475610733,
+      "num_tokens": 25679393.0,
+      "step": 11100
+    },
+    {
+      "epoch": 0.10186119006142844,
+      "learning_rate": 0.00017962959567250391,
+      "loss": 0.8769,
+      "mean_token_accuracy": 0.7353558778762818,
+      "num_tokens": 25702762.0,
+      "step": 11110
+    },
+    {
+      "epoch": 0.10195287430090767,
+      "learning_rate": 0.00017961125882460805,
+      "loss": 0.9116,
+      "mean_token_accuracy": 0.7379616916179657,
+      "num_tokens": 25725428.0,
+      "step": 11120
+    },
+    {
+      "epoch": 0.10204455854038691,
+      "learning_rate": 0.0001795929219767122,
+      "loss": 0.9091,
+      "mean_token_accuracy": 0.7295917868614197,
+      "num_tokens": 25748562.0,
+      "step": 11130
+    },
+    {
+      "epoch": 0.10213624277986615,
+      "learning_rate": 0.00017957458512881638,
+      "loss": 0.8869,
+      "mean_token_accuracy": 0.7402543067932129,
+      "num_tokens": 25772133.0,
+      "step": 11140
+    },
+    {
+      "epoch": 0.10222792701934537,
+      "learning_rate": 0.00017955624828092052,
+      "loss": 0.8982,
+      "mean_token_accuracy": 0.724835330247879,
+      "num_tokens": 25794433.0,
+      "step": 11150
+    },
+    {
+      "epoch": 0.1023196112588246,
+      "learning_rate": 0.0001795379114330247,
+      "loss": 0.8815,
+      "mean_token_accuracy": 0.7409479439258575,
+      "num_tokens": 25816895.0,
+      "step": 11160
+    },
+    {
+      "epoch": 0.10241129549830384,
+      "learning_rate": 0.00017951957458512883,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.7375987052917481,
+      "num_tokens": 25839641.0,
+      "step": 11170
+    },
+    {
+      "epoch": 0.10250297973778308,
+      "learning_rate": 0.00017950123773723296,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7427475035190583,
+      "num_tokens": 25862957.0,
+      "step": 11180
+    },
+    {
+      "epoch": 0.10259466397726232,
+      "learning_rate": 0.00017948290088933713,
+      "loss": 0.9123,
+      "mean_token_accuracy": 0.7334792375564575,
+      "num_tokens": 25886323.0,
+      "step": 11190
+    },
+    {
+      "epoch": 0.10268634821674154,
+      "learning_rate": 0.00017946456404144127,
+      "loss": 0.8911,
+      "mean_token_accuracy": 0.7416549921035767,
+      "num_tokens": 25909248.0,
+      "step": 11200
+    },
+    {
+      "epoch": 0.10277803245622077,
+      "learning_rate": 0.00017944622719354543,
+      "loss": 0.9048,
+      "mean_token_accuracy": 0.730409562587738,
+      "num_tokens": 25932875.0,
+      "step": 11210
+    },
+    {
+      "epoch": 0.10286971669570001,
+      "learning_rate": 0.0001794278903456496,
+      "loss": 0.86,
+      "mean_token_accuracy": 0.7465738594532013,
+      "num_tokens": 25955886.0,
+      "step": 11220
+    },
+    {
+      "epoch": 0.10296140093517925,
+      "learning_rate": 0.00017940955349775374,
+      "loss": 0.9391,
+      "mean_token_accuracy": 0.7274121463298797,
+      "num_tokens": 25979004.0,
+      "step": 11230
+    },
+    {
+      "epoch": 0.10305308517465847,
+      "learning_rate": 0.0001793912166498579,
+      "loss": 0.8983,
+      "mean_token_accuracy": 0.7318204581737519,
+      "num_tokens": 26002269.0,
+      "step": 11240
+    },
+    {
+      "epoch": 0.10314476941413771,
+      "learning_rate": 0.00017937287980196204,
+      "loss": 0.879,
+      "mean_token_accuracy": 0.7433506727218628,
+      "num_tokens": 26026515.0,
+      "step": 11250
+    },
+    {
+      "epoch": 0.10323645365361694,
+      "learning_rate": 0.0001793545429540662,
+      "loss": 0.8626,
+      "mean_token_accuracy": 0.7416462600231171,
+      "num_tokens": 26050493.0,
+      "step": 11260
+    },
+    {
+      "epoch": 0.10332813789309618,
+      "learning_rate": 0.00017933620610617037,
+      "loss": 0.9057,
+      "mean_token_accuracy": 0.7388849973678588,
+      "num_tokens": 26073510.0,
+      "step": 11270
+    },
+    {
+      "epoch": 0.1034198221325754,
+      "learning_rate": 0.0001793178692582745,
+      "loss": 0.9024,
+      "mean_token_accuracy": 0.7374939978122711,
+      "num_tokens": 26096348.0,
+      "step": 11280
+    },
+    {
+      "epoch": 0.10351150637205464,
+      "learning_rate": 0.00017929953241037868,
+      "loss": 0.8201,
+      "mean_token_accuracy": 0.7585200309753418,
+      "num_tokens": 26119821.0,
+      "step": 11290
+    },
+    {
+      "epoch": 0.10360319061153388,
+      "learning_rate": 0.0001792811955624828,
+      "loss": 0.8929,
+      "mean_token_accuracy": 0.7429881691932678,
+      "num_tokens": 26143672.0,
+      "step": 11300
+    },
+    {
+      "epoch": 0.10369487485101311,
+      "learning_rate": 0.00017926285871458698,
+      "loss": 0.8975,
+      "mean_token_accuracy": 0.7410037696361542,
+      "num_tokens": 26166411.0,
+      "step": 11310
+    },
+    {
+      "epoch": 0.10378655909049235,
+      "learning_rate": 0.00017924452186669112,
+      "loss": 0.9039,
+      "mean_token_accuracy": 0.7327254235744476,
+      "num_tokens": 26190002.0,
+      "step": 11320
+    },
+    {
+      "epoch": 0.10387824332997157,
+      "learning_rate": 0.00017922618501879526,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.739010089635849,
+      "num_tokens": 26213467.0,
+      "step": 11330
+    },
+    {
+      "epoch": 0.10396992756945081,
+      "learning_rate": 0.00017920784817089945,
+      "loss": 0.8971,
+      "mean_token_accuracy": 0.7366046905517578,
+      "num_tokens": 26237288.0,
+      "step": 11340
+    },
+    {
+      "epoch": 0.10406161180893005,
+      "learning_rate": 0.00017918951132300359,
+      "loss": 0.8914,
+      "mean_token_accuracy": 0.7392790913581848,
+      "num_tokens": 26260388.0,
+      "step": 11350
+    },
+    {
+      "epoch": 0.10415329604840928,
+      "learning_rate": 0.00017917117447510775,
+      "loss": 0.883,
+      "mean_token_accuracy": 0.7385203599929809,
+      "num_tokens": 26284016.0,
+      "step": 11360
+    },
+    {
+      "epoch": 0.1042449802878885,
+      "learning_rate": 0.0001791528376272119,
+      "loss": 0.9186,
+      "mean_token_accuracy": 0.7294337689876557,
+      "num_tokens": 26307725.0,
+      "step": 11370
+    },
+    {
+      "epoch": 0.10433666452736774,
+      "learning_rate": 0.00017913450077931603,
+      "loss": 0.872,
+      "mean_token_accuracy": 0.740067058801651,
+      "num_tokens": 26330659.0,
+      "step": 11380
+    },
+    {
+      "epoch": 0.10442834876684698,
+      "learning_rate": 0.0001791161639314202,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7400892734527588,
+      "num_tokens": 26353488.0,
+      "step": 11390
+    },
+    {
+      "epoch": 0.10452003300632622,
+      "learning_rate": 0.00017909782708352436,
+      "loss": 0.8859,
+      "mean_token_accuracy": 0.7375518560409546,
+      "num_tokens": 26376672.0,
+      "step": 11400
+    },
+    {
+      "epoch": 0.10461171724580545,
+      "learning_rate": 0.0001790794902356285,
+      "loss": 0.8924,
+      "mean_token_accuracy": 0.7375367045402527,
+      "num_tokens": 26400298.0,
+      "step": 11410
+    },
+    {
+      "epoch": 0.10470340148528468,
+      "learning_rate": 0.00017906115338773266,
+      "loss": 0.8716,
+      "mean_token_accuracy": 0.7399585485458374,
+      "num_tokens": 26422697.0,
+      "step": 11420
+    },
+    {
+      "epoch": 0.10479508572476391,
+      "learning_rate": 0.0001790428165398368,
+      "loss": 0.877,
+      "mean_token_accuracy": 0.7456361293792725,
+      "num_tokens": 26445665.0,
+      "step": 11430
+    },
+    {
+      "epoch": 0.10488676996424315,
+      "learning_rate": 0.00017902447969194097,
+      "loss": 0.8833,
+      "mean_token_accuracy": 0.7388562619686126,
+      "num_tokens": 26469096.0,
+      "step": 11440
+    },
+    {
+      "epoch": 0.10497845420372239,
+      "learning_rate": 0.0001790061428440451,
+      "loss": 0.8758,
+      "mean_token_accuracy": 0.744290167093277,
+      "num_tokens": 26492195.0,
+      "step": 11450
+    },
+    {
+      "epoch": 0.10507013844320161,
+      "learning_rate": 0.00017898780599614927,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7436373651027679,
+      "num_tokens": 26515123.0,
+      "step": 11460
+    },
+    {
+      "epoch": 0.10516182268268084,
+      "learning_rate": 0.00017896946914825344,
+      "loss": 0.8677,
+      "mean_token_accuracy": 0.7376021027565003,
+      "num_tokens": 26538101.0,
+      "step": 11470
+    },
+    {
+      "epoch": 0.10525350692216008,
+      "learning_rate": 0.00017895113230035757,
+      "loss": 0.9099,
+      "mean_token_accuracy": 0.7306402206420899,
+      "num_tokens": 26560984.0,
+      "step": 11480
+    },
+    {
+      "epoch": 0.10534519116163932,
+      "learning_rate": 0.00017893279545246174,
+      "loss": 0.8278,
+      "mean_token_accuracy": 0.7509335875511169,
+      "num_tokens": 26584906.0,
+      "step": 11490
+    },
+    {
+      "epoch": 0.10543687540111855,
+      "learning_rate": 0.00017891445860456588,
+      "loss": 0.8929,
+      "mean_token_accuracy": 0.7350357353687287,
+      "num_tokens": 26607470.0,
+      "step": 11500
+    },
+    {
+      "epoch": 0.10552855964059778,
+      "learning_rate": 0.00017889612175667004,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.7428408443927765,
+      "num_tokens": 26630872.0,
+      "step": 11510
+    },
+    {
+      "epoch": 0.10562024388007701,
+      "learning_rate": 0.00017887778490877418,
+      "loss": 0.9204,
+      "mean_token_accuracy": 0.7317827820777894,
+      "num_tokens": 26653759.0,
+      "step": 11520
+    },
+    {
+      "epoch": 0.10571192811955625,
+      "learning_rate": 0.00017885944806087835,
+      "loss": 0.9528,
+      "mean_token_accuracy": 0.7178638577461243,
+      "num_tokens": 26677121.0,
+      "step": 11530
+    },
+    {
+      "epoch": 0.10580361235903549,
+      "learning_rate": 0.0001788411112129825,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7380670189857483,
+      "num_tokens": 26700678.0,
+      "step": 11540
+    },
+    {
+      "epoch": 0.10589529659851471,
+      "learning_rate": 0.00017882277436508665,
+      "loss": 0.9528,
+      "mean_token_accuracy": 0.7202961683273316,
+      "num_tokens": 26724084.0,
+      "step": 11550
+    },
+    {
+      "epoch": 0.10598698083799395,
+      "learning_rate": 0.00017880443751719082,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7396840393543244,
+      "num_tokens": 26746786.0,
+      "step": 11560
+    },
+    {
+      "epoch": 0.10607866507747318,
+      "learning_rate": 0.00017878610066929495,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.7380404055118561,
+      "num_tokens": 26770909.0,
+      "step": 11570
+    },
+    {
+      "epoch": 0.10617034931695242,
+      "learning_rate": 0.0001787677638213991,
+      "loss": 0.8609,
+      "mean_token_accuracy": 0.7454233050346375,
+      "num_tokens": 26794119.0,
+      "step": 11580
+    },
+    {
+      "epoch": 0.10626203355643164,
+      "learning_rate": 0.00017874942697350326,
+      "loss": 0.8557,
+      "mean_token_accuracy": 0.7433999359607697,
+      "num_tokens": 26817096.0,
+      "step": 11590
+    },
+    {
+      "epoch": 0.10635371779591088,
+      "learning_rate": 0.00017873109012560742,
+      "loss": 0.8296,
+      "mean_token_accuracy": 0.7519002318382263,
+      "num_tokens": 26840201.0,
+      "step": 11600
+    },
+    {
+      "epoch": 0.10644540203539012,
+      "learning_rate": 0.00017871275327771156,
+      "loss": 0.8736,
+      "mean_token_accuracy": 0.7433995008468628,
+      "num_tokens": 26862945.0,
+      "step": 11610
+    },
+    {
+      "epoch": 0.10653708627486935,
+      "learning_rate": 0.00017869441642981573,
+      "loss": 0.9185,
+      "mean_token_accuracy": 0.7319000780582428,
+      "num_tokens": 26886709.0,
+      "step": 11620
+    },
+    {
+      "epoch": 0.10662877051434859,
+      "learning_rate": 0.00017867607958191987,
+      "loss": 0.8349,
+      "mean_token_accuracy": 0.7504717409610748,
+      "num_tokens": 26909972.0,
+      "step": 11630
+    },
+    {
+      "epoch": 0.10672045475382781,
+      "learning_rate": 0.00017865774273402403,
+      "loss": 0.8242,
+      "mean_token_accuracy": 0.7521531820297241,
+      "num_tokens": 26932887.0,
+      "step": 11640
+    },
+    {
+      "epoch": 0.10681213899330705,
+      "learning_rate": 0.00017863940588612817,
+      "loss": 0.8836,
+      "mean_token_accuracy": 0.7407630920410156,
+      "num_tokens": 26955995.0,
+      "step": 11650
+    },
+    {
+      "epoch": 0.10690382323278629,
+      "learning_rate": 0.00017862106903823233,
+      "loss": 0.8662,
+      "mean_token_accuracy": 0.7427942395210266,
+      "num_tokens": 26979091.0,
+      "step": 11660
+    },
+    {
+      "epoch": 0.10699550747226552,
+      "learning_rate": 0.0001786027321903365,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7429852724075318,
+      "num_tokens": 27001826.0,
+      "step": 11670
+    },
+    {
+      "epoch": 0.10708719171174474,
+      "learning_rate": 0.00017858439534244064,
+      "loss": 0.883,
+      "mean_token_accuracy": 0.7366941869258881,
+      "num_tokens": 27024878.0,
+      "step": 11680
+    },
+    {
+      "epoch": 0.10717887595122398,
+      "learning_rate": 0.0001785660584945448,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7402225613594056,
+      "num_tokens": 27047969.0,
+      "step": 11690
+    },
+    {
+      "epoch": 0.10727056019070322,
+      "learning_rate": 0.00017854772164664894,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.743378323316574,
+      "num_tokens": 27070948.0,
+      "step": 11700
+    },
+    {
+      "epoch": 0.10736224443018245,
+      "learning_rate": 0.0001785293847987531,
+      "loss": 0.9495,
+      "mean_token_accuracy": 0.7240637183189392,
+      "num_tokens": 27094089.0,
+      "step": 11710
+    },
+    {
+      "epoch": 0.10745392866966169,
+      "learning_rate": 0.00017851104795085725,
+      "loss": 0.8912,
+      "mean_token_accuracy": 0.7427623808383942,
+      "num_tokens": 27117513.0,
+      "step": 11720
+    },
+    {
+      "epoch": 0.10754561290914091,
+      "learning_rate": 0.0001784927111029614,
+      "loss": 0.8667,
+      "mean_token_accuracy": 0.7423351585865021,
+      "num_tokens": 27141568.0,
+      "step": 11730
+    },
+    {
+      "epoch": 0.10763729714862015,
+      "learning_rate": 0.00017847437425506558,
+      "loss": 0.881,
+      "mean_token_accuracy": 0.7310326814651489,
+      "num_tokens": 27163758.0,
+      "step": 11740
+    },
+    {
+      "epoch": 0.10772898138809939,
+      "learning_rate": 0.00017845603740716972,
+      "loss": 0.8859,
+      "mean_token_accuracy": 0.7359584927558899,
+      "num_tokens": 27187215.0,
+      "step": 11750
+    },
+    {
+      "epoch": 0.10782066562757862,
+      "learning_rate": 0.00017843770055927388,
+      "loss": 0.8947,
+      "mean_token_accuracy": 0.7310957431793212,
+      "num_tokens": 27210322.0,
+      "step": 11760
+    },
+    {
+      "epoch": 0.10791234986705785,
+      "learning_rate": 0.00017841936371137802,
+      "loss": 0.9138,
+      "mean_token_accuracy": 0.7403761386871338,
+      "num_tokens": 27233377.0,
+      "step": 11770
+    },
+    {
+      "epoch": 0.10800403410653708,
+      "learning_rate": 0.00017840102686348216,
+      "loss": 0.8994,
+      "mean_token_accuracy": 0.7380115389823914,
+      "num_tokens": 27257227.0,
+      "step": 11780
+    },
+    {
+      "epoch": 0.10809571834601632,
+      "learning_rate": 0.00017838269001558635,
+      "loss": 0.8864,
+      "mean_token_accuracy": 0.738403606414795,
+      "num_tokens": 27280152.0,
+      "step": 11790
+    },
+    {
+      "epoch": 0.10818740258549556,
+      "learning_rate": 0.0001783643531676905,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.7386031329631806,
+      "num_tokens": 27304010.0,
+      "step": 11800
+    },
+    {
+      "epoch": 0.1082790868249748,
+      "learning_rate": 0.00017834601631979463,
+      "loss": 0.8917,
+      "mean_token_accuracy": 0.7318453073501587,
+      "num_tokens": 27326802.0,
+      "step": 11810
+    },
+    {
+      "epoch": 0.10837077106445402,
+      "learning_rate": 0.0001783276794718988,
+      "loss": 0.8832,
+      "mean_token_accuracy": 0.7354330539703369,
+      "num_tokens": 27349795.0,
+      "step": 11820
+    },
+    {
+      "epoch": 0.10846245530393325,
+      "learning_rate": 0.00017830934262400293,
+      "loss": 0.903,
+      "mean_token_accuracy": 0.7374026656150818,
+      "num_tokens": 27373434.0,
+      "step": 11830
+    },
+    {
+      "epoch": 0.10855413954341249,
+      "learning_rate": 0.0001782910057761071,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7449547350406647,
+      "num_tokens": 27396953.0,
+      "step": 11840
+    },
+    {
+      "epoch": 0.10864582378289173,
+      "learning_rate": 0.00017827266892821123,
+      "loss": 0.8781,
+      "mean_token_accuracy": 0.7443978965282441,
+      "num_tokens": 27420440.0,
+      "step": 11850
+    },
+    {
+      "epoch": 0.10873750802237095,
+      "learning_rate": 0.0001782543320803154,
+      "loss": 0.8807,
+      "mean_token_accuracy": 0.7423559188842773,
+      "num_tokens": 27443543.0,
+      "step": 11860
+    },
+    {
+      "epoch": 0.10882919226185019,
+      "learning_rate": 0.00017823599523241956,
+      "loss": 0.8782,
+      "mean_token_accuracy": 0.7370914876461029,
+      "num_tokens": 27466587.0,
+      "step": 11870
+    },
+    {
+      "epoch": 0.10892087650132942,
+      "learning_rate": 0.0001782176583845237,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.7436290860176087,
+      "num_tokens": 27488972.0,
+      "step": 11880
+    },
+    {
+      "epoch": 0.10901256074080866,
+      "learning_rate": 0.00017819932153662787,
+      "loss": 0.882,
+      "mean_token_accuracy": 0.74011932015419,
+      "num_tokens": 27511518.0,
+      "step": 11890
+    },
+    {
+      "epoch": 0.10910424498028788,
+      "learning_rate": 0.000178180984688732,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7381210803985596,
+      "num_tokens": 27533827.0,
+      "step": 11900
+    },
+    {
+      "epoch": 0.10919592921976712,
+      "learning_rate": 0.00017816264784083617,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.74366534948349,
+      "num_tokens": 27557508.0,
+      "step": 11910
+    },
+    {
+      "epoch": 0.10928761345924635,
+      "learning_rate": 0.00017814431099294034,
+      "loss": 0.8958,
+      "mean_token_accuracy": 0.7379620313644409,
+      "num_tokens": 27580767.0,
+      "step": 11920
+    },
+    {
+      "epoch": 0.10937929769872559,
+      "learning_rate": 0.00017812597414504448,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7385945558547974,
+      "num_tokens": 27603997.0,
+      "step": 11930
+    },
+    {
+      "epoch": 0.10947098193820483,
+      "learning_rate": 0.00017810763729714864,
+      "loss": 0.8549,
+      "mean_token_accuracy": 0.7429787874221802,
+      "num_tokens": 27626885.0,
+      "step": 11940
+    },
+    {
+      "epoch": 0.10956266617768405,
+      "learning_rate": 0.00017808930044925278,
+      "loss": 0.8777,
+      "mean_token_accuracy": 0.7384230136871338,
+      "num_tokens": 27649938.0,
+      "step": 11950
+    },
+    {
+      "epoch": 0.10965435041716329,
+      "learning_rate": 0.00017807096360135695,
+      "loss": 0.8872,
+      "mean_token_accuracy": 0.7369516789913177,
+      "num_tokens": 27673030.0,
+      "step": 11960
+    },
+    {
+      "epoch": 0.10974603465664252,
+      "learning_rate": 0.00017805262675346108,
+      "loss": 0.8647,
+      "mean_token_accuracy": 0.7508829653263092,
+      "num_tokens": 27695459.0,
+      "step": 11970
+    },
+    {
+      "epoch": 0.10983771889612176,
+      "learning_rate": 0.00017803428990556522,
+      "loss": 0.8792,
+      "mean_token_accuracy": 0.7326285004615783,
+      "num_tokens": 27718179.0,
+      "step": 11980
+    },
+    {
+      "epoch": 0.10992940313560098,
+      "learning_rate": 0.00017801595305766941,
+      "loss": 0.8683,
+      "mean_token_accuracy": 0.7448980331420898,
+      "num_tokens": 27740608.0,
+      "step": 11990
+    },
+    {
+      "epoch": 0.11002108737508022,
+      "learning_rate": 0.00017799761620977355,
+      "loss": 0.8489,
+      "mean_token_accuracy": 0.7497345745563507,
+      "num_tokens": 27764201.0,
+      "step": 12000
+    },
+    {
+      "epoch": 0.11011277161455946,
+      "learning_rate": 0.00017797927936187772,
+      "loss": 0.8729,
+      "mean_token_accuracy": 0.7328741371631622,
+      "num_tokens": 27787189.0,
+      "step": 12010
+    },
+    {
+      "epoch": 0.1102044558540387,
+      "learning_rate": 0.00017796094251398186,
+      "loss": 0.9071,
+      "mean_token_accuracy": 0.7407233119010925,
+      "num_tokens": 27811354.0,
+      "step": 12020
+    },
+    {
+      "epoch": 0.11029614009351793,
+      "learning_rate": 0.000177942605666086,
+      "loss": 0.8717,
+      "mean_token_accuracy": 0.7401291847229003,
+      "num_tokens": 27834739.0,
+      "step": 12030
+    },
+    {
+      "epoch": 0.11038782433299715,
+      "learning_rate": 0.00017792426881819016,
+      "loss": 0.8729,
+      "mean_token_accuracy": 0.7361323416233063,
+      "num_tokens": 27857881.0,
+      "step": 12040
+    },
+    {
+      "epoch": 0.11047950857247639,
+      "learning_rate": 0.0001779059319702943,
+      "loss": 0.8754,
+      "mean_token_accuracy": 0.7374109506607056,
+      "num_tokens": 27880926.0,
+      "step": 12050
+    },
+    {
+      "epoch": 0.11057119281195563,
+      "learning_rate": 0.00017788759512239846,
+      "loss": 0.916,
+      "mean_token_accuracy": 0.7358609676361084,
+      "num_tokens": 27903447.0,
+      "step": 12060
+    },
+    {
+      "epoch": 0.11066287705143486,
+      "learning_rate": 0.00017786925827450263,
+      "loss": 0.8911,
+      "mean_token_accuracy": 0.734874165058136,
+      "num_tokens": 27926965.0,
+      "step": 12070
+    },
+    {
+      "epoch": 0.11075456129091409,
+      "learning_rate": 0.00017785092142660677,
+      "loss": 0.852,
+      "mean_token_accuracy": 0.743153166770935,
+      "num_tokens": 27949419.0,
+      "step": 12080
+    },
+    {
+      "epoch": 0.11084624553039332,
+      "learning_rate": 0.00017783258457871093,
+      "loss": 0.8693,
+      "mean_token_accuracy": 0.7415815830230713,
+      "num_tokens": 27973278.0,
+      "step": 12090
+    },
+    {
+      "epoch": 0.11093792976987256,
+      "learning_rate": 0.00017781424773081507,
+      "loss": 0.8617,
+      "mean_token_accuracy": 0.7533626735210419,
+      "num_tokens": 27995873.0,
+      "step": 12100
+    },
+    {
+      "epoch": 0.1110296140093518,
+      "learning_rate": 0.00017779591088291924,
+      "loss": 0.8951,
+      "mean_token_accuracy": 0.7394928812980652,
+      "num_tokens": 28019131.0,
+      "step": 12110
+    },
+    {
+      "epoch": 0.11112129824883103,
+      "learning_rate": 0.0001777775740350234,
+      "loss": 0.904,
+      "mean_token_accuracy": 0.7413673281669617,
+      "num_tokens": 28042400.0,
+      "step": 12120
+    },
+    {
+      "epoch": 0.11121298248831026,
+      "learning_rate": 0.00017775923718712754,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.7390436410903931,
+      "num_tokens": 28065576.0,
+      "step": 12130
+    },
+    {
+      "epoch": 0.11130466672778949,
+      "learning_rate": 0.0001777409003392317,
+      "loss": 0.8783,
+      "mean_token_accuracy": 0.7301335394382477,
+      "num_tokens": 28088931.0,
+      "step": 12140
+    },
+    {
+      "epoch": 0.11139635096726873,
+      "learning_rate": 0.00017772256349133584,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7469180881977081,
+      "num_tokens": 28112217.0,
+      "step": 12150
+    },
+    {
+      "epoch": 0.11148803520674797,
+      "learning_rate": 0.00017770422664344,
+      "loss": 0.8482,
+      "mean_token_accuracy": 0.7464567482471466,
+      "num_tokens": 28135744.0,
+      "step": 12160
+    },
+    {
+      "epoch": 0.11157971944622719,
+      "learning_rate": 0.00017768588979554415,
+      "loss": 0.907,
+      "mean_token_accuracy": 0.7300693154335022,
+      "num_tokens": 28158330.0,
+      "step": 12170
+    },
+    {
+      "epoch": 0.11167140368570642,
+      "learning_rate": 0.00017766755294764829,
+      "loss": 0.8557,
+      "mean_token_accuracy": 0.7448145389556885,
+      "num_tokens": 28182363.0,
+      "step": 12180
+    },
+    {
+      "epoch": 0.11176308792518566,
+      "learning_rate": 0.00017764921609975248,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.748042207956314,
+      "num_tokens": 28205894.0,
+      "step": 12190
+    },
+    {
+      "epoch": 0.1118547721646649,
+      "learning_rate": 0.00017763087925185662,
+      "loss": 0.894,
+      "mean_token_accuracy": 0.7306191265583039,
+      "num_tokens": 28228515.0,
+      "step": 12200
+    },
+    {
+      "epoch": 0.11194645640414412,
+      "learning_rate": 0.00017761254240396078,
+      "loss": 0.8945,
+      "mean_token_accuracy": 0.7436901509761811,
+      "num_tokens": 28251754.0,
+      "step": 12210
+    },
+    {
+      "epoch": 0.11203814064362336,
+      "learning_rate": 0.00017759420555606492,
+      "loss": 0.851,
+      "mean_token_accuracy": 0.7463106334209442,
+      "num_tokens": 28274595.0,
+      "step": 12220
+    },
+    {
+      "epoch": 0.1121298248831026,
+      "learning_rate": 0.00017757586870816906,
+      "loss": 0.9214,
+      "mean_token_accuracy": 0.7293797373771668,
+      "num_tokens": 28297840.0,
+      "step": 12230
+    },
+    {
+      "epoch": 0.11222150912258183,
+      "learning_rate": 0.00017755753186027322,
+      "loss": 0.8924,
+      "mean_token_accuracy": 0.7376773357391357,
+      "num_tokens": 28320452.0,
+      "step": 12240
+    },
+    {
+      "epoch": 0.11231319336206107,
+      "learning_rate": 0.0001775391950123774,
+      "loss": 0.8673,
+      "mean_token_accuracy": 0.7442586421966553,
+      "num_tokens": 28343099.0,
+      "step": 12250
+    },
+    {
+      "epoch": 0.11240487760154029,
+      "learning_rate": 0.00017752085816448153,
+      "loss": 0.8665,
+      "mean_token_accuracy": 0.7418373227119446,
+      "num_tokens": 28366627.0,
+      "step": 12260
+    },
+    {
+      "epoch": 0.11249656184101953,
+      "learning_rate": 0.0001775025213165857,
+      "loss": 0.8692,
+      "mean_token_accuracy": 0.7441459059715271,
+      "num_tokens": 28389832.0,
+      "step": 12270
+    },
+    {
+      "epoch": 0.11258824608049876,
+      "learning_rate": 0.00017748418446868983,
+      "loss": 0.9088,
+      "mean_token_accuracy": 0.7338151514530182,
+      "num_tokens": 28413462.0,
+      "step": 12280
+    },
+    {
+      "epoch": 0.112679930319978,
+      "learning_rate": 0.000177465847620794,
+      "loss": 0.909,
+      "mean_token_accuracy": 0.7330934405326843,
+      "num_tokens": 28437242.0,
+      "step": 12290
+    },
+    {
+      "epoch": 0.11277161455945722,
+      "learning_rate": 0.00017744751077289814,
+      "loss": 0.9082,
+      "mean_token_accuracy": 0.7337328672409058,
+      "num_tokens": 28460968.0,
+      "step": 12300
+    },
+    {
+      "epoch": 0.11286329879893646,
+      "learning_rate": 0.0001774291739250023,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7393056094646454,
+      "num_tokens": 28483382.0,
+      "step": 12310
+    },
+    {
+      "epoch": 0.1129549830384157,
+      "learning_rate": 0.00017741083707710647,
+      "loss": 0.8754,
+      "mean_token_accuracy": 0.7461185574531555,
+      "num_tokens": 28506229.0,
+      "step": 12320
+    },
+    {
+      "epoch": 0.11304666727789493,
+      "learning_rate": 0.0001773925002292106,
+      "loss": 0.9038,
+      "mean_token_accuracy": 0.7372637808322906,
+      "num_tokens": 28529589.0,
+      "step": 12330
+    },
+    {
+      "epoch": 0.11313835151737417,
+      "learning_rate": 0.00017737416338131477,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.7373740017414093,
+      "num_tokens": 28552097.0,
+      "step": 12340
+    },
+    {
+      "epoch": 0.11323003575685339,
+      "learning_rate": 0.0001773558265334189,
+      "loss": 0.8685,
+      "mean_token_accuracy": 0.7405568540096283,
+      "num_tokens": 28575170.0,
+      "step": 12350
+    },
+    {
+      "epoch": 0.11332171999633263,
+      "learning_rate": 0.00017733748968552307,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.744938200712204,
+      "num_tokens": 28597896.0,
+      "step": 12360
+    },
+    {
+      "epoch": 0.11341340423581187,
+      "learning_rate": 0.0001773191528376272,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7475294947624207,
+      "num_tokens": 28620245.0,
+      "step": 12370
+    },
+    {
+      "epoch": 0.1135050884752911,
+      "learning_rate": 0.00017730081598973138,
+      "loss": 0.8944,
+      "mean_token_accuracy": 0.7386161625385285,
+      "num_tokens": 28643082.0,
+      "step": 12380
+    },
+    {
+      "epoch": 0.11359677271477032,
+      "learning_rate": 0.00017728247914183554,
+      "loss": 0.8798,
+      "mean_token_accuracy": 0.7405800342559814,
+      "num_tokens": 28666153.0,
+      "step": 12390
+    },
+    {
+      "epoch": 0.11368845695424956,
+      "learning_rate": 0.00017726414229393968,
+      "loss": 0.8729,
+      "mean_token_accuracy": 0.7414615154266357,
+      "num_tokens": 28689137.0,
+      "step": 12400
+    },
+    {
+      "epoch": 0.1137801411937288,
+      "learning_rate": 0.00017724580544604385,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.7416299939155578,
+      "num_tokens": 28713125.0,
+      "step": 12410
+    },
+    {
+      "epoch": 0.11387182543320803,
+      "learning_rate": 0.00017722746859814799,
+      "loss": 0.8854,
+      "mean_token_accuracy": 0.74670529961586,
+      "num_tokens": 28736225.0,
+      "step": 12420
+    },
+    {
+      "epoch": 0.11396350967268727,
+      "learning_rate": 0.00017720913175025212,
+      "loss": 0.8994,
+      "mean_token_accuracy": 0.736379188299179,
+      "num_tokens": 28758825.0,
+      "step": 12430
+    },
+    {
+      "epoch": 0.1140551939121665,
+      "learning_rate": 0.0001771907949023563,
+      "loss": 0.8572,
+      "mean_token_accuracy": 0.742750632762909,
+      "num_tokens": 28782712.0,
+      "step": 12440
+    },
+    {
+      "epoch": 0.11414687815164573,
+      "learning_rate": 0.00017717245805446045,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.737398874759674,
+      "num_tokens": 28805646.0,
+      "step": 12450
+    },
+    {
+      "epoch": 0.11423856239112497,
+      "learning_rate": 0.0001771541212065646,
+      "loss": 0.8893,
+      "mean_token_accuracy": 0.736296546459198,
+      "num_tokens": 28828809.0,
+      "step": 12460
+    },
+    {
+      "epoch": 0.1143302466306042,
+      "learning_rate": 0.00017713578435866876,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.743328982591629,
+      "num_tokens": 28851700.0,
+      "step": 12470
+    },
+    {
+      "epoch": 0.11442193087008343,
+      "learning_rate": 0.0001771174475107729,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.7449788212776184,
+      "num_tokens": 28874778.0,
+      "step": 12480
+    },
+    {
+      "epoch": 0.11451361510956266,
+      "learning_rate": 0.00017709911066287706,
+      "loss": 0.8932,
+      "mean_token_accuracy": 0.7372696399688721,
+      "num_tokens": 28897409.0,
+      "step": 12490
+    },
+    {
+      "epoch": 0.1146052993490419,
+      "learning_rate": 0.0001770807738149812,
+      "loss": 0.8597,
+      "mean_token_accuracy": 0.7413606107234955,
+      "num_tokens": 28920429.0,
+      "step": 12500
+    },
+    {
+      "epoch": 0.11469698358852114,
+      "learning_rate": 0.00017706243696708537,
+      "loss": 0.8942,
+      "mean_token_accuracy": 0.7346774339675903,
+      "num_tokens": 28943335.0,
+      "step": 12510
+    },
+    {
+      "epoch": 0.11478866782800037,
+      "learning_rate": 0.00017704410011918953,
+      "loss": 0.906,
+      "mean_token_accuracy": 0.738807737827301,
+      "num_tokens": 28966256.0,
+      "step": 12520
+    },
+    {
+      "epoch": 0.1148803520674796,
+      "learning_rate": 0.00017702576327129367,
+      "loss": 0.9194,
+      "mean_token_accuracy": 0.7317074060440063,
+      "num_tokens": 28989502.0,
+      "step": 12530
+    },
+    {
+      "epoch": 0.11497203630695883,
+      "learning_rate": 0.00017700742642339783,
+      "loss": 0.9073,
+      "mean_token_accuracy": 0.7361348927021026,
+      "num_tokens": 29012615.0,
+      "step": 12540
+    },
+    {
+      "epoch": 0.11506372054643807,
+      "learning_rate": 0.00017698908957550197,
+      "loss": 0.899,
+      "mean_token_accuracy": 0.7338710904121399,
+      "num_tokens": 29036250.0,
+      "step": 12550
+    },
+    {
+      "epoch": 0.1151554047859173,
+      "learning_rate": 0.00017697075272760614,
+      "loss": 0.8771,
+      "mean_token_accuracy": 0.7368703424930573,
+      "num_tokens": 29060229.0,
+      "step": 12560
+    },
+    {
+      "epoch": 0.11524708902539653,
+      "learning_rate": 0.00017695241587971028,
+      "loss": 0.8591,
+      "mean_token_accuracy": 0.7381478071212768,
+      "num_tokens": 29082833.0,
+      "step": 12570
+    },
+    {
+      "epoch": 0.11533877326487577,
+      "learning_rate": 0.00017693407903181444,
+      "loss": 0.8957,
+      "mean_token_accuracy": 0.7396318912506104,
+      "num_tokens": 29105956.0,
+      "step": 12580
+    },
+    {
+      "epoch": 0.115430457504355,
+      "learning_rate": 0.0001769157421839186,
+      "loss": 0.904,
+      "mean_token_accuracy": 0.7330422759056091,
+      "num_tokens": 29129058.0,
+      "step": 12590
+    },
+    {
+      "epoch": 0.11552214174383424,
+      "learning_rate": 0.00017689740533602275,
+      "loss": 0.8699,
+      "mean_token_accuracy": 0.7421502351760865,
+      "num_tokens": 29152454.0,
+      "step": 12600
+    },
+    {
+      "epoch": 0.11561382598331346,
+      "learning_rate": 0.0001768790684881269,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7383241057395935,
+      "num_tokens": 29176404.0,
+      "step": 12610
+    },
+    {
+      "epoch": 0.1157055102227927,
+      "learning_rate": 0.00017686073164023105,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7425327897071838,
+      "num_tokens": 29199327.0,
+      "step": 12620
+    },
+    {
+      "epoch": 0.11579719446227194,
+      "learning_rate": 0.0001768423947923352,
+      "loss": 0.8878,
+      "mean_token_accuracy": 0.7418642282485962,
+      "num_tokens": 29222644.0,
+      "step": 12630
+    },
+    {
+      "epoch": 0.11588887870175117,
+      "learning_rate": 0.00017682405794443938,
+      "loss": 0.8574,
+      "mean_token_accuracy": 0.7389842867851257,
+      "num_tokens": 29245584.0,
+      "step": 12640
+    },
+    {
+      "epoch": 0.11598056294123041,
+      "learning_rate": 0.00017680572109654352,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7447432279586792,
+      "num_tokens": 29268366.0,
+      "step": 12650
+    },
+    {
+      "epoch": 0.11607224718070963,
+      "learning_rate": 0.00017678738424864766,
+      "loss": 0.8932,
+      "mean_token_accuracy": 0.731689739227295,
+      "num_tokens": 29292046.0,
+      "step": 12660
+    },
+    {
+      "epoch": 0.11616393142018887,
+      "learning_rate": 0.00017676904740075182,
+      "loss": 0.8623,
+      "mean_token_accuracy": 0.7423143684864044,
+      "num_tokens": 29314600.0,
+      "step": 12670
+    },
+    {
+      "epoch": 0.1162556156596681,
+      "learning_rate": 0.00017675071055285596,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.733033561706543,
+      "num_tokens": 29338132.0,
+      "step": 12680
+    },
+    {
+      "epoch": 0.11634729989914734,
+      "learning_rate": 0.00017673237370496013,
+      "loss": 0.8266,
+      "mean_token_accuracy": 0.7551371455192566,
+      "num_tokens": 29362224.0,
+      "step": 12690
+    },
+    {
+      "epoch": 0.11643898413862656,
+      "learning_rate": 0.00017671403685706426,
+      "loss": 0.8902,
+      "mean_token_accuracy": 0.7397430539131165,
+      "num_tokens": 29384588.0,
+      "step": 12700
+    },
+    {
+      "epoch": 0.1165306683781058,
+      "learning_rate": 0.00017669570000916843,
+      "loss": 0.881,
+      "mean_token_accuracy": 0.7354751825332642,
+      "num_tokens": 29408007.0,
+      "step": 12710
+    },
+    {
+      "epoch": 0.11662235261758504,
+      "learning_rate": 0.0001766773631612726,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.741284316778183,
+      "num_tokens": 29430634.0,
+      "step": 12720
+    },
+    {
+      "epoch": 0.11671403685706427,
+      "learning_rate": 0.00017665902631337673,
+      "loss": 0.8712,
+      "mean_token_accuracy": 0.7477241635322571,
+      "num_tokens": 29453603.0,
+      "step": 12730
+    },
+    {
+      "epoch": 0.11680572109654351,
+      "learning_rate": 0.0001766406894654809,
+      "loss": 0.8891,
+      "mean_token_accuracy": 0.7399328410625458,
+      "num_tokens": 29477245.0,
+      "step": 12740
+    },
+    {
+      "epoch": 0.11689740533602273,
+      "learning_rate": 0.00017662235261758504,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.738675570487976,
+      "num_tokens": 29500502.0,
+      "step": 12750
+    },
+    {
+      "epoch": 0.11698908957550197,
+      "learning_rate": 0.0001766040157696892,
+      "loss": 0.8714,
+      "mean_token_accuracy": 0.745803713798523,
+      "num_tokens": 29523209.0,
+      "step": 12760
+    },
+    {
+      "epoch": 0.1170807738149812,
+      "learning_rate": 0.00017658567892179337,
+      "loss": 0.8471,
+      "mean_token_accuracy": 0.743337082862854,
+      "num_tokens": 29545777.0,
+      "step": 12770
+    },
+    {
+      "epoch": 0.11717245805446044,
+      "learning_rate": 0.0001765673420738975,
+      "loss": 0.8358,
+      "mean_token_accuracy": 0.7475039005279541,
+      "num_tokens": 29569322.0,
+      "step": 12780
+    },
+    {
+      "epoch": 0.11726414229393967,
+      "learning_rate": 0.00017654900522600167,
+      "loss": 0.9131,
+      "mean_token_accuracy": 0.7342238783836365,
+      "num_tokens": 29592371.0,
+      "step": 12790
+    },
+    {
+      "epoch": 0.1173558265334189,
+      "learning_rate": 0.0001765306683781058,
+      "loss": 0.8874,
+      "mean_token_accuracy": 0.7430296838283539,
+      "num_tokens": 29615364.0,
+      "step": 12800
+    },
+    {
+      "epoch": 0.11744751077289814,
+      "learning_rate": 0.00017651233153020998,
+      "loss": 0.8642,
+      "mean_token_accuracy": 0.7376182734966278,
+      "num_tokens": 29638438.0,
+      "step": 12810
+    },
+    {
+      "epoch": 0.11753919501237738,
+      "learning_rate": 0.00017649399468231411,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7459731817245483,
+      "num_tokens": 29661365.0,
+      "step": 12820
+    },
+    {
+      "epoch": 0.11763087925185661,
+      "learning_rate": 0.00017647565783441825,
+      "loss": 0.8524,
+      "mean_token_accuracy": 0.742122745513916,
+      "num_tokens": 29685258.0,
+      "step": 12830
+    },
+    {
+      "epoch": 0.11772256349133584,
+      "learning_rate": 0.00017645732098652244,
+      "loss": 0.8673,
+      "mean_token_accuracy": 0.7420704424381256,
+      "num_tokens": 29708804.0,
+      "step": 12840
+    },
+    {
+      "epoch": 0.11781424773081507,
+      "learning_rate": 0.00017643898413862658,
+      "loss": 0.8571,
+      "mean_token_accuracy": 0.745457673072815,
+      "num_tokens": 29732101.0,
+      "step": 12850
+    },
+    {
+      "epoch": 0.11790593197029431,
+      "learning_rate": 0.00017642064729073072,
+      "loss": 0.8712,
+      "mean_token_accuracy": 0.7470958411693573,
+      "num_tokens": 29755433.0,
+      "step": 12860
+    },
+    {
+      "epoch": 0.11799761620977355,
+      "learning_rate": 0.0001764023104428349,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7413121581077575,
+      "num_tokens": 29778250.0,
+      "step": 12870
+    },
+    {
+      "epoch": 0.11808930044925277,
+      "learning_rate": 0.00017638397359493903,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7354129433631897,
+      "num_tokens": 29800997.0,
+      "step": 12880
+    },
+    {
+      "epoch": 0.118180984688732,
+      "learning_rate": 0.0001763656367470432,
+      "loss": 0.883,
+      "mean_token_accuracy": 0.7436269700527192,
+      "num_tokens": 29824078.0,
+      "step": 12890
+    },
+    {
+      "epoch": 0.11827266892821124,
+      "learning_rate": 0.00017634729989914736,
+      "loss": 0.8747,
+      "mean_token_accuracy": 0.7435226678848267,
+      "num_tokens": 29847400.0,
+      "step": 12900
+    },
+    {
+      "epoch": 0.11836435316769048,
+      "learning_rate": 0.0001763289630512515,
+      "loss": 0.8906,
+      "mean_token_accuracy": 0.7374583840370178,
+      "num_tokens": 29870619.0,
+      "step": 12910
+    },
+    {
+      "epoch": 0.1184560374071697,
+      "learning_rate": 0.00017631062620335566,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7423339128494263,
+      "num_tokens": 29893367.0,
+      "step": 12920
+    },
+    {
+      "epoch": 0.11854772164664894,
+      "learning_rate": 0.0001762922893554598,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7501285612583161,
+      "num_tokens": 29916432.0,
+      "step": 12930
+    },
+    {
+      "epoch": 0.11863940588612817,
+      "learning_rate": 0.00017627395250756396,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7497082769870758,
+      "num_tokens": 29939826.0,
+      "step": 12940
+    },
+    {
+      "epoch": 0.11873109012560741,
+      "learning_rate": 0.0001762556156596681,
+      "loss": 0.8713,
+      "mean_token_accuracy": 0.7454786598682404,
+      "num_tokens": 29962629.0,
+      "step": 12950
+    },
+    {
+      "epoch": 0.11882277436508665,
+      "learning_rate": 0.00017623727881177227,
+      "loss": 0.904,
+      "mean_token_accuracy": 0.7302988648414612,
+      "num_tokens": 29985639.0,
+      "step": 12960
+    },
+    {
+      "epoch": 0.11891445860456587,
+      "learning_rate": 0.00017621894196387643,
+      "loss": 0.9112,
+      "mean_token_accuracy": 0.7313469231128693,
+      "num_tokens": 30009256.0,
+      "step": 12970
+    },
+    {
+      "epoch": 0.1190061428440451,
+      "learning_rate": 0.00017620060511598057,
+      "loss": 0.9059,
+      "mean_token_accuracy": 0.7313918471336365,
+      "num_tokens": 30032244.0,
+      "step": 12980
+    },
+    {
+      "epoch": 0.11909782708352434,
+      "learning_rate": 0.00017618226826808474,
+      "loss": 0.9034,
+      "mean_token_accuracy": 0.7358958899974823,
+      "num_tokens": 30055592.0,
+      "step": 12990
+    },
+    {
+      "epoch": 0.11918951132300358,
+      "learning_rate": 0.00017616393142018887,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7436038911342621,
+      "num_tokens": 30078678.0,
+      "step": 13000
+    },
+    {
+      "epoch": 0.1192811955624828,
+      "learning_rate": 0.00017614559457229304,
+      "loss": 0.8779,
+      "mean_token_accuracy": 0.7430507123470307,
+      "num_tokens": 30101997.0,
+      "step": 13010
+    },
+    {
+      "epoch": 0.11937287980196204,
+      "learning_rate": 0.00017612725772439718,
+      "loss": 0.8259,
+      "mean_token_accuracy": 0.7528484761714935,
+      "num_tokens": 30124331.0,
+      "step": 13020
+    },
+    {
+      "epoch": 0.11946456404144128,
+      "learning_rate": 0.00017610892087650134,
+      "loss": 0.8707,
+      "mean_token_accuracy": 0.7411357223987579,
+      "num_tokens": 30147264.0,
+      "step": 13030
+    },
+    {
+      "epoch": 0.11955624828092051,
+      "learning_rate": 0.0001760905840286055,
+      "loss": 0.8674,
+      "mean_token_accuracy": 0.7380927324295044,
+      "num_tokens": 30170073.0,
+      "step": 13040
+    },
+    {
+      "epoch": 0.11964793252039975,
+      "learning_rate": 0.00017607224718070965,
+      "loss": 0.8856,
+      "mean_token_accuracy": 0.7355506956577301,
+      "num_tokens": 30192745.0,
+      "step": 13050
+    },
+    {
+      "epoch": 0.11973961675987897,
+      "learning_rate": 0.00017605391033281379,
+      "loss": 0.8428,
+      "mean_token_accuracy": 0.7427493333816528,
+      "num_tokens": 30215875.0,
+      "step": 13060
+    },
+    {
+      "epoch": 0.11983130099935821,
+      "learning_rate": 0.00017603557348491795,
+      "loss": 0.8605,
+      "mean_token_accuracy": 0.7437385082244873,
+      "num_tokens": 30239109.0,
+      "step": 13070
+    },
+    {
+      "epoch": 0.11992298523883745,
+      "learning_rate": 0.0001760172366370221,
+      "loss": 0.8769,
+      "mean_token_accuracy": 0.7371023893356323,
+      "num_tokens": 30262296.0,
+      "step": 13080
+    },
+    {
+      "epoch": 0.12001466947831668,
+      "learning_rate": 0.00017599889978912626,
+      "loss": 0.8807,
+      "mean_token_accuracy": 0.738269716501236,
+      "num_tokens": 30284904.0,
+      "step": 13090
+    },
+    {
+      "epoch": 0.1201063537177959,
+      "learning_rate": 0.00017598056294123042,
+      "loss": 0.8923,
+      "mean_token_accuracy": 0.7340485513210296,
+      "num_tokens": 30307147.0,
+      "step": 13100
+    },
+    {
+      "epoch": 0.12019803795727514,
+      "learning_rate": 0.00017596222609333456,
+      "loss": 0.9074,
+      "mean_token_accuracy": 0.7378047645092011,
+      "num_tokens": 30330172.0,
+      "step": 13110
+    },
+    {
+      "epoch": 0.12028972219675438,
+      "learning_rate": 0.00017594388924543872,
+      "loss": 0.8571,
+      "mean_token_accuracy": 0.7424475669860839,
+      "num_tokens": 30353920.0,
+      "step": 13120
+    },
+    {
+      "epoch": 0.12038140643623362,
+      "learning_rate": 0.00017592555239754286,
+      "loss": 0.8817,
+      "mean_token_accuracy": 0.7396665096282959,
+      "num_tokens": 30377352.0,
+      "step": 13130
+    },
+    {
+      "epoch": 0.12047309067571285,
+      "learning_rate": 0.00017590721554964703,
+      "loss": 0.8495,
+      "mean_token_accuracy": 0.7445613801479339,
+      "num_tokens": 30400876.0,
+      "step": 13140
+    },
+    {
+      "epoch": 0.12056477491519207,
+      "learning_rate": 0.00017588887870175117,
+      "loss": 0.8619,
+      "mean_token_accuracy": 0.7434922993183136,
+      "num_tokens": 30423565.0,
+      "step": 13150
+    },
+    {
+      "epoch": 0.12065645915467131,
+      "learning_rate": 0.00017587054185385533,
+      "loss": 0.8326,
+      "mean_token_accuracy": 0.7550764858722687,
+      "num_tokens": 30446652.0,
+      "step": 13160
+    },
+    {
+      "epoch": 0.12074814339415055,
+      "learning_rate": 0.0001758522050059595,
+      "loss": 0.8812,
+      "mean_token_accuracy": 0.7392385125160217,
+      "num_tokens": 30469711.0,
+      "step": 13170
+    },
+    {
+      "epoch": 0.12083982763362978,
+      "learning_rate": 0.00017583386815806364,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.7452516674995422,
+      "num_tokens": 30493225.0,
+      "step": 13180
+    },
+    {
+      "epoch": 0.12093151187310901,
+      "learning_rate": 0.0001758155313101678,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7458862066268921,
+      "num_tokens": 30516034.0,
+      "step": 13190
+    },
+    {
+      "epoch": 0.12102319611258824,
+      "learning_rate": 0.00017579719446227194,
+      "loss": 0.8811,
+      "mean_token_accuracy": 0.7418933987617493,
+      "num_tokens": 30540341.0,
+      "step": 13200
+    },
+    {
+      "epoch": 0.12111488035206748,
+      "learning_rate": 0.0001757788576143761,
+      "loss": 0.889,
+      "mean_token_accuracy": 0.7409767270088196,
+      "num_tokens": 30564017.0,
+      "step": 13210
+    },
+    {
+      "epoch": 0.12120656459154672,
+      "learning_rate": 0.00017576052076648024,
+      "loss": 0.8902,
+      "mean_token_accuracy": 0.7385667145252228,
+      "num_tokens": 30587035.0,
+      "step": 13220
+    },
+    {
+      "epoch": 0.12129824883102594,
+      "learning_rate": 0.0001757421839185844,
+      "loss": 0.8678,
+      "mean_token_accuracy": 0.7395294427871704,
+      "num_tokens": 30610000.0,
+      "step": 13230
+    },
+    {
+      "epoch": 0.12138993307050518,
+      "learning_rate": 0.00017572384707068857,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7460544407367706,
+      "num_tokens": 30633790.0,
+      "step": 13240
+    },
+    {
+      "epoch": 0.12148161730998441,
+      "learning_rate": 0.0001757055102227927,
+      "loss": 0.8696,
+      "mean_token_accuracy": 0.7419677972793579,
+      "num_tokens": 30655745.0,
+      "step": 13250
+    },
+    {
+      "epoch": 0.12157330154946365,
+      "learning_rate": 0.00017568717337489685,
+      "loss": 0.9234,
+      "mean_token_accuracy": 0.7229511618614197,
+      "num_tokens": 30678999.0,
+      "step": 13260
+    },
+    {
+      "epoch": 0.12166498578894289,
+      "learning_rate": 0.00017566883652700102,
+      "loss": 0.9156,
+      "mean_token_accuracy": 0.7283031642436981,
+      "num_tokens": 30701931.0,
+      "step": 13270
+    },
+    {
+      "epoch": 0.12175667002842211,
+      "learning_rate": 0.00017565049967910515,
+      "loss": 0.8696,
+      "mean_token_accuracy": 0.7463475167751312,
+      "num_tokens": 30725960.0,
+      "step": 13280
+    },
+    {
+      "epoch": 0.12184835426790135,
+      "learning_rate": 0.00017563216283120932,
+      "loss": 0.903,
+      "mean_token_accuracy": 0.7365835070610046,
+      "num_tokens": 30749486.0,
+      "step": 13290
+    },
+    {
+      "epoch": 0.12194003850738058,
+      "learning_rate": 0.00017561382598331348,
+      "loss": 0.8559,
+      "mean_token_accuracy": 0.7459181606769562,
+      "num_tokens": 30772928.0,
+      "step": 13300
+    },
+    {
+      "epoch": 0.12203172274685982,
+      "learning_rate": 0.00017559548913541762,
+      "loss": 0.8935,
+      "mean_token_accuracy": 0.7427541315555573,
+      "num_tokens": 30795424.0,
+      "step": 13310
+    },
+    {
+      "epoch": 0.12212340698633904,
+      "learning_rate": 0.0001755771522875218,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7493497610092164,
+      "num_tokens": 30817827.0,
+      "step": 13320
+    },
+    {
+      "epoch": 0.12221509122581828,
+      "learning_rate": 0.00017555881543962593,
+      "loss": 0.9287,
+      "mean_token_accuracy": 0.7289869010448455,
+      "num_tokens": 30840847.0,
+      "step": 13330
+    },
+    {
+      "epoch": 0.12230677546529752,
+      "learning_rate": 0.0001755404785917301,
+      "loss": 0.8713,
+      "mean_token_accuracy": 0.7393110394477844,
+      "num_tokens": 30863622.0,
+      "step": 13340
+    },
+    {
+      "epoch": 0.12239845970477675,
+      "learning_rate": 0.00017552214174383423,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7428774952888488,
+      "num_tokens": 30886248.0,
+      "step": 13350
+    },
+    {
+      "epoch": 0.12249014394425599,
+      "learning_rate": 0.0001755038048959384,
+      "loss": 0.9237,
+      "mean_token_accuracy": 0.7274897336959839,
+      "num_tokens": 30910208.0,
+      "step": 13360
+    },
+    {
+      "epoch": 0.12258182818373521,
+      "learning_rate": 0.00017548546804804256,
+      "loss": 0.8977,
+      "mean_token_accuracy": 0.7355661809444427,
+      "num_tokens": 30932555.0,
+      "step": 13370
+    },
+    {
+      "epoch": 0.12267351242321445,
+      "learning_rate": 0.0001754671312001467,
+      "loss": 0.8762,
+      "mean_token_accuracy": 0.741988730430603,
+      "num_tokens": 30955296.0,
+      "step": 13380
+    },
+    {
+      "epoch": 0.12276519666269368,
+      "learning_rate": 0.00017544879435225087,
+      "loss": 0.844,
+      "mean_token_accuracy": 0.7488784313201904,
+      "num_tokens": 30977767.0,
+      "step": 13390
+    },
+    {
+      "epoch": 0.12285688090217292,
+      "learning_rate": 0.000175430457504355,
+      "loss": 0.8796,
+      "mean_token_accuracy": 0.7436460018157959,
+      "num_tokens": 31001079.0,
+      "step": 13400
+    },
+    {
+      "epoch": 0.12294856514165214,
+      "learning_rate": 0.00017541212065645917,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.7348134934902191,
+      "num_tokens": 31023951.0,
+      "step": 13410
+    },
+    {
+      "epoch": 0.12304024938113138,
+      "learning_rate": 0.0001753937838085633,
+      "loss": 0.8382,
+      "mean_token_accuracy": 0.7484686076641083,
+      "num_tokens": 31046813.0,
+      "step": 13420
+    },
+    {
+      "epoch": 0.12313193362061062,
+      "learning_rate": 0.00017537544696066747,
+      "loss": 0.8896,
+      "mean_token_accuracy": 0.7325926005840302,
+      "num_tokens": 31069636.0,
+      "step": 13430
+    },
+    {
+      "epoch": 0.12322361786008985,
+      "learning_rate": 0.00017535711011277164,
+      "loss": 0.8812,
+      "mean_token_accuracy": 0.7350605428218842,
+      "num_tokens": 31092806.0,
+      "step": 13440
+    },
+    {
+      "epoch": 0.12331530209956909,
+      "learning_rate": 0.00017533877326487578,
+      "loss": 0.8582,
+      "mean_token_accuracy": 0.7459491133689881,
+      "num_tokens": 31115851.0,
+      "step": 13450
+    },
+    {
+      "epoch": 0.12340698633904831,
+      "learning_rate": 0.00017532043641697991,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7331504344940185,
+      "num_tokens": 31138514.0,
+      "step": 13460
+    },
+    {
+      "epoch": 0.12349867057852755,
+      "learning_rate": 0.00017530209956908408,
+      "loss": 0.8733,
+      "mean_token_accuracy": 0.7477503836154937,
+      "num_tokens": 31161440.0,
+      "step": 13470
+    },
+    {
+      "epoch": 0.12359035481800679,
+      "learning_rate": 0.00017528376272118822,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.732633912563324,
+      "num_tokens": 31184423.0,
+      "step": 13480
+    },
+    {
+      "epoch": 0.12368203905748602,
+      "learning_rate": 0.0001752654258732924,
+      "loss": 0.8959,
+      "mean_token_accuracy": 0.7294002056121827,
+      "num_tokens": 31207404.0,
+      "step": 13490
+    },
+    {
+      "epoch": 0.12377372329696525,
+      "learning_rate": 0.00017524708902539655,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.7389393508434295,
+      "num_tokens": 31230643.0,
+      "step": 13500
+    },
+    {
+      "epoch": 0.12386540753644448,
+      "learning_rate": 0.0001752287521775007,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.7482302010059356,
+      "num_tokens": 31254253.0,
+      "step": 13510
+    },
+    {
+      "epoch": 0.12395709177592372,
+      "learning_rate": 0.00017521041532960485,
+      "loss": 0.912,
+      "mean_token_accuracy": 0.7345802307128906,
+      "num_tokens": 31276987.0,
+      "step": 13520
+    },
+    {
+      "epoch": 0.12404877601540296,
+      "learning_rate": 0.000175192078481709,
+      "loss": 0.8782,
+      "mean_token_accuracy": 0.7386352062225342,
+      "num_tokens": 31300638.0,
+      "step": 13530
+    },
+    {
+      "epoch": 0.12414046025488218,
+      "learning_rate": 0.00017517374163381316,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7417355597019195,
+      "num_tokens": 31324326.0,
+      "step": 13540
+    },
+    {
+      "epoch": 0.12423214449436142,
+      "learning_rate": 0.0001751554047859173,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7393722474575043,
+      "num_tokens": 31348317.0,
+      "step": 13550
+    },
+    {
+      "epoch": 0.12432382873384065,
+      "learning_rate": 0.00017513706793802146,
+      "loss": 0.9122,
+      "mean_token_accuracy": 0.7346632480621338,
+      "num_tokens": 31371033.0,
+      "step": 13560
+    },
+    {
+      "epoch": 0.12441551297331989,
+      "learning_rate": 0.00017511873109012563,
+      "loss": 0.8524,
+      "mean_token_accuracy": 0.7437283635139466,
+      "num_tokens": 31394548.0,
+      "step": 13570
+    },
+    {
+      "epoch": 0.12450719721279913,
+      "learning_rate": 0.00017510039424222976,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7488535881042481,
+      "num_tokens": 31417408.0,
+      "step": 13580
+    },
+    {
+      "epoch": 0.12459888145227835,
+      "learning_rate": 0.00017508205739433393,
+      "loss": 0.8886,
+      "mean_token_accuracy": 0.7315831184387207,
+      "num_tokens": 31439844.0,
+      "step": 13590
+    },
+    {
+      "epoch": 0.12469056569175758,
+      "learning_rate": 0.00017506372054643807,
+      "loss": 0.8789,
+      "mean_token_accuracy": 0.7398853242397309,
+      "num_tokens": 31464053.0,
+      "step": 13600
+    },
+    {
+      "epoch": 0.12478224993123682,
+      "learning_rate": 0.00017504538369854223,
+      "loss": 0.9049,
+      "mean_token_accuracy": 0.7342214941978454,
+      "num_tokens": 31487565.0,
+      "step": 13610
+    },
+    {
+      "epoch": 0.12487393417071606,
+      "learning_rate": 0.0001750270468506464,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7285666584968566,
+      "num_tokens": 31511006.0,
+      "step": 13620
+    },
+    {
+      "epoch": 0.12496561841019528,
+      "learning_rate": 0.00017500871000275054,
+      "loss": 0.882,
+      "mean_token_accuracy": 0.7379772901535034,
+      "num_tokens": 31534480.0,
+      "step": 13630
+    },
+    {
+      "epoch": 0.12505730264967452,
+      "learning_rate": 0.0001749903731548547,
+      "loss": 0.8741,
+      "mean_token_accuracy": 0.7383758902549744,
+      "num_tokens": 31557449.0,
+      "step": 13640
+    },
+    {
+      "epoch": 0.12514898688915377,
+      "learning_rate": 0.00017497203630695884,
+      "loss": 0.8883,
+      "mean_token_accuracy": 0.7329118847846985,
+      "num_tokens": 31581519.0,
+      "step": 13650
+    },
+    {
+      "epoch": 0.125240671128633,
+      "learning_rate": 0.00017495369945906298,
+      "loss": 0.8653,
+      "mean_token_accuracy": 0.7431224584579468,
+      "num_tokens": 31603877.0,
+      "step": 13660
+    },
+    {
+      "epoch": 0.1253323553681122,
+      "learning_rate": 0.00017493536261116714,
+      "loss": 0.8817,
+      "mean_token_accuracy": 0.7401574850082397,
+      "num_tokens": 31627032.0,
+      "step": 13670
+    },
+    {
+      "epoch": 0.12542403960759146,
+      "learning_rate": 0.00017491702576327128,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7415537536144257,
+      "num_tokens": 31650209.0,
+      "step": 13680
+    },
+    {
+      "epoch": 0.1255157238470707,
+      "learning_rate": 0.00017489868891537548,
+      "loss": 0.8795,
+      "mean_token_accuracy": 0.7378593504428863,
+      "num_tokens": 31673550.0,
+      "step": 13690
+    },
+    {
+      "epoch": 0.1256074080865499,
+      "learning_rate": 0.00017488035206747961,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7466789901256561,
+      "num_tokens": 31696646.0,
+      "step": 13700
+    },
+    {
+      "epoch": 0.12569909232602916,
+      "learning_rate": 0.00017486201521958375,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.7511944532394409,
+      "num_tokens": 31719187.0,
+      "step": 13710
+    },
+    {
+      "epoch": 0.12579077656550838,
+      "learning_rate": 0.00017484367837168792,
+      "loss": 0.881,
+      "mean_token_accuracy": 0.7432884573936462,
+      "num_tokens": 31742580.0,
+      "step": 13720
+    },
+    {
+      "epoch": 0.12588246080498763,
+      "learning_rate": 0.00017482534152379206,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7441620707511902,
+      "num_tokens": 31765975.0,
+      "step": 13730
+    },
+    {
+      "epoch": 0.12597414504446686,
+      "learning_rate": 0.00017480700467589622,
+      "loss": 0.9079,
+      "mean_token_accuracy": 0.7359534800052643,
+      "num_tokens": 31789434.0,
+      "step": 13740
+    },
+    {
+      "epoch": 0.12606582928394608,
+      "learning_rate": 0.0001747886678280004,
+      "loss": 0.8799,
+      "mean_token_accuracy": 0.7369920551776886,
+      "num_tokens": 31812985.0,
+      "step": 13750
+    },
+    {
+      "epoch": 0.12615751352342533,
+      "learning_rate": 0.00017477033098010453,
+      "loss": 0.8825,
+      "mean_token_accuracy": 0.7377887725830078,
+      "num_tokens": 31835774.0,
+      "step": 13760
+    },
+    {
+      "epoch": 0.12624919776290455,
+      "learning_rate": 0.0001747519941322087,
+      "loss": 0.8391,
+      "mean_token_accuracy": 0.7521439611911773,
+      "num_tokens": 31859100.0,
+      "step": 13770
+    },
+    {
+      "epoch": 0.1263408820023838,
+      "learning_rate": 0.00017473365728431283,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7330224096775055,
+      "num_tokens": 31882262.0,
+      "step": 13780
+    },
+    {
+      "epoch": 0.12643256624186303,
+      "learning_rate": 0.000174715320436417,
+      "loss": 0.9049,
+      "mean_token_accuracy": 0.734117305278778,
+      "num_tokens": 31905080.0,
+      "step": 13790
+    },
+    {
+      "epoch": 0.12652425048134225,
+      "learning_rate": 0.00017469698358852113,
+      "loss": 0.8465,
+      "mean_token_accuracy": 0.744682377576828,
+      "num_tokens": 31927947.0,
+      "step": 13800
+    },
+    {
+      "epoch": 0.1266159347208215,
+      "learning_rate": 0.0001746786467406253,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7385300934314728,
+      "num_tokens": 31950836.0,
+      "step": 13810
+    },
+    {
+      "epoch": 0.12670761896030072,
+      "learning_rate": 0.00017466030989272946,
+      "loss": 0.8692,
+      "mean_token_accuracy": 0.7386554837226867,
+      "num_tokens": 31973742.0,
+      "step": 13820
+    },
+    {
+      "epoch": 0.12679930319977994,
+      "learning_rate": 0.0001746419730448336,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.7460804641246795,
+      "num_tokens": 31996494.0,
+      "step": 13830
+    },
+    {
+      "epoch": 0.1268909874392592,
+      "learning_rate": 0.00017462363619693777,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7330855011940003,
+      "num_tokens": 32019415.0,
+      "step": 13840
+    },
+    {
+      "epoch": 0.12698267167873842,
+      "learning_rate": 0.0001746052993490419,
+      "loss": 0.8693,
+      "mean_token_accuracy": 0.7430007636547089,
+      "num_tokens": 32043083.0,
+      "step": 13850
+    },
+    {
+      "epoch": 0.12707435591821767,
+      "learning_rate": 0.00017458696250114604,
+      "loss": 0.8624,
+      "mean_token_accuracy": 0.7419776797294617,
+      "num_tokens": 32065552.0,
+      "step": 13860
+    },
+    {
+      "epoch": 0.1271660401576969,
+      "learning_rate": 0.0001745686256532502,
+      "loss": 0.937,
+      "mean_token_accuracy": 0.7287083685398101,
+      "num_tokens": 32088367.0,
+      "step": 13870
+    },
+    {
+      "epoch": 0.12725772439717611,
+      "learning_rate": 0.00017455028880535437,
+      "loss": 0.925,
+      "mean_token_accuracy": 0.7303915798664093,
+      "num_tokens": 32110728.0,
+      "step": 13880
+    },
+    {
+      "epoch": 0.12734940863665536,
+      "learning_rate": 0.00017453195195745854,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7415268540382385,
+      "num_tokens": 32134295.0,
+      "step": 13890
+    },
+    {
+      "epoch": 0.1274410928761346,
+      "learning_rate": 0.00017451361510956268,
+      "loss": 0.8718,
+      "mean_token_accuracy": 0.7446969985961914,
+      "num_tokens": 32156688.0,
+      "step": 13900
+    },
+    {
+      "epoch": 0.12753277711561384,
+      "learning_rate": 0.00017449527826166682,
+      "loss": 0.8958,
+      "mean_token_accuracy": 0.7414741516113281,
+      "num_tokens": 32179971.0,
+      "step": 13910
+    },
+    {
+      "epoch": 0.12762446135509306,
+      "learning_rate": 0.00017447694141377098,
+      "loss": 0.8449,
+      "mean_token_accuracy": 0.7451531112194061,
+      "num_tokens": 32202781.0,
+      "step": 13920
+    },
+    {
+      "epoch": 0.12771614559457228,
+      "learning_rate": 0.00017445860456587512,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.7495111107826233,
+      "num_tokens": 32225524.0,
+      "step": 13930
+    },
+    {
+      "epoch": 0.12780782983405153,
+      "learning_rate": 0.00017444026771797929,
+      "loss": 0.8796,
+      "mean_token_accuracy": 0.7446104109287262,
+      "num_tokens": 32248427.0,
+      "step": 13940
+    },
+    {
+      "epoch": 0.12789951407353076,
+      "learning_rate": 0.00017442193087008345,
+      "loss": 0.9104,
+      "mean_token_accuracy": 0.7278967261314392,
+      "num_tokens": 32271092.0,
+      "step": 13950
+    },
+    {
+      "epoch": 0.12799119831301,
+      "learning_rate": 0.0001744035940221876,
+      "loss": 0.9051,
+      "mean_token_accuracy": 0.7343689262866974,
+      "num_tokens": 32294127.0,
+      "step": 13960
+    },
+    {
+      "epoch": 0.12808288255248923,
+      "learning_rate": 0.00017438525717429175,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7487718880176544,
+      "num_tokens": 32316594.0,
+      "step": 13970
+    },
+    {
+      "epoch": 0.12817456679196845,
+      "learning_rate": 0.0001743669203263959,
+      "loss": 0.9167,
+      "mean_token_accuracy": 0.7296574294567109,
+      "num_tokens": 32340409.0,
+      "step": 13980
+    },
+    {
+      "epoch": 0.1282662510314477,
+      "learning_rate": 0.00017434858347850006,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.743960189819336,
+      "num_tokens": 32363980.0,
+      "step": 13990
+    },
+    {
+      "epoch": 0.12835793527092693,
+      "learning_rate": 0.0001743302466306042,
+      "loss": 0.9075,
+      "mean_token_accuracy": 0.7316538751125335,
+      "num_tokens": 32386796.0,
+      "step": 14000
+    },
+    {
+      "epoch": 0.12844961951040615,
+      "learning_rate": 0.00017431190978270836,
+      "loss": 0.852,
+      "mean_token_accuracy": 0.7484723925590515,
+      "num_tokens": 32410086.0,
+      "step": 14010
+    },
+    {
+      "epoch": 0.1285413037498854,
+      "learning_rate": 0.00017429357293481253,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7521367490291595,
+      "num_tokens": 32432378.0,
+      "step": 14020
+    },
+    {
+      "epoch": 0.12863298798936462,
+      "learning_rate": 0.00017427523608691667,
+      "loss": 0.9017,
+      "mean_token_accuracy": 0.7342784345149994,
+      "num_tokens": 32454976.0,
+      "step": 14030
+    },
+    {
+      "epoch": 0.12872467222884387,
+      "learning_rate": 0.00017425689923902083,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7416263163089752,
+      "num_tokens": 32477821.0,
+      "step": 14040
+    },
+    {
+      "epoch": 0.1288163564683231,
+      "learning_rate": 0.00017423856239112497,
+      "loss": 0.893,
+      "mean_token_accuracy": 0.7359785735607147,
+      "num_tokens": 32501400.0,
+      "step": 14050
+    },
+    {
+      "epoch": 0.12890804070780232,
+      "learning_rate": 0.0001742202255432291,
+      "loss": 0.8906,
+      "mean_token_accuracy": 0.7366522014141083,
+      "num_tokens": 32524581.0,
+      "step": 14060
+    },
+    {
+      "epoch": 0.12899972494728157,
+      "learning_rate": 0.00017420188869533327,
+      "loss": 0.8726,
+      "mean_token_accuracy": 0.7413610994815827,
+      "num_tokens": 32546892.0,
+      "step": 14070
+    },
+    {
+      "epoch": 0.1290914091867608,
+      "learning_rate": 0.00017418355184743744,
+      "loss": 0.8903,
+      "mean_token_accuracy": 0.7370317041873932,
+      "num_tokens": 32570582.0,
+      "step": 14080
+    },
+    {
+      "epoch": 0.12918309342624004,
+      "learning_rate": 0.0001741652149995416,
+      "loss": 0.8827,
+      "mean_token_accuracy": 0.7373828768730164,
+      "num_tokens": 32594207.0,
+      "step": 14090
+    },
+    {
+      "epoch": 0.12927477766571926,
+      "learning_rate": 0.00017414687815164574,
+      "loss": 0.8813,
+      "mean_token_accuracy": 0.7411800384521484,
+      "num_tokens": 32617335.0,
+      "step": 14100
+    },
+    {
+      "epoch": 0.1293664619051985,
+      "learning_rate": 0.00017412854130374988,
+      "loss": 0.8307,
+      "mean_token_accuracy": 0.7452889561653138,
+      "num_tokens": 32640265.0,
+      "step": 14110
+    },
+    {
+      "epoch": 0.12945814614467774,
+      "learning_rate": 0.00017411020445585405,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7437470495700836,
+      "num_tokens": 32664621.0,
+      "step": 14120
+    },
+    {
+      "epoch": 0.12954983038415696,
+      "learning_rate": 0.00017409186760795818,
+      "loss": 0.8763,
+      "mean_token_accuracy": 0.7378992319107056,
+      "num_tokens": 32688438.0,
+      "step": 14130
+    },
+    {
+      "epoch": 0.12964151462363618,
+      "learning_rate": 0.00017407353076006238,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7478434562683105,
+      "num_tokens": 32711422.0,
+      "step": 14140
+    },
+    {
+      "epoch": 0.12973319886311543,
+      "learning_rate": 0.00017405519391216652,
+      "loss": 0.8866,
+      "mean_token_accuracy": 0.7385292470455169,
+      "num_tokens": 32734443.0,
+      "step": 14150
+    },
+    {
+      "epoch": 0.12982488310259466,
+      "learning_rate": 0.00017403685706427065,
+      "loss": 0.8914,
+      "mean_token_accuracy": 0.7368983030319214,
+      "num_tokens": 32757880.0,
+      "step": 14160
+    },
+    {
+      "epoch": 0.1299165673420739,
+      "learning_rate": 0.00017401852021637482,
+      "loss": 0.8566,
+      "mean_token_accuracy": 0.7447448313236237,
+      "num_tokens": 32781375.0,
+      "step": 14170
+    },
+    {
+      "epoch": 0.13000825158155313,
+      "learning_rate": 0.00017400018336847896,
+      "loss": 0.8912,
+      "mean_token_accuracy": 0.7378781378269196,
+      "num_tokens": 32804014.0,
+      "step": 14180
+    },
+    {
+      "epoch": 0.13009993582103235,
+      "learning_rate": 0.00017398184652058312,
+      "loss": 0.8675,
+      "mean_token_accuracy": 0.744933158159256,
+      "num_tokens": 32827445.0,
+      "step": 14190
+    },
+    {
+      "epoch": 0.1301916200605116,
+      "learning_rate": 0.00017396350967268726,
+      "loss": 0.883,
+      "mean_token_accuracy": 0.7406885266304016,
+      "num_tokens": 32850833.0,
+      "step": 14200
+    },
+    {
+      "epoch": 0.13028330429999083,
+      "learning_rate": 0.00017394517282479143,
+      "loss": 0.8786,
+      "mean_token_accuracy": 0.7334289312362671,
+      "num_tokens": 32874270.0,
+      "step": 14210
+    },
+    {
+      "epoch": 0.13037498853947008,
+      "learning_rate": 0.0001739268359768956,
+      "loss": 0.87,
+      "mean_token_accuracy": 0.7492436230182647,
+      "num_tokens": 32897659.0,
+      "step": 14220
+    },
+    {
+      "epoch": 0.1304666727789493,
+      "learning_rate": 0.00017390849912899973,
+      "loss": 0.8896,
+      "mean_token_accuracy": 0.7279127836227417,
+      "num_tokens": 32920966.0,
+      "step": 14230
+    },
+    {
+      "epoch": 0.13055835701842852,
+      "learning_rate": 0.0001738901622811039,
+      "loss": 0.8707,
+      "mean_token_accuracy": 0.7412879467010498,
+      "num_tokens": 32942964.0,
+      "step": 14240
+    },
+    {
+      "epoch": 0.13065004125790777,
+      "learning_rate": 0.00017387182543320803,
+      "loss": 0.8905,
+      "mean_token_accuracy": 0.7424752235412597,
+      "num_tokens": 32966348.0,
+      "step": 14250
+    },
+    {
+      "epoch": 0.130741725497387,
+      "learning_rate": 0.00017385348858531217,
+      "loss": 0.8767,
+      "mean_token_accuracy": 0.7338845789432525,
+      "num_tokens": 32989647.0,
+      "step": 14260
+    },
+    {
+      "epoch": 0.13083340973686625,
+      "learning_rate": 0.00017383515173741637,
+      "loss": 0.9067,
+      "mean_token_accuracy": 0.7268571972846984,
+      "num_tokens": 33012854.0,
+      "step": 14270
+    },
+    {
+      "epoch": 0.13092509397634547,
+      "learning_rate": 0.0001738168148895205,
+      "loss": 0.8825,
+      "mean_token_accuracy": 0.7357637226581574,
+      "num_tokens": 33035560.0,
+      "step": 14280
+    },
+    {
+      "epoch": 0.1310167782158247,
+      "learning_rate": 0.00017379847804162467,
+      "loss": 0.8935,
+      "mean_token_accuracy": 0.739679080247879,
+      "num_tokens": 33058348.0,
+      "step": 14290
+    },
+    {
+      "epoch": 0.13110846245530394,
+      "learning_rate": 0.0001737801411937288,
+      "loss": 0.8915,
+      "mean_token_accuracy": 0.7366253197193146,
+      "num_tokens": 33081400.0,
+      "step": 14300
+    },
+    {
+      "epoch": 0.13120014669478317,
+      "learning_rate": 0.00017376180434583295,
+      "loss": 0.8934,
+      "mean_token_accuracy": 0.7382798731327057,
+      "num_tokens": 33104888.0,
+      "step": 14310
+    },
+    {
+      "epoch": 0.1312918309342624,
+      "learning_rate": 0.0001737434674979371,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7401932656764985,
+      "num_tokens": 33128450.0,
+      "step": 14320
+    },
+    {
+      "epoch": 0.13138351517374164,
+      "learning_rate": 0.00017372513065004125,
+      "loss": 0.8823,
+      "mean_token_accuracy": 0.735636180639267,
+      "num_tokens": 33151650.0,
+      "step": 14330
+    },
+    {
+      "epoch": 0.13147519941322086,
+      "learning_rate": 0.00017370679380214544,
+      "loss": 0.8608,
+      "mean_token_accuracy": 0.7445764899253845,
+      "num_tokens": 33174893.0,
+      "step": 14340
+    },
+    {
+      "epoch": 0.1315668836527001,
+      "learning_rate": 0.00017368845695424958,
+      "loss": 0.9213,
+      "mean_token_accuracy": 0.7347506880760193,
+      "num_tokens": 33197781.0,
+      "step": 14350
+    },
+    {
+      "epoch": 0.13165856789217933,
+      "learning_rate": 0.00017367012010635372,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7362947225570678,
+      "num_tokens": 33221135.0,
+      "step": 14360
+    },
+    {
+      "epoch": 0.13175025213165856,
+      "learning_rate": 0.00017365178325845788,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.7380826473236084,
+      "num_tokens": 33244067.0,
+      "step": 14370
+    },
+    {
+      "epoch": 0.1318419363711378,
+      "learning_rate": 0.00017363344641056202,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7559607982635498,
+      "num_tokens": 33266712.0,
+      "step": 14380
+    },
+    {
+      "epoch": 0.13193362061061703,
+      "learning_rate": 0.0001736151095626662,
+      "loss": 0.9012,
+      "mean_token_accuracy": 0.7327978372573852,
+      "num_tokens": 33289080.0,
+      "step": 14390
+    },
+    {
+      "epoch": 0.13202530485009628,
+      "learning_rate": 0.00017359677271477033,
+      "loss": 0.8526,
+      "mean_token_accuracy": 0.7474592685699463,
+      "num_tokens": 33312643.0,
+      "step": 14400
+    },
+    {
+      "epoch": 0.1321169890895755,
+      "learning_rate": 0.0001735784358668745,
+      "loss": 0.8799,
+      "mean_token_accuracy": 0.7366613864898681,
+      "num_tokens": 33335828.0,
+      "step": 14410
+    },
+    {
+      "epoch": 0.13220867332905473,
+      "learning_rate": 0.00017356009901897866,
+      "loss": 0.8346,
+      "mean_token_accuracy": 0.7410920321941376,
+      "num_tokens": 33358712.0,
+      "step": 14420
+    },
+    {
+      "epoch": 0.13230035756853398,
+      "learning_rate": 0.0001735417621710828,
+      "loss": 0.9087,
+      "mean_token_accuracy": 0.7315150141716004,
+      "num_tokens": 33381941.0,
+      "step": 14430
+    },
+    {
+      "epoch": 0.1323920418080132,
+      "learning_rate": 0.00017352342532318696,
+      "loss": 0.8513,
+      "mean_token_accuracy": 0.746486222743988,
+      "num_tokens": 33404783.0,
+      "step": 14440
+    },
+    {
+      "epoch": 0.13248372604749242,
+      "learning_rate": 0.0001735050884752911,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7400833129882812,
+      "num_tokens": 33428194.0,
+      "step": 14450
+    },
+    {
+      "epoch": 0.13257541028697167,
+      "learning_rate": 0.00017348675162739524,
+      "loss": 0.8727,
+      "mean_token_accuracy": 0.7396412909030914,
+      "num_tokens": 33451989.0,
+      "step": 14460
+    },
+    {
+      "epoch": 0.1326670945264509,
+      "learning_rate": 0.00017346841477949943,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7379320859909058,
+      "num_tokens": 33475313.0,
+      "step": 14470
+    },
+    {
+      "epoch": 0.13275877876593015,
+      "learning_rate": 0.00017345007793160357,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.748997038602829,
+      "num_tokens": 33498554.0,
+      "step": 14480
+    },
+    {
+      "epoch": 0.13285046300540937,
+      "learning_rate": 0.00017343174108370773,
+      "loss": 0.8784,
+      "mean_token_accuracy": 0.7325134396553039,
+      "num_tokens": 33522514.0,
+      "step": 14490
+    },
+    {
+      "epoch": 0.1329421472448886,
+      "learning_rate": 0.00017341340423581187,
+      "loss": 0.8732,
+      "mean_token_accuracy": 0.7366645336151123,
+      "num_tokens": 33546435.0,
+      "step": 14500
+    },
+    {
+      "epoch": 0.13303383148436784,
+      "learning_rate": 0.000173395067387916,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7399717330932617,
+      "num_tokens": 33568916.0,
+      "step": 14510
+    },
+    {
+      "epoch": 0.13312551572384707,
+      "learning_rate": 0.00017337673054002018,
+      "loss": 0.8993,
+      "mean_token_accuracy": 0.7399784088134765,
+      "num_tokens": 33591657.0,
+      "step": 14520
+    },
+    {
+      "epoch": 0.13321719996332632,
+      "learning_rate": 0.00017335839369212431,
+      "loss": 0.9011,
+      "mean_token_accuracy": 0.7350584924221039,
+      "num_tokens": 33614933.0,
+      "step": 14530
+    },
+    {
+      "epoch": 0.13330888420280554,
+      "learning_rate": 0.0001733400568442285,
+      "loss": 0.8839,
+      "mean_token_accuracy": 0.7400415539741516,
+      "num_tokens": 33638132.0,
+      "step": 14540
+    },
+    {
+      "epoch": 0.13340056844228476,
+      "learning_rate": 0.00017332171999633264,
+      "loss": 0.8462,
+      "mean_token_accuracy": 0.744160383939743,
+      "num_tokens": 33661575.0,
+      "step": 14550
+    },
+    {
+      "epoch": 0.133492252681764,
+      "learning_rate": 0.00017330338314843678,
+      "loss": 0.8713,
+      "mean_token_accuracy": 0.7381963133811951,
+      "num_tokens": 33684064.0,
+      "step": 14560
+    },
+    {
+      "epoch": 0.13358393692124323,
+      "learning_rate": 0.00017328504630054095,
+      "loss": 0.8592,
+      "mean_token_accuracy": 0.744098824262619,
+      "num_tokens": 33706846.0,
+      "step": 14570
+    },
+    {
+      "epoch": 0.13367562116072249,
+      "learning_rate": 0.0001732667094526451,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7483438372611999,
+      "num_tokens": 33730542.0,
+      "step": 14580
+    },
+    {
+      "epoch": 0.1337673054002017,
+      "learning_rate": 0.00017324837260474925,
+      "loss": 0.8548,
+      "mean_token_accuracy": 0.7378258168697357,
+      "num_tokens": 33753327.0,
+      "step": 14590
+    },
+    {
+      "epoch": 0.13385898963968093,
+      "learning_rate": 0.00017323003575685342,
+      "loss": 0.8784,
+      "mean_token_accuracy": 0.737323272228241,
+      "num_tokens": 33775698.0,
+      "step": 14600
+    },
+    {
+      "epoch": 0.13395067387916018,
+      "learning_rate": 0.00017321169890895756,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7428741157054901,
+      "num_tokens": 33798575.0,
+      "step": 14610
+    },
+    {
+      "epoch": 0.1340423581186394,
+      "learning_rate": 0.00017319336206106172,
+      "loss": 0.8901,
+      "mean_token_accuracy": 0.7395589411258697,
+      "num_tokens": 33821900.0,
+      "step": 14620
+    },
+    {
+      "epoch": 0.13413404235811863,
+      "learning_rate": 0.00017317502521316586,
+      "loss": 0.9027,
+      "mean_token_accuracy": 0.7431897580623626,
+      "num_tokens": 33844451.0,
+      "step": 14630
+    },
+    {
+      "epoch": 0.13422572659759788,
+      "learning_rate": 0.00017315668836527002,
+      "loss": 0.8767,
+      "mean_token_accuracy": 0.7426425874233246,
+      "num_tokens": 33867120.0,
+      "step": 14640
+    },
+    {
+      "epoch": 0.1343174108370771,
+      "learning_rate": 0.00017313835151737416,
+      "loss": 0.8865,
+      "mean_token_accuracy": 0.7388552486896515,
+      "num_tokens": 33891226.0,
+      "step": 14650
+    },
+    {
+      "epoch": 0.13440909507655635,
+      "learning_rate": 0.0001731200146694783,
+      "loss": 0.8871,
+      "mean_token_accuracy": 0.7396299719810486,
+      "num_tokens": 33914381.0,
+      "step": 14660
+    },
+    {
+      "epoch": 0.13450077931603557,
+      "learning_rate": 0.0001731016778215825,
+      "loss": 0.9021,
+      "mean_token_accuracy": 0.7366632640361785,
+      "num_tokens": 33937847.0,
+      "step": 14670
+    },
+    {
+      "epoch": 0.1345924635555148,
+      "learning_rate": 0.00017308334097368663,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7463048279285431,
+      "num_tokens": 33961142.0,
+      "step": 14680
+    },
+    {
+      "epoch": 0.13468414779499405,
+      "learning_rate": 0.0001730650041257908,
+      "loss": 0.9041,
+      "mean_token_accuracy": 0.7357403934001923,
+      "num_tokens": 33983978.0,
+      "step": 14690
+    },
+    {
+      "epoch": 0.13477583203447327,
+      "learning_rate": 0.00017304666727789494,
+      "loss": 0.8969,
+      "mean_token_accuracy": 0.7384519755840302,
+      "num_tokens": 34006363.0,
+      "step": 14700
+    },
+    {
+      "epoch": 0.13486751627395252,
+      "learning_rate": 0.00017302833042999907,
+      "loss": 0.8922,
+      "mean_token_accuracy": 0.7325655400753022,
+      "num_tokens": 34029506.0,
+      "step": 14710
+    },
+    {
+      "epoch": 0.13495920051343174,
+      "learning_rate": 0.00017300999358210324,
+      "loss": 0.8747,
+      "mean_token_accuracy": 0.7357029318809509,
+      "num_tokens": 34052710.0,
+      "step": 14720
+    },
+    {
+      "epoch": 0.13505088475291097,
+      "learning_rate": 0.0001729916567342074,
+      "loss": 0.9007,
+      "mean_token_accuracy": 0.7338222444057465,
+      "num_tokens": 34076247.0,
+      "step": 14730
+    },
+    {
+      "epoch": 0.13514256899239022,
+      "learning_rate": 0.00017297331988631157,
+      "loss": 0.8823,
+      "mean_token_accuracy": 0.7370768666267395,
+      "num_tokens": 34098953.0,
+      "step": 14740
+    },
+    {
+      "epoch": 0.13523425323186944,
+      "learning_rate": 0.0001729549830384157,
+      "loss": 0.9181,
+      "mean_token_accuracy": 0.7265935063362121,
+      "num_tokens": 34122802.0,
+      "step": 14750
+    },
+    {
+      "epoch": 0.13532593747134866,
+      "learning_rate": 0.00017293664619051985,
+      "loss": 0.8771,
+      "mean_token_accuracy": 0.7369015514850616,
+      "num_tokens": 34145898.0,
+      "step": 14760
+    },
+    {
+      "epoch": 0.1354176217108279,
+      "learning_rate": 0.000172918309342624,
+      "loss": 0.8535,
+      "mean_token_accuracy": 0.7423493146896363,
+      "num_tokens": 34169195.0,
+      "step": 14770
+    },
+    {
+      "epoch": 0.13550930595030714,
+      "learning_rate": 0.00017289997249472815,
+      "loss": 0.8726,
+      "mean_token_accuracy": 0.7451719880104065,
+      "num_tokens": 34192593.0,
+      "step": 14780
+    },
+    {
+      "epoch": 0.13560099018978639,
+      "learning_rate": 0.00017288163564683232,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.7411118149757385,
+      "num_tokens": 34215861.0,
+      "step": 14790
+    },
+    {
+      "epoch": 0.1356926744292656,
+      "learning_rate": 0.00017286329879893648,
+      "loss": 0.8551,
+      "mean_token_accuracy": 0.7451322019100189,
+      "num_tokens": 34238554.0,
+      "step": 14800
+    },
+    {
+      "epoch": 0.13578435866874483,
+      "learning_rate": 0.00017284496195104062,
+      "loss": 0.8379,
+      "mean_token_accuracy": 0.738130247592926,
+      "num_tokens": 34262443.0,
+      "step": 14810
+    },
+    {
+      "epoch": 0.13587604290822408,
+      "learning_rate": 0.00017282662510314479,
+      "loss": 0.9108,
+      "mean_token_accuracy": 0.7346627950668335,
+      "num_tokens": 34285167.0,
+      "step": 14820
+    },
+    {
+      "epoch": 0.1359677271477033,
+      "learning_rate": 0.00017280828825524892,
+      "loss": 0.8718,
+      "mean_token_accuracy": 0.7400343775749206,
+      "num_tokens": 34308781.0,
+      "step": 14830
+    },
+    {
+      "epoch": 0.13605941138718256,
+      "learning_rate": 0.0001727899514073531,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7371185064315796,
+      "num_tokens": 34332493.0,
+      "step": 14840
+    },
+    {
+      "epoch": 0.13615109562666178,
+      "learning_rate": 0.00017277161455945723,
+      "loss": 0.8793,
+      "mean_token_accuracy": 0.7439113616943359,
+      "num_tokens": 34355974.0,
+      "step": 14850
+    },
+    {
+      "epoch": 0.136242779866141,
+      "learning_rate": 0.0001727532777115614,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.7329540967941284,
+      "num_tokens": 34378899.0,
+      "step": 14860
+    },
+    {
+      "epoch": 0.13633446410562025,
+      "learning_rate": 0.00017273494086366556,
+      "loss": 0.8871,
+      "mean_token_accuracy": 0.7310245156288147,
+      "num_tokens": 34401704.0,
+      "step": 14870
+    },
+    {
+      "epoch": 0.13642614834509947,
+      "learning_rate": 0.0001727166040157697,
+      "loss": 0.858,
+      "mean_token_accuracy": 0.7442012190818786,
+      "num_tokens": 34424568.0,
+      "step": 14880
+    },
+    {
+      "epoch": 0.13651783258457872,
+      "learning_rate": 0.00017269826716787386,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7480300724506378,
+      "num_tokens": 34447311.0,
+      "step": 14890
+    },
+    {
+      "epoch": 0.13660951682405795,
+      "learning_rate": 0.000172679930319978,
+      "loss": 0.8408,
+      "mean_token_accuracy": 0.7481724441051483,
+      "num_tokens": 34470874.0,
+      "step": 14900
+    },
+    {
+      "epoch": 0.13670120106353717,
+      "learning_rate": 0.00017266159347208214,
+      "loss": 0.9208,
+      "mean_token_accuracy": 0.7317855417728424,
+      "num_tokens": 34493724.0,
+      "step": 14910
+    },
+    {
+      "epoch": 0.13679288530301642,
+      "learning_rate": 0.0001726432566241863,
+      "loss": 0.8601,
+      "mean_token_accuracy": 0.7386585772037506,
+      "num_tokens": 34517071.0,
+      "step": 14920
+    },
+    {
+      "epoch": 0.13688456954249564,
+      "learning_rate": 0.00017262491977629047,
+      "loss": 0.8699,
+      "mean_token_accuracy": 0.7394436359405517,
+      "num_tokens": 34540311.0,
+      "step": 14930
+    },
+    {
+      "epoch": 0.13697625378197487,
+      "learning_rate": 0.00017260658292839464,
+      "loss": 0.8761,
+      "mean_token_accuracy": 0.7429749131202698,
+      "num_tokens": 34563600.0,
+      "step": 14940
+    },
+    {
+      "epoch": 0.13706793802145412,
+      "learning_rate": 0.00017258824608049877,
+      "loss": 0.8647,
+      "mean_token_accuracy": 0.7409651696681976,
+      "num_tokens": 34586227.0,
+      "step": 14950
+    },
+    {
+      "epoch": 0.13715962226093334,
+      "learning_rate": 0.0001725699092326029,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.7521701097488404,
+      "num_tokens": 34609307.0,
+      "step": 14960
+    },
+    {
+      "epoch": 0.1372513065004126,
+      "learning_rate": 0.00017255157238470708,
+      "loss": 0.8579,
+      "mean_token_accuracy": 0.7402417600154877,
+      "num_tokens": 34632401.0,
+      "step": 14970
+    },
+    {
+      "epoch": 0.1373429907398918,
+      "learning_rate": 0.00017253323553681122,
+      "loss": 0.8366,
+      "mean_token_accuracy": 0.7455971479415894,
+      "num_tokens": 34655572.0,
+      "step": 14980
+    },
+    {
+      "epoch": 0.13743467497937104,
+      "learning_rate": 0.00017251489868891538,
+      "loss": 0.8303,
+      "mean_token_accuracy": 0.7452256083488464,
+      "num_tokens": 34678111.0,
+      "step": 14990
+    },
+    {
+      "epoch": 0.13752635921885029,
+      "learning_rate": 0.00017249656184101955,
+      "loss": 0.8298,
+      "mean_token_accuracy": 0.7465719163417817,
+      "num_tokens": 34702553.0,
+      "step": 15000
+    },
+    {
+      "epoch": 0.1376180434583295,
+      "learning_rate": 0.00017247822499312368,
+      "loss": 0.8511,
+      "mean_token_accuracy": 0.7408022582530975,
+      "num_tokens": 34725268.0,
+      "step": 15010
+    },
+    {
+      "epoch": 0.13770972769780876,
+      "learning_rate": 0.00017245988814522785,
+      "loss": 0.8866,
+      "mean_token_accuracy": 0.7359605014324189,
+      "num_tokens": 34748284.0,
+      "step": 15020
+    },
+    {
+      "epoch": 0.13780141193728798,
+      "learning_rate": 0.000172441551297332,
+      "loss": 0.8828,
+      "mean_token_accuracy": 0.7411446869373322,
+      "num_tokens": 34772131.0,
+      "step": 15030
+    },
+    {
+      "epoch": 0.1378930961767672,
+      "learning_rate": 0.00017242321444943615,
+      "loss": 0.8496,
+      "mean_token_accuracy": 0.7462159633636475,
+      "num_tokens": 34795648.0,
+      "step": 15040
+    },
+    {
+      "epoch": 0.13798478041624646,
+      "learning_rate": 0.0001724048776015403,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7407910585403442,
+      "num_tokens": 34819991.0,
+      "step": 15050
+    },
+    {
+      "epoch": 0.13807646465572568,
+      "learning_rate": 0.00017238654075364446,
+      "loss": 0.8682,
+      "mean_token_accuracy": 0.742782711982727,
+      "num_tokens": 34843500.0,
+      "step": 15060
+    },
+    {
+      "epoch": 0.1381681488952049,
+      "learning_rate": 0.00017236820390574862,
+      "loss": 0.857,
+      "mean_token_accuracy": 0.7453341901302337,
+      "num_tokens": 34867460.0,
+      "step": 15070
+    },
+    {
+      "epoch": 0.13825983313468415,
+      "learning_rate": 0.00017234986705785276,
+      "loss": 0.8539,
+      "mean_token_accuracy": 0.7477288484573364,
+      "num_tokens": 34891119.0,
+      "step": 15080
+    },
+    {
+      "epoch": 0.13835151737416337,
+      "learning_rate": 0.00017233153020995693,
+      "loss": 0.8712,
+      "mean_token_accuracy": 0.7427833139896393,
+      "num_tokens": 34914637.0,
+      "step": 15090
+    },
+    {
+      "epoch": 0.13844320161364262,
+      "learning_rate": 0.00017231319336206107,
+      "loss": 0.8924,
+      "mean_token_accuracy": 0.7354365885257721,
+      "num_tokens": 34938035.0,
+      "step": 15100
+    },
+    {
+      "epoch": 0.13853488585312185,
+      "learning_rate": 0.0001722948565141652,
+      "loss": 0.8967,
+      "mean_token_accuracy": 0.7368919432163239,
+      "num_tokens": 34962046.0,
+      "step": 15110
+    },
+    {
+      "epoch": 0.13862657009260107,
+      "learning_rate": 0.0001722765196662694,
+      "loss": 0.8942,
+      "mean_token_accuracy": 0.7349571168422699,
+      "num_tokens": 34985902.0,
+      "step": 15120
+    },
+    {
+      "epoch": 0.13871825433208032,
+      "learning_rate": 0.00017225818281837353,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7445537507534027,
+      "num_tokens": 35008986.0,
+      "step": 15130
+    },
+    {
+      "epoch": 0.13880993857155954,
+      "learning_rate": 0.0001722398459704777,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7431390702724456,
+      "num_tokens": 35032371.0,
+      "step": 15140
+    },
+    {
+      "epoch": 0.1389016228110388,
+      "learning_rate": 0.00017222150912258184,
+      "loss": 0.9225,
+      "mean_token_accuracy": 0.7271012902259827,
+      "num_tokens": 35055932.0,
+      "step": 15150
+    },
+    {
+      "epoch": 0.13899330705051802,
+      "learning_rate": 0.00017220317227468598,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7437769711017609,
+      "num_tokens": 35079900.0,
+      "step": 15160
+    },
+    {
+      "epoch": 0.13908499128999724,
+      "learning_rate": 0.00017218483542679014,
+      "loss": 0.8699,
+      "mean_token_accuracy": 0.741841334104538,
+      "num_tokens": 35103658.0,
+      "step": 15170
+    },
+    {
+      "epoch": 0.1391766755294765,
+      "learning_rate": 0.00017216649857889428,
+      "loss": 0.8824,
+      "mean_token_accuracy": 0.7311523973941803,
+      "num_tokens": 35126704.0,
+      "step": 15180
+    },
+    {
+      "epoch": 0.1392683597689557,
+      "learning_rate": 0.00017214816173099845,
+      "loss": 0.8449,
+      "mean_token_accuracy": 0.7465055763721467,
+      "num_tokens": 35149803.0,
+      "step": 15190
+    },
+    {
+      "epoch": 0.13936004400843496,
+      "learning_rate": 0.0001721298248831026,
+      "loss": 0.8934,
+      "mean_token_accuracy": 0.7403610050678253,
+      "num_tokens": 35173329.0,
+      "step": 15200
+    },
+    {
+      "epoch": 0.1394517282479142,
+      "learning_rate": 0.00017211148803520675,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7426059722900391,
+      "num_tokens": 35196346.0,
+      "step": 15210
+    },
+    {
+      "epoch": 0.1395434124873934,
+      "learning_rate": 0.00017209315118731091,
+      "loss": 0.8415,
+      "mean_token_accuracy": 0.7491662502288818,
+      "num_tokens": 35219268.0,
+      "step": 15220
+    },
+    {
+      "epoch": 0.13963509672687266,
+      "learning_rate": 0.00017207481433941505,
+      "loss": 0.9077,
+      "mean_token_accuracy": 0.7325476408004761,
+      "num_tokens": 35242623.0,
+      "step": 15230
+    },
+    {
+      "epoch": 0.13972678096635188,
+      "learning_rate": 0.00017205647749151922,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.7378370702266693,
+      "num_tokens": 35266277.0,
+      "step": 15240
+    },
+    {
+      "epoch": 0.1398184652058311,
+      "learning_rate": 0.00017203814064362338,
+      "loss": 0.8741,
+      "mean_token_accuracy": 0.7393323838710785,
+      "num_tokens": 35289214.0,
+      "step": 15250
+    },
+    {
+      "epoch": 0.13991014944531036,
+      "learning_rate": 0.00017201980379572752,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7427632868289947,
+      "num_tokens": 35311458.0,
+      "step": 15260
+    },
+    {
+      "epoch": 0.14000183368478958,
+      "learning_rate": 0.0001720014669478317,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7470477044582366,
+      "num_tokens": 35333737.0,
+      "step": 15270
+    },
+    {
+      "epoch": 0.14009351792426883,
+      "learning_rate": 0.00017198313009993583,
+      "loss": 0.8898,
+      "mean_token_accuracy": 0.7392634630203248,
+      "num_tokens": 35357066.0,
+      "step": 15280
+    },
+    {
+      "epoch": 0.14018520216374805,
+      "learning_rate": 0.00017196479325204,
+      "loss": 0.891,
+      "mean_token_accuracy": 0.741010183095932,
+      "num_tokens": 35379496.0,
+      "step": 15290
+    },
+    {
+      "epoch": 0.14027688640322727,
+      "learning_rate": 0.00017194645640414413,
+      "loss": 0.8885,
+      "mean_token_accuracy": 0.7416484355926514,
+      "num_tokens": 35402315.0,
+      "step": 15300
+    },
+    {
+      "epoch": 0.14036857064270652,
+      "learning_rate": 0.00017192811955624827,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.736742228269577,
+      "num_tokens": 35425384.0,
+      "step": 15310
+    },
+    {
+      "epoch": 0.14046025488218575,
+      "learning_rate": 0.00017190978270835246,
+      "loss": 0.8352,
+      "mean_token_accuracy": 0.748482209444046,
+      "num_tokens": 35448273.0,
+      "step": 15320
+    },
+    {
+      "epoch": 0.140551939121665,
+      "learning_rate": 0.0001718914458604566,
+      "loss": 0.8616,
+      "mean_token_accuracy": 0.7401842474937439,
+      "num_tokens": 35471443.0,
+      "step": 15330
+    },
+    {
+      "epoch": 0.14064362336114422,
+      "learning_rate": 0.00017187310901256076,
+      "loss": 0.8454,
+      "mean_token_accuracy": 0.7439776420593261,
+      "num_tokens": 35494220.0,
+      "step": 15340
+    },
+    {
+      "epoch": 0.14073530760062344,
+      "learning_rate": 0.0001718547721646649,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.7468129754066467,
+      "num_tokens": 35516708.0,
+      "step": 15350
+    },
+    {
+      "epoch": 0.1408269918401027,
+      "learning_rate": 0.00017183643531676904,
+      "loss": 0.8568,
+      "mean_token_accuracy": 0.7475004255771637,
+      "num_tokens": 35539775.0,
+      "step": 15360
+    },
+    {
+      "epoch": 0.14091867607958192,
+      "learning_rate": 0.0001718180984688732,
+      "loss": 0.8814,
+      "mean_token_accuracy": 0.7429614782333374,
+      "num_tokens": 35564023.0,
+      "step": 15370
+    },
+    {
+      "epoch": 0.14101036031906114,
+      "learning_rate": 0.00017179976162097737,
+      "loss": 0.8494,
+      "mean_token_accuracy": 0.748023945093155,
+      "num_tokens": 35587842.0,
+      "step": 15380
+    },
+    {
+      "epoch": 0.1411020445585404,
+      "learning_rate": 0.0001717814247730815,
+      "loss": 0.894,
+      "mean_token_accuracy": 0.7352497637271881,
+      "num_tokens": 35610698.0,
+      "step": 15390
+    },
+    {
+      "epoch": 0.1411937287980196,
+      "learning_rate": 0.00017176308792518568,
+      "loss": 0.8624,
+      "mean_token_accuracy": 0.7445066869258881,
+      "num_tokens": 35633919.0,
+      "step": 15400
+    },
+    {
+      "epoch": 0.14128541303749886,
+      "learning_rate": 0.00017174475107728981,
+      "loss": 0.8812,
+      "mean_token_accuracy": 0.7427648067474365,
+      "num_tokens": 35656371.0,
+      "step": 15410
+    },
+    {
+      "epoch": 0.1413770972769781,
+      "learning_rate": 0.00017172641422939398,
+      "loss": 0.858,
+      "mean_token_accuracy": 0.7444013237953186,
+      "num_tokens": 35679926.0,
+      "step": 15420
+    },
+    {
+      "epoch": 0.1414687815164573,
+      "learning_rate": 0.00017170807738149812,
+      "loss": 0.8924,
+      "mean_token_accuracy": 0.7378455460071563,
+      "num_tokens": 35704186.0,
+      "step": 15430
+    },
+    {
+      "epoch": 0.14156046575593656,
+      "learning_rate": 0.00017168974053360228,
+      "loss": 0.8943,
+      "mean_token_accuracy": 0.7372198641300202,
+      "num_tokens": 35727393.0,
+      "step": 15440
+    },
+    {
+      "epoch": 0.14165214999541578,
+      "learning_rate": 0.00017167140368570645,
+      "loss": 0.8957,
+      "mean_token_accuracy": 0.734748101234436,
+      "num_tokens": 35749874.0,
+      "step": 15450
+    },
+    {
+      "epoch": 0.14174383423489503,
+      "learning_rate": 0.0001716530668378106,
+      "loss": 0.8487,
+      "mean_token_accuracy": 0.7426269173622131,
+      "num_tokens": 35772862.0,
+      "step": 15460
+    },
+    {
+      "epoch": 0.14183551847437426,
+      "learning_rate": 0.00017163472998991475,
+      "loss": 0.8364,
+      "mean_token_accuracy": 0.7502005517482757,
+      "num_tokens": 35795644.0,
+      "step": 15470
+    },
+    {
+      "epoch": 0.14192720271385348,
+      "learning_rate": 0.0001716163931420189,
+      "loss": 0.8928,
+      "mean_token_accuracy": 0.7369527220726013,
+      "num_tokens": 35818756.0,
+      "step": 15480
+    },
+    {
+      "epoch": 0.14201888695333273,
+      "learning_rate": 0.00017159805629412306,
+      "loss": 0.8422,
+      "mean_token_accuracy": 0.748214191198349,
+      "num_tokens": 35841885.0,
+      "step": 15490
+    },
+    {
+      "epoch": 0.14211057119281195,
+      "learning_rate": 0.0001715797194462272,
+      "loss": 0.8917,
+      "mean_token_accuracy": 0.7364135205745697,
+      "num_tokens": 35865058.0,
+      "step": 15500
+    },
+    {
+      "epoch": 0.1422022554322912,
+      "learning_rate": 0.00017156138259833136,
+      "loss": 0.8266,
+      "mean_token_accuracy": 0.7535389542579651,
+      "num_tokens": 35888439.0,
+      "step": 15510
+    },
+    {
+      "epoch": 0.14229393967177043,
+      "learning_rate": 0.00017154304575043552,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.736277061700821,
+      "num_tokens": 35912253.0,
+      "step": 15520
+    },
+    {
+      "epoch": 0.14238562391124965,
+      "learning_rate": 0.00017152470890253966,
+      "loss": 0.8887,
+      "mean_token_accuracy": 0.7365772306919098,
+      "num_tokens": 35935184.0,
+      "step": 15530
+    },
+    {
+      "epoch": 0.1424773081507289,
+      "learning_rate": 0.00017150637205464383,
+      "loss": 0.9029,
+      "mean_token_accuracy": 0.7392226040363312,
+      "num_tokens": 35957892.0,
+      "step": 15540
+    },
+    {
+      "epoch": 0.14256899239020812,
+      "learning_rate": 0.00017148803520674797,
+      "loss": 0.8513,
+      "mean_token_accuracy": 0.7421040892601013,
+      "num_tokens": 35980740.0,
+      "step": 15550
+    },
+    {
+      "epoch": 0.14266067662968734,
+      "learning_rate": 0.0001714696983588521,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.738554447889328,
+      "num_tokens": 36004105.0,
+      "step": 15560
+    },
+    {
+      "epoch": 0.1427523608691666,
+      "learning_rate": 0.00017145136151095627,
+      "loss": 0.8761,
+      "mean_token_accuracy": 0.7424564182758331,
+      "num_tokens": 36027396.0,
+      "step": 15570
+    },
+    {
+      "epoch": 0.14284404510864582,
+      "learning_rate": 0.00017143302466306044,
+      "loss": 0.8548,
+      "mean_token_accuracy": 0.741351866722107,
+      "num_tokens": 36050220.0,
+      "step": 15580
+    },
+    {
+      "epoch": 0.14293572934812507,
+      "learning_rate": 0.00017141468781516457,
+      "loss": 0.8722,
+      "mean_token_accuracy": 0.7424733400344848,
+      "num_tokens": 36074020.0,
+      "step": 15590
+    },
+    {
+      "epoch": 0.1430274135876043,
+      "learning_rate": 0.00017139635096726874,
+      "loss": 0.8819,
+      "mean_token_accuracy": 0.7411562919616699,
+      "num_tokens": 36096764.0,
+      "step": 15600
+    },
+    {
+      "epoch": 0.1431190978270835,
+      "learning_rate": 0.00017137801411937288,
+      "loss": 0.8361,
+      "mean_token_accuracy": 0.7521393835544586,
+      "num_tokens": 36119354.0,
+      "step": 15610
+    },
+    {
+      "epoch": 0.14321078206656276,
+      "learning_rate": 0.00017135967727147704,
+      "loss": 0.8811,
+      "mean_token_accuracy": 0.7374622523784637,
+      "num_tokens": 36143320.0,
+      "step": 15620
+    },
+    {
+      "epoch": 0.143302466306042,
+      "learning_rate": 0.00017134134042358118,
+      "loss": 0.9099,
+      "mean_token_accuracy": 0.7347093343734741,
+      "num_tokens": 36167441.0,
+      "step": 15630
+    },
+    {
+      "epoch": 0.14339415054552124,
+      "learning_rate": 0.00017132300357568535,
+      "loss": 0.8793,
+      "mean_token_accuracy": 0.7435518801212311,
+      "num_tokens": 36190208.0,
+      "step": 15640
+    },
+    {
+      "epoch": 0.14348583478500046,
+      "learning_rate": 0.0001713046667277895,
+      "loss": 0.8869,
+      "mean_token_accuracy": 0.7344530284404754,
+      "num_tokens": 36212747.0,
+      "step": 15650
+    },
+    {
+      "epoch": 0.14357751902447968,
+      "learning_rate": 0.00017128632987989365,
+      "loss": 0.9215,
+      "mean_token_accuracy": 0.7322555363178254,
+      "num_tokens": 36236226.0,
+      "step": 15660
+    },
+    {
+      "epoch": 0.14366920326395893,
+      "learning_rate": 0.00017126799303199782,
+      "loss": 0.916,
+      "mean_token_accuracy": 0.739119553565979,
+      "num_tokens": 36259439.0,
+      "step": 15670
+    },
+    {
+      "epoch": 0.14376088750343816,
+      "learning_rate": 0.00017124965618410195,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7416538894176483,
+      "num_tokens": 36282354.0,
+      "step": 15680
+    },
+    {
+      "epoch": 0.14385257174291738,
+      "learning_rate": 0.00017123131933620612,
+      "loss": 0.8778,
+      "mean_token_accuracy": 0.7373986840248108,
+      "num_tokens": 36305510.0,
+      "step": 15690
+    },
+    {
+      "epoch": 0.14394425598239663,
+      "learning_rate": 0.00017121298248831026,
+      "loss": 0.8462,
+      "mean_token_accuracy": 0.753288346529007,
+      "num_tokens": 36328607.0,
+      "step": 15700
+    },
+    {
+      "epoch": 0.14403594022187585,
+      "learning_rate": 0.00017119464564041442,
+      "loss": 0.8365,
+      "mean_token_accuracy": 0.7509838223457337,
+      "num_tokens": 36351930.0,
+      "step": 15710
+    },
+    {
+      "epoch": 0.1441276244613551,
+      "learning_rate": 0.0001711763087925186,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.7432153224945068,
+      "num_tokens": 36374693.0,
+      "step": 15720
+    },
+    {
+      "epoch": 0.14421930870083433,
+      "learning_rate": 0.00017115797194462273,
+      "loss": 0.8842,
+      "mean_token_accuracy": 0.7355160593986512,
+      "num_tokens": 36398007.0,
+      "step": 15730
+    },
+    {
+      "epoch": 0.14431099294031355,
+      "learning_rate": 0.0001711396350967269,
+      "loss": 0.8893,
+      "mean_token_accuracy": 0.7363717138767243,
+      "num_tokens": 36420952.0,
+      "step": 15740
+    },
+    {
+      "epoch": 0.1444026771797928,
+      "learning_rate": 0.00017112129824883103,
+      "loss": 0.8924,
+      "mean_token_accuracy": 0.7350456714630127,
+      "num_tokens": 36443933.0,
+      "step": 15750
+    },
+    {
+      "epoch": 0.14449436141927202,
+      "learning_rate": 0.00017110296140093517,
+      "loss": 0.9009,
+      "mean_token_accuracy": 0.7325533866882324,
+      "num_tokens": 36466978.0,
+      "step": 15760
+    },
+    {
+      "epoch": 0.14458604565875127,
+      "learning_rate": 0.00017108462455303934,
+      "loss": 0.8756,
+      "mean_token_accuracy": 0.7401955187320709,
+      "num_tokens": 36489975.0,
+      "step": 15770
+    },
+    {
+      "epoch": 0.1446777298982305,
+      "learning_rate": 0.0001710662877051435,
+      "loss": 0.856,
+      "mean_token_accuracy": 0.7392392814159393,
+      "num_tokens": 36513624.0,
+      "step": 15780
+    },
+    {
+      "epoch": 0.14476941413770972,
+      "learning_rate": 0.00017104795085724764,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7432485938072204,
+      "num_tokens": 36536843.0,
+      "step": 15790
+    },
+    {
+      "epoch": 0.14486109837718897,
+      "learning_rate": 0.0001710296140093518,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7388611733913422,
+      "num_tokens": 36559656.0,
+      "step": 15800
+    },
+    {
+      "epoch": 0.1449527826166682,
+      "learning_rate": 0.00017101127716145594,
+      "loss": 0.9033,
+      "mean_token_accuracy": 0.7361470818519592,
+      "num_tokens": 36583295.0,
+      "step": 15810
+    },
+    {
+      "epoch": 0.14504446685614744,
+      "learning_rate": 0.0001709929403135601,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.7397248089313507,
+      "num_tokens": 36605779.0,
+      "step": 15820
+    },
+    {
+      "epoch": 0.14513615109562666,
+      "learning_rate": 0.00017097460346566425,
+      "loss": 0.8691,
+      "mean_token_accuracy": 0.7428908169269561,
+      "num_tokens": 36628121.0,
+      "step": 15830
+    },
+    {
+      "epoch": 0.1452278353351059,
+      "learning_rate": 0.0001709562666177684,
+      "loss": 0.8702,
+      "mean_token_accuracy": 0.7383060693740845,
+      "num_tokens": 36651628.0,
+      "step": 15840
+    },
+    {
+      "epoch": 0.14531951957458514,
+      "learning_rate": 0.00017093792976987258,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7416314542293548,
+      "num_tokens": 36674567.0,
+      "step": 15850
+    },
+    {
+      "epoch": 0.14541120381406436,
+      "learning_rate": 0.00017091959292197672,
+      "loss": 0.9071,
+      "mean_token_accuracy": 0.734781700372696,
+      "num_tokens": 36697863.0,
+      "step": 15860
+    },
+    {
+      "epoch": 0.14550288805354358,
+      "learning_rate": 0.00017090125607408088,
+      "loss": 0.9083,
+      "mean_token_accuracy": 0.7320820689201355,
+      "num_tokens": 36721003.0,
+      "step": 15870
+    },
+    {
+      "epoch": 0.14559457229302283,
+      "learning_rate": 0.00017088291922618502,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7432838439941406,
+      "num_tokens": 36743910.0,
+      "step": 15880
+    },
+    {
+      "epoch": 0.14568625653250206,
+      "learning_rate": 0.00017086458237828918,
+      "loss": 0.8947,
+      "mean_token_accuracy": 0.739415991306305,
+      "num_tokens": 36766315.0,
+      "step": 15890
+    },
+    {
+      "epoch": 0.1457779407719813,
+      "learning_rate": 0.00017084624553039332,
+      "loss": 0.8755,
+      "mean_token_accuracy": 0.7403792202472687,
+      "num_tokens": 36788852.0,
+      "step": 15900
+    },
+    {
+      "epoch": 0.14586962501146053,
+      "learning_rate": 0.0001708279086824975,
+      "loss": 0.8744,
+      "mean_token_accuracy": 0.7431094586849213,
+      "num_tokens": 36811735.0,
+      "step": 15910
+    },
+    {
+      "epoch": 0.14596130925093975,
+      "learning_rate": 0.00017080957183460165,
+      "loss": 0.8407,
+      "mean_token_accuracy": 0.7473398447036743,
+      "num_tokens": 36834411.0,
+      "step": 15920
+    },
+    {
+      "epoch": 0.146052993490419,
+      "learning_rate": 0.0001707912349867058,
+      "loss": 0.8466,
+      "mean_token_accuracy": 0.749835342168808,
+      "num_tokens": 36856962.0,
+      "step": 15930
+    },
+    {
+      "epoch": 0.14614467772989823,
+      "learning_rate": 0.00017077289813880996,
+      "loss": 0.8417,
+      "mean_token_accuracy": 0.7449469566345215,
+      "num_tokens": 36880853.0,
+      "step": 15940
+    },
+    {
+      "epoch": 0.14623636196937748,
+      "learning_rate": 0.0001707545612909141,
+      "loss": 0.8946,
+      "mean_token_accuracy": 0.7358967542648316,
+      "num_tokens": 36904041.0,
+      "step": 15950
+    },
+    {
+      "epoch": 0.1463280462088567,
+      "learning_rate": 0.00017073622444301823,
+      "loss": 0.9127,
+      "mean_token_accuracy": 0.7398579776287079,
+      "num_tokens": 36927780.0,
+      "step": 15960
+    },
+    {
+      "epoch": 0.14641973044833592,
+      "learning_rate": 0.00017071788759512243,
+      "loss": 0.8529,
+      "mean_token_accuracy": 0.7463305592536926,
+      "num_tokens": 36950260.0,
+      "step": 15970
+    },
+    {
+      "epoch": 0.14651141468781517,
+      "learning_rate": 0.00017069955074722656,
+      "loss": 0.8771,
+      "mean_token_accuracy": 0.7401873469352722,
+      "num_tokens": 36973332.0,
+      "step": 15980
+    },
+    {
+      "epoch": 0.1466030989272944,
+      "learning_rate": 0.0001706812138993307,
+      "loss": 0.8793,
+      "mean_token_accuracy": 0.7394184529781341,
+      "num_tokens": 36996111.0,
+      "step": 15990
+    },
+    {
+      "epoch": 0.14669478316677362,
+      "learning_rate": 0.00017066287705143487,
+      "loss": 0.88,
+      "mean_token_accuracy": 0.7426718354225159,
+      "num_tokens": 37019427.0,
+      "step": 16000
+    },
+    {
+      "epoch": 0.14678646740625287,
+      "learning_rate": 0.000170644540203539,
+      "loss": 0.9024,
+      "mean_token_accuracy": 0.7295919239521027,
+      "num_tokens": 37043087.0,
+      "step": 16010
+    },
+    {
+      "epoch": 0.1468781516457321,
+      "learning_rate": 0.00017062620335564317,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7450354874134064,
+      "num_tokens": 37066364.0,
+      "step": 16020
+    },
+    {
+      "epoch": 0.14696983588521134,
+      "learning_rate": 0.0001706078665077473,
+      "loss": 0.8659,
+      "mean_token_accuracy": 0.7447131872177124,
+      "num_tokens": 37089369.0,
+      "step": 16030
+    },
+    {
+      "epoch": 0.14706152012469056,
+      "learning_rate": 0.00017058952965985148,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7468583405017852,
+      "num_tokens": 37112609.0,
+      "step": 16040
+    },
+    {
+      "epoch": 0.1471532043641698,
+      "learning_rate": 0.00017057119281195564,
+      "loss": 0.9012,
+      "mean_token_accuracy": 0.7315944612026215,
+      "num_tokens": 37135391.0,
+      "step": 16050
+    },
+    {
+      "epoch": 0.14724488860364904,
+      "learning_rate": 0.00017055285596405978,
+      "loss": 0.8897,
+      "mean_token_accuracy": 0.7404382467269898,
+      "num_tokens": 37157542.0,
+      "step": 16060
+    },
+    {
+      "epoch": 0.14733657284312826,
+      "learning_rate": 0.00017053451911616395,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7500638067722321,
+      "num_tokens": 37180347.0,
+      "step": 16070
+    },
+    {
+      "epoch": 0.1474282570826075,
+      "learning_rate": 0.00017051618226826808,
+      "loss": 0.865,
+      "mean_token_accuracy": 0.7455952882766723,
+      "num_tokens": 37203148.0,
+      "step": 16080
+    },
+    {
+      "epoch": 0.14751994132208673,
+      "learning_rate": 0.00017049784542037225,
+      "loss": 0.8444,
+      "mean_token_accuracy": 0.7497556746006012,
+      "num_tokens": 37226287.0,
+      "step": 16090
+    },
+    {
+      "epoch": 0.14761162556156596,
+      "learning_rate": 0.00017047950857247641,
+      "loss": 0.8825,
+      "mean_token_accuracy": 0.7477022767066955,
+      "num_tokens": 37249004.0,
+      "step": 16100
+    },
+    {
+      "epoch": 0.1477033098010452,
+      "learning_rate": 0.00017046117172458055,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7420090973377228,
+      "num_tokens": 37272304.0,
+      "step": 16110
+    },
+    {
+      "epoch": 0.14779499404052443,
+      "learning_rate": 0.00017044283487668472,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.7381621420383453,
+      "num_tokens": 37295281.0,
+      "step": 16120
+    },
+    {
+      "epoch": 0.14788667828000368,
+      "learning_rate": 0.00017042449802878886,
+      "loss": 0.8722,
+      "mean_token_accuracy": 0.740356034040451,
+      "num_tokens": 37317828.0,
+      "step": 16130
+    },
+    {
+      "epoch": 0.1479783625194829,
+      "learning_rate": 0.00017040616118089302,
+      "loss": 0.8633,
+      "mean_token_accuracy": 0.7404953300952911,
+      "num_tokens": 37340704.0,
+      "step": 16140
+    },
+    {
+      "epoch": 0.14807004675896213,
+      "learning_rate": 0.00017038782433299716,
+      "loss": 0.8389,
+      "mean_token_accuracy": 0.7452967226505279,
+      "num_tokens": 37362928.0,
+      "step": 16150
+    },
+    {
+      "epoch": 0.14816173099844138,
+      "learning_rate": 0.0001703694874851013,
+      "loss": 0.8689,
+      "mean_token_accuracy": 0.7365191996097564,
+      "num_tokens": 37385600.0,
+      "step": 16160
+    },
+    {
+      "epoch": 0.1482534152379206,
+      "learning_rate": 0.0001703511506372055,
+      "loss": 0.8949,
+      "mean_token_accuracy": 0.7305776953697205,
+      "num_tokens": 37408518.0,
+      "step": 16170
+    },
+    {
+      "epoch": 0.14834509947739982,
+      "learning_rate": 0.00017033281378930963,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.7508726000785828,
+      "num_tokens": 37431714.0,
+      "step": 16180
+    },
+    {
+      "epoch": 0.14843678371687907,
+      "learning_rate": 0.00017031447694141377,
+      "loss": 0.8877,
+      "mean_token_accuracy": 0.7442928493022919,
+      "num_tokens": 37455286.0,
+      "step": 16190
+    },
+    {
+      "epoch": 0.1485284679563583,
+      "learning_rate": 0.00017029614009351793,
+      "loss": 0.8895,
+      "mean_token_accuracy": 0.738189160823822,
+      "num_tokens": 37478469.0,
+      "step": 16200
+    },
+    {
+      "epoch": 0.14862015219583755,
+      "learning_rate": 0.00017027780324562207,
+      "loss": 0.8995,
+      "mean_token_accuracy": 0.7329420864582061,
+      "num_tokens": 37501267.0,
+      "step": 16210
+    },
+    {
+      "epoch": 0.14871183643531677,
+      "learning_rate": 0.00017025946639772624,
+      "loss": 0.8444,
+      "mean_token_accuracy": 0.7452007949352264,
+      "num_tokens": 37523554.0,
+      "step": 16220
+    },
+    {
+      "epoch": 0.148803520674796,
+      "learning_rate": 0.0001702411295498304,
+      "loss": 0.9022,
+      "mean_token_accuracy": 0.73763148188591,
+      "num_tokens": 37547044.0,
+      "step": 16230
+    },
+    {
+      "epoch": 0.14889520491427524,
+      "learning_rate": 0.00017022279270193454,
+      "loss": 0.8462,
+      "mean_token_accuracy": 0.7493780314922333,
+      "num_tokens": 37569517.0,
+      "step": 16240
+    },
+    {
+      "epoch": 0.14898688915375446,
+      "learning_rate": 0.0001702044558540387,
+      "loss": 0.9,
+      "mean_token_accuracy": 0.736665791273117,
+      "num_tokens": 37592746.0,
+      "step": 16250
+    },
+    {
+      "epoch": 0.14907857339323372,
+      "learning_rate": 0.00017018611900614284,
+      "loss": 0.9179,
+      "mean_token_accuracy": 0.7329012751579285,
+      "num_tokens": 37616892.0,
+      "step": 16260
+    },
+    {
+      "epoch": 0.14917025763271294,
+      "learning_rate": 0.000170167782158247,
+      "loss": 0.8869,
+      "mean_token_accuracy": 0.7348154604434967,
+      "num_tokens": 37640878.0,
+      "step": 16270
+    },
+    {
+      "epoch": 0.14926194187219216,
+      "learning_rate": 0.00017014944531035115,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.7386293411254883,
+      "num_tokens": 37664360.0,
+      "step": 16280
+    },
+    {
+      "epoch": 0.1493536261116714,
+      "learning_rate": 0.0001701311084624553,
+      "loss": 0.8402,
+      "mean_token_accuracy": 0.7535671114921569,
+      "num_tokens": 37688008.0,
+      "step": 16290
+    },
+    {
+      "epoch": 0.14944531035115063,
+      "learning_rate": 0.00017011277161455948,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7393059194087982,
+      "num_tokens": 37710843.0,
+      "step": 16300
+    },
+    {
+      "epoch": 0.14953699459062986,
+      "learning_rate": 0.00017009443476666362,
+      "loss": 0.8823,
+      "mean_token_accuracy": 0.7435824513435364,
+      "num_tokens": 37734174.0,
+      "step": 16310
+    },
+    {
+      "epoch": 0.1496286788301091,
+      "learning_rate": 0.00017007609791876778,
+      "loss": 0.8803,
+      "mean_token_accuracy": 0.7369657516479492,
+      "num_tokens": 37757725.0,
+      "step": 16320
+    },
+    {
+      "epoch": 0.14972036306958833,
+      "learning_rate": 0.00017005776107087192,
+      "loss": 0.8222,
+      "mean_token_accuracy": 0.7561168491840362,
+      "num_tokens": 37780735.0,
+      "step": 16330
+    },
+    {
+      "epoch": 0.14981204730906758,
+      "learning_rate": 0.00017003942422297609,
+      "loss": 0.8919,
+      "mean_token_accuracy": 0.7403030455112457,
+      "num_tokens": 37803988.0,
+      "step": 16340
+    },
+    {
+      "epoch": 0.1499037315485468,
+      "learning_rate": 0.00017002108737508022,
+      "loss": 0.8943,
+      "mean_token_accuracy": 0.7361368477344513,
+      "num_tokens": 37827326.0,
+      "step": 16350
+    },
+    {
+      "epoch": 0.14999541578802603,
+      "learning_rate": 0.0001700027505271844,
+      "loss": 0.9001,
+      "mean_token_accuracy": 0.7352225959300995,
+      "num_tokens": 37849799.0,
+      "step": 16360
+    },
+    {
+      "epoch": 0.15008710002750528,
+      "learning_rate": 0.00016998441367928856,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.7423892736434936,
+      "num_tokens": 37872842.0,
+      "step": 16370
+    },
+    {
+      "epoch": 0.1501787842669845,
+      "learning_rate": 0.0001699660768313927,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.74049591422081,
+      "num_tokens": 37896520.0,
+      "step": 16380
+    },
+    {
+      "epoch": 0.15027046850646375,
+      "learning_rate": 0.00016994773998349683,
+      "loss": 0.8559,
+      "mean_token_accuracy": 0.7443911910057068,
+      "num_tokens": 37919910.0,
+      "step": 16390
+    },
+    {
+      "epoch": 0.15036215274594297,
+      "learning_rate": 0.000169929403135601,
+      "loss": 0.8546,
+      "mean_token_accuracy": 0.7446241974830627,
+      "num_tokens": 37943244.0,
+      "step": 16400
+    },
+    {
+      "epoch": 0.1504538369854222,
+      "learning_rate": 0.00016991106628770514,
+      "loss": 0.8407,
+      "mean_token_accuracy": 0.7498221933841706,
+      "num_tokens": 37966098.0,
+      "step": 16410
+    },
+    {
+      "epoch": 0.15054552122490145,
+      "learning_rate": 0.0001698927294398093,
+      "loss": 0.9167,
+      "mean_token_accuracy": 0.7348417937755585,
+      "num_tokens": 37989078.0,
+      "step": 16420
+    },
+    {
+      "epoch": 0.15063720546438067,
+      "learning_rate": 0.00016987439259191347,
+      "loss": 0.9307,
+      "mean_token_accuracy": 0.7289546847343444,
+      "num_tokens": 38012011.0,
+      "step": 16430
+    },
+    {
+      "epoch": 0.15072888970385992,
+      "learning_rate": 0.0001698560557440176,
+      "loss": 0.8998,
+      "mean_token_accuracy": 0.7262084662914277,
+      "num_tokens": 38035169.0,
+      "step": 16440
+    },
+    {
+      "epoch": 0.15082057394333914,
+      "learning_rate": 0.00016983771889612177,
+      "loss": 0.8371,
+      "mean_token_accuracy": 0.7426431775093079,
+      "num_tokens": 38058471.0,
+      "step": 16450
+    },
+    {
+      "epoch": 0.15091225818281837,
+      "learning_rate": 0.0001698193820482259,
+      "loss": 0.8904,
+      "mean_token_accuracy": 0.7356981933116913,
+      "num_tokens": 38081511.0,
+      "step": 16460
+    },
+    {
+      "epoch": 0.15100394242229762,
+      "learning_rate": 0.00016980104520033007,
+      "loss": 0.8985,
+      "mean_token_accuracy": 0.736701226234436,
+      "num_tokens": 38104079.0,
+      "step": 16470
+    },
+    {
+      "epoch": 0.15109562666177684,
+      "learning_rate": 0.0001697827083524342,
+      "loss": 0.8682,
+      "mean_token_accuracy": 0.7417209446430206,
+      "num_tokens": 38126801.0,
+      "step": 16480
+    },
+    {
+      "epoch": 0.15118731090125606,
+      "learning_rate": 0.00016976437150453838,
+      "loss": 0.8731,
+      "mean_token_accuracy": 0.7413496732711792,
+      "num_tokens": 38149794.0,
+      "step": 16490
+    },
+    {
+      "epoch": 0.1512789951407353,
+      "learning_rate": 0.00016974603465664254,
+      "loss": 0.8699,
+      "mean_token_accuracy": 0.7438834309577942,
+      "num_tokens": 38172585.0,
+      "step": 16500
+    },
+    {
+      "epoch": 0.15137067938021453,
+      "learning_rate": 0.00016972769780874668,
+      "loss": 0.84,
+      "mean_token_accuracy": 0.7425670325756073,
+      "num_tokens": 38195448.0,
+      "step": 16510
+    },
+    {
+      "epoch": 0.15146236361969379,
+      "learning_rate": 0.00016970936096085085,
+      "loss": 0.8955,
+      "mean_token_accuracy": 0.7389983892440796,
+      "num_tokens": 38219907.0,
+      "step": 16520
+    },
+    {
+      "epoch": 0.151554047859173,
+      "learning_rate": 0.00016969102411295499,
+      "loss": 0.8762,
+      "mean_token_accuracy": 0.7366839945316315,
+      "num_tokens": 38242733.0,
+      "step": 16530
+    },
+    {
+      "epoch": 0.15164573209865223,
+      "learning_rate": 0.00016967268726505915,
+      "loss": 0.9292,
+      "mean_token_accuracy": 0.7314104080200196,
+      "num_tokens": 38265375.0,
+      "step": 16540
+    },
+    {
+      "epoch": 0.15173741633813148,
+      "learning_rate": 0.0001696543504171633,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7488788306713104,
+      "num_tokens": 38288239.0,
+      "step": 16550
+    },
+    {
+      "epoch": 0.1518291005776107,
+      "learning_rate": 0.00016963601356926745,
+      "loss": 0.8892,
+      "mean_token_accuracy": 0.7305476188659668,
+      "num_tokens": 38311485.0,
+      "step": 16560
+    },
+    {
+      "epoch": 0.15192078481708995,
+      "learning_rate": 0.00016961767672137162,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7440915644168854,
+      "num_tokens": 38334683.0,
+      "step": 16570
+    },
+    {
+      "epoch": 0.15201246905656918,
+      "learning_rate": 0.00016959933987347576,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7366457760334015,
+      "num_tokens": 38357877.0,
+      "step": 16580
+    },
+    {
+      "epoch": 0.1521041532960484,
+      "learning_rate": 0.0001695810030255799,
+      "loss": 0.8546,
+      "mean_token_accuracy": 0.7432753086090088,
+      "num_tokens": 38380922.0,
+      "step": 16590
+    },
+    {
+      "epoch": 0.15219583753552765,
+      "learning_rate": 0.00016956266617768406,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7470687568187714,
+      "num_tokens": 38403242.0,
+      "step": 16600
+    },
+    {
+      "epoch": 0.15228752177500687,
+      "learning_rate": 0.0001695443293297882,
+      "loss": 0.849,
+      "mean_token_accuracy": 0.7500737011432648,
+      "num_tokens": 38426605.0,
+      "step": 16610
+    },
+    {
+      "epoch": 0.1523792060144861,
+      "learning_rate": 0.0001695259924818924,
+      "loss": 0.8631,
+      "mean_token_accuracy": 0.747471010684967,
+      "num_tokens": 38449559.0,
+      "step": 16620
+    },
+    {
+      "epoch": 0.15247089025396535,
+      "learning_rate": 0.00016950765563399653,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7461052656173706,
+      "num_tokens": 38472792.0,
+      "step": 16630
+    },
+    {
+      "epoch": 0.15256257449344457,
+      "learning_rate": 0.00016948931878610067,
+      "loss": 0.888,
+      "mean_token_accuracy": 0.7389775276184082,
+      "num_tokens": 38496150.0,
+      "step": 16640
+    },
+    {
+      "epoch": 0.15265425873292382,
+      "learning_rate": 0.00016947098193820483,
+      "loss": 0.8231,
+      "mean_token_accuracy": 0.751199460029602,
+      "num_tokens": 38519388.0,
+      "step": 16650
+    },
+    {
+      "epoch": 0.15274594297240304,
+      "learning_rate": 0.00016945264509030897,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7645283997058868,
+      "num_tokens": 38541871.0,
+      "step": 16660
+    },
+    {
+      "epoch": 0.15283762721188227,
+      "learning_rate": 0.00016943430824241314,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7426731526851654,
+      "num_tokens": 38565786.0,
+      "step": 16670
+    },
+    {
+      "epoch": 0.15292931145136152,
+      "learning_rate": 0.00016941597139451728,
+      "loss": 0.8846,
+      "mean_token_accuracy": 0.7351303577423096,
+      "num_tokens": 38588275.0,
+      "step": 16680
+    },
+    {
+      "epoch": 0.15302099569084074,
+      "learning_rate": 0.00016939763454662144,
+      "loss": 0.8786,
+      "mean_token_accuracy": 0.7357687294483185,
+      "num_tokens": 38611373.0,
+      "step": 16690
+    },
+    {
+      "epoch": 0.15311267993032,
+      "learning_rate": 0.0001693792976987256,
+      "loss": 0.8672,
+      "mean_token_accuracy": 0.7499398589134216,
+      "num_tokens": 38635003.0,
+      "step": 16700
+    },
+    {
+      "epoch": 0.1532043641697992,
+      "learning_rate": 0.00016936096085082975,
+      "loss": 0.8927,
+      "mean_token_accuracy": 0.7349049627780915,
+      "num_tokens": 38657858.0,
+      "step": 16710
+    },
+    {
+      "epoch": 0.15329604840927843,
+      "learning_rate": 0.0001693426240029339,
+      "loss": 0.8161,
+      "mean_token_accuracy": 0.7554900467395782,
+      "num_tokens": 38680885.0,
+      "step": 16720
+    },
+    {
+      "epoch": 0.15338773264875769,
+      "learning_rate": 0.00016932428715503805,
+      "loss": 0.8772,
+      "mean_token_accuracy": 0.7395583987236023,
+      "num_tokens": 38704389.0,
+      "step": 16730
+    },
+    {
+      "epoch": 0.1534794168882369,
+      "learning_rate": 0.00016930595030714222,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.7397893249988556,
+      "num_tokens": 38726402.0,
+      "step": 16740
+    },
+    {
+      "epoch": 0.15357110112771616,
+      "learning_rate": 0.00016928761345924638,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7350054621696472,
+      "num_tokens": 38748572.0,
+      "step": 16750
+    },
+    {
+      "epoch": 0.15366278536719538,
+      "learning_rate": 0.00016926927661135052,
+      "loss": 0.8551,
+      "mean_token_accuracy": 0.7403785765171051,
+      "num_tokens": 38771042.0,
+      "step": 16760
+    },
+    {
+      "epoch": 0.1537544696066746,
+      "learning_rate": 0.00016925093976345468,
+      "loss": 0.8563,
+      "mean_token_accuracy": 0.7481821477413177,
+      "num_tokens": 38793985.0,
+      "step": 16770
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "learning_rate": 0.00016923260291555882,
+      "loss": 0.9078,
+      "mean_token_accuracy": 0.7312174201011657,
+      "num_tokens": 38817376.0,
+      "step": 16780
+    },
+    {
+      "epoch": 0.15393783808563308,
+      "learning_rate": 0.00016921426606766296,
+      "loss": 0.896,
+      "mean_token_accuracy": 0.7304867029190063,
+      "num_tokens": 38840997.0,
+      "step": 16790
+    },
+    {
+      "epoch": 0.1540295223251123,
+      "learning_rate": 0.00016919592921976713,
+      "loss": 0.9133,
+      "mean_token_accuracy": 0.7256821393966675,
+      "num_tokens": 38864395.0,
+      "step": 16800
+    },
+    {
+      "epoch": 0.15412120656459155,
+      "learning_rate": 0.00016917759237187126,
+      "loss": 0.8646,
+      "mean_token_accuracy": 0.7446182489395141,
+      "num_tokens": 38887982.0,
+      "step": 16810
+    },
+    {
+      "epoch": 0.15421289080407077,
+      "learning_rate": 0.00016915925552397546,
+      "loss": 0.8449,
+      "mean_token_accuracy": 0.7444073379039764,
+      "num_tokens": 38911351.0,
+      "step": 16820
+    },
+    {
+      "epoch": 0.15430457504355002,
+      "learning_rate": 0.0001691409186760796,
+      "loss": 0.8894,
+      "mean_token_accuracy": 0.7392773270606995,
+      "num_tokens": 38934444.0,
+      "step": 16830
+    },
+    {
+      "epoch": 0.15439625928302925,
+      "learning_rate": 0.00016912258182818373,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7407513499259949,
+      "num_tokens": 38957685.0,
+      "step": 16840
+    },
+    {
+      "epoch": 0.15448794352250847,
+      "learning_rate": 0.0001691042449802879,
+      "loss": 0.8771,
+      "mean_token_accuracy": 0.7445762813091278,
+      "num_tokens": 38980926.0,
+      "step": 16850
+    },
+    {
+      "epoch": 0.15457962776198772,
+      "learning_rate": 0.00016908590813239204,
+      "loss": 0.8764,
+      "mean_token_accuracy": 0.740376353263855,
+      "num_tokens": 39003981.0,
+      "step": 16860
+    },
+    {
+      "epoch": 0.15467131200146694,
+      "learning_rate": 0.0001690675712844962,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7444840729236603,
+      "num_tokens": 39026544.0,
+      "step": 16870
+    },
+    {
+      "epoch": 0.1547629962409462,
+      "learning_rate": 0.00016904923443660034,
+      "loss": 0.8746,
+      "mean_token_accuracy": 0.7461060881614685,
+      "num_tokens": 39050280.0,
+      "step": 16880
+    },
+    {
+      "epoch": 0.15485468048042542,
+      "learning_rate": 0.0001690308975887045,
+      "loss": 0.9001,
+      "mean_token_accuracy": 0.7330372750759124,
+      "num_tokens": 39073130.0,
+      "step": 16890
+    },
+    {
+      "epoch": 0.15494636471990464,
+      "learning_rate": 0.00016901256074080867,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7405720889568329,
+      "num_tokens": 39096834.0,
+      "step": 16900
+    },
+    {
+      "epoch": 0.1550380489593839,
+      "learning_rate": 0.0001689942238929128,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7495450854301453,
+      "num_tokens": 39119212.0,
+      "step": 16910
+    },
+    {
+      "epoch": 0.1551297331988631,
+      "learning_rate": 0.00016897588704501698,
+      "loss": 0.8744,
+      "mean_token_accuracy": 0.7428697288036347,
+      "num_tokens": 39142979.0,
+      "step": 16920
+    },
+    {
+      "epoch": 0.15522141743834236,
+      "learning_rate": 0.00016895755019712111,
+      "loss": 0.9012,
+      "mean_token_accuracy": 0.7362680971622467,
+      "num_tokens": 39165585.0,
+      "step": 16930
+    },
+    {
+      "epoch": 0.15531310167782159,
+      "learning_rate": 0.00016893921334922528,
+      "loss": 0.8311,
+      "mean_token_accuracy": 0.7509696185588837,
+      "num_tokens": 39188522.0,
+      "step": 16940
+    },
+    {
+      "epoch": 0.1554047859173008,
+      "learning_rate": 0.00016892087650132945,
+      "loss": 0.9017,
+      "mean_token_accuracy": 0.7352629005908966,
+      "num_tokens": 39211450.0,
+      "step": 16950
+    },
+    {
+      "epoch": 0.15549647015678006,
+      "learning_rate": 0.00016890253965343358,
+      "loss": 0.8318,
+      "mean_token_accuracy": 0.7483528077602386,
+      "num_tokens": 39234302.0,
+      "step": 16960
+    },
+    {
+      "epoch": 0.15558815439625928,
+      "learning_rate": 0.00016888420280553775,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7395950853824615,
+      "num_tokens": 39257127.0,
+      "step": 16970
+    },
+    {
+      "epoch": 0.1556798386357385,
+      "learning_rate": 0.0001688658659576419,
+      "loss": 0.8363,
+      "mean_token_accuracy": 0.7496638536453247,
+      "num_tokens": 39279972.0,
+      "step": 16980
+    },
+    {
+      "epoch": 0.15577152287521775,
+      "learning_rate": 0.00016884752910974603,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7437224447727203,
+      "num_tokens": 39304175.0,
+      "step": 16990
+    },
+    {
+      "epoch": 0.15586320711469698,
+      "learning_rate": 0.0001688291922618502,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.7452760875225067,
+      "num_tokens": 39328003.0,
+      "step": 17000
+    },
+    {
+      "epoch": 0.15595489135417623,
+      "learning_rate": 0.00016881085541395433,
+      "loss": 0.8996,
+      "mean_token_accuracy": 0.7387515842914582,
+      "num_tokens": 39350914.0,
+      "step": 17010
+    },
+    {
+      "epoch": 0.15604657559365545,
+      "learning_rate": 0.00016879251856605852,
+      "loss": 0.8378,
+      "mean_token_accuracy": 0.7472789049148559,
+      "num_tokens": 39373506.0,
+      "step": 17020
+    },
+    {
+      "epoch": 0.15613825983313467,
+      "learning_rate": 0.00016877418171816266,
+      "loss": 0.8911,
+      "mean_token_accuracy": 0.7346392273902893,
+      "num_tokens": 39396798.0,
+      "step": 17030
+    },
+    {
+      "epoch": 0.15622994407261392,
+      "learning_rate": 0.0001687558448702668,
+      "loss": 0.8814,
+      "mean_token_accuracy": 0.7421307742595673,
+      "num_tokens": 39420214.0,
+      "step": 17040
+    },
+    {
+      "epoch": 0.15632162831209315,
+      "learning_rate": 0.00016873750802237096,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7466699361801148,
+      "num_tokens": 39443028.0,
+      "step": 17050
+    },
+    {
+      "epoch": 0.1564133125515724,
+      "learning_rate": 0.0001687191711744751,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7446559250354767,
+      "num_tokens": 39466517.0,
+      "step": 17060
+    },
+    {
+      "epoch": 0.15650499679105162,
+      "learning_rate": 0.00016870083432657927,
+      "loss": 0.9331,
+      "mean_token_accuracy": 0.7287180542945861,
+      "num_tokens": 39490932.0,
+      "step": 17070
+    },
+    {
+      "epoch": 0.15659668103053084,
+      "learning_rate": 0.00016868249747868343,
+      "loss": 0.8499,
+      "mean_token_accuracy": 0.7485387921333313,
+      "num_tokens": 39513849.0,
+      "step": 17080
+    },
+    {
+      "epoch": 0.1566883652700101,
+      "learning_rate": 0.00016866416063078757,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.7490086317062378,
+      "num_tokens": 39538317.0,
+      "step": 17090
+    },
+    {
+      "epoch": 0.15678004950948932,
+      "learning_rate": 0.00016864582378289174,
+      "loss": 0.8786,
+      "mean_token_accuracy": 0.7363821864128113,
+      "num_tokens": 39561298.0,
+      "step": 17100
+    },
+    {
+      "epoch": 0.15687173374896854,
+      "learning_rate": 0.00016862748693499588,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7386886358261109,
+      "num_tokens": 39583735.0,
+      "step": 17110
+    },
+    {
+      "epoch": 0.1569634179884478,
+      "learning_rate": 0.00016860915008710004,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7353485643863678,
+      "num_tokens": 39606533.0,
+      "step": 17120
+    },
+    {
+      "epoch": 0.157055102227927,
+      "learning_rate": 0.00016859081323920418,
+      "loss": 0.9284,
+      "mean_token_accuracy": 0.7309687435626984,
+      "num_tokens": 39629313.0,
+      "step": 17130
+    },
+    {
+      "epoch": 0.15714678646740626,
+      "learning_rate": 0.00016857247639130834,
+      "loss": 0.8536,
+      "mean_token_accuracy": 0.7418226718902587,
+      "num_tokens": 39651940.0,
+      "step": 17140
+    },
+    {
+      "epoch": 0.15723847070688549,
+      "learning_rate": 0.0001685541395434125,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7322404265403748,
+      "num_tokens": 39674810.0,
+      "step": 17150
+    },
+    {
+      "epoch": 0.1573301549463647,
+      "learning_rate": 0.00016853580269551665,
+      "loss": 0.9,
+      "mean_token_accuracy": 0.7297082722187043,
+      "num_tokens": 39698837.0,
+      "step": 17160
+    },
+    {
+      "epoch": 0.15742183918584396,
+      "learning_rate": 0.0001685174658476208,
+      "loss": 0.8814,
+      "mean_token_accuracy": 0.7342394649982452,
+      "num_tokens": 39721751.0,
+      "step": 17170
+    },
+    {
+      "epoch": 0.15751352342532318,
+      "learning_rate": 0.00016849912899972495,
+      "loss": 0.8597,
+      "mean_token_accuracy": 0.750165456533432,
+      "num_tokens": 39744338.0,
+      "step": 17180
+    },
+    {
+      "epoch": 0.15760520766480243,
+      "learning_rate": 0.0001684807921518291,
+      "loss": 0.8828,
+      "mean_token_accuracy": 0.736766242980957,
+      "num_tokens": 39768072.0,
+      "step": 17190
+    },
+    {
+      "epoch": 0.15769689190428166,
+      "learning_rate": 0.00016846245530393326,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7397107303142547,
+      "num_tokens": 39791075.0,
+      "step": 17200
+    },
+    {
+      "epoch": 0.15778857614376088,
+      "learning_rate": 0.00016844411845603742,
+      "loss": 0.8736,
+      "mean_token_accuracy": 0.7415744304656983,
+      "num_tokens": 39814033.0,
+      "step": 17210
+    },
+    {
+      "epoch": 0.15788026038324013,
+      "learning_rate": 0.00016842578160814159,
+      "loss": 0.8237,
+      "mean_token_accuracy": 0.7556914150714874,
+      "num_tokens": 39838059.0,
+      "step": 17220
+    },
+    {
+      "epoch": 0.15797194462271935,
+      "learning_rate": 0.00016840744476024572,
+      "loss": 0.8579,
+      "mean_token_accuracy": 0.7471156656742096,
+      "num_tokens": 39861463.0,
+      "step": 17230
+    },
+    {
+      "epoch": 0.1580636288621986,
+      "learning_rate": 0.00016838910791234986,
+      "loss": 0.8737,
+      "mean_token_accuracy": 0.741560983657837,
+      "num_tokens": 39884397.0,
+      "step": 17240
+    },
+    {
+      "epoch": 0.15815531310167782,
+      "learning_rate": 0.00016837077106445403,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.738440752029419,
+      "num_tokens": 39907994.0,
+      "step": 17250
+    },
+    {
+      "epoch": 0.15824699734115705,
+      "learning_rate": 0.00016835243421655817,
+      "loss": 0.8741,
+      "mean_token_accuracy": 0.7396030426025391,
+      "num_tokens": 39930416.0,
+      "step": 17260
+    },
+    {
+      "epoch": 0.1583386815806363,
+      "learning_rate": 0.00016833409736866233,
+      "loss": 0.897,
+      "mean_token_accuracy": 0.7328746974468231,
+      "num_tokens": 39953552.0,
+      "step": 17270
+    },
+    {
+      "epoch": 0.15843036582011552,
+      "learning_rate": 0.0001683157605207665,
+      "loss": 0.8798,
+      "mean_token_accuracy": 0.7370184361934662,
+      "num_tokens": 39976687.0,
+      "step": 17280
+    },
+    {
+      "epoch": 0.15852205005959474,
+      "learning_rate": 0.00016829742367287064,
+      "loss": 0.8398,
+      "mean_token_accuracy": 0.7485899269580841,
+      "num_tokens": 40000792.0,
+      "step": 17290
+    },
+    {
+      "epoch": 0.158613734299074,
+      "learning_rate": 0.0001682790868249748,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7397284567356109,
+      "num_tokens": 40023544.0,
+      "step": 17300
+    },
+    {
+      "epoch": 0.15870541853855322,
+      "learning_rate": 0.00016826074997707894,
+      "loss": 0.8769,
+      "mean_token_accuracy": 0.7368694961071014,
+      "num_tokens": 40046236.0,
+      "step": 17310
+    },
+    {
+      "epoch": 0.15879710277803247,
+      "learning_rate": 0.0001682424131291831,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7377434968948364,
+      "num_tokens": 40069467.0,
+      "step": 17320
+    },
+    {
+      "epoch": 0.1588887870175117,
+      "learning_rate": 0.00016822407628128724,
+      "loss": 0.9228,
+      "mean_token_accuracy": 0.733688759803772,
+      "num_tokens": 40092677.0,
+      "step": 17330
+    },
+    {
+      "epoch": 0.1589804712569909,
+      "learning_rate": 0.0001682057394333914,
+      "loss": 0.8815,
+      "mean_token_accuracy": 0.7415893852710724,
+      "num_tokens": 40115888.0,
+      "step": 17340
+    },
+    {
+      "epoch": 0.15907215549647016,
+      "learning_rate": 0.00016818740258549557,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7410473346710205,
+      "num_tokens": 40138364.0,
+      "step": 17350
+    },
+    {
+      "epoch": 0.1591638397359494,
+      "learning_rate": 0.0001681690657375997,
+      "loss": 0.8853,
+      "mean_token_accuracy": 0.7327299773693084,
+      "num_tokens": 40161415.0,
+      "step": 17360
+    },
+    {
+      "epoch": 0.15925552397542864,
+      "learning_rate": 0.00016815072888970388,
+      "loss": 0.9074,
+      "mean_token_accuracy": 0.7273565292358398,
+      "num_tokens": 40184518.0,
+      "step": 17370
+    },
+    {
+      "epoch": 0.15934720821490786,
+      "learning_rate": 0.00016813239204180802,
+      "loss": 0.8358,
+      "mean_token_accuracy": 0.748268336057663,
+      "num_tokens": 40208537.0,
+      "step": 17380
+    },
+    {
+      "epoch": 0.15943889245438708,
+      "learning_rate": 0.00016811405519391215,
+      "loss": 0.8466,
+      "mean_token_accuracy": 0.7444461345672607,
+      "num_tokens": 40231126.0,
+      "step": 17390
+    },
+    {
+      "epoch": 0.15953057669386633,
+      "learning_rate": 0.00016809571834601632,
+      "loss": 0.8726,
+      "mean_token_accuracy": 0.7356184244155883,
+      "num_tokens": 40253514.0,
+      "step": 17400
+    },
+    {
+      "epoch": 0.15962226093334556,
+      "learning_rate": 0.00016807738149812049,
+      "loss": 0.8935,
+      "mean_token_accuracy": 0.7391709446907043,
+      "num_tokens": 40276077.0,
+      "step": 17410
+    },
+    {
+      "epoch": 0.15971394517282478,
+      "learning_rate": 0.00016805904465022465,
+      "loss": 0.8631,
+      "mean_token_accuracy": 0.7399099051952363,
+      "num_tokens": 40299076.0,
+      "step": 17420
+    },
+    {
+      "epoch": 0.15980562941230403,
+      "learning_rate": 0.0001680407078023288,
+      "loss": 0.8573,
+      "mean_token_accuracy": 0.7385023057460784,
+      "num_tokens": 40321935.0,
+      "step": 17430
+    },
+    {
+      "epoch": 0.15989731365178325,
+      "learning_rate": 0.00016802237095443293,
+      "loss": 0.8413,
+      "mean_token_accuracy": 0.7480131924152374,
+      "num_tokens": 40344781.0,
+      "step": 17440
+    },
+    {
+      "epoch": 0.1599889978912625,
+      "learning_rate": 0.0001680040341065371,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7477836430072784,
+      "num_tokens": 40368409.0,
+      "step": 17450
+    },
+    {
+      "epoch": 0.16008068213074172,
+      "learning_rate": 0.00016798569725864123,
+      "loss": 0.8624,
+      "mean_token_accuracy": 0.7396600663661956,
+      "num_tokens": 40391838.0,
+      "step": 17460
+    },
+    {
+      "epoch": 0.16017236637022095,
+      "learning_rate": 0.00016796736041074542,
+      "loss": 0.8757,
+      "mean_token_accuracy": 0.744785076379776,
+      "num_tokens": 40414894.0,
+      "step": 17470
+    },
+    {
+      "epoch": 0.1602640506097002,
+      "learning_rate": 0.00016794902356284956,
+      "loss": 0.8487,
+      "mean_token_accuracy": 0.74610915184021,
+      "num_tokens": 40437014.0,
+      "step": 17480
+    },
+    {
+      "epoch": 0.16035573484917942,
+      "learning_rate": 0.0001679306867149537,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7441764593124389,
+      "num_tokens": 40459884.0,
+      "step": 17490
+    },
+    {
+      "epoch": 0.16044741908865867,
+      "learning_rate": 0.00016791234986705787,
+      "loss": 0.8633,
+      "mean_token_accuracy": 0.7488744735717774,
+      "num_tokens": 40483174.0,
+      "step": 17500
+    },
+    {
+      "epoch": 0.1605391033281379,
+      "learning_rate": 0.000167894013019162,
+      "loss": 0.8736,
+      "mean_token_accuracy": 0.7425401151180268,
+      "num_tokens": 40505775.0,
+      "step": 17510
+    },
+    {
+      "epoch": 0.16063078756761712,
+      "learning_rate": 0.00016787567617126617,
+      "loss": 0.8412,
+      "mean_token_accuracy": 0.7526976048946381,
+      "num_tokens": 40528854.0,
+      "step": 17520
+    },
+    {
+      "epoch": 0.16072247180709637,
+      "learning_rate": 0.0001678573393233703,
+      "loss": 0.8519,
+      "mean_token_accuracy": 0.7444073617458343,
+      "num_tokens": 40552442.0,
+      "step": 17530
+    },
+    {
+      "epoch": 0.1608141560465756,
+      "learning_rate": 0.00016783900247547447,
+      "loss": 0.8215,
+      "mean_token_accuracy": 0.7446524620056152,
+      "num_tokens": 40575262.0,
+      "step": 17540
+    },
+    {
+      "epoch": 0.16090584028605484,
+      "learning_rate": 0.00016782066562757864,
+      "loss": 0.9047,
+      "mean_token_accuracy": 0.7284642696380615,
+      "num_tokens": 40598247.0,
+      "step": 17550
+    },
+    {
+      "epoch": 0.16099752452553406,
+      "learning_rate": 0.00016780232877968278,
+      "loss": 0.8789,
+      "mean_token_accuracy": 0.7382528483867645,
+      "num_tokens": 40621385.0,
+      "step": 17560
+    },
+    {
+      "epoch": 0.1610892087650133,
+      "learning_rate": 0.00016778399193178694,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7507353901863099,
+      "num_tokens": 40644241.0,
+      "step": 17570
+    },
+    {
+      "epoch": 0.16118089300449254,
+      "learning_rate": 0.00016776565508389108,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7480726540088654,
+      "num_tokens": 40667355.0,
+      "step": 17580
+    },
+    {
+      "epoch": 0.16127257724397176,
+      "learning_rate": 0.00016774731823599522,
+      "loss": 0.9176,
+      "mean_token_accuracy": 0.7309023559093475,
+      "num_tokens": 40690723.0,
+      "step": 17590
+    },
+    {
+      "epoch": 0.16136426148345098,
+      "learning_rate": 0.0001677289813880994,
+      "loss": 0.8678,
+      "mean_token_accuracy": 0.7350658476352692,
+      "num_tokens": 40713685.0,
+      "step": 17600
+    },
+    {
+      "epoch": 0.16145594572293023,
+      "learning_rate": 0.00016771064454020355,
+      "loss": 0.883,
+      "mean_token_accuracy": 0.7347937643527984,
+      "num_tokens": 40737807.0,
+      "step": 17610
+    },
+    {
+      "epoch": 0.16154762996240946,
+      "learning_rate": 0.00016769230769230772,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7383953213691712,
+      "num_tokens": 40760836.0,
+      "step": 17620
+    },
+    {
+      "epoch": 0.1616393142018887,
+      "learning_rate": 0.00016767397084441185,
+      "loss": 0.8989,
+      "mean_token_accuracy": 0.7323030471801758,
+      "num_tokens": 40783731.0,
+      "step": 17630
+    },
+    {
+      "epoch": 0.16173099844136793,
+      "learning_rate": 0.000167655633996516,
+      "loss": 0.8462,
+      "mean_token_accuracy": 0.7501704514026641,
+      "num_tokens": 40806410.0,
+      "step": 17640
+    },
+    {
+      "epoch": 0.16182268268084715,
+      "learning_rate": 0.00016763729714862016,
+      "loss": 0.8738,
+      "mean_token_accuracy": 0.7390379190444947,
+      "num_tokens": 40830395.0,
+      "step": 17650
+    },
+    {
+      "epoch": 0.1619143669203264,
+      "learning_rate": 0.0001676189603007243,
+      "loss": 0.8263,
+      "mean_token_accuracy": 0.7556713819503784,
+      "num_tokens": 40853533.0,
+      "step": 17660
+    },
+    {
+      "epoch": 0.16200605115980563,
+      "learning_rate": 0.0001676006234528285,
+      "loss": 0.8509,
+      "mean_token_accuracy": 0.7434868693351746,
+      "num_tokens": 40877452.0,
+      "step": 17670
+    },
+    {
+      "epoch": 0.16209773539928488,
+      "learning_rate": 0.00016758228660493263,
+      "loss": 0.8279,
+      "mean_token_accuracy": 0.7525975346565247,
+      "num_tokens": 40900679.0,
+      "step": 17680
+    },
+    {
+      "epoch": 0.1621894196387641,
+      "learning_rate": 0.00016756394975703676,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.7483399212360382,
+      "num_tokens": 40923667.0,
+      "step": 17690
+    },
+    {
+      "epoch": 0.16228110387824332,
+      "learning_rate": 0.00016754561290914093,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7391563951969147,
+      "num_tokens": 40946901.0,
+      "step": 17700
+    },
+    {
+      "epoch": 0.16237278811772257,
+      "learning_rate": 0.00016752727606124507,
+      "loss": 0.8968,
+      "mean_token_accuracy": 0.7328025579452515,
+      "num_tokens": 40970559.0,
+      "step": 17710
+    },
+    {
+      "epoch": 0.1624644723572018,
+      "learning_rate": 0.00016750893921334923,
+      "loss": 0.8767,
+      "mean_token_accuracy": 0.7393238782882691,
+      "num_tokens": 40994083.0,
+      "step": 17720
+    },
+    {
+      "epoch": 0.16255615659668102,
+      "learning_rate": 0.0001674906023654534,
+      "loss": 0.9021,
+      "mean_token_accuracy": 0.7304047882556916,
+      "num_tokens": 41017443.0,
+      "step": 17730
+    },
+    {
+      "epoch": 0.16264784083616027,
+      "learning_rate": 0.00016747226551755754,
+      "loss": 0.8565,
+      "mean_token_accuracy": 0.7497427701950073,
+      "num_tokens": 41040736.0,
+      "step": 17740
+    },
+    {
+      "epoch": 0.1627395250756395,
+      "learning_rate": 0.0001674539286696617,
+      "loss": 0.8527,
+      "mean_token_accuracy": 0.7382766842842102,
+      "num_tokens": 41063711.0,
+      "step": 17750
+    },
+    {
+      "epoch": 0.16283120931511874,
+      "learning_rate": 0.00016743559182176584,
+      "loss": 0.8953,
+      "mean_token_accuracy": 0.7397700190544129,
+      "num_tokens": 41086464.0,
+      "step": 17760
+    },
+    {
+      "epoch": 0.16292289355459796,
+      "learning_rate": 0.00016741725497387,
+      "loss": 0.8306,
+      "mean_token_accuracy": 0.7546618223190308,
+      "num_tokens": 41110300.0,
+      "step": 17770
+    },
+    {
+      "epoch": 0.1630145777940772,
+      "learning_rate": 0.00016739891812597415,
+      "loss": 0.9008,
+      "mean_token_accuracy": 0.7381853759288788,
+      "num_tokens": 41133380.0,
+      "step": 17780
+    },
+    {
+      "epoch": 0.16310626203355644,
+      "learning_rate": 0.00016738058127807828,
+      "loss": 0.8659,
+      "mean_token_accuracy": 0.7472920775413513,
+      "num_tokens": 41157213.0,
+      "step": 17790
+    },
+    {
+      "epoch": 0.16319794627303566,
+      "learning_rate": 0.00016736224443018248,
+      "loss": 0.8839,
+      "mean_token_accuracy": 0.7403927803039551,
+      "num_tokens": 41180107.0,
+      "step": 17800
+    },
+    {
+      "epoch": 0.1632896305125149,
+      "learning_rate": 0.00016734390758228661,
+      "loss": 0.8435,
+      "mean_token_accuracy": 0.7507363438606263,
+      "num_tokens": 41203089.0,
+      "step": 17810
+    },
+    {
+      "epoch": 0.16338131475199413,
+      "learning_rate": 0.00016732557073439078,
+      "loss": 0.8605,
+      "mean_token_accuracy": 0.7463716804981232,
+      "num_tokens": 41226464.0,
+      "step": 17820
+    },
+    {
+      "epoch": 0.16347299899147336,
+      "learning_rate": 0.00016730723388649492,
+      "loss": 0.8599,
+      "mean_token_accuracy": 0.7399094104766846,
+      "num_tokens": 41249910.0,
+      "step": 17830
+    },
+    {
+      "epoch": 0.1635646832309526,
+      "learning_rate": 0.00016728889703859906,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.744236272573471,
+      "num_tokens": 41273470.0,
+      "step": 17840
+    },
+    {
+      "epoch": 0.16365636747043183,
+      "learning_rate": 0.00016727056019070322,
+      "loss": 0.88,
+      "mean_token_accuracy": 0.7369423568248749,
+      "num_tokens": 41297073.0,
+      "step": 17850
+    },
+    {
+      "epoch": 0.16374805170991108,
+      "learning_rate": 0.0001672522233428074,
+      "loss": 0.8834,
+      "mean_token_accuracy": 0.7375329375267029,
+      "num_tokens": 41319820.0,
+      "step": 17860
+    },
+    {
+      "epoch": 0.1638397359493903,
+      "learning_rate": 0.00016723388649491155,
+      "loss": 0.8611,
+      "mean_token_accuracy": 0.7408622324466705,
+      "num_tokens": 41343302.0,
+      "step": 17870
+    },
+    {
+      "epoch": 0.16393142018886953,
+      "learning_rate": 0.0001672155496470157,
+      "loss": 0.8906,
+      "mean_token_accuracy": 0.7352072417736053,
+      "num_tokens": 41366200.0,
+      "step": 17880
+    },
+    {
+      "epoch": 0.16402310442834878,
+      "learning_rate": 0.00016719721279911983,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.7434074938297272,
+      "num_tokens": 41389922.0,
+      "step": 17890
+    },
+    {
+      "epoch": 0.164114788667828,
+      "learning_rate": 0.000167178875951224,
+      "loss": 0.872,
+      "mean_token_accuracy": 0.7438861548900604,
+      "num_tokens": 41412672.0,
+      "step": 17900
+    },
+    {
+      "epoch": 0.16420647290730722,
+      "learning_rate": 0.00016716053910332813,
+      "loss": 0.8794,
+      "mean_token_accuracy": 0.7390958905220032,
+      "num_tokens": 41435521.0,
+      "step": 17910
+    },
+    {
+      "epoch": 0.16429815714678647,
+      "learning_rate": 0.0001671422022554323,
+      "loss": 0.8893,
+      "mean_token_accuracy": 0.7348230957984925,
+      "num_tokens": 41458368.0,
+      "step": 17920
+    },
+    {
+      "epoch": 0.1643898413862657,
+      "learning_rate": 0.00016712386540753646,
+      "loss": 0.888,
+      "mean_token_accuracy": 0.7326744616031646,
+      "num_tokens": 41481181.0,
+      "step": 17930
+    },
+    {
+      "epoch": 0.16448152562574495,
+      "learning_rate": 0.0001671055285596406,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7451447486877442,
+      "num_tokens": 41504551.0,
+      "step": 17940
+    },
+    {
+      "epoch": 0.16457320986522417,
+      "learning_rate": 0.00016708719171174477,
+      "loss": 0.8908,
+      "mean_token_accuracy": 0.7369565606117249,
+      "num_tokens": 41527906.0,
+      "step": 17950
+    },
+    {
+      "epoch": 0.1646648941047034,
+      "learning_rate": 0.0001670688548638489,
+      "loss": 0.8823,
+      "mean_token_accuracy": 0.7395092904567718,
+      "num_tokens": 41550696.0,
+      "step": 17960
+    },
+    {
+      "epoch": 0.16475657834418264,
+      "learning_rate": 0.00016705051801595307,
+      "loss": 0.8643,
+      "mean_token_accuracy": 0.7378550350666047,
+      "num_tokens": 41573912.0,
+      "step": 17970
+    },
+    {
+      "epoch": 0.16484826258366186,
+      "learning_rate": 0.0001670321811680572,
+      "loss": 0.8576,
+      "mean_token_accuracy": 0.7506725192070007,
+      "num_tokens": 41596736.0,
+      "step": 17980
+    },
+    {
+      "epoch": 0.16493994682314111,
+      "learning_rate": 0.00016701384432016135,
+      "loss": 0.8617,
+      "mean_token_accuracy": 0.745305860042572,
+      "num_tokens": 41620143.0,
+      "step": 17990
+    },
+    {
+      "epoch": 0.16503163106262034,
+      "learning_rate": 0.00016699550747226554,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7396979570388794,
+      "num_tokens": 41642888.0,
+      "step": 18000
+    },
+    {
+      "epoch": 0.16512331530209956,
+      "learning_rate": 0.00016697717062436968,
+      "loss": 0.8763,
+      "mean_token_accuracy": 0.7461963713169097,
+      "num_tokens": 41666072.0,
+      "step": 18010
+    },
+    {
+      "epoch": 0.1652149995415788,
+      "learning_rate": 0.00016695883377647384,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.7509789228439331,
+      "num_tokens": 41689047.0,
+      "step": 18020
+    },
+    {
+      "epoch": 0.16530668378105803,
+      "learning_rate": 0.00016694049692857798,
+      "loss": 0.8528,
+      "mean_token_accuracy": 0.7466614663600921,
+      "num_tokens": 41711812.0,
+      "step": 18030
+    },
+    {
+      "epoch": 0.16539836802053726,
+      "learning_rate": 0.00016692216008068212,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7391000628471375,
+      "num_tokens": 41734735.0,
+      "step": 18040
+    },
+    {
+      "epoch": 0.1654900522600165,
+      "learning_rate": 0.00016690382323278629,
+      "loss": 0.8296,
+      "mean_token_accuracy": 0.7464460790157318,
+      "num_tokens": 41758220.0,
+      "step": 18050
+    },
+    {
+      "epoch": 0.16558173649949573,
+      "learning_rate": 0.00016688548638489045,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7436749279499054,
+      "num_tokens": 41780829.0,
+      "step": 18060
+    },
+    {
+      "epoch": 0.16567342073897498,
+      "learning_rate": 0.00016686714953699462,
+      "loss": 0.9127,
+      "mean_token_accuracy": 0.7377270340919495,
+      "num_tokens": 41804318.0,
+      "step": 18070
+    },
+    {
+      "epoch": 0.1657651049784542,
+      "learning_rate": 0.00016684881268909876,
+      "loss": 0.8736,
+      "mean_token_accuracy": 0.7382775187492371,
+      "num_tokens": 41827759.0,
+      "step": 18080
+    },
+    {
+      "epoch": 0.16585678921793343,
+      "learning_rate": 0.0001668304758412029,
+      "loss": 0.8922,
+      "mean_token_accuracy": 0.73809455037117,
+      "num_tokens": 41851849.0,
+      "step": 18090
+    },
+    {
+      "epoch": 0.16594847345741268,
+      "learning_rate": 0.00016681213899330706,
+      "loss": 0.8676,
+      "mean_token_accuracy": 0.7345297038555145,
+      "num_tokens": 41875334.0,
+      "step": 18100
+    },
+    {
+      "epoch": 0.1660401576968919,
+      "learning_rate": 0.0001667938021454112,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7380764663219452,
+      "num_tokens": 41898370.0,
+      "step": 18110
+    },
+    {
+      "epoch": 0.16613184193637115,
+      "learning_rate": 0.00016677546529751536,
+      "loss": 0.8565,
+      "mean_token_accuracy": 0.7484489917755127,
+      "num_tokens": 41920790.0,
+      "step": 18120
+    },
+    {
+      "epoch": 0.16622352617585037,
+      "learning_rate": 0.00016675712844961953,
+      "loss": 0.8562,
+      "mean_token_accuracy": 0.7416860640048981,
+      "num_tokens": 41943929.0,
+      "step": 18130
+    },
+    {
+      "epoch": 0.1663152104153296,
+      "learning_rate": 0.00016673879160172367,
+      "loss": 0.8774,
+      "mean_token_accuracy": 0.7420123457908631,
+      "num_tokens": 41967118.0,
+      "step": 18140
+    },
+    {
+      "epoch": 0.16640689465480885,
+      "learning_rate": 0.00016672045475382783,
+      "loss": 0.8264,
+      "mean_token_accuracy": 0.7512724161148071,
+      "num_tokens": 41990403.0,
+      "step": 18150
+    },
+    {
+      "epoch": 0.16649857889428807,
+      "learning_rate": 0.00016670211790593197,
+      "loss": 0.8437,
+      "mean_token_accuracy": 0.7459653854370117,
+      "num_tokens": 42013721.0,
+      "step": 18160
+    },
+    {
+      "epoch": 0.16659026313376732,
+      "learning_rate": 0.00016668378105803614,
+      "loss": 0.8881,
+      "mean_token_accuracy": 0.7412309527397156,
+      "num_tokens": 42037341.0,
+      "step": 18170
+    },
+    {
+      "epoch": 0.16668194737324654,
+      "learning_rate": 0.00016666544421014027,
+      "loss": 0.8493,
+      "mean_token_accuracy": 0.7402162849903107,
+      "num_tokens": 42060495.0,
+      "step": 18180
+    },
+    {
+      "epoch": 0.16677363161272576,
+      "learning_rate": 0.00016664710736224444,
+      "loss": 0.8624,
+      "mean_token_accuracy": 0.7407247722148895,
+      "num_tokens": 42084480.0,
+      "step": 18190
+    },
+    {
+      "epoch": 0.16686531585220502,
+      "learning_rate": 0.0001666287705143486,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.7436165332794189,
+      "num_tokens": 42107356.0,
+      "step": 18200
+    },
+    {
+      "epoch": 0.16695700009168424,
+      "learning_rate": 0.00016661043366645274,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7454732954502106,
+      "num_tokens": 42131261.0,
+      "step": 18210
+    },
+    {
+      "epoch": 0.16704868433116346,
+      "learning_rate": 0.0001665920968185569,
+      "loss": 0.8923,
+      "mean_token_accuracy": 0.7352973461151123,
+      "num_tokens": 42154529.0,
+      "step": 18220
+    },
+    {
+      "epoch": 0.1671403685706427,
+      "learning_rate": 0.00016657375997066105,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7395649433135987,
+      "num_tokens": 42177293.0,
+      "step": 18230
+    },
+    {
+      "epoch": 0.16723205281012193,
+      "learning_rate": 0.00016655542312276519,
+      "loss": 0.8969,
+      "mean_token_accuracy": 0.7318514168262482,
+      "num_tokens": 42201043.0,
+      "step": 18240
+    },
+    {
+      "epoch": 0.16732373704960118,
+      "learning_rate": 0.00016653708627486935,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7404728889465332,
+      "num_tokens": 42224080.0,
+      "step": 18250
+    },
+    {
+      "epoch": 0.1674154212890804,
+      "learning_rate": 0.00016651874942697352,
+      "loss": 0.901,
+      "mean_token_accuracy": 0.7297580242156982,
+      "num_tokens": 42246844.0,
+      "step": 18260
+    },
+    {
+      "epoch": 0.16750710552855963,
+      "learning_rate": 0.00016650041257907768,
+      "loss": 0.886,
+      "mean_token_accuracy": 0.737334668636322,
+      "num_tokens": 42270168.0,
+      "step": 18270
+    },
+    {
+      "epoch": 0.16759878976803888,
+      "learning_rate": 0.00016648207573118182,
+      "loss": 0.8799,
+      "mean_token_accuracy": 0.745305997133255,
+      "num_tokens": 42293055.0,
+      "step": 18280
+    },
+    {
+      "epoch": 0.1676904740075181,
+      "learning_rate": 0.00016646373888328596,
+      "loss": 0.8566,
+      "mean_token_accuracy": 0.7517912030220032,
+      "num_tokens": 42317077.0,
+      "step": 18290
+    },
+    {
+      "epoch": 0.16778215824699735,
+      "learning_rate": 0.00016644540203539012,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.739583021402359,
+      "num_tokens": 42340461.0,
+      "step": 18300
+    },
+    {
+      "epoch": 0.16787384248647658,
+      "learning_rate": 0.00016642706518749426,
+      "loss": 0.8967,
+      "mean_token_accuracy": 0.7339466094970704,
+      "num_tokens": 42363235.0,
+      "step": 18310
+    },
+    {
+      "epoch": 0.1679655267259558,
+      "learning_rate": 0.00016640872833959845,
+      "loss": 0.8746,
+      "mean_token_accuracy": 0.7352726817131042,
+      "num_tokens": 42385965.0,
+      "step": 18320
+    },
+    {
+      "epoch": 0.16805721096543505,
+      "learning_rate": 0.0001663903914917026,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.747681075334549,
+      "num_tokens": 42409057.0,
+      "step": 18330
+    },
+    {
+      "epoch": 0.16814889520491427,
+      "learning_rate": 0.00016637205464380673,
+      "loss": 0.8451,
+      "mean_token_accuracy": 0.7485225677490235,
+      "num_tokens": 42432307.0,
+      "step": 18340
+    },
+    {
+      "epoch": 0.1682405794443935,
+      "learning_rate": 0.0001663537177959109,
+      "loss": 0.883,
+      "mean_token_accuracy": 0.7471256136894227,
+      "num_tokens": 42455741.0,
+      "step": 18350
+    },
+    {
+      "epoch": 0.16833226368387275,
+      "learning_rate": 0.00016633538094801503,
+      "loss": 0.8828,
+      "mean_token_accuracy": 0.7374133229255676,
+      "num_tokens": 42479199.0,
+      "step": 18360
+    },
+    {
+      "epoch": 0.16842394792335197,
+      "learning_rate": 0.0001663170441001192,
+      "loss": 0.8592,
+      "mean_token_accuracy": 0.7457978844642639,
+      "num_tokens": 42502926.0,
+      "step": 18370
+    },
+    {
+      "epoch": 0.16851563216283122,
+      "learning_rate": 0.00016629870725222334,
+      "loss": 0.8568,
+      "mean_token_accuracy": 0.7415787398815155,
+      "num_tokens": 42525572.0,
+      "step": 18380
+    },
+    {
+      "epoch": 0.16860731640231044,
+      "learning_rate": 0.0001662803704043275,
+      "loss": 0.8406,
+      "mean_token_accuracy": 0.7437069296836853,
+      "num_tokens": 42548672.0,
+      "step": 18390
+    },
+    {
+      "epoch": 0.16869900064178966,
+      "learning_rate": 0.00016626203355643167,
+      "loss": 0.8605,
+      "mean_token_accuracy": 0.7456466615200043,
+      "num_tokens": 42571442.0,
+      "step": 18400
+    },
+    {
+      "epoch": 0.16879068488126892,
+      "learning_rate": 0.0001662436967085358,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.7428866684436798,
+      "num_tokens": 42595266.0,
+      "step": 18410
+    },
+    {
+      "epoch": 0.16888236912074814,
+      "learning_rate": 0.00016622535986063997,
+      "loss": 0.8631,
+      "mean_token_accuracy": 0.7364484310150147,
+      "num_tokens": 42618549.0,
+      "step": 18420
+    },
+    {
+      "epoch": 0.1689740533602274,
+      "learning_rate": 0.0001662070230127441,
+      "loss": 0.8511,
+      "mean_token_accuracy": 0.7413220882415772,
+      "num_tokens": 42641416.0,
+      "step": 18430
+    },
+    {
+      "epoch": 0.1690657375997066,
+      "learning_rate": 0.00016618868616484825,
+      "loss": 0.8551,
+      "mean_token_accuracy": 0.7441992104053498,
+      "num_tokens": 42664003.0,
+      "step": 18440
+    },
+    {
+      "epoch": 0.16915742183918583,
+      "learning_rate": 0.00016617034931695244,
+      "loss": 0.8622,
+      "mean_token_accuracy": 0.744612205028534,
+      "num_tokens": 42686954.0,
+      "step": 18450
+    },
+    {
+      "epoch": 0.16924910607866508,
+      "learning_rate": 0.00016615201246905658,
+      "loss": 0.8922,
+      "mean_token_accuracy": 0.7364725351333619,
+      "num_tokens": 42710534.0,
+      "step": 18460
+    },
+    {
+      "epoch": 0.1693407903181443,
+      "learning_rate": 0.00016613367562116075,
+      "loss": 0.8191,
+      "mean_token_accuracy": 0.7570809066295624,
+      "num_tokens": 42733186.0,
+      "step": 18470
+    },
+    {
+      "epoch": 0.16943247455762356,
+      "learning_rate": 0.00016611533877326488,
+      "loss": 0.8987,
+      "mean_token_accuracy": 0.7353633105754852,
+      "num_tokens": 42756712.0,
+      "step": 18480
+    },
+    {
+      "epoch": 0.16952415879710278,
+      "learning_rate": 0.00016609700192536902,
+      "loss": 0.8761,
+      "mean_token_accuracy": 0.734033203125,
+      "num_tokens": 42781296.0,
+      "step": 18490
+    },
+    {
+      "epoch": 0.169615843036582,
+      "learning_rate": 0.0001660786650774732,
+      "loss": 0.8536,
+      "mean_token_accuracy": 0.7415765881538391,
+      "num_tokens": 42804319.0,
+      "step": 18500
+    },
+    {
+      "epoch": 0.16970752727606125,
+      "learning_rate": 0.00016606032822957733,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7495999097824096,
+      "num_tokens": 42827149.0,
+      "step": 18510
+    },
+    {
+      "epoch": 0.16979921151554048,
+      "learning_rate": 0.00016604199138168152,
+      "loss": 0.8976,
+      "mean_token_accuracy": 0.7378803551197052,
+      "num_tokens": 42850539.0,
+      "step": 18520
+    },
+    {
+      "epoch": 0.1698908957550197,
+      "learning_rate": 0.00016602365453378566,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7473590672016144,
+      "num_tokens": 42873882.0,
+      "step": 18530
+    },
+    {
+      "epoch": 0.16998257999449895,
+      "learning_rate": 0.0001660053176858898,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7415569961071015,
+      "num_tokens": 42897100.0,
+      "step": 18540
+    },
+    {
+      "epoch": 0.17007426423397817,
+      "learning_rate": 0.00016598698083799396,
+      "loss": 0.8384,
+      "mean_token_accuracy": 0.7476524829864502,
+      "num_tokens": 42920015.0,
+      "step": 18550
+    },
+    {
+      "epoch": 0.17016594847345742,
+      "learning_rate": 0.0001659686439900981,
+      "loss": 0.8643,
+      "mean_token_accuracy": 0.7376882374286652,
+      "num_tokens": 42943162.0,
+      "step": 18560
+    },
+    {
+      "epoch": 0.17025763271293665,
+      "learning_rate": 0.00016595030714220226,
+      "loss": 0.8767,
+      "mean_token_accuracy": 0.7456963539123536,
+      "num_tokens": 42966756.0,
+      "step": 18570
+    },
+    {
+      "epoch": 0.17034931695241587,
+      "learning_rate": 0.00016593197029430643,
+      "loss": 0.8904,
+      "mean_token_accuracy": 0.745152348279953,
+      "num_tokens": 42989343.0,
+      "step": 18580
+    },
+    {
+      "epoch": 0.17044100119189512,
+      "learning_rate": 0.00016591363344641057,
+      "loss": 0.8785,
+      "mean_token_accuracy": 0.7373985052108765,
+      "num_tokens": 43012254.0,
+      "step": 18590
+    },
+    {
+      "epoch": 0.17053268543137434,
+      "learning_rate": 0.00016589529659851473,
+      "loss": 0.8575,
+      "mean_token_accuracy": 0.746576726436615,
+      "num_tokens": 43035654.0,
+      "step": 18600
+    },
+    {
+      "epoch": 0.1706243696708536,
+      "learning_rate": 0.00016587695975061887,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7436378777027131,
+      "num_tokens": 43058202.0,
+      "step": 18610
+    },
+    {
+      "epoch": 0.17071605391033282,
+      "learning_rate": 0.00016585862290272304,
+      "loss": 0.8553,
+      "mean_token_accuracy": 0.7430861413478851,
+      "num_tokens": 43081274.0,
+      "step": 18620
+    },
+    {
+      "epoch": 0.17080773814981204,
+      "learning_rate": 0.00016584028605482718,
+      "loss": 0.8635,
+      "mean_token_accuracy": 0.7420690774917602,
+      "num_tokens": 43104564.0,
+      "step": 18630
+    },
+    {
+      "epoch": 0.1708994223892913,
+      "learning_rate": 0.00016582194920693131,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7420691609382629,
+      "num_tokens": 43127185.0,
+      "step": 18640
+    },
+    {
+      "epoch": 0.1709911066287705,
+      "learning_rate": 0.0001658036123590355,
+      "loss": 0.8493,
+      "mean_token_accuracy": 0.7503941476345062,
+      "num_tokens": 43149937.0,
+      "step": 18650
+    },
+    {
+      "epoch": 0.17108279086824973,
+      "learning_rate": 0.00016578527551113964,
+      "loss": 0.8751,
+      "mean_token_accuracy": 0.7426930785179138,
+      "num_tokens": 43173374.0,
+      "step": 18660
+    },
+    {
+      "epoch": 0.17117447510772899,
+      "learning_rate": 0.0001657669386632438,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7485217332839966,
+      "num_tokens": 43196583.0,
+      "step": 18670
+    },
+    {
+      "epoch": 0.1712661593472082,
+      "learning_rate": 0.00016574860181534795,
+      "loss": 0.8932,
+      "mean_token_accuracy": 0.7311478614807129,
+      "num_tokens": 43219366.0,
+      "step": 18680
+    },
+    {
+      "epoch": 0.17135784358668746,
+      "learning_rate": 0.0001657302649674521,
+      "loss": 0.8403,
+      "mean_token_accuracy": 0.7468105971813201,
+      "num_tokens": 43242694.0,
+      "step": 18690
+    },
+    {
+      "epoch": 0.17144952782616668,
+      "learning_rate": 0.00016571192811955625,
+      "loss": 0.8767,
+      "mean_token_accuracy": 0.7375354588031768,
+      "num_tokens": 43265614.0,
+      "step": 18700
+    },
+    {
+      "epoch": 0.1715412120656459,
+      "learning_rate": 0.00016569359127166042,
+      "loss": 0.8935,
+      "mean_token_accuracy": 0.7387665510177612,
+      "num_tokens": 43289693.0,
+      "step": 18710
+    },
+    {
+      "epoch": 0.17163289630512515,
+      "learning_rate": 0.00016567525442376458,
+      "loss": 0.901,
+      "mean_token_accuracy": 0.7406913757324218,
+      "num_tokens": 43312562.0,
+      "step": 18720
+    },
+    {
+      "epoch": 0.17172458054460438,
+      "learning_rate": 0.00016565691757586872,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.7439317822456359,
+      "num_tokens": 43336606.0,
+      "step": 18730
+    },
+    {
+      "epoch": 0.17181626478408363,
+      "learning_rate": 0.00016563858072797286,
+      "loss": 0.8827,
+      "mean_token_accuracy": 0.745622968673706,
+      "num_tokens": 43359267.0,
+      "step": 18740
+    },
+    {
+      "epoch": 0.17190794902356285,
+      "learning_rate": 0.00016562024388007703,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7413673400878906,
+      "num_tokens": 43381707.0,
+      "step": 18750
+    },
+    {
+      "epoch": 0.17199963326304207,
+      "learning_rate": 0.00016560190703218116,
+      "loss": 0.8815,
+      "mean_token_accuracy": 0.7391298234462738,
+      "num_tokens": 43404642.0,
+      "step": 18760
+    },
+    {
+      "epoch": 0.17209131750252132,
+      "learning_rate": 0.00016558357018428533,
+      "loss": 0.8286,
+      "mean_token_accuracy": 0.749553632736206,
+      "num_tokens": 43427570.0,
+      "step": 18770
+    },
+    {
+      "epoch": 0.17218300174200055,
+      "learning_rate": 0.0001655652333363895,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.7399058640003204,
+      "num_tokens": 43450522.0,
+      "step": 18780
+    },
+    {
+      "epoch": 0.1722746859814798,
+      "learning_rate": 0.00016554689648849363,
+      "loss": 0.8819,
+      "mean_token_accuracy": 0.7367096960544586,
+      "num_tokens": 43474693.0,
+      "step": 18790
+    },
+    {
+      "epoch": 0.17236637022095902,
+      "learning_rate": 0.0001655285596405978,
+      "loss": 0.8765,
+      "mean_token_accuracy": 0.745033347606659,
+      "num_tokens": 43497467.0,
+      "step": 18800
+    },
+    {
+      "epoch": 0.17245805446043824,
+      "learning_rate": 0.00016551022279270194,
+      "loss": 0.8341,
+      "mean_token_accuracy": 0.7419469773769378,
+      "num_tokens": 43520604.0,
+      "step": 18810
+    },
+    {
+      "epoch": 0.1725497386999175,
+      "learning_rate": 0.0001654918859448061,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7410984992980957,
+      "num_tokens": 43543484.0,
+      "step": 18820
+    },
+    {
+      "epoch": 0.17264142293939672,
+      "learning_rate": 0.00016547354909691024,
+      "loss": 0.8622,
+      "mean_token_accuracy": 0.7435258150100708,
+      "num_tokens": 43566599.0,
+      "step": 18830
+    },
+    {
+      "epoch": 0.17273310717887594,
+      "learning_rate": 0.0001654552122490144,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.7390017509460449,
+      "num_tokens": 43589880.0,
+      "step": 18840
+    },
+    {
+      "epoch": 0.1728247914183552,
+      "learning_rate": 0.00016543687540111857,
+      "loss": 0.8222,
+      "mean_token_accuracy": 0.7479391753673553,
+      "num_tokens": 43613401.0,
+      "step": 18850
+    },
+    {
+      "epoch": 0.1729164756578344,
+      "learning_rate": 0.0001654185385532227,
+      "loss": 0.8804,
+      "mean_token_accuracy": 0.7409896969795227,
+      "num_tokens": 43636740.0,
+      "step": 18860
+    },
+    {
+      "epoch": 0.17300815989731366,
+      "learning_rate": 0.00016540020170532687,
+      "loss": 0.9086,
+      "mean_token_accuracy": 0.7395063579082489,
+      "num_tokens": 43659994.0,
+      "step": 18870
+    },
+    {
+      "epoch": 0.17309984413679289,
+      "learning_rate": 0.000165381864857431,
+      "loss": 0.8353,
+      "mean_token_accuracy": 0.7457105338573455,
+      "num_tokens": 43683450.0,
+      "step": 18880
+    },
+    {
+      "epoch": 0.1731915283762721,
+      "learning_rate": 0.00016536352800953515,
+      "loss": 0.8514,
+      "mean_token_accuracy": 0.7494057893753052,
+      "num_tokens": 43706137.0,
+      "step": 18890
+    },
+    {
+      "epoch": 0.17328321261575136,
+      "learning_rate": 0.00016534519116163932,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7364172041416168,
+      "num_tokens": 43729224.0,
+      "step": 18900
+    },
+    {
+      "epoch": 0.17337489685523058,
+      "learning_rate": 0.00016532685431374348,
+      "loss": 0.8394,
+      "mean_token_accuracy": 0.7496098220348358,
+      "num_tokens": 43751460.0,
+      "step": 18910
+    },
+    {
+      "epoch": 0.17346658109470983,
+      "learning_rate": 0.00016530851746584765,
+      "loss": 0.8153,
+      "mean_token_accuracy": 0.7558809697628022,
+      "num_tokens": 43774059.0,
+      "step": 18920
+    },
+    {
+      "epoch": 0.17355826533418905,
+      "learning_rate": 0.00016529018061795179,
+      "loss": 0.882,
+      "mean_token_accuracy": 0.7378742516040802,
+      "num_tokens": 43797321.0,
+      "step": 18930
+    },
+    {
+      "epoch": 0.17364994957366828,
+      "learning_rate": 0.00016527184377005592,
+      "loss": 0.8832,
+      "mean_token_accuracy": 0.7422385215759277,
+      "num_tokens": 43820304.0,
+      "step": 18940
+    },
+    {
+      "epoch": 0.17374163381314753,
+      "learning_rate": 0.0001652535069221601,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.7387481093406677,
+      "num_tokens": 43843406.0,
+      "step": 18950
+    },
+    {
+      "epoch": 0.17383331805262675,
+      "learning_rate": 0.00016523517007426423,
+      "loss": 0.8709,
+      "mean_token_accuracy": 0.7417144775390625,
+      "num_tokens": 43866202.0,
+      "step": 18960
+    },
+    {
+      "epoch": 0.17392500229210597,
+      "learning_rate": 0.0001652168332263684,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7436456739902496,
+      "num_tokens": 43889572.0,
+      "step": 18970
+    },
+    {
+      "epoch": 0.17401668653158522,
+      "learning_rate": 0.00016519849637847256,
+      "loss": 0.8575,
+      "mean_token_accuracy": 0.742885822057724,
+      "num_tokens": 43911886.0,
+      "step": 18980
+    },
+    {
+      "epoch": 0.17410837077106445,
+      "learning_rate": 0.0001651801595305767,
+      "loss": 0.8632,
+      "mean_token_accuracy": 0.7443060219287873,
+      "num_tokens": 43934848.0,
+      "step": 18990
+    },
+    {
+      "epoch": 0.1742000550105437,
+      "learning_rate": 0.00016516182268268086,
+      "loss": 0.8399,
+      "mean_token_accuracy": 0.7461564898490906,
+      "num_tokens": 43957831.0,
+      "step": 19000
+    },
+    {
+      "epoch": 0.17429173925002292,
+      "learning_rate": 0.000165143485834785,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.7413355529308319,
+      "num_tokens": 43981365.0,
+      "step": 19010
+    },
+    {
+      "epoch": 0.17438342348950214,
+      "learning_rate": 0.00016512514898688917,
+      "loss": 0.8653,
+      "mean_token_accuracy": 0.742383724451065,
+      "num_tokens": 44004122.0,
+      "step": 19020
+    },
+    {
+      "epoch": 0.1744751077289814,
+      "learning_rate": 0.0001651068121389933,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7481968820095062,
+      "num_tokens": 44027032.0,
+      "step": 19030
+    },
+    {
+      "epoch": 0.17456679196846062,
+      "learning_rate": 0.00016508847529109747,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7434242486953735,
+      "num_tokens": 44050243.0,
+      "step": 19040
+    },
+    {
+      "epoch": 0.17465847620793987,
+      "learning_rate": 0.00016507013844320164,
+      "loss": 0.8354,
+      "mean_token_accuracy": 0.75514035820961,
+      "num_tokens": 44073127.0,
+      "step": 19050
+    },
+    {
+      "epoch": 0.1747501604474191,
+      "learning_rate": 0.00016505180159530577,
+      "loss": 0.8855,
+      "mean_token_accuracy": 0.7421274006366729,
+      "num_tokens": 44095286.0,
+      "step": 19060
+    },
+    {
+      "epoch": 0.1748418446868983,
+      "learning_rate": 0.00016503346474740994,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.7372287750244141,
+      "num_tokens": 44118182.0,
+      "step": 19070
+    },
+    {
+      "epoch": 0.17493352892637756,
+      "learning_rate": 0.00016501512789951408,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7466937124729156,
+      "num_tokens": 44141274.0,
+      "step": 19080
+    },
+    {
+      "epoch": 0.17502521316585679,
+      "learning_rate": 0.00016499679105161822,
+      "loss": 0.8583,
+      "mean_token_accuracy": 0.7452625215053559,
+      "num_tokens": 44164500.0,
+      "step": 19090
+    },
+    {
+      "epoch": 0.17511689740533604,
+      "learning_rate": 0.0001649784542037224,
+      "loss": 0.8361,
+      "mean_token_accuracy": 0.7550947666168213,
+      "num_tokens": 44187497.0,
+      "step": 19100
+    },
+    {
+      "epoch": 0.17520858164481526,
+      "learning_rate": 0.00016496011735582655,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.7341136753559112,
+      "num_tokens": 44210448.0,
+      "step": 19110
+    },
+    {
+      "epoch": 0.17530026588429448,
+      "learning_rate": 0.0001649417805079307,
+      "loss": 0.8689,
+      "mean_token_accuracy": 0.7395574748516083,
+      "num_tokens": 44234122.0,
+      "step": 19120
+    },
+    {
+      "epoch": 0.17539195012377373,
+      "learning_rate": 0.00016492344366003485,
+      "loss": 0.872,
+      "mean_token_accuracy": 0.7390694618225098,
+      "num_tokens": 44257673.0,
+      "step": 19130
+    },
+    {
+      "epoch": 0.17548363436325295,
+      "learning_rate": 0.000164905106812139,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.74883993268013,
+      "num_tokens": 44281056.0,
+      "step": 19140
+    },
+    {
+      "epoch": 0.17557531860273218,
+      "learning_rate": 0.00016488676996424315,
+      "loss": 0.8948,
+      "mean_token_accuracy": 0.7357378482818604,
+      "num_tokens": 44303911.0,
+      "step": 19150
+    },
+    {
+      "epoch": 0.17566700284221143,
+      "learning_rate": 0.0001648684331163473,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.7400649547576904,
+      "num_tokens": 44327654.0,
+      "step": 19160
+    },
+    {
+      "epoch": 0.17575868708169065,
+      "learning_rate": 0.00016485009626845146,
+      "loss": 0.901,
+      "mean_token_accuracy": 0.7337875425815582,
+      "num_tokens": 44350895.0,
+      "step": 19170
+    },
+    {
+      "epoch": 0.1758503713211699,
+      "learning_rate": 0.00016483175942055562,
+      "loss": 0.8898,
+      "mean_token_accuracy": 0.7370629727840423,
+      "num_tokens": 44374201.0,
+      "step": 19180
+    },
+    {
+      "epoch": 0.17594205556064912,
+      "learning_rate": 0.00016481342257265976,
+      "loss": 0.8717,
+      "mean_token_accuracy": 0.7427584946155548,
+      "num_tokens": 44396954.0,
+      "step": 19190
+    },
+    {
+      "epoch": 0.17603373980012835,
+      "learning_rate": 0.00016479508572476393,
+      "loss": 0.8361,
+      "mean_token_accuracy": 0.7497299134731292,
+      "num_tokens": 44419306.0,
+      "step": 19200
+    },
+    {
+      "epoch": 0.1761254240396076,
+      "learning_rate": 0.00016477674887686807,
+      "loss": 0.8891,
+      "mean_token_accuracy": 0.73868328332901,
+      "num_tokens": 44442145.0,
+      "step": 19210
+    },
+    {
+      "epoch": 0.17621710827908682,
+      "learning_rate": 0.00016475841202897223,
+      "loss": 0.8564,
+      "mean_token_accuracy": 0.7455623626708985,
+      "num_tokens": 44465833.0,
+      "step": 19220
+    },
+    {
+      "epoch": 0.17630879251856607,
+      "learning_rate": 0.00016474007518107637,
+      "loss": 0.8942,
+      "mean_token_accuracy": 0.7397753775119782,
+      "num_tokens": 44488565.0,
+      "step": 19230
+    },
+    {
+      "epoch": 0.1764004767580453,
+      "learning_rate": 0.00016472173833318053,
+      "loss": 0.8786,
+      "mean_token_accuracy": 0.7374221444129944,
+      "num_tokens": 44512602.0,
+      "step": 19240
+    },
+    {
+      "epoch": 0.17649216099752452,
+      "learning_rate": 0.0001647034014852847,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7420086741447449,
+      "num_tokens": 44535602.0,
+      "step": 19250
+    },
+    {
+      "epoch": 0.17658384523700377,
+      "learning_rate": 0.00016468506463738884,
+      "loss": 0.8832,
+      "mean_token_accuracy": 0.7369343996047973,
+      "num_tokens": 44558799.0,
+      "step": 19260
+    },
+    {
+      "epoch": 0.176675529476483,
+      "learning_rate": 0.000164666727789493,
+      "loss": 0.8477,
+      "mean_token_accuracy": 0.7505451381206513,
+      "num_tokens": 44582549.0,
+      "step": 19270
+    },
+    {
+      "epoch": 0.1767672137159622,
+      "learning_rate": 0.00016464839094159714,
+      "loss": 0.8617,
+      "mean_token_accuracy": 0.7434860467910767,
+      "num_tokens": 44604766.0,
+      "step": 19280
+    },
+    {
+      "epoch": 0.17685889795544146,
+      "learning_rate": 0.00016463005409370128,
+      "loss": 0.836,
+      "mean_token_accuracy": 0.7471416652202606,
+      "num_tokens": 44627621.0,
+      "step": 19290
+    },
+    {
+      "epoch": 0.17695058219492069,
+      "learning_rate": 0.00016461171724580547,
+      "loss": 0.9164,
+      "mean_token_accuracy": 0.7298359870910645,
+      "num_tokens": 44650461.0,
+      "step": 19300
+    },
+    {
+      "epoch": 0.17704226643439994,
+      "learning_rate": 0.0001645933803979096,
+      "loss": 0.804,
+      "mean_token_accuracy": 0.7538255751132965,
+      "num_tokens": 44673964.0,
+      "step": 19310
+    },
+    {
+      "epoch": 0.17713395067387916,
+      "learning_rate": 0.00016457504355001378,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7465196669101715,
+      "num_tokens": 44696838.0,
+      "step": 19320
+    },
+    {
+      "epoch": 0.17722563491335838,
+      "learning_rate": 0.00016455670670211791,
+      "loss": 0.8774,
+      "mean_token_accuracy": 0.7344714820384979,
+      "num_tokens": 44719826.0,
+      "step": 19330
+    },
+    {
+      "epoch": 0.17731731915283763,
+      "learning_rate": 0.00016453836985422205,
+      "loss": 0.8914,
+      "mean_token_accuracy": 0.7317513585090637,
+      "num_tokens": 44743711.0,
+      "step": 19340
+    },
+    {
+      "epoch": 0.17740900339231686,
+      "learning_rate": 0.00016452003300632622,
+      "loss": 0.8777,
+      "mean_token_accuracy": 0.7445966899394989,
+      "num_tokens": 44766384.0,
+      "step": 19350
+    },
+    {
+      "epoch": 0.1775006876317961,
+      "learning_rate": 0.00016450169615843036,
+      "loss": 0.8465,
+      "mean_token_accuracy": 0.741133987903595,
+      "num_tokens": 44790028.0,
+      "step": 19360
+    },
+    {
+      "epoch": 0.17759237187127533,
+      "learning_rate": 0.00016448335931053452,
+      "loss": 0.8943,
+      "mean_token_accuracy": 0.7360158443450928,
+      "num_tokens": 44812679.0,
+      "step": 19370
+    },
+    {
+      "epoch": 0.17768405611075455,
+      "learning_rate": 0.0001644650224626387,
+      "loss": 0.9177,
+      "mean_token_accuracy": 0.7309502482414245,
+      "num_tokens": 44835430.0,
+      "step": 19380
+    },
+    {
+      "epoch": 0.1777757403502338,
+      "learning_rate": 0.00016444668561474283,
+      "loss": 0.8903,
+      "mean_token_accuracy": 0.7399446070194244,
+      "num_tokens": 44858385.0,
+      "step": 19390
+    },
+    {
+      "epoch": 0.17786742458971302,
+      "learning_rate": 0.000164428348766847,
+      "loss": 0.8501,
+      "mean_token_accuracy": 0.7484567940235138,
+      "num_tokens": 44881181.0,
+      "step": 19400
+    },
+    {
+      "epoch": 0.17795910882919228,
+      "learning_rate": 0.00016441001191895113,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7476139545440674,
+      "num_tokens": 44904429.0,
+      "step": 19410
+    },
+    {
+      "epoch": 0.1780507930686715,
+      "learning_rate": 0.0001643916750710553,
+      "loss": 0.8799,
+      "mean_token_accuracy": 0.7403489708900451,
+      "num_tokens": 44927631.0,
+      "step": 19420
+    },
+    {
+      "epoch": 0.17814247730815072,
+      "learning_rate": 0.00016437333822315946,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.732146555185318,
+      "num_tokens": 44951356.0,
+      "step": 19430
+    },
+    {
+      "epoch": 0.17823416154762997,
+      "learning_rate": 0.0001643550013752636,
+      "loss": 0.8373,
+      "mean_token_accuracy": 0.7548953115940094,
+      "num_tokens": 44974118.0,
+      "step": 19440
+    },
+    {
+      "epoch": 0.1783258457871092,
+      "learning_rate": 0.00016433666452736776,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.7457742691040039,
+      "num_tokens": 44996700.0,
+      "step": 19450
+    },
+    {
+      "epoch": 0.17841753002658842,
+      "learning_rate": 0.0001643183276794719,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7452894151210785,
+      "num_tokens": 45019881.0,
+      "step": 19460
+    },
+    {
+      "epoch": 0.17850921426606767,
+      "learning_rate": 0.00016429999083157607,
+      "loss": 0.8159,
+      "mean_token_accuracy": 0.7612899601459503,
+      "num_tokens": 45043074.0,
+      "step": 19470
+    },
+    {
+      "epoch": 0.1786008985055469,
+      "learning_rate": 0.0001642816539836802,
+      "loss": 0.8534,
+      "mean_token_accuracy": 0.7475025415420532,
+      "num_tokens": 45066291.0,
+      "step": 19480
+    },
+    {
+      "epoch": 0.17869258274502614,
+      "learning_rate": 0.00016426331713578434,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7391200125217438,
+      "num_tokens": 45090233.0,
+      "step": 19490
+    },
+    {
+      "epoch": 0.17878426698450536,
+      "learning_rate": 0.00016424498028788854,
+      "loss": 0.8746,
+      "mean_token_accuracy": 0.7415766716003418,
+      "num_tokens": 45113733.0,
+      "step": 19500
+    },
+    {
+      "epoch": 0.1788759512239846,
+      "learning_rate": 0.00016422664343999268,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7435136795043945,
+      "num_tokens": 45137173.0,
+      "step": 19510
+    },
+    {
+      "epoch": 0.17896763546346384,
+      "learning_rate": 0.00016420830659209684,
+      "loss": 0.8473,
+      "mean_token_accuracy": 0.748375552892685,
+      "num_tokens": 45159798.0,
+      "step": 19520
+    },
+    {
+      "epoch": 0.17905931970294306,
+      "learning_rate": 0.00016418996974420098,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7303220510482789,
+      "num_tokens": 45183235.0,
+      "step": 19530
+    },
+    {
+      "epoch": 0.1791510039424223,
+      "learning_rate": 0.00016417163289630512,
+      "loss": 0.888,
+      "mean_token_accuracy": 0.7358284950256347,
+      "num_tokens": 45206108.0,
+      "step": 19540
+    },
+    {
+      "epoch": 0.17924268818190153,
+      "learning_rate": 0.00016415329604840928,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7365753650665283,
+      "num_tokens": 45228871.0,
+      "step": 19550
+    },
+    {
+      "epoch": 0.17933437242138076,
+      "learning_rate": 0.00016413495920051345,
+      "loss": 0.8983,
+      "mean_token_accuracy": 0.7411333501338959,
+      "num_tokens": 45252530.0,
+      "step": 19560
+    },
+    {
+      "epoch": 0.17942605666086,
+      "learning_rate": 0.0001641166223526176,
+      "loss": 0.8737,
+      "mean_token_accuracy": 0.746991240978241,
+      "num_tokens": 45276643.0,
+      "step": 19570
+    },
+    {
+      "epoch": 0.17951774090033923,
+      "learning_rate": 0.00016409828550472175,
+      "loss": 0.8556,
+      "mean_token_accuracy": 0.7432246744632721,
+      "num_tokens": 45299791.0,
+      "step": 19580
+    },
+    {
+      "epoch": 0.17960942513981845,
+      "learning_rate": 0.0001640799486568259,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.7457501471042634,
+      "num_tokens": 45323666.0,
+      "step": 19590
+    },
+    {
+      "epoch": 0.1797011093792977,
+      "learning_rate": 0.00016406161180893006,
+      "loss": 0.8451,
+      "mean_token_accuracy": 0.7440292656421661,
+      "num_tokens": 45347025.0,
+      "step": 19600
+    },
+    {
+      "epoch": 0.17979279361877692,
+      "learning_rate": 0.0001640432749610342,
+      "loss": 0.8839,
+      "mean_token_accuracy": 0.7357450425624847,
+      "num_tokens": 45369527.0,
+      "step": 19610
+    },
+    {
+      "epoch": 0.17988447785825618,
+      "learning_rate": 0.00016402493811313836,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.7435458481311799,
+      "num_tokens": 45392440.0,
+      "step": 19620
+    },
+    {
+      "epoch": 0.1799761620977354,
+      "learning_rate": 0.00016400660126524253,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7475884079933166,
+      "num_tokens": 45416597.0,
+      "step": 19630
+    },
+    {
+      "epoch": 0.18006784633721462,
+      "learning_rate": 0.00016398826441734666,
+      "loss": 0.8506,
+      "mean_token_accuracy": 0.7514847576618194,
+      "num_tokens": 45439155.0,
+      "step": 19640
+    },
+    {
+      "epoch": 0.18015953057669387,
+      "learning_rate": 0.00016396992756945083,
+      "loss": 0.8856,
+      "mean_token_accuracy": 0.7473500847816468,
+      "num_tokens": 45461757.0,
+      "step": 19650
+    },
+    {
+      "epoch": 0.1802512148161731,
+      "learning_rate": 0.00016395159072155497,
+      "loss": 0.9023,
+      "mean_token_accuracy": 0.7286785840988159,
+      "num_tokens": 45484391.0,
+      "step": 19660
+    },
+    {
+      "epoch": 0.18034289905565234,
+      "learning_rate": 0.00016393325387365913,
+      "loss": 0.8871,
+      "mean_token_accuracy": 0.7373834609985351,
+      "num_tokens": 45507644.0,
+      "step": 19670
+    },
+    {
+      "epoch": 0.18043458329513157,
+      "learning_rate": 0.00016391491702576327,
+      "loss": 0.8757,
+      "mean_token_accuracy": 0.7394998490810394,
+      "num_tokens": 45530964.0,
+      "step": 19680
+    },
+    {
+      "epoch": 0.1805262675346108,
+      "learning_rate": 0.00016389658017786744,
+      "loss": 0.8708,
+      "mean_token_accuracy": 0.7445003032684326,
+      "num_tokens": 45554679.0,
+      "step": 19690
+    },
+    {
+      "epoch": 0.18061795177409004,
+      "learning_rate": 0.0001638782433299716,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7432986915111541,
+      "num_tokens": 45577356.0,
+      "step": 19700
+    },
+    {
+      "epoch": 0.18070963601356926,
+      "learning_rate": 0.00016385990648207574,
+      "loss": 0.8865,
+      "mean_token_accuracy": 0.740937077999115,
+      "num_tokens": 45600553.0,
+      "step": 19710
+    },
+    {
+      "epoch": 0.18080132025304851,
+      "learning_rate": 0.0001638415696341799,
+      "loss": 0.8362,
+      "mean_token_accuracy": 0.7443297207355499,
+      "num_tokens": 45623672.0,
+      "step": 19720
+    },
+    {
+      "epoch": 0.18089300449252774,
+      "learning_rate": 0.00016382323278628404,
+      "loss": 0.8856,
+      "mean_token_accuracy": 0.7394998967647552,
+      "num_tokens": 45647220.0,
+      "step": 19730
+    },
+    {
+      "epoch": 0.18098468873200696,
+      "learning_rate": 0.00016380489593838818,
+      "loss": 0.8825,
+      "mean_token_accuracy": 0.7407486975193024,
+      "num_tokens": 45670407.0,
+      "step": 19740
+    },
+    {
+      "epoch": 0.1810763729714862,
+      "learning_rate": 0.00016378655909049235,
+      "loss": 0.8626,
+      "mean_token_accuracy": 0.7361254632472992,
+      "num_tokens": 45693934.0,
+      "step": 19750
+    },
+    {
+      "epoch": 0.18116805721096543,
+      "learning_rate": 0.0001637682222425965,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7386096835136413,
+      "num_tokens": 45716903.0,
+      "step": 19760
+    },
+    {
+      "epoch": 0.18125974145044466,
+      "learning_rate": 0.00016374988539470065,
+      "loss": 0.8946,
+      "mean_token_accuracy": 0.7376744985580445,
+      "num_tokens": 45740340.0,
+      "step": 19770
+    },
+    {
+      "epoch": 0.1813514256899239,
+      "learning_rate": 0.00016373154854680482,
+      "loss": 0.8693,
+      "mean_token_accuracy": 0.7449719130992889,
+      "num_tokens": 45763197.0,
+      "step": 19780
+    },
+    {
+      "epoch": 0.18144310992940313,
+      "learning_rate": 0.00016371321169890895,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.7459929168224335,
+      "num_tokens": 45785869.0,
+      "step": 19790
+    },
+    {
+      "epoch": 0.18153479416888238,
+      "learning_rate": 0.00016369487485101312,
+      "loss": 0.8516,
+      "mean_token_accuracy": 0.7407790124416351,
+      "num_tokens": 45809432.0,
+      "step": 19800
+    },
+    {
+      "epoch": 0.1816264784083616,
+      "learning_rate": 0.00016367653800311726,
+      "loss": 0.8459,
+      "mean_token_accuracy": 0.744923597574234,
+      "num_tokens": 45832356.0,
+      "step": 19810
+    },
+    {
+      "epoch": 0.18171816264784083,
+      "learning_rate": 0.00016365820115522142,
+      "loss": 0.8828,
+      "mean_token_accuracy": 0.7451933920383453,
+      "num_tokens": 45855088.0,
+      "step": 19820
+    },
+    {
+      "epoch": 0.18180984688732008,
+      "learning_rate": 0.0001636398643073256,
+      "loss": 0.8364,
+      "mean_token_accuracy": 0.7445485055446625,
+      "num_tokens": 45877463.0,
+      "step": 19830
+    },
+    {
+      "epoch": 0.1819015311267993,
+      "learning_rate": 0.00016362152745942973,
+      "loss": 0.8548,
+      "mean_token_accuracy": 0.745529317855835,
+      "num_tokens": 45900717.0,
+      "step": 19840
+    },
+    {
+      "epoch": 0.18199321536627855,
+      "learning_rate": 0.0001636031906115339,
+      "loss": 0.9117,
+      "mean_token_accuracy": 0.7307271659374237,
+      "num_tokens": 45924046.0,
+      "step": 19850
+    },
+    {
+      "epoch": 0.18208489960575777,
+      "learning_rate": 0.00016358485376363803,
+      "loss": 0.9024,
+      "mean_token_accuracy": 0.7345387160778045,
+      "num_tokens": 45946762.0,
+      "step": 19860
+    },
+    {
+      "epoch": 0.182176583845237,
+      "learning_rate": 0.0001635665169157422,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.736064100265503,
+      "num_tokens": 45969741.0,
+      "step": 19870
+    },
+    {
+      "epoch": 0.18226826808471625,
+      "learning_rate": 0.00016354818006784634,
+      "loss": 0.8923,
+      "mean_token_accuracy": 0.7301272034645081,
+      "num_tokens": 45992270.0,
+      "step": 19880
+    },
+    {
+      "epoch": 0.18235995232419547,
+      "learning_rate": 0.0001635298432199505,
+      "loss": 0.8584,
+      "mean_token_accuracy": 0.7410679697990418,
+      "num_tokens": 46015207.0,
+      "step": 19890
+    },
+    {
+      "epoch": 0.1824516365636747,
+      "learning_rate": 0.00016351150637205467,
+      "loss": 0.8307,
+      "mean_token_accuracy": 0.7555568873882293,
+      "num_tokens": 46038371.0,
+      "step": 19900
+    },
+    {
+      "epoch": 0.18254332080315394,
+      "learning_rate": 0.0001634931695241588,
+      "loss": 0.8631,
+      "mean_token_accuracy": 0.7425933182239532,
+      "num_tokens": 46061087.0,
+      "step": 19910
+    },
+    {
+      "epoch": 0.18263500504263316,
+      "learning_rate": 0.00016347483267626297,
+      "loss": 0.8616,
+      "mean_token_accuracy": 0.740113353729248,
+      "num_tokens": 46084548.0,
+      "step": 19920
+    },
+    {
+      "epoch": 0.18272668928211241,
+      "learning_rate": 0.0001634564958283671,
+      "loss": 0.8058,
+      "mean_token_accuracy": 0.7532809972763062,
+      "num_tokens": 46107256.0,
+      "step": 19930
+    },
+    {
+      "epoch": 0.18281837352159164,
+      "learning_rate": 0.00016343815898047125,
+      "loss": 0.9007,
+      "mean_token_accuracy": 0.7346232056617736,
+      "num_tokens": 46130516.0,
+      "step": 19940
+    },
+    {
+      "epoch": 0.18291005776107086,
+      "learning_rate": 0.00016341982213257544,
+      "loss": 0.8798,
+      "mean_token_accuracy": 0.7400797307491302,
+      "num_tokens": 46154477.0,
+      "step": 19950
+    },
+    {
+      "epoch": 0.1830017420005501,
+      "learning_rate": 0.00016340148528467958,
+      "loss": 0.801,
+      "mean_token_accuracy": 0.7564774513244629,
+      "num_tokens": 46177456.0,
+      "step": 19960
+    },
+    {
+      "epoch": 0.18309342624002933,
+      "learning_rate": 0.00016338314843678372,
+      "loss": 0.8632,
+      "mean_token_accuracy": 0.7449011743068695,
+      "num_tokens": 46200615.0,
+      "step": 19970
+    },
+    {
+      "epoch": 0.18318511047950858,
+      "learning_rate": 0.00016336481158888788,
+      "loss": 0.8412,
+      "mean_token_accuracy": 0.7452260315418243,
+      "num_tokens": 46223213.0,
+      "step": 19980
+    },
+    {
+      "epoch": 0.1832767947189878,
+      "learning_rate": 0.00016334647474099202,
+      "loss": 0.8601,
+      "mean_token_accuracy": 0.746433675289154,
+      "num_tokens": 46246461.0,
+      "step": 19990
+    },
+    {
+      "epoch": 0.18336847895846703,
+      "learning_rate": 0.00016332813789309618,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7386339783668519,
+      "num_tokens": 46268440.0,
+      "step": 20000
+    },
+    {
+      "epoch": 0.18346016319794628,
+      "learning_rate": 0.00016330980104520032,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7404561281204224,
+      "num_tokens": 46291501.0,
+      "step": 20010
+    },
+    {
+      "epoch": 0.1835518474374255,
+      "learning_rate": 0.0001632914641973045,
+      "loss": 0.8874,
+      "mean_token_accuracy": 0.7369221687316895,
+      "num_tokens": 46314964.0,
+      "step": 20020
+    },
+    {
+      "epoch": 0.18364353167690475,
+      "learning_rate": 0.00016327312734940865,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.7409957587718964,
+      "num_tokens": 46337717.0,
+      "step": 20030
+    },
+    {
+      "epoch": 0.18373521591638398,
+      "learning_rate": 0.0001632547905015128,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.7375479519367218,
+      "num_tokens": 46360631.0,
+      "step": 20040
+    },
+    {
+      "epoch": 0.1838269001558632,
+      "learning_rate": 0.00016323645365361696,
+      "loss": 0.8746,
+      "mean_token_accuracy": 0.7424764573574066,
+      "num_tokens": 46384250.0,
+      "step": 20050
+    },
+    {
+      "epoch": 0.18391858439534245,
+      "learning_rate": 0.0001632181168057211,
+      "loss": 0.8493,
+      "mean_token_accuracy": 0.7477382957935333,
+      "num_tokens": 46407372.0,
+      "step": 20060
+    },
+    {
+      "epoch": 0.18401026863482167,
+      "learning_rate": 0.00016319977995782526,
+      "loss": 0.8902,
+      "mean_token_accuracy": 0.7340875267982483,
+      "num_tokens": 46430793.0,
+      "step": 20070
+    },
+    {
+      "epoch": 0.1841019528743009,
+      "learning_rate": 0.00016318144310992943,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7367632687091827,
+      "num_tokens": 46452956.0,
+      "step": 20080
+    },
+    {
+      "epoch": 0.18419363711378015,
+      "learning_rate": 0.00016316310626203357,
+      "loss": 0.9275,
+      "mean_token_accuracy": 0.7276939272880554,
+      "num_tokens": 46476194.0,
+      "step": 20090
+    },
+    {
+      "epoch": 0.18428532135325937,
+      "learning_rate": 0.00016314476941413773,
+      "loss": 0.8635,
+      "mean_token_accuracy": 0.7420652866363525,
+      "num_tokens": 46500028.0,
+      "step": 20100
+    },
+    {
+      "epoch": 0.18437700559273862,
+      "learning_rate": 0.00016312643256624187,
+      "loss": 0.8622,
+      "mean_token_accuracy": 0.7478643298149109,
+      "num_tokens": 46522328.0,
+      "step": 20110
+    },
+    {
+      "epoch": 0.18446868983221784,
+      "learning_rate": 0.00016310809571834603,
+      "loss": 0.9022,
+      "mean_token_accuracy": 0.739221578836441,
+      "num_tokens": 46544781.0,
+      "step": 20120
+    },
+    {
+      "epoch": 0.18456037407169706,
+      "learning_rate": 0.00016308975887045017,
+      "loss": 0.8737,
+      "mean_token_accuracy": 0.7402739584445953,
+      "num_tokens": 46567167.0,
+      "step": 20130
+    },
+    {
+      "epoch": 0.18465205831117631,
+      "learning_rate": 0.0001630714220225543,
+      "loss": 0.8891,
+      "mean_token_accuracy": 0.7409417510032654,
+      "num_tokens": 46590490.0,
+      "step": 20140
+    },
+    {
+      "epoch": 0.18474374255065554,
+      "learning_rate": 0.0001630530851746585,
+      "loss": 0.8535,
+      "mean_token_accuracy": 0.7470172762870788,
+      "num_tokens": 46613515.0,
+      "step": 20150
+    },
+    {
+      "epoch": 0.1848354267901348,
+      "learning_rate": 0.00016303474832676264,
+      "loss": 0.8709,
+      "mean_token_accuracy": 0.7364602446556091,
+      "num_tokens": 46636624.0,
+      "step": 20160
+    },
+    {
+      "epoch": 0.184927111029614,
+      "learning_rate": 0.00016301641147886678,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.7441396534442901,
+      "num_tokens": 46659294.0,
+      "step": 20170
+    },
+    {
+      "epoch": 0.18501879526909323,
+      "learning_rate": 0.00016299807463097095,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.751100218296051,
+      "num_tokens": 46682558.0,
+      "step": 20180
+    },
+    {
+      "epoch": 0.18511047950857248,
+      "learning_rate": 0.00016297973778307508,
+      "loss": 0.8435,
+      "mean_token_accuracy": 0.7442737102508545,
+      "num_tokens": 46706272.0,
+      "step": 20190
+    },
+    {
+      "epoch": 0.1852021637480517,
+      "learning_rate": 0.00016296140093517925,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7447192013263703,
+      "num_tokens": 46729004.0,
+      "step": 20200
+    },
+    {
+      "epoch": 0.18529384798753093,
+      "learning_rate": 0.00016294306408728341,
+      "loss": 0.8893,
+      "mean_token_accuracy": 0.737049913406372,
+      "num_tokens": 46751920.0,
+      "step": 20210
+    },
+    {
+      "epoch": 0.18538553222701018,
+      "learning_rate": 0.00016292472723938755,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.7449952661991119,
+      "num_tokens": 46775591.0,
+      "step": 20220
+    },
+    {
+      "epoch": 0.1854772164664894,
+      "learning_rate": 0.00016290639039149172,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7429184257984162,
+      "num_tokens": 46799177.0,
+      "step": 20230
+    },
+    {
+      "epoch": 0.18556890070596865,
+      "learning_rate": 0.00016288805354359586,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7441246688365937,
+      "num_tokens": 46822170.0,
+      "step": 20240
+    },
+    {
+      "epoch": 0.18566058494544788,
+      "learning_rate": 0.00016286971669570002,
+      "loss": 0.8754,
+      "mean_token_accuracy": 0.741225790977478,
+      "num_tokens": 46845590.0,
+      "step": 20250
+    },
+    {
+      "epoch": 0.1857522691849271,
+      "learning_rate": 0.00016285137984780416,
+      "loss": 0.8961,
+      "mean_token_accuracy": 0.739786970615387,
+      "num_tokens": 46868713.0,
+      "step": 20260
+    },
+    {
+      "epoch": 0.18584395342440635,
+      "learning_rate": 0.00016283304299990833,
+      "loss": 0.8766,
+      "mean_token_accuracy": 0.7376769185066223,
+      "num_tokens": 46891875.0,
+      "step": 20270
+    },
+    {
+      "epoch": 0.18593563766388557,
+      "learning_rate": 0.0001628147061520125,
+      "loss": 0.8516,
+      "mean_token_accuracy": 0.7486691176891327,
+      "num_tokens": 46915159.0,
+      "step": 20280
+    },
+    {
+      "epoch": 0.18602732190336482,
+      "learning_rate": 0.00016279636930411663,
+      "loss": 0.8967,
+      "mean_token_accuracy": 0.737358820438385,
+      "num_tokens": 46937783.0,
+      "step": 20290
+    },
+    {
+      "epoch": 0.18611900614284405,
+      "learning_rate": 0.0001627780324562208,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.7393028676509857,
+      "num_tokens": 46960877.0,
+      "step": 20300
+    },
+    {
+      "epoch": 0.18621069038232327,
+      "learning_rate": 0.00016275969560832493,
+      "loss": 0.8501,
+      "mean_token_accuracy": 0.7443229734897614,
+      "num_tokens": 46984237.0,
+      "step": 20310
+    },
+    {
+      "epoch": 0.18630237462180252,
+      "learning_rate": 0.0001627413587604291,
+      "loss": 0.8692,
+      "mean_token_accuracy": 0.7384073853492736,
+      "num_tokens": 47007071.0,
+      "step": 20320
+    },
+    {
+      "epoch": 0.18639405886128174,
+      "learning_rate": 0.00016272302191253324,
+      "loss": 0.8642,
+      "mean_token_accuracy": 0.7440657258033753,
+      "num_tokens": 47030529.0,
+      "step": 20330
+    },
+    {
+      "epoch": 0.186485743100761,
+      "learning_rate": 0.0001627046850646374,
+      "loss": 0.8998,
+      "mean_token_accuracy": 0.7337565958499909,
+      "num_tokens": 47054139.0,
+      "step": 20340
+    },
+    {
+      "epoch": 0.18657742734024022,
+      "learning_rate": 0.00016268634821674157,
+      "loss": 0.9173,
+      "mean_token_accuracy": 0.7255724966526031,
+      "num_tokens": 47077431.0,
+      "step": 20350
+    },
+    {
+      "epoch": 0.18666911157971944,
+      "learning_rate": 0.0001626680113688457,
+      "loss": 0.8224,
+      "mean_token_accuracy": 0.7518046200275421,
+      "num_tokens": 47100193.0,
+      "step": 20360
+    },
+    {
+      "epoch": 0.1867607958191987,
+      "learning_rate": 0.00016264967452094984,
+      "loss": 0.8833,
+      "mean_token_accuracy": 0.7428637504577636,
+      "num_tokens": 47123625.0,
+      "step": 20370
+    },
+    {
+      "epoch": 0.1868524800586779,
+      "learning_rate": 0.000162631337673054,
+      "loss": 0.8519,
+      "mean_token_accuracy": 0.742831003665924,
+      "num_tokens": 47146914.0,
+      "step": 20380
+    },
+    {
+      "epoch": 0.18694416429815713,
+      "learning_rate": 0.00016261300082515815,
+      "loss": 0.8822,
+      "mean_token_accuracy": 0.7372194111347199,
+      "num_tokens": 47170484.0,
+      "step": 20390
+    },
+    {
+      "epoch": 0.18703584853763638,
+      "learning_rate": 0.00016259466397726231,
+      "loss": 0.8827,
+      "mean_token_accuracy": 0.7353140771389007,
+      "num_tokens": 47193430.0,
+      "step": 20400
+    },
+    {
+      "epoch": 0.1871275327771156,
+      "learning_rate": 0.00016257632712936648,
+      "loss": 0.8743,
+      "mean_token_accuracy": 0.7365429759025574,
+      "num_tokens": 47216740.0,
+      "step": 20410
+    },
+    {
+      "epoch": 0.18721921701659486,
+      "learning_rate": 0.00016255799028147062,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7402590930461883,
+      "num_tokens": 47239893.0,
+      "step": 20420
+    },
+    {
+      "epoch": 0.18731090125607408,
+      "learning_rate": 0.00016253965343357478,
+      "loss": 0.8506,
+      "mean_token_accuracy": 0.7422642886638642,
+      "num_tokens": 47262839.0,
+      "step": 20430
+    },
+    {
+      "epoch": 0.1874025854955533,
+      "learning_rate": 0.00016252131658567892,
+      "loss": 0.9027,
+      "mean_token_accuracy": 0.7306043088436127,
+      "num_tokens": 47285612.0,
+      "step": 20440
+    },
+    {
+      "epoch": 0.18749426973503255,
+      "learning_rate": 0.0001625029797377831,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.7351073741912841,
+      "num_tokens": 47308929.0,
+      "step": 20450
+    },
+    {
+      "epoch": 0.18758595397451178,
+      "learning_rate": 0.00016248464288988722,
+      "loss": 0.8744,
+      "mean_token_accuracy": 0.7398530781269074,
+      "num_tokens": 47332396.0,
+      "step": 20460
+    },
+    {
+      "epoch": 0.18767763821399103,
+      "learning_rate": 0.0001624663060419914,
+      "loss": 0.8563,
+      "mean_token_accuracy": 0.7346394240856171,
+      "num_tokens": 47356014.0,
+      "step": 20470
+    },
+    {
+      "epoch": 0.18776932245347025,
+      "learning_rate": 0.00016244796919409556,
+      "loss": 0.8702,
+      "mean_token_accuracy": 0.7430578410625458,
+      "num_tokens": 47378518.0,
+      "step": 20480
+    },
+    {
+      "epoch": 0.18786100669294947,
+      "learning_rate": 0.0001624296323461997,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.7519625306129456,
+      "num_tokens": 47401956.0,
+      "step": 20490
+    },
+    {
+      "epoch": 0.18795269093242872,
+      "learning_rate": 0.00016241129549830386,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7472677826881409,
+      "num_tokens": 47424377.0,
+      "step": 20500
+    },
+    {
+      "epoch": 0.18804437517190795,
+      "learning_rate": 0.000162392958650408,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.7401869833469391,
+      "num_tokens": 47446951.0,
+      "step": 20510
+    },
+    {
+      "epoch": 0.18813605941138717,
+      "learning_rate": 0.00016237462180251216,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.739195829629898,
+      "num_tokens": 47469323.0,
+      "step": 20520
+    },
+    {
+      "epoch": 0.18822774365086642,
+      "learning_rate": 0.0001623562849546163,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7473972558975219,
+      "num_tokens": 47492517.0,
+      "step": 20530
+    },
+    {
+      "epoch": 0.18831942789034564,
+      "learning_rate": 0.00016233794810672047,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7496969938278198,
+      "num_tokens": 47515625.0,
+      "step": 20540
+    },
+    {
+      "epoch": 0.1884111121298249,
+      "learning_rate": 0.00016231961125882463,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7391631364822387,
+      "num_tokens": 47538126.0,
+      "step": 20550
+    },
+    {
+      "epoch": 0.18850279636930412,
+      "learning_rate": 0.00016230127441092877,
+      "loss": 0.8279,
+      "mean_token_accuracy": 0.7459759950637818,
+      "num_tokens": 47560266.0,
+      "step": 20560
+    },
+    {
+      "epoch": 0.18859448060878334,
+      "learning_rate": 0.0001622829375630329,
+      "loss": 0.8903,
+      "mean_token_accuracy": 0.7351974725723267,
+      "num_tokens": 47583398.0,
+      "step": 20570
+    },
+    {
+      "epoch": 0.1886861648482626,
+      "learning_rate": 0.00016226460071513707,
+      "loss": 0.8355,
+      "mean_token_accuracy": 0.7493352830410004,
+      "num_tokens": 47606618.0,
+      "step": 20580
+    },
+    {
+      "epoch": 0.1887778490877418,
+      "learning_rate": 0.0001622462638672412,
+      "loss": 0.8783,
+      "mean_token_accuracy": 0.7386853516101837,
+      "num_tokens": 47629504.0,
+      "step": 20590
+    },
+    {
+      "epoch": 0.18886953332722106,
+      "learning_rate": 0.00016222792701934538,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7422787070274353,
+      "num_tokens": 47652234.0,
+      "step": 20600
+    },
+    {
+      "epoch": 0.18896121756670028,
+      "learning_rate": 0.00016220959017144954,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.7353614032268524,
+      "num_tokens": 47675685.0,
+      "step": 20610
+    },
+    {
+      "epoch": 0.1890529018061795,
+      "learning_rate": 0.00016219125332355368,
+      "loss": 0.8815,
+      "mean_token_accuracy": 0.7416007101535798,
+      "num_tokens": 47698240.0,
+      "step": 20620
+    },
+    {
+      "epoch": 0.18914458604565876,
+      "learning_rate": 0.00016217291647565785,
+      "loss": 0.8834,
+      "mean_token_accuracy": 0.7420873403549194,
+      "num_tokens": 47722415.0,
+      "step": 20630
+    },
+    {
+      "epoch": 0.18923627028513798,
+      "learning_rate": 0.00016215457962776199,
+      "loss": 0.8507,
+      "mean_token_accuracy": 0.7479335606098175,
+      "num_tokens": 47745210.0,
+      "step": 20640
+    },
+    {
+      "epoch": 0.18932795452461723,
+      "learning_rate": 0.00016213624277986615,
+      "loss": 0.8028,
+      "mean_token_accuracy": 0.7593098104000091,
+      "num_tokens": 47768717.0,
+      "step": 20650
+    },
+    {
+      "epoch": 0.18941963876409645,
+      "learning_rate": 0.0001621179059319703,
+      "loss": 0.8568,
+      "mean_token_accuracy": 0.7413754463195801,
+      "num_tokens": 47791806.0,
+      "step": 20660
+    },
+    {
+      "epoch": 0.18951132300357568,
+      "learning_rate": 0.00016209956908407445,
+      "loss": 0.8905,
+      "mean_token_accuracy": 0.7388002455234528,
+      "num_tokens": 47814623.0,
+      "step": 20670
+    },
+    {
+      "epoch": 0.18960300724305493,
+      "learning_rate": 0.00016208123223617862,
+      "loss": 0.897,
+      "mean_token_accuracy": 0.7394916534423828,
+      "num_tokens": 47837359.0,
+      "step": 20680
+    },
+    {
+      "epoch": 0.18969469148253415,
+      "learning_rate": 0.00016206289538828276,
+      "loss": 0.8884,
+      "mean_token_accuracy": 0.7370145142078399,
+      "num_tokens": 47860716.0,
+      "step": 20690
+    },
+    {
+      "epoch": 0.18978637572201337,
+      "learning_rate": 0.00016204455854038692,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.7474730551242829,
+      "num_tokens": 47883540.0,
+      "step": 20700
+    },
+    {
+      "epoch": 0.18987805996149262,
+      "learning_rate": 0.00016202622169249106,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7456452190876007,
+      "num_tokens": 47905850.0,
+      "step": 20710
+    },
+    {
+      "epoch": 0.18996974420097185,
+      "learning_rate": 0.00016200788484459523,
+      "loss": 0.8965,
+      "mean_token_accuracy": 0.7329223453998566,
+      "num_tokens": 47928676.0,
+      "step": 20720
+    },
+    {
+      "epoch": 0.1900614284404511,
+      "learning_rate": 0.00016198954799669937,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.7428911566734314,
+      "num_tokens": 47951913.0,
+      "step": 20730
+    },
+    {
+      "epoch": 0.19015311267993032,
+      "learning_rate": 0.00016197121114880353,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7258612811565399,
+      "num_tokens": 47974742.0,
+      "step": 20740
+    },
+    {
+      "epoch": 0.19024479691940954,
+      "learning_rate": 0.0001619528743009077,
+      "loss": 0.8565,
+      "mean_token_accuracy": 0.7507954895496368,
+      "num_tokens": 47997674.0,
+      "step": 20750
+    },
+    {
+      "epoch": 0.1903364811588888,
+      "learning_rate": 0.00016193453745301184,
+      "loss": 0.8899,
+      "mean_token_accuracy": 0.741180008649826,
+      "num_tokens": 48019409.0,
+      "step": 20760
+    },
+    {
+      "epoch": 0.19042816539836802,
+      "learning_rate": 0.00016191620060511597,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7505606949329376,
+      "num_tokens": 48042169.0,
+      "step": 20770
+    },
+    {
+      "epoch": 0.19051984963784727,
+      "learning_rate": 0.00016189786375722014,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7380756914615632,
+      "num_tokens": 48065218.0,
+      "step": 20780
+    },
+    {
+      "epoch": 0.1906115338773265,
+      "learning_rate": 0.00016187952690932428,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7432824075222015,
+      "num_tokens": 48087671.0,
+      "step": 20790
+    },
+    {
+      "epoch": 0.1907032181168057,
+      "learning_rate": 0.00016186119006142847,
+      "loss": 0.8334,
+      "mean_token_accuracy": 0.7480520606040955,
+      "num_tokens": 48110601.0,
+      "step": 20800
+    },
+    {
+      "epoch": 0.19079490235628496,
+      "learning_rate": 0.0001618428532135326,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7470580756664276,
+      "num_tokens": 48134187.0,
+      "step": 20810
+    },
+    {
+      "epoch": 0.19088658659576418,
+      "learning_rate": 0.00016182451636563675,
+      "loss": 0.8579,
+      "mean_token_accuracy": 0.7445999026298523,
+      "num_tokens": 48156688.0,
+      "step": 20820
+    },
+    {
+      "epoch": 0.19097827083524344,
+      "learning_rate": 0.0001618061795177409,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7393371820449829,
+      "num_tokens": 48179961.0,
+      "step": 20830
+    },
+    {
+      "epoch": 0.19106995507472266,
+      "learning_rate": 0.00016178784266984505,
+      "loss": 0.8817,
+      "mean_token_accuracy": 0.7396546721458435,
+      "num_tokens": 48203535.0,
+      "step": 20840
+    },
+    {
+      "epoch": 0.19116163931420188,
+      "learning_rate": 0.00016176950582194922,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7402848124504089,
+      "num_tokens": 48227238.0,
+      "step": 20850
+    },
+    {
+      "epoch": 0.19125332355368113,
+      "learning_rate": 0.00016175116897405335,
+      "loss": 0.8888,
+      "mean_token_accuracy": 0.7421349585056305,
+      "num_tokens": 48250790.0,
+      "step": 20860
+    },
+    {
+      "epoch": 0.19134500779316035,
+      "learning_rate": 0.00016173283212615752,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.740914386510849,
+      "num_tokens": 48273532.0,
+      "step": 20870
+    },
+    {
+      "epoch": 0.19143669203263958,
+      "learning_rate": 0.00016171449527826168,
+      "loss": 0.8841,
+      "mean_token_accuracy": 0.7401702165603637,
+      "num_tokens": 48296239.0,
+      "step": 20880
+    },
+    {
+      "epoch": 0.19152837627211883,
+      "learning_rate": 0.00016169615843036582,
+      "loss": 0.8734,
+      "mean_token_accuracy": 0.7406877756118775,
+      "num_tokens": 48319770.0,
+      "step": 20890
+    },
+    {
+      "epoch": 0.19162006051159805,
+      "learning_rate": 0.00016167782158247,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7431667506694793,
+      "num_tokens": 48342994.0,
+      "step": 20900
+    },
+    {
+      "epoch": 0.1917117447510773,
+      "learning_rate": 0.00016165948473457413,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.743863171339035,
+      "num_tokens": 48366530.0,
+      "step": 20910
+    },
+    {
+      "epoch": 0.19180342899055652,
+      "learning_rate": 0.0001616411478866783,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7453465580940246,
+      "num_tokens": 48390329.0,
+      "step": 20920
+    },
+    {
+      "epoch": 0.19189511323003575,
+      "learning_rate": 0.00016162281103878246,
+      "loss": 0.8696,
+      "mean_token_accuracy": 0.7463473320007324,
+      "num_tokens": 48414045.0,
+      "step": 20930
+    },
+    {
+      "epoch": 0.191986797469515,
+      "learning_rate": 0.0001616044741908866,
+      "loss": 0.8403,
+      "mean_token_accuracy": 0.7469229459762573,
+      "num_tokens": 48437275.0,
+      "step": 20940
+    },
+    {
+      "epoch": 0.19207848170899422,
+      "learning_rate": 0.00016158613734299076,
+      "loss": 0.8809,
+      "mean_token_accuracy": 0.739329582452774,
+      "num_tokens": 48460011.0,
+      "step": 20950
+    },
+    {
+      "epoch": 0.19217016594847347,
+      "learning_rate": 0.0001615678004950949,
+      "loss": 0.8458,
+      "mean_token_accuracy": 0.7448213636875153,
+      "num_tokens": 48482742.0,
+      "step": 20960
+    },
+    {
+      "epoch": 0.1922618501879527,
+      "learning_rate": 0.00016154946364719904,
+      "loss": 0.8266,
+      "mean_token_accuracy": 0.7491724967956543,
+      "num_tokens": 48506138.0,
+      "step": 20970
+    },
+    {
+      "epoch": 0.19235353442743192,
+      "learning_rate": 0.0001615311267993032,
+      "loss": 0.8624,
+      "mean_token_accuracy": 0.7481300830841064,
+      "num_tokens": 48529314.0,
+      "step": 20980
+    },
+    {
+      "epoch": 0.19244521866691117,
+      "learning_rate": 0.00016151278995140734,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7472502052783966,
+      "num_tokens": 48552194.0,
+      "step": 20990
+    },
+    {
+      "epoch": 0.1925369029063904,
+      "learning_rate": 0.00016149445310351153,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7430356681346894,
+      "num_tokens": 48575035.0,
+      "step": 21000
+    },
+    {
+      "epoch": 0.1926285871458696,
+      "learning_rate": 0.00016147611625561567,
+      "loss": 0.857,
+      "mean_token_accuracy": 0.7412232458591461,
+      "num_tokens": 48597752.0,
+      "step": 21010
+    },
+    {
+      "epoch": 0.19272027138534886,
+      "learning_rate": 0.0001614577794077198,
+      "loss": 0.8431,
+      "mean_token_accuracy": 0.7491430640220642,
+      "num_tokens": 48620978.0,
+      "step": 21020
+    },
+    {
+      "epoch": 0.19281195562482809,
+      "learning_rate": 0.00016143944255982398,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7437418103218079,
+      "num_tokens": 48643813.0,
+      "step": 21030
+    },
+    {
+      "epoch": 0.19290363986430734,
+      "learning_rate": 0.00016142110571192811,
+      "loss": 0.8528,
+      "mean_token_accuracy": 0.7407807052135468,
+      "num_tokens": 48666585.0,
+      "step": 21040
+    },
+    {
+      "epoch": 0.19299532410378656,
+      "learning_rate": 0.00016140276886403228,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.753812450170517,
+      "num_tokens": 48689187.0,
+      "step": 21050
+    },
+    {
+      "epoch": 0.19308700834326578,
+      "learning_rate": 0.00016138443201613645,
+      "loss": 0.8609,
+      "mean_token_accuracy": 0.7366057932376862,
+      "num_tokens": 48712961.0,
+      "step": 21060
+    },
+    {
+      "epoch": 0.19317869258274503,
+      "learning_rate": 0.00016136609516824058,
+      "loss": 0.8319,
+      "mean_token_accuracy": 0.7512231767177582,
+      "num_tokens": 48735786.0,
+      "step": 21070
+    },
+    {
+      "epoch": 0.19327037682222425,
+      "learning_rate": 0.00016134775832034475,
+      "loss": 0.8723,
+      "mean_token_accuracy": 0.7407830059528351,
+      "num_tokens": 48759045.0,
+      "step": 21080
+    },
+    {
+      "epoch": 0.1933620610617035,
+      "learning_rate": 0.0001613294214724489,
+      "loss": 0.8911,
+      "mean_token_accuracy": 0.7348986148834229,
+      "num_tokens": 48781924.0,
+      "step": 21090
+    },
+    {
+      "epoch": 0.19345374530118273,
+      "learning_rate": 0.00016131108462455305,
+      "loss": 0.8808,
+      "mean_token_accuracy": 0.7405512809753418,
+      "num_tokens": 48805482.0,
+      "step": 21100
+    },
+    {
+      "epoch": 0.19354542954066195,
+      "learning_rate": 0.0001612927477766572,
+      "loss": 0.891,
+      "mean_token_accuracy": 0.7435926377773285,
+      "num_tokens": 48828544.0,
+      "step": 21110
+    },
+    {
+      "epoch": 0.1936371137801412,
+      "learning_rate": 0.00016127441092876136,
+      "loss": 0.8516,
+      "mean_token_accuracy": 0.7438839197158813,
+      "num_tokens": 48851860.0,
+      "step": 21120
+    },
+    {
+      "epoch": 0.19372879801962042,
+      "learning_rate": 0.00016125607408086552,
+      "loss": 0.8439,
+      "mean_token_accuracy": 0.7518651962280274,
+      "num_tokens": 48875109.0,
+      "step": 21130
+    },
+    {
+      "epoch": 0.19382048225909967,
+      "learning_rate": 0.00016123773723296966,
+      "loss": 0.8441,
+      "mean_token_accuracy": 0.740175086259842,
+      "num_tokens": 48897684.0,
+      "step": 21140
+    },
+    {
+      "epoch": 0.1939121664985789,
+      "learning_rate": 0.00016121940038507383,
+      "loss": 0.8888,
+      "mean_token_accuracy": 0.7309944331645966,
+      "num_tokens": 48920214.0,
+      "step": 21150
+    },
+    {
+      "epoch": 0.19400385073805812,
+      "learning_rate": 0.00016120106353717796,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.7339881181716919,
+      "num_tokens": 48943709.0,
+      "step": 21160
+    },
+    {
+      "epoch": 0.19409553497753737,
+      "learning_rate": 0.0001611827266892821,
+      "loss": 0.8611,
+      "mean_token_accuracy": 0.7382303059101105,
+      "num_tokens": 48966728.0,
+      "step": 21170
+    },
+    {
+      "epoch": 0.1941872192170166,
+      "learning_rate": 0.00016116438984138627,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7442708969116211,
+      "num_tokens": 48990006.0,
+      "step": 21180
+    },
+    {
+      "epoch": 0.19427890345649582,
+      "learning_rate": 0.00016114605299349043,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.7375147223472596,
+      "num_tokens": 49013324.0,
+      "step": 21190
+    },
+    {
+      "epoch": 0.19437058769597507,
+      "learning_rate": 0.0001611277161455946,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7424695074558259,
+      "num_tokens": 49036876.0,
+      "step": 21200
+    },
+    {
+      "epoch": 0.1944622719354543,
+      "learning_rate": 0.00016110937929769874,
+      "loss": 0.876,
+      "mean_token_accuracy": 0.733031016588211,
+      "num_tokens": 49059836.0,
+      "step": 21210
+    },
+    {
+      "epoch": 0.19455395617493354,
+      "learning_rate": 0.00016109104244980288,
+      "loss": 0.8763,
+      "mean_token_accuracy": 0.7428466975688934,
+      "num_tokens": 49083313.0,
+      "step": 21220
+    },
+    {
+      "epoch": 0.19464564041441276,
+      "learning_rate": 0.00016107270560190704,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.7427617132663726,
+      "num_tokens": 49106201.0,
+      "step": 21230
+    },
+    {
+      "epoch": 0.19473732465389199,
+      "learning_rate": 0.00016105436875401118,
+      "loss": 0.839,
+      "mean_token_accuracy": 0.7424100399017334,
+      "num_tokens": 49129637.0,
+      "step": 21240
+    },
+    {
+      "epoch": 0.19482900889337124,
+      "learning_rate": 0.00016103603190611534,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7467679858207703,
+      "num_tokens": 49152603.0,
+      "step": 21250
+    },
+    {
+      "epoch": 0.19492069313285046,
+      "learning_rate": 0.0001610176950582195,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7400746285915375,
+      "num_tokens": 49176624.0,
+      "step": 21260
+    },
+    {
+      "epoch": 0.1950123773723297,
+      "learning_rate": 0.00016099935821032365,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7443634688854217,
+      "num_tokens": 49200284.0,
+      "step": 21270
+    },
+    {
+      "epoch": 0.19510406161180893,
+      "learning_rate": 0.00016098102136242781,
+      "loss": 0.8775,
+      "mean_token_accuracy": 0.7443433940410614,
+      "num_tokens": 49223346.0,
+      "step": 21280
+    },
+    {
+      "epoch": 0.19519574585128815,
+      "learning_rate": 0.00016096268451453195,
+      "loss": 0.8679,
+      "mean_token_accuracy": 0.7394611775875092,
+      "num_tokens": 49246308.0,
+      "step": 21290
+    },
+    {
+      "epoch": 0.1952874300907674,
+      "learning_rate": 0.00016094434766663612,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.7369400084018707,
+      "num_tokens": 49269531.0,
+      "step": 21300
+    },
+    {
+      "epoch": 0.19537911433024663,
+      "learning_rate": 0.00016092601081874026,
+      "loss": 0.8431,
+      "mean_token_accuracy": 0.7498523473739624,
+      "num_tokens": 49292741.0,
+      "step": 21310
+    },
+    {
+      "epoch": 0.19547079856972585,
+      "learning_rate": 0.00016090767397084442,
+      "loss": 0.923,
+      "mean_token_accuracy": 0.7297579348087311,
+      "num_tokens": 49315473.0,
+      "step": 21320
+    },
+    {
+      "epoch": 0.1955624828092051,
+      "learning_rate": 0.0001608893371229486,
+      "loss": 0.8284,
+      "mean_token_accuracy": 0.7490214943885803,
+      "num_tokens": 49338023.0,
+      "step": 21330
+    },
+    {
+      "epoch": 0.19565416704868432,
+      "learning_rate": 0.00016087100027505272,
+      "loss": 0.8693,
+      "mean_token_accuracy": 0.740743488073349,
+      "num_tokens": 49360660.0,
+      "step": 21340
+    },
+    {
+      "epoch": 0.19574585128816357,
+      "learning_rate": 0.0001608526634271569,
+      "loss": 0.8527,
+      "mean_token_accuracy": 0.7407018840312958,
+      "num_tokens": 49384019.0,
+      "step": 21350
+    },
+    {
+      "epoch": 0.1958375355276428,
+      "learning_rate": 0.00016083432657926103,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7416827857494355,
+      "num_tokens": 49407251.0,
+      "step": 21360
+    },
+    {
+      "epoch": 0.19592921976712202,
+      "learning_rate": 0.00016081598973136517,
+      "loss": 0.8749,
+      "mean_token_accuracy": 0.7415823817253113,
+      "num_tokens": 49431233.0,
+      "step": 21370
+    },
+    {
+      "epoch": 0.19602090400660127,
+      "learning_rate": 0.00016079765288346933,
+      "loss": 0.8675,
+      "mean_token_accuracy": 0.7362518429756164,
+      "num_tokens": 49453887.0,
+      "step": 21380
+    },
+    {
+      "epoch": 0.1961125882460805,
+      "learning_rate": 0.0001607793160355735,
+      "loss": 0.8628,
+      "mean_token_accuracy": 0.7395012438297272,
+      "num_tokens": 49476830.0,
+      "step": 21390
+    },
+    {
+      "epoch": 0.19620427248555974,
+      "learning_rate": 0.00016076097918767766,
+      "loss": 0.887,
+      "mean_token_accuracy": 0.7354326903820038,
+      "num_tokens": 49500141.0,
+      "step": 21400
+    },
+    {
+      "epoch": 0.19629595672503897,
+      "learning_rate": 0.0001607426423397818,
+      "loss": 0.8573,
+      "mean_token_accuracy": 0.7437665939331055,
+      "num_tokens": 49524576.0,
+      "step": 21410
+    },
+    {
+      "epoch": 0.1963876409645182,
+      "learning_rate": 0.00016072430549188594,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7397708415985107,
+      "num_tokens": 49547077.0,
+      "step": 21420
+    },
+    {
+      "epoch": 0.19647932520399744,
+      "learning_rate": 0.0001607059686439901,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7386774241924285,
+      "num_tokens": 49569078.0,
+      "step": 21430
+    },
+    {
+      "epoch": 0.19657100944347666,
+      "learning_rate": 0.00016068763179609424,
+      "loss": 0.8996,
+      "mean_token_accuracy": 0.7338386654853821,
+      "num_tokens": 49591194.0,
+      "step": 21440
+    },
+    {
+      "epoch": 0.1966626936829559,
+      "learning_rate": 0.00016066929494819844,
+      "loss": 0.887,
+      "mean_token_accuracy": 0.7388228833675384,
+      "num_tokens": 49614734.0,
+      "step": 21450
+    },
+    {
+      "epoch": 0.19675437792243514,
+      "learning_rate": 0.00016065095810030257,
+      "loss": 0.8207,
+      "mean_token_accuracy": 0.7569953620433807,
+      "num_tokens": 49637603.0,
+      "step": 21460
+    },
+    {
+      "epoch": 0.19684606216191436,
+      "learning_rate": 0.0001606326212524067,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.7382251560688019,
+      "num_tokens": 49660518.0,
+      "step": 21470
+    },
+    {
+      "epoch": 0.1969377464013936,
+      "learning_rate": 0.00016061428440451088,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7341851830482483,
+      "num_tokens": 49683704.0,
+      "step": 21480
+    },
+    {
+      "epoch": 0.19702943064087283,
+      "learning_rate": 0.00016059594755661502,
+      "loss": 0.8925,
+      "mean_token_accuracy": 0.7335306823253631,
+      "num_tokens": 49707128.0,
+      "step": 21490
+    },
+    {
+      "epoch": 0.19712111488035206,
+      "learning_rate": 0.00016057761070871918,
+      "loss": 0.8705,
+      "mean_token_accuracy": 0.7388100624084473,
+      "num_tokens": 49730635.0,
+      "step": 21500
+    },
+    {
+      "epoch": 0.1972127991198313,
+      "learning_rate": 0.00016055927386082332,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.7437248170375824,
+      "num_tokens": 49753763.0,
+      "step": 21510
+    },
+    {
+      "epoch": 0.19730448335931053,
+      "learning_rate": 0.00016054093701292749,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.7463112950325013,
+      "num_tokens": 49777325.0,
+      "step": 21520
+    },
+    {
+      "epoch": 0.19739616759878978,
+      "learning_rate": 0.00016052260016503165,
+      "loss": 0.8478,
+      "mean_token_accuracy": 0.7517745077610016,
+      "num_tokens": 49799946.0,
+      "step": 21530
+    },
+    {
+      "epoch": 0.197487851838269,
+      "learning_rate": 0.0001605042633171358,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.7430775046348572,
+      "num_tokens": 49823209.0,
+      "step": 21540
+    },
+    {
+      "epoch": 0.19757953607774822,
+      "learning_rate": 0.00016048592646923995,
+      "loss": 0.8909,
+      "mean_token_accuracy": 0.7399197399616242,
+      "num_tokens": 49846491.0,
+      "step": 21550
+    },
+    {
+      "epoch": 0.19767122031722748,
+      "learning_rate": 0.0001604675896213441,
+      "loss": 0.8296,
+      "mean_token_accuracy": 0.7475386679172515,
+      "num_tokens": 49869883.0,
+      "step": 21560
+    },
+    {
+      "epoch": 0.1977629045567067,
+      "learning_rate": 0.00016044925277344823,
+      "loss": 0.8906,
+      "mean_token_accuracy": 0.7356217682361603,
+      "num_tokens": 49893339.0,
+      "step": 21570
+    },
+    {
+      "epoch": 0.19785458879618595,
+      "learning_rate": 0.0001604309159255524,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7522357940673828,
+      "num_tokens": 49915948.0,
+      "step": 21580
+    },
+    {
+      "epoch": 0.19794627303566517,
+      "learning_rate": 0.00016041257907765656,
+      "loss": 0.8696,
+      "mean_token_accuracy": 0.7433183133602143,
+      "num_tokens": 49939441.0,
+      "step": 21590
+    },
+    {
+      "epoch": 0.1980379572751444,
+      "learning_rate": 0.00016039424222976073,
+      "loss": 0.875,
+      "mean_token_accuracy": 0.7392898082733155,
+      "num_tokens": 49963324.0,
+      "step": 21600
+    },
+    {
+      "epoch": 0.19812964151462364,
+      "learning_rate": 0.00016037590538186487,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7357822299003601,
+      "num_tokens": 49986258.0,
+      "step": 21610
+    },
+    {
+      "epoch": 0.19822132575410287,
+      "learning_rate": 0.000160357568533969,
+      "loss": 0.8493,
+      "mean_token_accuracy": 0.743507343530655,
+      "num_tokens": 50009885.0,
+      "step": 21620
+    },
+    {
+      "epoch": 0.1983130099935821,
+      "learning_rate": 0.00016033923168607317,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7407045245170594,
+      "num_tokens": 50033198.0,
+      "step": 21630
+    },
+    {
+      "epoch": 0.19840469423306134,
+      "learning_rate": 0.0001603208948381773,
+      "loss": 0.8676,
+      "mean_token_accuracy": 0.7428215622901917,
+      "num_tokens": 50055204.0,
+      "step": 21640
+    },
+    {
+      "epoch": 0.19849637847254056,
+      "learning_rate": 0.0001603025579902815,
+      "loss": 0.8806,
+      "mean_token_accuracy": 0.7379589200019836,
+      "num_tokens": 50078253.0,
+      "step": 21650
+    },
+    {
+      "epoch": 0.1985880627120198,
+      "learning_rate": 0.00016028422114238564,
+      "loss": 0.849,
+      "mean_token_accuracy": 0.7472977638244629,
+      "num_tokens": 50101847.0,
+      "step": 21660
+    },
+    {
+      "epoch": 0.19867974695149904,
+      "learning_rate": 0.00016026588429448978,
+      "loss": 0.9058,
+      "mean_token_accuracy": 0.7337198078632354,
+      "num_tokens": 50125764.0,
+      "step": 21670
+    },
+    {
+      "epoch": 0.19877143119097826,
+      "learning_rate": 0.00016024754744659394,
+      "loss": 0.8531,
+      "mean_token_accuracy": 0.747775810956955,
+      "num_tokens": 50148979.0,
+      "step": 21680
+    },
+    {
+      "epoch": 0.1988631154304575,
+      "learning_rate": 0.00016022921059869808,
+      "loss": 0.8592,
+      "mean_token_accuracy": 0.7430618703365326,
+      "num_tokens": 50171941.0,
+      "step": 21690
+    },
+    {
+      "epoch": 0.19895479966993673,
+      "learning_rate": 0.00016021087375080225,
+      "loss": 0.8623,
+      "mean_token_accuracy": 0.7450705766677856,
+      "num_tokens": 50195318.0,
+      "step": 21700
+    },
+    {
+      "epoch": 0.19904648390941598,
+      "learning_rate": 0.00016019253690290638,
+      "loss": 0.8357,
+      "mean_token_accuracy": 0.7473804354667664,
+      "num_tokens": 50219248.0,
+      "step": 21710
+    },
+    {
+      "epoch": 0.1991381681488952,
+      "learning_rate": 0.00016017420005501055,
+      "loss": 0.8357,
+      "mean_token_accuracy": 0.7460615515708924,
+      "num_tokens": 50241914.0,
+      "step": 21720
+    },
+    {
+      "epoch": 0.19922985238837443,
+      "learning_rate": 0.00016015586320711472,
+      "loss": 0.863,
+      "mean_token_accuracy": 0.7443862438201905,
+      "num_tokens": 50264705.0,
+      "step": 21730
+    },
+    {
+      "epoch": 0.19932153662785368,
+      "learning_rate": 0.00016013752635921885,
+      "loss": 0.808,
+      "mean_token_accuracy": 0.7541323065757751,
+      "num_tokens": 50288411.0,
+      "step": 21740
+    },
+    {
+      "epoch": 0.1994132208673329,
+      "learning_rate": 0.00016011918951132302,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.7393250286579132,
+      "num_tokens": 50310524.0,
+      "step": 21750
+    },
+    {
+      "epoch": 0.19950490510681215,
+      "learning_rate": 0.00016010085266342716,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7456794321537018,
+      "num_tokens": 50332865.0,
+      "step": 21760
+    },
+    {
+      "epoch": 0.19959658934629138,
+      "learning_rate": 0.0001600825158155313,
+      "loss": 0.8613,
+      "mean_token_accuracy": 0.7444769442081451,
+      "num_tokens": 50356165.0,
+      "step": 21770
+    },
+    {
+      "epoch": 0.1996882735857706,
+      "learning_rate": 0.0001600641789676355,
+      "loss": 0.8371,
+      "mean_token_accuracy": 0.7507792174816131,
+      "num_tokens": 50378823.0,
+      "step": 21780
+    },
+    {
+      "epoch": 0.19977995782524985,
+      "learning_rate": 0.00016004584211973963,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7494022369384765,
+      "num_tokens": 50401408.0,
+      "step": 21790
+    },
+    {
+      "epoch": 0.19987164206472907,
+      "learning_rate": 0.0001600275052718438,
+      "loss": 0.8709,
+      "mean_token_accuracy": 0.7419447243213654,
+      "num_tokens": 50424655.0,
+      "step": 21800
+    },
+    {
+      "epoch": 0.1999633263042083,
+      "learning_rate": 0.00016000916842394793,
+      "loss": 0.8819,
+      "mean_token_accuracy": 0.7361635088920593,
+      "num_tokens": 50448104.0,
+      "step": 21810
+    },
+    {
+      "epoch": 0.20005501054368754,
+      "learning_rate": 0.00015999083157605207,
+      "loss": 0.8324,
+      "mean_token_accuracy": 0.7455726146697998,
+      "num_tokens": 50471452.0,
+      "step": 21820
+    },
+    {
+      "epoch": 0.20014669478316677,
+      "learning_rate": 0.00015997249472815623,
+      "loss": 0.8786,
+      "mean_token_accuracy": 0.7404028177261353,
+      "num_tokens": 50494949.0,
+      "step": 21830
+    },
+    {
+      "epoch": 0.20023837902264602,
+      "learning_rate": 0.00015995415788026037,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7435766935348511,
+      "num_tokens": 50518380.0,
+      "step": 21840
+    },
+    {
+      "epoch": 0.20033006326212524,
+      "learning_rate": 0.00015993582103236456,
+      "loss": 0.8832,
+      "mean_token_accuracy": 0.7396993160247802,
+      "num_tokens": 50541556.0,
+      "step": 21850
+    },
+    {
+      "epoch": 0.20042174750160446,
+      "learning_rate": 0.0001599174841844687,
+      "loss": 0.8682,
+      "mean_token_accuracy": 0.7403628826141357,
+      "num_tokens": 50565156.0,
+      "step": 21860
+    },
+    {
+      "epoch": 0.20051343174108371,
+      "learning_rate": 0.00015989914733657284,
+      "loss": 0.808,
+      "mean_token_accuracy": 0.7503013730049133,
+      "num_tokens": 50588472.0,
+      "step": 21870
+    },
+    {
+      "epoch": 0.20060511598056294,
+      "learning_rate": 0.000159880810488677,
+      "loss": 0.8249,
+      "mean_token_accuracy": 0.7554940998554229,
+      "num_tokens": 50611909.0,
+      "step": 21880
+    },
+    {
+      "epoch": 0.2006968002200422,
+      "learning_rate": 0.00015986247364078115,
+      "loss": 0.8751,
+      "mean_token_accuracy": 0.7415297389030456,
+      "num_tokens": 50635149.0,
+      "step": 21890
+    },
+    {
+      "epoch": 0.2007884844595214,
+      "learning_rate": 0.0001598441367928853,
+      "loss": 0.854,
+      "mean_token_accuracy": 0.7425120592117309,
+      "num_tokens": 50658808.0,
+      "step": 21900
+    },
+    {
+      "epoch": 0.20088016869900063,
+      "learning_rate": 0.00015982579994498948,
+      "loss": 0.883,
+      "mean_token_accuracy": 0.7333277583122253,
+      "num_tokens": 50682479.0,
+      "step": 21910
+    },
+    {
+      "epoch": 0.20097185293847988,
+      "learning_rate": 0.00015980746309709361,
+      "loss": 0.8342,
+      "mean_token_accuracy": 0.7389296174049378,
+      "num_tokens": 50705643.0,
+      "step": 21920
+    },
+    {
+      "epoch": 0.2010635371779591,
+      "learning_rate": 0.00015978912624919778,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7412992298603058,
+      "num_tokens": 50729130.0,
+      "step": 21930
+    },
+    {
+      "epoch": 0.20115522141743833,
+      "learning_rate": 0.00015977078940130192,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.7428201913833619,
+      "num_tokens": 50752283.0,
+      "step": 21940
+    },
+    {
+      "epoch": 0.20124690565691758,
+      "learning_rate": 0.00015975245255340608,
+      "loss": 0.8699,
+      "mean_token_accuracy": 0.7395201802253724,
+      "num_tokens": 50775140.0,
+      "step": 21950
+    },
+    {
+      "epoch": 0.2013385898963968,
+      "learning_rate": 0.00015973411570551022,
+      "loss": 0.8461,
+      "mean_token_accuracy": 0.7436147034168243,
+      "num_tokens": 50798527.0,
+      "step": 21960
+    },
+    {
+      "epoch": 0.20143027413587605,
+      "learning_rate": 0.00015971577885761436,
+      "loss": 0.8824,
+      "mean_token_accuracy": 0.7316768348217011,
+      "num_tokens": 50821745.0,
+      "step": 21970
+    },
+    {
+      "epoch": 0.20152195837535528,
+      "learning_rate": 0.00015969744200971855,
+      "loss": 0.8418,
+      "mean_token_accuracy": 0.7444996953010559,
+      "num_tokens": 50844569.0,
+      "step": 21980
+    },
+    {
+      "epoch": 0.2016136426148345,
+      "learning_rate": 0.0001596791051618227,
+      "loss": 0.8546,
+      "mean_token_accuracy": 0.7404028832912445,
+      "num_tokens": 50866994.0,
+      "step": 21990
+    },
+    {
+      "epoch": 0.20170532685431375,
+      "learning_rate": 0.00015966076831392686,
+      "loss": 0.8609,
+      "mean_token_accuracy": 0.7448885262012481,
+      "num_tokens": 50890498.0,
+      "step": 22000
+    },
+    {
+      "epoch": 0.20179701109379297,
+      "learning_rate": 0.000159642431466031,
+      "loss": 0.8859,
+      "mean_token_accuracy": 0.7408232808113098,
+      "num_tokens": 50913991.0,
+      "step": 22010
+    },
+    {
+      "epoch": 0.20188869533327222,
+      "learning_rate": 0.00015962409461813513,
+      "loss": 0.8137,
+      "mean_token_accuracy": 0.7524222314357758,
+      "num_tokens": 50936393.0,
+      "step": 22020
+    },
+    {
+      "epoch": 0.20198037957275145,
+      "learning_rate": 0.0001596057577702393,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7503765940666198,
+      "num_tokens": 50959777.0,
+      "step": 22030
+    },
+    {
+      "epoch": 0.20207206381223067,
+      "learning_rate": 0.00015958742092234346,
+      "loss": 0.8863,
+      "mean_token_accuracy": 0.7385656297206878,
+      "num_tokens": 50982461.0,
+      "step": 22040
+    },
+    {
+      "epoch": 0.20216374805170992,
+      "learning_rate": 0.00015956908407444763,
+      "loss": 0.8567,
+      "mean_token_accuracy": 0.7457262992858886,
+      "num_tokens": 51005599.0,
+      "step": 22050
+    },
+    {
+      "epoch": 0.20225543229118914,
+      "learning_rate": 0.00015955074722655177,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7431886255741119,
+      "num_tokens": 51027794.0,
+      "step": 22060
+    },
+    {
+      "epoch": 0.2023471165306684,
+      "learning_rate": 0.0001595324103786559,
+      "loss": 0.8501,
+      "mean_token_accuracy": 0.7480876386165619,
+      "num_tokens": 51050893.0,
+      "step": 22070
+    },
+    {
+      "epoch": 0.20243880077014761,
+      "learning_rate": 0.00015951407353076007,
+      "loss": 0.902,
+      "mean_token_accuracy": 0.7377268075942993,
+      "num_tokens": 51074288.0,
+      "step": 22080
+    },
+    {
+      "epoch": 0.20253048500962684,
+      "learning_rate": 0.0001594957366828642,
+      "loss": 0.8679,
+      "mean_token_accuracy": 0.7407731413841248,
+      "num_tokens": 51097035.0,
+      "step": 22090
+    },
+    {
+      "epoch": 0.2026221692491061,
+      "learning_rate": 0.00015947739983496838,
+      "loss": 0.8568,
+      "mean_token_accuracy": 0.7497326552867889,
+      "num_tokens": 51120737.0,
+      "step": 22100
+    },
+    {
+      "epoch": 0.2027138534885853,
+      "learning_rate": 0.00015945906298707254,
+      "loss": 0.8569,
+      "mean_token_accuracy": 0.749122929573059,
+      "num_tokens": 51143780.0,
+      "step": 22110
+    },
+    {
+      "epoch": 0.20280553772806453,
+      "learning_rate": 0.00015944072613917668,
+      "loss": 0.8325,
+      "mean_token_accuracy": 0.744272917509079,
+      "num_tokens": 51166455.0,
+      "step": 22120
+    },
+    {
+      "epoch": 0.20289722196754378,
+      "learning_rate": 0.00015942238929128084,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7408117949962616,
+      "num_tokens": 51189569.0,
+      "step": 22130
+    },
+    {
+      "epoch": 0.202988906207023,
+      "learning_rate": 0.00015940405244338498,
+      "loss": 0.866,
+      "mean_token_accuracy": 0.7365211248397827,
+      "num_tokens": 51212350.0,
+      "step": 22140
+    },
+    {
+      "epoch": 0.20308059044650226,
+      "learning_rate": 0.00015938571559548915,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.7488377571105957,
+      "num_tokens": 51235842.0,
+      "step": 22150
+    },
+    {
+      "epoch": 0.20317227468598148,
+      "learning_rate": 0.00015936737874759329,
+      "loss": 0.8661,
+      "mean_token_accuracy": 0.7402497053146362,
+      "num_tokens": 51259259.0,
+      "step": 22160
+    },
+    {
+      "epoch": 0.2032639589254607,
+      "learning_rate": 0.00015934904189969745,
+      "loss": 0.845,
+      "mean_token_accuracy": 0.7476921737194061,
+      "num_tokens": 51282364.0,
+      "step": 22170
+    },
+    {
+      "epoch": 0.20335564316493995,
+      "learning_rate": 0.00015933070505180162,
+      "loss": 0.8867,
+      "mean_token_accuracy": 0.7382035672664642,
+      "num_tokens": 51305531.0,
+      "step": 22180
+    },
+    {
+      "epoch": 0.20344732740441918,
+      "learning_rate": 0.00015931236820390576,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7433157622814178,
+      "num_tokens": 51329112.0,
+      "step": 22190
+    },
+    {
+      "epoch": 0.20353901164389843,
+      "learning_rate": 0.00015929403135600992,
+      "loss": 0.8683,
+      "mean_token_accuracy": 0.7413956165313721,
+      "num_tokens": 51351873.0,
+      "step": 22200
+    },
+    {
+      "epoch": 0.20363069588337765,
+      "learning_rate": 0.00015927569450811406,
+      "loss": 0.8801,
+      "mean_token_accuracy": 0.7334484934806824,
+      "num_tokens": 51375100.0,
+      "step": 22210
+    },
+    {
+      "epoch": 0.20372238012285687,
+      "learning_rate": 0.0001592573576602182,
+      "loss": 0.836,
+      "mean_token_accuracy": 0.7457320809364318,
+      "num_tokens": 51398266.0,
+      "step": 22220
+    },
+    {
+      "epoch": 0.20381406436233612,
+      "learning_rate": 0.00015923902081232236,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7472280204296112,
+      "num_tokens": 51421399.0,
+      "step": 22230
+    },
+    {
+      "epoch": 0.20390574860181535,
+      "learning_rate": 0.00015922068396442653,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7409771323204041,
+      "num_tokens": 51444216.0,
+      "step": 22240
+    },
+    {
+      "epoch": 0.20399743284129457,
+      "learning_rate": 0.0001592023471165307,
+      "loss": 0.8599,
+      "mean_token_accuracy": 0.7456480145454407,
+      "num_tokens": 51467897.0,
+      "step": 22250
+    },
+    {
+      "epoch": 0.20408911708077382,
+      "learning_rate": 0.00015918401026863483,
+      "loss": 0.9041,
+      "mean_token_accuracy": 0.7344803214073181,
+      "num_tokens": 51491731.0,
+      "step": 22260
+    },
+    {
+      "epoch": 0.20418080132025304,
+      "learning_rate": 0.00015916567342073897,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.7416087329387665,
+      "num_tokens": 51514964.0,
+      "step": 22270
+    },
+    {
+      "epoch": 0.2042724855597323,
+      "learning_rate": 0.00015914733657284314,
+      "loss": 0.8465,
+      "mean_token_accuracy": 0.7430679261684418,
+      "num_tokens": 51538211.0,
+      "step": 22280
+    },
+    {
+      "epoch": 0.20436416979921151,
+      "learning_rate": 0.00015912899972494727,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7441716909408569,
+      "num_tokens": 51561928.0,
+      "step": 22290
+    },
+    {
+      "epoch": 0.20445585403869074,
+      "learning_rate": 0.00015911066287705144,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7427517771720886,
+      "num_tokens": 51584876.0,
+      "step": 22300
+    },
+    {
+      "epoch": 0.20454753827817,
+      "learning_rate": 0.0001590923260291556,
+      "loss": 0.8532,
+      "mean_token_accuracy": 0.7431596398353577,
+      "num_tokens": 51607025.0,
+      "step": 22310
+    },
+    {
+      "epoch": 0.2046392225176492,
+      "learning_rate": 0.00015907398918125974,
+      "loss": 0.8465,
+      "mean_token_accuracy": 0.7427390575408935,
+      "num_tokens": 51630064.0,
+      "step": 22320
+    },
+    {
+      "epoch": 0.20473090675712846,
+      "learning_rate": 0.0001590556523333639,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7487075865268707,
+      "num_tokens": 51653929.0,
+      "step": 22330
+    },
+    {
+      "epoch": 0.20482259099660768,
+      "learning_rate": 0.00015903731548546805,
+      "loss": 0.8777,
+      "mean_token_accuracy": 0.7471191346645355,
+      "num_tokens": 51677171.0,
+      "step": 22340
+    },
+    {
+      "epoch": 0.2049142752360869,
+      "learning_rate": 0.0001590189786375722,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.7524676620960236,
+      "num_tokens": 51699778.0,
+      "step": 22350
+    },
+    {
+      "epoch": 0.20500595947556616,
+      "learning_rate": 0.00015900064178967635,
+      "loss": 0.8833,
+      "mean_token_accuracy": 0.7445943593978882,
+      "num_tokens": 51722784.0,
+      "step": 22360
+    },
+    {
+      "epoch": 0.20509764371504538,
+      "learning_rate": 0.00015898230494178052,
+      "loss": 0.8436,
+      "mean_token_accuracy": 0.7526306509971619,
+      "num_tokens": 51744504.0,
+      "step": 22370
+    },
+    {
+      "epoch": 0.20518932795452463,
+      "learning_rate": 0.00015896396809388468,
+      "loss": 0.8334,
+      "mean_token_accuracy": 0.7482061982154846,
+      "num_tokens": 51768151.0,
+      "step": 22380
+    },
+    {
+      "epoch": 0.20528101219400385,
+      "learning_rate": 0.00015894563124598882,
+      "loss": 0.8553,
+      "mean_token_accuracy": 0.746174567937851,
+      "num_tokens": 51790906.0,
+      "step": 22390
+    },
+    {
+      "epoch": 0.20537269643348308,
+      "learning_rate": 0.00015892729439809299,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7336604475975037,
+      "num_tokens": 51813564.0,
+      "step": 22400
+    },
+    {
+      "epoch": 0.20546438067296233,
+      "learning_rate": 0.00015890895755019712,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.7384875535964965,
+      "num_tokens": 51836941.0,
+      "step": 22410
+    },
+    {
+      "epoch": 0.20555606491244155,
+      "learning_rate": 0.00015889062070230126,
+      "loss": 0.8725,
+      "mean_token_accuracy": 0.7424970269203186,
+      "num_tokens": 51859858.0,
+      "step": 22420
+    },
+    {
+      "epoch": 0.20564774915192077,
+      "learning_rate": 0.00015887228385440545,
+      "loss": 0.8631,
+      "mean_token_accuracy": 0.7439145982265473,
+      "num_tokens": 51883311.0,
+      "step": 22430
+    },
+    {
+      "epoch": 0.20573943339140002,
+      "learning_rate": 0.0001588539470065096,
+      "loss": 0.8685,
+      "mean_token_accuracy": 0.7378161787986756,
+      "num_tokens": 51906335.0,
+      "step": 22440
+    },
+    {
+      "epoch": 0.20583111763087925,
+      "learning_rate": 0.00015883561015861376,
+      "loss": 0.8401,
+      "mean_token_accuracy": 0.7537043929100037,
+      "num_tokens": 51930121.0,
+      "step": 22450
+    },
+    {
+      "epoch": 0.2059228018703585,
+      "learning_rate": 0.0001588172733107179,
+      "loss": 0.8656,
+      "mean_token_accuracy": 0.7449964284896851,
+      "num_tokens": 51952639.0,
+      "step": 22460
+    },
+    {
+      "epoch": 0.20601448610983772,
+      "learning_rate": 0.00015879893646282203,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7512423753738403,
+      "num_tokens": 51975722.0,
+      "step": 22470
+    },
+    {
+      "epoch": 0.20610617034931694,
+      "learning_rate": 0.0001587805996149262,
+      "loss": 0.8424,
+      "mean_token_accuracy": 0.7491206347942352,
+      "num_tokens": 51998867.0,
+      "step": 22480
+    },
+    {
+      "epoch": 0.2061978545887962,
+      "learning_rate": 0.00015876226276703034,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7521736443042755,
+      "num_tokens": 52022136.0,
+      "step": 22490
+    },
+    {
+      "epoch": 0.20628953882827541,
+      "learning_rate": 0.0001587439259191345,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7509084165096283,
+      "num_tokens": 52045455.0,
+      "step": 22500
+    },
+    {
+      "epoch": 0.20638122306775467,
+      "learning_rate": 0.00015872558907123867,
+      "loss": 0.8828,
+      "mean_token_accuracy": 0.7422175109386444,
+      "num_tokens": 52069071.0,
+      "step": 22510
+    },
+    {
+      "epoch": 0.2064729073072339,
+      "learning_rate": 0.0001587072522233428,
+      "loss": 0.8857,
+      "mean_token_accuracy": 0.734280240535736,
+      "num_tokens": 52092525.0,
+      "step": 22520
+    },
+    {
+      "epoch": 0.2065645915467131,
+      "learning_rate": 0.00015868891537544697,
+      "loss": 0.8607,
+      "mean_token_accuracy": 0.746072632074356,
+      "num_tokens": 52115754.0,
+      "step": 22530
+    },
+    {
+      "epoch": 0.20665627578619236,
+      "learning_rate": 0.0001586705785275511,
+      "loss": 0.9128,
+      "mean_token_accuracy": 0.7358653903007507,
+      "num_tokens": 52138986.0,
+      "step": 22540
+    },
+    {
+      "epoch": 0.20674796002567158,
+      "learning_rate": 0.00015865224167965528,
+      "loss": 0.8529,
+      "mean_token_accuracy": 0.743468987941742,
+      "num_tokens": 52161991.0,
+      "step": 22550
+    },
+    {
+      "epoch": 0.2068396442651508,
+      "learning_rate": 0.00015863390483175944,
+      "loss": 0.868,
+      "mean_token_accuracy": 0.7468756318092347,
+      "num_tokens": 52185451.0,
+      "step": 22560
+    },
+    {
+      "epoch": 0.20693132850463006,
+      "learning_rate": 0.00015861556798386358,
+      "loss": 0.8723,
+      "mean_token_accuracy": 0.7450823128223419,
+      "num_tokens": 52208249.0,
+      "step": 22570
+    },
+    {
+      "epoch": 0.20702301274410928,
+      "learning_rate": 0.00015859723113596775,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7449024319648743,
+      "num_tokens": 52231571.0,
+      "step": 22580
+    },
+    {
+      "epoch": 0.20711469698358853,
+      "learning_rate": 0.00015857889428807188,
+      "loss": 0.8236,
+      "mean_token_accuracy": 0.7532026469707489,
+      "num_tokens": 52254458.0,
+      "step": 22590
+    },
+    {
+      "epoch": 0.20720638122306775,
+      "learning_rate": 0.00015856055744017605,
+      "loss": 0.8853,
+      "mean_token_accuracy": 0.7419590651988983,
+      "num_tokens": 52277353.0,
+      "step": 22600
+    },
+    {
+      "epoch": 0.20729806546254698,
+      "learning_rate": 0.0001585422205922802,
+      "loss": 0.8797,
+      "mean_token_accuracy": 0.7334124922752381,
+      "num_tokens": 52300646.0,
+      "step": 22610
+    },
+    {
+      "epoch": 0.20738974970202623,
+      "learning_rate": 0.00015852388374438433,
+      "loss": 0.8706,
+      "mean_token_accuracy": 0.741769301891327,
+      "num_tokens": 52323640.0,
+      "step": 22620
+    },
+    {
+      "epoch": 0.20748143394150545,
+      "learning_rate": 0.00015850554689648852,
+      "loss": 0.8591,
+      "mean_token_accuracy": 0.7348312199115753,
+      "num_tokens": 52347076.0,
+      "step": 22630
+    },
+    {
+      "epoch": 0.2075731181809847,
+      "learning_rate": 0.00015848721004859266,
+      "loss": 0.8921,
+      "mean_token_accuracy": 0.7361427903175354,
+      "num_tokens": 52370014.0,
+      "step": 22640
+    },
+    {
+      "epoch": 0.20766480242046392,
+      "learning_rate": 0.00015846887320069682,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.7475686013698578,
+      "num_tokens": 52393061.0,
+      "step": 22650
+    },
+    {
+      "epoch": 0.20775648665994315,
+      "learning_rate": 0.00015845053635280096,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7397183537483215,
+      "num_tokens": 52415796.0,
+      "step": 22660
+    },
+    {
+      "epoch": 0.2078481708994224,
+      "learning_rate": 0.0001584321995049051,
+      "loss": 0.8597,
+      "mean_token_accuracy": 0.7469383776187897,
+      "num_tokens": 52438316.0,
+      "step": 22670
+    },
+    {
+      "epoch": 0.20793985513890162,
+      "learning_rate": 0.00015841386265700926,
+      "loss": 0.8568,
+      "mean_token_accuracy": 0.7485611915588379,
+      "num_tokens": 52462059.0,
+      "step": 22680
+    },
+    {
+      "epoch": 0.20803153937838087,
+      "learning_rate": 0.00015839552580911343,
+      "loss": 0.8955,
+      "mean_token_accuracy": 0.7418225705623627,
+      "num_tokens": 52484563.0,
+      "step": 22690
+    },
+    {
+      "epoch": 0.2081232236178601,
+      "learning_rate": 0.00015837718896121757,
+      "loss": 0.896,
+      "mean_token_accuracy": 0.732795125246048,
+      "num_tokens": 52507688.0,
+      "step": 22700
+    },
+    {
+      "epoch": 0.20821490785733932,
+      "learning_rate": 0.00015835885211332173,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7425751388072968,
+      "num_tokens": 52531415.0,
+      "step": 22710
+    },
+    {
+      "epoch": 0.20830659209681857,
+      "learning_rate": 0.00015834051526542587,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7430127382278442,
+      "num_tokens": 52554075.0,
+      "step": 22720
+    },
+    {
+      "epoch": 0.2083982763362978,
+      "learning_rate": 0.00015832217841753004,
+      "loss": 0.8284,
+      "mean_token_accuracy": 0.74892458319664,
+      "num_tokens": 52576818.0,
+      "step": 22730
+    },
+    {
+      "epoch": 0.208489960575777,
+      "learning_rate": 0.00015830384156963418,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7368107259273529,
+      "num_tokens": 52600359.0,
+      "step": 22740
+    },
+    {
+      "epoch": 0.20858164481525626,
+      "learning_rate": 0.00015828550472173834,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.7400186717510223,
+      "num_tokens": 52624016.0,
+      "step": 22750
+    },
+    {
+      "epoch": 0.20867332905473548,
+      "learning_rate": 0.0001582671678738425,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7459397435188293,
+      "num_tokens": 52647758.0,
+      "step": 22760
+    },
+    {
+      "epoch": 0.20876501329421474,
+      "learning_rate": 0.00015824883102594665,
+      "loss": 0.8173,
+      "mean_token_accuracy": 0.7541181325912476,
+      "num_tokens": 52670209.0,
+      "step": 22770
+    },
+    {
+      "epoch": 0.20885669753369396,
+      "learning_rate": 0.0001582304941780508,
+      "loss": 0.8528,
+      "mean_token_accuracy": 0.7441991806030274,
+      "num_tokens": 52693886.0,
+      "step": 22780
+    },
+    {
+      "epoch": 0.20894838177317318,
+      "learning_rate": 0.00015821215733015495,
+      "loss": 0.8656,
+      "mean_token_accuracy": 0.7385231554508209,
+      "num_tokens": 52717927.0,
+      "step": 22790
+    },
+    {
+      "epoch": 0.20904006601265243,
+      "learning_rate": 0.00015819382048225911,
+      "loss": 0.8811,
+      "mean_token_accuracy": 0.7368879735469818,
+      "num_tokens": 52740786.0,
+      "step": 22800
+    },
+    {
+      "epoch": 0.20913175025213165,
+      "learning_rate": 0.00015817548363436325,
+      "loss": 0.8763,
+      "mean_token_accuracy": 0.7419000327587127,
+      "num_tokens": 52764240.0,
+      "step": 22810
+    },
+    {
+      "epoch": 0.2092234344916109,
+      "learning_rate": 0.0001581571467864674,
+      "loss": 0.8755,
+      "mean_token_accuracy": 0.7437607824802399,
+      "num_tokens": 52787198.0,
+      "step": 22820
+    },
+    {
+      "epoch": 0.20931511873109013,
+      "learning_rate": 0.00015813880993857158,
+      "loss": 0.8601,
+      "mean_token_accuracy": 0.7427997291088104,
+      "num_tokens": 52810972.0,
+      "step": 22830
+    },
+    {
+      "epoch": 0.20940680297056935,
+      "learning_rate": 0.00015812047309067572,
+      "loss": 0.825,
+      "mean_token_accuracy": 0.7468115806579589,
+      "num_tokens": 52833364.0,
+      "step": 22840
+    },
+    {
+      "epoch": 0.2094984872100486,
+      "learning_rate": 0.0001581021362427799,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7529172241687775,
+      "num_tokens": 52856738.0,
+      "step": 22850
+    },
+    {
+      "epoch": 0.20959017144952782,
+      "learning_rate": 0.00015808379939488403,
+      "loss": 0.8411,
+      "mean_token_accuracy": 0.7520411849021912,
+      "num_tokens": 52879726.0,
+      "step": 22860
+    },
+    {
+      "epoch": 0.20968185568900705,
+      "learning_rate": 0.00015806546254698816,
+      "loss": 0.8556,
+      "mean_token_accuracy": 0.7454691350460052,
+      "num_tokens": 52903323.0,
+      "step": 22870
+    },
+    {
+      "epoch": 0.2097735399284863,
+      "learning_rate": 0.00015804712569909233,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.7444688558578492,
+      "num_tokens": 52927232.0,
+      "step": 22880
+    },
+    {
+      "epoch": 0.20986522416796552,
+      "learning_rate": 0.0001580287888511965,
+      "loss": 0.8174,
+      "mean_token_accuracy": 0.751531845331192,
+      "num_tokens": 52950700.0,
+      "step": 22890
+    },
+    {
+      "epoch": 0.20995690840744477,
+      "learning_rate": 0.00015801045200330063,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7518243432044983,
+      "num_tokens": 52973978.0,
+      "step": 22900
+    },
+    {
+      "epoch": 0.210048592646924,
+      "learning_rate": 0.0001579921151554048,
+      "loss": 0.826,
+      "mean_token_accuracy": 0.7524587571620941,
+      "num_tokens": 52996751.0,
+      "step": 22910
+    },
+    {
+      "epoch": 0.21014027688640322,
+      "learning_rate": 0.00015797377830750894,
+      "loss": 0.8327,
+      "mean_token_accuracy": 0.7447079837322235,
+      "num_tokens": 53020346.0,
+      "step": 22920
+    },
+    {
+      "epoch": 0.21023196112588247,
+      "learning_rate": 0.0001579554414596131,
+      "loss": 0.8493,
+      "mean_token_accuracy": 0.7439808189868927,
+      "num_tokens": 53043552.0,
+      "step": 22930
+    },
+    {
+      "epoch": 0.2103236453653617,
+      "learning_rate": 0.00015793710461171724,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7447343707084656,
+      "num_tokens": 53066439.0,
+      "step": 22940
+    },
+    {
+      "epoch": 0.21041532960484094,
+      "learning_rate": 0.0001579187677638214,
+      "loss": 0.8901,
+      "mean_token_accuracy": 0.7344002246856689,
+      "num_tokens": 53089696.0,
+      "step": 22950
+    },
+    {
+      "epoch": 0.21050701384432016,
+      "learning_rate": 0.00015790043091592557,
+      "loss": 0.8451,
+      "mean_token_accuracy": 0.7429101884365081,
+      "num_tokens": 53112911.0,
+      "step": 22960
+    },
+    {
+      "epoch": 0.21059869808379938,
+      "learning_rate": 0.0001578820940680297,
+      "loss": 0.887,
+      "mean_token_accuracy": 0.7421387135982513,
+      "num_tokens": 53136508.0,
+      "step": 22970
+    },
+    {
+      "epoch": 0.21069038232327864,
+      "learning_rate": 0.00015786375722013388,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7438655614852905,
+      "num_tokens": 53159562.0,
+      "step": 22980
+    },
+    {
+      "epoch": 0.21078206656275786,
+      "learning_rate": 0.000157845420372238,
+      "loss": 0.8742,
+      "mean_token_accuracy": 0.738437169790268,
+      "num_tokens": 53182338.0,
+      "step": 22990
+    },
+    {
+      "epoch": 0.2108737508022371,
+      "learning_rate": 0.00015782708352434218,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7404647469520569,
+      "num_tokens": 53205760.0,
+      "step": 23000
+    },
+    {
+      "epoch": 0.21096543504171633,
+      "learning_rate": 0.00015780874667644632,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7402274906635284,
+      "num_tokens": 53229103.0,
+      "step": 23010
+    },
+    {
+      "epoch": 0.21105711928119555,
+      "learning_rate": 0.00015779040982855048,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.751193743944168,
+      "num_tokens": 53252856.0,
+      "step": 23020
+    },
+    {
+      "epoch": 0.2111488035206748,
+      "learning_rate": 0.00015777207298065465,
+      "loss": 0.8431,
+      "mean_token_accuracy": 0.7468126237392425,
+      "num_tokens": 53276558.0,
+      "step": 23030
+    },
+    {
+      "epoch": 0.21124048776015403,
+      "learning_rate": 0.00015775373613275879,
+      "loss": 0.8539,
+      "mean_token_accuracy": 0.7457605421543121,
+      "num_tokens": 53299878.0,
+      "step": 23040
+    },
+    {
+      "epoch": 0.21133217199963325,
+      "learning_rate": 0.00015773539928486295,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7381194353103637,
+      "num_tokens": 53322991.0,
+      "step": 23050
+    },
+    {
+      "epoch": 0.2114238562391125,
+      "learning_rate": 0.0001577170624369671,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.7466514110565186,
+      "num_tokens": 53346031.0,
+      "step": 23060
+    },
+    {
+      "epoch": 0.21151554047859172,
+      "learning_rate": 0.00015769872558907123,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.748447859287262,
+      "num_tokens": 53368430.0,
+      "step": 23070
+    },
+    {
+      "epoch": 0.21160722471807097,
+      "learning_rate": 0.0001576803887411754,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7409026741981506,
+      "num_tokens": 53391918.0,
+      "step": 23080
+    },
+    {
+      "epoch": 0.2116989089575502,
+      "learning_rate": 0.00015766205189327956,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.7559028267860413,
+      "num_tokens": 53413879.0,
+      "step": 23090
+    },
+    {
+      "epoch": 0.21179059319702942,
+      "learning_rate": 0.0001576437150453837,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7381623983383179,
+      "num_tokens": 53437667.0,
+      "step": 23100
+    },
+    {
+      "epoch": 0.21188227743650867,
+      "learning_rate": 0.00015762537819748786,
+      "loss": 0.7952,
+      "mean_token_accuracy": 0.7536631882190704,
+      "num_tokens": 53461367.0,
+      "step": 23110
+    },
+    {
+      "epoch": 0.2119739616759879,
+      "learning_rate": 0.000157607041349592,
+      "loss": 0.8783,
+      "mean_token_accuracy": 0.7351391434669494,
+      "num_tokens": 53483854.0,
+      "step": 23120
+    },
+    {
+      "epoch": 0.21206564591546714,
+      "learning_rate": 0.00015758870450169617,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7492637693881988,
+      "num_tokens": 53506978.0,
+      "step": 23130
+    },
+    {
+      "epoch": 0.21215733015494637,
+      "learning_rate": 0.0001575703676538003,
+      "loss": 0.8812,
+      "mean_token_accuracy": 0.7430793285369873,
+      "num_tokens": 53529964.0,
+      "step": 23140
+    },
+    {
+      "epoch": 0.2122490143944256,
+      "learning_rate": 0.00015755203080590447,
+      "loss": 0.8998,
+      "mean_token_accuracy": 0.734472805261612,
+      "num_tokens": 53553476.0,
+      "step": 23150
+    },
+    {
+      "epoch": 0.21234069863390484,
+      "learning_rate": 0.00015753369395800864,
+      "loss": 0.9035,
+      "mean_token_accuracy": 0.7372500181198121,
+      "num_tokens": 53576958.0,
+      "step": 23160
+    },
+    {
+      "epoch": 0.21243238287338406,
+      "learning_rate": 0.00015751535711011277,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.7355253279209137,
+      "num_tokens": 53599662.0,
+      "step": 23170
+    },
+    {
+      "epoch": 0.21252406711286329,
+      "learning_rate": 0.00015749702026221694,
+      "loss": 0.8762,
+      "mean_token_accuracy": 0.7412170886993408,
+      "num_tokens": 53622289.0,
+      "step": 23180
+    },
+    {
+      "epoch": 0.21261575135234254,
+      "learning_rate": 0.00015747868341432108,
+      "loss": 0.818,
+      "mean_token_accuracy": 0.7529448926448822,
+      "num_tokens": 53646308.0,
+      "step": 23190
+    },
+    {
+      "epoch": 0.21270743559182176,
+      "learning_rate": 0.00015746034656642524,
+      "loss": 0.8986,
+      "mean_token_accuracy": 0.7288304030895233,
+      "num_tokens": 53669551.0,
+      "step": 23200
+    },
+    {
+      "epoch": 0.212799119831301,
+      "learning_rate": 0.00015744200971852938,
+      "loss": 0.8444,
+      "mean_token_accuracy": 0.7498817265033721,
+      "num_tokens": 53693088.0,
+      "step": 23210
+    },
+    {
+      "epoch": 0.21289080407078023,
+      "learning_rate": 0.00015742367287063355,
+      "loss": 0.8121,
+      "mean_token_accuracy": 0.7579306125640869,
+      "num_tokens": 53715872.0,
+      "step": 23220
+    },
+    {
+      "epoch": 0.21298248831025945,
+      "learning_rate": 0.0001574053360227377,
+      "loss": 0.9312,
+      "mean_token_accuracy": 0.7323280274868011,
+      "num_tokens": 53739238.0,
+      "step": 23230
+    },
+    {
+      "epoch": 0.2130741725497387,
+      "learning_rate": 0.00015738699917484185,
+      "loss": 0.8538,
+      "mean_token_accuracy": 0.7449313700199127,
+      "num_tokens": 53762212.0,
+      "step": 23240
+    },
+    {
+      "epoch": 0.21316585678921793,
+      "learning_rate": 0.00015736866232694602,
+      "loss": 0.8477,
+      "mean_token_accuracy": 0.7526765108108521,
+      "num_tokens": 53785291.0,
+      "step": 23250
+    },
+    {
+      "epoch": 0.21325754102869718,
+      "learning_rate": 0.00015735032547905015,
+      "loss": 0.8426,
+      "mean_token_accuracy": 0.7461187839508057,
+      "num_tokens": 53807700.0,
+      "step": 23260
+    },
+    {
+      "epoch": 0.2133492252681764,
+      "learning_rate": 0.0001573319886311543,
+      "loss": 0.8775,
+      "mean_token_accuracy": 0.7434069573879242,
+      "num_tokens": 53830824.0,
+      "step": 23270
+    },
+    {
+      "epoch": 0.21344090950765562,
+      "learning_rate": 0.00015731365178325849,
+      "loss": 0.9066,
+      "mean_token_accuracy": 0.7361610054969787,
+      "num_tokens": 53855212.0,
+      "step": 23280
+    },
+    {
+      "epoch": 0.21353259374713487,
+      "learning_rate": 0.00015729531493536262,
+      "loss": 0.8844,
+      "mean_token_accuracy": 0.7391054093837738,
+      "num_tokens": 53877991.0,
+      "step": 23290
+    },
+    {
+      "epoch": 0.2136242779866141,
+      "learning_rate": 0.00015727697808746676,
+      "loss": 0.8158,
+      "mean_token_accuracy": 0.7555877804756165,
+      "num_tokens": 53900944.0,
+      "step": 23300
+    },
+    {
+      "epoch": 0.21371596222609335,
+      "learning_rate": 0.00015725864123957093,
+      "loss": 0.8358,
+      "mean_token_accuracy": 0.7458467245101928,
+      "num_tokens": 53923949.0,
+      "step": 23310
+    },
+    {
+      "epoch": 0.21380764646557257,
+      "learning_rate": 0.00015724030439167507,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7416692852973938,
+      "num_tokens": 53946905.0,
+      "step": 23320
+    },
+    {
+      "epoch": 0.2138993307050518,
+      "learning_rate": 0.00015722196754377923,
+      "loss": 0.8507,
+      "mean_token_accuracy": 0.7485923886299133,
+      "num_tokens": 53969557.0,
+      "step": 23330
+    },
+    {
+      "epoch": 0.21399101494453104,
+      "learning_rate": 0.00015720363069588337,
+      "loss": 0.8635,
+      "mean_token_accuracy": 0.7405233025550843,
+      "num_tokens": 53993251.0,
+      "step": 23340
+    },
+    {
+      "epoch": 0.21408269918401027,
+      "learning_rate": 0.00015718529384798753,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7482027769088745,
+      "num_tokens": 54016482.0,
+      "step": 23350
+    },
+    {
+      "epoch": 0.2141743834234895,
+      "learning_rate": 0.0001571669570000917,
+      "loss": 0.8301,
+      "mean_token_accuracy": 0.7466087222099305,
+      "num_tokens": 54039231.0,
+      "step": 23360
+    },
+    {
+      "epoch": 0.21426606766296874,
+      "learning_rate": 0.00015714862015219584,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.7439805567264557,
+      "num_tokens": 54062281.0,
+      "step": 23370
+    },
+    {
+      "epoch": 0.21435775190244796,
+      "learning_rate": 0.0001571302833043,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7457484066486358,
+      "num_tokens": 54085035.0,
+      "step": 23380
+    },
+    {
+      "epoch": 0.2144494361419272,
+      "learning_rate": 0.00015711194645640414,
+      "loss": 0.8878,
+      "mean_token_accuracy": 0.7381054818630218,
+      "num_tokens": 54108892.0,
+      "step": 23390
+    },
+    {
+      "epoch": 0.21454112038140644,
+      "learning_rate": 0.0001570936096085083,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7365092277526856,
+      "num_tokens": 54131651.0,
+      "step": 23400
+    },
+    {
+      "epoch": 0.21463280462088566,
+      "learning_rate": 0.00015707527276061247,
+      "loss": 0.8684,
+      "mean_token_accuracy": 0.7431603789329528,
+      "num_tokens": 54154336.0,
+      "step": 23410
+    },
+    {
+      "epoch": 0.2147244888603649,
+      "learning_rate": 0.0001570569359127166,
+      "loss": 0.883,
+      "mean_token_accuracy": 0.7376543760299683,
+      "num_tokens": 54177580.0,
+      "step": 23420
+    },
+    {
+      "epoch": 0.21481617309984413,
+      "learning_rate": 0.00015703859906482078,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7387879192829132,
+      "num_tokens": 54201126.0,
+      "step": 23430
+    },
+    {
+      "epoch": 0.21490785733932338,
+      "learning_rate": 0.00015702026221692492,
+      "loss": 0.8437,
+      "mean_token_accuracy": 0.748453801870346,
+      "num_tokens": 54224885.0,
+      "step": 23440
+    },
+    {
+      "epoch": 0.2149995415788026,
+      "learning_rate": 0.00015700192536902908,
+      "loss": 0.8623,
+      "mean_token_accuracy": 0.7423072993755341,
+      "num_tokens": 54247480.0,
+      "step": 23450
+    },
+    {
+      "epoch": 0.21509122581828183,
+      "learning_rate": 0.00015698358852113322,
+      "loss": 0.8959,
+      "mean_token_accuracy": 0.7366766691207886,
+      "num_tokens": 54269692.0,
+      "step": 23460
+    },
+    {
+      "epoch": 0.21518291005776108,
+      "learning_rate": 0.00015696525167323736,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.737098115682602,
+      "num_tokens": 54292968.0,
+      "step": 23470
+    },
+    {
+      "epoch": 0.2152745942972403,
+      "learning_rate": 0.00015694691482534155,
+      "loss": 0.8866,
+      "mean_token_accuracy": 0.7405371010303498,
+      "num_tokens": 54316741.0,
+      "step": 23480
+    },
+    {
+      "epoch": 0.21536627853671952,
+      "learning_rate": 0.0001569285779774457,
+      "loss": 0.8566,
+      "mean_token_accuracy": 0.7423771560192108,
+      "num_tokens": 54339565.0,
+      "step": 23490
+    },
+    {
+      "epoch": 0.21545796277619877,
+      "learning_rate": 0.00015691024112954983,
+      "loss": 0.8331,
+      "mean_token_accuracy": 0.7455996751785279,
+      "num_tokens": 54363044.0,
+      "step": 23500
+    },
+    {
+      "epoch": 0.215549647015678,
+      "learning_rate": 0.000156891904281654,
+      "loss": 0.8933,
+      "mean_token_accuracy": 0.7349541306495666,
+      "num_tokens": 54385937.0,
+      "step": 23510
+    },
+    {
+      "epoch": 0.21564133125515725,
+      "learning_rate": 0.00015687356743375813,
+      "loss": 0.8642,
+      "mean_token_accuracy": 0.7499670505523681,
+      "num_tokens": 54408753.0,
+      "step": 23520
+    },
+    {
+      "epoch": 0.21573301549463647,
+      "learning_rate": 0.0001568552305858623,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.7448411345481872,
+      "num_tokens": 54431866.0,
+      "step": 23530
+    },
+    {
+      "epoch": 0.2158246997341157,
+      "learning_rate": 0.00015683689373796646,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.747662615776062,
+      "num_tokens": 54454634.0,
+      "step": 23540
+    },
+    {
+      "epoch": 0.21591638397359494,
+      "learning_rate": 0.0001568185568900706,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.744737309217453,
+      "num_tokens": 54477758.0,
+      "step": 23550
+    },
+    {
+      "epoch": 0.21600806821307417,
+      "learning_rate": 0.00015680022004217476,
+      "loss": 0.8384,
+      "mean_token_accuracy": 0.7470738470554352,
+      "num_tokens": 54501201.0,
+      "step": 23560
+    },
+    {
+      "epoch": 0.21609975245255342,
+      "learning_rate": 0.0001567818831942789,
+      "loss": 0.8633,
+      "mean_token_accuracy": 0.7414463341236115,
+      "num_tokens": 54524678.0,
+      "step": 23570
+    },
+    {
+      "epoch": 0.21619143669203264,
+      "learning_rate": 0.00015676354634638307,
+      "loss": 0.8578,
+      "mean_token_accuracy": 0.7483405590057373,
+      "num_tokens": 54547753.0,
+      "step": 23580
+    },
+    {
+      "epoch": 0.21628312093151186,
+      "learning_rate": 0.0001567452094984872,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7444630563259125,
+      "num_tokens": 54570375.0,
+      "step": 23590
+    },
+    {
+      "epoch": 0.2163748051709911,
+      "learning_rate": 0.00015672687265059137,
+      "loss": 0.8685,
+      "mean_token_accuracy": 0.742591780424118,
+      "num_tokens": 54593404.0,
+      "step": 23600
+    },
+    {
+      "epoch": 0.21646648941047034,
+      "learning_rate": 0.00015670853580269554,
+      "loss": 0.857,
+      "mean_token_accuracy": 0.7420598983764648,
+      "num_tokens": 54617124.0,
+      "step": 23610
+    },
+    {
+      "epoch": 0.2165581736499496,
+      "learning_rate": 0.00015669019895479968,
+      "loss": 0.8402,
+      "mean_token_accuracy": 0.7527478635311127,
+      "num_tokens": 54640572.0,
+      "step": 23620
+    },
+    {
+      "epoch": 0.2166498578894288,
+      "learning_rate": 0.00015667186210690384,
+      "loss": 0.8273,
+      "mean_token_accuracy": 0.7498106837272644,
+      "num_tokens": 54664242.0,
+      "step": 23630
+    },
+    {
+      "epoch": 0.21674154212890803,
+      "learning_rate": 0.00015665352525900798,
+      "loss": 0.8275,
+      "mean_token_accuracy": 0.7492702424526214,
+      "num_tokens": 54687693.0,
+      "step": 23640
+    },
+    {
+      "epoch": 0.21683322636838728,
+      "learning_rate": 0.00015663518841111215,
+      "loss": 0.8802,
+      "mean_token_accuracy": 0.7409344494342804,
+      "num_tokens": 54711150.0,
+      "step": 23650
+    },
+    {
+      "epoch": 0.2169249106078665,
+      "learning_rate": 0.00015661685156321628,
+      "loss": 0.8867,
+      "mean_token_accuracy": 0.7304402410984039,
+      "num_tokens": 54733747.0,
+      "step": 23660
+    },
+    {
+      "epoch": 0.21701659484734573,
+      "learning_rate": 0.00015659851471532045,
+      "loss": 0.857,
+      "mean_token_accuracy": 0.747049993276596,
+      "num_tokens": 54756418.0,
+      "step": 23670
+    },
+    {
+      "epoch": 0.21710827908682498,
+      "learning_rate": 0.00015658017786742461,
+      "loss": 0.8784,
+      "mean_token_accuracy": 0.7382007956504821,
+      "num_tokens": 54779656.0,
+      "step": 23680
+    },
+    {
+      "epoch": 0.2171999633263042,
+      "learning_rate": 0.00015656184101952875,
+      "loss": 0.89,
+      "mean_token_accuracy": 0.7308126807212829,
+      "num_tokens": 54803066.0,
+      "step": 23690
+    },
+    {
+      "epoch": 0.21729164756578345,
+      "learning_rate": 0.0001565435041716329,
+      "loss": 0.8903,
+      "mean_token_accuracy": 0.7373201549053192,
+      "num_tokens": 54826372.0,
+      "step": 23700
+    },
+    {
+      "epoch": 0.21738333180526268,
+      "learning_rate": 0.00015652516732373706,
+      "loss": 0.8808,
+      "mean_token_accuracy": 0.7408906638622283,
+      "num_tokens": 54850113.0,
+      "step": 23710
+    },
+    {
+      "epoch": 0.2174750160447419,
+      "learning_rate": 0.0001565068304758412,
+      "loss": 0.8341,
+      "mean_token_accuracy": 0.7525183320045471,
+      "num_tokens": 54872989.0,
+      "step": 23720
+    },
+    {
+      "epoch": 0.21756670028422115,
+      "learning_rate": 0.00015648849362794536,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7476661264896393,
+      "num_tokens": 54896134.0,
+      "step": 23730
+    },
+    {
+      "epoch": 0.21765838452370037,
+      "learning_rate": 0.00015647015678004953,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.750644987821579,
+      "num_tokens": 54918632.0,
+      "step": 23740
+    },
+    {
+      "epoch": 0.21775006876317962,
+      "learning_rate": 0.00015645181993215366,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7441705107688904,
+      "num_tokens": 54941866.0,
+      "step": 23750
+    },
+    {
+      "epoch": 0.21784175300265884,
+      "learning_rate": 0.00015643348308425783,
+      "loss": 0.8242,
+      "mean_token_accuracy": 0.7528449773788453,
+      "num_tokens": 54964650.0,
+      "step": 23760
+    },
+    {
+      "epoch": 0.21793343724213807,
+      "learning_rate": 0.00015641514623636197,
+      "loss": 0.8757,
+      "mean_token_accuracy": 0.7425302565097809,
+      "num_tokens": 54987828.0,
+      "step": 23770
+    },
+    {
+      "epoch": 0.21802512148161732,
+      "learning_rate": 0.00015639680938846613,
+      "loss": 0.8637,
+      "mean_token_accuracy": 0.7403858184814454,
+      "num_tokens": 55011006.0,
+      "step": 23780
+    },
+    {
+      "epoch": 0.21811680572109654,
+      "learning_rate": 0.00015637847254057027,
+      "loss": 0.8804,
+      "mean_token_accuracy": 0.7361753046512604,
+      "num_tokens": 55033867.0,
+      "step": 23790
+    },
+    {
+      "epoch": 0.21820848996057576,
+      "learning_rate": 0.00015636013569267444,
+      "loss": 0.8673,
+      "mean_token_accuracy": 0.7374181807041168,
+      "num_tokens": 55057010.0,
+      "step": 23800
+    },
+    {
+      "epoch": 0.218300174200055,
+      "learning_rate": 0.0001563417988447786,
+      "loss": 0.8607,
+      "mean_token_accuracy": 0.7394064128398895,
+      "num_tokens": 55079738.0,
+      "step": 23810
+    },
+    {
+      "epoch": 0.21839185843953424,
+      "learning_rate": 0.00015632346199688274,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.7540367722511292,
+      "num_tokens": 55102910.0,
+      "step": 23820
+    },
+    {
+      "epoch": 0.2184835426790135,
+      "learning_rate": 0.0001563051251489869,
+      "loss": 0.8661,
+      "mean_token_accuracy": 0.7424493491649627,
+      "num_tokens": 55126616.0,
+      "step": 23830
+    },
+    {
+      "epoch": 0.2185752269184927,
+      "learning_rate": 0.00015628678830109104,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7432507991790771,
+      "num_tokens": 55150326.0,
+      "step": 23840
+    },
+    {
+      "epoch": 0.21866691115797193,
+      "learning_rate": 0.0001562684514531952,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7417873263359069,
+      "num_tokens": 55173701.0,
+      "step": 23850
+    },
+    {
+      "epoch": 0.21875859539745118,
+      "learning_rate": 0.00015625011460529935,
+      "loss": 0.8874,
+      "mean_token_accuracy": 0.7332236111164093,
+      "num_tokens": 55197126.0,
+      "step": 23860
+    },
+    {
+      "epoch": 0.2188502796369304,
+      "learning_rate": 0.0001562317777574035,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7481783926486969,
+      "num_tokens": 55220566.0,
+      "step": 23870
+    },
+    {
+      "epoch": 0.21894196387640966,
+      "learning_rate": 0.00015621344090950768,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.7439969122409821,
+      "num_tokens": 55243504.0,
+      "step": 23880
+    },
+    {
+      "epoch": 0.21903364811588888,
+      "learning_rate": 0.00015619510406161182,
+      "loss": 0.8057,
+      "mean_token_accuracy": 0.7523382008075714,
+      "num_tokens": 55267123.0,
+      "step": 23890
+    },
+    {
+      "epoch": 0.2191253323553681,
+      "learning_rate": 0.00015617676721371596,
+      "loss": 0.8726,
+      "mean_token_accuracy": 0.7446614146232605,
+      "num_tokens": 55291060.0,
+      "step": 23900
+    },
+    {
+      "epoch": 0.21921701659484735,
+      "learning_rate": 0.00015615843036582012,
+      "loss": 0.89,
+      "mean_token_accuracy": 0.7387046813964844,
+      "num_tokens": 55314005.0,
+      "step": 23910
+    },
+    {
+      "epoch": 0.21930870083432658,
+      "learning_rate": 0.00015614009351792426,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.7483787298202514,
+      "num_tokens": 55336522.0,
+      "step": 23920
+    },
+    {
+      "epoch": 0.21940038507380583,
+      "learning_rate": 0.00015612175667002845,
+      "loss": 0.89,
+      "mean_token_accuracy": 0.728826767206192,
+      "num_tokens": 55360200.0,
+      "step": 23930
+    },
+    {
+      "epoch": 0.21949206931328505,
+      "learning_rate": 0.0001561034198221326,
+      "loss": 0.8282,
+      "mean_token_accuracy": 0.7571542203426361,
+      "num_tokens": 55383552.0,
+      "step": 23940
+    },
+    {
+      "epoch": 0.21958375355276427,
+      "learning_rate": 0.00015608508297423673,
+      "loss": 0.9024,
+      "mean_token_accuracy": 0.7327570796012879,
+      "num_tokens": 55406816.0,
+      "step": 23950
+    },
+    {
+      "epoch": 0.21967543779224352,
+      "learning_rate": 0.0001560667461263409,
+      "loss": 0.8751,
+      "mean_token_accuracy": 0.7369073569774628,
+      "num_tokens": 55429402.0,
+      "step": 23960
+    },
+    {
+      "epoch": 0.21976712203172274,
+      "learning_rate": 0.00015604840927844503,
+      "loss": 0.8308,
+      "mean_token_accuracy": 0.753199428319931,
+      "num_tokens": 55452121.0,
+      "step": 23970
+    },
+    {
+      "epoch": 0.21985880627120197,
+      "learning_rate": 0.0001560300724305492,
+      "loss": 0.8376,
+      "mean_token_accuracy": 0.750776594877243,
+      "num_tokens": 55476260.0,
+      "step": 23980
+    },
+    {
+      "epoch": 0.21995049051068122,
+      "learning_rate": 0.00015601173558265334,
+      "loss": 0.8573,
+      "mean_token_accuracy": 0.7445099115371704,
+      "num_tokens": 55499888.0,
+      "step": 23990
+    },
+    {
+      "epoch": 0.22004217475016044,
+      "learning_rate": 0.0001559933987347575,
+      "loss": 0.9013,
+      "mean_token_accuracy": 0.7359369695186615,
+      "num_tokens": 55522509.0,
+      "step": 24000
+    },
+    {
+      "epoch": 0.2201338589896397,
+      "learning_rate": 0.00015597506188686167,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.75285684466362,
+      "num_tokens": 55545077.0,
+      "step": 24010
+    },
+    {
+      "epoch": 0.22022554322911891,
+      "learning_rate": 0.0001559567250389658,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7412333250045776,
+      "num_tokens": 55567446.0,
+      "step": 24020
+    },
+    {
+      "epoch": 0.22031722746859814,
+      "learning_rate": 0.00015593838819106997,
+      "loss": 0.8819,
+      "mean_token_accuracy": 0.7352995276451111,
+      "num_tokens": 55590440.0,
+      "step": 24030
+    },
+    {
+      "epoch": 0.2204089117080774,
+      "learning_rate": 0.0001559200513431741,
+      "loss": 0.8723,
+      "mean_token_accuracy": 0.7348103106021882,
+      "num_tokens": 55613620.0,
+      "step": 24040
+    },
+    {
+      "epoch": 0.2205005959475566,
+      "learning_rate": 0.00015590171449527827,
+      "loss": 0.8783,
+      "mean_token_accuracy": 0.7398612320423126,
+      "num_tokens": 55637043.0,
+      "step": 24050
+    },
+    {
+      "epoch": 0.22059228018703586,
+      "learning_rate": 0.0001558833776473824,
+      "loss": 0.8474,
+      "mean_token_accuracy": 0.745047253370285,
+      "num_tokens": 55659449.0,
+      "step": 24060
+    },
+    {
+      "epoch": 0.22068396442651508,
+      "learning_rate": 0.00015586504079948658,
+      "loss": 0.887,
+      "mean_token_accuracy": 0.7352611005306244,
+      "num_tokens": 55682019.0,
+      "step": 24070
+    },
+    {
+      "epoch": 0.2207756486659943,
+      "learning_rate": 0.00015584670395159074,
+      "loss": 0.8948,
+      "mean_token_accuracy": 0.7356932163238525,
+      "num_tokens": 55705342.0,
+      "step": 24080
+    },
+    {
+      "epoch": 0.22086733290547356,
+      "learning_rate": 0.00015582836710369488,
+      "loss": 0.8175,
+      "mean_token_accuracy": 0.7523733794689178,
+      "num_tokens": 55729072.0,
+      "step": 24090
+    },
+    {
+      "epoch": 0.22095901714495278,
+      "learning_rate": 0.00015581003025579902,
+      "loss": 0.8956,
+      "mean_token_accuracy": 0.7405926644802093,
+      "num_tokens": 55752285.0,
+      "step": 24100
+    },
+    {
+      "epoch": 0.221050701384432,
+      "learning_rate": 0.00015579169340790319,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7390296995639801,
+      "num_tokens": 55774982.0,
+      "step": 24110
+    },
+    {
+      "epoch": 0.22114238562391125,
+      "learning_rate": 0.00015577335656000732,
+      "loss": 0.8418,
+      "mean_token_accuracy": 0.744842278957367,
+      "num_tokens": 55797962.0,
+      "step": 24120
+    },
+    {
+      "epoch": 0.22123406986339048,
+      "learning_rate": 0.00015575501971211152,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7467646896839142,
+      "num_tokens": 55820234.0,
+      "step": 24130
+    },
+    {
+      "epoch": 0.22132575410286973,
+      "learning_rate": 0.00015573668286421565,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7507136106491089,
+      "num_tokens": 55843204.0,
+      "step": 24140
+    },
+    {
+      "epoch": 0.22141743834234895,
+      "learning_rate": 0.0001557183460163198,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.7322293877601623,
+      "num_tokens": 55866244.0,
+      "step": 24150
+    },
+    {
+      "epoch": 0.22150912258182817,
+      "learning_rate": 0.00015570000916842396,
+      "loss": 0.8653,
+      "mean_token_accuracy": 0.740640789270401,
+      "num_tokens": 55888300.0,
+      "step": 24160
+    },
+    {
+      "epoch": 0.22160080682130742,
+      "learning_rate": 0.0001556816723205281,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7410377204418183,
+      "num_tokens": 55911949.0,
+      "step": 24170
+    },
+    {
+      "epoch": 0.22169249106078665,
+      "learning_rate": 0.00015566333547263226,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7393958389759063,
+      "num_tokens": 55935051.0,
+      "step": 24180
+    },
+    {
+      "epoch": 0.2217841753002659,
+      "learning_rate": 0.0001556449986247364,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7424891471862793,
+      "num_tokens": 55957905.0,
+      "step": 24190
+    },
+    {
+      "epoch": 0.22187585953974512,
+      "learning_rate": 0.00015562666177684057,
+      "loss": 0.8742,
+      "mean_token_accuracy": 0.7418850660324097,
+      "num_tokens": 55981382.0,
+      "step": 24200
+    },
+    {
+      "epoch": 0.22196754377922434,
+      "learning_rate": 0.00015560832492894473,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7461317777633667,
+      "num_tokens": 56004358.0,
+      "step": 24210
+    },
+    {
+      "epoch": 0.2220592280187036,
+      "learning_rate": 0.00015558998808104887,
+      "loss": 0.8637,
+      "mean_token_accuracy": 0.7400383591651917,
+      "num_tokens": 56027254.0,
+      "step": 24220
+    },
+    {
+      "epoch": 0.22215091225818281,
+      "learning_rate": 0.00015557165123315303,
+      "loss": 0.8915,
+      "mean_token_accuracy": 0.7401159465312958,
+      "num_tokens": 56050493.0,
+      "step": 24230
+    },
+    {
+      "epoch": 0.22224259649766206,
+      "learning_rate": 0.00015555331438525717,
+      "loss": 0.8385,
+      "mean_token_accuracy": 0.7505717515945435,
+      "num_tokens": 56074138.0,
+      "step": 24240
+    },
+    {
+      "epoch": 0.2223342807371413,
+      "learning_rate": 0.00015553497753736134,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7438802659511566,
+      "num_tokens": 56097304.0,
+      "step": 24250
+    },
+    {
+      "epoch": 0.2224259649766205,
+      "learning_rate": 0.0001555166406894655,
+      "loss": 0.8582,
+      "mean_token_accuracy": 0.7439675509929657,
+      "num_tokens": 56120300.0,
+      "step": 24260
+    },
+    {
+      "epoch": 0.22251764921609976,
+      "learning_rate": 0.00015549830384156964,
+      "loss": 0.84,
+      "mean_token_accuracy": 0.7437430799007416,
+      "num_tokens": 56143176.0,
+      "step": 24270
+    },
+    {
+      "epoch": 0.22260933345557898,
+      "learning_rate": 0.0001554799669936738,
+      "loss": 0.8819,
+      "mean_token_accuracy": 0.7356357038021087,
+      "num_tokens": 56166239.0,
+      "step": 24280
+    },
+    {
+      "epoch": 0.2227010176950582,
+      "learning_rate": 0.00015546163014577795,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.741468071937561,
+      "num_tokens": 56189490.0,
+      "step": 24290
+    },
+    {
+      "epoch": 0.22279270193453746,
+      "learning_rate": 0.00015544329329788208,
+      "loss": 0.8789,
+      "mean_token_accuracy": 0.7427528321743011,
+      "num_tokens": 56213041.0,
+      "step": 24300
+    },
+    {
+      "epoch": 0.22288438617401668,
+      "learning_rate": 0.00015542495644998625,
+      "loss": 0.88,
+      "mean_token_accuracy": 0.7418298780918121,
+      "num_tokens": 56236414.0,
+      "step": 24310
+    },
+    {
+      "epoch": 0.22297607041349593,
+      "learning_rate": 0.0001554066196020904,
+      "loss": 0.8399,
+      "mean_token_accuracy": 0.7472315311431885,
+      "num_tokens": 56259296.0,
+      "step": 24320
+    },
+    {
+      "epoch": 0.22306775465297515,
+      "learning_rate": 0.00015538828275419458,
+      "loss": 0.8759,
+      "mean_token_accuracy": 0.7376609802246094,
+      "num_tokens": 56282709.0,
+      "step": 24330
+    },
+    {
+      "epoch": 0.22315943889245438,
+      "learning_rate": 0.00015536994590629872,
+      "loss": 0.8295,
+      "mean_token_accuracy": 0.7500767946243286,
+      "num_tokens": 56305399.0,
+      "step": 24340
+    },
+    {
+      "epoch": 0.22325112313193363,
+      "learning_rate": 0.00015535160905840286,
+      "loss": 0.8399,
+      "mean_token_accuracy": 0.7495501339435577,
+      "num_tokens": 56329027.0,
+      "step": 24350
+    },
+    {
+      "epoch": 0.22334280737141285,
+      "learning_rate": 0.00015533327221050702,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.747114610671997,
+      "num_tokens": 56352401.0,
+      "step": 24360
+    },
+    {
+      "epoch": 0.2234344916108921,
+      "learning_rate": 0.00015531493536261116,
+      "loss": 0.8055,
+      "mean_token_accuracy": 0.7586235523223877,
+      "num_tokens": 56375613.0,
+      "step": 24370
+    },
+    {
+      "epoch": 0.22352617585037132,
+      "learning_rate": 0.00015529659851471533,
+      "loss": 0.869,
+      "mean_token_accuracy": 0.7420877635478973,
+      "num_tokens": 56398680.0,
+      "step": 24380
+    },
+    {
+      "epoch": 0.22361786008985055,
+      "learning_rate": 0.0001552782616668195,
+      "loss": 0.8528,
+      "mean_token_accuracy": 0.7443232476711273,
+      "num_tokens": 56422318.0,
+      "step": 24390
+    },
+    {
+      "epoch": 0.2237095443293298,
+      "learning_rate": 0.00015525992481892363,
+      "loss": 0.8817,
+      "mean_token_accuracy": 0.74188734292984,
+      "num_tokens": 56445584.0,
+      "step": 24400
+    },
+    {
+      "epoch": 0.22380122856880902,
+      "learning_rate": 0.0001552415879710278,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.7461014330387116,
+      "num_tokens": 56469280.0,
+      "step": 24410
+    },
+    {
+      "epoch": 0.22389291280828824,
+      "learning_rate": 0.00015522325112313193,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7505100846290589,
+      "num_tokens": 56492747.0,
+      "step": 24420
+    },
+    {
+      "epoch": 0.2239845970477675,
+      "learning_rate": 0.0001552049142752361,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7417403936386109,
+      "num_tokens": 56516060.0,
+      "step": 24430
+    },
+    {
+      "epoch": 0.22407628128724671,
+      "learning_rate": 0.00015518657742734024,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.7434927344322204,
+      "num_tokens": 56539230.0,
+      "step": 24440
+    },
+    {
+      "epoch": 0.22416796552672597,
+      "learning_rate": 0.0001551682405794444,
+      "loss": 0.8751,
+      "mean_token_accuracy": 0.7428310811519623,
+      "num_tokens": 56561651.0,
+      "step": 24450
+    },
+    {
+      "epoch": 0.2242596497662052,
+      "learning_rate": 0.00015514990373154857,
+      "loss": 0.8644,
+      "mean_token_accuracy": 0.7391244351863862,
+      "num_tokens": 56584795.0,
+      "step": 24460
+    },
+    {
+      "epoch": 0.2243513340056844,
+      "learning_rate": 0.0001551315668836527,
+      "loss": 0.8515,
+      "mean_token_accuracy": 0.7435534298419952,
+      "num_tokens": 56607654.0,
+      "step": 24470
+    },
+    {
+      "epoch": 0.22444301824516366,
+      "learning_rate": 0.00015511323003575687,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.743599671125412,
+      "num_tokens": 56631807.0,
+      "step": 24480
+    },
+    {
+      "epoch": 0.22453470248464288,
+      "learning_rate": 0.000155094893187861,
+      "loss": 0.8604,
+      "mean_token_accuracy": 0.7444347977638245,
+      "num_tokens": 56655863.0,
+      "step": 24490
+    },
+    {
+      "epoch": 0.22462638672412213,
+      "learning_rate": 0.00015507655633996515,
+      "loss": 0.8759,
+      "mean_token_accuracy": 0.7387280941009522,
+      "num_tokens": 56678861.0,
+      "step": 24500
+    },
+    {
+      "epoch": 0.22471807096360136,
+      "learning_rate": 0.00015505821949206931,
+      "loss": 0.8874,
+      "mean_token_accuracy": 0.7401228547096252,
+      "num_tokens": 56702575.0,
+      "step": 24510
+    },
+    {
+      "epoch": 0.22480975520308058,
+      "learning_rate": 0.00015503988264417348,
+      "loss": 0.8294,
+      "mean_token_accuracy": 0.7471656143665314,
+      "num_tokens": 56725452.0,
+      "step": 24520
+    },
+    {
+      "epoch": 0.22490143944255983,
+      "learning_rate": 0.00015502154579627764,
+      "loss": 0.8424,
+      "mean_token_accuracy": 0.7398037910461426,
+      "num_tokens": 56748631.0,
+      "step": 24530
+    },
+    {
+      "epoch": 0.22499312368203905,
+      "learning_rate": 0.00015500320894838178,
+      "loss": 0.8798,
+      "mean_token_accuracy": 0.7348233699798584,
+      "num_tokens": 56771624.0,
+      "step": 24540
+    },
+    {
+      "epoch": 0.2250848079215183,
+      "learning_rate": 0.00015498487210048592,
+      "loss": 0.8239,
+      "mean_token_accuracy": 0.7494547724723816,
+      "num_tokens": 56794665.0,
+      "step": 24550
+    },
+    {
+      "epoch": 0.22517649216099753,
+      "learning_rate": 0.0001549665352525901,
+      "loss": 0.8837,
+      "mean_token_accuracy": 0.7364970028400422,
+      "num_tokens": 56817625.0,
+      "step": 24560
+    },
+    {
+      "epoch": 0.22526817640047675,
+      "learning_rate": 0.00015494819840469423,
+      "loss": 0.8248,
+      "mean_token_accuracy": 0.7507057189941406,
+      "num_tokens": 56840334.0,
+      "step": 24570
+    },
+    {
+      "epoch": 0.225359860639956,
+      "learning_rate": 0.0001549298615567984,
+      "loss": 0.8114,
+      "mean_token_accuracy": 0.7491330444812775,
+      "num_tokens": 56863710.0,
+      "step": 24580
+    },
+    {
+      "epoch": 0.22545154487943522,
+      "learning_rate": 0.00015491152470890256,
+      "loss": 0.8821,
+      "mean_token_accuracy": 0.7415892958641053,
+      "num_tokens": 56886699.0,
+      "step": 24590
+    },
+    {
+      "epoch": 0.22554322911891445,
+      "learning_rate": 0.0001548931878610067,
+      "loss": 0.8379,
+      "mean_token_accuracy": 0.7512984931468963,
+      "num_tokens": 56909298.0,
+      "step": 24600
+    },
+    {
+      "epoch": 0.2256349133583937,
+      "learning_rate": 0.00015487485101311086,
+      "loss": 0.8739,
+      "mean_token_accuracy": 0.7323919415473938,
+      "num_tokens": 56933073.0,
+      "step": 24610
+    },
+    {
+      "epoch": 0.22572659759787292,
+      "learning_rate": 0.000154856514165215,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7439683377742767,
+      "num_tokens": 56956139.0,
+      "step": 24620
+    },
+    {
+      "epoch": 0.22581828183735217,
+      "learning_rate": 0.00015483817731731916,
+      "loss": 0.8689,
+      "mean_token_accuracy": 0.7440510332584381,
+      "num_tokens": 56979088.0,
+      "step": 24630
+    },
+    {
+      "epoch": 0.2259099660768314,
+      "learning_rate": 0.0001548198404694233,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7423623442649842,
+      "num_tokens": 57002535.0,
+      "step": 24640
+    },
+    {
+      "epoch": 0.22600165031631061,
+      "learning_rate": 0.00015480150362152747,
+      "loss": 0.8944,
+      "mean_token_accuracy": 0.7390952825546264,
+      "num_tokens": 57025765.0,
+      "step": 24650
+    },
+    {
+      "epoch": 0.22609333455578987,
+      "learning_rate": 0.00015478316677363163,
+      "loss": 0.8493,
+      "mean_token_accuracy": 0.7484005272388459,
+      "num_tokens": 57048734.0,
+      "step": 24660
+    },
+    {
+      "epoch": 0.2261850187952691,
+      "learning_rate": 0.00015476482992573577,
+      "loss": 0.8722,
+      "mean_token_accuracy": 0.736942982673645,
+      "num_tokens": 57071929.0,
+      "step": 24670
+    },
+    {
+      "epoch": 0.22627670303474834,
+      "learning_rate": 0.00015474649307783994,
+      "loss": 0.8824,
+      "mean_token_accuracy": 0.7420193493366242,
+      "num_tokens": 57095542.0,
+      "step": 24680
+    },
+    {
+      "epoch": 0.22636838727422756,
+      "learning_rate": 0.00015472815622994407,
+      "loss": 0.8378,
+      "mean_token_accuracy": 0.7449758231639863,
+      "num_tokens": 57118565.0,
+      "step": 24690
+    },
+    {
+      "epoch": 0.22646007151370678,
+      "learning_rate": 0.00015470981938204824,
+      "loss": 0.8741,
+      "mean_token_accuracy": 0.7404014468193054,
+      "num_tokens": 57141420.0,
+      "step": 24700
+    },
+    {
+      "epoch": 0.22655175575318603,
+      "learning_rate": 0.00015469148253415238,
+      "loss": 0.8519,
+      "mean_token_accuracy": 0.7445592880249023,
+      "num_tokens": 57164619.0,
+      "step": 24710
+    },
+    {
+      "epoch": 0.22664343999266526,
+      "learning_rate": 0.00015467314568625654,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7371529340744019,
+      "num_tokens": 57188234.0,
+      "step": 24720
+    },
+    {
+      "epoch": 0.2267351242321445,
+      "learning_rate": 0.0001546548088383607,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.735043853521347,
+      "num_tokens": 57211321.0,
+      "step": 24730
+    },
+    {
+      "epoch": 0.22682680847162373,
+      "learning_rate": 0.00015463647199046485,
+      "loss": 0.8483,
+      "mean_token_accuracy": 0.7475723206996918,
+      "num_tokens": 57234545.0,
+      "step": 24740
+    },
+    {
+      "epoch": 0.22691849271110295,
+      "learning_rate": 0.00015461813514256899,
+      "loss": 0.8822,
+      "mean_token_accuracy": 0.7385982275009155,
+      "num_tokens": 57257569.0,
+      "step": 24750
+    },
+    {
+      "epoch": 0.2270101769505822,
+      "learning_rate": 0.00015459979829467315,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.742702966928482,
+      "num_tokens": 57279975.0,
+      "step": 24760
+    },
+    {
+      "epoch": 0.22710186119006143,
+      "learning_rate": 0.0001545814614467773,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.7445092737674713,
+      "num_tokens": 57303376.0,
+      "step": 24770
+    },
+    {
+      "epoch": 0.22719354542954065,
+      "learning_rate": 0.00015456312459888148,
+      "loss": 0.8751,
+      "mean_token_accuracy": 0.7360867619514465,
+      "num_tokens": 57326850.0,
+      "step": 24780
+    },
+    {
+      "epoch": 0.2272852296690199,
+      "learning_rate": 0.00015454478775098562,
+      "loss": 0.8735,
+      "mean_token_accuracy": 0.7465667426586151,
+      "num_tokens": 57350406.0,
+      "step": 24790
+    },
+    {
+      "epoch": 0.22737691390849912,
+      "learning_rate": 0.00015452645090308976,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7381657600402832,
+      "num_tokens": 57373930.0,
+      "step": 24800
+    },
+    {
+      "epoch": 0.22746859814797837,
+      "learning_rate": 0.00015450811405519392,
+      "loss": 0.866,
+      "mean_token_accuracy": 0.7426710903644562,
+      "num_tokens": 57397601.0,
+      "step": 24810
+    },
+    {
+      "epoch": 0.2275602823874576,
+      "learning_rate": 0.00015448977720729806,
+      "loss": 0.8336,
+      "mean_token_accuracy": 0.7462375402450562,
+      "num_tokens": 57420395.0,
+      "step": 24820
+    },
+    {
+      "epoch": 0.22765196662693682,
+      "learning_rate": 0.00015447144035940223,
+      "loss": 0.827,
+      "mean_token_accuracy": 0.746272224187851,
+      "num_tokens": 57443718.0,
+      "step": 24830
+    },
+    {
+      "epoch": 0.22774365086641607,
+      "learning_rate": 0.00015445310351150637,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.7406410813331604,
+      "num_tokens": 57467203.0,
+      "step": 24840
+    },
+    {
+      "epoch": 0.2278353351058953,
+      "learning_rate": 0.00015443476666361053,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.748191624879837,
+      "num_tokens": 57490304.0,
+      "step": 24850
+    },
+    {
+      "epoch": 0.22792701934537454,
+      "learning_rate": 0.0001544164298157147,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.74213125705719,
+      "num_tokens": 57513518.0,
+      "step": 24860
+    },
+    {
+      "epoch": 0.22801870358485377,
+      "learning_rate": 0.00015439809296781884,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7463255822658539,
+      "num_tokens": 57536445.0,
+      "step": 24870
+    },
+    {
+      "epoch": 0.228110387824333,
+      "learning_rate": 0.000154379756119923,
+      "loss": 0.9032,
+      "mean_token_accuracy": 0.7351026475429535,
+      "num_tokens": 57559747.0,
+      "step": 24880
+    },
+    {
+      "epoch": 0.22820207206381224,
+      "learning_rate": 0.00015436141927202714,
+      "loss": 0.8471,
+      "mean_token_accuracy": 0.748066633939743,
+      "num_tokens": 57581966.0,
+      "step": 24890
+    },
+    {
+      "epoch": 0.22829375630329146,
+      "learning_rate": 0.0001543430824241313,
+      "loss": 0.8504,
+      "mean_token_accuracy": 0.7467912256717681,
+      "num_tokens": 57604905.0,
+      "step": 24900
+    },
+    {
+      "epoch": 0.22838544054277068,
+      "learning_rate": 0.00015432474557623547,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.7365090548992157,
+      "num_tokens": 57628518.0,
+      "step": 24910
+    },
+    {
+      "epoch": 0.22847712478224994,
+      "learning_rate": 0.0001543064087283396,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7403376162052154,
+      "num_tokens": 57651665.0,
+      "step": 24920
+    },
+    {
+      "epoch": 0.22856880902172916,
+      "learning_rate": 0.00015428807188044377,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7411530137062072,
+      "num_tokens": 57674991.0,
+      "step": 24930
+    },
+    {
+      "epoch": 0.2286604932612084,
+      "learning_rate": 0.0001542697350325479,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7386019885540008,
+      "num_tokens": 57698144.0,
+      "step": 24940
+    },
+    {
+      "epoch": 0.22875217750068763,
+      "learning_rate": 0.00015425139818465205,
+      "loss": 0.8329,
+      "mean_token_accuracy": 0.7490780770778656,
+      "num_tokens": 57721894.0,
+      "step": 24950
+    },
+    {
+      "epoch": 0.22884386174016685,
+      "learning_rate": 0.00015423306133675622,
+      "loss": 0.8425,
+      "mean_token_accuracy": 0.7458144187927246,
+      "num_tokens": 57744725.0,
+      "step": 24960
+    },
+    {
+      "epoch": 0.2289355459796461,
+      "learning_rate": 0.00015421472448886035,
+      "loss": 0.8907,
+      "mean_token_accuracy": 0.7429078042507171,
+      "num_tokens": 57768097.0,
+      "step": 24970
+    },
+    {
+      "epoch": 0.22902723021912533,
+      "learning_rate": 0.00015419638764096455,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7372221231460572,
+      "num_tokens": 57790831.0,
+      "step": 24980
+    },
+    {
+      "epoch": 0.22911891445860458,
+      "learning_rate": 0.00015417805079306869,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.737917172908783,
+      "num_tokens": 57813922.0,
+      "step": 24990
+    },
+    {
+      "epoch": 0.2292105986980838,
+      "learning_rate": 0.00015415971394517282,
+      "loss": 0.8627,
+      "mean_token_accuracy": 0.7416988790035248,
+      "num_tokens": 57837143.0,
+      "step": 25000
+    },
+    {
+      "epoch": 0.22930228293756302,
+      "learning_rate": 0.000154141377097277,
+      "loss": 0.9013,
+      "mean_token_accuracy": 0.7348793506622314,
+      "num_tokens": 57859442.0,
+      "step": 25010
+    },
+    {
+      "epoch": 0.22939396717704227,
+      "learning_rate": 0.00015412304024938113,
+      "loss": 0.8776,
+      "mean_token_accuracy": 0.7403935253620147,
+      "num_tokens": 57882885.0,
+      "step": 25020
+    },
+    {
+      "epoch": 0.2294856514165215,
+      "learning_rate": 0.0001541047034014853,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7390576839447022,
+      "num_tokens": 57905922.0,
+      "step": 25030
+    },
+    {
+      "epoch": 0.22957733565600075,
+      "learning_rate": 0.00015408636655358946,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.7414995014667511,
+      "num_tokens": 57928918.0,
+      "step": 25040
+    },
+    {
+      "epoch": 0.22966901989547997,
+      "learning_rate": 0.0001540680297056936,
+      "loss": 0.8782,
+      "mean_token_accuracy": 0.7407548666000366,
+      "num_tokens": 57951626.0,
+      "step": 25050
+    },
+    {
+      "epoch": 0.2297607041349592,
+      "learning_rate": 0.00015404969285779776,
+      "loss": 0.8767,
+      "mean_token_accuracy": 0.735957533121109,
+      "num_tokens": 57975618.0,
+      "step": 25060
+    },
+    {
+      "epoch": 0.22985238837443844,
+      "learning_rate": 0.0001540313560099019,
+      "loss": 0.8424,
+      "mean_token_accuracy": 0.7482416808605195,
+      "num_tokens": 57998609.0,
+      "step": 25070
+    },
+    {
+      "epoch": 0.22994407261391767,
+      "learning_rate": 0.00015401301916200607,
+      "loss": 0.8899,
+      "mean_token_accuracy": 0.735405319929123,
+      "num_tokens": 58021340.0,
+      "step": 25080
+    },
+    {
+      "epoch": 0.2300357568533969,
+      "learning_rate": 0.0001539946823141102,
+      "loss": 0.8619,
+      "mean_token_accuracy": 0.7451351046562195,
+      "num_tokens": 58044855.0,
+      "step": 25090
+    },
+    {
+      "epoch": 0.23012744109287614,
+      "learning_rate": 0.00015397634546621437,
+      "loss": 0.8689,
+      "mean_token_accuracy": 0.7421709656715393,
+      "num_tokens": 58067997.0,
+      "step": 25100
+    },
+    {
+      "epoch": 0.23021912533235536,
+      "learning_rate": 0.00015395800861831853,
+      "loss": 0.8743,
+      "mean_token_accuracy": 0.7373112738132477,
+      "num_tokens": 58091078.0,
+      "step": 25110
+    },
+    {
+      "epoch": 0.2303108095718346,
+      "learning_rate": 0.00015393967177042267,
+      "loss": 0.8873,
+      "mean_token_accuracy": 0.738627690076828,
+      "num_tokens": 58114277.0,
+      "step": 25120
+    },
+    {
+      "epoch": 0.23040249381131384,
+      "learning_rate": 0.00015392133492252684,
+      "loss": 0.8799,
+      "mean_token_accuracy": 0.7405313074588775,
+      "num_tokens": 58137929.0,
+      "step": 25130
+    },
+    {
+      "epoch": 0.23049417805079306,
+      "learning_rate": 0.00015390299807463098,
+      "loss": 0.9035,
+      "mean_token_accuracy": 0.7274326145648956,
+      "num_tokens": 58160912.0,
+      "step": 25140
+    },
+    {
+      "epoch": 0.2305858622902723,
+      "learning_rate": 0.00015388466122673511,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7468173027038574,
+      "num_tokens": 58183086.0,
+      "step": 25150
+    },
+    {
+      "epoch": 0.23067754652975153,
+      "learning_rate": 0.00015386632437883928,
+      "loss": 0.8665,
+      "mean_token_accuracy": 0.7416611790657044,
+      "num_tokens": 58206392.0,
+      "step": 25160
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "learning_rate": 0.00015384798753094345,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7480005204677582,
+      "num_tokens": 58229758.0,
+      "step": 25170
+    },
+    {
+      "epoch": 0.23086091500871,
+      "learning_rate": 0.0001538296506830476,
+      "loss": 0.8819,
+      "mean_token_accuracy": 0.7375615358352661,
+      "num_tokens": 58252065.0,
+      "step": 25180
+    },
+    {
+      "epoch": 0.23095259924818923,
+      "learning_rate": 0.00015381131383515175,
+      "loss": 0.8661,
+      "mean_token_accuracy": 0.7426708579063416,
+      "num_tokens": 58276073.0,
+      "step": 25190
+    },
+    {
+      "epoch": 0.23104428348766848,
+      "learning_rate": 0.0001537929769872559,
+      "loss": 0.9221,
+      "mean_token_accuracy": 0.7264938712120056,
+      "num_tokens": 58299222.0,
+      "step": 25200
+    },
+    {
+      "epoch": 0.2311359677271477,
+      "learning_rate": 0.00015377464013936005,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7471257090568543,
+      "num_tokens": 58321658.0,
+      "step": 25210
+    },
+    {
+      "epoch": 0.23122765196662692,
+      "learning_rate": 0.0001537563032914642,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7433595061302185,
+      "num_tokens": 58344246.0,
+      "step": 25220
+    },
+    {
+      "epoch": 0.23131933620610617,
+      "learning_rate": 0.00015373796644356836,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7436851263046265,
+      "num_tokens": 58367350.0,
+      "step": 25230
+    },
+    {
+      "epoch": 0.2314110204455854,
+      "learning_rate": 0.00015371962959567252,
+      "loss": 0.8295,
+      "mean_token_accuracy": 0.7513671100139618,
+      "num_tokens": 58390349.0,
+      "step": 25240
+    },
+    {
+      "epoch": 0.23150270468506465,
+      "learning_rate": 0.00015370129274777666,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7378962755203247,
+      "num_tokens": 58413732.0,
+      "step": 25250
+    },
+    {
+      "epoch": 0.23159438892454387,
+      "learning_rate": 0.00015368295589988083,
+      "loss": 0.8173,
+      "mean_token_accuracy": 0.7515727698802948,
+      "num_tokens": 58436637.0,
+      "step": 25260
+    },
+    {
+      "epoch": 0.2316860731640231,
+      "learning_rate": 0.00015366461905198496,
+      "loss": 0.8461,
+      "mean_token_accuracy": 0.7481701791286468,
+      "num_tokens": 58459711.0,
+      "step": 25270
+    },
+    {
+      "epoch": 0.23177775740350234,
+      "learning_rate": 0.00015364628220408913,
+      "loss": 0.8758,
+      "mean_token_accuracy": 0.7404216885566711,
+      "num_tokens": 58482154.0,
+      "step": 25280
+    },
+    {
+      "epoch": 0.23186944164298157,
+      "learning_rate": 0.00015362794535619327,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7502171277999878,
+      "num_tokens": 58505395.0,
+      "step": 25290
+    },
+    {
+      "epoch": 0.23196112588246082,
+      "learning_rate": 0.00015360960850829743,
+      "loss": 0.8883,
+      "mean_token_accuracy": 0.7388078510761261,
+      "num_tokens": 58528371.0,
+      "step": 25300
+    },
+    {
+      "epoch": 0.23205281012194004,
+      "learning_rate": 0.0001535912716604016,
+      "loss": 0.8237,
+      "mean_token_accuracy": 0.747081845998764,
+      "num_tokens": 58551990.0,
+      "step": 25310
+    },
+    {
+      "epoch": 0.23214449436141926,
+      "learning_rate": 0.00015357293481250574,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7440243124961853,
+      "num_tokens": 58574792.0,
+      "step": 25320
+    },
+    {
+      "epoch": 0.2322361786008985,
+      "learning_rate": 0.0001535545979646099,
+      "loss": 0.8504,
+      "mean_token_accuracy": 0.7496018052101135,
+      "num_tokens": 58596797.0,
+      "step": 25330
+    },
+    {
+      "epoch": 0.23232786284037774,
+      "learning_rate": 0.00015353626111671404,
+      "loss": 0.8148,
+      "mean_token_accuracy": 0.7489557206630707,
+      "num_tokens": 58619683.0,
+      "step": 25340
+    },
+    {
+      "epoch": 0.232419547079857,
+      "learning_rate": 0.00015351792426881818,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.7465729355812073,
+      "num_tokens": 58642595.0,
+      "step": 25350
+    },
+    {
+      "epoch": 0.2325112313193362,
+      "learning_rate": 0.00015349958742092234,
+      "loss": 0.8488,
+      "mean_token_accuracy": 0.7383763372898102,
+      "num_tokens": 58665394.0,
+      "step": 25360
+    },
+    {
+      "epoch": 0.23260291555881543,
+      "learning_rate": 0.0001534812505730265,
+      "loss": 0.8403,
+      "mean_token_accuracy": 0.7497602224349975,
+      "num_tokens": 58688035.0,
+      "step": 25370
+    },
+    {
+      "epoch": 0.23269459979829468,
+      "learning_rate": 0.00015346291372513068,
+      "loss": 0.8458,
+      "mean_token_accuracy": 0.7423629879951477,
+      "num_tokens": 58711086.0,
+      "step": 25380
+    },
+    {
+      "epoch": 0.2327862840377739,
+      "learning_rate": 0.00015344457687723481,
+      "loss": 0.847,
+      "mean_token_accuracy": 0.7445569574832916,
+      "num_tokens": 58733477.0,
+      "step": 25390
+    },
+    {
+      "epoch": 0.23287796827725313,
+      "learning_rate": 0.00015342624002933895,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7452936112880707,
+      "num_tokens": 58757107.0,
+      "step": 25400
+    },
+    {
+      "epoch": 0.23296965251673238,
+      "learning_rate": 0.00015340790318144312,
+      "loss": 0.8601,
+      "mean_token_accuracy": 0.741677588224411,
+      "num_tokens": 58780288.0,
+      "step": 25410
+    },
+    {
+      "epoch": 0.2330613367562116,
+      "learning_rate": 0.00015338956633354726,
+      "loss": 0.8434,
+      "mean_token_accuracy": 0.7475954532623291,
+      "num_tokens": 58803007.0,
+      "step": 25420
+    },
+    {
+      "epoch": 0.23315302099569085,
+      "learning_rate": 0.00015337122948565142,
+      "loss": 0.84,
+      "mean_token_accuracy": 0.7521816074848175,
+      "num_tokens": 58825954.0,
+      "step": 25430
+    },
+    {
+      "epoch": 0.23324470523517007,
+      "learning_rate": 0.0001533528926377556,
+      "loss": 0.8641,
+      "mean_token_accuracy": 0.7487018287181855,
+      "num_tokens": 58848635.0,
+      "step": 25440
+    },
+    {
+      "epoch": 0.2333363894746493,
+      "learning_rate": 0.00015333455578985973,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.7424684166908264,
+      "num_tokens": 58872077.0,
+      "step": 25450
+    },
+    {
+      "epoch": 0.23342807371412855,
+      "learning_rate": 0.0001533162189419639,
+      "loss": 0.8704,
+      "mean_token_accuracy": 0.7372147023677826,
+      "num_tokens": 58894889.0,
+      "step": 25460
+    },
+    {
+      "epoch": 0.23351975795360777,
+      "learning_rate": 0.00015329788209406803,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.7474406838417054,
+      "num_tokens": 58918355.0,
+      "step": 25470
+    },
+    {
+      "epoch": 0.23361144219308702,
+      "learning_rate": 0.0001532795452461722,
+      "loss": 0.8816,
+      "mean_token_accuracy": 0.744608473777771,
+      "num_tokens": 58941341.0,
+      "step": 25480
+    },
+    {
+      "epoch": 0.23370312643256624,
+      "learning_rate": 0.00015326120839827633,
+      "loss": 0.9057,
+      "mean_token_accuracy": 0.7318639755249023,
+      "num_tokens": 58964375.0,
+      "step": 25490
+    },
+    {
+      "epoch": 0.23379481067204547,
+      "learning_rate": 0.0001532428715503805,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7439403295516968,
+      "num_tokens": 58988214.0,
+      "step": 25500
+    },
+    {
+      "epoch": 0.23388649491152472,
+      "learning_rate": 0.00015322453470248466,
+      "loss": 0.8057,
+      "mean_token_accuracy": 0.7566617786884308,
+      "num_tokens": 59010918.0,
+      "step": 25510
+    },
+    {
+      "epoch": 0.23397817915100394,
+      "learning_rate": 0.0001532061978545888,
+      "loss": 0.835,
+      "mean_token_accuracy": 0.7533825874328614,
+      "num_tokens": 59034500.0,
+      "step": 25520
+    },
+    {
+      "epoch": 0.23406986339048316,
+      "learning_rate": 0.00015318786100669297,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.74126296043396,
+      "num_tokens": 59057221.0,
+      "step": 25530
+    },
+    {
+      "epoch": 0.2341615476299624,
+      "learning_rate": 0.0001531695241587971,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7496389210224151,
+      "num_tokens": 59080168.0,
+      "step": 25540
+    },
+    {
+      "epoch": 0.23425323186944164,
+      "learning_rate": 0.00015315118731090124,
+      "loss": 0.854,
+      "mean_token_accuracy": 0.7466756463050842,
+      "num_tokens": 59103732.0,
+      "step": 25550
+    },
+    {
+      "epoch": 0.2343449161089209,
+      "learning_rate": 0.0001531328504630054,
+      "loss": 0.8872,
+      "mean_token_accuracy": 0.7361054718494415,
+      "num_tokens": 59127703.0,
+      "step": 25560
+    },
+    {
+      "epoch": 0.2344366003484001,
+      "learning_rate": 0.00015311451361510957,
+      "loss": 0.8779,
+      "mean_token_accuracy": 0.7358223915100097,
+      "num_tokens": 59150810.0,
+      "step": 25570
+    },
+    {
+      "epoch": 0.23452828458787933,
+      "learning_rate": 0.00015309617676721374,
+      "loss": 0.8278,
+      "mean_token_accuracy": 0.7478847682476044,
+      "num_tokens": 59173205.0,
+      "step": 25580
+    },
+    {
+      "epoch": 0.23461996882735858,
+      "learning_rate": 0.00015307783991931788,
+      "loss": 0.8674,
+      "mean_token_accuracy": 0.7413980424404144,
+      "num_tokens": 59196150.0,
+      "step": 25590
+    },
+    {
+      "epoch": 0.2347116530668378,
+      "learning_rate": 0.00015305950307142202,
+      "loss": 0.8171,
+      "mean_token_accuracy": 0.7523119330406189,
+      "num_tokens": 59218947.0,
+      "step": 25600
+    },
+    {
+      "epoch": 0.23480333730631706,
+      "learning_rate": 0.00015304116622352618,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.746193128824234,
+      "num_tokens": 59242644.0,
+      "step": 25610
+    },
+    {
+      "epoch": 0.23489502154579628,
+      "learning_rate": 0.00015302282937563032,
+      "loss": 0.7993,
+      "mean_token_accuracy": 0.7595108270645141,
+      "num_tokens": 59266159.0,
+      "step": 25620
+    },
+    {
+      "epoch": 0.2349867057852755,
+      "learning_rate": 0.0001530044925277345,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.7411499857902527,
+      "num_tokens": 59288880.0,
+      "step": 25630
+    },
+    {
+      "epoch": 0.23507839002475475,
+      "learning_rate": 0.00015298615567983865,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7382328033447265,
+      "num_tokens": 59311283.0,
+      "step": 25640
+    },
+    {
+      "epoch": 0.23517007426423397,
+      "learning_rate": 0.0001529678188319428,
+      "loss": 0.8174,
+      "mean_token_accuracy": 0.7555031061172486,
+      "num_tokens": 59335183.0,
+      "step": 25650
+    },
+    {
+      "epoch": 0.23526175850371323,
+      "learning_rate": 0.00015294948198404696,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7349155366420745,
+      "num_tokens": 59357863.0,
+      "step": 25660
+    },
+    {
+      "epoch": 0.23535344274319245,
+      "learning_rate": 0.0001529311451361511,
+      "loss": 0.8725,
+      "mean_token_accuracy": 0.7460060000419617,
+      "num_tokens": 59381002.0,
+      "step": 25670
+    },
+    {
+      "epoch": 0.23544512698267167,
+      "learning_rate": 0.00015291280828825526,
+      "loss": 0.8683,
+      "mean_token_accuracy": 0.7456463515758515,
+      "num_tokens": 59404668.0,
+      "step": 25680
+    },
+    {
+      "epoch": 0.23553681122215092,
+      "learning_rate": 0.0001528944714403594,
+      "loss": 0.8966,
+      "mean_token_accuracy": 0.7353382349014282,
+      "num_tokens": 59427709.0,
+      "step": 25690
+    },
+    {
+      "epoch": 0.23562849546163014,
+      "learning_rate": 0.00015287613459246356,
+      "loss": 0.8186,
+      "mean_token_accuracy": 0.754264110326767,
+      "num_tokens": 59450281.0,
+      "step": 25700
+    },
+    {
+      "epoch": 0.23572017970110937,
+      "learning_rate": 0.00015285779774456773,
+      "loss": 0.8902,
+      "mean_token_accuracy": 0.7390115857124329,
+      "num_tokens": 59473837.0,
+      "step": 25710
+    },
+    {
+      "epoch": 0.23581186394058862,
+      "learning_rate": 0.00015283946089667187,
+      "loss": 0.899,
+      "mean_token_accuracy": 0.7373830497264862,
+      "num_tokens": 59497061.0,
+      "step": 25720
+    },
+    {
+      "epoch": 0.23590354818006784,
+      "learning_rate": 0.00015282112404877603,
+      "loss": 0.8389,
+      "mean_token_accuracy": 0.7503412425518036,
+      "num_tokens": 59519429.0,
+      "step": 25730
+    },
+    {
+      "epoch": 0.2359952324195471,
+      "learning_rate": 0.00015280278720088017,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7404552578926087,
+      "num_tokens": 59542369.0,
+      "step": 25740
+    },
+    {
+      "epoch": 0.2360869166590263,
+      "learning_rate": 0.0001527844503529843,
+      "loss": 0.8885,
+      "mean_token_accuracy": 0.7375787854194641,
+      "num_tokens": 59566727.0,
+      "step": 25750
+    },
+    {
+      "epoch": 0.23617860089850554,
+      "learning_rate": 0.0001527661135050885,
+      "loss": 0.8165,
+      "mean_token_accuracy": 0.7525578200817108,
+      "num_tokens": 59589340.0,
+      "step": 25760
+    },
+    {
+      "epoch": 0.2362702851379848,
+      "learning_rate": 0.00015274777665719264,
+      "loss": 0.8746,
+      "mean_token_accuracy": 0.7354776978492736,
+      "num_tokens": 59612169.0,
+      "step": 25770
+    },
+    {
+      "epoch": 0.236361969377464,
+      "learning_rate": 0.0001527294398092968,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.7442165195941925,
+      "num_tokens": 59634687.0,
+      "step": 25780
+    },
+    {
+      "epoch": 0.23645365361694326,
+      "learning_rate": 0.00015271110296140094,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.7429732918739319,
+      "num_tokens": 59657665.0,
+      "step": 25790
+    },
+    {
+      "epoch": 0.23654533785642248,
+      "learning_rate": 0.00015269276611350508,
+      "loss": 0.8892,
+      "mean_token_accuracy": 0.7356120705604553,
+      "num_tokens": 59680038.0,
+      "step": 25800
+    },
+    {
+      "epoch": 0.2366370220959017,
+      "learning_rate": 0.00015267442926560925,
+      "loss": 0.8803,
+      "mean_token_accuracy": 0.7332807183265686,
+      "num_tokens": 59703503.0,
+      "step": 25810
+    },
+    {
+      "epoch": 0.23672870633538096,
+      "learning_rate": 0.00015265609241771338,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7459983766078949,
+      "num_tokens": 59726455.0,
+      "step": 25820
+    },
+    {
+      "epoch": 0.23682039057486018,
+      "learning_rate": 0.00015263775556981758,
+      "loss": 0.8394,
+      "mean_token_accuracy": 0.7493344068527221,
+      "num_tokens": 59749853.0,
+      "step": 25830
+    },
+    {
+      "epoch": 0.2369120748143394,
+      "learning_rate": 0.00015261941872192172,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7432567596435546,
+      "num_tokens": 59772394.0,
+      "step": 25840
+    },
+    {
+      "epoch": 0.23700375905381865,
+      "learning_rate": 0.00015260108187402585,
+      "loss": 0.8616,
+      "mean_token_accuracy": 0.7459872603416443,
+      "num_tokens": 59795906.0,
+      "step": 25850
+    },
+    {
+      "epoch": 0.23709544329329788,
+      "learning_rate": 0.00015258274502613002,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.740733414888382,
+      "num_tokens": 59818436.0,
+      "step": 25860
+    },
+    {
+      "epoch": 0.23718712753277713,
+      "learning_rate": 0.00015256440817823416,
+      "loss": 0.8277,
+      "mean_token_accuracy": 0.7478637576103211,
+      "num_tokens": 59842154.0,
+      "step": 25870
+    },
+    {
+      "epoch": 0.23727881177225635,
+      "learning_rate": 0.00015254607133033832,
+      "loss": 0.9151,
+      "mean_token_accuracy": 0.7274119138717652,
+      "num_tokens": 59864635.0,
+      "step": 25880
+    },
+    {
+      "epoch": 0.23737049601173557,
+      "learning_rate": 0.0001525277344824425,
+      "loss": 0.8764,
+      "mean_token_accuracy": 0.7428054809570312,
+      "num_tokens": 59888780.0,
+      "step": 25890
+    },
+    {
+      "epoch": 0.23746218025121482,
+      "learning_rate": 0.00015250939763454663,
+      "loss": 0.8731,
+      "mean_token_accuracy": 0.7317515194416047,
+      "num_tokens": 59912268.0,
+      "step": 25900
+    },
+    {
+      "epoch": 0.23755386449069404,
+      "learning_rate": 0.0001524910607866508,
+      "loss": 0.8546,
+      "mean_token_accuracy": 0.7477369487285614,
+      "num_tokens": 59935411.0,
+      "step": 25910
+    },
+    {
+      "epoch": 0.2376455487301733,
+      "learning_rate": 0.00015247272393875493,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7396930694580078,
+      "num_tokens": 59958666.0,
+      "step": 25920
+    },
+    {
+      "epoch": 0.23773723296965252,
+      "learning_rate": 0.0001524543870908591,
+      "loss": 0.8279,
+      "mean_token_accuracy": 0.7489610254764557,
+      "num_tokens": 59981887.0,
+      "step": 25930
+    },
+    {
+      "epoch": 0.23782891720913174,
+      "learning_rate": 0.00015243605024296323,
+      "loss": 0.8792,
+      "mean_token_accuracy": 0.7404627442359925,
+      "num_tokens": 60006042.0,
+      "step": 25940
+    },
+    {
+      "epoch": 0.237920601448611,
+      "learning_rate": 0.00015241771339506737,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7439409255981445,
+      "num_tokens": 60030124.0,
+      "step": 25950
+    },
+    {
+      "epoch": 0.2380122856880902,
+      "learning_rate": 0.00015239937654717157,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7414111018180847,
+      "num_tokens": 60053152.0,
+      "step": 25960
+    },
+    {
+      "epoch": 0.23810396992756946,
+      "learning_rate": 0.0001523810396992757,
+      "loss": 0.8807,
+      "mean_token_accuracy": 0.7420764863491058,
+      "num_tokens": 60076507.0,
+      "step": 25970
+    },
+    {
+      "epoch": 0.2381956541670487,
+      "learning_rate": 0.00015236270285137987,
+      "loss": 0.8557,
+      "mean_token_accuracy": 0.7429322481155396,
+      "num_tokens": 60100421.0,
+      "step": 25980
+    },
+    {
+      "epoch": 0.2382873384065279,
+      "learning_rate": 0.000152344366003484,
+      "loss": 0.8628,
+      "mean_token_accuracy": 0.7405757248401642,
+      "num_tokens": 60122979.0,
+      "step": 25990
+    },
+    {
+      "epoch": 0.23837902264600716,
+      "learning_rate": 0.00015232602915558815,
+      "loss": 0.8441,
+      "mean_token_accuracy": 0.7469919562339783,
+      "num_tokens": 60145681.0,
+      "step": 26000
+    },
+    {
+      "epoch": 0.23847070688548638,
+      "learning_rate": 0.0001523076923076923,
+      "loss": 0.8747,
+      "mean_token_accuracy": 0.7382449448108673,
+      "num_tokens": 60168676.0,
+      "step": 26010
+    },
+    {
+      "epoch": 0.2385623911249656,
+      "learning_rate": 0.00015228935545979648,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7435623288154602,
+      "num_tokens": 60192045.0,
+      "step": 26020
+    },
+    {
+      "epoch": 0.23865407536444486,
+      "learning_rate": 0.00015227101861190064,
+      "loss": 0.8608,
+      "mean_token_accuracy": 0.7379026055335999,
+      "num_tokens": 60214867.0,
+      "step": 26030
+    },
+    {
+      "epoch": 0.23874575960392408,
+      "learning_rate": 0.00015225268176400478,
+      "loss": 0.8149,
+      "mean_token_accuracy": 0.7526860177516937,
+      "num_tokens": 60238420.0,
+      "step": 26040
+    },
+    {
+      "epoch": 0.23883744384340333,
+      "learning_rate": 0.00015223434491610892,
+      "loss": 0.8896,
+      "mean_token_accuracy": 0.7381485342979431,
+      "num_tokens": 60261502.0,
+      "step": 26050
+    },
+    {
+      "epoch": 0.23892912808288255,
+      "learning_rate": 0.00015221600806821308,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7382685422897339,
+      "num_tokens": 60285200.0,
+      "step": 26060
+    },
+    {
+      "epoch": 0.23902081232236178,
+      "learning_rate": 0.00015219767122031722,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7376444458961486,
+      "num_tokens": 60309846.0,
+      "step": 26070
+    },
+    {
+      "epoch": 0.23911249656184103,
+      "learning_rate": 0.0001521793343724214,
+      "loss": 0.8757,
+      "mean_token_accuracy": 0.7434535145759582,
+      "num_tokens": 60332933.0,
+      "step": 26080
+    },
+    {
+      "epoch": 0.23920418080132025,
+      "learning_rate": 0.00015216099752452555,
+      "loss": 0.9186,
+      "mean_token_accuracy": 0.7292102634906769,
+      "num_tokens": 60356626.0,
+      "step": 26090
+    },
+    {
+      "epoch": 0.2392958650407995,
+      "learning_rate": 0.0001521426606766297,
+      "loss": 0.8738,
+      "mean_token_accuracy": 0.7410090923309326,
+      "num_tokens": 60380610.0,
+      "step": 26100
+    },
+    {
+      "epoch": 0.23938754928027872,
+      "learning_rate": 0.00015212432382873386,
+      "loss": 0.8837,
+      "mean_token_accuracy": 0.7353364884853363,
+      "num_tokens": 60402937.0,
+      "step": 26110
+    },
+    {
+      "epoch": 0.23947923351975794,
+      "learning_rate": 0.000152105986980838,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7505240142345428,
+      "num_tokens": 60426106.0,
+      "step": 26120
+    },
+    {
+      "epoch": 0.2395709177592372,
+      "learning_rate": 0.00015208765013294216,
+      "loss": 0.8653,
+      "mean_token_accuracy": 0.7438559412956238,
+      "num_tokens": 60448246.0,
+      "step": 26130
+    },
+    {
+      "epoch": 0.23966260199871642,
+      "learning_rate": 0.0001520693132850463,
+      "loss": 0.8354,
+      "mean_token_accuracy": 0.7442316055297852,
+      "num_tokens": 60471616.0,
+      "step": 26140
+    },
+    {
+      "epoch": 0.23975428623819564,
+      "learning_rate": 0.00015205097643715046,
+      "loss": 0.8599,
+      "mean_token_accuracy": 0.7502242922782898,
+      "num_tokens": 60495373.0,
+      "step": 26150
+    },
+    {
+      "epoch": 0.2398459704776749,
+      "learning_rate": 0.00015203263958925463,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7505661010742187,
+      "num_tokens": 60519805.0,
+      "step": 26160
+    },
+    {
+      "epoch": 0.23993765471715411,
+      "learning_rate": 0.00015201430274135877,
+      "loss": 0.8457,
+      "mean_token_accuracy": 0.750264585018158,
+      "num_tokens": 60543982.0,
+      "step": 26170
+    },
+    {
+      "epoch": 0.24002933895663336,
+      "learning_rate": 0.00015199596589346293,
+      "loss": 0.8764,
+      "mean_token_accuracy": 0.7347649276256562,
+      "num_tokens": 60567493.0,
+      "step": 26180
+    },
+    {
+      "epoch": 0.2401210231961126,
+      "learning_rate": 0.00015197762904556707,
+      "loss": 0.856,
+      "mean_token_accuracy": 0.7419838309288025,
+      "num_tokens": 60589884.0,
+      "step": 26190
+    },
+    {
+      "epoch": 0.2402127074355918,
+      "learning_rate": 0.0001519592921976712,
+      "loss": 0.8774,
+      "mean_token_accuracy": 0.7397713422775268,
+      "num_tokens": 60613225.0,
+      "step": 26200
+    },
+    {
+      "epoch": 0.24030439167507106,
+      "learning_rate": 0.00015194095534977538,
+      "loss": 0.8617,
+      "mean_token_accuracy": 0.743641597032547,
+      "num_tokens": 60637045.0,
+      "step": 26210
+    },
+    {
+      "epoch": 0.24039607591455028,
+      "learning_rate": 0.00015192261850187954,
+      "loss": 0.832,
+      "mean_token_accuracy": 0.754779440164566,
+      "num_tokens": 60660055.0,
+      "step": 26220
+    },
+    {
+      "epoch": 0.24048776015402953,
+      "learning_rate": 0.0001519042816539837,
+      "loss": 0.8741,
+      "mean_token_accuracy": 0.7382874011993408,
+      "num_tokens": 60683596.0,
+      "step": 26230
+    },
+    {
+      "epoch": 0.24057944439350876,
+      "learning_rate": 0.00015188594480608784,
+      "loss": 0.8374,
+      "mean_token_accuracy": 0.7444458305835724,
+      "num_tokens": 60706186.0,
+      "step": 26240
+    },
+    {
+      "epoch": 0.24067112863298798,
+      "learning_rate": 0.00015186760795819198,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7413602232933044,
+      "num_tokens": 60729319.0,
+      "step": 26250
+    },
+    {
+      "epoch": 0.24076281287246723,
+      "learning_rate": 0.00015184927111029615,
+      "loss": 0.8076,
+      "mean_token_accuracy": 0.7561198234558105,
+      "num_tokens": 60752702.0,
+      "step": 26260
+    },
+    {
+      "epoch": 0.24085449711194645,
+      "learning_rate": 0.0001518309342624003,
+      "loss": 0.91,
+      "mean_token_accuracy": 0.7317551493644714,
+      "num_tokens": 60775931.0,
+      "step": 26270
+    },
+    {
+      "epoch": 0.2409461813514257,
+      "learning_rate": 0.00015181259741450445,
+      "loss": 0.8449,
+      "mean_token_accuracy": 0.7484565675258636,
+      "num_tokens": 60798616.0,
+      "step": 26280
+    },
+    {
+      "epoch": 0.24103786559090493,
+      "learning_rate": 0.00015179426056660862,
+      "loss": 0.8874,
+      "mean_token_accuracy": 0.7473255813121795,
+      "num_tokens": 60821762.0,
+      "step": 26290
+    },
+    {
+      "epoch": 0.24112954983038415,
+      "learning_rate": 0.00015177592371871276,
+      "loss": 0.8769,
+      "mean_token_accuracy": 0.7485851407051086,
+      "num_tokens": 60844276.0,
+      "step": 26300
+    },
+    {
+      "epoch": 0.2412212340698634,
+      "learning_rate": 0.00015175758687081692,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.741224330663681,
+      "num_tokens": 60867688.0,
+      "step": 26310
+    },
+    {
+      "epoch": 0.24131291830934262,
+      "learning_rate": 0.00015173925002292106,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7429096281528473,
+      "num_tokens": 60890773.0,
+      "step": 26320
+    },
+    {
+      "epoch": 0.24140460254882184,
+      "learning_rate": 0.00015172091317502522,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.7413500130176545,
+      "num_tokens": 60913280.0,
+      "step": 26330
+    },
+    {
+      "epoch": 0.2414962867883011,
+      "learning_rate": 0.00015170257632712936,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7433875560760498,
+      "num_tokens": 60936647.0,
+      "step": 26340
+    },
+    {
+      "epoch": 0.24158797102778032,
+      "learning_rate": 0.00015168423947923353,
+      "loss": 0.8056,
+      "mean_token_accuracy": 0.7550627827644348,
+      "num_tokens": 60959907.0,
+      "step": 26350
+    },
+    {
+      "epoch": 0.24167965526725957,
+      "learning_rate": 0.0001516659026313377,
+      "loss": 0.8746,
+      "mean_token_accuracy": 0.7450252413749695,
+      "num_tokens": 60982569.0,
+      "step": 26360
+    },
+    {
+      "epoch": 0.2417713395067388,
+      "learning_rate": 0.00015164756578344183,
+      "loss": 0.8338,
+      "mean_token_accuracy": 0.7551529586315155,
+      "num_tokens": 61005896.0,
+      "step": 26370
+    },
+    {
+      "epoch": 0.24186302374621801,
+      "learning_rate": 0.000151629228935546,
+      "loss": 0.8407,
+      "mean_token_accuracy": 0.7414148271083831,
+      "num_tokens": 61029129.0,
+      "step": 26380
+    },
+    {
+      "epoch": 0.24195470798569726,
+      "learning_rate": 0.00015161089208765014,
+      "loss": 0.8037,
+      "mean_token_accuracy": 0.7530421197414399,
+      "num_tokens": 61051764.0,
+      "step": 26390
+    },
+    {
+      "epoch": 0.2420463922251765,
+      "learning_rate": 0.00015159255523975427,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7420929193496704,
+      "num_tokens": 61074638.0,
+      "step": 26400
+    },
+    {
+      "epoch": 0.24213807646465574,
+      "learning_rate": 0.00015157421839185844,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7520487010478973,
+      "num_tokens": 61098008.0,
+      "step": 26410
+    },
+    {
+      "epoch": 0.24222976070413496,
+      "learning_rate": 0.0001515558815439626,
+      "loss": 0.8774,
+      "mean_token_accuracy": 0.7372593939304352,
+      "num_tokens": 61120785.0,
+      "step": 26420
+    },
+    {
+      "epoch": 0.24232144494361418,
+      "learning_rate": 0.00015153754469606677,
+      "loss": 0.9286,
+      "mean_token_accuracy": 0.7267246842384338,
+      "num_tokens": 61143530.0,
+      "step": 26430
+    },
+    {
+      "epoch": 0.24241312918309343,
+      "learning_rate": 0.0001515192078481709,
+      "loss": 0.8977,
+      "mean_token_accuracy": 0.7310905039310456,
+      "num_tokens": 61166665.0,
+      "step": 26440
+    },
+    {
+      "epoch": 0.24250481342257266,
+      "learning_rate": 0.00015150087100027505,
+      "loss": 0.9056,
+      "mean_token_accuracy": 0.7390459179878235,
+      "num_tokens": 61190312.0,
+      "step": 26450
+    },
+    {
+      "epoch": 0.24259649766205188,
+      "learning_rate": 0.0001514825341523792,
+      "loss": 0.8584,
+      "mean_token_accuracy": 0.7411555409431457,
+      "num_tokens": 61212338.0,
+      "step": 26460
+    },
+    {
+      "epoch": 0.24268818190153113,
+      "learning_rate": 0.00015146419730448335,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7455422639846802,
+      "num_tokens": 61235359.0,
+      "step": 26470
+    },
+    {
+      "epoch": 0.24277986614101035,
+      "learning_rate": 0.00015144586045658752,
+      "loss": 0.8569,
+      "mean_token_accuracy": 0.7438354790210724,
+      "num_tokens": 61258182.0,
+      "step": 26480
+    },
+    {
+      "epoch": 0.2428715503804896,
+      "learning_rate": 0.00015142752360869168,
+      "loss": 0.8839,
+      "mean_token_accuracy": 0.7364487528800965,
+      "num_tokens": 61280806.0,
+      "step": 26490
+    },
+    {
+      "epoch": 0.24296323461996883,
+      "learning_rate": 0.00015140918676079582,
+      "loss": 0.8501,
+      "mean_token_accuracy": 0.7426234364509583,
+      "num_tokens": 61304501.0,
+      "step": 26500
+    },
+    {
+      "epoch": 0.24305491885944805,
+      "learning_rate": 0.00015139084991289999,
+      "loss": 0.8674,
+      "mean_token_accuracy": 0.7410211980342865,
+      "num_tokens": 61327555.0,
+      "step": 26510
+    },
+    {
+      "epoch": 0.2431466030989273,
+      "learning_rate": 0.00015137251306500412,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7516920983791351,
+      "num_tokens": 61351270.0,
+      "step": 26520
+    },
+    {
+      "epoch": 0.24323828733840652,
+      "learning_rate": 0.0001513541762171083,
+      "loss": 0.8623,
+      "mean_token_accuracy": 0.7415857791900635,
+      "num_tokens": 61374573.0,
+      "step": 26530
+    },
+    {
+      "epoch": 0.24332997157788577,
+      "learning_rate": 0.00015133583936921243,
+      "loss": 0.8741,
+      "mean_token_accuracy": 0.7398116827011109,
+      "num_tokens": 61398198.0,
+      "step": 26540
+    },
+    {
+      "epoch": 0.243421655817365,
+      "learning_rate": 0.0001513175025213166,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7457467496395112,
+      "num_tokens": 61421330.0,
+      "step": 26550
+    },
+    {
+      "epoch": 0.24351334005684422,
+      "learning_rate": 0.00015129916567342076,
+      "loss": 0.8626,
+      "mean_token_accuracy": 0.7473425567150116,
+      "num_tokens": 61444872.0,
+      "step": 26560
+    },
+    {
+      "epoch": 0.24360502429632347,
+      "learning_rate": 0.0001512808288255249,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7467374980449677,
+      "num_tokens": 61468539.0,
+      "step": 26570
+    },
+    {
+      "epoch": 0.2436967085358027,
+      "learning_rate": 0.00015126249197762906,
+      "loss": 0.8967,
+      "mean_token_accuracy": 0.7340288460254669,
+      "num_tokens": 61491513.0,
+      "step": 26580
+    },
+    {
+      "epoch": 0.24378839277528194,
+      "learning_rate": 0.0001512441551297332,
+      "loss": 0.8572,
+      "mean_token_accuracy": 0.7451032221317291,
+      "num_tokens": 61514131.0,
+      "step": 26590
+    },
+    {
+      "epoch": 0.24388007701476117,
+      "learning_rate": 0.00015122581828183734,
+      "loss": 0.8729,
+      "mean_token_accuracy": 0.7386572897434235,
+      "num_tokens": 61538498.0,
+      "step": 26600
+    },
+    {
+      "epoch": 0.2439717612542404,
+      "learning_rate": 0.00015120748143394153,
+      "loss": 0.8251,
+      "mean_token_accuracy": 0.7478527426719666,
+      "num_tokens": 61562105.0,
+      "step": 26610
+    },
+    {
+      "epoch": 0.24406344549371964,
+      "learning_rate": 0.00015118914458604567,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.737487506866455,
+      "num_tokens": 61585307.0,
+      "step": 26620
+    },
+    {
+      "epoch": 0.24415512973319886,
+      "learning_rate": 0.00015117080773814984,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7511949062347412,
+      "num_tokens": 61608507.0,
+      "step": 26630
+    },
+    {
+      "epoch": 0.24424681397267808,
+      "learning_rate": 0.00015115247089025397,
+      "loss": 0.8279,
+      "mean_token_accuracy": 0.7499415636062622,
+      "num_tokens": 61631639.0,
+      "step": 26640
+    },
+    {
+      "epoch": 0.24433849821215733,
+      "learning_rate": 0.0001511341340423581,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7521292626857757,
+      "num_tokens": 61654814.0,
+      "step": 26650
+    },
+    {
+      "epoch": 0.24443018245163656,
+      "learning_rate": 0.00015111579719446228,
+      "loss": 0.8268,
+      "mean_token_accuracy": 0.7496983349323273,
+      "num_tokens": 61678882.0,
+      "step": 26660
+    },
+    {
+      "epoch": 0.2445218666911158,
+      "learning_rate": 0.00015109746034656642,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7474999368190766,
+      "num_tokens": 61702379.0,
+      "step": 26670
+    },
+    {
+      "epoch": 0.24461355093059503,
+      "learning_rate": 0.00015107912349867058,
+      "loss": 0.8533,
+      "mean_token_accuracy": 0.7445842027664185,
+      "num_tokens": 61726096.0,
+      "step": 26680
+    },
+    {
+      "epoch": 0.24470523517007425,
+      "learning_rate": 0.00015106078665077475,
+      "loss": 0.9065,
+      "mean_token_accuracy": 0.7378456234931946,
+      "num_tokens": 61748683.0,
+      "step": 26690
+    },
+    {
+      "epoch": 0.2447969194095535,
+      "learning_rate": 0.00015104244980287888,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7429121255874633,
+      "num_tokens": 61772500.0,
+      "step": 26700
+    },
+    {
+      "epoch": 0.24488860364903273,
+      "learning_rate": 0.00015102411295498305,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.7383490979671479,
+      "num_tokens": 61796587.0,
+      "step": 26710
+    },
+    {
+      "epoch": 0.24498028788851198,
+      "learning_rate": 0.0001510057761070872,
+      "loss": 0.893,
+      "mean_token_accuracy": 0.7369406044483184,
+      "num_tokens": 61819163.0,
+      "step": 26720
+    },
+    {
+      "epoch": 0.2450719721279912,
+      "learning_rate": 0.00015098743925919135,
+      "loss": 0.8101,
+      "mean_token_accuracy": 0.7523659229278564,
+      "num_tokens": 61842815.0,
+      "step": 26730
+    },
+    {
+      "epoch": 0.24516365636747042,
+      "learning_rate": 0.00015096910241129552,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7446376144886017,
+      "num_tokens": 61866346.0,
+      "step": 26740
+    },
+    {
+      "epoch": 0.24525534060694967,
+      "learning_rate": 0.00015095076556339966,
+      "loss": 0.881,
+      "mean_token_accuracy": 0.7377575397491455,
+      "num_tokens": 61890179.0,
+      "step": 26750
+    },
+    {
+      "epoch": 0.2453470248464289,
+      "learning_rate": 0.00015093242871550382,
+      "loss": 0.8813,
+      "mean_token_accuracy": 0.7363186776638031,
+      "num_tokens": 61913063.0,
+      "step": 26760
+    },
+    {
+      "epoch": 0.24543870908590812,
+      "learning_rate": 0.00015091409186760796,
+      "loss": 0.8611,
+      "mean_token_accuracy": 0.7483425676822663,
+      "num_tokens": 61935806.0,
+      "step": 26770
+    },
+    {
+      "epoch": 0.24553039332538737,
+      "learning_rate": 0.00015089575501971213,
+      "loss": 0.8642,
+      "mean_token_accuracy": 0.7431134819984436,
+      "num_tokens": 61959200.0,
+      "step": 26780
+    },
+    {
+      "epoch": 0.2456220775648666,
+      "learning_rate": 0.00015087741817181627,
+      "loss": 0.8142,
+      "mean_token_accuracy": 0.7511698305606842,
+      "num_tokens": 61982728.0,
+      "step": 26790
+    },
+    {
+      "epoch": 0.24571376180434584,
+      "learning_rate": 0.0001508590813239204,
+      "loss": 0.8368,
+      "mean_token_accuracy": 0.7514487862586975,
+      "num_tokens": 62006017.0,
+      "step": 26800
+    },
+    {
+      "epoch": 0.24580544604382507,
+      "learning_rate": 0.0001508407444760246,
+      "loss": 0.8538,
+      "mean_token_accuracy": 0.7411257028579712,
+      "num_tokens": 62029398.0,
+      "step": 26810
+    },
+    {
+      "epoch": 0.2458971302833043,
+      "learning_rate": 0.00015082240762812873,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.745817244052887,
+      "num_tokens": 62052495.0,
+      "step": 26820
+    },
+    {
+      "epoch": 0.24598881452278354,
+      "learning_rate": 0.0001508040707802329,
+      "loss": 0.8226,
+      "mean_token_accuracy": 0.75605508685112,
+      "num_tokens": 62075613.0,
+      "step": 26830
+    },
+    {
+      "epoch": 0.24608049876226276,
+      "learning_rate": 0.00015078573393233704,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7498820841312408,
+      "num_tokens": 62099819.0,
+      "step": 26840
+    },
+    {
+      "epoch": 0.246172183001742,
+      "learning_rate": 0.00015076739708444118,
+      "loss": 0.8277,
+      "mean_token_accuracy": 0.7518484711647033,
+      "num_tokens": 62121997.0,
+      "step": 26850
+    },
+    {
+      "epoch": 0.24626386724122123,
+      "learning_rate": 0.00015074906023654534,
+      "loss": 0.8767,
+      "mean_token_accuracy": 0.7354409396648407,
+      "num_tokens": 62145068.0,
+      "step": 26860
+    },
+    {
+      "epoch": 0.24635555148070046,
+      "learning_rate": 0.0001507307233886495,
+      "loss": 0.8836,
+      "mean_token_accuracy": 0.7420814752578735,
+      "num_tokens": 62168099.0,
+      "step": 26870
+    },
+    {
+      "epoch": 0.2464472357201797,
+      "learning_rate": 0.00015071238654075365,
+      "loss": 0.8605,
+      "mean_token_accuracy": 0.7410089373588562,
+      "num_tokens": 62191380.0,
+      "step": 26880
+    },
+    {
+      "epoch": 0.24653891995965893,
+      "learning_rate": 0.0001506940496928578,
+      "loss": 0.8738,
+      "mean_token_accuracy": 0.7459089696407318,
+      "num_tokens": 62214322.0,
+      "step": 26890
+    },
+    {
+      "epoch": 0.24663060419913818,
+      "learning_rate": 0.00015067571284496195,
+      "loss": 0.8597,
+      "mean_token_accuracy": 0.743206363916397,
+      "num_tokens": 62237955.0,
+      "step": 26900
+    },
+    {
+      "epoch": 0.2467222884386174,
+      "learning_rate": 0.00015065737599706611,
+      "loss": 0.895,
+      "mean_token_accuracy": 0.7290444374084473,
+      "num_tokens": 62260323.0,
+      "step": 26910
+    },
+    {
+      "epoch": 0.24681397267809663,
+      "learning_rate": 0.00015063903914917025,
+      "loss": 0.8891,
+      "mean_token_accuracy": 0.731582248210907,
+      "num_tokens": 62283170.0,
+      "step": 26920
+    },
+    {
+      "epoch": 0.24690565691757588,
+      "learning_rate": 0.00015062070230127442,
+      "loss": 0.8526,
+      "mean_token_accuracy": 0.7428878247737885,
+      "num_tokens": 62306287.0,
+      "step": 26930
+    },
+    {
+      "epoch": 0.2469973411570551,
+      "learning_rate": 0.00015060236545337858,
+      "loss": 0.875,
+      "mean_token_accuracy": 0.7361538946628571,
+      "num_tokens": 62329846.0,
+      "step": 26940
+    },
+    {
+      "epoch": 0.24708902539653432,
+      "learning_rate": 0.00015058402860548272,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.744611668586731,
+      "num_tokens": 62352767.0,
+      "step": 26950
+    },
+    {
+      "epoch": 0.24718070963601357,
+      "learning_rate": 0.0001505656917575869,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7452895998954773,
+      "num_tokens": 62375546.0,
+      "step": 26960
+    },
+    {
+      "epoch": 0.2472723938754928,
+      "learning_rate": 0.00015054735490969103,
+      "loss": 0.9066,
+      "mean_token_accuracy": 0.7247574865818024,
+      "num_tokens": 62397916.0,
+      "step": 26970
+    },
+    {
+      "epoch": 0.24736407811497205,
+      "learning_rate": 0.0001505290180617952,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7422787070274353,
+      "num_tokens": 62421526.0,
+      "step": 26980
+    },
+    {
+      "epoch": 0.24745576235445127,
+      "learning_rate": 0.00015051068121389933,
+      "loss": 0.8953,
+      "mean_token_accuracy": 0.736442220211029,
+      "num_tokens": 62444172.0,
+      "step": 26990
+    },
+    {
+      "epoch": 0.2475474465939305,
+      "learning_rate": 0.0001504923443660035,
+      "loss": 0.8931,
+      "mean_token_accuracy": 0.7396362960338593,
+      "num_tokens": 62466877.0,
+      "step": 27000
+    },
+    {
+      "epoch": 0.24763913083340974,
+      "learning_rate": 0.00015047400751810766,
+      "loss": 0.8734,
+      "mean_token_accuracy": 0.7421592712402344,
+      "num_tokens": 62489935.0,
+      "step": 27010
+    },
+    {
+      "epoch": 0.24773081507288897,
+      "learning_rate": 0.0001504556706702118,
+      "loss": 0.8624,
+      "mean_token_accuracy": 0.7441258907318116,
+      "num_tokens": 62513308.0,
+      "step": 27020
+    },
+    {
+      "epoch": 0.24782249931236822,
+      "learning_rate": 0.00015043733382231596,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7449506878852844,
+      "num_tokens": 62537081.0,
+      "step": 27030
+    },
+    {
+      "epoch": 0.24791418355184744,
+      "learning_rate": 0.0001504189969744201,
+      "loss": 0.8491,
+      "mean_token_accuracy": 0.7461920797824859,
+      "num_tokens": 62560291.0,
+      "step": 27040
+    },
+    {
+      "epoch": 0.24800586779132666,
+      "learning_rate": 0.00015040066012652424,
+      "loss": 0.8852,
+      "mean_token_accuracy": 0.7368533611297607,
+      "num_tokens": 62582713.0,
+      "step": 27050
+    },
+    {
+      "epoch": 0.2480975520308059,
+      "learning_rate": 0.0001503823232786284,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.7470174729824066,
+      "num_tokens": 62604820.0,
+      "step": 27060
+    },
+    {
+      "epoch": 0.24818923627028514,
+      "learning_rate": 0.00015036398643073257,
+      "loss": 0.8722,
+      "mean_token_accuracy": 0.7433541059494019,
+      "num_tokens": 62628708.0,
+      "step": 27070
+    },
+    {
+      "epoch": 0.24828092050976436,
+      "learning_rate": 0.0001503456495828367,
+      "loss": 0.8394,
+      "mean_token_accuracy": 0.7433184921741486,
+      "num_tokens": 62651787.0,
+      "step": 27080
+    },
+    {
+      "epoch": 0.2483726047492436,
+      "learning_rate": 0.00015032731273494088,
+      "loss": 0.8791,
+      "mean_token_accuracy": 0.7418513059616089,
+      "num_tokens": 62674438.0,
+      "step": 27090
+    },
+    {
+      "epoch": 0.24846428898872283,
+      "learning_rate": 0.00015030897588704501,
+      "loss": 0.8435,
+      "mean_token_accuracy": 0.745457261800766,
+      "num_tokens": 62697937.0,
+      "step": 27100
+    },
+    {
+      "epoch": 0.24855597322820208,
+      "learning_rate": 0.00015029063903914918,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7362743437290191,
+      "num_tokens": 62721792.0,
+      "step": 27110
+    },
+    {
+      "epoch": 0.2486476574676813,
+      "learning_rate": 0.00015027230219125332,
+      "loss": 0.8763,
+      "mean_token_accuracy": 0.7412759482860565,
+      "num_tokens": 62745298.0,
+      "step": 27120
+    },
+    {
+      "epoch": 0.24873934170716053,
+      "learning_rate": 0.00015025396534335748,
+      "loss": 0.8571,
+      "mean_token_accuracy": 0.7426306903362274,
+      "num_tokens": 62768695.0,
+      "step": 27130
+    },
+    {
+      "epoch": 0.24883102594663978,
+      "learning_rate": 0.00015023562849546165,
+      "loss": 0.8402,
+      "mean_token_accuracy": 0.7535669267177582,
+      "num_tokens": 62791289.0,
+      "step": 27140
+    },
+    {
+      "epoch": 0.248922710186119,
+      "learning_rate": 0.0001502172916475658,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.742111599445343,
+      "num_tokens": 62814872.0,
+      "step": 27150
+    },
+    {
+      "epoch": 0.24901439442559825,
+      "learning_rate": 0.00015019895479966995,
+      "loss": 0.8392,
+      "mean_token_accuracy": 0.7452911853790283,
+      "num_tokens": 62838316.0,
+      "step": 27160
+    },
+    {
+      "epoch": 0.24910607866507747,
+      "learning_rate": 0.0001501806179517741,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.7459239482879638,
+      "num_tokens": 62861706.0,
+      "step": 27170
+    },
+    {
+      "epoch": 0.2491977629045567,
+      "learning_rate": 0.00015016228110387826,
+      "loss": 0.9038,
+      "mean_token_accuracy": 0.7337956607341767,
+      "num_tokens": 62884501.0,
+      "step": 27180
+    },
+    {
+      "epoch": 0.24928944714403595,
+      "learning_rate": 0.0001501439442559824,
+      "loss": 0.8656,
+      "mean_token_accuracy": 0.7398990273475647,
+      "num_tokens": 62908622.0,
+      "step": 27190
+    },
+    {
+      "epoch": 0.24938113138351517,
+      "learning_rate": 0.00015012560740808656,
+      "loss": 0.8806,
+      "mean_token_accuracy": 0.7421919882297516,
+      "num_tokens": 62931700.0,
+      "step": 27200
+    },
+    {
+      "epoch": 0.24947281562299442,
+      "learning_rate": 0.00015010727056019072,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.737982589006424,
+      "num_tokens": 62954404.0,
+      "step": 27210
+    },
+    {
+      "epoch": 0.24956449986247364,
+      "learning_rate": 0.00015008893371229486,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7433018922805786,
+      "num_tokens": 62977991.0,
+      "step": 27220
+    },
+    {
+      "epoch": 0.24965618410195287,
+      "learning_rate": 0.00015007059686439903,
+      "loss": 0.7948,
+      "mean_token_accuracy": 0.7564925909042358,
+      "num_tokens": 63000780.0,
+      "step": 27230
+    },
+    {
+      "epoch": 0.24974786834143212,
+      "learning_rate": 0.00015005226001650317,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.7499097645282745,
+      "num_tokens": 63024223.0,
+      "step": 27240
+    },
+    {
+      "epoch": 0.24983955258091134,
+      "learning_rate": 0.0001500339231686073,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7465799510478973,
+      "num_tokens": 63047655.0,
+      "step": 27250
+    },
+    {
+      "epoch": 0.24993123682039056,
+      "learning_rate": 0.0001500155863207115,
+      "loss": 0.8161,
+      "mean_token_accuracy": 0.7452511370182038,
+      "num_tokens": 63070714.0,
+      "step": 27260
+    },
+    {
+      "epoch": 0.2500229210598698,
+      "learning_rate": 0.00014999724947281564,
+      "loss": 0.817,
+      "mean_token_accuracy": 0.7468233466148376,
+      "num_tokens": 63093706.0,
+      "step": 27270
+    },
+    {
+      "epoch": 0.25011460529934904,
+      "learning_rate": 0.00014997891262491977,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7532055139541626,
+      "num_tokens": 63116413.0,
+      "step": 27280
+    },
+    {
+      "epoch": 0.2502062895388283,
+      "learning_rate": 0.00014996057577702394,
+      "loss": 0.841,
+      "mean_token_accuracy": 0.7506026148796081,
+      "num_tokens": 63140020.0,
+      "step": 27290
+    },
+    {
+      "epoch": 0.25029797377830754,
+      "learning_rate": 0.00014994223892912808,
+      "loss": 0.9062,
+      "mean_token_accuracy": 0.732699579000473,
+      "num_tokens": 63163403.0,
+      "step": 27300
+    },
+    {
+      "epoch": 0.25038965801778673,
+      "learning_rate": 0.00014992390208123224,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7432677268981933,
+      "num_tokens": 63186589.0,
+      "step": 27310
+    },
+    {
+      "epoch": 0.250481342257266,
+      "learning_rate": 0.00014990556523333638,
+      "loss": 0.8401,
+      "mean_token_accuracy": 0.7463052451610566,
+      "num_tokens": 63211032.0,
+      "step": 27320
+    },
+    {
+      "epoch": 0.25057302649674523,
+      "learning_rate": 0.00014988722838544055,
+      "loss": 0.8661,
+      "mean_token_accuracy": 0.7454921841621399,
+      "num_tokens": 63233942.0,
+      "step": 27330
+    },
+    {
+      "epoch": 0.2506647107362244,
+      "learning_rate": 0.0001498688915375447,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7432247877120972,
+      "num_tokens": 63256252.0,
+      "step": 27340
+    },
+    {
+      "epoch": 0.2507563949757037,
+      "learning_rate": 0.00014985055468964885,
+      "loss": 0.8242,
+      "mean_token_accuracy": 0.7477229058742523,
+      "num_tokens": 63280037.0,
+      "step": 27350
+    },
+    {
+      "epoch": 0.25084807921518293,
+      "learning_rate": 0.00014983221784175302,
+      "loss": 0.8703,
+      "mean_token_accuracy": 0.7464641690254211,
+      "num_tokens": 63302481.0,
+      "step": 27360
+    },
+    {
+      "epoch": 0.2509397634546621,
+      "learning_rate": 0.00014981388099385715,
+      "loss": 0.8309,
+      "mean_token_accuracy": 0.7471476495265961,
+      "num_tokens": 63325954.0,
+      "step": 27370
+    },
+    {
+      "epoch": 0.2510314476941414,
+      "learning_rate": 0.00014979554414596132,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.7405181527137756,
+      "num_tokens": 63348730.0,
+      "step": 27380
+    },
+    {
+      "epoch": 0.2511231319336206,
+      "learning_rate": 0.00014977720729806549,
+      "loss": 0.8874,
+      "mean_token_accuracy": 0.7401892364025116,
+      "num_tokens": 63370900.0,
+      "step": 27390
+    },
+    {
+      "epoch": 0.2512148161730998,
+      "learning_rate": 0.00014975887045016962,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.743671715259552,
+      "num_tokens": 63393906.0,
+      "step": 27400
+    },
+    {
+      "epoch": 0.25130650041257907,
+      "learning_rate": 0.0001497405336022738,
+      "loss": 0.8673,
+      "mean_token_accuracy": 0.7424066245555878,
+      "num_tokens": 63417663.0,
+      "step": 27410
+    },
+    {
+      "epoch": 0.2513981846520583,
+      "learning_rate": 0.00014972219675437793,
+      "loss": 0.8321,
+      "mean_token_accuracy": 0.7520144999027252,
+      "num_tokens": 63440724.0,
+      "step": 27420
+    },
+    {
+      "epoch": 0.25148986889153757,
+      "learning_rate": 0.0001497038599064821,
+      "loss": 0.859,
+      "mean_token_accuracy": 0.741386330127716,
+      "num_tokens": 63462970.0,
+      "step": 27430
+    },
+    {
+      "epoch": 0.25158155313101677,
+      "learning_rate": 0.00014968552305858623,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7527381777763367,
+      "num_tokens": 63485825.0,
+      "step": 27440
+    },
+    {
+      "epoch": 0.251673237370496,
+      "learning_rate": 0.00014966718621069037,
+      "loss": 0.8802,
+      "mean_token_accuracy": 0.739150595664978,
+      "num_tokens": 63508316.0,
+      "step": 27450
+    },
+    {
+      "epoch": 0.25176492160997527,
+      "learning_rate": 0.00014964884936279456,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7439385056495667,
+      "num_tokens": 63531956.0,
+      "step": 27460
+    },
+    {
+      "epoch": 0.25185660584945446,
+      "learning_rate": 0.0001496305125148987,
+      "loss": 0.8215,
+      "mean_token_accuracy": 0.7559373676776886,
+      "num_tokens": 63554944.0,
+      "step": 27470
+    },
+    {
+      "epoch": 0.2519482900889337,
+      "learning_rate": 0.00014961217566700284,
+      "loss": 0.8358,
+      "mean_token_accuracy": 0.7509729623794555,
+      "num_tokens": 63578468.0,
+      "step": 27480
+    },
+    {
+      "epoch": 0.25203997432841296,
+      "learning_rate": 0.000149593838819107,
+      "loss": 0.8713,
+      "mean_token_accuracy": 0.7407134234905243,
+      "num_tokens": 63601170.0,
+      "step": 27490
+    },
+    {
+      "epoch": 0.25213165856789216,
+      "learning_rate": 0.00014957550197121114,
+      "loss": 0.8394,
+      "mean_token_accuracy": 0.7518621146678924,
+      "num_tokens": 63625539.0,
+      "step": 27500
+    },
+    {
+      "epoch": 0.2522233428073714,
+      "learning_rate": 0.0001495571651233153,
+      "loss": 0.8454,
+      "mean_token_accuracy": 0.7450213611125946,
+      "num_tokens": 63649273.0,
+      "step": 27510
+    },
+    {
+      "epoch": 0.25231502704685066,
+      "learning_rate": 0.00014953882827541947,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.740081399679184,
+      "num_tokens": 63672100.0,
+      "step": 27520
+    },
+    {
+      "epoch": 0.25240671128632985,
+      "learning_rate": 0.0001495204914275236,
+      "loss": 0.8434,
+      "mean_token_accuracy": 0.7457730114459992,
+      "num_tokens": 63695148.0,
+      "step": 27530
+    },
+    {
+      "epoch": 0.2524983955258091,
+      "learning_rate": 0.00014950215457962778,
+      "loss": 0.8739,
+      "mean_token_accuracy": 0.739480984210968,
+      "num_tokens": 63718328.0,
+      "step": 27540
+    },
+    {
+      "epoch": 0.25259007976528836,
+      "learning_rate": 0.00014948381773173192,
+      "loss": 0.814,
+      "mean_token_accuracy": 0.7560616672039032,
+      "num_tokens": 63741194.0,
+      "step": 27550
+    },
+    {
+      "epoch": 0.2526817640047676,
+      "learning_rate": 0.00014946548088383608,
+      "loss": 0.8461,
+      "mean_token_accuracy": 0.7488047420978546,
+      "num_tokens": 63764382.0,
+      "step": 27560
+    },
+    {
+      "epoch": 0.2527734482442468,
+      "learning_rate": 0.00014944714403594022,
+      "loss": 0.8337,
+      "mean_token_accuracy": 0.7544873893260956,
+      "num_tokens": 63788497.0,
+      "step": 27570
+    },
+    {
+      "epoch": 0.25286513248372605,
+      "learning_rate": 0.00014942880718804438,
+      "loss": 0.8687,
+      "mean_token_accuracy": 0.7343391597270965,
+      "num_tokens": 63811426.0,
+      "step": 27580
+    },
+    {
+      "epoch": 0.2529568167232053,
+      "learning_rate": 0.00014941047034014855,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.7463156521320343,
+      "num_tokens": 63833993.0,
+      "step": 27590
+    },
+    {
+      "epoch": 0.2530485009626845,
+      "learning_rate": 0.0001493921334922527,
+      "loss": 0.8813,
+      "mean_token_accuracy": 0.7421612560749054,
+      "num_tokens": 63856580.0,
+      "step": 27600
+    },
+    {
+      "epoch": 0.25314018520216375,
+      "learning_rate": 0.00014937379664435685,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.7461703479290008,
+      "num_tokens": 63879804.0,
+      "step": 27610
+    },
+    {
+      "epoch": 0.253231869441643,
+      "learning_rate": 0.000149355459796461,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.744089663028717,
+      "num_tokens": 63902235.0,
+      "step": 27620
+    },
+    {
+      "epoch": 0.2533235536811222,
+      "learning_rate": 0.00014933712294856516,
+      "loss": 0.8367,
+      "mean_token_accuracy": 0.7552065253257751,
+      "num_tokens": 63924451.0,
+      "step": 27630
+    },
+    {
+      "epoch": 0.25341523792060144,
+      "learning_rate": 0.0001493187861006693,
+      "loss": 0.8366,
+      "mean_token_accuracy": 0.7422115445137024,
+      "num_tokens": 63947429.0,
+      "step": 27640
+    },
+    {
+      "epoch": 0.2535069221600807,
+      "learning_rate": 0.00014930044925277343,
+      "loss": 0.8403,
+      "mean_token_accuracy": 0.7481965363025666,
+      "num_tokens": 63970539.0,
+      "step": 27650
+    },
+    {
+      "epoch": 0.2535986063995599,
+      "learning_rate": 0.00014928211240487763,
+      "loss": 0.8954,
+      "mean_token_accuracy": 0.7365795850753785,
+      "num_tokens": 63994385.0,
+      "step": 27660
+    },
+    {
+      "epoch": 0.25369029063903914,
+      "learning_rate": 0.00014926377555698176,
+      "loss": 0.8895,
+      "mean_token_accuracy": 0.736483770608902,
+      "num_tokens": 64017109.0,
+      "step": 27670
+    },
+    {
+      "epoch": 0.2537819748785184,
+      "learning_rate": 0.0001492454387090859,
+      "loss": 0.794,
+      "mean_token_accuracy": 0.7664094746112824,
+      "num_tokens": 64040056.0,
+      "step": 27680
+    },
+    {
+      "epoch": 0.25387365911799764,
+      "learning_rate": 0.00014922710186119007,
+      "loss": 0.8667,
+      "mean_token_accuracy": 0.7422010958194732,
+      "num_tokens": 64063168.0,
+      "step": 27690
+    },
+    {
+      "epoch": 0.25396534335747684,
+      "learning_rate": 0.0001492087650132942,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.73812175989151,
+      "num_tokens": 64086260.0,
+      "step": 27700
+    },
+    {
+      "epoch": 0.2540570275969561,
+      "learning_rate": 0.00014919042816539837,
+      "loss": 0.8224,
+      "mean_token_accuracy": 0.7560637831687927,
+      "num_tokens": 64110765.0,
+      "step": 27710
+    },
+    {
+      "epoch": 0.25414871183643534,
+      "learning_rate": 0.00014917209131750254,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.7410921573638916,
+      "num_tokens": 64133797.0,
+      "step": 27720
+    },
+    {
+      "epoch": 0.25424039607591453,
+      "learning_rate": 0.00014915375446960668,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7404982089996338,
+      "num_tokens": 64156972.0,
+      "step": 27730
+    },
+    {
+      "epoch": 0.2543320803153938,
+      "learning_rate": 0.00014913541762171084,
+      "loss": 0.8336,
+      "mean_token_accuracy": 0.7471336543560028,
+      "num_tokens": 64180484.0,
+      "step": 27740
+    },
+    {
+      "epoch": 0.25442376455487303,
+      "learning_rate": 0.00014911708077381498,
+      "loss": 0.902,
+      "mean_token_accuracy": 0.7338102877140045,
+      "num_tokens": 64203458.0,
+      "step": 27750
+    },
+    {
+      "epoch": 0.25451544879435223,
+      "learning_rate": 0.00014909874392591915,
+      "loss": 0.8794,
+      "mean_token_accuracy": 0.742933839559555,
+      "num_tokens": 64226780.0,
+      "step": 27760
+    },
+    {
+      "epoch": 0.2546071330338315,
+      "learning_rate": 0.00014908040707802328,
+      "loss": 0.8968,
+      "mean_token_accuracy": 0.7303280770778656,
+      "num_tokens": 64250066.0,
+      "step": 27770
+    },
+    {
+      "epoch": 0.25469881727331073,
+      "learning_rate": 0.00014906207023012745,
+      "loss": 0.8363,
+      "mean_token_accuracy": 0.7506760716438293,
+      "num_tokens": 64272875.0,
+      "step": 27780
+    },
+    {
+      "epoch": 0.2547905015127899,
+      "learning_rate": 0.00014904373338223161,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7442705988883972,
+      "num_tokens": 64296047.0,
+      "step": 27790
+    },
+    {
+      "epoch": 0.2548821857522692,
+      "learning_rate": 0.00014902539653433575,
+      "loss": 0.8578,
+      "mean_token_accuracy": 0.749461966753006,
+      "num_tokens": 64318818.0,
+      "step": 27800
+    },
+    {
+      "epoch": 0.2549738699917484,
+      "learning_rate": 0.00014900705968643992,
+      "loss": 0.8371,
+      "mean_token_accuracy": 0.7498965322971344,
+      "num_tokens": 64342462.0,
+      "step": 27810
+    },
+    {
+      "epoch": 0.2550655542312277,
+      "learning_rate": 0.00014898872283854406,
+      "loss": 0.8697,
+      "mean_token_accuracy": 0.7387650489807129,
+      "num_tokens": 64365382.0,
+      "step": 27820
+    },
+    {
+      "epoch": 0.25515723847070687,
+      "learning_rate": 0.00014897038599064822,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.7320538461208344,
+      "num_tokens": 64388493.0,
+      "step": 27830
+    },
+    {
+      "epoch": 0.2552489227101861,
+      "learning_rate": 0.00014895204914275236,
+      "loss": 0.8112,
+      "mean_token_accuracy": 0.7556831479072571,
+      "num_tokens": 64411274.0,
+      "step": 27840
+    },
+    {
+      "epoch": 0.25534060694966537,
+      "learning_rate": 0.00014893371229485653,
+      "loss": 0.8541,
+      "mean_token_accuracy": 0.7366828501224518,
+      "num_tokens": 64433665.0,
+      "step": 27850
+    },
+    {
+      "epoch": 0.25543229118914457,
+      "learning_rate": 0.0001489153754469607,
+      "loss": 0.8689,
+      "mean_token_accuracy": 0.7413061380386352,
+      "num_tokens": 64456799.0,
+      "step": 27860
+    },
+    {
+      "epoch": 0.2555239754286238,
+      "learning_rate": 0.00014889703859906483,
+      "loss": 0.8824,
+      "mean_token_accuracy": 0.7397329449653626,
+      "num_tokens": 64480108.0,
+      "step": 27870
+    },
+    {
+      "epoch": 0.25561565966810307,
+      "learning_rate": 0.00014887870175116897,
+      "loss": 0.847,
+      "mean_token_accuracy": 0.749979555606842,
+      "num_tokens": 64503395.0,
+      "step": 27880
+    },
+    {
+      "epoch": 0.25570734390758226,
+      "learning_rate": 0.00014886036490327313,
+      "loss": 0.8703,
+      "mean_token_accuracy": 0.7492643415927887,
+      "num_tokens": 64526547.0,
+      "step": 27890
+    },
+    {
+      "epoch": 0.2557990281470615,
+      "learning_rate": 0.00014884202805537727,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7489019095897674,
+      "num_tokens": 64549687.0,
+      "step": 27900
+    },
+    {
+      "epoch": 0.25589071238654076,
+      "learning_rate": 0.00014882369120748144,
+      "loss": 0.8737,
+      "mean_token_accuracy": 0.7379236459732056,
+      "num_tokens": 64572307.0,
+      "step": 27910
+    },
+    {
+      "epoch": 0.25598239662602,
+      "learning_rate": 0.0001488053543595856,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.7451977670192719,
+      "num_tokens": 64595770.0,
+      "step": 27920
+    },
+    {
+      "epoch": 0.2560740808654992,
+      "learning_rate": 0.00014878701751168974,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7364241421222687,
+      "num_tokens": 64618894.0,
+      "step": 27930
+    },
+    {
+      "epoch": 0.25616576510497846,
+      "learning_rate": 0.0001487686806637939,
+      "loss": 0.8322,
+      "mean_token_accuracy": 0.7494477868080139,
+      "num_tokens": 64641565.0,
+      "step": 27940
+    },
+    {
+      "epoch": 0.2562574493444577,
+      "learning_rate": 0.00014875034381589804,
+      "loss": 0.8448,
+      "mean_token_accuracy": 0.7500940978527069,
+      "num_tokens": 64665238.0,
+      "step": 27950
+    },
+    {
+      "epoch": 0.2563491335839369,
+      "learning_rate": 0.0001487320069680022,
+      "loss": 0.8758,
+      "mean_token_accuracy": 0.7441661059856415,
+      "num_tokens": 64688977.0,
+      "step": 27960
+    },
+    {
+      "epoch": 0.25644081782341616,
+      "learning_rate": 0.00014871367012010635,
+      "loss": 0.832,
+      "mean_token_accuracy": 0.7484582185745239,
+      "num_tokens": 64711971.0,
+      "step": 27970
+    },
+    {
+      "epoch": 0.2565325020628954,
+      "learning_rate": 0.0001486953332722105,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7460843920707703,
+      "num_tokens": 64735269.0,
+      "step": 27980
+    },
+    {
+      "epoch": 0.2566241863023746,
+      "learning_rate": 0.00014867699642431468,
+      "loss": 0.9121,
+      "mean_token_accuracy": 0.7320519149303436,
+      "num_tokens": 64758426.0,
+      "step": 27990
+    },
+    {
+      "epoch": 0.25671587054185385,
+      "learning_rate": 0.00014865865957641882,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7345442593097686,
+      "num_tokens": 64781597.0,
+      "step": 28000
+    },
+    {
+      "epoch": 0.2568075547813331,
+      "learning_rate": 0.00014864032272852298,
+      "loss": 0.8328,
+      "mean_token_accuracy": 0.7472794055938721,
+      "num_tokens": 64805020.0,
+      "step": 28010
+    },
+    {
+      "epoch": 0.2568992390208123,
+      "learning_rate": 0.00014862198588062712,
+      "loss": 0.8096,
+      "mean_token_accuracy": 0.7494252622127533,
+      "num_tokens": 64828278.0,
+      "step": 28020
+    },
+    {
+      "epoch": 0.25699092326029155,
+      "learning_rate": 0.00014860364903273129,
+      "loss": 0.869,
+      "mean_token_accuracy": 0.7407665729522706,
+      "num_tokens": 64850841.0,
+      "step": 28030
+    },
+    {
+      "epoch": 0.2570826074997708,
+      "learning_rate": 0.00014858531218483542,
+      "loss": 0.8318,
+      "mean_token_accuracy": 0.7525594234466553,
+      "num_tokens": 64873107.0,
+      "step": 28040
+    },
+    {
+      "epoch": 0.25717429173925005,
+      "learning_rate": 0.0001485669753369396,
+      "loss": 0.8449,
+      "mean_token_accuracy": 0.7486140429973602,
+      "num_tokens": 64896635.0,
+      "step": 28050
+    },
+    {
+      "epoch": 0.25726597597872924,
+      "learning_rate": 0.00014854863848904376,
+      "loss": 0.9136,
+      "mean_token_accuracy": 0.7302045464515686,
+      "num_tokens": 64920819.0,
+      "step": 28060
+    },
+    {
+      "epoch": 0.2573576602182085,
+      "learning_rate": 0.0001485303016411479,
+      "loss": 0.8725,
+      "mean_token_accuracy": 0.7373661279678345,
+      "num_tokens": 64943333.0,
+      "step": 28070
+    },
+    {
+      "epoch": 0.25744934445768775,
+      "learning_rate": 0.00014851196479325203,
+      "loss": 0.8991,
+      "mean_token_accuracy": 0.7330242574214936,
+      "num_tokens": 64966510.0,
+      "step": 28080
+    },
+    {
+      "epoch": 0.25754102869716694,
+      "learning_rate": 0.0001484936279453562,
+      "loss": 0.8844,
+      "mean_token_accuracy": 0.7403602242469788,
+      "num_tokens": 64989669.0,
+      "step": 28090
+    },
+    {
+      "epoch": 0.2576327129366462,
+      "learning_rate": 0.00014847529109746034,
+      "loss": 0.8483,
+      "mean_token_accuracy": 0.7522640228271484,
+      "num_tokens": 65012931.0,
+      "step": 28100
+    },
+    {
+      "epoch": 0.25772439717612544,
+      "learning_rate": 0.00014845695424956453,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7439845263957977,
+      "num_tokens": 65035544.0,
+      "step": 28110
+    },
+    {
+      "epoch": 0.25781608141560464,
+      "learning_rate": 0.00014843861740166867,
+      "loss": 0.868,
+      "mean_token_accuracy": 0.7458803355693817,
+      "num_tokens": 65058158.0,
+      "step": 28120
+    },
+    {
+      "epoch": 0.2579077656550839,
+      "learning_rate": 0.0001484202805537728,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.7457037091255188,
+      "num_tokens": 65081277.0,
+      "step": 28130
+    },
+    {
+      "epoch": 0.25799944989456314,
+      "learning_rate": 0.00014840194370587697,
+      "loss": 0.8278,
+      "mean_token_accuracy": 0.756156575679779,
+      "num_tokens": 65104023.0,
+      "step": 28140
+    },
+    {
+      "epoch": 0.25809113413404233,
+      "learning_rate": 0.0001483836068579811,
+      "loss": 0.7927,
+      "mean_token_accuracy": 0.7523608386516571,
+      "num_tokens": 65127045.0,
+      "step": 28150
+    },
+    {
+      "epoch": 0.2581828183735216,
+      "learning_rate": 0.00014836527001008527,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.7322318434715271,
+      "num_tokens": 65150294.0,
+      "step": 28160
+    },
+    {
+      "epoch": 0.25827450261300083,
+      "learning_rate": 0.0001483469331621894,
+      "loss": 0.8567,
+      "mean_token_accuracy": 0.7436087250709533,
+      "num_tokens": 65172773.0,
+      "step": 28170
+    },
+    {
+      "epoch": 0.2583661868524801,
+      "learning_rate": 0.00014832859631429358,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7448115289211273,
+      "num_tokens": 65196296.0,
+      "step": 28180
+    },
+    {
+      "epoch": 0.2584578710919593,
+      "learning_rate": 0.00014831025946639774,
+      "loss": 0.8195,
+      "mean_token_accuracy": 0.7525208175182343,
+      "num_tokens": 65219469.0,
+      "step": 28190
+    },
+    {
+      "epoch": 0.25854955533143853,
+      "learning_rate": 0.00014829192261850188,
+      "loss": 0.8904,
+      "mean_token_accuracy": 0.735933917760849,
+      "num_tokens": 65242975.0,
+      "step": 28200
+    },
+    {
+      "epoch": 0.2586412395709178,
+      "learning_rate": 0.00014827358577060605,
+      "loss": 0.9345,
+      "mean_token_accuracy": 0.7288546979427337,
+      "num_tokens": 65265797.0,
+      "step": 28210
+    },
+    {
+      "epoch": 0.258732923810397,
+      "learning_rate": 0.00014825524892271019,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.736388212442398,
+      "num_tokens": 65288984.0,
+      "step": 28220
+    },
+    {
+      "epoch": 0.2588246080498762,
+      "learning_rate": 0.00014823691207481435,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7434767723083496,
+      "num_tokens": 65313165.0,
+      "step": 28230
+    },
+    {
+      "epoch": 0.2589162922893555,
+      "learning_rate": 0.00014821857522691852,
+      "loss": 0.8781,
+      "mean_token_accuracy": 0.7393793523311615,
+      "num_tokens": 65335543.0,
+      "step": 28240
+    },
+    {
+      "epoch": 0.25900797652883467,
+      "learning_rate": 0.00014820023837902265,
+      "loss": 0.8239,
+      "mean_token_accuracy": 0.7503474652767181,
+      "num_tokens": 65359806.0,
+      "step": 28250
+    },
+    {
+      "epoch": 0.2590996607683139,
+      "learning_rate": 0.00014818190153112682,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7347579777240754,
+      "num_tokens": 65382585.0,
+      "step": 28260
+    },
+    {
+      "epoch": 0.2591913450077932,
+      "learning_rate": 0.00014816356468323096,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7446154773235321,
+      "num_tokens": 65406208.0,
+      "step": 28270
+    },
+    {
+      "epoch": 0.25928302924727237,
+      "learning_rate": 0.0001481452278353351,
+      "loss": 0.8457,
+      "mean_token_accuracy": 0.7462586760520935,
+      "num_tokens": 65429819.0,
+      "step": 28280
+    },
+    {
+      "epoch": 0.2593747134867516,
+      "learning_rate": 0.00014812689098743926,
+      "loss": 0.8707,
+      "mean_token_accuracy": 0.7396560609340668,
+      "num_tokens": 65452921.0,
+      "step": 28290
+    },
+    {
+      "epoch": 0.25946639772623087,
+      "learning_rate": 0.0001481085541395434,
+      "loss": 0.8768,
+      "mean_token_accuracy": 0.7417271196842193,
+      "num_tokens": 65476641.0,
+      "step": 28300
+    },
+    {
+      "epoch": 0.2595580819657101,
+      "learning_rate": 0.0001480902172916476,
+      "loss": 0.8541,
+      "mean_token_accuracy": 0.7493411481380463,
+      "num_tokens": 65499848.0,
+      "step": 28310
+    },
+    {
+      "epoch": 0.2596497662051893,
+      "learning_rate": 0.00014807188044375173,
+      "loss": 0.8282,
+      "mean_token_accuracy": 0.7527835428714752,
+      "num_tokens": 65521786.0,
+      "step": 28320
+    },
+    {
+      "epoch": 0.25974145044466856,
+      "learning_rate": 0.00014805354359585587,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7430724859237671,
+      "num_tokens": 65544046.0,
+      "step": 28330
+    },
+    {
+      "epoch": 0.2598331346841478,
+      "learning_rate": 0.00014803520674796003,
+      "loss": 0.8215,
+      "mean_token_accuracy": 0.7531533479690552,
+      "num_tokens": 65566841.0,
+      "step": 28340
+    },
+    {
+      "epoch": 0.259924818923627,
+      "learning_rate": 0.00014801686990006417,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.7496873915195466,
+      "num_tokens": 65589363.0,
+      "step": 28350
+    },
+    {
+      "epoch": 0.26001650316310626,
+      "learning_rate": 0.00014799853305216834,
+      "loss": 0.8627,
+      "mean_token_accuracy": 0.7484847903251648,
+      "num_tokens": 65612743.0,
+      "step": 28360
+    },
+    {
+      "epoch": 0.2601081874025855,
+      "learning_rate": 0.0001479801962042725,
+      "loss": 0.8304,
+      "mean_token_accuracy": 0.7515036821365356,
+      "num_tokens": 65635628.0,
+      "step": 28370
+    },
+    {
+      "epoch": 0.2601998716420647,
+      "learning_rate": 0.00014796185935637664,
+      "loss": 0.8287,
+      "mean_token_accuracy": 0.7580148875713348,
+      "num_tokens": 65659904.0,
+      "step": 28380
+    },
+    {
+      "epoch": 0.26029155588154396,
+      "learning_rate": 0.0001479435225084808,
+      "loss": 0.8814,
+      "mean_token_accuracy": 0.7388691902160645,
+      "num_tokens": 65682954.0,
+      "step": 28390
+    },
+    {
+      "epoch": 0.2603832401210232,
+      "learning_rate": 0.00014792518566058495,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7522849917411805,
+      "num_tokens": 65706837.0,
+      "step": 28400
+    },
+    {
+      "epoch": 0.2604749243605024,
+      "learning_rate": 0.0001479068488126891,
+      "loss": 0.8412,
+      "mean_token_accuracy": 0.750131368637085,
+      "num_tokens": 65729766.0,
+      "step": 28410
+    },
+    {
+      "epoch": 0.26056660859998165,
+      "learning_rate": 0.00014788851196479325,
+      "loss": 0.8731,
+      "mean_token_accuracy": 0.7444917023181915,
+      "num_tokens": 65752106.0,
+      "step": 28420
+    },
+    {
+      "epoch": 0.2606582928394609,
+      "learning_rate": 0.00014787017511689742,
+      "loss": 0.8513,
+      "mean_token_accuracy": 0.7475113570690155,
+      "num_tokens": 65774994.0,
+      "step": 28430
+    },
+    {
+      "epoch": 0.26074997707894015,
+      "learning_rate": 0.00014785183826900158,
+      "loss": 0.8466,
+      "mean_token_accuracy": 0.7454447269439697,
+      "num_tokens": 65798226.0,
+      "step": 28440
+    },
+    {
+      "epoch": 0.26084166131841935,
+      "learning_rate": 0.00014783350142110572,
+      "loss": 0.8825,
+      "mean_token_accuracy": 0.7391950786113739,
+      "num_tokens": 65821982.0,
+      "step": 28450
+    },
+    {
+      "epoch": 0.2609333455578986,
+      "learning_rate": 0.00014781516457320988,
+      "loss": 0.8632,
+      "mean_token_accuracy": 0.7417172074317933,
+      "num_tokens": 65845979.0,
+      "step": 28460
+    },
+    {
+      "epoch": 0.26102502979737785,
+      "learning_rate": 0.00014779682772531402,
+      "loss": 0.8217,
+      "mean_token_accuracy": 0.7534918189048767,
+      "num_tokens": 65868720.0,
+      "step": 28470
+    },
+    {
+      "epoch": 0.26111671403685704,
+      "learning_rate": 0.00014777849087741816,
+      "loss": 0.8573,
+      "mean_token_accuracy": 0.740533709526062,
+      "num_tokens": 65893097.0,
+      "step": 28480
+    },
+    {
+      "epoch": 0.2612083982763363,
+      "learning_rate": 0.00014776015402952233,
+      "loss": 0.8681,
+      "mean_token_accuracy": 0.7428515791893006,
+      "num_tokens": 65916086.0,
+      "step": 28490
+    },
+    {
+      "epoch": 0.26130008251581555,
+      "learning_rate": 0.0001477418171816265,
+      "loss": 0.8632,
+      "mean_token_accuracy": 0.7424728989601135,
+      "num_tokens": 65938757.0,
+      "step": 28500
+    },
+    {
+      "epoch": 0.26139176675529474,
+      "learning_rate": 0.00014772348033373066,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.7380162298679351,
+      "num_tokens": 65962216.0,
+      "step": 28510
+    },
+    {
+      "epoch": 0.261483450994774,
+      "learning_rate": 0.0001477051434858348,
+      "loss": 0.8754,
+      "mean_token_accuracy": 0.7504049718379975,
+      "num_tokens": 65984995.0,
+      "step": 28520
+    },
+    {
+      "epoch": 0.26157513523425324,
+      "learning_rate": 0.00014768680663793893,
+      "loss": 0.8608,
+      "mean_token_accuracy": 0.7417503893375397,
+      "num_tokens": 66008025.0,
+      "step": 28530
+    },
+    {
+      "epoch": 0.2616668194737325,
+      "learning_rate": 0.0001476684697900431,
+      "loss": 0.8235,
+      "mean_token_accuracy": 0.7488840878009796,
+      "num_tokens": 66031154.0,
+      "step": 28540
+    },
+    {
+      "epoch": 0.2617585037132117,
+      "learning_rate": 0.00014765013294214724,
+      "loss": 0.8707,
+      "mean_token_accuracy": 0.7523537755012513,
+      "num_tokens": 66054178.0,
+      "step": 28550
+    },
+    {
+      "epoch": 0.26185018795269094,
+      "learning_rate": 0.0001476317960942514,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7428346157073975,
+      "num_tokens": 66077158.0,
+      "step": 28560
+    },
+    {
+      "epoch": 0.2619418721921702,
+      "learning_rate": 0.00014761345924635557,
+      "loss": 0.8502,
+      "mean_token_accuracy": 0.7426634967327118,
+      "num_tokens": 66099979.0,
+      "step": 28570
+    },
+    {
+      "epoch": 0.2620335564316494,
+      "learning_rate": 0.0001475951223984597,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.745317006111145,
+      "num_tokens": 66122506.0,
+      "step": 28580
+    },
+    {
+      "epoch": 0.26212524067112863,
+      "learning_rate": 0.00014757678555056387,
+      "loss": 0.8215,
+      "mean_token_accuracy": 0.7506672024726868,
+      "num_tokens": 66145201.0,
+      "step": 28590
+    },
+    {
+      "epoch": 0.2622169249106079,
+      "learning_rate": 0.000147558448702668,
+      "loss": 0.8259,
+      "mean_token_accuracy": 0.750843733549118,
+      "num_tokens": 66168813.0,
+      "step": 28600
+    },
+    {
+      "epoch": 0.2623086091500871,
+      "learning_rate": 0.00014754011185477218,
+      "loss": 0.821,
+      "mean_token_accuracy": 0.7530823886394501,
+      "num_tokens": 66192111.0,
+      "step": 28610
+    },
+    {
+      "epoch": 0.26240029338956633,
+      "learning_rate": 0.00014752177500687631,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.7440512835979461,
+      "num_tokens": 66215335.0,
+      "step": 28620
+    },
+    {
+      "epoch": 0.2624919776290456,
+      "learning_rate": 0.00014750343815898048,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.739682936668396,
+      "num_tokens": 66238485.0,
+      "step": 28630
+    },
+    {
+      "epoch": 0.2625836618685248,
+      "learning_rate": 0.00014748510131108465,
+      "loss": 0.8329,
+      "mean_token_accuracy": 0.7496993005275726,
+      "num_tokens": 66261317.0,
+      "step": 28640
+    },
+    {
+      "epoch": 0.262675346108004,
+      "learning_rate": 0.00014746676446318878,
+      "loss": 0.8359,
+      "mean_token_accuracy": 0.7488134264945984,
+      "num_tokens": 66285067.0,
+      "step": 28650
+    },
+    {
+      "epoch": 0.2627670303474833,
+      "learning_rate": 0.00014744842761529295,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.7370449006557465,
+      "num_tokens": 66308060.0,
+      "step": 28660
+    },
+    {
+      "epoch": 0.2628587145869625,
+      "learning_rate": 0.0001474300907673971,
+      "loss": 0.893,
+      "mean_token_accuracy": 0.739249175786972,
+      "num_tokens": 66332005.0,
+      "step": 28670
+    },
+    {
+      "epoch": 0.2629503988264417,
+      "learning_rate": 0.00014741175391950123,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7420450985431671,
+      "num_tokens": 66355210.0,
+      "step": 28680
+    },
+    {
+      "epoch": 0.263042083065921,
+      "learning_rate": 0.0001473934170716054,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.744170343875885,
+      "num_tokens": 66378389.0,
+      "step": 28690
+    },
+    {
+      "epoch": 0.2631337673054002,
+      "learning_rate": 0.00014737508022370956,
+      "loss": 0.8733,
+      "mean_token_accuracy": 0.7426165461540222,
+      "num_tokens": 66401536.0,
+      "step": 28700
+    },
+    {
+      "epoch": 0.2632254515448794,
+      "learning_rate": 0.00014735674337581372,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7362983167171478,
+      "num_tokens": 66424685.0,
+      "step": 28710
+    },
+    {
+      "epoch": 0.26331713578435867,
+      "learning_rate": 0.00014733840652791786,
+      "loss": 0.852,
+      "mean_token_accuracy": 0.74595428109169,
+      "num_tokens": 66447837.0,
+      "step": 28720
+    },
+    {
+      "epoch": 0.2634088200238379,
+      "learning_rate": 0.000147320069680022,
+      "loss": 0.8692,
+      "mean_token_accuracy": 0.7449594378471375,
+      "num_tokens": 66471085.0,
+      "step": 28730
+    },
+    {
+      "epoch": 0.2635005042633171,
+      "learning_rate": 0.00014730173283212616,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.7552474617958069,
+      "num_tokens": 66493931.0,
+      "step": 28740
+    },
+    {
+      "epoch": 0.26359218850279637,
+      "learning_rate": 0.0001472833959842303,
+      "loss": 0.852,
+      "mean_token_accuracy": 0.7436244904994964,
+      "num_tokens": 66516820.0,
+      "step": 28750
+    },
+    {
+      "epoch": 0.2636838727422756,
+      "learning_rate": 0.0001472650591363345,
+      "loss": 0.8601,
+      "mean_token_accuracy": 0.7436582863330841,
+      "num_tokens": 66540533.0,
+      "step": 28760
+    },
+    {
+      "epoch": 0.2637755569817548,
+      "learning_rate": 0.00014724672228843863,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.7424706399440766,
+      "num_tokens": 66563600.0,
+      "step": 28770
+    },
+    {
+      "epoch": 0.26386724122123406,
+      "learning_rate": 0.00014722838544054277,
+      "loss": 0.8647,
+      "mean_token_accuracy": 0.7430243372917176,
+      "num_tokens": 66586069.0,
+      "step": 28780
+    },
+    {
+      "epoch": 0.2639589254607133,
+      "learning_rate": 0.00014721004859264694,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7367045223712921,
+      "num_tokens": 66609021.0,
+      "step": 28790
+    },
+    {
+      "epoch": 0.26405060970019256,
+      "learning_rate": 0.00014719171174475108,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.741569983959198,
+      "num_tokens": 66631758.0,
+      "step": 28800
+    },
+    {
+      "epoch": 0.26414229393967176,
+      "learning_rate": 0.00014717337489685524,
+      "loss": 0.8512,
+      "mean_token_accuracy": 0.7436561405658721,
+      "num_tokens": 66654784.0,
+      "step": 28810
+    },
+    {
+      "epoch": 0.264233978179151,
+      "learning_rate": 0.00014715503804895938,
+      "loss": 0.8296,
+      "mean_token_accuracy": 0.748553729057312,
+      "num_tokens": 66677839.0,
+      "step": 28820
+    },
+    {
+      "epoch": 0.26432566241863026,
+      "learning_rate": 0.00014713670120106354,
+      "loss": 0.8694,
+      "mean_token_accuracy": 0.7424539923667908,
+      "num_tokens": 66699875.0,
+      "step": 28830
+    },
+    {
+      "epoch": 0.26441734665810945,
+      "learning_rate": 0.0001471183643531677,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7472703456878662,
+      "num_tokens": 66722994.0,
+      "step": 28840
+    },
+    {
+      "epoch": 0.2645090308975887,
+      "learning_rate": 0.00014710002750527185,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7443240582942963,
+      "num_tokens": 66745235.0,
+      "step": 28850
+    },
+    {
+      "epoch": 0.26460071513706795,
+      "learning_rate": 0.000147081690657376,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7454345524311066,
+      "num_tokens": 66768823.0,
+      "step": 28860
+    },
+    {
+      "epoch": 0.26469239937654715,
+      "learning_rate": 0.00014706335380948015,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7457144618034363,
+      "num_tokens": 66791689.0,
+      "step": 28870
+    },
+    {
+      "epoch": 0.2647840836160264,
+      "learning_rate": 0.0001470450169615843,
+      "loss": 0.906,
+      "mean_token_accuracy": 0.7367794573307037,
+      "num_tokens": 66814321.0,
+      "step": 28880
+    },
+    {
+      "epoch": 0.26487576785550565,
+      "learning_rate": 0.00014702668011368846,
+      "loss": 0.8623,
+      "mean_token_accuracy": 0.7381490409374237,
+      "num_tokens": 66837741.0,
+      "step": 28890
+    },
+    {
+      "epoch": 0.26496745209498485,
+      "learning_rate": 0.00014700834326579262,
+      "loss": 0.8665,
+      "mean_token_accuracy": 0.7440053343772888,
+      "num_tokens": 66860678.0,
+      "step": 28900
+    },
+    {
+      "epoch": 0.2650591363344641,
+      "learning_rate": 0.00014699000641789679,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7494698226451874,
+      "num_tokens": 66883296.0,
+      "step": 28910
+    },
+    {
+      "epoch": 0.26515082057394335,
+      "learning_rate": 0.00014697166957000092,
+      "loss": 0.8732,
+      "mean_token_accuracy": 0.7404204905033112,
+      "num_tokens": 66906013.0,
+      "step": 28920
+    },
+    {
+      "epoch": 0.2652425048134226,
+      "learning_rate": 0.00014695333272210506,
+      "loss": 0.8465,
+      "mean_token_accuracy": 0.7487307369709015,
+      "num_tokens": 66928597.0,
+      "step": 28930
+    },
+    {
+      "epoch": 0.2653341890529018,
+      "learning_rate": 0.00014693499587420923,
+      "loss": 0.8337,
+      "mean_token_accuracy": 0.7548036456108094,
+      "num_tokens": 66952014.0,
+      "step": 28940
+    },
+    {
+      "epoch": 0.26542587329238104,
+      "learning_rate": 0.00014691665902631337,
+      "loss": 0.8708,
+      "mean_token_accuracy": 0.7427916049957275,
+      "num_tokens": 66974983.0,
+      "step": 28950
+    },
+    {
+      "epoch": 0.2655175575318603,
+      "learning_rate": 0.00014689832217841756,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7510434329509735,
+      "num_tokens": 66997248.0,
+      "step": 28960
+    },
+    {
+      "epoch": 0.2656092417713395,
+      "learning_rate": 0.0001468799853305217,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7415893733501434,
+      "num_tokens": 67019587.0,
+      "step": 28970
+    },
+    {
+      "epoch": 0.26570092601081874,
+      "learning_rate": 0.00014686164848262584,
+      "loss": 0.8838,
+      "mean_token_accuracy": 0.7404287934303284,
+      "num_tokens": 67042656.0,
+      "step": 28980
+    },
+    {
+      "epoch": 0.265792610250298,
+      "learning_rate": 0.00014684331163473,
+      "loss": 0.8627,
+      "mean_token_accuracy": 0.7427705228328705,
+      "num_tokens": 67064991.0,
+      "step": 28990
+    },
+    {
+      "epoch": 0.2658842944897772,
+      "learning_rate": 0.00014682497478683414,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7465137422084809,
+      "num_tokens": 67088364.0,
+      "step": 29000
+    },
+    {
+      "epoch": 0.26597597872925643,
+      "learning_rate": 0.0001468066379389383,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7433292806148529,
+      "num_tokens": 67111344.0,
+      "step": 29010
+    },
+    {
+      "epoch": 0.2660676629687357,
+      "learning_rate": 0.00014678830109104244,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7543451845645904,
+      "num_tokens": 67134783.0,
+      "step": 29020
+    },
+    {
+      "epoch": 0.2661593472082149,
+      "learning_rate": 0.0001467699642431466,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7442370295524597,
+      "num_tokens": 67158060.0,
+      "step": 29030
+    },
+    {
+      "epoch": 0.26625103144769413,
+      "learning_rate": 0.00014675162739525077,
+      "loss": 0.8478,
+      "mean_token_accuracy": 0.7446286261081696,
+      "num_tokens": 67182099.0,
+      "step": 29040
+    },
+    {
+      "epoch": 0.2663427156871734,
+      "learning_rate": 0.0001467332905473549,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7392730414867401,
+      "num_tokens": 67205297.0,
+      "step": 29050
+    },
+    {
+      "epoch": 0.26643439992665263,
+      "learning_rate": 0.00014671495369945908,
+      "loss": 0.8619,
+      "mean_token_accuracy": 0.7441462278366089,
+      "num_tokens": 67227530.0,
+      "step": 29060
+    },
+    {
+      "epoch": 0.2665260841661318,
+      "learning_rate": 0.00014669661685156322,
+      "loss": 0.8509,
+      "mean_token_accuracy": 0.752511179447174,
+      "num_tokens": 67250416.0,
+      "step": 29070
+    },
+    {
+      "epoch": 0.2666177684056111,
+      "learning_rate": 0.00014667828000366735,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.7423738956451416,
+      "num_tokens": 67273483.0,
+      "step": 29080
+    },
+    {
+      "epoch": 0.26670945264509033,
+      "learning_rate": 0.00014665994315577155,
+      "loss": 0.888,
+      "mean_token_accuracy": 0.7406603693962097,
+      "num_tokens": 67296409.0,
+      "step": 29090
+    },
+    {
+      "epoch": 0.2668011368845695,
+      "learning_rate": 0.00014664160630787569,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7430307030677795,
+      "num_tokens": 67320120.0,
+      "step": 29100
+    },
+    {
+      "epoch": 0.2668928211240488,
+      "learning_rate": 0.00014662326945997985,
+      "loss": 0.8729,
+      "mean_token_accuracy": 0.7388255298137665,
+      "num_tokens": 67342508.0,
+      "step": 29110
+    },
+    {
+      "epoch": 0.266984505363528,
+      "learning_rate": 0.000146604932612084,
+      "loss": 0.8512,
+      "mean_token_accuracy": 0.7403953552246094,
+      "num_tokens": 67366231.0,
+      "step": 29120
+    },
+    {
+      "epoch": 0.2670761896030072,
+      "learning_rate": 0.00014658659576418813,
+      "loss": 0.845,
+      "mean_token_accuracy": 0.7455290079116821,
+      "num_tokens": 67389728.0,
+      "step": 29130
+    },
+    {
+      "epoch": 0.26716787384248647,
+      "learning_rate": 0.0001465682589162923,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7436563789844512,
+      "num_tokens": 67411895.0,
+      "step": 29140
+    },
+    {
+      "epoch": 0.2672595580819657,
+      "learning_rate": 0.00014654992206839643,
+      "loss": 0.858,
+      "mean_token_accuracy": 0.7372546017169952,
+      "num_tokens": 67435493.0,
+      "step": 29150
+    },
+    {
+      "epoch": 0.26735124232144497,
+      "learning_rate": 0.00014653158522050062,
+      "loss": 0.9159,
+      "mean_token_accuracy": 0.7308765232563019,
+      "num_tokens": 67458961.0,
+      "step": 29160
+    },
+    {
+      "epoch": 0.26744292656092417,
+      "learning_rate": 0.00014651324837260476,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7504374504089355,
+      "num_tokens": 67481489.0,
+      "step": 29170
+    },
+    {
+      "epoch": 0.2675346108004034,
+      "learning_rate": 0.0001464949115247089,
+      "loss": 0.8749,
+      "mean_token_accuracy": 0.7418615996837616,
+      "num_tokens": 67504256.0,
+      "step": 29180
+    },
+    {
+      "epoch": 0.26762629503988267,
+      "learning_rate": 0.00014647657467681307,
+      "loss": 0.8883,
+      "mean_token_accuracy": 0.7359951794147491,
+      "num_tokens": 67527126.0,
+      "step": 29190
+    },
+    {
+      "epoch": 0.26771797927936186,
+      "learning_rate": 0.0001464582378289172,
+      "loss": 0.8367,
+      "mean_token_accuracy": 0.746992540359497,
+      "num_tokens": 67550006.0,
+      "step": 29200
+    },
+    {
+      "epoch": 0.2678096635188411,
+      "learning_rate": 0.00014643990098102137,
+      "loss": 0.8372,
+      "mean_token_accuracy": 0.742882889509201,
+      "num_tokens": 67573479.0,
+      "step": 29210
+    },
+    {
+      "epoch": 0.26790134775832036,
+      "learning_rate": 0.00014642156413312553,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.7388453662395478,
+      "num_tokens": 67596265.0,
+      "step": 29220
+    },
+    {
+      "epoch": 0.26799303199779956,
+      "learning_rate": 0.00014640322728522967,
+      "loss": 0.8482,
+      "mean_token_accuracy": 0.750929081439972,
+      "num_tokens": 67619372.0,
+      "step": 29230
+    },
+    {
+      "epoch": 0.2680847162372788,
+      "learning_rate": 0.00014638489043733384,
+      "loss": 0.8274,
+      "mean_token_accuracy": 0.7511480689048767,
+      "num_tokens": 67642384.0,
+      "step": 29240
+    },
+    {
+      "epoch": 0.26817640047675806,
+      "learning_rate": 0.00014636655358943798,
+      "loss": 0.8599,
+      "mean_token_accuracy": 0.7407106041908265,
+      "num_tokens": 67665331.0,
+      "step": 29250
+    },
+    {
+      "epoch": 0.26826808471623725,
+      "learning_rate": 0.00014634821674154214,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7456808745861053,
+      "num_tokens": 67688857.0,
+      "step": 29260
+    },
+    {
+      "epoch": 0.2683597689557165,
+      "learning_rate": 0.00014632987989364628,
+      "loss": 0.8359,
+      "mean_token_accuracy": 0.7532157719135284,
+      "num_tokens": 67711592.0,
+      "step": 29270
+    },
+    {
+      "epoch": 0.26845145319519575,
+      "learning_rate": 0.00014631154304575042,
+      "loss": 0.8353,
+      "mean_token_accuracy": 0.7491818904876709,
+      "num_tokens": 67733929.0,
+      "step": 29280
+    },
+    {
+      "epoch": 0.268543137434675,
+      "learning_rate": 0.0001462932061978546,
+      "loss": 0.8838,
+      "mean_token_accuracy": 0.746704387664795,
+      "num_tokens": 67757078.0,
+      "step": 29290
+    },
+    {
+      "epoch": 0.2686348216741542,
+      "learning_rate": 0.00014627486934995875,
+      "loss": 0.9128,
+      "mean_token_accuracy": 0.736178743839264,
+      "num_tokens": 67780851.0,
+      "step": 29300
+    },
+    {
+      "epoch": 0.26872650591363345,
+      "learning_rate": 0.00014625653250206292,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7438143193721771,
+      "num_tokens": 67803841.0,
+      "step": 29310
+    },
+    {
+      "epoch": 0.2688181901531127,
+      "learning_rate": 0.00014623819565416705,
+      "loss": 0.8135,
+      "mean_token_accuracy": 0.7553632915019989,
+      "num_tokens": 67826609.0,
+      "step": 29320
+    },
+    {
+      "epoch": 0.2689098743925919,
+      "learning_rate": 0.0001462198588062712,
+      "loss": 0.8803,
+      "mean_token_accuracy": 0.7446336030960083,
+      "num_tokens": 67849713.0,
+      "step": 29330
+    },
+    {
+      "epoch": 0.26900155863207115,
+      "learning_rate": 0.00014620152195837536,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7467943906784058,
+      "num_tokens": 67872427.0,
+      "step": 29340
+    },
+    {
+      "epoch": 0.2690932428715504,
+      "learning_rate": 0.00014618318511047952,
+      "loss": 0.8214,
+      "mean_token_accuracy": 0.7474032640457153,
+      "num_tokens": 67895742.0,
+      "step": 29350
+    },
+    {
+      "epoch": 0.2691849271110296,
+      "learning_rate": 0.0001461648482625837,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7436695754528045,
+      "num_tokens": 67918365.0,
+      "step": 29360
+    },
+    {
+      "epoch": 0.26927661135050884,
+      "learning_rate": 0.00014614651141468783,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7457526862621308,
+      "num_tokens": 67941987.0,
+      "step": 29370
+    },
+    {
+      "epoch": 0.2693682955899881,
+      "learning_rate": 0.00014612817456679196,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.7480398297309876,
+      "num_tokens": 67963766.0,
+      "step": 29380
+    },
+    {
+      "epoch": 0.2694599798294673,
+      "learning_rate": 0.00014610983771889613,
+      "loss": 0.8155,
+      "mean_token_accuracy": 0.7545328974723816,
+      "num_tokens": 67986842.0,
+      "step": 29390
+    },
+    {
+      "epoch": 0.26955166406894654,
+      "learning_rate": 0.00014609150087100027,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.7504626631736755,
+      "num_tokens": 68010395.0,
+      "step": 29400
+    },
+    {
+      "epoch": 0.2696433483084258,
+      "learning_rate": 0.00014607316402310443,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.7495385110378265,
+      "num_tokens": 68033735.0,
+      "step": 29410
+    },
+    {
+      "epoch": 0.26973503254790504,
+      "learning_rate": 0.0001460548271752086,
+      "loss": 0.9072,
+      "mean_token_accuracy": 0.7287107169628143,
+      "num_tokens": 68056965.0,
+      "step": 29420
+    },
+    {
+      "epoch": 0.26982671678738424,
+      "learning_rate": 0.00014603649032731274,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7426390469074249,
+      "num_tokens": 68079781.0,
+      "step": 29430
+    },
+    {
+      "epoch": 0.2699184010268635,
+      "learning_rate": 0.0001460181534794169,
+      "loss": 0.8716,
+      "mean_token_accuracy": 0.7354677617549896,
+      "num_tokens": 68102197.0,
+      "step": 29440
+    },
+    {
+      "epoch": 0.27001008526634274,
+      "learning_rate": 0.00014599981663152104,
+      "loss": 0.8747,
+      "mean_token_accuracy": 0.7406946778297424,
+      "num_tokens": 68125530.0,
+      "step": 29450
+    },
+    {
+      "epoch": 0.27010176950582193,
+      "learning_rate": 0.0001459814797836252,
+      "loss": 0.8222,
+      "mean_token_accuracy": 0.749953418970108,
+      "num_tokens": 68148332.0,
+      "step": 29460
+    },
+    {
+      "epoch": 0.2701934537453012,
+      "learning_rate": 0.00014596314293572935,
+      "loss": 0.8235,
+      "mean_token_accuracy": 0.7534517705440521,
+      "num_tokens": 68171579.0,
+      "step": 29470
+    },
+    {
+      "epoch": 0.27028513798478043,
+      "learning_rate": 0.0001459448060878335,
+      "loss": 0.8509,
+      "mean_token_accuracy": 0.7438921391963959,
+      "num_tokens": 68195314.0,
+      "step": 29480
+    },
+    {
+      "epoch": 0.2703768222242596,
+      "learning_rate": 0.00014592646923993768,
+      "loss": 0.8843,
+      "mean_token_accuracy": 0.7390227317810059,
+      "num_tokens": 68218487.0,
+      "step": 29490
+    },
+    {
+      "epoch": 0.2704685064637389,
+      "learning_rate": 0.00014590813239204181,
+      "loss": 0.8721,
+      "mean_token_accuracy": 0.7434373795986176,
+      "num_tokens": 68242370.0,
+      "step": 29500
+    },
+    {
+      "epoch": 0.27056019070321813,
+      "learning_rate": 0.00014588979554414598,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7449617803096771,
+      "num_tokens": 68265183.0,
+      "step": 29510
+    },
+    {
+      "epoch": 0.2706518749426973,
+      "learning_rate": 0.00014587145869625012,
+      "loss": 0.8058,
+      "mean_token_accuracy": 0.755420982837677,
+      "num_tokens": 68287846.0,
+      "step": 29520
+    },
+    {
+      "epoch": 0.2707435591821766,
+      "learning_rate": 0.00014585312184835426,
+      "loss": 0.8157,
+      "mean_token_accuracy": 0.7547839224338532,
+      "num_tokens": 68311104.0,
+      "step": 29530
+    },
+    {
+      "epoch": 0.2708352434216558,
+      "learning_rate": 0.00014583478500045842,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7439712226390839,
+      "num_tokens": 68334408.0,
+      "step": 29540
+    },
+    {
+      "epoch": 0.2709269276611351,
+      "learning_rate": 0.0001458164481525626,
+      "loss": 0.8527,
+      "mean_token_accuracy": 0.7445021390914917,
+      "num_tokens": 68357512.0,
+      "step": 29550
+    },
+    {
+      "epoch": 0.27101861190061427,
+      "learning_rate": 0.00014579811130466675,
+      "loss": 0.8511,
+      "mean_token_accuracy": 0.7417372643947602,
+      "num_tokens": 68381227.0,
+      "step": 29560
+    },
+    {
+      "epoch": 0.2711102961400935,
+      "learning_rate": 0.0001457797744567709,
+      "loss": 0.8108,
+      "mean_token_accuracy": 0.7528151869773865,
+      "num_tokens": 68403866.0,
+      "step": 29570
+    },
+    {
+      "epoch": 0.27120198037957277,
+      "learning_rate": 0.00014576143760887503,
+      "loss": 0.8788,
+      "mean_token_accuracy": 0.745859557390213,
+      "num_tokens": 68426608.0,
+      "step": 29580
+    },
+    {
+      "epoch": 0.27129366461905197,
+      "learning_rate": 0.0001457431007609792,
+      "loss": 0.8505,
+      "mean_token_accuracy": 0.7478248178958893,
+      "num_tokens": 68449694.0,
+      "step": 29590
+    },
+    {
+      "epoch": 0.2713853488585312,
+      "learning_rate": 0.00014572476391308333,
+      "loss": 0.8461,
+      "mean_token_accuracy": 0.7508221387863159,
+      "num_tokens": 68472316.0,
+      "step": 29600
+    },
+    {
+      "epoch": 0.27147703309801047,
+      "learning_rate": 0.0001457064270651875,
+      "loss": 0.8502,
+      "mean_token_accuracy": 0.7456076145172119,
+      "num_tokens": 68494968.0,
+      "step": 29610
+    },
+    {
+      "epoch": 0.27156871733748966,
+      "learning_rate": 0.00014568809021729166,
+      "loss": 0.8447,
+      "mean_token_accuracy": 0.747329032421112,
+      "num_tokens": 68517740.0,
+      "step": 29620
+    },
+    {
+      "epoch": 0.2716604015769689,
+      "learning_rate": 0.0001456697533693958,
+      "loss": 0.8425,
+      "mean_token_accuracy": 0.7470325529575348,
+      "num_tokens": 68540265.0,
+      "step": 29630
+    },
+    {
+      "epoch": 0.27175208581644816,
+      "learning_rate": 0.00014565141652149997,
+      "loss": 0.8355,
+      "mean_token_accuracy": 0.7455933690071106,
+      "num_tokens": 68563435.0,
+      "step": 29640
+    },
+    {
+      "epoch": 0.27184377005592736,
+      "learning_rate": 0.0001456330796736041,
+      "loss": 0.8221,
+      "mean_token_accuracy": 0.7530646741390228,
+      "num_tokens": 68587537.0,
+      "step": 29650
+    },
+    {
+      "epoch": 0.2719354542954066,
+      "learning_rate": 0.00014561474282570827,
+      "loss": 0.8706,
+      "mean_token_accuracy": 0.7441581964492798,
+      "num_tokens": 68610865.0,
+      "step": 29660
+    },
+    {
+      "epoch": 0.27202713853488586,
+      "learning_rate": 0.0001455964059778124,
+      "loss": 0.8854,
+      "mean_token_accuracy": 0.729330575466156,
+      "num_tokens": 68633724.0,
+      "step": 29670
+    },
+    {
+      "epoch": 0.2721188227743651,
+      "learning_rate": 0.00014557806912991657,
+      "loss": 0.8566,
+      "mean_token_accuracy": 0.7444162905216217,
+      "num_tokens": 68656674.0,
+      "step": 29680
+    },
+    {
+      "epoch": 0.2722105070138443,
+      "learning_rate": 0.00014555973228202074,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.7447042107582093,
+      "num_tokens": 68680344.0,
+      "step": 29690
+    },
+    {
+      "epoch": 0.27230219125332356,
+      "learning_rate": 0.00014554139543412488,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7424675345420837,
+      "num_tokens": 68703847.0,
+      "step": 29700
+    },
+    {
+      "epoch": 0.2723938754928028,
+      "learning_rate": 0.00014552305858622904,
+      "loss": 0.8795,
+      "mean_token_accuracy": 0.7416525959968567,
+      "num_tokens": 68726794.0,
+      "step": 29710
+    },
+    {
+      "epoch": 0.272485559732282,
+      "learning_rate": 0.00014550472173833318,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7532901585102081,
+      "num_tokens": 68749658.0,
+      "step": 29720
+    },
+    {
+      "epoch": 0.27257724397176125,
+      "learning_rate": 0.00014548638489043732,
+      "loss": 0.8869,
+      "mean_token_accuracy": 0.7452648997306823,
+      "num_tokens": 68772465.0,
+      "step": 29730
+    },
+    {
+      "epoch": 0.2726689282112405,
+      "learning_rate": 0.0001454680480425415,
+      "loss": 0.894,
+      "mean_token_accuracy": 0.737343817949295,
+      "num_tokens": 68795707.0,
+      "step": 29740
+    },
+    {
+      "epoch": 0.2727606124507197,
+      "learning_rate": 0.00014544971119464565,
+      "loss": 0.8197,
+      "mean_token_accuracy": 0.7482154846191407,
+      "num_tokens": 68818929.0,
+      "step": 29750
+    },
+    {
+      "epoch": 0.27285229669019895,
+      "learning_rate": 0.00014543137434674982,
+      "loss": 0.828,
+      "mean_token_accuracy": 0.7502427756786346,
+      "num_tokens": 68841705.0,
+      "step": 29760
+    },
+    {
+      "epoch": 0.2729439809296782,
+      "learning_rate": 0.00014541303749885396,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7512379825115204,
+      "num_tokens": 68864949.0,
+      "step": 29770
+    },
+    {
+      "epoch": 0.27303566516915745,
+      "learning_rate": 0.0001453947006509581,
+      "loss": 0.9004,
+      "mean_token_accuracy": 0.7296261727809906,
+      "num_tokens": 68888372.0,
+      "step": 29780
+    },
+    {
+      "epoch": 0.27312734940863664,
+      "learning_rate": 0.00014537636380306226,
+      "loss": 0.8255,
+      "mean_token_accuracy": 0.7524441063404084,
+      "num_tokens": 68911681.0,
+      "step": 29790
+    },
+    {
+      "epoch": 0.2732190336481159,
+      "learning_rate": 0.0001453580269551664,
+      "loss": 0.8377,
+      "mean_token_accuracy": 0.7447475910186767,
+      "num_tokens": 68934440.0,
+      "step": 29800
+    },
+    {
+      "epoch": 0.27331071788759514,
+      "learning_rate": 0.0001453396901072706,
+      "loss": 0.9117,
+      "mean_token_accuracy": 0.7330490410327911,
+      "num_tokens": 68957962.0,
+      "step": 29810
+    },
+    {
+      "epoch": 0.27340240212707434,
+      "learning_rate": 0.00014532135325937473,
+      "loss": 0.8504,
+      "mean_token_accuracy": 0.7409867823123932,
+      "num_tokens": 68982125.0,
+      "step": 29820
+    },
+    {
+      "epoch": 0.2734940863665536,
+      "learning_rate": 0.00014530301641147887,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7477595150470734,
+      "num_tokens": 69005424.0,
+      "step": 29830
+    },
+    {
+      "epoch": 0.27358577060603284,
+      "learning_rate": 0.00014528467956358303,
+      "loss": 0.8487,
+      "mean_token_accuracy": 0.7439844191074372,
+      "num_tokens": 69028157.0,
+      "step": 29840
+    },
+    {
+      "epoch": 0.27367745484551204,
+      "learning_rate": 0.00014526634271568717,
+      "loss": 0.7956,
+      "mean_token_accuracy": 0.7536679446697235,
+      "num_tokens": 69051438.0,
+      "step": 29850
+    },
+    {
+      "epoch": 0.2737691390849913,
+      "learning_rate": 0.00014524800586779134,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7462322473526001,
+      "num_tokens": 69074184.0,
+      "step": 29860
+    },
+    {
+      "epoch": 0.27386082332447054,
+      "learning_rate": 0.0001452296690198955,
+      "loss": 0.8447,
+      "mean_token_accuracy": 0.7494982481002808,
+      "num_tokens": 69096820.0,
+      "step": 29870
+    },
+    {
+      "epoch": 0.27395250756394973,
+      "learning_rate": 0.00014521133217199964,
+      "loss": 0.8507,
+      "mean_token_accuracy": 0.7408902049064636,
+      "num_tokens": 69119900.0,
+      "step": 29880
+    },
+    {
+      "epoch": 0.274044191803429,
+      "learning_rate": 0.0001451929953241038,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.744701224565506,
+      "num_tokens": 69143006.0,
+      "step": 29890
+    },
+    {
+      "epoch": 0.27413587604290823,
+      "learning_rate": 0.00014517465847620794,
+      "loss": 0.8518,
+      "mean_token_accuracy": 0.746039605140686,
+      "num_tokens": 69166044.0,
+      "step": 29900
+    },
+    {
+      "epoch": 0.2742275602823875,
+      "learning_rate": 0.0001451563216283121,
+      "loss": 0.8569,
+      "mean_token_accuracy": 0.7490433156490326,
+      "num_tokens": 69188816.0,
+      "step": 29910
+    },
+    {
+      "epoch": 0.2743192445218667,
+      "learning_rate": 0.00014513798478041625,
+      "loss": 0.8696,
+      "mean_token_accuracy": 0.7380292892456055,
+      "num_tokens": 69211601.0,
+      "step": 29920
+    },
+    {
+      "epoch": 0.27441092876134593,
+      "learning_rate": 0.00014511964793252039,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7535680830478668,
+      "num_tokens": 69234842.0,
+      "step": 29930
+    },
+    {
+      "epoch": 0.2745026130008252,
+      "learning_rate": 0.00014510131108462458,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.7445323169231415,
+      "num_tokens": 69257837.0,
+      "step": 29940
+    },
+    {
+      "epoch": 0.2745942972403044,
+      "learning_rate": 0.00014508297423672872,
+      "loss": 0.8413,
+      "mean_token_accuracy": 0.7428589642047883,
+      "num_tokens": 69281484.0,
+      "step": 29950
+    },
+    {
+      "epoch": 0.2746859814797836,
+      "learning_rate": 0.00014506463738883288,
+      "loss": 0.8465,
+      "mean_token_accuracy": 0.7460168719291687,
+      "num_tokens": 69304230.0,
+      "step": 29960
+    },
+    {
+      "epoch": 0.2747776657192629,
+      "learning_rate": 0.00014504630054093702,
+      "loss": 0.8601,
+      "mean_token_accuracy": 0.7441195726394654,
+      "num_tokens": 69326555.0,
+      "step": 29970
+    },
+    {
+      "epoch": 0.27486934995874207,
+      "learning_rate": 0.00014502796369304116,
+      "loss": 0.8582,
+      "mean_token_accuracy": 0.7398589491844177,
+      "num_tokens": 69350255.0,
+      "step": 29980
+    },
+    {
+      "epoch": 0.2749610341982213,
+      "learning_rate": 0.00014500962684514532,
+      "loss": 0.9022,
+      "mean_token_accuracy": 0.7317766427993775,
+      "num_tokens": 69372911.0,
+      "step": 29990
+    },
+    {
+      "epoch": 0.27505271843770057,
+      "learning_rate": 0.00014499128999724946,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.7391605913639069,
+      "num_tokens": 69395940.0,
+      "step": 30000
+    },
+    {
+      "epoch": 0.27514440267717977,
+      "learning_rate": 0.00014497295314935365,
+      "loss": 0.9062,
+      "mean_token_accuracy": 0.7369853496551514,
+      "num_tokens": 69419661.0,
+      "step": 30010
+    },
+    {
+      "epoch": 0.275236086916659,
+      "learning_rate": 0.0001449546163014578,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7369670629501343,
+      "num_tokens": 69442493.0,
+      "step": 30020
+    },
+    {
+      "epoch": 0.27532777115613827,
+      "learning_rate": 0.00014493627945356193,
+      "loss": 0.8738,
+      "mean_token_accuracy": 0.7404657423496246,
+      "num_tokens": 69464509.0,
+      "step": 30030
+    },
+    {
+      "epoch": 0.2754194553956175,
+      "learning_rate": 0.0001449179426056661,
+      "loss": 0.8409,
+      "mean_token_accuracy": 0.7477413415908813,
+      "num_tokens": 69487946.0,
+      "step": 30040
+    },
+    {
+      "epoch": 0.2755111396350967,
+      "learning_rate": 0.00014489960575777023,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7451659798622131,
+      "num_tokens": 69510802.0,
+      "step": 30050
+    },
+    {
+      "epoch": 0.27560282387457596,
+      "learning_rate": 0.0001448812689098744,
+      "loss": 0.8637,
+      "mean_token_accuracy": 0.7422425389289856,
+      "num_tokens": 69533141.0,
+      "step": 30060
+    },
+    {
+      "epoch": 0.2756945081140552,
+      "learning_rate": 0.00014486293206197857,
+      "loss": 0.8407,
+      "mean_token_accuracy": 0.748996478319168,
+      "num_tokens": 69556322.0,
+      "step": 30070
+    },
+    {
+      "epoch": 0.2757861923535344,
+      "learning_rate": 0.0001448445952140827,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7472386837005616,
+      "num_tokens": 69579630.0,
+      "step": 30080
+    },
+    {
+      "epoch": 0.27587787659301366,
+      "learning_rate": 0.00014482625836618687,
+      "loss": 0.841,
+      "mean_token_accuracy": 0.7489966690540314,
+      "num_tokens": 69603221.0,
+      "step": 30090
+    },
+    {
+      "epoch": 0.2759695608324929,
+      "learning_rate": 0.000144807921518291,
+      "loss": 0.8907,
+      "mean_token_accuracy": 0.7398915767669678,
+      "num_tokens": 69626369.0,
+      "step": 30100
+    },
+    {
+      "epoch": 0.2760612450719721,
+      "learning_rate": 0.00014478958467039517,
+      "loss": 0.8266,
+      "mean_token_accuracy": 0.7532822728157044,
+      "num_tokens": 69649118.0,
+      "step": 30110
+    },
+    {
+      "epoch": 0.27615292931145136,
+      "learning_rate": 0.0001447712478224993,
+      "loss": 0.8502,
+      "mean_token_accuracy": 0.7412570297718049,
+      "num_tokens": 69671698.0,
+      "step": 30120
+    },
+    {
+      "epoch": 0.2762446135509306,
+      "learning_rate": 0.00014475291097460345,
+      "loss": 0.8868,
+      "mean_token_accuracy": 0.7416863083839417,
+      "num_tokens": 69694071.0,
+      "step": 30130
+    },
+    {
+      "epoch": 0.2763362977904098,
+      "learning_rate": 0.00014473457412670764,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7469041466712951,
+      "num_tokens": 69716804.0,
+      "step": 30140
+    },
+    {
+      "epoch": 0.27642798202988905,
+      "learning_rate": 0.00014471623727881178,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.7424174904823303,
+      "num_tokens": 69739564.0,
+      "step": 30150
+    },
+    {
+      "epoch": 0.2765196662693683,
+      "learning_rate": 0.00014469790043091595,
+      "loss": 0.8536,
+      "mean_token_accuracy": 0.7470351040363312,
+      "num_tokens": 69762953.0,
+      "step": 30160
+    },
+    {
+      "epoch": 0.27661135050884755,
+      "learning_rate": 0.00014467956358302008,
+      "loss": 0.847,
+      "mean_token_accuracy": 0.7427092969417572,
+      "num_tokens": 69786264.0,
+      "step": 30170
+    },
+    {
+      "epoch": 0.27670303474832675,
+      "learning_rate": 0.00014466122673512422,
+      "loss": 0.8849,
+      "mean_token_accuracy": 0.7354965507984161,
+      "num_tokens": 69809539.0,
+      "step": 30180
+    },
+    {
+      "epoch": 0.276794718987806,
+      "learning_rate": 0.0001446428898872284,
+      "loss": 0.8127,
+      "mean_token_accuracy": 0.7516207277774811,
+      "num_tokens": 69832795.0,
+      "step": 30190
+    },
+    {
+      "epoch": 0.27688640322728525,
+      "learning_rate": 0.00014462455303933255,
+      "loss": 0.8366,
+      "mean_token_accuracy": 0.7521384298801422,
+      "num_tokens": 69855803.0,
+      "step": 30200
+    },
+    {
+      "epoch": 0.27697808746676444,
+      "learning_rate": 0.00014460621619143672,
+      "loss": 0.8735,
+      "mean_token_accuracy": 0.7349007666110993,
+      "num_tokens": 69879872.0,
+      "step": 30210
+    },
+    {
+      "epoch": 0.2770697717062437,
+      "learning_rate": 0.00014458787934354086,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7428090989589691,
+      "num_tokens": 69903014.0,
+      "step": 30220
+    },
+    {
+      "epoch": 0.27716145594572295,
+      "learning_rate": 0.000144569542495645,
+      "loss": 0.8473,
+      "mean_token_accuracy": 0.7449588358402253,
+      "num_tokens": 69926856.0,
+      "step": 30230
+    },
+    {
+      "epoch": 0.27725314018520214,
+      "learning_rate": 0.00014455120564774916,
+      "loss": 0.8619,
+      "mean_token_accuracy": 0.7457121670246124,
+      "num_tokens": 69949635.0,
+      "step": 30240
+    },
+    {
+      "epoch": 0.2773448244246814,
+      "learning_rate": 0.0001445328687998533,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7396000623703003,
+      "num_tokens": 69972313.0,
+      "step": 30250
+    },
+    {
+      "epoch": 0.27743650866416064,
+      "learning_rate": 0.00014451453195195746,
+      "loss": 0.813,
+      "mean_token_accuracy": 0.7499053537845611,
+      "num_tokens": 69995105.0,
+      "step": 30260
+    },
+    {
+      "epoch": 0.2775281929036399,
+      "learning_rate": 0.00014449619510406163,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7377649426460267,
+      "num_tokens": 70018594.0,
+      "step": 30270
+    },
+    {
+      "epoch": 0.2776198771431191,
+      "learning_rate": 0.00014447785825616577,
+      "loss": 0.8321,
+      "mean_token_accuracy": 0.753245210647583,
+      "num_tokens": 70041959.0,
+      "step": 30280
+    },
+    {
+      "epoch": 0.27771156138259834,
+      "learning_rate": 0.00014445952140826993,
+      "loss": 0.8702,
+      "mean_token_accuracy": 0.7458824634552002,
+      "num_tokens": 70064932.0,
+      "step": 30290
+    },
+    {
+      "epoch": 0.2778032456220776,
+      "learning_rate": 0.00014444118456037407,
+      "loss": 0.8276,
+      "mean_token_accuracy": 0.7527470052242279,
+      "num_tokens": 70087530.0,
+      "step": 30300
+    },
+    {
+      "epoch": 0.2778949298615568,
+      "learning_rate": 0.00014442284771247824,
+      "loss": 0.8575,
+      "mean_token_accuracy": 0.7428588151931763,
+      "num_tokens": 70111207.0,
+      "step": 30310
+    },
+    {
+      "epoch": 0.27798661410103603,
+      "learning_rate": 0.00014440451086458238,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7492354273796081,
+      "num_tokens": 70133716.0,
+      "step": 30320
+    },
+    {
+      "epoch": 0.2780782983405153,
+      "learning_rate": 0.00014438617401668654,
+      "loss": 0.8378,
+      "mean_token_accuracy": 0.7398899555206299,
+      "num_tokens": 70156836.0,
+      "step": 30330
+    },
+    {
+      "epoch": 0.2781699825799945,
+      "learning_rate": 0.0001443678371687907,
+      "loss": 0.8777,
+      "mean_token_accuracy": 0.7388945102691651,
+      "num_tokens": 70179296.0,
+      "step": 30340
+    },
+    {
+      "epoch": 0.27826166681947373,
+      "learning_rate": 0.00014434950032089484,
+      "loss": 0.8441,
+      "mean_token_accuracy": 0.7435081481933594,
+      "num_tokens": 70202823.0,
+      "step": 30350
+    },
+    {
+      "epoch": 0.278353351058953,
+      "learning_rate": 0.000144331163472999,
+      "loss": 0.8823,
+      "mean_token_accuracy": 0.7377229452133178,
+      "num_tokens": 70226026.0,
+      "step": 30360
+    },
+    {
+      "epoch": 0.2784450352984322,
+      "learning_rate": 0.00014431282662510315,
+      "loss": 0.8192,
+      "mean_token_accuracy": 0.7503212690353394,
+      "num_tokens": 70249365.0,
+      "step": 30370
+    },
+    {
+      "epoch": 0.2785367195379114,
+      "learning_rate": 0.0001442944897772073,
+      "loss": 0.8929,
+      "mean_token_accuracy": 0.729200690984726,
+      "num_tokens": 70272518.0,
+      "step": 30380
+    },
+    {
+      "epoch": 0.2786284037773907,
+      "learning_rate": 0.00014427615292931145,
+      "loss": 0.8414,
+      "mean_token_accuracy": 0.7466009974479675,
+      "num_tokens": 70296084.0,
+      "step": 30390
+    },
+    {
+      "epoch": 0.2787200880168699,
+      "learning_rate": 0.00014425781608141562,
+      "loss": 0.81,
+      "mean_token_accuracy": 0.7537119925022125,
+      "num_tokens": 70319467.0,
+      "step": 30400
+    },
+    {
+      "epoch": 0.2788117722563491,
+      "learning_rate": 0.00014423947923351978,
+      "loss": 0.9073,
+      "mean_token_accuracy": 0.7421898543834686,
+      "num_tokens": 70342795.0,
+      "step": 30410
+    },
+    {
+      "epoch": 0.2789034564958284,
+      "learning_rate": 0.00014422114238562392,
+      "loss": 0.8277,
+      "mean_token_accuracy": 0.7475308477878571,
+      "num_tokens": 70366224.0,
+      "step": 30420
+    },
+    {
+      "epoch": 0.2789951407353076,
+      "learning_rate": 0.00014420280553772806,
+      "loss": 0.841,
+      "mean_token_accuracy": 0.751807564496994,
+      "num_tokens": 70389878.0,
+      "step": 30430
+    },
+    {
+      "epoch": 0.2790868249747868,
+      "learning_rate": 0.00014418446868983223,
+      "loss": 0.8571,
+      "mean_token_accuracy": 0.7408379077911377,
+      "num_tokens": 70413175.0,
+      "step": 30440
+    },
+    {
+      "epoch": 0.27917850921426607,
+      "learning_rate": 0.00014416613184193636,
+      "loss": 0.8473,
+      "mean_token_accuracy": 0.7510758996009826,
+      "num_tokens": 70436352.0,
+      "step": 30450
+    },
+    {
+      "epoch": 0.2792701934537453,
+      "learning_rate": 0.00014414779499404053,
+      "loss": 0.8255,
+      "mean_token_accuracy": 0.7521364748477936,
+      "num_tokens": 70459425.0,
+      "step": 30460
+    },
+    {
+      "epoch": 0.2793618776932245,
+      "learning_rate": 0.0001441294581461447,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.744724178314209,
+      "num_tokens": 70482852.0,
+      "step": 30470
+    },
+    {
+      "epoch": 0.27945356193270376,
+      "learning_rate": 0.00014411112129824883,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.7340844094753265,
+      "num_tokens": 70506594.0,
+      "step": 30480
+    },
+    {
+      "epoch": 0.279545246172183,
+      "learning_rate": 0.000144092784450353,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.744951730966568,
+      "num_tokens": 70529541.0,
+      "step": 30490
+    },
+    {
+      "epoch": 0.2796369304116622,
+      "learning_rate": 0.00014407444760245714,
+      "loss": 0.8694,
+      "mean_token_accuracy": 0.7414206326007843,
+      "num_tokens": 70553084.0,
+      "step": 30500
+    },
+    {
+      "epoch": 0.27972861465114146,
+      "learning_rate": 0.0001440561107545613,
+      "loss": 0.8501,
+      "mean_token_accuracy": 0.7483701586723328,
+      "num_tokens": 70576154.0,
+      "step": 30510
+    },
+    {
+      "epoch": 0.2798202988906207,
+      "learning_rate": 0.00014403777390666544,
+      "loss": 0.8383,
+      "mean_token_accuracy": 0.7501836717128754,
+      "num_tokens": 70599045.0,
+      "step": 30520
+    },
+    {
+      "epoch": 0.27991198313009996,
+      "learning_rate": 0.0001440194370587696,
+      "loss": 0.8635,
+      "mean_token_accuracy": 0.7398757040500641,
+      "num_tokens": 70622118.0,
+      "step": 30530
+    },
+    {
+      "epoch": 0.28000366736957916,
+      "learning_rate": 0.00014400110021087377,
+      "loss": 0.8549,
+      "mean_token_accuracy": 0.744711983203888,
+      "num_tokens": 70645721.0,
+      "step": 30540
+    },
+    {
+      "epoch": 0.2800953516090584,
+      "learning_rate": 0.0001439827633629779,
+      "loss": 0.8319,
+      "mean_token_accuracy": 0.7482663452625274,
+      "num_tokens": 70669074.0,
+      "step": 30550
+    },
+    {
+      "epoch": 0.28018703584853766,
+      "learning_rate": 0.00014396442651508207,
+      "loss": 0.8221,
+      "mean_token_accuracy": 0.7502211153507232,
+      "num_tokens": 70691889.0,
+      "step": 30560
+    },
+    {
+      "epoch": 0.28027872008801685,
+      "learning_rate": 0.0001439460896671862,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7460961759090423,
+      "num_tokens": 70715519.0,
+      "step": 30570
+    },
+    {
+      "epoch": 0.2803704043274961,
+      "learning_rate": 0.00014392775281929035,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7456342220306397,
+      "num_tokens": 70738545.0,
+      "step": 30580
+    },
+    {
+      "epoch": 0.28046208856697535,
+      "learning_rate": 0.00014390941597139454,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7371834278106689,
+      "num_tokens": 70761485.0,
+      "step": 30590
+    },
+    {
+      "epoch": 0.28055377280645455,
+      "learning_rate": 0.00014389107912349868,
+      "loss": 0.858,
+      "mean_token_accuracy": 0.7476982533931732,
+      "num_tokens": 70785250.0,
+      "step": 30600
+    },
+    {
+      "epoch": 0.2806454570459338,
+      "learning_rate": 0.00014387274227560285,
+      "loss": 0.8226,
+      "mean_token_accuracy": 0.7469619154930115,
+      "num_tokens": 70808223.0,
+      "step": 30610
+    },
+    {
+      "epoch": 0.28073714128541305,
+      "learning_rate": 0.00014385440542770699,
+      "loss": 0.8946,
+      "mean_token_accuracy": 0.7383210480213165,
+      "num_tokens": 70830986.0,
+      "step": 30620
+    },
+    {
+      "epoch": 0.28082882552489224,
+      "learning_rate": 0.00014383606857981112,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7438196122646332,
+      "num_tokens": 70853956.0,
+      "step": 30630
+    },
+    {
+      "epoch": 0.2809205097643715,
+      "learning_rate": 0.0001438177317319153,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7453059732913971,
+      "num_tokens": 70877787.0,
+      "step": 30640
+    },
+    {
+      "epoch": 0.28101219400385075,
+      "learning_rate": 0.00014379939488401943,
+      "loss": 0.8188,
+      "mean_token_accuracy": 0.75405233502388,
+      "num_tokens": 70900938.0,
+      "step": 30650
+    },
+    {
+      "epoch": 0.28110387824333,
+      "learning_rate": 0.0001437810580361236,
+      "loss": 0.8804,
+      "mean_token_accuracy": 0.7383061647415161,
+      "num_tokens": 70924212.0,
+      "step": 30660
+    },
+    {
+      "epoch": 0.2811955624828092,
+      "learning_rate": 0.00014376272118822776,
+      "loss": 0.8409,
+      "mean_token_accuracy": 0.745840048789978,
+      "num_tokens": 70947002.0,
+      "step": 30670
+    },
+    {
+      "epoch": 0.28128724672228844,
+      "learning_rate": 0.0001437443843403319,
+      "loss": 0.8286,
+      "mean_token_accuracy": 0.7478355526924133,
+      "num_tokens": 70969438.0,
+      "step": 30680
+    },
+    {
+      "epoch": 0.2813789309617677,
+      "learning_rate": 0.00014372604749243606,
+      "loss": 0.842,
+      "mean_token_accuracy": 0.74234339594841,
+      "num_tokens": 70992775.0,
+      "step": 30690
+    },
+    {
+      "epoch": 0.2814706152012469,
+      "learning_rate": 0.0001437077106445402,
+      "loss": 0.8366,
+      "mean_token_accuracy": 0.7444370150566101,
+      "num_tokens": 71016071.0,
+      "step": 30700
+    },
+    {
+      "epoch": 0.28156229944072614,
+      "learning_rate": 0.00014368937379664437,
+      "loss": 0.8425,
+      "mean_token_accuracy": 0.7464881598949432,
+      "num_tokens": 71038736.0,
+      "step": 30710
+    },
+    {
+      "epoch": 0.2816539836802054,
+      "learning_rate": 0.00014367103694874853,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.7483361184597015,
+      "num_tokens": 71061500.0,
+      "step": 30720
+    },
+    {
+      "epoch": 0.2817456679196846,
+      "learning_rate": 0.00014365270010085267,
+      "loss": 0.8815,
+      "mean_token_accuracy": 0.7391713559627533,
+      "num_tokens": 71084834.0,
+      "step": 30730
+    },
+    {
+      "epoch": 0.28183735215916383,
+      "learning_rate": 0.00014363436325295684,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7401086568832398,
+      "num_tokens": 71107733.0,
+      "step": 30740
+    },
+    {
+      "epoch": 0.2819290363986431,
+      "learning_rate": 0.00014361602640506097,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7415320634841919,
+      "num_tokens": 71130038.0,
+      "step": 30750
+    },
+    {
+      "epoch": 0.2820207206381223,
+      "learning_rate": 0.00014359768955716514,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7403928756713867,
+      "num_tokens": 71153764.0,
+      "step": 30760
+    },
+    {
+      "epoch": 0.28211240487760153,
+      "learning_rate": 0.00014357935270926928,
+      "loss": 0.8647,
+      "mean_token_accuracy": 0.7437114596366883,
+      "num_tokens": 71176764.0,
+      "step": 30770
+    },
+    {
+      "epoch": 0.2822040891170808,
+      "learning_rate": 0.00014356101586137342,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7416842520236969,
+      "num_tokens": 71200580.0,
+      "step": 30780
+    },
+    {
+      "epoch": 0.28229577335656003,
+      "learning_rate": 0.0001435426790134776,
+      "loss": 0.8533,
+      "mean_token_accuracy": 0.7468991339206695,
+      "num_tokens": 71223847.0,
+      "step": 30790
+    },
+    {
+      "epoch": 0.2823874575960392,
+      "learning_rate": 0.00014352434216558175,
+      "loss": 0.8709,
+      "mean_token_accuracy": 0.7413143992424012,
+      "num_tokens": 71247311.0,
+      "step": 30800
+    },
+    {
+      "epoch": 0.2824791418355185,
+      "learning_rate": 0.0001435060053176859,
+      "loss": 0.8798,
+      "mean_token_accuracy": 0.7401417315006256,
+      "num_tokens": 71270464.0,
+      "step": 30810
+    },
+    {
+      "epoch": 0.2825708260749977,
+      "learning_rate": 0.00014348766846979005,
+      "loss": 0.8277,
+      "mean_token_accuracy": 0.7541962921619415,
+      "num_tokens": 71293295.0,
+      "step": 30820
+    },
+    {
+      "epoch": 0.2826625103144769,
+      "learning_rate": 0.0001434693316218942,
+      "loss": 0.839,
+      "mean_token_accuracy": 0.751328992843628,
+      "num_tokens": 71316894.0,
+      "step": 30830
+    },
+    {
+      "epoch": 0.2827541945539562,
+      "learning_rate": 0.00014345099477399835,
+      "loss": 0.8572,
+      "mean_token_accuracy": 0.746560251712799,
+      "num_tokens": 71339751.0,
+      "step": 30840
+    },
+    {
+      "epoch": 0.2828458787934354,
+      "learning_rate": 0.00014343265792610252,
+      "loss": 0.8152,
+      "mean_token_accuracy": 0.7542898833751679,
+      "num_tokens": 71362966.0,
+      "step": 30850
+    },
+    {
+      "epoch": 0.2829375630329146,
+      "learning_rate": 0.00014341432107820666,
+      "loss": 0.8849,
+      "mean_token_accuracy": 0.7338407158851623,
+      "num_tokens": 71386109.0,
+      "step": 30860
+    },
+    {
+      "epoch": 0.28302924727239387,
+      "learning_rate": 0.00014339598423031082,
+      "loss": 0.8696,
+      "mean_token_accuracy": 0.7444013357162476,
+      "num_tokens": 71409278.0,
+      "step": 30870
+    },
+    {
+      "epoch": 0.2831209315118731,
+      "learning_rate": 0.00014337764738241496,
+      "loss": 0.8533,
+      "mean_token_accuracy": 0.7491775214672088,
+      "num_tokens": 71433147.0,
+      "step": 30880
+    },
+    {
+      "epoch": 0.28321261575135237,
+      "learning_rate": 0.00014335931053451913,
+      "loss": 0.8571,
+      "mean_token_accuracy": 0.7481815874576568,
+      "num_tokens": 71455969.0,
+      "step": 30890
+    },
+    {
+      "epoch": 0.28330429999083157,
+      "learning_rate": 0.00014334097368662327,
+      "loss": 0.8394,
+      "mean_token_accuracy": 0.740744435787201,
+      "num_tokens": 71479000.0,
+      "step": 30900
+    },
+    {
+      "epoch": 0.2833959842303108,
+      "learning_rate": 0.00014332263683872743,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7422500550746918,
+      "num_tokens": 71502807.0,
+      "step": 30910
+    },
+    {
+      "epoch": 0.28348766846979007,
+      "learning_rate": 0.0001433042999908316,
+      "loss": 0.8567,
+      "mean_token_accuracy": 0.7460370779037475,
+      "num_tokens": 71525740.0,
+      "step": 30920
+    },
+    {
+      "epoch": 0.28357935270926926,
+      "learning_rate": 0.00014328596314293573,
+      "loss": 0.9152,
+      "mean_token_accuracy": 0.7335561573505401,
+      "num_tokens": 71548360.0,
+      "step": 30930
+    },
+    {
+      "epoch": 0.2836710369487485,
+      "learning_rate": 0.0001432676262950399,
+      "loss": 0.8688,
+      "mean_token_accuracy": 0.7373440802097321,
+      "num_tokens": 71571150.0,
+      "step": 30940
+    },
+    {
+      "epoch": 0.28376272118822776,
+      "learning_rate": 0.00014324928944714404,
+      "loss": 0.8736,
+      "mean_token_accuracy": 0.7447324335575104,
+      "num_tokens": 71594344.0,
+      "step": 30950
+    },
+    {
+      "epoch": 0.28385440542770696,
+      "learning_rate": 0.0001432309525992482,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.7510844051837922,
+      "num_tokens": 71616975.0,
+      "step": 30960
+    },
+    {
+      "epoch": 0.2839460896671862,
+      "learning_rate": 0.00014321261575135234,
+      "loss": 0.8865,
+      "mean_token_accuracy": 0.7364141762256622,
+      "num_tokens": 71640641.0,
+      "step": 30970
+    },
+    {
+      "epoch": 0.28403777390666546,
+      "learning_rate": 0.0001431942789034565,
+      "loss": 0.8659,
+      "mean_token_accuracy": 0.7428589820861816,
+      "num_tokens": 71663679.0,
+      "step": 30980
+    },
+    {
+      "epoch": 0.28412945814614465,
+      "learning_rate": 0.00014317594205556067,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.7394986510276794,
+      "num_tokens": 71686389.0,
+      "step": 30990
+    },
+    {
+      "epoch": 0.2842211423856239,
+      "learning_rate": 0.0001431576052076648,
+      "loss": 0.8758,
+      "mean_token_accuracy": 0.7456152856349945,
+      "num_tokens": 71709668.0,
+      "step": 31000
+    },
+    {
+      "epoch": 0.28431282662510315,
+      "learning_rate": 0.00014313926835976898,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7445572733879089,
+      "num_tokens": 71732192.0,
+      "step": 31010
+    },
+    {
+      "epoch": 0.2844045108645824,
+      "learning_rate": 0.00014312093151187311,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7428758382797241,
+      "num_tokens": 71755527.0,
+      "step": 31020
+    },
+    {
+      "epoch": 0.2844961951040616,
+      "learning_rate": 0.00014310259466397725,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7395639896392823,
+      "num_tokens": 71779069.0,
+      "step": 31030
+    },
+    {
+      "epoch": 0.28458787934354085,
+      "learning_rate": 0.00014308425781608142,
+      "loss": 0.8513,
+      "mean_token_accuracy": 0.742822241783142,
+      "num_tokens": 71802376.0,
+      "step": 31040
+    },
+    {
+      "epoch": 0.2846795635830201,
+      "learning_rate": 0.00014306592096818558,
+      "loss": 0.8782,
+      "mean_token_accuracy": 0.741634726524353,
+      "num_tokens": 71825712.0,
+      "step": 31050
+    },
+    {
+      "epoch": 0.2847712478224993,
+      "learning_rate": 0.00014304758412028972,
+      "loss": 0.8739,
+      "mean_token_accuracy": 0.7346738338470459,
+      "num_tokens": 71848767.0,
+      "step": 31060
+    },
+    {
+      "epoch": 0.28486293206197855,
+      "learning_rate": 0.0001430292472723939,
+      "loss": 0.8763,
+      "mean_token_accuracy": 0.7358915567398071,
+      "num_tokens": 71871158.0,
+      "step": 31070
+    },
+    {
+      "epoch": 0.2849546163014578,
+      "learning_rate": 0.00014301091042449803,
+      "loss": 0.8511,
+      "mean_token_accuracy": 0.7476229250431061,
+      "num_tokens": 71894272.0,
+      "step": 31080
+    },
+    {
+      "epoch": 0.285046300540937,
+      "learning_rate": 0.0001429925735766022,
+      "loss": 0.8733,
+      "mean_token_accuracy": 0.7406140685081481,
+      "num_tokens": 71918164.0,
+      "step": 31090
+    },
+    {
+      "epoch": 0.28513798478041624,
+      "learning_rate": 0.00014297423672870633,
+      "loss": 0.8656,
+      "mean_token_accuracy": 0.7448238968849182,
+      "num_tokens": 71941865.0,
+      "step": 31100
+    },
+    {
+      "epoch": 0.2852296690198955,
+      "learning_rate": 0.0001429558998808105,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.7404263973236084,
+      "num_tokens": 71965151.0,
+      "step": 31110
+    },
+    {
+      "epoch": 0.2853213532593747,
+      "learning_rate": 0.00014293756303291466,
+      "loss": 0.8156,
+      "mean_token_accuracy": 0.7532539010047913,
+      "num_tokens": 71988453.0,
+      "step": 31120
+    },
+    {
+      "epoch": 0.28541303749885394,
+      "learning_rate": 0.0001429192261850188,
+      "loss": 0.8983,
+      "mean_token_accuracy": 0.7389238178730011,
+      "num_tokens": 72011386.0,
+      "step": 31130
+    },
+    {
+      "epoch": 0.2855047217383332,
+      "learning_rate": 0.00014290088933712296,
+      "loss": 0.9089,
+      "mean_token_accuracy": 0.7329099833965301,
+      "num_tokens": 72034207.0,
+      "step": 31140
+    },
+    {
+      "epoch": 0.28559640597781244,
+      "learning_rate": 0.0001428825524892271,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7471947193145752,
+      "num_tokens": 72057621.0,
+      "step": 31150
+    },
+    {
+      "epoch": 0.28568809021729163,
+      "learning_rate": 0.00014286421564133127,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7492359936237335,
+      "num_tokens": 72079852.0,
+      "step": 31160
+    },
+    {
+      "epoch": 0.2857797744567709,
+      "learning_rate": 0.0001428458787934354,
+      "loss": 0.8466,
+      "mean_token_accuracy": 0.747854334115982,
+      "num_tokens": 72104097.0,
+      "step": 31170
+    },
+    {
+      "epoch": 0.28587145869625014,
+      "learning_rate": 0.00014282754194553957,
+      "loss": 0.9065,
+      "mean_token_accuracy": 0.7415757656097413,
+      "num_tokens": 72128268.0,
+      "step": 31180
+    },
+    {
+      "epoch": 0.28596314293572933,
+      "learning_rate": 0.00014280920509764374,
+      "loss": 0.83,
+      "mean_token_accuracy": 0.7463529765605926,
+      "num_tokens": 72152418.0,
+      "step": 31190
+    },
+    {
+      "epoch": 0.2860548271752086,
+      "learning_rate": 0.00014279086824974788,
+      "loss": 0.818,
+      "mean_token_accuracy": 0.7579379498958587,
+      "num_tokens": 72175496.0,
+      "step": 31200
+    },
+    {
+      "epoch": 0.28614651141468783,
+      "learning_rate": 0.00014277253140185204,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7471842169761658,
+      "num_tokens": 72198625.0,
+      "step": 31210
+    },
+    {
+      "epoch": 0.286238195654167,
+      "learning_rate": 0.00014275419455395618,
+      "loss": 0.8186,
+      "mean_token_accuracy": 0.7483863472938538,
+      "num_tokens": 72222112.0,
+      "step": 31220
+    },
+    {
+      "epoch": 0.2863298798936463,
+      "learning_rate": 0.00014273585770606032,
+      "loss": 0.84,
+      "mean_token_accuracy": 0.7470630824565887,
+      "num_tokens": 72244957.0,
+      "step": 31230
+    },
+    {
+      "epoch": 0.28642156413312553,
+      "learning_rate": 0.00014271752085816448,
+      "loss": 0.87,
+      "mean_token_accuracy": 0.7364301800727844,
+      "num_tokens": 72267893.0,
+      "step": 31240
+    },
+    {
+      "epoch": 0.2865132483726047,
+      "learning_rate": 0.00014269918401026865,
+      "loss": 0.8761,
+      "mean_token_accuracy": 0.7397234499454498,
+      "num_tokens": 72291577.0,
+      "step": 31250
+    },
+    {
+      "epoch": 0.286604932612084,
+      "learning_rate": 0.0001426808471623728,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7466777324676513,
+      "num_tokens": 72314451.0,
+      "step": 31260
+    },
+    {
+      "epoch": 0.2866966168515632,
+      "learning_rate": 0.00014266251031447695,
+      "loss": 0.8717,
+      "mean_token_accuracy": 0.7358124434947968,
+      "num_tokens": 72337215.0,
+      "step": 31270
+    },
+    {
+      "epoch": 0.2867883010910425,
+      "learning_rate": 0.0001426441734665811,
+      "loss": 0.8797,
+      "mean_token_accuracy": 0.7377731680870057,
+      "num_tokens": 72360521.0,
+      "step": 31280
+    },
+    {
+      "epoch": 0.28687998533052167,
+      "learning_rate": 0.00014262583661868526,
+      "loss": 0.9012,
+      "mean_token_accuracy": 0.7384210169315338,
+      "num_tokens": 72384074.0,
+      "step": 31290
+    },
+    {
+      "epoch": 0.2869716695700009,
+      "learning_rate": 0.0001426074997707894,
+      "loss": 0.8515,
+      "mean_token_accuracy": 0.7453174591064453,
+      "num_tokens": 72406413.0,
+      "step": 31300
+    },
+    {
+      "epoch": 0.28706335380948017,
+      "learning_rate": 0.00014258916292289356,
+      "loss": 0.896,
+      "mean_token_accuracy": 0.7345102667808533,
+      "num_tokens": 72430027.0,
+      "step": 31310
+    },
+    {
+      "epoch": 0.28715503804895937,
+      "learning_rate": 0.00014257082607499773,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7416463077068329,
+      "num_tokens": 72452820.0,
+      "step": 31320
+    },
+    {
+      "epoch": 0.2872467222884386,
+      "learning_rate": 0.00014255248922710186,
+      "loss": 0.8291,
+      "mean_token_accuracy": 0.7540943801403046,
+      "num_tokens": 72475923.0,
+      "step": 31330
+    },
+    {
+      "epoch": 0.28733840652791787,
+      "learning_rate": 0.00014253415237920603,
+      "loss": 0.821,
+      "mean_token_accuracy": 0.748418390750885,
+      "num_tokens": 72498435.0,
+      "step": 31340
+    },
+    {
+      "epoch": 0.28743009076739706,
+      "learning_rate": 0.00014251581553131017,
+      "loss": 0.8536,
+      "mean_token_accuracy": 0.7409149229526519,
+      "num_tokens": 72521049.0,
+      "step": 31350
+    },
+    {
+      "epoch": 0.2875217750068763,
+      "learning_rate": 0.00014249747868341433,
+      "loss": 0.8704,
+      "mean_token_accuracy": 0.7396306455135345,
+      "num_tokens": 72544866.0,
+      "step": 31360
+    },
+    {
+      "epoch": 0.28761345924635556,
+      "learning_rate": 0.00014247914183551847,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7428924560546875,
+      "num_tokens": 72568158.0,
+      "step": 31370
+    },
+    {
+      "epoch": 0.28770514348583476,
+      "learning_rate": 0.00014246080498762264,
+      "loss": 0.8989,
+      "mean_token_accuracy": 0.7329227685928345,
+      "num_tokens": 72591398.0,
+      "step": 31380
+    },
+    {
+      "epoch": 0.287796827725314,
+      "learning_rate": 0.0001424424681397268,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.742163461446762,
+      "num_tokens": 72614580.0,
+      "step": 31390
+    },
+    {
+      "epoch": 0.28788851196479326,
+      "learning_rate": 0.00014242413129183094,
+      "loss": 0.8503,
+      "mean_token_accuracy": 0.7476768851280212,
+      "num_tokens": 72637245.0,
+      "step": 31400
+    },
+    {
+      "epoch": 0.2879801962042725,
+      "learning_rate": 0.0001424057944439351,
+      "loss": 0.8731,
+      "mean_token_accuracy": 0.7394823253154754,
+      "num_tokens": 72659984.0,
+      "step": 31410
+    },
+    {
+      "epoch": 0.2880718804437517,
+      "learning_rate": 0.00014238745759603924,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7433938205242157,
+      "num_tokens": 72683009.0,
+      "step": 31420
+    },
+    {
+      "epoch": 0.28816356468323095,
+      "learning_rate": 0.00014236912074814338,
+      "loss": 0.8716,
+      "mean_token_accuracy": 0.7435907065868378,
+      "num_tokens": 72706440.0,
+      "step": 31430
+    },
+    {
+      "epoch": 0.2882552489227102,
+      "learning_rate": 0.00014235078390024757,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7424872159957886,
+      "num_tokens": 72729641.0,
+      "step": 31440
+    },
+    {
+      "epoch": 0.2883469331621894,
+      "learning_rate": 0.0001423324470523517,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7388116776943207,
+      "num_tokens": 72751920.0,
+      "step": 31450
+    },
+    {
+      "epoch": 0.28843861740166865,
+      "learning_rate": 0.00014231411020445585,
+      "loss": 0.834,
+      "mean_token_accuracy": 0.7525705695152283,
+      "num_tokens": 72775401.0,
+      "step": 31460
+    },
+    {
+      "epoch": 0.2885303016411479,
+      "learning_rate": 0.00014229577335656002,
+      "loss": 0.8488,
+      "mean_token_accuracy": 0.748732203245163,
+      "num_tokens": 72798287.0,
+      "step": 31470
+    },
+    {
+      "epoch": 0.2886219858806271,
+      "learning_rate": 0.00014227743650866416,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7422996640205384,
+      "num_tokens": 72821016.0,
+      "step": 31480
+    },
+    {
+      "epoch": 0.28871367012010635,
+      "learning_rate": 0.00014225909966076832,
+      "loss": 0.9165,
+      "mean_token_accuracy": 0.733242517709732,
+      "num_tokens": 72843729.0,
+      "step": 31490
+    },
+    {
+      "epoch": 0.2888053543595856,
+      "learning_rate": 0.00014224076281287246,
+      "loss": 0.8356,
+      "mean_token_accuracy": 0.7454355776309967,
+      "num_tokens": 72867442.0,
+      "step": 31500
+    },
+    {
+      "epoch": 0.28889703859906485,
+      "learning_rate": 0.00014222242596497662,
+      "loss": 0.8887,
+      "mean_token_accuracy": 0.7318439543247223,
+      "num_tokens": 72890400.0,
+      "step": 31510
+    },
+    {
+      "epoch": 0.28898872283854404,
+      "learning_rate": 0.0001422040891170808,
+      "loss": 0.8894,
+      "mean_token_accuracy": 0.7362122118473053,
+      "num_tokens": 72913503.0,
+      "step": 31520
+    },
+    {
+      "epoch": 0.2890804070780233,
+      "learning_rate": 0.00014218575226918493,
+      "loss": 0.9052,
+      "mean_token_accuracy": 0.7309132397174836,
+      "num_tokens": 72936459.0,
+      "step": 31530
+    },
+    {
+      "epoch": 0.28917209131750254,
+      "learning_rate": 0.0001421674154212891,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.7345360696315766,
+      "num_tokens": 72959987.0,
+      "step": 31540
+    },
+    {
+      "epoch": 0.28926377555698174,
+      "learning_rate": 0.00014214907857339323,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.739002925157547,
+      "num_tokens": 72982358.0,
+      "step": 31550
+    },
+    {
+      "epoch": 0.289355459796461,
+      "learning_rate": 0.0001421307417254974,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7455204069614411,
+      "num_tokens": 73005658.0,
+      "step": 31560
+    },
+    {
+      "epoch": 0.28944714403594024,
+      "learning_rate": 0.00014211240487760156,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7440772473812103,
+      "num_tokens": 73029475.0,
+      "step": 31570
+    },
+    {
+      "epoch": 0.28953882827541944,
+      "learning_rate": 0.0001420940680297057,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.7433251142501831,
+      "num_tokens": 73052807.0,
+      "step": 31580
+    },
+    {
+      "epoch": 0.2896305125148987,
+      "learning_rate": 0.00014207573118180987,
+      "loss": 0.8068,
+      "mean_token_accuracy": 0.7552599966526031,
+      "num_tokens": 73075825.0,
+      "step": 31590
+    },
+    {
+      "epoch": 0.28972219675437794,
+      "learning_rate": 0.000142057394333914,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7442369043827057,
+      "num_tokens": 73100114.0,
+      "step": 31600
+    },
+    {
+      "epoch": 0.28981388099385713,
+      "learning_rate": 0.00014203905748601817,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7419051170349121,
+      "num_tokens": 73122940.0,
+      "step": 31610
+    },
+    {
+      "epoch": 0.2899055652333364,
+      "learning_rate": 0.0001420207206381223,
+      "loss": 0.8841,
+      "mean_token_accuracy": 0.7381398737430572,
+      "num_tokens": 73145978.0,
+      "step": 31620
+    },
+    {
+      "epoch": 0.28999724947281563,
+      "learning_rate": 0.00014200238379022645,
+      "loss": 0.8814,
+      "mean_token_accuracy": 0.7453715085983277,
+      "num_tokens": 73168783.0,
+      "step": 31630
+    },
+    {
+      "epoch": 0.2900889337122949,
+      "learning_rate": 0.00014198404694233064,
+      "loss": 0.8441,
+      "mean_token_accuracy": 0.7435619235038757,
+      "num_tokens": 73191613.0,
+      "step": 31640
+    },
+    {
+      "epoch": 0.2901806179517741,
+      "learning_rate": 0.00014196571009443478,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7409064710140228,
+      "num_tokens": 73214562.0,
+      "step": 31650
+    },
+    {
+      "epoch": 0.29027230219125333,
+      "learning_rate": 0.00014194737324653892,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.7437946617603302,
+      "num_tokens": 73237810.0,
+      "step": 31660
+    },
+    {
+      "epoch": 0.2903639864307326,
+      "learning_rate": 0.00014192903639864308,
+      "loss": 0.8256,
+      "mean_token_accuracy": 0.7505256474018097,
+      "num_tokens": 73260321.0,
+      "step": 31670
+    },
+    {
+      "epoch": 0.2904556706702118,
+      "learning_rate": 0.00014191069955074722,
+      "loss": 0.8271,
+      "mean_token_accuracy": 0.7476964652538299,
+      "num_tokens": 73283163.0,
+      "step": 31680
+    },
+    {
+      "epoch": 0.290547354909691,
+      "learning_rate": 0.00014189236270285138,
+      "loss": 0.8846,
+      "mean_token_accuracy": 0.7330077946186065,
+      "num_tokens": 73306138.0,
+      "step": 31690
+    },
+    {
+      "epoch": 0.2906390391491703,
+      "learning_rate": 0.00014187402585495555,
+      "loss": 0.8843,
+      "mean_token_accuracy": 0.7401116073131562,
+      "num_tokens": 73330472.0,
+      "step": 31700
+    },
+    {
+      "epoch": 0.29073072338864947,
+      "learning_rate": 0.0001418556890070597,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.744620555639267,
+      "num_tokens": 73353624.0,
+      "step": 31710
+    },
+    {
+      "epoch": 0.2908224076281287,
+      "learning_rate": 0.00014183735215916385,
+      "loss": 0.8006,
+      "mean_token_accuracy": 0.7572574019432068,
+      "num_tokens": 73375890.0,
+      "step": 31720
+    },
+    {
+      "epoch": 0.29091409186760797,
+      "learning_rate": 0.000141819015311268,
+      "loss": 0.8572,
+      "mean_token_accuracy": 0.7402051568031311,
+      "num_tokens": 73399379.0,
+      "step": 31730
+    },
+    {
+      "epoch": 0.29100577610708717,
+      "learning_rate": 0.00014180067846337216,
+      "loss": 0.8812,
+      "mean_token_accuracy": 0.7400589942932129,
+      "num_tokens": 73422434.0,
+      "step": 31740
+    },
+    {
+      "epoch": 0.2910974603465664,
+      "learning_rate": 0.0001417823416154763,
+      "loss": 0.8361,
+      "mean_token_accuracy": 0.7483542859554291,
+      "num_tokens": 73445375.0,
+      "step": 31750
+    },
+    {
+      "epoch": 0.29118914458604567,
+      "learning_rate": 0.00014176400476758046,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7430842816829681,
+      "num_tokens": 73469386.0,
+      "step": 31760
+    },
+    {
+      "epoch": 0.2912808288255249,
+      "learning_rate": 0.00014174566791968463,
+      "loss": 0.8514,
+      "mean_token_accuracy": 0.7490638792514801,
+      "num_tokens": 73493032.0,
+      "step": 31770
+    },
+    {
+      "epoch": 0.2913725130650041,
+      "learning_rate": 0.00014172733107178877,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7369000732898712,
+      "num_tokens": 73516028.0,
+      "step": 31780
+    },
+    {
+      "epoch": 0.29146419730448336,
+      "learning_rate": 0.00014170899422389293,
+      "loss": 0.8378,
+      "mean_token_accuracy": 0.7542305409908294,
+      "num_tokens": 73538543.0,
+      "step": 31790
+    },
+    {
+      "epoch": 0.2915558815439626,
+      "learning_rate": 0.00014169065737599707,
+      "loss": 0.8599,
+      "mean_token_accuracy": 0.7383139252662658,
+      "num_tokens": 73561062.0,
+      "step": 31800
+    },
+    {
+      "epoch": 0.2916475657834418,
+      "learning_rate": 0.00014167232052810123,
+      "loss": 0.8321,
+      "mean_token_accuracy": 0.7473494529724121,
+      "num_tokens": 73583815.0,
+      "step": 31810
+    },
+    {
+      "epoch": 0.29173925002292106,
+      "learning_rate": 0.00014165398368020537,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.741040050983429,
+      "num_tokens": 73607221.0,
+      "step": 31820
+    },
+    {
+      "epoch": 0.2918309342624003,
+      "learning_rate": 0.00014163564683230954,
+      "loss": 0.8251,
+      "mean_token_accuracy": 0.75463907122612,
+      "num_tokens": 73630164.0,
+      "step": 31830
+    },
+    {
+      "epoch": 0.2919226185018795,
+      "learning_rate": 0.0001416173099844137,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.7428629636764527,
+      "num_tokens": 73653482.0,
+      "step": 31840
+    },
+    {
+      "epoch": 0.29201430274135876,
+      "learning_rate": 0.00014159897313651784,
+      "loss": 0.8516,
+      "mean_token_accuracy": 0.7402035355567932,
+      "num_tokens": 73676116.0,
+      "step": 31850
+    },
+    {
+      "epoch": 0.292105986980838,
+      "learning_rate": 0.00014158063628862198,
+      "loss": 0.8303,
+      "mean_token_accuracy": 0.7501801729202271,
+      "num_tokens": 73699175.0,
+      "step": 31860
+    },
+    {
+      "epoch": 0.2921976712203172,
+      "learning_rate": 0.00014156229944072615,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7425274848937988,
+      "num_tokens": 73722392.0,
+      "step": 31870
+    },
+    {
+      "epoch": 0.29228935545979645,
+      "learning_rate": 0.00014154396259283028,
+      "loss": 0.8236,
+      "mean_token_accuracy": 0.7515719771385193,
+      "num_tokens": 73745307.0,
+      "step": 31880
+    },
+    {
+      "epoch": 0.2923810396992757,
+      "learning_rate": 0.00014152562574493445,
+      "loss": 0.8413,
+      "mean_token_accuracy": 0.7502724707126618,
+      "num_tokens": 73768673.0,
+      "step": 31890
+    },
+    {
+      "epoch": 0.29247272393875495,
+      "learning_rate": 0.00014150728889703861,
+      "loss": 0.8771,
+      "mean_token_accuracy": 0.7392754435539246,
+      "num_tokens": 73791261.0,
+      "step": 31900
+    },
+    {
+      "epoch": 0.29256440817823415,
+      "learning_rate": 0.00014148895204914275,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7486158907413483,
+      "num_tokens": 73813887.0,
+      "step": 31910
+    },
+    {
+      "epoch": 0.2926560924177134,
+      "learning_rate": 0.00014147061520124692,
+      "loss": 0.8244,
+      "mean_token_accuracy": 0.7549403071403503,
+      "num_tokens": 73836544.0,
+      "step": 31920
+    },
+    {
+      "epoch": 0.29274777665719265,
+      "learning_rate": 0.00014145227835335106,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7428262293338775,
+      "num_tokens": 73859671.0,
+      "step": 31930
+    },
+    {
+      "epoch": 0.29283946089667184,
+      "learning_rate": 0.00014143394150545522,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7449959874153137,
+      "num_tokens": 73882330.0,
+      "step": 31940
+    },
+    {
+      "epoch": 0.2929311451361511,
+      "learning_rate": 0.00014141560465755936,
+      "loss": 0.8951,
+      "mean_token_accuracy": 0.7362458229064941,
+      "num_tokens": 73905308.0,
+      "step": 31950
+    },
+    {
+      "epoch": 0.29302282937563034,
+      "learning_rate": 0.00014139726780966353,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7479972541332245,
+      "num_tokens": 73928426.0,
+      "step": 31960
+    },
+    {
+      "epoch": 0.29311451361510954,
+      "learning_rate": 0.0001413789309617677,
+      "loss": 0.8455,
+      "mean_token_accuracy": 0.7483917891979217,
+      "num_tokens": 73952447.0,
+      "step": 31970
+    },
+    {
+      "epoch": 0.2932061978545888,
+      "learning_rate": 0.00014136059411387183,
+      "loss": 0.8445,
+      "mean_token_accuracy": 0.7439047753810882,
+      "num_tokens": 73975454.0,
+      "step": 31980
+    },
+    {
+      "epoch": 0.29329788209406804,
+      "learning_rate": 0.000141342257265976,
+      "loss": 0.8764,
+      "mean_token_accuracy": 0.7404709458351135,
+      "num_tokens": 73999058.0,
+      "step": 31990
+    },
+    {
+      "epoch": 0.29338956633354724,
+      "learning_rate": 0.00014132392041808013,
+      "loss": 0.8678,
+      "mean_token_accuracy": 0.7416110396385193,
+      "num_tokens": 74021635.0,
+      "step": 32000
+    },
+    {
+      "epoch": 0.2934812505730265,
+      "learning_rate": 0.0001413055835701843,
+      "loss": 0.8866,
+      "mean_token_accuracy": 0.7399464905261993,
+      "num_tokens": 74044911.0,
+      "step": 32010
+    },
+    {
+      "epoch": 0.29357293481250574,
+      "learning_rate": 0.00014128724672228844,
+      "loss": 0.8308,
+      "mean_token_accuracy": 0.7489681363105773,
+      "num_tokens": 74068136.0,
+      "step": 32020
+    },
+    {
+      "epoch": 0.293664619051985,
+      "learning_rate": 0.0001412689098743926,
+      "loss": 0.8597,
+      "mean_token_accuracy": 0.7433870613574982,
+      "num_tokens": 74090667.0,
+      "step": 32030
+    },
+    {
+      "epoch": 0.2937563032914642,
+      "learning_rate": 0.00014125057302649677,
+      "loss": 0.898,
+      "mean_token_accuracy": 0.7378264725208282,
+      "num_tokens": 74114025.0,
+      "step": 32040
+    },
+    {
+      "epoch": 0.29384798753094343,
+      "learning_rate": 0.0001412322361786009,
+      "loss": 0.8597,
+      "mean_token_accuracy": 0.7399658620357513,
+      "num_tokens": 74136459.0,
+      "step": 32050
+    },
+    {
+      "epoch": 0.2939396717704227,
+      "learning_rate": 0.00014121389933070504,
+      "loss": 0.8756,
+      "mean_token_accuracy": 0.7452867448329925,
+      "num_tokens": 74159395.0,
+      "step": 32060
+    },
+    {
+      "epoch": 0.2940313560099019,
+      "learning_rate": 0.0001411955624828092,
+      "loss": 0.8425,
+      "mean_token_accuracy": 0.7508084893226623,
+      "num_tokens": 74182500.0,
+      "step": 32070
+    },
+    {
+      "epoch": 0.29412304024938113,
+      "learning_rate": 0.00014117722563491335,
+      "loss": 0.8487,
+      "mean_token_accuracy": 0.7505961298942566,
+      "num_tokens": 74205656.0,
+      "step": 32080
+    },
+    {
+      "epoch": 0.2942147244888604,
+      "learning_rate": 0.00014115888878701754,
+      "loss": 0.88,
+      "mean_token_accuracy": 0.7409342706203461,
+      "num_tokens": 74228330.0,
+      "step": 32090
+    },
+    {
+      "epoch": 0.2943064087283396,
+      "learning_rate": 0.00014114055193912168,
+      "loss": 0.8368,
+      "mean_token_accuracy": 0.7509453773498536,
+      "num_tokens": 74251978.0,
+      "step": 32100
+    },
+    {
+      "epoch": 0.2943980929678188,
+      "learning_rate": 0.00014112221509122582,
+      "loss": 0.8987,
+      "mean_token_accuracy": 0.7340470790863037,
+      "num_tokens": 74275035.0,
+      "step": 32110
+    },
+    {
+      "epoch": 0.2944897772072981,
+      "learning_rate": 0.00014110387824332998,
+      "loss": 0.8391,
+      "mean_token_accuracy": 0.7468198537826538,
+      "num_tokens": 74298321.0,
+      "step": 32120
+    },
+    {
+      "epoch": 0.2945814614467773,
+      "learning_rate": 0.00014108554139543412,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7403269171714782,
+      "num_tokens": 74320919.0,
+      "step": 32130
+    },
+    {
+      "epoch": 0.2946731456862565,
+      "learning_rate": 0.0001410672045475383,
+      "loss": 0.827,
+      "mean_token_accuracy": 0.7486122071743011,
+      "num_tokens": 74343472.0,
+      "step": 32140
+    },
+    {
+      "epoch": 0.29476482992573577,
+      "learning_rate": 0.00014104886769964243,
+      "loss": 0.9174,
+      "mean_token_accuracy": 0.7354245066642762,
+      "num_tokens": 74366550.0,
+      "step": 32150
+    },
+    {
+      "epoch": 0.294856514165215,
+      "learning_rate": 0.0001410305308517466,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.7446018040180207,
+      "num_tokens": 74389540.0,
+      "step": 32160
+    },
+    {
+      "epoch": 0.2949481984046942,
+      "learning_rate": 0.00014101219400385076,
+      "loss": 0.8509,
+      "mean_token_accuracy": 0.7483080565929413,
+      "num_tokens": 74412838.0,
+      "step": 32170
+    },
+    {
+      "epoch": 0.29503988264417347,
+      "learning_rate": 0.0001409938571559549,
+      "loss": 0.8479,
+      "mean_token_accuracy": 0.7413890719413757,
+      "num_tokens": 74436548.0,
+      "step": 32180
+    },
+    {
+      "epoch": 0.2951315668836527,
+      "learning_rate": 0.00014097552030805906,
+      "loss": 0.8748,
+      "mean_token_accuracy": 0.74229536652565,
+      "num_tokens": 74460687.0,
+      "step": 32190
+    },
+    {
+      "epoch": 0.2952232511231319,
+      "learning_rate": 0.0001409571834601632,
+      "loss": 0.8407,
+      "mean_token_accuracy": 0.7461408078670502,
+      "num_tokens": 74483701.0,
+      "step": 32200
+    },
+    {
+      "epoch": 0.29531493536261116,
+      "learning_rate": 0.00014093884661226736,
+      "loss": 0.8704,
+      "mean_token_accuracy": 0.7388374507427216,
+      "num_tokens": 74506773.0,
+      "step": 32210
+    },
+    {
+      "epoch": 0.2954066196020904,
+      "learning_rate": 0.00014092050976437153,
+      "loss": 0.8081,
+      "mean_token_accuracy": 0.755785596370697,
+      "num_tokens": 74529366.0,
+      "step": 32220
+    },
+    {
+      "epoch": 0.2954983038415696,
+      "learning_rate": 0.00014090217291647567,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.741152960062027,
+      "num_tokens": 74552644.0,
+      "step": 32230
+    },
+    {
+      "epoch": 0.29558998808104886,
+      "learning_rate": 0.00014088383606857983,
+      "loss": 0.8309,
+      "mean_token_accuracy": 0.7479057490825654,
+      "num_tokens": 74575039.0,
+      "step": 32240
+    },
+    {
+      "epoch": 0.2956816723205281,
+      "learning_rate": 0.00014086549922068397,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7457516610622406,
+      "num_tokens": 74598000.0,
+      "step": 32250
+    },
+    {
+      "epoch": 0.29577335656000736,
+      "learning_rate": 0.0001408471623727881,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7434446692466736,
+      "num_tokens": 74620147.0,
+      "step": 32260
+    },
+    {
+      "epoch": 0.29586504079948656,
+      "learning_rate": 0.00014082882552489227,
+      "loss": 0.8512,
+      "mean_token_accuracy": 0.7454415023326874,
+      "num_tokens": 74642345.0,
+      "step": 32270
+    },
+    {
+      "epoch": 0.2959567250389658,
+      "learning_rate": 0.0001408104886769964,
+      "loss": 0.8257,
+      "mean_token_accuracy": 0.7547140061855316,
+      "num_tokens": 74666155.0,
+      "step": 32280
+    },
+    {
+      "epoch": 0.29604840927844506,
+      "learning_rate": 0.0001407921518291006,
+      "loss": 0.8205,
+      "mean_token_accuracy": 0.7539973735809327,
+      "num_tokens": 74688943.0,
+      "step": 32290
+    },
+    {
+      "epoch": 0.29614009351792425,
+      "learning_rate": 0.00014077381498120474,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7428905308246613,
+      "num_tokens": 74711307.0,
+      "step": 32300
+    },
+    {
+      "epoch": 0.2962317777574035,
+      "learning_rate": 0.00014075547813330888,
+      "loss": 0.8432,
+      "mean_token_accuracy": 0.7513552963733673,
+      "num_tokens": 74733617.0,
+      "step": 32310
+    },
+    {
+      "epoch": 0.29632346199688275,
+      "learning_rate": 0.00014073714128541305,
+      "loss": 0.8984,
+      "mean_token_accuracy": 0.732017582654953,
+      "num_tokens": 74756660.0,
+      "step": 32320
+    },
+    {
+      "epoch": 0.29641514623636195,
+      "learning_rate": 0.00014071880443751719,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.750505793094635,
+      "num_tokens": 74779023.0,
+      "step": 32330
+    },
+    {
+      "epoch": 0.2965068304758412,
+      "learning_rate": 0.00014070046758962135,
+      "loss": 0.8321,
+      "mean_token_accuracy": 0.745517760515213,
+      "num_tokens": 74802343.0,
+      "step": 32340
+    },
+    {
+      "epoch": 0.29659851471532045,
+      "learning_rate": 0.00014068213074172552,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.7482177495956421,
+      "num_tokens": 74825804.0,
+      "step": 32350
+    },
+    {
+      "epoch": 0.29669019895479964,
+      "learning_rate": 0.00014066379389382965,
+      "loss": 0.839,
+      "mean_token_accuracy": 0.7464603304862976,
+      "num_tokens": 74850115.0,
+      "step": 32360
+    },
+    {
+      "epoch": 0.2967818831942789,
+      "learning_rate": 0.00014064545704593382,
+      "loss": 0.8827,
+      "mean_token_accuracy": 0.7317130923271179,
+      "num_tokens": 74873543.0,
+      "step": 32370
+    },
+    {
+      "epoch": 0.29687356743375815,
+      "learning_rate": 0.00014062712019803796,
+      "loss": 0.8646,
+      "mean_token_accuracy": 0.7418008387088776,
+      "num_tokens": 74896263.0,
+      "step": 32380
+    },
+    {
+      "epoch": 0.2969652516732374,
+      "learning_rate": 0.00014060878335014212,
+      "loss": 0.8314,
+      "mean_token_accuracy": 0.7491295874118805,
+      "num_tokens": 74918428.0,
+      "step": 32390
+    },
+    {
+      "epoch": 0.2970569359127166,
+      "learning_rate": 0.00014059044650224626,
+      "loss": 0.805,
+      "mean_token_accuracy": 0.7593595862388611,
+      "num_tokens": 74941764.0,
+      "step": 32400
+    },
+    {
+      "epoch": 0.29714862015219584,
+      "learning_rate": 0.00014057210965435043,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.7525422215461731,
+      "num_tokens": 74965617.0,
+      "step": 32410
+    },
+    {
+      "epoch": 0.2972403043916751,
+      "learning_rate": 0.0001405537728064546,
+      "loss": 0.8279,
+      "mean_token_accuracy": 0.747933852672577,
+      "num_tokens": 74988959.0,
+      "step": 32420
+    },
+    {
+      "epoch": 0.2973319886311543,
+      "learning_rate": 0.00014053543595855873,
+      "loss": 0.8478,
+      "mean_token_accuracy": 0.7457592070102692,
+      "num_tokens": 75012040.0,
+      "step": 32430
+    },
+    {
+      "epoch": 0.29742367287063354,
+      "learning_rate": 0.0001405170991106629,
+      "loss": 0.8385,
+      "mean_token_accuracy": 0.7494657516479493,
+      "num_tokens": 75035529.0,
+      "step": 32440
+    },
+    {
+      "epoch": 0.2975153571101128,
+      "learning_rate": 0.00014049876226276704,
+      "loss": 0.8677,
+      "mean_token_accuracy": 0.7418507039546967,
+      "num_tokens": 75058556.0,
+      "step": 32450
+    },
+    {
+      "epoch": 0.297607041349592,
+      "learning_rate": 0.00014048042541487117,
+      "loss": 0.8996,
+      "mean_token_accuracy": 0.7299916386604309,
+      "num_tokens": 75081452.0,
+      "step": 32460
+    },
+    {
+      "epoch": 0.29769872558907123,
+      "learning_rate": 0.00014046208856697534,
+      "loss": 0.8431,
+      "mean_token_accuracy": 0.7505478620529175,
+      "num_tokens": 75104868.0,
+      "step": 32470
+    },
+    {
+      "epoch": 0.2977904098285505,
+      "learning_rate": 0.00014044375171907948,
+      "loss": 0.8368,
+      "mean_token_accuracy": 0.7543375313282012,
+      "num_tokens": 75128512.0,
+      "step": 32480
+    },
+    {
+      "epoch": 0.2978820940680297,
+      "learning_rate": 0.00014042541487118367,
+      "loss": 0.8516,
+      "mean_token_accuracy": 0.7499530851840973,
+      "num_tokens": 75151795.0,
+      "step": 32490
+    },
+    {
+      "epoch": 0.29797377830750893,
+      "learning_rate": 0.0001404070780232878,
+      "loss": 0.8384,
+      "mean_token_accuracy": 0.7483568310737609,
+      "num_tokens": 75174689.0,
+      "step": 32500
+    },
+    {
+      "epoch": 0.2980654625469882,
+      "learning_rate": 0.00014038874117539195,
+      "loss": 0.8768,
+      "mean_token_accuracy": 0.7385094165802002,
+      "num_tokens": 75197103.0,
+      "step": 32510
+    },
+    {
+      "epoch": 0.29815714678646743,
+      "learning_rate": 0.0001403704043274961,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7431336998939514,
+      "num_tokens": 75220996.0,
+      "step": 32520
+    },
+    {
+      "epoch": 0.2982488310259466,
+      "learning_rate": 0.00014035206747960025,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.7378594398498535,
+      "num_tokens": 75244686.0,
+      "step": 32530
+    },
+    {
+      "epoch": 0.2983405152654259,
+      "learning_rate": 0.00014033373063170442,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.7430776119232178,
+      "num_tokens": 75266932.0,
+      "step": 32540
+    },
+    {
+      "epoch": 0.2984321995049051,
+      "learning_rate": 0.00014031539378380858,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.7432925641536713,
+      "num_tokens": 75290657.0,
+      "step": 32550
+    },
+    {
+      "epoch": 0.2985238837443843,
+      "learning_rate": 0.00014029705693591272,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7460358500480652,
+      "num_tokens": 75314025.0,
+      "step": 32560
+    },
+    {
+      "epoch": 0.2986155679838636,
+      "learning_rate": 0.00014027872008801688,
+      "loss": 0.8363,
+      "mean_token_accuracy": 0.7493690609931946,
+      "num_tokens": 75336336.0,
+      "step": 32570
+    },
+    {
+      "epoch": 0.2987072522233428,
+      "learning_rate": 0.00014026038324012102,
+      "loss": 0.906,
+      "mean_token_accuracy": 0.7391089081764222,
+      "num_tokens": 75359158.0,
+      "step": 32580
+    },
+    {
+      "epoch": 0.298798936462822,
+      "learning_rate": 0.0001402420463922252,
+      "loss": 0.8353,
+      "mean_token_accuracy": 0.7448742389678955,
+      "num_tokens": 75381898.0,
+      "step": 32590
+    },
+    {
+      "epoch": 0.29889062070230127,
+      "learning_rate": 0.00014022370954432933,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7458133399486542,
+      "num_tokens": 75405062.0,
+      "step": 32600
+    },
+    {
+      "epoch": 0.2989823049417805,
+      "learning_rate": 0.0001402053726964335,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7488585948944092,
+      "num_tokens": 75428130.0,
+      "step": 32610
+    },
+    {
+      "epoch": 0.2990739891812597,
+      "learning_rate": 0.00014018703584853766,
+      "loss": 0.8337,
+      "mean_token_accuracy": 0.7478954017162323,
+      "num_tokens": 75450520.0,
+      "step": 32620
+    },
+    {
+      "epoch": 0.29916567342073896,
+      "learning_rate": 0.0001401686990006418,
+      "loss": 0.8863,
+      "mean_token_accuracy": 0.7290149629116058,
+      "num_tokens": 75473367.0,
+      "step": 32630
+    },
+    {
+      "epoch": 0.2992573576602182,
+      "learning_rate": 0.00014015036215274596,
+      "loss": 0.856,
+      "mean_token_accuracy": 0.7386962234973907,
+      "num_tokens": 75496775.0,
+      "step": 32640
+    },
+    {
+      "epoch": 0.29934904189969747,
+      "learning_rate": 0.0001401320253048501,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7430225610733032,
+      "num_tokens": 75520106.0,
+      "step": 32650
+    },
+    {
+      "epoch": 0.29944072613917666,
+      "learning_rate": 0.00014011368845695424,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7468016922473908,
+      "num_tokens": 75543050.0,
+      "step": 32660
+    },
+    {
+      "epoch": 0.2995324103786559,
+      "learning_rate": 0.0001400953516090584,
+      "loss": 0.8398,
+      "mean_token_accuracy": 0.7432979285717011,
+      "num_tokens": 75566710.0,
+      "step": 32670
+    },
+    {
+      "epoch": 0.29962409461813516,
+      "learning_rate": 0.00014007701476116257,
+      "loss": 0.8289,
+      "mean_token_accuracy": 0.7504425466060638,
+      "num_tokens": 75589624.0,
+      "step": 32680
+    },
+    {
+      "epoch": 0.29971577885761436,
+      "learning_rate": 0.00014005867791326673,
+      "loss": 0.8293,
+      "mean_token_accuracy": 0.747108769416809,
+      "num_tokens": 75612434.0,
+      "step": 32690
+    },
+    {
+      "epoch": 0.2998074630970936,
+      "learning_rate": 0.00014004034106537087,
+      "loss": 0.8254,
+      "mean_token_accuracy": 0.754813802242279,
+      "num_tokens": 75636279.0,
+      "step": 32700
+    },
+    {
+      "epoch": 0.29989914733657286,
+      "learning_rate": 0.000140022004217475,
+      "loss": 0.8211,
+      "mean_token_accuracy": 0.7556751549243927,
+      "num_tokens": 75659171.0,
+      "step": 32710
+    },
+    {
+      "epoch": 0.29999083157605205,
+      "learning_rate": 0.00014000366736957918,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.7435480654239655,
+      "num_tokens": 75682235.0,
+      "step": 32720
+    },
+    {
+      "epoch": 0.3000825158155313,
+      "learning_rate": 0.00013998533052168331,
+      "loss": 0.8572,
+      "mean_token_accuracy": 0.7503647685050965,
+      "num_tokens": 75705256.0,
+      "step": 32730
+    },
+    {
+      "epoch": 0.30017420005501055,
+      "learning_rate": 0.00013996699367378748,
+      "loss": 0.849,
+      "mean_token_accuracy": 0.7448776602745056,
+      "num_tokens": 75728535.0,
+      "step": 32740
+    },
+    {
+      "epoch": 0.3002658842944898,
+      "learning_rate": 0.00013994865682589165,
+      "loss": 0.8409,
+      "mean_token_accuracy": 0.7476956903934479,
+      "num_tokens": 75752096.0,
+      "step": 32750
+    },
+    {
+      "epoch": 0.300357568533969,
+      "learning_rate": 0.00013993031997799578,
+      "loss": 0.8838,
+      "mean_token_accuracy": 0.7399802386760712,
+      "num_tokens": 75774942.0,
+      "step": 32760
+    },
+    {
+      "epoch": 0.30044925277344825,
+      "learning_rate": 0.00013991198313009995,
+      "loss": 0.8464,
+      "mean_token_accuracy": 0.752990335226059,
+      "num_tokens": 75797111.0,
+      "step": 32770
+    },
+    {
+      "epoch": 0.3005409370129275,
+      "learning_rate": 0.0001398936462822041,
+      "loss": 0.8609,
+      "mean_token_accuracy": 0.736510944366455,
+      "num_tokens": 75819608.0,
+      "step": 32780
+    },
+    {
+      "epoch": 0.3006326212524067,
+      "learning_rate": 0.00013987530943430825,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.7408483803272248,
+      "num_tokens": 75842093.0,
+      "step": 32790
+    },
+    {
+      "epoch": 0.30072430549188595,
+      "learning_rate": 0.0001398569725864124,
+      "loss": 0.8363,
+      "mean_token_accuracy": 0.7478606402873993,
+      "num_tokens": 75865258.0,
+      "step": 32800
+    },
+    {
+      "epoch": 0.3008159897313652,
+      "learning_rate": 0.00013983863573851656,
+      "loss": 0.8739,
+      "mean_token_accuracy": 0.7317023873329163,
+      "num_tokens": 75888126.0,
+      "step": 32810
+    },
+    {
+      "epoch": 0.3009076739708444,
+      "learning_rate": 0.00013982029889062072,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7460320651531219,
+      "num_tokens": 75911310.0,
+      "step": 32820
+    },
+    {
+      "epoch": 0.30099935821032364,
+      "learning_rate": 0.00013980196204272486,
+      "loss": 0.8217,
+      "mean_token_accuracy": 0.7521295309066772,
+      "num_tokens": 75933653.0,
+      "step": 32830
+    },
+    {
+      "epoch": 0.3010910424498029,
+      "learning_rate": 0.00013978362519482903,
+      "loss": 0.8238,
+      "mean_token_accuracy": 0.7496893882751465,
+      "num_tokens": 75955649.0,
+      "step": 32840
+    },
+    {
+      "epoch": 0.3011827266892821,
+      "learning_rate": 0.00013976528834693316,
+      "loss": 0.8675,
+      "mean_token_accuracy": 0.7385407745838165,
+      "num_tokens": 75978503.0,
+      "step": 32850
+    },
+    {
+      "epoch": 0.30127441092876134,
+      "learning_rate": 0.0001397469514990373,
+      "loss": 0.851,
+      "mean_token_accuracy": 0.7472930252552032,
+      "num_tokens": 76001930.0,
+      "step": 32860
+    },
+    {
+      "epoch": 0.3013660951682406,
+      "learning_rate": 0.00013972861465114147,
+      "loss": 0.8601,
+      "mean_token_accuracy": 0.7451577663421631,
+      "num_tokens": 76024711.0,
+      "step": 32870
+    },
+    {
+      "epoch": 0.30145777940771984,
+      "learning_rate": 0.00013971027780324563,
+      "loss": 0.8251,
+      "mean_token_accuracy": 0.7494319319725037,
+      "num_tokens": 76047325.0,
+      "step": 32880
+    },
+    {
+      "epoch": 0.30154946364719903,
+      "learning_rate": 0.0001396919409553498,
+      "loss": 0.8759,
+      "mean_token_accuracy": 0.7339219331741333,
+      "num_tokens": 76070650.0,
+      "step": 32890
+    },
+    {
+      "epoch": 0.3016411478866783,
+      "learning_rate": 0.00013967360410745394,
+      "loss": 0.8827,
+      "mean_token_accuracy": 0.7390013694763183,
+      "num_tokens": 76093518.0,
+      "step": 32900
+    },
+    {
+      "epoch": 0.30173283212615754,
+      "learning_rate": 0.00013965526725955808,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.7473307371139526,
+      "num_tokens": 76116985.0,
+      "step": 32910
+    },
+    {
+      "epoch": 0.30182451636563673,
+      "learning_rate": 0.00013963693041166224,
+      "loss": 0.87,
+      "mean_token_accuracy": 0.7436900436878204,
+      "num_tokens": 76140208.0,
+      "step": 32920
+    },
+    {
+      "epoch": 0.301916200605116,
+      "learning_rate": 0.00013961859356376638,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7430337429046631,
+      "num_tokens": 76163444.0,
+      "step": 32930
+    },
+    {
+      "epoch": 0.30200788484459523,
+      "learning_rate": 0.00013960025671587057,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7484816789627076,
+      "num_tokens": 76186322.0,
+      "step": 32940
+    },
+    {
+      "epoch": 0.3020995690840744,
+      "learning_rate": 0.0001395819198679747,
+      "loss": 0.8329,
+      "mean_token_accuracy": 0.7511256515979767,
+      "num_tokens": 76208826.0,
+      "step": 32950
+    },
+    {
+      "epoch": 0.3021912533235537,
+      "learning_rate": 0.00013956358302007885,
+      "loss": 0.8322,
+      "mean_token_accuracy": 0.752809476852417,
+      "num_tokens": 76232270.0,
+      "step": 32960
+    },
+    {
+      "epoch": 0.3022829375630329,
+      "learning_rate": 0.00013954524617218301,
+      "loss": 0.844,
+      "mean_token_accuracy": 0.7424205303192138,
+      "num_tokens": 76255069.0,
+      "step": 32970
+    },
+    {
+      "epoch": 0.3023746218025121,
+      "learning_rate": 0.00013952690932428715,
+      "loss": 0.8378,
+      "mean_token_accuracy": 0.7539783895015717,
+      "num_tokens": 76278489.0,
+      "step": 32980
+    },
+    {
+      "epoch": 0.3024663060419914,
+      "learning_rate": 0.00013950857247639132,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.7450504243373871,
+      "num_tokens": 76300346.0,
+      "step": 32990
+    },
+    {
+      "epoch": 0.3025579902814706,
+      "learning_rate": 0.00013949023562849546,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.7447242200374603,
+      "num_tokens": 76323112.0,
+      "step": 33000
+    },
+    {
+      "epoch": 0.3026496745209499,
+      "learning_rate": 0.00013947189878059962,
+      "loss": 0.8101,
+      "mean_token_accuracy": 0.7504319071769714,
+      "num_tokens": 76345720.0,
+      "step": 33010
+    },
+    {
+      "epoch": 0.30274135876042907,
+      "learning_rate": 0.0001394535619327038,
+      "loss": 0.859,
+      "mean_token_accuracy": 0.7491183936595917,
+      "num_tokens": 76368925.0,
+      "step": 33020
+    },
+    {
+      "epoch": 0.3028330429999083,
+      "learning_rate": 0.00013943522508480792,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.7520512044429779,
+      "num_tokens": 76392125.0,
+      "step": 33030
+    },
+    {
+      "epoch": 0.30292472723938757,
+      "learning_rate": 0.0001394168882369121,
+      "loss": 0.8565,
+      "mean_token_accuracy": 0.7417470753192902,
+      "num_tokens": 76414784.0,
+      "step": 33040
+    },
+    {
+      "epoch": 0.30301641147886677,
+      "learning_rate": 0.00013939855138901623,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7458143174648285,
+      "num_tokens": 76438553.0,
+      "step": 33050
+    },
+    {
+      "epoch": 0.303108095718346,
+      "learning_rate": 0.00013938021454112037,
+      "loss": 0.8357,
+      "mean_token_accuracy": 0.7528502702713012,
+      "num_tokens": 76461596.0,
+      "step": 33060
+    },
+    {
+      "epoch": 0.30319977995782527,
+      "learning_rate": 0.00013936187769322456,
+      "loss": 0.8864,
+      "mean_token_accuracy": 0.7383783102035523,
+      "num_tokens": 76484693.0,
+      "step": 33070
+    },
+    {
+      "epoch": 0.30329146419730446,
+      "learning_rate": 0.0001393435408453287,
+      "loss": 0.8461,
+      "mean_token_accuracy": 0.7431750297546387,
+      "num_tokens": 76507530.0,
+      "step": 33080
+    },
+    {
+      "epoch": 0.3033831484367837,
+      "learning_rate": 0.00013932520399743286,
+      "loss": 0.8776,
+      "mean_token_accuracy": 0.738301956653595,
+      "num_tokens": 76529391.0,
+      "step": 33090
+    },
+    {
+      "epoch": 0.30347483267626296,
+      "learning_rate": 0.000139306867149537,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.7473569512367249,
+      "num_tokens": 76551653.0,
+      "step": 33100
+    },
+    {
+      "epoch": 0.30356651691574216,
+      "learning_rate": 0.00013928853030164114,
+      "loss": 0.8513,
+      "mean_token_accuracy": 0.7458438456058503,
+      "num_tokens": 76574931.0,
+      "step": 33110
+    },
+    {
+      "epoch": 0.3036582011552214,
+      "learning_rate": 0.0001392701934537453,
+      "loss": 0.8563,
+      "mean_token_accuracy": 0.7456990718841553,
+      "num_tokens": 76598160.0,
+      "step": 33120
+    },
+    {
+      "epoch": 0.30374988539470066,
+      "learning_rate": 0.00013925185660584944,
+      "loss": 0.8362,
+      "mean_token_accuracy": 0.7468683779239654,
+      "num_tokens": 76621926.0,
+      "step": 33130
+    },
+    {
+      "epoch": 0.3038415696341799,
+      "learning_rate": 0.00013923351975795364,
+      "loss": 0.8364,
+      "mean_token_accuracy": 0.7583596706390381,
+      "num_tokens": 76644602.0,
+      "step": 33140
+    },
+    {
+      "epoch": 0.3039332538736591,
+      "learning_rate": 0.00013921518291005777,
+      "loss": 0.839,
+      "mean_token_accuracy": 0.7520948767662048,
+      "num_tokens": 76668096.0,
+      "step": 33150
+    },
+    {
+      "epoch": 0.30402493811313835,
+      "learning_rate": 0.0001391968460621619,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7482321858406067,
+      "num_tokens": 76690471.0,
+      "step": 33160
+    },
+    {
+      "epoch": 0.3041166223526176,
+      "learning_rate": 0.00013917850921426608,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7384559750556946,
+      "num_tokens": 76713138.0,
+      "step": 33170
+    },
+    {
+      "epoch": 0.3042083065920968,
+      "learning_rate": 0.00013916017236637022,
+      "loss": 0.8736,
+      "mean_token_accuracy": 0.738041752576828,
+      "num_tokens": 76735680.0,
+      "step": 33180
+    },
+    {
+      "epoch": 0.30429999083157605,
+      "learning_rate": 0.00013914183551847438,
+      "loss": 0.8289,
+      "mean_token_accuracy": 0.7484506666660309,
+      "num_tokens": 76758523.0,
+      "step": 33190
+    },
+    {
+      "epoch": 0.3043916750710553,
+      "learning_rate": 0.00013912349867057855,
+      "loss": 0.8823,
+      "mean_token_accuracy": 0.7356573045253754,
+      "num_tokens": 76780316.0,
+      "step": 33200
+    },
+    {
+      "epoch": 0.3044833593105345,
+      "learning_rate": 0.00013910516182268269,
+      "loss": 0.8146,
+      "mean_token_accuracy": 0.749269676208496,
+      "num_tokens": 76803997.0,
+      "step": 33210
+    },
+    {
+      "epoch": 0.30457504355001375,
+      "learning_rate": 0.00013908682497478685,
+      "loss": 0.8394,
+      "mean_token_accuracy": 0.7488521456718444,
+      "num_tokens": 76826201.0,
+      "step": 33220
+    },
+    {
+      "epoch": 0.304666727789493,
+      "learning_rate": 0.000139068488126891,
+      "loss": 0.8409,
+      "mean_token_accuracy": 0.7477450549602509,
+      "num_tokens": 76850018.0,
+      "step": 33230
+    },
+    {
+      "epoch": 0.3047584120289722,
+      "learning_rate": 0.00013905015127899515,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7425000071525574,
+      "num_tokens": 76872969.0,
+      "step": 33240
+    },
+    {
+      "epoch": 0.30485009626845144,
+      "learning_rate": 0.0001390318144310993,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7348301351070404,
+      "num_tokens": 76896052.0,
+      "step": 33250
+    },
+    {
+      "epoch": 0.3049417805079307,
+      "learning_rate": 0.00013901347758320343,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7494641184806824,
+      "num_tokens": 76917979.0,
+      "step": 33260
+    },
+    {
+      "epoch": 0.30503346474740994,
+      "learning_rate": 0.00013899514073530762,
+      "loss": 0.8406,
+      "mean_token_accuracy": 0.7472612977027893,
+      "num_tokens": 76941167.0,
+      "step": 33270
+    },
+    {
+      "epoch": 0.30512514898688914,
+      "learning_rate": 0.00013897680388741176,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7480234622955322,
+      "num_tokens": 76964086.0,
+      "step": 33280
+    },
+    {
+      "epoch": 0.3052168332263684,
+      "learning_rate": 0.00013895846703951593,
+      "loss": 0.8368,
+      "mean_token_accuracy": 0.7458792328834534,
+      "num_tokens": 76987183.0,
+      "step": 33290
+    },
+    {
+      "epoch": 0.30530851746584764,
+      "learning_rate": 0.00013894013019162007,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.742641395330429,
+      "num_tokens": 77010631.0,
+      "step": 33300
+    },
+    {
+      "epoch": 0.30540020170532683,
+      "learning_rate": 0.0001389217933437242,
+      "loss": 0.8646,
+      "mean_token_accuracy": 0.7419748723506927,
+      "num_tokens": 77033406.0,
+      "step": 33310
+    },
+    {
+      "epoch": 0.3054918859448061,
+      "learning_rate": 0.00013890345649582837,
+      "loss": 0.7912,
+      "mean_token_accuracy": 0.760988461971283,
+      "num_tokens": 77056069.0,
+      "step": 33320
+    },
+    {
+      "epoch": 0.30558357018428534,
+      "learning_rate": 0.00013888511964793254,
+      "loss": 0.8489,
+      "mean_token_accuracy": 0.7463546633720398,
+      "num_tokens": 77079187.0,
+      "step": 33330
+    },
+    {
+      "epoch": 0.30567525442376453,
+      "learning_rate": 0.0001388667828000367,
+      "loss": 0.8792,
+      "mean_token_accuracy": 0.7405191779136657,
+      "num_tokens": 77101817.0,
+      "step": 33340
+    },
+    {
+      "epoch": 0.3057669386632438,
+      "learning_rate": 0.00013884844595214084,
+      "loss": 0.8568,
+      "mean_token_accuracy": 0.743732112646103,
+      "num_tokens": 77124917.0,
+      "step": 33350
+    },
+    {
+      "epoch": 0.30585862290272303,
+      "learning_rate": 0.00013883010910424498,
+      "loss": 0.8591,
+      "mean_token_accuracy": 0.7493031442165374,
+      "num_tokens": 77148323.0,
+      "step": 33360
+    },
+    {
+      "epoch": 0.3059503071422023,
+      "learning_rate": 0.00013881177225634914,
+      "loss": 0.828,
+      "mean_token_accuracy": 0.7545134902000428,
+      "num_tokens": 77171971.0,
+      "step": 33370
+    },
+    {
+      "epoch": 0.3060419913816815,
+      "learning_rate": 0.00013879343540845328,
+      "loss": 0.8334,
+      "mean_token_accuracy": 0.7538052439689636,
+      "num_tokens": 77195015.0,
+      "step": 33380
+    },
+    {
+      "epoch": 0.30613367562116073,
+      "learning_rate": 0.00013877509856055745,
+      "loss": 0.8962,
+      "mean_token_accuracy": 0.7372683942317962,
+      "num_tokens": 77218283.0,
+      "step": 33390
+    },
+    {
+      "epoch": 0.30622535986064,
+      "learning_rate": 0.0001387567617126616,
+      "loss": 0.8271,
+      "mean_token_accuracy": 0.7487876236438751,
+      "num_tokens": 77241440.0,
+      "step": 33400
+    },
+    {
+      "epoch": 0.3063170441001192,
+      "learning_rate": 0.00013873842486476575,
+      "loss": 0.8656,
+      "mean_token_accuracy": 0.7421483814716339,
+      "num_tokens": 77264426.0,
+      "step": 33410
+    },
+    {
+      "epoch": 0.3064087283395984,
+      "learning_rate": 0.00013872008801686992,
+      "loss": 0.8591,
+      "mean_token_accuracy": 0.7459542334079743,
+      "num_tokens": 77287335.0,
+      "step": 33420
+    },
+    {
+      "epoch": 0.3065004125790777,
+      "learning_rate": 0.00013870175116897405,
+      "loss": 0.8505,
+      "mean_token_accuracy": 0.7433769702911377,
+      "num_tokens": 77310740.0,
+      "step": 33430
+    },
+    {
+      "epoch": 0.30659209681855687,
+      "learning_rate": 0.00013868341432107822,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7411685645580292,
+      "num_tokens": 77333516.0,
+      "step": 33440
+    },
+    {
+      "epoch": 0.3066837810580361,
+      "learning_rate": 0.00013866507747318236,
+      "loss": 0.8347,
+      "mean_token_accuracy": 0.7427520096302033,
+      "num_tokens": 77356228.0,
+      "step": 33450
+    },
+    {
+      "epoch": 0.30677546529751537,
+      "learning_rate": 0.00013864674062528652,
+      "loss": 0.8709,
+      "mean_token_accuracy": 0.7422504723072052,
+      "num_tokens": 77379003.0,
+      "step": 33460
+    },
+    {
+      "epoch": 0.30686714953699457,
+      "learning_rate": 0.0001386284037773907,
+      "loss": 0.8177,
+      "mean_token_accuracy": 0.7486840307712554,
+      "num_tokens": 77402464.0,
+      "step": 33470
+    },
+    {
+      "epoch": 0.3069588337764738,
+      "learning_rate": 0.00013861006692949483,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7467386662960053,
+      "num_tokens": 77425347.0,
+      "step": 33480
+    },
+    {
+      "epoch": 0.30705051801595307,
+      "learning_rate": 0.000138591730081599,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7501772046089172,
+      "num_tokens": 77447614.0,
+      "step": 33490
+    },
+    {
+      "epoch": 0.3071422022554323,
+      "learning_rate": 0.00013857339323370313,
+      "loss": 0.8269,
+      "mean_token_accuracy": 0.7498204708099365,
+      "num_tokens": 77471207.0,
+      "step": 33500
+    },
+    {
+      "epoch": 0.3072338864949115,
+      "learning_rate": 0.00013855505638580727,
+      "loss": 0.841,
+      "mean_token_accuracy": 0.7512203574180603,
+      "num_tokens": 77494344.0,
+      "step": 33510
+    },
+    {
+      "epoch": 0.30732557073439076,
+      "learning_rate": 0.00013853671953791143,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7417186796665192,
+      "num_tokens": 77516961.0,
+      "step": 33520
+    },
+    {
+      "epoch": 0.30741725497387,
+      "learning_rate": 0.0001385183826900156,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7555020093917847,
+      "num_tokens": 77539976.0,
+      "step": 33530
+    },
+    {
+      "epoch": 0.3075089392133492,
+      "learning_rate": 0.00013850004584211976,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7431936383247375,
+      "num_tokens": 77563528.0,
+      "step": 33540
+    },
+    {
+      "epoch": 0.30760062345282846,
+      "learning_rate": 0.0001384817089942239,
+      "loss": 0.8348,
+      "mean_token_accuracy": 0.7480647802352905,
+      "num_tokens": 77586766.0,
+      "step": 33550
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "learning_rate": 0.00013846337214632804,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.7537217140197754,
+      "num_tokens": 77609032.0,
+      "step": 33560
+    },
+    {
+      "epoch": 0.3077839919317869,
+      "learning_rate": 0.0001384450352984322,
+      "loss": 0.8445,
+      "mean_token_accuracy": 0.7477411210536957,
+      "num_tokens": 77632016.0,
+      "step": 33570
+    },
+    {
+      "epoch": 0.30787567617126615,
+      "learning_rate": 0.00013842669845053635,
+      "loss": 0.906,
+      "mean_token_accuracy": 0.7402415156364441,
+      "num_tokens": 77654913.0,
+      "step": 33580
+    },
+    {
+      "epoch": 0.3079673604107454,
+      "learning_rate": 0.0001384083616026405,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7412020742893219,
+      "num_tokens": 77677818.0,
+      "step": 33590
+    },
+    {
+      "epoch": 0.3080590446502246,
+      "learning_rate": 0.00013839002475474468,
+      "loss": 0.8731,
+      "mean_token_accuracy": 0.7371971726417541,
+      "num_tokens": 77701930.0,
+      "step": 33600
+    },
+    {
+      "epoch": 0.30815072888970385,
+      "learning_rate": 0.00013837168790684881,
+      "loss": 0.844,
+      "mean_token_accuracy": 0.7502223968505859,
+      "num_tokens": 77724335.0,
+      "step": 33610
+    },
+    {
+      "epoch": 0.3082424131291831,
+      "learning_rate": 0.00013835335105895298,
+      "loss": 0.8432,
+      "mean_token_accuracy": 0.7439646542072296,
+      "num_tokens": 77747638.0,
+      "step": 33620
+    },
+    {
+      "epoch": 0.30833409736866235,
+      "learning_rate": 0.00013833501421105712,
+      "loss": 0.8704,
+      "mean_token_accuracy": 0.7407457113265992,
+      "num_tokens": 77770369.0,
+      "step": 33630
+    },
+    {
+      "epoch": 0.30842578160814155,
+      "learning_rate": 0.00013831667736316128,
+      "loss": 0.7746,
+      "mean_token_accuracy": 0.7646201491355896,
+      "num_tokens": 77793521.0,
+      "step": 33640
+    },
+    {
+      "epoch": 0.3085174658476208,
+      "learning_rate": 0.00013829834051526542,
+      "loss": 0.8796,
+      "mean_token_accuracy": 0.7416764795780182,
+      "num_tokens": 77816667.0,
+      "step": 33650
+    },
+    {
+      "epoch": 0.30860915008710005,
+      "learning_rate": 0.0001382800036673696,
+      "loss": 0.8788,
+      "mean_token_accuracy": 0.7355660200119019,
+      "num_tokens": 77840037.0,
+      "step": 33660
+    },
+    {
+      "epoch": 0.30870083432657924,
+      "learning_rate": 0.00013826166681947375,
+      "loss": 0.8006,
+      "mean_token_accuracy": 0.7525620520114898,
+      "num_tokens": 77862699.0,
+      "step": 33670
+    },
+    {
+      "epoch": 0.3087925185660585,
+      "learning_rate": 0.0001382433299715779,
+      "loss": 0.8562,
+      "mean_token_accuracy": 0.7428406774997711,
+      "num_tokens": 77885136.0,
+      "step": 33680
+    },
+    {
+      "epoch": 0.30888420280553774,
+      "learning_rate": 0.00013822499312368206,
+      "loss": 0.8449,
+      "mean_token_accuracy": 0.739268159866333,
+      "num_tokens": 77907679.0,
+      "step": 33690
+    },
+    {
+      "epoch": 0.30897588704501694,
+      "learning_rate": 0.0001382066562757862,
+      "loss": 0.8431,
+      "mean_token_accuracy": 0.7458111703395843,
+      "num_tokens": 77930616.0,
+      "step": 33700
+    },
+    {
+      "epoch": 0.3090675712844962,
+      "learning_rate": 0.00013818831942789033,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7450494110584259,
+      "num_tokens": 77953639.0,
+      "step": 33710
+    },
+    {
+      "epoch": 0.30915925552397544,
+      "learning_rate": 0.0001381699825799945,
+      "loss": 0.8643,
+      "mean_token_accuracy": 0.7464971423149109,
+      "num_tokens": 77976914.0,
+      "step": 33720
+    },
+    {
+      "epoch": 0.30925093976345464,
+      "learning_rate": 0.00013815164573209866,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.7457862615585327,
+      "num_tokens": 77999509.0,
+      "step": 33730
+    },
+    {
+      "epoch": 0.3093426240029339,
+      "learning_rate": 0.00013813330888420283,
+      "loss": 0.8424,
+      "mean_token_accuracy": 0.7435806334018707,
+      "num_tokens": 78023950.0,
+      "step": 33740
+    },
+    {
+      "epoch": 0.30943430824241314,
+      "learning_rate": 0.00013811497203630697,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.7419386982917786,
+      "num_tokens": 78047833.0,
+      "step": 33750
+    },
+    {
+      "epoch": 0.3095259924818924,
+      "learning_rate": 0.0001380966351884111,
+      "loss": 0.8387,
+      "mean_token_accuracy": 0.7498943448066712,
+      "num_tokens": 78070834.0,
+      "step": 33760
+    },
+    {
+      "epoch": 0.3096176767213716,
+      "learning_rate": 0.00013807829834051527,
+      "loss": 0.8471,
+      "mean_token_accuracy": 0.7432628989219665,
+      "num_tokens": 78093543.0,
+      "step": 33770
+    },
+    {
+      "epoch": 0.30970936096085083,
+      "learning_rate": 0.0001380599614926194,
+      "loss": 0.8613,
+      "mean_token_accuracy": 0.7407761573791504,
+      "num_tokens": 78116708.0,
+      "step": 33780
+    },
+    {
+      "epoch": 0.3098010452003301,
+      "learning_rate": 0.00013804162464472358,
+      "loss": 0.8765,
+      "mean_token_accuracy": 0.7389798283576965,
+      "num_tokens": 78140055.0,
+      "step": 33790
+    },
+    {
+      "epoch": 0.3098927294398093,
+      "learning_rate": 0.00013802328779682774,
+      "loss": 0.8564,
+      "mean_token_accuracy": 0.750953084230423,
+      "num_tokens": 78163584.0,
+      "step": 33800
+    },
+    {
+      "epoch": 0.30998441367928853,
+      "learning_rate": 0.00013800495094893188,
+      "loss": 0.8509,
+      "mean_token_accuracy": 0.7473524689674378,
+      "num_tokens": 78186526.0,
+      "step": 33810
+    },
+    {
+      "epoch": 0.3100760979187678,
+      "learning_rate": 0.00013798661410103604,
+      "loss": 0.8337,
+      "mean_token_accuracy": 0.7452644348144531,
+      "num_tokens": 78209865.0,
+      "step": 33820
+    },
+    {
+      "epoch": 0.310167782158247,
+      "learning_rate": 0.00013796827725314018,
+      "loss": 0.8362,
+      "mean_token_accuracy": 0.7420852065086365,
+      "num_tokens": 78232857.0,
+      "step": 33830
+    },
+    {
+      "epoch": 0.3102594663977262,
+      "learning_rate": 0.00013794994040524435,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7465553224086762,
+      "num_tokens": 78256639.0,
+      "step": 33840
+    },
+    {
+      "epoch": 0.3103511506372055,
+      "learning_rate": 0.00013793160355734849,
+      "loss": 0.8656,
+      "mean_token_accuracy": 0.7351106941699982,
+      "num_tokens": 78279723.0,
+      "step": 33850
+    },
+    {
+      "epoch": 0.3104428348766847,
+      "learning_rate": 0.00013791326670945265,
+      "loss": 0.8627,
+      "mean_token_accuracy": 0.7368688642978668,
+      "num_tokens": 78302587.0,
+      "step": 33860
+    },
+    {
+      "epoch": 0.3105345191161639,
+      "learning_rate": 0.00013789492986155682,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.7496148109436035,
+      "num_tokens": 78325276.0,
+      "step": 33870
+    },
+    {
+      "epoch": 0.31062620335564317,
+      "learning_rate": 0.00013787659301366096,
+      "loss": 0.8181,
+      "mean_token_accuracy": 0.7458925127983094,
+      "num_tokens": 78348670.0,
+      "step": 33880
+    },
+    {
+      "epoch": 0.3107178875951224,
+      "learning_rate": 0.00013785825616576512,
+      "loss": 0.8495,
+      "mean_token_accuracy": 0.7495343565940857,
+      "num_tokens": 78371576.0,
+      "step": 33890
+    },
+    {
+      "epoch": 0.3108095718346016,
+      "learning_rate": 0.00013783991931786926,
+      "loss": 0.872,
+      "mean_token_accuracy": 0.7415913641452789,
+      "num_tokens": 78394561.0,
+      "step": 33900
+    },
+    {
+      "epoch": 0.31090125607408087,
+      "learning_rate": 0.0001378215824699734,
+      "loss": 0.8622,
+      "mean_token_accuracy": 0.7382225751876831,
+      "num_tokens": 78417398.0,
+      "step": 33910
+    },
+    {
+      "epoch": 0.3109929403135601,
+      "learning_rate": 0.0001378032456220776,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7446352064609527,
+      "num_tokens": 78439842.0,
+      "step": 33920
+    },
+    {
+      "epoch": 0.3110846245530393,
+      "learning_rate": 0.00013778490877418173,
+      "loss": 0.8918,
+      "mean_token_accuracy": 0.7458747148513794,
+      "num_tokens": 78462547.0,
+      "step": 33930
+    },
+    {
+      "epoch": 0.31117630879251856,
+      "learning_rate": 0.0001377665719262859,
+      "loss": 0.8929,
+      "mean_token_accuracy": 0.7326566874980927,
+      "num_tokens": 78485928.0,
+      "step": 33940
+    },
+    {
+      "epoch": 0.3112679930319978,
+      "learning_rate": 0.00013774823507839003,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7494336009025574,
+      "num_tokens": 78508363.0,
+      "step": 33950
+    },
+    {
+      "epoch": 0.311359677271477,
+      "learning_rate": 0.00013772989823049417,
+      "loss": 0.8582,
+      "mean_token_accuracy": 0.7438672244548797,
+      "num_tokens": 78531327.0,
+      "step": 33960
+    },
+    {
+      "epoch": 0.31145136151095626,
+      "learning_rate": 0.00013771156138259834,
+      "loss": 0.8786,
+      "mean_token_accuracy": 0.7443876504898072,
+      "num_tokens": 78555046.0,
+      "step": 33970
+    },
+    {
+      "epoch": 0.3115430457504355,
+      "learning_rate": 0.00013769322453470247,
+      "loss": 0.8256,
+      "mean_token_accuracy": 0.7507291913032532,
+      "num_tokens": 78577902.0,
+      "step": 33980
+    },
+    {
+      "epoch": 0.31163472998991476,
+      "learning_rate": 0.00013767488768680664,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7446135640144348,
+      "num_tokens": 78600485.0,
+      "step": 33990
+    },
+    {
+      "epoch": 0.31172641422939396,
+      "learning_rate": 0.0001376565508389108,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7453547239303588,
+      "num_tokens": 78622655.0,
+      "step": 34000
+    },
+    {
+      "epoch": 0.3118180984688732,
+      "learning_rate": 0.00013763821399101494,
+      "loss": 0.8801,
+      "mean_token_accuracy": 0.7399561107158661,
+      "num_tokens": 78646145.0,
+      "step": 34010
+    },
+    {
+      "epoch": 0.31190978270835246,
+      "learning_rate": 0.0001376198771431191,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7371825695037841,
+      "num_tokens": 78669868.0,
+      "step": 34020
+    },
+    {
+      "epoch": 0.31200146694783165,
+      "learning_rate": 0.00013760154029522325,
+      "loss": 0.8354,
+      "mean_token_accuracy": 0.7505948424339295,
+      "num_tokens": 78692541.0,
+      "step": 34030
+    },
+    {
+      "epoch": 0.3120931511873109,
+      "learning_rate": 0.0001375832034473274,
+      "loss": 0.8324,
+      "mean_token_accuracy": 0.7473336160182953,
+      "num_tokens": 78715340.0,
+      "step": 34040
+    },
+    {
+      "epoch": 0.31218483542679015,
+      "learning_rate": 0.00013756486659943158,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.7392273724079133,
+      "num_tokens": 78737797.0,
+      "step": 34050
+    },
+    {
+      "epoch": 0.31227651966626935,
+      "learning_rate": 0.00013754652975153572,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.7455706655979156,
+      "num_tokens": 78760741.0,
+      "step": 34060
+    },
+    {
+      "epoch": 0.3123682039057486,
+      "learning_rate": 0.00013752819290363988,
+      "loss": 0.8897,
+      "mean_token_accuracy": 0.7384982705116272,
+      "num_tokens": 78785235.0,
+      "step": 34070
+    },
+    {
+      "epoch": 0.31245988814522785,
+      "learning_rate": 0.00013750985605574402,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7436144709587097,
+      "num_tokens": 78808390.0,
+      "step": 34080
+    },
+    {
+      "epoch": 0.31255157238470704,
+      "learning_rate": 0.00013749151920784819,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.742193204164505,
+      "num_tokens": 78831392.0,
+      "step": 34090
+    },
+    {
+      "epoch": 0.3126432566241863,
+      "learning_rate": 0.00013747318235995232,
+      "loss": 0.8323,
+      "mean_token_accuracy": 0.741822338104248,
+      "num_tokens": 78854868.0,
+      "step": 34100
+    },
+    {
+      "epoch": 0.31273494086366554,
+      "learning_rate": 0.00013745484551205646,
+      "loss": 0.7973,
+      "mean_token_accuracy": 0.7539364516735076,
+      "num_tokens": 78877421.0,
+      "step": 34110
+    },
+    {
+      "epoch": 0.3128266251031448,
+      "learning_rate": 0.00013743650866416065,
+      "loss": 0.8239,
+      "mean_token_accuracy": 0.7552946329116821,
+      "num_tokens": 78901198.0,
+      "step": 34120
+    },
+    {
+      "epoch": 0.312918309342624,
+      "learning_rate": 0.0001374181718162648,
+      "loss": 0.8978,
+      "mean_token_accuracy": 0.7341907978057861,
+      "num_tokens": 78923534.0,
+      "step": 34130
+    },
+    {
+      "epoch": 0.31300999358210324,
+      "learning_rate": 0.00013739983496836896,
+      "loss": 0.8733,
+      "mean_token_accuracy": 0.7404236376285553,
+      "num_tokens": 78946443.0,
+      "step": 34140
+    },
+    {
+      "epoch": 0.3131016778215825,
+      "learning_rate": 0.0001373814981204731,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.7449717223644257,
+      "num_tokens": 78970146.0,
+      "step": 34150
+    },
+    {
+      "epoch": 0.3131933620610617,
+      "learning_rate": 0.00013736316127257723,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.7426335871219635,
+      "num_tokens": 78992758.0,
+      "step": 34160
+    },
+    {
+      "epoch": 0.31328504630054094,
+      "learning_rate": 0.0001373448244246814,
+      "loss": 0.8367,
+      "mean_token_accuracy": 0.750715833902359,
+      "num_tokens": 79015249.0,
+      "step": 34170
+    },
+    {
+      "epoch": 0.3133767305400202,
+      "learning_rate": 0.00013732648757678557,
+      "loss": 0.8978,
+      "mean_token_accuracy": 0.7321100652217865,
+      "num_tokens": 79037518.0,
+      "step": 34180
+    },
+    {
+      "epoch": 0.3134684147794994,
+      "learning_rate": 0.0001373081507288897,
+      "loss": 0.8428,
+      "mean_token_accuracy": 0.7446274638175965,
+      "num_tokens": 79060436.0,
+      "step": 34190
+    },
+    {
+      "epoch": 0.31356009901897863,
+      "learning_rate": 0.00013728981388099387,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.743911737203598,
+      "num_tokens": 79083372.0,
+      "step": 34200
+    },
+    {
+      "epoch": 0.3136517832584579,
+      "learning_rate": 0.000137271477033098,
+      "loss": 0.8306,
+      "mean_token_accuracy": 0.7476195931434632,
+      "num_tokens": 79106779.0,
+      "step": 34210
+    },
+    {
+      "epoch": 0.3137434674979371,
+      "learning_rate": 0.00013725314018520217,
+      "loss": 0.8751,
+      "mean_token_accuracy": 0.7391022264957428,
+      "num_tokens": 79129618.0,
+      "step": 34220
+    },
+    {
+      "epoch": 0.31383515173741633,
+      "learning_rate": 0.0001372348033373063,
+      "loss": 0.8712,
+      "mean_token_accuracy": 0.7366578221321106,
+      "num_tokens": 79152007.0,
+      "step": 34230
+    },
+    {
+      "epoch": 0.3139268359768956,
+      "learning_rate": 0.00013721646648941048,
+      "loss": 0.7959,
+      "mean_token_accuracy": 0.7588099181652069,
+      "num_tokens": 79175125.0,
+      "step": 34240
+    },
+    {
+      "epoch": 0.31401852021637483,
+      "learning_rate": 0.00013719812964151464,
+      "loss": 0.8432,
+      "mean_token_accuracy": 0.7426735818386078,
+      "num_tokens": 79198722.0,
+      "step": 34250
+    },
+    {
+      "epoch": 0.314110204455854,
+      "learning_rate": 0.00013717979279361878,
+      "loss": 0.8843,
+      "mean_token_accuracy": 0.734452360868454,
+      "num_tokens": 79221976.0,
+      "step": 34260
+    },
+    {
+      "epoch": 0.3142018886953333,
+      "learning_rate": 0.00013716145594572295,
+      "loss": 0.8364,
+      "mean_token_accuracy": 0.7462021052837372,
+      "num_tokens": 79245526.0,
+      "step": 34270
+    },
+    {
+      "epoch": 0.3142935729348125,
+      "learning_rate": 0.00013714311909782708,
+      "loss": 0.8156,
+      "mean_token_accuracy": 0.7524182617664337,
+      "num_tokens": 79268613.0,
+      "step": 34280
+    },
+    {
+      "epoch": 0.3143852571742917,
+      "learning_rate": 0.00013712478224993125,
+      "loss": 0.8682,
+      "mean_token_accuracy": 0.7409781634807586,
+      "num_tokens": 79291659.0,
+      "step": 34290
+    },
+    {
+      "epoch": 0.31447694141377097,
+      "learning_rate": 0.0001371064454020354,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7476778864860535,
+      "num_tokens": 79315501.0,
+      "step": 34300
+    },
+    {
+      "epoch": 0.3145686256532502,
+      "learning_rate": 0.00013708810855413955,
+      "loss": 0.86,
+      "mean_token_accuracy": 0.7427622258663178,
+      "num_tokens": 79338230.0,
+      "step": 34310
+    },
+    {
+      "epoch": 0.3146603098927294,
+      "learning_rate": 0.00013706977170624372,
+      "loss": 0.8422,
+      "mean_token_accuracy": 0.7434896886348724,
+      "num_tokens": 79361510.0,
+      "step": 34320
+    },
+    {
+      "epoch": 0.31475199413220867,
+      "learning_rate": 0.00013705143485834786,
+      "loss": 0.8829,
+      "mean_token_accuracy": 0.7420975625514984,
+      "num_tokens": 79384636.0,
+      "step": 34330
+    },
+    {
+      "epoch": 0.3148436783716879,
+      "learning_rate": 0.00013703309801045202,
+      "loss": 0.8748,
+      "mean_token_accuracy": 0.739655327796936,
+      "num_tokens": 79407919.0,
+      "step": 34340
+    },
+    {
+      "epoch": 0.3149353626111671,
+      "learning_rate": 0.00013701476116255616,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7402777910232544,
+      "num_tokens": 79430699.0,
+      "step": 34350
+    },
+    {
+      "epoch": 0.31502704685064636,
+      "learning_rate": 0.0001369964243146603,
+      "loss": 0.8899,
+      "mean_token_accuracy": 0.736571878194809,
+      "num_tokens": 79453750.0,
+      "step": 34360
+    },
+    {
+      "epoch": 0.3151187310901256,
+      "learning_rate": 0.00013697808746676446,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7471187770366668,
+      "num_tokens": 79476905.0,
+      "step": 34370
+    },
+    {
+      "epoch": 0.31521041532960486,
+      "learning_rate": 0.00013695975061886863,
+      "loss": 0.812,
+      "mean_token_accuracy": 0.7540494322776794,
+      "num_tokens": 79499697.0,
+      "step": 34380
+    },
+    {
+      "epoch": 0.31530209956908406,
+      "learning_rate": 0.00013694141377097277,
+      "loss": 0.8571,
+      "mean_token_accuracy": 0.7447143197059631,
+      "num_tokens": 79523491.0,
+      "step": 34390
+    },
+    {
+      "epoch": 0.3153937838085633,
+      "learning_rate": 0.00013692307692307693,
+      "loss": 0.8662,
+      "mean_token_accuracy": 0.7435140669345855,
+      "num_tokens": 79546953.0,
+      "step": 34400
+    },
+    {
+      "epoch": 0.31548546804804256,
+      "learning_rate": 0.00013690474007518107,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7524021387100219,
+      "num_tokens": 79569717.0,
+      "step": 34410
+    },
+    {
+      "epoch": 0.31557715228752176,
+      "learning_rate": 0.00013688640322728524,
+      "loss": 0.8186,
+      "mean_token_accuracy": 0.7480797469615936,
+      "num_tokens": 79593407.0,
+      "step": 34420
+    },
+    {
+      "epoch": 0.315668836527001,
+      "learning_rate": 0.00013686806637938938,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7463244736194611,
+      "num_tokens": 79616375.0,
+      "step": 34430
+    },
+    {
+      "epoch": 0.31576052076648026,
+      "learning_rate": 0.00013684972953149354,
+      "loss": 0.8918,
+      "mean_token_accuracy": 0.7361148715019226,
+      "num_tokens": 79639236.0,
+      "step": 34440
+    },
+    {
+      "epoch": 0.31585220500595945,
+      "learning_rate": 0.0001368313926835977,
+      "loss": 0.8359,
+      "mean_token_accuracy": 0.7510080814361573,
+      "num_tokens": 79663088.0,
+      "step": 34450
+    },
+    {
+      "epoch": 0.3159438892454387,
+      "learning_rate": 0.00013681305583570185,
+      "loss": 0.8547,
+      "mean_token_accuracy": 0.7443218410015107,
+      "num_tokens": 79686905.0,
+      "step": 34460
+    },
+    {
+      "epoch": 0.31603557348491795,
+      "learning_rate": 0.000136794718987806,
+      "loss": 0.856,
+      "mean_token_accuracy": 0.7426484644412994,
+      "num_tokens": 79709982.0,
+      "step": 34470
+    },
+    {
+      "epoch": 0.3161272577243972,
+      "learning_rate": 0.00013677638213991015,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.7364355206489563,
+      "num_tokens": 79732331.0,
+      "step": 34480
+    },
+    {
+      "epoch": 0.3162189419638764,
+      "learning_rate": 0.00013675804529201431,
+      "loss": 0.8169,
+      "mean_token_accuracy": 0.7513714134693146,
+      "num_tokens": 79755980.0,
+      "step": 34490
+    },
+    {
+      "epoch": 0.31631062620335565,
+      "learning_rate": 0.00013673970844411845,
+      "loss": 0.7929,
+      "mean_token_accuracy": 0.7554454922676086,
+      "num_tokens": 79779575.0,
+      "step": 34500
+    },
+    {
+      "epoch": 0.3164023104428349,
+      "learning_rate": 0.00013672137159622262,
+      "loss": 0.8262,
+      "mean_token_accuracy": 0.7452967226505279,
+      "num_tokens": 79803051.0,
+      "step": 34510
+    },
+    {
+      "epoch": 0.3164939946823141,
+      "learning_rate": 0.00013670303474832678,
+      "loss": 0.7819,
+      "mean_token_accuracy": 0.7652263939380646,
+      "num_tokens": 79825345.0,
+      "step": 34520
+    },
+    {
+      "epoch": 0.31658567892179335,
+      "learning_rate": 0.00013668469790043092,
+      "loss": 0.8494,
+      "mean_token_accuracy": 0.7425295948982239,
+      "num_tokens": 79848108.0,
+      "step": 34530
+    },
+    {
+      "epoch": 0.3166773631612726,
+      "learning_rate": 0.0001366663610525351,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.7419913232326507,
+      "num_tokens": 79871056.0,
+      "step": 34540
+    },
+    {
+      "epoch": 0.3167690474007518,
+      "learning_rate": 0.00013664802420463923,
+      "loss": 0.8669,
+      "mean_token_accuracy": 0.7456808865070343,
+      "num_tokens": 79894582.0,
+      "step": 34550
+    },
+    {
+      "epoch": 0.31686073164023104,
+      "learning_rate": 0.00013662968735674336,
+      "loss": 0.8165,
+      "mean_token_accuracy": 0.7550671517848968,
+      "num_tokens": 79917591.0,
+      "step": 34560
+    },
+    {
+      "epoch": 0.3169524158797103,
+      "learning_rate": 0.00013661135050884756,
+      "loss": 0.8877,
+      "mean_token_accuracy": 0.7403483629226685,
+      "num_tokens": 79940644.0,
+      "step": 34570
+    },
+    {
+      "epoch": 0.3170441001191895,
+      "learning_rate": 0.0001365930136609517,
+      "loss": 0.8556,
+      "mean_token_accuracy": 0.7453246176242828,
+      "num_tokens": 79964170.0,
+      "step": 34580
+    },
+    {
+      "epoch": 0.31713578435866874,
+      "learning_rate": 0.00013657467681305583,
+      "loss": 0.8926,
+      "mean_token_accuracy": 0.7324411630630493,
+      "num_tokens": 79987016.0,
+      "step": 34590
+    },
+    {
+      "epoch": 0.317227468598148,
+      "learning_rate": 0.00013655633996516,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7431699335575104,
+      "num_tokens": 80010734.0,
+      "step": 34600
+    },
+    {
+      "epoch": 0.31731915283762724,
+      "learning_rate": 0.00013653800311726414,
+      "loss": 0.8842,
+      "mean_token_accuracy": 0.7378191828727723,
+      "num_tokens": 80033339.0,
+      "step": 34610
+    },
+    {
+      "epoch": 0.31741083707710643,
+      "learning_rate": 0.0001365196662693683,
+      "loss": 0.8356,
+      "mean_token_accuracy": 0.7426839649677277,
+      "num_tokens": 80056594.0,
+      "step": 34620
+    },
+    {
+      "epoch": 0.3175025213165857,
+      "learning_rate": 0.00013650132942147244,
+      "loss": 0.844,
+      "mean_token_accuracy": 0.7452988803386689,
+      "num_tokens": 80079577.0,
+      "step": 34630
+    },
+    {
+      "epoch": 0.31759420555606493,
+      "learning_rate": 0.0001364829925735766,
+      "loss": 0.8824,
+      "mean_token_accuracy": 0.7362828552722931,
+      "num_tokens": 80102624.0,
+      "step": 34640
+    },
+    {
+      "epoch": 0.31768588979554413,
+      "learning_rate": 0.00013646465572568077,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7441802263259888,
+      "num_tokens": 80125360.0,
+      "step": 34650
+    },
+    {
+      "epoch": 0.3177775740350234,
+      "learning_rate": 0.0001364463188777849,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7502237677574157,
+      "num_tokens": 80148354.0,
+      "step": 34660
+    },
+    {
+      "epoch": 0.31786925827450263,
+      "learning_rate": 0.00013642798202988908,
+      "loss": 0.8152,
+      "mean_token_accuracy": 0.755932092666626,
+      "num_tokens": 80171617.0,
+      "step": 34670
+    },
+    {
+      "epoch": 0.3179609425139818,
+      "learning_rate": 0.0001364096451819932,
+      "loss": 0.8789,
+      "mean_token_accuracy": 0.7385014951229095,
+      "num_tokens": 80194206.0,
+      "step": 34680
+    },
+    {
+      "epoch": 0.3180526267534611,
+      "learning_rate": 0.00013639130833409738,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7461274921894073,
+      "num_tokens": 80217109.0,
+      "step": 34690
+    },
+    {
+      "epoch": 0.3181443109929403,
+      "learning_rate": 0.00013637297148620154,
+      "loss": 0.8276,
+      "mean_token_accuracy": 0.7556744992733002,
+      "num_tokens": 80241261.0,
+      "step": 34700
+    },
+    {
+      "epoch": 0.3182359952324195,
+      "learning_rate": 0.00013635463463830568,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.744313532114029,
+      "num_tokens": 80263934.0,
+      "step": 34710
+    },
+    {
+      "epoch": 0.3183276794718988,
+      "learning_rate": 0.00013633629779040985,
+      "loss": 0.8051,
+      "mean_token_accuracy": 0.7545833706855773,
+      "num_tokens": 80286978.0,
+      "step": 34720
+    },
+    {
+      "epoch": 0.318419363711378,
+      "learning_rate": 0.00013631796094251399,
+      "loss": 0.8466,
+      "mean_token_accuracy": 0.7389474272727966,
+      "num_tokens": 80309883.0,
+      "step": 34730
+    },
+    {
+      "epoch": 0.3185110479508573,
+      "learning_rate": 0.00013629962409461815,
+      "loss": 0.8129,
+      "mean_token_accuracy": 0.7531138718128204,
+      "num_tokens": 80333232.0,
+      "step": 34740
+    },
+    {
+      "epoch": 0.31860273219033647,
+      "learning_rate": 0.0001362812872467223,
+      "loss": 0.8768,
+      "mean_token_accuracy": 0.7434490025043488,
+      "num_tokens": 80356688.0,
+      "step": 34750
+    },
+    {
+      "epoch": 0.3186944164298157,
+      "learning_rate": 0.00013626295039882643,
+      "loss": 0.8695,
+      "mean_token_accuracy": 0.745382410287857,
+      "num_tokens": 80380400.0,
+      "step": 34760
+    },
+    {
+      "epoch": 0.31878610066929497,
+      "learning_rate": 0.00013624461355093062,
+      "loss": 0.8694,
+      "mean_token_accuracy": 0.7431151747703553,
+      "num_tokens": 80403509.0,
+      "step": 34770
+    },
+    {
+      "epoch": 0.31887778490877416,
+      "learning_rate": 0.00013622627670303476,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.744390344619751,
+      "num_tokens": 80426896.0,
+      "step": 34780
+    },
+    {
+      "epoch": 0.3189694691482534,
+      "learning_rate": 0.0001362079398551389,
+      "loss": 0.8722,
+      "mean_token_accuracy": 0.7442276298999786,
+      "num_tokens": 80448749.0,
+      "step": 34790
+    },
+    {
+      "epoch": 0.31906115338773267,
+      "learning_rate": 0.00013618960300724306,
+      "loss": 0.8527,
+      "mean_token_accuracy": 0.7401118993759155,
+      "num_tokens": 80472387.0,
+      "step": 34800
+    },
+    {
+      "epoch": 0.31915283762721186,
+      "learning_rate": 0.0001361712661593472,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7415448069572449,
+      "num_tokens": 80495804.0,
+      "step": 34810
+    },
+    {
+      "epoch": 0.3192445218666911,
+      "learning_rate": 0.00013615292931145137,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.7414491415023804,
+      "num_tokens": 80518247.0,
+      "step": 34820
+    },
+    {
+      "epoch": 0.31933620610617036,
+      "learning_rate": 0.0001361345924635555,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7462976157665253,
+      "num_tokens": 80540939.0,
+      "step": 34830
+    },
+    {
+      "epoch": 0.31942789034564956,
+      "learning_rate": 0.00013611625561565967,
+      "loss": 0.8488,
+      "mean_token_accuracy": 0.7435784935951233,
+      "num_tokens": 80563840.0,
+      "step": 34840
+    },
+    {
+      "epoch": 0.3195195745851288,
+      "learning_rate": 0.00013609791876776384,
+      "loss": 0.8363,
+      "mean_token_accuracy": 0.7504896581172943,
+      "num_tokens": 80586742.0,
+      "step": 34850
+    },
+    {
+      "epoch": 0.31961125882460806,
+      "learning_rate": 0.00013607958191986797,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.7462678611278534,
+      "num_tokens": 80609657.0,
+      "step": 34860
+    },
+    {
+      "epoch": 0.3197029430640873,
+      "learning_rate": 0.00013606124507197214,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7398136258125305,
+      "num_tokens": 80632678.0,
+      "step": 34870
+    },
+    {
+      "epoch": 0.3197946273035665,
+      "learning_rate": 0.00013604290822407628,
+      "loss": 0.8364,
+      "mean_token_accuracy": 0.7386039614677429,
+      "num_tokens": 80655452.0,
+      "step": 34880
+    },
+    {
+      "epoch": 0.31988631154304575,
+      "learning_rate": 0.00013602457137618044,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7428508758544922,
+      "num_tokens": 80678008.0,
+      "step": 34890
+    },
+    {
+      "epoch": 0.319977995782525,
+      "learning_rate": 0.0001360062345282846,
+      "loss": 0.8642,
+      "mean_token_accuracy": 0.7445224344730377,
+      "num_tokens": 80701095.0,
+      "step": 34900
+    },
+    {
+      "epoch": 0.3200696800220042,
+      "learning_rate": 0.00013598789768038875,
+      "loss": 0.8355,
+      "mean_token_accuracy": 0.751128089427948,
+      "num_tokens": 80723753.0,
+      "step": 34910
+    },
+    {
+      "epoch": 0.32016136426148345,
+      "learning_rate": 0.0001359695608324929,
+      "loss": 0.8031,
+      "mean_token_accuracy": 0.760913074016571,
+      "num_tokens": 80746750.0,
+      "step": 34920
+    },
+    {
+      "epoch": 0.3202530485009627,
+      "learning_rate": 0.00013595122398459705,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7404983341693878,
+      "num_tokens": 80770039.0,
+      "step": 34930
+    },
+    {
+      "epoch": 0.3203447327404419,
+      "learning_rate": 0.00013593288713670122,
+      "loss": 0.865,
+      "mean_token_accuracy": 0.7448597013950348,
+      "num_tokens": 80793367.0,
+      "step": 34940
+    },
+    {
+      "epoch": 0.32043641697992115,
+      "learning_rate": 0.00013591455028880535,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7469914615154266,
+      "num_tokens": 80815752.0,
+      "step": 34950
+    },
+    {
+      "epoch": 0.3205281012194004,
+      "learning_rate": 0.0001358962134409095,
+      "loss": 0.8223,
+      "mean_token_accuracy": 0.750765734910965,
+      "num_tokens": 80838887.0,
+      "step": 34960
+    },
+    {
+      "epoch": 0.3206197854588796,
+      "learning_rate": 0.00013587787659301369,
+      "loss": 0.8391,
+      "mean_token_accuracy": 0.7528128325939178,
+      "num_tokens": 80862533.0,
+      "step": 34970
+    },
+    {
+      "epoch": 0.32071146969835884,
+      "learning_rate": 0.00013585953974511782,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7497487127780914,
+      "num_tokens": 80886573.0,
+      "step": 34980
+    },
+    {
+      "epoch": 0.3208031539378381,
+      "learning_rate": 0.00013584120289722196,
+      "loss": 0.8568,
+      "mean_token_accuracy": 0.7458373665809631,
+      "num_tokens": 80909588.0,
+      "step": 34990
+    },
+    {
+      "epoch": 0.32089483817731734,
+      "learning_rate": 0.00013582286604932613,
+      "loss": 0.8015,
+      "mean_token_accuracy": 0.7568274974822998,
+      "num_tokens": 80932614.0,
+      "step": 35000
+    },
+    {
+      "epoch": 0.32098652241679654,
+      "learning_rate": 0.00013580452920143027,
+      "loss": 0.879,
+      "mean_token_accuracy": 0.7383477210998535,
+      "num_tokens": 80955131.0,
+      "step": 35010
+    },
+    {
+      "epoch": 0.3210782066562758,
+      "learning_rate": 0.00013578619235353443,
+      "loss": 0.8513,
+      "mean_token_accuracy": 0.744678795337677,
+      "num_tokens": 80978554.0,
+      "step": 35020
+    },
+    {
+      "epoch": 0.32116989089575504,
+      "learning_rate": 0.0001357678555056386,
+      "loss": 0.7966,
+      "mean_token_accuracy": 0.7619555592536926,
+      "num_tokens": 81001515.0,
+      "step": 35030
+    },
+    {
+      "epoch": 0.32126157513523423,
+      "learning_rate": 0.00013574951865774273,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7381498634815216,
+      "num_tokens": 81023827.0,
+      "step": 35040
+    },
+    {
+      "epoch": 0.3213532593747135,
+      "learning_rate": 0.0001357311818098469,
+      "loss": 0.8705,
+      "mean_token_accuracy": 0.7359326004981994,
+      "num_tokens": 81047193.0,
+      "step": 35050
+    },
+    {
+      "epoch": 0.32144494361419274,
+      "learning_rate": 0.00013571284496195104,
+      "loss": 0.8265,
+      "mean_token_accuracy": 0.7469355463981628,
+      "num_tokens": 81069044.0,
+      "step": 35060
+    },
+    {
+      "epoch": 0.32153662785367193,
+      "learning_rate": 0.0001356945081140552,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.7443061113357544,
+      "num_tokens": 81092424.0,
+      "step": 35070
+    },
+    {
+      "epoch": 0.3216283120931512,
+      "learning_rate": 0.00013567617126615934,
+      "loss": 0.8384,
+      "mean_token_accuracy": 0.7469860255718231,
+      "num_tokens": 81114963.0,
+      "step": 35080
+    },
+    {
+      "epoch": 0.32171999633263043,
+      "learning_rate": 0.0001356578344182635,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7420911133289337,
+      "num_tokens": 81138069.0,
+      "step": 35090
+    },
+    {
+      "epoch": 0.3218116805721097,
+      "learning_rate": 0.00013563949757036767,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7411840558052063,
+      "num_tokens": 81160431.0,
+      "step": 35100
+    },
+    {
+      "epoch": 0.3219033648115889,
+      "learning_rate": 0.0001356211607224718,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7432519495487213,
+      "num_tokens": 81182957.0,
+      "step": 35110
+    },
+    {
+      "epoch": 0.3219950490510681,
+      "learning_rate": 0.00013560282387457598,
+      "loss": 0.8548,
+      "mean_token_accuracy": 0.748400354385376,
+      "num_tokens": 81205478.0,
+      "step": 35120
+    },
+    {
+      "epoch": 0.3220867332905474,
+      "learning_rate": 0.00013558448702668012,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7453448891639709,
+      "num_tokens": 81229441.0,
+      "step": 35130
+    },
+    {
+      "epoch": 0.3221784175300266,
+      "learning_rate": 0.00013556615017878428,
+      "loss": 0.8525,
+      "mean_token_accuracy": 0.7415412187576294,
+      "num_tokens": 81251897.0,
+      "step": 35140
+    },
+    {
+      "epoch": 0.3222701017695058,
+      "learning_rate": 0.00013554781333088842,
+      "loss": 0.8785,
+      "mean_token_accuracy": 0.7355118870735169,
+      "num_tokens": 81274628.0,
+      "step": 35150
+    },
+    {
+      "epoch": 0.3223617860089851,
+      "learning_rate": 0.00013552947648299258,
+      "loss": 0.8675,
+      "mean_token_accuracy": 0.7326271176338196,
+      "num_tokens": 81297257.0,
+      "step": 35160
+    },
+    {
+      "epoch": 0.32245347024846427,
+      "learning_rate": 0.00013551113963509675,
+      "loss": 0.8415,
+      "mean_token_accuracy": 0.7444237768650055,
+      "num_tokens": 81320642.0,
+      "step": 35170
+    },
+    {
+      "epoch": 0.3225451544879435,
+      "learning_rate": 0.0001354928027872009,
+      "loss": 0.8489,
+      "mean_token_accuracy": 0.7437173664569855,
+      "num_tokens": 81343977.0,
+      "step": 35180
+    },
+    {
+      "epoch": 0.32263683872742277,
+      "learning_rate": 0.00013547446593930503,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.7386248350143433,
+      "num_tokens": 81366711.0,
+      "step": 35190
+    },
+    {
+      "epoch": 0.32272852296690197,
+      "learning_rate": 0.0001354561290914092,
+      "loss": 0.857,
+      "mean_token_accuracy": 0.7472717225551605,
+      "num_tokens": 81389605.0,
+      "step": 35200
+    },
+    {
+      "epoch": 0.3228202072063812,
+      "learning_rate": 0.00013543779224351333,
+      "loss": 0.8268,
+      "mean_token_accuracy": 0.7485568761825562,
+      "num_tokens": 81412276.0,
+      "step": 35210
+    },
+    {
+      "epoch": 0.32291189144586047,
+      "learning_rate": 0.0001354194553956175,
+      "loss": 0.8081,
+      "mean_token_accuracy": 0.7577523529529572,
+      "num_tokens": 81435050.0,
+      "step": 35220
+    },
+    {
+      "epoch": 0.3230035756853397,
+      "learning_rate": 0.00013540111854772166,
+      "loss": 0.8157,
+      "mean_token_accuracy": 0.7532997727394104,
+      "num_tokens": 81458130.0,
+      "step": 35230
+    },
+    {
+      "epoch": 0.3230952599248189,
+      "learning_rate": 0.0001353827816998258,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7457041919231415,
+      "num_tokens": 81481495.0,
+      "step": 35240
+    },
+    {
+      "epoch": 0.32318694416429816,
+      "learning_rate": 0.00013536444485192996,
+      "loss": 0.8819,
+      "mean_token_accuracy": 0.7388751924037933,
+      "num_tokens": 81504352.0,
+      "step": 35250
+    },
+    {
+      "epoch": 0.3232786284037774,
+      "learning_rate": 0.0001353461080040341,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.7405800104141236,
+      "num_tokens": 81527779.0,
+      "step": 35260
+    },
+    {
+      "epoch": 0.3233703126432566,
+      "learning_rate": 0.00013532777115613827,
+      "loss": 0.8478,
+      "mean_token_accuracy": 0.7437808692455292,
+      "num_tokens": 81550871.0,
+      "step": 35270
+    },
+    {
+      "epoch": 0.32346199688273586,
+      "learning_rate": 0.0001353094343082424,
+      "loss": 0.8436,
+      "mean_token_accuracy": 0.7423400282859802,
+      "num_tokens": 81574427.0,
+      "step": 35280
+    },
+    {
+      "epoch": 0.3235536811222151,
+      "learning_rate": 0.00013529109746034657,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7487216591835022,
+      "num_tokens": 81596950.0,
+      "step": 35290
+    },
+    {
+      "epoch": 0.3236453653616943,
+      "learning_rate": 0.00013527276061245074,
+      "loss": 0.8621,
+      "mean_token_accuracy": 0.7433628737926483,
+      "num_tokens": 81619335.0,
+      "step": 35300
+    },
+    {
+      "epoch": 0.32373704960117355,
+      "learning_rate": 0.00013525442376455488,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.7557617902755738,
+      "num_tokens": 81642327.0,
+      "step": 35310
+    },
+    {
+      "epoch": 0.3238287338406528,
+      "learning_rate": 0.00013523608691665904,
+      "loss": 0.8274,
+      "mean_token_accuracy": 0.7449274480342865,
+      "num_tokens": 81665850.0,
+      "step": 35320
+    },
+    {
+      "epoch": 0.323920418080132,
+      "learning_rate": 0.00013521775006876318,
+      "loss": 0.8801,
+      "mean_token_accuracy": 0.743822056055069,
+      "num_tokens": 81688793.0,
+      "step": 35330
+    },
+    {
+      "epoch": 0.32401210231961125,
+      "learning_rate": 0.00013519941322086735,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7431145966053009,
+      "num_tokens": 81711704.0,
+      "step": 35340
+    },
+    {
+      "epoch": 0.3241037865590905,
+      "learning_rate": 0.00013518107637297148,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.7453645706176758,
+      "num_tokens": 81734610.0,
+      "step": 35350
+    },
+    {
+      "epoch": 0.32419547079856975,
+      "learning_rate": 0.00013516273952507565,
+      "loss": 0.8195,
+      "mean_token_accuracy": 0.7508151352405548,
+      "num_tokens": 81757684.0,
+      "step": 35360
+    },
+    {
+      "epoch": 0.32428715503804895,
+      "learning_rate": 0.00013514440267717981,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.7492231905460358,
+      "num_tokens": 81781581.0,
+      "step": 35370
+    },
+    {
+      "epoch": 0.3243788392775282,
+      "learning_rate": 0.00013512606582928395,
+      "loss": 0.795,
+      "mean_token_accuracy": 0.7528245329856873,
+      "num_tokens": 81803708.0,
+      "step": 35380
+    },
+    {
+      "epoch": 0.32447052351700745,
+      "learning_rate": 0.0001351077289813881,
+      "loss": 0.8331,
+      "mean_token_accuracy": 0.7523864388465882,
+      "num_tokens": 81826149.0,
+      "step": 35390
+    },
+    {
+      "epoch": 0.32456220775648664,
+      "learning_rate": 0.00013508939213349226,
+      "loss": 0.8414,
+      "mean_token_accuracy": 0.7468234360218048,
+      "num_tokens": 81849109.0,
+      "step": 35400
+    },
+    {
+      "epoch": 0.3246538919959659,
+      "learning_rate": 0.0001350710552855964,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7513863444328308,
+      "num_tokens": 81872404.0,
+      "step": 35410
+    },
+    {
+      "epoch": 0.32474557623544514,
+      "learning_rate": 0.0001350527184377006,
+      "loss": 0.8337,
+      "mean_token_accuracy": 0.7513627886772156,
+      "num_tokens": 81895299.0,
+      "step": 35420
+    },
+    {
+      "epoch": 0.32483726047492434,
+      "learning_rate": 0.00013503438158980473,
+      "loss": 0.8311,
+      "mean_token_accuracy": 0.7515579104423523,
+      "num_tokens": 81918391.0,
+      "step": 35430
+    },
+    {
+      "epoch": 0.3249289447144036,
+      "learning_rate": 0.00013501604474190886,
+      "loss": 0.8473,
+      "mean_token_accuracy": 0.7471840679645538,
+      "num_tokens": 81941487.0,
+      "step": 35440
+    },
+    {
+      "epoch": 0.32502062895388284,
+      "learning_rate": 0.00013499770789401303,
+      "loss": 0.8646,
+      "mean_token_accuracy": 0.7400823950767517,
+      "num_tokens": 81964909.0,
+      "step": 35450
+    },
+    {
+      "epoch": 0.32511231319336203,
+      "learning_rate": 0.00013497937104611717,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7515281498432159,
+      "num_tokens": 81988848.0,
+      "step": 35460
+    },
+    {
+      "epoch": 0.3252039974328413,
+      "learning_rate": 0.00013496103419822133,
+      "loss": 0.8309,
+      "mean_token_accuracy": 0.7513287782669067,
+      "num_tokens": 82012459.0,
+      "step": 35470
+    },
+    {
+      "epoch": 0.32529568167232054,
+      "learning_rate": 0.00013494269735032547,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.740870189666748,
+      "num_tokens": 82036273.0,
+      "step": 35480
+    },
+    {
+      "epoch": 0.3253873659117998,
+      "learning_rate": 0.00013492436050242964,
+      "loss": 0.8206,
+      "mean_token_accuracy": 0.7488671004772186,
+      "num_tokens": 82058919.0,
+      "step": 35490
+    },
+    {
+      "epoch": 0.325479050151279,
+      "learning_rate": 0.0001349060236545338,
+      "loss": 0.8227,
+      "mean_token_accuracy": 0.7483937859535217,
+      "num_tokens": 82081852.0,
+      "step": 35500
+    },
+    {
+      "epoch": 0.32557073439075823,
+      "learning_rate": 0.00013488768680663794,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7361494719982147,
+      "num_tokens": 82104310.0,
+      "step": 35510
+    },
+    {
+      "epoch": 0.3256624186302375,
+      "learning_rate": 0.0001348693499587421,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.7421862840652466,
+      "num_tokens": 82127545.0,
+      "step": 35520
+    },
+    {
+      "epoch": 0.3257541028697167,
+      "learning_rate": 0.00013485101311084624,
+      "loss": 0.8176,
+      "mean_token_accuracy": 0.7536823272705078,
+      "num_tokens": 82150812.0,
+      "step": 35530
+    },
+    {
+      "epoch": 0.32584578710919593,
+      "learning_rate": 0.0001348326762629504,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7387829601764679,
+      "num_tokens": 82174944.0,
+      "step": 35540
+    },
+    {
+      "epoch": 0.3259374713486752,
+      "learning_rate": 0.00013481433941505457,
+      "loss": 0.8474,
+      "mean_token_accuracy": 0.7454278707504273,
+      "num_tokens": 82197846.0,
+      "step": 35550
+    },
+    {
+      "epoch": 0.3260291555881544,
+      "learning_rate": 0.0001347960025671587,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7468889653682709,
+      "num_tokens": 82221244.0,
+      "step": 35560
+    },
+    {
+      "epoch": 0.3261208398276336,
+      "learning_rate": 0.00013477766571926288,
+      "loss": 0.8299,
+      "mean_token_accuracy": 0.7487314581871033,
+      "num_tokens": 82244019.0,
+      "step": 35570
+    },
+    {
+      "epoch": 0.3262125240671129,
+      "learning_rate": 0.00013475932887136702,
+      "loss": 0.9003,
+      "mean_token_accuracy": 0.7275493264198303,
+      "num_tokens": 82267283.0,
+      "step": 35580
+    },
+    {
+      "epoch": 0.32630420830659207,
+      "learning_rate": 0.00013474099202347116,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.736548376083374,
+      "num_tokens": 82290406.0,
+      "step": 35590
+    },
+    {
+      "epoch": 0.3263958925460713,
+      "learning_rate": 0.00013472265517557532,
+      "loss": 0.8571,
+      "mean_token_accuracy": 0.7379210770130158,
+      "num_tokens": 82313558.0,
+      "step": 35600
+    },
+    {
+      "epoch": 0.32648757678555057,
+      "learning_rate": 0.00013470431832767946,
+      "loss": 0.8441,
+      "mean_token_accuracy": 0.7455053985118866,
+      "num_tokens": 82335906.0,
+      "step": 35610
+    },
+    {
+      "epoch": 0.3265792610250298,
+      "learning_rate": 0.00013468598147978365,
+      "loss": 0.8212,
+      "mean_token_accuracy": 0.7589194655418396,
+      "num_tokens": 82358530.0,
+      "step": 35620
+    },
+    {
+      "epoch": 0.326670945264509,
+      "learning_rate": 0.0001346676446318878,
+      "loss": 0.8665,
+      "mean_token_accuracy": 0.7412411212921143,
+      "num_tokens": 82382048.0,
+      "step": 35630
+    },
+    {
+      "epoch": 0.32676262950398827,
+      "learning_rate": 0.00013464930778399193,
+      "loss": 0.838,
+      "mean_token_accuracy": 0.7476840674877167,
+      "num_tokens": 82404846.0,
+      "step": 35640
+    },
+    {
+      "epoch": 0.3268543137434675,
+      "learning_rate": 0.0001346309709360961,
+      "loss": 0.8966,
+      "mean_token_accuracy": 0.7345963656902313,
+      "num_tokens": 82427674.0,
+      "step": 35650
+    },
+    {
+      "epoch": 0.3269459979829467,
+      "learning_rate": 0.00013461263408820023,
+      "loss": 0.8015,
+      "mean_token_accuracy": 0.75493882894516,
+      "num_tokens": 82450146.0,
+      "step": 35660
+    },
+    {
+      "epoch": 0.32703768222242596,
+      "learning_rate": 0.0001345942972403044,
+      "loss": 0.8497,
+      "mean_token_accuracy": 0.7464676558971405,
+      "num_tokens": 82472582.0,
+      "step": 35670
+    },
+    {
+      "epoch": 0.3271293664619052,
+      "learning_rate": 0.00013457596039240856,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7504710197448731,
+      "num_tokens": 82495685.0,
+      "step": 35680
+    },
+    {
+      "epoch": 0.3272210507013844,
+      "learning_rate": 0.0001345576235445127,
+      "loss": 0.8464,
+      "mean_token_accuracy": 0.7482772767543793,
+      "num_tokens": 82518711.0,
+      "step": 35690
+    },
+    {
+      "epoch": 0.32731273494086366,
+      "learning_rate": 0.00013453928669661687,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7443857192993164,
+      "num_tokens": 82541345.0,
+      "step": 35700
+    },
+    {
+      "epoch": 0.3274044191803429,
+      "learning_rate": 0.000134520949848721,
+      "loss": 0.8553,
+      "mean_token_accuracy": 0.7438439548015594,
+      "num_tokens": 82563974.0,
+      "step": 35710
+    },
+    {
+      "epoch": 0.32749610341982216,
+      "learning_rate": 0.00013450261300082517,
+      "loss": 0.8499,
+      "mean_token_accuracy": 0.7503586053848267,
+      "num_tokens": 82587554.0,
+      "step": 35720
+    },
+    {
+      "epoch": 0.32758778765930135,
+      "learning_rate": 0.0001344842761529293,
+      "loss": 0.8833,
+      "mean_token_accuracy": 0.739331042766571,
+      "num_tokens": 82610042.0,
+      "step": 35730
+    },
+    {
+      "epoch": 0.3276794718987806,
+      "learning_rate": 0.00013446593930503347,
+      "loss": 0.8032,
+      "mean_token_accuracy": 0.7542295634746552,
+      "num_tokens": 82633565.0,
+      "step": 35740
+    },
+    {
+      "epoch": 0.32777115613825986,
+      "learning_rate": 0.00013444760245713764,
+      "loss": 0.878,
+      "mean_token_accuracy": 0.7337488234043121,
+      "num_tokens": 82656625.0,
+      "step": 35750
+    },
+    {
+      "epoch": 0.32786284037773905,
+      "learning_rate": 0.00013442926560924178,
+      "loss": 0.8909,
+      "mean_token_accuracy": 0.7295827150344849,
+      "num_tokens": 82679780.0,
+      "step": 35760
+    },
+    {
+      "epoch": 0.3279545246172183,
+      "learning_rate": 0.00013441092876134594,
+      "loss": 0.8363,
+      "mean_token_accuracy": 0.7473319292068481,
+      "num_tokens": 82704065.0,
+      "step": 35770
+    },
+    {
+      "epoch": 0.32804620885669755,
+      "learning_rate": 0.00013439259191345008,
+      "loss": 0.8143,
+      "mean_token_accuracy": 0.7503743290901184,
+      "num_tokens": 82727035.0,
+      "step": 35780
+    },
+    {
+      "epoch": 0.32813789309617675,
+      "learning_rate": 0.00013437425506555422,
+      "loss": 0.8387,
+      "mean_token_accuracy": 0.755956894159317,
+      "num_tokens": 82750698.0,
+      "step": 35790
+    },
+    {
+      "epoch": 0.328229577335656,
+      "learning_rate": 0.00013435591821765839,
+      "loss": 0.8515,
+      "mean_token_accuracy": 0.7519905805587769,
+      "num_tokens": 82773181.0,
+      "step": 35800
+    },
+    {
+      "epoch": 0.32832126157513525,
+      "learning_rate": 0.00013433758136976255,
+      "loss": 0.8463,
+      "mean_token_accuracy": 0.7504700422286987,
+      "num_tokens": 82796677.0,
+      "step": 35810
+    },
+    {
+      "epoch": 0.32841294581461444,
+      "learning_rate": 0.00013431924452186672,
+      "loss": 0.8313,
+      "mean_token_accuracy": 0.7485404849052429,
+      "num_tokens": 82820405.0,
+      "step": 35820
+    },
+    {
+      "epoch": 0.3285046300540937,
+      "learning_rate": 0.00013430090767397085,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7402553975582122,
+      "num_tokens": 82844118.0,
+      "step": 35830
+    },
+    {
+      "epoch": 0.32859631429357294,
+      "learning_rate": 0.000134282570826075,
+      "loss": 0.8349,
+      "mean_token_accuracy": 0.7491594135761261,
+      "num_tokens": 82867156.0,
+      "step": 35840
+    },
+    {
+      "epoch": 0.3286879985330522,
+      "learning_rate": 0.00013426423397817916,
+      "loss": 0.8042,
+      "mean_token_accuracy": 0.7536715269088745,
+      "num_tokens": 82889919.0,
+      "step": 35850
+    },
+    {
+      "epoch": 0.3287796827725314,
+      "learning_rate": 0.0001342458971302833,
+      "loss": 0.8796,
+      "mean_token_accuracy": 0.7440406084060669,
+      "num_tokens": 82913503.0,
+      "step": 35860
+    },
+    {
+      "epoch": 0.32887136701201064,
+      "learning_rate": 0.00013422756028238746,
+      "loss": 0.8564,
+      "mean_token_accuracy": 0.7469022214412689,
+      "num_tokens": 82936700.0,
+      "step": 35870
+    },
+    {
+      "epoch": 0.3289630512514899,
+      "learning_rate": 0.00013420922343449163,
+      "loss": 0.8166,
+      "mean_token_accuracy": 0.7591893672943115,
+      "num_tokens": 82959989.0,
+      "step": 35880
+    },
+    {
+      "epoch": 0.3290547354909691,
+      "learning_rate": 0.00013419088658659577,
+      "loss": 0.8533,
+      "mean_token_accuracy": 0.7430768609046936,
+      "num_tokens": 82982688.0,
+      "step": 35890
+    },
+    {
+      "epoch": 0.32914641973044834,
+      "learning_rate": 0.00013417254973869993,
+      "loss": 0.8211,
+      "mean_token_accuracy": 0.7535585284233093,
+      "num_tokens": 83004976.0,
+      "step": 35900
+    },
+    {
+      "epoch": 0.3292381039699276,
+      "learning_rate": 0.00013415421289080407,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7439225792884827,
+      "num_tokens": 83028886.0,
+      "step": 35910
+    },
+    {
+      "epoch": 0.3293297882094068,
+      "learning_rate": 0.00013413587604290823,
+      "loss": 0.8806,
+      "mean_token_accuracy": 0.7404143691062928,
+      "num_tokens": 83051375.0,
+      "step": 35920
+    },
+    {
+      "epoch": 0.32942147244888603,
+      "learning_rate": 0.00013411753919501237,
+      "loss": 0.8078,
+      "mean_token_accuracy": 0.7531822741031646,
+      "num_tokens": 83075443.0,
+      "step": 35930
+    },
+    {
+      "epoch": 0.3295131566883653,
+      "learning_rate": 0.00013409920234711654,
+      "loss": 0.8464,
+      "mean_token_accuracy": 0.7476143658161163,
+      "num_tokens": 83097861.0,
+      "step": 35940
+    },
+    {
+      "epoch": 0.3296048409278445,
+      "learning_rate": 0.0001340808654992207,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7435589492321014,
+      "num_tokens": 83121036.0,
+      "step": 35950
+    },
+    {
+      "epoch": 0.32969652516732373,
+      "learning_rate": 0.00013406252865132484,
+      "loss": 0.8407,
+      "mean_token_accuracy": 0.7459194183349609,
+      "num_tokens": 83144332.0,
+      "step": 35960
+    },
+    {
+      "epoch": 0.329788209406803,
+      "learning_rate": 0.000134044191803429,
+      "loss": 0.8256,
+      "mean_token_accuracy": 0.752760899066925,
+      "num_tokens": 83168003.0,
+      "step": 35970
+    },
+    {
+      "epoch": 0.32987989364628223,
+      "learning_rate": 0.00013402585495553315,
+      "loss": 0.8251,
+      "mean_token_accuracy": 0.7474977254867554,
+      "num_tokens": 83191122.0,
+      "step": 35980
+    },
+    {
+      "epoch": 0.3299715778857614,
+      "learning_rate": 0.00013400751810763728,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.7548482060432434,
+      "num_tokens": 83214237.0,
+      "step": 35990
+    },
+    {
+      "epoch": 0.3300632621252407,
+      "learning_rate": 0.00013398918125974145,
+      "loss": 0.8481,
+      "mean_token_accuracy": 0.7422476232051849,
+      "num_tokens": 83237553.0,
+      "step": 36000
+    },
+    {
+      "epoch": 0.3301549463647199,
+      "learning_rate": 0.00013397084441184562,
+      "loss": 0.86,
+      "mean_token_accuracy": 0.7420591831207275,
+      "num_tokens": 83261411.0,
+      "step": 36010
+    },
+    {
+      "epoch": 0.3302466306041991,
+      "learning_rate": 0.00013395250756394978,
+      "loss": 0.8796,
+      "mean_token_accuracy": 0.7426454842090606,
+      "num_tokens": 83284089.0,
+      "step": 36020
+    },
+    {
+      "epoch": 0.33033831484367837,
+      "learning_rate": 0.00013393417071605392,
+      "loss": 0.8342,
+      "mean_token_accuracy": 0.7506460428237915,
+      "num_tokens": 83306866.0,
+      "step": 36030
+    },
+    {
+      "epoch": 0.3304299990831576,
+      "learning_rate": 0.00013391583386815806,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.733001071214676,
+      "num_tokens": 83329752.0,
+      "step": 36040
+    },
+    {
+      "epoch": 0.3305216833226368,
+      "learning_rate": 0.00013389749702026222,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.748653793334961,
+      "num_tokens": 83353323.0,
+      "step": 36050
+    },
+    {
+      "epoch": 0.33061336756211607,
+      "learning_rate": 0.00013387916017236636,
+      "loss": 0.8249,
+      "mean_token_accuracy": 0.7484886705875397,
+      "num_tokens": 83376495.0,
+      "step": 36060
+    },
+    {
+      "epoch": 0.3307050518015953,
+      "learning_rate": 0.00013386082332447053,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7476909399032593,
+      "num_tokens": 83399869.0,
+      "step": 36070
+    },
+    {
+      "epoch": 0.3307967360410745,
+      "learning_rate": 0.0001338424864765747,
+      "loss": 0.8113,
+      "mean_token_accuracy": 0.7545444250106812,
+      "num_tokens": 83422697.0,
+      "step": 36080
+    },
+    {
+      "epoch": 0.33088842028055376,
+      "learning_rate": 0.00013382414962867883,
+      "loss": 0.8669,
+      "mean_token_accuracy": 0.7394547402858734,
+      "num_tokens": 83445919.0,
+      "step": 36090
+    },
+    {
+      "epoch": 0.330980104520033,
+      "learning_rate": 0.000133805812780783,
+      "loss": 0.8522,
+      "mean_token_accuracy": 0.7406936287879944,
+      "num_tokens": 83469760.0,
+      "step": 36100
+    },
+    {
+      "epoch": 0.33107178875951226,
+      "learning_rate": 0.00013378747593288713,
+      "loss": 0.8207,
+      "mean_token_accuracy": 0.7470863163471222,
+      "num_tokens": 83491764.0,
+      "step": 36110
+    },
+    {
+      "epoch": 0.33116347299899146,
+      "learning_rate": 0.0001337691390849913,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7513814866542816,
+      "num_tokens": 83514568.0,
+      "step": 36120
+    },
+    {
+      "epoch": 0.3312551572384707,
+      "learning_rate": 0.00013375080223709544,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.7450028181076049,
+      "num_tokens": 83536772.0,
+      "step": 36130
+    },
+    {
+      "epoch": 0.33134684147794996,
+      "learning_rate": 0.0001337324653891996,
+      "loss": 0.8695,
+      "mean_token_accuracy": 0.7432056784629821,
+      "num_tokens": 83559175.0,
+      "step": 36140
+    },
+    {
+      "epoch": 0.33143852571742916,
+      "learning_rate": 0.00013371412854130377,
+      "loss": 0.8098,
+      "mean_token_accuracy": 0.7504313349723816,
+      "num_tokens": 83581727.0,
+      "step": 36150
+    },
+    {
+      "epoch": 0.3315302099569084,
+      "learning_rate": 0.0001336957916934079,
+      "loss": 0.8428,
+      "mean_token_accuracy": 0.7397641062736511,
+      "num_tokens": 83605865.0,
+      "step": 36160
+    },
+    {
+      "epoch": 0.33162189419638766,
+      "learning_rate": 0.00013367745484551207,
+      "loss": 0.8529,
+      "mean_token_accuracy": 0.7477608442306518,
+      "num_tokens": 83629565.0,
+      "step": 36170
+    },
+    {
+      "epoch": 0.33171357843586685,
+      "learning_rate": 0.0001336591179976162,
+      "loss": 0.8611,
+      "mean_token_accuracy": 0.7459795713424683,
+      "num_tokens": 83651876.0,
+      "step": 36180
+    },
+    {
+      "epoch": 0.3318052626753461,
+      "learning_rate": 0.00013364078114972035,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7443089842796325,
+      "num_tokens": 83674656.0,
+      "step": 36190
+    },
+    {
+      "epoch": 0.33189694691482535,
+      "learning_rate": 0.00013362244430182451,
+      "loss": 0.8757,
+      "mean_token_accuracy": 0.7393298149108887,
+      "num_tokens": 83697678.0,
+      "step": 36200
+    },
+    {
+      "epoch": 0.33198863115430455,
+      "learning_rate": 0.00013360410745392868,
+      "loss": 0.8946,
+      "mean_token_accuracy": 0.7454830348491669,
+      "num_tokens": 83720947.0,
+      "step": 36210
+    },
+    {
+      "epoch": 0.3320803153937838,
+      "learning_rate": 0.00013358577060603284,
+      "loss": 0.8145,
+      "mean_token_accuracy": 0.7509842216968536,
+      "num_tokens": 83743924.0,
+      "step": 36220
+    },
+    {
+      "epoch": 0.33217199963326305,
+      "learning_rate": 0.00013356743375813698,
+      "loss": 0.8559,
+      "mean_token_accuracy": 0.7424210250377655,
+      "num_tokens": 83767174.0,
+      "step": 36230
+    },
+    {
+      "epoch": 0.3322636838727423,
+      "learning_rate": 0.00013354909691024112,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7500545799732208,
+      "num_tokens": 83789080.0,
+      "step": 36240
+    },
+    {
+      "epoch": 0.3323553681122215,
+      "learning_rate": 0.0001335307600623453,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7432078897953034,
+      "num_tokens": 83812054.0,
+      "step": 36250
+    },
+    {
+      "epoch": 0.33244705235170074,
+      "learning_rate": 0.00013351242321444943,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7478015542030334,
+      "num_tokens": 83834880.0,
+      "step": 36260
+    },
+    {
+      "epoch": 0.33253873659118,
+      "learning_rate": 0.00013349408636655362,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7472139954566955,
+      "num_tokens": 83858308.0,
+      "step": 36270
+    },
+    {
+      "epoch": 0.3326304208306592,
+      "learning_rate": 0.00013347574951865776,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7430963695049286,
+      "num_tokens": 83881817.0,
+      "step": 36280
+    },
+    {
+      "epoch": 0.33272210507013844,
+      "learning_rate": 0.0001334574126707619,
+      "loss": 0.8463,
+      "mean_token_accuracy": 0.7436040818691254,
+      "num_tokens": 83905127.0,
+      "step": 36290
+    },
+    {
+      "epoch": 0.3328137893096177,
+      "learning_rate": 0.00013343907582286606,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7438998699188233,
+      "num_tokens": 83927600.0,
+      "step": 36300
+    },
+    {
+      "epoch": 0.3329054735490969,
+      "learning_rate": 0.0001334207389749702,
+      "loss": 0.8501,
+      "mean_token_accuracy": 0.7480362892150879,
+      "num_tokens": 83950550.0,
+      "step": 36310
+    },
+    {
+      "epoch": 0.33299715778857614,
+      "learning_rate": 0.00013340240212707436,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.752432930469513,
+      "num_tokens": 83973474.0,
+      "step": 36320
+    },
+    {
+      "epoch": 0.3330888420280554,
+      "learning_rate": 0.0001333840652791785,
+      "loss": 0.8488,
+      "mean_token_accuracy": 0.7455214023590088,
+      "num_tokens": 83996424.0,
+      "step": 36330
+    },
+    {
+      "epoch": 0.33318052626753464,
+      "learning_rate": 0.00013336572843128267,
+      "loss": 0.859,
+      "mean_token_accuracy": 0.7452872037887573,
+      "num_tokens": 84019412.0,
+      "step": 36340
+    },
+    {
+      "epoch": 0.33327221050701383,
+      "learning_rate": 0.00013334739158338683,
+      "loss": 0.8098,
+      "mean_token_accuracy": 0.756863534450531,
+      "num_tokens": 84042655.0,
+      "step": 36350
+    },
+    {
+      "epoch": 0.3333638947464931,
+      "learning_rate": 0.00013332905473549097,
+      "loss": 0.8562,
+      "mean_token_accuracy": 0.7415922701358795,
+      "num_tokens": 84066406.0,
+      "step": 36360
+    },
+    {
+      "epoch": 0.33345557898597233,
+      "learning_rate": 0.00013331071788759514,
+      "loss": 0.8106,
+      "mean_token_accuracy": 0.7486303687095642,
+      "num_tokens": 84090247.0,
+      "step": 36370
+    },
+    {
+      "epoch": 0.33354726322545153,
+      "learning_rate": 0.00013329238103969927,
+      "loss": 0.8572,
+      "mean_token_accuracy": 0.740035605430603,
+      "num_tokens": 84113829.0,
+      "step": 36380
+    },
+    {
+      "epoch": 0.3336389474649308,
+      "learning_rate": 0.0001332740441918034,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7442825198173523,
+      "num_tokens": 84137550.0,
+      "step": 36390
+    },
+    {
+      "epoch": 0.33373063170441003,
+      "learning_rate": 0.0001332557073439076,
+      "loss": 0.872,
+      "mean_token_accuracy": 0.7430775403976441,
+      "num_tokens": 84159965.0,
+      "step": 36400
+    },
+    {
+      "epoch": 0.3338223159438892,
+      "learning_rate": 0.00013323737049601174,
+      "loss": 0.8291,
+      "mean_token_accuracy": 0.7472260773181916,
+      "num_tokens": 84183451.0,
+      "step": 36410
+    },
+    {
+      "epoch": 0.3339140001833685,
+      "learning_rate": 0.0001332190336481159,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.7508554220199585,
+      "num_tokens": 84206829.0,
+      "step": 36420
+    },
+    {
+      "epoch": 0.3340056844228477,
+      "learning_rate": 0.00013320069680022005,
+      "loss": 0.852,
+      "mean_token_accuracy": 0.7439524888992309,
+      "num_tokens": 84229964.0,
+      "step": 36430
+    },
+    {
+      "epoch": 0.3340973686623269,
+      "learning_rate": 0.00013318235995232419,
+      "loss": 0.8965,
+      "mean_token_accuracy": 0.7367576241493226,
+      "num_tokens": 84252935.0,
+      "step": 36440
+    },
+    {
+      "epoch": 0.33418905290180617,
+      "learning_rate": 0.00013316402310442835,
+      "loss": 0.8265,
+      "mean_token_accuracy": 0.744022423028946,
+      "num_tokens": 84276312.0,
+      "step": 36450
+    },
+    {
+      "epoch": 0.3342807371412854,
+      "learning_rate": 0.0001331456862565325,
+      "loss": 0.8534,
+      "mean_token_accuracy": 0.7482629716396332,
+      "num_tokens": 84299175.0,
+      "step": 36460
+    },
+    {
+      "epoch": 0.3343724213807647,
+      "learning_rate": 0.00013312734940863668,
+      "loss": 0.8202,
+      "mean_token_accuracy": 0.7586467385292053,
+      "num_tokens": 84322218.0,
+      "step": 36470
+    },
+    {
+      "epoch": 0.33446410562024387,
+      "learning_rate": 0.00013310901256074082,
+      "loss": 0.8985,
+      "mean_token_accuracy": 0.7379003882408142,
+      "num_tokens": 84344952.0,
+      "step": 36480
+    },
+    {
+      "epoch": 0.3345557898597231,
+      "learning_rate": 0.00013309067571284496,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7478548049926758,
+      "num_tokens": 84368107.0,
+      "step": 36490
+    },
+    {
+      "epoch": 0.33464747409920237,
+      "learning_rate": 0.00013307233886494912,
+      "loss": 0.8743,
+      "mean_token_accuracy": 0.7421445190906525,
+      "num_tokens": 84391482.0,
+      "step": 36500
+    },
+    {
+      "epoch": 0.33473915833868156,
+      "learning_rate": 0.00013305400201705326,
+      "loss": 0.8836,
+      "mean_token_accuracy": 0.7406928241252899,
+      "num_tokens": 84414405.0,
+      "step": 36510
+    },
+    {
+      "epoch": 0.3348308425781608,
+      "learning_rate": 0.00013303566516915743,
+      "loss": 0.8516,
+      "mean_token_accuracy": 0.7462628364562989,
+      "num_tokens": 84437923.0,
+      "step": 36520
+    },
+    {
+      "epoch": 0.33492252681764006,
+      "learning_rate": 0.0001330173283212616,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7484513282775879,
+      "num_tokens": 84460627.0,
+      "step": 36530
+    },
+    {
+      "epoch": 0.33501421105711926,
+      "learning_rate": 0.00013299899147336573,
+      "loss": 0.8863,
+      "mean_token_accuracy": 0.7424584090709686,
+      "num_tokens": 84483322.0,
+      "step": 36540
+    },
+    {
+      "epoch": 0.3351058952965985,
+      "learning_rate": 0.0001329806546254699,
+      "loss": 0.8841,
+      "mean_token_accuracy": 0.7359366238117218,
+      "num_tokens": 84505527.0,
+      "step": 36550
+    },
+    {
+      "epoch": 0.33519757953607776,
+      "learning_rate": 0.00013296231777757404,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7437104701995849,
+      "num_tokens": 84529547.0,
+      "step": 36560
+    },
+    {
+      "epoch": 0.33528926377555696,
+      "learning_rate": 0.0001329439809296782,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7423426568508148,
+      "num_tokens": 84552247.0,
+      "step": 36570
+    },
+    {
+      "epoch": 0.3353809480150362,
+      "learning_rate": 0.00013292564408178234,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.738763564825058,
+      "num_tokens": 84574062.0,
+      "step": 36580
+    },
+    {
+      "epoch": 0.33547263225451546,
+      "learning_rate": 0.00013290730723388648,
+      "loss": 0.8632,
+      "mean_token_accuracy": 0.7434098660945893,
+      "num_tokens": 84597243.0,
+      "step": 36590
+    },
+    {
+      "epoch": 0.3355643164939947,
+      "learning_rate": 0.00013288897038599067,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7464635312557221,
+      "num_tokens": 84619817.0,
+      "step": 36600
+    },
+    {
+      "epoch": 0.3356560007334739,
+      "learning_rate": 0.0001328706335380948,
+      "loss": 0.8464,
+      "mean_token_accuracy": 0.7427412450313569,
+      "num_tokens": 84642830.0,
+      "step": 36610
+    },
+    {
+      "epoch": 0.33574768497295315,
+      "learning_rate": 0.00013285229669019897,
+      "loss": 0.8609,
+      "mean_token_accuracy": 0.747695368528366,
+      "num_tokens": 84665975.0,
+      "step": 36620
+    },
+    {
+      "epoch": 0.3358393692124324,
+      "learning_rate": 0.0001328339598423031,
+      "loss": 0.8137,
+      "mean_token_accuracy": 0.750172358751297,
+      "num_tokens": 84688862.0,
+      "step": 36630
+    },
+    {
+      "epoch": 0.3359310534519116,
+      "learning_rate": 0.00013281562299440725,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7464560210704804,
+      "num_tokens": 84711845.0,
+      "step": 36640
+    },
+    {
+      "epoch": 0.33602273769139085,
+      "learning_rate": 0.00013279728614651142,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.7443154096603394,
+      "num_tokens": 84735577.0,
+      "step": 36650
+    },
+    {
+      "epoch": 0.3361144219308701,
+      "learning_rate": 0.00013277894929861558,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.7442899823188782,
+      "num_tokens": 84758130.0,
+      "step": 36660
+    },
+    {
+      "epoch": 0.3362061061703493,
+      "learning_rate": 0.00013276061245071975,
+      "loss": 0.863,
+      "mean_token_accuracy": 0.7436336934566498,
+      "num_tokens": 84781808.0,
+      "step": 36670
+    },
+    {
+      "epoch": 0.33629779040982855,
+      "learning_rate": 0.00013274227560282389,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7438279747962951,
+      "num_tokens": 84804876.0,
+      "step": 36680
+    },
+    {
+      "epoch": 0.3363894746493078,
+      "learning_rate": 0.00013272393875492802,
+      "loss": 0.8524,
+      "mean_token_accuracy": 0.7418655335903168,
+      "num_tokens": 84828253.0,
+      "step": 36690
+    },
+    {
+      "epoch": 0.336481158888787,
+      "learning_rate": 0.0001327056019070322,
+      "loss": 0.8481,
+      "mean_token_accuracy": 0.7425264775753021,
+      "num_tokens": 84852049.0,
+      "step": 36700
+    },
+    {
+      "epoch": 0.33657284312826624,
+      "learning_rate": 0.00013268726505913633,
+      "loss": 0.8949,
+      "mean_token_accuracy": 0.7352565705776215,
+      "num_tokens": 84874657.0,
+      "step": 36710
+    },
+    {
+      "epoch": 0.3366645273677455,
+      "learning_rate": 0.0001326689282112405,
+      "loss": 0.8674,
+      "mean_token_accuracy": 0.7446515262126923,
+      "num_tokens": 84897173.0,
+      "step": 36720
+    },
+    {
+      "epoch": 0.33675621160722474,
+      "learning_rate": 0.00013265059136334466,
+      "loss": 0.8592,
+      "mean_token_accuracy": 0.7412495791912079,
+      "num_tokens": 84920633.0,
+      "step": 36730
+    },
+    {
+      "epoch": 0.33684789584670394,
+      "learning_rate": 0.0001326322545154488,
+      "loss": 0.8179,
+      "mean_token_accuracy": 0.7505267918109894,
+      "num_tokens": 84943480.0,
+      "step": 36740
+    },
+    {
+      "epoch": 0.3369395800861832,
+      "learning_rate": 0.00013261391766755296,
+      "loss": 0.8415,
+      "mean_token_accuracy": 0.7481511950492858,
+      "num_tokens": 84966045.0,
+      "step": 36750
+    },
+    {
+      "epoch": 0.33703126432566244,
+      "learning_rate": 0.0001325955808196571,
+      "loss": 0.8725,
+      "mean_token_accuracy": 0.7443264424800873,
+      "num_tokens": 84988586.0,
+      "step": 36760
+    },
+    {
+      "epoch": 0.33712294856514163,
+      "learning_rate": 0.00013257724397176127,
+      "loss": 0.8471,
+      "mean_token_accuracy": 0.747825825214386,
+      "num_tokens": 85011896.0,
+      "step": 36770
+    },
+    {
+      "epoch": 0.3372146328046209,
+      "learning_rate": 0.0001325589071238654,
+      "loss": 0.8326,
+      "mean_token_accuracy": 0.7540049493312836,
+      "num_tokens": 85035434.0,
+      "step": 36780
+    },
+    {
+      "epoch": 0.33730631704410013,
+      "learning_rate": 0.00013254057027596957,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7456261396408081,
+      "num_tokens": 85058819.0,
+      "step": 36790
+    },
+    {
+      "epoch": 0.33739800128357933,
+      "learning_rate": 0.00013252223342807373,
+      "loss": 0.8358,
+      "mean_token_accuracy": 0.751392924785614,
+      "num_tokens": 85082363.0,
+      "step": 36800
+    },
+    {
+      "epoch": 0.3374896855230586,
+      "learning_rate": 0.00013250389658017787,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7476738095283508,
+      "num_tokens": 85104936.0,
+      "step": 36810
+    },
+    {
+      "epoch": 0.33758136976253783,
+      "learning_rate": 0.00013248555973228204,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7451462864875793,
+      "num_tokens": 85128653.0,
+      "step": 36820
+    },
+    {
+      "epoch": 0.337673054002017,
+      "learning_rate": 0.00013246722288438618,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7442201793193817,
+      "num_tokens": 85151880.0,
+      "step": 36830
+    },
+    {
+      "epoch": 0.3377647382414963,
+      "learning_rate": 0.00013244888603649031,
+      "loss": 0.8735,
+      "mean_token_accuracy": 0.7356188356876373,
+      "num_tokens": 85174037.0,
+      "step": 36840
+    },
+    {
+      "epoch": 0.3378564224809755,
+      "learning_rate": 0.00013243054918859448,
+      "loss": 0.8309,
+      "mean_token_accuracy": 0.75159912109375,
+      "num_tokens": 85198024.0,
+      "step": 36850
+    },
+    {
+      "epoch": 0.3379481067204548,
+      "learning_rate": 0.00013241221234069865,
+      "loss": 0.8691,
+      "mean_token_accuracy": 0.7484756886959076,
+      "num_tokens": 85220739.0,
+      "step": 36860
+    },
+    {
+      "epoch": 0.338039790959934,
+      "learning_rate": 0.0001323938754928028,
+      "loss": 0.8399,
+      "mean_token_accuracy": 0.7407955288887024,
+      "num_tokens": 85243291.0,
+      "step": 36870
+    },
+    {
+      "epoch": 0.3381314751994132,
+      "learning_rate": 0.00013237553864490695,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.7445390164852143,
+      "num_tokens": 85266951.0,
+      "step": 36880
+    },
+    {
+      "epoch": 0.3382231594388925,
+      "learning_rate": 0.0001323572017970111,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.7566348731517791,
+      "num_tokens": 85290143.0,
+      "step": 36890
+    },
+    {
+      "epoch": 0.33831484367837167,
+      "learning_rate": 0.00013233886494911525,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7452965676784515,
+      "num_tokens": 85312440.0,
+      "step": 36900
+    },
+    {
+      "epoch": 0.3384065279178509,
+      "learning_rate": 0.0001323205281012194,
+      "loss": 0.8679,
+      "mean_token_accuracy": 0.7389244675636292,
+      "num_tokens": 85335363.0,
+      "step": 36910
+    },
+    {
+      "epoch": 0.33849821215733017,
+      "learning_rate": 0.00013230219125332358,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7493228614330292,
+      "num_tokens": 85358520.0,
+      "step": 36920
+    },
+    {
+      "epoch": 0.33858989639680936,
+      "learning_rate": 0.00013228385440542772,
+      "loss": 0.8919,
+      "mean_token_accuracy": 0.7353790402412415,
+      "num_tokens": 85381886.0,
+      "step": 36930
+    },
+    {
+      "epoch": 0.3386815806362886,
+      "learning_rate": 0.00013226551755753186,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7530224323272705,
+      "num_tokens": 85405009.0,
+      "step": 36940
+    },
+    {
+      "epoch": 0.33877326487576787,
+      "learning_rate": 0.00013224718070963603,
+      "loss": 0.8832,
+      "mean_token_accuracy": 0.7367229163646698,
+      "num_tokens": 85428120.0,
+      "step": 36950
+    },
+    {
+      "epoch": 0.3388649491152471,
+      "learning_rate": 0.00013222884386174016,
+      "loss": 0.8951,
+      "mean_token_accuracy": 0.7369018435478211,
+      "num_tokens": 85451075.0,
+      "step": 36960
+    },
+    {
+      "epoch": 0.3389566333547263,
+      "learning_rate": 0.00013221050701384433,
+      "loss": 0.8685,
+      "mean_token_accuracy": 0.7381076693534852,
+      "num_tokens": 85474371.0,
+      "step": 36970
+    },
+    {
+      "epoch": 0.33904831759420556,
+      "learning_rate": 0.00013219217016594847,
+      "loss": 0.8571,
+      "mean_token_accuracy": 0.7486586391925811,
+      "num_tokens": 85497716.0,
+      "step": 36980
+    },
+    {
+      "epoch": 0.3391400018336848,
+      "learning_rate": 0.00013217383331805263,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7552816092967987,
+      "num_tokens": 85520387.0,
+      "step": 36990
+    },
+    {
+      "epoch": 0.339231686073164,
+      "learning_rate": 0.0001321554964701568,
+      "loss": 0.8121,
+      "mean_token_accuracy": 0.7532137870788574,
+      "num_tokens": 85543418.0,
+      "step": 37000
+    },
+    {
+      "epoch": 0.33932337031264326,
+      "learning_rate": 0.00013213715962226094,
+      "loss": 0.8275,
+      "mean_token_accuracy": 0.7536680340766907,
+      "num_tokens": 85567146.0,
+      "step": 37010
+    },
+    {
+      "epoch": 0.3394150545521225,
+      "learning_rate": 0.0001321188227743651,
+      "loss": 0.8708,
+      "mean_token_accuracy": 0.740613979101181,
+      "num_tokens": 85590525.0,
+      "step": 37020
+    },
+    {
+      "epoch": 0.3395067387916017,
+      "learning_rate": 0.00013210048592646924,
+      "loss": 0.8343,
+      "mean_token_accuracy": 0.7506062805652618,
+      "num_tokens": 85612864.0,
+      "step": 37030
+    },
+    {
+      "epoch": 0.33959842303108095,
+      "learning_rate": 0.00013208214907857338,
+      "loss": 0.8686,
+      "mean_token_accuracy": 0.7389248132705688,
+      "num_tokens": 85636267.0,
+      "step": 37040
+    },
+    {
+      "epoch": 0.3396901072705602,
+      "learning_rate": 0.00013206381223067757,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.7465624272823334,
+      "num_tokens": 85659474.0,
+      "step": 37050
+    },
+    {
+      "epoch": 0.3397817915100394,
+      "learning_rate": 0.0001320454753827817,
+      "loss": 0.8035,
+      "mean_token_accuracy": 0.757887351512909,
+      "num_tokens": 85683234.0,
+      "step": 37060
+    },
+    {
+      "epoch": 0.33987347574951865,
+      "learning_rate": 0.00013202713853488588,
+      "loss": 0.8781,
+      "mean_token_accuracy": 0.7341601371765136,
+      "num_tokens": 85707234.0,
+      "step": 37070
+    },
+    {
+      "epoch": 0.3399651599889979,
+      "learning_rate": 0.00013200880168699001,
+      "loss": 0.8493,
+      "mean_token_accuracy": 0.744634622335434,
+      "num_tokens": 85729868.0,
+      "step": 37080
+    },
+    {
+      "epoch": 0.34005684422847715,
+      "learning_rate": 0.00013199046483909415,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.7524881303310395,
+      "num_tokens": 85752871.0,
+      "step": 37090
+    },
+    {
+      "epoch": 0.34014852846795635,
+      "learning_rate": 0.00013197212799119832,
+      "loss": 0.8437,
+      "mean_token_accuracy": 0.7475317716598511,
+      "num_tokens": 85776016.0,
+      "step": 37100
+    },
+    {
+      "epoch": 0.3402402127074356,
+      "learning_rate": 0.00013195379114330246,
+      "loss": 0.836,
+      "mean_token_accuracy": 0.7491052448749542,
+      "num_tokens": 85799713.0,
+      "step": 37110
+    },
+    {
+      "epoch": 0.34033189694691485,
+      "learning_rate": 0.00013193545429540665,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7442010939121246,
+      "num_tokens": 85822017.0,
+      "step": 37120
+    },
+    {
+      "epoch": 0.34042358118639404,
+      "learning_rate": 0.0001319171174475108,
+      "loss": 0.8389,
+      "mean_token_accuracy": 0.7481133580207825,
+      "num_tokens": 85845360.0,
+      "step": 37130
+    },
+    {
+      "epoch": 0.3405152654258733,
+      "learning_rate": 0.00013189878059961493,
+      "loss": 0.8417,
+      "mean_token_accuracy": 0.7507116019725799,
+      "num_tokens": 85868020.0,
+      "step": 37140
+    },
+    {
+      "epoch": 0.34060694966535254,
+      "learning_rate": 0.0001318804437517191,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7394050121307373,
+      "num_tokens": 85891078.0,
+      "step": 37150
+    },
+    {
+      "epoch": 0.34069863390483174,
+      "learning_rate": 0.00013186210690382323,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7420512437820435,
+      "num_tokens": 85914547.0,
+      "step": 37160
+    },
+    {
+      "epoch": 0.340790318144311,
+      "learning_rate": 0.0001318437700559274,
+      "loss": 0.8293,
+      "mean_token_accuracy": 0.7506530165672303,
+      "num_tokens": 85937453.0,
+      "step": 37170
+    },
+    {
+      "epoch": 0.34088200238379024,
+      "learning_rate": 0.00013182543320803153,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7584544062614441,
+      "num_tokens": 85960622.0,
+      "step": 37180
+    },
+    {
+      "epoch": 0.34097368662326943,
+      "learning_rate": 0.0001318070963601357,
+      "loss": 0.8418,
+      "mean_token_accuracy": 0.7440663695335388,
+      "num_tokens": 85984150.0,
+      "step": 37190
+    },
+    {
+      "epoch": 0.3410653708627487,
+      "learning_rate": 0.00013178875951223986,
+      "loss": 0.8015,
+      "mean_token_accuracy": 0.7545207738876343,
+      "num_tokens": 86006654.0,
+      "step": 37200
+    },
+    {
+      "epoch": 0.34115705510222794,
+      "learning_rate": 0.000131770422664344,
+      "loss": 0.8287,
+      "mean_token_accuracy": 0.745617938041687,
+      "num_tokens": 86029823.0,
+      "step": 37210
+    },
+    {
+      "epoch": 0.3412487393417072,
+      "learning_rate": 0.00013175208581644817,
+      "loss": 0.8011,
+      "mean_token_accuracy": 0.7545137763023376,
+      "num_tokens": 86053544.0,
+      "step": 37220
+    },
+    {
+      "epoch": 0.3413404235811864,
+      "learning_rate": 0.0001317337489685523,
+      "loss": 0.8584,
+      "mean_token_accuracy": 0.7426408231258392,
+      "num_tokens": 86076110.0,
+      "step": 37230
+    },
+    {
+      "epoch": 0.34143210782066563,
+      "learning_rate": 0.00013171541212065644,
+      "loss": 0.8628,
+      "mean_token_accuracy": 0.7432761430740357,
+      "num_tokens": 86099801.0,
+      "step": 37240
+    },
+    {
+      "epoch": 0.3415237920601449,
+      "learning_rate": 0.00013169707527276064,
+      "loss": 0.8363,
+      "mean_token_accuracy": 0.7503327965736389,
+      "num_tokens": 86121711.0,
+      "step": 37250
+    },
+    {
+      "epoch": 0.3416154762996241,
+      "learning_rate": 0.00013167873842486477,
+      "loss": 0.8342,
+      "mean_token_accuracy": 0.7504224061965943,
+      "num_tokens": 86144332.0,
+      "step": 37260
+    },
+    {
+      "epoch": 0.3417071605391033,
+      "learning_rate": 0.00013166040157696894,
+      "loss": 0.8687,
+      "mean_token_accuracy": 0.7470543205738067,
+      "num_tokens": 86167057.0,
+      "step": 37270
+    },
+    {
+      "epoch": 0.3417988447785826,
+      "learning_rate": 0.00013164206472907308,
+      "loss": 0.8734,
+      "mean_token_accuracy": 0.7430323839187623,
+      "num_tokens": 86190290.0,
+      "step": 37280
+    },
+    {
+      "epoch": 0.3418905290180618,
+      "learning_rate": 0.00013162372788117722,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.7559139013290406,
+      "num_tokens": 86213506.0,
+      "step": 37290
+    },
+    {
+      "epoch": 0.341982213257541,
+      "learning_rate": 0.00013160539103328138,
+      "loss": 0.8784,
+      "mean_token_accuracy": 0.7396174609661103,
+      "num_tokens": 86236711.0,
+      "step": 37300
+    },
+    {
+      "epoch": 0.3420738974970203,
+      "learning_rate": 0.00013158705418538552,
+      "loss": 0.8313,
+      "mean_token_accuracy": 0.7533867299556732,
+      "num_tokens": 86259182.0,
+      "step": 37310
+    },
+    {
+      "epoch": 0.34216558173649947,
+      "learning_rate": 0.0001315687173374897,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.7477795958518982,
+      "num_tokens": 86282316.0,
+      "step": 37320
+    },
+    {
+      "epoch": 0.3422572659759787,
+      "learning_rate": 0.00013155038048959385,
+      "loss": 0.8353,
+      "mean_token_accuracy": 0.7535763680934906,
+      "num_tokens": 86305575.0,
+      "step": 37330
+    },
+    {
+      "epoch": 0.34234895021545797,
+      "learning_rate": 0.000131532043641698,
+      "loss": 0.8176,
+      "mean_token_accuracy": 0.7536270260810852,
+      "num_tokens": 86329018.0,
+      "step": 37340
+    },
+    {
+      "epoch": 0.3424406344549372,
+      "learning_rate": 0.00013151370679380216,
+      "loss": 0.8741,
+      "mean_token_accuracy": 0.7423791468143464,
+      "num_tokens": 86352916.0,
+      "step": 37350
+    },
+    {
+      "epoch": 0.3425323186944164,
+      "learning_rate": 0.0001314953699459063,
+      "loss": 0.8224,
+      "mean_token_accuracy": 0.7576164901256561,
+      "num_tokens": 86375967.0,
+      "step": 37360
+    },
+    {
+      "epoch": 0.34262400293389567,
+      "learning_rate": 0.00013147703309801046,
+      "loss": 0.8772,
+      "mean_token_accuracy": 0.7445416212081909,
+      "num_tokens": 86398528.0,
+      "step": 37370
+    },
+    {
+      "epoch": 0.3427156871733749,
+      "learning_rate": 0.00013145869625011462,
+      "loss": 0.8697,
+      "mean_token_accuracy": 0.744277560710907,
+      "num_tokens": 86421102.0,
+      "step": 37380
+    },
+    {
+      "epoch": 0.3428073714128541,
+      "learning_rate": 0.00013144035940221876,
+      "loss": 0.8242,
+      "mean_token_accuracy": 0.7569944679737091,
+      "num_tokens": 86444589.0,
+      "step": 37390
+    },
+    {
+      "epoch": 0.34289905565233336,
+      "learning_rate": 0.00013142202255432293,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.7452614784240723,
+      "num_tokens": 86468504.0,
+      "step": 37400
+    },
+    {
+      "epoch": 0.3429907398918126,
+      "learning_rate": 0.00013140368570642707,
+      "loss": 0.8209,
+      "mean_token_accuracy": 0.7482340097427368,
+      "num_tokens": 86491017.0,
+      "step": 37410
+    },
+    {
+      "epoch": 0.3430824241312918,
+      "learning_rate": 0.00013138534885853123,
+      "loss": 0.8459,
+      "mean_token_accuracy": 0.7495341420173645,
+      "num_tokens": 86513521.0,
+      "step": 37420
+    },
+    {
+      "epoch": 0.34317410837077106,
+      "learning_rate": 0.00013136701201063537,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7421615958213806,
+      "num_tokens": 86536788.0,
+      "step": 37430
+    },
+    {
+      "epoch": 0.3432657926102503,
+      "learning_rate": 0.0001313486751627395,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.7407772898674011,
+      "num_tokens": 86559493.0,
+      "step": 37440
+    },
+    {
+      "epoch": 0.3433574768497295,
+      "learning_rate": 0.0001313303383148437,
+      "loss": 0.8773,
+      "mean_token_accuracy": 0.7429486989974976,
+      "num_tokens": 86582716.0,
+      "step": 37450
+    },
+    {
+      "epoch": 0.34344916108920875,
+      "learning_rate": 0.00013131200146694784,
+      "loss": 0.8771,
+      "mean_token_accuracy": 0.7355626225471497,
+      "num_tokens": 86605360.0,
+      "step": 37460
+    },
+    {
+      "epoch": 0.343540845328688,
+      "learning_rate": 0.000131293664619052,
+      "loss": 0.823,
+      "mean_token_accuracy": 0.7481252431869507,
+      "num_tokens": 86629032.0,
+      "step": 37470
+    },
+    {
+      "epoch": 0.34363252956816726,
+      "learning_rate": 0.00013127532777115614,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.7385040998458863,
+      "num_tokens": 86651654.0,
+      "step": 37480
+    },
+    {
+      "epoch": 0.34372421380764645,
+      "learning_rate": 0.00013125699092326028,
+      "loss": 0.8448,
+      "mean_token_accuracy": 0.7508082270622254,
+      "num_tokens": 86674558.0,
+      "step": 37490
+    },
+    {
+      "epoch": 0.3438158980471257,
+      "learning_rate": 0.00013123865407536445,
+      "loss": 0.8501,
+      "mean_token_accuracy": 0.7485431969165802,
+      "num_tokens": 86697438.0,
+      "step": 37500
+    },
+    {
+      "epoch": 0.34390758228660495,
+      "learning_rate": 0.0001312203172274686,
+      "loss": 0.8621,
+      "mean_token_accuracy": 0.7432440876960754,
+      "num_tokens": 86720289.0,
+      "step": 37510
+    },
+    {
+      "epoch": 0.34399926652608415,
+      "learning_rate": 0.00013120198037957278,
+      "loss": 0.857,
+      "mean_token_accuracy": 0.7458439469337463,
+      "num_tokens": 86743169.0,
+      "step": 37520
+    },
+    {
+      "epoch": 0.3440909507655634,
+      "learning_rate": 0.00013118364353167692,
+      "loss": 0.8592,
+      "mean_token_accuracy": 0.7406644225120544,
+      "num_tokens": 86765674.0,
+      "step": 37530
+    },
+    {
+      "epoch": 0.34418263500504265,
+      "learning_rate": 0.00013116530668378105,
+      "loss": 0.8136,
+      "mean_token_accuracy": 0.7552064299583435,
+      "num_tokens": 86788919.0,
+      "step": 37540
+    },
+    {
+      "epoch": 0.34427431924452184,
+      "learning_rate": 0.00013114696983588522,
+      "loss": 0.8192,
+      "mean_token_accuracy": 0.7529111206531525,
+      "num_tokens": 86811614.0,
+      "step": 37550
+    },
+    {
+      "epoch": 0.3443660034840011,
+      "learning_rate": 0.00013112863298798936,
+      "loss": 0.8744,
+      "mean_token_accuracy": 0.7411521255970002,
+      "num_tokens": 86834577.0,
+      "step": 37560
+    },
+    {
+      "epoch": 0.34445768772348034,
+      "learning_rate": 0.00013111029614009352,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7424495041370391,
+      "num_tokens": 86857501.0,
+      "step": 37570
+    },
+    {
+      "epoch": 0.3445493719629596,
+      "learning_rate": 0.0001310919592921977,
+      "loss": 0.8342,
+      "mean_token_accuracy": 0.7537790715694428,
+      "num_tokens": 86880464.0,
+      "step": 37580
+    },
+    {
+      "epoch": 0.3446410562024388,
+      "learning_rate": 0.00013107362244430183,
+      "loss": 0.8501,
+      "mean_token_accuracy": 0.742199969291687,
+      "num_tokens": 86902963.0,
+      "step": 37590
+    },
+    {
+      "epoch": 0.34473274044191804,
+      "learning_rate": 0.000131055285596406,
+      "loss": 0.8852,
+      "mean_token_accuracy": 0.7372816562652588,
+      "num_tokens": 86927118.0,
+      "step": 37600
+    },
+    {
+      "epoch": 0.3448244246813973,
+      "learning_rate": 0.00013103694874851013,
+      "loss": 0.9019,
+      "mean_token_accuracy": 0.7293107867240906,
+      "num_tokens": 86950308.0,
+      "step": 37610
+    },
+    {
+      "epoch": 0.3449161089208765,
+      "learning_rate": 0.0001310186119006143,
+      "loss": 0.824,
+      "mean_token_accuracy": 0.7526019752025604,
+      "num_tokens": 86973391.0,
+      "step": 37620
+    },
+    {
+      "epoch": 0.34500779316035574,
+      "learning_rate": 0.00013100027505271843,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.7496634900569916,
+      "num_tokens": 86997271.0,
+      "step": 37630
+    },
+    {
+      "epoch": 0.345099477399835,
+      "learning_rate": 0.0001309819382048226,
+      "loss": 0.8726,
+      "mean_token_accuracy": 0.7416471123695374,
+      "num_tokens": 87020380.0,
+      "step": 37640
+    },
+    {
+      "epoch": 0.3451911616393142,
+      "learning_rate": 0.00013096360135692677,
+      "loss": 0.8231,
+      "mean_token_accuracy": 0.7499141097068787,
+      "num_tokens": 87043012.0,
+      "step": 37650
+    },
+    {
+      "epoch": 0.34528284587879343,
+      "learning_rate": 0.0001309452645090309,
+      "loss": 0.8591,
+      "mean_token_accuracy": 0.7471584260463715,
+      "num_tokens": 87066014.0,
+      "step": 37660
+    },
+    {
+      "epoch": 0.3453745301182727,
+      "learning_rate": 0.00013092692766113507,
+      "loss": 0.858,
+      "mean_token_accuracy": 0.7428099274635315,
+      "num_tokens": 87089163.0,
+      "step": 37670
+    },
+    {
+      "epoch": 0.3454662143577519,
+      "learning_rate": 0.0001309085908132392,
+      "loss": 0.8687,
+      "mean_token_accuracy": 0.7425667703151703,
+      "num_tokens": 87111221.0,
+      "step": 37680
+    },
+    {
+      "epoch": 0.34555789859723113,
+      "learning_rate": 0.00013089025396534335,
+      "loss": 0.8447,
+      "mean_token_accuracy": 0.7500782668590545,
+      "num_tokens": 87134521.0,
+      "step": 37690
+    },
+    {
+      "epoch": 0.3456495828367104,
+      "learning_rate": 0.0001308719171174475,
+      "loss": 0.8828,
+      "mean_token_accuracy": 0.7427931904792786,
+      "num_tokens": 87157449.0,
+      "step": 37700
+    },
+    {
+      "epoch": 0.34574126707618963,
+      "learning_rate": 0.00013085358026955168,
+      "loss": 0.8773,
+      "mean_token_accuracy": 0.7426659226417541,
+      "num_tokens": 87180357.0,
+      "step": 37710
+    },
+    {
+      "epoch": 0.3458329513156688,
+      "learning_rate": 0.00013083524342165584,
+      "loss": 0.8455,
+      "mean_token_accuracy": 0.7439717769622802,
+      "num_tokens": 87203970.0,
+      "step": 37720
+    },
+    {
+      "epoch": 0.3459246355551481,
+      "learning_rate": 0.00013081690657375998,
+      "loss": 0.8538,
+      "mean_token_accuracy": 0.7418691635131835,
+      "num_tokens": 87227687.0,
+      "step": 37730
+    },
+    {
+      "epoch": 0.3460163197946273,
+      "learning_rate": 0.00013079856972586412,
+      "loss": 0.8219,
+      "mean_token_accuracy": 0.744043642282486,
+      "num_tokens": 87251269.0,
+      "step": 37740
+    },
+    {
+      "epoch": 0.3461080040341065,
+      "learning_rate": 0.00013078023287796828,
+      "loss": 0.8685,
+      "mean_token_accuracy": 0.744477504491806,
+      "num_tokens": 87273693.0,
+      "step": 37750
+    },
+    {
+      "epoch": 0.34619968827358577,
+      "learning_rate": 0.00013076189603007242,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7536784887313843,
+      "num_tokens": 87296790.0,
+      "step": 37760
+    },
+    {
+      "epoch": 0.346291372513065,
+      "learning_rate": 0.0001307435591821766,
+      "loss": 0.8365,
+      "mean_token_accuracy": 0.7495584964752198,
+      "num_tokens": 87319613.0,
+      "step": 37770
+    },
+    {
+      "epoch": 0.3463830567525442,
+      "learning_rate": 0.00013072522233428075,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.741416972875595,
+      "num_tokens": 87342338.0,
+      "step": 37780
+    },
+    {
+      "epoch": 0.34647474099202347,
+      "learning_rate": 0.0001307068854863849,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.7562924802303315,
+      "num_tokens": 87365148.0,
+      "step": 37790
+    },
+    {
+      "epoch": 0.3465664252315027,
+      "learning_rate": 0.00013068854863848906,
+      "loss": 0.8291,
+      "mean_token_accuracy": 0.7532803475856781,
+      "num_tokens": 87388057.0,
+      "step": 37800
+    },
+    {
+      "epoch": 0.3466581094709819,
+      "learning_rate": 0.0001306702117905932,
+      "loss": 0.8232,
+      "mean_token_accuracy": 0.7507896840572357,
+      "num_tokens": 87411028.0,
+      "step": 37810
+    },
+    {
+      "epoch": 0.34674979371046116,
+      "learning_rate": 0.00013065187494269736,
+      "loss": 0.8323,
+      "mean_token_accuracy": 0.7489738523960113,
+      "num_tokens": 87435038.0,
+      "step": 37820
+    },
+    {
+      "epoch": 0.3468414779499404,
+      "learning_rate": 0.0001306335380948015,
+      "loss": 0.8617,
+      "mean_token_accuracy": 0.7449505984783172,
+      "num_tokens": 87458597.0,
+      "step": 37830
+    },
+    {
+      "epoch": 0.34693316218941966,
+      "learning_rate": 0.00013061520124690566,
+      "loss": 0.8692,
+      "mean_token_accuracy": 0.7353573739528656,
+      "num_tokens": 87481060.0,
+      "step": 37840
+    },
+    {
+      "epoch": 0.34702484642889886,
+      "learning_rate": 0.00013059686439900983,
+      "loss": 0.8923,
+      "mean_token_accuracy": 0.7383727550506591,
+      "num_tokens": 87504648.0,
+      "step": 37850
+    },
+    {
+      "epoch": 0.3471165306683781,
+      "learning_rate": 0.00013057852755111397,
+      "loss": 0.8376,
+      "mean_token_accuracy": 0.7521066486835479,
+      "num_tokens": 87527636.0,
+      "step": 37860
+    },
+    {
+      "epoch": 0.34720821490785736,
+      "learning_rate": 0.00013056019070321813,
+      "loss": 0.8213,
+      "mean_token_accuracy": 0.748869800567627,
+      "num_tokens": 87550078.0,
+      "step": 37870
+    },
+    {
+      "epoch": 0.34729989914733655,
+      "learning_rate": 0.00013054185385532227,
+      "loss": 0.8356,
+      "mean_token_accuracy": 0.7438906967639923,
+      "num_tokens": 87572798.0,
+      "step": 37880
+    },
+    {
+      "epoch": 0.3473915833868158,
+      "learning_rate": 0.0001305235170074264,
+      "loss": 0.8193,
+      "mean_token_accuracy": 0.7509337663650513,
+      "num_tokens": 87595912.0,
+      "step": 37890
+    },
+    {
+      "epoch": 0.34748326762629506,
+      "learning_rate": 0.0001305051801595306,
+      "loss": 0.8567,
+      "mean_token_accuracy": 0.7424492061138153,
+      "num_tokens": 87618768.0,
+      "step": 37900
+    },
+    {
+      "epoch": 0.34757495186577425,
+      "learning_rate": 0.00013048684331163474,
+      "loss": 0.8203,
+      "mean_token_accuracy": 0.7546485662460327,
+      "num_tokens": 87642943.0,
+      "step": 37910
+    },
+    {
+      "epoch": 0.3476666361052535,
+      "learning_rate": 0.0001304685064637389,
+      "loss": 0.8419,
+      "mean_token_accuracy": 0.7400360822677612,
+      "num_tokens": 87666185.0,
+      "step": 37920
+    },
+    {
+      "epoch": 0.34775832034473275,
+      "learning_rate": 0.00013045016961584304,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7454596698284149,
+      "num_tokens": 87688712.0,
+      "step": 37930
+    },
+    {
+      "epoch": 0.34785000458421195,
+      "learning_rate": 0.00013043183276794718,
+      "loss": 0.8901,
+      "mean_token_accuracy": 0.7401292622089386,
+      "num_tokens": 87711469.0,
+      "step": 37940
+    },
+    {
+      "epoch": 0.3479416888236912,
+      "learning_rate": 0.00013041349592005135,
+      "loss": 0.8863,
+      "mean_token_accuracy": 0.7317825496196747,
+      "num_tokens": 87733921.0,
+      "step": 37950
+    },
+    {
+      "epoch": 0.34803337306317045,
+      "learning_rate": 0.0001303951590721555,
+      "loss": 0.8743,
+      "mean_token_accuracy": 0.7413212358951569,
+      "num_tokens": 87756943.0,
+      "step": 37960
+    },
+    {
+      "epoch": 0.3481250573026497,
+      "learning_rate": 0.00013037682222425965,
+      "loss": 0.8531,
+      "mean_token_accuracy": 0.7407483100891114,
+      "num_tokens": 87779530.0,
+      "step": 37970
+    },
+    {
+      "epoch": 0.3482167415421289,
+      "learning_rate": 0.00013035848537636382,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7409225344657898,
+      "num_tokens": 87802282.0,
+      "step": 37980
+    },
+    {
+      "epoch": 0.34830842578160814,
+      "learning_rate": 0.00013034014852846796,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7477506041526795,
+      "num_tokens": 87824812.0,
+      "step": 37990
+    },
+    {
+      "epoch": 0.3484001100210874,
+      "learning_rate": 0.00013032181168057212,
+      "loss": 0.8426,
+      "mean_token_accuracy": 0.7458292841911316,
+      "num_tokens": 87847989.0,
+      "step": 38000
+    },
+    {
+      "epoch": 0.3484917942605666,
+      "learning_rate": 0.00013030347483267626,
+      "loss": 0.8286,
+      "mean_token_accuracy": 0.7489347279071807,
+      "num_tokens": 87870754.0,
+      "step": 38010
+    },
+    {
+      "epoch": 0.34858347850004584,
+      "learning_rate": 0.00013028513798478043,
+      "loss": 0.8477,
+      "mean_token_accuracy": 0.7397971987724304,
+      "num_tokens": 87894222.0,
+      "step": 38020
+    },
+    {
+      "epoch": 0.3486751627395251,
+      "learning_rate": 0.0001302668011368846,
+      "loss": 0.8408,
+      "mean_token_accuracy": 0.7475711643695832,
+      "num_tokens": 87917482.0,
+      "step": 38030
+    },
+    {
+      "epoch": 0.3487668469790043,
+      "learning_rate": 0.00013024846428898873,
+      "loss": 0.8524,
+      "mean_token_accuracy": 0.7466792702674866,
+      "num_tokens": 87939990.0,
+      "step": 38040
+    },
+    {
+      "epoch": 0.34885853121848354,
+      "learning_rate": 0.0001302301274410929,
+      "loss": 0.8713,
+      "mean_token_accuracy": 0.740998363494873,
+      "num_tokens": 87963274.0,
+      "step": 38050
+    },
+    {
+      "epoch": 0.3489502154579628,
+      "learning_rate": 0.00013021179059319703,
+      "loss": 0.7988,
+      "mean_token_accuracy": 0.7589530646800995,
+      "num_tokens": 87985674.0,
+      "step": 38060
+    },
+    {
+      "epoch": 0.34904189969744204,
+      "learning_rate": 0.0001301934537453012,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.7426394581794739,
+      "num_tokens": 88008767.0,
+      "step": 38070
+    },
+    {
+      "epoch": 0.34913358393692123,
+      "learning_rate": 0.00013017511689740534,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.740776801109314,
+      "num_tokens": 88032679.0,
+      "step": 38080
+    },
+    {
+      "epoch": 0.3492252681764005,
+      "learning_rate": 0.00013015678004950947,
+      "loss": 0.8563,
+      "mean_token_accuracy": 0.7444301605224609,
+      "num_tokens": 88055877.0,
+      "step": 38090
+    },
+    {
+      "epoch": 0.34931695241587973,
+      "learning_rate": 0.00013013844320161367,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.7415159523487092,
+      "num_tokens": 88079298.0,
+      "step": 38100
+    },
+    {
+      "epoch": 0.34940863665535893,
+      "learning_rate": 0.0001301201063537178,
+      "loss": 0.8382,
+      "mean_token_accuracy": 0.7479581534862518,
+      "num_tokens": 88102502.0,
+      "step": 38110
+    },
+    {
+      "epoch": 0.3495003208948382,
+      "learning_rate": 0.00013010176950582197,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7455908417701721,
+      "num_tokens": 88125584.0,
+      "step": 38120
+    },
+    {
+      "epoch": 0.34959200513431743,
+      "learning_rate": 0.0001300834326579261,
+      "loss": 0.8383,
+      "mean_token_accuracy": 0.7464239716529846,
+      "num_tokens": 88148301.0,
+      "step": 38130
+    },
+    {
+      "epoch": 0.3496836893737966,
+      "learning_rate": 0.00013006509581003025,
+      "loss": 0.8461,
+      "mean_token_accuracy": 0.7454171359539032,
+      "num_tokens": 88171274.0,
+      "step": 38140
+    },
+    {
+      "epoch": 0.3497753736132759,
+      "learning_rate": 0.0001300467589621344,
+      "loss": 0.8529,
+      "mean_token_accuracy": 0.750511234998703,
+      "num_tokens": 88194919.0,
+      "step": 38150
+    },
+    {
+      "epoch": 0.3498670578527551,
+      "learning_rate": 0.00013002842211423858,
+      "loss": 0.8374,
+      "mean_token_accuracy": 0.7474356830120087,
+      "num_tokens": 88217787.0,
+      "step": 38160
+    },
+    {
+      "epoch": 0.3499587420922343,
+      "learning_rate": 0.00013001008526634272,
+      "loss": 0.8546,
+      "mean_token_accuracy": 0.7470196425914765,
+      "num_tokens": 88240788.0,
+      "step": 38170
+    },
+    {
+      "epoch": 0.35005042633171357,
+      "learning_rate": 0.00012999174841844688,
+      "loss": 0.8642,
+      "mean_token_accuracy": 0.7350736796855927,
+      "num_tokens": 88262889.0,
+      "step": 38180
+    },
+    {
+      "epoch": 0.3501421105711928,
+      "learning_rate": 0.00012997341157055102,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.7441868960857392,
+      "num_tokens": 88285565.0,
+      "step": 38190
+    },
+    {
+      "epoch": 0.35023379481067207,
+      "learning_rate": 0.00012995507472265519,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7338118314743042,
+      "num_tokens": 88308280.0,
+      "step": 38200
+    },
+    {
+      "epoch": 0.35032547905015127,
+      "learning_rate": 0.00012993673787475932,
+      "loss": 0.9061,
+      "mean_token_accuracy": 0.7339636385440826,
+      "num_tokens": 88330976.0,
+      "step": 38210
+    },
+    {
+      "epoch": 0.3504171632896305,
+      "learning_rate": 0.0001299184010268635,
+      "loss": 0.8417,
+      "mean_token_accuracy": 0.7496311962604523,
+      "num_tokens": 88353606.0,
+      "step": 38220
+    },
+    {
+      "epoch": 0.35050884752910977,
+      "learning_rate": 0.00012990006417896765,
+      "loss": 0.8338,
+      "mean_token_accuracy": 0.7457334578037262,
+      "num_tokens": 88375937.0,
+      "step": 38230
+    },
+    {
+      "epoch": 0.35060053176858896,
+      "learning_rate": 0.0001298817273310718,
+      "loss": 0.8359,
+      "mean_token_accuracy": 0.7450960040092468,
+      "num_tokens": 88398913.0,
+      "step": 38240
+    },
+    {
+      "epoch": 0.3506922160080682,
+      "learning_rate": 0.00012986339048317596,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.7556888043880463,
+      "num_tokens": 88422414.0,
+      "step": 38250
+    },
+    {
+      "epoch": 0.35078390024754746,
+      "learning_rate": 0.0001298450536352801,
+      "loss": 0.8166,
+      "mean_token_accuracy": 0.7548144578933715,
+      "num_tokens": 88445186.0,
+      "step": 38260
+    },
+    {
+      "epoch": 0.35087558448702666,
+      "learning_rate": 0.00012982671678738426,
+      "loss": 0.8977,
+      "mean_token_accuracy": 0.7276447057723999,
+      "num_tokens": 88468663.0,
+      "step": 38270
+    },
+    {
+      "epoch": 0.3509672687265059,
+      "learning_rate": 0.0001298083799394884,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.7385667145252228,
+      "num_tokens": 88491986.0,
+      "step": 38280
+    },
+    {
+      "epoch": 0.35105895296598516,
+      "learning_rate": 0.00012979004309159257,
+      "loss": 0.8491,
+      "mean_token_accuracy": 0.7474011480808258,
+      "num_tokens": 88515616.0,
+      "step": 38290
+    },
+    {
+      "epoch": 0.35115063720546436,
+      "learning_rate": 0.00012977170624369673,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.7485447883605957,
+      "num_tokens": 88538853.0,
+      "step": 38300
+    },
+    {
+      "epoch": 0.3512423214449436,
+      "learning_rate": 0.00012975336939580087,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7485349535942077,
+      "num_tokens": 88562617.0,
+      "step": 38310
+    },
+    {
+      "epoch": 0.35133400568442286,
+      "learning_rate": 0.00012973503254790504,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.749411553144455,
+      "num_tokens": 88585428.0,
+      "step": 38320
+    },
+    {
+      "epoch": 0.3514256899239021,
+      "learning_rate": 0.00012971669570000917,
+      "loss": 0.8661,
+      "mean_token_accuracy": 0.7448081552982331,
+      "num_tokens": 88609616.0,
+      "step": 38330
+    },
+    {
+      "epoch": 0.3515173741633813,
+      "learning_rate": 0.0001296983588521133,
+      "loss": 0.7972,
+      "mean_token_accuracy": 0.762670373916626,
+      "num_tokens": 88632131.0,
+      "step": 38340
+    },
+    {
+      "epoch": 0.35160905840286055,
+      "learning_rate": 0.00012968002200421748,
+      "loss": 0.8628,
+      "mean_token_accuracy": 0.7470655024051667,
+      "num_tokens": 88655781.0,
+      "step": 38350
+    },
+    {
+      "epoch": 0.3517007426423398,
+      "learning_rate": 0.00012966168515632164,
+      "loss": 0.8415,
+      "mean_token_accuracy": 0.7441945314407349,
+      "num_tokens": 88679327.0,
+      "step": 38360
+    },
+    {
+      "epoch": 0.351792426881819,
+      "learning_rate": 0.00012964334830842578,
+      "loss": 0.8707,
+      "mean_token_accuracy": 0.7448734819889069,
+      "num_tokens": 88703165.0,
+      "step": 38370
+    },
+    {
+      "epoch": 0.35188411112129825,
+      "learning_rate": 0.00012962501146052995,
+      "loss": 0.8312,
+      "mean_token_accuracy": 0.7476518809795379,
+      "num_tokens": 88726363.0,
+      "step": 38380
+    },
+    {
+      "epoch": 0.3519757953607775,
+      "learning_rate": 0.00012960667461263408,
+      "loss": 0.8249,
+      "mean_token_accuracy": 0.757578718662262,
+      "num_tokens": 88750254.0,
+      "step": 38390
+    },
+    {
+      "epoch": 0.3520674796002567,
+      "learning_rate": 0.00012958833776473825,
+      "loss": 0.8694,
+      "mean_token_accuracy": 0.7445272147655487,
+      "num_tokens": 88774885.0,
+      "step": 38400
+    },
+    {
+      "epoch": 0.35215916383973594,
+      "learning_rate": 0.0001295700009168424,
+      "loss": 0.854,
+      "mean_token_accuracy": 0.7485789954662323,
+      "num_tokens": 88798150.0,
+      "step": 38410
+    },
+    {
+      "epoch": 0.3522508480792152,
+      "learning_rate": 0.00012955166406894655,
+      "loss": 0.8862,
+      "mean_token_accuracy": 0.7371294021606445,
+      "num_tokens": 88820933.0,
+      "step": 38420
+    },
+    {
+      "epoch": 0.3523425323186944,
+      "learning_rate": 0.00012953332722105072,
+      "loss": 0.8318,
+      "mean_token_accuracy": 0.7497548401355744,
+      "num_tokens": 88844233.0,
+      "step": 38430
+    },
+    {
+      "epoch": 0.35243421655817364,
+      "learning_rate": 0.00012951499037315486,
+      "loss": 0.8868,
+      "mean_token_accuracy": 0.744812297821045,
+      "num_tokens": 88867769.0,
+      "step": 38440
+    },
+    {
+      "epoch": 0.3525259007976529,
+      "learning_rate": 0.00012949665352525902,
+      "loss": 0.8434,
+      "mean_token_accuracy": 0.7472955286502838,
+      "num_tokens": 88891327.0,
+      "step": 38450
+    },
+    {
+      "epoch": 0.35261758503713214,
+      "learning_rate": 0.00012947831667736316,
+      "loss": 0.8129,
+      "mean_token_accuracy": 0.761379873752594,
+      "num_tokens": 88914666.0,
+      "step": 38460
+    },
+    {
+      "epoch": 0.35270926927661134,
+      "learning_rate": 0.00012945997982946733,
+      "loss": 0.7881,
+      "mean_token_accuracy": 0.7566261053085327,
+      "num_tokens": 88936929.0,
+      "step": 38470
+    },
+    {
+      "epoch": 0.3528009535160906,
+      "learning_rate": 0.00012944164298157147,
+      "loss": 0.8293,
+      "mean_token_accuracy": 0.7504927217960358,
+      "num_tokens": 88960216.0,
+      "step": 38480
+    },
+    {
+      "epoch": 0.35289263775556984,
+      "learning_rate": 0.00012942330613367563,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.7413956522941589,
+      "num_tokens": 88983905.0,
+      "step": 38490
+    },
+    {
+      "epoch": 0.35298432199504903,
+      "learning_rate": 0.0001294049692857798,
+      "loss": 0.8044,
+      "mean_token_accuracy": 0.7506045043468476,
+      "num_tokens": 89007115.0,
+      "step": 38500
+    },
+    {
+      "epoch": 0.3530760062345283,
+      "learning_rate": 0.00012938663243788393,
+      "loss": 0.8731,
+      "mean_token_accuracy": 0.7386382818222046,
+      "num_tokens": 89030435.0,
+      "step": 38510
+    },
+    {
+      "epoch": 0.35316769047400753,
+      "learning_rate": 0.0001293682955899881,
+      "loss": 0.8282,
+      "mean_token_accuracy": 0.7485121607780456,
+      "num_tokens": 89053089.0,
+      "step": 38520
+    },
+    {
+      "epoch": 0.35325937471348673,
+      "learning_rate": 0.00012934995874209224,
+      "loss": 0.8346,
+      "mean_token_accuracy": 0.7461729288101197,
+      "num_tokens": 89075910.0,
+      "step": 38530
+    },
+    {
+      "epoch": 0.353351058952966,
+      "learning_rate": 0.00012933162189419638,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.7396175980567932,
+      "num_tokens": 89098818.0,
+      "step": 38540
+    },
+    {
+      "epoch": 0.35344274319244523,
+      "learning_rate": 0.00012931328504630054,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7416675686836243,
+      "num_tokens": 89121559.0,
+      "step": 38550
+    },
+    {
+      "epoch": 0.3535344274319244,
+      "learning_rate": 0.0001292949481984047,
+      "loss": 0.8591,
+      "mean_token_accuracy": 0.7444033920764923,
+      "num_tokens": 89144569.0,
+      "step": 38560
+    },
+    {
+      "epoch": 0.3536261116714037,
+      "learning_rate": 0.00012927661135050885,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.7391893625259399,
+      "num_tokens": 89167488.0,
+      "step": 38570
+    },
+    {
+      "epoch": 0.3537177959108829,
+      "learning_rate": 0.000129258274502613,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7431459367275238,
+      "num_tokens": 89191087.0,
+      "step": 38580
+    },
+    {
+      "epoch": 0.3538094801503622,
+      "learning_rate": 0.00012923993765471715,
+      "loss": 0.8943,
+      "mean_token_accuracy": 0.7316861569881439,
+      "num_tokens": 89214751.0,
+      "step": 38590
+    },
+    {
+      "epoch": 0.35390116438984137,
+      "learning_rate": 0.00012922160080682131,
+      "loss": 0.8157,
+      "mean_token_accuracy": 0.7547654986381531,
+      "num_tokens": 89237541.0,
+      "step": 38600
+    },
+    {
+      "epoch": 0.3539928486293206,
+      "learning_rate": 0.00012920326395892545,
+      "loss": 0.8402,
+      "mean_token_accuracy": 0.7504114627838134,
+      "num_tokens": 89261000.0,
+      "step": 38610
+    },
+    {
+      "epoch": 0.3540845328687999,
+      "learning_rate": 0.00012918492711102962,
+      "loss": 0.8761,
+      "mean_token_accuracy": 0.7410715520381927,
+      "num_tokens": 89284352.0,
+      "step": 38620
+    },
+    {
+      "epoch": 0.35417621710827907,
+      "learning_rate": 0.00012916659026313378,
+      "loss": 0.8579,
+      "mean_token_accuracy": 0.7444826424121856,
+      "num_tokens": 89307059.0,
+      "step": 38630
+    },
+    {
+      "epoch": 0.3542679013477583,
+      "learning_rate": 0.00012914825341523792,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7434938609600067,
+      "num_tokens": 89329848.0,
+      "step": 38640
+    },
+    {
+      "epoch": 0.35435958558723757,
+      "learning_rate": 0.0001291299165673421,
+      "loss": 0.8214,
+      "mean_token_accuracy": 0.7490093350410462,
+      "num_tokens": 89352610.0,
+      "step": 38650
+    },
+    {
+      "epoch": 0.35445126982671676,
+      "learning_rate": 0.00012911157971944623,
+      "loss": 0.8965,
+      "mean_token_accuracy": 0.7354145407676697,
+      "num_tokens": 89375348.0,
+      "step": 38660
+    },
+    {
+      "epoch": 0.354542954066196,
+      "learning_rate": 0.0001290932428715504,
+      "loss": 0.8358,
+      "mean_token_accuracy": 0.7552769839763641,
+      "num_tokens": 89398362.0,
+      "step": 38670
+    },
+    {
+      "epoch": 0.35463463830567526,
+      "learning_rate": 0.00012907490602365453,
+      "loss": 0.8434,
+      "mean_token_accuracy": 0.7459128439426422,
+      "num_tokens": 89421632.0,
+      "step": 38680
+    },
+    {
+      "epoch": 0.3547263225451545,
+      "learning_rate": 0.0001290565691757587,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7391178250312805,
+      "num_tokens": 89444958.0,
+      "step": 38690
+    },
+    {
+      "epoch": 0.3548180067846337,
+      "learning_rate": 0.00012903823232786286,
+      "loss": 0.8234,
+      "mean_token_accuracy": 0.7508709132671356,
+      "num_tokens": 89468517.0,
+      "step": 38700
+    },
+    {
+      "epoch": 0.35490969102411296,
+      "learning_rate": 0.000129019895479967,
+      "loss": 0.8756,
+      "mean_token_accuracy": 0.7424369275569915,
+      "num_tokens": 89491319.0,
+      "step": 38710
+    },
+    {
+      "epoch": 0.3550013752635922,
+      "learning_rate": 0.00012900155863207116,
+      "loss": 0.859,
+      "mean_token_accuracy": 0.741323322057724,
+      "num_tokens": 89513835.0,
+      "step": 38720
+    },
+    {
+      "epoch": 0.3550930595030714,
+      "learning_rate": 0.0001289832217841753,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7459016382694245,
+      "num_tokens": 89535923.0,
+      "step": 38730
+    },
+    {
+      "epoch": 0.35518474374255066,
+      "learning_rate": 0.00012896488493627944,
+      "loss": 0.8435,
+      "mean_token_accuracy": 0.7381949424743652,
+      "num_tokens": 89558352.0,
+      "step": 38740
+    },
+    {
+      "epoch": 0.3552764279820299,
+      "learning_rate": 0.00012894654808838363,
+      "loss": 0.8768,
+      "mean_token_accuracy": 0.7390536606311798,
+      "num_tokens": 89581438.0,
+      "step": 38750
+    },
+    {
+      "epoch": 0.3553681122215091,
+      "learning_rate": 0.00012892821124048777,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7429502248764038,
+      "num_tokens": 89604468.0,
+      "step": 38760
+    },
+    {
+      "epoch": 0.35545979646098835,
+      "learning_rate": 0.0001289098743925919,
+      "loss": 0.8377,
+      "mean_token_accuracy": 0.7495685458183289,
+      "num_tokens": 89627744.0,
+      "step": 38770
+    },
+    {
+      "epoch": 0.3555514807004676,
+      "learning_rate": 0.00012889153754469608,
+      "loss": 0.7933,
+      "mean_token_accuracy": 0.760120975971222,
+      "num_tokens": 89650348.0,
+      "step": 38780
+    },
+    {
+      "epoch": 0.3556431649399468,
+      "learning_rate": 0.00012887320069680021,
+      "loss": 0.8458,
+      "mean_token_accuracy": 0.7500900089740753,
+      "num_tokens": 89672996.0,
+      "step": 38790
+    },
+    {
+      "epoch": 0.35573484917942605,
+      "learning_rate": 0.00012885486384890438,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7481849253177643,
+      "num_tokens": 89695909.0,
+      "step": 38800
+    },
+    {
+      "epoch": 0.3558265334189053,
+      "learning_rate": 0.00012883652700100852,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.7435585975646972,
+      "num_tokens": 89719193.0,
+      "step": 38810
+    },
+    {
+      "epoch": 0.35591821765838455,
+      "learning_rate": 0.00012881819015311268,
+      "loss": 0.8693,
+      "mean_token_accuracy": 0.7380140542984008,
+      "num_tokens": 89742571.0,
+      "step": 38820
+    },
+    {
+      "epoch": 0.35600990189786375,
+      "learning_rate": 0.00012879985330521685,
+      "loss": 0.8536,
+      "mean_token_accuracy": 0.7438553154468537,
+      "num_tokens": 89765541.0,
+      "step": 38830
+    },
+    {
+      "epoch": 0.356101586137343,
+      "learning_rate": 0.000128781516457321,
+      "loss": 0.8287,
+      "mean_token_accuracy": 0.7518167436122895,
+      "num_tokens": 89788947.0,
+      "step": 38840
+    },
+    {
+      "epoch": 0.35619327037682225,
+      "learning_rate": 0.00012876317960942515,
+      "loss": 0.8549,
+      "mean_token_accuracy": 0.7409891963005066,
+      "num_tokens": 89812581.0,
+      "step": 38850
+    },
+    {
+      "epoch": 0.35628495461630144,
+      "learning_rate": 0.0001287448427615293,
+      "loss": 0.8367,
+      "mean_token_accuracy": 0.7552560925483703,
+      "num_tokens": 89835979.0,
+      "step": 38860
+    },
+    {
+      "epoch": 0.3563766388557807,
+      "learning_rate": 0.00012872650591363346,
+      "loss": 0.8899,
+      "mean_token_accuracy": 0.738469660282135,
+      "num_tokens": 89859430.0,
+      "step": 38870
+    },
+    {
+      "epoch": 0.35646832309525994,
+      "learning_rate": 0.00012870816906573762,
+      "loss": 0.8423,
+      "mean_token_accuracy": 0.7447143912315368,
+      "num_tokens": 89881839.0,
+      "step": 38880
+    },
+    {
+      "epoch": 0.35656000733473914,
+      "learning_rate": 0.00012868983221784176,
+      "loss": 0.8659,
+      "mean_token_accuracy": 0.7414678514003754,
+      "num_tokens": 89905307.0,
+      "step": 38890
+    },
+    {
+      "epoch": 0.3566516915742184,
+      "learning_rate": 0.00012867149536994592,
+      "loss": 0.8785,
+      "mean_token_accuracy": 0.7430894374847412,
+      "num_tokens": 89927163.0,
+      "step": 38900
+    },
+    {
+      "epoch": 0.35674337581369764,
+      "learning_rate": 0.00012865315852205006,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7401298344135284,
+      "num_tokens": 89949534.0,
+      "step": 38910
+    },
+    {
+      "epoch": 0.35683506005317683,
+      "learning_rate": 0.00012863482167415423,
+      "loss": 0.8889,
+      "mean_token_accuracy": 0.732297396659851,
+      "num_tokens": 89972443.0,
+      "step": 38920
+    },
+    {
+      "epoch": 0.3569267442926561,
+      "learning_rate": 0.00012861648482625837,
+      "loss": 0.8917,
+      "mean_token_accuracy": 0.7350900173187256,
+      "num_tokens": 89996057.0,
+      "step": 38930
+    },
+    {
+      "epoch": 0.35701842853213533,
+      "learning_rate": 0.0001285981479783625,
+      "loss": 0.8232,
+      "mean_token_accuracy": 0.75139200091362,
+      "num_tokens": 90019554.0,
+      "step": 38940
+    },
+    {
+      "epoch": 0.3571101127716146,
+      "learning_rate": 0.0001285798111304667,
+      "loss": 0.8193,
+      "mean_token_accuracy": 0.7521109402179718,
+      "num_tokens": 90042731.0,
+      "step": 38950
+    },
+    {
+      "epoch": 0.3572017970110938,
+      "learning_rate": 0.00012856147428257084,
+      "loss": 0.8265,
+      "mean_token_accuracy": 0.7502789199352264,
+      "num_tokens": 90066579.0,
+      "step": 38960
+    },
+    {
+      "epoch": 0.35729348125057303,
+      "learning_rate": 0.00012854313743467497,
+      "loss": 0.8414,
+      "mean_token_accuracy": 0.7500373899936676,
+      "num_tokens": 90089836.0,
+      "step": 38970
+    },
+    {
+      "epoch": 0.3573851654900523,
+      "learning_rate": 0.00012852480058677914,
+      "loss": 0.8224,
+      "mean_token_accuracy": 0.7452053368091583,
+      "num_tokens": 90113076.0,
+      "step": 38980
+    },
+    {
+      "epoch": 0.3574768497295315,
+      "learning_rate": 0.00012850646373888328,
+      "loss": 0.8497,
+      "mean_token_accuracy": 0.7411587238311768,
+      "num_tokens": 90135710.0,
+      "step": 38990
+    },
+    {
+      "epoch": 0.3575685339690107,
+      "learning_rate": 0.00012848812689098744,
+      "loss": 0.8635,
+      "mean_token_accuracy": 0.746865326166153,
+      "num_tokens": 90158250.0,
+      "step": 39000
+    },
+    {
+      "epoch": 0.35766021820849,
+      "learning_rate": 0.0001284697900430916,
+      "loss": 0.8973,
+      "mean_token_accuracy": 0.7384992718696595,
+      "num_tokens": 90180860.0,
+      "step": 39010
+    },
+    {
+      "epoch": 0.3577519024479692,
+      "learning_rate": 0.00012845145319519575,
+      "loss": 0.8197,
+      "mean_token_accuracy": 0.7542255163192749,
+      "num_tokens": 90205006.0,
+      "step": 39020
+    },
+    {
+      "epoch": 0.3578435866874484,
+      "learning_rate": 0.0001284331163472999,
+      "loss": 0.8778,
+      "mean_token_accuracy": 0.7382643461227417,
+      "num_tokens": 90228296.0,
+      "step": 39030
+    },
+    {
+      "epoch": 0.3579352709269277,
+      "learning_rate": 0.00012841477949940405,
+      "loss": 0.8675,
+      "mean_token_accuracy": 0.7422235667705536,
+      "num_tokens": 90251008.0,
+      "step": 39040
+    },
+    {
+      "epoch": 0.35802695516640687,
+      "learning_rate": 0.00012839644265150822,
+      "loss": 0.8676,
+      "mean_token_accuracy": 0.7396080613136291,
+      "num_tokens": 90273592.0,
+      "step": 39050
+    },
+    {
+      "epoch": 0.3581186394058861,
+      "learning_rate": 0.00012837810580361235,
+      "loss": 0.8192,
+      "mean_token_accuracy": 0.7484216511249542,
+      "num_tokens": 90295922.0,
+      "step": 39060
+    },
+    {
+      "epoch": 0.35821032364536537,
+      "learning_rate": 0.00012835976895571652,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7439599335193634,
+      "num_tokens": 90319196.0,
+      "step": 39070
+    },
+    {
+      "epoch": 0.3583020078848446,
+      "learning_rate": 0.00012834143210782069,
+      "loss": 0.8107,
+      "mean_token_accuracy": 0.7551554799079895,
+      "num_tokens": 90342472.0,
+      "step": 39080
+    },
+    {
+      "epoch": 0.3583936921243238,
+      "learning_rate": 0.00012832309525992482,
+      "loss": 0.8435,
+      "mean_token_accuracy": 0.7491034090518951,
+      "num_tokens": 90365535.0,
+      "step": 39090
+    },
+    {
+      "epoch": 0.35848537636380307,
+      "learning_rate": 0.000128304758412029,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.7451542317867279,
+      "num_tokens": 90389256.0,
+      "step": 39100
+    },
+    {
+      "epoch": 0.3585770606032823,
+      "learning_rate": 0.00012828642156413313,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7479837834835052,
+      "num_tokens": 90412471.0,
+      "step": 39110
+    },
+    {
+      "epoch": 0.3586687448427615,
+      "learning_rate": 0.0001282680847162373,
+      "loss": 0.8556,
+      "mean_token_accuracy": 0.7468072712421417,
+      "num_tokens": 90435449.0,
+      "step": 39120
+    },
+    {
+      "epoch": 0.35876042908224076,
+      "learning_rate": 0.00012824974786834143,
+      "loss": 0.8422,
+      "mean_token_accuracy": 0.7487893998622894,
+      "num_tokens": 90458550.0,
+      "step": 39130
+    },
+    {
+      "epoch": 0.35885211332172,
+      "learning_rate": 0.0001282314110204456,
+      "loss": 0.8225,
+      "mean_token_accuracy": 0.7485275208950043,
+      "num_tokens": 90481708.0,
+      "step": 39140
+    },
+    {
+      "epoch": 0.3589437975611992,
+      "learning_rate": 0.00012821307417254976,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7405048310756683,
+      "num_tokens": 90504383.0,
+      "step": 39150
+    },
+    {
+      "epoch": 0.35903548180067846,
+      "learning_rate": 0.0001281947373246539,
+      "loss": 0.8815,
+      "mean_token_accuracy": 0.7343416690826416,
+      "num_tokens": 90526789.0,
+      "step": 39160
+    },
+    {
+      "epoch": 0.3591271660401577,
+      "learning_rate": 0.00012817640047675804,
+      "loss": 0.8402,
+      "mean_token_accuracy": 0.7461209952831268,
+      "num_tokens": 90549569.0,
+      "step": 39170
+    },
+    {
+      "epoch": 0.3592188502796369,
+      "learning_rate": 0.0001281580636288622,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7476036727428437,
+      "num_tokens": 90572528.0,
+      "step": 39180
+    },
+    {
+      "epoch": 0.35931053451911615,
+      "learning_rate": 0.00012813972678096634,
+      "loss": 0.8527,
+      "mean_token_accuracy": 0.7453954100608826,
+      "num_tokens": 90595236.0,
+      "step": 39190
+    },
+    {
+      "epoch": 0.3594022187585954,
+      "learning_rate": 0.0001281213899330705,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7483029246330262,
+      "num_tokens": 90618319.0,
+      "step": 39200
+    },
+    {
+      "epoch": 0.35949390299807465,
+      "learning_rate": 0.00012810305308517467,
+      "loss": 0.8381,
+      "mean_token_accuracy": 0.7419435441493988,
+      "num_tokens": 90642358.0,
+      "step": 39210
+    },
+    {
+      "epoch": 0.35958558723755385,
+      "learning_rate": 0.0001280847162372788,
+      "loss": 0.824,
+      "mean_token_accuracy": 0.7481161892414093,
+      "num_tokens": 90665259.0,
+      "step": 39220
+    },
+    {
+      "epoch": 0.3596772714770331,
+      "learning_rate": 0.00012806637938938298,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.7428594410419465,
+      "num_tokens": 90688499.0,
+      "step": 39230
+    },
+    {
+      "epoch": 0.35976895571651235,
+      "learning_rate": 0.00012804804254148712,
+      "loss": 0.8357,
+      "mean_token_accuracy": 0.7452928125858307,
+      "num_tokens": 90711019.0,
+      "step": 39240
+    },
+    {
+      "epoch": 0.35986063995599155,
+      "learning_rate": 0.00012802970569359128,
+      "loss": 0.8387,
+      "mean_token_accuracy": 0.7483808994293213,
+      "num_tokens": 90734314.0,
+      "step": 39250
+    },
+    {
+      "epoch": 0.3599523241954708,
+      "learning_rate": 0.00012801136884569542,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7443465471267701,
+      "num_tokens": 90756865.0,
+      "step": 39260
+    },
+    {
+      "epoch": 0.36004400843495005,
+      "learning_rate": 0.00012799303199779958,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7470487833023072,
+      "num_tokens": 90779856.0,
+      "step": 39270
+    },
+    {
+      "epoch": 0.36013569267442924,
+      "learning_rate": 0.00012797469514990375,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.745301365852356,
+      "num_tokens": 90802740.0,
+      "step": 39280
+    },
+    {
+      "epoch": 0.3602273769139085,
+      "learning_rate": 0.0001279563583020079,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.7416181266307831,
+      "num_tokens": 90825374.0,
+      "step": 39290
+    },
+    {
+      "epoch": 0.36031906115338774,
+      "learning_rate": 0.00012793802145411205,
+      "loss": 0.8394,
+      "mean_token_accuracy": 0.7457209348678588,
+      "num_tokens": 90848466.0,
+      "step": 39300
+    },
+    {
+      "epoch": 0.360410745392867,
+      "learning_rate": 0.0001279196846062162,
+      "loss": 0.8263,
+      "mean_token_accuracy": 0.7445708155632019,
+      "num_tokens": 90871587.0,
+      "step": 39310
+    },
+    {
+      "epoch": 0.3605024296323462,
+      "learning_rate": 0.00012790134775832036,
+      "loss": 0.8219,
+      "mean_token_accuracy": 0.7543377220630646,
+      "num_tokens": 90894651.0,
+      "step": 39320
+    },
+    {
+      "epoch": 0.36059411387182544,
+      "learning_rate": 0.0001278830109104245,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7494737327098846,
+      "num_tokens": 90916604.0,
+      "step": 39330
+    },
+    {
+      "epoch": 0.3606857981113047,
+      "learning_rate": 0.00012786467406252866,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.7450239300727844,
+      "num_tokens": 90940320.0,
+      "step": 39340
+    },
+    {
+      "epoch": 0.3607774823507839,
+      "learning_rate": 0.00012784633721463283,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7413978159427643,
+      "num_tokens": 90963151.0,
+      "step": 39350
+    },
+    {
+      "epoch": 0.36086916659026314,
+      "learning_rate": 0.00012782800036673697,
+      "loss": 0.8646,
+      "mean_token_accuracy": 0.7412473142147065,
+      "num_tokens": 90985936.0,
+      "step": 39360
+    },
+    {
+      "epoch": 0.3609608508297424,
+      "learning_rate": 0.0001278096635188411,
+      "loss": 0.877,
+      "mean_token_accuracy": 0.7413740694522858,
+      "num_tokens": 91009272.0,
+      "step": 39370
+    },
+    {
+      "epoch": 0.3610525350692216,
+      "learning_rate": 0.00012779132667094527,
+      "loss": 0.8859,
+      "mean_token_accuracy": 0.7399205446243287,
+      "num_tokens": 91031775.0,
+      "step": 39380
+    },
+    {
+      "epoch": 0.36114421930870083,
+      "learning_rate": 0.0001277729898230494,
+      "loss": 0.8553,
+      "mean_token_accuracy": 0.7363078713417053,
+      "num_tokens": 91054082.0,
+      "step": 39390
+    },
+    {
+      "epoch": 0.3612359035481801,
+      "learning_rate": 0.0001277546529751536,
+      "loss": 0.8494,
+      "mean_token_accuracy": 0.7414499998092652,
+      "num_tokens": 91077776.0,
+      "step": 39400
+    },
+    {
+      "epoch": 0.3613275877876593,
+      "learning_rate": 0.00012773631612725774,
+      "loss": 0.8359,
+      "mean_token_accuracy": 0.7497224688529969,
+      "num_tokens": 91100582.0,
+      "step": 39410
+    },
+    {
+      "epoch": 0.3614192720271385,
+      "learning_rate": 0.00012771797927936188,
+      "loss": 0.8398,
+      "mean_token_accuracy": 0.7529937386512756,
+      "num_tokens": 91123815.0,
+      "step": 39420
+    },
+    {
+      "epoch": 0.3615109562666178,
+      "learning_rate": 0.00012769964243146604,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7540438890457153,
+      "num_tokens": 91146444.0,
+      "step": 39430
+    },
+    {
+      "epoch": 0.36160264050609703,
+      "learning_rate": 0.00012768130558357018,
+      "loss": 0.8569,
+      "mean_token_accuracy": 0.7441222786903381,
+      "num_tokens": 91168597.0,
+      "step": 39440
+    },
+    {
+      "epoch": 0.3616943247455762,
+      "learning_rate": 0.00012766296873567435,
+      "loss": 0.8169,
+      "mean_token_accuracy": 0.7493405520915986,
+      "num_tokens": 91192109.0,
+      "step": 39450
+    },
+    {
+      "epoch": 0.3617860089850555,
+      "learning_rate": 0.00012764463188777848,
+      "loss": 0.9045,
+      "mean_token_accuracy": 0.7368192732334137,
+      "num_tokens": 91215439.0,
+      "step": 39460
+    },
+    {
+      "epoch": 0.3618776932245347,
+      "learning_rate": 0.00012762629503988265,
+      "loss": 0.8497,
+      "mean_token_accuracy": 0.7431175529956817,
+      "num_tokens": 91238840.0,
+      "step": 39470
+    },
+    {
+      "epoch": 0.3619693774640139,
+      "learning_rate": 0.00012760795819198681,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7466048240661621,
+      "num_tokens": 91261077.0,
+      "step": 39480
+    },
+    {
+      "epoch": 0.36206106170349317,
+      "learning_rate": 0.00012758962134409095,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7444063425064087,
+      "num_tokens": 91283721.0,
+      "step": 39490
+    },
+    {
+      "epoch": 0.3621527459429724,
+      "learning_rate": 0.00012757128449619512,
+      "loss": 0.8679,
+      "mean_token_accuracy": 0.7448558211326599,
+      "num_tokens": 91306497.0,
+      "step": 39500
+    },
+    {
+      "epoch": 0.3622444301824516,
+      "learning_rate": 0.00012755294764829926,
+      "loss": 0.8287,
+      "mean_token_accuracy": 0.7477375030517578,
+      "num_tokens": 91329031.0,
+      "step": 39510
+    },
+    {
+      "epoch": 0.36233611442193087,
+      "learning_rate": 0.00012753461080040342,
+      "loss": 0.852,
+      "mean_token_accuracy": 0.741021329164505,
+      "num_tokens": 91352555.0,
+      "step": 39520
+    },
+    {
+      "epoch": 0.3624277986614101,
+      "learning_rate": 0.0001275162739525076,
+      "loss": 0.7893,
+      "mean_token_accuracy": 0.7620260953903198,
+      "num_tokens": 91375262.0,
+      "step": 39530
+    },
+    {
+      "epoch": 0.3625194829008893,
+      "learning_rate": 0.00012749793710461173,
+      "loss": 0.8846,
+      "mean_token_accuracy": 0.7371660232543945,
+      "num_tokens": 91398662.0,
+      "step": 39540
+    },
+    {
+      "epoch": 0.36261116714036856,
+      "learning_rate": 0.0001274796002567159,
+      "loss": 0.8849,
+      "mean_token_accuracy": 0.7396804332733155,
+      "num_tokens": 91422616.0,
+      "step": 39550
+    },
+    {
+      "epoch": 0.3627028513798478,
+      "learning_rate": 0.00012746126340882003,
+      "loss": 0.8744,
+      "mean_token_accuracy": 0.7417828977108002,
+      "num_tokens": 91445879.0,
+      "step": 39560
+    },
+    {
+      "epoch": 0.36279453561932706,
+      "learning_rate": 0.00012744292656092417,
+      "loss": 0.8211,
+      "mean_token_accuracy": 0.7551993131637573,
+      "num_tokens": 91469407.0,
+      "step": 39570
+    },
+    {
+      "epoch": 0.36288621985880626,
+      "learning_rate": 0.00012742458971302833,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7422672986984253,
+      "num_tokens": 91493572.0,
+      "step": 39580
+    },
+    {
+      "epoch": 0.3629779040982855,
+      "learning_rate": 0.00012740625286513247,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7383472084999084,
+      "num_tokens": 91517109.0,
+      "step": 39590
+    },
+    {
+      "epoch": 0.36306958833776476,
+      "learning_rate": 0.00012738791601723666,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.7447839200496673,
+      "num_tokens": 91540272.0,
+      "step": 39600
+    },
+    {
+      "epoch": 0.36316127257724395,
+      "learning_rate": 0.0001273695791693408,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.750486308336258,
+      "num_tokens": 91563517.0,
+      "step": 39610
+    },
+    {
+      "epoch": 0.3632529568167232,
+      "learning_rate": 0.00012735124232144494,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.7488204300403595,
+      "num_tokens": 91586299.0,
+      "step": 39620
+    },
+    {
+      "epoch": 0.36334464105620246,
+      "learning_rate": 0.0001273329054735491,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7387410640716553,
+      "num_tokens": 91609314.0,
+      "step": 39630
+    },
+    {
+      "epoch": 0.36343632529568165,
+      "learning_rate": 0.00012731456862565324,
+      "loss": 0.8755,
+      "mean_token_accuracy": 0.7395784020423889,
+      "num_tokens": 91631534.0,
+      "step": 39640
+    },
+    {
+      "epoch": 0.3635280095351609,
+      "learning_rate": 0.0001272962317777574,
+      "loss": 0.8138,
+      "mean_token_accuracy": 0.7593177497386933,
+      "num_tokens": 91654197.0,
+      "step": 39650
+    },
+    {
+      "epoch": 0.36361969377464015,
+      "learning_rate": 0.00012727789492986155,
+      "loss": 0.901,
+      "mean_token_accuracy": 0.7381822168827057,
+      "num_tokens": 91677741.0,
+      "step": 39660
+    },
+    {
+      "epoch": 0.36371137801411935,
+      "learning_rate": 0.00012725955808196571,
+      "loss": 0.8208,
+      "mean_token_accuracy": 0.7579244554042817,
+      "num_tokens": 91700968.0,
+      "step": 39670
+    },
+    {
+      "epoch": 0.3638030622535986,
+      "learning_rate": 0.00012724122123406988,
+      "loss": 0.8199,
+      "mean_token_accuracy": 0.7514577746391297,
+      "num_tokens": 91723811.0,
+      "step": 39680
+    },
+    {
+      "epoch": 0.36389474649307785,
+      "learning_rate": 0.00012722288438617402,
+      "loss": 0.849,
+      "mean_token_accuracy": 0.7445736348628997,
+      "num_tokens": 91747070.0,
+      "step": 39690
+    },
+    {
+      "epoch": 0.3639864307325571,
+      "learning_rate": 0.00012720454753827818,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7425487399101257,
+      "num_tokens": 91770766.0,
+      "step": 39700
+    },
+    {
+      "epoch": 0.3640781149720363,
+      "learning_rate": 0.00012718621069038232,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7428417444229126,
+      "num_tokens": 91793856.0,
+      "step": 39710
+    },
+    {
+      "epoch": 0.36416979921151554,
+      "learning_rate": 0.0001271678738424865,
+      "loss": 0.8573,
+      "mean_token_accuracy": 0.7450142085552216,
+      "num_tokens": 91816416.0,
+      "step": 39720
+    },
+    {
+      "epoch": 0.3642614834509948,
+      "learning_rate": 0.00012714953699459065,
+      "loss": 0.8599,
+      "mean_token_accuracy": 0.7451660454273223,
+      "num_tokens": 91839904.0,
+      "step": 39730
+    },
+    {
+      "epoch": 0.364353167690474,
+      "learning_rate": 0.0001271312001466948,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7544918477535247,
+      "num_tokens": 91862889.0,
+      "step": 39740
+    },
+    {
+      "epoch": 0.36444485192995324,
+      "learning_rate": 0.00012711286329879896,
+      "loss": 0.8108,
+      "mean_token_accuracy": 0.7567556619644165,
+      "num_tokens": 91886440.0,
+      "step": 39750
+    },
+    {
+      "epoch": 0.3645365361694325,
+      "learning_rate": 0.0001270945264509031,
+      "loss": 0.8637,
+      "mean_token_accuracy": 0.741011118888855,
+      "num_tokens": 91908793.0,
+      "step": 39760
+    },
+    {
+      "epoch": 0.3646282204089117,
+      "learning_rate": 0.00012707618960300723,
+      "loss": 0.8299,
+      "mean_token_accuracy": 0.7522927403450013,
+      "num_tokens": 91931469.0,
+      "step": 39770
+    },
+    {
+      "epoch": 0.36471990464839094,
+      "learning_rate": 0.0001270578527551114,
+      "loss": 0.8267,
+      "mean_token_accuracy": 0.7494844019412994,
+      "num_tokens": 91955670.0,
+      "step": 39780
+    },
+    {
+      "epoch": 0.3648115888878702,
+      "learning_rate": 0.00012703951590721554,
+      "loss": 0.8347,
+      "mean_token_accuracy": 0.7501931846141815,
+      "num_tokens": 91977964.0,
+      "step": 39790
+    },
+    {
+      "epoch": 0.3649032731273494,
+      "learning_rate": 0.00012702117905931973,
+      "loss": 0.8576,
+      "mean_token_accuracy": 0.7422442376613617,
+      "num_tokens": 92000707.0,
+      "step": 39800
+    },
+    {
+      "epoch": 0.36499495736682863,
+      "learning_rate": 0.00012700284221142387,
+      "loss": 0.8159,
+      "mean_token_accuracy": 0.7515161991119385,
+      "num_tokens": 92023331.0,
+      "step": 39810
+    },
+    {
+      "epoch": 0.3650866416063079,
+      "learning_rate": 0.000126984505363528,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7397204816341401,
+      "num_tokens": 92046328.0,
+      "step": 39820
+    },
+    {
+      "epoch": 0.36517832584578713,
+      "learning_rate": 0.00012696616851563217,
+      "loss": 0.8626,
+      "mean_token_accuracy": 0.7454255521297455,
+      "num_tokens": 92069562.0,
+      "step": 39830
+    },
+    {
+      "epoch": 0.36527001008526633,
+      "learning_rate": 0.0001269478316677363,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.747641122341156,
+      "num_tokens": 92091851.0,
+      "step": 39840
+    },
+    {
+      "epoch": 0.3653616943247456,
+      "learning_rate": 0.00012692949481984047,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7518446266651153,
+      "num_tokens": 92115268.0,
+      "step": 39850
+    },
+    {
+      "epoch": 0.36545337856422483,
+      "learning_rate": 0.00012691115797194464,
+      "loss": 0.8191,
+      "mean_token_accuracy": 0.7559412837028503,
+      "num_tokens": 92138997.0,
+      "step": 39860
+    },
+    {
+      "epoch": 0.365545062803704,
+      "learning_rate": 0.00012689282112404878,
+      "loss": 0.8185,
+      "mean_token_accuracy": 0.7551422715187073,
+      "num_tokens": 92162308.0,
+      "step": 39870
+    },
+    {
+      "epoch": 0.3656367470431833,
+      "learning_rate": 0.00012687448427615294,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.742394644021988,
+      "num_tokens": 92186312.0,
+      "step": 39880
+    },
+    {
+      "epoch": 0.3657284312826625,
+      "learning_rate": 0.00012685614742825708,
+      "loss": 0.8675,
+      "mean_token_accuracy": 0.743462860584259,
+      "num_tokens": 92209528.0,
+      "step": 39890
+    },
+    {
+      "epoch": 0.3658201155221417,
+      "learning_rate": 0.00012683781058036125,
+      "loss": 0.8764,
+      "mean_token_accuracy": 0.7435184717178345,
+      "num_tokens": 92232266.0,
+      "step": 39900
+    },
+    {
+      "epoch": 0.36591179976162097,
+      "learning_rate": 0.00012681947373246539,
+      "loss": 0.8281,
+      "mean_token_accuracy": 0.7465691208839417,
+      "num_tokens": 92255286.0,
+      "step": 39910
+    },
+    {
+      "epoch": 0.3660034840011002,
+      "learning_rate": 0.00012680113688456955,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.7478818774223328,
+      "num_tokens": 92278139.0,
+      "step": 39920
+    },
+    {
+      "epoch": 0.36609516824057947,
+      "learning_rate": 0.00012678280003667372,
+      "loss": 0.8633,
+      "mean_token_accuracy": 0.7438245713710785,
+      "num_tokens": 92301488.0,
+      "step": 39930
+    },
+    {
+      "epoch": 0.36618685248005867,
+      "learning_rate": 0.00012676446318877785,
+      "loss": 0.8329,
+      "mean_token_accuracy": 0.7494479298591614,
+      "num_tokens": 92324312.0,
+      "step": 39940
+    },
+    {
+      "epoch": 0.3662785367195379,
+      "learning_rate": 0.00012674612634088202,
+      "loss": 0.856,
+      "mean_token_accuracy": 0.7483045279979705,
+      "num_tokens": 92347224.0,
+      "step": 39950
+    },
+    {
+      "epoch": 0.36637022095901717,
+      "learning_rate": 0.00012672778949298616,
+      "loss": 0.8827,
+      "mean_token_accuracy": 0.7397355735301971,
+      "num_tokens": 92369762.0,
+      "step": 39960
+    },
+    {
+      "epoch": 0.36646190519849636,
+      "learning_rate": 0.0001267094526450903,
+      "loss": 0.838,
+      "mean_token_accuracy": 0.7478506326675415,
+      "num_tokens": 92393210.0,
+      "step": 39970
+    },
+    {
+      "epoch": 0.3665535894379756,
+      "learning_rate": 0.00012669111579719446,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.7567574977874756,
+      "num_tokens": 92416100.0,
+      "step": 39980
+    },
+    {
+      "epoch": 0.36664527367745486,
+      "learning_rate": 0.00012667277894929863,
+      "loss": 0.8177,
+      "mean_token_accuracy": 0.7546796023845672,
+      "num_tokens": 92439552.0,
+      "step": 39990
+    },
+    {
+      "epoch": 0.36673695791693406,
+      "learning_rate": 0.0001266544421014028,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7529680073261261,
+      "num_tokens": 92462410.0,
+      "step": 40000
+    },
+    {
+      "epoch": 0.3668286421564133,
+      "learning_rate": 0.00012663610525350693,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.7406797587871552,
+      "num_tokens": 92485957.0,
+      "step": 40010
+    },
+    {
+      "epoch": 0.36692032639589256,
+      "learning_rate": 0.00012661776840561107,
+      "loss": 0.8346,
+      "mean_token_accuracy": 0.7488115429878235,
+      "num_tokens": 92508695.0,
+      "step": 40020
+    },
+    {
+      "epoch": 0.36701201063537175,
+      "learning_rate": 0.00012659943155771524,
+      "loss": 0.8482,
+      "mean_token_accuracy": 0.7431809842586518,
+      "num_tokens": 92532395.0,
+      "step": 40030
+    },
+    {
+      "epoch": 0.367103694874851,
+      "learning_rate": 0.00012658109470981937,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.7561107218265534,
+      "num_tokens": 92555286.0,
+      "step": 40040
+    },
+    {
+      "epoch": 0.36719537911433026,
+      "learning_rate": 0.00012656275786192354,
+      "loss": 0.8408,
+      "mean_token_accuracy": 0.7459770798683166,
+      "num_tokens": 92577634.0,
+      "step": 40050
+    },
+    {
+      "epoch": 0.3672870633538095,
+      "learning_rate": 0.0001265444210140277,
+      "loss": 0.8849,
+      "mean_token_accuracy": 0.7432604789733886,
+      "num_tokens": 92600191.0,
+      "step": 40060
+    },
+    {
+      "epoch": 0.3673787475932887,
+      "learning_rate": 0.00012652608416613184,
+      "loss": 0.8035,
+      "mean_token_accuracy": 0.7561726629734039,
+      "num_tokens": 92623965.0,
+      "step": 40070
+    },
+    {
+      "epoch": 0.36747043183276795,
+      "learning_rate": 0.000126507747318236,
+      "loss": 0.8311,
+      "mean_token_accuracy": 0.7523778915405274,
+      "num_tokens": 92647741.0,
+      "step": 40080
+    },
+    {
+      "epoch": 0.3675621160722472,
+      "learning_rate": 0.00012648941047034015,
+      "loss": 0.821,
+      "mean_token_accuracy": 0.7464438676834106,
+      "num_tokens": 92670825.0,
+      "step": 40090
+    },
+    {
+      "epoch": 0.3676538003117264,
+      "learning_rate": 0.0001264710736224443,
+      "loss": 0.8582,
+      "mean_token_accuracy": 0.7455631375312806,
+      "num_tokens": 92693320.0,
+      "step": 40100
+    },
+    {
+      "epoch": 0.36774548455120565,
+      "learning_rate": 0.00012645273677454845,
+      "loss": 0.7979,
+      "mean_token_accuracy": 0.7619030833244324,
+      "num_tokens": 92715172.0,
+      "step": 40110
+    },
+    {
+      "epoch": 0.3678371687906849,
+      "learning_rate": 0.00012643439992665262,
+      "loss": 0.8307,
+      "mean_token_accuracy": 0.7564894676208496,
+      "num_tokens": 92738453.0,
+      "step": 40120
+    },
+    {
+      "epoch": 0.3679288530301641,
+      "learning_rate": 0.00012641606307875678,
+      "loss": 0.8251,
+      "mean_token_accuracy": 0.7496735274791717,
+      "num_tokens": 92761941.0,
+      "step": 40130
+    },
+    {
+      "epoch": 0.36802053726964334,
+      "learning_rate": 0.00012639772623086092,
+      "loss": 0.8662,
+      "mean_token_accuracy": 0.7428528904914856,
+      "num_tokens": 92784551.0,
+      "step": 40140
+    },
+    {
+      "epoch": 0.3681122215091226,
+      "learning_rate": 0.00012637938938296508,
+      "loss": 0.8338,
+      "mean_token_accuracy": 0.7540151894092559,
+      "num_tokens": 92806983.0,
+      "step": 40150
+    },
+    {
+      "epoch": 0.3682039057486018,
+      "learning_rate": 0.00012636105253506922,
+      "loss": 0.8424,
+      "mean_token_accuracy": 0.7459520876407624,
+      "num_tokens": 92829925.0,
+      "step": 40160
+    },
+    {
+      "epoch": 0.36829558998808104,
+      "learning_rate": 0.00012634271568717336,
+      "loss": 0.8298,
+      "mean_token_accuracy": 0.7497051417827606,
+      "num_tokens": 92853052.0,
+      "step": 40170
+    },
+    {
+      "epoch": 0.3683872742275603,
+      "learning_rate": 0.00012632437883927753,
+      "loss": 0.8425,
+      "mean_token_accuracy": 0.7553964674472808,
+      "num_tokens": 92875672.0,
+      "step": 40180
+    },
+    {
+      "epoch": 0.36847895846703954,
+      "learning_rate": 0.0001263060419913817,
+      "loss": 0.8324,
+      "mean_token_accuracy": 0.7483945846557617,
+      "num_tokens": 92898021.0,
+      "step": 40190
+    },
+    {
+      "epoch": 0.36857064270651874,
+      "learning_rate": 0.00012628770514348586,
+      "loss": 0.7824,
+      "mean_token_accuracy": 0.7633194088935852,
+      "num_tokens": 92921031.0,
+      "step": 40200
+    },
+    {
+      "epoch": 0.368662326945998,
+      "learning_rate": 0.00012626936829559,
+      "loss": 0.8493,
+      "mean_token_accuracy": 0.7487386226654053,
+      "num_tokens": 92943211.0,
+      "step": 40210
+    },
+    {
+      "epoch": 0.36875401118547724,
+      "learning_rate": 0.00012625103144769413,
+      "loss": 0.8353,
+      "mean_token_accuracy": 0.7439074158668518,
+      "num_tokens": 92965929.0,
+      "step": 40220
+    },
+    {
+      "epoch": 0.36884569542495643,
+      "learning_rate": 0.0001262326945997983,
+      "loss": 0.8342,
+      "mean_token_accuracy": 0.7512854933738708,
+      "num_tokens": 92989508.0,
+      "step": 40230
+    },
+    {
+      "epoch": 0.3689373796644357,
+      "learning_rate": 0.00012621435775190244,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7368911981582642,
+      "num_tokens": 93012147.0,
+      "step": 40240
+    },
+    {
+      "epoch": 0.36902906390391493,
+      "learning_rate": 0.00012619602090400663,
+      "loss": 0.8342,
+      "mean_token_accuracy": 0.7457172989845275,
+      "num_tokens": 93035091.0,
+      "step": 40250
+    },
+    {
+      "epoch": 0.36912074814339413,
+      "learning_rate": 0.00012617768405611077,
+      "loss": 0.8514,
+      "mean_token_accuracy": 0.7521363794803619,
+      "num_tokens": 93058530.0,
+      "step": 40260
+    },
+    {
+      "epoch": 0.3692124323828734,
+      "learning_rate": 0.0001261593472082149,
+      "loss": 0.813,
+      "mean_token_accuracy": 0.7531933188438416,
+      "num_tokens": 93081509.0,
+      "step": 40270
+    },
+    {
+      "epoch": 0.36930411662235263,
+      "learning_rate": 0.00012614101036031907,
+      "loss": 0.8282,
+      "mean_token_accuracy": 0.7440933346748352,
+      "num_tokens": 93103968.0,
+      "step": 40280
+    },
+    {
+      "epoch": 0.3693958008618318,
+      "learning_rate": 0.0001261226735124232,
+      "loss": 0.8138,
+      "mean_token_accuracy": 0.753024297952652,
+      "num_tokens": 93126975.0,
+      "step": 40290
+    },
+    {
+      "epoch": 0.3694874851013111,
+      "learning_rate": 0.00012610433666452738,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7481602966785431,
+      "num_tokens": 93150114.0,
+      "step": 40300
+    },
+    {
+      "epoch": 0.3695791693407903,
+      "learning_rate": 0.00012608599981663151,
+      "loss": 0.9202,
+      "mean_token_accuracy": 0.7269593954086304,
+      "num_tokens": 93173992.0,
+      "step": 40310
+    },
+    {
+      "epoch": 0.3696708535802696,
+      "learning_rate": 0.00012606766296873568,
+      "loss": 0.8513,
+      "mean_token_accuracy": 0.7426384091377258,
+      "num_tokens": 93197403.0,
+      "step": 40320
+    },
+    {
+      "epoch": 0.36976253781974877,
+      "learning_rate": 0.00012604932612083985,
+      "loss": 0.859,
+      "mean_token_accuracy": 0.7421603322029113,
+      "num_tokens": 93220655.0,
+      "step": 40330
+    },
+    {
+      "epoch": 0.369854222059228,
+      "learning_rate": 0.00012603098927294398,
+      "loss": 0.8047,
+      "mean_token_accuracy": 0.7526649475097656,
+      "num_tokens": 93243287.0,
+      "step": 40340
+    },
+    {
+      "epoch": 0.36994590629870727,
+      "learning_rate": 0.00012601265242504815,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.747111189365387,
+      "num_tokens": 93266655.0,
+      "step": 40350
+    },
+    {
+      "epoch": 0.37003759053818647,
+      "learning_rate": 0.0001259943155771523,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7481015443801879,
+      "num_tokens": 93289581.0,
+      "step": 40360
+    },
+    {
+      "epoch": 0.3701292747776657,
+      "learning_rate": 0.00012597597872925643,
+      "loss": 0.8106,
+      "mean_token_accuracy": 0.7533900260925293,
+      "num_tokens": 93312663.0,
+      "step": 40370
+    },
+    {
+      "epoch": 0.37022095901714497,
+      "learning_rate": 0.00012595764188136062,
+      "loss": 0.8559,
+      "mean_token_accuracy": 0.7490554809570312,
+      "num_tokens": 93335413.0,
+      "step": 40380
+    },
+    {
+      "epoch": 0.37031264325662416,
+      "learning_rate": 0.00012593930503346476,
+      "loss": 0.8592,
+      "mean_token_accuracy": 0.7416788399219513,
+      "num_tokens": 93358953.0,
+      "step": 40390
+    },
+    {
+      "epoch": 0.3704043274961034,
+      "learning_rate": 0.00012592096818556892,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.744015908241272,
+      "num_tokens": 93382002.0,
+      "step": 40400
+    },
+    {
+      "epoch": 0.37049601173558266,
+      "learning_rate": 0.00012590263133767306,
+      "loss": 0.8255,
+      "mean_token_accuracy": 0.75286785364151,
+      "num_tokens": 93405306.0,
+      "step": 40410
+    },
+    {
+      "epoch": 0.37058769597506186,
+      "learning_rate": 0.0001258842944897772,
+      "loss": 0.8513,
+      "mean_token_accuracy": 0.7488247215747833,
+      "num_tokens": 93429457.0,
+      "step": 40420
+    },
+    {
+      "epoch": 0.3706793802145411,
+      "learning_rate": 0.00012586595764188136,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7498377323150635,
+      "num_tokens": 93452354.0,
+      "step": 40430
+    },
+    {
+      "epoch": 0.37077106445402036,
+      "learning_rate": 0.0001258476207939855,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7410698115825654,
+      "num_tokens": 93474616.0,
+      "step": 40440
+    },
+    {
+      "epoch": 0.3708627486934996,
+      "learning_rate": 0.0001258292839460897,
+      "loss": 0.8004,
+      "mean_token_accuracy": 0.7547740161418914,
+      "num_tokens": 93498594.0,
+      "step": 40450
+    },
+    {
+      "epoch": 0.3709544329329788,
+      "learning_rate": 0.00012581094709819383,
+      "loss": 0.8481,
+      "mean_token_accuracy": 0.7415642082691193,
+      "num_tokens": 93521499.0,
+      "step": 40460
+    },
+    {
+      "epoch": 0.37104611717245806,
+      "learning_rate": 0.00012579261025029797,
+      "loss": 0.8715,
+      "mean_token_accuracy": 0.7353518903255463,
+      "num_tokens": 93544054.0,
+      "step": 40470
+    },
+    {
+      "epoch": 0.3711378014119373,
+      "learning_rate": 0.00012577427340240214,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.7446906924247741,
+      "num_tokens": 93567225.0,
+      "step": 40480
+    },
+    {
+      "epoch": 0.3712294856514165,
+      "learning_rate": 0.00012575593655450628,
+      "loss": 0.8502,
+      "mean_token_accuracy": 0.7465663194656372,
+      "num_tokens": 93590500.0,
+      "step": 40490
+    },
+    {
+      "epoch": 0.37132116989089575,
+      "learning_rate": 0.00012573759970661044,
+      "loss": 0.8437,
+      "mean_token_accuracy": 0.7428981006145478,
+      "num_tokens": 93613735.0,
+      "step": 40500
+    },
+    {
+      "epoch": 0.371412854130375,
+      "learning_rate": 0.0001257192628587146,
+      "loss": 0.8992,
+      "mean_token_accuracy": 0.7355528891086578,
+      "num_tokens": 93637224.0,
+      "step": 40510
+    },
+    {
+      "epoch": 0.3715045383698542,
+      "learning_rate": 0.00012570092601081874,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.742870831489563,
+      "num_tokens": 93660064.0,
+      "step": 40520
+    },
+    {
+      "epoch": 0.37159622260933345,
+      "learning_rate": 0.0001256825891629229,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.7555842518806457,
+      "num_tokens": 93683197.0,
+      "step": 40530
+    },
+    {
+      "epoch": 0.3716879068488127,
+      "learning_rate": 0.00012566425231502705,
+      "loss": 0.8352,
+      "mean_token_accuracy": 0.7442308604717255,
+      "num_tokens": 93706907.0,
+      "step": 40540
+    },
+    {
+      "epoch": 0.37177959108829195,
+      "learning_rate": 0.0001256459154671312,
+      "loss": 0.8294,
+      "mean_token_accuracy": 0.7532436072826385,
+      "num_tokens": 93730090.0,
+      "step": 40550
+    },
+    {
+      "epoch": 0.37187127532777114,
+      "learning_rate": 0.00012562757861923535,
+      "loss": 0.8732,
+      "mean_token_accuracy": 0.7416025876998902,
+      "num_tokens": 93753397.0,
+      "step": 40560
+    },
+    {
+      "epoch": 0.3719629595672504,
+      "learning_rate": 0.0001256092417713395,
+      "loss": 0.8535,
+      "mean_token_accuracy": 0.7419710040092469,
+      "num_tokens": 93776825.0,
+      "step": 40570
+    },
+    {
+      "epoch": 0.37205464380672965,
+      "learning_rate": 0.00012559090492344368,
+      "loss": 0.863,
+      "mean_token_accuracy": 0.7388189852237701,
+      "num_tokens": 93800186.0,
+      "step": 40580
+    },
+    {
+      "epoch": 0.37214632804620884,
+      "learning_rate": 0.00012557256807554782,
+      "loss": 0.8313,
+      "mean_token_accuracy": 0.7578201413154602,
+      "num_tokens": 93823162.0,
+      "step": 40590
+    },
+    {
+      "epoch": 0.3722380122856881,
+      "learning_rate": 0.00012555423122765199,
+      "loss": 0.8293,
+      "mean_token_accuracy": 0.7469756484031678,
+      "num_tokens": 93846024.0,
+      "step": 40600
+    },
+    {
+      "epoch": 0.37232969652516734,
+      "learning_rate": 0.00012553589437975612,
+      "loss": 0.8495,
+      "mean_token_accuracy": 0.7459544360637664,
+      "num_tokens": 93869454.0,
+      "step": 40610
+    },
+    {
+      "epoch": 0.37242138076464654,
+      "learning_rate": 0.00012551755753186026,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7407496690750122,
+      "num_tokens": 93892403.0,
+      "step": 40620
+    },
+    {
+      "epoch": 0.3725130650041258,
+      "learning_rate": 0.00012549922068396443,
+      "loss": 0.8914,
+      "mean_token_accuracy": 0.7309635579586029,
+      "num_tokens": 93915122.0,
+      "step": 40630
+    },
+    {
+      "epoch": 0.37260474924360504,
+      "learning_rate": 0.0001254808838360686,
+      "loss": 0.8264,
+      "mean_token_accuracy": 0.7473390102386475,
+      "num_tokens": 93938613.0,
+      "step": 40640
+    },
+    {
+      "epoch": 0.37269643348308423,
+      "learning_rate": 0.00012546254698817276,
+      "loss": 0.8518,
+      "mean_token_accuracy": 0.748279732465744,
+      "num_tokens": 93961859.0,
+      "step": 40650
+    },
+    {
+      "epoch": 0.3727881177225635,
+      "learning_rate": 0.0001254442101402769,
+      "loss": 0.8539,
+      "mean_token_accuracy": 0.7445384502410889,
+      "num_tokens": 93984682.0,
+      "step": 40660
+    },
+    {
+      "epoch": 0.37287980196204273,
+      "learning_rate": 0.00012542587329238104,
+      "loss": 0.8455,
+      "mean_token_accuracy": 0.7497119009494781,
+      "num_tokens": 94007612.0,
+      "step": 40670
+    },
+    {
+      "epoch": 0.372971486201522,
+      "learning_rate": 0.0001254075364444852,
+      "loss": 0.8628,
+      "mean_token_accuracy": 0.7399684607982635,
+      "num_tokens": 94030804.0,
+      "step": 40680
+    },
+    {
+      "epoch": 0.3730631704410012,
+      "learning_rate": 0.00012538919959658934,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7462611377239228,
+      "num_tokens": 94054300.0,
+      "step": 40690
+    },
+    {
+      "epoch": 0.37315485468048043,
+      "learning_rate": 0.0001253708627486935,
+      "loss": 0.8748,
+      "mean_token_accuracy": 0.7407118141651153,
+      "num_tokens": 94077828.0,
+      "step": 40700
+    },
+    {
+      "epoch": 0.3732465389199597,
+      "learning_rate": 0.00012535252590079767,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7392879724502563,
+      "num_tokens": 94100971.0,
+      "step": 40710
+    },
+    {
+      "epoch": 0.3733382231594389,
+      "learning_rate": 0.0001253341890529018,
+      "loss": 0.8399,
+      "mean_token_accuracy": 0.7479075014591217,
+      "num_tokens": 94124078.0,
+      "step": 40720
+    },
+    {
+      "epoch": 0.3734299073989181,
+      "learning_rate": 0.00012531585220500597,
+      "loss": 0.8186,
+      "mean_token_accuracy": 0.7510039865970611,
+      "num_tokens": 94147238.0,
+      "step": 40730
+    },
+    {
+      "epoch": 0.3735215916383974,
+      "learning_rate": 0.0001252975153571101,
+      "loss": 0.8604,
+      "mean_token_accuracy": 0.741525012254715,
+      "num_tokens": 94170910.0,
+      "step": 40740
+    },
+    {
+      "epoch": 0.37361327587787657,
+      "learning_rate": 0.00012527917850921428,
+      "loss": 0.8155,
+      "mean_token_accuracy": 0.756772369146347,
+      "num_tokens": 94194339.0,
+      "step": 40750
+    },
+    {
+      "epoch": 0.3737049601173558,
+      "learning_rate": 0.00012526084166131842,
+      "loss": 0.8514,
+      "mean_token_accuracy": 0.74683518409729,
+      "num_tokens": 94217202.0,
+      "step": 40760
+    },
+    {
+      "epoch": 0.3737966443568351,
+      "learning_rate": 0.00012524250481342255,
+      "loss": 0.8497,
+      "mean_token_accuracy": 0.7476775407791137,
+      "num_tokens": 94239737.0,
+      "step": 40770
+    },
+    {
+      "epoch": 0.37388832859631427,
+      "learning_rate": 0.00012522416796552675,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7390937447547913,
+      "num_tokens": 94262098.0,
+      "step": 40780
+    },
+    {
+      "epoch": 0.3739800128357935,
+      "learning_rate": 0.00012520583111763089,
+      "loss": 0.8444,
+      "mean_token_accuracy": 0.7504557430744171,
+      "num_tokens": 94285742.0,
+      "step": 40790
+    },
+    {
+      "epoch": 0.37407169707527277,
+      "learning_rate": 0.00012518749426973505,
+      "loss": 0.8141,
+      "mean_token_accuracy": 0.755222475528717,
+      "num_tokens": 94308696.0,
+      "step": 40800
+    },
+    {
+      "epoch": 0.374163381314752,
+      "learning_rate": 0.0001251691574218392,
+      "loss": 0.8436,
+      "mean_token_accuracy": 0.7472757160663605,
+      "num_tokens": 94331831.0,
+      "step": 40810
+    },
+    {
+      "epoch": 0.3742550655542312,
+      "learning_rate": 0.00012515082057394333,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.7493879616260528,
+      "num_tokens": 94355344.0,
+      "step": 40820
+    },
+    {
+      "epoch": 0.37434674979371046,
+      "learning_rate": 0.0001251324837260475,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7448008000850678,
+      "num_tokens": 94377772.0,
+      "step": 40830
+    },
+    {
+      "epoch": 0.3744384340331897,
+      "learning_rate": 0.00012511414687815166,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.7426619470119477,
+      "num_tokens": 94400584.0,
+      "step": 40840
+    },
+    {
+      "epoch": 0.3745301182726689,
+      "learning_rate": 0.00012509581003025582,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7521373689174652,
+      "num_tokens": 94423858.0,
+      "step": 40850
+    },
+    {
+      "epoch": 0.37462180251214816,
+      "learning_rate": 0.00012507747318235996,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7479946434497833,
+      "num_tokens": 94447180.0,
+      "step": 40860
+    },
+    {
+      "epoch": 0.3747134867516274,
+      "learning_rate": 0.0001250591363344641,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.7471463203430175,
+      "num_tokens": 94469854.0,
+      "step": 40870
+    },
+    {
+      "epoch": 0.3748051709911066,
+      "learning_rate": 0.00012504079948656827,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7519271194934845,
+      "num_tokens": 94492674.0,
+      "step": 40880
+    },
+    {
+      "epoch": 0.37489685523058586,
+      "learning_rate": 0.0001250224626386724,
+      "loss": 0.8688,
+      "mean_token_accuracy": 0.738123744726181,
+      "num_tokens": 94515867.0,
+      "step": 40890
+    },
+    {
+      "epoch": 0.3749885394700651,
+      "learning_rate": 0.00012500412579077657,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7527352869510651,
+      "num_tokens": 94539391.0,
+      "step": 40900
+    },
+    {
+      "epoch": 0.3750802237095443,
+      "learning_rate": 0.00012498578894288073,
+      "loss": 0.8203,
+      "mean_token_accuracy": 0.7547052025794982,
+      "num_tokens": 94562710.0,
+      "step": 40910
+    },
+    {
+      "epoch": 0.37517190794902355,
+      "learning_rate": 0.00012496745209498487,
+      "loss": 0.8207,
+      "mean_token_accuracy": 0.7532356142997741,
+      "num_tokens": 94585833.0,
+      "step": 40920
+    },
+    {
+      "epoch": 0.3752635921885028,
+      "learning_rate": 0.00012494911524708904,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7508581697940826,
+      "num_tokens": 94608891.0,
+      "step": 40930
+    },
+    {
+      "epoch": 0.37535527642798205,
+      "learning_rate": 0.00012493077839919318,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.7545634806156158,
+      "num_tokens": 94632468.0,
+      "step": 40940
+    },
+    {
+      "epoch": 0.37544696066746125,
+      "learning_rate": 0.00012491244155129734,
+      "loss": 0.8566,
+      "mean_token_accuracy": 0.7451045572757721,
+      "num_tokens": 94655745.0,
+      "step": 40950
+    },
+    {
+      "epoch": 0.3755386449069405,
+      "learning_rate": 0.00012489410470340148,
+      "loss": 0.8644,
+      "mean_token_accuracy": 0.7480914235115051,
+      "num_tokens": 94679039.0,
+      "step": 40960
+    },
+    {
+      "epoch": 0.37563032914641975,
+      "learning_rate": 0.00012487576785550565,
+      "loss": 0.835,
+      "mean_token_accuracy": 0.7503280758857727,
+      "num_tokens": 94701980.0,
+      "step": 40970
+    },
+    {
+      "epoch": 0.37572201338589895,
+      "learning_rate": 0.0001248574310076098,
+      "loss": 0.8683,
+      "mean_token_accuracy": 0.7421993672847748,
+      "num_tokens": 94724636.0,
+      "step": 40980
+    },
+    {
+      "epoch": 0.3758136976253782,
+      "learning_rate": 0.00012483909415971395,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.7421487331390381,
+      "num_tokens": 94747249.0,
+      "step": 40990
+    },
+    {
+      "epoch": 0.37590538186485745,
+      "learning_rate": 0.00012482075731181812,
+      "loss": 0.8223,
+      "mean_token_accuracy": 0.7498030722141266,
+      "num_tokens": 94770674.0,
+      "step": 41000
+    },
+    {
+      "epoch": 0.37599706610433664,
+      "learning_rate": 0.00012480242046392225,
+      "loss": 0.8823,
+      "mean_token_accuracy": 0.7403291165828705,
+      "num_tokens": 94792841.0,
+      "step": 41010
+    },
+    {
+      "epoch": 0.3760887503438159,
+      "learning_rate": 0.0001247840836160264,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7503290355205536,
+      "num_tokens": 94816525.0,
+      "step": 41020
+    },
+    {
+      "epoch": 0.37618043458329514,
+      "learning_rate": 0.00012476574676813056,
+      "loss": 0.8758,
+      "mean_token_accuracy": 0.7423287034034729,
+      "num_tokens": 94839607.0,
+      "step": 41030
+    },
+    {
+      "epoch": 0.37627211882277434,
+      "learning_rate": 0.00012474740992023472,
+      "loss": 0.8336,
+      "mean_token_accuracy": 0.7480650782585144,
+      "num_tokens": 94862380.0,
+      "step": 41040
+    },
+    {
+      "epoch": 0.3763638030622536,
+      "learning_rate": 0.0001247290730723389,
+      "loss": 0.8816,
+      "mean_token_accuracy": 0.7368262827396392,
+      "num_tokens": 94884977.0,
+      "step": 41050
+    },
+    {
+      "epoch": 0.37645548730173284,
+      "learning_rate": 0.00012471073622444303,
+      "loss": 0.8737,
+      "mean_token_accuracy": 0.7356580853462219,
+      "num_tokens": 94907461.0,
+      "step": 41060
+    },
+    {
+      "epoch": 0.3765471715412121,
+      "learning_rate": 0.00012469239937654716,
+      "loss": 0.8217,
+      "mean_token_accuracy": 0.7550493061542511,
+      "num_tokens": 94930311.0,
+      "step": 41070
+    },
+    {
+      "epoch": 0.3766388557806913,
+      "learning_rate": 0.00012467406252865133,
+      "loss": 0.7902,
+      "mean_token_accuracy": 0.7596096932888031,
+      "num_tokens": 94954141.0,
+      "step": 41080
+    },
+    {
+      "epoch": 0.37673054002017053,
+      "learning_rate": 0.00012465572568075547,
+      "loss": 0.7835,
+      "mean_token_accuracy": 0.7641208648681641,
+      "num_tokens": 94977436.0,
+      "step": 41090
+    },
+    {
+      "epoch": 0.3768222242596498,
+      "learning_rate": 0.00012463738883285963,
+      "loss": 0.9044,
+      "mean_token_accuracy": 0.7317468404769898,
+      "num_tokens": 95000214.0,
+      "step": 41100
+    },
+    {
+      "epoch": 0.376913908499129,
+      "learning_rate": 0.0001246190519849638,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.7414962947368622,
+      "num_tokens": 95023249.0,
+      "step": 41110
+    },
+    {
+      "epoch": 0.37700559273860823,
+      "learning_rate": 0.00012460071513706794,
+      "loss": 0.8281,
+      "mean_token_accuracy": 0.7514614582061767,
+      "num_tokens": 95047427.0,
+      "step": 41120
+    },
+    {
+      "epoch": 0.3770972769780875,
+      "learning_rate": 0.0001245823782891721,
+      "loss": 0.8769,
+      "mean_token_accuracy": 0.746800833940506,
+      "num_tokens": 95070850.0,
+      "step": 41130
+    },
+    {
+      "epoch": 0.3771889612175667,
+      "learning_rate": 0.00012456404144127624,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.7497871875762939,
+      "num_tokens": 95094166.0,
+      "step": 41140
+    },
+    {
+      "epoch": 0.3772806454570459,
+      "learning_rate": 0.0001245457045933804,
+      "loss": 0.8387,
+      "mean_token_accuracy": 0.7449946641921997,
+      "num_tokens": 95117487.0,
+      "step": 41150
+    },
+    {
+      "epoch": 0.3773723296965252,
+      "learning_rate": 0.00012452736774548455,
+      "loss": 0.7937,
+      "mean_token_accuracy": 0.7537588477134705,
+      "num_tokens": 95139646.0,
+      "step": 41160
+    },
+    {
+      "epoch": 0.3774640139360044,
+      "learning_rate": 0.0001245090308975887,
+      "loss": 0.8948,
+      "mean_token_accuracy": 0.7397721230983734,
+      "num_tokens": 95162508.0,
+      "step": 41170
+    },
+    {
+      "epoch": 0.3775556981754836,
+      "learning_rate": 0.00012449069404969288,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.7402627348899842,
+      "num_tokens": 95186041.0,
+      "step": 41180
+    },
+    {
+      "epoch": 0.3776473824149629,
+      "learning_rate": 0.00012447235720179701,
+      "loss": 0.8489,
+      "mean_token_accuracy": 0.7446037411689759,
+      "num_tokens": 95209303.0,
+      "step": 41190
+    },
+    {
+      "epoch": 0.3777390666544421,
+      "learning_rate": 0.00012445402035390118,
+      "loss": 0.8362,
+      "mean_token_accuracy": 0.7498924195766449,
+      "num_tokens": 95231958.0,
+      "step": 41200
+    },
+    {
+      "epoch": 0.3778307508939213,
+      "learning_rate": 0.00012443568350600532,
+      "loss": 0.8527,
+      "mean_token_accuracy": 0.7455955564975738,
+      "num_tokens": 95255840.0,
+      "step": 41210
+    },
+    {
+      "epoch": 0.37792243513340057,
+      "learning_rate": 0.00012441734665810946,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7489126145839691,
+      "num_tokens": 95277953.0,
+      "step": 41220
+    },
+    {
+      "epoch": 0.3780141193728798,
+      "learning_rate": 0.00012439900981021365,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.7478990614414215,
+      "num_tokens": 95300794.0,
+      "step": 41230
+    },
+    {
+      "epoch": 0.378105803612359,
+      "learning_rate": 0.0001243806729623178,
+      "loss": 0.8459,
+      "mean_token_accuracy": 0.7479574739933014,
+      "num_tokens": 95324437.0,
+      "step": 41240
+    },
+    {
+      "epoch": 0.37819748785183827,
+      "learning_rate": 0.00012436233611442195,
+      "loss": 0.8491,
+      "mean_token_accuracy": 0.7490546584129334,
+      "num_tokens": 95347068.0,
+      "step": 41250
+    },
+    {
+      "epoch": 0.3782891720913175,
+      "learning_rate": 0.0001243439992665261,
+      "loss": 0.8327,
+      "mean_token_accuracy": 0.7576212704181671,
+      "num_tokens": 95370337.0,
+      "step": 41260
+    },
+    {
+      "epoch": 0.3783808563307967,
+      "learning_rate": 0.00012432566241863023,
+      "loss": 0.8434,
+      "mean_token_accuracy": 0.7524004936218261,
+      "num_tokens": 95392832.0,
+      "step": 41270
+    },
+    {
+      "epoch": 0.37847254057027596,
+      "learning_rate": 0.0001243073255707344,
+      "loss": 0.8465,
+      "mean_token_accuracy": 0.7434422254562378,
+      "num_tokens": 95415356.0,
+      "step": 41280
+    },
+    {
+      "epoch": 0.3785642248097552,
+      "learning_rate": 0.00012428898872283853,
+      "loss": 0.8459,
+      "mean_token_accuracy": 0.7438011527061462,
+      "num_tokens": 95438070.0,
+      "step": 41290
+    },
+    {
+      "epoch": 0.37865590904923446,
+      "learning_rate": 0.0001242706518749427,
+      "loss": 0.828,
+      "mean_token_accuracy": 0.750016987323761,
+      "num_tokens": 95460620.0,
+      "step": 41300
+    },
+    {
+      "epoch": 0.37874759328871366,
+      "learning_rate": 0.00012425231502704686,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.7443816423416137,
+      "num_tokens": 95483925.0,
+      "step": 41310
+    },
+    {
+      "epoch": 0.3788392775281929,
+      "learning_rate": 0.000124233978179151,
+      "loss": 0.8584,
+      "mean_token_accuracy": 0.7468372225761414,
+      "num_tokens": 95507638.0,
+      "step": 41320
+    },
+    {
+      "epoch": 0.37893096176767216,
+      "learning_rate": 0.00012421564133125517,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7462254405021668,
+      "num_tokens": 95530682.0,
+      "step": 41330
+    },
+    {
+      "epoch": 0.37902264600715135,
+      "learning_rate": 0.0001241973044833593,
+      "loss": 0.8772,
+      "mean_token_accuracy": 0.7407940268516541,
+      "num_tokens": 95553642.0,
+      "step": 41340
+    },
+    {
+      "epoch": 0.3791143302466306,
+      "learning_rate": 0.00012417896763546347,
+      "loss": 0.8379,
+      "mean_token_accuracy": 0.7433828294277192,
+      "num_tokens": 95576261.0,
+      "step": 41350
+    },
+    {
+      "epoch": 0.37920601448610985,
+      "learning_rate": 0.00012416063078756764,
+      "loss": 0.8466,
+      "mean_token_accuracy": 0.7472116649150848,
+      "num_tokens": 95599140.0,
+      "step": 41360
+    },
+    {
+      "epoch": 0.37929769872558905,
+      "learning_rate": 0.00012414229393967177,
+      "loss": 0.8284,
+      "mean_token_accuracy": 0.747962474822998,
+      "num_tokens": 95622279.0,
+      "step": 41370
+    },
+    {
+      "epoch": 0.3793893829650683,
+      "learning_rate": 0.00012412395709177594,
+      "loss": 0.8466,
+      "mean_token_accuracy": 0.7496549546718597,
+      "num_tokens": 95645558.0,
+      "step": 41380
+    },
+    {
+      "epoch": 0.37948106720454755,
+      "learning_rate": 0.00012410562024388008,
+      "loss": 0.8786,
+      "mean_token_accuracy": 0.7420814752578735,
+      "num_tokens": 95669066.0,
+      "step": 41390
+    },
+    {
+      "epoch": 0.37957275144402675,
+      "learning_rate": 0.00012408728339598424,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.75220627784729,
+      "num_tokens": 95692111.0,
+      "step": 41400
+    },
+    {
+      "epoch": 0.379664435683506,
+      "learning_rate": 0.00012406894654808838,
+      "loss": 0.8597,
+      "mean_token_accuracy": 0.7449198603630066,
+      "num_tokens": 95715528.0,
+      "step": 41410
+    },
+    {
+      "epoch": 0.37975611992298525,
+      "learning_rate": 0.00012405060970019252,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7405585646629333,
+      "num_tokens": 95738027.0,
+      "step": 41420
+    },
+    {
+      "epoch": 0.3798478041624645,
+      "learning_rate": 0.0001240322728522967,
+      "loss": 0.8591,
+      "mean_token_accuracy": 0.7457330167293549,
+      "num_tokens": 95760789.0,
+      "step": 41430
+    },
+    {
+      "epoch": 0.3799394884019437,
+      "learning_rate": 0.00012401393600440085,
+      "loss": 0.8421,
+      "mean_token_accuracy": 0.7469176411628723,
+      "num_tokens": 95784396.0,
+      "step": 41440
+    },
+    {
+      "epoch": 0.38003117264142294,
+      "learning_rate": 0.00012399559915650502,
+      "loss": 0.8676,
+      "mean_token_accuracy": 0.7385257601737976,
+      "num_tokens": 95807642.0,
+      "step": 41450
+    },
+    {
+      "epoch": 0.3801228568809022,
+      "learning_rate": 0.00012397726230860916,
+      "loss": 0.8541,
+      "mean_token_accuracy": 0.7452293395996094,
+      "num_tokens": 95831408.0,
+      "step": 41460
+    },
+    {
+      "epoch": 0.3802145411203814,
+      "learning_rate": 0.0001239589254607133,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.7429750680923461,
+      "num_tokens": 95853824.0,
+      "step": 41470
+    },
+    {
+      "epoch": 0.38030622535986064,
+      "learning_rate": 0.00012394058861281746,
+      "loss": 0.8191,
+      "mean_token_accuracy": 0.7523462414741516,
+      "num_tokens": 95876263.0,
+      "step": 41480
+    },
+    {
+      "epoch": 0.3803979095993399,
+      "learning_rate": 0.00012392225176492162,
+      "loss": 0.8441,
+      "mean_token_accuracy": 0.7436273038387299,
+      "num_tokens": 95898382.0,
+      "step": 41490
+    },
+    {
+      "epoch": 0.3804895938388191,
+      "learning_rate": 0.00012390391491702576,
+      "loss": 0.8362,
+      "mean_token_accuracy": 0.7424362659454345,
+      "num_tokens": 95921189.0,
+      "step": 41500
+    },
+    {
+      "epoch": 0.38058127807829834,
+      "learning_rate": 0.00012388557806912993,
+      "loss": 0.8746,
+      "mean_token_accuracy": 0.7430182814598083,
+      "num_tokens": 95944401.0,
+      "step": 41510
+    },
+    {
+      "epoch": 0.3806729623177776,
+      "learning_rate": 0.00012386724122123407,
+      "loss": 0.8622,
+      "mean_token_accuracy": 0.7408500671386719,
+      "num_tokens": 95968077.0,
+      "step": 41520
+    },
+    {
+      "epoch": 0.3807646465572568,
+      "learning_rate": 0.00012384890437333823,
+      "loss": 0.8196,
+      "mean_token_accuracy": 0.7536414504051209,
+      "num_tokens": 95991717.0,
+      "step": 41530
+    },
+    {
+      "epoch": 0.38085633079673603,
+      "learning_rate": 0.00012383056752544237,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.745907062292099,
+      "num_tokens": 96014936.0,
+      "step": 41540
+    },
+    {
+      "epoch": 0.3809480150362153,
+      "learning_rate": 0.00012381223067754654,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.7377768397331238,
+      "num_tokens": 96038452.0,
+      "step": 41550
+    },
+    {
+      "epoch": 0.38103969927569453,
+      "learning_rate": 0.0001237938938296507,
+      "loss": 0.8583,
+      "mean_token_accuracy": 0.7457735657691955,
+      "num_tokens": 96061547.0,
+      "step": 41560
+    },
+    {
+      "epoch": 0.3811313835151737,
+      "learning_rate": 0.00012377555698175484,
+      "loss": 0.8483,
+      "mean_token_accuracy": 0.7477504968643188,
+      "num_tokens": 96084469.0,
+      "step": 41570
+    },
+    {
+      "epoch": 0.381223067754653,
+      "learning_rate": 0.000123757220133859,
+      "loss": 0.8195,
+      "mean_token_accuracy": 0.7502038061618805,
+      "num_tokens": 96106614.0,
+      "step": 41580
+    },
+    {
+      "epoch": 0.38131475199413223,
+      "learning_rate": 0.00012373888328596314,
+      "loss": 0.8362,
+      "mean_token_accuracy": 0.7560655534267425,
+      "num_tokens": 96129823.0,
+      "step": 41590
+    },
+    {
+      "epoch": 0.3814064362336114,
+      "learning_rate": 0.0001237205464380673,
+      "loss": 0.8237,
+      "mean_token_accuracy": 0.7563638985157013,
+      "num_tokens": 96152566.0,
+      "step": 41600
+    },
+    {
+      "epoch": 0.3814981204730907,
+      "learning_rate": 0.00012370220959017145,
+      "loss": 0.8346,
+      "mean_token_accuracy": 0.7453992128372192,
+      "num_tokens": 96175222.0,
+      "step": 41610
+    },
+    {
+      "epoch": 0.3815898047125699,
+      "learning_rate": 0.0001236838727422756,
+      "loss": 0.8499,
+      "mean_token_accuracy": 0.7355525553226471,
+      "num_tokens": 96197591.0,
+      "step": 41620
+    },
+    {
+      "epoch": 0.3816814889520491,
+      "learning_rate": 0.00012366553589437978,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.7429073810577392,
+      "num_tokens": 96221013.0,
+      "step": 41630
+    },
+    {
+      "epoch": 0.38177317319152837,
+      "learning_rate": 0.00012364719904648392,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7413643419742584,
+      "num_tokens": 96243836.0,
+      "step": 41640
+    },
+    {
+      "epoch": 0.3818648574310076,
+      "learning_rate": 0.00012362886219858808,
+      "loss": 0.8604,
+      "mean_token_accuracy": 0.7440150916576386,
+      "num_tokens": 96267157.0,
+      "step": 41650
+    },
+    {
+      "epoch": 0.38195654167048687,
+      "learning_rate": 0.00012361052535069222,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.7463220000267029,
+      "num_tokens": 96289822.0,
+      "step": 41660
+    },
+    {
+      "epoch": 0.38204822590996607,
+      "learning_rate": 0.00012359218850279636,
+      "loss": 0.8804,
+      "mean_token_accuracy": 0.7378134787082672,
+      "num_tokens": 96313109.0,
+      "step": 41670
+    },
+    {
+      "epoch": 0.3821399101494453,
+      "learning_rate": 0.00012357385165490052,
+      "loss": 0.836,
+      "mean_token_accuracy": 0.7468629121780396,
+      "num_tokens": 96335956.0,
+      "step": 41680
+    },
+    {
+      "epoch": 0.38223159438892457,
+      "learning_rate": 0.0001235555148070047,
+      "loss": 0.8304,
+      "mean_token_accuracy": 0.7488134324550628,
+      "num_tokens": 96359066.0,
+      "step": 41690
+    },
+    {
+      "epoch": 0.38232327862840376,
+      "learning_rate": 0.00012353717795910883,
+      "loss": 0.8768,
+      "mean_token_accuracy": 0.7380219399929047,
+      "num_tokens": 96383087.0,
+      "step": 41700
+    },
+    {
+      "epoch": 0.382414962867883,
+      "learning_rate": 0.000123518841111213,
+      "loss": 0.8441,
+      "mean_token_accuracy": 0.7522395133972168,
+      "num_tokens": 96405689.0,
+      "step": 41710
+    },
+    {
+      "epoch": 0.38250664710736226,
+      "learning_rate": 0.00012350050426331713,
+      "loss": 0.8496,
+      "mean_token_accuracy": 0.7440075397491455,
+      "num_tokens": 96428809.0,
+      "step": 41720
+    },
+    {
+      "epoch": 0.38259833134684146,
+      "learning_rate": 0.0001234821674154213,
+      "loss": 0.8331,
+      "mean_token_accuracy": 0.7506649017333984,
+      "num_tokens": 96452688.0,
+      "step": 41730
+    },
+    {
+      "epoch": 0.3826900155863207,
+      "learning_rate": 0.00012346383056752543,
+      "loss": 0.8637,
+      "mean_token_accuracy": 0.7435618579387665,
+      "num_tokens": 96475313.0,
+      "step": 41740
+    },
+    {
+      "epoch": 0.38278169982579996,
+      "learning_rate": 0.0001234454937196296,
+      "loss": 0.8683,
+      "mean_token_accuracy": 0.7351320087909698,
+      "num_tokens": 96498150.0,
+      "step": 41750
+    },
+    {
+      "epoch": 0.38287338406527915,
+      "learning_rate": 0.00012342715687173377,
+      "loss": 0.8123,
+      "mean_token_accuracy": 0.7541450798511505,
+      "num_tokens": 96520766.0,
+      "step": 41760
+    },
+    {
+      "epoch": 0.3829650683047584,
+      "learning_rate": 0.0001234088200238379,
+      "loss": 0.8532,
+      "mean_token_accuracy": 0.7484757244586945,
+      "num_tokens": 96544237.0,
+      "step": 41770
+    },
+    {
+      "epoch": 0.38305675254423766,
+      "learning_rate": 0.00012339048317594207,
+      "loss": 0.8635,
+      "mean_token_accuracy": 0.740970355272293,
+      "num_tokens": 96566726.0,
+      "step": 41780
+    },
+    {
+      "epoch": 0.3831484367837169,
+      "learning_rate": 0.0001233721463280462,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.745781946182251,
+      "num_tokens": 96589099.0,
+      "step": 41790
+    },
+    {
+      "epoch": 0.3832401210231961,
+      "learning_rate": 0.00012335380948015037,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7469968736171723,
+      "num_tokens": 96611204.0,
+      "step": 41800
+    },
+    {
+      "epoch": 0.38333180526267535,
+      "learning_rate": 0.0001233354726322545,
+      "loss": 0.8642,
+      "mean_token_accuracy": 0.7485895335674286,
+      "num_tokens": 96634671.0,
+      "step": 41810
+    },
+    {
+      "epoch": 0.3834234895021546,
+      "learning_rate": 0.00012331713578435868,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.7442315816879272,
+      "num_tokens": 96658050.0,
+      "step": 41820
+    },
+    {
+      "epoch": 0.3835151737416338,
+      "learning_rate": 0.00012329879893646284,
+      "loss": 0.8792,
+      "mean_token_accuracy": 0.7427907228469849,
+      "num_tokens": 96680366.0,
+      "step": 41830
+    },
+    {
+      "epoch": 0.38360685798111305,
+      "learning_rate": 0.00012328046208856698,
+      "loss": 0.8269,
+      "mean_token_accuracy": 0.7465570032596588,
+      "num_tokens": 96704047.0,
+      "step": 41840
+    },
+    {
+      "epoch": 0.3836985422205923,
+      "learning_rate": 0.00012326212524067115,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7422377943992615,
+      "num_tokens": 96726945.0,
+      "step": 41850
+    },
+    {
+      "epoch": 0.3837902264600715,
+      "learning_rate": 0.00012324378839277528,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7436909973621368,
+      "num_tokens": 96749645.0,
+      "step": 41860
+    },
+    {
+      "epoch": 0.38388191069955074,
+      "learning_rate": 0.00012322545154487942,
+      "loss": 0.8324,
+      "mean_token_accuracy": 0.7500499904155731,
+      "num_tokens": 96772768.0,
+      "step": 41870
+    },
+    {
+      "epoch": 0.38397359493903,
+      "learning_rate": 0.00012320711469698362,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.743410587310791,
+      "num_tokens": 96796061.0,
+      "step": 41880
+    },
+    {
+      "epoch": 0.3840652791785092,
+      "learning_rate": 0.00012318877784908775,
+      "loss": 0.8362,
+      "mean_token_accuracy": 0.7453331232070923,
+      "num_tokens": 96819674.0,
+      "step": 41890
+    },
+    {
+      "epoch": 0.38415696341798844,
+      "learning_rate": 0.0001231704410011919,
+      "loss": 0.8502,
+      "mean_token_accuracy": 0.7472075521945953,
+      "num_tokens": 96842618.0,
+      "step": 41900
+    },
+    {
+      "epoch": 0.3842486476574677,
+      "learning_rate": 0.00012315210415329606,
+      "loss": 0.8565,
+      "mean_token_accuracy": 0.7406310021877289,
+      "num_tokens": 96865875.0,
+      "step": 41910
+    },
+    {
+      "epoch": 0.38434033189694694,
+      "learning_rate": 0.0001231337673054002,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.7463303685188294,
+      "num_tokens": 96888689.0,
+      "step": 41920
+    },
+    {
+      "epoch": 0.38443201613642614,
+      "learning_rate": 0.00012311543045750436,
+      "loss": 0.8324,
+      "mean_token_accuracy": 0.7508059799671173,
+      "num_tokens": 96911639.0,
+      "step": 41930
+    },
+    {
+      "epoch": 0.3845237003759054,
+      "learning_rate": 0.0001230970936096085,
+      "loss": 0.8267,
+      "mean_token_accuracy": 0.7515237748622894,
+      "num_tokens": 96934886.0,
+      "step": 41940
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "learning_rate": 0.00012307875676171266,
+      "loss": 0.8098,
+      "mean_token_accuracy": 0.7531907796859741,
+      "num_tokens": 96958264.0,
+      "step": 41950
+    },
+    {
+      "epoch": 0.38470706885486383,
+      "learning_rate": 0.00012306041991381683,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7565134763717651,
+      "num_tokens": 96981073.0,
+      "step": 41960
+    },
+    {
+      "epoch": 0.3847987530943431,
+      "learning_rate": 0.00012304208306592097,
+      "loss": 0.838,
+      "mean_token_accuracy": 0.7485580503940582,
+      "num_tokens": 97004097.0,
+      "step": 41970
+    },
+    {
+      "epoch": 0.38489043733382233,
+      "learning_rate": 0.00012302374621802513,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7474245131015778,
+      "num_tokens": 97028029.0,
+      "step": 41980
+    },
+    {
+      "epoch": 0.38498212157330153,
+      "learning_rate": 0.00012300540937012927,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7395976901054382,
+      "num_tokens": 97051585.0,
+      "step": 41990
+    },
+    {
+      "epoch": 0.3850738058127808,
+      "learning_rate": 0.00012298707252223344,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7424962103366852,
+      "num_tokens": 97075015.0,
+      "step": 42000
+    },
+    {
+      "epoch": 0.38516549005226003,
+      "learning_rate": 0.00012296873567433758,
+      "loss": 0.8325,
+      "mean_token_accuracy": 0.7579595148563385,
+      "num_tokens": 97098684.0,
+      "step": 42010
+    },
+    {
+      "epoch": 0.3852571742917392,
+      "learning_rate": 0.00012295039882644174,
+      "loss": 0.8374,
+      "mean_token_accuracy": 0.7485345542430878,
+      "num_tokens": 97121634.0,
+      "step": 42020
+    },
+    {
+      "epoch": 0.3853488585312185,
+      "learning_rate": 0.0001229320619785459,
+      "loss": 0.904,
+      "mean_token_accuracy": 0.7320353865623475,
+      "num_tokens": 97143979.0,
+      "step": 42030
+    },
+    {
+      "epoch": 0.3854405427706977,
+      "learning_rate": 0.00012291372513065004,
+      "loss": 0.8389,
+      "mean_token_accuracy": 0.7421582698822021,
+      "num_tokens": 97167154.0,
+      "step": 42040
+    },
+    {
+      "epoch": 0.385532227010177,
+      "learning_rate": 0.0001228953882827542,
+      "loss": 0.8432,
+      "mean_token_accuracy": 0.7446771323680877,
+      "num_tokens": 97190056.0,
+      "step": 42050
+    },
+    {
+      "epoch": 0.38562391124965617,
+      "learning_rate": 0.00012287705143485835,
+      "loss": 0.854,
+      "mean_token_accuracy": 0.7453037738800049,
+      "num_tokens": 97212306.0,
+      "step": 42060
+    },
+    {
+      "epoch": 0.3857155954891354,
+      "learning_rate": 0.0001228587145869625,
+      "loss": 0.8621,
+      "mean_token_accuracy": 0.740193498134613,
+      "num_tokens": 97235252.0,
+      "step": 42070
+    },
+    {
+      "epoch": 0.38580727972861467,
+      "learning_rate": 0.00012284037773906668,
+      "loss": 0.8512,
+      "mean_token_accuracy": 0.7503459095954895,
+      "num_tokens": 97258122.0,
+      "step": 42080
+    },
+    {
+      "epoch": 0.38589896396809387,
+      "learning_rate": 0.00012282204089117082,
+      "loss": 0.8551,
+      "mean_token_accuracy": 0.7457744300365448,
+      "num_tokens": 97281193.0,
+      "step": 42090
+    },
+    {
+      "epoch": 0.3859906482075731,
+      "learning_rate": 0.00012280370404327496,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.7550775706768036,
+      "num_tokens": 97304317.0,
+      "step": 42100
+    },
+    {
+      "epoch": 0.38608233244705237,
+      "learning_rate": 0.00012278536719537912,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.7475060760974884,
+      "num_tokens": 97326943.0,
+      "step": 42110
+    },
+    {
+      "epoch": 0.38617401668653156,
+      "learning_rate": 0.00012276703034748326,
+      "loss": 0.8289,
+      "mean_token_accuracy": 0.7514943778514862,
+      "num_tokens": 97350240.0,
+      "step": 42120
+    },
+    {
+      "epoch": 0.3862657009260108,
+      "learning_rate": 0.00012274869349958743,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.7483797132968902,
+      "num_tokens": 97373501.0,
+      "step": 42130
+    },
+    {
+      "epoch": 0.38635738516549006,
+      "learning_rate": 0.00012273035665169156,
+      "loss": 0.8326,
+      "mean_token_accuracy": 0.7438017189502716,
+      "num_tokens": 97396653.0,
+      "step": 42140
+    },
+    {
+      "epoch": 0.38644906940496926,
+      "learning_rate": 0.00012271201980379573,
+      "loss": 0.8892,
+      "mean_token_accuracy": 0.7413406610488892,
+      "num_tokens": 97420576.0,
+      "step": 42150
+    },
+    {
+      "epoch": 0.3865407536444485,
+      "learning_rate": 0.0001226936829558999,
+      "loss": 0.7911,
+      "mean_token_accuracy": 0.7549246847629547,
+      "num_tokens": 97443503.0,
+      "step": 42160
+    },
+    {
+      "epoch": 0.38663243788392776,
+      "learning_rate": 0.00012267534610800403,
+      "loss": 0.8077,
+      "mean_token_accuracy": 0.7582757711410523,
+      "num_tokens": 97466317.0,
+      "step": 42170
+    },
+    {
+      "epoch": 0.386724122123407,
+      "learning_rate": 0.0001226570092601082,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7399112820625305,
+      "num_tokens": 97489520.0,
+      "step": 42180
+    },
+    {
+      "epoch": 0.3868158063628862,
+      "learning_rate": 0.00012263867241221234,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.747317761182785,
+      "num_tokens": 97512132.0,
+      "step": 42190
+    },
+    {
+      "epoch": 0.38690749060236546,
+      "learning_rate": 0.0001226203355643165,
+      "loss": 0.8308,
+      "mean_token_accuracy": 0.7511660397052765,
+      "num_tokens": 97535004.0,
+      "step": 42200
+    },
+    {
+      "epoch": 0.3869991748418447,
+      "learning_rate": 0.00012260199871642067,
+      "loss": 0.8471,
+      "mean_token_accuracy": 0.7478880822658539,
+      "num_tokens": 97557136.0,
+      "step": 42210
+    },
+    {
+      "epoch": 0.3870908590813239,
+      "learning_rate": 0.0001225836618685248,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7463126242160797,
+      "num_tokens": 97579596.0,
+      "step": 42220
+    },
+    {
+      "epoch": 0.38718254332080315,
+      "learning_rate": 0.00012256532502062897,
+      "loss": 0.8444,
+      "mean_token_accuracy": 0.7474619388580322,
+      "num_tokens": 97602562.0,
+      "step": 42230
+    },
+    {
+      "epoch": 0.3872742275602824,
+      "learning_rate": 0.0001225469881727331,
+      "loss": 0.8184,
+      "mean_token_accuracy": 0.7536703109741211,
+      "num_tokens": 97625331.0,
+      "step": 42240
+    },
+    {
+      "epoch": 0.3873659117997616,
+      "learning_rate": 0.00012252865132483727,
+      "loss": 0.8307,
+      "mean_token_accuracy": 0.7487288653850556,
+      "num_tokens": 97648543.0,
+      "step": 42250
+    },
+    {
+      "epoch": 0.38745759603924085,
+      "learning_rate": 0.0001225103144769414,
+      "loss": 0.9086,
+      "mean_token_accuracy": 0.7395607650279998,
+      "num_tokens": 97671348.0,
+      "step": 42260
+    },
+    {
+      "epoch": 0.3875492802787201,
+      "learning_rate": 0.00012249197762904555,
+      "loss": 0.827,
+      "mean_token_accuracy": 0.7564811944961548,
+      "num_tokens": 97694750.0,
+      "step": 42270
+    },
+    {
+      "epoch": 0.38764096451819935,
+      "learning_rate": 0.00012247364078114974,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7427230477333069,
+      "num_tokens": 97717795.0,
+      "step": 42280
+    },
+    {
+      "epoch": 0.38773264875767854,
+      "learning_rate": 0.00012245530393325388,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7396143674850464,
+      "num_tokens": 97741096.0,
+      "step": 42290
+    },
+    {
+      "epoch": 0.3878243329971578,
+      "learning_rate": 0.00012243696708535802,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.7423577010631561,
+      "num_tokens": 97764715.0,
+      "step": 42300
+    },
+    {
+      "epoch": 0.38791601723663705,
+      "learning_rate": 0.00012241863023746219,
+      "loss": 0.8201,
+      "mean_token_accuracy": 0.7547329008579254,
+      "num_tokens": 97787789.0,
+      "step": 42310
+    },
+    {
+      "epoch": 0.38800770147611624,
+      "learning_rate": 0.00012240029338956632,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.7471260011196137,
+      "num_tokens": 97811110.0,
+      "step": 42320
+    },
+    {
+      "epoch": 0.3880993857155955,
+      "learning_rate": 0.0001223819565416705,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7432671666145325,
+      "num_tokens": 97833659.0,
+      "step": 42330
+    },
+    {
+      "epoch": 0.38819106995507474,
+      "learning_rate": 0.00012236361969377466,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7468033790588379,
+      "num_tokens": 97856353.0,
+      "step": 42340
+    },
+    {
+      "epoch": 0.38828275419455394,
+      "learning_rate": 0.0001223452828458788,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7481363534927368,
+      "num_tokens": 97878954.0,
+      "step": 42350
+    },
+    {
+      "epoch": 0.3883744384340332,
+      "learning_rate": 0.00012232694599798296,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.747006356716156,
+      "num_tokens": 97901800.0,
+      "step": 42360
+    },
+    {
+      "epoch": 0.38846612267351244,
+      "learning_rate": 0.0001223086091500871,
+      "loss": 0.8725,
+      "mean_token_accuracy": 0.739853686094284,
+      "num_tokens": 97925070.0,
+      "step": 42370
+    },
+    {
+      "epoch": 0.38855780691299163,
+      "learning_rate": 0.00012229027230219126,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7503976643085479,
+      "num_tokens": 97948335.0,
+      "step": 42380
+    },
+    {
+      "epoch": 0.3886494911524709,
+      "learning_rate": 0.0001222719354542954,
+      "loss": 0.8567,
+      "mean_token_accuracy": 0.7516186773777008,
+      "num_tokens": 97971069.0,
+      "step": 42390
+    },
+    {
+      "epoch": 0.38874117539195013,
+      "learning_rate": 0.00012225359860639957,
+      "loss": 0.8584,
+      "mean_token_accuracy": 0.7411632359027862,
+      "num_tokens": 97993235.0,
+      "step": 42400
+    },
+    {
+      "epoch": 0.3888328596314294,
+      "learning_rate": 0.00012223526175850373,
+      "loss": 0.847,
+      "mean_token_accuracy": 0.7408832550048828,
+      "num_tokens": 98016281.0,
+      "step": 42410
+    },
+    {
+      "epoch": 0.3889245438709086,
+      "learning_rate": 0.00012221692491060787,
+      "loss": 0.9311,
+      "mean_token_accuracy": 0.7320229709148407,
+      "num_tokens": 98039308.0,
+      "step": 42420
+    },
+    {
+      "epoch": 0.38901622811038783,
+      "learning_rate": 0.00012219858806271204,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.744017893075943,
+      "num_tokens": 98062055.0,
+      "step": 42430
+    },
+    {
+      "epoch": 0.3891079123498671,
+      "learning_rate": 0.00012218025121481617,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7475646376609802,
+      "num_tokens": 98085320.0,
+      "step": 42440
+    },
+    {
+      "epoch": 0.3891995965893463,
+      "learning_rate": 0.00012216191436692034,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.7441819250583649,
+      "num_tokens": 98108024.0,
+      "step": 42450
+    },
+    {
+      "epoch": 0.3892912808288255,
+      "learning_rate": 0.00012214357751902448,
+      "loss": 0.8901,
+      "mean_token_accuracy": 0.7354893386363983,
+      "num_tokens": 98130336.0,
+      "step": 42460
+    },
+    {
+      "epoch": 0.3893829650683048,
+      "learning_rate": 0.00012212524067112864,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7383145391941071,
+      "num_tokens": 98153502.0,
+      "step": 42470
+    },
+    {
+      "epoch": 0.38947464930778397,
+      "learning_rate": 0.0001221069038232328,
+      "loss": 0.8717,
+      "mean_token_accuracy": 0.7394768238067627,
+      "num_tokens": 98177078.0,
+      "step": 42480
+    },
+    {
+      "epoch": 0.3895663335472632,
+      "learning_rate": 0.00012208856697533695,
+      "loss": 0.8478,
+      "mean_token_accuracy": 0.744846111536026,
+      "num_tokens": 98200383.0,
+      "step": 42490
+    },
+    {
+      "epoch": 0.38965801778674247,
+      "learning_rate": 0.0001220702301274411,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7416116893291473,
+      "num_tokens": 98223102.0,
+      "step": 42500
+    },
+    {
+      "epoch": 0.38974970202622167,
+      "learning_rate": 0.00012205189327954525,
+      "loss": 0.8751,
+      "mean_token_accuracy": 0.7336749017238617,
+      "num_tokens": 98246797.0,
+      "step": 42510
+    },
+    {
+      "epoch": 0.3898413862657009,
+      "learning_rate": 0.0001220335564316494,
+      "loss": 0.8081,
+      "mean_token_accuracy": 0.7532801032066345,
+      "num_tokens": 98269820.0,
+      "step": 42520
+    },
+    {
+      "epoch": 0.38993307050518017,
+      "learning_rate": 0.00012201521958375355,
+      "loss": 0.8313,
+      "mean_token_accuracy": 0.7471497416496277,
+      "num_tokens": 98293072.0,
+      "step": 42530
+    },
+    {
+      "epoch": 0.3900247547446594,
+      "learning_rate": 0.00012199688273585772,
+      "loss": 0.8637,
+      "mean_token_accuracy": 0.7445648729801178,
+      "num_tokens": 98315948.0,
+      "step": 42540
+    },
+    {
+      "epoch": 0.3901164389841386,
+      "learning_rate": 0.00012197854588796187,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7450841248035431,
+      "num_tokens": 98338378.0,
+      "step": 42550
+    },
+    {
+      "epoch": 0.39020812322361786,
+      "learning_rate": 0.00012196020904006602,
+      "loss": 0.8267,
+      "mean_token_accuracy": 0.7513948380947113,
+      "num_tokens": 98362016.0,
+      "step": 42560
+    },
+    {
+      "epoch": 0.3902998074630971,
+      "learning_rate": 0.00012194187219217018,
+      "loss": 0.8697,
+      "mean_token_accuracy": 0.7414669454097748,
+      "num_tokens": 98385630.0,
+      "step": 42570
+    },
+    {
+      "epoch": 0.3903914917025763,
+      "learning_rate": 0.00012192353534427431,
+      "loss": 0.8283,
+      "mean_token_accuracy": 0.7504765570163727,
+      "num_tokens": 98408996.0,
+      "step": 42580
+    },
+    {
+      "epoch": 0.39048317594205556,
+      "learning_rate": 0.00012190519849637847,
+      "loss": 0.8619,
+      "mean_token_accuracy": 0.7417500793933869,
+      "num_tokens": 98432049.0,
+      "step": 42590
+    },
+    {
+      "epoch": 0.3905748601815348,
+      "learning_rate": 0.00012188686164848264,
+      "loss": 0.8381,
+      "mean_token_accuracy": 0.7420496046543121,
+      "num_tokens": 98455519.0,
+      "step": 42600
+    },
+    {
+      "epoch": 0.390666544421014,
+      "learning_rate": 0.0001218685248005868,
+      "loss": 0.8385,
+      "mean_token_accuracy": 0.7466835856437684,
+      "num_tokens": 98478520.0,
+      "step": 42610
+    },
+    {
+      "epoch": 0.39075822866049326,
+      "learning_rate": 0.00012185018795269095,
+      "loss": 0.7995,
+      "mean_token_accuracy": 0.7615115284919739,
+      "num_tokens": 98501855.0,
+      "step": 42620
+    },
+    {
+      "epoch": 0.3908499128999725,
+      "learning_rate": 0.00012183185110479509,
+      "loss": 0.8389,
+      "mean_token_accuracy": 0.7464279592037201,
+      "num_tokens": 98524679.0,
+      "step": 42630
+    },
+    {
+      "epoch": 0.3909415971394517,
+      "learning_rate": 0.00012181351425689924,
+      "loss": 0.8199,
+      "mean_token_accuracy": 0.7543918073177338,
+      "num_tokens": 98547887.0,
+      "step": 42640
+    },
+    {
+      "epoch": 0.39103328137893095,
+      "learning_rate": 0.00012179517740900339,
+      "loss": 0.8726,
+      "mean_token_accuracy": 0.7465454757213592,
+      "num_tokens": 98570755.0,
+      "step": 42650
+    },
+    {
+      "epoch": 0.3911249656184102,
+      "learning_rate": 0.00012177684056110754,
+      "loss": 0.8359,
+      "mean_token_accuracy": 0.7499042809009552,
+      "num_tokens": 98593619.0,
+      "step": 42660
+    },
+    {
+      "epoch": 0.39121664985788945,
+      "learning_rate": 0.00012175850371321171,
+      "loss": 0.8672,
+      "mean_token_accuracy": 0.739033991098404,
+      "num_tokens": 98617118.0,
+      "step": 42670
+    },
+    {
+      "epoch": 0.39130833409736865,
+      "learning_rate": 0.00012174016686531586,
+      "loss": 0.8327,
+      "mean_token_accuracy": 0.7493605077266693,
+      "num_tokens": 98640269.0,
+      "step": 42680
+    },
+    {
+      "epoch": 0.3914000183368479,
+      "learning_rate": 0.00012172183001742001,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7385443985462189,
+      "num_tokens": 98663620.0,
+      "step": 42690
+    },
+    {
+      "epoch": 0.39149170257632715,
+      "learning_rate": 0.00012170349316952416,
+      "loss": 0.878,
+      "mean_token_accuracy": 0.7378088414669037,
+      "num_tokens": 98685736.0,
+      "step": 42700
+    },
+    {
+      "epoch": 0.39158338681580634,
+      "learning_rate": 0.00012168515632162831,
+      "loss": 0.8515,
+      "mean_token_accuracy": 0.7487938463687897,
+      "num_tokens": 98709117.0,
+      "step": 42710
+    },
+    {
+      "epoch": 0.3916750710552856,
+      "learning_rate": 0.00012166681947373247,
+      "loss": 0.8203,
+      "mean_token_accuracy": 0.749659925699234,
+      "num_tokens": 98731923.0,
+      "step": 42720
+    },
+    {
+      "epoch": 0.39176675529476485,
+      "learning_rate": 0.00012164848262583663,
+      "loss": 0.8418,
+      "mean_token_accuracy": 0.7471487343311309,
+      "num_tokens": 98755131.0,
+      "step": 42730
+    },
+    {
+      "epoch": 0.39185843953424404,
+      "learning_rate": 0.00012163014577794078,
+      "loss": 0.8276,
+      "mean_token_accuracy": 0.7469102501869201,
+      "num_tokens": 98777466.0,
+      "step": 42740
+    },
+    {
+      "epoch": 0.3919501237737233,
+      "learning_rate": 0.00012161180893004494,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7467431008815766,
+      "num_tokens": 98800678.0,
+      "step": 42750
+    },
+    {
+      "epoch": 0.39204180801320254,
+      "learning_rate": 0.00012159347208214909,
+      "loss": 0.8514,
+      "mean_token_accuracy": 0.7480627417564392,
+      "num_tokens": 98823655.0,
+      "step": 42760
+    },
+    {
+      "epoch": 0.39213349225268174,
+      "learning_rate": 0.00012157513523425324,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7421223044395446,
+      "num_tokens": 98846518.0,
+      "step": 42770
+    },
+    {
+      "epoch": 0.392225176492161,
+      "learning_rate": 0.00012155679838635738,
+      "loss": 0.8477,
+      "mean_token_accuracy": 0.7416431248188019,
+      "num_tokens": 98869526.0,
+      "step": 42780
+    },
+    {
+      "epoch": 0.39231686073164024,
+      "learning_rate": 0.00012153846153846153,
+      "loss": 0.8112,
+      "mean_token_accuracy": 0.7553065061569214,
+      "num_tokens": 98892933.0,
+      "step": 42790
+    },
+    {
+      "epoch": 0.3924085449711195,
+      "learning_rate": 0.00012152012469056571,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7412650406360626,
+      "num_tokens": 98917050.0,
+      "step": 42800
+    },
+    {
+      "epoch": 0.3925002292105987,
+      "learning_rate": 0.00012150178784266986,
+      "loss": 0.8334,
+      "mean_token_accuracy": 0.7481203973293304,
+      "num_tokens": 98940712.0,
+      "step": 42810
+    },
+    {
+      "epoch": 0.39259191345007793,
+      "learning_rate": 0.00012148345099477401,
+      "loss": 0.8293,
+      "mean_token_accuracy": 0.7494286298751831,
+      "num_tokens": 98963947.0,
+      "step": 42820
+    },
+    {
+      "epoch": 0.3926835976895572,
+      "learning_rate": 0.00012146511414687815,
+      "loss": 0.8074,
+      "mean_token_accuracy": 0.750173556804657,
+      "num_tokens": 98987200.0,
+      "step": 42830
+    },
+    {
+      "epoch": 0.3927752819290364,
+      "learning_rate": 0.0001214467772989823,
+      "loss": 0.868,
+      "mean_token_accuracy": 0.7378338158130646,
+      "num_tokens": 99009578.0,
+      "step": 42840
+    },
+    {
+      "epoch": 0.39286696616851563,
+      "learning_rate": 0.00012142844045108645,
+      "loss": 0.8632,
+      "mean_token_accuracy": 0.7450571894645691,
+      "num_tokens": 99032320.0,
+      "step": 42850
+    },
+    {
+      "epoch": 0.3929586504079949,
+      "learning_rate": 0.00012141010360319063,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7450294673442841,
+      "num_tokens": 99055223.0,
+      "step": 42860
+    },
+    {
+      "epoch": 0.3930503346474741,
+      "learning_rate": 0.00012139176675529477,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7470239520072937,
+      "num_tokens": 99078331.0,
+      "step": 42870
+    },
+    {
+      "epoch": 0.3931420188869533,
+      "learning_rate": 0.00012137342990739892,
+      "loss": 0.8092,
+      "mean_token_accuracy": 0.7598694741725922,
+      "num_tokens": 99101696.0,
+      "step": 42880
+    },
+    {
+      "epoch": 0.3932337031264326,
+      "learning_rate": 0.00012135509305950308,
+      "loss": 0.827,
+      "mean_token_accuracy": 0.7489305198192596,
+      "num_tokens": 99124755.0,
+      "step": 42890
+    },
+    {
+      "epoch": 0.3933253873659118,
+      "learning_rate": 0.00012133675621160723,
+      "loss": 0.8434,
+      "mean_token_accuracy": 0.7529129922389984,
+      "num_tokens": 99147473.0,
+      "step": 42900
+    },
+    {
+      "epoch": 0.393417071605391,
+      "learning_rate": 0.00012131841936371138,
+      "loss": 0.8458,
+      "mean_token_accuracy": 0.7490242481231689,
+      "num_tokens": 99171225.0,
+      "step": 42910
+    },
+    {
+      "epoch": 0.3935087558448703,
+      "learning_rate": 0.00012130008251581553,
+      "loss": 0.8284,
+      "mean_token_accuracy": 0.7435774505138397,
+      "num_tokens": 99195536.0,
+      "step": 42920
+    },
+    {
+      "epoch": 0.3936004400843495,
+      "learning_rate": 0.0001212817456679197,
+      "loss": 0.8307,
+      "mean_token_accuracy": 0.7494882762432098,
+      "num_tokens": 99218147.0,
+      "step": 42930
+    },
+    {
+      "epoch": 0.3936921243238287,
+      "learning_rate": 0.00012126340882002385,
+      "loss": 0.8304,
+      "mean_token_accuracy": 0.7515585660934448,
+      "num_tokens": 99240637.0,
+      "step": 42940
+    },
+    {
+      "epoch": 0.39378380856330797,
+      "learning_rate": 0.000121245071972128,
+      "loss": 0.845,
+      "mean_token_accuracy": 0.7488064289093017,
+      "num_tokens": 99263733.0,
+      "step": 42950
+    },
+    {
+      "epoch": 0.3938754928027872,
+      "learning_rate": 0.00012122673512423215,
+      "loss": 0.9101,
+      "mean_token_accuracy": 0.7396009862422943,
+      "num_tokens": 99286473.0,
+      "step": 42960
+    },
+    {
+      "epoch": 0.3939671770422664,
+      "learning_rate": 0.0001212083982763363,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.743252569437027,
+      "num_tokens": 99309354.0,
+      "step": 42970
+    },
+    {
+      "epoch": 0.39405886128174566,
+      "learning_rate": 0.00012119006142844044,
+      "loss": 0.816,
+      "mean_token_accuracy": 0.7507643580436707,
+      "num_tokens": 99331905.0,
+      "step": 42980
+    },
+    {
+      "epoch": 0.3941505455212249,
+      "learning_rate": 0.00012117172458054462,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.746544623374939,
+      "num_tokens": 99354264.0,
+      "step": 42990
+    },
+    {
+      "epoch": 0.3942422297607041,
+      "learning_rate": 0.00012115338773264877,
+      "loss": 0.8609,
+      "mean_token_accuracy": 0.7427237629890442,
+      "num_tokens": 99377357.0,
+      "step": 43000
+    },
+    {
+      "epoch": 0.39433391400018336,
+      "learning_rate": 0.00012113505088475293,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7417212784290313,
+      "num_tokens": 99400242.0,
+      "step": 43010
+    },
+    {
+      "epoch": 0.3944255982396626,
+      "learning_rate": 0.00012111671403685708,
+      "loss": 0.842,
+      "mean_token_accuracy": 0.7445471584796906,
+      "num_tokens": 99423294.0,
+      "step": 43020
+    },
+    {
+      "epoch": 0.39451728247914186,
+      "learning_rate": 0.00012109837718896122,
+      "loss": 0.8659,
+      "mean_token_accuracy": 0.7416339635848999,
+      "num_tokens": 99445823.0,
+      "step": 43030
+    },
+    {
+      "epoch": 0.39460896671862106,
+      "learning_rate": 0.00012108004034106537,
+      "loss": 0.8688,
+      "mean_token_accuracy": 0.7422155618667603,
+      "num_tokens": 99468454.0,
+      "step": 43040
+    },
+    {
+      "epoch": 0.3947006509581003,
+      "learning_rate": 0.00012106170349316952,
+      "loss": 0.8361,
+      "mean_token_accuracy": 0.7583410799503326,
+      "num_tokens": 99491409.0,
+      "step": 43050
+    },
+    {
+      "epoch": 0.39479233519757956,
+      "learning_rate": 0.0001210433666452737,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.7550324678421021,
+      "num_tokens": 99514341.0,
+      "step": 43060
+    },
+    {
+      "epoch": 0.39488401943705875,
+      "learning_rate": 0.00012102502979737784,
+      "loss": 0.8803,
+      "mean_token_accuracy": 0.7441314160823822,
+      "num_tokens": 99537581.0,
+      "step": 43070
+    },
+    {
+      "epoch": 0.394975703676538,
+      "learning_rate": 0.00012100669294948199,
+      "loss": 0.8515,
+      "mean_token_accuracy": 0.7438615500926972,
+      "num_tokens": 99560268.0,
+      "step": 43080
+    },
+    {
+      "epoch": 0.39506738791601725,
+      "learning_rate": 0.00012098835610158614,
+      "loss": 0.8457,
+      "mean_token_accuracy": 0.7410091102123261,
+      "num_tokens": 99583320.0,
+      "step": 43090
+    },
+    {
+      "epoch": 0.39515907215549645,
+      "learning_rate": 0.00012097001925369029,
+      "loss": 0.8483,
+      "mean_token_accuracy": 0.7494796335697174,
+      "num_tokens": 99606247.0,
+      "step": 43100
+    },
+    {
+      "epoch": 0.3952507563949757,
+      "learning_rate": 0.00012095168240579444,
+      "loss": 0.8149,
+      "mean_token_accuracy": 0.7595623373985291,
+      "num_tokens": 99629914.0,
+      "step": 43110
+    },
+    {
+      "epoch": 0.39534244063445495,
+      "learning_rate": 0.00012093334555789861,
+      "loss": 0.8512,
+      "mean_token_accuracy": 0.7407776236534118,
+      "num_tokens": 99652816.0,
+      "step": 43120
+    },
+    {
+      "epoch": 0.39543412487393415,
+      "learning_rate": 0.00012091500871000276,
+      "loss": 0.8373,
+      "mean_token_accuracy": 0.7508548200130463,
+      "num_tokens": 99676031.0,
+      "step": 43130
+    },
+    {
+      "epoch": 0.3955258091134134,
+      "learning_rate": 0.00012089667186210691,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7446913063526154,
+      "num_tokens": 99699036.0,
+      "step": 43140
+    },
+    {
+      "epoch": 0.39561749335289265,
+      "learning_rate": 0.00012087833501421106,
+      "loss": 0.8713,
+      "mean_token_accuracy": 0.7402331829071045,
+      "num_tokens": 99722190.0,
+      "step": 43150
+    },
+    {
+      "epoch": 0.3957091775923719,
+      "learning_rate": 0.00012085999816631522,
+      "loss": 0.8153,
+      "mean_token_accuracy": 0.750926959514618,
+      "num_tokens": 99745313.0,
+      "step": 43160
+    },
+    {
+      "epoch": 0.3958008618318511,
+      "learning_rate": 0.00012084166131841937,
+      "loss": 0.8101,
+      "mean_token_accuracy": 0.7540000259876252,
+      "num_tokens": 99768305.0,
+      "step": 43170
+    },
+    {
+      "epoch": 0.39589254607133034,
+      "learning_rate": 0.00012082332447052351,
+      "loss": 0.8608,
+      "mean_token_accuracy": 0.7495555579662323,
+      "num_tokens": 99792280.0,
+      "step": 43180
+    },
+    {
+      "epoch": 0.3959842303108096,
+      "learning_rate": 0.00012080498762262769,
+      "loss": 0.8031,
+      "mean_token_accuracy": 0.7534538269042969,
+      "num_tokens": 99815372.0,
+      "step": 43190
+    },
+    {
+      "epoch": 0.3960759145502888,
+      "learning_rate": 0.00012078665077473184,
+      "loss": 0.88,
+      "mean_token_accuracy": 0.7427812039852142,
+      "num_tokens": 99838466.0,
+      "step": 43200
+    },
+    {
+      "epoch": 0.39616759878976804,
+      "learning_rate": 0.00012076831392683599,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.741842633485794,
+      "num_tokens": 99862328.0,
+      "step": 43210
+    },
+    {
+      "epoch": 0.3962592830292473,
+      "learning_rate": 0.00012074997707894014,
+      "loss": 0.8568,
+      "mean_token_accuracy": 0.7447823584079742,
+      "num_tokens": 99884941.0,
+      "step": 43220
+    },
+    {
+      "epoch": 0.3963509672687265,
+      "learning_rate": 0.00012073164023104428,
+      "loss": 0.8504,
+      "mean_token_accuracy": 0.7410430669784546,
+      "num_tokens": 99907642.0,
+      "step": 43230
+    },
+    {
+      "epoch": 0.39644265150820573,
+      "learning_rate": 0.00012071330338314843,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7453603327274323,
+      "num_tokens": 99930851.0,
+      "step": 43240
+    },
+    {
+      "epoch": 0.396534335747685,
+      "learning_rate": 0.00012069496653525258,
+      "loss": 0.8579,
+      "mean_token_accuracy": 0.7414930880069732,
+      "num_tokens": 99953855.0,
+      "step": 43250
+    },
+    {
+      "epoch": 0.3966260199871642,
+      "learning_rate": 0.00012067662968735676,
+      "loss": 0.8482,
+      "mean_token_accuracy": 0.7489110469818115,
+      "num_tokens": 99976786.0,
+      "step": 43260
+    },
+    {
+      "epoch": 0.39671770422664343,
+      "learning_rate": 0.0001206582928394609,
+      "loss": 0.8079,
+      "mean_token_accuracy": 0.7556454241275787,
+      "num_tokens": 99999249.0,
+      "step": 43270
+    },
+    {
+      "epoch": 0.3968093884661227,
+      "learning_rate": 0.00012063995599156505,
+      "loss": 0.8448,
+      "mean_token_accuracy": 0.7377425312995911,
+      "num_tokens": 100021933.0,
+      "step": 43280
+    },
+    {
+      "epoch": 0.39690107270560193,
+      "learning_rate": 0.0001206216191436692,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.760702782869339,
+      "num_tokens": 100045196.0,
+      "step": 43290
+    },
+    {
+      "epoch": 0.3969927569450811,
+      "learning_rate": 0.00012060328229577336,
+      "loss": 0.819,
+      "mean_token_accuracy": 0.744375479221344,
+      "num_tokens": 100068677.0,
+      "step": 43300
+    },
+    {
+      "epoch": 0.3970844411845604,
+      "learning_rate": 0.00012058494544787751,
+      "loss": 0.8532,
+      "mean_token_accuracy": 0.7433301329612731,
+      "num_tokens": 100092304.0,
+      "step": 43310
+    },
+    {
+      "epoch": 0.3971761254240396,
+      "learning_rate": 0.00012056660859998167,
+      "loss": 0.8163,
+      "mean_token_accuracy": 0.7581689298152924,
+      "num_tokens": 100115907.0,
+      "step": 43320
+    },
+    {
+      "epoch": 0.3972678096635188,
+      "learning_rate": 0.00012054827175208583,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7469379425048828,
+      "num_tokens": 100139640.0,
+      "step": 43330
+    },
+    {
+      "epoch": 0.3973594939029981,
+      "learning_rate": 0.00012052993490418998,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7534853637218475,
+      "num_tokens": 100162485.0,
+      "step": 43340
+    },
+    {
+      "epoch": 0.3974511781424773,
+      "learning_rate": 0.00012051159805629413,
+      "loss": 0.8403,
+      "mean_token_accuracy": 0.7501318275928497,
+      "num_tokens": 100185141.0,
+      "step": 43350
+    },
+    {
+      "epoch": 0.3975428623819565,
+      "learning_rate": 0.00012049326120839828,
+      "loss": 0.8592,
+      "mean_token_accuracy": 0.7472609996795654,
+      "num_tokens": 100207347.0,
+      "step": 43360
+    },
+    {
+      "epoch": 0.39763454662143577,
+      "learning_rate": 0.00012047492436050243,
+      "loss": 0.887,
+      "mean_token_accuracy": 0.7339904844760895,
+      "num_tokens": 100230845.0,
+      "step": 43370
+    },
+    {
+      "epoch": 0.397726230860915,
+      "learning_rate": 0.00012045658751260657,
+      "loss": 0.8466,
+      "mean_token_accuracy": 0.7437448918819427,
+      "num_tokens": 100253777.0,
+      "step": 43380
+    },
+    {
+      "epoch": 0.3978179151003942,
+      "learning_rate": 0.00012043825066471075,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7450472593307496,
+      "num_tokens": 100277783.0,
+      "step": 43390
+    },
+    {
+      "epoch": 0.39790959933987347,
+      "learning_rate": 0.0001204199138168149,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7582185566425323,
+      "num_tokens": 100300341.0,
+      "step": 43400
+    },
+    {
+      "epoch": 0.3980012835793527,
+      "learning_rate": 0.00012040157696891905,
+      "loss": 0.8411,
+      "mean_token_accuracy": 0.7432398974895478,
+      "num_tokens": 100323475.0,
+      "step": 43410
+    },
+    {
+      "epoch": 0.39809296781883197,
+      "learning_rate": 0.0001203832401210232,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7457428812980652,
+      "num_tokens": 100346552.0,
+      "step": 43420
+    },
+    {
+      "epoch": 0.39818465205831116,
+      "learning_rate": 0.00012036490327312734,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7546516418457031,
+      "num_tokens": 100370155.0,
+      "step": 43430
+    },
+    {
+      "epoch": 0.3982763362977904,
+      "learning_rate": 0.0001203465664252315,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7470997512340546,
+      "num_tokens": 100393240.0,
+      "step": 43440
+    },
+    {
+      "epoch": 0.39836802053726966,
+      "learning_rate": 0.00012032822957733568,
+      "loss": 0.8607,
+      "mean_token_accuracy": 0.7425906300544739,
+      "num_tokens": 100416700.0,
+      "step": 43450
+    },
+    {
+      "epoch": 0.39845970477674886,
+      "learning_rate": 0.00012030989272943983,
+      "loss": 0.8386,
+      "mean_token_accuracy": 0.7485748767852783,
+      "num_tokens": 100441329.0,
+      "step": 43460
+    },
+    {
+      "epoch": 0.3985513890162281,
+      "learning_rate": 0.00012029155588154397,
+      "loss": 0.8715,
+      "mean_token_accuracy": 0.7444190859794617,
+      "num_tokens": 100464254.0,
+      "step": 43470
+    },
+    {
+      "epoch": 0.39864307325570736,
+      "learning_rate": 0.00012027321903364812,
+      "loss": 0.8355,
+      "mean_token_accuracy": 0.7469911396503448,
+      "num_tokens": 100486921.0,
+      "step": 43480
+    },
+    {
+      "epoch": 0.39873475749518655,
+      "learning_rate": 0.00012025488218575227,
+      "loss": 0.8213,
+      "mean_token_accuracy": 0.748364919424057,
+      "num_tokens": 100509643.0,
+      "step": 43490
+    },
+    {
+      "epoch": 0.3988264417346658,
+      "learning_rate": 0.00012023654533785642,
+      "loss": 0.8507,
+      "mean_token_accuracy": 0.7430125772953033,
+      "num_tokens": 100533152.0,
+      "step": 43500
+    },
+    {
+      "epoch": 0.39891812597414505,
+      "learning_rate": 0.00012021820848996057,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7426909685134888,
+      "num_tokens": 100556596.0,
+      "step": 43510
+    },
+    {
+      "epoch": 0.3990098102136243,
+      "learning_rate": 0.00012019987164206474,
+      "loss": 0.8448,
+      "mean_token_accuracy": 0.7450658738613128,
+      "num_tokens": 100580296.0,
+      "step": 43520
+    },
+    {
+      "epoch": 0.3991014944531035,
+      "learning_rate": 0.00012018153479416889,
+      "loss": 0.8528,
+      "mean_token_accuracy": 0.7435936272144318,
+      "num_tokens": 100602819.0,
+      "step": 43530
+    },
+    {
+      "epoch": 0.39919317869258275,
+      "learning_rate": 0.00012016319794627304,
+      "loss": 0.8497,
+      "mean_token_accuracy": 0.7447150588035584,
+      "num_tokens": 100625938.0,
+      "step": 43540
+    },
+    {
+      "epoch": 0.399284862932062,
+      "learning_rate": 0.0001201448610983772,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7477009654045105,
+      "num_tokens": 100649270.0,
+      "step": 43550
+    },
+    {
+      "epoch": 0.3993765471715412,
+      "learning_rate": 0.00012012652425048135,
+      "loss": 0.803,
+      "mean_token_accuracy": 0.7549422085285187,
+      "num_tokens": 100671672.0,
+      "step": 43560
+    },
+    {
+      "epoch": 0.39946823141102045,
+      "learning_rate": 0.0001201081874025855,
+      "loss": 0.8285,
+      "mean_token_accuracy": 0.7511558353900909,
+      "num_tokens": 100694873.0,
+      "step": 43570
+    },
+    {
+      "epoch": 0.3995599156504997,
+      "learning_rate": 0.00012008985055468966,
+      "loss": 0.821,
+      "mean_token_accuracy": 0.7527211785316468,
+      "num_tokens": 100718075.0,
+      "step": 43580
+    },
+    {
+      "epoch": 0.3996515998899789,
+      "learning_rate": 0.00012007151370679381,
+      "loss": 0.8583,
+      "mean_token_accuracy": 0.7468614757061005,
+      "num_tokens": 100741696.0,
+      "step": 43590
+    },
+    {
+      "epoch": 0.39974328412945814,
+      "learning_rate": 0.00012005317685889797,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7396007180213928,
+      "num_tokens": 100765060.0,
+      "step": 43600
+    },
+    {
+      "epoch": 0.3998349683689374,
+      "learning_rate": 0.00012003484001100212,
+      "loss": 0.8714,
+      "mean_token_accuracy": 0.7453867971897126,
+      "num_tokens": 100787963.0,
+      "step": 43610
+    },
+    {
+      "epoch": 0.3999266526084166,
+      "learning_rate": 0.00012001650316310627,
+      "loss": 0.8341,
+      "mean_token_accuracy": 0.747481620311737,
+      "num_tokens": 100811178.0,
+      "step": 43620
+    },
+    {
+      "epoch": 0.40001833684789584,
+      "learning_rate": 0.00011999816631521041,
+      "loss": 0.7975,
+      "mean_token_accuracy": 0.7568912744522095,
+      "num_tokens": 100834528.0,
+      "step": 43630
+    },
+    {
+      "epoch": 0.4001100210873751,
+      "learning_rate": 0.00011997982946731456,
+      "loss": 0.8208,
+      "mean_token_accuracy": 0.7569917798042297,
+      "num_tokens": 100857446.0,
+      "step": 43640
+    },
+    {
+      "epoch": 0.40020170532685434,
+      "learning_rate": 0.00011996149261941874,
+      "loss": 0.8304,
+      "mean_token_accuracy": 0.7474347651004791,
+      "num_tokens": 100880370.0,
+      "step": 43650
+    },
+    {
+      "epoch": 0.40029338956633354,
+      "learning_rate": 0.00011994315577152289,
+      "loss": 0.8183,
+      "mean_token_accuracy": 0.7509926438331604,
+      "num_tokens": 100903945.0,
+      "step": 43660
+    },
+    {
+      "epoch": 0.4003850738058128,
+      "learning_rate": 0.00011992481892362703,
+      "loss": 0.8228,
+      "mean_token_accuracy": 0.7485302925109864,
+      "num_tokens": 100928168.0,
+      "step": 43670
+    },
+    {
+      "epoch": 0.40047675804529204,
+      "learning_rate": 0.00011990648207573118,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.7437901318073272,
+      "num_tokens": 100951180.0,
+      "step": 43680
+    },
+    {
+      "epoch": 0.40056844228477123,
+      "learning_rate": 0.00011988814522783533,
+      "loss": 0.8741,
+      "mean_token_accuracy": 0.7450493574142456,
+      "num_tokens": 100975076.0,
+      "step": 43690
+    },
+    {
+      "epoch": 0.4006601265242505,
+      "learning_rate": 0.00011986980837993949,
+      "loss": 0.8482,
+      "mean_token_accuracy": 0.7441963791847229,
+      "num_tokens": 100997457.0,
+      "step": 43700
+    },
+    {
+      "epoch": 0.40075181076372973,
+      "learning_rate": 0.00011985147153204366,
+      "loss": 0.8509,
+      "mean_token_accuracy": 0.7498200833797455,
+      "num_tokens": 101019465.0,
+      "step": 43710
+    },
+    {
+      "epoch": 0.4008434950032089,
+      "learning_rate": 0.0001198331346841478,
+      "loss": 0.8061,
+      "mean_token_accuracy": 0.7549359560012817,
+      "num_tokens": 101041935.0,
+      "step": 43720
+    },
+    {
+      "epoch": 0.4009351792426882,
+      "learning_rate": 0.00011981479783625195,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.7457621693611145,
+      "num_tokens": 101064399.0,
+      "step": 43730
+    },
+    {
+      "epoch": 0.40102686348216743,
+      "learning_rate": 0.0001197964609883561,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7421365141868591,
+      "num_tokens": 101087211.0,
+      "step": 43740
+    },
+    {
+      "epoch": 0.4011185477216466,
+      "learning_rate": 0.00011977812414046026,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7481885135173798,
+      "num_tokens": 101110590.0,
+      "step": 43750
+    },
+    {
+      "epoch": 0.4012102319611259,
+      "learning_rate": 0.00011975978729256441,
+      "loss": 0.8816,
+      "mean_token_accuracy": 0.7407104909420014,
+      "num_tokens": 101132886.0,
+      "step": 43760
+    },
+    {
+      "epoch": 0.4013019162006051,
+      "learning_rate": 0.00011974145044466856,
+      "loss": 0.8582,
+      "mean_token_accuracy": 0.7441464364528656,
+      "num_tokens": 101156067.0,
+      "step": 43770
+    },
+    {
+      "epoch": 0.4013936004400844,
+      "learning_rate": 0.00011972311359677273,
+      "loss": 0.8721,
+      "mean_token_accuracy": 0.7364722907543182,
+      "num_tokens": 101178662.0,
+      "step": 43780
+    },
+    {
+      "epoch": 0.40148528467956357,
+      "learning_rate": 0.00011970477674887688,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7392441511154175,
+      "num_tokens": 101201467.0,
+      "step": 43790
+    },
+    {
+      "epoch": 0.4015769689190428,
+      "learning_rate": 0.00011968643990098103,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7440224051475525,
+      "num_tokens": 101223977.0,
+      "step": 43800
+    },
+    {
+      "epoch": 0.40166865315852207,
+      "learning_rate": 0.00011966810305308518,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7479540884494782,
+      "num_tokens": 101247890.0,
+      "step": 43810
+    },
+    {
+      "epoch": 0.40176033739800127,
+      "learning_rate": 0.00011964976620518933,
+      "loss": 0.8483,
+      "mean_token_accuracy": 0.7450882971286774,
+      "num_tokens": 101271332.0,
+      "step": 43820
+    },
+    {
+      "epoch": 0.4018520216374805,
+      "learning_rate": 0.00011963142935729347,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7435620367527008,
+      "num_tokens": 101294299.0,
+      "step": 43830
+    },
+    {
+      "epoch": 0.40194370587695977,
+      "learning_rate": 0.00011961309250939765,
+      "loss": 0.8206,
+      "mean_token_accuracy": 0.7506714522838592,
+      "num_tokens": 101316883.0,
+      "step": 43840
+    },
+    {
+      "epoch": 0.40203539011643896,
+      "learning_rate": 0.0001195947556615018,
+      "loss": 0.8718,
+      "mean_token_accuracy": 0.7427827298641205,
+      "num_tokens": 101339963.0,
+      "step": 43850
+    },
+    {
+      "epoch": 0.4021270743559182,
+      "learning_rate": 0.00011957641881360596,
+      "loss": 0.8667,
+      "mean_token_accuracy": 0.7400134563446045,
+      "num_tokens": 101362611.0,
+      "step": 43860
+    },
+    {
+      "epoch": 0.40221875859539746,
+      "learning_rate": 0.0001195580819657101,
+      "loss": 0.806,
+      "mean_token_accuracy": 0.7647567331790924,
+      "num_tokens": 101385865.0,
+      "step": 43870
+    },
+    {
+      "epoch": 0.40231044283487666,
+      "learning_rate": 0.00011953974511781425,
+      "loss": 0.8297,
+      "mean_token_accuracy": 0.7499696552753449,
+      "num_tokens": 101409036.0,
+      "step": 43880
+    },
+    {
+      "epoch": 0.4024021270743559,
+      "learning_rate": 0.0001195214082699184,
+      "loss": 0.8476,
+      "mean_token_accuracy": 0.7489574372768402,
+      "num_tokens": 101432714.0,
+      "step": 43890
+    },
+    {
+      "epoch": 0.40249381131383516,
+      "learning_rate": 0.00011950307142202255,
+      "loss": 0.838,
+      "mean_token_accuracy": 0.7476186215877533,
+      "num_tokens": 101455873.0,
+      "step": 43900
+    },
+    {
+      "epoch": 0.4025854955533144,
+      "learning_rate": 0.00011948473457412673,
+      "loss": 0.8823,
+      "mean_token_accuracy": 0.7406718492507934,
+      "num_tokens": 101479168.0,
+      "step": 43910
+    },
+    {
+      "epoch": 0.4026771797927936,
+      "learning_rate": 0.00011946639772623087,
+      "loss": 0.802,
+      "mean_token_accuracy": 0.7554150402545929,
+      "num_tokens": 101502172.0,
+      "step": 43920
+    },
+    {
+      "epoch": 0.40276886403227286,
+      "learning_rate": 0.00011944806087833502,
+      "loss": 0.8512,
+      "mean_token_accuracy": 0.7441450178623199,
+      "num_tokens": 101524795.0,
+      "step": 43930
+    },
+    {
+      "epoch": 0.4028605482717521,
+      "learning_rate": 0.00011942972403043917,
+      "loss": 0.8607,
+      "mean_token_accuracy": 0.7382725477218628,
+      "num_tokens": 101547878.0,
+      "step": 43940
+    },
+    {
+      "epoch": 0.4029522325112313,
+      "learning_rate": 0.00011941138718254332,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7430664479732514,
+      "num_tokens": 101571696.0,
+      "step": 43950
+    },
+    {
+      "epoch": 0.40304391675071055,
+      "learning_rate": 0.00011939305033464747,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7453457295894623,
+      "num_tokens": 101595070.0,
+      "step": 43960
+    },
+    {
+      "epoch": 0.4031356009901898,
+      "learning_rate": 0.00011937471348675164,
+      "loss": 0.8905,
+      "mean_token_accuracy": 0.7316078066825866,
+      "num_tokens": 101618211.0,
+      "step": 43970
+    },
+    {
+      "epoch": 0.403227285229669,
+      "learning_rate": 0.00011935637663885579,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7397210478782654,
+      "num_tokens": 101641912.0,
+      "step": 43980
+    },
+    {
+      "epoch": 0.40331896946914825,
+      "learning_rate": 0.00011933803979095994,
+      "loss": 0.8556,
+      "mean_token_accuracy": 0.7466451168060303,
+      "num_tokens": 101665133.0,
+      "step": 43990
+    },
+    {
+      "epoch": 0.4034106537086275,
+      "learning_rate": 0.0001193197029430641,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7544361591339112,
+      "num_tokens": 101687839.0,
+      "step": 44000
+    },
+    {
+      "epoch": 0.4035023379481067,
+      "learning_rate": 0.00011930136609516825,
+      "loss": 0.83,
+      "mean_token_accuracy": 0.7543080925941468,
+      "num_tokens": 101711127.0,
+      "step": 44010
+    },
+    {
+      "epoch": 0.40359402218758594,
+      "learning_rate": 0.0001192830292472724,
+      "loss": 0.8526,
+      "mean_token_accuracy": 0.7467760264873504,
+      "num_tokens": 101734011.0,
+      "step": 44020
+    },
+    {
+      "epoch": 0.4036857064270652,
+      "learning_rate": 0.00011926469239937654,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7457446813583374,
+      "num_tokens": 101756746.0,
+      "step": 44030
+    },
+    {
+      "epoch": 0.40377739066654444,
+      "learning_rate": 0.00011924635555148072,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.747233772277832,
+      "num_tokens": 101779213.0,
+      "step": 44040
+    },
+    {
+      "epoch": 0.40386907490602364,
+      "learning_rate": 0.00011922801870358487,
+      "loss": 0.8582,
+      "mean_token_accuracy": 0.745703935623169,
+      "num_tokens": 101802158.0,
+      "step": 44050
+    },
+    {
+      "epoch": 0.4039607591455029,
+      "learning_rate": 0.00011920968185568902,
+      "loss": 0.8628,
+      "mean_token_accuracy": 0.7423049509525299,
+      "num_tokens": 101826496.0,
+      "step": 44060
+    },
+    {
+      "epoch": 0.40405244338498214,
+      "learning_rate": 0.00011919134500779316,
+      "loss": 0.8436,
+      "mean_token_accuracy": 0.7474491834640503,
+      "num_tokens": 101850270.0,
+      "step": 44070
+    },
+    {
+      "epoch": 0.40414412762446134,
+      "learning_rate": 0.00011917300815989731,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7463626205921173,
+      "num_tokens": 101873556.0,
+      "step": 44080
+    },
+    {
+      "epoch": 0.4042358118639406,
+      "learning_rate": 0.00011915467131200146,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7486437380313873,
+      "num_tokens": 101895943.0,
+      "step": 44090
+    },
+    {
+      "epoch": 0.40432749610341984,
+      "learning_rate": 0.00011913633446410564,
+      "loss": 0.8374,
+      "mean_token_accuracy": 0.7513238906860351,
+      "num_tokens": 101918519.0,
+      "step": 44100
+    },
+    {
+      "epoch": 0.40441918034289903,
+      "learning_rate": 0.0001191179976162098,
+      "loss": 0.8215,
+      "mean_token_accuracy": 0.7529733896255493,
+      "num_tokens": 101941660.0,
+      "step": 44110
+    },
+    {
+      "epoch": 0.4045108645823783,
+      "learning_rate": 0.00011909966076831393,
+      "loss": 0.824,
+      "mean_token_accuracy": 0.7548643171787262,
+      "num_tokens": 101965299.0,
+      "step": 44120
+    },
+    {
+      "epoch": 0.40460254882185753,
+      "learning_rate": 0.00011908132392041808,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7420718908309937,
+      "num_tokens": 101987681.0,
+      "step": 44130
+    },
+    {
+      "epoch": 0.4046942330613368,
+      "learning_rate": 0.00011906298707252224,
+      "loss": 0.8319,
+      "mean_token_accuracy": 0.746285080909729,
+      "num_tokens": 102011072.0,
+      "step": 44140
+    },
+    {
+      "epoch": 0.404785917300816,
+      "learning_rate": 0.00011904465022462639,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.734488719701767,
+      "num_tokens": 102034292.0,
+      "step": 44150
+    },
+    {
+      "epoch": 0.40487760154029523,
+      "learning_rate": 0.00011902631337673054,
+      "loss": 0.8133,
+      "mean_token_accuracy": 0.7553673982620239,
+      "num_tokens": 102057192.0,
+      "step": 44160
+    },
+    {
+      "epoch": 0.4049692857797745,
+      "learning_rate": 0.0001190079765288347,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.747709047794342,
+      "num_tokens": 102081059.0,
+      "step": 44170
+    },
+    {
+      "epoch": 0.4050609700192537,
+      "learning_rate": 0.00011898963968093886,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7444964587688446,
+      "num_tokens": 102103572.0,
+      "step": 44180
+    },
+    {
+      "epoch": 0.4051526542587329,
+      "learning_rate": 0.00011897130283304301,
+      "loss": 0.8652,
+      "mean_token_accuracy": 0.7443179070949555,
+      "num_tokens": 102126442.0,
+      "step": 44190
+    },
+    {
+      "epoch": 0.4052443384982122,
+      "learning_rate": 0.00011895296598514716,
+      "loss": 0.8333,
+      "mean_token_accuracy": 0.7456013679504394,
+      "num_tokens": 102150062.0,
+      "step": 44200
+    },
+    {
+      "epoch": 0.40533602273769137,
+      "learning_rate": 0.00011893462913725131,
+      "loss": 0.8189,
+      "mean_token_accuracy": 0.7510856688022614,
+      "num_tokens": 102172765.0,
+      "step": 44210
+    },
+    {
+      "epoch": 0.4054277069771706,
+      "learning_rate": 0.00011891629228935546,
+      "loss": 0.8832,
+      "mean_token_accuracy": 0.7396148025989533,
+      "num_tokens": 102195726.0,
+      "step": 44220
+    },
+    {
+      "epoch": 0.40551939121664987,
+      "learning_rate": 0.00011889795544145963,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7436016976833344,
+      "num_tokens": 102218803.0,
+      "step": 44230
+    },
+    {
+      "epoch": 0.40561107545612907,
+      "learning_rate": 0.00011887961859356378,
+      "loss": 0.8285,
+      "mean_token_accuracy": 0.7495310842990875,
+      "num_tokens": 102241640.0,
+      "step": 44240
+    },
+    {
+      "epoch": 0.4057027596956083,
+      "learning_rate": 0.00011886128174566793,
+      "loss": 0.84,
+      "mean_token_accuracy": 0.744154554605484,
+      "num_tokens": 102264526.0,
+      "step": 44250
+    },
+    {
+      "epoch": 0.40579444393508757,
+      "learning_rate": 0.00011884294489777208,
+      "loss": 0.86,
+      "mean_token_accuracy": 0.7446904242038727,
+      "num_tokens": 102287369.0,
+      "step": 44260
+    },
+    {
+      "epoch": 0.4058861281745668,
+      "learning_rate": 0.00011882460804987622,
+      "loss": 0.8619,
+      "mean_token_accuracy": 0.7469142735004425,
+      "num_tokens": 102309206.0,
+      "step": 44270
+    },
+    {
+      "epoch": 0.405977812414046,
+      "learning_rate": 0.00011880627120198037,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7514234960079194,
+      "num_tokens": 102332607.0,
+      "step": 44280
+    },
+    {
+      "epoch": 0.40606949665352526,
+      "learning_rate": 0.00011878793435408453,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7492182672023773,
+      "num_tokens": 102355420.0,
+      "step": 44290
+    },
+    {
+      "epoch": 0.4061611808930045,
+      "learning_rate": 0.0001187695975061887,
+      "loss": 0.8296,
+      "mean_token_accuracy": 0.7488252580165863,
+      "num_tokens": 102378099.0,
+      "step": 44300
+    },
+    {
+      "epoch": 0.4062528651324837,
+      "learning_rate": 0.00011875126065829286,
+      "loss": 0.8359,
+      "mean_token_accuracy": 0.7438942015171051,
+      "num_tokens": 102400457.0,
+      "step": 44310
+    },
+    {
+      "epoch": 0.40634454937196296,
+      "learning_rate": 0.000118732923810397,
+      "loss": 0.8376,
+      "mean_token_accuracy": 0.7458237946033478,
+      "num_tokens": 102423634.0,
+      "step": 44320
+    },
+    {
+      "epoch": 0.4064362336114422,
+      "learning_rate": 0.00011871458696250115,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7501737952232361,
+      "num_tokens": 102446648.0,
+      "step": 44330
+    },
+    {
+      "epoch": 0.4065279178509214,
+      "learning_rate": 0.0001186962501146053,
+      "loss": 0.8217,
+      "mean_token_accuracy": 0.7535607576370239,
+      "num_tokens": 102469286.0,
+      "step": 44340
+    },
+    {
+      "epoch": 0.40661960209040066,
+      "learning_rate": 0.00011867791326670945,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.745497316122055,
+      "num_tokens": 102492018.0,
+      "step": 44350
+    },
+    {
+      "epoch": 0.4067112863298799,
+      "learning_rate": 0.0001186595764188136,
+      "loss": 0.8541,
+      "mean_token_accuracy": 0.7416003167629241,
+      "num_tokens": 102515303.0,
+      "step": 44360
+    },
+    {
+      "epoch": 0.4068029705693591,
+      "learning_rate": 0.00011864123957091777,
+      "loss": 0.8378,
+      "mean_token_accuracy": 0.7473686873912812,
+      "num_tokens": 102538681.0,
+      "step": 44370
+    },
+    {
+      "epoch": 0.40689465480883835,
+      "learning_rate": 0.00011862290272302192,
+      "loss": 0.8094,
+      "mean_token_accuracy": 0.7564398288726807,
+      "num_tokens": 102561937.0,
+      "step": 44380
+    },
+    {
+      "epoch": 0.4069863390483176,
+      "learning_rate": 0.00011860456587512607,
+      "loss": 0.8623,
+      "mean_token_accuracy": 0.7400487780570983,
+      "num_tokens": 102586449.0,
+      "step": 44390
+    },
+    {
+      "epoch": 0.40707802328779685,
+      "learning_rate": 0.00011858622902723022,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7529193341732026,
+      "num_tokens": 102609825.0,
+      "step": 44400
+    },
+    {
+      "epoch": 0.40716970752727605,
+      "learning_rate": 0.00011856789217933438,
+      "loss": 0.8672,
+      "mean_token_accuracy": 0.7442286550998688,
+      "num_tokens": 102632464.0,
+      "step": 44410
+    },
+    {
+      "epoch": 0.4072613917667553,
+      "learning_rate": 0.00011854955533143853,
+      "loss": 0.8463,
+      "mean_token_accuracy": 0.7453698098659516,
+      "num_tokens": 102655420.0,
+      "step": 44420
+    },
+    {
+      "epoch": 0.40735307600623455,
+      "learning_rate": 0.0001185312184835427,
+      "loss": 0.8175,
+      "mean_token_accuracy": 0.7565670073032379,
+      "num_tokens": 102677858.0,
+      "step": 44430
+    },
+    {
+      "epoch": 0.40744476024571374,
+      "learning_rate": 0.00011851288163564685,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7486736238002777,
+      "num_tokens": 102701173.0,
+      "step": 44440
+    },
+    {
+      "epoch": 0.407536444485193,
+      "learning_rate": 0.000118494544787751,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.7465475678443909,
+      "num_tokens": 102723990.0,
+      "step": 44450
+    },
+    {
+      "epoch": 0.40762812872467225,
+      "learning_rate": 0.00011847620793985515,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7493509948253632,
+      "num_tokens": 102746923.0,
+      "step": 44460
+    },
+    {
+      "epoch": 0.40771981296415144,
+      "learning_rate": 0.00011845787109195929,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.7481965899467469,
+      "num_tokens": 102769325.0,
+      "step": 44470
+    },
+    {
+      "epoch": 0.4078114972036307,
+      "learning_rate": 0.00011843953424406344,
+      "loss": 0.8743,
+      "mean_token_accuracy": 0.739555686712265,
+      "num_tokens": 102791877.0,
+      "step": 44480
+    },
+    {
+      "epoch": 0.40790318144310994,
+      "learning_rate": 0.00011842119739616759,
+      "loss": 0.8601,
+      "mean_token_accuracy": 0.7448561131954193,
+      "num_tokens": 102815231.0,
+      "step": 44490
+    },
+    {
+      "epoch": 0.40799486568258914,
+      "learning_rate": 0.00011840286054827177,
+      "loss": 0.8747,
+      "mean_token_accuracy": 0.7401957511901855,
+      "num_tokens": 102838475.0,
+      "step": 44500
+    },
+    {
+      "epoch": 0.4080865499220684,
+      "learning_rate": 0.00011838452370037592,
+      "loss": 0.8054,
+      "mean_token_accuracy": 0.7538772940635681,
+      "num_tokens": 102862031.0,
+      "step": 44510
+    },
+    {
+      "epoch": 0.40817823416154764,
+      "learning_rate": 0.00011836618685248006,
+      "loss": 0.8771,
+      "mean_token_accuracy": 0.7395708858966827,
+      "num_tokens": 102884888.0,
+      "step": 44520
+    },
+    {
+      "epoch": 0.4082699184010269,
+      "learning_rate": 0.00011834785000458421,
+      "loss": 0.8437,
+      "mean_token_accuracy": 0.7475594997406005,
+      "num_tokens": 102907801.0,
+      "step": 44530
+    },
+    {
+      "epoch": 0.4083616026405061,
+      "learning_rate": 0.00011832951315668836,
+      "loss": 0.8337,
+      "mean_token_accuracy": 0.750956678390503,
+      "num_tokens": 102930883.0,
+      "step": 44540
+    },
+    {
+      "epoch": 0.40845328687998533,
+      "learning_rate": 0.00011831117630879252,
+      "loss": 0.8071,
+      "mean_token_accuracy": 0.756844425201416,
+      "num_tokens": 102954198.0,
+      "step": 44550
+    },
+    {
+      "epoch": 0.4085449711194646,
+      "learning_rate": 0.00011829283946089668,
+      "loss": 0.831,
+      "mean_token_accuracy": 0.7487766325473786,
+      "num_tokens": 102977390.0,
+      "step": 44560
+    },
+    {
+      "epoch": 0.4086366553589438,
+      "learning_rate": 0.00011827450261300083,
+      "loss": 0.8609,
+      "mean_token_accuracy": 0.7401157557964325,
+      "num_tokens": 102999947.0,
+      "step": 44570
+    },
+    {
+      "epoch": 0.40872833959842303,
+      "learning_rate": 0.00011825616576510499,
+      "loss": 0.8476,
+      "mean_token_accuracy": 0.748603880405426,
+      "num_tokens": 103023087.0,
+      "step": 44580
+    },
+    {
+      "epoch": 0.4088200238379023,
+      "learning_rate": 0.00011823782891720914,
+      "loss": 0.8179,
+      "mean_token_accuracy": 0.7481927216053009,
+      "num_tokens": 103046768.0,
+      "step": 44590
+    },
+    {
+      "epoch": 0.4089117080773815,
+      "learning_rate": 0.00011821949206931329,
+      "loss": 0.8276,
+      "mean_token_accuracy": 0.7536837041378022,
+      "num_tokens": 103069457.0,
+      "step": 44600
+    },
+    {
+      "epoch": 0.4090033923168607,
+      "learning_rate": 0.00011820115522141744,
+      "loss": 0.8702,
+      "mean_token_accuracy": 0.7421279847621918,
+      "num_tokens": 103092706.0,
+      "step": 44610
+    },
+    {
+      "epoch": 0.40909507655634,
+      "learning_rate": 0.00011818281837352159,
+      "loss": 0.8373,
+      "mean_token_accuracy": 0.7506707310676575,
+      "num_tokens": 103116056.0,
+      "step": 44620
+    },
+    {
+      "epoch": 0.40918676079581917,
+      "learning_rate": 0.00011816448152562576,
+      "loss": 0.8236,
+      "mean_token_accuracy": 0.7536875903606415,
+      "num_tokens": 103138614.0,
+      "step": 44630
+    },
+    {
+      "epoch": 0.4092784450352984,
+      "learning_rate": 0.00011814614467772991,
+      "loss": 0.8237,
+      "mean_token_accuracy": 0.7470583975315094,
+      "num_tokens": 103161558.0,
+      "step": 44640
+    },
+    {
+      "epoch": 0.40937012927477767,
+      "learning_rate": 0.00011812780782983406,
+      "loss": 0.8478,
+      "mean_token_accuracy": 0.74755819439888,
+      "num_tokens": 103183648.0,
+      "step": 44650
+    },
+    {
+      "epoch": 0.4094618135142569,
+      "learning_rate": 0.00011810947098193821,
+      "loss": 0.8389,
+      "mean_token_accuracy": 0.7449398815631867,
+      "num_tokens": 103207033.0,
+      "step": 44660
+    },
+    {
+      "epoch": 0.4095534977537361,
+      "learning_rate": 0.00011809113413404235,
+      "loss": 0.8576,
+      "mean_token_accuracy": 0.7453439235687256,
+      "num_tokens": 103230407.0,
+      "step": 44670
+    },
+    {
+      "epoch": 0.40964518199321537,
+      "learning_rate": 0.0001180727972861465,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.7497498154640198,
+      "num_tokens": 103254064.0,
+      "step": 44680
+    },
+    {
+      "epoch": 0.4097368662326946,
+      "learning_rate": 0.00011805446043825068,
+      "loss": 0.8721,
+      "mean_token_accuracy": 0.7430652916431427,
+      "num_tokens": 103277093.0,
+      "step": 44690
+    },
+    {
+      "epoch": 0.4098285504721738,
+      "learning_rate": 0.00011803612359035483,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.7504394054412842,
+      "num_tokens": 103300286.0,
+      "step": 44700
+    },
+    {
+      "epoch": 0.40992023471165306,
+      "learning_rate": 0.00011801778674245899,
+      "loss": 0.8309,
+      "mean_token_accuracy": 0.7491260170936584,
+      "num_tokens": 103323115.0,
+      "step": 44710
+    },
+    {
+      "epoch": 0.4100119189511323,
+      "learning_rate": 0.00011799944989456312,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.743862408399582,
+      "num_tokens": 103346547.0,
+      "step": 44720
+    },
+    {
+      "epoch": 0.4101036031906115,
+      "learning_rate": 0.00011798111304666728,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7455031991004943,
+      "num_tokens": 103369733.0,
+      "step": 44730
+    },
+    {
+      "epoch": 0.41019528743009076,
+      "learning_rate": 0.00011796277619877143,
+      "loss": 0.8236,
+      "mean_token_accuracy": 0.75127472281456,
+      "num_tokens": 103393322.0,
+      "step": 44740
+    },
+    {
+      "epoch": 0.41028697166957,
+      "learning_rate": 0.00011794443935087558,
+      "loss": 0.84,
+      "mean_token_accuracy": 0.7439639151096344,
+      "num_tokens": 103416118.0,
+      "step": 44750
+    },
+    {
+      "epoch": 0.41037865590904926,
+      "learning_rate": 0.00011792610250297975,
+      "loss": 0.8613,
+      "mean_token_accuracy": 0.7476392447948456,
+      "num_tokens": 103438843.0,
+      "step": 44760
+    },
+    {
+      "epoch": 0.41047034014852846,
+      "learning_rate": 0.0001179077656550839,
+      "loss": 0.8525,
+      "mean_token_accuracy": 0.7453845322132111,
+      "num_tokens": 103461877.0,
+      "step": 44770
+    },
+    {
+      "epoch": 0.4105620243880077,
+      "learning_rate": 0.00011788942880718805,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7461943209171296,
+      "num_tokens": 103484793.0,
+      "step": 44780
+    },
+    {
+      "epoch": 0.41065370862748696,
+      "learning_rate": 0.0001178710919592922,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7418680310249328,
+      "num_tokens": 103507457.0,
+      "step": 44790
+    },
+    {
+      "epoch": 0.41074539286696615,
+      "learning_rate": 0.00011785275511139635,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7420167028903961,
+      "num_tokens": 103530696.0,
+      "step": 44800
+    },
+    {
+      "epoch": 0.4108370771064454,
+      "learning_rate": 0.0001178344182635005,
+      "loss": 0.843,
+      "mean_token_accuracy": 0.7444317817687989,
+      "num_tokens": 103554093.0,
+      "step": 44810
+    },
+    {
+      "epoch": 0.41092876134592465,
+      "learning_rate": 0.00011781608141560467,
+      "loss": 0.8196,
+      "mean_token_accuracy": 0.7494815468788147,
+      "num_tokens": 103577543.0,
+      "step": 44820
+    },
+    {
+      "epoch": 0.41102044558540385,
+      "learning_rate": 0.00011779774456770882,
+      "loss": 0.8898,
+      "mean_token_accuracy": 0.7315061092376709,
+      "num_tokens": 103600184.0,
+      "step": 44830
+    },
+    {
+      "epoch": 0.4111121298248831,
+      "learning_rate": 0.00011777940771981297,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.7381754636764526,
+      "num_tokens": 103622962.0,
+      "step": 44840
+    },
+    {
+      "epoch": 0.41120381406436235,
+      "learning_rate": 0.00011776107087191713,
+      "loss": 0.8139,
+      "mean_token_accuracy": 0.7545563101768493,
+      "num_tokens": 103646700.0,
+      "step": 44850
+    },
+    {
+      "epoch": 0.41129549830384154,
+      "learning_rate": 0.00011774273402402128,
+      "loss": 0.8044,
+      "mean_token_accuracy": 0.7531153202056885,
+      "num_tokens": 103669148.0,
+      "step": 44860
+    },
+    {
+      "epoch": 0.4113871825433208,
+      "learning_rate": 0.00011772439717612542,
+      "loss": 0.8984,
+      "mean_token_accuracy": 0.7359162330627441,
+      "num_tokens": 103692313.0,
+      "step": 44870
+    },
+    {
+      "epoch": 0.41147886678280005,
+      "learning_rate": 0.00011770606032822957,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.7378570258617401,
+      "num_tokens": 103714806.0,
+      "step": 44880
+    },
+    {
+      "epoch": 0.4115705510222793,
+      "learning_rate": 0.00011768772348033375,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7548257350921631,
+      "num_tokens": 103737699.0,
+      "step": 44890
+    },
+    {
+      "epoch": 0.4116622352617585,
+      "learning_rate": 0.0001176693866324379,
+      "loss": 0.809,
+      "mean_token_accuracy": 0.7526199340820312,
+      "num_tokens": 103760162.0,
+      "step": 44900
+    },
+    {
+      "epoch": 0.41175391950123774,
+      "learning_rate": 0.00011765104978454205,
+      "loss": 0.8173,
+      "mean_token_accuracy": 0.7497694373130799,
+      "num_tokens": 103782907.0,
+      "step": 44910
+    },
+    {
+      "epoch": 0.411845603740717,
+      "learning_rate": 0.00011763271293664619,
+      "loss": 0.8624,
+      "mean_token_accuracy": 0.740943992137909,
+      "num_tokens": 103804971.0,
+      "step": 44920
+    },
+    {
+      "epoch": 0.4119372879801962,
+      "learning_rate": 0.00011761437608875034,
+      "loss": 0.8649,
+      "mean_token_accuracy": 0.736336624622345,
+      "num_tokens": 103828003.0,
+      "step": 44930
+    },
+    {
+      "epoch": 0.41202897221967544,
+      "learning_rate": 0.00011759603924085449,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.7434169828891755,
+      "num_tokens": 103850733.0,
+      "step": 44940
+    },
+    {
+      "epoch": 0.4121206564591547,
+      "learning_rate": 0.00011757770239295867,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.7394767940044403,
+      "num_tokens": 103873738.0,
+      "step": 44950
+    },
+    {
+      "epoch": 0.4122123406986339,
+      "learning_rate": 0.00011755936554506281,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.7482432544231414,
+      "num_tokens": 103897402.0,
+      "step": 44960
+    },
+    {
+      "epoch": 0.41230402493811313,
+      "learning_rate": 0.00011754102869716696,
+      "loss": 0.8512,
+      "mean_token_accuracy": 0.7492608785629272,
+      "num_tokens": 103919812.0,
+      "step": 44970
+    },
+    {
+      "epoch": 0.4123957091775924,
+      "learning_rate": 0.00011752269184927111,
+      "loss": 0.8331,
+      "mean_token_accuracy": 0.7446746230125427,
+      "num_tokens": 103943004.0,
+      "step": 44980
+    },
+    {
+      "epoch": 0.4124873934170716,
+      "learning_rate": 0.00011750435500137527,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.742868846654892,
+      "num_tokens": 103966092.0,
+      "step": 44990
+    },
+    {
+      "epoch": 0.41257907765655083,
+      "learning_rate": 0.00011748601815347942,
+      "loss": 0.8425,
+      "mean_token_accuracy": 0.7450195848941803,
+      "num_tokens": 103988479.0,
+      "step": 45000
+    },
+    {
+      "epoch": 0.4126707618960301,
+      "learning_rate": 0.00011746768130558357,
+      "loss": 0.8052,
+      "mean_token_accuracy": 0.7531399846076965,
+      "num_tokens": 104012140.0,
+      "step": 45010
+    },
+    {
+      "epoch": 0.41276244613550933,
+      "learning_rate": 0.00011744934445768774,
+      "loss": 0.8481,
+      "mean_token_accuracy": 0.7497302830219269,
+      "num_tokens": 104035348.0,
+      "step": 45020
+    },
+    {
+      "epoch": 0.4128541303749885,
+      "learning_rate": 0.00011743100760979189,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7362090647220612,
+      "num_tokens": 104058566.0,
+      "step": 45030
+    },
+    {
+      "epoch": 0.4129458146144678,
+      "learning_rate": 0.00011741267076189604,
+      "loss": 0.8406,
+      "mean_token_accuracy": 0.7523486912250519,
+      "num_tokens": 104081960.0,
+      "step": 45040
+    },
+    {
+      "epoch": 0.413037498853947,
+      "learning_rate": 0.00011739433391400019,
+      "loss": 0.8669,
+      "mean_token_accuracy": 0.7415791153907776,
+      "num_tokens": 104104438.0,
+      "step": 45050
+    },
+    {
+      "epoch": 0.4131291830934262,
+      "learning_rate": 0.00011737599706610434,
+      "loss": 0.8777,
+      "mean_token_accuracy": 0.7370117008686066,
+      "num_tokens": 104127169.0,
+      "step": 45060
+    },
+    {
+      "epoch": 0.4132208673329055,
+      "learning_rate": 0.00011735766021820848,
+      "loss": 0.857,
+      "mean_token_accuracy": 0.7467028617858886,
+      "num_tokens": 104149718.0,
+      "step": 45070
+    },
+    {
+      "epoch": 0.4133125515723847,
+      "learning_rate": 0.00011733932337031266,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7402410984039307,
+      "num_tokens": 104172687.0,
+      "step": 45080
+    },
+    {
+      "epoch": 0.4134042358118639,
+      "learning_rate": 0.00011732098652241681,
+      "loss": 0.868,
+      "mean_token_accuracy": 0.7320907115936279,
+      "num_tokens": 104195917.0,
+      "step": 45090
+    },
+    {
+      "epoch": 0.41349592005134317,
+      "learning_rate": 0.00011730264967452096,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7465797960758209,
+      "num_tokens": 104218361.0,
+      "step": 45100
+    },
+    {
+      "epoch": 0.4135876042908224,
+      "learning_rate": 0.00011728431282662512,
+      "loss": 0.8551,
+      "mean_token_accuracy": 0.74462109208107,
+      "num_tokens": 104241534.0,
+      "step": 45110
+    },
+    {
+      "epoch": 0.4136792885303016,
+      "learning_rate": 0.00011726597597872925,
+      "loss": 0.8565,
+      "mean_token_accuracy": 0.7477043271064758,
+      "num_tokens": 104263476.0,
+      "step": 45120
+    },
+    {
+      "epoch": 0.41377097276978086,
+      "learning_rate": 0.0001172476391308334,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7380394160747528,
+      "num_tokens": 104286942.0,
+      "step": 45130
+    },
+    {
+      "epoch": 0.4138626570092601,
+      "learning_rate": 0.00011722930228293756,
+      "loss": 0.8499,
+      "mean_token_accuracy": 0.7433103680610657,
+      "num_tokens": 104310207.0,
+      "step": 45140
+    },
+    {
+      "epoch": 0.41395434124873937,
+      "learning_rate": 0.00011721096543504174,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.748098623752594,
+      "num_tokens": 104333051.0,
+      "step": 45150
+    },
+    {
+      "epoch": 0.41404602548821856,
+      "learning_rate": 0.00011719262858714587,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7461417078971863,
+      "num_tokens": 104357050.0,
+      "step": 45160
+    },
+    {
+      "epoch": 0.4141377097276978,
+      "learning_rate": 0.00011717429173925003,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7391665399074554,
+      "num_tokens": 104379551.0,
+      "step": 45170
+    },
+    {
+      "epoch": 0.41422939396717706,
+      "learning_rate": 0.00011715595489135418,
+      "loss": 0.8391,
+      "mean_token_accuracy": 0.7471110463142395,
+      "num_tokens": 104404067.0,
+      "step": 45180
+    },
+    {
+      "epoch": 0.41432107820665626,
+      "learning_rate": 0.00011713761804345833,
+      "loss": 0.8575,
+      "mean_token_accuracy": 0.744124048948288,
+      "num_tokens": 104427315.0,
+      "step": 45190
+    },
+    {
+      "epoch": 0.4144127624461355,
+      "learning_rate": 0.00011711928119556248,
+      "loss": 0.8241,
+      "mean_token_accuracy": 0.7467672049999237,
+      "num_tokens": 104450719.0,
+      "step": 45200
+    },
+    {
+      "epoch": 0.41450444668561476,
+      "learning_rate": 0.00011710094434766665,
+      "loss": 0.8226,
+      "mean_token_accuracy": 0.7560986161231995,
+      "num_tokens": 104474191.0,
+      "step": 45210
+    },
+    {
+      "epoch": 0.41459613092509395,
+      "learning_rate": 0.0001170826074997708,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7443042755126953,
+      "num_tokens": 104497735.0,
+      "step": 45220
+    },
+    {
+      "epoch": 0.4146878151645732,
+      "learning_rate": 0.00011706427065187495,
+      "loss": 0.8714,
+      "mean_token_accuracy": 0.7403197705745697,
+      "num_tokens": 104520274.0,
+      "step": 45230
+    },
+    {
+      "epoch": 0.41477949940405245,
+      "learning_rate": 0.0001170459338039791,
+      "loss": 0.8774,
+      "mean_token_accuracy": 0.7455156743526459,
+      "num_tokens": 104543077.0,
+      "step": 45240
+    },
+    {
+      "epoch": 0.4148711836435317,
+      "learning_rate": 0.00011702759695608326,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7446125626564026,
+      "num_tokens": 104566045.0,
+      "step": 45250
+    },
+    {
+      "epoch": 0.4149628678830109,
+      "learning_rate": 0.00011700926010818741,
+      "loss": 0.8182,
+      "mean_token_accuracy": 0.7474395215511322,
+      "num_tokens": 104588489.0,
+      "step": 45260
+    },
+    {
+      "epoch": 0.41505455212249015,
+      "learning_rate": 0.00011699092326029155,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7484601020812989,
+      "num_tokens": 104611950.0,
+      "step": 45270
+    },
+    {
+      "epoch": 0.4151462363619694,
+      "learning_rate": 0.00011697258641239572,
+      "loss": 0.8277,
+      "mean_token_accuracy": 0.7526186227798461,
+      "num_tokens": 104635604.0,
+      "step": 45280
+    },
+    {
+      "epoch": 0.4152379206014486,
+      "learning_rate": 0.00011695424956449988,
+      "loss": 0.8319,
+      "mean_token_accuracy": 0.7496239125728608,
+      "num_tokens": 104658393.0,
+      "step": 45290
+    },
+    {
+      "epoch": 0.41532960484092785,
+      "learning_rate": 0.00011693591271660403,
+      "loss": 0.8414,
+      "mean_token_accuracy": 0.7488942444324493,
+      "num_tokens": 104681563.0,
+      "step": 45300
+    },
+    {
+      "epoch": 0.4154212890804071,
+      "learning_rate": 0.00011691757586870818,
+      "loss": 0.8186,
+      "mean_token_accuracy": 0.7454311132431031,
+      "num_tokens": 104705044.0,
+      "step": 45310
+    },
+    {
+      "epoch": 0.4155129733198863,
+      "learning_rate": 0.00011689923902081232,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7493055939674378,
+      "num_tokens": 104727656.0,
+      "step": 45320
+    },
+    {
+      "epoch": 0.41560465755936554,
+      "learning_rate": 0.00011688090217291647,
+      "loss": 0.8314,
+      "mean_token_accuracy": 0.74879270195961,
+      "num_tokens": 104751145.0,
+      "step": 45330
+    },
+    {
+      "epoch": 0.4156963417988448,
+      "learning_rate": 0.00011686256532502065,
+      "loss": 0.8336,
+      "mean_token_accuracy": 0.7552003026008606,
+      "num_tokens": 104774820.0,
+      "step": 45340
+    },
+    {
+      "epoch": 0.415788026038324,
+      "learning_rate": 0.0001168442284771248,
+      "loss": 0.8507,
+      "mean_token_accuracy": 0.748703908920288,
+      "num_tokens": 104797916.0,
+      "step": 45350
+    },
+    {
+      "epoch": 0.41587971027780324,
+      "learning_rate": 0.00011682589162922894,
+      "loss": 0.8563,
+      "mean_token_accuracy": 0.7438113152980804,
+      "num_tokens": 104821565.0,
+      "step": 45360
+    },
+    {
+      "epoch": 0.4159713945172825,
+      "learning_rate": 0.00011680755478133309,
+      "loss": 0.8423,
+      "mean_token_accuracy": 0.7446358323097229,
+      "num_tokens": 104845272.0,
+      "step": 45370
+    },
+    {
+      "epoch": 0.41606307875676174,
+      "learning_rate": 0.00011678921793343724,
+      "loss": 0.8141,
+      "mean_token_accuracy": 0.7505999803543091,
+      "num_tokens": 104868137.0,
+      "step": 45380
+    },
+    {
+      "epoch": 0.41615476299624093,
+      "learning_rate": 0.0001167708810855414,
+      "loss": 0.8547,
+      "mean_token_accuracy": 0.7448664903640747,
+      "num_tokens": 104892107.0,
+      "step": 45390
+    },
+    {
+      "epoch": 0.4162464472357202,
+      "learning_rate": 0.00011675254423764555,
+      "loss": 0.8255,
+      "mean_token_accuracy": 0.7546456217765808,
+      "num_tokens": 104915193.0,
+      "step": 45400
+    },
+    {
+      "epoch": 0.41633813147519944,
+      "learning_rate": 0.00011673420738974971,
+      "loss": 0.8628,
+      "mean_token_accuracy": 0.7427400827407837,
+      "num_tokens": 104937698.0,
+      "step": 45410
+    },
+    {
+      "epoch": 0.41642981571467863,
+      "learning_rate": 0.00011671587054185386,
+      "loss": 0.8295,
+      "mean_token_accuracy": 0.7500802636146545,
+      "num_tokens": 104960515.0,
+      "step": 45420
+    },
+    {
+      "epoch": 0.4165214999541579,
+      "learning_rate": 0.00011669753369395802,
+      "loss": 0.8087,
+      "mean_token_accuracy": 0.7531710028648376,
+      "num_tokens": 104984037.0,
+      "step": 45430
+    },
+    {
+      "epoch": 0.41661318419363713,
+      "learning_rate": 0.00011667919684606217,
+      "loss": 0.8506,
+      "mean_token_accuracy": 0.7413691818714142,
+      "num_tokens": 105007977.0,
+      "step": 45440
+    },
+    {
+      "epoch": 0.4167048684331163,
+      "learning_rate": 0.00011666085999816632,
+      "loss": 0.8368,
+      "mean_token_accuracy": 0.7456383883953095,
+      "num_tokens": 105031155.0,
+      "step": 45450
+    },
+    {
+      "epoch": 0.4167965526725956,
+      "learning_rate": 0.00011664252315027047,
+      "loss": 0.8203,
+      "mean_token_accuracy": 0.7559475481510163,
+      "num_tokens": 105054640.0,
+      "step": 45460
+    },
+    {
+      "epoch": 0.4168882369120748,
+      "learning_rate": 0.00011662418630237464,
+      "loss": 0.8705,
+      "mean_token_accuracy": 0.7432197690010071,
+      "num_tokens": 105077830.0,
+      "step": 45470
+    },
+    {
+      "epoch": 0.416979921151554,
+      "learning_rate": 0.00011660584945447879,
+      "loss": 0.8142,
+      "mean_token_accuracy": 0.7596755623817444,
+      "num_tokens": 105100700.0,
+      "step": 45480
+    },
+    {
+      "epoch": 0.4170716053910333,
+      "learning_rate": 0.00011658751260658294,
+      "loss": 0.812,
+      "mean_token_accuracy": 0.7577692210674286,
+      "num_tokens": 105123057.0,
+      "step": 45490
+    },
+    {
+      "epoch": 0.4171632896305125,
+      "learning_rate": 0.00011656917575868709,
+      "loss": 0.8562,
+      "mean_token_accuracy": 0.7477727413177491,
+      "num_tokens": 105146667.0,
+      "step": 45500
+    },
+    {
+      "epoch": 0.4172549738699918,
+      "learning_rate": 0.00011655083891079124,
+      "loss": 0.8796,
+      "mean_token_accuracy": 0.7398671507835388,
+      "num_tokens": 105169653.0,
+      "step": 45510
+    },
+    {
+      "epoch": 0.41734665810947097,
+      "learning_rate": 0.00011653250206289538,
+      "loss": 0.8182,
+      "mean_token_accuracy": 0.7536071181297302,
+      "num_tokens": 105193215.0,
+      "step": 45520
+    },
+    {
+      "epoch": 0.4174383423489502,
+      "learning_rate": 0.00011651416521499953,
+      "loss": 0.8194,
+      "mean_token_accuracy": 0.7570696473121643,
+      "num_tokens": 105216210.0,
+      "step": 45530
+    },
+    {
+      "epoch": 0.41753002658842947,
+      "learning_rate": 0.00011649582836710371,
+      "loss": 0.8548,
+      "mean_token_accuracy": 0.7465242505073547,
+      "num_tokens": 105239689.0,
+      "step": 45540
+    },
+    {
+      "epoch": 0.41762171082790867,
+      "learning_rate": 0.00011647749151920787,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.75292107462883,
+      "num_tokens": 105262978.0,
+      "step": 45550
+    },
+    {
+      "epoch": 0.4177133950673879,
+      "learning_rate": 0.000116459154671312,
+      "loss": 0.8722,
+      "mean_token_accuracy": 0.7375808358192444,
+      "num_tokens": 105287163.0,
+      "step": 45560
+    },
+    {
+      "epoch": 0.41780507930686717,
+      "learning_rate": 0.00011644081782341616,
+      "loss": 0.8894,
+      "mean_token_accuracy": 0.7363746345043183,
+      "num_tokens": 105310174.0,
+      "step": 45570
+    },
+    {
+      "epoch": 0.41789676354634636,
+      "learning_rate": 0.00011642248097552031,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.7463148236274719,
+      "num_tokens": 105333603.0,
+      "step": 45580
+    },
+    {
+      "epoch": 0.4179884477858256,
+      "learning_rate": 0.00011640414412762446,
+      "loss": 0.8357,
+      "mean_token_accuracy": 0.7453543424606324,
+      "num_tokens": 105356793.0,
+      "step": 45590
+    },
+    {
+      "epoch": 0.41808013202530486,
+      "learning_rate": 0.00011638580727972861,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.7480204463005066,
+      "num_tokens": 105379835.0,
+      "step": 45600
+    },
+    {
+      "epoch": 0.41817181626478406,
+      "learning_rate": 0.00011636747043183278,
+      "loss": 0.8423,
+      "mean_token_accuracy": 0.7426742017269135,
+      "num_tokens": 105403266.0,
+      "step": 45610
+    },
+    {
+      "epoch": 0.4182635005042633,
+      "learning_rate": 0.00011634913358393693,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.7503256618976593,
+      "num_tokens": 105425953.0,
+      "step": 45620
+    },
+    {
+      "epoch": 0.41835518474374256,
+      "learning_rate": 0.00011633079673604108,
+      "loss": 0.8778,
+      "mean_token_accuracy": 0.7438302993774414,
+      "num_tokens": 105447892.0,
+      "step": 45630
+    },
+    {
+      "epoch": 0.4184468689832218,
+      "learning_rate": 0.00011631245988814523,
+      "loss": 0.8788,
+      "mean_token_accuracy": 0.7388303697109222,
+      "num_tokens": 105470830.0,
+      "step": 45640
+    },
+    {
+      "epoch": 0.418538553222701,
+      "learning_rate": 0.00011629412304024938,
+      "loss": 0.8189,
+      "mean_token_accuracy": 0.750828355550766,
+      "num_tokens": 105493417.0,
+      "step": 45650
+    },
+    {
+      "epoch": 0.41863023746218025,
+      "learning_rate": 0.00011627578619235354,
+      "loss": 0.8262,
+      "mean_token_accuracy": 0.7559551656246185,
+      "num_tokens": 105516431.0,
+      "step": 45660
+    },
+    {
+      "epoch": 0.4187219217016595,
+      "learning_rate": 0.0001162574493444577,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.752439683675766,
+      "num_tokens": 105539949.0,
+      "step": 45670
+    },
+    {
+      "epoch": 0.4188136059411387,
+      "learning_rate": 0.00011623911249656185,
+      "loss": 0.8461,
+      "mean_token_accuracy": 0.7447866261005401,
+      "num_tokens": 105564062.0,
+      "step": 45680
+    },
+    {
+      "epoch": 0.41890529018061795,
+      "learning_rate": 0.000116220775648666,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7445887506008149,
+      "num_tokens": 105587012.0,
+      "step": 45690
+    },
+    {
+      "epoch": 0.4189969744200972,
+      "learning_rate": 0.00011620243880077016,
+      "loss": 0.8744,
+      "mean_token_accuracy": 0.7401760339736938,
+      "num_tokens": 105610455.0,
+      "step": 45700
+    },
+    {
+      "epoch": 0.4190886586595764,
+      "learning_rate": 0.00011618410195287431,
+      "loss": 0.8279,
+      "mean_token_accuracy": 0.7462787568569184,
+      "num_tokens": 105632633.0,
+      "step": 45710
+    },
+    {
+      "epoch": 0.41918034289905565,
+      "learning_rate": 0.00011616576510497845,
+      "loss": 0.8732,
+      "mean_token_accuracy": 0.7452632546424866,
+      "num_tokens": 105656151.0,
+      "step": 45720
+    },
+    {
+      "epoch": 0.4192720271385349,
+      "learning_rate": 0.0001161474282570826,
+      "loss": 0.8808,
+      "mean_token_accuracy": 0.7367396593093872,
+      "num_tokens": 105679462.0,
+      "step": 45730
+    },
+    {
+      "epoch": 0.4193637113780141,
+      "learning_rate": 0.00011612909140918678,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7499977886676789,
+      "num_tokens": 105702901.0,
+      "step": 45740
+    },
+    {
+      "epoch": 0.41945539561749334,
+      "learning_rate": 0.00011611075456129093,
+      "loss": 0.7976,
+      "mean_token_accuracy": 0.7562902629375458,
+      "num_tokens": 105726473.0,
+      "step": 45750
+    },
+    {
+      "epoch": 0.4195470798569726,
+      "learning_rate": 0.00011609241771339507,
+      "loss": 0.8624,
+      "mean_token_accuracy": 0.7390582978725433,
+      "num_tokens": 105749821.0,
+      "step": 45760
+    },
+    {
+      "epoch": 0.41963876409645184,
+      "learning_rate": 0.00011607408086549922,
+      "loss": 0.8321,
+      "mean_token_accuracy": 0.7496195077896118,
+      "num_tokens": 105773048.0,
+      "step": 45770
+    },
+    {
+      "epoch": 0.41973044833593104,
+      "learning_rate": 0.00011605574401760337,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7463629901409149,
+      "num_tokens": 105796028.0,
+      "step": 45780
+    },
+    {
+      "epoch": 0.4198221325754103,
+      "learning_rate": 0.00011603740716970752,
+      "loss": 0.8727,
+      "mean_token_accuracy": 0.7390254437923431,
+      "num_tokens": 105819424.0,
+      "step": 45790
+    },
+    {
+      "epoch": 0.41991381681488954,
+      "learning_rate": 0.00011601907032181169,
+      "loss": 0.8553,
+      "mean_token_accuracy": 0.7427088797092438,
+      "num_tokens": 105842506.0,
+      "step": 45800
+    },
+    {
+      "epoch": 0.42000550105436874,
+      "learning_rate": 0.00011600073347391584,
+      "loss": 0.8335,
+      "mean_token_accuracy": 0.7487434446811676,
+      "num_tokens": 105865914.0,
+      "step": 45810
+    },
+    {
+      "epoch": 0.420097185293848,
+      "learning_rate": 0.00011598239662601999,
+      "loss": 0.8925,
+      "mean_token_accuracy": 0.7376970946788788,
+      "num_tokens": 105889179.0,
+      "step": 45820
+    },
+    {
+      "epoch": 0.42018886953332724,
+      "learning_rate": 0.00011596405977812414,
+      "loss": 0.7837,
+      "mean_token_accuracy": 0.7546100616455078,
+      "num_tokens": 105912235.0,
+      "step": 45830
+    },
+    {
+      "epoch": 0.42028055377280643,
+      "learning_rate": 0.0001159457229302283,
+      "loss": 0.8801,
+      "mean_token_accuracy": 0.7406101167201996,
+      "num_tokens": 105935843.0,
+      "step": 45840
+    },
+    {
+      "epoch": 0.4203722380122857,
+      "learning_rate": 0.00011592738608233245,
+      "loss": 0.8263,
+      "mean_token_accuracy": 0.750576651096344,
+      "num_tokens": 105959465.0,
+      "step": 45850
+    },
+    {
+      "epoch": 0.42046392225176493,
+      "learning_rate": 0.0001159090492344366,
+      "loss": 0.8815,
+      "mean_token_accuracy": 0.7370054364204407,
+      "num_tokens": 105983307.0,
+      "step": 45860
+    },
+    {
+      "epoch": 0.4205556064912442,
+      "learning_rate": 0.00011589071238654077,
+      "loss": 0.8643,
+      "mean_token_accuracy": 0.7418716788291931,
+      "num_tokens": 106006639.0,
+      "step": 45870
+    },
+    {
+      "epoch": 0.4206472907307234,
+      "learning_rate": 0.00011587237553864492,
+      "loss": 0.8715,
+      "mean_token_accuracy": 0.7429073452949524,
+      "num_tokens": 106029369.0,
+      "step": 45880
+    },
+    {
+      "epoch": 0.42073897497020263,
+      "learning_rate": 0.00011585403869074907,
+      "loss": 0.8253,
+      "mean_token_accuracy": 0.7510428845882415,
+      "num_tokens": 106052040.0,
+      "step": 45890
+    },
+    {
+      "epoch": 0.4208306592096819,
+      "learning_rate": 0.00011583570184285322,
+      "loss": 0.8779,
+      "mean_token_accuracy": 0.7394703567028046,
+      "num_tokens": 106075073.0,
+      "step": 45900
+    },
+    {
+      "epoch": 0.4209223434491611,
+      "learning_rate": 0.00011581736499495737,
+      "loss": 0.8392,
+      "mean_token_accuracy": 0.7552196383476257,
+      "num_tokens": 106098575.0,
+      "step": 45910
+    },
+    {
+      "epoch": 0.4210140276886403,
+      "learning_rate": 0.00011579902814706151,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7435707151889801,
+      "num_tokens": 106121685.0,
+      "step": 45920
+    },
+    {
+      "epoch": 0.4211057119281196,
+      "learning_rate": 0.00011578069129916569,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7399836182594299,
+      "num_tokens": 106144497.0,
+      "step": 45930
+    },
+    {
+      "epoch": 0.42119739616759877,
+      "learning_rate": 0.00011576235445126984,
+      "loss": 0.8864,
+      "mean_token_accuracy": 0.7383939027786255,
+      "num_tokens": 106167585.0,
+      "step": 45940
+    },
+    {
+      "epoch": 0.421289080407078,
+      "learning_rate": 0.000115744017603374,
+      "loss": 0.8677,
+      "mean_token_accuracy": 0.7396584928035737,
+      "num_tokens": 106190120.0,
+      "step": 45950
+    },
+    {
+      "epoch": 0.42138076464655727,
+      "learning_rate": 0.00011572568075547813,
+      "loss": 0.8402,
+      "mean_token_accuracy": 0.7479067862033844,
+      "num_tokens": 106212767.0,
+      "step": 45960
+    },
+    {
+      "epoch": 0.42147244888603647,
+      "learning_rate": 0.00011570734390758228,
+      "loss": 0.8778,
+      "mean_token_accuracy": 0.7405851066112519,
+      "num_tokens": 106234721.0,
+      "step": 45970
+    },
+    {
+      "epoch": 0.4215641331255157,
+      "learning_rate": 0.00011568900705968644,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.7409386932849884,
+      "num_tokens": 106257402.0,
+      "step": 45980
+    },
+    {
+      "epoch": 0.42165581736499497,
+      "learning_rate": 0.00011567067021179059,
+      "loss": 0.8444,
+      "mean_token_accuracy": 0.7518729507923126,
+      "num_tokens": 106280272.0,
+      "step": 45990
+    },
+    {
+      "epoch": 0.4217475016044742,
+      "learning_rate": 0.00011565233336389475,
+      "loss": 0.8635,
+      "mean_token_accuracy": 0.7428579866886139,
+      "num_tokens": 106303374.0,
+      "step": 46000
+    },
+    {
+      "epoch": 0.4218391858439534,
+      "learning_rate": 0.0001156339965159989,
+      "loss": 0.8603,
+      "mean_token_accuracy": 0.7387055456638336,
+      "num_tokens": 106326608.0,
+      "step": 46010
+    },
+    {
+      "epoch": 0.42193087008343266,
+      "learning_rate": 0.00011561565966810306,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7429966807365418,
+      "num_tokens": 106350235.0,
+      "step": 46020
+    },
+    {
+      "epoch": 0.4220225543229119,
+      "learning_rate": 0.00011559732282020721,
+      "loss": 0.8423,
+      "mean_token_accuracy": 0.7473647713661193,
+      "num_tokens": 106373390.0,
+      "step": 46030
+    },
+    {
+      "epoch": 0.4221142385623911,
+      "learning_rate": 0.00011557898597231136,
+      "loss": 0.8421,
+      "mean_token_accuracy": 0.7498080492019653,
+      "num_tokens": 106397098.0,
+      "step": 46040
+    },
+    {
+      "epoch": 0.42220592280187036,
+      "learning_rate": 0.00011556064912441551,
+      "loss": 0.8412,
+      "mean_token_accuracy": 0.7453709840774536,
+      "num_tokens": 106420285.0,
+      "step": 46050
+    },
+    {
+      "epoch": 0.4222976070413496,
+      "learning_rate": 0.00011554231227651968,
+      "loss": 0.8463,
+      "mean_token_accuracy": 0.7433944940567017,
+      "num_tokens": 106442894.0,
+      "step": 46060
+    },
+    {
+      "epoch": 0.4223892912808288,
+      "learning_rate": 0.00011552397542862383,
+      "loss": 0.8824,
+      "mean_token_accuracy": 0.7417986690998077,
+      "num_tokens": 106465957.0,
+      "step": 46070
+    },
+    {
+      "epoch": 0.42248097552030806,
+      "learning_rate": 0.00011550563858072798,
+      "loss": 0.8672,
+      "mean_token_accuracy": 0.7395024836063385,
+      "num_tokens": 106488880.0,
+      "step": 46080
+    },
+    {
+      "epoch": 0.4225726597597873,
+      "learning_rate": 0.00011548730173283213,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.7445531606674194,
+      "num_tokens": 106512511.0,
+      "step": 46090
+    },
+    {
+      "epoch": 0.4226643439992665,
+      "learning_rate": 0.00011546896488493629,
+      "loss": 0.8571,
+      "mean_token_accuracy": 0.7428897380828857,
+      "num_tokens": 106536804.0,
+      "step": 46100
+    },
+    {
+      "epoch": 0.42275602823874575,
+      "learning_rate": 0.00011545062803704044,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7499698102474213,
+      "num_tokens": 106559217.0,
+      "step": 46110
+    },
+    {
+      "epoch": 0.422847712478225,
+      "learning_rate": 0.00011543229118914458,
+      "loss": 0.8661,
+      "mean_token_accuracy": 0.738968962430954,
+      "num_tokens": 106582047.0,
+      "step": 46120
+    },
+    {
+      "epoch": 0.42293939671770425,
+      "learning_rate": 0.00011541395434124876,
+      "loss": 0.8288,
+      "mean_token_accuracy": 0.7504612803459167,
+      "num_tokens": 106605527.0,
+      "step": 46130
+    },
+    {
+      "epoch": 0.42303108095718345,
+      "learning_rate": 0.00011539561749335291,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7508462846279145,
+      "num_tokens": 106628365.0,
+      "step": 46140
+    },
+    {
+      "epoch": 0.4231227651966627,
+      "learning_rate": 0.00011537728064545706,
+      "loss": 0.8522,
+      "mean_token_accuracy": 0.748188054561615,
+      "num_tokens": 106652326.0,
+      "step": 46150
+    },
+    {
+      "epoch": 0.42321444943614195,
+      "learning_rate": 0.0001153589437975612,
+      "loss": 0.8356,
+      "mean_token_accuracy": 0.7493544459342957,
+      "num_tokens": 106675436.0,
+      "step": 46160
+    },
+    {
+      "epoch": 0.42330613367562114,
+      "learning_rate": 0.00011534060694966535,
+      "loss": 0.8426,
+      "mean_token_accuracy": 0.7411046922206879,
+      "num_tokens": 106698878.0,
+      "step": 46170
+    },
+    {
+      "epoch": 0.4233978179151004,
+      "learning_rate": 0.0001153222701017695,
+      "loss": 0.8772,
+      "mean_token_accuracy": 0.7387505054473877,
+      "num_tokens": 106721574.0,
+      "step": 46180
+    },
+    {
+      "epoch": 0.42348950215457964,
+      "learning_rate": 0.00011530393325387368,
+      "loss": 0.8331,
+      "mean_token_accuracy": 0.7473115742206573,
+      "num_tokens": 106744381.0,
+      "step": 46190
+    },
+    {
+      "epoch": 0.42358118639405884,
+      "learning_rate": 0.00011528559640597782,
+      "loss": 0.8495,
+      "mean_token_accuracy": 0.7447142958641052,
+      "num_tokens": 106767371.0,
+      "step": 46200
+    },
+    {
+      "epoch": 0.4236728706335381,
+      "learning_rate": 0.00011526725955808197,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.750130033493042,
+      "num_tokens": 106790745.0,
+      "step": 46210
+    },
+    {
+      "epoch": 0.42376455487301734,
+      "learning_rate": 0.00011524892271018612,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7497813165187835,
+      "num_tokens": 106814397.0,
+      "step": 46220
+    },
+    {
+      "epoch": 0.42385623911249654,
+      "learning_rate": 0.00011523058586229027,
+      "loss": 0.8386,
+      "mean_token_accuracy": 0.7483627617359161,
+      "num_tokens": 106837580.0,
+      "step": 46230
+    },
+    {
+      "epoch": 0.4239479233519758,
+      "learning_rate": 0.00011521224901439443,
+      "loss": 0.8055,
+      "mean_token_accuracy": 0.753358781337738,
+      "num_tokens": 106860740.0,
+      "step": 46240
+    },
+    {
+      "epoch": 0.42403960759145504,
+      "learning_rate": 0.00011519391216649858,
+      "loss": 0.835,
+      "mean_token_accuracy": 0.751039469242096,
+      "num_tokens": 106883649.0,
+      "step": 46250
+    },
+    {
+      "epoch": 0.4241312918309343,
+      "learning_rate": 0.00011517557531860274,
+      "loss": 0.8451,
+      "mean_token_accuracy": 0.7448641002178192,
+      "num_tokens": 106906797.0,
+      "step": 46260
+    },
+    {
+      "epoch": 0.4242229760704135,
+      "learning_rate": 0.0001151572384707069,
+      "loss": 0.8388,
+      "mean_token_accuracy": 0.7540587186813354,
+      "num_tokens": 106929867.0,
+      "step": 46270
+    },
+    {
+      "epoch": 0.42431466030989273,
+      "learning_rate": 0.00011513890162281105,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7385248601436615,
+      "num_tokens": 106952653.0,
+      "step": 46280
+    },
+    {
+      "epoch": 0.424406344549372,
+      "learning_rate": 0.0001151205647749152,
+      "loss": 0.8113,
+      "mean_token_accuracy": 0.7544450163841248,
+      "num_tokens": 106975920.0,
+      "step": 46290
+    },
+    {
+      "epoch": 0.4244980287888512,
+      "learning_rate": 0.00011510222792701935,
+      "loss": 0.8078,
+      "mean_token_accuracy": 0.7519317746162415,
+      "num_tokens": 106998058.0,
+      "step": 46300
+    },
+    {
+      "epoch": 0.42458971302833043,
+      "learning_rate": 0.0001150838910791235,
+      "loss": 0.8371,
+      "mean_token_accuracy": 0.7469861268997192,
+      "num_tokens": 107021147.0,
+      "step": 46310
+    },
+    {
+      "epoch": 0.4246813972678097,
+      "learning_rate": 0.00011506555423122767,
+      "loss": 0.7939,
+      "mean_token_accuracy": 0.754490852355957,
+      "num_tokens": 107045031.0,
+      "step": 46320
+    },
+    {
+      "epoch": 0.4247730815072889,
+      "learning_rate": 0.00011504721738333182,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7441570162773132,
+      "num_tokens": 107068091.0,
+      "step": 46330
+    },
+    {
+      "epoch": 0.4248647657467681,
+      "learning_rate": 0.00011502888053543597,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7424165308475494,
+      "num_tokens": 107090966.0,
+      "step": 46340
+    },
+    {
+      "epoch": 0.4249564499862474,
+      "learning_rate": 0.00011501054368754012,
+      "loss": 0.8389,
+      "mean_token_accuracy": 0.7543610811233521,
+      "num_tokens": 107113144.0,
+      "step": 46350
+    },
+    {
+      "epoch": 0.42504813422572657,
+      "learning_rate": 0.00011499220683964426,
+      "loss": 0.8709,
+      "mean_token_accuracy": 0.7408024132251739,
+      "num_tokens": 107136129.0,
+      "step": 46360
+    },
+    {
+      "epoch": 0.4251398184652058,
+      "learning_rate": 0.00011497386999174841,
+      "loss": 0.8228,
+      "mean_token_accuracy": 0.7498598039150238,
+      "num_tokens": 107159147.0,
+      "step": 46370
+    },
+    {
+      "epoch": 0.42523150270468507,
+      "learning_rate": 0.00011495553314385257,
+      "loss": 0.8403,
+      "mean_token_accuracy": 0.7475099742412568,
+      "num_tokens": 107181963.0,
+      "step": 46380
+    },
+    {
+      "epoch": 0.4253231869441643,
+      "learning_rate": 0.00011493719629595674,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7424833416938782,
+      "num_tokens": 107204721.0,
+      "step": 46390
+    },
+    {
+      "epoch": 0.4254148711836435,
+      "learning_rate": 0.00011491885944806088,
+      "loss": 0.8519,
+      "mean_token_accuracy": 0.7513519704341889,
+      "num_tokens": 107228509.0,
+      "step": 46400
+    },
+    {
+      "epoch": 0.42550655542312277,
+      "learning_rate": 0.00011490052260016503,
+      "loss": 0.8497,
+      "mean_token_accuracy": 0.7518217444419861,
+      "num_tokens": 107251226.0,
+      "step": 46410
+    },
+    {
+      "epoch": 0.425598239662602,
+      "learning_rate": 0.00011488218575226919,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7468670606613159,
+      "num_tokens": 107273883.0,
+      "step": 46420
+    },
+    {
+      "epoch": 0.4256899239020812,
+      "learning_rate": 0.00011486384890437334,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7417753159999847,
+      "num_tokens": 107297263.0,
+      "step": 46430
+    },
+    {
+      "epoch": 0.42578160814156046,
+      "learning_rate": 0.00011484551205647749,
+      "loss": 0.8383,
+      "mean_token_accuracy": 0.7486030995845795,
+      "num_tokens": 107320290.0,
+      "step": 46440
+    },
+    {
+      "epoch": 0.4258732923810397,
+      "learning_rate": 0.00011482717520858166,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.7476086497306824,
+      "num_tokens": 107343053.0,
+      "step": 46450
+    },
+    {
+      "epoch": 0.4259649766205189,
+      "learning_rate": 0.00011480883836068581,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.75301753282547,
+      "num_tokens": 107365837.0,
+      "step": 46460
+    },
+    {
+      "epoch": 0.42605666085999816,
+      "learning_rate": 0.00011479050151278996,
+      "loss": 0.8494,
+      "mean_token_accuracy": 0.7412908494472503,
+      "num_tokens": 107388408.0,
+      "step": 46470
+    },
+    {
+      "epoch": 0.4261483450994774,
+      "learning_rate": 0.00011477216466489411,
+      "loss": 0.8784,
+      "mean_token_accuracy": 0.7370311617851257,
+      "num_tokens": 107411871.0,
+      "step": 46480
+    },
+    {
+      "epoch": 0.42624002933895666,
+      "learning_rate": 0.00011475382781699826,
+      "loss": 0.858,
+      "mean_token_accuracy": 0.7447152078151703,
+      "num_tokens": 107435014.0,
+      "step": 46490
+    },
+    {
+      "epoch": 0.42633171357843586,
+      "learning_rate": 0.00011473549096910241,
+      "loss": 0.8148,
+      "mean_token_accuracy": 0.7514746487140656,
+      "num_tokens": 107458939.0,
+      "step": 46500
+    },
+    {
+      "epoch": 0.4264233978179151,
+      "learning_rate": 0.00011471715412120657,
+      "loss": 0.865,
+      "mean_token_accuracy": 0.7420465648174286,
+      "num_tokens": 107482667.0,
+      "step": 46510
+    },
+    {
+      "epoch": 0.42651508205739436,
+      "learning_rate": 0.00011469881727331073,
+      "loss": 0.8504,
+      "mean_token_accuracy": 0.742823350429535,
+      "num_tokens": 107505392.0,
+      "step": 46520
+    },
+    {
+      "epoch": 0.42660676629687355,
+      "learning_rate": 0.00011468048042541488,
+      "loss": 0.8307,
+      "mean_token_accuracy": 0.7505095362663269,
+      "num_tokens": 107528397.0,
+      "step": 46530
+    },
+    {
+      "epoch": 0.4266984505363528,
+      "learning_rate": 0.00011466214357751904,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.7376496374607087,
+      "num_tokens": 107551304.0,
+      "step": 46540
+    },
+    {
+      "epoch": 0.42679013477583205,
+      "learning_rate": 0.00011464380672962319,
+      "loss": 0.844,
+      "mean_token_accuracy": 0.7526470005512238,
+      "num_tokens": 107575383.0,
+      "step": 46550
+    },
+    {
+      "epoch": 0.42688181901531125,
+      "learning_rate": 0.00011462546988172733,
+      "loss": 0.8477,
+      "mean_token_accuracy": 0.745954406261444,
+      "num_tokens": 107598795.0,
+      "step": 46560
+    },
+    {
+      "epoch": 0.4269735032547905,
+      "learning_rate": 0.00011460713303383148,
+      "loss": 0.8275,
+      "mean_token_accuracy": 0.752588552236557,
+      "num_tokens": 107622572.0,
+      "step": 46570
+    },
+    {
+      "epoch": 0.42706518749426975,
+      "learning_rate": 0.00011458879618593566,
+      "loss": 0.8334,
+      "mean_token_accuracy": 0.7562140345573425,
+      "num_tokens": 107645640.0,
+      "step": 46580
+    },
+    {
+      "epoch": 0.42715687173374894,
+      "learning_rate": 0.00011457045933803981,
+      "loss": 0.8255,
+      "mean_token_accuracy": 0.7416290700435638,
+      "num_tokens": 107667678.0,
+      "step": 46590
+    },
+    {
+      "epoch": 0.4272485559732282,
+      "learning_rate": 0.00011455212249014395,
+      "loss": 0.8682,
+      "mean_token_accuracy": 0.7378032803535461,
+      "num_tokens": 107690796.0,
+      "step": 46600
+    },
+    {
+      "epoch": 0.42734024021270745,
+      "learning_rate": 0.0001145337856422481,
+      "loss": 0.8114,
+      "mean_token_accuracy": 0.7644253432750702,
+      "num_tokens": 107713308.0,
+      "step": 46610
+    },
+    {
+      "epoch": 0.4274319244521867,
+      "learning_rate": 0.00011451544879435225,
+      "loss": 0.8479,
+      "mean_token_accuracy": 0.7421428322792053,
+      "num_tokens": 107735691.0,
+      "step": 46620
+    },
+    {
+      "epoch": 0.4275236086916659,
+      "learning_rate": 0.0001144971119464564,
+      "loss": 0.863,
+      "mean_token_accuracy": 0.7498755097389221,
+      "num_tokens": 107758652.0,
+      "step": 46630
+    },
+    {
+      "epoch": 0.42761529293114514,
+      "learning_rate": 0.00011447877509856055,
+      "loss": 0.8194,
+      "mean_token_accuracy": 0.7523223459720612,
+      "num_tokens": 107781493.0,
+      "step": 46640
+    },
+    {
+      "epoch": 0.4277069771706244,
+      "learning_rate": 0.00011446043825066472,
+      "loss": 0.8894,
+      "mean_token_accuracy": 0.7344420850276947,
+      "num_tokens": 107804722.0,
+      "step": 46650
+    },
+    {
+      "epoch": 0.4277986614101036,
+      "learning_rate": 0.00011444210140276887,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.745812463760376,
+      "num_tokens": 107827444.0,
+      "step": 46660
+    },
+    {
+      "epoch": 0.42789034564958284,
+      "learning_rate": 0.00011442376455487302,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7476184010505676,
+      "num_tokens": 107850960.0,
+      "step": 46670
+    },
+    {
+      "epoch": 0.4279820298890621,
+      "learning_rate": 0.00011440542770697718,
+      "loss": 0.8065,
+      "mean_token_accuracy": 0.7576253831386566,
+      "num_tokens": 107874184.0,
+      "step": 46680
+    },
+    {
+      "epoch": 0.4280737141285413,
+      "learning_rate": 0.00011438709085908133,
+      "loss": 0.8677,
+      "mean_token_accuracy": 0.7421083569526672,
+      "num_tokens": 107897879.0,
+      "step": 46690
+    },
+    {
+      "epoch": 0.42816539836802053,
+      "learning_rate": 0.00011436875401118548,
+      "loss": 0.8013,
+      "mean_token_accuracy": 0.756723302602768,
+      "num_tokens": 107920538.0,
+      "step": 46700
+    },
+    {
+      "epoch": 0.4282570826074998,
+      "learning_rate": 0.00011435041716328964,
+      "loss": 0.8377,
+      "mean_token_accuracy": 0.7451123595237732,
+      "num_tokens": 107943645.0,
+      "step": 46710
+    },
+    {
+      "epoch": 0.428348766846979,
+      "learning_rate": 0.0001143320803153938,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7516084432601928,
+      "num_tokens": 107966827.0,
+      "step": 46720
+    },
+    {
+      "epoch": 0.42844045108645823,
+      "learning_rate": 0.00011431374346749795,
+      "loss": 0.8058,
+      "mean_token_accuracy": 0.7576744019985199,
+      "num_tokens": 107990085.0,
+      "step": 46730
+    },
+    {
+      "epoch": 0.4285321353259375,
+      "learning_rate": 0.0001142954066196021,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.7461348176002502,
+      "num_tokens": 108013237.0,
+      "step": 46740
+    },
+    {
+      "epoch": 0.42862381956541673,
+      "learning_rate": 0.00011427706977170625,
+      "loss": 0.8565,
+      "mean_token_accuracy": 0.7457659125328064,
+      "num_tokens": 108035642.0,
+      "step": 46750
+    },
+    {
+      "epoch": 0.4287155038048959,
+      "learning_rate": 0.00011425873292381039,
+      "loss": 0.8009,
+      "mean_token_accuracy": 0.7604243099689484,
+      "num_tokens": 108058652.0,
+      "step": 46760
+    },
+    {
+      "epoch": 0.4288071880443752,
+      "learning_rate": 0.00011424039607591454,
+      "loss": 0.8551,
+      "mean_token_accuracy": 0.7479537963867188,
+      "num_tokens": 108082719.0,
+      "step": 46770
+    },
+    {
+      "epoch": 0.4288988722838544,
+      "learning_rate": 0.00011422205922801872,
+      "loss": 0.8723,
+      "mean_token_accuracy": 0.7449828326702118,
+      "num_tokens": 108105907.0,
+      "step": 46780
+    },
+    {
+      "epoch": 0.4289905565233336,
+      "learning_rate": 0.00011420372238012287,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.747146052122116,
+      "num_tokens": 108128613.0,
+      "step": 46790
+    },
+    {
+      "epoch": 0.42908224076281287,
+      "learning_rate": 0.00011418538553222701,
+      "loss": 0.8723,
+      "mean_token_accuracy": 0.7421761274337768,
+      "num_tokens": 108151929.0,
+      "step": 46800
+    },
+    {
+      "epoch": 0.4291739250022921,
+      "learning_rate": 0.00011416704868433116,
+      "loss": 0.8338,
+      "mean_token_accuracy": 0.7516462743282318,
+      "num_tokens": 108174638.0,
+      "step": 46810
+    },
+    {
+      "epoch": 0.4292656092417713,
+      "learning_rate": 0.00011414871183643532,
+      "loss": 0.8717,
+      "mean_token_accuracy": 0.7379578471183776,
+      "num_tokens": 108198260.0,
+      "step": 46820
+    },
+    {
+      "epoch": 0.42935729348125057,
+      "learning_rate": 0.00011413037498853947,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.745039564371109,
+      "num_tokens": 108221876.0,
+      "step": 46830
+    },
+    {
+      "epoch": 0.4294489777207298,
+      "learning_rate": 0.00011411203814064362,
+      "loss": 0.8577,
+      "mean_token_accuracy": 0.7417080342769623,
+      "num_tokens": 108245244.0,
+      "step": 46840
+    },
+    {
+      "epoch": 0.429540661960209,
+      "learning_rate": 0.00011409370129274778,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7539389371871948,
+      "num_tokens": 108268712.0,
+      "step": 46850
+    },
+    {
+      "epoch": 0.42963234619968826,
+      "learning_rate": 0.00011407536444485194,
+      "loss": 0.8444,
+      "mean_token_accuracy": 0.7451859951019287,
+      "num_tokens": 108291201.0,
+      "step": 46860
+    },
+    {
+      "epoch": 0.4297240304391675,
+      "learning_rate": 0.00011405702759695609,
+      "loss": 0.8364,
+      "mean_token_accuracy": 0.7501066982746124,
+      "num_tokens": 108313510.0,
+      "step": 46870
+    },
+    {
+      "epoch": 0.42981571467864677,
+      "learning_rate": 0.00011403869074906024,
+      "loss": 0.8919,
+      "mean_token_accuracy": 0.736325454711914,
+      "num_tokens": 108336439.0,
+      "step": 46880
+    },
+    {
+      "epoch": 0.42990739891812596,
+      "learning_rate": 0.00011402035390116439,
+      "loss": 0.8355,
+      "mean_token_accuracy": 0.7490274965763092,
+      "num_tokens": 108359350.0,
+      "step": 46890
+    },
+    {
+      "epoch": 0.4299990831576052,
+      "learning_rate": 0.00011400201705326854,
+      "loss": 0.8167,
+      "mean_token_accuracy": 0.747568941116333,
+      "num_tokens": 108382543.0,
+      "step": 46900
+    },
+    {
+      "epoch": 0.43009076739708446,
+      "learning_rate": 0.00011398368020537271,
+      "loss": 0.8634,
+      "mean_token_accuracy": 0.7424772560596467,
+      "num_tokens": 108406241.0,
+      "step": 46910
+    },
+    {
+      "epoch": 0.43018245163656366,
+      "learning_rate": 0.00011396534335747686,
+      "loss": 0.8325,
+      "mean_token_accuracy": 0.7476962924003601,
+      "num_tokens": 108429123.0,
+      "step": 46920
+    },
+    {
+      "epoch": 0.4302741358760429,
+      "learning_rate": 0.00011394700650958101,
+      "loss": 0.8509,
+      "mean_token_accuracy": 0.7468082726001739,
+      "num_tokens": 108452949.0,
+      "step": 46930
+    },
+    {
+      "epoch": 0.43036582011552216,
+      "learning_rate": 0.00011392866966168516,
+      "loss": 0.8464,
+      "mean_token_accuracy": 0.7468827307224274,
+      "num_tokens": 108475489.0,
+      "step": 46940
+    },
+    {
+      "epoch": 0.43045750435500135,
+      "learning_rate": 0.00011391033281378932,
+      "loss": 0.844,
+      "mean_token_accuracy": 0.7473788142204285,
+      "num_tokens": 108498587.0,
+      "step": 46950
+    },
+    {
+      "epoch": 0.4305491885944806,
+      "learning_rate": 0.00011389199596589345,
+      "loss": 0.8111,
+      "mean_token_accuracy": 0.7529711425304413,
+      "num_tokens": 108522167.0,
+      "step": 46960
+    },
+    {
+      "epoch": 0.43064087283395985,
+      "learning_rate": 0.00011387365911799761,
+      "loss": 0.8235,
+      "mean_token_accuracy": 0.7489188611507416,
+      "num_tokens": 108545631.0,
+      "step": 46970
+    },
+    {
+      "epoch": 0.43073255707343905,
+      "learning_rate": 0.00011385532227010179,
+      "loss": 0.8478,
+      "mean_token_accuracy": 0.7511611342430115,
+      "num_tokens": 108568542.0,
+      "step": 46980
+    },
+    {
+      "epoch": 0.4308242413129183,
+      "learning_rate": 0.00011383698542220594,
+      "loss": 0.8559,
+      "mean_token_accuracy": 0.7441620111465455,
+      "num_tokens": 108591717.0,
+      "step": 46990
+    },
+    {
+      "epoch": 0.43091592555239755,
+      "learning_rate": 0.00011381864857431009,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.745752078294754,
+      "num_tokens": 108615225.0,
+      "step": 47000
+    },
+    {
+      "epoch": 0.4310076097918768,
+      "learning_rate": 0.00011380031172641423,
+      "loss": 0.8852,
+      "mean_token_accuracy": 0.7369329392910003,
+      "num_tokens": 108637645.0,
+      "step": 47010
+    },
+    {
+      "epoch": 0.431099294031356,
+      "learning_rate": 0.00011378197487851838,
+      "loss": 0.8178,
+      "mean_token_accuracy": 0.7573180019855499,
+      "num_tokens": 108661040.0,
+      "step": 47020
+    },
+    {
+      "epoch": 0.43119097827083525,
+      "learning_rate": 0.00011376363803062253,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7456544458866119,
+      "num_tokens": 108684243.0,
+      "step": 47030
+    },
+    {
+      "epoch": 0.4312826625103145,
+      "learning_rate": 0.00011374530118272671,
+      "loss": 0.8712,
+      "mean_token_accuracy": 0.7465338408946991,
+      "num_tokens": 108707164.0,
+      "step": 47040
+    },
+    {
+      "epoch": 0.4313743467497937,
+      "learning_rate": 0.00011372696433483085,
+      "loss": 0.8297,
+      "mean_token_accuracy": 0.7447070181369781,
+      "num_tokens": 108729998.0,
+      "step": 47050
+    },
+    {
+      "epoch": 0.43146603098927294,
+      "learning_rate": 0.000113708627486935,
+      "loss": 0.8165,
+      "mean_token_accuracy": 0.7562990963459015,
+      "num_tokens": 108753120.0,
+      "step": 47060
+    },
+    {
+      "epoch": 0.4315577152287522,
+      "learning_rate": 0.00011369029063903915,
+      "loss": 0.8185,
+      "mean_token_accuracy": 0.7538254737854004,
+      "num_tokens": 108776661.0,
+      "step": 47070
+    },
+    {
+      "epoch": 0.4316493994682314,
+      "learning_rate": 0.0001136719537911433,
+      "loss": 0.8541,
+      "mean_token_accuracy": 0.7439543306827545,
+      "num_tokens": 108799474.0,
+      "step": 47080
+    },
+    {
+      "epoch": 0.43174108370771064,
+      "learning_rate": 0.00011365361694324746,
+      "loss": 0.8215,
+      "mean_token_accuracy": 0.7555378258228302,
+      "num_tokens": 108822375.0,
+      "step": 47090
+    },
+    {
+      "epoch": 0.4318327679471899,
+      "learning_rate": 0.00011363528009535161,
+      "loss": 0.8082,
+      "mean_token_accuracy": 0.749610298871994,
+      "num_tokens": 108845344.0,
+      "step": 47100
+    },
+    {
+      "epoch": 0.43192445218666914,
+      "learning_rate": 0.00011361694324745577,
+      "loss": 0.8228,
+      "mean_token_accuracy": 0.7503890573978425,
+      "num_tokens": 108868464.0,
+      "step": 47110
+    },
+    {
+      "epoch": 0.43201613642614833,
+      "learning_rate": 0.00011359860639955993,
+      "loss": 0.8153,
+      "mean_token_accuracy": 0.7527883887290955,
+      "num_tokens": 108892714.0,
+      "step": 47120
+    },
+    {
+      "epoch": 0.4321078206656276,
+      "learning_rate": 0.00011358026955166408,
+      "loss": 0.8289,
+      "mean_token_accuracy": 0.7499531209468842,
+      "num_tokens": 108915592.0,
+      "step": 47130
+    },
+    {
+      "epoch": 0.43219950490510683,
+      "learning_rate": 0.00011356193270376823,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.744117820262909,
+      "num_tokens": 108939067.0,
+      "step": 47140
+    },
+    {
+      "epoch": 0.43229118914458603,
+      "learning_rate": 0.00011354359585587238,
+      "loss": 0.8515,
+      "mean_token_accuracy": 0.7455109179019928,
+      "num_tokens": 108962684.0,
+      "step": 47150
+    },
+    {
+      "epoch": 0.4323828733840653,
+      "learning_rate": 0.00011352525900797652,
+      "loss": 0.809,
+      "mean_token_accuracy": 0.7546250522136688,
+      "num_tokens": 108985648.0,
+      "step": 47160
+    },
+    {
+      "epoch": 0.43247455762354453,
+      "learning_rate": 0.0001135069221600807,
+      "loss": 0.8522,
+      "mean_token_accuracy": 0.7464926838874817,
+      "num_tokens": 109009257.0,
+      "step": 47170
+    },
+    {
+      "epoch": 0.4325662418630237,
+      "learning_rate": 0.00011348858531218485,
+      "loss": 0.8255,
+      "mean_token_accuracy": 0.7509628474712372,
+      "num_tokens": 109032839.0,
+      "step": 47180
+    },
+    {
+      "epoch": 0.432657926102503,
+      "learning_rate": 0.000113470248464289,
+      "loss": 0.8287,
+      "mean_token_accuracy": 0.7467759847640991,
+      "num_tokens": 109056222.0,
+      "step": 47190
+    },
+    {
+      "epoch": 0.4327496103419822,
+      "learning_rate": 0.00011345191161639315,
+      "loss": 0.8233,
+      "mean_token_accuracy": 0.7523932933807373,
+      "num_tokens": 109079559.0,
+      "step": 47200
+    },
+    {
+      "epoch": 0.4328412945814614,
+      "learning_rate": 0.00011343357476849729,
+      "loss": 0.8488,
+      "mean_token_accuracy": 0.751216036081314,
+      "num_tokens": 109103218.0,
+      "step": 47210
+    },
+    {
+      "epoch": 0.4329329788209407,
+      "learning_rate": 0.00011341523792060144,
+      "loss": 0.8852,
+      "mean_token_accuracy": 0.7350303649902343,
+      "num_tokens": 109126005.0,
+      "step": 47220
+    },
+    {
+      "epoch": 0.4330246630604199,
+      "learning_rate": 0.0001133969010727056,
+      "loss": 0.8303,
+      "mean_token_accuracy": 0.7465072870254517,
+      "num_tokens": 109148997.0,
+      "step": 47230
+    },
+    {
+      "epoch": 0.4331163472998992,
+      "learning_rate": 0.00011337856422480978,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7420282483100891,
+      "num_tokens": 109172288.0,
+      "step": 47240
+    },
+    {
+      "epoch": 0.43320803153937837,
+      "learning_rate": 0.00011336022737691391,
+      "loss": 0.8146,
+      "mean_token_accuracy": 0.7545742690563202,
+      "num_tokens": 109196546.0,
+      "step": 47250
+    },
+    {
+      "epoch": 0.4332997157788576,
+      "learning_rate": 0.00011334189052901807,
+      "loss": 0.8921,
+      "mean_token_accuracy": 0.7386726856231689,
+      "num_tokens": 109219914.0,
+      "step": 47260
+    },
+    {
+      "epoch": 0.43339140001833687,
+      "learning_rate": 0.00011332355368112222,
+      "loss": 0.8304,
+      "mean_token_accuracy": 0.7491811335086822,
+      "num_tokens": 109243472.0,
+      "step": 47270
+    },
+    {
+      "epoch": 0.43348308425781606,
+      "learning_rate": 0.00011330521683322637,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7429771661758423,
+      "num_tokens": 109266778.0,
+      "step": 47280
+    },
+    {
+      "epoch": 0.4335747684972953,
+      "learning_rate": 0.00011328687998533052,
+      "loss": 0.8425,
+      "mean_token_accuracy": 0.7458553075790405,
+      "num_tokens": 109290212.0,
+      "step": 47290
+    },
+    {
+      "epoch": 0.43366645273677457,
+      "learning_rate": 0.00011326854313743469,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7425900459289551,
+      "num_tokens": 109313187.0,
+      "step": 47300
+    },
+    {
+      "epoch": 0.43375813697625376,
+      "learning_rate": 0.00011325020628953884,
+      "loss": 0.8091,
+      "mean_token_accuracy": 0.756443053483963,
+      "num_tokens": 109336723.0,
+      "step": 47310
+    },
+    {
+      "epoch": 0.433849821215733,
+      "learning_rate": 0.00011323186944164299,
+      "loss": 0.8696,
+      "mean_token_accuracy": 0.7415278375148773,
+      "num_tokens": 109360238.0,
+      "step": 47320
+    },
+    {
+      "epoch": 0.43394150545521226,
+      "learning_rate": 0.00011321353259374714,
+      "loss": 0.8426,
+      "mean_token_accuracy": 0.7460782110691071,
+      "num_tokens": 109383338.0,
+      "step": 47330
+    },
+    {
+      "epoch": 0.43403318969469146,
+      "learning_rate": 0.0001131951957458513,
+      "loss": 0.8582,
+      "mean_token_accuracy": 0.7424134850502014,
+      "num_tokens": 109406490.0,
+      "step": 47340
+    },
+    {
+      "epoch": 0.4341248739341707,
+      "learning_rate": 0.00011317685889795545,
+      "loss": 0.8301,
+      "mean_token_accuracy": 0.7553377032279969,
+      "num_tokens": 109429540.0,
+      "step": 47350
+    },
+    {
+      "epoch": 0.43421655817364996,
+      "learning_rate": 0.00011315852205005958,
+      "loss": 0.8116,
+      "mean_token_accuracy": 0.7546046912670136,
+      "num_tokens": 109452428.0,
+      "step": 47360
+    },
+    {
+      "epoch": 0.4343082424131292,
+      "learning_rate": 0.00011314018520216376,
+      "loss": 0.8363,
+      "mean_token_accuracy": 0.7467213153839112,
+      "num_tokens": 109475244.0,
+      "step": 47370
+    },
+    {
+      "epoch": 0.4343999266526084,
+      "learning_rate": 0.00011312184835426791,
+      "loss": 0.8499,
+      "mean_token_accuracy": 0.7427000403404236,
+      "num_tokens": 109498832.0,
+      "step": 47380
+    },
+    {
+      "epoch": 0.43449161089208765,
+      "learning_rate": 0.00011310351150637207,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7458879351615906,
+      "num_tokens": 109521803.0,
+      "step": 47390
+    },
+    {
+      "epoch": 0.4345832951315669,
+      "learning_rate": 0.00011308517465847622,
+      "loss": 0.8439,
+      "mean_token_accuracy": 0.7500184535980224,
+      "num_tokens": 109544973.0,
+      "step": 47400
+    },
+    {
+      "epoch": 0.4346749793710461,
+      "learning_rate": 0.00011306683781058036,
+      "loss": 0.8311,
+      "mean_token_accuracy": 0.7486188590526581,
+      "num_tokens": 109567935.0,
+      "step": 47410
+    },
+    {
+      "epoch": 0.43476666361052535,
+      "learning_rate": 0.00011304850096268451,
+      "loss": 0.8035,
+      "mean_token_accuracy": 0.7516610980033874,
+      "num_tokens": 109591707.0,
+      "step": 47420
+    },
+    {
+      "epoch": 0.4348583478500046,
+      "learning_rate": 0.00011303016411478869,
+      "loss": 0.8036,
+      "mean_token_accuracy": 0.7554397463798523,
+      "num_tokens": 109614822.0,
+      "step": 47430
+    },
+    {
+      "epoch": 0.4349500320894838,
+      "learning_rate": 0.00011301182726689284,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.7391930520534515,
+      "num_tokens": 109637931.0,
+      "step": 47440
+    },
+    {
+      "epoch": 0.43504171632896305,
+      "learning_rate": 0.00011299349041899698,
+      "loss": 0.8772,
+      "mean_token_accuracy": 0.7359247624874115,
+      "num_tokens": 109661019.0,
+      "step": 47450
+    },
+    {
+      "epoch": 0.4351334005684423,
+      "learning_rate": 0.00011297515357110113,
+      "loss": 0.8285,
+      "mean_token_accuracy": 0.7506414473056793,
+      "num_tokens": 109683416.0,
+      "step": 47460
+    },
+    {
+      "epoch": 0.4352250848079215,
+      "learning_rate": 0.00011295681672320528,
+      "loss": 0.8491,
+      "mean_token_accuracy": 0.7429089784622193,
+      "num_tokens": 109706303.0,
+      "step": 47470
+    },
+    {
+      "epoch": 0.43531676904740074,
+      "learning_rate": 0.00011293847987530943,
+      "loss": 0.8288,
+      "mean_token_accuracy": 0.7507230579853058,
+      "num_tokens": 109729845.0,
+      "step": 47480
+    },
+    {
+      "epoch": 0.43540845328688,
+      "learning_rate": 0.00011292014302741359,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.7348084867000579,
+      "num_tokens": 109753109.0,
+      "step": 47490
+    },
+    {
+      "epoch": 0.43550013752635924,
+      "learning_rate": 0.00011290180617951775,
+      "loss": 0.8278,
+      "mean_token_accuracy": 0.7435014486312866,
+      "num_tokens": 109775944.0,
+      "step": 47500
+    },
+    {
+      "epoch": 0.43559182176583844,
+      "learning_rate": 0.0001128834693316219,
+      "loss": 0.8863,
+      "mean_token_accuracy": 0.7433479726314545,
+      "num_tokens": 109799022.0,
+      "step": 47510
+    },
+    {
+      "epoch": 0.4356835060053177,
+      "learning_rate": 0.00011286513248372605,
+      "loss": 0.8715,
+      "mean_token_accuracy": 0.7416722118854523,
+      "num_tokens": 109822320.0,
+      "step": 47520
+    },
+    {
+      "epoch": 0.43577519024479694,
+      "learning_rate": 0.0001128467956358302,
+      "loss": 0.8673,
+      "mean_token_accuracy": 0.7449494242668152,
+      "num_tokens": 109845031.0,
+      "step": 47530
+    },
+    {
+      "epoch": 0.43586687448427613,
+      "learning_rate": 0.00011282845878793436,
+      "loss": 0.835,
+      "mean_token_accuracy": 0.7494809627532959,
+      "num_tokens": 109869083.0,
+      "step": 47540
+    },
+    {
+      "epoch": 0.4359585587237554,
+      "learning_rate": 0.00011281012194003851,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.7433875679969788,
+      "num_tokens": 109892372.0,
+      "step": 47550
+    },
+    {
+      "epoch": 0.43605024296323464,
+      "learning_rate": 0.00011279178509214268,
+      "loss": 0.8527,
+      "mean_token_accuracy": 0.7452048778533935,
+      "num_tokens": 109914468.0,
+      "step": 47560
+    },
+    {
+      "epoch": 0.43614192720271383,
+      "learning_rate": 0.00011277344824424683,
+      "loss": 0.8421,
+      "mean_token_accuracy": 0.7487707376480103,
+      "num_tokens": 109937209.0,
+      "step": 47570
+    },
+    {
+      "epoch": 0.4362336114421931,
+      "learning_rate": 0.00011275511139635098,
+      "loss": 0.8534,
+      "mean_token_accuracy": 0.7431705534458161,
+      "num_tokens": 109960483.0,
+      "step": 47580
+    },
+    {
+      "epoch": 0.43632529568167233,
+      "learning_rate": 0.00011273677454845513,
+      "loss": 0.8795,
+      "mean_token_accuracy": 0.7445425689220428,
+      "num_tokens": 109984301.0,
+      "step": 47590
+    },
+    {
+      "epoch": 0.4364169799211515,
+      "learning_rate": 0.00011271843770055928,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7453936636447906,
+      "num_tokens": 110007046.0,
+      "step": 47600
+    },
+    {
+      "epoch": 0.4365086641606308,
+      "learning_rate": 0.00011270010085266342,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7403316795825958,
+      "num_tokens": 110030704.0,
+      "step": 47610
+    },
+    {
+      "epoch": 0.43660034840011,
+      "learning_rate": 0.00011268176400476757,
+      "loss": 0.8027,
+      "mean_token_accuracy": 0.7546936690807342,
+      "num_tokens": 110053814.0,
+      "step": 47620
+    },
+    {
+      "epoch": 0.4366920326395893,
+      "learning_rate": 0.00011266342715687175,
+      "loss": 0.84,
+      "mean_token_accuracy": 0.7457985043525696,
+      "num_tokens": 110076518.0,
+      "step": 47630
+    },
+    {
+      "epoch": 0.4367837168790685,
+      "learning_rate": 0.0001126450903089759,
+      "loss": 0.8349,
+      "mean_token_accuracy": 0.7517504334449768,
+      "num_tokens": 110099414.0,
+      "step": 47640
+    },
+    {
+      "epoch": 0.4368754011185477,
+      "learning_rate": 0.00011262675346108004,
+      "loss": 0.8605,
+      "mean_token_accuracy": 0.7411771178245544,
+      "num_tokens": 110122327.0,
+      "step": 47650
+    },
+    {
+      "epoch": 0.436967085358027,
+      "learning_rate": 0.0001126084166131842,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7514030754566192,
+      "num_tokens": 110144843.0,
+      "step": 47660
+    },
+    {
+      "epoch": 0.43705876959750617,
+      "learning_rate": 0.00011259007976528835,
+      "loss": 0.7993,
+      "mean_token_accuracy": 0.7621033012866973,
+      "num_tokens": 110167395.0,
+      "step": 47670
+    },
+    {
+      "epoch": 0.4371504538369854,
+      "learning_rate": 0.0001125717429173925,
+      "loss": 0.8609,
+      "mean_token_accuracy": 0.743192583322525,
+      "num_tokens": 110190771.0,
+      "step": 47680
+    },
+    {
+      "epoch": 0.43724213807646467,
+      "learning_rate": 0.00011255340606949666,
+      "loss": 0.8391,
+      "mean_token_accuracy": 0.7415359079837799,
+      "num_tokens": 110213809.0,
+      "step": 47690
+    },
+    {
+      "epoch": 0.43733382231594387,
+      "learning_rate": 0.00011253506922160082,
+      "loss": 0.8687,
+      "mean_token_accuracy": 0.7418878674507141,
+      "num_tokens": 110237147.0,
+      "step": 47700
+    },
+    {
+      "epoch": 0.4374255065554231,
+      "learning_rate": 0.00011251673237370497,
+      "loss": 0.8464,
+      "mean_token_accuracy": 0.744903302192688,
+      "num_tokens": 110260620.0,
+      "step": 47710
+    },
+    {
+      "epoch": 0.43751719079490237,
+      "learning_rate": 0.00011249839552580912,
+      "loss": 0.8511,
+      "mean_token_accuracy": 0.7429839372634888,
+      "num_tokens": 110283629.0,
+      "step": 47720
+    },
+    {
+      "epoch": 0.4376088750343816,
+      "learning_rate": 0.00011248005867791327,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.7538785874843598,
+      "num_tokens": 110307301.0,
+      "step": 47730
+    },
+    {
+      "epoch": 0.4377005592738608,
+      "learning_rate": 0.00011246172183001742,
+      "loss": 0.8326,
+      "mean_token_accuracy": 0.7453486859798432,
+      "num_tokens": 110331156.0,
+      "step": 47740
+    },
+    {
+      "epoch": 0.43779224351334006,
+      "learning_rate": 0.00011244338498212157,
+      "loss": 0.8229,
+      "mean_token_accuracy": 0.7505684137344361,
+      "num_tokens": 110353859.0,
+      "step": 47750
+    },
+    {
+      "epoch": 0.4378839277528193,
+      "learning_rate": 0.00011242504813422574,
+      "loss": 0.8459,
+      "mean_token_accuracy": 0.7491546094417572,
+      "num_tokens": 110376640.0,
+      "step": 47760
+    },
+    {
+      "epoch": 0.4379756119922985,
+      "learning_rate": 0.00011240671128632989,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.737398773431778,
+      "num_tokens": 110398873.0,
+      "step": 47770
+    },
+    {
+      "epoch": 0.43806729623177776,
+      "learning_rate": 0.00011238837443843404,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7408490121364594,
+      "num_tokens": 110421782.0,
+      "step": 47780
+    },
+    {
+      "epoch": 0.438158980471257,
+      "learning_rate": 0.0001123700375905382,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7413858413696289,
+      "num_tokens": 110444264.0,
+      "step": 47790
+    },
+    {
+      "epoch": 0.4382506647107362,
+      "learning_rate": 0.00011235170074264235,
+      "loss": 0.8096,
+      "mean_token_accuracy": 0.7543434023857116,
+      "num_tokens": 110468139.0,
+      "step": 47800
+    },
+    {
+      "epoch": 0.43834234895021545,
+      "learning_rate": 0.00011233336389474649,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7495468199253082,
+      "num_tokens": 110491294.0,
+      "step": 47810
+    },
+    {
+      "epoch": 0.4384340331896947,
+      "learning_rate": 0.00011231502704685066,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7498344779014587,
+      "num_tokens": 110514341.0,
+      "step": 47820
+    },
+    {
+      "epoch": 0.4385257174291739,
+      "learning_rate": 0.00011229669019895482,
+      "loss": 0.8297,
+      "mean_token_accuracy": 0.7459732294082642,
+      "num_tokens": 110536919.0,
+      "step": 47830
+    },
+    {
+      "epoch": 0.43861740166865315,
+      "learning_rate": 0.00011227835335105897,
+      "loss": 0.8193,
+      "mean_token_accuracy": 0.7531251728534698,
+      "num_tokens": 110559772.0,
+      "step": 47840
+    },
+    {
+      "epoch": 0.4387090859081324,
+      "learning_rate": 0.0001122600165031631,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.74596666097641,
+      "num_tokens": 110583155.0,
+      "step": 47850
+    },
+    {
+      "epoch": 0.43880077014761165,
+      "learning_rate": 0.00011224167965526726,
+      "loss": 0.8042,
+      "mean_token_accuracy": 0.7572456061840057,
+      "num_tokens": 110606133.0,
+      "step": 47860
+    },
+    {
+      "epoch": 0.43889245438709085,
+      "learning_rate": 0.00011222334280737141,
+      "loss": 0.86,
+      "mean_token_accuracy": 0.7468728244304657,
+      "num_tokens": 110629386.0,
+      "step": 47870
+    },
+    {
+      "epoch": 0.4389841386265701,
+      "learning_rate": 0.00011220500595947556,
+      "loss": 0.8094,
+      "mean_token_accuracy": 0.7484618306159974,
+      "num_tokens": 110652706.0,
+      "step": 47880
+    },
+    {
+      "epoch": 0.43907582286604935,
+      "learning_rate": 0.00011218666911157973,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7431293547153472,
+      "num_tokens": 110676470.0,
+      "step": 47890
+    },
+    {
+      "epoch": 0.43916750710552854,
+      "learning_rate": 0.00011216833226368388,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7359496176242828,
+      "num_tokens": 110698917.0,
+      "step": 47900
+    },
+    {
+      "epoch": 0.4392591913450078,
+      "learning_rate": 0.00011214999541578803,
+      "loss": 0.8782,
+      "mean_token_accuracy": 0.7332504093647003,
+      "num_tokens": 110722163.0,
+      "step": 47910
+    },
+    {
+      "epoch": 0.43935087558448704,
+      "learning_rate": 0.00011213165856789218,
+      "loss": 0.842,
+      "mean_token_accuracy": 0.7502342283725738,
+      "num_tokens": 110744412.0,
+      "step": 47920
+    },
+    {
+      "epoch": 0.43944255982396624,
+      "learning_rate": 0.00011211332171999634,
+      "loss": 0.8358,
+      "mean_token_accuracy": 0.7493318498134613,
+      "num_tokens": 110767155.0,
+      "step": 47930
+    },
+    {
+      "epoch": 0.4395342440634455,
+      "learning_rate": 0.00011209498487210049,
+      "loss": 0.8393,
+      "mean_token_accuracy": 0.7504944324493408,
+      "num_tokens": 110790698.0,
+      "step": 47940
+    },
+    {
+      "epoch": 0.43962592830292474,
+      "learning_rate": 0.00011207664802420464,
+      "loss": 0.8611,
+      "mean_token_accuracy": 0.74017373919487,
+      "num_tokens": 110813105.0,
+      "step": 47950
+    },
+    {
+      "epoch": 0.43971761254240394,
+      "learning_rate": 0.0001120583111763088,
+      "loss": 0.8906,
+      "mean_token_accuracy": 0.7391743063926697,
+      "num_tokens": 110836414.0,
+      "step": 47960
+    },
+    {
+      "epoch": 0.4398092967818832,
+      "learning_rate": 0.00011203997432841296,
+      "loss": 0.849,
+      "mean_token_accuracy": 0.7519432365894317,
+      "num_tokens": 110859412.0,
+      "step": 47970
+    },
+    {
+      "epoch": 0.43990098102136244,
+      "learning_rate": 0.00011202163748051711,
+      "loss": 0.8219,
+      "mean_token_accuracy": 0.7501432836055756,
+      "num_tokens": 110882565.0,
+      "step": 47980
+    },
+    {
+      "epoch": 0.4399926652608417,
+      "learning_rate": 0.00011200330063262126,
+      "loss": 0.8565,
+      "mean_token_accuracy": 0.7413366377353668,
+      "num_tokens": 110905873.0,
+      "step": 47990
+    },
+    {
+      "epoch": 0.4400843495003209,
+      "learning_rate": 0.00011198496378472541,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7434598743915558,
+      "num_tokens": 110928593.0,
+      "step": 48000
+    },
+    {
+      "epoch": 0.44017603373980013,
+      "learning_rate": 0.00011196662693682955,
+      "loss": 0.8569,
+      "mean_token_accuracy": 0.7399765908718109,
+      "num_tokens": 110951927.0,
+      "step": 48010
+    },
+    {
+      "epoch": 0.4402677179792794,
+      "learning_rate": 0.00011194829008893373,
+      "loss": 0.8959,
+      "mean_token_accuracy": 0.7379552721977234,
+      "num_tokens": 110974728.0,
+      "step": 48020
+    },
+    {
+      "epoch": 0.4403594022187586,
+      "learning_rate": 0.00011192995324103788,
+      "loss": 0.8563,
+      "mean_token_accuracy": 0.7470605075359344,
+      "num_tokens": 110997767.0,
+      "step": 48030
+    },
+    {
+      "epoch": 0.44045108645823783,
+      "learning_rate": 0.00011191161639314203,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7417555212974548,
+      "num_tokens": 111020601.0,
+      "step": 48040
+    },
+    {
+      "epoch": 0.4405427706977171,
+      "learning_rate": 0.00011189327954524617,
+      "loss": 0.8309,
+      "mean_token_accuracy": 0.7526868402957916,
+      "num_tokens": 111044160.0,
+      "step": 48050
+    },
+    {
+      "epoch": 0.4406344549371963,
+      "learning_rate": 0.00011187494269735032,
+      "loss": 0.8659,
+      "mean_token_accuracy": 0.7407997131347657,
+      "num_tokens": 111066463.0,
+      "step": 48060
+    },
+    {
+      "epoch": 0.4407261391766755,
+      "learning_rate": 0.00011185660584945447,
+      "loss": 0.8841,
+      "mean_token_accuracy": 0.7373033761978149,
+      "num_tokens": 111089205.0,
+      "step": 48070
+    },
+    {
+      "epoch": 0.4408178234161548,
+      "learning_rate": 0.00011183826900155863,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7385069906711579,
+      "num_tokens": 111113153.0,
+      "step": 48080
+    },
+    {
+      "epoch": 0.44090950765563397,
+      "learning_rate": 0.00011181993215366279,
+      "loss": 0.8454,
+      "mean_token_accuracy": 0.7498517215251923,
+      "num_tokens": 111136309.0,
+      "step": 48090
+    },
+    {
+      "epoch": 0.4410011918951132,
+      "learning_rate": 0.00011180159530576694,
+      "loss": 0.8045,
+      "mean_token_accuracy": 0.752761310338974,
+      "num_tokens": 111159628.0,
+      "step": 48100
+    },
+    {
+      "epoch": 0.44109287613459247,
+      "learning_rate": 0.0001117832584578711,
+      "loss": 0.8274,
+      "mean_token_accuracy": 0.7500851690769196,
+      "num_tokens": 111181301.0,
+      "step": 48110
+    },
+    {
+      "epoch": 0.4411845603740717,
+      "learning_rate": 0.00011176492160997525,
+      "loss": 0.8878,
+      "mean_token_accuracy": 0.73563272356987,
+      "num_tokens": 111205095.0,
+      "step": 48120
+    },
+    {
+      "epoch": 0.4412762446135509,
+      "learning_rate": 0.0001117465847620794,
+      "loss": 0.8276,
+      "mean_token_accuracy": 0.7553383767604828,
+      "num_tokens": 111229087.0,
+      "step": 48130
+    },
+    {
+      "epoch": 0.44136792885303017,
+      "learning_rate": 0.00011172824791418355,
+      "loss": 0.8027,
+      "mean_token_accuracy": 0.7565086603164672,
+      "num_tokens": 111252412.0,
+      "step": 48140
+    },
+    {
+      "epoch": 0.4414596130925094,
+      "learning_rate": 0.00011170991106628772,
+      "loss": 0.8535,
+      "mean_token_accuracy": 0.7428046345710755,
+      "num_tokens": 111276166.0,
+      "step": 48150
+    },
+    {
+      "epoch": 0.4415512973319886,
+      "learning_rate": 0.00011169157421839187,
+      "loss": 0.8743,
+      "mean_token_accuracy": 0.7416279554367066,
+      "num_tokens": 111299985.0,
+      "step": 48160
+    },
+    {
+      "epoch": 0.44164298157146786,
+      "learning_rate": 0.00011167323737049602,
+      "loss": 0.8538,
+      "mean_token_accuracy": 0.7420009493827819,
+      "num_tokens": 111322962.0,
+      "step": 48170
+    },
+    {
+      "epoch": 0.4417346658109471,
+      "learning_rate": 0.00011165490052260017,
+      "loss": 0.8049,
+      "mean_token_accuracy": 0.7578908443450928,
+      "num_tokens": 111346803.0,
+      "step": 48180
+    },
+    {
+      "epoch": 0.4418263500504263,
+      "learning_rate": 0.00011163656367470432,
+      "loss": 0.8412,
+      "mean_token_accuracy": 0.7449650347232819,
+      "num_tokens": 111370770.0,
+      "step": 48190
+    },
+    {
+      "epoch": 0.44191803428990556,
+      "learning_rate": 0.00011161822682680848,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7434915244579315,
+      "num_tokens": 111393568.0,
+      "step": 48200
+    },
+    {
+      "epoch": 0.4420097185293848,
+      "learning_rate": 0.00011159988997891261,
+      "loss": 0.8212,
+      "mean_token_accuracy": 0.7529119253158569,
+      "num_tokens": 111415946.0,
+      "step": 48210
+    },
+    {
+      "epoch": 0.442101402768864,
+      "learning_rate": 0.0001115815531310168,
+      "loss": 0.8207,
+      "mean_token_accuracy": 0.7478813529014587,
+      "num_tokens": 111439298.0,
+      "step": 48220
+    },
+    {
+      "epoch": 0.44219308700834326,
+      "learning_rate": 0.00011156321628312095,
+      "loss": 0.8414,
+      "mean_token_accuracy": 0.7486931622028351,
+      "num_tokens": 111461606.0,
+      "step": 48230
+    },
+    {
+      "epoch": 0.4422847712478225,
+      "learning_rate": 0.0001115448794352251,
+      "loss": 0.8721,
+      "mean_token_accuracy": 0.734430581331253,
+      "num_tokens": 111485670.0,
+      "step": 48240
+    },
+    {
+      "epoch": 0.44237645548730176,
+      "learning_rate": 0.00011152654258732924,
+      "loss": 0.803,
+      "mean_token_accuracy": 0.752145254611969,
+      "num_tokens": 111508206.0,
+      "step": 48250
+    },
+    {
+      "epoch": 0.44246813972678095,
+      "learning_rate": 0.00011150820573943339,
+      "loss": 0.8162,
+      "mean_token_accuracy": 0.7478123664855957,
+      "num_tokens": 111531283.0,
+      "step": 48260
+    },
+    {
+      "epoch": 0.4425598239662602,
+      "learning_rate": 0.00011148986889153754,
+      "loss": 0.8832,
+      "mean_token_accuracy": 0.732526296377182,
+      "num_tokens": 111554499.0,
+      "step": 48270
+    },
+    {
+      "epoch": 0.44265150820573945,
+      "learning_rate": 0.00011147153204364172,
+      "loss": 0.8333,
+      "mean_token_accuracy": 0.7498092830181122,
+      "num_tokens": 111577813.0,
+      "step": 48280
+    },
+    {
+      "epoch": 0.44274319244521865,
+      "learning_rate": 0.00011145319519574586,
+      "loss": 0.8205,
+      "mean_token_accuracy": 0.7563820898532867,
+      "num_tokens": 111600687.0,
+      "step": 48290
+    },
+    {
+      "epoch": 0.4428348766846979,
+      "learning_rate": 0.00011143485834785001,
+      "loss": 0.8373,
+      "mean_token_accuracy": 0.7509080648422242,
+      "num_tokens": 111624675.0,
+      "step": 48300
+    },
+    {
+      "epoch": 0.44292656092417715,
+      "learning_rate": 0.00011141652149995416,
+      "loss": 0.876,
+      "mean_token_accuracy": 0.745772922039032,
+      "num_tokens": 111648130.0,
+      "step": 48310
+    },
+    {
+      "epoch": 0.44301824516365634,
+      "learning_rate": 0.00011139818465205831,
+      "loss": 0.8116,
+      "mean_token_accuracy": 0.7508679330348969,
+      "num_tokens": 111671383.0,
+      "step": 48320
+    },
+    {
+      "epoch": 0.4431099294031356,
+      "learning_rate": 0.00011137984780416246,
+      "loss": 0.8242,
+      "mean_token_accuracy": 0.744922548532486,
+      "num_tokens": 111694055.0,
+      "step": 48330
+    },
+    {
+      "epoch": 0.44320161364261484,
+      "learning_rate": 0.00011136151095626662,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.739057207107544,
+      "num_tokens": 111716887.0,
+      "step": 48340
+    },
+    {
+      "epoch": 0.4432932978820941,
+      "learning_rate": 0.00011134317410837078,
+      "loss": 0.8019,
+      "mean_token_accuracy": 0.755261492729187,
+      "num_tokens": 111740165.0,
+      "step": 48350
+    },
+    {
+      "epoch": 0.4433849821215733,
+      "learning_rate": 0.00011132483726047493,
+      "loss": 0.8231,
+      "mean_token_accuracy": 0.755534029006958,
+      "num_tokens": 111762941.0,
+      "step": 48360
+    },
+    {
+      "epoch": 0.44347666636105254,
+      "learning_rate": 0.00011130650041257909,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.7432640075683594,
+      "num_tokens": 111786033.0,
+      "step": 48370
+    },
+    {
+      "epoch": 0.4435683506005318,
+      "learning_rate": 0.00011128816356468324,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7367870032787323,
+      "num_tokens": 111809438.0,
+      "step": 48380
+    },
+    {
+      "epoch": 0.443660034840011,
+      "learning_rate": 0.00011126982671678739,
+      "loss": 0.8162,
+      "mean_token_accuracy": 0.7507876813411712,
+      "num_tokens": 111832186.0,
+      "step": 48390
+    },
+    {
+      "epoch": 0.44375171907949024,
+      "learning_rate": 0.00011125148986889154,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.7526997089385986,
+      "num_tokens": 111855065.0,
+      "step": 48400
+    },
+    {
+      "epoch": 0.4438434033189695,
+      "learning_rate": 0.0001112331530209957,
+      "loss": 0.8694,
+      "mean_token_accuracy": 0.7352891981601715,
+      "num_tokens": 111878264.0,
+      "step": 48410
+    },
+    {
+      "epoch": 0.4439350875584487,
+      "learning_rate": 0.00011121481617309986,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7428710043430329,
+      "num_tokens": 111900964.0,
+      "step": 48420
+    },
+    {
+      "epoch": 0.44402677179792793,
+      "learning_rate": 0.00011119647932520401,
+      "loss": 0.8669,
+      "mean_token_accuracy": 0.7485485553741456,
+      "num_tokens": 111924686.0,
+      "step": 48430
+    },
+    {
+      "epoch": 0.4441184560374072,
+      "learning_rate": 0.00011117814247730816,
+      "loss": 0.8378,
+      "mean_token_accuracy": 0.7454011261463165,
+      "num_tokens": 111947085.0,
+      "step": 48440
+    },
+    {
+      "epoch": 0.4442101402768864,
+      "learning_rate": 0.0001111598056294123,
+      "loss": 0.7867,
+      "mean_token_accuracy": 0.762846690416336,
+      "num_tokens": 111969613.0,
+      "step": 48450
+    },
+    {
+      "epoch": 0.44430182451636563,
+      "learning_rate": 0.00011114146878151645,
+      "loss": 0.8488,
+      "mean_token_accuracy": 0.747734671831131,
+      "num_tokens": 111992591.0,
+      "step": 48460
+    },
+    {
+      "epoch": 0.4443935087558449,
+      "learning_rate": 0.0001111231319336206,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7515955924987793,
+      "num_tokens": 112015781.0,
+      "step": 48470
+    },
+    {
+      "epoch": 0.44448519299532413,
+      "learning_rate": 0.00011110479508572478,
+      "loss": 0.8622,
+      "mean_token_accuracy": 0.739305853843689,
+      "num_tokens": 112038556.0,
+      "step": 48480
+    },
+    {
+      "epoch": 0.4445768772348033,
+      "learning_rate": 0.00011108645823782892,
+      "loss": 0.8199,
+      "mean_token_accuracy": 0.7545050203800201,
+      "num_tokens": 112061245.0,
+      "step": 48490
+    },
+    {
+      "epoch": 0.4446685614742826,
+      "learning_rate": 0.00011106812138993307,
+      "loss": 0.8519,
+      "mean_token_accuracy": 0.7491310596466064,
+      "num_tokens": 112084157.0,
+      "step": 48500
+    },
+    {
+      "epoch": 0.4447602457137618,
+      "learning_rate": 0.00011104978454203722,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.743950366973877,
+      "num_tokens": 112106816.0,
+      "step": 48510
+    },
+    {
+      "epoch": 0.444851929953241,
+      "learning_rate": 0.00011103144769414138,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.7435440301895142,
+      "num_tokens": 112130547.0,
+      "step": 48520
+    },
+    {
+      "epoch": 0.44494361419272027,
+      "learning_rate": 0.00011101311084624553,
+      "loss": 0.8379,
+      "mean_token_accuracy": 0.7546590387821197,
+      "num_tokens": 112153080.0,
+      "step": 48530
+    },
+    {
+      "epoch": 0.4450352984321995,
+      "learning_rate": 0.0001109947739983497,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.7477194607257843,
+      "num_tokens": 112175546.0,
+      "step": 48540
+    },
+    {
+      "epoch": 0.4451269826716787,
+      "learning_rate": 0.00011097643715045385,
+      "loss": 0.8207,
+      "mean_token_accuracy": 0.7554192423820496,
+      "num_tokens": 112199320.0,
+      "step": 48550
+    },
+    {
+      "epoch": 0.44521866691115797,
+      "learning_rate": 0.000110958100302558,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.7486271440982819,
+      "num_tokens": 112222922.0,
+      "step": 48560
+    },
+    {
+      "epoch": 0.4453103511506372,
+      "learning_rate": 0.00011093976345466215,
+      "loss": 0.8,
+      "mean_token_accuracy": 0.7520759165287018,
+      "num_tokens": 112246848.0,
+      "step": 48570
+    },
+    {
+      "epoch": 0.4454020353901164,
+      "learning_rate": 0.0001109214266067663,
+      "loss": 0.8281,
+      "mean_token_accuracy": 0.7456780135631561,
+      "num_tokens": 112270313.0,
+      "step": 48580
+    },
+    {
+      "epoch": 0.44549371962959566,
+      "learning_rate": 0.00011090308975887045,
+      "loss": 0.8382,
+      "mean_token_accuracy": 0.7441405475139617,
+      "num_tokens": 112293406.0,
+      "step": 48590
+    },
+    {
+      "epoch": 0.4455854038690749,
+      "learning_rate": 0.0001108847529109746,
+      "loss": 0.885,
+      "mean_token_accuracy": 0.7378743767738343,
+      "num_tokens": 112316479.0,
+      "step": 48600
+    },
+    {
+      "epoch": 0.44567708810855416,
+      "learning_rate": 0.00011086641606307877,
+      "loss": 0.8191,
+      "mean_token_accuracy": 0.7532954156398773,
+      "num_tokens": 112339542.0,
+      "step": 48610
+    },
+    {
+      "epoch": 0.44576877234803336,
+      "learning_rate": 0.00011084807921518292,
+      "loss": 0.8242,
+      "mean_token_accuracy": 0.7503843426704406,
+      "num_tokens": 112362658.0,
+      "step": 48620
+    },
+    {
+      "epoch": 0.4458604565875126,
+      "learning_rate": 0.00011082974236728707,
+      "loss": 0.8274,
+      "mean_token_accuracy": 0.7489453911781311,
+      "num_tokens": 112385581.0,
+      "step": 48630
+    },
+    {
+      "epoch": 0.44595214082699186,
+      "learning_rate": 0.00011081140551939123,
+      "loss": 0.8532,
+      "mean_token_accuracy": 0.7436773359775544,
+      "num_tokens": 112408418.0,
+      "step": 48640
+    },
+    {
+      "epoch": 0.44604382506647106,
+      "learning_rate": 0.00011079306867149536,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7423886835575104,
+      "num_tokens": 112431512.0,
+      "step": 48650
+    },
+    {
+      "epoch": 0.4461355093059503,
+      "learning_rate": 0.00011077473182359952,
+      "loss": 0.8121,
+      "mean_token_accuracy": 0.7494610130786896,
+      "num_tokens": 112454697.0,
+      "step": 48660
+    },
+    {
+      "epoch": 0.44622719354542956,
+      "learning_rate": 0.0001107563949757037,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.746228039264679,
+      "num_tokens": 112478021.0,
+      "step": 48670
+    },
+    {
+      "epoch": 0.44631887778490875,
+      "learning_rate": 0.00011073805812780785,
+      "loss": 0.8319,
+      "mean_token_accuracy": 0.7489956200122834,
+      "num_tokens": 112500980.0,
+      "step": 48680
+    },
+    {
+      "epoch": 0.446410562024388,
+      "learning_rate": 0.00011071972127991199,
+      "loss": 0.8488,
+      "mean_token_accuracy": 0.746698260307312,
+      "num_tokens": 112524369.0,
+      "step": 48690
+    },
+    {
+      "epoch": 0.44650224626386725,
+      "learning_rate": 0.00011070138443201614,
+      "loss": 0.8455,
+      "mean_token_accuracy": 0.7442999124526978,
+      "num_tokens": 112548066.0,
+      "step": 48700
+    },
+    {
+      "epoch": 0.44659393050334645,
+      "learning_rate": 0.00011068304758412029,
+      "loss": 0.8301,
+      "mean_token_accuracy": 0.7480234801769257,
+      "num_tokens": 112571076.0,
+      "step": 48710
+    },
+    {
+      "epoch": 0.4466856147428257,
+      "learning_rate": 0.00011066471073622444,
+      "loss": 0.842,
+      "mean_token_accuracy": 0.7521705389022827,
+      "num_tokens": 112594507.0,
+      "step": 48720
+    },
+    {
+      "epoch": 0.44677729898230495,
+      "learning_rate": 0.00011064637388832859,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7517581045627594,
+      "num_tokens": 112618156.0,
+      "step": 48730
+    },
+    {
+      "epoch": 0.4468689832217842,
+      "learning_rate": 0.00011062803704043276,
+      "loss": 0.8267,
+      "mean_token_accuracy": 0.7492414057254791,
+      "num_tokens": 112642309.0,
+      "step": 48740
+    },
+    {
+      "epoch": 0.4469606674612634,
+      "learning_rate": 0.00011060970019253691,
+      "loss": 0.7948,
+      "mean_token_accuracy": 0.7600102841854095,
+      "num_tokens": 112665549.0,
+      "step": 48750
+    },
+    {
+      "epoch": 0.44705235170074265,
+      "learning_rate": 0.00011059136334464106,
+      "loss": 0.8529,
+      "mean_token_accuracy": 0.7448238372802735,
+      "num_tokens": 112688820.0,
+      "step": 48760
+    },
+    {
+      "epoch": 0.4471440359402219,
+      "learning_rate": 0.00011057302649674521,
+      "loss": 0.7746,
+      "mean_token_accuracy": 0.7631296992301941,
+      "num_tokens": 112711825.0,
+      "step": 48770
+    },
+    {
+      "epoch": 0.4472357201797011,
+      "learning_rate": 0.00011055468964884937,
+      "loss": 0.8449,
+      "mean_token_accuracy": 0.7509827733039856,
+      "num_tokens": 112735456.0,
+      "step": 48780
+    },
+    {
+      "epoch": 0.44732740441918034,
+      "learning_rate": 0.00011053635280095352,
+      "loss": 0.8539,
+      "mean_token_accuracy": 0.7425627887248993,
+      "num_tokens": 112758342.0,
+      "step": 48790
+    },
+    {
+      "epoch": 0.4474190886586596,
+      "learning_rate": 0.00011051801595305768,
+      "loss": 0.8925,
+      "mean_token_accuracy": 0.7421483218669891,
+      "num_tokens": 112781609.0,
+      "step": 48800
+    },
+    {
+      "epoch": 0.4475107728981388,
+      "learning_rate": 0.00011049967910516184,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7388948678970337,
+      "num_tokens": 112804584.0,
+      "step": 48810
+    },
+    {
+      "epoch": 0.44760245713761804,
+      "learning_rate": 0.00011048134225726599,
+      "loss": 0.8518,
+      "mean_token_accuracy": 0.7474654912948608,
+      "num_tokens": 112827395.0,
+      "step": 48820
+    },
+    {
+      "epoch": 0.4476941413770973,
+      "learning_rate": 0.00011046300540937014,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7417679548263549,
+      "num_tokens": 112851109.0,
+      "step": 48830
+    },
+    {
+      "epoch": 0.4477858256165765,
+      "learning_rate": 0.00011044466856147429,
+      "loss": 0.8388,
+      "mean_token_accuracy": 0.7523158729076386,
+      "num_tokens": 112874602.0,
+      "step": 48840
+    },
+    {
+      "epoch": 0.44787750985605573,
+      "learning_rate": 0.00011042633171357843,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7469957709312439,
+      "num_tokens": 112897642.0,
+      "step": 48850
+    },
+    {
+      "epoch": 0.447969194095535,
+      "learning_rate": 0.00011040799486568258,
+      "loss": 0.8798,
+      "mean_token_accuracy": 0.7374691367149353,
+      "num_tokens": 112921366.0,
+      "step": 48860
+    },
+    {
+      "epoch": 0.44806087833501423,
+      "learning_rate": 0.00011038965801778676,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7467567145824432,
+      "num_tokens": 112944561.0,
+      "step": 48870
+    },
+    {
+      "epoch": 0.44815256257449343,
+      "learning_rate": 0.00011037132116989091,
+      "loss": 0.8436,
+      "mean_token_accuracy": 0.747118490934372,
+      "num_tokens": 112967862.0,
+      "step": 48880
+    },
+    {
+      "epoch": 0.4482442468139727,
+      "learning_rate": 0.00011035298432199505,
+      "loss": 0.8251,
+      "mean_token_accuracy": 0.7517585754394531,
+      "num_tokens": 112990840.0,
+      "step": 48890
+    },
+    {
+      "epoch": 0.44833593105345193,
+      "learning_rate": 0.0001103346474740992,
+      "loss": 0.8541,
+      "mean_token_accuracy": 0.7512996613979339,
+      "num_tokens": 113014034.0,
+      "step": 48900
+    },
+    {
+      "epoch": 0.4484276152929311,
+      "learning_rate": 0.00011031631062620335,
+      "loss": 0.8539,
+      "mean_token_accuracy": 0.7437214195728302,
+      "num_tokens": 113037423.0,
+      "step": 48910
+    },
+    {
+      "epoch": 0.4485192995324104,
+      "learning_rate": 0.0001102979737783075,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.7485715508460998,
+      "num_tokens": 113061167.0,
+      "step": 48920
+    },
+    {
+      "epoch": 0.4486109837718896,
+      "learning_rate": 0.00011027963693041168,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.7500149130821228,
+      "num_tokens": 113084455.0,
+      "step": 48930
+    },
+    {
+      "epoch": 0.4487026680113688,
+      "learning_rate": 0.00011026130008251582,
+      "loss": 0.8333,
+      "mean_token_accuracy": 0.742855429649353,
+      "num_tokens": 113107016.0,
+      "step": 48940
+    },
+    {
+      "epoch": 0.44879435225084807,
+      "learning_rate": 0.00011024296323461997,
+      "loss": 0.8603,
+      "mean_token_accuracy": 0.743027514219284,
+      "num_tokens": 113129446.0,
+      "step": 48950
+    },
+    {
+      "epoch": 0.4488860364903273,
+      "learning_rate": 0.00011022462638672413,
+      "loss": 0.8164,
+      "mean_token_accuracy": 0.7533604979515076,
+      "num_tokens": 113152457.0,
+      "step": 48960
+    },
+    {
+      "epoch": 0.4489777207298066,
+      "learning_rate": 0.00011020628953882828,
+      "loss": 0.8203,
+      "mean_token_accuracy": 0.7508922159671784,
+      "num_tokens": 113176081.0,
+      "step": 48970
+    },
+    {
+      "epoch": 0.44906940496928577,
+      "learning_rate": 0.00011018795269093243,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7423801779747009,
+      "num_tokens": 113199096.0,
+      "step": 48980
+    },
+    {
+      "epoch": 0.449161089208765,
+      "learning_rate": 0.00011016961584303658,
+      "loss": 0.8227,
+      "mean_token_accuracy": 0.7502870976924896,
+      "num_tokens": 113223453.0,
+      "step": 48990
+    },
+    {
+      "epoch": 0.44925277344824427,
+      "learning_rate": 0.00011015127899514075,
+      "loss": 0.8709,
+      "mean_token_accuracy": 0.7444837629795075,
+      "num_tokens": 113246436.0,
+      "step": 49000
+    },
+    {
+      "epoch": 0.44934445768772346,
+      "learning_rate": 0.0001101329421472449,
+      "loss": 0.8286,
+      "mean_token_accuracy": 0.7501603424549103,
+      "num_tokens": 113269700.0,
+      "step": 49010
+    },
+    {
+      "epoch": 0.4494361419272027,
+      "learning_rate": 0.00011011460529934905,
+      "loss": 0.8225,
+      "mean_token_accuracy": 0.7534201204776764,
+      "num_tokens": 113293917.0,
+      "step": 49020
+    },
+    {
+      "epoch": 0.44952782616668197,
+      "learning_rate": 0.0001100962684514532,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.748278695344925,
+      "num_tokens": 113316884.0,
+      "step": 49030
+    },
+    {
+      "epoch": 0.44961951040616116,
+      "learning_rate": 0.00011007793160355736,
+      "loss": 0.8707,
+      "mean_token_accuracy": 0.744801926612854,
+      "num_tokens": 113340314.0,
+      "step": 49040
+    },
+    {
+      "epoch": 0.4497111946456404,
+      "learning_rate": 0.0001100595947556615,
+      "loss": 0.8447,
+      "mean_token_accuracy": 0.7505634129047394,
+      "num_tokens": 113363471.0,
+      "step": 49050
+    },
+    {
+      "epoch": 0.44980287888511966,
+      "learning_rate": 0.00011004125790776567,
+      "loss": 0.8275,
+      "mean_token_accuracy": 0.7532466113567352,
+      "num_tokens": 113386679.0,
+      "step": 49060
+    },
+    {
+      "epoch": 0.44989456312459886,
+      "learning_rate": 0.00011002292105986982,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7395869612693786,
+      "num_tokens": 113410078.0,
+      "step": 49070
+    },
+    {
+      "epoch": 0.4499862473640781,
+      "learning_rate": 0.00011000458421197398,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7496163547039032,
+      "num_tokens": 113433602.0,
+      "step": 49080
+    },
+    {
+      "epoch": 0.45007793160355736,
+      "learning_rate": 0.00010998624736407811,
+      "loss": 0.8174,
+      "mean_token_accuracy": 0.7479480087757111,
+      "num_tokens": 113455785.0,
+      "step": 49090
+    },
+    {
+      "epoch": 0.4501696158430366,
+      "learning_rate": 0.00010996791051618227,
+      "loss": 0.8109,
+      "mean_token_accuracy": 0.7530344069004059,
+      "num_tokens": 113479378.0,
+      "step": 49100
+    },
+    {
+      "epoch": 0.4502613000825158,
+      "learning_rate": 0.00010994957366828642,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.7382154881954193,
+      "num_tokens": 113503006.0,
+      "step": 49110
+    },
+    {
+      "epoch": 0.45035298432199505,
+      "learning_rate": 0.00010993123682039057,
+      "loss": 0.8161,
+      "mean_token_accuracy": 0.7520354568958283,
+      "num_tokens": 113526408.0,
+      "step": 49120
+    },
+    {
+      "epoch": 0.4504446685614743,
+      "learning_rate": 0.00010991289997249475,
+      "loss": 0.8194,
+      "mean_token_accuracy": 0.7565617859363556,
+      "num_tokens": 113549645.0,
+      "step": 49130
+    },
+    {
+      "epoch": 0.4505363528009535,
+      "learning_rate": 0.00010989456312459889,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7475900232791901,
+      "num_tokens": 113572660.0,
+      "step": 49140
+    },
+    {
+      "epoch": 0.45062803704043275,
+      "learning_rate": 0.00010987622627670304,
+      "loss": 0.842,
+      "mean_token_accuracy": 0.7509126484394073,
+      "num_tokens": 113595738.0,
+      "step": 49150
+    },
+    {
+      "epoch": 0.450719721279912,
+      "learning_rate": 0.00010985788942880719,
+      "loss": 0.8846,
+      "mean_token_accuracy": 0.7325856447219848,
+      "num_tokens": 113618373.0,
+      "step": 49160
+    },
+    {
+      "epoch": 0.4508114055193912,
+      "learning_rate": 0.00010983955258091134,
+      "loss": 0.8355,
+      "mean_token_accuracy": 0.755406379699707,
+      "num_tokens": 113641563.0,
+      "step": 49170
+    },
+    {
+      "epoch": 0.45090308975887045,
+      "learning_rate": 0.0001098212157330155,
+      "loss": 0.8411,
+      "mean_token_accuracy": 0.7422263979911804,
+      "num_tokens": 113665643.0,
+      "step": 49180
+    },
+    {
+      "epoch": 0.4509947739983497,
+      "learning_rate": 0.00010980287888511965,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7456914603710174,
+      "num_tokens": 113688658.0,
+      "step": 49190
+    },
+    {
+      "epoch": 0.4510864582378289,
+      "learning_rate": 0.00010978454203722381,
+      "loss": 0.8761,
+      "mean_token_accuracy": 0.7379150867462159,
+      "num_tokens": 113711771.0,
+      "step": 49200
+    },
+    {
+      "epoch": 0.45117814247730814,
+      "learning_rate": 0.00010976620518932796,
+      "loss": 0.8219,
+      "mean_token_accuracy": 0.750760293006897,
+      "num_tokens": 113734772.0,
+      "step": 49210
+    },
+    {
+      "epoch": 0.4512698267167874,
+      "learning_rate": 0.00010974786834143212,
+      "loss": 0.8573,
+      "mean_token_accuracy": 0.7350454568862915,
+      "num_tokens": 113757914.0,
+      "step": 49220
+    },
+    {
+      "epoch": 0.45136151095626664,
+      "learning_rate": 0.00010972953149353627,
+      "loss": 0.797,
+      "mean_token_accuracy": 0.7552305936813355,
+      "num_tokens": 113780439.0,
+      "step": 49230
+    },
+    {
+      "epoch": 0.45145319519574584,
+      "learning_rate": 0.00010971119464564042,
+      "loss": 0.8196,
+      "mean_token_accuracy": 0.7575195908546448,
+      "num_tokens": 113803385.0,
+      "step": 49240
+    },
+    {
+      "epoch": 0.4515448794352251,
+      "learning_rate": 0.00010969285779774456,
+      "loss": 0.8887,
+      "mean_token_accuracy": 0.7384458303451538,
+      "num_tokens": 113826477.0,
+      "step": 49250
+    },
+    {
+      "epoch": 0.45163656367470434,
+      "learning_rate": 0.00010967452094984874,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.7467671036720276,
+      "num_tokens": 113850545.0,
+      "step": 49260
+    },
+    {
+      "epoch": 0.45172824791418353,
+      "learning_rate": 0.00010965618410195289,
+      "loss": 0.8208,
+      "mean_token_accuracy": 0.7519316792488098,
+      "num_tokens": 113874981.0,
+      "step": 49270
+    },
+    {
+      "epoch": 0.4518199321536628,
+      "learning_rate": 0.00010963784725405704,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7524272322654724,
+      "num_tokens": 113897540.0,
+      "step": 49280
+    },
+    {
+      "epoch": 0.45191161639314203,
+      "learning_rate": 0.00010961951040616118,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.739130049943924,
+      "num_tokens": 113921403.0,
+      "step": 49290
+    },
+    {
+      "epoch": 0.45200330063262123,
+      "learning_rate": 0.00010960117355826533,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.7519500851631165,
+      "num_tokens": 113944458.0,
+      "step": 49300
+    },
+    {
+      "epoch": 0.4520949848721005,
+      "learning_rate": 0.00010958283671036948,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.751862770318985,
+      "num_tokens": 113967690.0,
+      "step": 49310
+    },
+    {
+      "epoch": 0.45218666911157973,
+      "learning_rate": 0.00010956449986247363,
+      "loss": 0.8458,
+      "mean_token_accuracy": 0.7447650969028473,
+      "num_tokens": 113991343.0,
+      "step": 49320
+    },
+    {
+      "epoch": 0.4522783533510589,
+      "learning_rate": 0.00010954616301457781,
+      "loss": 0.8986,
+      "mean_token_accuracy": 0.7406126022338867,
+      "num_tokens": 114014515.0,
+      "step": 49330
+    },
+    {
+      "epoch": 0.4523700375905382,
+      "learning_rate": 0.00010952782616668195,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.749345576763153,
+      "num_tokens": 114037608.0,
+      "step": 49340
+    },
+    {
+      "epoch": 0.4524617218300174,
+      "learning_rate": 0.0001095094893187861,
+      "loss": 0.8354,
+      "mean_token_accuracy": 0.7491923153400422,
+      "num_tokens": 114061200.0,
+      "step": 49350
+    },
+    {
+      "epoch": 0.4525534060694967,
+      "learning_rate": 0.00010949115247089026,
+      "loss": 0.8482,
+      "mean_token_accuracy": 0.7426029682159424,
+      "num_tokens": 114084074.0,
+      "step": 49360
+    },
+    {
+      "epoch": 0.4526450903089759,
+      "learning_rate": 0.00010947281562299441,
+      "loss": 0.8104,
+      "mean_token_accuracy": 0.7577181577682495,
+      "num_tokens": 114107740.0,
+      "step": 49370
+    },
+    {
+      "epoch": 0.4527367745484551,
+      "learning_rate": 0.00010945447877509856,
+      "loss": 0.823,
+      "mean_token_accuracy": 0.7490709841251373,
+      "num_tokens": 114130569.0,
+      "step": 49380
+    },
+    {
+      "epoch": 0.4528284587879344,
+      "learning_rate": 0.00010943614192720272,
+      "loss": 0.812,
+      "mean_token_accuracy": 0.7540494084358216,
+      "num_tokens": 114154651.0,
+      "step": 49390
+    },
+    {
+      "epoch": 0.45292014302741357,
+      "learning_rate": 0.00010941780507930688,
+      "loss": 0.8759,
+      "mean_token_accuracy": 0.7451411008834838,
+      "num_tokens": 114177721.0,
+      "step": 49400
+    },
+    {
+      "epoch": 0.4530118272668928,
+      "learning_rate": 0.00010939946823141103,
+      "loss": 0.8165,
+      "mean_token_accuracy": 0.7556455373764038,
+      "num_tokens": 114200656.0,
+      "step": 49410
+    },
+    {
+      "epoch": 0.45310351150637207,
+      "learning_rate": 0.00010938113138351518,
+      "loss": 0.8431,
+      "mean_token_accuracy": 0.7506132960319519,
+      "num_tokens": 114224071.0,
+      "step": 49420
+    },
+    {
+      "epoch": 0.45319519574585126,
+      "learning_rate": 0.00010936279453561933,
+      "loss": 0.8505,
+      "mean_token_accuracy": 0.746786379814148,
+      "num_tokens": 114247298.0,
+      "step": 49430
+    },
+    {
+      "epoch": 0.4532868799853305,
+      "learning_rate": 0.00010934445768772348,
+      "loss": 0.8104,
+      "mean_token_accuracy": 0.7564125180244445,
+      "num_tokens": 114269942.0,
+      "step": 49440
+    },
+    {
+      "epoch": 0.45337856422480977,
+      "learning_rate": 0.00010932612083982762,
+      "loss": 0.8259,
+      "mean_token_accuracy": 0.7480632960796356,
+      "num_tokens": 114293166.0,
+      "step": 49450
+    },
+    {
+      "epoch": 0.453470248464289,
+      "learning_rate": 0.0001093077839919318,
+      "loss": 0.8311,
+      "mean_token_accuracy": 0.7524271965026855,
+      "num_tokens": 114316551.0,
+      "step": 49460
+    },
+    {
+      "epoch": 0.4535619327037682,
+      "learning_rate": 0.00010928944714403595,
+      "loss": 0.8592,
+      "mean_token_accuracy": 0.742966377735138,
+      "num_tokens": 114339708.0,
+      "step": 49470
+    },
+    {
+      "epoch": 0.45365361694324746,
+      "learning_rate": 0.0001092711102961401,
+      "loss": 0.8483,
+      "mean_token_accuracy": 0.7440168797969818,
+      "num_tokens": 114363328.0,
+      "step": 49480
+    },
+    {
+      "epoch": 0.4537453011827267,
+      "learning_rate": 0.00010925277344824424,
+      "loss": 0.8145,
+      "mean_token_accuracy": 0.7530664384365082,
+      "num_tokens": 114386281.0,
+      "step": 49490
+    },
+    {
+      "epoch": 0.4538369854222059,
+      "learning_rate": 0.0001092344366003484,
+      "loss": 0.8144,
+      "mean_token_accuracy": 0.7575432062149048,
+      "num_tokens": 114409297.0,
+      "step": 49500
+    },
+    {
+      "epoch": 0.45392866966168516,
+      "learning_rate": 0.00010921609975245255,
+      "loss": 0.8343,
+      "mean_token_accuracy": 0.7455037117004395,
+      "num_tokens": 114432655.0,
+      "step": 49510
+    },
+    {
+      "epoch": 0.4540203539011644,
+      "learning_rate": 0.00010919776290455673,
+      "loss": 0.8576,
+      "mean_token_accuracy": 0.7393884003162384,
+      "num_tokens": 114455275.0,
+      "step": 49520
+    },
+    {
+      "epoch": 0.4541120381406436,
+      "learning_rate": 0.00010917942605666088,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.746986186504364,
+      "num_tokens": 114478223.0,
+      "step": 49530
+    },
+    {
+      "epoch": 0.45420372238012285,
+      "learning_rate": 0.00010916108920876502,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.7462375462055206,
+      "num_tokens": 114501432.0,
+      "step": 49540
+    },
+    {
+      "epoch": 0.4542954066196021,
+      "learning_rate": 0.00010914275236086917,
+      "loss": 0.8068,
+      "mean_token_accuracy": 0.7507108628749848,
+      "num_tokens": 114524940.0,
+      "step": 49550
+    },
+    {
+      "epoch": 0.4543870908590813,
+      "learning_rate": 0.00010912441551297332,
+      "loss": 0.8599,
+      "mean_token_accuracy": 0.7398683190345764,
+      "num_tokens": 114547775.0,
+      "step": 49560
+    },
+    {
+      "epoch": 0.45447877509856055,
+      "learning_rate": 0.00010910607866507747,
+      "loss": 0.851,
+      "mean_token_accuracy": 0.7424595952033997,
+      "num_tokens": 114570577.0,
+      "step": 49570
+    },
+    {
+      "epoch": 0.4545704593380398,
+      "learning_rate": 0.00010908774181718162,
+      "loss": 0.8313,
+      "mean_token_accuracy": 0.7421681165695191,
+      "num_tokens": 114593620.0,
+      "step": 49580
+    },
+    {
+      "epoch": 0.45466214357751905,
+      "learning_rate": 0.00010906940496928579,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7452627182006836,
+      "num_tokens": 114616443.0,
+      "step": 49590
+    },
+    {
+      "epoch": 0.45475382781699825,
+      "learning_rate": 0.00010905106812138994,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7472487986087799,
+      "num_tokens": 114640011.0,
+      "step": 49600
+    },
+    {
+      "epoch": 0.4548455120564775,
+      "learning_rate": 0.00010903273127349409,
+      "loss": 0.8251,
+      "mean_token_accuracy": 0.7492351531982422,
+      "num_tokens": 114663381.0,
+      "step": 49610
+    },
+    {
+      "epoch": 0.45493719629595675,
+      "learning_rate": 0.00010901439442559824,
+      "loss": 0.8166,
+      "mean_token_accuracy": 0.753437340259552,
+      "num_tokens": 114687242.0,
+      "step": 49620
+    },
+    {
+      "epoch": 0.45502888053543594,
+      "learning_rate": 0.0001089960575777024,
+      "loss": 0.8655,
+      "mean_token_accuracy": 0.7422393739223481,
+      "num_tokens": 114709454.0,
+      "step": 49630
+    },
+    {
+      "epoch": 0.4551205647749152,
+      "learning_rate": 0.00010897772072980655,
+      "loss": 0.8238,
+      "mean_token_accuracy": 0.7529425501823426,
+      "num_tokens": 114732151.0,
+      "step": 49640
+    },
+    {
+      "epoch": 0.45521224901439444,
+      "learning_rate": 0.00010895938388191071,
+      "loss": 0.8713,
+      "mean_token_accuracy": 0.7446607828140259,
+      "num_tokens": 114755040.0,
+      "step": 49650
+    },
+    {
+      "epoch": 0.45530393325387364,
+      "learning_rate": 0.00010894104703401487,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.7471251487731934,
+      "num_tokens": 114778102.0,
+      "step": 49660
+    },
+    {
+      "epoch": 0.4553956174933529,
+      "learning_rate": 0.00010892271018611902,
+      "loss": 0.8518,
+      "mean_token_accuracy": 0.7474866032600402,
+      "num_tokens": 114800788.0,
+      "step": 49670
+    },
+    {
+      "epoch": 0.45548730173283214,
+      "learning_rate": 0.00010890437333822317,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7449249029159546,
+      "num_tokens": 114823909.0,
+      "step": 49680
+    },
+    {
+      "epoch": 0.45557898597231133,
+      "learning_rate": 0.00010888603649032731,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7440057277679444,
+      "num_tokens": 114847934.0,
+      "step": 49690
+    },
+    {
+      "epoch": 0.4556706702117906,
+      "learning_rate": 0.00010886769964243146,
+      "loss": 0.8459,
+      "mean_token_accuracy": 0.7479400277137757,
+      "num_tokens": 114871114.0,
+      "step": 49700
+    },
+    {
+      "epoch": 0.45576235445126984,
+      "learning_rate": 0.00010884936279453561,
+      "loss": 0.8441,
+      "mean_token_accuracy": 0.745722359418869,
+      "num_tokens": 114895228.0,
+      "step": 49710
+    },
+    {
+      "epoch": 0.4558540386907491,
+      "learning_rate": 0.00010883102594663979,
+      "loss": 0.8146,
+      "mean_token_accuracy": 0.7478174984455108,
+      "num_tokens": 114919367.0,
+      "step": 49720
+    },
+    {
+      "epoch": 0.4559457229302283,
+      "learning_rate": 0.00010881268909874394,
+      "loss": 0.8631,
+      "mean_token_accuracy": 0.7402715504169464,
+      "num_tokens": 114942509.0,
+      "step": 49730
+    },
+    {
+      "epoch": 0.45603740716970753,
+      "learning_rate": 0.00010879435225084808,
+      "loss": 0.8297,
+      "mean_token_accuracy": 0.7462266325950623,
+      "num_tokens": 114966233.0,
+      "step": 49740
+    },
+    {
+      "epoch": 0.4561290914091868,
+      "learning_rate": 0.00010877601540295223,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.7475348711013794,
+      "num_tokens": 114989497.0,
+      "step": 49750
+    },
+    {
+      "epoch": 0.456220775648666,
+      "learning_rate": 0.00010875767855505638,
+      "loss": 0.8924,
+      "mean_token_accuracy": 0.7382708609104156,
+      "num_tokens": 115012636.0,
+      "step": 49760
+    },
+    {
+      "epoch": 0.4563124598881452,
+      "learning_rate": 0.00010873934170716054,
+      "loss": 0.8237,
+      "mean_token_accuracy": 0.754587596654892,
+      "num_tokens": 115035701.0,
+      "step": 49770
+    },
+    {
+      "epoch": 0.4564041441276245,
+      "learning_rate": 0.0001087210048592647,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7473047614097595,
+      "num_tokens": 115058173.0,
+      "step": 49780
+    },
+    {
+      "epoch": 0.4564958283671037,
+      "learning_rate": 0.00010870266801136885,
+      "loss": 0.8303,
+      "mean_token_accuracy": 0.7516798675060272,
+      "num_tokens": 115080946.0,
+      "step": 49790
+    },
+    {
+      "epoch": 0.4565875126065829,
+      "learning_rate": 0.000108684331163473,
+      "loss": 0.9069,
+      "mean_token_accuracy": 0.7308391451835632,
+      "num_tokens": 115104503.0,
+      "step": 49800
+    },
+    {
+      "epoch": 0.4566791968460622,
+      "learning_rate": 0.00010866599431557716,
+      "loss": 0.8685,
+      "mean_token_accuracy": 0.741806811094284,
+      "num_tokens": 115128179.0,
+      "step": 49810
+    },
+    {
+      "epoch": 0.45677088108554137,
+      "learning_rate": 0.00010864765746768131,
+      "loss": 0.8707,
+      "mean_token_accuracy": 0.7365410387516022,
+      "num_tokens": 115152009.0,
+      "step": 49820
+    },
+    {
+      "epoch": 0.4568625653250206,
+      "learning_rate": 0.00010862932061978546,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7453692495822907,
+      "num_tokens": 115174952.0,
+      "step": 49830
+    },
+    {
+      "epoch": 0.45695424956449987,
+      "learning_rate": 0.00010861098377188961,
+      "loss": 0.8778,
+      "mean_token_accuracy": 0.7436991155147552,
+      "num_tokens": 115197963.0,
+      "step": 49840
+    },
+    {
+      "epoch": 0.4570459338039791,
+      "learning_rate": 0.00010859264692399378,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7575132727622986,
+      "num_tokens": 115220780.0,
+      "step": 49850
+    },
+    {
+      "epoch": 0.4571376180434583,
+      "learning_rate": 0.00010857431007609793,
+      "loss": 0.8327,
+      "mean_token_accuracy": 0.7503166556358337,
+      "num_tokens": 115242933.0,
+      "step": 49860
+    },
+    {
+      "epoch": 0.45722930228293757,
+      "learning_rate": 0.00010855597322820208,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7496700286865234,
+      "num_tokens": 115266172.0,
+      "step": 49870
+    },
+    {
+      "epoch": 0.4573209865224168,
+      "learning_rate": 0.00010853763638030623,
+      "loss": 0.8497,
+      "mean_token_accuracy": 0.7433329045772552,
+      "num_tokens": 115289280.0,
+      "step": 49880
+    },
+    {
+      "epoch": 0.457412670761896,
+      "learning_rate": 0.00010851929953241037,
+      "loss": 0.8299,
+      "mean_token_accuracy": 0.7510054588317872,
+      "num_tokens": 115312699.0,
+      "step": 49890
+    },
+    {
+      "epoch": 0.45750435500137526,
+      "learning_rate": 0.00010850096268451452,
+      "loss": 0.8395,
+      "mean_token_accuracy": 0.7443045854568482,
+      "num_tokens": 115335620.0,
+      "step": 49900
+    },
+    {
+      "epoch": 0.4575960392408545,
+      "learning_rate": 0.0001084826258366187,
+      "loss": 0.8996,
+      "mean_token_accuracy": 0.7296887338161469,
+      "num_tokens": 115358098.0,
+      "step": 49910
+    },
+    {
+      "epoch": 0.4576877234803337,
+      "learning_rate": 0.00010846428898872285,
+      "loss": 0.8163,
+      "mean_token_accuracy": 0.751824027299881,
+      "num_tokens": 115381361.0,
+      "step": 49920
+    },
+    {
+      "epoch": 0.45777940771981296,
+      "learning_rate": 0.00010844595214082701,
+      "loss": 0.8277,
+      "mean_token_accuracy": 0.7506865561008453,
+      "num_tokens": 115404842.0,
+      "step": 49930
+    },
+    {
+      "epoch": 0.4578710919592922,
+      "learning_rate": 0.00010842761529293115,
+      "loss": 0.8266,
+      "mean_token_accuracy": 0.7501639425754547,
+      "num_tokens": 115427625.0,
+      "step": 49940
+    },
+    {
+      "epoch": 0.4579627761987714,
+      "learning_rate": 0.0001084092784450353,
+      "loss": 0.8644,
+      "mean_token_accuracy": 0.7454055070877075,
+      "num_tokens": 115451230.0,
+      "step": 49950
+    },
+    {
+      "epoch": 0.45805446043825065,
+      "learning_rate": 0.00010839094159713945,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.7487506866455078,
+      "num_tokens": 115474683.0,
+      "step": 49960
+    },
+    {
+      "epoch": 0.4581461446777299,
+      "learning_rate": 0.0001083726047492436,
+      "loss": 0.8183,
+      "mean_token_accuracy": 0.7492614328861237,
+      "num_tokens": 115497894.0,
+      "step": 49970
+    },
+    {
+      "epoch": 0.45823782891720916,
+      "learning_rate": 0.00010835426790134777,
+      "loss": 0.8423,
+      "mean_token_accuracy": 0.749496477842331,
+      "num_tokens": 115521237.0,
+      "step": 49980
+    },
+    {
+      "epoch": 0.45832951315668835,
+      "learning_rate": 0.00010833593105345192,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.746222198009491,
+      "num_tokens": 115544763.0,
+      "step": 49990
+    },
+    {
+      "epoch": 0.4584211973961676,
+      "learning_rate": 0.00010831759420555607,
+      "loss": 0.8175,
+      "mean_token_accuracy": 0.7521841049194335,
+      "num_tokens": 115568240.0,
+      "step": 50000
+    },
+    {
+      "epoch": 0.45851288163564685,
+      "learning_rate": 0.00010829925735766022,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7403563261032104,
+      "num_tokens": 115591813.0,
+      "step": 50010
+    },
+    {
+      "epoch": 0.45860456587512605,
+      "learning_rate": 0.00010828092050976437,
+      "loss": 0.8313,
+      "mean_token_accuracy": 0.7544568061828614,
+      "num_tokens": 115614726.0,
+      "step": 50020
+    },
+    {
+      "epoch": 0.4586962501146053,
+      "learning_rate": 0.00010826258366186853,
+      "loss": 0.8242,
+      "mean_token_accuracy": 0.7494811654090882,
+      "num_tokens": 115638111.0,
+      "step": 50030
+    },
+    {
+      "epoch": 0.45878793435408455,
+      "learning_rate": 0.00010824424681397269,
+      "loss": 0.8091,
+      "mean_token_accuracy": 0.7566789090633392,
+      "num_tokens": 115661738.0,
+      "step": 50040
+    },
+    {
+      "epoch": 0.45887961859356374,
+      "learning_rate": 0.00010822590996607684,
+      "loss": 0.8165,
+      "mean_token_accuracy": 0.7553297936916351,
+      "num_tokens": 115684907.0,
+      "step": 50050
+    },
+    {
+      "epoch": 0.458971302833043,
+      "learning_rate": 0.000108207573118181,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7515370666980743,
+      "num_tokens": 115707567.0,
+      "step": 50060
+    },
+    {
+      "epoch": 0.45906298707252224,
+      "learning_rate": 0.00010818923627028515,
+      "loss": 0.8275,
+      "mean_token_accuracy": 0.754007738828659,
+      "num_tokens": 115730192.0,
+      "step": 50070
+    },
+    {
+      "epoch": 0.4591546713120015,
+      "learning_rate": 0.0001081708994223893,
+      "loss": 0.8023,
+      "mean_token_accuracy": 0.757024222612381,
+      "num_tokens": 115752620.0,
+      "step": 50080
+    },
+    {
+      "epoch": 0.4592463555514807,
+      "learning_rate": 0.00010815256257449344,
+      "loss": 0.8688,
+      "mean_token_accuracy": 0.7419677913188935,
+      "num_tokens": 115776581.0,
+      "step": 50090
+    },
+    {
+      "epoch": 0.45933803979095994,
+      "learning_rate": 0.00010813422572659759,
+      "loss": 0.8853,
+      "mean_token_accuracy": 0.7376410245895386,
+      "num_tokens": 115799978.0,
+      "step": 50100
+    },
+    {
+      "epoch": 0.4594297240304392,
+      "learning_rate": 0.00010811588887870177,
+      "loss": 0.8144,
+      "mean_token_accuracy": 0.7546159029006958,
+      "num_tokens": 115823279.0,
+      "step": 50110
+    },
+    {
+      "epoch": 0.4595214082699184,
+      "learning_rate": 0.00010809755203080592,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7492096900939942,
+      "num_tokens": 115846249.0,
+      "step": 50120
+    },
+    {
+      "epoch": 0.45961309250939764,
+      "learning_rate": 0.00010807921518291007,
+      "loss": 0.8244,
+      "mean_token_accuracy": 0.7569547533988953,
+      "num_tokens": 115869287.0,
+      "step": 50130
+    },
+    {
+      "epoch": 0.4597047767488769,
+      "learning_rate": 0.00010806087833501421,
+      "loss": 0.8025,
+      "mean_token_accuracy": 0.7493510842323303,
+      "num_tokens": 115892147.0,
+      "step": 50140
+    },
+    {
+      "epoch": 0.4597964609883561,
+      "learning_rate": 0.00010804254148711836,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7442805290222168,
+      "num_tokens": 115915570.0,
+      "step": 50150
+    },
+    {
+      "epoch": 0.45988814522783533,
+      "learning_rate": 0.00010802420463922251,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7487399339675903,
+      "num_tokens": 115938993.0,
+      "step": 50160
+    },
+    {
+      "epoch": 0.4599798294673146,
+      "learning_rate": 0.00010800586779132669,
+      "loss": 0.8402,
+      "mean_token_accuracy": 0.7479621529579162,
+      "num_tokens": 115961349.0,
+      "step": 50170
+    },
+    {
+      "epoch": 0.4600715137067938,
+      "learning_rate": 0.00010798753094343083,
+      "loss": 0.8231,
+      "mean_token_accuracy": 0.7495421350002289,
+      "num_tokens": 115984834.0,
+      "step": 50180
+    },
+    {
+      "epoch": 0.46016319794627303,
+      "learning_rate": 0.00010796919409553498,
+      "loss": 0.8541,
+      "mean_token_accuracy": 0.7462188839912415,
+      "num_tokens": 116007818.0,
+      "step": 50190
+    },
+    {
+      "epoch": 0.4602548821857523,
+      "learning_rate": 0.00010795085724763913,
+      "loss": 0.8535,
+      "mean_token_accuracy": 0.7470133304595947,
+      "num_tokens": 116031305.0,
+      "step": 50200
+    },
+    {
+      "epoch": 0.46034656642523153,
+      "learning_rate": 0.00010793252039974329,
+      "loss": 0.8249,
+      "mean_token_accuracy": 0.7510017812252044,
+      "num_tokens": 116054213.0,
+      "step": 50210
+    },
+    {
+      "epoch": 0.4604382506647107,
+      "learning_rate": 0.00010791418355184744,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7504799664020538,
+      "num_tokens": 116076506.0,
+      "step": 50220
+    },
+    {
+      "epoch": 0.46052993490419,
+      "learning_rate": 0.00010789584670395159,
+      "loss": 0.8627,
+      "mean_token_accuracy": 0.7414623498916626,
+      "num_tokens": 116100639.0,
+      "step": 50230
+    },
+    {
+      "epoch": 0.4606216191436692,
+      "learning_rate": 0.00010787750985605576,
+      "loss": 0.8896,
+      "mean_token_accuracy": 0.7352072656154632,
+      "num_tokens": 116124090.0,
+      "step": 50240
+    },
+    {
+      "epoch": 0.4607133033831484,
+      "learning_rate": 0.00010785917300815991,
+      "loss": 0.8479,
+      "mean_token_accuracy": 0.7485179603099823,
+      "num_tokens": 116146498.0,
+      "step": 50250
+    },
+    {
+      "epoch": 0.46080498762262767,
+      "learning_rate": 0.00010784083616026406,
+      "loss": 0.8164,
+      "mean_token_accuracy": 0.7551505208015442,
+      "num_tokens": 116169667.0,
+      "step": 50260
+    },
+    {
+      "epoch": 0.4608966718621069,
+      "learning_rate": 0.00010782249931236821,
+      "loss": 0.8514,
+      "mean_token_accuracy": 0.7464648723602295,
+      "num_tokens": 116192840.0,
+      "step": 50270
+    },
+    {
+      "epoch": 0.4609883561015861,
+      "learning_rate": 0.00010780416246447236,
+      "loss": 0.824,
+      "mean_token_accuracy": 0.7510474324226379,
+      "num_tokens": 116215831.0,
+      "step": 50280
+    },
+    {
+      "epoch": 0.46108004034106537,
+      "learning_rate": 0.0001077858256165765,
+      "loss": 0.8225,
+      "mean_token_accuracy": 0.7547004103660584,
+      "num_tokens": 116238647.0,
+      "step": 50290
+    },
+    {
+      "epoch": 0.4611717245805446,
+      "learning_rate": 0.00010776748876868068,
+      "loss": 0.8619,
+      "mean_token_accuracy": 0.7416106343269349,
+      "num_tokens": 116261417.0,
+      "step": 50300
+    },
+    {
+      "epoch": 0.4612634088200238,
+      "learning_rate": 0.00010774915192078483,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7462671041488648,
+      "num_tokens": 116285092.0,
+      "step": 50310
+    },
+    {
+      "epoch": 0.46135509305950306,
+      "learning_rate": 0.00010773081507288898,
+      "loss": 0.8562,
+      "mean_token_accuracy": 0.7486227571964263,
+      "num_tokens": 116308572.0,
+      "step": 50320
+    },
+    {
+      "epoch": 0.4614467772989823,
+      "learning_rate": 0.00010771247822499314,
+      "loss": 0.8318,
+      "mean_token_accuracy": 0.7512663900852203,
+      "num_tokens": 116331203.0,
+      "step": 50330
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "learning_rate": 0.00010769414137709727,
+      "loss": 0.8212,
+      "mean_token_accuracy": 0.7521014273166656,
+      "num_tokens": 116353705.0,
+      "step": 50340
+    },
+    {
+      "epoch": 0.46163014577794076,
+      "learning_rate": 0.00010767580452920143,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7383383989334107,
+      "num_tokens": 116376058.0,
+      "step": 50350
+    },
+    {
+      "epoch": 0.46172183001742,
+      "learning_rate": 0.00010765746768130558,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.7493795573711395,
+      "num_tokens": 116399235.0,
+      "step": 50360
+    },
+    {
+      "epoch": 0.46181351425689926,
+      "learning_rate": 0.00010763913083340976,
+      "loss": 0.8411,
+      "mean_token_accuracy": 0.7499105215072632,
+      "num_tokens": 116422093.0,
+      "step": 50370
+    },
+    {
+      "epoch": 0.46190519849637846,
+      "learning_rate": 0.0001076207939855139,
+      "loss": 0.8264,
+      "mean_token_accuracy": 0.7485403299331665,
+      "num_tokens": 116444540.0,
+      "step": 50380
+    },
+    {
+      "epoch": 0.4619968827358577,
+      "learning_rate": 0.00010760245713761805,
+      "loss": 0.8763,
+      "mean_token_accuracy": 0.7494730412960052,
+      "num_tokens": 116468067.0,
+      "step": 50390
+    },
+    {
+      "epoch": 0.46208856697533696,
+      "learning_rate": 0.0001075841202897222,
+      "loss": 0.8229,
+      "mean_token_accuracy": 0.7540560901165009,
+      "num_tokens": 116490872.0,
+      "step": 50400
+    },
+    {
+      "epoch": 0.46218025121481615,
+      "learning_rate": 0.00010756578344182635,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7454632878303528,
+      "num_tokens": 116513739.0,
+      "step": 50410
+    },
+    {
+      "epoch": 0.4622719354542954,
+      "learning_rate": 0.0001075474465939305,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.7500062763690949,
+      "num_tokens": 116536617.0,
+      "step": 50420
+    },
+    {
+      "epoch": 0.46236361969377465,
+      "learning_rate": 0.00010752910974603465,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7448347449302674,
+      "num_tokens": 116559527.0,
+      "step": 50430
+    },
+    {
+      "epoch": 0.46245530393325385,
+      "learning_rate": 0.00010751077289813882,
+      "loss": 0.8291,
+      "mean_token_accuracy": 0.7494826912879944,
+      "num_tokens": 116582109.0,
+      "step": 50440
+    },
+    {
+      "epoch": 0.4625469881727331,
+      "learning_rate": 0.00010749243605024297,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7425334811210632,
+      "num_tokens": 116605397.0,
+      "step": 50450
+    },
+    {
+      "epoch": 0.46263867241221235,
+      "learning_rate": 0.00010747409920234712,
+      "loss": 0.8183,
+      "mean_token_accuracy": 0.7540872037410736,
+      "num_tokens": 116627798.0,
+      "step": 50460
+    },
+    {
+      "epoch": 0.4627303566516916,
+      "learning_rate": 0.00010745576235445128,
+      "loss": 0.8242,
+      "mean_token_accuracy": 0.7543343484401703,
+      "num_tokens": 116650549.0,
+      "step": 50470
+    },
+    {
+      "epoch": 0.4628220408911708,
+      "learning_rate": 0.00010743742550655543,
+      "loss": 0.8319,
+      "mean_token_accuracy": 0.7507722616195679,
+      "num_tokens": 116673922.0,
+      "step": 50480
+    },
+    {
+      "epoch": 0.46291372513065004,
+      "learning_rate": 0.00010741908865865957,
+      "loss": 0.8355,
+      "mean_token_accuracy": 0.7464515626430511,
+      "num_tokens": 116697639.0,
+      "step": 50490
+    },
+    {
+      "epoch": 0.4630054093701293,
+      "learning_rate": 0.00010740075181076374,
+      "loss": 0.8437,
+      "mean_token_accuracy": 0.7505419135093689,
+      "num_tokens": 116720789.0,
+      "step": 50500
+    },
+    {
+      "epoch": 0.4630970936096085,
+      "learning_rate": 0.0001073824149628679,
+      "loss": 0.8655,
+      "mean_token_accuracy": 0.7391029357910156,
+      "num_tokens": 116744169.0,
+      "step": 50510
+    },
+    {
+      "epoch": 0.46318877784908774,
+      "learning_rate": 0.00010736407811497205,
+      "loss": 0.8311,
+      "mean_token_accuracy": 0.7503938496112823,
+      "num_tokens": 116766188.0,
+      "step": 50520
+    },
+    {
+      "epoch": 0.463280462088567,
+      "learning_rate": 0.0001073457412670762,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.744571739435196,
+      "num_tokens": 116788861.0,
+      "step": 50530
+    },
+    {
+      "epoch": 0.4633721463280462,
+      "learning_rate": 0.00010732740441918034,
+      "loss": 0.8259,
+      "mean_token_accuracy": 0.749758392572403,
+      "num_tokens": 116811487.0,
+      "step": 50540
+    },
+    {
+      "epoch": 0.46346383056752544,
+      "learning_rate": 0.00010730906757128449,
+      "loss": 0.8064,
+      "mean_token_accuracy": 0.753113043308258,
+      "num_tokens": 116835181.0,
+      "step": 50550
+    },
+    {
+      "epoch": 0.4635555148070047,
+      "learning_rate": 0.00010729073072338864,
+      "loss": 0.8395,
+      "mean_token_accuracy": 0.7491614937782287,
+      "num_tokens": 116857859.0,
+      "step": 50560
+    },
+    {
+      "epoch": 0.4636471990464839,
+      "learning_rate": 0.00010727239387549282,
+      "loss": 0.8222,
+      "mean_token_accuracy": 0.7530656456947327,
+      "num_tokens": 116881321.0,
+      "step": 50570
+    },
+    {
+      "epoch": 0.46373888328596313,
+      "learning_rate": 0.00010725405702759696,
+      "loss": 0.793,
+      "mean_token_accuracy": 0.7553530097007751,
+      "num_tokens": 116904790.0,
+      "step": 50580
+    },
+    {
+      "epoch": 0.4638305675254424,
+      "learning_rate": 0.00010723572017970111,
+      "loss": 0.8549,
+      "mean_token_accuracy": 0.7461807429790497,
+      "num_tokens": 116928194.0,
+      "step": 50590
+    },
+    {
+      "epoch": 0.46392225176492163,
+      "learning_rate": 0.00010721738333180526,
+      "loss": 0.8175,
+      "mean_token_accuracy": 0.7489982545375824,
+      "num_tokens": 116951407.0,
+      "step": 50600
+    },
+    {
+      "epoch": 0.46401393600440083,
+      "learning_rate": 0.00010719904648390942,
+      "loss": 0.8439,
+      "mean_token_accuracy": 0.7496041774749755,
+      "num_tokens": 116975147.0,
+      "step": 50610
+    },
+    {
+      "epoch": 0.4641056202438801,
+      "learning_rate": 0.00010718070963601357,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7445078730583191,
+      "num_tokens": 116999365.0,
+      "step": 50620
+    },
+    {
+      "epoch": 0.46419730448335933,
+      "learning_rate": 0.00010716237278811773,
+      "loss": 0.8451,
+      "mean_token_accuracy": 0.7497836112976074,
+      "num_tokens": 117022386.0,
+      "step": 50630
+    },
+    {
+      "epoch": 0.4642889887228385,
+      "learning_rate": 0.00010714403594022188,
+      "loss": 0.8437,
+      "mean_token_accuracy": 0.7496808290481567,
+      "num_tokens": 117045041.0,
+      "step": 50640
+    },
+    {
+      "epoch": 0.4643806729623178,
+      "learning_rate": 0.00010712569909232604,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7458152651786805,
+      "num_tokens": 117068307.0,
+      "step": 50650
+    },
+    {
+      "epoch": 0.464472357201797,
+      "learning_rate": 0.00010710736224443019,
+      "loss": 0.8496,
+      "mean_token_accuracy": 0.7513097286224365,
+      "num_tokens": 117091801.0,
+      "step": 50660
+    },
+    {
+      "epoch": 0.4645640414412762,
+      "learning_rate": 0.00010708902539653434,
+      "loss": 0.7991,
+      "mean_token_accuracy": 0.7584814131259918,
+      "num_tokens": 117114593.0,
+      "step": 50670
+    },
+    {
+      "epoch": 0.46465572568075547,
+      "learning_rate": 0.00010707068854863849,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7379404366016388,
+      "num_tokens": 117137486.0,
+      "step": 50680
+    },
+    {
+      "epoch": 0.4647474099202347,
+      "learning_rate": 0.00010705235170074264,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.7391998946666718,
+      "num_tokens": 117160033.0,
+      "step": 50690
+    },
+    {
+      "epoch": 0.464839094159714,
+      "learning_rate": 0.00010703401485284681,
+      "loss": 0.8627,
+      "mean_token_accuracy": 0.7434830665588379,
+      "num_tokens": 117183427.0,
+      "step": 50700
+    },
+    {
+      "epoch": 0.46493077839919317,
+      "learning_rate": 0.00010701567800495096,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7426767408847809,
+      "num_tokens": 117206345.0,
+      "step": 50710
+    },
+    {
+      "epoch": 0.4650224626386724,
+      "learning_rate": 0.00010699734115705511,
+      "loss": 0.8835,
+      "mean_token_accuracy": 0.7401986122131348,
+      "num_tokens": 117229219.0,
+      "step": 50720
+    },
+    {
+      "epoch": 0.46511414687815167,
+      "learning_rate": 0.00010697900430915926,
+      "loss": 0.8578,
+      "mean_token_accuracy": 0.7393283247947693,
+      "num_tokens": 117252991.0,
+      "step": 50730
+    },
+    {
+      "epoch": 0.46520583111763086,
+      "learning_rate": 0.0001069606674612634,
+      "loss": 0.7975,
+      "mean_token_accuracy": 0.7621454775333405,
+      "num_tokens": 117275853.0,
+      "step": 50740
+    },
+    {
+      "epoch": 0.4652975153571101,
+      "learning_rate": 0.00010694233061336755,
+      "loss": 0.823,
+      "mean_token_accuracy": 0.7500155448913575,
+      "num_tokens": 117298624.0,
+      "step": 50750
+    },
+    {
+      "epoch": 0.46538919959658936,
+      "learning_rate": 0.00010692399376547173,
+      "loss": 0.8217,
+      "mean_token_accuracy": 0.7510518133640289,
+      "num_tokens": 117322138.0,
+      "step": 50760
+    },
+    {
+      "epoch": 0.46548088383606856,
+      "learning_rate": 0.00010690565691757589,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.748747569322586,
+      "num_tokens": 117345049.0,
+      "step": 50770
+    },
+    {
+      "epoch": 0.4655725680755478,
+      "learning_rate": 0.00010688732006968002,
+      "loss": 0.8154,
+      "mean_token_accuracy": 0.7526210367679596,
+      "num_tokens": 117367844.0,
+      "step": 50780
+    },
+    {
+      "epoch": 0.46566425231502706,
+      "learning_rate": 0.00010686898322178418,
+      "loss": 0.891,
+      "mean_token_accuracy": 0.7360921025276184,
+      "num_tokens": 117391359.0,
+      "step": 50790
+    },
+    {
+      "epoch": 0.46575593655450626,
+      "learning_rate": 0.00010685064637388833,
+      "loss": 0.8708,
+      "mean_token_accuracy": 0.7436611413955688,
+      "num_tokens": 117414531.0,
+      "step": 50800
+    },
+    {
+      "epoch": 0.4658476207939855,
+      "learning_rate": 0.00010683230952599248,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7603997647762298,
+      "num_tokens": 117437361.0,
+      "step": 50810
+    },
+    {
+      "epoch": 0.46593930503346476,
+      "learning_rate": 0.00010681397267809663,
+      "loss": 0.8376,
+      "mean_token_accuracy": 0.7506824672222138,
+      "num_tokens": 117460283.0,
+      "step": 50820
+    },
+    {
+      "epoch": 0.466030989272944,
+      "learning_rate": 0.0001067956358302008,
+      "loss": 0.8133,
+      "mean_token_accuracy": 0.7531838774681091,
+      "num_tokens": 117482509.0,
+      "step": 50830
+    },
+    {
+      "epoch": 0.4661226735124232,
+      "learning_rate": 0.00010677729898230495,
+      "loss": 0.8531,
+      "mean_token_accuracy": 0.7507034957408905,
+      "num_tokens": 117506382.0,
+      "step": 50840
+    },
+    {
+      "epoch": 0.46621435775190245,
+      "learning_rate": 0.0001067589621344091,
+      "loss": 0.8281,
+      "mean_token_accuracy": 0.7464346587657928,
+      "num_tokens": 117529661.0,
+      "step": 50850
+    },
+    {
+      "epoch": 0.4663060419913817,
+      "learning_rate": 0.00010674062528651325,
+      "loss": 0.8491,
+      "mean_token_accuracy": 0.7451833069324494,
+      "num_tokens": 117552202.0,
+      "step": 50860
+    },
+    {
+      "epoch": 0.4663977262308609,
+      "learning_rate": 0.0001067222884386174,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.748199737071991,
+      "num_tokens": 117576254.0,
+      "step": 50870
+    },
+    {
+      "epoch": 0.46648941047034015,
+      "learning_rate": 0.00010670395159072156,
+      "loss": 0.8353,
+      "mean_token_accuracy": 0.7499429762363434,
+      "num_tokens": 117599490.0,
+      "step": 50880
+    },
+    {
+      "epoch": 0.4665810947098194,
+      "learning_rate": 0.00010668561474282572,
+      "loss": 0.8031,
+      "mean_token_accuracy": 0.7485783219337463,
+      "num_tokens": 117622567.0,
+      "step": 50890
+    },
+    {
+      "epoch": 0.4666727789492986,
+      "learning_rate": 0.00010666727789492987,
+      "loss": 0.8374,
+      "mean_token_accuracy": 0.7473032593727111,
+      "num_tokens": 117645775.0,
+      "step": 50900
+    },
+    {
+      "epoch": 0.46676446318877785,
+      "learning_rate": 0.00010664894104703403,
+      "loss": 0.8202,
+      "mean_token_accuracy": 0.7585340678691864,
+      "num_tokens": 117669341.0,
+      "step": 50910
+    },
+    {
+      "epoch": 0.4668561474282571,
+      "learning_rate": 0.00010663060419913818,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7439858019351959,
+      "num_tokens": 117691666.0,
+      "step": 50920
+    },
+    {
+      "epoch": 0.4669478316677363,
+      "learning_rate": 0.00010661226735124233,
+      "loss": 0.8275,
+      "mean_token_accuracy": 0.7561112999916076,
+      "num_tokens": 117715316.0,
+      "step": 50930
+    },
+    {
+      "epoch": 0.46703951590721554,
+      "learning_rate": 0.00010659393050334647,
+      "loss": 0.8734,
+      "mean_token_accuracy": 0.740497225522995,
+      "num_tokens": 117738787.0,
+      "step": 50940
+    },
+    {
+      "epoch": 0.4671312001466948,
+      "learning_rate": 0.00010657559365545062,
+      "loss": 0.8505,
+      "mean_token_accuracy": 0.7452231168746948,
+      "num_tokens": 117761967.0,
+      "step": 50950
+    },
+    {
+      "epoch": 0.46722288438617404,
+      "learning_rate": 0.0001065572568075548,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7463249921798706,
+      "num_tokens": 117784543.0,
+      "step": 50960
+    },
+    {
+      "epoch": 0.46731456862565324,
+      "learning_rate": 0.00010653891995965895,
+      "loss": 0.8127,
+      "mean_token_accuracy": 0.7490747451782227,
+      "num_tokens": 117806932.0,
+      "step": 50970
+    },
+    {
+      "epoch": 0.4674062528651325,
+      "learning_rate": 0.00010652058311176309,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7450135529041291,
+      "num_tokens": 117830057.0,
+      "step": 50980
+    },
+    {
+      "epoch": 0.46749793710461174,
+      "learning_rate": 0.00010650224626386724,
+      "loss": 0.8774,
+      "mean_token_accuracy": 0.7434897541999816,
+      "num_tokens": 117853793.0,
+      "step": 50990
+    },
+    {
+      "epoch": 0.46758962134409093,
+      "learning_rate": 0.00010648390941597139,
+      "loss": 0.8635,
+      "mean_token_accuracy": 0.7430111706256867,
+      "num_tokens": 117876448.0,
+      "step": 51000
+    },
+    {
+      "epoch": 0.4676813055835702,
+      "learning_rate": 0.00010646557256807554,
+      "loss": 0.8815,
+      "mean_token_accuracy": 0.7391182601451873,
+      "num_tokens": 117899109.0,
+      "step": 51010
+    },
+    {
+      "epoch": 0.46777298982304943,
+      "learning_rate": 0.00010644723572017972,
+      "loss": 0.8334,
+      "mean_token_accuracy": 0.7455284893512726,
+      "num_tokens": 117923303.0,
+      "step": 51020
+    },
+    {
+      "epoch": 0.46786467406252863,
+      "learning_rate": 0.00010642889887228386,
+      "loss": 0.8003,
+      "mean_token_accuracy": 0.7555950045585632,
+      "num_tokens": 117945852.0,
+      "step": 51030
+    },
+    {
+      "epoch": 0.4679563583020079,
+      "learning_rate": 0.00010641056202438801,
+      "loss": 0.8312,
+      "mean_token_accuracy": 0.7510682940483093,
+      "num_tokens": 117970004.0,
+      "step": 51040
+    },
+    {
+      "epoch": 0.46804804254148713,
+      "learning_rate": 0.00010639222517649217,
+      "loss": 0.8312,
+      "mean_token_accuracy": 0.7443981349468232,
+      "num_tokens": 117991660.0,
+      "step": 51050
+    },
+    {
+      "epoch": 0.4681397267809663,
+      "learning_rate": 0.00010637388832859632,
+      "loss": 0.8511,
+      "mean_token_accuracy": 0.7525094747543335,
+      "num_tokens": 118014556.0,
+      "step": 51060
+    },
+    {
+      "epoch": 0.4682314110204456,
+      "learning_rate": 0.00010635555148070047,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.7467299461364746,
+      "num_tokens": 118037560.0,
+      "step": 51070
+    },
+    {
+      "epoch": 0.4683230952599248,
+      "learning_rate": 0.00010633721463280462,
+      "loss": 0.8672,
+      "mean_token_accuracy": 0.7387905538082122,
+      "num_tokens": 118060953.0,
+      "step": 51080
+    },
+    {
+      "epoch": 0.4684147794994041,
+      "learning_rate": 0.00010631887778490879,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7508937537670135,
+      "num_tokens": 118084704.0,
+      "step": 51090
+    },
+    {
+      "epoch": 0.46850646373888327,
+      "learning_rate": 0.00010630054093701294,
+      "loss": 0.858,
+      "mean_token_accuracy": 0.7452883899211884,
+      "num_tokens": 118108778.0,
+      "step": 51100
+    },
+    {
+      "epoch": 0.4685981479783625,
+      "learning_rate": 0.00010628220408911709,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7486472249031066,
+      "num_tokens": 118131700.0,
+      "step": 51110
+    },
+    {
+      "epoch": 0.4686898322178418,
+      "learning_rate": 0.00010626386724122124,
+      "loss": 0.8769,
+      "mean_token_accuracy": 0.7451722025871277,
+      "num_tokens": 118155078.0,
+      "step": 51120
+    },
+    {
+      "epoch": 0.46878151645732097,
+      "learning_rate": 0.0001062455303933254,
+      "loss": 0.8744,
+      "mean_token_accuracy": 0.743290901184082,
+      "num_tokens": 118178973.0,
+      "step": 51130
+    },
+    {
+      "epoch": 0.4688732006968002,
+      "learning_rate": 0.00010622719354542953,
+      "loss": 0.8354,
+      "mean_token_accuracy": 0.7442569673061371,
+      "num_tokens": 118202027.0,
+      "step": 51140
+    },
+    {
+      "epoch": 0.46896488493627947,
+      "learning_rate": 0.00010620885669753371,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7528537273406982,
+      "num_tokens": 118225208.0,
+      "step": 51150
+    },
+    {
+      "epoch": 0.46905656917575866,
+      "learning_rate": 0.00010619051984963786,
+      "loss": 0.8307,
+      "mean_token_accuracy": 0.7523713350296021,
+      "num_tokens": 118248682.0,
+      "step": 51160
+    },
+    {
+      "epoch": 0.4691482534152379,
+      "learning_rate": 0.00010617218300174201,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.7432997345924377,
+      "num_tokens": 118271357.0,
+      "step": 51170
+    },
+    {
+      "epoch": 0.46923993765471717,
+      "learning_rate": 0.00010615384615384615,
+      "loss": 0.8127,
+      "mean_token_accuracy": 0.7542767465114594,
+      "num_tokens": 118293294.0,
+      "step": 51180
+    },
+    {
+      "epoch": 0.46933162189419636,
+      "learning_rate": 0.0001061355093059503,
+      "loss": 0.8512,
+      "mean_token_accuracy": 0.7457583487033844,
+      "num_tokens": 118316071.0,
+      "step": 51190
+    },
+    {
+      "epoch": 0.4694233061336756,
+      "learning_rate": 0.00010611717245805446,
+      "loss": 0.805,
+      "mean_token_accuracy": 0.7538500905036927,
+      "num_tokens": 118338236.0,
+      "step": 51200
+    },
+    {
+      "epoch": 0.46951499037315486,
+      "learning_rate": 0.00010609883561015861,
+      "loss": 0.8382,
+      "mean_token_accuracy": 0.7423735499382019,
+      "num_tokens": 118362457.0,
+      "step": 51210
+    },
+    {
+      "epoch": 0.4696066746126341,
+      "learning_rate": 0.00010608049876226279,
+      "loss": 0.8256,
+      "mean_token_accuracy": 0.7476546823978424,
+      "num_tokens": 118386486.0,
+      "step": 51220
+    },
+    {
+      "epoch": 0.4696983588521133,
+      "learning_rate": 0.00010606216191436693,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.7486424028873444,
+      "num_tokens": 118410419.0,
+      "step": 51230
+    },
+    {
+      "epoch": 0.46979004309159256,
+      "learning_rate": 0.00010604382506647108,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7476514160633088,
+      "num_tokens": 118433801.0,
+      "step": 51240
+    },
+    {
+      "epoch": 0.4698817273310718,
+      "learning_rate": 0.00010602548821857523,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7491062879562378,
+      "num_tokens": 118456782.0,
+      "step": 51250
+    },
+    {
+      "epoch": 0.469973411570551,
+      "learning_rate": 0.00010600715137067938,
+      "loss": 0.8196,
+      "mean_token_accuracy": 0.7526462316513062,
+      "num_tokens": 118479643.0,
+      "step": 51260
+    },
+    {
+      "epoch": 0.47006509581003025,
+      "learning_rate": 0.00010598881452278353,
+      "loss": 0.8924,
+      "mean_token_accuracy": 0.7386376023292541,
+      "num_tokens": 118502213.0,
+      "step": 51270
+    },
+    {
+      "epoch": 0.4701567800495095,
+      "learning_rate": 0.0001059704776748877,
+      "loss": 0.799,
+      "mean_token_accuracy": 0.7570102334022522,
+      "num_tokens": 118525687.0,
+      "step": 51280
+    },
+    {
+      "epoch": 0.4702484642889887,
+      "learning_rate": 0.00010595214082699185,
+      "loss": 0.8706,
+      "mean_token_accuracy": 0.7403916001319886,
+      "num_tokens": 118549338.0,
+      "step": 51290
+    },
+    {
+      "epoch": 0.47034014852846795,
+      "learning_rate": 0.000105933803979096,
+      "loss": 0.8349,
+      "mean_token_accuracy": 0.7466923415660858,
+      "num_tokens": 118572386.0,
+      "step": 51300
+    },
+    {
+      "epoch": 0.4704318327679472,
+      "learning_rate": 0.00010591546713120015,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.7625176906585693,
+      "num_tokens": 118595336.0,
+      "step": 51310
+    },
+    {
+      "epoch": 0.47052351700742645,
+      "learning_rate": 0.0001058971302833043,
+      "loss": 0.8196,
+      "mean_token_accuracy": 0.7529705107212067,
+      "num_tokens": 118619430.0,
+      "step": 51320
+    },
+    {
+      "epoch": 0.47061520124690565,
+      "learning_rate": 0.00010587879343540846,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.7511988520622254,
+      "num_tokens": 118642805.0,
+      "step": 51330
+    },
+    {
+      "epoch": 0.4707068854863849,
+      "learning_rate": 0.0001058604565875126,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.7438337802886963,
+      "num_tokens": 118665617.0,
+      "step": 51340
+    },
+    {
+      "epoch": 0.47079856972586415,
+      "learning_rate": 0.00010584211973961678,
+      "loss": 0.851,
+      "mean_token_accuracy": 0.7440335154533386,
+      "num_tokens": 118688621.0,
+      "step": 51350
+    },
+    {
+      "epoch": 0.47089025396534334,
+      "learning_rate": 0.00010582378289172093,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.7429041504859925,
+      "num_tokens": 118711355.0,
+      "step": 51360
+    },
+    {
+      "epoch": 0.4709819382048226,
+      "learning_rate": 0.00010580544604382508,
+      "loss": 0.8499,
+      "mean_token_accuracy": 0.7501009702682495,
+      "num_tokens": 118733886.0,
+      "step": 51370
+    },
+    {
+      "epoch": 0.47107362244430184,
+      "learning_rate": 0.00010578710919592922,
+      "loss": 0.817,
+      "mean_token_accuracy": 0.7575704097747803,
+      "num_tokens": 118756994.0,
+      "step": 51380
+    },
+    {
+      "epoch": 0.47116530668378104,
+      "learning_rate": 0.00010576877234803337,
+      "loss": 0.8123,
+      "mean_token_accuracy": 0.7541246056556702,
+      "num_tokens": 118779958.0,
+      "step": 51390
+    },
+    {
+      "epoch": 0.4712569909232603,
+      "learning_rate": 0.00010575043550013752,
+      "loss": 0.8106,
+      "mean_token_accuracy": 0.7564858138561249,
+      "num_tokens": 118803401.0,
+      "step": 51400
+    },
+    {
+      "epoch": 0.47134867516273954,
+      "learning_rate": 0.0001057320986522417,
+      "loss": 0.8505,
+      "mean_token_accuracy": 0.7455509245395661,
+      "num_tokens": 118825733.0,
+      "step": 51410
+    },
+    {
+      "epoch": 0.47144035940221873,
+      "learning_rate": 0.00010571376180434585,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.7454495429992676,
+      "num_tokens": 118848401.0,
+      "step": 51420
+    },
+    {
+      "epoch": 0.471532043641698,
+      "learning_rate": 0.00010569542495644999,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.738635241985321,
+      "num_tokens": 118871766.0,
+      "step": 51430
+    },
+    {
+      "epoch": 0.47162372788117723,
+      "learning_rate": 0.00010567708810855414,
+      "loss": 0.8149,
+      "mean_token_accuracy": 0.7502297759056091,
+      "num_tokens": 118895310.0,
+      "step": 51440
+    },
+    {
+      "epoch": 0.4717154121206565,
+      "learning_rate": 0.0001056587512606583,
+      "loss": 0.8037,
+      "mean_token_accuracy": 0.7565677106380463,
+      "num_tokens": 118918312.0,
+      "step": 51450
+    },
+    {
+      "epoch": 0.4718070963601357,
+      "learning_rate": 0.00010564041441276245,
+      "loss": 0.8372,
+      "mean_token_accuracy": 0.7516836285591125,
+      "num_tokens": 118940763.0,
+      "step": 51460
+    },
+    {
+      "epoch": 0.47189878059961493,
+      "learning_rate": 0.0001056220775648666,
+      "loss": 0.8483,
+      "mean_token_accuracy": 0.7425761103630066,
+      "num_tokens": 118963901.0,
+      "step": 51470
+    },
+    {
+      "epoch": 0.4719904648390942,
+      "learning_rate": 0.00010560374071697076,
+      "loss": 0.8102,
+      "mean_token_accuracy": 0.7571872532367706,
+      "num_tokens": 118987206.0,
+      "step": 51480
+    },
+    {
+      "epoch": 0.4720821490785734,
+      "learning_rate": 0.00010558540386907491,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7401663601398468,
+      "num_tokens": 119010643.0,
+      "step": 51490
+    },
+    {
+      "epoch": 0.4721738333180526,
+      "learning_rate": 0.00010556706702117907,
+      "loss": 0.8282,
+      "mean_token_accuracy": 0.7522598505020142,
+      "num_tokens": 119034049.0,
+      "step": 51500
+    },
+    {
+      "epoch": 0.4722655175575319,
+      "learning_rate": 0.00010554873017328322,
+      "loss": 0.81,
+      "mean_token_accuracy": 0.7589615702629089,
+      "num_tokens": 119056837.0,
+      "step": 51510
+    },
+    {
+      "epoch": 0.4723572017970111,
+      "learning_rate": 0.00010553039332538737,
+      "loss": 0.8775,
+      "mean_token_accuracy": 0.7378575384616852,
+      "num_tokens": 119079421.0,
+      "step": 51520
+    },
+    {
+      "epoch": 0.4724488860364903,
+      "learning_rate": 0.00010551205647749152,
+      "loss": 0.8418,
+      "mean_token_accuracy": 0.7503543257713318,
+      "num_tokens": 119102716.0,
+      "step": 51530
+    },
+    {
+      "epoch": 0.4725405702759696,
+      "learning_rate": 0.00010549371962959566,
+      "loss": 0.8419,
+      "mean_token_accuracy": 0.7465055823326111,
+      "num_tokens": 119125546.0,
+      "step": 51540
+    },
+    {
+      "epoch": 0.47263225451544877,
+      "learning_rate": 0.00010547538278169984,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7422538876533509,
+      "num_tokens": 119147921.0,
+      "step": 51550
+    },
+    {
+      "epoch": 0.472723938754928,
+      "learning_rate": 0.00010545704593380399,
+      "loss": 0.8145,
+      "mean_token_accuracy": 0.7520704388618469,
+      "num_tokens": 119171297.0,
+      "step": 51560
+    },
+    {
+      "epoch": 0.47281562299440727,
+      "learning_rate": 0.00010543870908590814,
+      "loss": 0.8284,
+      "mean_token_accuracy": 0.7501891314983368,
+      "num_tokens": 119194705.0,
+      "step": 51570
+    },
+    {
+      "epoch": 0.4729073072338865,
+      "learning_rate": 0.00010542037223801228,
+      "loss": 0.8495,
+      "mean_token_accuracy": 0.7503993451595307,
+      "num_tokens": 119218328.0,
+      "step": 51580
+    },
+    {
+      "epoch": 0.4729989914733657,
+      "learning_rate": 0.00010540203539011643,
+      "loss": 0.8372,
+      "mean_token_accuracy": 0.7520356714725495,
+      "num_tokens": 119241496.0,
+      "step": 51590
+    },
+    {
+      "epoch": 0.47309067571284497,
+      "learning_rate": 0.00010538369854222059,
+      "loss": 0.8557,
+      "mean_token_accuracy": 0.74845569729805,
+      "num_tokens": 119264681.0,
+      "step": 51600
+    },
+    {
+      "epoch": 0.4731823599523242,
+      "learning_rate": 0.00010536536169432476,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7576924562454224,
+      "num_tokens": 119288006.0,
+      "step": 51610
+    },
+    {
+      "epoch": 0.4732740441918034,
+      "learning_rate": 0.00010534702484642892,
+      "loss": 0.808,
+      "mean_token_accuracy": 0.7529437422752381,
+      "num_tokens": 119310643.0,
+      "step": 51620
+    },
+    {
+      "epoch": 0.47336572843128266,
+      "learning_rate": 0.00010532868799853305,
+      "loss": 0.8412,
+      "mean_token_accuracy": 0.7463583409786224,
+      "num_tokens": 119333862.0,
+      "step": 51630
+    },
+    {
+      "epoch": 0.4734574126707619,
+      "learning_rate": 0.0001053103511506372,
+      "loss": 0.8165,
+      "mean_token_accuracy": 0.7486286044120789,
+      "num_tokens": 119357247.0,
+      "step": 51640
+    },
+    {
+      "epoch": 0.4735490969102411,
+      "learning_rate": 0.00010529201430274136,
+      "loss": 0.8608,
+      "mean_token_accuracy": 0.7437616527080536,
+      "num_tokens": 119380397.0,
+      "step": 51650
+    },
+    {
+      "epoch": 0.47364078114972036,
+      "learning_rate": 0.00010527367745484551,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.752439022064209,
+      "num_tokens": 119403250.0,
+      "step": 51660
+    },
+    {
+      "epoch": 0.4737324653891996,
+      "learning_rate": 0.00010525534060694966,
+      "loss": 0.8784,
+      "mean_token_accuracy": 0.7407907843589783,
+      "num_tokens": 119426124.0,
+      "step": 51670
+    },
+    {
+      "epoch": 0.4738241496286788,
+      "learning_rate": 0.00010523700375905383,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7457657277584075,
+      "num_tokens": 119448976.0,
+      "step": 51680
+    },
+    {
+      "epoch": 0.47391583386815805,
+      "learning_rate": 0.00010521866691115798,
+      "loss": 0.8079,
+      "mean_token_accuracy": 0.7516883432865142,
+      "num_tokens": 119472277.0,
+      "step": 51690
+    },
+    {
+      "epoch": 0.4740075181076373,
+      "learning_rate": 0.00010520033006326213,
+      "loss": 0.8496,
+      "mean_token_accuracy": 0.7472087979316712,
+      "num_tokens": 119496364.0,
+      "step": 51700
+    },
+    {
+      "epoch": 0.47409920234711656,
+      "learning_rate": 0.00010518199321536628,
+      "loss": 0.7881,
+      "mean_token_accuracy": 0.761373096704483,
+      "num_tokens": 119519462.0,
+      "step": 51710
+    },
+    {
+      "epoch": 0.47419088658659575,
+      "learning_rate": 0.00010516365636747044,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.7540312886238099,
+      "num_tokens": 119543215.0,
+      "step": 51720
+    },
+    {
+      "epoch": 0.474282570826075,
+      "learning_rate": 0.00010514531951957459,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7419944047927857,
+      "num_tokens": 119565829.0,
+      "step": 51730
+    },
+    {
+      "epoch": 0.47437425506555425,
+      "learning_rate": 0.00010512698267167875,
+      "loss": 0.8483,
+      "mean_token_accuracy": 0.7489297389984131,
+      "num_tokens": 119588959.0,
+      "step": 51740
+    },
+    {
+      "epoch": 0.47446593930503345,
+      "learning_rate": 0.0001051086458237829,
+      "loss": 0.8264,
+      "mean_token_accuracy": 0.7514209091663361,
+      "num_tokens": 119611914.0,
+      "step": 51750
+    },
+    {
+      "epoch": 0.4745576235445127,
+      "learning_rate": 0.00010509030897588706,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.745944368839264,
+      "num_tokens": 119635176.0,
+      "step": 51760
+    },
+    {
+      "epoch": 0.47464930778399195,
+      "learning_rate": 0.00010507197212799121,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7430785715579986,
+      "num_tokens": 119657685.0,
+      "step": 51770
+    },
+    {
+      "epoch": 0.47474099202347114,
+      "learning_rate": 0.00010505363528009535,
+      "loss": 0.8348,
+      "mean_token_accuracy": 0.7516210675239563,
+      "num_tokens": 119680720.0,
+      "step": 51780
+    },
+    {
+      "epoch": 0.4748326762629504,
+      "learning_rate": 0.0001050352984321995,
+      "loss": 0.8301,
+      "mean_token_accuracy": 0.7496432542800904,
+      "num_tokens": 119703871.0,
+      "step": 51790
+    },
+    {
+      "epoch": 0.47492436050242964,
+      "learning_rate": 0.00010501696158430365,
+      "loss": 0.8379,
+      "mean_token_accuracy": 0.7499487519264221,
+      "num_tokens": 119726436.0,
+      "step": 51800
+    },
+    {
+      "epoch": 0.47501604474190884,
+      "learning_rate": 0.00010499862473640783,
+      "loss": 0.8547,
+      "mean_token_accuracy": 0.7385568976402282,
+      "num_tokens": 119749559.0,
+      "step": 51810
+    },
+    {
+      "epoch": 0.4751077289813881,
+      "learning_rate": 0.00010498028788851198,
+      "loss": 0.8496,
+      "mean_token_accuracy": 0.7408250153064728,
+      "num_tokens": 119772661.0,
+      "step": 51820
+    },
+    {
+      "epoch": 0.47519941322086734,
+      "learning_rate": 0.00010496195104061612,
+      "loss": 0.8455,
+      "mean_token_accuracy": 0.7459851145744324,
+      "num_tokens": 119795952.0,
+      "step": 51830
+    },
+    {
+      "epoch": 0.4752910974603466,
+      "learning_rate": 0.00010494361419272027,
+      "loss": 0.8395,
+      "mean_token_accuracy": 0.7483327984809875,
+      "num_tokens": 119819648.0,
+      "step": 51840
+    },
+    {
+      "epoch": 0.4753827816998258,
+      "learning_rate": 0.00010492527734482442,
+      "loss": 0.8633,
+      "mean_token_accuracy": 0.7430233657360077,
+      "num_tokens": 119843115.0,
+      "step": 51850
+    },
+    {
+      "epoch": 0.47547446593930504,
+      "learning_rate": 0.00010490694049692857,
+      "loss": 0.8822,
+      "mean_token_accuracy": 0.7394209146499634,
+      "num_tokens": 119866523.0,
+      "step": 51860
+    },
+    {
+      "epoch": 0.4755661501787843,
+      "learning_rate": 0.00010488860364903274,
+      "loss": 0.8444,
+      "mean_token_accuracy": 0.7485830128192902,
+      "num_tokens": 119890122.0,
+      "step": 51870
+    },
+    {
+      "epoch": 0.4756578344182635,
+      "learning_rate": 0.00010487026680113689,
+      "loss": 0.8181,
+      "mean_token_accuracy": 0.7529316782951355,
+      "num_tokens": 119913204.0,
+      "step": 51880
+    },
+    {
+      "epoch": 0.47574951865774273,
+      "learning_rate": 0.00010485192995324104,
+      "loss": 0.8844,
+      "mean_token_accuracy": 0.7386928558349609,
+      "num_tokens": 119937298.0,
+      "step": 51890
+    },
+    {
+      "epoch": 0.475841202897222,
+      "learning_rate": 0.0001048335931053452,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.7530156075954437,
+      "num_tokens": 119960987.0,
+      "step": 51900
+    },
+    {
+      "epoch": 0.4759328871367012,
+      "learning_rate": 0.00010481525625744935,
+      "loss": 0.8356,
+      "mean_token_accuracy": 0.7472600281238556,
+      "num_tokens": 119984511.0,
+      "step": 51910
+    },
+    {
+      "epoch": 0.4760245713761804,
+      "learning_rate": 0.0001047969194095535,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7455800354480744,
+      "num_tokens": 120007621.0,
+      "step": 51920
+    },
+    {
+      "epoch": 0.4761162556156597,
+      "learning_rate": 0.00010477858256165765,
+      "loss": 0.8178,
+      "mean_token_accuracy": 0.7557486116886138,
+      "num_tokens": 120030561.0,
+      "step": 51930
+    },
+    {
+      "epoch": 0.47620793985513893,
+      "learning_rate": 0.00010476024571376182,
+      "loss": 0.86,
+      "mean_token_accuracy": 0.7437947511672973,
+      "num_tokens": 120053788.0,
+      "step": 51940
+    },
+    {
+      "epoch": 0.4762996240946181,
+      "learning_rate": 0.00010474190886586597,
+      "loss": 0.8248,
+      "mean_token_accuracy": 0.7512517213821411,
+      "num_tokens": 120077500.0,
+      "step": 51950
+    },
+    {
+      "epoch": 0.4763913083340974,
+      "learning_rate": 0.00010472357201797012,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7436775326728821,
+      "num_tokens": 120100246.0,
+      "step": 51960
+    },
+    {
+      "epoch": 0.4764829925735766,
+      "learning_rate": 0.00010470523517007427,
+      "loss": 0.8117,
+      "mean_token_accuracy": 0.7532978773117065,
+      "num_tokens": 120123221.0,
+      "step": 51970
+    },
+    {
+      "epoch": 0.4765746768130558,
+      "learning_rate": 0.00010468689832217841,
+      "loss": 0.8155,
+      "mean_token_accuracy": 0.7556072354316712,
+      "num_tokens": 120145683.0,
+      "step": 51980
+    },
+    {
+      "epoch": 0.47666636105253507,
+      "learning_rate": 0.00010466856147428256,
+      "loss": 0.8219,
+      "mean_token_accuracy": 0.7522577106952667,
+      "num_tokens": 120169455.0,
+      "step": 51990
+    },
+    {
+      "epoch": 0.4767580452920143,
+      "learning_rate": 0.00010465022462638674,
+      "loss": 0.8117,
+      "mean_token_accuracy": 0.7569294095039367,
+      "num_tokens": 120192384.0,
+      "step": 52000
+    },
+    {
+      "epoch": 0.4768497295314935,
+      "learning_rate": 0.0001046318877784909,
+      "loss": 0.8605,
+      "mean_token_accuracy": 0.742826646566391,
+      "num_tokens": 120215179.0,
+      "step": 52010
+    },
+    {
+      "epoch": 0.47694141377097277,
+      "learning_rate": 0.00010461355093059505,
+      "loss": 0.8067,
+      "mean_token_accuracy": 0.7528352260589599,
+      "num_tokens": 120239227.0,
+      "step": 52020
+    },
+    {
+      "epoch": 0.477033098010452,
+      "learning_rate": 0.00010459521408269918,
+      "loss": 0.8394,
+      "mean_token_accuracy": 0.7538626313209533,
+      "num_tokens": 120261731.0,
+      "step": 52030
+    },
+    {
+      "epoch": 0.4771247822499312,
+      "learning_rate": 0.00010457687723480334,
+      "loss": 0.7721,
+      "mean_token_accuracy": 0.7627759158611298,
+      "num_tokens": 120285007.0,
+      "step": 52040
+    },
+    {
+      "epoch": 0.47721646648941046,
+      "learning_rate": 0.00010455854038690749,
+      "loss": 0.8051,
+      "mean_token_accuracy": 0.7523185551166535,
+      "num_tokens": 120308883.0,
+      "step": 52050
+    },
+    {
+      "epoch": 0.4773081507288897,
+      "learning_rate": 0.00010454020353901164,
+      "loss": 0.7912,
+      "mean_token_accuracy": 0.7587815761566162,
+      "num_tokens": 120332633.0,
+      "step": 52060
+    },
+    {
+      "epoch": 0.47739983496836896,
+      "learning_rate": 0.0001045218666911158,
+      "loss": 0.8111,
+      "mean_token_accuracy": 0.7487365424633026,
+      "num_tokens": 120355728.0,
+      "step": 52070
+    },
+    {
+      "epoch": 0.47749151920784816,
+      "learning_rate": 0.00010450352984321996,
+      "loss": 0.8223,
+      "mean_token_accuracy": 0.7584529757499695,
+      "num_tokens": 120378645.0,
+      "step": 52080
+    },
+    {
+      "epoch": 0.4775832034473274,
+      "learning_rate": 0.00010448519299532411,
+      "loss": 0.8527,
+      "mean_token_accuracy": 0.7448618650436402,
+      "num_tokens": 120401944.0,
+      "step": 52090
+    },
+    {
+      "epoch": 0.47767488768680666,
+      "learning_rate": 0.00010446685614742826,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7441773533821106,
+      "num_tokens": 120425306.0,
+      "step": 52100
+    },
+    {
+      "epoch": 0.47776657192628585,
+      "learning_rate": 0.00010444851929953241,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.7408665239810943,
+      "num_tokens": 120448261.0,
+      "step": 52110
+    },
+    {
+      "epoch": 0.4778582561657651,
+      "learning_rate": 0.00010443018245163656,
+      "loss": 0.8511,
+      "mean_token_accuracy": 0.7457766473293305,
+      "num_tokens": 120471511.0,
+      "step": 52120
+    },
+    {
+      "epoch": 0.47794994040524436,
+      "learning_rate": 0.00010441184560374073,
+      "loss": 0.8259,
+      "mean_token_accuracy": 0.7492813110351563,
+      "num_tokens": 120493351.0,
+      "step": 52130
+    },
+    {
+      "epoch": 0.47804162464472355,
+      "learning_rate": 0.00010439350875584488,
+      "loss": 0.8023,
+      "mean_token_accuracy": 0.7554517388343811,
+      "num_tokens": 120515568.0,
+      "step": 52140
+    },
+    {
+      "epoch": 0.4781333088842028,
+      "learning_rate": 0.00010437517190794903,
+      "loss": 0.8378,
+      "mean_token_accuracy": 0.7453085303306579,
+      "num_tokens": 120538742.0,
+      "step": 52150
+    },
+    {
+      "epoch": 0.47822499312368205,
+      "learning_rate": 0.00010435683506005318,
+      "loss": 0.8355,
+      "mean_token_accuracy": 0.7500856280326843,
+      "num_tokens": 120561821.0,
+      "step": 52160
+    },
+    {
+      "epoch": 0.47831667736316125,
+      "learning_rate": 0.00010433849821215734,
+      "loss": 0.8482,
+      "mean_token_accuracy": 0.7485159277915955,
+      "num_tokens": 120585029.0,
+      "step": 52170
+    },
+    {
+      "epoch": 0.4784083616026405,
+      "learning_rate": 0.00010432016136426148,
+      "loss": 0.8381,
+      "mean_token_accuracy": 0.747643381357193,
+      "num_tokens": 120607937.0,
+      "step": 52180
+    },
+    {
+      "epoch": 0.47850004584211975,
+      "learning_rate": 0.00010430182451636563,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7510312020778656,
+      "num_tokens": 120629692.0,
+      "step": 52190
+    },
+    {
+      "epoch": 0.478591730081599,
+      "learning_rate": 0.0001042834876684698,
+      "loss": 0.7832,
+      "mean_token_accuracy": 0.7608694970607758,
+      "num_tokens": 120652411.0,
+      "step": 52200
+    },
+    {
+      "epoch": 0.4786834143210782,
+      "learning_rate": 0.00010426515082057396,
+      "loss": 0.8531,
+      "mean_token_accuracy": 0.7451915502548218,
+      "num_tokens": 120676303.0,
+      "step": 52210
+    },
+    {
+      "epoch": 0.47877509856055744,
+      "learning_rate": 0.00010424681397267811,
+      "loss": 0.8264,
+      "mean_token_accuracy": 0.7469135582447052,
+      "num_tokens": 120699746.0,
+      "step": 52220
+    },
+    {
+      "epoch": 0.4788667828000367,
+      "learning_rate": 0.00010422847712478225,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7460567653179169,
+      "num_tokens": 120722778.0,
+      "step": 52230
+    },
+    {
+      "epoch": 0.4789584670395159,
+      "learning_rate": 0.0001042101402768864,
+      "loss": 0.8275,
+      "mean_token_accuracy": 0.7463421106338501,
+      "num_tokens": 120746157.0,
+      "step": 52240
+    },
+    {
+      "epoch": 0.47905015127899514,
+      "learning_rate": 0.00010419180342899055,
+      "loss": 0.8263,
+      "mean_token_accuracy": 0.7499868214130402,
+      "num_tokens": 120769751.0,
+      "step": 52250
+    },
+    {
+      "epoch": 0.4791418355184744,
+      "learning_rate": 0.00010417346658109473,
+      "loss": 0.806,
+      "mean_token_accuracy": 0.7564871966838836,
+      "num_tokens": 120793248.0,
+      "step": 52260
+    },
+    {
+      "epoch": 0.4792335197579536,
+      "learning_rate": 0.00010415512973319887,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7389548480510711,
+      "num_tokens": 120816385.0,
+      "step": 52270
+    },
+    {
+      "epoch": 0.47932520399743284,
+      "learning_rate": 0.00010413679288530302,
+      "loss": 0.8147,
+      "mean_token_accuracy": 0.7481178760528564,
+      "num_tokens": 120839881.0,
+      "step": 52280
+    },
+    {
+      "epoch": 0.4794168882369121,
+      "learning_rate": 0.00010411845603740717,
+      "loss": 0.8121,
+      "mean_token_accuracy": 0.7531205475330353,
+      "num_tokens": 120862336.0,
+      "step": 52290
+    },
+    {
+      "epoch": 0.4795085724763913,
+      "learning_rate": 0.00010410011918951132,
+      "loss": 0.8156,
+      "mean_token_accuracy": 0.7549703240394592,
+      "num_tokens": 120886121.0,
+      "step": 52300
+    },
+    {
+      "epoch": 0.47960025671587053,
+      "learning_rate": 0.00010408178234161548,
+      "loss": 0.815,
+      "mean_token_accuracy": 0.7477683067321778,
+      "num_tokens": 120908940.0,
+      "step": 52310
+    },
+    {
+      "epoch": 0.4796919409553498,
+      "learning_rate": 0.00010406344549371963,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7484028577804566,
+      "num_tokens": 120932037.0,
+      "step": 52320
+    },
+    {
+      "epoch": 0.47978362519482903,
+      "learning_rate": 0.0001040451086458238,
+      "loss": 0.8617,
+      "mean_token_accuracy": 0.7395271182060241,
+      "num_tokens": 120954654.0,
+      "step": 52330
+    },
+    {
+      "epoch": 0.47987530943430823,
+      "learning_rate": 0.00010402677179792795,
+      "loss": 0.8259,
+      "mean_token_accuracy": 0.7498848557472229,
+      "num_tokens": 120978417.0,
+      "step": 52340
+    },
+    {
+      "epoch": 0.4799669936737875,
+      "learning_rate": 0.0001040084349500321,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7381206154823303,
+      "num_tokens": 121001792.0,
+      "step": 52350
+    },
+    {
+      "epoch": 0.48005867791326673,
+      "learning_rate": 0.00010399009810213625,
+      "loss": 0.7991,
+      "mean_token_accuracy": 0.7535934984683991,
+      "num_tokens": 121024543.0,
+      "step": 52360
+    },
+    {
+      "epoch": 0.4801503621527459,
+      "learning_rate": 0.0001039717612542404,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7542675793170929,
+      "num_tokens": 121048134.0,
+      "step": 52370
+    },
+    {
+      "epoch": 0.4802420463922252,
+      "learning_rate": 0.00010395342440634454,
+      "loss": 0.8424,
+      "mean_token_accuracy": 0.7429106175899506,
+      "num_tokens": 121071475.0,
+      "step": 52380
+    },
+    {
+      "epoch": 0.4803337306317044,
+      "learning_rate": 0.00010393508755844872,
+      "loss": 0.796,
+      "mean_token_accuracy": 0.7548368990421295,
+      "num_tokens": 121094800.0,
+      "step": 52390
+    },
+    {
+      "epoch": 0.4804254148711836,
+      "learning_rate": 0.00010391675071055287,
+      "loss": 0.8326,
+      "mean_token_accuracy": 0.7476666033267975,
+      "num_tokens": 121116980.0,
+      "step": 52400
+    },
+    {
+      "epoch": 0.48051709911066287,
+      "learning_rate": 0.00010389841386265702,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7479492425918579,
+      "num_tokens": 121139950.0,
+      "step": 52410
+    },
+    {
+      "epoch": 0.4806087833501421,
+      "learning_rate": 0.00010388007701476117,
+      "loss": 0.8385,
+      "mean_token_accuracy": 0.7493346989154815,
+      "num_tokens": 121163291.0,
+      "step": 52420
+    },
+    {
+      "epoch": 0.4807004675896213,
+      "learning_rate": 0.00010386174016686531,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.7546171486377716,
+      "num_tokens": 121185902.0,
+      "step": 52430
+    },
+    {
+      "epoch": 0.48079215182910057,
+      "learning_rate": 0.00010384340331896946,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.743706864118576,
+      "num_tokens": 121208699.0,
+      "step": 52440
+    },
+    {
+      "epoch": 0.4808838360685798,
+      "learning_rate": 0.00010382506647107362,
+      "loss": 0.8489,
+      "mean_token_accuracy": 0.7450501918792725,
+      "num_tokens": 121231698.0,
+      "step": 52450
+    },
+    {
+      "epoch": 0.48097552030805907,
+      "learning_rate": 0.0001038067296231778,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7432117164134979,
+      "num_tokens": 121256217.0,
+      "step": 52460
+    },
+    {
+      "epoch": 0.48106720454753826,
+      "learning_rate": 0.00010378839277528193,
+      "loss": 0.8337,
+      "mean_token_accuracy": 0.743570214509964,
+      "num_tokens": 121278164.0,
+      "step": 52470
+    },
+    {
+      "epoch": 0.4811588887870175,
+      "learning_rate": 0.00010377005592738609,
+      "loss": 0.8142,
+      "mean_token_accuracy": 0.7559535920619964,
+      "num_tokens": 121301395.0,
+      "step": 52480
+    },
+    {
+      "epoch": 0.48125057302649676,
+      "learning_rate": 0.00010375171907949024,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.751169091463089,
+      "num_tokens": 121324729.0,
+      "step": 52490
+    },
+    {
+      "epoch": 0.48134225726597596,
+      "learning_rate": 0.00010373338223159439,
+      "loss": 0.8318,
+      "mean_token_accuracy": 0.7524395287036896,
+      "num_tokens": 121347953.0,
+      "step": 52500
+    },
+    {
+      "epoch": 0.4814339415054552,
+      "learning_rate": 0.00010371504538369854,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.7450460493564606,
+      "num_tokens": 121371802.0,
+      "step": 52510
+    },
+    {
+      "epoch": 0.48152562574493446,
+      "learning_rate": 0.0001036967085358027,
+      "loss": 0.8391,
+      "mean_token_accuracy": 0.7463997006416321,
+      "num_tokens": 121394010.0,
+      "step": 52520
+    },
+    {
+      "epoch": 0.48161730998441366,
+      "learning_rate": 0.00010367837168790686,
+      "loss": 0.8553,
+      "mean_token_accuracy": 0.743210768699646,
+      "num_tokens": 121416354.0,
+      "step": 52530
+    },
+    {
+      "epoch": 0.4817089942238929,
+      "learning_rate": 0.00010366003484001101,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.7479791879653931,
+      "num_tokens": 121439253.0,
+      "step": 52540
+    },
+    {
+      "epoch": 0.48180067846337216,
+      "learning_rate": 0.00010364169799211516,
+      "loss": 0.8827,
+      "mean_token_accuracy": 0.7378830492496491,
+      "num_tokens": 121462252.0,
+      "step": 52550
+    },
+    {
+      "epoch": 0.4818923627028514,
+      "learning_rate": 0.00010362336114421931,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7453905045986176,
+      "num_tokens": 121484757.0,
+      "step": 52560
+    },
+    {
+      "epoch": 0.4819840469423306,
+      "learning_rate": 0.00010360502429632347,
+      "loss": 0.8349,
+      "mean_token_accuracy": 0.7503317594528198,
+      "num_tokens": 121508367.0,
+      "step": 52570
+    },
+    {
+      "epoch": 0.48207573118180985,
+      "learning_rate": 0.0001035866874484276,
+      "loss": 0.8188,
+      "mean_token_accuracy": 0.7495253026485443,
+      "num_tokens": 121531651.0,
+      "step": 52580
+    },
+    {
+      "epoch": 0.4821674154212891,
+      "learning_rate": 0.00010356835060053178,
+      "loss": 0.8608,
+      "mean_token_accuracy": 0.7381142377853394,
+      "num_tokens": 121554517.0,
+      "step": 52590
+    },
+    {
+      "epoch": 0.4822590996607683,
+      "learning_rate": 0.00010355001375263593,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.7543053805828095,
+      "num_tokens": 121576594.0,
+      "step": 52600
+    },
+    {
+      "epoch": 0.48235078390024755,
+      "learning_rate": 0.00010353167690474009,
+      "loss": 0.8567,
+      "mean_token_accuracy": 0.7421742618083954,
+      "num_tokens": 121599915.0,
+      "step": 52610
+    },
+    {
+      "epoch": 0.4824424681397268,
+      "learning_rate": 0.00010351334005684424,
+      "loss": 0.8347,
+      "mean_token_accuracy": 0.7479613125324249,
+      "num_tokens": 121623606.0,
+      "step": 52620
+    },
+    {
+      "epoch": 0.482534152379206,
+      "learning_rate": 0.00010349500320894838,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7455990254878998,
+      "num_tokens": 121647035.0,
+      "step": 52630
+    },
+    {
+      "epoch": 0.48262583661868524,
+      "learning_rate": 0.00010347666636105253,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7386673331260681,
+      "num_tokens": 121670147.0,
+      "step": 52640
+    },
+    {
+      "epoch": 0.4827175208581645,
+      "learning_rate": 0.00010345832951315671,
+      "loss": 0.8354,
+      "mean_token_accuracy": 0.7507193684577942,
+      "num_tokens": 121693605.0,
+      "step": 52650
+    },
+    {
+      "epoch": 0.4828092050976437,
+      "learning_rate": 0.00010343999266526086,
+      "loss": 0.8244,
+      "mean_token_accuracy": 0.7527222037315369,
+      "num_tokens": 121717183.0,
+      "step": 52660
+    },
+    {
+      "epoch": 0.48290088933712294,
+      "learning_rate": 0.000103421655817365,
+      "loss": 0.8295,
+      "mean_token_accuracy": 0.7455935955047608,
+      "num_tokens": 121739687.0,
+      "step": 52670
+    },
+    {
+      "epoch": 0.4829925735766022,
+      "learning_rate": 0.00010340331896946915,
+      "loss": 0.8202,
+      "mean_token_accuracy": 0.7583676457405091,
+      "num_tokens": 121761628.0,
+      "step": 52680
+    },
+    {
+      "epoch": 0.48308425781608144,
+      "learning_rate": 0.0001033849821215733,
+      "loss": 0.8408,
+      "mean_token_accuracy": 0.746461832523346,
+      "num_tokens": 121783805.0,
+      "step": 52690
+    },
+    {
+      "epoch": 0.48317594205556064,
+      "learning_rate": 0.00010336664527367745,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.7479347705841064,
+      "num_tokens": 121808213.0,
+      "step": 52700
+    },
+    {
+      "epoch": 0.4832676262950399,
+      "learning_rate": 0.0001033483084257816,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.7482814371585846,
+      "num_tokens": 121830679.0,
+      "step": 52710
+    },
+    {
+      "epoch": 0.48335931053451914,
+      "learning_rate": 0.00010332997157788577,
+      "loss": 0.8133,
+      "mean_token_accuracy": 0.752324515581131,
+      "num_tokens": 121853673.0,
+      "step": 52720
+    },
+    {
+      "epoch": 0.48345099477399833,
+      "learning_rate": 0.00010331163472998992,
+      "loss": 0.8391,
+      "mean_token_accuracy": 0.7527458786964416,
+      "num_tokens": 121877208.0,
+      "step": 52730
+    },
+    {
+      "epoch": 0.4835426790134776,
+      "learning_rate": 0.00010329329788209407,
+      "loss": 0.8248,
+      "mean_token_accuracy": 0.7487370789051055,
+      "num_tokens": 121900046.0,
+      "step": 52740
+    },
+    {
+      "epoch": 0.48363436325295683,
+      "learning_rate": 0.00010327496103419823,
+      "loss": 0.8196,
+      "mean_token_accuracy": 0.7506524980068207,
+      "num_tokens": 121923400.0,
+      "step": 52750
+    },
+    {
+      "epoch": 0.48372604749243603,
+      "learning_rate": 0.00010325662418630238,
+      "loss": 0.8323,
+      "mean_token_accuracy": 0.7495845139026642,
+      "num_tokens": 121946713.0,
+      "step": 52760
+    },
+    {
+      "epoch": 0.4838177317319153,
+      "learning_rate": 0.00010323828733840653,
+      "loss": 0.8181,
+      "mean_token_accuracy": 0.7542740881443024,
+      "num_tokens": 121970201.0,
+      "step": 52770
+    },
+    {
+      "epoch": 0.48390941597139453,
+      "learning_rate": 0.00010321995049051067,
+      "loss": 0.8181,
+      "mean_token_accuracy": 0.7539512097835541,
+      "num_tokens": 121992894.0,
+      "step": 52780
+    },
+    {
+      "epoch": 0.4840011002108737,
+      "learning_rate": 0.00010320161364261485,
+      "loss": 0.8546,
+      "mean_token_accuracy": 0.7451876163482666,
+      "num_tokens": 122015688.0,
+      "step": 52790
+    },
+    {
+      "epoch": 0.484092784450353,
+      "learning_rate": 0.000103183276794719,
+      "loss": 0.8287,
+      "mean_token_accuracy": 0.7509006202220917,
+      "num_tokens": 122038967.0,
+      "step": 52800
+    },
+    {
+      "epoch": 0.4841844686898322,
+      "learning_rate": 0.00010316493994682315,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7380962252616883,
+      "num_tokens": 122061353.0,
+      "step": 52810
+    },
+    {
+      "epoch": 0.4842761529293115,
+      "learning_rate": 0.0001031466030989273,
+      "loss": 0.8237,
+      "mean_token_accuracy": 0.754196149110794,
+      "num_tokens": 122084477.0,
+      "step": 52820
+    },
+    {
+      "epoch": 0.48436783716879067,
+      "learning_rate": 0.00010312826625103144,
+      "loss": 0.8154,
+      "mean_token_accuracy": 0.753213620185852,
+      "num_tokens": 122107703.0,
+      "step": 52830
+    },
+    {
+      "epoch": 0.4844595214082699,
+      "learning_rate": 0.0001031099294031356,
+      "loss": 0.8276,
+      "mean_token_accuracy": 0.7469814538955688,
+      "num_tokens": 122131131.0,
+      "step": 52840
+    },
+    {
+      "epoch": 0.4845512056477492,
+      "learning_rate": 0.00010309159255523977,
+      "loss": 0.8162,
+      "mean_token_accuracy": 0.7541975140571594,
+      "num_tokens": 122153827.0,
+      "step": 52850
+    },
+    {
+      "epoch": 0.48464288988722837,
+      "learning_rate": 0.00010307325570734392,
+      "loss": 0.8775,
+      "mean_token_accuracy": 0.7420593559741974,
+      "num_tokens": 122176540.0,
+      "step": 52860
+    },
+    {
+      "epoch": 0.4847345741267076,
+      "learning_rate": 0.00010305491885944806,
+      "loss": 0.8406,
+      "mean_token_accuracy": 0.7503218650817871,
+      "num_tokens": 122200077.0,
+      "step": 52870
+    },
+    {
+      "epoch": 0.48482625836618687,
+      "learning_rate": 0.00010303658201155221,
+      "loss": 0.8824,
+      "mean_token_accuracy": 0.7431121945381165,
+      "num_tokens": 122222774.0,
+      "step": 52880
+    },
+    {
+      "epoch": 0.48491794260566606,
+      "learning_rate": 0.00010301824516365637,
+      "loss": 0.8375,
+      "mean_token_accuracy": 0.7453800857067108,
+      "num_tokens": 122245482.0,
+      "step": 52890
+    },
+    {
+      "epoch": 0.4850096268451453,
+      "learning_rate": 0.00010299990831576052,
+      "loss": 0.8432,
+      "mean_token_accuracy": 0.7508934795856476,
+      "num_tokens": 122268827.0,
+      "step": 52900
+    },
+    {
+      "epoch": 0.48510131108462456,
+      "learning_rate": 0.00010298157146786467,
+      "loss": 0.8268,
+      "mean_token_accuracy": 0.7549352049827576,
+      "num_tokens": 122292940.0,
+      "step": 52910
+    },
+    {
+      "epoch": 0.48519299532410376,
+      "learning_rate": 0.00010296323461996884,
+      "loss": 0.7945,
+      "mean_token_accuracy": 0.7612595319747925,
+      "num_tokens": 122315859.0,
+      "step": 52920
+    },
+    {
+      "epoch": 0.485284679563583,
+      "learning_rate": 0.00010294489777207299,
+      "loss": 0.8001,
+      "mean_token_accuracy": 0.7555343866348266,
+      "num_tokens": 122338418.0,
+      "step": 52930
+    },
+    {
+      "epoch": 0.48537636380306226,
+      "learning_rate": 0.00010292656092417714,
+      "loss": 0.818,
+      "mean_token_accuracy": 0.7511711657047272,
+      "num_tokens": 122362571.0,
+      "step": 52940
+    },
+    {
+      "epoch": 0.4854680480425415,
+      "learning_rate": 0.00010290822407628129,
+      "loss": 0.8462,
+      "mean_token_accuracy": 0.7403854608535767,
+      "num_tokens": 122386044.0,
+      "step": 52950
+    },
+    {
+      "epoch": 0.4855597322820207,
+      "learning_rate": 0.00010288988722838544,
+      "loss": 0.8326,
+      "mean_token_accuracy": 0.7513818204402923,
+      "num_tokens": 122409734.0,
+      "step": 52960
+    },
+    {
+      "epoch": 0.48565141652149996,
+      "learning_rate": 0.0001028715503804896,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.748537564277649,
+      "num_tokens": 122432414.0,
+      "step": 52970
+    },
+    {
+      "epoch": 0.4857431007609792,
+      "learning_rate": 0.00010285321353259376,
+      "loss": 0.8434,
+      "mean_token_accuracy": 0.7477980315685272,
+      "num_tokens": 122455862.0,
+      "step": 52980
+    },
+    {
+      "epoch": 0.4858347850004584,
+      "learning_rate": 0.00010283487668469791,
+      "loss": 0.8376,
+      "mean_token_accuracy": 0.7481094300746918,
+      "num_tokens": 122478776.0,
+      "step": 52990
+    },
+    {
+      "epoch": 0.48592646923993765,
+      "learning_rate": 0.00010281653983680206,
+      "loss": 0.7979,
+      "mean_token_accuracy": 0.7605419218540191,
+      "num_tokens": 122502337.0,
+      "step": 53000
+    },
+    {
+      "epoch": 0.4860181534794169,
+      "learning_rate": 0.00010279820298890622,
+      "loss": 0.8217,
+      "mean_token_accuracy": 0.7488216757774353,
+      "num_tokens": 122527045.0,
+      "step": 53010
+    },
+    {
+      "epoch": 0.4861098377188961,
+      "learning_rate": 0.00010277986614101037,
+      "loss": 0.8262,
+      "mean_token_accuracy": 0.7520414590835571,
+      "num_tokens": 122550180.0,
+      "step": 53020
+    },
+    {
+      "epoch": 0.48620152195837535,
+      "learning_rate": 0.0001027615292931145,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7416727066040039,
+      "num_tokens": 122573166.0,
+      "step": 53030
+    },
+    {
+      "epoch": 0.4862932061978546,
+      "learning_rate": 0.00010274319244521866,
+      "loss": 0.838,
+      "mean_token_accuracy": 0.7475881159305573,
+      "num_tokens": 122597039.0,
+      "step": 53040
+    },
+    {
+      "epoch": 0.48638489043733385,
+      "learning_rate": 0.00010272485559732284,
+      "loss": 0.8538,
+      "mean_token_accuracy": 0.7462023854255676,
+      "num_tokens": 122620075.0,
+      "step": 53050
+    },
+    {
+      "epoch": 0.48647657467681304,
+      "learning_rate": 0.00010270651874942699,
+      "loss": 0.8641,
+      "mean_token_accuracy": 0.7412254214286804,
+      "num_tokens": 122643227.0,
+      "step": 53060
+    },
+    {
+      "epoch": 0.4865682589162923,
+      "learning_rate": 0.00010268818190153113,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7480344295501709,
+      "num_tokens": 122666324.0,
+      "step": 53070
+    },
+    {
+      "epoch": 0.48665994315577155,
+      "learning_rate": 0.00010266984505363528,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7418996036052704,
+      "num_tokens": 122690440.0,
+      "step": 53080
+    },
+    {
+      "epoch": 0.48675162739525074,
+      "learning_rate": 0.00010265150820573943,
+      "loss": 0.8128,
+      "mean_token_accuracy": 0.7535100996494293,
+      "num_tokens": 122714101.0,
+      "step": 53090
+    },
+    {
+      "epoch": 0.48684331163473,
+      "learning_rate": 0.00010263317135784358,
+      "loss": 0.8647,
+      "mean_token_accuracy": 0.7431944847106934,
+      "num_tokens": 122737056.0,
+      "step": 53100
+    },
+    {
+      "epoch": 0.48693499587420924,
+      "learning_rate": 0.00010261483450994776,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7404684543609619,
+      "num_tokens": 122759933.0,
+      "step": 53110
+    },
+    {
+      "epoch": 0.48702668011368844,
+      "learning_rate": 0.0001025964976620519,
+      "loss": 0.8289,
+      "mean_token_accuracy": 0.7526221632957458,
+      "num_tokens": 122783701.0,
+      "step": 53120
+    },
+    {
+      "epoch": 0.4871183643531677,
+      "learning_rate": 0.00010257816081415605,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7498015403747559,
+      "num_tokens": 122806347.0,
+      "step": 53130
+    },
+    {
+      "epoch": 0.48721004859264694,
+      "learning_rate": 0.0001025598239662602,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7467487394809723,
+      "num_tokens": 122829849.0,
+      "step": 53140
+    },
+    {
+      "epoch": 0.48730173283212613,
+      "learning_rate": 0.00010254148711836436,
+      "loss": 0.8365,
+      "mean_token_accuracy": 0.7487476229667663,
+      "num_tokens": 122852790.0,
+      "step": 53150
+    },
+    {
+      "epoch": 0.4873934170716054,
+      "learning_rate": 0.00010252315027046851,
+      "loss": 0.8493,
+      "mean_token_accuracy": 0.7416711390018463,
+      "num_tokens": 122876160.0,
+      "step": 53160
+    },
+    {
+      "epoch": 0.48748510131108463,
+      "learning_rate": 0.00010250481342257266,
+      "loss": 0.8168,
+      "mean_token_accuracy": 0.7511836946010589,
+      "num_tokens": 122899198.0,
+      "step": 53170
+    },
+    {
+      "epoch": 0.4875767855505639,
+      "learning_rate": 0.00010248647657467682,
+      "loss": 0.8041,
+      "mean_token_accuracy": 0.7555497586727142,
+      "num_tokens": 122922624.0,
+      "step": 53180
+    },
+    {
+      "epoch": 0.4876684697900431,
+      "learning_rate": 0.00010246813972678098,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.7509292483329773,
+      "num_tokens": 122945964.0,
+      "step": 53190
+    },
+    {
+      "epoch": 0.48776015402952233,
+      "learning_rate": 0.00010244980287888513,
+      "loss": 0.8352,
+      "mean_token_accuracy": 0.7502477407455445,
+      "num_tokens": 122968305.0,
+      "step": 53200
+    },
+    {
+      "epoch": 0.4878518382690016,
+      "learning_rate": 0.00010243146603098928,
+      "loss": 0.8207,
+      "mean_token_accuracy": 0.7479778409004212,
+      "num_tokens": 122991402.0,
+      "step": 53210
+    },
+    {
+      "epoch": 0.4879435225084808,
+      "learning_rate": 0.00010241312918309343,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7482169210910797,
+      "num_tokens": 123013809.0,
+      "step": 53220
+    },
+    {
+      "epoch": 0.48803520674796,
+      "learning_rate": 0.00010239479233519757,
+      "loss": 0.8012,
+      "mean_token_accuracy": 0.7504085719585418,
+      "num_tokens": 123037355.0,
+      "step": 53230
+    },
+    {
+      "epoch": 0.4881268909874393,
+      "learning_rate": 0.00010237645548730175,
+      "loss": 0.8286,
+      "mean_token_accuracy": 0.754689735174179,
+      "num_tokens": 123060883.0,
+      "step": 53240
+    },
+    {
+      "epoch": 0.48821857522691847,
+      "learning_rate": 0.0001023581186394059,
+      "loss": 0.8821,
+      "mean_token_accuracy": 0.7471439480781555,
+      "num_tokens": 123083718.0,
+      "step": 53250
+    },
+    {
+      "epoch": 0.4883102594663977,
+      "learning_rate": 0.00010233978179151005,
+      "loss": 0.8211,
+      "mean_token_accuracy": 0.7488535463809967,
+      "num_tokens": 123108286.0,
+      "step": 53260
+    },
+    {
+      "epoch": 0.488401943705877,
+      "learning_rate": 0.00010232144494361419,
+      "loss": 0.8447,
+      "mean_token_accuracy": 0.7493493914604187,
+      "num_tokens": 123130644.0,
+      "step": 53270
+    },
+    {
+      "epoch": 0.48849362794535617,
+      "learning_rate": 0.00010230310809571834,
+      "loss": 0.8265,
+      "mean_token_accuracy": 0.7495214402675628,
+      "num_tokens": 123153605.0,
+      "step": 53280
+    },
+    {
+      "epoch": 0.4885853121848354,
+      "learning_rate": 0.0001022847712478225,
+      "loss": 0.852,
+      "mean_token_accuracy": 0.7408915996551514,
+      "num_tokens": 123176943.0,
+      "step": 53290
+    },
+    {
+      "epoch": 0.48867699642431467,
+      "learning_rate": 0.00010226643439992665,
+      "loss": 0.8617,
+      "mean_token_accuracy": 0.7433248281478881,
+      "num_tokens": 123200250.0,
+      "step": 53300
+    },
+    {
+      "epoch": 0.4887686806637939,
+      "learning_rate": 0.00010224809755203083,
+      "loss": 0.8116,
+      "mean_token_accuracy": 0.7507937252521515,
+      "num_tokens": 123223883.0,
+      "step": 53310
+    },
+    {
+      "epoch": 0.4888603649032731,
+      "learning_rate": 0.00010222976070413496,
+      "loss": 0.7952,
+      "mean_token_accuracy": 0.7581035494804382,
+      "num_tokens": 123246988.0,
+      "step": 53320
+    },
+    {
+      "epoch": 0.48895204914275237,
+      "learning_rate": 0.00010221142385623912,
+      "loss": 0.8105,
+      "mean_token_accuracy": 0.7543392360210419,
+      "num_tokens": 123269938.0,
+      "step": 53330
+    },
+    {
+      "epoch": 0.4890437333822316,
+      "learning_rate": 0.00010219308700834327,
+      "loss": 0.8474,
+      "mean_token_accuracy": 0.7448423445224762,
+      "num_tokens": 123292687.0,
+      "step": 53340
+    },
+    {
+      "epoch": 0.4891354176217108,
+      "learning_rate": 0.00010217475016044742,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.7385833621025085,
+      "num_tokens": 123315907.0,
+      "step": 53350
+    },
+    {
+      "epoch": 0.48922710186119006,
+      "learning_rate": 0.00010215641331255157,
+      "loss": 0.8205,
+      "mean_token_accuracy": 0.7468352377414703,
+      "num_tokens": 123339611.0,
+      "step": 53360
+    },
+    {
+      "epoch": 0.4893187861006693,
+      "learning_rate": 0.00010213807646465574,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.7495156407356263,
+      "num_tokens": 123362280.0,
+      "step": 53370
+    },
+    {
+      "epoch": 0.4894104703401485,
+      "learning_rate": 0.00010211973961675989,
+      "loss": 0.8107,
+      "mean_token_accuracy": 0.7533836543560029,
+      "num_tokens": 123385607.0,
+      "step": 53380
+    },
+    {
+      "epoch": 0.48950215457962776,
+      "learning_rate": 0.00010210140276886404,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7377766191959381,
+      "num_tokens": 123408402.0,
+      "step": 53390
+    },
+    {
+      "epoch": 0.489593838819107,
+      "learning_rate": 0.00010208306592096819,
+      "loss": 0.8772,
+      "mean_token_accuracy": 0.7404467165470123,
+      "num_tokens": 123430836.0,
+      "step": 53400
+    },
+    {
+      "epoch": 0.4896855230585862,
+      "learning_rate": 0.00010206472907307234,
+      "loss": 0.8328,
+      "mean_token_accuracy": 0.7528629839420319,
+      "num_tokens": 123454242.0,
+      "step": 53410
+    },
+    {
+      "epoch": 0.48977720729806545,
+      "learning_rate": 0.0001020463922251765,
+      "loss": 0.8507,
+      "mean_token_accuracy": 0.745625376701355,
+      "num_tokens": 123476968.0,
+      "step": 53420
+    },
+    {
+      "epoch": 0.4898688915375447,
+      "learning_rate": 0.00010202805537728063,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.7525958836078643,
+      "num_tokens": 123500200.0,
+      "step": 53430
+    },
+    {
+      "epoch": 0.48996057577702395,
+      "learning_rate": 0.00010200971852938481,
+      "loss": 0.8293,
+      "mean_token_accuracy": 0.7544655859470367,
+      "num_tokens": 123522379.0,
+      "step": 53440
+    },
+    {
+      "epoch": 0.49005226001650315,
+      "learning_rate": 0.00010199138168148897,
+      "loss": 0.8846,
+      "mean_token_accuracy": 0.7418582677841187,
+      "num_tokens": 123545390.0,
+      "step": 53450
+    },
+    {
+      "epoch": 0.4901439442559824,
+      "learning_rate": 0.00010197304483359312,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.7502604067325592,
+      "num_tokens": 123568907.0,
+      "step": 53460
+    },
+    {
+      "epoch": 0.49023562849546165,
+      "learning_rate": 0.00010195470798569726,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.7447330892086029,
+      "num_tokens": 123591722.0,
+      "step": 53470
+    },
+    {
+      "epoch": 0.49032731273494085,
+      "learning_rate": 0.00010193637113780141,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.7416792869567871,
+      "num_tokens": 123615097.0,
+      "step": 53480
+    },
+    {
+      "epoch": 0.4904189969744201,
+      "learning_rate": 0.00010191803428990556,
+      "loss": 0.8703,
+      "mean_token_accuracy": 0.7404252707958221,
+      "num_tokens": 123638149.0,
+      "step": 53490
+    },
+    {
+      "epoch": 0.49051068121389935,
+      "learning_rate": 0.00010189969744200974,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7452394545078278,
+      "num_tokens": 123661313.0,
+      "step": 53500
+    },
+    {
+      "epoch": 0.49060236545337854,
+      "learning_rate": 0.00010188136059411389,
+      "loss": 0.8757,
+      "mean_token_accuracy": 0.7413649797439575,
+      "num_tokens": 123683722.0,
+      "step": 53510
+    },
+    {
+      "epoch": 0.4906940496928578,
+      "learning_rate": 0.00010186302374621803,
+      "loss": 0.8553,
+      "mean_token_accuracy": 0.7450942695140839,
+      "num_tokens": 123706555.0,
+      "step": 53520
+    },
+    {
+      "epoch": 0.49078573393233704,
+      "learning_rate": 0.00010184468689832218,
+      "loss": 0.8547,
+      "mean_token_accuracy": 0.7429875314235688,
+      "num_tokens": 123729654.0,
+      "step": 53530
+    },
+    {
+      "epoch": 0.49087741817181624,
+      "learning_rate": 0.00010182635005042633,
+      "loss": 0.8255,
+      "mean_token_accuracy": 0.7533956587314605,
+      "num_tokens": 123752547.0,
+      "step": 53540
+    },
+    {
+      "epoch": 0.4909691024112955,
+      "learning_rate": 0.00010180801320253048,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.7497077524662018,
+      "num_tokens": 123775174.0,
+      "step": 53550
+    },
+    {
+      "epoch": 0.49106078665077474,
+      "learning_rate": 0.00010178967635463464,
+      "loss": 0.8499,
+      "mean_token_accuracy": 0.748982959985733,
+      "num_tokens": 123798753.0,
+      "step": 53560
+    },
+    {
+      "epoch": 0.491152470890254,
+      "learning_rate": 0.0001017713395067388,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7488219320774079,
+      "num_tokens": 123821854.0,
+      "step": 53570
+    },
+    {
+      "epoch": 0.4912441551297332,
+      "learning_rate": 0.00010175300265884295,
+      "loss": 0.8274,
+      "mean_token_accuracy": 0.7435190796852111,
+      "num_tokens": 123845987.0,
+      "step": 53580
+    },
+    {
+      "epoch": 0.49133583936921243,
+      "learning_rate": 0.0001017346658109471,
+      "loss": 0.8145,
+      "mean_token_accuracy": 0.7536229133605957,
+      "num_tokens": 123869171.0,
+      "step": 53590
+    },
+    {
+      "epoch": 0.4914275236086917,
+      "learning_rate": 0.00010171632896305126,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7433777451515198,
+      "num_tokens": 123891472.0,
+      "step": 53600
+    },
+    {
+      "epoch": 0.4915192078481709,
+      "learning_rate": 0.00010169799211515541,
+      "loss": 0.8101,
+      "mean_token_accuracy": 0.7552849471569061,
+      "num_tokens": 123914707.0,
+      "step": 53610
+    },
+    {
+      "epoch": 0.49161089208765013,
+      "learning_rate": 0.00010167965526725956,
+      "loss": 0.8462,
+      "mean_token_accuracy": 0.7565894305706025,
+      "num_tokens": 123938036.0,
+      "step": 53620
+    },
+    {
+      "epoch": 0.4917025763271294,
+      "learning_rate": 0.00010166131841936373,
+      "loss": 0.8168,
+      "mean_token_accuracy": 0.7446473956108093,
+      "num_tokens": 123960850.0,
+      "step": 53630
+    },
+    {
+      "epoch": 0.4917942605666086,
+      "learning_rate": 0.00010164298157146788,
+      "loss": 0.9007,
+      "mean_token_accuracy": 0.7369200766086579,
+      "num_tokens": 123983908.0,
+      "step": 53640
+    },
+    {
+      "epoch": 0.4918859448060878,
+      "learning_rate": 0.00010162464472357203,
+      "loss": 0.8374,
+      "mean_token_accuracy": 0.7470629513263702,
+      "num_tokens": 124006825.0,
+      "step": 53650
+    },
+    {
+      "epoch": 0.4919776290455671,
+      "learning_rate": 0.00010160630787567618,
+      "loss": 0.8299,
+      "mean_token_accuracy": 0.7514701068401337,
+      "num_tokens": 124030247.0,
+      "step": 53660
+    },
+    {
+      "epoch": 0.49206931328504633,
+      "learning_rate": 0.00010158797102778032,
+      "loss": 0.813,
+      "mean_token_accuracy": 0.7538914799690246,
+      "num_tokens": 124053670.0,
+      "step": 53670
+    },
+    {
+      "epoch": 0.4921609975245255,
+      "learning_rate": 0.00010156963417988447,
+      "loss": 0.8503,
+      "mean_token_accuracy": 0.7473560988903045,
+      "num_tokens": 124077408.0,
+      "step": 53680
+    },
+    {
+      "epoch": 0.4922526817640048,
+      "learning_rate": 0.00010155129733198862,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7408625245094299,
+      "num_tokens": 124100296.0,
+      "step": 53690
+    },
+    {
+      "epoch": 0.492344366003484,
+      "learning_rate": 0.0001015329604840928,
+      "loss": 0.8428,
+      "mean_token_accuracy": 0.7433066725730896,
+      "num_tokens": 124124254.0,
+      "step": 53700
+    },
+    {
+      "epoch": 0.4924360502429632,
+      "learning_rate": 0.00010151462363619695,
+      "loss": 0.8383,
+      "mean_token_accuracy": 0.752090334892273,
+      "num_tokens": 124147370.0,
+      "step": 53710
+    },
+    {
+      "epoch": 0.49252773448244247,
+      "learning_rate": 0.00010149628678830109,
+      "loss": 0.8395,
+      "mean_token_accuracy": 0.7413933634757995,
+      "num_tokens": 124170302.0,
+      "step": 53720
+    },
+    {
+      "epoch": 0.4926194187219217,
+      "learning_rate": 0.00010147794994040525,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7450426995754242,
+      "num_tokens": 124192812.0,
+      "step": 53730
+    },
+    {
+      "epoch": 0.4927111029614009,
+      "learning_rate": 0.0001014596130925094,
+      "loss": 0.8206,
+      "mean_token_accuracy": 0.7579657852649688,
+      "num_tokens": 124216287.0,
+      "step": 53740
+    },
+    {
+      "epoch": 0.49280278720088017,
+      "learning_rate": 0.00010144127624461355,
+      "loss": 0.8173,
+      "mean_token_accuracy": 0.7462433695793151,
+      "num_tokens": 124240432.0,
+      "step": 53750
+    },
+    {
+      "epoch": 0.4928944714403594,
+      "learning_rate": 0.00010142293939671771,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7429296374320984,
+      "num_tokens": 124263116.0,
+      "step": 53760
+    },
+    {
+      "epoch": 0.4929861556798386,
+      "learning_rate": 0.00010140460254882187,
+      "loss": 0.8656,
+      "mean_token_accuracy": 0.7424785375595093,
+      "num_tokens": 124285638.0,
+      "step": 53770
+    },
+    {
+      "epoch": 0.49307783991931786,
+      "learning_rate": 0.00010138626570092602,
+      "loss": 0.8563,
+      "mean_token_accuracy": 0.7357562601566314,
+      "num_tokens": 124308847.0,
+      "step": 53780
+    },
+    {
+      "epoch": 0.4931695241587971,
+      "learning_rate": 0.00010136792885303017,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.7508998870849609,
+      "num_tokens": 124331689.0,
+      "step": 53790
+    },
+    {
+      "epoch": 0.49326120839827636,
+      "learning_rate": 0.00010134959200513432,
+      "loss": 0.8947,
+      "mean_token_accuracy": 0.7394472539424897,
+      "num_tokens": 124354817.0,
+      "step": 53800
+    },
+    {
+      "epoch": 0.49335289263775556,
+      "learning_rate": 0.00010133125515723847,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.7496876776218414,
+      "num_tokens": 124379230.0,
+      "step": 53810
+    },
+    {
+      "epoch": 0.4934445768772348,
+      "learning_rate": 0.00010131291830934263,
+      "loss": 0.8401,
+      "mean_token_accuracy": 0.7517239868640899,
+      "num_tokens": 124401885.0,
+      "step": 53820
+    },
+    {
+      "epoch": 0.49353626111671406,
+      "learning_rate": 0.00010129458146144679,
+      "loss": 0.9133,
+      "mean_token_accuracy": 0.7294662594795227,
+      "num_tokens": 124424534.0,
+      "step": 53830
+    },
+    {
+      "epoch": 0.49362794535619325,
+      "learning_rate": 0.00010127624461355094,
+      "loss": 0.8519,
+      "mean_token_accuracy": 0.7492198050022125,
+      "num_tokens": 124448015.0,
+      "step": 53840
+    },
+    {
+      "epoch": 0.4937196295956725,
+      "learning_rate": 0.0001012579077656551,
+      "loss": 0.8727,
+      "mean_token_accuracy": 0.7425671398639679,
+      "num_tokens": 124470790.0,
+      "step": 53850
+    },
+    {
+      "epoch": 0.49381131383515176,
+      "learning_rate": 0.00010123957091775925,
+      "loss": 0.8732,
+      "mean_token_accuracy": 0.7382482171058655,
+      "num_tokens": 124493253.0,
+      "step": 53860
+    },
+    {
+      "epoch": 0.49390299807463095,
+      "learning_rate": 0.00010122123406986338,
+      "loss": 0.845,
+      "mean_token_accuracy": 0.7456029891967774,
+      "num_tokens": 124516663.0,
+      "step": 53870
+    },
+    {
+      "epoch": 0.4939946823141102,
+      "learning_rate": 0.00010120289722196754,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7506387591361999,
+      "num_tokens": 124539985.0,
+      "step": 53880
+    },
+    {
+      "epoch": 0.49408636655358945,
+      "learning_rate": 0.00010118456037407172,
+      "loss": 0.8231,
+      "mean_token_accuracy": 0.7518276274204254,
+      "num_tokens": 124562657.0,
+      "step": 53890
+    },
+    {
+      "epoch": 0.49417805079306865,
+      "learning_rate": 0.00010116622352617587,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.7578824937343598,
+      "num_tokens": 124586201.0,
+      "step": 53900
+    },
+    {
+      "epoch": 0.4942697350325479,
+      "learning_rate": 0.00010114788667828002,
+      "loss": 0.8031,
+      "mean_token_accuracy": 0.7554262220859528,
+      "num_tokens": 124610078.0,
+      "step": 53910
+    },
+    {
+      "epoch": 0.49436141927202715,
+      "learning_rate": 0.00010112954983038416,
+      "loss": 0.8343,
+      "mean_token_accuracy": 0.7464954614639282,
+      "num_tokens": 124633249.0,
+      "step": 53920
+    },
+    {
+      "epoch": 0.4944531035115064,
+      "learning_rate": 0.00010111121298248831,
+      "loss": 0.8002,
+      "mean_token_accuracy": 0.7635982513427735,
+      "num_tokens": 124656943.0,
+      "step": 53930
+    },
+    {
+      "epoch": 0.4945447877509856,
+      "learning_rate": 0.00010109287613459246,
+      "loss": 0.8482,
+      "mean_token_accuracy": 0.7445980846881867,
+      "num_tokens": 124680625.0,
+      "step": 53940
+    },
+    {
+      "epoch": 0.49463647199046484,
+      "learning_rate": 0.00010107453928669661,
+      "loss": 0.815,
+      "mean_token_accuracy": 0.7551271378993988,
+      "num_tokens": 124703669.0,
+      "step": 53950
+    },
+    {
+      "epoch": 0.4947281562299441,
+      "learning_rate": 0.00010105620243880078,
+      "loss": 0.8597,
+      "mean_token_accuracy": 0.7429558396339416,
+      "num_tokens": 124726885.0,
+      "step": 53960
+    },
+    {
+      "epoch": 0.4948198404694233,
+      "learning_rate": 0.00010103786559090493,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7451183259487152,
+      "num_tokens": 124750499.0,
+      "step": 53970
+    },
+    {
+      "epoch": 0.49491152470890254,
+      "learning_rate": 0.00010101952874300908,
+      "loss": 0.8971,
+      "mean_token_accuracy": 0.7334162175655365,
+      "num_tokens": 124773676.0,
+      "step": 53980
+    },
+    {
+      "epoch": 0.4950032089483818,
+      "learning_rate": 0.00010100119189511323,
+      "loss": 0.8321,
+      "mean_token_accuracy": 0.7451906800270081,
+      "num_tokens": 124796337.0,
+      "step": 53990
+    },
+    {
+      "epoch": 0.495094893187861,
+      "learning_rate": 0.00010098285504721739,
+      "loss": 0.7969,
+      "mean_token_accuracy": 0.7585631549358368,
+      "num_tokens": 124819339.0,
+      "step": 54000
+    },
+    {
+      "epoch": 0.49518657742734024,
+      "learning_rate": 0.00010096451819932154,
+      "loss": 0.827,
+      "mean_token_accuracy": 0.7474742352962493,
+      "num_tokens": 124842774.0,
+      "step": 54010
+    },
+    {
+      "epoch": 0.4952782616668195,
+      "learning_rate": 0.00010094618135142569,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7424811363220215,
+      "num_tokens": 124866459.0,
+      "step": 54020
+    },
+    {
+      "epoch": 0.4953699459062987,
+      "learning_rate": 0.00010092784450352986,
+      "loss": 0.8353,
+      "mean_token_accuracy": 0.7505001127719879,
+      "num_tokens": 124889371.0,
+      "step": 54030
+    },
+    {
+      "epoch": 0.49546163014577793,
+      "learning_rate": 0.00010090950765563401,
+      "loss": 0.8579,
+      "mean_token_accuracy": 0.7487920939922332,
+      "num_tokens": 124913372.0,
+      "step": 54040
+    },
+    {
+      "epoch": 0.4955533143852572,
+      "learning_rate": 0.00010089117080773816,
+      "loss": 0.8328,
+      "mean_token_accuracy": 0.7490865528583527,
+      "num_tokens": 124936404.0,
+      "step": 54050
+    },
+    {
+      "epoch": 0.49564499862473643,
+      "learning_rate": 0.00010087283395984231,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7462813854217529,
+      "num_tokens": 124960577.0,
+      "step": 54060
+    },
+    {
+      "epoch": 0.4957366828642156,
+      "learning_rate": 0.00010085449711194645,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7449421465396882,
+      "num_tokens": 124983439.0,
+      "step": 54070
+    },
+    {
+      "epoch": 0.4958283671036949,
+      "learning_rate": 0.0001008361602640506,
+      "loss": 0.8138,
+      "mean_token_accuracy": 0.7517531335353851,
+      "num_tokens": 125006546.0,
+      "step": 54080
+    },
+    {
+      "epoch": 0.49592005134317413,
+      "learning_rate": 0.00010081782341615478,
+      "loss": 0.839,
+      "mean_token_accuracy": 0.7510481238365173,
+      "num_tokens": 125029454.0,
+      "step": 54090
+    },
+    {
+      "epoch": 0.4960117355826533,
+      "learning_rate": 0.00010079948656825893,
+      "loss": 0.8402,
+      "mean_token_accuracy": 0.7541648924350739,
+      "num_tokens": 125051516.0,
+      "step": 54100
+    },
+    {
+      "epoch": 0.4961034198221326,
+      "learning_rate": 0.00010078114972036308,
+      "loss": 0.8322,
+      "mean_token_accuracy": 0.748639988899231,
+      "num_tokens": 125074623.0,
+      "step": 54110
+    },
+    {
+      "epoch": 0.4961951040616118,
+      "learning_rate": 0.00010076281287246722,
+      "loss": 0.8382,
+      "mean_token_accuracy": 0.751268619298935,
+      "num_tokens": 125097413.0,
+      "step": 54120
+    },
+    {
+      "epoch": 0.496286788301091,
+      "learning_rate": 0.00010074447602457137,
+      "loss": 0.8972,
+      "mean_token_accuracy": 0.7367906272411346,
+      "num_tokens": 125120365.0,
+      "step": 54130
+    },
+    {
+      "epoch": 0.49637847254057027,
+      "learning_rate": 0.00010072613917667553,
+      "loss": 0.8189,
+      "mean_token_accuracy": 0.7522039532661438,
+      "num_tokens": 125143300.0,
+      "step": 54140
+    },
+    {
+      "epoch": 0.4964701567800495,
+      "learning_rate": 0.00010070780232877968,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7452004492282868,
+      "num_tokens": 125165646.0,
+      "step": 54150
+    },
+    {
+      "epoch": 0.4965618410195287,
+      "learning_rate": 0.00010068946548088384,
+      "loss": 0.8293,
+      "mean_token_accuracy": 0.7507408201694489,
+      "num_tokens": 125187671.0,
+      "step": 54160
+    },
+    {
+      "epoch": 0.49665352525900797,
+      "learning_rate": 0.000100671128632988,
+      "loss": 0.8436,
+      "mean_token_accuracy": 0.7435282468795776,
+      "num_tokens": 125209892.0,
+      "step": 54170
+    },
+    {
+      "epoch": 0.4967452094984872,
+      "learning_rate": 0.00010065279178509215,
+      "loss": 0.8135,
+      "mean_token_accuracy": 0.754814600944519,
+      "num_tokens": 125232710.0,
+      "step": 54180
+    },
+    {
+      "epoch": 0.49683689373796647,
+      "learning_rate": 0.0001006344549371963,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7491725504398346,
+      "num_tokens": 125255023.0,
+      "step": 54190
+    },
+    {
+      "epoch": 0.49692857797744566,
+      "learning_rate": 0.00010061611808930045,
+      "loss": 0.8191,
+      "mean_token_accuracy": 0.7464327573776245,
+      "num_tokens": 125278048.0,
+      "step": 54200
+    },
+    {
+      "epoch": 0.4970202622169249,
+      "learning_rate": 0.0001005977812414046,
+      "loss": 0.8353,
+      "mean_token_accuracy": 0.7493406713008881,
+      "num_tokens": 125302133.0,
+      "step": 54210
+    },
+    {
+      "epoch": 0.49711194645640416,
+      "learning_rate": 0.00010057944439350877,
+      "loss": 0.8163,
+      "mean_token_accuracy": 0.7528886139392853,
+      "num_tokens": 125325039.0,
+      "step": 54220
+    },
+    {
+      "epoch": 0.49720363069588336,
+      "learning_rate": 0.00010056110754561292,
+      "loss": 0.7919,
+      "mean_token_accuracy": 0.7638591527938843,
+      "num_tokens": 125348243.0,
+      "step": 54230
+    },
+    {
+      "epoch": 0.4972953149353626,
+      "learning_rate": 0.00010054277069771707,
+      "loss": 0.82,
+      "mean_token_accuracy": 0.7532366871833801,
+      "num_tokens": 125370526.0,
+      "step": 54240
+    },
+    {
+      "epoch": 0.49738699917484186,
+      "learning_rate": 0.00010052443384982122,
+      "loss": 0.8248,
+      "mean_token_accuracy": 0.7459402322769165,
+      "num_tokens": 125393716.0,
+      "step": 54250
+    },
+    {
+      "epoch": 0.49747868341432105,
+      "learning_rate": 0.00010050609700192538,
+      "loss": 0.8025,
+      "mean_token_accuracy": 0.759676319360733,
+      "num_tokens": 125416768.0,
+      "step": 54260
+    },
+    {
+      "epoch": 0.4975703676538003,
+      "learning_rate": 0.00010048776015402951,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7408665359020233,
+      "num_tokens": 125439528.0,
+      "step": 54270
+    },
+    {
+      "epoch": 0.49766205189327956,
+      "learning_rate": 0.00010046942330613367,
+      "loss": 0.8084,
+      "mean_token_accuracy": 0.7522037148475647,
+      "num_tokens": 125462834.0,
+      "step": 54280
+    },
+    {
+      "epoch": 0.4977537361327588,
+      "learning_rate": 0.00010045108645823784,
+      "loss": 0.8482,
+      "mean_token_accuracy": 0.7438068509101867,
+      "num_tokens": 125485982.0,
+      "step": 54290
+    },
+    {
+      "epoch": 0.497845420372238,
+      "learning_rate": 0.000100432749610342,
+      "loss": 0.8153,
+      "mean_token_accuracy": 0.7500245273113251,
+      "num_tokens": 125509648.0,
+      "step": 54300
+    },
+    {
+      "epoch": 0.49793710461171725,
+      "learning_rate": 0.00010041441276244615,
+      "loss": 0.8459,
+      "mean_token_accuracy": 0.7480392217636108,
+      "num_tokens": 125532751.0,
+      "step": 54310
+    },
+    {
+      "epoch": 0.4980287888511965,
+      "learning_rate": 0.00010039607591455029,
+      "loss": 0.8357,
+      "mean_token_accuracy": 0.7492863476276398,
+      "num_tokens": 125555742.0,
+      "step": 54320
+    },
+    {
+      "epoch": 0.4981204730906757,
+      "learning_rate": 0.00010037773906665444,
+      "loss": 0.8396,
+      "mean_token_accuracy": 0.7488640606403351,
+      "num_tokens": 125578749.0,
+      "step": 54330
+    },
+    {
+      "epoch": 0.49821215733015495,
+      "learning_rate": 0.00010035940221875859,
+      "loss": 0.8007,
+      "mean_token_accuracy": 0.7627981245517731,
+      "num_tokens": 125601938.0,
+      "step": 54340
+    },
+    {
+      "epoch": 0.4983038415696342,
+      "learning_rate": 0.00010034106537086277,
+      "loss": 0.8868,
+      "mean_token_accuracy": 0.7371641397476196,
+      "num_tokens": 125624452.0,
+      "step": 54350
+    },
+    {
+      "epoch": 0.4983955258091134,
+      "learning_rate": 0.00010032272852296691,
+      "loss": 0.8319,
+      "mean_token_accuracy": 0.7496842861175537,
+      "num_tokens": 125648034.0,
+      "step": 54360
+    },
+    {
+      "epoch": 0.49848721004859264,
+      "learning_rate": 0.00010030439167507106,
+      "loss": 0.8401,
+      "mean_token_accuracy": 0.7501612901687622,
+      "num_tokens": 125672256.0,
+      "step": 54370
+    },
+    {
+      "epoch": 0.4985788942880719,
+      "learning_rate": 0.00010028605482717521,
+      "loss": 0.8522,
+      "mean_token_accuracy": 0.7450436055660248,
+      "num_tokens": 125694924.0,
+      "step": 54380
+    },
+    {
+      "epoch": 0.4986705785275511,
+      "learning_rate": 0.00010026771797927936,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7515852391719818,
+      "num_tokens": 125718388.0,
+      "step": 54390
+    },
+    {
+      "epoch": 0.49876226276703034,
+      "learning_rate": 0.00010024938113138351,
+      "loss": 0.8366,
+      "mean_token_accuracy": 0.7515949666500091,
+      "num_tokens": 125741631.0,
+      "step": 54400
+    },
+    {
+      "epoch": 0.4988539470065096,
+      "learning_rate": 0.00010023104428348767,
+      "loss": 0.8088,
+      "mean_token_accuracy": 0.7510480940341949,
+      "num_tokens": 125764628.0,
+      "step": 54410
+    },
+    {
+      "epoch": 0.49894563124598884,
+      "learning_rate": 0.00010021270743559183,
+      "loss": 0.8109,
+      "mean_token_accuracy": 0.7558558702468872,
+      "num_tokens": 125787253.0,
+      "step": 54420
+    },
+    {
+      "epoch": 0.49903731548546804,
+      "learning_rate": 0.00010019437058769598,
+      "loss": 0.8274,
+      "mean_token_accuracy": 0.7478065371513367,
+      "num_tokens": 125810156.0,
+      "step": 54430
+    },
+    {
+      "epoch": 0.4991289997249473,
+      "learning_rate": 0.00010017603373980014,
+      "loss": 0.8462,
+      "mean_token_accuracy": 0.7445954740047455,
+      "num_tokens": 125833313.0,
+      "step": 54440
+    },
+    {
+      "epoch": 0.49922068396442654,
+      "learning_rate": 0.00010015769689190429,
+      "loss": 0.8254,
+      "mean_token_accuracy": 0.7470640480518341,
+      "num_tokens": 125855964.0,
+      "step": 54450
+    },
+    {
+      "epoch": 0.49931236820390573,
+      "learning_rate": 0.00010013936004400844,
+      "loss": 0.8168,
+      "mean_token_accuracy": 0.755207484960556,
+      "num_tokens": 125878786.0,
+      "step": 54460
+    },
+    {
+      "epoch": 0.499404052443385,
+      "learning_rate": 0.00010012102319611258,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7470451116561889,
+      "num_tokens": 125901422.0,
+      "step": 54470
+    },
+    {
+      "epoch": 0.49949573668286423,
+      "learning_rate": 0.00010010268634821676,
+      "loss": 0.8195,
+      "mean_token_accuracy": 0.7487702548503876,
+      "num_tokens": 125924383.0,
+      "step": 54480
+    },
+    {
+      "epoch": 0.49958742092234343,
+      "learning_rate": 0.00010008434950032091,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7458652555942535,
+      "num_tokens": 125947372.0,
+      "step": 54490
+    },
+    {
+      "epoch": 0.4996791051618227,
+      "learning_rate": 0.00010006601265242506,
+      "loss": 0.808,
+      "mean_token_accuracy": 0.7584410846233368,
+      "num_tokens": 125970240.0,
+      "step": 54500
+    },
+    {
+      "epoch": 0.49977078940130193,
+      "learning_rate": 0.00010004767580452921,
+      "loss": 0.8181,
+      "mean_token_accuracy": 0.7533949971199035,
+      "num_tokens": 125993205.0,
+      "step": 54510
+    },
+    {
+      "epoch": 0.4998624736407811,
+      "learning_rate": 0.00010002933895663335,
+      "loss": 0.8365,
+      "mean_token_accuracy": 0.7492292523384094,
+      "num_tokens": 126016370.0,
+      "step": 54520
+    },
+    {
+      "epoch": 0.4999541578802604,
+      "learning_rate": 0.0001000110021087375,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.743920087814331,
+      "num_tokens": 126040237.0,
+      "step": 54530
+    },
+    {
+      "epoch": 0.5000458421197396,
+      "learning_rate": 9.999266526084167e-05,
+      "loss": 0.8374,
+      "mean_token_accuracy": 0.7522234141826629,
+      "num_tokens": 126063868.0,
+      "step": 54540
+    },
+    {
+      "epoch": 0.5001375263592188,
+      "learning_rate": 9.997432841294582e-05,
+      "loss": 0.9003,
+      "mean_token_accuracy": 0.7352230846881866,
+      "num_tokens": 126086241.0,
+      "step": 54550
+    },
+    {
+      "epoch": 0.5002292105986981,
+      "learning_rate": 9.995599156504997e-05,
+      "loss": 0.8216,
+      "mean_token_accuracy": 0.7472166180610657,
+      "num_tokens": 126108299.0,
+      "step": 54560
+    },
+    {
+      "epoch": 0.5003208948381773,
+      "learning_rate": 9.993765471715412e-05,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.7412724792957306,
+      "num_tokens": 126131054.0,
+      "step": 54570
+    },
+    {
+      "epoch": 0.5004125790776566,
+      "learning_rate": 9.991931786925828e-05,
+      "loss": 0.795,
+      "mean_token_accuracy": 0.7548500001430511,
+      "num_tokens": 126153660.0,
+      "step": 54580
+    },
+    {
+      "epoch": 0.5005042633171358,
+      "learning_rate": 9.990098102136243e-05,
+      "loss": 0.8226,
+      "mean_token_accuracy": 0.7520363807678223,
+      "num_tokens": 126176936.0,
+      "step": 54590
+    },
+    {
+      "epoch": 0.5005959475566151,
+      "learning_rate": 9.988264417346659e-05,
+      "loss": 0.8002,
+      "mean_token_accuracy": 0.7580593407154084,
+      "num_tokens": 126200221.0,
+      "step": 54600
+    },
+    {
+      "epoch": 0.5006876317960942,
+      "learning_rate": 9.986430732557074e-05,
+      "loss": 0.8076,
+      "mean_token_accuracy": 0.7551198363304138,
+      "num_tokens": 126224001.0,
+      "step": 54610
+    },
+    {
+      "epoch": 0.5007793160355735,
+      "learning_rate": 9.984597047767488e-05,
+      "loss": 0.8287,
+      "mean_token_accuracy": 0.7408681035041809,
+      "num_tokens": 126246677.0,
+      "step": 54620
+    },
+    {
+      "epoch": 0.5008710002750527,
+      "learning_rate": 9.982763362977905e-05,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.7422008037567138,
+      "num_tokens": 126269735.0,
+      "step": 54630
+    },
+    {
+      "epoch": 0.500962684514532,
+      "learning_rate": 9.98092967818832e-05,
+      "loss": 0.7771,
+      "mean_token_accuracy": 0.7584976851940155,
+      "num_tokens": 126292808.0,
+      "step": 54640
+    },
+    {
+      "epoch": 0.5010543687540112,
+      "learning_rate": 9.979095993398735e-05,
+      "loss": 0.8238,
+      "mean_token_accuracy": 0.7523404061794281,
+      "num_tokens": 126315748.0,
+      "step": 54650
+    },
+    {
+      "epoch": 0.5011460529934905,
+      "learning_rate": 9.97726230860915e-05,
+      "loss": 0.8015,
+      "mean_token_accuracy": 0.7549440562725067,
+      "num_tokens": 126338980.0,
+      "step": 54660
+    },
+    {
+      "epoch": 0.5012377372329696,
+      "learning_rate": 9.975428623819566e-05,
+      "loss": 0.8345,
+      "mean_token_accuracy": 0.7477507054805755,
+      "num_tokens": 126362205.0,
+      "step": 54670
+    },
+    {
+      "epoch": 0.5013294214724489,
+      "learning_rate": 9.973594939029981e-05,
+      "loss": 0.8341,
+      "mean_token_accuracy": 0.7542234659194946,
+      "num_tokens": 126385467.0,
+      "step": 54680
+    },
+    {
+      "epoch": 0.5014211057119281,
+      "learning_rate": 9.971761254240397e-05,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7496972739696502,
+      "num_tokens": 126408886.0,
+      "step": 54690
+    },
+    {
+      "epoch": 0.5015127899514074,
+      "learning_rate": 9.969927569450813e-05,
+      "loss": 0.8107,
+      "mean_token_accuracy": 0.7551493108272552,
+      "num_tokens": 126431755.0,
+      "step": 54700
+    },
+    {
+      "epoch": 0.5016044741908866,
+      "learning_rate": 9.968093884661228e-05,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7427143335342408,
+      "num_tokens": 126454755.0,
+      "step": 54710
+    },
+    {
+      "epoch": 0.5016961584303659,
+      "learning_rate": 9.966260199871642e-05,
+      "loss": 0.8604,
+      "mean_token_accuracy": 0.739615398645401,
+      "num_tokens": 126477920.0,
+      "step": 54720
+    },
+    {
+      "epoch": 0.5017878426698451,
+      "learning_rate": 9.964426515082058e-05,
+      "loss": 0.8436,
+      "mean_token_accuracy": 0.7475731015205384,
+      "num_tokens": 126501745.0,
+      "step": 54730
+    },
+    {
+      "epoch": 0.5018795269093242,
+      "learning_rate": 9.962592830292473e-05,
+      "loss": 0.8653,
+      "mean_token_accuracy": 0.7458015322685242,
+      "num_tokens": 126524704.0,
+      "step": 54740
+    },
+    {
+      "epoch": 0.5019712111488035,
+      "learning_rate": 9.960759145502888e-05,
+      "loss": 0.9123,
+      "mean_token_accuracy": 0.731933981180191,
+      "num_tokens": 126547338.0,
+      "step": 54750
+    },
+    {
+      "epoch": 0.5020628953882827,
+      "learning_rate": 9.958925460713304e-05,
+      "loss": 0.8152,
+      "mean_token_accuracy": 0.7562912046909332,
+      "num_tokens": 126570696.0,
+      "step": 54760
+    },
+    {
+      "epoch": 0.502154579627762,
+      "learning_rate": 9.957091775923719e-05,
+      "loss": 0.8748,
+      "mean_token_accuracy": 0.7461549878120423,
+      "num_tokens": 126594498.0,
+      "step": 54770
+    },
+    {
+      "epoch": 0.5022462638672412,
+      "learning_rate": 9.955258091134134e-05,
+      "loss": 0.8205,
+      "mean_token_accuracy": 0.7527951240539551,
+      "num_tokens": 126617710.0,
+      "step": 54780
+    },
+    {
+      "epoch": 0.5023379481067205,
+      "learning_rate": 9.95342440634455e-05,
+      "loss": 0.8289,
+      "mean_token_accuracy": 0.7487452089786529,
+      "num_tokens": 126641340.0,
+      "step": 54790
+    },
+    {
+      "epoch": 0.5024296323461996,
+      "learning_rate": 9.951590721554966e-05,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7519077181816101,
+      "num_tokens": 126663634.0,
+      "step": 54800
+    },
+    {
+      "epoch": 0.5025213165856789,
+      "learning_rate": 9.949757036765381e-05,
+      "loss": 0.8628,
+      "mean_token_accuracy": 0.744753384590149,
+      "num_tokens": 126686917.0,
+      "step": 54810
+    },
+    {
+      "epoch": 0.5026130008251581,
+      "learning_rate": 9.947923351975796e-05,
+      "loss": 0.8132,
+      "mean_token_accuracy": 0.7549303352832795,
+      "num_tokens": 126710460.0,
+      "step": 54820
+    },
+    {
+      "epoch": 0.5027046850646374,
+      "learning_rate": 9.946089667186211e-05,
+      "loss": 0.8384,
+      "mean_token_accuracy": 0.7446047544479371,
+      "num_tokens": 126733935.0,
+      "step": 54830
+    },
+    {
+      "epoch": 0.5027963693041166,
+      "learning_rate": 9.944255982396626e-05,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7540520310401917,
+      "num_tokens": 126756564.0,
+      "step": 54840
+    },
+    {
+      "epoch": 0.5028880535435959,
+      "learning_rate": 9.942422297607042e-05,
+      "loss": 0.8569,
+      "mean_token_accuracy": 0.743932569026947,
+      "num_tokens": 126780825.0,
+      "step": 54850
+    },
+    {
+      "epoch": 0.5029797377830751,
+      "learning_rate": 9.940588612817457e-05,
+      "loss": 0.8118,
+      "mean_token_accuracy": 0.7560799717903137,
+      "num_tokens": 126804400.0,
+      "step": 54860
+    },
+    {
+      "epoch": 0.5030714220225543,
+      "learning_rate": 9.938754928027872e-05,
+      "loss": 0.8321,
+      "mean_token_accuracy": 0.7500090837478638,
+      "num_tokens": 126828220.0,
+      "step": 54870
+    },
+    {
+      "epoch": 0.5031631062620335,
+      "learning_rate": 9.936921243238287e-05,
+      "loss": 0.8669,
+      "mean_token_accuracy": 0.7424058318138123,
+      "num_tokens": 126851194.0,
+      "step": 54880
+    },
+    {
+      "epoch": 0.5032547905015128,
+      "learning_rate": 9.935087558448704e-05,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7454552531242371,
+      "num_tokens": 126874381.0,
+      "step": 54890
+    },
+    {
+      "epoch": 0.503346474740992,
+      "learning_rate": 9.933253873659119e-05,
+      "loss": 0.7986,
+      "mean_token_accuracy": 0.7588015139102936,
+      "num_tokens": 126897302.0,
+      "step": 54900
+    },
+    {
+      "epoch": 0.5034381589804713,
+      "learning_rate": 9.931420188869534e-05,
+      "loss": 0.8631,
+      "mean_token_accuracy": 0.7387021481990814,
+      "num_tokens": 126920953.0,
+      "step": 54910
+    },
+    {
+      "epoch": 0.5035298432199505,
+      "learning_rate": 9.92958650407995e-05,
+      "loss": 0.816,
+      "mean_token_accuracy": 0.7493123829364776,
+      "num_tokens": 126943968.0,
+      "step": 54920
+    },
+    {
+      "epoch": 0.5036215274594297,
+      "learning_rate": 9.927752819290365e-05,
+      "loss": 0.8326,
+      "mean_token_accuracy": 0.743408453464508,
+      "num_tokens": 126966237.0,
+      "step": 54930
+    },
+    {
+      "epoch": 0.5037132116989089,
+      "learning_rate": 9.92591913450078e-05,
+      "loss": 0.8511,
+      "mean_token_accuracy": 0.7374984502792359,
+      "num_tokens": 126989481.0,
+      "step": 54940
+    },
+    {
+      "epoch": 0.5038048959383882,
+      "learning_rate": 9.924085449711195e-05,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.7659353673458099,
+      "num_tokens": 127012831.0,
+      "step": 54950
+    },
+    {
+      "epoch": 0.5038965801778674,
+      "learning_rate": 9.92225176492161e-05,
+      "loss": 0.8699,
+      "mean_token_accuracy": 0.7392362534999848,
+      "num_tokens": 127036098.0,
+      "step": 54960
+    },
+    {
+      "epoch": 0.5039882644173467,
+      "learning_rate": 9.920418080132025e-05,
+      "loss": 0.8157,
+      "mean_token_accuracy": 0.7554765939712524,
+      "num_tokens": 127059532.0,
+      "step": 54970
+    },
+    {
+      "epoch": 0.5040799486568259,
+      "learning_rate": 9.91858439534244e-05,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.740082061290741,
+      "num_tokens": 127083105.0,
+      "step": 54980
+    },
+    {
+      "epoch": 0.5041716328963052,
+      "learning_rate": 9.916750710552857e-05,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7439696490764618,
+      "num_tokens": 127106250.0,
+      "step": 54990
+    },
+    {
+      "epoch": 0.5042633171357843,
+      "learning_rate": 9.914917025763272e-05,
+      "loss": 0.8385,
+      "mean_token_accuracy": 0.7487026512622833,
+      "num_tokens": 127130489.0,
+      "step": 55000
+    },
+    {
+      "epoch": 0.5043550013752636,
+      "learning_rate": 9.913083340973687e-05,
+      "loss": 0.8679,
+      "mean_token_accuracy": 0.7465303182601929,
+      "num_tokens": 127153623.0,
+      "step": 55010
+    },
+    {
+      "epoch": 0.5044466856147428,
+      "learning_rate": 9.911249656184103e-05,
+      "loss": 0.8631,
+      "mean_token_accuracy": 0.7382868230342865,
+      "num_tokens": 127176977.0,
+      "step": 55020
+    },
+    {
+      "epoch": 0.5045383698542221,
+      "learning_rate": 9.909415971394518e-05,
+      "loss": 0.8424,
+      "mean_token_accuracy": 0.7415868818759919,
+      "num_tokens": 127200079.0,
+      "step": 55030
+    },
+    {
+      "epoch": 0.5046300540937013,
+      "learning_rate": 9.907582286604933e-05,
+      "loss": 0.8239,
+      "mean_token_accuracy": 0.7546780347824097,
+      "num_tokens": 127224392.0,
+      "step": 55040
+    },
+    {
+      "epoch": 0.5047217383331806,
+      "learning_rate": 9.90574860181535e-05,
+      "loss": 0.8825,
+      "mean_token_accuracy": 0.7388218402862549,
+      "num_tokens": 127247293.0,
+      "step": 55050
+    },
+    {
+      "epoch": 0.5048134225726597,
+      "learning_rate": 9.903914917025763e-05,
+      "loss": 0.8748,
+      "mean_token_accuracy": 0.741418582201004,
+      "num_tokens": 127271290.0,
+      "step": 55060
+    },
+    {
+      "epoch": 0.504905106812139,
+      "learning_rate": 9.902081232236178e-05,
+      "loss": 0.831,
+      "mean_token_accuracy": 0.7555666148662568,
+      "num_tokens": 127294268.0,
+      "step": 55070
+    },
+    {
+      "epoch": 0.5049967910516182,
+      "learning_rate": 9.900247547446594e-05,
+      "loss": 0.8298,
+      "mean_token_accuracy": 0.7488799631595612,
+      "num_tokens": 127317141.0,
+      "step": 55080
+    },
+    {
+      "epoch": 0.5050884752910975,
+      "learning_rate": 9.89841386265701e-05,
+      "loss": 0.8157,
+      "mean_token_accuracy": 0.7507079720497132,
+      "num_tokens": 127339832.0,
+      "step": 55090
+    },
+    {
+      "epoch": 0.5051801595305767,
+      "learning_rate": 9.896580177867425e-05,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.7446078598499298,
+      "num_tokens": 127363129.0,
+      "step": 55100
+    },
+    {
+      "epoch": 0.505271843770056,
+      "learning_rate": 9.89474649307784e-05,
+      "loss": 0.7936,
+      "mean_token_accuracy": 0.7580273449420929,
+      "num_tokens": 127386426.0,
+      "step": 55110
+    },
+    {
+      "epoch": 0.5053635280095352,
+      "learning_rate": 9.892912808288256e-05,
+      "loss": 0.8335,
+      "mean_token_accuracy": 0.7495563566684723,
+      "num_tokens": 127409625.0,
+      "step": 55120
+    },
+    {
+      "epoch": 0.5054552122490144,
+      "learning_rate": 9.891079123498671e-05,
+      "loss": 0.8048,
+      "mean_token_accuracy": 0.7522223949432373,
+      "num_tokens": 127432061.0,
+      "step": 55130
+    },
+    {
+      "epoch": 0.5055468964884936,
+      "learning_rate": 9.889245438709086e-05,
+      "loss": 0.8334,
+      "mean_token_accuracy": 0.744785737991333,
+      "num_tokens": 127455714.0,
+      "step": 55140
+    },
+    {
+      "epoch": 0.5056385807279729,
+      "learning_rate": 9.887411753919503e-05,
+      "loss": 0.8419,
+      "mean_token_accuracy": 0.7535149455070496,
+      "num_tokens": 127478644.0,
+      "step": 55150
+    },
+    {
+      "epoch": 0.5057302649674521,
+      "learning_rate": 9.885578069129917e-05,
+      "loss": 0.8201,
+      "mean_token_accuracy": 0.7557021617889405,
+      "num_tokens": 127501827.0,
+      "step": 55160
+    },
+    {
+      "epoch": 0.5058219492069314,
+      "learning_rate": 9.883744384340332e-05,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7471792221069335,
+      "num_tokens": 127526083.0,
+      "step": 55170
+    },
+    {
+      "epoch": 0.5059136334464106,
+      "learning_rate": 9.881910699550748e-05,
+      "loss": 0.8333,
+      "mean_token_accuracy": 0.7493178725242615,
+      "num_tokens": 127550299.0,
+      "step": 55180
+    },
+    {
+      "epoch": 0.5060053176858897,
+      "learning_rate": 9.880077014761163e-05,
+      "loss": 0.81,
+      "mean_token_accuracy": 0.7530982732772827,
+      "num_tokens": 127573181.0,
+      "step": 55190
+    },
+    {
+      "epoch": 0.506097001925369,
+      "learning_rate": 9.878243329971579e-05,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7522773265838623,
+      "num_tokens": 127596723.0,
+      "step": 55200
+    },
+    {
+      "epoch": 0.5061886861648482,
+      "learning_rate": 9.876409645181994e-05,
+      "loss": 0.8377,
+      "mean_token_accuracy": 0.7426759779453278,
+      "num_tokens": 127620142.0,
+      "step": 55210
+    },
+    {
+      "epoch": 0.5062803704043275,
+      "learning_rate": 9.874575960392409e-05,
+      "loss": 0.8365,
+      "mean_token_accuracy": 0.7444867551326751,
+      "num_tokens": 127643905.0,
+      "step": 55220
+    },
+    {
+      "epoch": 0.5063720546438067,
+      "learning_rate": 9.872742275602824e-05,
+      "loss": 0.8287,
+      "mean_token_accuracy": 0.7522559106349945,
+      "num_tokens": 127666297.0,
+      "step": 55230
+    },
+    {
+      "epoch": 0.506463738883286,
+      "learning_rate": 9.87090859081324e-05,
+      "loss": 0.8852,
+      "mean_token_accuracy": 0.7422611057758332,
+      "num_tokens": 127689578.0,
+      "step": 55240
+    },
+    {
+      "epoch": 0.5065554231227652,
+      "learning_rate": 9.869074906023656e-05,
+      "loss": 0.8448,
+      "mean_token_accuracy": 0.7505880951881408,
+      "num_tokens": 127712282.0,
+      "step": 55250
+    },
+    {
+      "epoch": 0.5066471073622444,
+      "learning_rate": 9.86724122123407e-05,
+      "loss": 0.8387,
+      "mean_token_accuracy": 0.7525406897068023,
+      "num_tokens": 127735198.0,
+      "step": 55260
+    },
+    {
+      "epoch": 0.5067387916017236,
+      "learning_rate": 9.865407536444485e-05,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7460958778858184,
+      "num_tokens": 127758639.0,
+      "step": 55270
+    },
+    {
+      "epoch": 0.5068304758412029,
+      "learning_rate": 9.863573851654901e-05,
+      "loss": 0.8362,
+      "mean_token_accuracy": 0.7530414164066315,
+      "num_tokens": 127781784.0,
+      "step": 55280
+    },
+    {
+      "epoch": 0.5069221600806821,
+      "learning_rate": 9.861740166865317e-05,
+      "loss": 0.8662,
+      "mean_token_accuracy": 0.7414700984954834,
+      "num_tokens": 127804452.0,
+      "step": 55290
+    },
+    {
+      "epoch": 0.5070138443201614,
+      "learning_rate": 9.859906482075732e-05,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7503695964813233,
+      "num_tokens": 127827346.0,
+      "step": 55300
+    },
+    {
+      "epoch": 0.5071055285596406,
+      "learning_rate": 9.858072797286147e-05,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7402788043022156,
+      "num_tokens": 127849687.0,
+      "step": 55310
+    },
+    {
+      "epoch": 0.5071972127991198,
+      "learning_rate": 9.856239112496562e-05,
+      "loss": 0.8159,
+      "mean_token_accuracy": 0.7513509571552277,
+      "num_tokens": 127872183.0,
+      "step": 55320
+    },
+    {
+      "epoch": 0.507288897038599,
+      "learning_rate": 9.854405427706977e-05,
+      "loss": 0.8247,
+      "mean_token_accuracy": 0.756741201877594,
+      "num_tokens": 127895640.0,
+      "step": 55330
+    },
+    {
+      "epoch": 0.5073805812780783,
+      "learning_rate": 9.852571742917393e-05,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.754223370552063,
+      "num_tokens": 127919095.0,
+      "step": 55340
+    },
+    {
+      "epoch": 0.5074722655175575,
+      "learning_rate": 9.850738058127809e-05,
+      "loss": 0.8002,
+      "mean_token_accuracy": 0.7614878475666046,
+      "num_tokens": 127942701.0,
+      "step": 55350
+    },
+    {
+      "epoch": 0.5075639497570368,
+      "learning_rate": 9.848904373338223e-05,
+      "loss": 0.8173,
+      "mean_token_accuracy": 0.749773132801056,
+      "num_tokens": 127965305.0,
+      "step": 55360
+    },
+    {
+      "epoch": 0.507655633996516,
+      "learning_rate": 9.847070688548638e-05,
+      "loss": 0.8757,
+      "mean_token_accuracy": 0.7366073966026306,
+      "num_tokens": 127988287.0,
+      "step": 55370
+    },
+    {
+      "epoch": 0.5077473182359953,
+      "learning_rate": 9.845237003759055e-05,
+      "loss": 0.831,
+      "mean_token_accuracy": 0.7543848633766175,
+      "num_tokens": 128011174.0,
+      "step": 55380
+    },
+    {
+      "epoch": 0.5078390024754744,
+      "learning_rate": 9.84340331896947e-05,
+      "loss": 0.8514,
+      "mean_token_accuracy": 0.7410233438014984,
+      "num_tokens": 128034072.0,
+      "step": 55390
+    },
+    {
+      "epoch": 0.5079306867149537,
+      "learning_rate": 9.841569634179885e-05,
+      "loss": 0.8779,
+      "mean_token_accuracy": 0.7453109204769135,
+      "num_tokens": 128057754.0,
+      "step": 55400
+    },
+    {
+      "epoch": 0.5080223709544329,
+      "learning_rate": 9.8397359493903e-05,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7473021626472474,
+      "num_tokens": 128080279.0,
+      "step": 55410
+    },
+    {
+      "epoch": 0.5081140551939122,
+      "learning_rate": 9.837902264600715e-05,
+      "loss": 0.8355,
+      "mean_token_accuracy": 0.7450722455978394,
+      "num_tokens": 128104021.0,
+      "step": 55420
+    },
+    {
+      "epoch": 0.5082057394333914,
+      "learning_rate": 9.83606857981113e-05,
+      "loss": 0.7832,
+      "mean_token_accuracy": 0.7592859029769897,
+      "num_tokens": 128127786.0,
+      "step": 55430
+    },
+    {
+      "epoch": 0.5082974236728707,
+      "learning_rate": 9.834234895021546e-05,
+      "loss": 0.8529,
+      "mean_token_accuracy": 0.7397254407405853,
+      "num_tokens": 128150101.0,
+      "step": 55440
+    },
+    {
+      "epoch": 0.5083891079123498,
+      "learning_rate": 9.832401210231962e-05,
+      "loss": 0.8281,
+      "mean_token_accuracy": 0.7477978467941284,
+      "num_tokens": 128173294.0,
+      "step": 55450
+    },
+    {
+      "epoch": 0.5084807921518291,
+      "learning_rate": 9.830567525442376e-05,
+      "loss": 0.8327,
+      "mean_token_accuracy": 0.7507203817367554,
+      "num_tokens": 128196072.0,
+      "step": 55460
+    },
+    {
+      "epoch": 0.5085724763913083,
+      "learning_rate": 9.828733840652791e-05,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.7511986076831818,
+      "num_tokens": 128218837.0,
+      "step": 55470
+    },
+    {
+      "epoch": 0.5086641606307876,
+      "learning_rate": 9.826900155863208e-05,
+      "loss": 0.8161,
+      "mean_token_accuracy": 0.7469117105007171,
+      "num_tokens": 128241361.0,
+      "step": 55480
+    },
+    {
+      "epoch": 0.5087558448702668,
+      "learning_rate": 9.825066471073623e-05,
+      "loss": 0.8451,
+      "mean_token_accuracy": 0.7483738541603089,
+      "num_tokens": 128265047.0,
+      "step": 55490
+    },
+    {
+      "epoch": 0.5088475291097461,
+      "learning_rate": 9.823232786284038e-05,
+      "loss": 0.8965,
+      "mean_token_accuracy": 0.7301852464675903,
+      "num_tokens": 128288379.0,
+      "step": 55500
+    },
+    {
+      "epoch": 0.5089392133492253,
+      "learning_rate": 9.821399101494453e-05,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7519036054611206,
+      "num_tokens": 128311210.0,
+      "step": 55510
+    },
+    {
+      "epoch": 0.5090308975887045,
+      "learning_rate": 9.819565416704869e-05,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7539181113243103,
+      "num_tokens": 128333412.0,
+      "step": 55520
+    },
+    {
+      "epoch": 0.5091225818281837,
+      "learning_rate": 9.817731731915284e-05,
+      "loss": 0.824,
+      "mean_token_accuracy": 0.7523548245429993,
+      "num_tokens": 128355933.0,
+      "step": 55530
+    },
+    {
+      "epoch": 0.509214266067663,
+      "learning_rate": 9.8158980471257e-05,
+      "loss": 0.8388,
+      "mean_token_accuracy": 0.749217939376831,
+      "num_tokens": 128379533.0,
+      "step": 55540
+    },
+    {
+      "epoch": 0.5093059503071422,
+      "learning_rate": 9.814064362336116e-05,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7430170595645904,
+      "num_tokens": 128402664.0,
+      "step": 55550
+    },
+    {
+      "epoch": 0.5093976345466215,
+      "learning_rate": 9.81223067754653e-05,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.743842476606369,
+      "num_tokens": 128425739.0,
+      "step": 55560
+    },
+    {
+      "epoch": 0.5094893187861007,
+      "learning_rate": 9.810396992756945e-05,
+      "loss": 0.8464,
+      "mean_token_accuracy": 0.7461042940616608,
+      "num_tokens": 128447730.0,
+      "step": 55570
+    },
+    {
+      "epoch": 0.5095810030255798,
+      "learning_rate": 9.808563307967361e-05,
+      "loss": 0.8227,
+      "mean_token_accuracy": 0.7504740059375763,
+      "num_tokens": 128470146.0,
+      "step": 55580
+    },
+    {
+      "epoch": 0.5096726872650591,
+      "learning_rate": 9.806729623177776e-05,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7417598307132721,
+      "num_tokens": 128493164.0,
+      "step": 55590
+    },
+    {
+      "epoch": 0.5097643715045383,
+      "learning_rate": 9.804895938388192e-05,
+      "loss": 0.8203,
+      "mean_token_accuracy": 0.7573263347148895,
+      "num_tokens": 128516133.0,
+      "step": 55600
+    },
+    {
+      "epoch": 0.5098560557440176,
+      "learning_rate": 9.803062253598607e-05,
+      "loss": 0.8476,
+      "mean_token_accuracy": 0.7501390933990478,
+      "num_tokens": 128539886.0,
+      "step": 55610
+    },
+    {
+      "epoch": 0.5099477399834969,
+      "learning_rate": 9.801228568809022e-05,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7500462651252746,
+      "num_tokens": 128562059.0,
+      "step": 55620
+    },
+    {
+      "epoch": 0.5100394242229761,
+      "learning_rate": 9.799394884019437e-05,
+      "loss": 0.8409,
+      "mean_token_accuracy": 0.7515318751335144,
+      "num_tokens": 128585511.0,
+      "step": 55630
+    },
+    {
+      "epoch": 0.5101311084624554,
+      "learning_rate": 9.797561199229854e-05,
+      "loss": 0.881,
+      "mean_token_accuracy": 0.7405971586704254,
+      "num_tokens": 128608428.0,
+      "step": 55640
+    },
+    {
+      "epoch": 0.5102227927019345,
+      "learning_rate": 9.795727514440269e-05,
+      "loss": 0.8368,
+      "mean_token_accuracy": 0.7500964045524597,
+      "num_tokens": 128632418.0,
+      "step": 55650
+    },
+    {
+      "epoch": 0.5103144769414137,
+      "learning_rate": 9.793893829650683e-05,
+      "loss": 0.8704,
+      "mean_token_accuracy": 0.7406009614467621,
+      "num_tokens": 128655669.0,
+      "step": 55660
+    },
+    {
+      "epoch": 0.510406161180893,
+      "learning_rate": 9.792060144861099e-05,
+      "loss": 0.832,
+      "mean_token_accuracy": 0.7474497854709625,
+      "num_tokens": 128678512.0,
+      "step": 55670
+    },
+    {
+      "epoch": 0.5104978454203722,
+      "learning_rate": 9.790226460071514e-05,
+      "loss": 0.8726,
+      "mean_token_accuracy": 0.7440742075443267,
+      "num_tokens": 128701958.0,
+      "step": 55680
+    },
+    {
+      "epoch": 0.5105895296598515,
+      "learning_rate": 9.78839277528193e-05,
+      "loss": 0.8136,
+      "mean_token_accuracy": 0.757123863697052,
+      "num_tokens": 128725537.0,
+      "step": 55690
+    },
+    {
+      "epoch": 0.5106812138993307,
+      "learning_rate": 9.786559090492345e-05,
+      "loss": 0.8167,
+      "mean_token_accuracy": 0.7601333737373352,
+      "num_tokens": 128748847.0,
+      "step": 55700
+    },
+    {
+      "epoch": 0.5107728981388099,
+      "learning_rate": 9.78472540570276e-05,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7527754962444305,
+      "num_tokens": 128771671.0,
+      "step": 55710
+    },
+    {
+      "epoch": 0.5108645823782891,
+      "learning_rate": 9.782891720913175e-05,
+      "loss": 0.819,
+      "mean_token_accuracy": 0.7536244928836823,
+      "num_tokens": 128794737.0,
+      "step": 55720
+    },
+    {
+      "epoch": 0.5109562666177684,
+      "learning_rate": 9.78105803612359e-05,
+      "loss": 0.8791,
+      "mean_token_accuracy": 0.7345493853092193,
+      "num_tokens": 128818320.0,
+      "step": 55730
+    },
+    {
+      "epoch": 0.5110479508572476,
+      "learning_rate": 9.779224351334007e-05,
+      "loss": 0.8131,
+      "mean_token_accuracy": 0.7515747964382171,
+      "num_tokens": 128841468.0,
+      "step": 55740
+    },
+    {
+      "epoch": 0.5111396350967269,
+      "learning_rate": 9.777390666544422e-05,
+      "loss": 0.8233,
+      "mean_token_accuracy": 0.7469059705734253,
+      "num_tokens": 128864809.0,
+      "step": 55750
+    },
+    {
+      "epoch": 0.5112313193362061,
+      "learning_rate": 9.775556981754836e-05,
+      "loss": 0.8099,
+      "mean_token_accuracy": 0.7587299227714539,
+      "num_tokens": 128888020.0,
+      "step": 55760
+    },
+    {
+      "epoch": 0.5113230035756854,
+      "learning_rate": 9.773723296965252e-05,
+      "loss": 0.8205,
+      "mean_token_accuracy": 0.7524077892303467,
+      "num_tokens": 128911578.0,
+      "step": 55770
+    },
+    {
+      "epoch": 0.5114146878151645,
+      "learning_rate": 9.771889612175668e-05,
+      "loss": 0.8379,
+      "mean_token_accuracy": 0.753923523426056,
+      "num_tokens": 128935301.0,
+      "step": 55780
+    },
+    {
+      "epoch": 0.5115063720546438,
+      "learning_rate": 9.770055927386083e-05,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.7583781242370605,
+      "num_tokens": 128958541.0,
+      "step": 55790
+    },
+    {
+      "epoch": 0.511598056294123,
+      "learning_rate": 9.7682222425965e-05,
+      "loss": 0.8825,
+      "mean_token_accuracy": 0.7409592568874359,
+      "num_tokens": 128981377.0,
+      "step": 55800
+    },
+    {
+      "epoch": 0.5116897405336023,
+      "learning_rate": 9.766388557806913e-05,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.7621455550193786,
+      "num_tokens": 129003669.0,
+      "step": 55810
+    },
+    {
+      "epoch": 0.5117814247730815,
+      "learning_rate": 9.764554873017328e-05,
+      "loss": 0.8672,
+      "mean_token_accuracy": 0.745113855600357,
+      "num_tokens": 129026295.0,
+      "step": 55820
+    },
+    {
+      "epoch": 0.5118731090125608,
+      "learning_rate": 9.762721188227744e-05,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.7521910727024078,
+      "num_tokens": 129049756.0,
+      "step": 55830
+    },
+    {
+      "epoch": 0.51196479325204,
+      "learning_rate": 9.76088750343816e-05,
+      "loss": 0.8182,
+      "mean_token_accuracy": 0.7492237269878388,
+      "num_tokens": 129072490.0,
+      "step": 55840
+    },
+    {
+      "epoch": 0.5120564774915192,
+      "learning_rate": 9.759053818648575e-05,
+      "loss": 0.836,
+      "mean_token_accuracy": 0.7493184745311737,
+      "num_tokens": 129095824.0,
+      "step": 55850
+    },
+    {
+      "epoch": 0.5121481617309984,
+      "learning_rate": 9.757220133858989e-05,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7501515507698059,
+      "num_tokens": 129118889.0,
+      "step": 55860
+    },
+    {
+      "epoch": 0.5122398459704777,
+      "learning_rate": 9.755386449069406e-05,
+      "loss": 0.8599,
+      "mean_token_accuracy": 0.745026308298111,
+      "num_tokens": 129142115.0,
+      "step": 55870
+    },
+    {
+      "epoch": 0.5123315302099569,
+      "learning_rate": 9.753552764279821e-05,
+      "loss": 0.8507,
+      "mean_token_accuracy": 0.7484850764274598,
+      "num_tokens": 129164757.0,
+      "step": 55880
+    },
+    {
+      "epoch": 0.5124232144494362,
+      "learning_rate": 9.751719079490236e-05,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7456823706626892,
+      "num_tokens": 129187540.0,
+      "step": 55890
+    },
+    {
+      "epoch": 0.5125148986889154,
+      "learning_rate": 9.749885394700653e-05,
+      "loss": 0.8286,
+      "mean_token_accuracy": 0.7571347236633301,
+      "num_tokens": 129210427.0,
+      "step": 55900
+    },
+    {
+      "epoch": 0.5126065829283946,
+      "learning_rate": 9.748051709911066e-05,
+      "loss": 0.8518,
+      "mean_token_accuracy": 0.7484507083892822,
+      "num_tokens": 129234739.0,
+      "step": 55910
+    },
+    {
+      "epoch": 0.5126982671678738,
+      "learning_rate": 9.746218025121482e-05,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7446218013763428,
+      "num_tokens": 129258461.0,
+      "step": 55920
+    },
+    {
+      "epoch": 0.5127899514073531,
+      "learning_rate": 9.744384340331898e-05,
+      "loss": 0.8088,
+      "mean_token_accuracy": 0.7495130300521851,
+      "num_tokens": 129281228.0,
+      "step": 55930
+    },
+    {
+      "epoch": 0.5128816356468323,
+      "learning_rate": 9.742550655542313e-05,
+      "loss": 0.8407,
+      "mean_token_accuracy": 0.7451620817184448,
+      "num_tokens": 129304144.0,
+      "step": 55940
+    },
+    {
+      "epoch": 0.5129733198863116,
+      "learning_rate": 9.740716970752728e-05,
+      "loss": 0.8748,
+      "mean_token_accuracy": 0.7495053827762603,
+      "num_tokens": 129327151.0,
+      "step": 55950
+    },
+    {
+      "epoch": 0.5130650041257908,
+      "learning_rate": 9.738883285963142e-05,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7442684769630432,
+      "num_tokens": 129349965.0,
+      "step": 55960
+    },
+    {
+      "epoch": 0.5131566883652701,
+      "learning_rate": 9.737049601173559e-05,
+      "loss": 0.8463,
+      "mean_token_accuracy": 0.7464066505432129,
+      "num_tokens": 129372206.0,
+      "step": 55970
+    },
+    {
+      "epoch": 0.5132483726047492,
+      "learning_rate": 9.735215916383974e-05,
+      "loss": 0.792,
+      "mean_token_accuracy": 0.7607055842876435,
+      "num_tokens": 129395855.0,
+      "step": 55980
+    },
+    {
+      "epoch": 0.5133400568442285,
+      "learning_rate": 9.733382231594389e-05,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.743760633468628,
+      "num_tokens": 129418954.0,
+      "step": 55990
+    },
+    {
+      "epoch": 0.5134317410837077,
+      "learning_rate": 9.731548546804806e-05,
+      "loss": 0.8685,
+      "mean_token_accuracy": 0.7486790597438813,
+      "num_tokens": 129441879.0,
+      "step": 56000
+    },
+    {
+      "epoch": 0.513523425323187,
+      "learning_rate": 9.72971486201522e-05,
+      "loss": 0.8007,
+      "mean_token_accuracy": 0.7590884685516357,
+      "num_tokens": 129465720.0,
+      "step": 56010
+    },
+    {
+      "epoch": 0.5136151095626662,
+      "learning_rate": 9.727881177225635e-05,
+      "loss": 0.8273,
+      "mean_token_accuracy": 0.74710493683815,
+      "num_tokens": 129489680.0,
+      "step": 56020
+    },
+    {
+      "epoch": 0.5137067938021455,
+      "learning_rate": 9.726047492436051e-05,
+      "loss": 0.8471,
+      "mean_token_accuracy": 0.7541328251361847,
+      "num_tokens": 129513337.0,
+      "step": 56030
+    },
+    {
+      "epoch": 0.5137984780416246,
+      "learning_rate": 9.724213807646467e-05,
+      "loss": 0.8607,
+      "mean_token_accuracy": 0.7480553388595581,
+      "num_tokens": 129536815.0,
+      "step": 56040
+    },
+    {
+      "epoch": 0.5138901622811038,
+      "learning_rate": 9.722380122856882e-05,
+      "loss": 0.8584,
+      "mean_token_accuracy": 0.7457198381423951,
+      "num_tokens": 129559278.0,
+      "step": 56050
+    },
+    {
+      "epoch": 0.5139818465205831,
+      "learning_rate": 9.720546438067296e-05,
+      "loss": 0.841,
+      "mean_token_accuracy": 0.7460310757160187,
+      "num_tokens": 129581530.0,
+      "step": 56060
+    },
+    {
+      "epoch": 0.5140735307600623,
+      "learning_rate": 9.718712753277712e-05,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7485625565052032,
+      "num_tokens": 129604600.0,
+      "step": 56070
+    },
+    {
+      "epoch": 0.5141652149995416,
+      "learning_rate": 9.716879068488127e-05,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7470367312431335,
+      "num_tokens": 129627948.0,
+      "step": 56080
+    },
+    {
+      "epoch": 0.5142568992390208,
+      "learning_rate": 9.715045383698542e-05,
+      "loss": 0.8491,
+      "mean_token_accuracy": 0.7488150477409363,
+      "num_tokens": 129651538.0,
+      "step": 56090
+    },
+    {
+      "epoch": 0.5143485834785001,
+      "learning_rate": 9.713211698908959e-05,
+      "loss": 0.8513,
+      "mean_token_accuracy": 0.7482627809047699,
+      "num_tokens": 129674323.0,
+      "step": 56100
+    },
+    {
+      "epoch": 0.5144402677179792,
+      "learning_rate": 9.711378014119373e-05,
+      "loss": 0.8308,
+      "mean_token_accuracy": 0.7516775965690613,
+      "num_tokens": 129696668.0,
+      "step": 56110
+    },
+    {
+      "epoch": 0.5145319519574585,
+      "learning_rate": 9.709544329329788e-05,
+      "loss": 0.8254,
+      "mean_token_accuracy": 0.7606424748897552,
+      "num_tokens": 129720306.0,
+      "step": 56120
+    },
+    {
+      "epoch": 0.5146236361969377,
+      "learning_rate": 9.707710644540205e-05,
+      "loss": 0.8582,
+      "mean_token_accuracy": 0.7417102813720703,
+      "num_tokens": 129742964.0,
+      "step": 56130
+    },
+    {
+      "epoch": 0.514715320436417,
+      "learning_rate": 9.70587695975062e-05,
+      "loss": 0.8562,
+      "mean_token_accuracy": 0.7418065547943116,
+      "num_tokens": 129766227.0,
+      "step": 56140
+    },
+    {
+      "epoch": 0.5148070046758962,
+      "learning_rate": 9.704043274961035e-05,
+      "loss": 0.876,
+      "mean_token_accuracy": 0.745196384191513,
+      "num_tokens": 129789345.0,
+      "step": 56150
+    },
+    {
+      "epoch": 0.5148986889153755,
+      "learning_rate": 9.70220959017145e-05,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.7529929518699646,
+      "num_tokens": 129813158.0,
+      "step": 56160
+    },
+    {
+      "epoch": 0.5149903731548546,
+      "learning_rate": 9.700375905381865e-05,
+      "loss": 0.8494,
+      "mean_token_accuracy": 0.7480137288570404,
+      "num_tokens": 129836146.0,
+      "step": 56170
+    },
+    {
+      "epoch": 0.5150820573943339,
+      "learning_rate": 9.69854222059228e-05,
+      "loss": 0.8431,
+      "mean_token_accuracy": 0.7475702226161957,
+      "num_tokens": 129859853.0,
+      "step": 56180
+    },
+    {
+      "epoch": 0.5151737416338131,
+      "learning_rate": 9.696708535802696e-05,
+      "loss": 0.8415,
+      "mean_token_accuracy": 0.7543336629867554,
+      "num_tokens": 129882598.0,
+      "step": 56190
+    },
+    {
+      "epoch": 0.5152654258732924,
+      "learning_rate": 9.694874851013112e-05,
+      "loss": 0.8129,
+      "mean_token_accuracy": 0.7548528909683228,
+      "num_tokens": 129905096.0,
+      "step": 56200
+    },
+    {
+      "epoch": 0.5153571101127716,
+      "learning_rate": 9.693041166223526e-05,
+      "loss": 0.7859,
+      "mean_token_accuracy": 0.7606727242469787,
+      "num_tokens": 129928093.0,
+      "step": 56210
+    },
+    {
+      "epoch": 0.5154487943522509,
+      "learning_rate": 9.691207481433941e-05,
+      "loss": 0.8208,
+      "mean_token_accuracy": 0.7520423114299775,
+      "num_tokens": 129951594.0,
+      "step": 56220
+    },
+    {
+      "epoch": 0.5155404785917301,
+      "learning_rate": 9.689373796644358e-05,
+      "loss": 0.836,
+      "mean_token_accuracy": 0.7486515760421752,
+      "num_tokens": 129974314.0,
+      "step": 56230
+    },
+    {
+      "epoch": 0.5156321628312093,
+      "learning_rate": 9.687540111854773e-05,
+      "loss": 0.7994,
+      "mean_token_accuracy": 0.7629447162151337,
+      "num_tokens": 129997542.0,
+      "step": 56240
+    },
+    {
+      "epoch": 0.5157238470706885,
+      "learning_rate": 9.685706427065188e-05,
+      "loss": 0.881,
+      "mean_token_accuracy": 0.7391123592853546,
+      "num_tokens": 130020459.0,
+      "step": 56250
+    },
+    {
+      "epoch": 0.5158155313101678,
+      "learning_rate": 9.683872742275603e-05,
+      "loss": 0.8586,
+      "mean_token_accuracy": 0.7472450077533722,
+      "num_tokens": 130043649.0,
+      "step": 56260
+    },
+    {
+      "epoch": 0.515907215549647,
+      "learning_rate": 9.682039057486019e-05,
+      "loss": 0.8366,
+      "mean_token_accuracy": 0.7542475044727326,
+      "num_tokens": 130067388.0,
+      "step": 56270
+    },
+    {
+      "epoch": 0.5159988997891263,
+      "learning_rate": 9.680205372696434e-05,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7474272787570954,
+      "num_tokens": 130090318.0,
+      "step": 56280
+    },
+    {
+      "epoch": 0.5160905840286055,
+      "learning_rate": 9.67837168790685e-05,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7500754296779633,
+      "num_tokens": 130113308.0,
+      "step": 56290
+    },
+    {
+      "epoch": 0.5161822682680847,
+      "learning_rate": 9.676538003117265e-05,
+      "loss": 0.8895,
+      "mean_token_accuracy": 0.7304994821548462,
+      "num_tokens": 130136726.0,
+      "step": 56300
+    },
+    {
+      "epoch": 0.5162739525075639,
+      "learning_rate": 9.674704318327679e-05,
+      "loss": 0.8225,
+      "mean_token_accuracy": 0.7477835178375244,
+      "num_tokens": 130159769.0,
+      "step": 56310
+    },
+    {
+      "epoch": 0.5163656367470432,
+      "learning_rate": 9.672870633538094e-05,
+      "loss": 0.8408,
+      "mean_token_accuracy": 0.743791788816452,
+      "num_tokens": 130182757.0,
+      "step": 56320
+    },
+    {
+      "epoch": 0.5164573209865224,
+      "learning_rate": 9.671036948748511e-05,
+      "loss": 0.8262,
+      "mean_token_accuracy": 0.7528278768062592,
+      "num_tokens": 130206495.0,
+      "step": 56330
+    },
+    {
+      "epoch": 0.5165490052260017,
+      "learning_rate": 9.669203263958926e-05,
+      "loss": 0.8291,
+      "mean_token_accuracy": 0.7492689609527587,
+      "num_tokens": 130229215.0,
+      "step": 56340
+    },
+    {
+      "epoch": 0.5166406894654809,
+      "learning_rate": 9.667369579169341e-05,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.7553637564182282,
+      "num_tokens": 130252431.0,
+      "step": 56350
+    },
+    {
+      "epoch": 0.5167323737049602,
+      "learning_rate": 9.665535894379757e-05,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7451859712600708,
+      "num_tokens": 130275970.0,
+      "step": 56360
+    },
+    {
+      "epoch": 0.5168240579444393,
+      "learning_rate": 9.663702209590172e-05,
+      "loss": 0.8074,
+      "mean_token_accuracy": 0.7540444791316986,
+      "num_tokens": 130299393.0,
+      "step": 56370
+    },
+    {
+      "epoch": 0.5169157421839186,
+      "learning_rate": 9.661868524800587e-05,
+      "loss": 0.8422,
+      "mean_token_accuracy": 0.7524396538734436,
+      "num_tokens": 130322460.0,
+      "step": 56380
+    },
+    {
+      "epoch": 0.5170074264233978,
+      "learning_rate": 9.660034840011003e-05,
+      "loss": 0.8684,
+      "mean_token_accuracy": 0.7433933138847351,
+      "num_tokens": 130345701.0,
+      "step": 56390
+    },
+    {
+      "epoch": 0.5170991106628771,
+      "learning_rate": 9.658201155221419e-05,
+      "loss": 0.8552,
+      "mean_token_accuracy": 0.7409780025482178,
+      "num_tokens": 130369050.0,
+      "step": 56400
+    },
+    {
+      "epoch": 0.5171907949023563,
+      "learning_rate": 9.656367470431832e-05,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7443040788173676,
+      "num_tokens": 130392154.0,
+      "step": 56410
+    },
+    {
+      "epoch": 0.5172824791418356,
+      "learning_rate": 9.654533785642249e-05,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.752303522825241,
+      "num_tokens": 130414408.0,
+      "step": 56420
+    },
+    {
+      "epoch": 0.5173741633813147,
+      "learning_rate": 9.652700100852664e-05,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7437770307064057,
+      "num_tokens": 130436886.0,
+      "step": 56430
+    },
+    {
+      "epoch": 0.517465847620794,
+      "learning_rate": 9.65086641606308e-05,
+      "loss": 0.8502,
+      "mean_token_accuracy": 0.7493125796318054,
+      "num_tokens": 130459919.0,
+      "step": 56440
+    },
+    {
+      "epoch": 0.5175575318602732,
+      "learning_rate": 9.649032731273495e-05,
+      "loss": 0.8256,
+      "mean_token_accuracy": 0.7514883935451507,
+      "num_tokens": 130483189.0,
+      "step": 56450
+    },
+    {
+      "epoch": 0.5176492160997525,
+      "learning_rate": 9.64719904648391e-05,
+      "loss": 0.8247,
+      "mean_token_accuracy": 0.7569907665252685,
+      "num_tokens": 130505681.0,
+      "step": 56460
+    },
+    {
+      "epoch": 0.5177409003392317,
+      "learning_rate": 9.645365361694325e-05,
+      "loss": 0.7985,
+      "mean_token_accuracy": 0.7550045609474182,
+      "num_tokens": 130528841.0,
+      "step": 56470
+    },
+    {
+      "epoch": 0.517832584578711,
+      "learning_rate": 9.64353167690474e-05,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.7549361646175384,
+      "num_tokens": 130552409.0,
+      "step": 56480
+    },
+    {
+      "epoch": 0.5179242688181902,
+      "learning_rate": 9.641697992115157e-05,
+      "loss": 0.8145,
+      "mean_token_accuracy": 0.758808434009552,
+      "num_tokens": 130575061.0,
+      "step": 56490
+    },
+    {
+      "epoch": 0.5180159530576693,
+      "learning_rate": 9.639864307325572e-05,
+      "loss": 0.8206,
+      "mean_token_accuracy": 0.7493482887744903,
+      "num_tokens": 130598765.0,
+      "step": 56500
+    },
+    {
+      "epoch": 0.5181076372971486,
+      "learning_rate": 9.638030622535986e-05,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7403877735137939,
+      "num_tokens": 130622746.0,
+      "step": 56510
+    },
+    {
+      "epoch": 0.5181993215366278,
+      "learning_rate": 9.636196937746402e-05,
+      "loss": 0.8264,
+      "mean_token_accuracy": 0.7492237091064453,
+      "num_tokens": 130645671.0,
+      "step": 56520
+    },
+    {
+      "epoch": 0.5182910057761071,
+      "learning_rate": 9.634363252956817e-05,
+      "loss": 0.8388,
+      "mean_token_accuracy": 0.7474077999591827,
+      "num_tokens": 130668535.0,
+      "step": 56530
+    },
+    {
+      "epoch": 0.5183826900155863,
+      "learning_rate": 9.632529568167233e-05,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7496312081813812,
+      "num_tokens": 130691741.0,
+      "step": 56540
+    },
+    {
+      "epoch": 0.5184743742550656,
+      "learning_rate": 9.630695883377648e-05,
+      "loss": 0.8177,
+      "mean_token_accuracy": 0.7515130519866944,
+      "num_tokens": 130714872.0,
+      "step": 56550
+    },
+    {
+      "epoch": 0.5185660584945447,
+      "learning_rate": 9.628862198588063e-05,
+      "loss": 0.8172,
+      "mean_token_accuracy": 0.7503711342811584,
+      "num_tokens": 130737646.0,
+      "step": 56560
+    },
+    {
+      "epoch": 0.518657742734024,
+      "learning_rate": 9.627028513798478e-05,
+      "loss": 0.8013,
+      "mean_token_accuracy": 0.7572340548038483,
+      "num_tokens": 130760567.0,
+      "step": 56570
+    },
+    {
+      "epoch": 0.5187494269735032,
+      "learning_rate": 9.625194829008893e-05,
+      "loss": 0.8347,
+      "mean_token_accuracy": 0.7530829191207886,
+      "num_tokens": 130783564.0,
+      "step": 56580
+    },
+    {
+      "epoch": 0.5188411112129825,
+      "learning_rate": 9.62336114421931e-05,
+      "loss": 0.8625,
+      "mean_token_accuracy": 0.7468912541866303,
+      "num_tokens": 130806111.0,
+      "step": 56590
+    },
+    {
+      "epoch": 0.5189327954524617,
+      "learning_rate": 9.621527459429725e-05,
+      "loss": 0.8209,
+      "mean_token_accuracy": 0.7515967309474945,
+      "num_tokens": 130829362.0,
+      "step": 56600
+    },
+    {
+      "epoch": 0.519024479691941,
+      "learning_rate": 9.619693774640139e-05,
+      "loss": 0.8133,
+      "mean_token_accuracy": 0.7559102058410645,
+      "num_tokens": 130852671.0,
+      "step": 56610
+    },
+    {
+      "epoch": 0.5191161639314202,
+      "learning_rate": 9.617860089850555e-05,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7443732738494873,
+      "num_tokens": 130875717.0,
+      "step": 56620
+    },
+    {
+      "epoch": 0.5192078481708994,
+      "learning_rate": 9.61602640506097e-05,
+      "loss": 0.8813,
+      "mean_token_accuracy": 0.7381378471851349,
+      "num_tokens": 130898979.0,
+      "step": 56630
+    },
+    {
+      "epoch": 0.5192995324103786,
+      "learning_rate": 9.614192720271386e-05,
+      "loss": 0.8489,
+      "mean_token_accuracy": 0.742695790529251,
+      "num_tokens": 130922449.0,
+      "step": 56640
+    },
+    {
+      "epoch": 0.5193912166498579,
+      "learning_rate": 9.612359035481801e-05,
+      "loss": 0.8137,
+      "mean_token_accuracy": 0.7539062261581421,
+      "num_tokens": 130944965.0,
+      "step": 56650
+    },
+    {
+      "epoch": 0.5194829008893371,
+      "learning_rate": 9.610525350692216e-05,
+      "loss": 0.8681,
+      "mean_token_accuracy": 0.7439112007617951,
+      "num_tokens": 130968046.0,
+      "step": 56660
+    },
+    {
+      "epoch": 0.5195745851288164,
+      "learning_rate": 9.608691665902631e-05,
+      "loss": 0.8179,
+      "mean_token_accuracy": 0.7543026506900787,
+      "num_tokens": 130991273.0,
+      "step": 56670
+    },
+    {
+      "epoch": 0.5196662693682956,
+      "learning_rate": 9.606857981113047e-05,
+      "loss": 0.8213,
+      "mean_token_accuracy": 0.7530774056911469,
+      "num_tokens": 131015586.0,
+      "step": 56680
+    },
+    {
+      "epoch": 0.5197579536077748,
+      "learning_rate": 9.605024296323463e-05,
+      "loss": 0.8061,
+      "mean_token_accuracy": 0.7528681635856629,
+      "num_tokens": 131038648.0,
+      "step": 56690
+    },
+    {
+      "epoch": 0.519849637847254,
+      "learning_rate": 9.603190611533878e-05,
+      "loss": 0.8134,
+      "mean_token_accuracy": 0.7553314149379731,
+      "num_tokens": 131061932.0,
+      "step": 56700
+    },
+    {
+      "epoch": 0.5199413220867333,
+      "learning_rate": 9.601356926744292e-05,
+      "loss": 0.8223,
+      "mean_token_accuracy": 0.7534793615341187,
+      "num_tokens": 131084572.0,
+      "step": 56710
+    },
+    {
+      "epoch": 0.5200330063262125,
+      "learning_rate": 9.599523241954709e-05,
+      "loss": 0.8142,
+      "mean_token_accuracy": 0.7502929627895355,
+      "num_tokens": 131107331.0,
+      "step": 56720
+    },
+    {
+      "epoch": 0.5201246905656918,
+      "learning_rate": 9.597689557165124e-05,
+      "loss": 0.8519,
+      "mean_token_accuracy": 0.7471710503101349,
+      "num_tokens": 131130301.0,
+      "step": 56730
+    },
+    {
+      "epoch": 0.520216374805171,
+      "learning_rate": 9.595855872375539e-05,
+      "loss": 0.8415,
+      "mean_token_accuracy": 0.7527761101722718,
+      "num_tokens": 131153232.0,
+      "step": 56740
+    },
+    {
+      "epoch": 0.5203080590446503,
+      "learning_rate": 9.594022187585954e-05,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7562039017677307,
+      "num_tokens": 131176569.0,
+      "step": 56750
+    },
+    {
+      "epoch": 0.5203997432841294,
+      "learning_rate": 9.59218850279637e-05,
+      "loss": 0.8489,
+      "mean_token_accuracy": 0.7481797993183136,
+      "num_tokens": 131199560.0,
+      "step": 56760
+    },
+    {
+      "epoch": 0.5204914275236087,
+      "learning_rate": 9.590354818006785e-05,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.7534015774726868,
+      "num_tokens": 131222235.0,
+      "step": 56770
+    },
+    {
+      "epoch": 0.5205831117630879,
+      "learning_rate": 9.588521133217201e-05,
+      "loss": 0.7896,
+      "mean_token_accuracy": 0.761231142282486,
+      "num_tokens": 131244707.0,
+      "step": 56780
+    },
+    {
+      "epoch": 0.5206747960025672,
+      "learning_rate": 9.586687448427616e-05,
+      "loss": 0.8674,
+      "mean_token_accuracy": 0.7412490248680115,
+      "num_tokens": 131267957.0,
+      "step": 56790
+    },
+    {
+      "epoch": 0.5207664802420464,
+      "learning_rate": 9.584853763638032e-05,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7470519006252289,
+      "num_tokens": 131290722.0,
+      "step": 56800
+    },
+    {
+      "epoch": 0.5208581644815257,
+      "learning_rate": 9.583020078848445e-05,
+      "loss": 0.8579,
+      "mean_token_accuracy": 0.7428442060947418,
+      "num_tokens": 131314094.0,
+      "step": 56810
+    },
+    {
+      "epoch": 0.5209498487210048,
+      "learning_rate": 9.581186394058862e-05,
+      "loss": 0.8172,
+      "mean_token_accuracy": 0.7486804187297821,
+      "num_tokens": 131337454.0,
+      "step": 56820
+    },
+    {
+      "epoch": 0.521041532960484,
+      "learning_rate": 9.579352709269277e-05,
+      "loss": 0.8326,
+      "mean_token_accuracy": 0.7509289622306824,
+      "num_tokens": 131360423.0,
+      "step": 56830
+    },
+    {
+      "epoch": 0.5211332171999633,
+      "learning_rate": 9.577519024479692e-05,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.743102389574051,
+      "num_tokens": 131383916.0,
+      "step": 56840
+    },
+    {
+      "epoch": 0.5212249014394426,
+      "learning_rate": 9.575685339690107e-05,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7442952692508698,
+      "num_tokens": 131407662.0,
+      "step": 56850
+    },
+    {
+      "epoch": 0.5213165856789218,
+      "learning_rate": 9.573851654900523e-05,
+      "loss": 0.8257,
+      "mean_token_accuracy": 0.7501402974128724,
+      "num_tokens": 131430601.0,
+      "step": 56860
+    },
+    {
+      "epoch": 0.5214082699184011,
+      "learning_rate": 9.572017970110938e-05,
+      "loss": 0.8251,
+      "mean_token_accuracy": 0.7523505985736847,
+      "num_tokens": 131453829.0,
+      "step": 56870
+    },
+    {
+      "epoch": 0.5214999541578803,
+      "learning_rate": 9.570184285321354e-05,
+      "loss": 0.804,
+      "mean_token_accuracy": 0.7546593546867371,
+      "num_tokens": 131477027.0,
+      "step": 56880
+    },
+    {
+      "epoch": 0.5215916383973594,
+      "learning_rate": 9.56835060053177e-05,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.7541169285774231,
+      "num_tokens": 131500478.0,
+      "step": 56890
+    },
+    {
+      "epoch": 0.5216833226368387,
+      "learning_rate": 9.566516915742185e-05,
+      "loss": 0.8466,
+      "mean_token_accuracy": 0.7427525162696839,
+      "num_tokens": 131523953.0,
+      "step": 56900
+    },
+    {
+      "epoch": 0.521775006876318,
+      "learning_rate": 9.5646832309526e-05,
+      "loss": 0.8428,
+      "mean_token_accuracy": 0.742903733253479,
+      "num_tokens": 131547167.0,
+      "step": 56910
+    },
+    {
+      "epoch": 0.5218666911157972,
+      "learning_rate": 9.562849546163015e-05,
+      "loss": 0.8455,
+      "mean_token_accuracy": 0.7451618432998657,
+      "num_tokens": 131570197.0,
+      "step": 56920
+    },
+    {
+      "epoch": 0.5219583753552764,
+      "learning_rate": 9.56101586137343e-05,
+      "loss": 0.8208,
+      "mean_token_accuracy": 0.7520460546016693,
+      "num_tokens": 131594129.0,
+      "step": 56930
+    },
+    {
+      "epoch": 0.5220500595947557,
+      "learning_rate": 9.559182176583846e-05,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7443790912628174,
+      "num_tokens": 131616888.0,
+      "step": 56940
+    },
+    {
+      "epoch": 0.522141743834235,
+      "learning_rate": 9.557348491794261e-05,
+      "loss": 0.8314,
+      "mean_token_accuracy": 0.7462843596935272,
+      "num_tokens": 131639627.0,
+      "step": 56950
+    },
+    {
+      "epoch": 0.5222334280737141,
+      "learning_rate": 9.555514807004676e-05,
+      "loss": 0.8737,
+      "mean_token_accuracy": 0.7411675214767456,
+      "num_tokens": 131662072.0,
+      "step": 56960
+    },
+    {
+      "epoch": 0.5223251123131933,
+      "learning_rate": 9.553681122215091e-05,
+      "loss": 0.851,
+      "mean_token_accuracy": 0.745289546251297,
+      "num_tokens": 131685280.0,
+      "step": 56970
+    },
+    {
+      "epoch": 0.5224167965526726,
+      "learning_rate": 9.551847437425508e-05,
+      "loss": 0.8357,
+      "mean_token_accuracy": 0.7507482826709747,
+      "num_tokens": 131708358.0,
+      "step": 56980
+    },
+    {
+      "epoch": 0.5225084807921518,
+      "learning_rate": 9.550013752635923e-05,
+      "loss": 0.8474,
+      "mean_token_accuracy": 0.7514139413833618,
+      "num_tokens": 131732494.0,
+      "step": 56990
+    },
+    {
+      "epoch": 0.5226001650316311,
+      "learning_rate": 9.548180067846338e-05,
+      "loss": 0.8186,
+      "mean_token_accuracy": 0.7497166395187378,
+      "num_tokens": 131755048.0,
+      "step": 57000
+    },
+    {
+      "epoch": 0.5226918492711103,
+      "learning_rate": 9.546346383056753e-05,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7481898128986358,
+      "num_tokens": 131778522.0,
+      "step": 57010
+    },
+    {
+      "epoch": 0.5227835335105895,
+      "learning_rate": 9.544512698267168e-05,
+      "loss": 0.8311,
+      "mean_token_accuracy": 0.7491007447242737,
+      "num_tokens": 131801658.0,
+      "step": 57020
+    },
+    {
+      "epoch": 0.5228752177500687,
+      "learning_rate": 9.542679013477584e-05,
+      "loss": 0.8458,
+      "mean_token_accuracy": 0.7505529463291168,
+      "num_tokens": 131824486.0,
+      "step": 57030
+    },
+    {
+      "epoch": 0.522966901989548,
+      "learning_rate": 9.540845328688e-05,
+      "loss": 0.7944,
+      "mean_token_accuracy": 0.7576842725276947,
+      "num_tokens": 131847717.0,
+      "step": 57040
+    },
+    {
+      "epoch": 0.5230585862290272,
+      "learning_rate": 9.539011643898414e-05,
+      "loss": 0.8911,
+      "mean_token_accuracy": 0.7397165119647979,
+      "num_tokens": 131871073.0,
+      "step": 57050
+    },
+    {
+      "epoch": 0.5231502704685065,
+      "learning_rate": 9.537177959108829e-05,
+      "loss": 0.8399,
+      "mean_token_accuracy": 0.7458792924880981,
+      "num_tokens": 131893726.0,
+      "step": 57060
+    },
+    {
+      "epoch": 0.5232419547079857,
+      "learning_rate": 9.535344274319244e-05,
+      "loss": 0.8409,
+      "mean_token_accuracy": 0.7471013247966767,
+      "num_tokens": 131917471.0,
+      "step": 57070
+    },
+    {
+      "epoch": 0.523333638947465,
+      "learning_rate": 9.533510589529661e-05,
+      "loss": 0.8717,
+      "mean_token_accuracy": 0.7374906480312348,
+      "num_tokens": 131940815.0,
+      "step": 57080
+    },
+    {
+      "epoch": 0.5234253231869441,
+      "learning_rate": 9.531676904740076e-05,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7464725852012635,
+      "num_tokens": 131964724.0,
+      "step": 57090
+    },
+    {
+      "epoch": 0.5235170074264234,
+      "learning_rate": 9.529843219950491e-05,
+      "loss": 0.8448,
+      "mean_token_accuracy": 0.741492235660553,
+      "num_tokens": 131987976.0,
+      "step": 57100
+    },
+    {
+      "epoch": 0.5236086916659026,
+      "learning_rate": 9.528009535160906e-05,
+      "loss": 0.8083,
+      "mean_token_accuracy": 0.758763724565506,
+      "num_tokens": 132011836.0,
+      "step": 57110
+    },
+    {
+      "epoch": 0.5237003759053819,
+      "learning_rate": 9.526175850371322e-05,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.7485464990139008,
+      "num_tokens": 132034853.0,
+      "step": 57120
+    },
+    {
+      "epoch": 0.5237920601448611,
+      "learning_rate": 9.524342165581737e-05,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7454329550266265,
+      "num_tokens": 132058644.0,
+      "step": 57130
+    },
+    {
+      "epoch": 0.5238837443843404,
+      "learning_rate": 9.522508480792153e-05,
+      "loss": 0.8604,
+      "mean_token_accuracy": 0.7422268509864807,
+      "num_tokens": 132081896.0,
+      "step": 57140
+    },
+    {
+      "epoch": 0.5239754286238195,
+      "learning_rate": 9.520674796002567e-05,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7491759121417999,
+      "num_tokens": 132105666.0,
+      "step": 57150
+    },
+    {
+      "epoch": 0.5240671128632988,
+      "learning_rate": 9.518841111212982e-05,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7442149877548218,
+      "num_tokens": 132128795.0,
+      "step": 57160
+    },
+    {
+      "epoch": 0.524158797102778,
+      "learning_rate": 9.517007426423399e-05,
+      "loss": 0.8123,
+      "mean_token_accuracy": 0.7535880506038666,
+      "num_tokens": 132152913.0,
+      "step": 57170
+    },
+    {
+      "epoch": 0.5242504813422573,
+      "learning_rate": 9.515173741633814e-05,
+      "loss": 0.8256,
+      "mean_token_accuracy": 0.7542143046855927,
+      "num_tokens": 132176758.0,
+      "step": 57180
+    },
+    {
+      "epoch": 0.5243421655817365,
+      "learning_rate": 9.513340056844229e-05,
+      "loss": 0.8217,
+      "mean_token_accuracy": 0.7579224288463593,
+      "num_tokens": 132199520.0,
+      "step": 57190
+    },
+    {
+      "epoch": 0.5244338498212158,
+      "learning_rate": 9.511506372054644e-05,
+      "loss": 0.8376,
+      "mean_token_accuracy": 0.7474627912044525,
+      "num_tokens": 132223405.0,
+      "step": 57200
+    },
+    {
+      "epoch": 0.524525534060695,
+      "learning_rate": 9.50967268726506e-05,
+      "loss": 0.8274,
+      "mean_token_accuracy": 0.7523048102855683,
+      "num_tokens": 132246499.0,
+      "step": 57210
+    },
+    {
+      "epoch": 0.5246172183001742,
+      "learning_rate": 9.507839002475475e-05,
+      "loss": 0.8278,
+      "mean_token_accuracy": 0.7563997805118561,
+      "num_tokens": 132270309.0,
+      "step": 57220
+    },
+    {
+      "epoch": 0.5247089025396534,
+      "learning_rate": 9.50600531768589e-05,
+      "loss": 0.812,
+      "mean_token_accuracy": 0.7527232706546784,
+      "num_tokens": 132293903.0,
+      "step": 57230
+    },
+    {
+      "epoch": 0.5248005867791327,
+      "learning_rate": 9.504171632896307e-05,
+      "loss": 0.8095,
+      "mean_token_accuracy": 0.7493672370910645,
+      "num_tokens": 132317030.0,
+      "step": 57240
+    },
+    {
+      "epoch": 0.5248922710186119,
+      "learning_rate": 9.50233794810672e-05,
+      "loss": 0.8288,
+      "mean_token_accuracy": 0.7502146303653717,
+      "num_tokens": 132340928.0,
+      "step": 57250
+    },
+    {
+      "epoch": 0.5249839552580912,
+      "learning_rate": 9.500504263317136e-05,
+      "loss": 0.8286,
+      "mean_token_accuracy": 0.7479620635509491,
+      "num_tokens": 132364758.0,
+      "step": 57260
+    },
+    {
+      "epoch": 0.5250756394975704,
+      "learning_rate": 9.498670578527552e-05,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7462909758090973,
+      "num_tokens": 132388386.0,
+      "step": 57270
+    },
+    {
+      "epoch": 0.5251673237370496,
+      "learning_rate": 9.496836893737967e-05,
+      "loss": 0.8087,
+      "mean_token_accuracy": 0.7573596954345703,
+      "num_tokens": 132412261.0,
+      "step": 57280
+    },
+    {
+      "epoch": 0.5252590079765288,
+      "learning_rate": 9.495003208948382e-05,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7508314967155456,
+      "num_tokens": 132435035.0,
+      "step": 57290
+    },
+    {
+      "epoch": 0.525350692216008,
+      "learning_rate": 9.493169524158798e-05,
+      "loss": 0.8335,
+      "mean_token_accuracy": 0.7515981912612915,
+      "num_tokens": 132458573.0,
+      "step": 57300
+    },
+    {
+      "epoch": 0.5254423764554873,
+      "learning_rate": 9.491335839369213e-05,
+      "loss": 0.8131,
+      "mean_token_accuracy": 0.7545001804828644,
+      "num_tokens": 132481783.0,
+      "step": 57310
+    },
+    {
+      "epoch": 0.5255340606949666,
+      "learning_rate": 9.489502154579628e-05,
+      "loss": 0.8037,
+      "mean_token_accuracy": 0.756865781545639,
+      "num_tokens": 132505196.0,
+      "step": 57320
+    },
+    {
+      "epoch": 0.5256257449344458,
+      "learning_rate": 9.487668469790043e-05,
+      "loss": 0.8196,
+      "mean_token_accuracy": 0.7529776632785797,
+      "num_tokens": 132529169.0,
+      "step": 57330
+    },
+    {
+      "epoch": 0.525717429173925,
+      "learning_rate": 9.48583478500046e-05,
+      "loss": 0.8247,
+      "mean_token_accuracy": 0.7593657732009887,
+      "num_tokens": 132552626.0,
+      "step": 57340
+    },
+    {
+      "epoch": 0.5258091134134042,
+      "learning_rate": 9.484001100210874e-05,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7545748353004456,
+      "num_tokens": 132576850.0,
+      "step": 57350
+    },
+    {
+      "epoch": 0.5259007976528834,
+      "learning_rate": 9.482167415421289e-05,
+      "loss": 0.7969,
+      "mean_token_accuracy": 0.7564575612545014,
+      "num_tokens": 132601601.0,
+      "step": 57360
+    },
+    {
+      "epoch": 0.5259924818923627,
+      "learning_rate": 9.480333730631705e-05,
+      "loss": 0.8186,
+      "mean_token_accuracy": 0.7548130989074707,
+      "num_tokens": 132625568.0,
+      "step": 57370
+    },
+    {
+      "epoch": 0.526084166131842,
+      "learning_rate": 9.47850004584212e-05,
+      "loss": 0.8422,
+      "mean_token_accuracy": 0.7476212620735169,
+      "num_tokens": 132649507.0,
+      "step": 57380
+    },
+    {
+      "epoch": 0.5261758503713212,
+      "learning_rate": 9.476666361052536e-05,
+      "loss": 0.8363,
+      "mean_token_accuracy": 0.7549296379089355,
+      "num_tokens": 132672725.0,
+      "step": 57390
+    },
+    {
+      "epoch": 0.5262675346108004,
+      "learning_rate": 9.474832676262951e-05,
+      "loss": 0.8237,
+      "mean_token_accuracy": 0.7542741537094116,
+      "num_tokens": 132696060.0,
+      "step": 57400
+    },
+    {
+      "epoch": 0.5263592188502796,
+      "learning_rate": 9.472998991473366e-05,
+      "loss": 0.8367,
+      "mean_token_accuracy": 0.7523957431316376,
+      "num_tokens": 132719765.0,
+      "step": 57410
+    },
+    {
+      "epoch": 0.5264509030897588,
+      "learning_rate": 9.471165306683781e-05,
+      "loss": 0.8058,
+      "mean_token_accuracy": 0.755360621213913,
+      "num_tokens": 132743859.0,
+      "step": 57420
+    },
+    {
+      "epoch": 0.5265425873292381,
+      "learning_rate": 9.469331621894196e-05,
+      "loss": 0.8484,
+      "mean_token_accuracy": 0.7548549771308899,
+      "num_tokens": 132766432.0,
+      "step": 57430
+    },
+    {
+      "epoch": 0.5266342715687173,
+      "learning_rate": 9.467497937104613e-05,
+      "loss": 0.8446,
+      "mean_token_accuracy": 0.7448384761810303,
+      "num_tokens": 132789432.0,
+      "step": 57440
+    },
+    {
+      "epoch": 0.5267259558081966,
+      "learning_rate": 9.465664252315027e-05,
+      "loss": 0.7871,
+      "mean_token_accuracy": 0.7622412264347076,
+      "num_tokens": 132813301.0,
+      "step": 57450
+    },
+    {
+      "epoch": 0.5268176400476758,
+      "learning_rate": 9.463830567525442e-05,
+      "loss": 0.8309,
+      "mean_token_accuracy": 0.750828206539154,
+      "num_tokens": 132836450.0,
+      "step": 57460
+    },
+    {
+      "epoch": 0.5269093242871551,
+      "learning_rate": 9.461996882735859e-05,
+      "loss": 0.8423,
+      "mean_token_accuracy": 0.7530226111412048,
+      "num_tokens": 132859850.0,
+      "step": 57470
+    },
+    {
+      "epoch": 0.5270010085266342,
+      "learning_rate": 9.460163197946274e-05,
+      "loss": 0.8388,
+      "mean_token_accuracy": 0.7607247173786164,
+      "num_tokens": 132883759.0,
+      "step": 57480
+    },
+    {
+      "epoch": 0.5270926927661135,
+      "learning_rate": 9.458329513156689e-05,
+      "loss": 0.8297,
+      "mean_token_accuracy": 0.752485191822052,
+      "num_tokens": 132907839.0,
+      "step": 57490
+    },
+    {
+      "epoch": 0.5271843770055927,
+      "learning_rate": 9.456495828367104e-05,
+      "loss": 0.8071,
+      "mean_token_accuracy": 0.7524581611156463,
+      "num_tokens": 132931824.0,
+      "step": 57500
+    },
+    {
+      "epoch": 0.527276061245072,
+      "learning_rate": 9.454662143577519e-05,
+      "loss": 0.8011,
+      "mean_token_accuracy": 0.7583715319633484,
+      "num_tokens": 132955307.0,
+      "step": 57510
+    },
+    {
+      "epoch": 0.5273677454845512,
+      "learning_rate": 9.452828458787934e-05,
+      "loss": 0.8086,
+      "mean_token_accuracy": 0.7547164380550384,
+      "num_tokens": 132978556.0,
+      "step": 57520
+    },
+    {
+      "epoch": 0.5274594297240305,
+      "learning_rate": 9.450994773998351e-05,
+      "loss": 0.8672,
+      "mean_token_accuracy": 0.7438210904598236,
+      "num_tokens": 133001261.0,
+      "step": 57530
+    },
+    {
+      "epoch": 0.5275511139635096,
+      "learning_rate": 9.449161089208766e-05,
+      "loss": 0.8208,
+      "mean_token_accuracy": 0.7512878775596619,
+      "num_tokens": 133025584.0,
+      "step": 57540
+    },
+    {
+      "epoch": 0.5276427982029889,
+      "learning_rate": 9.44732740441918e-05,
+      "loss": 0.8079,
+      "mean_token_accuracy": 0.7574107646942139,
+      "num_tokens": 133048385.0,
+      "step": 57550
+    },
+    {
+      "epoch": 0.5277344824424681,
+      "learning_rate": 9.445493719629595e-05,
+      "loss": 0.8154,
+      "mean_token_accuracy": 0.7554541051387786,
+      "num_tokens": 133072294.0,
+      "step": 57560
+    },
+    {
+      "epoch": 0.5278261666819474,
+      "learning_rate": 9.443660034840012e-05,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7651211798191071,
+      "num_tokens": 133095880.0,
+      "step": 57570
+    },
+    {
+      "epoch": 0.5279178509214266,
+      "learning_rate": 9.441826350050427e-05,
+      "loss": 0.8231,
+      "mean_token_accuracy": 0.7492182612419128,
+      "num_tokens": 133120385.0,
+      "step": 57580
+    },
+    {
+      "epoch": 0.5280095351609059,
+      "learning_rate": 9.439992665260842e-05,
+      "loss": 0.7875,
+      "mean_token_accuracy": 0.7622422754764557,
+      "num_tokens": 133144058.0,
+      "step": 57590
+    },
+    {
+      "epoch": 0.5281012194003851,
+      "learning_rate": 9.438158980471257e-05,
+      "loss": 0.7835,
+      "mean_token_accuracy": 0.7637467741966247,
+      "num_tokens": 133167657.0,
+      "step": 57600
+    },
+    {
+      "epoch": 0.5281929036398643,
+      "learning_rate": 9.436325295681673e-05,
+      "loss": 0.8643,
+      "mean_token_accuracy": 0.7401656031608581,
+      "num_tokens": 133192252.0,
+      "step": 57610
+    },
+    {
+      "epoch": 0.5282845878793435,
+      "learning_rate": 9.434491610892088e-05,
+      "loss": 0.8026,
+      "mean_token_accuracy": 0.756588363647461,
+      "num_tokens": 133216402.0,
+      "step": 57620
+    },
+    {
+      "epoch": 0.5283762721188228,
+      "learning_rate": 9.432657926102504e-05,
+      "loss": 0.7983,
+      "mean_token_accuracy": 0.763013631105423,
+      "num_tokens": 133239776.0,
+      "step": 57630
+    },
+    {
+      "epoch": 0.528467956358302,
+      "learning_rate": 9.43082424131292e-05,
+      "loss": 0.8524,
+      "mean_token_accuracy": 0.7439938604831695,
+      "num_tokens": 133262456.0,
+      "step": 57640
+    },
+    {
+      "epoch": 0.5285596405977813,
+      "learning_rate": 9.428990556523333e-05,
+      "loss": 0.7994,
+      "mean_token_accuracy": 0.7561735332012176,
+      "num_tokens": 133285602.0,
+      "step": 57650
+    },
+    {
+      "epoch": 0.5286513248372605,
+      "learning_rate": 9.42715687173375e-05,
+      "loss": 0.7467,
+      "mean_token_accuracy": 0.7689444124698639,
+      "num_tokens": 133309649.0,
+      "step": 57660
+    },
+    {
+      "epoch": 0.5287430090767397,
+      "learning_rate": 9.425323186944165e-05,
+      "loss": 0.7982,
+      "mean_token_accuracy": 0.7575224041938782,
+      "num_tokens": 133333004.0,
+      "step": 57670
+    },
+    {
+      "epoch": 0.5288346933162189,
+      "learning_rate": 9.42348950215458e-05,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7395725250244141,
+      "num_tokens": 133356516.0,
+      "step": 57680
+    },
+    {
+      "epoch": 0.5289263775556982,
+      "learning_rate": 9.421655817364995e-05,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.7581106424331665,
+      "num_tokens": 133381215.0,
+      "step": 57690
+    },
+    {
+      "epoch": 0.5290180617951774,
+      "learning_rate": 9.41982213257541e-05,
+      "loss": 0.805,
+      "mean_token_accuracy": 0.7568289577960968,
+      "num_tokens": 133405564.0,
+      "step": 57700
+    },
+    {
+      "epoch": 0.5291097460346567,
+      "learning_rate": 9.417988447785826e-05,
+      "loss": 0.8248,
+      "mean_token_accuracy": 0.7545980334281921,
+      "num_tokens": 133427933.0,
+      "step": 57710
+    },
+    {
+      "epoch": 0.5292014302741359,
+      "learning_rate": 9.416154762996241e-05,
+      "loss": 0.8056,
+      "mean_token_accuracy": 0.7599722623825074,
+      "num_tokens": 133451518.0,
+      "step": 57720
+    },
+    {
+      "epoch": 0.5292931145136152,
+      "learning_rate": 9.414321078206657e-05,
+      "loss": 0.8205,
+      "mean_token_accuracy": 0.7493108093738556,
+      "num_tokens": 133475683.0,
+      "step": 57730
+    },
+    {
+      "epoch": 0.5293847987530943,
+      "learning_rate": 9.412487393417073e-05,
+      "loss": 0.7904,
+      "mean_token_accuracy": 0.7551456451416015,
+      "num_tokens": 133498559.0,
+      "step": 57740
+    },
+    {
+      "epoch": 0.5294764829925735,
+      "learning_rate": 9.410653708627486e-05,
+      "loss": 0.7369,
+      "mean_token_accuracy": 0.7759769082069397,
+      "num_tokens": 133522020.0,
+      "step": 57750
+    },
+    {
+      "epoch": 0.5295681672320528,
+      "learning_rate": 9.408820023837903e-05,
+      "loss": 0.7993,
+      "mean_token_accuracy": 0.7615691244602203,
+      "num_tokens": 133545112.0,
+      "step": 57760
+    },
+    {
+      "epoch": 0.529659851471532,
+      "learning_rate": 9.406986339048318e-05,
+      "loss": 0.8141,
+      "mean_token_accuracy": 0.7508325040340423,
+      "num_tokens": 133569653.0,
+      "step": 57770
+    },
+    {
+      "epoch": 0.5297515357110113,
+      "learning_rate": 9.405152654258733e-05,
+      "loss": 0.7997,
+      "mean_token_accuracy": 0.7645523369312286,
+      "num_tokens": 133593225.0,
+      "step": 57780
+    },
+    {
+      "epoch": 0.5298432199504906,
+      "learning_rate": 9.403318969469149e-05,
+      "loss": 0.8089,
+      "mean_token_accuracy": 0.7540140330791474,
+      "num_tokens": 133617163.0,
+      "step": 57790
+    },
+    {
+      "epoch": 0.5299349041899697,
+      "learning_rate": 9.401485284679564e-05,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.756240439414978,
+      "num_tokens": 133641186.0,
+      "step": 57800
+    },
+    {
+      "epoch": 0.5300265884294489,
+      "learning_rate": 9.399651599889979e-05,
+      "loss": 0.798,
+      "mean_token_accuracy": 0.7593181669712067,
+      "num_tokens": 133664755.0,
+      "step": 57810
+    },
+    {
+      "epoch": 0.5301182726689282,
+      "learning_rate": 9.397817915100394e-05,
+      "loss": 0.789,
+      "mean_token_accuracy": 0.7571062684059143,
+      "num_tokens": 133688104.0,
+      "step": 57820
+    },
+    {
+      "epoch": 0.5302099569084074,
+      "learning_rate": 9.395984230310811e-05,
+      "loss": 0.8109,
+      "mean_token_accuracy": 0.7545394539833069,
+      "num_tokens": 133712739.0,
+      "step": 57830
+    },
+    {
+      "epoch": 0.5303016411478867,
+      "learning_rate": 9.394150545521226e-05,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.7724260568618775,
+      "num_tokens": 133737768.0,
+      "step": 57840
+    },
+    {
+      "epoch": 0.5303933253873659,
+      "learning_rate": 9.39231686073164e-05,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7640564382076264,
+      "num_tokens": 133761290.0,
+      "step": 57850
+    },
+    {
+      "epoch": 0.5304850096268452,
+      "learning_rate": 9.390483175942056e-05,
+      "loss": 0.7987,
+      "mean_token_accuracy": 0.757305258512497,
+      "num_tokens": 133785048.0,
+      "step": 57860
+    },
+    {
+      "epoch": 0.5305766938663243,
+      "learning_rate": 9.388649491152471e-05,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7601220607757568,
+      "num_tokens": 133808910.0,
+      "step": 57870
+    },
+    {
+      "epoch": 0.5306683781058036,
+      "learning_rate": 9.386815806362887e-05,
+      "loss": 0.7884,
+      "mean_token_accuracy": 0.76932772397995,
+      "num_tokens": 133832965.0,
+      "step": 57880
+    },
+    {
+      "epoch": 0.5307600623452828,
+      "learning_rate": 9.384982121573302e-05,
+      "loss": 0.8328,
+      "mean_token_accuracy": 0.7507218062877655,
+      "num_tokens": 133856049.0,
+      "step": 57890
+    },
+    {
+      "epoch": 0.5308517465847621,
+      "learning_rate": 9.383148436783717e-05,
+      "loss": 0.7905,
+      "mean_token_accuracy": 0.7680697858333587,
+      "num_tokens": 133880539.0,
+      "step": 57900
+    },
+    {
+      "epoch": 0.5309434308242413,
+      "learning_rate": 9.381314751994132e-05,
+      "loss": 0.792,
+      "mean_token_accuracy": 0.7629115223884583,
+      "num_tokens": 133904651.0,
+      "step": 57910
+    },
+    {
+      "epoch": 0.5310351150637206,
+      "learning_rate": 9.379481067204547e-05,
+      "loss": 0.828,
+      "mean_token_accuracy": 0.7504708647727967,
+      "num_tokens": 133928674.0,
+      "step": 57920
+    },
+    {
+      "epoch": 0.5311267993031997,
+      "learning_rate": 9.377647382414964e-05,
+      "loss": 0.7775,
+      "mean_token_accuracy": 0.7598287761211395,
+      "num_tokens": 133952675.0,
+      "step": 57930
+    },
+    {
+      "epoch": 0.531218483542679,
+      "learning_rate": 9.375813697625379e-05,
+      "loss": 0.7971,
+      "mean_token_accuracy": 0.7638172924518585,
+      "num_tokens": 133976567.0,
+      "step": 57940
+    },
+    {
+      "epoch": 0.5313101677821582,
+      "learning_rate": 9.373980012835793e-05,
+      "loss": 0.7929,
+      "mean_token_accuracy": 0.7637137115001679,
+      "num_tokens": 134000416.0,
+      "step": 57950
+    },
+    {
+      "epoch": 0.5314018520216375,
+      "learning_rate": 9.37214632804621e-05,
+      "loss": 0.8031,
+      "mean_token_accuracy": 0.7616167545318604,
+      "num_tokens": 134024444.0,
+      "step": 57960
+    },
+    {
+      "epoch": 0.5314935362611167,
+      "learning_rate": 9.370312643256625e-05,
+      "loss": 0.8212,
+      "mean_token_accuracy": 0.749068558216095,
+      "num_tokens": 134047944.0,
+      "step": 57970
+    },
+    {
+      "epoch": 0.531585220500596,
+      "learning_rate": 9.36847895846704e-05,
+      "loss": 0.8304,
+      "mean_token_accuracy": 0.752483582496643,
+      "num_tokens": 134071264.0,
+      "step": 57980
+    },
+    {
+      "epoch": 0.5316769047400752,
+      "learning_rate": 9.366645273677455e-05,
+      "loss": 0.7875,
+      "mean_token_accuracy": 0.7603633165359497,
+      "num_tokens": 134095233.0,
+      "step": 57990
+    },
+    {
+      "epoch": 0.5317685889795544,
+      "learning_rate": 9.36481158888787e-05,
+      "loss": 0.8491,
+      "mean_token_accuracy": 0.7498716950416565,
+      "num_tokens": 134119169.0,
+      "step": 58000
+    },
+    {
+      "epoch": 0.5318602732190336,
+      "learning_rate": 9.362977904098285e-05,
+      "loss": 0.8036,
+      "mean_token_accuracy": 0.7566041052341461,
+      "num_tokens": 134142485.0,
+      "step": 58010
+    },
+    {
+      "epoch": 0.5319519574585129,
+      "learning_rate": 9.361144219308702e-05,
+      "loss": 0.7884,
+      "mean_token_accuracy": 0.7657165229320526,
+      "num_tokens": 134167175.0,
+      "step": 58020
+    },
+    {
+      "epoch": 0.5320436416979921,
+      "learning_rate": 9.359310534519117e-05,
+      "loss": 0.7991,
+      "mean_token_accuracy": 0.7578857183456421,
+      "num_tokens": 134190810.0,
+      "step": 58030
+    },
+    {
+      "epoch": 0.5321353259374714,
+      "learning_rate": 9.357476849729532e-05,
+      "loss": 0.7823,
+      "mean_token_accuracy": 0.7648557901382447,
+      "num_tokens": 134215342.0,
+      "step": 58040
+    },
+    {
+      "epoch": 0.5322270101769506,
+      "learning_rate": 9.355643164939946e-05,
+      "loss": 0.8028,
+      "mean_token_accuracy": 0.7602374613285064,
+      "num_tokens": 134239388.0,
+      "step": 58050
+    },
+    {
+      "epoch": 0.5323186944164298,
+      "learning_rate": 9.353809480150363e-05,
+      "loss": 0.7839,
+      "mean_token_accuracy": 0.7690163731575013,
+      "num_tokens": 134262982.0,
+      "step": 58060
+    },
+    {
+      "epoch": 0.532410378655909,
+      "learning_rate": 9.351975795360778e-05,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7536212086677552,
+      "num_tokens": 134287338.0,
+      "step": 58070
+    },
+    {
+      "epoch": 0.5325020628953883,
+      "learning_rate": 9.350142110571193e-05,
+      "loss": 0.7853,
+      "mean_token_accuracy": 0.7642540335655212,
+      "num_tokens": 134311080.0,
+      "step": 58080
+    },
+    {
+      "epoch": 0.5325937471348675,
+      "learning_rate": 9.348308425781608e-05,
+      "loss": 0.7322,
+      "mean_token_accuracy": 0.769903838634491,
+      "num_tokens": 134334833.0,
+      "step": 58090
+    },
+    {
+      "epoch": 0.5326854313743468,
+      "learning_rate": 9.346474740992023e-05,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.7454536139965058,
+      "num_tokens": 134358283.0,
+      "step": 58100
+    },
+    {
+      "epoch": 0.532777115613826,
+      "learning_rate": 9.344641056202439e-05,
+      "loss": 0.8399,
+      "mean_token_accuracy": 0.7469364881515503,
+      "num_tokens": 134381677.0,
+      "step": 58110
+    },
+    {
+      "epoch": 0.5328687998533053,
+      "learning_rate": 9.342807371412855e-05,
+      "loss": 0.7936,
+      "mean_token_accuracy": 0.7565255045890809,
+      "num_tokens": 134405319.0,
+      "step": 58120
+    },
+    {
+      "epoch": 0.5329604840927844,
+      "learning_rate": 9.34097368662327e-05,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.7440931677818299,
+      "num_tokens": 134429221.0,
+      "step": 58130
+    },
+    {
+      "epoch": 0.5330521683322637,
+      "learning_rate": 9.339140001833686e-05,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7703822612762451,
+      "num_tokens": 134454037.0,
+      "step": 58140
+    },
+    {
+      "epoch": 0.5331438525717429,
+      "learning_rate": 9.337306317044101e-05,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7670372784137726,
+      "num_tokens": 134478373.0,
+      "step": 58150
+    },
+    {
+      "epoch": 0.5332355368112222,
+      "learning_rate": 9.335472632254516e-05,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7652393400669097,
+      "num_tokens": 134502345.0,
+      "step": 58160
+    },
+    {
+      "epoch": 0.5333272210507014,
+      "learning_rate": 9.333638947464931e-05,
+      "loss": 0.8045,
+      "mean_token_accuracy": 0.7603592276573181,
+      "num_tokens": 134526756.0,
+      "step": 58170
+    },
+    {
+      "epoch": 0.5334189052901807,
+      "learning_rate": 9.331805262675346e-05,
+      "loss": 0.7985,
+      "mean_token_accuracy": 0.7629460453987121,
+      "num_tokens": 134551619.0,
+      "step": 58180
+    },
+    {
+      "epoch": 0.5335105895296599,
+      "learning_rate": 9.329971577885761e-05,
+      "loss": 0.7833,
+      "mean_token_accuracy": 0.7612346410751343,
+      "num_tokens": 134575385.0,
+      "step": 58190
+    },
+    {
+      "epoch": 0.533602273769139,
+      "learning_rate": 9.328137893096177e-05,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.766508835554123,
+      "num_tokens": 134599990.0,
+      "step": 58200
+    },
+    {
+      "epoch": 0.5336939580086183,
+      "learning_rate": 9.326304208306592e-05,
+      "loss": 0.7578,
+      "mean_token_accuracy": 0.7712429940700531,
+      "num_tokens": 134623671.0,
+      "step": 58210
+    },
+    {
+      "epoch": 0.5337856422480975,
+      "learning_rate": 9.324470523517008e-05,
+      "loss": 0.8112,
+      "mean_token_accuracy": 0.7547142028808593,
+      "num_tokens": 134647528.0,
+      "step": 58220
+    },
+    {
+      "epoch": 0.5338773264875768,
+      "learning_rate": 9.322636838727424e-05,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7550437688827515,
+      "num_tokens": 134671590.0,
+      "step": 58230
+    },
+    {
+      "epoch": 0.533969010727056,
+      "learning_rate": 9.320803153937839e-05,
+      "loss": 0.833,
+      "mean_token_accuracy": 0.7531205773353576,
+      "num_tokens": 134694681.0,
+      "step": 58240
+    },
+    {
+      "epoch": 0.5340606949665353,
+      "learning_rate": 9.318969469148254e-05,
+      "loss": 0.8013,
+      "mean_token_accuracy": 0.7594623863697052,
+      "num_tokens": 134718792.0,
+      "step": 58250
+    },
+    {
+      "epoch": 0.5341523792060144,
+      "learning_rate": 9.317135784358669e-05,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.7689113557338715,
+      "num_tokens": 134742825.0,
+      "step": 58260
+    },
+    {
+      "epoch": 0.5342440634454937,
+      "learning_rate": 9.315302099569084e-05,
+      "loss": 0.8206,
+      "mean_token_accuracy": 0.7558790862560272,
+      "num_tokens": 134765833.0,
+      "step": 58270
+    },
+    {
+      "epoch": 0.5343357476849729,
+      "learning_rate": 9.313468414779501e-05,
+      "loss": 0.8419,
+      "mean_token_accuracy": 0.7577371656894684,
+      "num_tokens": 134789050.0,
+      "step": 58280
+    },
+    {
+      "epoch": 0.5344274319244522,
+      "learning_rate": 9.311634729989915e-05,
+      "loss": 0.7838,
+      "mean_token_accuracy": 0.7634777903556824,
+      "num_tokens": 134813759.0,
+      "step": 58290
+    },
+    {
+      "epoch": 0.5345191161639314,
+      "learning_rate": 9.30980104520033e-05,
+      "loss": 0.8025,
+      "mean_token_accuracy": 0.7581336975097657,
+      "num_tokens": 134836989.0,
+      "step": 58300
+    },
+    {
+      "epoch": 0.5346108004034107,
+      "learning_rate": 9.307967360410745e-05,
+      "loss": 0.817,
+      "mean_token_accuracy": 0.7554428398609161,
+      "num_tokens": 134860209.0,
+      "step": 58310
+    },
+    {
+      "epoch": 0.5347024846428899,
+      "learning_rate": 9.306133675621162e-05,
+      "loss": 0.7948,
+      "mean_token_accuracy": 0.7627366423606873,
+      "num_tokens": 134884291.0,
+      "step": 58320
+    },
+    {
+      "epoch": 0.5347941688823691,
+      "learning_rate": 9.304299990831577e-05,
+      "loss": 0.8063,
+      "mean_token_accuracy": 0.7612611770629882,
+      "num_tokens": 134908434.0,
+      "step": 58330
+    },
+    {
+      "epoch": 0.5348858531218483,
+      "learning_rate": 9.302466306041992e-05,
+      "loss": 0.792,
+      "mean_token_accuracy": 0.7629269242286683,
+      "num_tokens": 134931790.0,
+      "step": 58340
+    },
+    {
+      "epoch": 0.5349775373613276,
+      "learning_rate": 9.300632621252407e-05,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.7589909791946411,
+      "num_tokens": 134955206.0,
+      "step": 58350
+    },
+    {
+      "epoch": 0.5350692216008068,
+      "learning_rate": 9.298798936462822e-05,
+      "loss": 0.7313,
+      "mean_token_accuracy": 0.7748064637184143,
+      "num_tokens": 134978977.0,
+      "step": 58360
+    },
+    {
+      "epoch": 0.5351609058402861,
+      "learning_rate": 9.296965251673238e-05,
+      "loss": 0.775,
+      "mean_token_accuracy": 0.7637354612350464,
+      "num_tokens": 135003333.0,
+      "step": 58370
+    },
+    {
+      "epoch": 0.5352525900797653,
+      "learning_rate": 9.295131566883654e-05,
+      "loss": 0.8168,
+      "mean_token_accuracy": 0.7580026090145111,
+      "num_tokens": 135026926.0,
+      "step": 58380
+    },
+    {
+      "epoch": 0.5353442743192445,
+      "learning_rate": 9.293297882094068e-05,
+      "loss": 0.7187,
+      "mean_token_accuracy": 0.774970269203186,
+      "num_tokens": 135050773.0,
+      "step": 58390
+    },
+    {
+      "epoch": 0.5354359585587237,
+      "learning_rate": 9.291464197304483e-05,
+      "loss": 0.784,
+      "mean_token_accuracy": 0.7690203487873077,
+      "num_tokens": 135074446.0,
+      "step": 58400
+    },
+    {
+      "epoch": 0.535527642798203,
+      "learning_rate": 9.2896305125149e-05,
+      "loss": 0.7838,
+      "mean_token_accuracy": 0.7571030974388122,
+      "num_tokens": 135098895.0,
+      "step": 58410
+    },
+    {
+      "epoch": 0.5356193270376822,
+      "learning_rate": 9.287796827725315e-05,
+      "loss": 0.7929,
+      "mean_token_accuracy": 0.7633040070533752,
+      "num_tokens": 135122402.0,
+      "step": 58420
+    },
+    {
+      "epoch": 0.5357110112771615,
+      "learning_rate": 9.28596314293573e-05,
+      "loss": 0.7882,
+      "mean_token_accuracy": 0.7635120272636413,
+      "num_tokens": 135145957.0,
+      "step": 58430
+    },
+    {
+      "epoch": 0.5358026955166407,
+      "learning_rate": 9.284129458146145e-05,
+      "loss": 0.7675,
+      "mean_token_accuracy": 0.7690262854099273,
+      "num_tokens": 135170375.0,
+      "step": 58440
+    },
+    {
+      "epoch": 0.53589437975612,
+      "learning_rate": 9.28229577335656e-05,
+      "loss": 0.7844,
+      "mean_token_accuracy": 0.7598381519317627,
+      "num_tokens": 135193863.0,
+      "step": 58450
+    },
+    {
+      "epoch": 0.5359860639955991,
+      "learning_rate": 9.280462088566976e-05,
+      "loss": 0.7855,
+      "mean_token_accuracy": 0.7660774469375611,
+      "num_tokens": 135218057.0,
+      "step": 58460
+    },
+    {
+      "epoch": 0.5360777482350784,
+      "learning_rate": 9.278628403777391e-05,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.7679777979850769,
+      "num_tokens": 135242161.0,
+      "step": 58470
+    },
+    {
+      "epoch": 0.5361694324745576,
+      "learning_rate": 9.276794718987807e-05,
+      "loss": 0.7931,
+      "mean_token_accuracy": 0.7602312386035919,
+      "num_tokens": 135266361.0,
+      "step": 58480
+    },
+    {
+      "epoch": 0.5362611167140369,
+      "learning_rate": 9.274961034198221e-05,
+      "loss": 0.7761,
+      "mean_token_accuracy": 0.7651449203491211,
+      "num_tokens": 135290715.0,
+      "step": 58490
+    },
+    {
+      "epoch": 0.5363528009535161,
+      "learning_rate": 9.273127349408636e-05,
+      "loss": 0.7851,
+      "mean_token_accuracy": 0.7679320812225342,
+      "num_tokens": 135315508.0,
+      "step": 58500
+    },
+    {
+      "epoch": 0.5364444851929954,
+      "learning_rate": 9.271293664619053e-05,
+      "loss": 0.7975,
+      "mean_token_accuracy": 0.7625534296035766,
+      "num_tokens": 135340418.0,
+      "step": 58510
+    },
+    {
+      "epoch": 0.5365361694324745,
+      "learning_rate": 9.269459979829468e-05,
+      "loss": 0.7957,
+      "mean_token_accuracy": 0.756694632768631,
+      "num_tokens": 135364927.0,
+      "step": 58520
+    },
+    {
+      "epoch": 0.5366278536719538,
+      "learning_rate": 9.267626295039883e-05,
+      "loss": 0.7727,
+      "mean_token_accuracy": 0.7667570650577545,
+      "num_tokens": 135389743.0,
+      "step": 58530
+    },
+    {
+      "epoch": 0.536719537911433,
+      "learning_rate": 9.265792610250298e-05,
+      "loss": 0.7956,
+      "mean_token_accuracy": 0.7599010944366456,
+      "num_tokens": 135413486.0,
+      "step": 58540
+    },
+    {
+      "epoch": 0.5368112221509123,
+      "learning_rate": 9.263958925460714e-05,
+      "loss": 0.72,
+      "mean_token_accuracy": 0.7795994818210602,
+      "num_tokens": 135439216.0,
+      "step": 58550
+    },
+    {
+      "epoch": 0.5369029063903915,
+      "learning_rate": 9.262125240671129e-05,
+      "loss": 0.7862,
+      "mean_token_accuracy": 0.7627681434154511,
+      "num_tokens": 135463329.0,
+      "step": 58560
+    },
+    {
+      "epoch": 0.5369945906298708,
+      "learning_rate": 9.260291555881544e-05,
+      "loss": 0.818,
+      "mean_token_accuracy": 0.7461243689060211,
+      "num_tokens": 135487244.0,
+      "step": 58570
+    },
+    {
+      "epoch": 0.53708627486935,
+      "learning_rate": 9.25845787109196e-05,
+      "loss": 0.7813,
+      "mean_token_accuracy": 0.7591448843479156,
+      "num_tokens": 135511644.0,
+      "step": 58580
+    },
+    {
+      "epoch": 0.5371779591088292,
+      "learning_rate": 9.256624186302374e-05,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.7616844713687897,
+      "num_tokens": 135536113.0,
+      "step": 58590
+    },
+    {
+      "epoch": 0.5372696433483084,
+      "learning_rate": 9.25479050151279e-05,
+      "loss": 0.7709,
+      "mean_token_accuracy": 0.765330970287323,
+      "num_tokens": 135560046.0,
+      "step": 58600
+    },
+    {
+      "epoch": 0.5373613275877877,
+      "learning_rate": 9.252956816723206e-05,
+      "loss": 0.7878,
+      "mean_token_accuracy": 0.7632694005966186,
+      "num_tokens": 135584148.0,
+      "step": 58610
+    },
+    {
+      "epoch": 0.5374530118272669,
+      "learning_rate": 9.251123131933621e-05,
+      "loss": 0.7176,
+      "mean_token_accuracy": 0.7732840240001678,
+      "num_tokens": 135609022.0,
+      "step": 58620
+    },
+    {
+      "epoch": 0.5375446960667462,
+      "learning_rate": 9.249289447144036e-05,
+      "loss": 0.7673,
+      "mean_token_accuracy": 0.7693818151950836,
+      "num_tokens": 135632757.0,
+      "step": 58630
+    },
+    {
+      "epoch": 0.5376363803062254,
+      "learning_rate": 9.247455762354452e-05,
+      "loss": 0.8176,
+      "mean_token_accuracy": 0.756648164987564,
+      "num_tokens": 135657576.0,
+      "step": 58640
+    },
+    {
+      "epoch": 0.5377280645457045,
+      "learning_rate": 9.245622077564867e-05,
+      "loss": 0.819,
+      "mean_token_accuracy": 0.7565523028373718,
+      "num_tokens": 135682154.0,
+      "step": 58650
+    },
+    {
+      "epoch": 0.5378197487851838,
+      "learning_rate": 9.243788392775282e-05,
+      "loss": 0.8146,
+      "mean_token_accuracy": 0.7552489995956421,
+      "num_tokens": 135706285.0,
+      "step": 58660
+    },
+    {
+      "epoch": 0.537911433024663,
+      "learning_rate": 9.241954707985697e-05,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7434813380241394,
+      "num_tokens": 135729514.0,
+      "step": 58670
+    },
+    {
+      "epoch": 0.5380031172641423,
+      "learning_rate": 9.240121023196114e-05,
+      "loss": 0.7479,
+      "mean_token_accuracy": 0.7707680761814117,
+      "num_tokens": 135753859.0,
+      "step": 58680
+    },
+    {
+      "epoch": 0.5380948015036215,
+      "learning_rate": 9.238287338406528e-05,
+      "loss": 0.7827,
+      "mean_token_accuracy": 0.7675798773765564,
+      "num_tokens": 135778482.0,
+      "step": 58690
+    },
+    {
+      "epoch": 0.5381864857431008,
+      "learning_rate": 9.236453653616943e-05,
+      "loss": 0.7784,
+      "mean_token_accuracy": 0.7681465446949005,
+      "num_tokens": 135802610.0,
+      "step": 58700
+    },
+    {
+      "epoch": 0.53827816998258,
+      "learning_rate": 9.23461996882736e-05,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.7572859346866607,
+      "num_tokens": 135826238.0,
+      "step": 58710
+    },
+    {
+      "epoch": 0.5383698542220592,
+      "learning_rate": 9.232786284037775e-05,
+      "loss": 0.8454,
+      "mean_token_accuracy": 0.7488239765167236,
+      "num_tokens": 135849888.0,
+      "step": 58720
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "learning_rate": 9.23095259924819e-05,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7731451570987702,
+      "num_tokens": 135874611.0,
+      "step": 58730
+    },
+    {
+      "epoch": 0.5385532227010177,
+      "learning_rate": 9.229118914458605e-05,
+      "loss": 0.762,
+      "mean_token_accuracy": 0.7702429115772247,
+      "num_tokens": 135899097.0,
+      "step": 58740
+    },
+    {
+      "epoch": 0.5386449069404969,
+      "learning_rate": 9.22728522966902e-05,
+      "loss": 0.8045,
+      "mean_token_accuracy": 0.7573400974273682,
+      "num_tokens": 135923927.0,
+      "step": 58750
+    },
+    {
+      "epoch": 0.5387365911799762,
+      "learning_rate": 9.225451544879435e-05,
+      "loss": 0.8197,
+      "mean_token_accuracy": 0.7608265519142151,
+      "num_tokens": 135948051.0,
+      "step": 58760
+    },
+    {
+      "epoch": 0.5388282754194554,
+      "learning_rate": 9.223617860089852e-05,
+      "loss": 0.7554,
+      "mean_token_accuracy": 0.7739845752716065,
+      "num_tokens": 135972671.0,
+      "step": 58770
+    },
+    {
+      "epoch": 0.5389199596589346,
+      "learning_rate": 9.221784175300267e-05,
+      "loss": 0.7566,
+      "mean_token_accuracy": 0.7717774510383606,
+      "num_tokens": 135998399.0,
+      "step": 58780
+    },
+    {
+      "epoch": 0.5390116438984138,
+      "learning_rate": 9.219950490510681e-05,
+      "loss": 0.7871,
+      "mean_token_accuracy": 0.7645931363105773,
+      "num_tokens": 136022794.0,
+      "step": 58790
+    },
+    {
+      "epoch": 0.5391033281378931,
+      "learning_rate": 9.218116805721096e-05,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7741548418998718,
+      "num_tokens": 136047505.0,
+      "step": 58800
+    },
+    {
+      "epoch": 0.5391950123773723,
+      "learning_rate": 9.216283120931513e-05,
+      "loss": 0.7894,
+      "mean_token_accuracy": 0.7640251398086548,
+      "num_tokens": 136071039.0,
+      "step": 58810
+    },
+    {
+      "epoch": 0.5392866966168516,
+      "learning_rate": 9.214449436141928e-05,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7764576375484467,
+      "num_tokens": 136095555.0,
+      "step": 58820
+    },
+    {
+      "epoch": 0.5393783808563308,
+      "learning_rate": 9.212615751352343e-05,
+      "loss": 0.7591,
+      "mean_token_accuracy": 0.7665407061576843,
+      "num_tokens": 136119708.0,
+      "step": 58830
+    },
+    {
+      "epoch": 0.5394700650958101,
+      "learning_rate": 9.210782066562758e-05,
+      "loss": 0.803,
+      "mean_token_accuracy": 0.7570671677589417,
+      "num_tokens": 136143469.0,
+      "step": 58840
+    },
+    {
+      "epoch": 0.5395617493352892,
+      "learning_rate": 9.208948381773173e-05,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.7553102552890778,
+      "num_tokens": 136165962.0,
+      "step": 58850
+    },
+    {
+      "epoch": 0.5396534335747685,
+      "learning_rate": 9.207114696983588e-05,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.767716646194458,
+      "num_tokens": 136189494.0,
+      "step": 58860
+    },
+    {
+      "epoch": 0.5397451178142477,
+      "learning_rate": 9.205281012194005e-05,
+      "loss": 0.7714,
+      "mean_token_accuracy": 0.7691974937915802,
+      "num_tokens": 136213295.0,
+      "step": 58870
+    },
+    {
+      "epoch": 0.539836802053727,
+      "learning_rate": 9.20344732740442e-05,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7664900839328765,
+      "num_tokens": 136235677.0,
+      "step": 58880
+    },
+    {
+      "epoch": 0.5399284862932062,
+      "learning_rate": 9.201613642614834e-05,
+      "loss": 0.78,
+      "mean_token_accuracy": 0.7662824213504791,
+      "num_tokens": 136259347.0,
+      "step": 58890
+    },
+    {
+      "epoch": 0.5400201705326855,
+      "learning_rate": 9.19977995782525e-05,
+      "loss": 0.8049,
+      "mean_token_accuracy": 0.7642119467258454,
+      "num_tokens": 136284294.0,
+      "step": 58900
+    },
+    {
+      "epoch": 0.5401118547721646,
+      "learning_rate": 9.197946273035666e-05,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7709306061267853,
+      "num_tokens": 136308767.0,
+      "step": 58910
+    },
+    {
+      "epoch": 0.5402035390116439,
+      "learning_rate": 9.196112588246081e-05,
+      "loss": 0.7474,
+      "mean_token_accuracy": 0.7727955460548401,
+      "num_tokens": 136332794.0,
+      "step": 58920
+    },
+    {
+      "epoch": 0.5402952232511231,
+      "learning_rate": 9.194278903456496e-05,
+      "loss": 0.7869,
+      "mean_token_accuracy": 0.7615944862365722,
+      "num_tokens": 136356879.0,
+      "step": 58930
+    },
+    {
+      "epoch": 0.5403869074906024,
+      "learning_rate": 9.192445218666911e-05,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.7636140763759613,
+      "num_tokens": 136380377.0,
+      "step": 58940
+    },
+    {
+      "epoch": 0.5404785917300816,
+      "learning_rate": 9.190611533877327e-05,
+      "loss": 0.804,
+      "mean_token_accuracy": 0.7576845228672028,
+      "num_tokens": 136404768.0,
+      "step": 58950
+    },
+    {
+      "epoch": 0.5405702759695609,
+      "learning_rate": 9.188777849087742e-05,
+      "loss": 0.8015,
+      "mean_token_accuracy": 0.7631287217140198,
+      "num_tokens": 136429418.0,
+      "step": 58960
+    },
+    {
+      "epoch": 0.5406619602090401,
+      "learning_rate": 9.186944164298158e-05,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.760073584318161,
+      "num_tokens": 136453803.0,
+      "step": 58970
+    },
+    {
+      "epoch": 0.5407536444485193,
+      "learning_rate": 9.185110479508573e-05,
+      "loss": 0.7551,
+      "mean_token_accuracy": 0.7727975487709046,
+      "num_tokens": 136479297.0,
+      "step": 58980
+    },
+    {
+      "epoch": 0.5408453286879985,
+      "learning_rate": 9.183276794718987e-05,
+      "loss": 0.7564,
+      "mean_token_accuracy": 0.7678837537765503,
+      "num_tokens": 136503667.0,
+      "step": 58990
+    },
+    {
+      "epoch": 0.5409370129274778,
+      "learning_rate": 9.181443109929404e-05,
+      "loss": 0.7837,
+      "mean_token_accuracy": 0.7612081289291381,
+      "num_tokens": 136527150.0,
+      "step": 59000
+    },
+    {
+      "epoch": 0.541028697166957,
+      "learning_rate": 9.179609425139819e-05,
+      "loss": 0.8105,
+      "mean_token_accuracy": 0.7585221111774445,
+      "num_tokens": 136550805.0,
+      "step": 59010
+    },
+    {
+      "epoch": 0.5411203814064363,
+      "learning_rate": 9.177775740350234e-05,
+      "loss": 0.7932,
+      "mean_token_accuracy": 0.7661087095737458,
+      "num_tokens": 136575553.0,
+      "step": 59020
+    },
+    {
+      "epoch": 0.5412120656459155,
+      "learning_rate": 9.175942055560651e-05,
+      "loss": 0.7622,
+      "mean_token_accuracy": 0.763846355676651,
+      "num_tokens": 136600504.0,
+      "step": 59030
+    },
+    {
+      "epoch": 0.5413037498853946,
+      "learning_rate": 9.174108370771065e-05,
+      "loss": 0.789,
+      "mean_token_accuracy": 0.7613443434238434,
+      "num_tokens": 136624694.0,
+      "step": 59040
+    },
+    {
+      "epoch": 0.5413954341248739,
+      "learning_rate": 9.17227468598148e-05,
+      "loss": 0.7518,
+      "mean_token_accuracy": 0.7728749930858612,
+      "num_tokens": 136648358.0,
+      "step": 59050
+    },
+    {
+      "epoch": 0.5414871183643531,
+      "learning_rate": 9.170441001191895e-05,
+      "loss": 0.8121,
+      "mean_token_accuracy": 0.7570838332176208,
+      "num_tokens": 136672057.0,
+      "step": 59060
+    },
+    {
+      "epoch": 0.5415788026038324,
+      "learning_rate": 9.168607316402311e-05,
+      "loss": 0.7983,
+      "mean_token_accuracy": 0.7637819230556488,
+      "num_tokens": 136695211.0,
+      "step": 59070
+    },
+    {
+      "epoch": 0.5416704868433116,
+      "learning_rate": 9.166773631612727e-05,
+      "loss": 0.7879,
+      "mean_token_accuracy": 0.7602478384971618,
+      "num_tokens": 136718855.0,
+      "step": 59080
+    },
+    {
+      "epoch": 0.5417621710827909,
+      "learning_rate": 9.16493994682314e-05,
+      "loss": 0.761,
+      "mean_token_accuracy": 0.7691253364086151,
+      "num_tokens": 136743036.0,
+      "step": 59090
+    },
+    {
+      "epoch": 0.5418538553222702,
+      "learning_rate": 9.163106262033557e-05,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7667005777359008,
+      "num_tokens": 136767331.0,
+      "step": 59100
+    },
+    {
+      "epoch": 0.5419455395617493,
+      "learning_rate": 9.161272577243972e-05,
+      "loss": 0.8674,
+      "mean_token_accuracy": 0.7464340209960938,
+      "num_tokens": 136790075.0,
+      "step": 59110
+    },
+    {
+      "epoch": 0.5420372238012285,
+      "learning_rate": 9.159438892454387e-05,
+      "loss": 0.832,
+      "mean_token_accuracy": 0.747848105430603,
+      "num_tokens": 136815522.0,
+      "step": 59120
+    },
+    {
+      "epoch": 0.5421289080407078,
+      "learning_rate": 9.157605207664804e-05,
+      "loss": 0.7869,
+      "mean_token_accuracy": 0.763346540927887,
+      "num_tokens": 136839737.0,
+      "step": 59130
+    },
+    {
+      "epoch": 0.542220592280187,
+      "learning_rate": 9.155771522875218e-05,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.7635484337806702,
+      "num_tokens": 136863491.0,
+      "step": 59140
+    },
+    {
+      "epoch": 0.5423122765196663,
+      "learning_rate": 9.153937838085633e-05,
+      "loss": 0.7502,
+      "mean_token_accuracy": 0.7703371524810791,
+      "num_tokens": 136887844.0,
+      "step": 59150
+    },
+    {
+      "epoch": 0.5424039607591455,
+      "learning_rate": 9.152104153296048e-05,
+      "loss": 0.7536,
+      "mean_token_accuracy": 0.770036643743515,
+      "num_tokens": 136912731.0,
+      "step": 59160
+    },
+    {
+      "epoch": 0.5424956449986247,
+      "learning_rate": 9.150270468506465e-05,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.769990760087967,
+      "num_tokens": 136936070.0,
+      "step": 59170
+    },
+    {
+      "epoch": 0.5425873292381039,
+      "learning_rate": 9.14843678371688e-05,
+      "loss": 0.7442,
+      "mean_token_accuracy": 0.773495864868164,
+      "num_tokens": 136959936.0,
+      "step": 59180
+    },
+    {
+      "epoch": 0.5426790134775832,
+      "learning_rate": 9.146603098927294e-05,
+      "loss": 0.7833,
+      "mean_token_accuracy": 0.7637665331363678,
+      "num_tokens": 136983865.0,
+      "step": 59190
+    },
+    {
+      "epoch": 0.5427706977170624,
+      "learning_rate": 9.14476941413771e-05,
+      "loss": 0.7712,
+      "mean_token_accuracy": 0.7694948494434357,
+      "num_tokens": 137007165.0,
+      "step": 59200
+    },
+    {
+      "epoch": 0.5428623819565417,
+      "learning_rate": 9.142935729348125e-05,
+      "loss": 0.793,
+      "mean_token_accuracy": 0.7622173428535461,
+      "num_tokens": 137031800.0,
+      "step": 59210
+    },
+    {
+      "epoch": 0.5429540661960209,
+      "learning_rate": 9.14110204455854e-05,
+      "loss": 0.7354,
+      "mean_token_accuracy": 0.7782780170440674,
+      "num_tokens": 137056100.0,
+      "step": 59220
+    },
+    {
+      "epoch": 0.5430457504355002,
+      "learning_rate": 9.139268359768957e-05,
+      "loss": 0.7661,
+      "mean_token_accuracy": 0.7641168296337127,
+      "num_tokens": 137080410.0,
+      "step": 59230
+    },
+    {
+      "epoch": 0.5431374346749793,
+      "learning_rate": 9.137434674979371e-05,
+      "loss": 0.7607,
+      "mean_token_accuracy": 0.7686898529529571,
+      "num_tokens": 137104366.0,
+      "step": 59240
+    },
+    {
+      "epoch": 0.5432291189144586,
+      "learning_rate": 9.135600990189786e-05,
+      "loss": 0.7683,
+      "mean_token_accuracy": 0.7621020674705505,
+      "num_tokens": 137129134.0,
+      "step": 59250
+    },
+    {
+      "epoch": 0.5433208031539378,
+      "learning_rate": 9.133767305400203e-05,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.7702788650989533,
+      "num_tokens": 137154523.0,
+      "step": 59260
+    },
+    {
+      "epoch": 0.5434124873934171,
+      "learning_rate": 9.131933620610618e-05,
+      "loss": 0.7955,
+      "mean_token_accuracy": 0.7685214102268219,
+      "num_tokens": 137179556.0,
+      "step": 59270
+    },
+    {
+      "epoch": 0.5435041716328963,
+      "learning_rate": 9.130099935821033e-05,
+      "loss": 0.7984,
+      "mean_token_accuracy": 0.7586545467376709,
+      "num_tokens": 137203707.0,
+      "step": 59280
+    },
+    {
+      "epoch": 0.5435958558723756,
+      "learning_rate": 9.128266251031447e-05,
+      "loss": 0.7535,
+      "mean_token_accuracy": 0.7715396046638489,
+      "num_tokens": 137227419.0,
+      "step": 59290
+    },
+    {
+      "epoch": 0.5436875401118547,
+      "learning_rate": 9.126432566241863e-05,
+      "loss": 0.7477,
+      "mean_token_accuracy": 0.7667481362819671,
+      "num_tokens": 137251769.0,
+      "step": 59300
+    },
+    {
+      "epoch": 0.543779224351334,
+      "learning_rate": 9.124598881452279e-05,
+      "loss": 0.7586,
+      "mean_token_accuracy": 0.7621158480644226,
+      "num_tokens": 137276167.0,
+      "step": 59310
+    },
+    {
+      "epoch": 0.5438709085908132,
+      "learning_rate": 9.122765196662694e-05,
+      "loss": 0.7935,
+      "mean_token_accuracy": 0.761205005645752,
+      "num_tokens": 137300360.0,
+      "step": 59320
+    },
+    {
+      "epoch": 0.5439625928302925,
+      "learning_rate": 9.12093151187311e-05,
+      "loss": 0.7793,
+      "mean_token_accuracy": 0.7672877252101898,
+      "num_tokens": 137325123.0,
+      "step": 59330
+    },
+    {
+      "epoch": 0.5440542770697717,
+      "learning_rate": 9.119097827083524e-05,
+      "loss": 0.7474,
+      "mean_token_accuracy": 0.773924195766449,
+      "num_tokens": 137349627.0,
+      "step": 59340
+    },
+    {
+      "epoch": 0.544145961309251,
+      "learning_rate": 9.11726414229394e-05,
+      "loss": 0.7736,
+      "mean_token_accuracy": 0.7691892743110657,
+      "num_tokens": 137374587.0,
+      "step": 59350
+    },
+    {
+      "epoch": 0.5442376455487302,
+      "learning_rate": 9.115430457504356e-05,
+      "loss": 0.7446,
+      "mean_token_accuracy": 0.7727368533611297,
+      "num_tokens": 137398954.0,
+      "step": 59360
+    },
+    {
+      "epoch": 0.5443293297882094,
+      "learning_rate": 9.113596772714771e-05,
+      "loss": 0.7869,
+      "mean_token_accuracy": 0.7626693844795227,
+      "num_tokens": 137422734.0,
+      "step": 59370
+    },
+    {
+      "epoch": 0.5444210140276886,
+      "learning_rate": 9.111763087925186e-05,
+      "loss": 0.781,
+      "mean_token_accuracy": 0.7663978457450866,
+      "num_tokens": 137446891.0,
+      "step": 59380
+    },
+    {
+      "epoch": 0.5445126982671679,
+      "learning_rate": 9.109929403135602e-05,
+      "loss": 0.7744,
+      "mean_token_accuracy": 0.7736323773860931,
+      "num_tokens": 137471015.0,
+      "step": 59390
+    },
+    {
+      "epoch": 0.5446043825066471,
+      "learning_rate": 9.108095718346017e-05,
+      "loss": 0.7985,
+      "mean_token_accuracy": 0.7581858396530151,
+      "num_tokens": 137495436.0,
+      "step": 59400
+    },
+    {
+      "epoch": 0.5446960667461264,
+      "learning_rate": 9.106262033556432e-05,
+      "loss": 0.8202,
+      "mean_token_accuracy": 0.7503768920898437,
+      "num_tokens": 137519383.0,
+      "step": 59410
+    },
+    {
+      "epoch": 0.5447877509856056,
+      "learning_rate": 9.104428348766847e-05,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.768260782957077,
+      "num_tokens": 137544364.0,
+      "step": 59420
+    },
+    {
+      "epoch": 0.5448794352250849,
+      "learning_rate": 9.102594663977264e-05,
+      "loss": 0.7689,
+      "mean_token_accuracy": 0.7657553672790527,
+      "num_tokens": 137568472.0,
+      "step": 59430
+    },
+    {
+      "epoch": 0.544971119464564,
+      "learning_rate": 9.100760979187677e-05,
+      "loss": 0.8023,
+      "mean_token_accuracy": 0.7602219462394715,
+      "num_tokens": 137592014.0,
+      "step": 59440
+    },
+    {
+      "epoch": 0.5450628037040433,
+      "learning_rate": 9.098927294398093e-05,
+      "loss": 0.7362,
+      "mean_token_accuracy": 0.7781449675559997,
+      "num_tokens": 137616381.0,
+      "step": 59450
+    },
+    {
+      "epoch": 0.5451544879435225,
+      "learning_rate": 9.097093609608509e-05,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.7706298887729645,
+      "num_tokens": 137641531.0,
+      "step": 59460
+    },
+    {
+      "epoch": 0.5452461721830018,
+      "learning_rate": 9.095259924818924e-05,
+      "loss": 0.766,
+      "mean_token_accuracy": 0.7670444369316101,
+      "num_tokens": 137665465.0,
+      "step": 59470
+    },
+    {
+      "epoch": 0.545337856422481,
+      "learning_rate": 9.09342624002934e-05,
+      "loss": 0.7875,
+      "mean_token_accuracy": 0.7642770171165466,
+      "num_tokens": 137689269.0,
+      "step": 59480
+    },
+    {
+      "epoch": 0.5454295406619603,
+      "learning_rate": 9.091592555239755e-05,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7682712256908417,
+      "num_tokens": 137713898.0,
+      "step": 59490
+    },
+    {
+      "epoch": 0.5455212249014394,
+      "learning_rate": 9.08975887045017e-05,
+      "loss": 0.7915,
+      "mean_token_accuracy": 0.7614459991455078,
+      "num_tokens": 137738114.0,
+      "step": 59500
+    },
+    {
+      "epoch": 0.5456129091409186,
+      "learning_rate": 9.087925185660585e-05,
+      "loss": 0.7396,
+      "mean_token_accuracy": 0.776121324300766,
+      "num_tokens": 137762528.0,
+      "step": 59510
+    },
+    {
+      "epoch": 0.5457045933803979,
+      "learning_rate": 9.086091500871002e-05,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.7770536065101623,
+      "num_tokens": 137786929.0,
+      "step": 59520
+    },
+    {
+      "epoch": 0.5457962776198771,
+      "learning_rate": 9.084257816081417e-05,
+      "loss": 0.8456,
+      "mean_token_accuracy": 0.7492571771144867,
+      "num_tokens": 137811416.0,
+      "step": 59530
+    },
+    {
+      "epoch": 0.5458879618593564,
+      "learning_rate": 9.08242413129183e-05,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.7638105869293212,
+      "num_tokens": 137835907.0,
+      "step": 59540
+    },
+    {
+      "epoch": 0.5459796460988356,
+      "learning_rate": 9.080590446502246e-05,
+      "loss": 0.7762,
+      "mean_token_accuracy": 0.7648620069026947,
+      "num_tokens": 137858858.0,
+      "step": 59550
+    },
+    {
+      "epoch": 0.5460713303383149,
+      "learning_rate": 9.078756761712662e-05,
+      "loss": 0.7516,
+      "mean_token_accuracy": 0.767018860578537,
+      "num_tokens": 137883066.0,
+      "step": 59560
+    },
+    {
+      "epoch": 0.546163014577794,
+      "learning_rate": 9.076923076923078e-05,
+      "loss": 0.7508,
+      "mean_token_accuracy": 0.772193866968155,
+      "num_tokens": 137907861.0,
+      "step": 59570
+    },
+    {
+      "epoch": 0.5462546988172733,
+      "learning_rate": 9.075089392133493e-05,
+      "loss": 0.7818,
+      "mean_token_accuracy": 0.7609135806560516,
+      "num_tokens": 137931671.0,
+      "step": 59580
+    },
+    {
+      "epoch": 0.5463463830567525,
+      "learning_rate": 9.073255707343908e-05,
+      "loss": 0.7457,
+      "mean_token_accuracy": 0.7700284779071808,
+      "num_tokens": 137955637.0,
+      "step": 59590
+    },
+    {
+      "epoch": 0.5464380672962318,
+      "learning_rate": 9.071422022554323e-05,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7626480221748352,
+      "num_tokens": 137981083.0,
+      "step": 59600
+    },
+    {
+      "epoch": 0.546529751535711,
+      "learning_rate": 9.069588337764738e-05,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7701726138591767,
+      "num_tokens": 138005180.0,
+      "step": 59610
+    },
+    {
+      "epoch": 0.5466214357751903,
+      "learning_rate": 9.067754652975155e-05,
+      "loss": 0.7834,
+      "mean_token_accuracy": 0.7668368339538574,
+      "num_tokens": 138028986.0,
+      "step": 59620
+    },
+    {
+      "epoch": 0.5467131200146694,
+      "learning_rate": 9.06592096818557e-05,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.7639792084693908,
+      "num_tokens": 138052877.0,
+      "step": 59630
+    },
+    {
+      "epoch": 0.5468048042541487,
+      "learning_rate": 9.064087283395984e-05,
+      "loss": 0.7363,
+      "mean_token_accuracy": 0.7714833319187164,
+      "num_tokens": 138077288.0,
+      "step": 59640
+    },
+    {
+      "epoch": 0.5468964884936279,
+      "learning_rate": 9.062253598606399e-05,
+      "loss": 0.8167,
+      "mean_token_accuracy": 0.7511677742004395,
+      "num_tokens": 138101702.0,
+      "step": 59650
+    },
+    {
+      "epoch": 0.5469881727331072,
+      "learning_rate": 9.060419913816816e-05,
+      "loss": 0.748,
+      "mean_token_accuracy": 0.7644962072372437,
+      "num_tokens": 138125949.0,
+      "step": 59660
+    },
+    {
+      "epoch": 0.5470798569725864,
+      "learning_rate": 9.058586229027231e-05,
+      "loss": 0.7367,
+      "mean_token_accuracy": 0.7778648614883423,
+      "num_tokens": 138150429.0,
+      "step": 59670
+    },
+    {
+      "epoch": 0.5471715412120657,
+      "learning_rate": 9.056752544237646e-05,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.7757183253765106,
+      "num_tokens": 138174301.0,
+      "step": 59680
+    },
+    {
+      "epoch": 0.5472632254515449,
+      "learning_rate": 9.054918859448061e-05,
+      "loss": 0.7876,
+      "mean_token_accuracy": 0.7617397546768189,
+      "num_tokens": 138198872.0,
+      "step": 59690
+    },
+    {
+      "epoch": 0.5473549096910241,
+      "learning_rate": 9.053085174658476e-05,
+      "loss": 0.7193,
+      "mean_token_accuracy": 0.7781609416007995,
+      "num_tokens": 138223673.0,
+      "step": 59700
+    },
+    {
+      "epoch": 0.5474465939305033,
+      "learning_rate": 9.051251489868892e-05,
+      "loss": 0.7674,
+      "mean_token_accuracy": 0.7675589799880982,
+      "num_tokens": 138247443.0,
+      "step": 59710
+    },
+    {
+      "epoch": 0.5475382781699826,
+      "learning_rate": 9.049417805079308e-05,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7680305480957031,
+      "num_tokens": 138271451.0,
+      "step": 59720
+    },
+    {
+      "epoch": 0.5476299624094618,
+      "learning_rate": 9.047584120289723e-05,
+      "loss": 0.7624,
+      "mean_token_accuracy": 0.7677345633506775,
+      "num_tokens": 138296088.0,
+      "step": 59730
+    },
+    {
+      "epoch": 0.5477216466489411,
+      "learning_rate": 9.045750435500137e-05,
+      "loss": 0.7328,
+      "mean_token_accuracy": 0.7792566597461701,
+      "num_tokens": 138320598.0,
+      "step": 59740
+    },
+    {
+      "epoch": 0.5478133308884203,
+      "learning_rate": 9.043916750710554e-05,
+      "loss": 0.7246,
+      "mean_token_accuracy": 0.7782040536403656,
+      "num_tokens": 138345654.0,
+      "step": 59750
+    },
+    {
+      "epoch": 0.5479050151278995,
+      "learning_rate": 9.042083065920969e-05,
+      "loss": 0.7558,
+      "mean_token_accuracy": 0.7744125485420227,
+      "num_tokens": 138369920.0,
+      "step": 59760
+    },
+    {
+      "epoch": 0.5479966993673787,
+      "learning_rate": 9.040249381131384e-05,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.764605587720871,
+      "num_tokens": 138395104.0,
+      "step": 59770
+    },
+    {
+      "epoch": 0.548088383606858,
+      "learning_rate": 9.038415696341799e-05,
+      "loss": 0.7749,
+      "mean_token_accuracy": 0.7652956008911133,
+      "num_tokens": 138419545.0,
+      "step": 59780
+    },
+    {
+      "epoch": 0.5481800678463372,
+      "learning_rate": 9.036582011552214e-05,
+      "loss": 0.7712,
+      "mean_token_accuracy": 0.7687505304813385,
+      "num_tokens": 138443242.0,
+      "step": 59790
+    },
+    {
+      "epoch": 0.5482717520858165,
+      "learning_rate": 9.03474832676263e-05,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7709036409854889,
+      "num_tokens": 138467198.0,
+      "step": 59800
+    },
+    {
+      "epoch": 0.5483634363252957,
+      "learning_rate": 9.032914641973045e-05,
+      "loss": 0.7685,
+      "mean_token_accuracy": 0.7652524054050446,
+      "num_tokens": 138491073.0,
+      "step": 59810
+    },
+    {
+      "epoch": 0.548455120564775,
+      "learning_rate": 9.031080957183461e-05,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.7733059167861939,
+      "num_tokens": 138516540.0,
+      "step": 59820
+    },
+    {
+      "epoch": 0.5485468048042541,
+      "learning_rate": 9.029247272393877e-05,
+      "loss": 0.7985,
+      "mean_token_accuracy": 0.7557156145572662,
+      "num_tokens": 138539771.0,
+      "step": 59830
+    },
+    {
+      "epoch": 0.5486384890437334,
+      "learning_rate": 9.02741358760429e-05,
+      "loss": 0.7918,
+      "mean_token_accuracy": 0.7638346672058105,
+      "num_tokens": 138564028.0,
+      "step": 59840
+    },
+    {
+      "epoch": 0.5487301732832126,
+      "learning_rate": 9.025579902814707e-05,
+      "loss": 0.7895,
+      "mean_token_accuracy": 0.7664784848690033,
+      "num_tokens": 138587648.0,
+      "step": 59850
+    },
+    {
+      "epoch": 0.5488218575226919,
+      "learning_rate": 9.023746218025122e-05,
+      "loss": 0.797,
+      "mean_token_accuracy": 0.764485776424408,
+      "num_tokens": 138611625.0,
+      "step": 59860
+    },
+    {
+      "epoch": 0.5489135417621711,
+      "learning_rate": 9.021912533235537e-05,
+      "loss": 0.7629,
+      "mean_token_accuracy": 0.7619397938251495,
+      "num_tokens": 138636265.0,
+      "step": 59870
+    },
+    {
+      "epoch": 0.5490052260016504,
+      "learning_rate": 9.020078848445952e-05,
+      "loss": 0.753,
+      "mean_token_accuracy": 0.7711450815200805,
+      "num_tokens": 138660743.0,
+      "step": 59880
+    },
+    {
+      "epoch": 0.5490969102411295,
+      "learning_rate": 9.018245163656368e-05,
+      "loss": 0.7708,
+      "mean_token_accuracy": 0.7716257810592652,
+      "num_tokens": 138685011.0,
+      "step": 59890
+    },
+    {
+      "epoch": 0.5491885944806087,
+      "learning_rate": 9.016411478866783e-05,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7772553920745849,
+      "num_tokens": 138709378.0,
+      "step": 59900
+    },
+    {
+      "epoch": 0.549280278720088,
+      "learning_rate": 9.014577794077198e-05,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7682043612003326,
+      "num_tokens": 138733271.0,
+      "step": 59910
+    },
+    {
+      "epoch": 0.5493719629595673,
+      "learning_rate": 9.012744109287615e-05,
+      "loss": 0.7461,
+      "mean_token_accuracy": 0.7688748776912689,
+      "num_tokens": 138757825.0,
+      "step": 59920
+    },
+    {
+      "epoch": 0.5494636471990465,
+      "learning_rate": 9.01091042449803e-05,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.7701102018356323,
+      "num_tokens": 138782084.0,
+      "step": 59930
+    },
+    {
+      "epoch": 0.5495553314385258,
+      "learning_rate": 9.009076739708444e-05,
+      "loss": 0.7337,
+      "mean_token_accuracy": 0.7770924091339111,
+      "num_tokens": 138806224.0,
+      "step": 59940
+    },
+    {
+      "epoch": 0.549647015678005,
+      "learning_rate": 9.00724305491886e-05,
+      "loss": 0.7905,
+      "mean_token_accuracy": 0.7614732563495636,
+      "num_tokens": 138830323.0,
+      "step": 59950
+    },
+    {
+      "epoch": 0.5497386999174841,
+      "learning_rate": 9.005409370129275e-05,
+      "loss": 0.7217,
+      "mean_token_accuracy": 0.7798113286495209,
+      "num_tokens": 138854012.0,
+      "step": 59960
+    },
+    {
+      "epoch": 0.5498303841569634,
+      "learning_rate": 9.00357568533969e-05,
+      "loss": 0.757,
+      "mean_token_accuracy": 0.7711302220821381,
+      "num_tokens": 138878505.0,
+      "step": 59970
+    },
+    {
+      "epoch": 0.5499220683964426,
+      "learning_rate": 9.001742000550106e-05,
+      "loss": 0.7008,
+      "mean_token_accuracy": 0.7812945306301117,
+      "num_tokens": 138903120.0,
+      "step": 59980
+    },
+    {
+      "epoch": 0.5500137526359219,
+      "learning_rate": 8.999908315760521e-05,
+      "loss": 0.7091,
+      "mean_token_accuracy": 0.7743261158466339,
+      "num_tokens": 138927734.0,
+      "step": 59990
+    },
+    {
+      "epoch": 0.5501054368754011,
+      "learning_rate": 8.998074630970936e-05,
+      "loss": 0.7105,
+      "mean_token_accuracy": 0.7869706869125366,
+      "num_tokens": 138953168.0,
+      "step": 60000
+    },
+    {
+      "epoch": 0.5501971211148804,
+      "learning_rate": 8.996240946181353e-05,
+      "loss": 0.8146,
+      "mean_token_accuracy": 0.7537143945693969,
+      "num_tokens": 138977485.0,
+      "step": 60010
+    },
+    {
+      "epoch": 0.5502888053543595,
+      "learning_rate": 8.994407261391768e-05,
+      "loss": 0.7735,
+      "mean_token_accuracy": 0.7683954477310181,
+      "num_tokens": 139001454.0,
+      "step": 60020
+    },
+    {
+      "epoch": 0.5503804895938388,
+      "learning_rate": 8.992573576602183e-05,
+      "loss": 0.781,
+      "mean_token_accuracy": 0.7664335370063782,
+      "num_tokens": 139025848.0,
+      "step": 60030
+    },
+    {
+      "epoch": 0.550472173833318,
+      "learning_rate": 8.990739891812597e-05,
+      "loss": 0.7821,
+      "mean_token_accuracy": 0.764576894044876,
+      "num_tokens": 139050485.0,
+      "step": 60040
+    },
+    {
+      "epoch": 0.5505638580727973,
+      "learning_rate": 8.988906207023013e-05,
+      "loss": 0.7389,
+      "mean_token_accuracy": 0.7736314237117767,
+      "num_tokens": 139074877.0,
+      "step": 60050
+    },
+    {
+      "epoch": 0.5506555423122765,
+      "learning_rate": 8.987072522233429e-05,
+      "loss": 0.7397,
+      "mean_token_accuracy": 0.7742312371730804,
+      "num_tokens": 139098559.0,
+      "step": 60060
+    },
+    {
+      "epoch": 0.5507472265517558,
+      "learning_rate": 8.985238837443844e-05,
+      "loss": 0.7574,
+      "mean_token_accuracy": 0.76783966422081,
+      "num_tokens": 139123948.0,
+      "step": 60070
+    },
+    {
+      "epoch": 0.550838910791235,
+      "learning_rate": 8.983405152654259e-05,
+      "loss": 0.7296,
+      "mean_token_accuracy": 0.7761675238609314,
+      "num_tokens": 139148419.0,
+      "step": 60080
+    },
+    {
+      "epoch": 0.5509305950307142,
+      "learning_rate": 8.981571467864674e-05,
+      "loss": 0.7365,
+      "mean_token_accuracy": 0.7745570719242096,
+      "num_tokens": 139173088.0,
+      "step": 60090
+    },
+    {
+      "epoch": 0.5510222792701934,
+      "learning_rate": 8.979737783075089e-05,
+      "loss": 0.7389,
+      "mean_token_accuracy": 0.779655647277832,
+      "num_tokens": 139198318.0,
+      "step": 60100
+    },
+    {
+      "epoch": 0.5511139635096727,
+      "learning_rate": 8.977904098285506e-05,
+      "loss": 0.6958,
+      "mean_token_accuracy": 0.7848239302635193,
+      "num_tokens": 139222660.0,
+      "step": 60110
+    },
+    {
+      "epoch": 0.5512056477491519,
+      "learning_rate": 8.976070413495921e-05,
+      "loss": 0.7385,
+      "mean_token_accuracy": 0.7779003560543061,
+      "num_tokens": 139247176.0,
+      "step": 60120
+    },
+    {
+      "epoch": 0.5512973319886312,
+      "learning_rate": 8.974236728706336e-05,
+      "loss": 0.7765,
+      "mean_token_accuracy": 0.7620017051696777,
+      "num_tokens": 139271464.0,
+      "step": 60130
+    },
+    {
+      "epoch": 0.5513890162281104,
+      "learning_rate": 8.972403043916751e-05,
+      "loss": 0.7618,
+      "mean_token_accuracy": 0.7723858892917633,
+      "num_tokens": 139294980.0,
+      "step": 60140
+    },
+    {
+      "epoch": 0.5514807004675896,
+      "learning_rate": 8.970569359127167e-05,
+      "loss": 0.7605,
+      "mean_token_accuracy": 0.7708096027374267,
+      "num_tokens": 139319266.0,
+      "step": 60150
+    },
+    {
+      "epoch": 0.5515723847070688,
+      "learning_rate": 8.968735674337582e-05,
+      "loss": 0.7718,
+      "mean_token_accuracy": 0.7658926963806152,
+      "num_tokens": 139343814.0,
+      "step": 60160
+    },
+    {
+      "epoch": 0.5516640689465481,
+      "learning_rate": 8.966901989547997e-05,
+      "loss": 0.7608,
+      "mean_token_accuracy": 0.7604470908641815,
+      "num_tokens": 139367529.0,
+      "step": 60170
+    },
+    {
+      "epoch": 0.5517557531860273,
+      "learning_rate": 8.965068304758412e-05,
+      "loss": 0.7396,
+      "mean_token_accuracy": 0.7774915635585785,
+      "num_tokens": 139391982.0,
+      "step": 60180
+    },
+    {
+      "epoch": 0.5518474374255066,
+      "learning_rate": 8.963234619968827e-05,
+      "loss": 0.7581,
+      "mean_token_accuracy": 0.7741159200668335,
+      "num_tokens": 139416174.0,
+      "step": 60190
+    },
+    {
+      "epoch": 0.5519391216649858,
+      "learning_rate": 8.961400935179242e-05,
+      "loss": 0.7495,
+      "mean_token_accuracy": 0.7707136988639831,
+      "num_tokens": 139441176.0,
+      "step": 60200
+    },
+    {
+      "epoch": 0.5520308059044651,
+      "learning_rate": 8.959567250389659e-05,
+      "loss": 0.7823,
+      "mean_token_accuracy": 0.7647032737731934,
+      "num_tokens": 139465616.0,
+      "step": 60210
+    },
+    {
+      "epoch": 0.5521224901439442,
+      "learning_rate": 8.957733565600074e-05,
+      "loss": 0.7424,
+      "mean_token_accuracy": 0.7688823223114014,
+      "num_tokens": 139489643.0,
+      "step": 60220
+    },
+    {
+      "epoch": 0.5522141743834235,
+      "learning_rate": 8.95589988081049e-05,
+      "loss": 0.7255,
+      "mean_token_accuracy": 0.7780394971370697,
+      "num_tokens": 139513584.0,
+      "step": 60230
+    },
+    {
+      "epoch": 0.5523058586229027,
+      "learning_rate": 8.954066196020905e-05,
+      "loss": 0.7664,
+      "mean_token_accuracy": 0.7680767595767974,
+      "num_tokens": 139537927.0,
+      "step": 60240
+    },
+    {
+      "epoch": 0.552397542862382,
+      "learning_rate": 8.95223251123132e-05,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.7776258885860443,
+      "num_tokens": 139562292.0,
+      "step": 60250
+    },
+    {
+      "epoch": 0.5524892271018612,
+      "learning_rate": 8.950398826441735e-05,
+      "loss": 0.7758,
+      "mean_token_accuracy": 0.7627845525741577,
+      "num_tokens": 139586080.0,
+      "step": 60260
+    },
+    {
+      "epoch": 0.5525809113413405,
+      "learning_rate": 8.94856514165215e-05,
+      "loss": 0.8077,
+      "mean_token_accuracy": 0.7642762005329132,
+      "num_tokens": 139609576.0,
+      "step": 60270
+    },
+    {
+      "epoch": 0.5526725955808196,
+      "learning_rate": 8.946731456862565e-05,
+      "loss": 0.747,
+      "mean_token_accuracy": 0.7752709448337555,
+      "num_tokens": 139634017.0,
+      "step": 60280
+    },
+    {
+      "epoch": 0.5527642798202989,
+      "learning_rate": 8.94489777207298e-05,
+      "loss": 0.7868,
+      "mean_token_accuracy": 0.7587838709354401,
+      "num_tokens": 139658406.0,
+      "step": 60290
+    },
+    {
+      "epoch": 0.5528559640597781,
+      "learning_rate": 8.943064087283396e-05,
+      "loss": 0.7345,
+      "mean_token_accuracy": 0.7784680187702179,
+      "num_tokens": 139683175.0,
+      "step": 60300
+    },
+    {
+      "epoch": 0.5529476482992574,
+      "learning_rate": 8.941230402493812e-05,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7650320649147033,
+      "num_tokens": 139708314.0,
+      "step": 60310
+    },
+    {
+      "epoch": 0.5530393325387366,
+      "learning_rate": 8.939396717704227e-05,
+      "loss": 0.7389,
+      "mean_token_accuracy": 0.7763622403144836,
+      "num_tokens": 139733482.0,
+      "step": 60320
+    },
+    {
+      "epoch": 0.5531310167782159,
+      "learning_rate": 8.937563032914643e-05,
+      "loss": 0.7976,
+      "mean_token_accuracy": 0.7633031368255615,
+      "num_tokens": 139757652.0,
+      "step": 60330
+    },
+    {
+      "epoch": 0.5532227010176951,
+      "learning_rate": 8.935729348125058e-05,
+      "loss": 0.7675,
+      "mean_token_accuracy": 0.766411405801773,
+      "num_tokens": 139782342.0,
+      "step": 60340
+    },
+    {
+      "epoch": 0.5533143852571742,
+      "learning_rate": 8.933895663335473e-05,
+      "loss": 0.7221,
+      "mean_token_accuracy": 0.7828215837478638,
+      "num_tokens": 139806368.0,
+      "step": 60350
+    },
+    {
+      "epoch": 0.5534060694966535,
+      "learning_rate": 8.932061978545888e-05,
+      "loss": 0.7424,
+      "mean_token_accuracy": 0.7767630159854889,
+      "num_tokens": 139830790.0,
+      "step": 60360
+    },
+    {
+      "epoch": 0.5534977537361327,
+      "learning_rate": 8.930228293756305e-05,
+      "loss": 0.7327,
+      "mean_token_accuracy": 0.7781683087348938,
+      "num_tokens": 139854776.0,
+      "step": 60370
+    },
+    {
+      "epoch": 0.553589437975612,
+      "learning_rate": 8.928394608966719e-05,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.7682272791862488,
+      "num_tokens": 139878902.0,
+      "step": 60380
+    },
+    {
+      "epoch": 0.5536811222150912,
+      "learning_rate": 8.926560924177134e-05,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7684300899505615,
+      "num_tokens": 139904005.0,
+      "step": 60390
+    },
+    {
+      "epoch": 0.5537728064545705,
+      "learning_rate": 8.924727239387549e-05,
+      "loss": 0.6881,
+      "mean_token_accuracy": 0.7928227484226227,
+      "num_tokens": 139928142.0,
+      "step": 60400
+    },
+    {
+      "epoch": 0.5538644906940496,
+      "learning_rate": 8.922893554597965e-05,
+      "loss": 0.7226,
+      "mean_token_accuracy": 0.7775552093982696,
+      "num_tokens": 139952589.0,
+      "step": 60410
+    },
+    {
+      "epoch": 0.5539561749335289,
+      "learning_rate": 8.92105986980838e-05,
+      "loss": 0.7156,
+      "mean_token_accuracy": 0.7812934756278992,
+      "num_tokens": 139977972.0,
+      "step": 60420
+    },
+    {
+      "epoch": 0.5540478591730081,
+      "learning_rate": 8.919226185018796e-05,
+      "loss": 0.7428,
+      "mean_token_accuracy": 0.7776130139827728,
+      "num_tokens": 140003070.0,
+      "step": 60430
+    },
+    {
+      "epoch": 0.5541395434124874,
+      "learning_rate": 8.917392500229211e-05,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.7657585442066193,
+      "num_tokens": 140027538.0,
+      "step": 60440
+    },
+    {
+      "epoch": 0.5542312276519666,
+      "learning_rate": 8.915558815439626e-05,
+      "loss": 0.769,
+      "mean_token_accuracy": 0.773032295703888,
+      "num_tokens": 140052026.0,
+      "step": 60450
+    },
+    {
+      "epoch": 0.5543229118914459,
+      "learning_rate": 8.913725130650041e-05,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.7604465782642365,
+      "num_tokens": 140076037.0,
+      "step": 60460
+    },
+    {
+      "epoch": 0.5544145961309251,
+      "learning_rate": 8.911891445860458e-05,
+      "loss": 0.7558,
+      "mean_token_accuracy": 0.7691322803497315,
+      "num_tokens": 140100191.0,
+      "step": 60470
+    },
+    {
+      "epoch": 0.5545062803704043,
+      "learning_rate": 8.910057761070872e-05,
+      "loss": 0.6975,
+      "mean_token_accuracy": 0.7837631106376648,
+      "num_tokens": 140125905.0,
+      "step": 60480
+    },
+    {
+      "epoch": 0.5545979646098835,
+      "learning_rate": 8.908224076281287e-05,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.7748779773712158,
+      "num_tokens": 140150936.0,
+      "step": 60490
+    },
+    {
+      "epoch": 0.5546896488493628,
+      "learning_rate": 8.906390391491704e-05,
+      "loss": 0.7967,
+      "mean_token_accuracy": 0.7652240335941315,
+      "num_tokens": 140175616.0,
+      "step": 60500
+    },
+    {
+      "epoch": 0.554781333088842,
+      "learning_rate": 8.904556706702119e-05,
+      "loss": 0.7376,
+      "mean_token_accuracy": 0.7764686822891236,
+      "num_tokens": 140200453.0,
+      "step": 60510
+    },
+    {
+      "epoch": 0.5548730173283213,
+      "learning_rate": 8.902723021912534e-05,
+      "loss": 0.7359,
+      "mean_token_accuracy": 0.7744403719902039,
+      "num_tokens": 140224911.0,
+      "step": 60520
+    },
+    {
+      "epoch": 0.5549647015678005,
+      "learning_rate": 8.900889337122949e-05,
+      "loss": 0.7671,
+      "mean_token_accuracy": 0.7694585621356964,
+      "num_tokens": 140250349.0,
+      "step": 60530
+    },
+    {
+      "epoch": 0.5550563858072798,
+      "learning_rate": 8.899055652333364e-05,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7733817458152771,
+      "num_tokens": 140275980.0,
+      "step": 60540
+    },
+    {
+      "epoch": 0.5551480700467589,
+      "learning_rate": 8.89722196754378e-05,
+      "loss": 0.7971,
+      "mean_token_accuracy": 0.7652704894542695,
+      "num_tokens": 140299793.0,
+      "step": 60550
+    },
+    {
+      "epoch": 0.5552397542862382,
+      "learning_rate": 8.895388282754195e-05,
+      "loss": 0.7787,
+      "mean_token_accuracy": 0.7674022853374481,
+      "num_tokens": 140323896.0,
+      "step": 60560
+    },
+    {
+      "epoch": 0.5553314385257174,
+      "learning_rate": 8.893554597964611e-05,
+      "loss": 0.7441,
+      "mean_token_accuracy": 0.7775610744953155,
+      "num_tokens": 140348789.0,
+      "step": 60570
+    },
+    {
+      "epoch": 0.5554231227651967,
+      "learning_rate": 8.891720913175025e-05,
+      "loss": 0.7536,
+      "mean_token_accuracy": 0.7698719739913941,
+      "num_tokens": 140373299.0,
+      "step": 60580
+    },
+    {
+      "epoch": 0.5555148070046759,
+      "learning_rate": 8.88988722838544e-05,
+      "loss": 0.7086,
+      "mean_token_accuracy": 0.7829937219619751,
+      "num_tokens": 140398560.0,
+      "step": 60590
+    },
+    {
+      "epoch": 0.5556064912441552,
+      "learning_rate": 8.888053543595857e-05,
+      "loss": 0.7261,
+      "mean_token_accuracy": 0.7843235313892365,
+      "num_tokens": 140424358.0,
+      "step": 60600
+    },
+    {
+      "epoch": 0.5556981754836343,
+      "learning_rate": 8.886219858806272e-05,
+      "loss": 0.8054,
+      "mean_token_accuracy": 0.7546370446681976,
+      "num_tokens": 140447908.0,
+      "step": 60610
+    },
+    {
+      "epoch": 0.5557898597231136,
+      "learning_rate": 8.884386174016687e-05,
+      "loss": 0.7462,
+      "mean_token_accuracy": 0.7767966389656067,
+      "num_tokens": 140472309.0,
+      "step": 60620
+    },
+    {
+      "epoch": 0.5558815439625928,
+      "learning_rate": 8.882552489227102e-05,
+      "loss": 0.7348,
+      "mean_token_accuracy": 0.7720439910888672,
+      "num_tokens": 140498085.0,
+      "step": 60630
+    },
+    {
+      "epoch": 0.5559732282020721,
+      "learning_rate": 8.880718804437517e-05,
+      "loss": 0.7322,
+      "mean_token_accuracy": 0.7770056009292603,
+      "num_tokens": 140522371.0,
+      "step": 60640
+    },
+    {
+      "epoch": 0.5560649124415513,
+      "learning_rate": 8.878885119647933e-05,
+      "loss": 0.7247,
+      "mean_token_accuracy": 0.7749279797077179,
+      "num_tokens": 140546948.0,
+      "step": 60650
+    },
+    {
+      "epoch": 0.5561565966810306,
+      "learning_rate": 8.877051434858348e-05,
+      "loss": 0.7111,
+      "mean_token_accuracy": 0.778215354681015,
+      "num_tokens": 140572536.0,
+      "step": 60660
+    },
+    {
+      "epoch": 0.5562482809205098,
+      "learning_rate": 8.875217750068764e-05,
+      "loss": 0.744,
+      "mean_token_accuracy": 0.7755861580371857,
+      "num_tokens": 140596979.0,
+      "step": 60670
+    },
+    {
+      "epoch": 0.556339965159989,
+      "learning_rate": 8.873384065279178e-05,
+      "loss": 0.7289,
+      "mean_token_accuracy": 0.7835552096366882,
+      "num_tokens": 140621150.0,
+      "step": 60680
+    },
+    {
+      "epoch": 0.5564316493994682,
+      "learning_rate": 8.871550380489593e-05,
+      "loss": 0.7297,
+      "mean_token_accuracy": 0.7749859631061554,
+      "num_tokens": 140645648.0,
+      "step": 60690
+    },
+    {
+      "epoch": 0.5565233336389475,
+      "learning_rate": 8.86971669570001e-05,
+      "loss": 0.7374,
+      "mean_token_accuracy": 0.7716366291046143,
+      "num_tokens": 140670477.0,
+      "step": 60700
+    },
+    {
+      "epoch": 0.5566150178784267,
+      "learning_rate": 8.867883010910425e-05,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.7810007274150849,
+      "num_tokens": 140695695.0,
+      "step": 60710
+    },
+    {
+      "epoch": 0.556706702117906,
+      "learning_rate": 8.86604932612084e-05,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.7683439433574677,
+      "num_tokens": 140720847.0,
+      "step": 60720
+    },
+    {
+      "epoch": 0.5567983863573852,
+      "learning_rate": 8.864215641331256e-05,
+      "loss": 0.7931,
+      "mean_token_accuracy": 0.7599188685417175,
+      "num_tokens": 140745014.0,
+      "step": 60730
+    },
+    {
+      "epoch": 0.5568900705968644,
+      "learning_rate": 8.862381956541671e-05,
+      "loss": 0.7312,
+      "mean_token_accuracy": 0.7761836469173431,
+      "num_tokens": 140769222.0,
+      "step": 60740
+    },
+    {
+      "epoch": 0.5569817548363436,
+      "learning_rate": 8.860548271752086e-05,
+      "loss": 0.6983,
+      "mean_token_accuracy": 0.7895478844642639,
+      "num_tokens": 140793364.0,
+      "step": 60750
+    },
+    {
+      "epoch": 0.5570734390758229,
+      "learning_rate": 8.858714586962502e-05,
+      "loss": 0.7291,
+      "mean_token_accuracy": 0.7726792871952057,
+      "num_tokens": 140817259.0,
+      "step": 60760
+    },
+    {
+      "epoch": 0.5571651233153021,
+      "learning_rate": 8.856880902172918e-05,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.7737743556499481,
+      "num_tokens": 140842288.0,
+      "step": 60770
+    },
+    {
+      "epoch": 0.5572568075547814,
+      "learning_rate": 8.855047217383331e-05,
+      "loss": 0.7161,
+      "mean_token_accuracy": 0.7794773995876312,
+      "num_tokens": 140866603.0,
+      "step": 60780
+    },
+    {
+      "epoch": 0.5573484917942606,
+      "learning_rate": 8.853213532593747e-05,
+      "loss": 0.7478,
+      "mean_token_accuracy": 0.7692714989185333,
+      "num_tokens": 140891440.0,
+      "step": 60790
+    },
+    {
+      "epoch": 0.5574401760337399,
+      "learning_rate": 8.851379847804163e-05,
+      "loss": 0.7435,
+      "mean_token_accuracy": 0.7763656795024871,
+      "num_tokens": 140916817.0,
+      "step": 60800
+    },
+    {
+      "epoch": 0.557531860273219,
+      "learning_rate": 8.849546163014578e-05,
+      "loss": 0.7549,
+      "mean_token_accuracy": 0.7707635164260864,
+      "num_tokens": 140941500.0,
+      "step": 60810
+    },
+    {
+      "epoch": 0.5576235445126982,
+      "learning_rate": 8.847712478224994e-05,
+      "loss": 0.726,
+      "mean_token_accuracy": 0.7727028369903565,
+      "num_tokens": 140966174.0,
+      "step": 60820
+    },
+    {
+      "epoch": 0.5577152287521775,
+      "learning_rate": 8.845878793435409e-05,
+      "loss": 0.6845,
+      "mean_token_accuracy": 0.7924252450466156,
+      "num_tokens": 140991763.0,
+      "step": 60830
+    },
+    {
+      "epoch": 0.5578069129916567,
+      "learning_rate": 8.844045108645824e-05,
+      "loss": 0.6939,
+      "mean_token_accuracy": 0.7871376514434815,
+      "num_tokens": 141017056.0,
+      "step": 60840
+    },
+    {
+      "epoch": 0.557898597231136,
+      "learning_rate": 8.842211423856239e-05,
+      "loss": 0.7453,
+      "mean_token_accuracy": 0.7725602924823761,
+      "num_tokens": 141042096.0,
+      "step": 60850
+    },
+    {
+      "epoch": 0.5579902814706152,
+      "learning_rate": 8.840377739066656e-05,
+      "loss": 0.7449,
+      "mean_token_accuracy": 0.7751950621604919,
+      "num_tokens": 141066226.0,
+      "step": 60860
+    },
+    {
+      "epoch": 0.5580819657100944,
+      "learning_rate": 8.838544054277071e-05,
+      "loss": 0.7084,
+      "mean_token_accuracy": 0.7858047962188721,
+      "num_tokens": 141090053.0,
+      "step": 60870
+    },
+    {
+      "epoch": 0.5581736499495736,
+      "learning_rate": 8.836710369487485e-05,
+      "loss": 0.7624,
+      "mean_token_accuracy": 0.7668496429920196,
+      "num_tokens": 141114729.0,
+      "step": 60880
+    },
+    {
+      "epoch": 0.5582653341890529,
+      "learning_rate": 8.8348766846979e-05,
+      "loss": 0.7123,
+      "mean_token_accuracy": 0.7860641777515411,
+      "num_tokens": 141140151.0,
+      "step": 60890
+    },
+    {
+      "epoch": 0.5583570184285321,
+      "learning_rate": 8.833042999908316e-05,
+      "loss": 0.738,
+      "mean_token_accuracy": 0.7760488331317902,
+      "num_tokens": 141165360.0,
+      "step": 60900
+    },
+    {
+      "epoch": 0.5584487026680114,
+      "learning_rate": 8.831209315118732e-05,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7627298414707184,
+      "num_tokens": 141189187.0,
+      "step": 60910
+    },
+    {
+      "epoch": 0.5585403869074906,
+      "learning_rate": 8.829375630329147e-05,
+      "loss": 0.7083,
+      "mean_token_accuracy": 0.7848528623580933,
+      "num_tokens": 141213872.0,
+      "step": 60920
+    },
+    {
+      "epoch": 0.5586320711469699,
+      "learning_rate": 8.827541945539562e-05,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7646727144718171,
+      "num_tokens": 141238522.0,
+      "step": 60930
+    },
+    {
+      "epoch": 0.558723755386449,
+      "learning_rate": 8.825708260749977e-05,
+      "loss": 0.734,
+      "mean_token_accuracy": 0.7805249214172363,
+      "num_tokens": 141263361.0,
+      "step": 60940
+    },
+    {
+      "epoch": 0.5588154396259283,
+      "learning_rate": 8.823874575960392e-05,
+      "loss": 0.7259,
+      "mean_token_accuracy": 0.7775725722312927,
+      "num_tokens": 141287671.0,
+      "step": 60950
+    },
+    {
+      "epoch": 0.5589071238654075,
+      "learning_rate": 8.822040891170809e-05,
+      "loss": 0.7474,
+      "mean_token_accuracy": 0.7717169404029847,
+      "num_tokens": 141311997.0,
+      "step": 60960
+    },
+    {
+      "epoch": 0.5589988081048868,
+      "learning_rate": 8.820207206381224e-05,
+      "loss": 0.6986,
+      "mean_token_accuracy": 0.7866757810115814,
+      "num_tokens": 141336475.0,
+      "step": 60970
+    },
+    {
+      "epoch": 0.559090492344366,
+      "learning_rate": 8.818373521591638e-05,
+      "loss": 0.7521,
+      "mean_token_accuracy": 0.7763409078121185,
+      "num_tokens": 141361489.0,
+      "step": 60980
+    },
+    {
+      "epoch": 0.5591821765838453,
+      "learning_rate": 8.816539836802054e-05,
+      "loss": 0.7253,
+      "mean_token_accuracy": 0.7778802812099457,
+      "num_tokens": 141386479.0,
+      "step": 60990
+    },
+    {
+      "epoch": 0.5592738608233244,
+      "learning_rate": 8.81470615201247e-05,
+      "loss": 0.764,
+      "mean_token_accuracy": 0.771779590845108,
+      "num_tokens": 141411513.0,
+      "step": 61000
+    },
+    {
+      "epoch": 0.5593655450628037,
+      "learning_rate": 8.812872467222885e-05,
+      "loss": 0.7154,
+      "mean_token_accuracy": 0.7872254371643066,
+      "num_tokens": 141436243.0,
+      "step": 61010
+    },
+    {
+      "epoch": 0.5594572293022829,
+      "learning_rate": 8.8110387824333e-05,
+      "loss": 0.703,
+      "mean_token_accuracy": 0.7816587388515472,
+      "num_tokens": 141461458.0,
+      "step": 61020
+    },
+    {
+      "epoch": 0.5595489135417622,
+      "learning_rate": 8.809205097643715e-05,
+      "loss": 0.772,
+      "mean_token_accuracy": 0.7643976628780365,
+      "num_tokens": 141485690.0,
+      "step": 61030
+    },
+    {
+      "epoch": 0.5596405977812414,
+      "learning_rate": 8.80737141285413e-05,
+      "loss": 0.7591,
+      "mean_token_accuracy": 0.7725388169288635,
+      "num_tokens": 141510135.0,
+      "step": 61040
+    },
+    {
+      "epoch": 0.5597322820207207,
+      "learning_rate": 8.805537728064546e-05,
+      "loss": 0.7009,
+      "mean_token_accuracy": 0.7810416221618652,
+      "num_tokens": 141535028.0,
+      "step": 61050
+    },
+    {
+      "epoch": 0.5598239662601999,
+      "learning_rate": 8.803704043274962e-05,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7759714722633362,
+      "num_tokens": 141559996.0,
+      "step": 61060
+    },
+    {
+      "epoch": 0.5599156504996791,
+      "learning_rate": 8.801870358485377e-05,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.7873109877109528,
+      "num_tokens": 141584710.0,
+      "step": 61070
+    },
+    {
+      "epoch": 0.5600073347391583,
+      "learning_rate": 8.800036673695791e-05,
+      "loss": 0.7179,
+      "mean_token_accuracy": 0.781514710187912,
+      "num_tokens": 141609681.0,
+      "step": 61080
+    },
+    {
+      "epoch": 0.5600990189786376,
+      "learning_rate": 8.798202988906208e-05,
+      "loss": 0.742,
+      "mean_token_accuracy": 0.7753199696540832,
+      "num_tokens": 141633912.0,
+      "step": 61090
+    },
+    {
+      "epoch": 0.5601907032181168,
+      "learning_rate": 8.796369304116623e-05,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7787988245487213,
+      "num_tokens": 141658106.0,
+      "step": 61100
+    },
+    {
+      "epoch": 0.5602823874575961,
+      "learning_rate": 8.794535619327038e-05,
+      "loss": 0.7315,
+      "mean_token_accuracy": 0.7788116216659546,
+      "num_tokens": 141682789.0,
+      "step": 61110
+    },
+    {
+      "epoch": 0.5603740716970753,
+      "learning_rate": 8.792701934537455e-05,
+      "loss": 0.7377,
+      "mean_token_accuracy": 0.7705783665180206,
+      "num_tokens": 141708360.0,
+      "step": 61120
+    },
+    {
+      "epoch": 0.5604657559365545,
+      "learning_rate": 8.790868249747868e-05,
+      "loss": 0.7175,
+      "mean_token_accuracy": 0.788539856672287,
+      "num_tokens": 141733260.0,
+      "step": 61130
+    },
+    {
+      "epoch": 0.5605574401760337,
+      "learning_rate": 8.789034564958284e-05,
+      "loss": 0.7515,
+      "mean_token_accuracy": 0.7713378965854645,
+      "num_tokens": 141757752.0,
+      "step": 61140
+    },
+    {
+      "epoch": 0.560649124415513,
+      "learning_rate": 8.787200880168699e-05,
+      "loss": 0.7288,
+      "mean_token_accuracy": 0.7729795396327972,
+      "num_tokens": 141782487.0,
+      "step": 61150
+    },
+    {
+      "epoch": 0.5607408086549922,
+      "learning_rate": 8.785367195379115e-05,
+      "loss": 0.7418,
+      "mean_token_accuracy": 0.774861353635788,
+      "num_tokens": 141807675.0,
+      "step": 61160
+    },
+    {
+      "epoch": 0.5608324928944715,
+      "learning_rate": 8.78353351058953e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.7917165219783783,
+      "num_tokens": 141832580.0,
+      "step": 61170
+    },
+    {
+      "epoch": 0.5609241771339507,
+      "learning_rate": 8.781699825799944e-05,
+      "loss": 0.7139,
+      "mean_token_accuracy": 0.7839303493499756,
+      "num_tokens": 141857734.0,
+      "step": 61180
+    },
+    {
+      "epoch": 0.56101586137343,
+      "learning_rate": 8.779866141010361e-05,
+      "loss": 0.7126,
+      "mean_token_accuracy": 0.7835335195064544,
+      "num_tokens": 141882208.0,
+      "step": 61190
+    },
+    {
+      "epoch": 0.5611075456129091,
+      "learning_rate": 8.778032456220776e-05,
+      "loss": 0.724,
+      "mean_token_accuracy": 0.783144611120224,
+      "num_tokens": 141906932.0,
+      "step": 61200
+    },
+    {
+      "epoch": 0.5611992298523883,
+      "learning_rate": 8.776198771431191e-05,
+      "loss": 0.7332,
+      "mean_token_accuracy": 0.7786126017570496,
+      "num_tokens": 141931412.0,
+      "step": 61210
+    },
+    {
+      "epoch": 0.5612909140918676,
+      "learning_rate": 8.774365086641608e-05,
+      "loss": 0.7439,
+      "mean_token_accuracy": 0.7730164468288422,
+      "num_tokens": 141956073.0,
+      "step": 61220
+    },
+    {
+      "epoch": 0.5613825983313468,
+      "learning_rate": 8.772531401852022e-05,
+      "loss": 0.7244,
+      "mean_token_accuracy": 0.7842257678508758,
+      "num_tokens": 141981534.0,
+      "step": 61230
+    },
+    {
+      "epoch": 0.5614742825708261,
+      "learning_rate": 8.770697717062437e-05,
+      "loss": 0.7428,
+      "mean_token_accuracy": 0.7746395111083985,
+      "num_tokens": 142005366.0,
+      "step": 61240
+    },
+    {
+      "epoch": 0.5615659668103054,
+      "learning_rate": 8.768864032272853e-05,
+      "loss": 0.7043,
+      "mean_token_accuracy": 0.7825834572315216,
+      "num_tokens": 142031163.0,
+      "step": 61250
+    },
+    {
+      "epoch": 0.5616576510497845,
+      "learning_rate": 8.767030347483269e-05,
+      "loss": 0.7326,
+      "mean_token_accuracy": 0.7777396857738494,
+      "num_tokens": 142056016.0,
+      "step": 61260
+    },
+    {
+      "epoch": 0.5617493352892637,
+      "learning_rate": 8.765196662693684e-05,
+      "loss": 0.7161,
+      "mean_token_accuracy": 0.7755993962287903,
+      "num_tokens": 142080906.0,
+      "step": 61270
+    },
+    {
+      "epoch": 0.561841019528743,
+      "learning_rate": 8.763362977904098e-05,
+      "loss": 0.7509,
+      "mean_token_accuracy": 0.7720266282558441,
+      "num_tokens": 142105483.0,
+      "step": 61280
+    },
+    {
+      "epoch": 0.5619327037682222,
+      "learning_rate": 8.761529293114514e-05,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7713659226894378,
+      "num_tokens": 142129440.0,
+      "step": 61290
+    },
+    {
+      "epoch": 0.5620243880077015,
+      "learning_rate": 8.759695608324929e-05,
+      "loss": 0.7376,
+      "mean_token_accuracy": 0.7746829330921173,
+      "num_tokens": 142154420.0,
+      "step": 61300
+    },
+    {
+      "epoch": 0.5621160722471807,
+      "learning_rate": 8.757861923535344e-05,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7759364008903503,
+      "num_tokens": 142179364.0,
+      "step": 61310
+    },
+    {
+      "epoch": 0.56220775648666,
+      "learning_rate": 8.756028238745761e-05,
+      "loss": 0.7467,
+      "mean_token_accuracy": 0.772876912355423,
+      "num_tokens": 142203813.0,
+      "step": 61320
+    },
+    {
+      "epoch": 0.5622994407261391,
+      "learning_rate": 8.754194553956175e-05,
+      "loss": 0.7143,
+      "mean_token_accuracy": 0.7848659813404083,
+      "num_tokens": 142229131.0,
+      "step": 61330
+    },
+    {
+      "epoch": 0.5623911249656184,
+      "learning_rate": 8.75236086916659e-05,
+      "loss": 0.7897,
+      "mean_token_accuracy": 0.7600663006305695,
+      "num_tokens": 142253859.0,
+      "step": 61340
+    },
+    {
+      "epoch": 0.5624828092050976,
+      "learning_rate": 8.750527184377007e-05,
+      "loss": 0.7186,
+      "mean_token_accuracy": 0.7817135035991669,
+      "num_tokens": 142279014.0,
+      "step": 61350
+    },
+    {
+      "epoch": 0.5625744934445769,
+      "learning_rate": 8.748693499587422e-05,
+      "loss": 0.7369,
+      "mean_token_accuracy": 0.7738412320613861,
+      "num_tokens": 142304369.0,
+      "step": 61360
+    },
+    {
+      "epoch": 0.5626661776840561,
+      "learning_rate": 8.746859814797837e-05,
+      "loss": 0.7277,
+      "mean_token_accuracy": 0.7809462487697602,
+      "num_tokens": 142329310.0,
+      "step": 61370
+    },
+    {
+      "epoch": 0.5627578619235354,
+      "learning_rate": 8.745026130008252e-05,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.7787182092666626,
+      "num_tokens": 142354353.0,
+      "step": 61380
+    },
+    {
+      "epoch": 0.5628495461630145,
+      "learning_rate": 8.743192445218667e-05,
+      "loss": 0.7509,
+      "mean_token_accuracy": 0.7639503538608551,
+      "num_tokens": 142378367.0,
+      "step": 61390
+    },
+    {
+      "epoch": 0.5629412304024938,
+      "learning_rate": 8.741358760429083e-05,
+      "loss": 0.729,
+      "mean_token_accuracy": 0.7805909156799317,
+      "num_tokens": 142403374.0,
+      "step": 61400
+    },
+    {
+      "epoch": 0.563032914641973,
+      "learning_rate": 8.739525075639498e-05,
+      "loss": 0.779,
+      "mean_token_accuracy": 0.765592074394226,
+      "num_tokens": 142428627.0,
+      "step": 61410
+    },
+    {
+      "epoch": 0.5631245988814523,
+      "learning_rate": 8.737691390849914e-05,
+      "loss": 0.7701,
+      "mean_token_accuracy": 0.7750620067119598,
+      "num_tokens": 142453343.0,
+      "step": 61420
+    },
+    {
+      "epoch": 0.5632162831209315,
+      "learning_rate": 8.735857706060328e-05,
+      "loss": 0.7202,
+      "mean_token_accuracy": 0.7815834820270539,
+      "num_tokens": 142478190.0,
+      "step": 61430
+    },
+    {
+      "epoch": 0.5633079673604108,
+      "learning_rate": 8.734024021270743e-05,
+      "loss": 0.7378,
+      "mean_token_accuracy": 0.7742599308490753,
+      "num_tokens": 142504276.0,
+      "step": 61440
+    },
+    {
+      "epoch": 0.56339965159989,
+      "learning_rate": 8.73219033648116e-05,
+      "loss": 0.7224,
+      "mean_token_accuracy": 0.7844827890396118,
+      "num_tokens": 142529124.0,
+      "step": 61450
+    },
+    {
+      "epoch": 0.5634913358393692,
+      "learning_rate": 8.730356651691575e-05,
+      "loss": 0.717,
+      "mean_token_accuracy": 0.7813876330852508,
+      "num_tokens": 142553543.0,
+      "step": 61460
+    },
+    {
+      "epoch": 0.5635830200788484,
+      "learning_rate": 8.72852296690199e-05,
+      "loss": 0.7452,
+      "mean_token_accuracy": 0.773801612854004,
+      "num_tokens": 142579118.0,
+      "step": 61470
+    },
+    {
+      "epoch": 0.5636747043183277,
+      "learning_rate": 8.726689282112405e-05,
+      "loss": 0.7275,
+      "mean_token_accuracy": 0.7764965236186981,
+      "num_tokens": 142603774.0,
+      "step": 61480
+    },
+    {
+      "epoch": 0.5637663885578069,
+      "learning_rate": 8.72485559732282e-05,
+      "loss": 0.7466,
+      "mean_token_accuracy": 0.7742263793945312,
+      "num_tokens": 142629028.0,
+      "step": 61490
+    },
+    {
+      "epoch": 0.5638580727972862,
+      "learning_rate": 8.723021912533236e-05,
+      "loss": 0.7397,
+      "mean_token_accuracy": 0.7819108486175537,
+      "num_tokens": 142654163.0,
+      "step": 61500
+    },
+    {
+      "epoch": 0.5639497570367654,
+      "learning_rate": 8.721188227743651e-05,
+      "loss": 0.7516,
+      "mean_token_accuracy": 0.772611802816391,
+      "num_tokens": 142678161.0,
+      "step": 61510
+    },
+    {
+      "epoch": 0.5640414412762446,
+      "learning_rate": 8.719354542954067e-05,
+      "loss": 0.7223,
+      "mean_token_accuracy": 0.7816067576408386,
+      "num_tokens": 142702549.0,
+      "step": 61520
+    },
+    {
+      "epoch": 0.5641331255157238,
+      "learning_rate": 8.717520858164481e-05,
+      "loss": 0.7228,
+      "mean_token_accuracy": 0.7792167603969574,
+      "num_tokens": 142728113.0,
+      "step": 61530
+    },
+    {
+      "epoch": 0.5642248097552031,
+      "learning_rate": 8.715687173374896e-05,
+      "loss": 0.7361,
+      "mean_token_accuracy": 0.780789577960968,
+      "num_tokens": 142753801.0,
+      "step": 61540
+    },
+    {
+      "epoch": 0.5643164939946823,
+      "learning_rate": 8.713853488585313e-05,
+      "loss": 0.7092,
+      "mean_token_accuracy": 0.7850222945213318,
+      "num_tokens": 142778988.0,
+      "step": 61550
+    },
+    {
+      "epoch": 0.5644081782341616,
+      "learning_rate": 8.712019803795728e-05,
+      "loss": 0.7199,
+      "mean_token_accuracy": 0.7778768181800843,
+      "num_tokens": 142804062.0,
+      "step": 61560
+    },
+    {
+      "epoch": 0.5644998624736408,
+      "learning_rate": 8.710186119006143e-05,
+      "loss": 0.7766,
+      "mean_token_accuracy": 0.7678954005241394,
+      "num_tokens": 142828820.0,
+      "step": 61570
+    },
+    {
+      "epoch": 0.5645915467131201,
+      "learning_rate": 8.708352434216559e-05,
+      "loss": 0.7465,
+      "mean_token_accuracy": 0.7716767072677613,
+      "num_tokens": 142853280.0,
+      "step": 61580
+    },
+    {
+      "epoch": 0.5646832309525992,
+      "learning_rate": 8.706518749426974e-05,
+      "loss": 0.7153,
+      "mean_token_accuracy": 0.7797940015792847,
+      "num_tokens": 142877595.0,
+      "step": 61590
+    },
+    {
+      "epoch": 0.5647749151920785,
+      "learning_rate": 8.704685064637389e-05,
+      "loss": 0.7345,
+      "mean_token_accuracy": 0.7749894678592681,
+      "num_tokens": 142902572.0,
+      "step": 61600
+    },
+    {
+      "epoch": 0.5648665994315577,
+      "learning_rate": 8.702851379847805e-05,
+      "loss": 0.7219,
+      "mean_token_accuracy": 0.7763688921928406,
+      "num_tokens": 142927733.0,
+      "step": 61610
+    },
+    {
+      "epoch": 0.564958283671037,
+      "learning_rate": 8.701017695058221e-05,
+      "loss": 0.7349,
+      "mean_token_accuracy": 0.7736750721931458,
+      "num_tokens": 142952978.0,
+      "step": 61620
+    },
+    {
+      "epoch": 0.5650499679105162,
+      "learning_rate": 8.699184010268635e-05,
+      "loss": 0.7156,
+      "mean_token_accuracy": 0.786289119720459,
+      "num_tokens": 142977716.0,
+      "step": 61630
+    },
+    {
+      "epoch": 0.5651416521499955,
+      "learning_rate": 8.69735032547905e-05,
+      "loss": 0.7285,
+      "mean_token_accuracy": 0.7771040678024292,
+      "num_tokens": 143002377.0,
+      "step": 61640
+    },
+    {
+      "epoch": 0.5652333363894746,
+      "learning_rate": 8.695516640689466e-05,
+      "loss": 0.7564,
+      "mean_token_accuracy": 0.7740037858486175,
+      "num_tokens": 143027062.0,
+      "step": 61650
+    },
+    {
+      "epoch": 0.5653250206289538,
+      "learning_rate": 8.693682955899881e-05,
+      "loss": 0.6785,
+      "mean_token_accuracy": 0.7890504896640778,
+      "num_tokens": 143052356.0,
+      "step": 61660
+    },
+    {
+      "epoch": 0.5654167048684331,
+      "learning_rate": 8.691849271110297e-05,
+      "loss": 0.7515,
+      "mean_token_accuracy": 0.7712706089019775,
+      "num_tokens": 143077597.0,
+      "step": 61670
+    },
+    {
+      "epoch": 0.5655083891079123,
+      "learning_rate": 8.690015586320712e-05,
+      "loss": 0.7017,
+      "mean_token_accuracy": 0.7918259799480438,
+      "num_tokens": 143102384.0,
+      "step": 61680
+    },
+    {
+      "epoch": 0.5656000733473916,
+      "learning_rate": 8.688181901531127e-05,
+      "loss": 0.718,
+      "mean_token_accuracy": 0.7809239447116851,
+      "num_tokens": 143127265.0,
+      "step": 61690
+    },
+    {
+      "epoch": 0.5656917575868708,
+      "learning_rate": 8.686348216741542e-05,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.7870328426361084,
+      "num_tokens": 143152889.0,
+      "step": 61700
+    },
+    {
+      "epoch": 0.5657834418263501,
+      "learning_rate": 8.684514531951959e-05,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.7889711618423462,
+      "num_tokens": 143178063.0,
+      "step": 61710
+    },
+    {
+      "epoch": 0.5658751260658292,
+      "learning_rate": 8.682680847162374e-05,
+      "loss": 0.7462,
+      "mean_token_accuracy": 0.7709170818328858,
+      "num_tokens": 143203681.0,
+      "step": 61720
+    },
+    {
+      "epoch": 0.5659668103053085,
+      "learning_rate": 8.680847162372788e-05,
+      "loss": 0.7135,
+      "mean_token_accuracy": 0.786488926410675,
+      "num_tokens": 143229020.0,
+      "step": 61730
+    },
+    {
+      "epoch": 0.5660584945447877,
+      "learning_rate": 8.679013477583204e-05,
+      "loss": 0.7145,
+      "mean_token_accuracy": 0.7814090788364411,
+      "num_tokens": 143253751.0,
+      "step": 61740
+    },
+    {
+      "epoch": 0.566150178784267,
+      "learning_rate": 8.67717979279362e-05,
+      "loss": 0.6934,
+      "mean_token_accuracy": 0.7881740212440491,
+      "num_tokens": 143278772.0,
+      "step": 61750
+    },
+    {
+      "epoch": 0.5662418630237462,
+      "learning_rate": 8.675346108004035e-05,
+      "loss": 0.7345,
+      "mean_token_accuracy": 0.7782626450061798,
+      "num_tokens": 143303803.0,
+      "step": 61760
+    },
+    {
+      "epoch": 0.5663335472632255,
+      "learning_rate": 8.67351242321445e-05,
+      "loss": 0.696,
+      "mean_token_accuracy": 0.7860243439674377,
+      "num_tokens": 143328999.0,
+      "step": 61770
+    },
+    {
+      "epoch": 0.5664252315027047,
+      "learning_rate": 8.671678738424865e-05,
+      "loss": 0.7314,
+      "mean_token_accuracy": 0.7823576271533966,
+      "num_tokens": 143354398.0,
+      "step": 61780
+    },
+    {
+      "epoch": 0.5665169157421839,
+      "learning_rate": 8.66984505363528e-05,
+      "loss": 0.7147,
+      "mean_token_accuracy": 0.7784162044525147,
+      "num_tokens": 143378353.0,
+      "step": 61790
+    },
+    {
+      "epoch": 0.5666085999816631,
+      "learning_rate": 8.668011368845695e-05,
+      "loss": 0.7071,
+      "mean_token_accuracy": 0.7821170151233673,
+      "num_tokens": 143403959.0,
+      "step": 61800
+    },
+    {
+      "epoch": 0.5667002842211424,
+      "learning_rate": 8.666177684056112e-05,
+      "loss": 0.7434,
+      "mean_token_accuracy": 0.7774074018001557,
+      "num_tokens": 143429416.0,
+      "step": 61810
+    },
+    {
+      "epoch": 0.5667919684606216,
+      "learning_rate": 8.664343999266527e-05,
+      "loss": 0.7473,
+      "mean_token_accuracy": 0.7764745056629181,
+      "num_tokens": 143455050.0,
+      "step": 61820
+    },
+    {
+      "epoch": 0.5668836527001009,
+      "learning_rate": 8.662510314476941e-05,
+      "loss": 0.7254,
+      "mean_token_accuracy": 0.7757602453231811,
+      "num_tokens": 143480438.0,
+      "step": 61830
+    },
+    {
+      "epoch": 0.5669753369395801,
+      "learning_rate": 8.660676629687358e-05,
+      "loss": 0.7415,
+      "mean_token_accuracy": 0.7818674862384796,
+      "num_tokens": 143503868.0,
+      "step": 61840
+    },
+    {
+      "epoch": 0.5670670211790593,
+      "learning_rate": 8.658842944897773e-05,
+      "loss": 0.715,
+      "mean_token_accuracy": 0.7772590398788453,
+      "num_tokens": 143527856.0,
+      "step": 61850
+    },
+    {
+      "epoch": 0.5671587054185385,
+      "learning_rate": 8.657009260108188e-05,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.7781527757644653,
+      "num_tokens": 143552042.0,
+      "step": 61860
+    },
+    {
+      "epoch": 0.5672503896580178,
+      "learning_rate": 8.655175575318603e-05,
+      "loss": 0.7236,
+      "mean_token_accuracy": 0.7846739411354064,
+      "num_tokens": 143577375.0,
+      "step": 61870
+    },
+    {
+      "epoch": 0.567342073897497,
+      "learning_rate": 8.653341890529018e-05,
+      "loss": 0.7422,
+      "mean_token_accuracy": 0.7735691845417023,
+      "num_tokens": 143602524.0,
+      "step": 61880
+    },
+    {
+      "epoch": 0.5674337581369763,
+      "learning_rate": 8.651508205739433e-05,
+      "loss": 0.6834,
+      "mean_token_accuracy": 0.7884851515293121,
+      "num_tokens": 143627104.0,
+      "step": 61890
+    },
+    {
+      "epoch": 0.5675254423764555,
+      "learning_rate": 8.649674520949849e-05,
+      "loss": 0.7026,
+      "mean_token_accuracy": 0.7839809358119965,
+      "num_tokens": 143652664.0,
+      "step": 61900
+    },
+    {
+      "epoch": 0.5676171266159348,
+      "learning_rate": 8.647840836160265e-05,
+      "loss": 0.7315,
+      "mean_token_accuracy": 0.7812272548675537,
+      "num_tokens": 143677592.0,
+      "step": 61910
+    },
+    {
+      "epoch": 0.5677088108554139,
+      "learning_rate": 8.64600715137068e-05,
+      "loss": 0.7349,
+      "mean_token_accuracy": 0.7826238691806793,
+      "num_tokens": 143702338.0,
+      "step": 61920
+    },
+    {
+      "epoch": 0.5678004950948932,
+      "learning_rate": 8.644173466581094e-05,
+      "loss": 0.7132,
+      "mean_token_accuracy": 0.7844032406806946,
+      "num_tokens": 143727529.0,
+      "step": 61930
+    },
+    {
+      "epoch": 0.5678921793343724,
+      "learning_rate": 8.642339781791511e-05,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.7779807567596435,
+      "num_tokens": 143752375.0,
+      "step": 61940
+    },
+    {
+      "epoch": 0.5679838635738517,
+      "learning_rate": 8.640506097001926e-05,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.7860865294933319,
+      "num_tokens": 143776787.0,
+      "step": 61950
+    },
+    {
+      "epoch": 0.5680755478133309,
+      "learning_rate": 8.638672412212341e-05,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7720317900180816,
+      "num_tokens": 143801991.0,
+      "step": 61960
+    },
+    {
+      "epoch": 0.5681672320528102,
+      "learning_rate": 8.636838727422756e-05,
+      "loss": 0.7483,
+      "mean_token_accuracy": 0.7743564248085022,
+      "num_tokens": 143827644.0,
+      "step": 61970
+    },
+    {
+      "epoch": 0.5682589162922893,
+      "learning_rate": 8.635005042633171e-05,
+      "loss": 0.6841,
+      "mean_token_accuracy": 0.7905368804931641,
+      "num_tokens": 143851690.0,
+      "step": 61980
+    },
+    {
+      "epoch": 0.5683506005317686,
+      "learning_rate": 8.633171357843587e-05,
+      "loss": 0.7342,
+      "mean_token_accuracy": 0.7730114579200744,
+      "num_tokens": 143876895.0,
+      "step": 61990
+    },
+    {
+      "epoch": 0.5684422847712478,
+      "learning_rate": 8.631337673054003e-05,
+      "loss": 0.7443,
+      "mean_token_accuracy": 0.7728031933307647,
+      "num_tokens": 143902310.0,
+      "step": 62000
+    },
+    {
+      "epoch": 0.5685339690107271,
+      "learning_rate": 8.629503988264418e-05,
+      "loss": 0.7007,
+      "mean_token_accuracy": 0.7833173513412476,
+      "num_tokens": 143927201.0,
+      "step": 62010
+    },
+    {
+      "epoch": 0.5686256532502063,
+      "learning_rate": 8.627670303474834e-05,
+      "loss": 0.7427,
+      "mean_token_accuracy": 0.7764660060405731,
+      "num_tokens": 143951563.0,
+      "step": 62020
+    },
+    {
+      "epoch": 0.5687173374896856,
+      "learning_rate": 8.625836618685247e-05,
+      "loss": 0.6761,
+      "mean_token_accuracy": 0.7847448766231537,
+      "num_tokens": 143976856.0,
+      "step": 62030
+    },
+    {
+      "epoch": 0.5688090217291648,
+      "learning_rate": 8.624002933895664e-05,
+      "loss": 0.7023,
+      "mean_token_accuracy": 0.7841219067573547,
+      "num_tokens": 144001330.0,
+      "step": 62040
+    },
+    {
+      "epoch": 0.568900705968644,
+      "learning_rate": 8.622169249106079e-05,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.773005610704422,
+      "num_tokens": 144027144.0,
+      "step": 62050
+    },
+    {
+      "epoch": 0.5689923902081232,
+      "learning_rate": 8.620335564316494e-05,
+      "loss": 0.6943,
+      "mean_token_accuracy": 0.7877039730548858,
+      "num_tokens": 144052308.0,
+      "step": 62060
+    },
+    {
+      "epoch": 0.5690840744476025,
+      "learning_rate": 8.61850187952691e-05,
+      "loss": 0.7176,
+      "mean_token_accuracy": 0.78382129073143,
+      "num_tokens": 144077742.0,
+      "step": 62070
+    },
+    {
+      "epoch": 0.5691757586870817,
+      "learning_rate": 8.616668194737325e-05,
+      "loss": 0.7193,
+      "mean_token_accuracy": 0.7807154059410095,
+      "num_tokens": 144103004.0,
+      "step": 62080
+    },
+    {
+      "epoch": 0.569267442926561,
+      "learning_rate": 8.61483450994774e-05,
+      "loss": 0.7071,
+      "mean_token_accuracy": 0.7818564057350159,
+      "num_tokens": 144127832.0,
+      "step": 62090
+    },
+    {
+      "epoch": 0.5693591271660402,
+      "learning_rate": 8.613000825158156e-05,
+      "loss": 0.7316,
+      "mean_token_accuracy": 0.7813744127750397,
+      "num_tokens": 144152562.0,
+      "step": 62100
+    },
+    {
+      "epoch": 0.5694508114055193,
+      "learning_rate": 8.611167140368572e-05,
+      "loss": 0.7125,
+      "mean_token_accuracy": 0.7842976093292237,
+      "num_tokens": 144177810.0,
+      "step": 62110
+    },
+    {
+      "epoch": 0.5695424956449986,
+      "learning_rate": 8.609333455578987e-05,
+      "loss": 0.7142,
+      "mean_token_accuracy": 0.7869325280189514,
+      "num_tokens": 144202645.0,
+      "step": 62120
+    },
+    {
+      "epoch": 0.5696341798844778,
+      "learning_rate": 8.6074997707894e-05,
+      "loss": 0.6733,
+      "mean_token_accuracy": 0.7955866277217865,
+      "num_tokens": 144227971.0,
+      "step": 62130
+    },
+    {
+      "epoch": 0.5697258641239571,
+      "learning_rate": 8.605666085999817e-05,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.7834891974925995,
+      "num_tokens": 144252837.0,
+      "step": 62140
+    },
+    {
+      "epoch": 0.5698175483634363,
+      "learning_rate": 8.603832401210232e-05,
+      "loss": 0.7068,
+      "mean_token_accuracy": 0.7873168587684631,
+      "num_tokens": 144278083.0,
+      "step": 62150
+    },
+    {
+      "epoch": 0.5699092326029156,
+      "learning_rate": 8.601998716420648e-05,
+      "loss": 0.6749,
+      "mean_token_accuracy": 0.7899958074092865,
+      "num_tokens": 144302066.0,
+      "step": 62160
+    },
+    {
+      "epoch": 0.5700009168423948,
+      "learning_rate": 8.600165031631063e-05,
+      "loss": 0.7229,
+      "mean_token_accuracy": 0.7795552492141724,
+      "num_tokens": 144326831.0,
+      "step": 62170
+    },
+    {
+      "epoch": 0.570092601081874,
+      "learning_rate": 8.598331346841478e-05,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.7803704261779785,
+      "num_tokens": 144351891.0,
+      "step": 62180
+    },
+    {
+      "epoch": 0.5701842853213532,
+      "learning_rate": 8.596497662051893e-05,
+      "loss": 0.7318,
+      "mean_token_accuracy": 0.7827034533023834,
+      "num_tokens": 144377216.0,
+      "step": 62190
+    },
+    {
+      "epoch": 0.5702759695608325,
+      "learning_rate": 8.59466397726231e-05,
+      "loss": 0.7335,
+      "mean_token_accuracy": 0.7783604204654694,
+      "num_tokens": 144402222.0,
+      "step": 62200
+    },
+    {
+      "epoch": 0.5703676538003117,
+      "learning_rate": 8.592830292472725e-05,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.7828047335147857,
+      "num_tokens": 144427150.0,
+      "step": 62210
+    },
+    {
+      "epoch": 0.570459338039791,
+      "learning_rate": 8.59099660768314e-05,
+      "loss": 0.7695,
+      "mean_token_accuracy": 0.763064694404602,
+      "num_tokens": 144452635.0,
+      "step": 62220
+    },
+    {
+      "epoch": 0.5705510222792702,
+      "learning_rate": 8.589162922893555e-05,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.7895868837833404,
+      "num_tokens": 144477485.0,
+      "step": 62230
+    },
+    {
+      "epoch": 0.5706427065187494,
+      "learning_rate": 8.58732923810397e-05,
+      "loss": 0.7187,
+      "mean_token_accuracy": 0.7818442523479462,
+      "num_tokens": 144501435.0,
+      "step": 62240
+    },
+    {
+      "epoch": 0.5707343907582286,
+      "learning_rate": 8.585495553314386e-05,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.7730175495147705,
+      "num_tokens": 144526511.0,
+      "step": 62250
+    },
+    {
+      "epoch": 0.5708260749977079,
+      "learning_rate": 8.583661868524801e-05,
+      "loss": 0.6967,
+      "mean_token_accuracy": 0.7853508651256561,
+      "num_tokens": 144551349.0,
+      "step": 62260
+    },
+    {
+      "epoch": 0.5709177592371871,
+      "learning_rate": 8.581828183735216e-05,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7731032490730285,
+      "num_tokens": 144576097.0,
+      "step": 62270
+    },
+    {
+      "epoch": 0.5710094434766664,
+      "learning_rate": 8.579994498945631e-05,
+      "loss": 0.737,
+      "mean_token_accuracy": 0.78244908452034,
+      "num_tokens": 144601027.0,
+      "step": 62280
+    },
+    {
+      "epoch": 0.5711011277161456,
+      "learning_rate": 8.578160814156046e-05,
+      "loss": 0.6968,
+      "mean_token_accuracy": 0.7845434546470642,
+      "num_tokens": 144626786.0,
+      "step": 62290
+    },
+    {
+      "epoch": 0.5711928119556249,
+      "learning_rate": 8.576327129366463e-05,
+      "loss": 0.6985,
+      "mean_token_accuracy": 0.7877167165279388,
+      "num_tokens": 144651979.0,
+      "step": 62300
+    },
+    {
+      "epoch": 0.571284496195104,
+      "learning_rate": 8.574493444576878e-05,
+      "loss": 0.7196,
+      "mean_token_accuracy": 0.7870518088340759,
+      "num_tokens": 144676752.0,
+      "step": 62310
+    },
+    {
+      "epoch": 0.5713761804345833,
+      "learning_rate": 8.572659759787293e-05,
+      "loss": 0.6851,
+      "mean_token_accuracy": 0.788501638174057,
+      "num_tokens": 144701565.0,
+      "step": 62320
+    },
+    {
+      "epoch": 0.5714678646740625,
+      "learning_rate": 8.570826074997708e-05,
+      "loss": 0.7642,
+      "mean_token_accuracy": 0.7776337265968323,
+      "num_tokens": 144726064.0,
+      "step": 62330
+    },
+    {
+      "epoch": 0.5715595489135418,
+      "learning_rate": 8.568992390208124e-05,
+      "loss": 0.7043,
+      "mean_token_accuracy": 0.7870524108409882,
+      "num_tokens": 144751769.0,
+      "step": 62340
+    },
+    {
+      "epoch": 0.571651233153021,
+      "learning_rate": 8.567158705418539e-05,
+      "loss": 0.7156,
+      "mean_token_accuracy": 0.7824094951152801,
+      "num_tokens": 144776972.0,
+      "step": 62350
+    },
+    {
+      "epoch": 0.5717429173925003,
+      "learning_rate": 8.565325020628955e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.7920378804206848,
+      "num_tokens": 144802057.0,
+      "step": 62360
+    },
+    {
+      "epoch": 0.5718346016319794,
+      "learning_rate": 8.563491335839369e-05,
+      "loss": 0.7267,
+      "mean_token_accuracy": 0.7790842652320862,
+      "num_tokens": 144826664.0,
+      "step": 62370
+    },
+    {
+      "epoch": 0.5719262858714587,
+      "learning_rate": 8.561657651049784e-05,
+      "loss": 0.6857,
+      "mean_token_accuracy": 0.7862694978713989,
+      "num_tokens": 144851285.0,
+      "step": 62380
+    },
+    {
+      "epoch": 0.5720179701109379,
+      "learning_rate": 8.5598239662602e-05,
+      "loss": 0.7211,
+      "mean_token_accuracy": 0.7760497152805328,
+      "num_tokens": 144875440.0,
+      "step": 62390
+    },
+    {
+      "epoch": 0.5721096543504172,
+      "learning_rate": 8.557990281470616e-05,
+      "loss": 0.7187,
+      "mean_token_accuracy": 0.7806708574295044,
+      "num_tokens": 144900548.0,
+      "step": 62400
+    },
+    {
+      "epoch": 0.5722013385898964,
+      "learning_rate": 8.556156596681031e-05,
+      "loss": 0.6712,
+      "mean_token_accuracy": 0.7920472621917725,
+      "num_tokens": 144925434.0,
+      "step": 62410
+    },
+    {
+      "epoch": 0.5722930228293757,
+      "learning_rate": 8.554322911891446e-05,
+      "loss": 0.6869,
+      "mean_token_accuracy": 0.7868934035301208,
+      "num_tokens": 144951111.0,
+      "step": 62420
+    },
+    {
+      "epoch": 0.5723847070688549,
+      "learning_rate": 8.552489227101862e-05,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.7944657862186432,
+      "num_tokens": 144976139.0,
+      "step": 62430
+    },
+    {
+      "epoch": 0.572476391308334,
+      "learning_rate": 8.550655542312277e-05,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.7857164978981018,
+      "num_tokens": 145001543.0,
+      "step": 62440
+    },
+    {
+      "epoch": 0.5725680755478133,
+      "learning_rate": 8.548821857522692e-05,
+      "loss": 0.7243,
+      "mean_token_accuracy": 0.7788842976093292,
+      "num_tokens": 145026031.0,
+      "step": 62450
+    },
+    {
+      "epoch": 0.5726597597872926,
+      "learning_rate": 8.546988172733109e-05,
+      "loss": 0.6661,
+      "mean_token_accuracy": 0.7954781174659729,
+      "num_tokens": 145051023.0,
+      "step": 62460
+    },
+    {
+      "epoch": 0.5727514440267718,
+      "learning_rate": 8.545154487943522e-05,
+      "loss": 0.7176,
+      "mean_token_accuracy": 0.7817997395992279,
+      "num_tokens": 145076003.0,
+      "step": 62470
+    },
+    {
+      "epoch": 0.5728431282662511,
+      "learning_rate": 8.543320803153938e-05,
+      "loss": 0.749,
+      "mean_token_accuracy": 0.7724713504314422,
+      "num_tokens": 145100516.0,
+      "step": 62480
+    },
+    {
+      "epoch": 0.5729348125057303,
+      "learning_rate": 8.541487118364354e-05,
+      "loss": 0.7602,
+      "mean_token_accuracy": 0.7740296423435211,
+      "num_tokens": 145124950.0,
+      "step": 62490
+    },
+    {
+      "epoch": 0.5730264967452094,
+      "learning_rate": 8.539653433574769e-05,
+      "loss": 0.7428,
+      "mean_token_accuracy": 0.7710140287876129,
+      "num_tokens": 145149959.0,
+      "step": 62500
+    },
+    {
+      "epoch": 0.5731181809846887,
+      "learning_rate": 8.537819748785185e-05,
+      "loss": 0.7218,
+      "mean_token_accuracy": 0.7826135754585266,
+      "num_tokens": 145175594.0,
+      "step": 62510
+    },
+    {
+      "epoch": 0.573209865224168,
+      "learning_rate": 8.5359860639956e-05,
+      "loss": 0.689,
+      "mean_token_accuracy": 0.7823807060718536,
+      "num_tokens": 145200677.0,
+      "step": 62520
+    },
+    {
+      "epoch": 0.5733015494636472,
+      "learning_rate": 8.534152379206015e-05,
+      "loss": 0.7006,
+      "mean_token_accuracy": 0.7830764055252075,
+      "num_tokens": 145226189.0,
+      "step": 62530
+    },
+    {
+      "epoch": 0.5733932337031264,
+      "learning_rate": 8.53231869441643e-05,
+      "loss": 0.6916,
+      "mean_token_accuracy": 0.7884008705615997,
+      "num_tokens": 145250934.0,
+      "step": 62540
+    },
+    {
+      "epoch": 0.5734849179426057,
+      "learning_rate": 8.530485009626845e-05,
+      "loss": 0.7046,
+      "mean_token_accuracy": 0.7841234266757965,
+      "num_tokens": 145276332.0,
+      "step": 62550
+    },
+    {
+      "epoch": 0.573576602182085,
+      "learning_rate": 8.528651324837262e-05,
+      "loss": 0.7187,
+      "mean_token_accuracy": 0.7814329028129577,
+      "num_tokens": 145301519.0,
+      "step": 62560
+    },
+    {
+      "epoch": 0.5736682864215641,
+      "learning_rate": 8.526817640047676e-05,
+      "loss": 0.6778,
+      "mean_token_accuracy": 0.7847923219203949,
+      "num_tokens": 145325973.0,
+      "step": 62570
+    },
+    {
+      "epoch": 0.5737599706610433,
+      "learning_rate": 8.524983955258091e-05,
+      "loss": 0.7402,
+      "mean_token_accuracy": 0.7722189664840698,
+      "num_tokens": 145350919.0,
+      "step": 62580
+    },
+    {
+      "epoch": 0.5738516549005226,
+      "learning_rate": 8.523150270468507e-05,
+      "loss": 0.7065,
+      "mean_token_accuracy": 0.7750476062297821,
+      "num_tokens": 145375526.0,
+      "step": 62590
+    },
+    {
+      "epoch": 0.5739433391400018,
+      "learning_rate": 8.521316585678923e-05,
+      "loss": 0.6907,
+      "mean_token_accuracy": 0.782847660779953,
+      "num_tokens": 145399627.0,
+      "step": 62600
+    },
+    {
+      "epoch": 0.5740350233794811,
+      "learning_rate": 8.519482900889338e-05,
+      "loss": 0.7037,
+      "mean_token_accuracy": 0.7839486539363861,
+      "num_tokens": 145425176.0,
+      "step": 62610
+    },
+    {
+      "epoch": 0.5741267076189603,
+      "learning_rate": 8.517649216099753e-05,
+      "loss": 0.7247,
+      "mean_token_accuracy": 0.7804330945014953,
+      "num_tokens": 145450645.0,
+      "step": 62620
+    },
+    {
+      "epoch": 0.5742183918584395,
+      "learning_rate": 8.515815531310168e-05,
+      "loss": 0.7109,
+      "mean_token_accuracy": 0.7823985755443573,
+      "num_tokens": 145476768.0,
+      "step": 62630
+    },
+    {
+      "epoch": 0.5743100760979187,
+      "learning_rate": 8.513981846520583e-05,
+      "loss": 0.6857,
+      "mean_token_accuracy": 0.7896067500114441,
+      "num_tokens": 145502852.0,
+      "step": 62640
+    },
+    {
+      "epoch": 0.574401760337398,
+      "learning_rate": 8.512148161730998e-05,
+      "loss": 0.7065,
+      "mean_token_accuracy": 0.7823691308498383,
+      "num_tokens": 145527394.0,
+      "step": 62650
+    },
+    {
+      "epoch": 0.5744934445768772,
+      "learning_rate": 8.510314476941415e-05,
+      "loss": 0.727,
+      "mean_token_accuracy": 0.7741926550865174,
+      "num_tokens": 145552203.0,
+      "step": 62660
+    },
+    {
+      "epoch": 0.5745851288163565,
+      "learning_rate": 8.508480792151829e-05,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.7900151491165162,
+      "num_tokens": 145577248.0,
+      "step": 62670
+    },
+    {
+      "epoch": 0.5746768130558357,
+      "learning_rate": 8.506647107362244e-05,
+      "loss": 0.7498,
+      "mean_token_accuracy": 0.7721031606197357,
+      "num_tokens": 145601599.0,
+      "step": 62680
+    },
+    {
+      "epoch": 0.574768497295315,
+      "learning_rate": 8.50481342257266e-05,
+      "loss": 0.7289,
+      "mean_token_accuracy": 0.7765326738357544,
+      "num_tokens": 145626537.0,
+      "step": 62690
+    },
+    {
+      "epoch": 0.5748601815347941,
+      "learning_rate": 8.502979737783076e-05,
+      "loss": 0.6884,
+      "mean_token_accuracy": 0.7924788475036622,
+      "num_tokens": 145651313.0,
+      "step": 62700
+    },
+    {
+      "epoch": 0.5749518657742734,
+      "learning_rate": 8.501146052993491e-05,
+      "loss": 0.7404,
+      "mean_token_accuracy": 0.7778672039508819,
+      "num_tokens": 145676890.0,
+      "step": 62710
+    },
+    {
+      "epoch": 0.5750435500137526,
+      "learning_rate": 8.499312368203906e-05,
+      "loss": 0.7038,
+      "mean_token_accuracy": 0.7784171462059021,
+      "num_tokens": 145701903.0,
+      "step": 62720
+    },
+    {
+      "epoch": 0.5751352342532319,
+      "learning_rate": 8.497478683414321e-05,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.7763013780117035,
+      "num_tokens": 145726327.0,
+      "step": 62730
+    },
+    {
+      "epoch": 0.5752269184927111,
+      "learning_rate": 8.495644998624737e-05,
+      "loss": 0.704,
+      "mean_token_accuracy": 0.7827533066272736,
+      "num_tokens": 145750743.0,
+      "step": 62740
+    },
+    {
+      "epoch": 0.5753186027321904,
+      "learning_rate": 8.493811313835152e-05,
+      "loss": 0.6647,
+      "mean_token_accuracy": 0.7939302027225494,
+      "num_tokens": 145776675.0,
+      "step": 62750
+    },
+    {
+      "epoch": 0.5754102869716695,
+      "learning_rate": 8.491977629045568e-05,
+      "loss": 0.7105,
+      "mean_token_accuracy": 0.7774204075336456,
+      "num_tokens": 145801596.0,
+      "step": 62760
+    },
+    {
+      "epoch": 0.5755019712111488,
+      "learning_rate": 8.490143944255982e-05,
+      "loss": 0.7042,
+      "mean_token_accuracy": 0.7891483724117279,
+      "num_tokens": 145826322.0,
+      "step": 62770
+    },
+    {
+      "epoch": 0.575593655450628,
+      "learning_rate": 8.488310259466397e-05,
+      "loss": 0.723,
+      "mean_token_accuracy": 0.7788620531558991,
+      "num_tokens": 145850580.0,
+      "step": 62780
+    },
+    {
+      "epoch": 0.5756853396901073,
+      "learning_rate": 8.486476574676814e-05,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7751010417938232,
+      "num_tokens": 145875264.0,
+      "step": 62790
+    },
+    {
+      "epoch": 0.5757770239295865,
+      "learning_rate": 8.484642889887229e-05,
+      "loss": 0.748,
+      "mean_token_accuracy": 0.7660605370998382,
+      "num_tokens": 145899837.0,
+      "step": 62800
+    },
+    {
+      "epoch": 0.5758687081690658,
+      "learning_rate": 8.482809205097644e-05,
+      "loss": 0.7417,
+      "mean_token_accuracy": 0.7785557568073272,
+      "num_tokens": 145925004.0,
+      "step": 62810
+    },
+    {
+      "epoch": 0.575960392408545,
+      "learning_rate": 8.48097552030806e-05,
+      "loss": 0.7149,
+      "mean_token_accuracy": 0.7836163759231567,
+      "num_tokens": 145949662.0,
+      "step": 62820
+    },
+    {
+      "epoch": 0.5760520766480242,
+      "learning_rate": 8.479141835518475e-05,
+      "loss": 0.7236,
+      "mean_token_accuracy": 0.7806240856647492,
+      "num_tokens": 145973929.0,
+      "step": 62830
+    },
+    {
+      "epoch": 0.5761437608875034,
+      "learning_rate": 8.47730815072889e-05,
+      "loss": 0.744,
+      "mean_token_accuracy": 0.7761898577213288,
+      "num_tokens": 145999163.0,
+      "step": 62840
+    },
+    {
+      "epoch": 0.5762354451269827,
+      "learning_rate": 8.475474465939306e-05,
+      "loss": 0.7314,
+      "mean_token_accuracy": 0.7833976149559021,
+      "num_tokens": 146023082.0,
+      "step": 62850
+    },
+    {
+      "epoch": 0.5763271293664619,
+      "learning_rate": 8.473640781149721e-05,
+      "loss": 0.7209,
+      "mean_token_accuracy": 0.7851142287254333,
+      "num_tokens": 146048387.0,
+      "step": 62860
+    },
+    {
+      "epoch": 0.5764188136059412,
+      "learning_rate": 8.471807096360135e-05,
+      "loss": 0.7106,
+      "mean_token_accuracy": 0.782892906665802,
+      "num_tokens": 146073149.0,
+      "step": 62870
+    },
+    {
+      "epoch": 0.5765104978454204,
+      "learning_rate": 8.46997341157055e-05,
+      "loss": 0.714,
+      "mean_token_accuracy": 0.782066923379898,
+      "num_tokens": 146098305.0,
+      "step": 62880
+    },
+    {
+      "epoch": 0.5766021820848996,
+      "learning_rate": 8.468139726780967e-05,
+      "loss": 0.7266,
+      "mean_token_accuracy": 0.7785180389881134,
+      "num_tokens": 146123449.0,
+      "step": 62890
+    },
+    {
+      "epoch": 0.5766938663243788,
+      "learning_rate": 8.466306041991382e-05,
+      "loss": 0.7007,
+      "mean_token_accuracy": 0.7885490775108337,
+      "num_tokens": 146148490.0,
+      "step": 62900
+    },
+    {
+      "epoch": 0.576785550563858,
+      "learning_rate": 8.464472357201797e-05,
+      "loss": 0.6833,
+      "mean_token_accuracy": 0.788060975074768,
+      "num_tokens": 146174302.0,
+      "step": 62910
+    },
+    {
+      "epoch": 0.5768772348033373,
+      "learning_rate": 8.462638672412213e-05,
+      "loss": 0.7283,
+      "mean_token_accuracy": 0.7802075624465943,
+      "num_tokens": 146199530.0,
+      "step": 62920
+    },
+    {
+      "epoch": 0.5769689190428166,
+      "learning_rate": 8.460804987622628e-05,
+      "loss": 0.7556,
+      "mean_token_accuracy": 0.7705292403697968,
+      "num_tokens": 146224084.0,
+      "step": 62930
+    },
+    {
+      "epoch": 0.5770606032822958,
+      "learning_rate": 8.458971302833043e-05,
+      "loss": 0.7149,
+      "mean_token_accuracy": 0.7895412623882294,
+      "num_tokens": 146249644.0,
+      "step": 62940
+    },
+    {
+      "epoch": 0.577152287521775,
+      "learning_rate": 8.45713761804346e-05,
+      "loss": 0.7132,
+      "mean_token_accuracy": 0.7768167138099671,
+      "num_tokens": 146274811.0,
+      "step": 62950
+    },
+    {
+      "epoch": 0.5772439717612542,
+      "learning_rate": 8.455303933253875e-05,
+      "loss": 0.7043,
+      "mean_token_accuracy": 0.7833518564701081,
+      "num_tokens": 146299642.0,
+      "step": 62960
+    },
+    {
+      "epoch": 0.5773356560007334,
+      "learning_rate": 8.453470248464289e-05,
+      "loss": 0.6982,
+      "mean_token_accuracy": 0.78038569688797,
+      "num_tokens": 146324475.0,
+      "step": 62970
+    },
+    {
+      "epoch": 0.5774273402402127,
+      "learning_rate": 8.451636563674705e-05,
+      "loss": 0.706,
+      "mean_token_accuracy": 0.7805178344249726,
+      "num_tokens": 146349802.0,
+      "step": 62980
+    },
+    {
+      "epoch": 0.577519024479692,
+      "learning_rate": 8.44980287888512e-05,
+      "loss": 0.71,
+      "mean_token_accuracy": 0.7821848571300507,
+      "num_tokens": 146375188.0,
+      "step": 62990
+    },
+    {
+      "epoch": 0.5776107087191712,
+      "learning_rate": 8.447969194095535e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.7967598080635071,
+      "num_tokens": 146400355.0,
+      "step": 63000
+    },
+    {
+      "epoch": 0.5777023929586504,
+      "learning_rate": 8.44613550930595e-05,
+      "loss": 0.7242,
+      "mean_token_accuracy": 0.7823297023773194,
+      "num_tokens": 146425836.0,
+      "step": 63010
+    },
+    {
+      "epoch": 0.5777940771981297,
+      "learning_rate": 8.444301824516366e-05,
+      "loss": 0.7146,
+      "mean_token_accuracy": 0.7926157057285309,
+      "num_tokens": 146451347.0,
+      "step": 63020
+    },
+    {
+      "epoch": 0.5778857614376088,
+      "learning_rate": 8.442468139726781e-05,
+      "loss": 0.6819,
+      "mean_token_accuracy": 0.7876127183437347,
+      "num_tokens": 146476917.0,
+      "step": 63030
+    },
+    {
+      "epoch": 0.5779774456770881,
+      "learning_rate": 8.440634454937196e-05,
+      "loss": 0.6825,
+      "mean_token_accuracy": 0.7935700833797454,
+      "num_tokens": 146501084.0,
+      "step": 63040
+    },
+    {
+      "epoch": 0.5780691299165673,
+      "learning_rate": 8.438800770147613e-05,
+      "loss": 0.7367,
+      "mean_token_accuracy": 0.7797764360904693,
+      "num_tokens": 146525975.0,
+      "step": 63050
+    },
+    {
+      "epoch": 0.5781608141560466,
+      "learning_rate": 8.436967085358028e-05,
+      "loss": 0.7,
+      "mean_token_accuracy": 0.7848899662494659,
+      "num_tokens": 146550896.0,
+      "step": 63060
+    },
+    {
+      "epoch": 0.5782524983955258,
+      "learning_rate": 8.435133400568442e-05,
+      "loss": 0.7035,
+      "mean_token_accuracy": 0.782321286201477,
+      "num_tokens": 146575765.0,
+      "step": 63070
+    },
+    {
+      "epoch": 0.5783441826350051,
+      "learning_rate": 8.433299715778858e-05,
+      "loss": 0.6938,
+      "mean_token_accuracy": 0.7841570675373077,
+      "num_tokens": 146600469.0,
+      "step": 63080
+    },
+    {
+      "epoch": 0.5784358668744842,
+      "learning_rate": 8.431466030989273e-05,
+      "loss": 0.7374,
+      "mean_token_accuracy": 0.7760751903057098,
+      "num_tokens": 146625258.0,
+      "step": 63090
+    },
+    {
+      "epoch": 0.5785275511139635,
+      "learning_rate": 8.429632346199689e-05,
+      "loss": 0.7013,
+      "mean_token_accuracy": 0.7842355668544769,
+      "num_tokens": 146649761.0,
+      "step": 63100
+    },
+    {
+      "epoch": 0.5786192353534427,
+      "learning_rate": 8.427798661410105e-05,
+      "loss": 0.7238,
+      "mean_token_accuracy": 0.7829664289951325,
+      "num_tokens": 146674748.0,
+      "step": 63110
+    },
+    {
+      "epoch": 0.578710919592922,
+      "learning_rate": 8.425964976620519e-05,
+      "loss": 0.6774,
+      "mean_token_accuracy": 0.7827253043651581,
+      "num_tokens": 146699220.0,
+      "step": 63120
+    },
+    {
+      "epoch": 0.5788026038324012,
+      "learning_rate": 8.424131291830934e-05,
+      "loss": 0.72,
+      "mean_token_accuracy": 0.7813050031661988,
+      "num_tokens": 146724612.0,
+      "step": 63130
+    },
+    {
+      "epoch": 0.5788942880718805,
+      "learning_rate": 8.42229760704135e-05,
+      "loss": 0.6826,
+      "mean_token_accuracy": 0.7871101498603821,
+      "num_tokens": 146749692.0,
+      "step": 63140
+    },
+    {
+      "epoch": 0.5789859723113597,
+      "learning_rate": 8.420463922251766e-05,
+      "loss": 0.7123,
+      "mean_token_accuracy": 0.7785660862922669,
+      "num_tokens": 146775434.0,
+      "step": 63150
+    },
+    {
+      "epoch": 0.5790776565508389,
+      "learning_rate": 8.418630237462181e-05,
+      "loss": 0.7478,
+      "mean_token_accuracy": 0.7785642802715301,
+      "num_tokens": 146799815.0,
+      "step": 63160
+    },
+    {
+      "epoch": 0.5791693407903181,
+      "learning_rate": 8.416796552672595e-05,
+      "loss": 0.7073,
+      "mean_token_accuracy": 0.7834717035293579,
+      "num_tokens": 146824601.0,
+      "step": 63170
+    },
+    {
+      "epoch": 0.5792610250297974,
+      "learning_rate": 8.414962867883012e-05,
+      "loss": 0.7155,
+      "mean_token_accuracy": 0.7826242804527282,
+      "num_tokens": 146849643.0,
+      "step": 63180
+    },
+    {
+      "epoch": 0.5793527092692766,
+      "learning_rate": 8.413129183093427e-05,
+      "loss": 0.6899,
+      "mean_token_accuracy": 0.7854468107223511,
+      "num_tokens": 146874481.0,
+      "step": 63190
+    },
+    {
+      "epoch": 0.5794443935087559,
+      "learning_rate": 8.411295498303842e-05,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.7812145709991455,
+      "num_tokens": 146899449.0,
+      "step": 63200
+    },
+    {
+      "epoch": 0.5795360777482351,
+      "learning_rate": 8.409461813514258e-05,
+      "loss": 0.7341,
+      "mean_token_accuracy": 0.779036957025528,
+      "num_tokens": 146924375.0,
+      "step": 63210
+    },
+    {
+      "epoch": 0.5796277619877143,
+      "learning_rate": 8.407628128724672e-05,
+      "loss": 0.6983,
+      "mean_token_accuracy": 0.7868231296539306,
+      "num_tokens": 146949424.0,
+      "step": 63220
+    },
+    {
+      "epoch": 0.5797194462271935,
+      "learning_rate": 8.405794443935087e-05,
+      "loss": 0.7047,
+      "mean_token_accuracy": 0.7889563262462616,
+      "num_tokens": 146973962.0,
+      "step": 63230
+    },
+    {
+      "epoch": 0.5798111304666728,
+      "learning_rate": 8.403960759145503e-05,
+      "loss": 0.6897,
+      "mean_token_accuracy": 0.7857967853546143,
+      "num_tokens": 147000350.0,
+      "step": 63240
+    },
+    {
+      "epoch": 0.579902814706152,
+      "learning_rate": 8.402127074355919e-05,
+      "loss": 0.6933,
+      "mean_token_accuracy": 0.7878205120563507,
+      "num_tokens": 147025143.0,
+      "step": 63250
+    },
+    {
+      "epoch": 0.5799944989456313,
+      "learning_rate": 8.400293389566334e-05,
+      "loss": 0.6937,
+      "mean_token_accuracy": 0.7922333836555481,
+      "num_tokens": 147050652.0,
+      "step": 63260
+    },
+    {
+      "epoch": 0.5800861831851105,
+      "learning_rate": 8.398459704776748e-05,
+      "loss": 0.6654,
+      "mean_token_accuracy": 0.7915775239467621,
+      "num_tokens": 147075218.0,
+      "step": 63270
+    },
+    {
+      "epoch": 0.5801778674245898,
+      "learning_rate": 8.396626019987165e-05,
+      "loss": 0.7133,
+      "mean_token_accuracy": 0.7793969333171844,
+      "num_tokens": 147099487.0,
+      "step": 63280
+    },
+    {
+      "epoch": 0.5802695516640689,
+      "learning_rate": 8.39479233519758e-05,
+      "loss": 0.67,
+      "mean_token_accuracy": 0.7927051484584808,
+      "num_tokens": 147125016.0,
+      "step": 63290
+    },
+    {
+      "epoch": 0.5803612359035482,
+      "learning_rate": 8.392958650407995e-05,
+      "loss": 0.6875,
+      "mean_token_accuracy": 0.78766930103302,
+      "num_tokens": 147149589.0,
+      "step": 63300
+    },
+    {
+      "epoch": 0.5804529201430274,
+      "learning_rate": 8.391124965618412e-05,
+      "loss": 0.6921,
+      "mean_token_accuracy": 0.7830041944980621,
+      "num_tokens": 147175048.0,
+      "step": 63310
+    },
+    {
+      "epoch": 0.5805446043825067,
+      "learning_rate": 8.389291280828825e-05,
+      "loss": 0.6755,
+      "mean_token_accuracy": 0.7940853297710418,
+      "num_tokens": 147199705.0,
+      "step": 63320
+    },
+    {
+      "epoch": 0.5806362886219859,
+      "learning_rate": 8.38745759603924e-05,
+      "loss": 0.7063,
+      "mean_token_accuracy": 0.7832659006118774,
+      "num_tokens": 147224235.0,
+      "step": 63330
+    },
+    {
+      "epoch": 0.5807279728614652,
+      "learning_rate": 8.385623911249657e-05,
+      "loss": 0.7003,
+      "mean_token_accuracy": 0.7910520970821381,
+      "num_tokens": 147248895.0,
+      "step": 63340
+    },
+    {
+      "epoch": 0.5808196571009443,
+      "learning_rate": 8.383790226460072e-05,
+      "loss": 0.7207,
+      "mean_token_accuracy": 0.7776158273220062,
+      "num_tokens": 147273937.0,
+      "step": 63350
+    },
+    {
+      "epoch": 0.5809113413404235,
+      "learning_rate": 8.381956541670488e-05,
+      "loss": 0.7092,
+      "mean_token_accuracy": 0.7816914856433869,
+      "num_tokens": 147298958.0,
+      "step": 63360
+    },
+    {
+      "epoch": 0.5810030255799028,
+      "learning_rate": 8.380122856880901e-05,
+      "loss": 0.7231,
+      "mean_token_accuracy": 0.7773494303226471,
+      "num_tokens": 147324865.0,
+      "step": 63370
+    },
+    {
+      "epoch": 0.581094709819382,
+      "learning_rate": 8.378289172091318e-05,
+      "loss": 0.7104,
+      "mean_token_accuracy": 0.7845860540866851,
+      "num_tokens": 147349623.0,
+      "step": 63380
+    },
+    {
+      "epoch": 0.5811863940588613,
+      "learning_rate": 8.376455487301733e-05,
+      "loss": 0.6921,
+      "mean_token_accuracy": 0.7901294887065887,
+      "num_tokens": 147374692.0,
+      "step": 63390
+    },
+    {
+      "epoch": 0.5812780782983406,
+      "learning_rate": 8.374621802512148e-05,
+      "loss": 0.7043,
+      "mean_token_accuracy": 0.7841442286968231,
+      "num_tokens": 147399633.0,
+      "step": 63400
+    },
+    {
+      "epoch": 0.5813697625378198,
+      "learning_rate": 8.372788117722565e-05,
+      "loss": 0.7403,
+      "mean_token_accuracy": 0.7764342486858368,
+      "num_tokens": 147424120.0,
+      "step": 63410
+    },
+    {
+      "epoch": 0.5814614467772989,
+      "learning_rate": 8.370954432932979e-05,
+      "loss": 0.716,
+      "mean_token_accuracy": 0.7835156202316285,
+      "num_tokens": 147448945.0,
+      "step": 63420
+    },
+    {
+      "epoch": 0.5815531310167782,
+      "learning_rate": 8.369120748143394e-05,
+      "loss": 0.7196,
+      "mean_token_accuracy": 0.7828741073608398,
+      "num_tokens": 147473602.0,
+      "step": 63430
+    },
+    {
+      "epoch": 0.5816448152562574,
+      "learning_rate": 8.36728706335381e-05,
+      "loss": 0.7073,
+      "mean_token_accuracy": 0.781675374507904,
+      "num_tokens": 147499587.0,
+      "step": 63440
+    },
+    {
+      "epoch": 0.5817364994957367,
+      "learning_rate": 8.365453378564226e-05,
+      "loss": 0.6948,
+      "mean_token_accuracy": 0.7856381833553314,
+      "num_tokens": 147525153.0,
+      "step": 63450
+    },
+    {
+      "epoch": 0.5818281837352159,
+      "learning_rate": 8.363619693774641e-05,
+      "loss": 0.709,
+      "mean_token_accuracy": 0.7836725771427154,
+      "num_tokens": 147550029.0,
+      "step": 63460
+    },
+    {
+      "epoch": 0.5819198679746952,
+      "learning_rate": 8.361786008985056e-05,
+      "loss": 0.6773,
+      "mean_token_accuracy": 0.7861979365348816,
+      "num_tokens": 147575441.0,
+      "step": 63470
+    },
+    {
+      "epoch": 0.5820115522141743,
+      "learning_rate": 8.359952324195471e-05,
+      "loss": 0.7022,
+      "mean_token_accuracy": 0.7886620819568634,
+      "num_tokens": 147600306.0,
+      "step": 63480
+    },
+    {
+      "epoch": 0.5821032364536536,
+      "learning_rate": 8.358118639405886e-05,
+      "loss": 0.7168,
+      "mean_token_accuracy": 0.7807242631912231,
+      "num_tokens": 147626353.0,
+      "step": 63490
+    },
+    {
+      "epoch": 0.5821949206931328,
+      "learning_rate": 8.356284954616302e-05,
+      "loss": 0.7286,
+      "mean_token_accuracy": 0.7790759265422821,
+      "num_tokens": 147651737.0,
+      "step": 63500
+    },
+    {
+      "epoch": 0.5822866049326121,
+      "learning_rate": 8.354451269826718e-05,
+      "loss": 0.6747,
+      "mean_token_accuracy": 0.7957220733165741,
+      "num_tokens": 147677184.0,
+      "step": 63510
+    },
+    {
+      "epoch": 0.5823782891720913,
+      "learning_rate": 8.352617585037132e-05,
+      "loss": 0.7245,
+      "mean_token_accuracy": 0.7769837498664856,
+      "num_tokens": 147702055.0,
+      "step": 63520
+    },
+    {
+      "epoch": 0.5824699734115706,
+      "learning_rate": 8.350783900247547e-05,
+      "loss": 0.74,
+      "mean_token_accuracy": 0.7733043849468231,
+      "num_tokens": 147727300.0,
+      "step": 63530
+    },
+    {
+      "epoch": 0.5825616576510498,
+      "learning_rate": 8.348950215457964e-05,
+      "loss": 0.7145,
+      "mean_token_accuracy": 0.7838710427284241,
+      "num_tokens": 147752010.0,
+      "step": 63540
+    },
+    {
+      "epoch": 0.582653341890529,
+      "learning_rate": 8.347116530668379e-05,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.7834075629711151,
+      "num_tokens": 147777880.0,
+      "step": 63550
+    },
+    {
+      "epoch": 0.5827450261300082,
+      "learning_rate": 8.345282845878794e-05,
+      "loss": 0.7124,
+      "mean_token_accuracy": 0.7875271916389466,
+      "num_tokens": 147802601.0,
+      "step": 63560
+    },
+    {
+      "epoch": 0.5828367103694875,
+      "learning_rate": 8.343449161089209e-05,
+      "loss": 0.6932,
+      "mean_token_accuracy": 0.7834922552108765,
+      "num_tokens": 147827355.0,
+      "step": 63570
+    },
+    {
+      "epoch": 0.5829283946089667,
+      "learning_rate": 8.341615476299624e-05,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.7847265064716339,
+      "num_tokens": 147851731.0,
+      "step": 63580
+    },
+    {
+      "epoch": 0.583020078848446,
+      "learning_rate": 8.33978179151004e-05,
+      "loss": 0.7342,
+      "mean_token_accuracy": 0.7769253015518188,
+      "num_tokens": 147876658.0,
+      "step": 63590
+    },
+    {
+      "epoch": 0.5831117630879252,
+      "learning_rate": 8.337948106720456e-05,
+      "loss": 0.7238,
+      "mean_token_accuracy": 0.7845895290374756,
+      "num_tokens": 147901547.0,
+      "step": 63600
+    },
+    {
+      "epoch": 0.5832034473274044,
+      "learning_rate": 8.336114421930871e-05,
+      "loss": 0.6858,
+      "mean_token_accuracy": 0.7881172716617584,
+      "num_tokens": 147927705.0,
+      "step": 63610
+    },
+    {
+      "epoch": 0.5832951315668836,
+      "learning_rate": 8.334280737141285e-05,
+      "loss": 0.6666,
+      "mean_token_accuracy": 0.7966143131256104,
+      "num_tokens": 147952671.0,
+      "step": 63620
+    },
+    {
+      "epoch": 0.5833868158063629,
+      "learning_rate": 8.3324470523517e-05,
+      "loss": 0.7122,
+      "mean_token_accuracy": 0.7883368730545044,
+      "num_tokens": 147977885.0,
+      "step": 63630
+    },
+    {
+      "epoch": 0.5834785000458421,
+      "learning_rate": 8.330613367562117e-05,
+      "loss": 0.7077,
+      "mean_token_accuracy": 0.7859804272651673,
+      "num_tokens": 148002936.0,
+      "step": 63640
+    },
+    {
+      "epoch": 0.5835701842853214,
+      "learning_rate": 8.328779682772532e-05,
+      "loss": 0.7072,
+      "mean_token_accuracy": 0.7861575543880462,
+      "num_tokens": 148028204.0,
+      "step": 63650
+    },
+    {
+      "epoch": 0.5836618685248006,
+      "learning_rate": 8.326945997982947e-05,
+      "loss": 0.7054,
+      "mean_token_accuracy": 0.7846588015556335,
+      "num_tokens": 148053259.0,
+      "step": 63660
+    },
+    {
+      "epoch": 0.5837535527642799,
+      "learning_rate": 8.325112313193362e-05,
+      "loss": 0.7082,
+      "mean_token_accuracy": 0.7851105570793152,
+      "num_tokens": 148077422.0,
+      "step": 63670
+    },
+    {
+      "epoch": 0.583845237003759,
+      "learning_rate": 8.323278628403778e-05,
+      "loss": 0.6979,
+      "mean_token_accuracy": 0.7855910003185272,
+      "num_tokens": 148102286.0,
+      "step": 63680
+    },
+    {
+      "epoch": 0.5839369212432383,
+      "learning_rate": 8.321444943614193e-05,
+      "loss": 0.6712,
+      "mean_token_accuracy": 0.7927249848842621,
+      "num_tokens": 148127324.0,
+      "step": 63690
+    },
+    {
+      "epoch": 0.5840286054827175,
+      "learning_rate": 8.31961125882461e-05,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.777326476573944,
+      "num_tokens": 148153074.0,
+      "step": 63700
+    },
+    {
+      "epoch": 0.5841202897221968,
+      "learning_rate": 8.317777574035025e-05,
+      "loss": 0.7108,
+      "mean_token_accuracy": 0.7857874810695649,
+      "num_tokens": 148178756.0,
+      "step": 63710
+    },
+    {
+      "epoch": 0.584211973961676,
+      "learning_rate": 8.315943889245438e-05,
+      "loss": 0.6583,
+      "mean_token_accuracy": 0.7944095849990844,
+      "num_tokens": 148204506.0,
+      "step": 63720
+    },
+    {
+      "epoch": 0.5843036582011553,
+      "learning_rate": 8.314110204455855e-05,
+      "loss": 0.7002,
+      "mean_token_accuracy": 0.7815732777118682,
+      "num_tokens": 148229454.0,
+      "step": 63730
+    },
+    {
+      "epoch": 0.5843953424406344,
+      "learning_rate": 8.31227651966627e-05,
+      "loss": 0.7299,
+      "mean_token_accuracy": 0.7798981308937073,
+      "num_tokens": 148254914.0,
+      "step": 63740
+    },
+    {
+      "epoch": 0.5844870266801137,
+      "learning_rate": 8.310442834876685e-05,
+      "loss": 0.6703,
+      "mean_token_accuracy": 0.7941282093524933,
+      "num_tokens": 148279383.0,
+      "step": 63750
+    },
+    {
+      "epoch": 0.5845787109195929,
+      "learning_rate": 8.3086091500871e-05,
+      "loss": 0.7003,
+      "mean_token_accuracy": 0.7854337990283966,
+      "num_tokens": 148304264.0,
+      "step": 63760
+    },
+    {
+      "epoch": 0.5846703951590722,
+      "learning_rate": 8.306775465297516e-05,
+      "loss": 0.734,
+      "mean_token_accuracy": 0.7756215274333954,
+      "num_tokens": 148329299.0,
+      "step": 63770
+    },
+    {
+      "epoch": 0.5847620793985514,
+      "learning_rate": 8.304941780507931e-05,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.7873515009880065,
+      "num_tokens": 148354155.0,
+      "step": 63780
+    },
+    {
+      "epoch": 0.5848537636380307,
+      "learning_rate": 8.303108095718346e-05,
+      "loss": 0.7098,
+      "mean_token_accuracy": 0.78629270195961,
+      "num_tokens": 148378644.0,
+      "step": 63790
+    },
+    {
+      "epoch": 0.5849454478775099,
+      "learning_rate": 8.301274410928763e-05,
+      "loss": 0.6794,
+      "mean_token_accuracy": 0.7919967710971832,
+      "num_tokens": 148403780.0,
+      "step": 63800
+    },
+    {
+      "epoch": 0.585037132116989,
+      "learning_rate": 8.299440726139178e-05,
+      "loss": 0.7437,
+      "mean_token_accuracy": 0.7779891729354859,
+      "num_tokens": 148428929.0,
+      "step": 63810
+    },
+    {
+      "epoch": 0.5851288163564683,
+      "learning_rate": 8.297607041349592e-05,
+      "loss": 0.7112,
+      "mean_token_accuracy": 0.784506893157959,
+      "num_tokens": 148454712.0,
+      "step": 63820
+    },
+    {
+      "epoch": 0.5852205005959475,
+      "learning_rate": 8.295773356560008e-05,
+      "loss": 0.6677,
+      "mean_token_accuracy": 0.7959107160568237,
+      "num_tokens": 148480565.0,
+      "step": 63830
+    },
+    {
+      "epoch": 0.5853121848354268,
+      "learning_rate": 8.293939671770423e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.7842214405536652,
+      "num_tokens": 148505519.0,
+      "step": 63840
+    },
+    {
+      "epoch": 0.585403869074906,
+      "learning_rate": 8.292105986980839e-05,
+      "loss": 0.6732,
+      "mean_token_accuracy": 0.7953565776348114,
+      "num_tokens": 148530671.0,
+      "step": 63850
+    },
+    {
+      "epoch": 0.5854955533143853,
+      "learning_rate": 8.290272302191254e-05,
+      "loss": 0.7078,
+      "mean_token_accuracy": 0.7895129323005676,
+      "num_tokens": 148555245.0,
+      "step": 63860
+    },
+    {
+      "epoch": 0.5855872375538644,
+      "learning_rate": 8.288438617401669e-05,
+      "loss": 0.6757,
+      "mean_token_accuracy": 0.789818549156189,
+      "num_tokens": 148580095.0,
+      "step": 63870
+    },
+    {
+      "epoch": 0.5856789217933437,
+      "learning_rate": 8.286604932612084e-05,
+      "loss": 0.7043,
+      "mean_token_accuracy": 0.7819573879241943,
+      "num_tokens": 148605190.0,
+      "step": 63880
+    },
+    {
+      "epoch": 0.5857706060328229,
+      "learning_rate": 8.284771247822499e-05,
+      "loss": 0.7285,
+      "mean_token_accuracy": 0.7796127200126648,
+      "num_tokens": 148629971.0,
+      "step": 63890
+    },
+    {
+      "epoch": 0.5858622902723022,
+      "learning_rate": 8.282937563032916e-05,
+      "loss": 0.703,
+      "mean_token_accuracy": 0.7765686392784119,
+      "num_tokens": 148654816.0,
+      "step": 63900
+    },
+    {
+      "epoch": 0.5859539745117814,
+      "learning_rate": 8.281103878243331e-05,
+      "loss": 0.706,
+      "mean_token_accuracy": 0.7846543312072753,
+      "num_tokens": 148679096.0,
+      "step": 63910
+    },
+    {
+      "epoch": 0.5860456587512607,
+      "learning_rate": 8.279270193453745e-05,
+      "loss": 0.7324,
+      "mean_token_accuracy": 0.7754747748374939,
+      "num_tokens": 148704015.0,
+      "step": 63920
+    },
+    {
+      "epoch": 0.5861373429907399,
+      "learning_rate": 8.277436508664161e-05,
+      "loss": 0.6661,
+      "mean_token_accuracy": 0.7943182706832885,
+      "num_tokens": 148728561.0,
+      "step": 63930
+    },
+    {
+      "epoch": 0.5862290272302191,
+      "learning_rate": 8.275602823874577e-05,
+      "loss": 0.6813,
+      "mean_token_accuracy": 0.7870662689208985,
+      "num_tokens": 148753393.0,
+      "step": 63940
+    },
+    {
+      "epoch": 0.5863207114696983,
+      "learning_rate": 8.273769139084992e-05,
+      "loss": 0.6776,
+      "mean_token_accuracy": 0.7936405122280121,
+      "num_tokens": 148778269.0,
+      "step": 63950
+    },
+    {
+      "epoch": 0.5864123957091776,
+      "learning_rate": 8.271935454295407e-05,
+      "loss": 0.687,
+      "mean_token_accuracy": 0.7924061715602875,
+      "num_tokens": 148803149.0,
+      "step": 63960
+    },
+    {
+      "epoch": 0.5865040799486568,
+      "learning_rate": 8.270101769505822e-05,
+      "loss": 0.7006,
+      "mean_token_accuracy": 0.7836643159389496,
+      "num_tokens": 148828126.0,
+      "step": 63970
+    },
+    {
+      "epoch": 0.5865957641881361,
+      "learning_rate": 8.268268084716237e-05,
+      "loss": 0.6778,
+      "mean_token_accuracy": 0.7899862587451935,
+      "num_tokens": 148852811.0,
+      "step": 63980
+    },
+    {
+      "epoch": 0.5866874484276153,
+      "learning_rate": 8.266434399926652e-05,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.7958472669124603,
+      "num_tokens": 148877989.0,
+      "step": 63990
+    },
+    {
+      "epoch": 0.5867791326670945,
+      "learning_rate": 8.264600715137069e-05,
+      "loss": 0.7003,
+      "mean_token_accuracy": 0.7905132412910462,
+      "num_tokens": 148903389.0,
+      "step": 64000
+    },
+    {
+      "epoch": 0.5868708169065737,
+      "learning_rate": 8.262767030347484e-05,
+      "loss": 0.7212,
+      "mean_token_accuracy": 0.7804872572422028,
+      "num_tokens": 148928501.0,
+      "step": 64010
+    },
+    {
+      "epoch": 0.586962501146053,
+      "learning_rate": 8.260933345557898e-05,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.7765512049198151,
+      "num_tokens": 148953804.0,
+      "step": 64020
+    },
+    {
+      "epoch": 0.5870541853855322,
+      "learning_rate": 8.259099660768315e-05,
+      "loss": 0.7128,
+      "mean_token_accuracy": 0.7861216723918915,
+      "num_tokens": 148979129.0,
+      "step": 64030
+    },
+    {
+      "epoch": 0.5871458696250115,
+      "learning_rate": 8.25726597597873e-05,
+      "loss": 0.7306,
+      "mean_token_accuracy": 0.775581157207489,
+      "num_tokens": 149003538.0,
+      "step": 64040
+    },
+    {
+      "epoch": 0.5872375538644907,
+      "learning_rate": 8.255432291189145e-05,
+      "loss": 0.6787,
+      "mean_token_accuracy": 0.7887977957725525,
+      "num_tokens": 149028401.0,
+      "step": 64050
+    },
+    {
+      "epoch": 0.58732923810397,
+      "learning_rate": 8.25359860639956e-05,
+      "loss": 0.6916,
+      "mean_token_accuracy": 0.7851101934909821,
+      "num_tokens": 149053731.0,
+      "step": 64060
+    },
+    {
+      "epoch": 0.5874209223434491,
+      "learning_rate": 8.251764921609975e-05,
+      "loss": 0.6748,
+      "mean_token_accuracy": 0.7852842628955841,
+      "num_tokens": 149079284.0,
+      "step": 64070
+    },
+    {
+      "epoch": 0.5875126065829284,
+      "learning_rate": 8.24993123682039e-05,
+      "loss": 0.676,
+      "mean_token_accuracy": 0.7941822052001953,
+      "num_tokens": 149104286.0,
+      "step": 64080
+    },
+    {
+      "epoch": 0.5876042908224076,
+      "learning_rate": 8.248097552030807e-05,
+      "loss": 0.7135,
+      "mean_token_accuracy": 0.7870282888412475,
+      "num_tokens": 149129240.0,
+      "step": 64090
+    },
+    {
+      "epoch": 0.5876959750618869,
+      "learning_rate": 8.246263867241222e-05,
+      "loss": 0.7072,
+      "mean_token_accuracy": 0.7809190571308136,
+      "num_tokens": 149154121.0,
+      "step": 64100
+    },
+    {
+      "epoch": 0.5877876593013661,
+      "learning_rate": 8.244430182451637e-05,
+      "loss": 0.703,
+      "mean_token_accuracy": 0.7839613497257233,
+      "num_tokens": 149179463.0,
+      "step": 64110
+    },
+    {
+      "epoch": 0.5878793435408454,
+      "learning_rate": 8.242596497662051e-05,
+      "loss": 0.7044,
+      "mean_token_accuracy": 0.7832949101924896,
+      "num_tokens": 149204670.0,
+      "step": 64120
+    },
+    {
+      "epoch": 0.5879710277803246,
+      "learning_rate": 8.240762812872468e-05,
+      "loss": 0.6786,
+      "mean_token_accuracy": 0.7926870584487915,
+      "num_tokens": 149230302.0,
+      "step": 64130
+    },
+    {
+      "epoch": 0.5880627120198038,
+      "learning_rate": 8.238929128082883e-05,
+      "loss": 0.7348,
+      "mean_token_accuracy": 0.7770215153694153,
+      "num_tokens": 149255432.0,
+      "step": 64140
+    },
+    {
+      "epoch": 0.588154396259283,
+      "learning_rate": 8.237095443293298e-05,
+      "loss": 0.7225,
+      "mean_token_accuracy": 0.7816013813018798,
+      "num_tokens": 149280035.0,
+      "step": 64150
+    },
+    {
+      "epoch": 0.5882460804987623,
+      "learning_rate": 8.235261758503713e-05,
+      "loss": 0.6891,
+      "mean_token_accuracy": 0.784260493516922,
+      "num_tokens": 149305482.0,
+      "step": 64160
+    },
+    {
+      "epoch": 0.5883377647382415,
+      "learning_rate": 8.233428073714129e-05,
+      "loss": 0.6798,
+      "mean_token_accuracy": 0.7883141458034515,
+      "num_tokens": 149330366.0,
+      "step": 64170
+    },
+    {
+      "epoch": 0.5884294489777208,
+      "learning_rate": 8.231594388924544e-05,
+      "loss": 0.752,
+      "mean_token_accuracy": 0.7755284488201142,
+      "num_tokens": 149356050.0,
+      "step": 64180
+    },
+    {
+      "epoch": 0.5885211332172,
+      "learning_rate": 8.22976070413496e-05,
+      "loss": 0.6753,
+      "mean_token_accuracy": 0.7889199912548065,
+      "num_tokens": 149380828.0,
+      "step": 64190
+    },
+    {
+      "epoch": 0.5886128174566791,
+      "learning_rate": 8.227927019345375e-05,
+      "loss": 0.6886,
+      "mean_token_accuracy": 0.7897959649562836,
+      "num_tokens": 149406675.0,
+      "step": 64200
+    },
+    {
+      "epoch": 0.5887045016961584,
+      "learning_rate": 8.22609333455579e-05,
+      "loss": 0.7024,
+      "mean_token_accuracy": 0.7864479064941406,
+      "num_tokens": 149431617.0,
+      "step": 64210
+    },
+    {
+      "epoch": 0.5887961859356377,
+      "learning_rate": 8.224259649766206e-05,
+      "loss": 0.6716,
+      "mean_token_accuracy": 0.7913666784763336,
+      "num_tokens": 149457665.0,
+      "step": 64220
+    },
+    {
+      "epoch": 0.5888878701751169,
+      "learning_rate": 8.222425964976621e-05,
+      "loss": 0.7335,
+      "mean_token_accuracy": 0.7786946952342987,
+      "num_tokens": 149482827.0,
+      "step": 64230
+    },
+    {
+      "epoch": 0.5889795544145962,
+      "learning_rate": 8.220592280187036e-05,
+      "loss": 0.7222,
+      "mean_token_accuracy": 0.7779923200607299,
+      "num_tokens": 149507331.0,
+      "step": 64240
+    },
+    {
+      "epoch": 0.5890712386540754,
+      "learning_rate": 8.218758595397451e-05,
+      "loss": 0.7003,
+      "mean_token_accuracy": 0.7880372524261474,
+      "num_tokens": 149532621.0,
+      "step": 64250
+    },
+    {
+      "epoch": 0.5891629228935547,
+      "learning_rate": 8.216924910607867e-05,
+      "loss": 0.7143,
+      "mean_token_accuracy": 0.7853023648262024,
+      "num_tokens": 149557682.0,
+      "step": 64260
+    },
+    {
+      "epoch": 0.5892546071330338,
+      "learning_rate": 8.215091225818282e-05,
+      "loss": 0.7574,
+      "mean_token_accuracy": 0.7721137225627899,
+      "num_tokens": 149583441.0,
+      "step": 64270
+    },
+    {
+      "epoch": 0.589346291372513,
+      "learning_rate": 8.213257541028697e-05,
+      "loss": 0.7037,
+      "mean_token_accuracy": 0.7882411360740662,
+      "num_tokens": 149608127.0,
+      "step": 64280
+    },
+    {
+      "epoch": 0.5894379756119923,
+      "learning_rate": 8.211423856239113e-05,
+      "loss": 0.7061,
+      "mean_token_accuracy": 0.7803534269332886,
+      "num_tokens": 149632863.0,
+      "step": 64290
+    },
+    {
+      "epoch": 0.5895296598514715,
+      "learning_rate": 8.209590171449529e-05,
+      "loss": 0.7187,
+      "mean_token_accuracy": 0.7781393647193908,
+      "num_tokens": 149658488.0,
+      "step": 64300
+    },
+    {
+      "epoch": 0.5896213440909508,
+      "learning_rate": 8.207756486659944e-05,
+      "loss": 0.6871,
+      "mean_token_accuracy": 0.7872996270656586,
+      "num_tokens": 149683311.0,
+      "step": 64310
+    },
+    {
+      "epoch": 0.58971302833043,
+      "learning_rate": 8.205922801870359e-05,
+      "loss": 0.6876,
+      "mean_token_accuracy": 0.7906221568584442,
+      "num_tokens": 149709057.0,
+      "step": 64320
+    },
+    {
+      "epoch": 0.5898047125699092,
+      "learning_rate": 8.204089117080774e-05,
+      "loss": 0.7203,
+      "mean_token_accuracy": 0.7827765464782714,
+      "num_tokens": 149733195.0,
+      "step": 64330
+    },
+    {
+      "epoch": 0.5898963968093884,
+      "learning_rate": 8.20225543229119e-05,
+      "loss": 0.694,
+      "mean_token_accuracy": 0.7838902354240418,
+      "num_tokens": 149757775.0,
+      "step": 64340
+    },
+    {
+      "epoch": 0.5899880810488677,
+      "learning_rate": 8.200421747501606e-05,
+      "loss": 0.7291,
+      "mean_token_accuracy": 0.7774064660072326,
+      "num_tokens": 149782232.0,
+      "step": 64350
+    },
+    {
+      "epoch": 0.5900797652883469,
+      "learning_rate": 8.19858806271202e-05,
+      "loss": 0.6934,
+      "mean_token_accuracy": 0.7859097242355346,
+      "num_tokens": 149807883.0,
+      "step": 64360
+    },
+    {
+      "epoch": 0.5901714495278262,
+      "learning_rate": 8.196754377922435e-05,
+      "loss": 0.7679,
+      "mean_token_accuracy": 0.7702005982398987,
+      "num_tokens": 149833750.0,
+      "step": 64370
+    },
+    {
+      "epoch": 0.5902631337673054,
+      "learning_rate": 8.19492069313285e-05,
+      "loss": 0.6779,
+      "mean_token_accuracy": 0.7937423825263977,
+      "num_tokens": 149858563.0,
+      "step": 64380
+    },
+    {
+      "epoch": 0.5903548180067847,
+      "learning_rate": 8.193087008343267e-05,
+      "loss": 0.7196,
+      "mean_token_accuracy": 0.7906647562980652,
+      "num_tokens": 149883174.0,
+      "step": 64390
+    },
+    {
+      "epoch": 0.5904465022462638,
+      "learning_rate": 8.191253323553682e-05,
+      "loss": 0.7145,
+      "mean_token_accuracy": 0.7835176408290863,
+      "num_tokens": 149907710.0,
+      "step": 64400
+    },
+    {
+      "epoch": 0.5905381864857431,
+      "learning_rate": 8.189419638764097e-05,
+      "loss": 0.6796,
+      "mean_token_accuracy": 0.7890723168849945,
+      "num_tokens": 149932311.0,
+      "step": 64410
+    },
+    {
+      "epoch": 0.5906298707252223,
+      "learning_rate": 8.187585953974512e-05,
+      "loss": 0.676,
+      "mean_token_accuracy": 0.7912966549396515,
+      "num_tokens": 149956459.0,
+      "step": 64420
+    },
+    {
+      "epoch": 0.5907215549647016,
+      "learning_rate": 8.185752269184927e-05,
+      "loss": 0.6672,
+      "mean_token_accuracy": 0.7880989968776703,
+      "num_tokens": 149981931.0,
+      "step": 64430
+    },
+    {
+      "epoch": 0.5908132392041808,
+      "learning_rate": 8.183918584395343e-05,
+      "loss": 0.704,
+      "mean_token_accuracy": 0.7908606708049775,
+      "num_tokens": 150007268.0,
+      "step": 64440
+    },
+    {
+      "epoch": 0.5909049234436601,
+      "learning_rate": 8.182084899605759e-05,
+      "loss": 0.6759,
+      "mean_token_accuracy": 0.7926512897014618,
+      "num_tokens": 150032489.0,
+      "step": 64450
+    },
+    {
+      "epoch": 0.5909966076831392,
+      "learning_rate": 8.180251214816173e-05,
+      "loss": 0.7026,
+      "mean_token_accuracy": 0.7807465970516205,
+      "num_tokens": 150057536.0,
+      "step": 64460
+    },
+    {
+      "epoch": 0.5910882919226185,
+      "learning_rate": 8.178417530026588e-05,
+      "loss": 0.7167,
+      "mean_token_accuracy": 0.7739458680152893,
+      "num_tokens": 150083170.0,
+      "step": 64470
+    },
+    {
+      "epoch": 0.5911799761620977,
+      "learning_rate": 8.176583845237003e-05,
+      "loss": 0.6609,
+      "mean_token_accuracy": 0.7943176209926606,
+      "num_tokens": 150108355.0,
+      "step": 64480
+    },
+    {
+      "epoch": 0.591271660401577,
+      "learning_rate": 8.17475016044742e-05,
+      "loss": 0.7059,
+      "mean_token_accuracy": 0.7872118592262268,
+      "num_tokens": 150133293.0,
+      "step": 64490
+    },
+    {
+      "epoch": 0.5913633446410562,
+      "learning_rate": 8.172916475657835e-05,
+      "loss": 0.7191,
+      "mean_token_accuracy": 0.7783481657505036,
+      "num_tokens": 150158436.0,
+      "step": 64500
+    },
+    {
+      "epoch": 0.5914550288805355,
+      "learning_rate": 8.17108279086825e-05,
+      "loss": 0.7325,
+      "mean_token_accuracy": 0.7804322838783264,
+      "num_tokens": 150183183.0,
+      "step": 64510
+    },
+    {
+      "epoch": 0.5915467131200147,
+      "learning_rate": 8.169249106078666e-05,
+      "loss": 0.6876,
+      "mean_token_accuracy": 0.7907378137111664,
+      "num_tokens": 150208272.0,
+      "step": 64520
+    },
+    {
+      "epoch": 0.5916383973594939,
+      "learning_rate": 8.167415421289081e-05,
+      "loss": 0.7189,
+      "mean_token_accuracy": 0.7767588555812835,
+      "num_tokens": 150232546.0,
+      "step": 64530
+    },
+    {
+      "epoch": 0.5917300815989731,
+      "learning_rate": 8.165581736499496e-05,
+      "loss": 0.7075,
+      "mean_token_accuracy": 0.7842768251895904,
+      "num_tokens": 150257428.0,
+      "step": 64540
+    },
+    {
+      "epoch": 0.5918217658384524,
+      "learning_rate": 8.163748051709912e-05,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.7891097903251648,
+      "num_tokens": 150282125.0,
+      "step": 64550
+    },
+    {
+      "epoch": 0.5919134500779316,
+      "learning_rate": 8.161914366920326e-05,
+      "loss": 0.7235,
+      "mean_token_accuracy": 0.7783652007579803,
+      "num_tokens": 150307180.0,
+      "step": 64560
+    },
+    {
+      "epoch": 0.5920051343174109,
+      "learning_rate": 8.160080682130741e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.7973542153835297,
+      "num_tokens": 150332280.0,
+      "step": 64570
+    },
+    {
+      "epoch": 0.5920968185568901,
+      "learning_rate": 8.158246997341158e-05,
+      "loss": 0.7015,
+      "mean_token_accuracy": 0.7862518548965454,
+      "num_tokens": 150357272.0,
+      "step": 64580
+    },
+    {
+      "epoch": 0.5921885027963693,
+      "learning_rate": 8.156413312551573e-05,
+      "loss": 0.7327,
+      "mean_token_accuracy": 0.7767649173736573,
+      "num_tokens": 150382115.0,
+      "step": 64590
+    },
+    {
+      "epoch": 0.5922801870358485,
+      "learning_rate": 8.154579627761988e-05,
+      "loss": 0.7009,
+      "mean_token_accuracy": 0.788267582654953,
+      "num_tokens": 150405832.0,
+      "step": 64600
+    },
+    {
+      "epoch": 0.5923718712753278,
+      "learning_rate": 8.152745942972404e-05,
+      "loss": 0.6938,
+      "mean_token_accuracy": 0.7892520785331726,
+      "num_tokens": 150430587.0,
+      "step": 64610
+    },
+    {
+      "epoch": 0.592463555514807,
+      "learning_rate": 8.150912258182819e-05,
+      "loss": 0.7184,
+      "mean_token_accuracy": 0.7759264647960663,
+      "num_tokens": 150455270.0,
+      "step": 64620
+    },
+    {
+      "epoch": 0.5925552397542863,
+      "learning_rate": 8.149078573393234e-05,
+      "loss": 0.6988,
+      "mean_token_accuracy": 0.7858653008937836,
+      "num_tokens": 150480147.0,
+      "step": 64630
+    },
+    {
+      "epoch": 0.5926469239937655,
+      "learning_rate": 8.147244888603649e-05,
+      "loss": 0.721,
+      "mean_token_accuracy": 0.7873132884502411,
+      "num_tokens": 150504742.0,
+      "step": 64640
+    },
+    {
+      "epoch": 0.5927386082332448,
+      "learning_rate": 8.145411203814066e-05,
+      "loss": 0.7336,
+      "mean_token_accuracy": 0.7803522884845734,
+      "num_tokens": 150529394.0,
+      "step": 64650
+    },
+    {
+      "epoch": 0.5928302924727239,
+      "learning_rate": 8.14357751902448e-05,
+      "loss": 0.699,
+      "mean_token_accuracy": 0.7888843536376953,
+      "num_tokens": 150554556.0,
+      "step": 64660
+    },
+    {
+      "epoch": 0.5929219767122031,
+      "learning_rate": 8.141743834234895e-05,
+      "loss": 0.6712,
+      "mean_token_accuracy": 0.7859473407268525,
+      "num_tokens": 150579300.0,
+      "step": 64670
+    },
+    {
+      "epoch": 0.5930136609516824,
+      "learning_rate": 8.139910149445311e-05,
+      "loss": 0.6879,
+      "mean_token_accuracy": 0.7903903961181641,
+      "num_tokens": 150605010.0,
+      "step": 64680
+    },
+    {
+      "epoch": 0.5931053451911616,
+      "learning_rate": 8.138076464655726e-05,
+      "loss": 0.6907,
+      "mean_token_accuracy": 0.7918246030807495,
+      "num_tokens": 150630010.0,
+      "step": 64690
+    },
+    {
+      "epoch": 0.5931970294306409,
+      "learning_rate": 8.136242779866142e-05,
+      "loss": 0.6718,
+      "mean_token_accuracy": 0.7947119235992431,
+      "num_tokens": 150654790.0,
+      "step": 64700
+    },
+    {
+      "epoch": 0.5932887136701201,
+      "learning_rate": 8.134409095076557e-05,
+      "loss": 0.7054,
+      "mean_token_accuracy": 0.7794699013233185,
+      "num_tokens": 150680688.0,
+      "step": 64710
+    },
+    {
+      "epoch": 0.5933803979095993,
+      "learning_rate": 8.132575410286972e-05,
+      "loss": 0.6995,
+      "mean_token_accuracy": 0.7933349549770355,
+      "num_tokens": 150705217.0,
+      "step": 64720
+    },
+    {
+      "epoch": 0.5934720821490785,
+      "learning_rate": 8.130741725497387e-05,
+      "loss": 0.7355,
+      "mean_token_accuracy": 0.7805914580821991,
+      "num_tokens": 150731051.0,
+      "step": 64730
+    },
+    {
+      "epoch": 0.5935637663885578,
+      "learning_rate": 8.128908040707802e-05,
+      "loss": 0.6802,
+      "mean_token_accuracy": 0.7900279283523559,
+      "num_tokens": 150755700.0,
+      "step": 64740
+    },
+    {
+      "epoch": 0.593655450628037,
+      "learning_rate": 8.127074355918219e-05,
+      "loss": 0.6721,
+      "mean_token_accuracy": 0.7904886305332184,
+      "num_tokens": 150780723.0,
+      "step": 64750
+    },
+    {
+      "epoch": 0.5937471348675163,
+      "learning_rate": 8.125240671128633e-05,
+      "loss": 0.7406,
+      "mean_token_accuracy": 0.7759748756885528,
+      "num_tokens": 150806416.0,
+      "step": 64760
+    },
+    {
+      "epoch": 0.5938388191069955,
+      "learning_rate": 8.123406986339048e-05,
+      "loss": 0.7001,
+      "mean_token_accuracy": 0.7837151050567627,
+      "num_tokens": 150832653.0,
+      "step": 64770
+    },
+    {
+      "epoch": 0.5939305033464748,
+      "learning_rate": 8.121573301549464e-05,
+      "loss": 0.7388,
+      "mean_token_accuracy": 0.7705853223800659,
+      "num_tokens": 150857682.0,
+      "step": 64780
+    },
+    {
+      "epoch": 0.5940221875859539,
+      "learning_rate": 8.11973961675988e-05,
+      "loss": 0.6735,
+      "mean_token_accuracy": 0.7909241139888763,
+      "num_tokens": 150883588.0,
+      "step": 64790
+    },
+    {
+      "epoch": 0.5941138718254332,
+      "learning_rate": 8.117905931970295e-05,
+      "loss": 0.712,
+      "mean_token_accuracy": 0.7861241042613983,
+      "num_tokens": 150908475.0,
+      "step": 64800
+    },
+    {
+      "epoch": 0.5942055560649124,
+      "learning_rate": 8.11607224718071e-05,
+      "loss": 0.6678,
+      "mean_token_accuracy": 0.7927547752857208,
+      "num_tokens": 150933485.0,
+      "step": 64810
+    },
+    {
+      "epoch": 0.5942972403043917,
+      "learning_rate": 8.114238562391125e-05,
+      "loss": 0.7102,
+      "mean_token_accuracy": 0.7864205956459045,
+      "num_tokens": 150958465.0,
+      "step": 64820
+    },
+    {
+      "epoch": 0.5943889245438709,
+      "learning_rate": 8.11240487760154e-05,
+      "loss": 0.677,
+      "mean_token_accuracy": 0.7920351564884186,
+      "num_tokens": 150983351.0,
+      "step": 64830
+    },
+    {
+      "epoch": 0.5944806087833502,
+      "learning_rate": 8.110571192811957e-05,
+      "loss": 0.6978,
+      "mean_token_accuracy": 0.7859266996383667,
+      "num_tokens": 151008309.0,
+      "step": 64840
+    },
+    {
+      "epoch": 0.5945722930228293,
+      "learning_rate": 8.108737508022372e-05,
+      "loss": 0.6822,
+      "mean_token_accuracy": 0.7874178767204285,
+      "num_tokens": 151032966.0,
+      "step": 64850
+    },
+    {
+      "epoch": 0.5946639772623086,
+      "learning_rate": 8.106903823232786e-05,
+      "loss": 0.6714,
+      "mean_token_accuracy": 0.7888278841972352,
+      "num_tokens": 151057177.0,
+      "step": 64860
+    },
+    {
+      "epoch": 0.5947556615017878,
+      "learning_rate": 8.105070138443201e-05,
+      "loss": 0.712,
+      "mean_token_accuracy": 0.7801290571689605,
+      "num_tokens": 151082202.0,
+      "step": 64870
+    },
+    {
+      "epoch": 0.5948473457412671,
+      "learning_rate": 8.103236453653618e-05,
+      "loss": 0.7165,
+      "mean_token_accuracy": 0.7849847555160523,
+      "num_tokens": 151107226.0,
+      "step": 64880
+    },
+    {
+      "epoch": 0.5949390299807463,
+      "learning_rate": 8.101402768864033e-05,
+      "loss": 0.7182,
+      "mean_token_accuracy": 0.7808296024799347,
+      "num_tokens": 151132624.0,
+      "step": 64890
+    },
+    {
+      "epoch": 0.5950307142202256,
+      "learning_rate": 8.099569084074448e-05,
+      "loss": 0.6896,
+      "mean_token_accuracy": 0.7926852285861969,
+      "num_tokens": 151158280.0,
+      "step": 64900
+    },
+    {
+      "epoch": 0.5951223984597048,
+      "learning_rate": 8.097735399284863e-05,
+      "loss": 0.6973,
+      "mean_token_accuracy": 0.7888705492019653,
+      "num_tokens": 151183554.0,
+      "step": 64910
+    },
+    {
+      "epoch": 0.595214082699184,
+      "learning_rate": 8.095901714495278e-05,
+      "loss": 0.6543,
+      "mean_token_accuracy": 0.7970582187175751,
+      "num_tokens": 151208225.0,
+      "step": 64920
+    },
+    {
+      "epoch": 0.5953057669386632,
+      "learning_rate": 8.094068029705694e-05,
+      "loss": 0.6996,
+      "mean_token_accuracy": 0.7853852927684783,
+      "num_tokens": 151233791.0,
+      "step": 64930
+    },
+    {
+      "epoch": 0.5953974511781425,
+      "learning_rate": 8.09223434491611e-05,
+      "loss": 0.6992,
+      "mean_token_accuracy": 0.7901147663593292,
+      "num_tokens": 151258661.0,
+      "step": 64940
+    },
+    {
+      "epoch": 0.5954891354176217,
+      "learning_rate": 8.090400660126525e-05,
+      "loss": 0.6817,
+      "mean_token_accuracy": 0.7880674540996552,
+      "num_tokens": 151283145.0,
+      "step": 64950
+    },
+    {
+      "epoch": 0.595580819657101,
+      "learning_rate": 8.088566975336939e-05,
+      "loss": 0.6976,
+      "mean_token_accuracy": 0.7901171863079071,
+      "num_tokens": 151307149.0,
+      "step": 64960
+    },
+    {
+      "epoch": 0.5956725038965802,
+      "learning_rate": 8.086733290547356e-05,
+      "loss": 0.6381,
+      "mean_token_accuracy": 0.8069639921188354,
+      "num_tokens": 151332250.0,
+      "step": 64970
+    },
+    {
+      "epoch": 0.5957641881360594,
+      "learning_rate": 8.084899605757771e-05,
+      "loss": 0.6729,
+      "mean_token_accuracy": 0.7924875020980835,
+      "num_tokens": 151356917.0,
+      "step": 64980
+    },
+    {
+      "epoch": 0.5958558723755386,
+      "learning_rate": 8.083065920968186e-05,
+      "loss": 0.6723,
+      "mean_token_accuracy": 0.7953984498977661,
+      "num_tokens": 151381471.0,
+      "step": 64990
+    },
+    {
+      "epoch": 0.5959475566150179,
+      "learning_rate": 8.081232236178601e-05,
+      "loss": 0.721,
+      "mean_token_accuracy": 0.779432862997055,
+      "num_tokens": 151405820.0,
+      "step": 65000
+    },
+    {
+      "epoch": 0.5960392408544971,
+      "learning_rate": 8.079398551389016e-05,
+      "loss": 0.704,
+      "mean_token_accuracy": 0.7881818771362304,
+      "num_tokens": 151431025.0,
+      "step": 65010
+    },
+    {
+      "epoch": 0.5961309250939764,
+      "learning_rate": 8.077564866599432e-05,
+      "loss": 0.6858,
+      "mean_token_accuracy": 0.7907904803752899,
+      "num_tokens": 151455337.0,
+      "step": 65020
+    },
+    {
+      "epoch": 0.5962226093334556,
+      "learning_rate": 8.075731181809847e-05,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.7922715544700623,
+      "num_tokens": 151479919.0,
+      "step": 65030
+    },
+    {
+      "epoch": 0.5963142935729349,
+      "learning_rate": 8.073897497020263e-05,
+      "loss": 0.689,
+      "mean_token_accuracy": 0.7919419050216675,
+      "num_tokens": 151505252.0,
+      "step": 65040
+    },
+    {
+      "epoch": 0.596405977812414,
+      "learning_rate": 8.072063812230679e-05,
+      "loss": 0.6699,
+      "mean_token_accuracy": 0.7882711529731751,
+      "num_tokens": 151530038.0,
+      "step": 65050
+    },
+    {
+      "epoch": 0.5964976620518933,
+      "learning_rate": 8.070230127441092e-05,
+      "loss": 0.7049,
+      "mean_token_accuracy": 0.7847719013690948,
+      "num_tokens": 151555326.0,
+      "step": 65060
+    },
+    {
+      "epoch": 0.5965893462913725,
+      "learning_rate": 8.068396442651509e-05,
+      "loss": 0.7061,
+      "mean_token_accuracy": 0.7804329514503479,
+      "num_tokens": 151580855.0,
+      "step": 65070
+    },
+    {
+      "epoch": 0.5966810305308518,
+      "learning_rate": 8.066562757861924e-05,
+      "loss": 0.7034,
+      "mean_token_accuracy": 0.7878237545490265,
+      "num_tokens": 151605551.0,
+      "step": 65080
+    },
+    {
+      "epoch": 0.596772714770331,
+      "learning_rate": 8.064729073072339e-05,
+      "loss": 0.706,
+      "mean_token_accuracy": 0.7816583752632141,
+      "num_tokens": 151629801.0,
+      "step": 65090
+    },
+    {
+      "epoch": 0.5968643990098103,
+      "learning_rate": 8.062895388282754e-05,
+      "loss": 0.6935,
+      "mean_token_accuracy": 0.7866775929927826,
+      "num_tokens": 151655179.0,
+      "step": 65100
+    },
+    {
+      "epoch": 0.5969560832492894,
+      "learning_rate": 8.06106170349317e-05,
+      "loss": 0.6813,
+      "mean_token_accuracy": 0.79028679728508,
+      "num_tokens": 151680104.0,
+      "step": 65110
+    },
+    {
+      "epoch": 0.5970477674887686,
+      "learning_rate": 8.059228018703585e-05,
+      "loss": 0.6809,
+      "mean_token_accuracy": 0.7855668783187866,
+      "num_tokens": 151705046.0,
+      "step": 65120
+    },
+    {
+      "epoch": 0.5971394517282479,
+      "learning_rate": 8.057394333914e-05,
+      "loss": 0.7323,
+      "mean_token_accuracy": 0.7846117556095124,
+      "num_tokens": 151729319.0,
+      "step": 65130
+    },
+    {
+      "epoch": 0.5972311359677271,
+      "learning_rate": 8.055560649124417e-05,
+      "loss": 0.7051,
+      "mean_token_accuracy": 0.7806686639785767,
+      "num_tokens": 151754957.0,
+      "step": 65140
+    },
+    {
+      "epoch": 0.5973228202072064,
+      "learning_rate": 8.053726964334832e-05,
+      "loss": 0.6877,
+      "mean_token_accuracy": 0.7896865606307983,
+      "num_tokens": 151779890.0,
+      "step": 65150
+    },
+    {
+      "epoch": 0.5974145044466856,
+      "learning_rate": 8.051893279545246e-05,
+      "loss": 0.6815,
+      "mean_token_accuracy": 0.7892807781696319,
+      "num_tokens": 151804724.0,
+      "step": 65160
+    },
+    {
+      "epoch": 0.5975061886861649,
+      "learning_rate": 8.050059594755662e-05,
+      "loss": 0.728,
+      "mean_token_accuracy": 0.7809356033802033,
+      "num_tokens": 151830000.0,
+      "step": 65170
+    },
+    {
+      "epoch": 0.597597872925644,
+      "learning_rate": 8.048225909966077e-05,
+      "loss": 0.6906,
+      "mean_token_accuracy": 0.7841889500617981,
+      "num_tokens": 151855194.0,
+      "step": 65180
+    },
+    {
+      "epoch": 0.5976895571651233,
+      "learning_rate": 8.046392225176492e-05,
+      "loss": 0.6852,
+      "mean_token_accuracy": 0.7840856671333313,
+      "num_tokens": 151879696.0,
+      "step": 65190
+    },
+    {
+      "epoch": 0.5977812414046025,
+      "learning_rate": 8.044558540386909e-05,
+      "loss": 0.6867,
+      "mean_token_accuracy": 0.7938149988651275,
+      "num_tokens": 151904775.0,
+      "step": 65200
+    },
+    {
+      "epoch": 0.5978729256440818,
+      "learning_rate": 8.042724855597323e-05,
+      "loss": 0.683,
+      "mean_token_accuracy": 0.7929098248481751,
+      "num_tokens": 151930309.0,
+      "step": 65210
+    },
+    {
+      "epoch": 0.597964609883561,
+      "learning_rate": 8.040891170807738e-05,
+      "loss": 0.6739,
+      "mean_token_accuracy": 0.785862410068512,
+      "num_tokens": 151955790.0,
+      "step": 65220
+    },
+    {
+      "epoch": 0.5980562941230403,
+      "learning_rate": 8.039057486018153e-05,
+      "loss": 0.676,
+      "mean_token_accuracy": 0.7941699206829071,
+      "num_tokens": 151981094.0,
+      "step": 65230
+    },
+    {
+      "epoch": 0.5981479783625194,
+      "learning_rate": 8.03722380122857e-05,
+      "loss": 0.6917,
+      "mean_token_accuracy": 0.793646776676178,
+      "num_tokens": 152005920.0,
+      "step": 65240
+    },
+    {
+      "epoch": 0.5982396626019987,
+      "learning_rate": 8.035390116438985e-05,
+      "loss": 0.6736,
+      "mean_token_accuracy": 0.7904787123203277,
+      "num_tokens": 152031097.0,
+      "step": 65250
+    },
+    {
+      "epoch": 0.5983313468414779,
+      "learning_rate": 8.033556431649399e-05,
+      "loss": 0.7083,
+      "mean_token_accuracy": 0.7841505885124207,
+      "num_tokens": 152055763.0,
+      "step": 65260
+    },
+    {
+      "epoch": 0.5984230310809572,
+      "learning_rate": 8.031722746859815e-05,
+      "loss": 0.6915,
+      "mean_token_accuracy": 0.7877865970134735,
+      "num_tokens": 152079670.0,
+      "step": 65270
+    },
+    {
+      "epoch": 0.5985147153204364,
+      "learning_rate": 8.02988906207023e-05,
+      "loss": 0.7102,
+      "mean_token_accuracy": 0.7813880145549774,
+      "num_tokens": 152104967.0,
+      "step": 65280
+    },
+    {
+      "epoch": 0.5986063995599157,
+      "learning_rate": 8.028055377280646e-05,
+      "loss": 0.6715,
+      "mean_token_accuracy": 0.7874277710914612,
+      "num_tokens": 152129490.0,
+      "step": 65290
+    },
+    {
+      "epoch": 0.5986980837993949,
+      "learning_rate": 8.026221692491062e-05,
+      "loss": 0.6921,
+      "mean_token_accuracy": 0.786574387550354,
+      "num_tokens": 152155020.0,
+      "step": 65300
+    },
+    {
+      "epoch": 0.5987897680388741,
+      "learning_rate": 8.024388007701476e-05,
+      "loss": 0.7322,
+      "mean_token_accuracy": 0.7764752984046936,
+      "num_tokens": 152180178.0,
+      "step": 65310
+    },
+    {
+      "epoch": 0.5988814522783533,
+      "learning_rate": 8.022554322911891e-05,
+      "loss": 0.6781,
+      "mean_token_accuracy": 0.787101149559021,
+      "num_tokens": 152204476.0,
+      "step": 65320
+    },
+    {
+      "epoch": 0.5989731365178326,
+      "learning_rate": 8.020720638122308e-05,
+      "loss": 0.7171,
+      "mean_token_accuracy": 0.7777836680412292,
+      "num_tokens": 152230885.0,
+      "step": 65330
+    },
+    {
+      "epoch": 0.5990648207573118,
+      "learning_rate": 8.018886953332723e-05,
+      "loss": 0.6962,
+      "mean_token_accuracy": 0.7864459693431854,
+      "num_tokens": 152256924.0,
+      "step": 65340
+    },
+    {
+      "epoch": 0.5991565049967911,
+      "learning_rate": 8.017053268543138e-05,
+      "loss": 0.6956,
+      "mean_token_accuracy": 0.7886278092861175,
+      "num_tokens": 152281849.0,
+      "step": 65350
+    },
+    {
+      "epoch": 0.5992481892362703,
+      "learning_rate": 8.015219583753552e-05,
+      "loss": 0.6831,
+      "mean_token_accuracy": 0.7853272080421447,
+      "num_tokens": 152306272.0,
+      "step": 65360
+    },
+    {
+      "epoch": 0.5993398734757496,
+      "learning_rate": 8.013385898963969e-05,
+      "loss": 0.7205,
+      "mean_token_accuracy": 0.7834070980548858,
+      "num_tokens": 152331405.0,
+      "step": 65370
+    },
+    {
+      "epoch": 0.5994315577152287,
+      "learning_rate": 8.011552214174384e-05,
+      "loss": 0.7185,
+      "mean_token_accuracy": 0.7829317331314087,
+      "num_tokens": 152356810.0,
+      "step": 65380
+    },
+    {
+      "epoch": 0.599523241954708,
+      "learning_rate": 8.009718529384799e-05,
+      "loss": 0.7058,
+      "mean_token_accuracy": 0.7811625599861145,
+      "num_tokens": 152381447.0,
+      "step": 65390
+    },
+    {
+      "epoch": 0.5996149261941872,
+      "learning_rate": 8.007884844595215e-05,
+      "loss": 0.6892,
+      "mean_token_accuracy": 0.7936155915260314,
+      "num_tokens": 152406264.0,
+      "step": 65400
+    },
+    {
+      "epoch": 0.5997066104336665,
+      "learning_rate": 8.006051159805629e-05,
+      "loss": 0.6826,
+      "mean_token_accuracy": 0.7890279054641723,
+      "num_tokens": 152431032.0,
+      "step": 65410
+    },
+    {
+      "epoch": 0.5997982946731457,
+      "learning_rate": 8.004217475016045e-05,
+      "loss": 0.6958,
+      "mean_token_accuracy": 0.7899737179279327,
+      "num_tokens": 152456170.0,
+      "step": 65420
+    },
+    {
+      "epoch": 0.599889978912625,
+      "learning_rate": 8.002383790226461e-05,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.7748258411884308,
+      "num_tokens": 152480781.0,
+      "step": 65430
+    },
+    {
+      "epoch": 0.5999816631521041,
+      "learning_rate": 8.000550105436876e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.7952584981918335,
+      "num_tokens": 152505840.0,
+      "step": 65440
+    },
+    {
+      "epoch": 0.6000733473915834,
+      "learning_rate": 7.998716420647291e-05,
+      "loss": 0.7077,
+      "mean_token_accuracy": 0.7831570863723755,
+      "num_tokens": 152531592.0,
+      "step": 65450
+    },
+    {
+      "epoch": 0.6001650316310626,
+      "learning_rate": 7.996882735857707e-05,
+      "loss": 0.7068,
+      "mean_token_accuracy": 0.7849279761314392,
+      "num_tokens": 152556504.0,
+      "step": 65460
+    },
+    {
+      "epoch": 0.6002567158705419,
+      "learning_rate": 7.995049051068122e-05,
+      "loss": 0.7487,
+      "mean_token_accuracy": 0.7772552847862244,
+      "num_tokens": 152581558.0,
+      "step": 65470
+    },
+    {
+      "epoch": 0.6003484001100211,
+      "learning_rate": 7.993215366278537e-05,
+      "loss": 0.6702,
+      "mean_token_accuracy": 0.791909259557724,
+      "num_tokens": 152605845.0,
+      "step": 65480
+    },
+    {
+      "epoch": 0.6004400843495004,
+      "learning_rate": 7.991381681488952e-05,
+      "loss": 0.687,
+      "mean_token_accuracy": 0.7882726311683654,
+      "num_tokens": 152630447.0,
+      "step": 65490
+    },
+    {
+      "epoch": 0.6005317685889796,
+      "learning_rate": 7.989547996699369e-05,
+      "loss": 0.6964,
+      "mean_token_accuracy": 0.7861719489097595,
+      "num_tokens": 152656099.0,
+      "step": 65500
+    },
+    {
+      "epoch": 0.6006234528284587,
+      "learning_rate": 7.987714311909783e-05,
+      "loss": 0.6742,
+      "mean_token_accuracy": 0.7913962602615356,
+      "num_tokens": 152681331.0,
+      "step": 65510
+    },
+    {
+      "epoch": 0.600715137067938,
+      "learning_rate": 7.985880627120198e-05,
+      "loss": 0.7271,
+      "mean_token_accuracy": 0.776265412569046,
+      "num_tokens": 152706181.0,
+      "step": 65520
+    },
+    {
+      "epoch": 0.6008068213074172,
+      "learning_rate": 7.984046942330614e-05,
+      "loss": 0.6754,
+      "mean_token_accuracy": 0.7891752243041992,
+      "num_tokens": 152731027.0,
+      "step": 65530
+    },
+    {
+      "epoch": 0.6008985055468965,
+      "learning_rate": 7.98221325754103e-05,
+      "loss": 0.733,
+      "mean_token_accuracy": 0.7765580773353576,
+      "num_tokens": 152756104.0,
+      "step": 65540
+    },
+    {
+      "epoch": 0.6009901897863758,
+      "learning_rate": 7.980379572751445e-05,
+      "loss": 0.6609,
+      "mean_token_accuracy": 0.794046550989151,
+      "num_tokens": 152781495.0,
+      "step": 65550
+    },
+    {
+      "epoch": 0.601081874025855,
+      "learning_rate": 7.97854588796186e-05,
+      "loss": 0.679,
+      "mean_token_accuracy": 0.789538037776947,
+      "num_tokens": 152806749.0,
+      "step": 65560
+    },
+    {
+      "epoch": 0.6011735582653341,
+      "learning_rate": 7.976712203172275e-05,
+      "loss": 0.6738,
+      "mean_token_accuracy": 0.7955724179744721,
+      "num_tokens": 152832054.0,
+      "step": 65570
+    },
+    {
+      "epoch": 0.6012652425048134,
+      "learning_rate": 7.97487851838269e-05,
+      "loss": 0.7199,
+      "mean_token_accuracy": 0.7878151297569275,
+      "num_tokens": 152857367.0,
+      "step": 65580
+    },
+    {
+      "epoch": 0.6013569267442926,
+      "learning_rate": 7.973044833593107e-05,
+      "loss": 0.7094,
+      "mean_token_accuracy": 0.7771391034126282,
+      "num_tokens": 152881428.0,
+      "step": 65590
+    },
+    {
+      "epoch": 0.6014486109837719,
+      "learning_rate": 7.971211148803522e-05,
+      "loss": 0.6544,
+      "mean_token_accuracy": 0.7940707743167877,
+      "num_tokens": 152906365.0,
+      "step": 65600
+    },
+    {
+      "epoch": 0.6015402952232511,
+      "learning_rate": 7.969377464013936e-05,
+      "loss": 0.6791,
+      "mean_token_accuracy": 0.7848674297332764,
+      "num_tokens": 152932233.0,
+      "step": 65610
+    },
+    {
+      "epoch": 0.6016319794627304,
+      "learning_rate": 7.967543779224351e-05,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.7926348149776459,
+      "num_tokens": 152956550.0,
+      "step": 65620
+    },
+    {
+      "epoch": 0.6017236637022096,
+      "learning_rate": 7.965710094434767e-05,
+      "loss": 0.7123,
+      "mean_token_accuracy": 0.7827710986137391,
+      "num_tokens": 152982137.0,
+      "step": 65630
+    },
+    {
+      "epoch": 0.6018153479416888,
+      "learning_rate": 7.963876409645183e-05,
+      "loss": 0.7092,
+      "mean_token_accuracy": 0.782976770401001,
+      "num_tokens": 153008140.0,
+      "step": 65640
+    },
+    {
+      "epoch": 0.601907032181168,
+      "learning_rate": 7.962042724855598e-05,
+      "loss": 0.6876,
+      "mean_token_accuracy": 0.7896133840084076,
+      "num_tokens": 153033161.0,
+      "step": 65650
+    },
+    {
+      "epoch": 0.6019987164206473,
+      "learning_rate": 7.960209040066013e-05,
+      "loss": 0.701,
+      "mean_token_accuracy": 0.7877236247062683,
+      "num_tokens": 153058576.0,
+      "step": 65660
+    },
+    {
+      "epoch": 0.6020904006601265,
+      "learning_rate": 7.958375355276428e-05,
+      "loss": 0.7046,
+      "mean_token_accuracy": 0.7810765504837036,
+      "num_tokens": 153083607.0,
+      "step": 65670
+    },
+    {
+      "epoch": 0.6021820848996058,
+      "learning_rate": 7.956541670486843e-05,
+      "loss": 0.6827,
+      "mean_token_accuracy": 0.7892462730407714,
+      "num_tokens": 153109400.0,
+      "step": 65680
+    },
+    {
+      "epoch": 0.602273769139085,
+      "learning_rate": 7.95470798569726e-05,
+      "loss": 0.6888,
+      "mean_token_accuracy": 0.7928320884704589,
+      "num_tokens": 153134951.0,
+      "step": 65690
+    },
+    {
+      "epoch": 0.6023654533785642,
+      "learning_rate": 7.952874300907675e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.7948405981063843,
+      "num_tokens": 153160933.0,
+      "step": 65700
+    },
+    {
+      "epoch": 0.6024571376180434,
+      "learning_rate": 7.951040616118089e-05,
+      "loss": 0.7027,
+      "mean_token_accuracy": 0.7873086512088776,
+      "num_tokens": 153186376.0,
+      "step": 65710
+    },
+    {
+      "epoch": 0.6025488218575227,
+      "learning_rate": 7.949206931328504e-05,
+      "loss": 0.6842,
+      "mean_token_accuracy": 0.790751975774765,
+      "num_tokens": 153212680.0,
+      "step": 65720
+    },
+    {
+      "epoch": 0.6026405060970019,
+      "learning_rate": 7.947373246538921e-05,
+      "loss": 0.6753,
+      "mean_token_accuracy": 0.7919830679893494,
+      "num_tokens": 153237542.0,
+      "step": 65730
+    },
+    {
+      "epoch": 0.6027321903364812,
+      "learning_rate": 7.945539561749336e-05,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.7838497817516327,
+      "num_tokens": 153262276.0,
+      "step": 65740
+    },
+    {
+      "epoch": 0.6028238745759604,
+      "learning_rate": 7.943705876959751e-05,
+      "loss": 0.6617,
+      "mean_token_accuracy": 0.791743528842926,
+      "num_tokens": 153287441.0,
+      "step": 65750
+    },
+    {
+      "epoch": 0.6029155588154397,
+      "learning_rate": 7.941872192170166e-05,
+      "loss": 0.7073,
+      "mean_token_accuracy": 0.7812867045402527,
+      "num_tokens": 153312416.0,
+      "step": 65760
+    },
+    {
+      "epoch": 0.6030072430549188,
+      "learning_rate": 7.940038507380581e-05,
+      "loss": 0.6934,
+      "mean_token_accuracy": 0.7891874372959137,
+      "num_tokens": 153338303.0,
+      "step": 65770
+    },
+    {
+      "epoch": 0.6030989272943981,
+      "learning_rate": 7.938204822590997e-05,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.7873879969120026,
+      "num_tokens": 153363072.0,
+      "step": 65780
+    },
+    {
+      "epoch": 0.6031906115338773,
+      "learning_rate": 7.936371137801413e-05,
+      "loss": 0.705,
+      "mean_token_accuracy": 0.7892035901546478,
+      "num_tokens": 153388420.0,
+      "step": 65790
+    },
+    {
+      "epoch": 0.6032822957733566,
+      "learning_rate": 7.934537453011828e-05,
+      "loss": 0.6981,
+      "mean_token_accuracy": 0.7878020882606507,
+      "num_tokens": 153413679.0,
+      "step": 65800
+    },
+    {
+      "epoch": 0.6033739800128358,
+      "learning_rate": 7.932703768222242e-05,
+      "loss": 0.7285,
+      "mean_token_accuracy": 0.7841762900352478,
+      "num_tokens": 153438648.0,
+      "step": 65810
+    },
+    {
+      "epoch": 0.6034656642523151,
+      "learning_rate": 7.930870083432659e-05,
+      "loss": 0.719,
+      "mean_token_accuracy": 0.779742956161499,
+      "num_tokens": 153462828.0,
+      "step": 65820
+    },
+    {
+      "epoch": 0.6035573484917942,
+      "learning_rate": 7.929036398643074e-05,
+      "loss": 0.6949,
+      "mean_token_accuracy": 0.7837848901748657,
+      "num_tokens": 153488382.0,
+      "step": 65830
+    },
+    {
+      "epoch": 0.6036490327312735,
+      "learning_rate": 7.927202713853489e-05,
+      "loss": 0.7072,
+      "mean_token_accuracy": 0.7886743426322937,
+      "num_tokens": 153512937.0,
+      "step": 65840
+    },
+    {
+      "epoch": 0.6037407169707527,
+      "learning_rate": 7.925369029063904e-05,
+      "loss": 0.6816,
+      "mean_token_accuracy": 0.7898770630359649,
+      "num_tokens": 153538815.0,
+      "step": 65850
+    },
+    {
+      "epoch": 0.603832401210232,
+      "learning_rate": 7.92353534427432e-05,
+      "loss": 0.7123,
+      "mean_token_accuracy": 0.7815713942050934,
+      "num_tokens": 153563671.0,
+      "step": 65860
+    },
+    {
+      "epoch": 0.6039240854497112,
+      "learning_rate": 7.921701659484735e-05,
+      "loss": 0.6893,
+      "mean_token_accuracy": 0.7910423338413238,
+      "num_tokens": 153588938.0,
+      "step": 65870
+    },
+    {
+      "epoch": 0.6040157696891905,
+      "learning_rate": 7.91986797469515e-05,
+      "loss": 0.6869,
+      "mean_token_accuracy": 0.7885134279727936,
+      "num_tokens": 153614313.0,
+      "step": 65880
+    },
+    {
+      "epoch": 0.6041074539286697,
+      "learning_rate": 7.918034289905566e-05,
+      "loss": 0.6876,
+      "mean_token_accuracy": 0.7845796167850494,
+      "num_tokens": 153638843.0,
+      "step": 65890
+    },
+    {
+      "epoch": 0.6041991381681489,
+      "learning_rate": 7.916200605115982e-05,
+      "loss": 0.6624,
+      "mean_token_accuracy": 0.7969128489494324,
+      "num_tokens": 153662982.0,
+      "step": 65900
+    },
+    {
+      "epoch": 0.6042908224076281,
+      "learning_rate": 7.914366920326395e-05,
+      "loss": 0.6855,
+      "mean_token_accuracy": 0.7880733489990235,
+      "num_tokens": 153688288.0,
+      "step": 65910
+    },
+    {
+      "epoch": 0.6043825066471074,
+      "learning_rate": 7.912533235536812e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.7834518909454345,
+      "num_tokens": 153714107.0,
+      "step": 65920
+    },
+    {
+      "epoch": 0.6044741908865866,
+      "learning_rate": 7.910699550747227e-05,
+      "loss": 0.675,
+      "mean_token_accuracy": 0.7943445026874543,
+      "num_tokens": 153739601.0,
+      "step": 65930
+    },
+    {
+      "epoch": 0.6045658751260659,
+      "learning_rate": 7.908865865957642e-05,
+      "loss": 0.6808,
+      "mean_token_accuracy": 0.7905770540237427,
+      "num_tokens": 153764907.0,
+      "step": 65940
+    },
+    {
+      "epoch": 0.6046575593655451,
+      "learning_rate": 7.907032181168058e-05,
+      "loss": 0.6786,
+      "mean_token_accuracy": 0.7863018035888671,
+      "num_tokens": 153789890.0,
+      "step": 65950
+    },
+    {
+      "epoch": 0.6047492436050242,
+      "learning_rate": 7.905198496378473e-05,
+      "loss": 0.6935,
+      "mean_token_accuracy": 0.7891980350017548,
+      "num_tokens": 153814873.0,
+      "step": 65960
+    },
+    {
+      "epoch": 0.6048409278445035,
+      "learning_rate": 7.903364811588888e-05,
+      "loss": 0.6963,
+      "mean_token_accuracy": 0.7839482605457306,
+      "num_tokens": 153839615.0,
+      "step": 65970
+    },
+    {
+      "epoch": 0.6049326120839827,
+      "learning_rate": 7.901531126799303e-05,
+      "loss": 0.6444,
+      "mean_token_accuracy": 0.7941844642162323,
+      "num_tokens": 153865150.0,
+      "step": 65980
+    },
+    {
+      "epoch": 0.605024296323462,
+      "learning_rate": 7.89969744200972e-05,
+      "loss": 0.7063,
+      "mean_token_accuracy": 0.7747714757919312,
+      "num_tokens": 153890668.0,
+      "step": 65990
+    },
+    {
+      "epoch": 0.6051159805629412,
+      "learning_rate": 7.897863757220135e-05,
+      "loss": 0.7315,
+      "mean_token_accuracy": 0.7796947598457337,
+      "num_tokens": 153914890.0,
+      "step": 66000
+    },
+    {
+      "epoch": 0.6052076648024205,
+      "learning_rate": 7.896030072430549e-05,
+      "loss": 0.6805,
+      "mean_token_accuracy": 0.7874993741512298,
+      "num_tokens": 153940568.0,
+      "step": 66010
+    },
+    {
+      "epoch": 0.6052993490418997,
+      "learning_rate": 7.894196387640965e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.7885223090648651,
+      "num_tokens": 153966239.0,
+      "step": 66020
+    },
+    {
+      "epoch": 0.6053910332813789,
+      "learning_rate": 7.89236270285138e-05,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.792020708322525,
+      "num_tokens": 153991411.0,
+      "step": 66030
+    },
+    {
+      "epoch": 0.6054827175208581,
+      "learning_rate": 7.890529018061796e-05,
+      "loss": 0.7079,
+      "mean_token_accuracy": 0.7849093317985535,
+      "num_tokens": 154016435.0,
+      "step": 66040
+    },
+    {
+      "epoch": 0.6055744017603374,
+      "learning_rate": 7.888695333272211e-05,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.7888876736164093,
+      "num_tokens": 154041285.0,
+      "step": 66050
+    },
+    {
+      "epoch": 0.6056660859998166,
+      "learning_rate": 7.886861648482626e-05,
+      "loss": 0.7135,
+      "mean_token_accuracy": 0.7864153563976288,
+      "num_tokens": 154066652.0,
+      "step": 66060
+    },
+    {
+      "epoch": 0.6057577702392959,
+      "learning_rate": 7.885027963693041e-05,
+      "loss": 0.6402,
+      "mean_token_accuracy": 0.7941450238227844,
+      "num_tokens": 154091648.0,
+      "step": 66070
+    },
+    {
+      "epoch": 0.6058494544787751,
+      "learning_rate": 7.883194278903458e-05,
+      "loss": 0.687,
+      "mean_token_accuracy": 0.789392215013504,
+      "num_tokens": 154115807.0,
+      "step": 66080
+    },
+    {
+      "epoch": 0.6059411387182543,
+      "learning_rate": 7.881360594113873e-05,
+      "loss": 0.7184,
+      "mean_token_accuracy": 0.7835983037948608,
+      "num_tokens": 154141467.0,
+      "step": 66090
+    },
+    {
+      "epoch": 0.6060328229577335,
+      "learning_rate": 7.879526909324288e-05,
+      "loss": 0.7087,
+      "mean_token_accuracy": 0.7765124142169952,
+      "num_tokens": 154166574.0,
+      "step": 66100
+    },
+    {
+      "epoch": 0.6061245071972128,
+      "learning_rate": 7.877693224534702e-05,
+      "loss": 0.6813,
+      "mean_token_accuracy": 0.7879556477069855,
+      "num_tokens": 154192557.0,
+      "step": 66110
+    },
+    {
+      "epoch": 0.606216191436692,
+      "learning_rate": 7.875859539745118e-05,
+      "loss": 0.7186,
+      "mean_token_accuracy": 0.7812003195285797,
+      "num_tokens": 154217456.0,
+      "step": 66120
+    },
+    {
+      "epoch": 0.6063078756761713,
+      "learning_rate": 7.874025854955534e-05,
+      "loss": 0.6783,
+      "mean_token_accuracy": 0.7895727813243866,
+      "num_tokens": 154241983.0,
+      "step": 66130
+    },
+    {
+      "epoch": 0.6063995599156505,
+      "learning_rate": 7.872192170165949e-05,
+      "loss": 0.72,
+      "mean_token_accuracy": 0.7837452471256257,
+      "num_tokens": 154267125.0,
+      "step": 66140
+    },
+    {
+      "epoch": 0.6064912441551298,
+      "learning_rate": 7.870358485376364e-05,
+      "loss": 0.6925,
+      "mean_token_accuracy": 0.7894594967365265,
+      "num_tokens": 154292480.0,
+      "step": 66150
+    },
+    {
+      "epoch": 0.6065829283946089,
+      "learning_rate": 7.868524800586779e-05,
+      "loss": 0.7277,
+      "mean_token_accuracy": 0.777235871553421,
+      "num_tokens": 154317003.0,
+      "step": 66160
+    },
+    {
+      "epoch": 0.6066746126340882,
+      "learning_rate": 7.866691115797194e-05,
+      "loss": 0.6992,
+      "mean_token_accuracy": 0.78583984375,
+      "num_tokens": 154341664.0,
+      "step": 66170
+    },
+    {
+      "epoch": 0.6067662968735674,
+      "learning_rate": 7.864857431007611e-05,
+      "loss": 0.6981,
+      "mean_token_accuracy": 0.7841795980930328,
+      "num_tokens": 154367685.0,
+      "step": 66180
+    },
+    {
+      "epoch": 0.6068579811130467,
+      "learning_rate": 7.863023746218026e-05,
+      "loss": 0.7056,
+      "mean_token_accuracy": 0.7853238999843597,
+      "num_tokens": 154392983.0,
+      "step": 66190
+    },
+    {
+      "epoch": 0.6069496653525259,
+      "learning_rate": 7.861190061428441e-05,
+      "loss": 0.662,
+      "mean_token_accuracy": 0.788264536857605,
+      "num_tokens": 154417989.0,
+      "step": 66200
+    },
+    {
+      "epoch": 0.6070413495920052,
+      "learning_rate": 7.859356376638856e-05,
+      "loss": 0.6915,
+      "mean_token_accuracy": 0.7927573800086976,
+      "num_tokens": 154442933.0,
+      "step": 66210
+    },
+    {
+      "epoch": 0.6071330338314843,
+      "learning_rate": 7.857522691849272e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.7889310359954834,
+      "num_tokens": 154468468.0,
+      "step": 66220
+    },
+    {
+      "epoch": 0.6072247180709636,
+      "learning_rate": 7.855689007059687e-05,
+      "loss": 0.711,
+      "mean_token_accuracy": 0.781570988893509,
+      "num_tokens": 154494333.0,
+      "step": 66230
+    },
+    {
+      "epoch": 0.6073164023104428,
+      "learning_rate": 7.853855322270102e-05,
+      "loss": 0.7277,
+      "mean_token_accuracy": 0.7829683005809784,
+      "num_tokens": 154518972.0,
+      "step": 66240
+    },
+    {
+      "epoch": 0.6074080865499221,
+      "learning_rate": 7.852021637480517e-05,
+      "loss": 0.6766,
+      "mean_token_accuracy": 0.7889496207237243,
+      "num_tokens": 154544166.0,
+      "step": 66250
+    },
+    {
+      "epoch": 0.6074997707894013,
+      "learning_rate": 7.850187952690932e-05,
+      "loss": 0.6857,
+      "mean_token_accuracy": 0.787873649597168,
+      "num_tokens": 154569209.0,
+      "step": 66260
+    },
+    {
+      "epoch": 0.6075914550288806,
+      "learning_rate": 7.848354267901348e-05,
+      "loss": 0.6607,
+      "mean_token_accuracy": 0.7953478693962097,
+      "num_tokens": 154595147.0,
+      "step": 66270
+    },
+    {
+      "epoch": 0.6076831392683598,
+      "learning_rate": 7.846520583111764e-05,
+      "loss": 0.6955,
+      "mean_token_accuracy": 0.786862713098526,
+      "num_tokens": 154620154.0,
+      "step": 66280
+    },
+    {
+      "epoch": 0.607774823507839,
+      "learning_rate": 7.844686898322179e-05,
+      "loss": 0.6978,
+      "mean_token_accuracy": 0.7850846529006958,
+      "num_tokens": 154646012.0,
+      "step": 66290
+    },
+    {
+      "epoch": 0.6078665077473182,
+      "learning_rate": 7.842853213532594e-05,
+      "loss": 0.6983,
+      "mean_token_accuracy": 0.7846573770046235,
+      "num_tokens": 154670715.0,
+      "step": 66300
+    },
+    {
+      "epoch": 0.6079581919867975,
+      "learning_rate": 7.84101952874301e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.7916235387325287,
+      "num_tokens": 154695808.0,
+      "step": 66310
+    },
+    {
+      "epoch": 0.6080498762262767,
+      "learning_rate": 7.839185843953425e-05,
+      "loss": 0.7,
+      "mean_token_accuracy": 0.7809223353862762,
+      "num_tokens": 154720924.0,
+      "step": 66320
+    },
+    {
+      "epoch": 0.608141560465756,
+      "learning_rate": 7.83735215916384e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.7868905007839203,
+      "num_tokens": 154746165.0,
+      "step": 66330
+    },
+    {
+      "epoch": 0.6082332447052352,
+      "learning_rate": 7.835518474374255e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.7884917736053467,
+      "num_tokens": 154770724.0,
+      "step": 66340
+    },
+    {
+      "epoch": 0.6083249289447143,
+      "learning_rate": 7.83368478958467e-05,
+      "loss": 0.6779,
+      "mean_token_accuracy": 0.7910633623600006,
+      "num_tokens": 154795434.0,
+      "step": 66350
+    },
+    {
+      "epoch": 0.6084166131841936,
+      "learning_rate": 7.831851104795086e-05,
+      "loss": 0.7047,
+      "mean_token_accuracy": 0.7864468812942504,
+      "num_tokens": 154820367.0,
+      "step": 66360
+    },
+    {
+      "epoch": 0.6085082974236729,
+      "learning_rate": 7.830017420005501e-05,
+      "loss": 0.7404,
+      "mean_token_accuracy": 0.774083936214447,
+      "num_tokens": 154845031.0,
+      "step": 66370
+    },
+    {
+      "epoch": 0.6085999816631521,
+      "learning_rate": 7.828183735215917e-05,
+      "loss": 0.705,
+      "mean_token_accuracy": 0.7764290571212769,
+      "num_tokens": 154870638.0,
+      "step": 66380
+    },
+    {
+      "epoch": 0.6086916659026314,
+      "learning_rate": 7.826350050426333e-05,
+      "loss": 0.7305,
+      "mean_token_accuracy": 0.7796914041042328,
+      "num_tokens": 154894919.0,
+      "step": 66390
+    },
+    {
+      "epoch": 0.6087833501421106,
+      "learning_rate": 7.824516365636748e-05,
+      "loss": 0.6968,
+      "mean_token_accuracy": 0.786124074459076,
+      "num_tokens": 154920073.0,
+      "step": 66400
+    },
+    {
+      "epoch": 0.6088750343815899,
+      "learning_rate": 7.822682680847163e-05,
+      "loss": 0.7448,
+      "mean_token_accuracy": 0.7776680946350097,
+      "num_tokens": 154945245.0,
+      "step": 66410
+    },
+    {
+      "epoch": 0.608966718621069,
+      "learning_rate": 7.820848996057578e-05,
+      "loss": 0.7104,
+      "mean_token_accuracy": 0.789386659860611,
+      "num_tokens": 154970398.0,
+      "step": 66420
+    },
+    {
+      "epoch": 0.6090584028605482,
+      "learning_rate": 7.819015311267993e-05,
+      "loss": 0.6997,
+      "mean_token_accuracy": 0.7849281609058381,
+      "num_tokens": 154995479.0,
+      "step": 66430
+    },
+    {
+      "epoch": 0.6091500871000275,
+      "learning_rate": 7.81718162647841e-05,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.7780871987342834,
+      "num_tokens": 155019548.0,
+      "step": 66440
+    },
+    {
+      "epoch": 0.6092417713395067,
+      "learning_rate": 7.815347941688824e-05,
+      "loss": 0.7068,
+      "mean_token_accuracy": 0.7843786537647247,
+      "num_tokens": 155044277.0,
+      "step": 66450
+    },
+    {
+      "epoch": 0.609333455578986,
+      "learning_rate": 7.813514256899239e-05,
+      "loss": 0.6721,
+      "mean_token_accuracy": 0.7907335042953492,
+      "num_tokens": 155069265.0,
+      "step": 66460
+    },
+    {
+      "epoch": 0.6094251398184652,
+      "learning_rate": 7.811680572109654e-05,
+      "loss": 0.6991,
+      "mean_token_accuracy": 0.7900092720985412,
+      "num_tokens": 155093391.0,
+      "step": 66470
+    },
+    {
+      "epoch": 0.6095168240579444,
+      "learning_rate": 7.80984688732007e-05,
+      "loss": 0.6776,
+      "mean_token_accuracy": 0.79387868642807,
+      "num_tokens": 155118283.0,
+      "step": 66480
+    },
+    {
+      "epoch": 0.6096085082974236,
+      "learning_rate": 7.808013202530486e-05,
+      "loss": 0.7238,
+      "mean_token_accuracy": 0.7838686764240265,
+      "num_tokens": 155143267.0,
+      "step": 66490
+    },
+    {
+      "epoch": 0.6097001925369029,
+      "learning_rate": 7.806179517740901e-05,
+      "loss": 0.6896,
+      "mean_token_accuracy": 0.7886497557163239,
+      "num_tokens": 155169101.0,
+      "step": 66500
+    },
+    {
+      "epoch": 0.6097918767763821,
+      "learning_rate": 7.804345832951316e-05,
+      "loss": 0.7015,
+      "mean_token_accuracy": 0.78909552693367,
+      "num_tokens": 155194197.0,
+      "step": 66510
+    },
+    {
+      "epoch": 0.6098835610158614,
+      "learning_rate": 7.802512148161731e-05,
+      "loss": 0.7024,
+      "mean_token_accuracy": 0.7856522023677825,
+      "num_tokens": 155218838.0,
+      "step": 66520
+    },
+    {
+      "epoch": 0.6099752452553406,
+      "learning_rate": 7.800678463372146e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.781846672296524,
+      "num_tokens": 155244430.0,
+      "step": 66530
+    },
+    {
+      "epoch": 0.6100669294948199,
+      "learning_rate": 7.798844778582563e-05,
+      "loss": 0.71,
+      "mean_token_accuracy": 0.7862145066261291,
+      "num_tokens": 155269316.0,
+      "step": 66540
+    },
+    {
+      "epoch": 0.610158613734299,
+      "learning_rate": 7.797011093792977e-05,
+      "loss": 0.6943,
+      "mean_token_accuracy": 0.7887844622135163,
+      "num_tokens": 155294074.0,
+      "step": 66550
+    },
+    {
+      "epoch": 0.6102502979737783,
+      "learning_rate": 7.795177409003392e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.7871222496032715,
+      "num_tokens": 155319491.0,
+      "step": 66560
+    },
+    {
+      "epoch": 0.6103419822132575,
+      "learning_rate": 7.793343724213809e-05,
+      "loss": 0.6749,
+      "mean_token_accuracy": 0.7936574578285217,
+      "num_tokens": 155344162.0,
+      "step": 66570
+    },
+    {
+      "epoch": 0.6104336664527368,
+      "learning_rate": 7.791510039424224e-05,
+      "loss": 0.7191,
+      "mean_token_accuracy": 0.7807522296905518,
+      "num_tokens": 155369480.0,
+      "step": 66580
+    },
+    {
+      "epoch": 0.610525350692216,
+      "learning_rate": 7.789676354634639e-05,
+      "loss": 0.7317,
+      "mean_token_accuracy": 0.7801062822341919,
+      "num_tokens": 155394359.0,
+      "step": 66590
+    },
+    {
+      "epoch": 0.6106170349316953,
+      "learning_rate": 7.787842669845054e-05,
+      "loss": 0.7029,
+      "mean_token_accuracy": 0.7821816086769104,
+      "num_tokens": 155420199.0,
+      "step": 66600
+    },
+    {
+      "epoch": 0.6107087191711745,
+      "learning_rate": 7.78600898505547e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.7910413682460785,
+      "num_tokens": 155445553.0,
+      "step": 66610
+    },
+    {
+      "epoch": 0.6108004034106537,
+      "learning_rate": 7.784175300265885e-05,
+      "loss": 0.6781,
+      "mean_token_accuracy": 0.7924752593040466,
+      "num_tokens": 155471458.0,
+      "step": 66620
+    },
+    {
+      "epoch": 0.6108920876501329,
+      "learning_rate": 7.7823416154763e-05,
+      "loss": 0.6906,
+      "mean_token_accuracy": 0.7882384598255158,
+      "num_tokens": 155496276.0,
+      "step": 66630
+    },
+    {
+      "epoch": 0.6109837718896122,
+      "learning_rate": 7.780507930686716e-05,
+      "loss": 0.7126,
+      "mean_token_accuracy": 0.7846039772033692,
+      "num_tokens": 155522026.0,
+      "step": 66640
+    },
+    {
+      "epoch": 0.6110754561290914,
+      "learning_rate": 7.77867424589713e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.7927641093730926,
+      "num_tokens": 155547187.0,
+      "step": 66650
+    },
+    {
+      "epoch": 0.6111671403685707,
+      "learning_rate": 7.776840561107545e-05,
+      "loss": 0.7098,
+      "mean_token_accuracy": 0.7821052551269532,
+      "num_tokens": 155572669.0,
+      "step": 66660
+    },
+    {
+      "epoch": 0.6112588246080499,
+      "learning_rate": 7.775006876317962e-05,
+      "loss": 0.6805,
+      "mean_token_accuracy": 0.7912693440914154,
+      "num_tokens": 155597605.0,
+      "step": 66670
+    },
+    {
+      "epoch": 0.6113505088475291,
+      "learning_rate": 7.773173191528377e-05,
+      "loss": 0.7139,
+      "mean_token_accuracy": 0.7910090684890747,
+      "num_tokens": 155621597.0,
+      "step": 66680
+    },
+    {
+      "epoch": 0.6114421930870083,
+      "learning_rate": 7.771339506738792e-05,
+      "loss": 0.6892,
+      "mean_token_accuracy": 0.7859422266483307,
+      "num_tokens": 155646764.0,
+      "step": 66690
+    },
+    {
+      "epoch": 0.6115338773264876,
+      "learning_rate": 7.769505821949207e-05,
+      "loss": 0.6584,
+      "mean_token_accuracy": 0.7977186799049377,
+      "num_tokens": 155671194.0,
+      "step": 66700
+    },
+    {
+      "epoch": 0.6116255615659668,
+      "learning_rate": 7.767672137159623e-05,
+      "loss": 0.7097,
+      "mean_token_accuracy": 0.7835528433322907,
+      "num_tokens": 155695446.0,
+      "step": 66710
+    },
+    {
+      "epoch": 0.6117172458054461,
+      "learning_rate": 7.765838452370038e-05,
+      "loss": 0.6863,
+      "mean_token_accuracy": 0.7886665225028991,
+      "num_tokens": 155720593.0,
+      "step": 66720
+    },
+    {
+      "epoch": 0.6118089300449253,
+      "learning_rate": 7.764004767580453e-05,
+      "loss": 0.7051,
+      "mean_token_accuracy": 0.7883149862289429,
+      "num_tokens": 155745718.0,
+      "step": 66730
+    },
+    {
+      "epoch": 0.6119006142844046,
+      "learning_rate": 7.76217108279087e-05,
+      "loss": 0.684,
+      "mean_token_accuracy": 0.7954345703125,
+      "num_tokens": 155771621.0,
+      "step": 66740
+    },
+    {
+      "epoch": 0.6119922985238837,
+      "learning_rate": 7.760337398001283e-05,
+      "loss": 0.6599,
+      "mean_token_accuracy": 0.7985195457935333,
+      "num_tokens": 155797266.0,
+      "step": 66750
+    },
+    {
+      "epoch": 0.612083982763363,
+      "learning_rate": 7.758503713211699e-05,
+      "loss": 0.6536,
+      "mean_token_accuracy": 0.7971826255321502,
+      "num_tokens": 155822315.0,
+      "step": 66760
+    },
+    {
+      "epoch": 0.6121756670028422,
+      "learning_rate": 7.756670028422115e-05,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7700275838375091,
+      "num_tokens": 155847309.0,
+      "step": 66770
+    },
+    {
+      "epoch": 0.6122673512423215,
+      "learning_rate": 7.75483634363253e-05,
+      "loss": 0.6882,
+      "mean_token_accuracy": 0.7865087866783143,
+      "num_tokens": 155872658.0,
+      "step": 66780
+    },
+    {
+      "epoch": 0.6123590354818007,
+      "learning_rate": 7.753002658842945e-05,
+      "loss": 0.7319,
+      "mean_token_accuracy": 0.7749264895915985,
+      "num_tokens": 155897432.0,
+      "step": 66790
+    },
+    {
+      "epoch": 0.61245071972128,
+      "learning_rate": 7.75116897405336e-05,
+      "loss": 0.6767,
+      "mean_token_accuracy": 0.7922833502292633,
+      "num_tokens": 155923215.0,
+      "step": 66800
+    },
+    {
+      "epoch": 0.6125424039607591,
+      "learning_rate": 7.749335289263776e-05,
+      "loss": 0.7127,
+      "mean_token_accuracy": 0.7818540394306183,
+      "num_tokens": 155947542.0,
+      "step": 66810
+    },
+    {
+      "epoch": 0.6126340882002383,
+      "learning_rate": 7.747501604474191e-05,
+      "loss": 0.6779,
+      "mean_token_accuracy": 0.7884433686733245,
+      "num_tokens": 155972351.0,
+      "step": 66820
+    },
+    {
+      "epoch": 0.6127257724397176,
+      "learning_rate": 7.745667919684606e-05,
+      "loss": 0.6881,
+      "mean_token_accuracy": 0.7872499525547028,
+      "num_tokens": 155997837.0,
+      "step": 66830
+    },
+    {
+      "epoch": 0.6128174566791968,
+      "learning_rate": 7.743834234895023e-05,
+      "loss": 0.7038,
+      "mean_token_accuracy": 0.7881186842918396,
+      "num_tokens": 156023628.0,
+      "step": 66840
+    },
+    {
+      "epoch": 0.6129091409186761,
+      "learning_rate": 7.742000550105437e-05,
+      "loss": 0.6833,
+      "mean_token_accuracy": 0.7897855699062347,
+      "num_tokens": 156048835.0,
+      "step": 66850
+    },
+    {
+      "epoch": 0.6130008251581553,
+      "learning_rate": 7.740166865315852e-05,
+      "loss": 0.7076,
+      "mean_token_accuracy": 0.7861597776412964,
+      "num_tokens": 156074913.0,
+      "step": 66860
+    },
+    {
+      "epoch": 0.6130925093976346,
+      "learning_rate": 7.738333180526268e-05,
+      "loss": 0.7162,
+      "mean_token_accuracy": 0.7880714237689972,
+      "num_tokens": 156100667.0,
+      "step": 66870
+    },
+    {
+      "epoch": 0.6131841936371137,
+      "learning_rate": 7.736499495736683e-05,
+      "loss": 0.6888,
+      "mean_token_accuracy": 0.7895596146583557,
+      "num_tokens": 156125793.0,
+      "step": 66880
+    },
+    {
+      "epoch": 0.613275877876593,
+      "learning_rate": 7.734665810947099e-05,
+      "loss": 0.7233,
+      "mean_token_accuracy": 0.7790824055671692,
+      "num_tokens": 156150583.0,
+      "step": 66890
+    },
+    {
+      "epoch": 0.6133675621160722,
+      "learning_rate": 7.732832126157514e-05,
+      "loss": 0.6943,
+      "mean_token_accuracy": 0.7864419519901276,
+      "num_tokens": 156176158.0,
+      "step": 66900
+    },
+    {
+      "epoch": 0.6134592463555515,
+      "learning_rate": 7.730998441367929e-05,
+      "loss": 0.667,
+      "mean_token_accuracy": 0.7933214724063873,
+      "num_tokens": 156201836.0,
+      "step": 66910
+    },
+    {
+      "epoch": 0.6135509305950307,
+      "learning_rate": 7.729164756578344e-05,
+      "loss": 0.7022,
+      "mean_token_accuracy": 0.7909168124198913,
+      "num_tokens": 156226603.0,
+      "step": 66920
+    },
+    {
+      "epoch": 0.61364261483451,
+      "learning_rate": 7.727331071788761e-05,
+      "loss": 0.6734,
+      "mean_token_accuracy": 0.7896043121814728,
+      "num_tokens": 156251709.0,
+      "step": 66930
+    },
+    {
+      "epoch": 0.6137342990739891,
+      "learning_rate": 7.725497386999176e-05,
+      "loss": 0.7271,
+      "mean_token_accuracy": 0.781232750415802,
+      "num_tokens": 156277127.0,
+      "step": 66940
+    },
+    {
+      "epoch": 0.6138259833134684,
+      "learning_rate": 7.72366370220959e-05,
+      "loss": 0.6989,
+      "mean_token_accuracy": 0.7875811159610748,
+      "num_tokens": 156302300.0,
+      "step": 66950
+    },
+    {
+      "epoch": 0.6139176675529476,
+      "learning_rate": 7.721830017420005e-05,
+      "loss": 0.7139,
+      "mean_token_accuracy": 0.7815785825252533,
+      "num_tokens": 156327265.0,
+      "step": 66960
+    },
+    {
+      "epoch": 0.6140093517924269,
+      "learning_rate": 7.719996332630421e-05,
+      "loss": 0.6665,
+      "mean_token_accuracy": 0.7898998618125915,
+      "num_tokens": 156353015.0,
+      "step": 66970
+    },
+    {
+      "epoch": 0.6141010360319061,
+      "learning_rate": 7.718162647840837e-05,
+      "loss": 0.7116,
+      "mean_token_accuracy": 0.7818050742149353,
+      "num_tokens": 156378368.0,
+      "step": 66980
+    },
+    {
+      "epoch": 0.6141927202713854,
+      "learning_rate": 7.716328963051252e-05,
+      "loss": 0.6938,
+      "mean_token_accuracy": 0.7857789099216461,
+      "num_tokens": 156403537.0,
+      "step": 66990
+    },
+    {
+      "epoch": 0.6142844045108646,
+      "learning_rate": 7.714495278261667e-05,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.7888389229774475,
+      "num_tokens": 156428130.0,
+      "step": 67000
+    },
+    {
+      "epoch": 0.6143760887503438,
+      "learning_rate": 7.712661593472082e-05,
+      "loss": 0.7437,
+      "mean_token_accuracy": 0.7711461544036865,
+      "num_tokens": 156453116.0,
+      "step": 67010
+    },
+    {
+      "epoch": 0.614467772989823,
+      "learning_rate": 7.710827908682497e-05,
+      "loss": 0.7387,
+      "mean_token_accuracy": 0.7768138706684112,
+      "num_tokens": 156478578.0,
+      "step": 67020
+    },
+    {
+      "epoch": 0.6145594572293023,
+      "learning_rate": 7.708994223892914e-05,
+      "loss": 0.6782,
+      "mean_token_accuracy": 0.7902256846427917,
+      "num_tokens": 156503763.0,
+      "step": 67030
+    },
+    {
+      "epoch": 0.6146511414687815,
+      "learning_rate": 7.707160539103329e-05,
+      "loss": 0.7186,
+      "mean_token_accuracy": 0.7766977071762085,
+      "num_tokens": 156529907.0,
+      "step": 67040
+    },
+    {
+      "epoch": 0.6147428257082608,
+      "learning_rate": 7.705326854313743e-05,
+      "loss": 0.6821,
+      "mean_token_accuracy": 0.7826661825180053,
+      "num_tokens": 156554295.0,
+      "step": 67050
+    },
+    {
+      "epoch": 0.61483450994774,
+      "learning_rate": 7.70349316952416e-05,
+      "loss": 0.7177,
+      "mean_token_accuracy": 0.7832845389842987,
+      "num_tokens": 156579377.0,
+      "step": 67060
+    },
+    {
+      "epoch": 0.6149261941872192,
+      "learning_rate": 7.701659484734575e-05,
+      "loss": 0.6498,
+      "mean_token_accuracy": 0.7975553512573242,
+      "num_tokens": 156604321.0,
+      "step": 67070
+    },
+    {
+      "epoch": 0.6150178784266984,
+      "learning_rate": 7.69982579994499e-05,
+      "loss": 0.676,
+      "mean_token_accuracy": 0.7933587968349457,
+      "num_tokens": 156630279.0,
+      "step": 67080
+    },
+    {
+      "epoch": 0.6151095626661777,
+      "learning_rate": 7.697992115155405e-05,
+      "loss": 0.6865,
+      "mean_token_accuracy": 0.7899942874908448,
+      "num_tokens": 156655172.0,
+      "step": 67090
+    },
+    {
+      "epoch": 0.6152012469056569,
+      "learning_rate": 7.69615843036582e-05,
+      "loss": 0.6712,
+      "mean_token_accuracy": 0.7936025261878967,
+      "num_tokens": 156680564.0,
+      "step": 67100
+    },
+    {
+      "epoch": 0.6152929311451362,
+      "learning_rate": 7.694324745576235e-05,
+      "loss": 0.6335,
+      "mean_token_accuracy": 0.7987747192382812,
+      "num_tokens": 156705787.0,
+      "step": 67110
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "learning_rate": 7.69249106078665e-05,
+      "loss": 0.7345,
+      "mean_token_accuracy": 0.7793562829494476,
+      "num_tokens": 156730527.0,
+      "step": 67120
+    },
+    {
+      "epoch": 0.6154762996240947,
+      "learning_rate": 7.690657375997067e-05,
+      "loss": 0.7091,
+      "mean_token_accuracy": 0.783651489019394,
+      "num_tokens": 156755189.0,
+      "step": 67130
+    },
+    {
+      "epoch": 0.6155679838635738,
+      "learning_rate": 7.688823691207482e-05,
+      "loss": 0.6564,
+      "mean_token_accuracy": 0.7978248178958893,
+      "num_tokens": 156779933.0,
+      "step": 67140
+    },
+    {
+      "epoch": 0.6156596681030531,
+      "learning_rate": 7.686990006417896e-05,
+      "loss": 0.7086,
+      "mean_token_accuracy": 0.7844275593757629,
+      "num_tokens": 156805423.0,
+      "step": 67150
+    },
+    {
+      "epoch": 0.6157513523425323,
+      "learning_rate": 7.685156321628313e-05,
+      "loss": 0.7118,
+      "mean_token_accuracy": 0.7816203474998474,
+      "num_tokens": 156829540.0,
+      "step": 67160
+    },
+    {
+      "epoch": 0.6158430365820116,
+      "learning_rate": 7.683322636838728e-05,
+      "loss": 0.7119,
+      "mean_token_accuracy": 0.7806913375854492,
+      "num_tokens": 156854815.0,
+      "step": 67170
+    },
+    {
+      "epoch": 0.6159347208214908,
+      "learning_rate": 7.681488952049143e-05,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.7896641671657563,
+      "num_tokens": 156879989.0,
+      "step": 67180
+    },
+    {
+      "epoch": 0.6160264050609701,
+      "learning_rate": 7.679655267259558e-05,
+      "loss": 0.683,
+      "mean_token_accuracy": 0.7872275531291961,
+      "num_tokens": 156905207.0,
+      "step": 67190
+    },
+    {
+      "epoch": 0.6161180893004492,
+      "learning_rate": 7.677821582469973e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.7908446371555329,
+      "num_tokens": 156930267.0,
+      "step": 67200
+    },
+    {
+      "epoch": 0.6162097735399285,
+      "learning_rate": 7.675987897680389e-05,
+      "loss": 0.6878,
+      "mean_token_accuracy": 0.7871572077274323,
+      "num_tokens": 156955934.0,
+      "step": 67210
+    },
+    {
+      "epoch": 0.6163014577794077,
+      "learning_rate": 7.674154212890804e-05,
+      "loss": 0.7319,
+      "mean_token_accuracy": 0.7791761040687561,
+      "num_tokens": 156980544.0,
+      "step": 67220
+    },
+    {
+      "epoch": 0.616393142018887,
+      "learning_rate": 7.67232052810122e-05,
+      "loss": 0.7133,
+      "mean_token_accuracy": 0.786835378408432,
+      "num_tokens": 157005548.0,
+      "step": 67230
+    },
+    {
+      "epoch": 0.6164848262583662,
+      "learning_rate": 7.670486843311636e-05,
+      "loss": 0.7041,
+      "mean_token_accuracy": 0.7829742968082428,
+      "num_tokens": 157030166.0,
+      "step": 67240
+    },
+    {
+      "epoch": 0.6165765104978455,
+      "learning_rate": 7.66865315852205e-05,
+      "loss": 0.6677,
+      "mean_token_accuracy": 0.790830647945404,
+      "num_tokens": 157055006.0,
+      "step": 67250
+    },
+    {
+      "epoch": 0.6166681947373247,
+      "learning_rate": 7.666819473732466e-05,
+      "loss": 0.673,
+      "mean_token_accuracy": 0.7910701930522919,
+      "num_tokens": 157079972.0,
+      "step": 67260
+    },
+    {
+      "epoch": 0.6167598789768038,
+      "learning_rate": 7.664985788942881e-05,
+      "loss": 0.6681,
+      "mean_token_accuracy": 0.7905817687511444,
+      "num_tokens": 157104639.0,
+      "step": 67270
+    },
+    {
+      "epoch": 0.6168515632162831,
+      "learning_rate": 7.663152104153296e-05,
+      "loss": 0.7024,
+      "mean_token_accuracy": 0.7871637105941772,
+      "num_tokens": 157129677.0,
+      "step": 67280
+    },
+    {
+      "epoch": 0.6169432474557623,
+      "learning_rate": 7.661318419363712e-05,
+      "loss": 0.6734,
+      "mean_token_accuracy": 0.7909211337566375,
+      "num_tokens": 157154585.0,
+      "step": 67290
+    },
+    {
+      "epoch": 0.6170349316952416,
+      "learning_rate": 7.659484734574127e-05,
+      "loss": 0.6731,
+      "mean_token_accuracy": 0.7936107039451599,
+      "num_tokens": 157179804.0,
+      "step": 67300
+    },
+    {
+      "epoch": 0.6171266159347208,
+      "learning_rate": 7.657651049784542e-05,
+      "loss": 0.6973,
+      "mean_token_accuracy": 0.7843657255172729,
+      "num_tokens": 157205066.0,
+      "step": 67310
+    },
+    {
+      "epoch": 0.6172183001742001,
+      "learning_rate": 7.655817364994958e-05,
+      "loss": 0.7115,
+      "mean_token_accuracy": 0.7843078374862671,
+      "num_tokens": 157229961.0,
+      "step": 67320
+    },
+    {
+      "epoch": 0.6173099844136792,
+      "learning_rate": 7.653983680205374e-05,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.788511061668396,
+      "num_tokens": 157255586.0,
+      "step": 67330
+    },
+    {
+      "epoch": 0.6174016686531585,
+      "learning_rate": 7.652149995415789e-05,
+      "loss": 0.6678,
+      "mean_token_accuracy": 0.7950996577739715,
+      "num_tokens": 157281970.0,
+      "step": 67340
+    },
+    {
+      "epoch": 0.6174933528926377,
+      "learning_rate": 7.650316310626203e-05,
+      "loss": 0.6811,
+      "mean_token_accuracy": 0.7874180376529694,
+      "num_tokens": 157307710.0,
+      "step": 67350
+    },
+    {
+      "epoch": 0.617585037132117,
+      "learning_rate": 7.648482625836619e-05,
+      "loss": 0.65,
+      "mean_token_accuracy": 0.7945042431354523,
+      "num_tokens": 157332924.0,
+      "step": 67360
+    },
+    {
+      "epoch": 0.6176767213715962,
+      "learning_rate": 7.646648941047034e-05,
+      "loss": 0.6894,
+      "mean_token_accuracy": 0.7862263977527618,
+      "num_tokens": 157358222.0,
+      "step": 67370
+    },
+    {
+      "epoch": 0.6177684056110755,
+      "learning_rate": 7.64481525625745e-05,
+      "loss": 0.7443,
+      "mean_token_accuracy": 0.7707981467247009,
+      "num_tokens": 157382854.0,
+      "step": 67380
+    },
+    {
+      "epoch": 0.6178600898505547,
+      "learning_rate": 7.642981571467865e-05,
+      "loss": 0.7035,
+      "mean_token_accuracy": 0.7888554453849792,
+      "num_tokens": 157407841.0,
+      "step": 67390
+    },
+    {
+      "epoch": 0.6179517740900339,
+      "learning_rate": 7.64114788667828e-05,
+      "loss": 0.6663,
+      "mean_token_accuracy": 0.7951247453689575,
+      "num_tokens": 157432579.0,
+      "step": 67400
+    },
+    {
+      "epoch": 0.6180434583295131,
+      "learning_rate": 7.639314201888695e-05,
+      "loss": 0.6863,
+      "mean_token_accuracy": 0.7897823691368103,
+      "num_tokens": 157456865.0,
+      "step": 67410
+    },
+    {
+      "epoch": 0.6181351425689924,
+      "learning_rate": 7.637480517099112e-05,
+      "loss": 0.7004,
+      "mean_token_accuracy": 0.7874565184116363,
+      "num_tokens": 157482026.0,
+      "step": 67420
+    },
+    {
+      "epoch": 0.6182268268084716,
+      "learning_rate": 7.635646832309527e-05,
+      "loss": 0.7098,
+      "mean_token_accuracy": 0.780177116394043,
+      "num_tokens": 157507826.0,
+      "step": 67430
+    },
+    {
+      "epoch": 0.6183185110479509,
+      "learning_rate": 7.633813147519942e-05,
+      "loss": 0.6708,
+      "mean_token_accuracy": 0.7879915475845337,
+      "num_tokens": 157533046.0,
+      "step": 67440
+    },
+    {
+      "epoch": 0.6184101952874301,
+      "learning_rate": 7.631979462730356e-05,
+      "loss": 0.6771,
+      "mean_token_accuracy": 0.7879619240760803,
+      "num_tokens": 157558964.0,
+      "step": 67450
+    },
+    {
+      "epoch": 0.6185018795269093,
+      "learning_rate": 7.630145777940772e-05,
+      "loss": 0.6794,
+      "mean_token_accuracy": 0.7913801431655884,
+      "num_tokens": 157583569.0,
+      "step": 67460
+    },
+    {
+      "epoch": 0.6185935637663885,
+      "learning_rate": 7.628312093151188e-05,
+      "loss": 0.6729,
+      "mean_token_accuracy": 0.794334328174591,
+      "num_tokens": 157607935.0,
+      "step": 67470
+    },
+    {
+      "epoch": 0.6186852480058678,
+      "learning_rate": 7.626478408361603e-05,
+      "loss": 0.6404,
+      "mean_token_accuracy": 0.7993609726428985,
+      "num_tokens": 157633464.0,
+      "step": 67480
+    },
+    {
+      "epoch": 0.618776932245347,
+      "learning_rate": 7.624644723572018e-05,
+      "loss": 0.6785,
+      "mean_token_accuracy": 0.7915915846824646,
+      "num_tokens": 157659110.0,
+      "step": 67490
+    },
+    {
+      "epoch": 0.6188686164848263,
+      "learning_rate": 7.622811038782433e-05,
+      "loss": 0.6898,
+      "mean_token_accuracy": 0.7868053615093231,
+      "num_tokens": 157684742.0,
+      "step": 67500
+    },
+    {
+      "epoch": 0.6189603007243055,
+      "learning_rate": 7.620977353992848e-05,
+      "loss": 0.6743,
+      "mean_token_accuracy": 0.7934269189834595,
+      "num_tokens": 157708553.0,
+      "step": 67510
+    },
+    {
+      "epoch": 0.6190519849637848,
+      "learning_rate": 7.619143669203265e-05,
+      "loss": 0.6754,
+      "mean_token_accuracy": 0.7853560388088227,
+      "num_tokens": 157733195.0,
+      "step": 67520
+    },
+    {
+      "epoch": 0.6191436692032639,
+      "learning_rate": 7.61730998441368e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.792363291978836,
+      "num_tokens": 157758076.0,
+      "step": 67530
+    },
+    {
+      "epoch": 0.6192353534427432,
+      "learning_rate": 7.615476299624095e-05,
+      "loss": 0.6926,
+      "mean_token_accuracy": 0.7883324027061462,
+      "num_tokens": 157783227.0,
+      "step": 67540
+    },
+    {
+      "epoch": 0.6193270376822224,
+      "learning_rate": 7.61364261483451e-05,
+      "loss": 0.6745,
+      "mean_token_accuracy": 0.7916837930679321,
+      "num_tokens": 157808657.0,
+      "step": 67550
+    },
+    {
+      "epoch": 0.6194187219217017,
+      "learning_rate": 7.611808930044926e-05,
+      "loss": 0.659,
+      "mean_token_accuracy": 0.7989928483963012,
+      "num_tokens": 157833941.0,
+      "step": 67560
+    },
+    {
+      "epoch": 0.6195104061611809,
+      "learning_rate": 7.609975245255341e-05,
+      "loss": 0.7092,
+      "mean_token_accuracy": 0.7815617918968201,
+      "num_tokens": 157858955.0,
+      "step": 67570
+    },
+    {
+      "epoch": 0.6196020904006602,
+      "learning_rate": 7.608141560465756e-05,
+      "loss": 0.6927,
+      "mean_token_accuracy": 0.789283150434494,
+      "num_tokens": 157884222.0,
+      "step": 67580
+    },
+    {
+      "epoch": 0.6196937746401393,
+      "learning_rate": 7.606307875676171e-05,
+      "loss": 0.6934,
+      "mean_token_accuracy": 0.7907307505607605,
+      "num_tokens": 157909958.0,
+      "step": 67590
+    },
+    {
+      "epoch": 0.6197854588796186,
+      "learning_rate": 7.604474190886586e-05,
+      "loss": 0.7398,
+      "mean_token_accuracy": 0.774592113494873,
+      "num_tokens": 157935871.0,
+      "step": 67600
+    },
+    {
+      "epoch": 0.6198771431190978,
+      "learning_rate": 7.602640506097002e-05,
+      "loss": 0.7288,
+      "mean_token_accuracy": 0.7738948345184327,
+      "num_tokens": 157962152.0,
+      "step": 67610
+    },
+    {
+      "epoch": 0.6199688273585771,
+      "learning_rate": 7.600806821307418e-05,
+      "loss": 0.7182,
+      "mean_token_accuracy": 0.7898809969425201,
+      "num_tokens": 157987051.0,
+      "step": 67620
+    },
+    {
+      "epoch": 0.6200605115980563,
+      "learning_rate": 7.598973136517833e-05,
+      "loss": 0.7003,
+      "mean_token_accuracy": 0.7832739412784576,
+      "num_tokens": 158012377.0,
+      "step": 67630
+    },
+    {
+      "epoch": 0.6201521958375356,
+      "learning_rate": 7.597139451728248e-05,
+      "loss": 0.6528,
+      "mean_token_accuracy": 0.7969177067279816,
+      "num_tokens": 158036551.0,
+      "step": 67640
+    },
+    {
+      "epoch": 0.6202438800770148,
+      "learning_rate": 7.595305766938664e-05,
+      "loss": 0.6872,
+      "mean_token_accuracy": 0.7888841569423676,
+      "num_tokens": 158061939.0,
+      "step": 67650
+    },
+    {
+      "epoch": 0.620335564316494,
+      "learning_rate": 7.593472082149079e-05,
+      "loss": 0.7137,
+      "mean_token_accuracy": 0.7862693428993225,
+      "num_tokens": 158087081.0,
+      "step": 67660
+    },
+    {
+      "epoch": 0.6204272485559732,
+      "learning_rate": 7.591638397359494e-05,
+      "loss": 0.6574,
+      "mean_token_accuracy": 0.79624143242836,
+      "num_tokens": 158112517.0,
+      "step": 67670
+    },
+    {
+      "epoch": 0.6205189327954524,
+      "learning_rate": 7.58980471256991e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.7950810492038727,
+      "num_tokens": 158138209.0,
+      "step": 67680
+    },
+    {
+      "epoch": 0.6206106170349317,
+      "learning_rate": 7.587971027780324e-05,
+      "loss": 0.7043,
+      "mean_token_accuracy": 0.7870907127857208,
+      "num_tokens": 158163371.0,
+      "step": 67690
+    },
+    {
+      "epoch": 0.620702301274411,
+      "learning_rate": 7.58613734299074e-05,
+      "loss": 0.7176,
+      "mean_token_accuracy": 0.7789473712444306,
+      "num_tokens": 158189126.0,
+      "step": 67700
+    },
+    {
+      "epoch": 0.6207939855138902,
+      "learning_rate": 7.584303658201155e-05,
+      "loss": 0.7079,
+      "mean_token_accuracy": 0.7856812477111816,
+      "num_tokens": 158213851.0,
+      "step": 67710
+    },
+    {
+      "epoch": 0.6208856697533695,
+      "learning_rate": 7.582469973411571e-05,
+      "loss": 0.6688,
+      "mean_token_accuracy": 0.7929436683654785,
+      "num_tokens": 158238262.0,
+      "step": 67720
+    },
+    {
+      "epoch": 0.6209773539928486,
+      "learning_rate": 7.580636288621987e-05,
+      "loss": 0.6784,
+      "mean_token_accuracy": 0.793372493982315,
+      "num_tokens": 158263671.0,
+      "step": 67730
+    },
+    {
+      "epoch": 0.6210690382323278,
+      "learning_rate": 7.578802603832402e-05,
+      "loss": 0.6735,
+      "mean_token_accuracy": 0.7910674631595611,
+      "num_tokens": 158289257.0,
+      "step": 67740
+    },
+    {
+      "epoch": 0.6211607224718071,
+      "learning_rate": 7.576968919042817e-05,
+      "loss": 0.6598,
+      "mean_token_accuracy": 0.7980355381965637,
+      "num_tokens": 158314173.0,
+      "step": 67750
+    },
+    {
+      "epoch": 0.6212524067112863,
+      "learning_rate": 7.575135234253232e-05,
+      "loss": 0.7411,
+      "mean_token_accuracy": 0.7710571587085724,
+      "num_tokens": 158339097.0,
+      "step": 67760
+    },
+    {
+      "epoch": 0.6213440909507656,
+      "learning_rate": 7.573301549463647e-05,
+      "loss": 0.6912,
+      "mean_token_accuracy": 0.7877136707305908,
+      "num_tokens": 158364302.0,
+      "step": 67770
+    },
+    {
+      "epoch": 0.6214357751902448,
+      "learning_rate": 7.571467864674064e-05,
+      "loss": 0.6822,
+      "mean_token_accuracy": 0.784754729270935,
+      "num_tokens": 158389615.0,
+      "step": 67780
+    },
+    {
+      "epoch": 0.621527459429724,
+      "learning_rate": 7.569634179884478e-05,
+      "loss": 0.6901,
+      "mean_token_accuracy": 0.7908022820949554,
+      "num_tokens": 158414937.0,
+      "step": 67790
+    },
+    {
+      "epoch": 0.6216191436692032,
+      "learning_rate": 7.567800495094893e-05,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.7832824528217316,
+      "num_tokens": 158440457.0,
+      "step": 67800
+    },
+    {
+      "epoch": 0.6217108279086825,
+      "learning_rate": 7.56596681030531e-05,
+      "loss": 0.7132,
+      "mean_token_accuracy": 0.7835166871547699,
+      "num_tokens": 158465443.0,
+      "step": 67810
+    },
+    {
+      "epoch": 0.6218025121481617,
+      "learning_rate": 7.564133125515725e-05,
+      "loss": 0.7072,
+      "mean_token_accuracy": 0.7831239640712738,
+      "num_tokens": 158490287.0,
+      "step": 67820
+    },
+    {
+      "epoch": 0.621894196387641,
+      "learning_rate": 7.56229944072614e-05,
+      "loss": 0.7148,
+      "mean_token_accuracy": 0.7799515902996064,
+      "num_tokens": 158515253.0,
+      "step": 67830
+    },
+    {
+      "epoch": 0.6219858806271202,
+      "learning_rate": 7.560465755936555e-05,
+      "loss": 0.6924,
+      "mean_token_accuracy": 0.7871422111988068,
+      "num_tokens": 158540757.0,
+      "step": 67840
+    },
+    {
+      "epoch": 0.6220775648665995,
+      "learning_rate": 7.55863207114697e-05,
+      "loss": 0.7205,
+      "mean_token_accuracy": 0.7767223000526429,
+      "num_tokens": 158565672.0,
+      "step": 67850
+    },
+    {
+      "epoch": 0.6221692491060786,
+      "learning_rate": 7.556798386357385e-05,
+      "loss": 0.657,
+      "mean_token_accuracy": 0.7968750596046448,
+      "num_tokens": 158590417.0,
+      "step": 67860
+    },
+    {
+      "epoch": 0.6222609333455579,
+      "learning_rate": 7.5549647015678e-05,
+      "loss": 0.6996,
+      "mean_token_accuracy": 0.7919406414031982,
+      "num_tokens": 158616112.0,
+      "step": 67870
+    },
+    {
+      "epoch": 0.6223526175850371,
+      "learning_rate": 7.553131016778217e-05,
+      "loss": 0.6979,
+      "mean_token_accuracy": 0.7862891614437103,
+      "num_tokens": 158641112.0,
+      "step": 67880
+    },
+    {
+      "epoch": 0.6224443018245164,
+      "learning_rate": 7.551297331988631e-05,
+      "loss": 0.6621,
+      "mean_token_accuracy": 0.7982531368732453,
+      "num_tokens": 158666558.0,
+      "step": 67890
+    },
+    {
+      "epoch": 0.6225359860639956,
+      "learning_rate": 7.549463647199046e-05,
+      "loss": 0.7145,
+      "mean_token_accuracy": 0.7796647191047669,
+      "num_tokens": 158691011.0,
+      "step": 67900
+    },
+    {
+      "epoch": 0.6226276703034749,
+      "learning_rate": 7.547629962409463e-05,
+      "loss": 0.6755,
+      "mean_token_accuracy": 0.7932925701141358,
+      "num_tokens": 158716251.0,
+      "step": 67910
+    },
+    {
+      "epoch": 0.622719354542954,
+      "learning_rate": 7.545796277619878e-05,
+      "loss": 0.6513,
+      "mean_token_accuracy": 0.7988629639148712,
+      "num_tokens": 158740514.0,
+      "step": 67920
+    },
+    {
+      "epoch": 0.6228110387824333,
+      "learning_rate": 7.543962592830293e-05,
+      "loss": 0.6702,
+      "mean_token_accuracy": 0.7902235269546509,
+      "num_tokens": 158765137.0,
+      "step": 67930
+    },
+    {
+      "epoch": 0.6229027230219125,
+      "learning_rate": 7.542128908040708e-05,
+      "loss": 0.7062,
+      "mean_token_accuracy": 0.7847516477108002,
+      "num_tokens": 158789701.0,
+      "step": 67940
+    },
+    {
+      "epoch": 0.6229944072613918,
+      "learning_rate": 7.540295223251123e-05,
+      "loss": 0.6742,
+      "mean_token_accuracy": 0.7943823873996735,
+      "num_tokens": 158814902.0,
+      "step": 67950
+    },
+    {
+      "epoch": 0.623086091500871,
+      "learning_rate": 7.538461538461539e-05,
+      "loss": 0.7007,
+      "mean_token_accuracy": 0.7827964663505554,
+      "num_tokens": 158839992.0,
+      "step": 67960
+    },
+    {
+      "epoch": 0.6231777757403503,
+      "learning_rate": 7.536627853671954e-05,
+      "loss": 0.6837,
+      "mean_token_accuracy": 0.7875102579593658,
+      "num_tokens": 158864454.0,
+      "step": 67970
+    },
+    {
+      "epoch": 0.6232694599798295,
+      "learning_rate": 7.53479416888237e-05,
+      "loss": 0.6534,
+      "mean_token_accuracy": 0.7980010867118835,
+      "num_tokens": 158889217.0,
+      "step": 67980
+    },
+    {
+      "epoch": 0.6233611442193087,
+      "learning_rate": 7.532960484092784e-05,
+      "loss": 0.685,
+      "mean_token_accuracy": 0.7899937689304352,
+      "num_tokens": 158913924.0,
+      "step": 67990
+    },
+    {
+      "epoch": 0.6234528284587879,
+      "learning_rate": 7.531126799303199e-05,
+      "loss": 0.6946,
+      "mean_token_accuracy": 0.78940549492836,
+      "num_tokens": 158939065.0,
+      "step": 68000
+    },
+    {
+      "epoch": 0.6235445126982672,
+      "learning_rate": 7.529293114513616e-05,
+      "loss": 0.6984,
+      "mean_token_accuracy": 0.7817919611930847,
+      "num_tokens": 158965309.0,
+      "step": 68010
+    },
+    {
+      "epoch": 0.6236361969377464,
+      "learning_rate": 7.527459429724031e-05,
+      "loss": 0.6909,
+      "mean_token_accuracy": 0.781625235080719,
+      "num_tokens": 158990683.0,
+      "step": 68020
+    },
+    {
+      "epoch": 0.6237278811772257,
+      "learning_rate": 7.525625744934446e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.7947934448719025,
+      "num_tokens": 159016280.0,
+      "step": 68030
+    },
+    {
+      "epoch": 0.6238195654167049,
+      "learning_rate": 7.523792060144861e-05,
+      "loss": 0.6937,
+      "mean_token_accuracy": 0.7849284529685974,
+      "num_tokens": 159041934.0,
+      "step": 68040
+    },
+    {
+      "epoch": 0.623911249656184,
+      "learning_rate": 7.521958375355277e-05,
+      "loss": 0.6812,
+      "mean_token_accuracy": 0.789916741847992,
+      "num_tokens": 159067684.0,
+      "step": 68050
+    },
+    {
+      "epoch": 0.6240029338956633,
+      "learning_rate": 7.520124690565692e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.7881591260433197,
+      "num_tokens": 159093100.0,
+      "step": 68060
+    },
+    {
+      "epoch": 0.6240946181351426,
+      "learning_rate": 7.518291005776107e-05,
+      "loss": 0.7146,
+      "mean_token_accuracy": 0.7809472620487213,
+      "num_tokens": 159118988.0,
+      "step": 68070
+    },
+    {
+      "epoch": 0.6241863023746218,
+      "learning_rate": 7.516457320986523e-05,
+      "loss": 0.6568,
+      "mean_token_accuracy": 0.7976008296012879,
+      "num_tokens": 159144479.0,
+      "step": 68080
+    },
+    {
+      "epoch": 0.624277986614101,
+      "learning_rate": 7.514623636196937e-05,
+      "loss": 0.6609,
+      "mean_token_accuracy": 0.7967592477798462,
+      "num_tokens": 159169705.0,
+      "step": 68090
+    },
+    {
+      "epoch": 0.6243696708535803,
+      "learning_rate": 7.512789951407353e-05,
+      "loss": 0.6936,
+      "mean_token_accuracy": 0.7881568074226379,
+      "num_tokens": 159195729.0,
+      "step": 68100
+    },
+    {
+      "epoch": 0.6244613550930596,
+      "learning_rate": 7.510956266617769e-05,
+      "loss": 0.721,
+      "mean_token_accuracy": 0.781345933675766,
+      "num_tokens": 159219812.0,
+      "step": 68110
+    },
+    {
+      "epoch": 0.6245530393325387,
+      "learning_rate": 7.509122581828184e-05,
+      "loss": 0.6658,
+      "mean_token_accuracy": 0.7950010359287262,
+      "num_tokens": 159244728.0,
+      "step": 68120
+    },
+    {
+      "epoch": 0.624644723572018,
+      "learning_rate": 7.5072888970386e-05,
+      "loss": 0.663,
+      "mean_token_accuracy": 0.7935277819633484,
+      "num_tokens": 159271032.0,
+      "step": 68130
+    },
+    {
+      "epoch": 0.6247364078114972,
+      "learning_rate": 7.505455212249015e-05,
+      "loss": 0.6694,
+      "mean_token_accuracy": 0.7925786733627319,
+      "num_tokens": 159295563.0,
+      "step": 68140
+    },
+    {
+      "epoch": 0.6248280920509764,
+      "learning_rate": 7.50362152745943e-05,
+      "loss": 0.6793,
+      "mean_token_accuracy": 0.7868499159812927,
+      "num_tokens": 159320547.0,
+      "step": 68150
+    },
+    {
+      "epoch": 0.6249197762904557,
+      "learning_rate": 7.501787842669845e-05,
+      "loss": 0.6916,
+      "mean_token_accuracy": 0.7876942098140717,
+      "num_tokens": 159345078.0,
+      "step": 68160
+    },
+    {
+      "epoch": 0.625011460529935,
+      "learning_rate": 7.499954157880262e-05,
+      "loss": 0.7137,
+      "mean_token_accuracy": 0.787305474281311,
+      "num_tokens": 159370032.0,
+      "step": 68170
+    },
+    {
+      "epoch": 0.6251031447694141,
+      "learning_rate": 7.498120473090677e-05,
+      "loss": 0.7139,
+      "mean_token_accuracy": 0.7832253873348236,
+      "num_tokens": 159395624.0,
+      "step": 68180
+    },
+    {
+      "epoch": 0.6251948290088933,
+      "learning_rate": 7.49628678830109e-05,
+      "loss": 0.6923,
+      "mean_token_accuracy": 0.7877730906009675,
+      "num_tokens": 159421516.0,
+      "step": 68190
+    },
+    {
+      "epoch": 0.6252865132483726,
+      "learning_rate": 7.494453103511506e-05,
+      "loss": 0.7026,
+      "mean_token_accuracy": 0.7831994831562042,
+      "num_tokens": 159446982.0,
+      "step": 68200
+    },
+    {
+      "epoch": 0.6253781974878518,
+      "learning_rate": 7.492619418721922e-05,
+      "loss": 0.6846,
+      "mean_token_accuracy": 0.792252266407013,
+      "num_tokens": 159472086.0,
+      "step": 68210
+    },
+    {
+      "epoch": 0.6254698817273311,
+      "learning_rate": 7.490785733932337e-05,
+      "loss": 0.6524,
+      "mean_token_accuracy": 0.7992007553577423,
+      "num_tokens": 159497352.0,
+      "step": 68220
+    },
+    {
+      "epoch": 0.6255615659668103,
+      "learning_rate": 7.488952049142753e-05,
+      "loss": 0.6822,
+      "mean_token_accuracy": 0.7904492437839508,
+      "num_tokens": 159522219.0,
+      "step": 68230
+    },
+    {
+      "epoch": 0.6256532502062896,
+      "learning_rate": 7.487118364353168e-05,
+      "loss": 0.7184,
+      "mean_token_accuracy": 0.7877139449119568,
+      "num_tokens": 159547681.0,
+      "step": 68240
+    },
+    {
+      "epoch": 0.6257449344457687,
+      "learning_rate": 7.485284679563583e-05,
+      "loss": 0.7304,
+      "mean_token_accuracy": 0.7756365776062012,
+      "num_tokens": 159574205.0,
+      "step": 68250
+    },
+    {
+      "epoch": 0.625836618685248,
+      "learning_rate": 7.483450994773998e-05,
+      "loss": 0.7168,
+      "mean_token_accuracy": 0.7812732100486756,
+      "num_tokens": 159599703.0,
+      "step": 68260
+    },
+    {
+      "epoch": 0.6259283029247272,
+      "learning_rate": 7.481617309984415e-05,
+      "loss": 0.6825,
+      "mean_token_accuracy": 0.7851893424987793,
+      "num_tokens": 159625739.0,
+      "step": 68270
+    },
+    {
+      "epoch": 0.6260199871642065,
+      "learning_rate": 7.47978362519483e-05,
+      "loss": 0.7585,
+      "mean_token_accuracy": 0.7742206215858459,
+      "num_tokens": 159651381.0,
+      "step": 68280
+    },
+    {
+      "epoch": 0.6261116714036857,
+      "learning_rate": 7.477949940405244e-05,
+      "loss": 0.6818,
+      "mean_token_accuracy": 0.7892438113689423,
+      "num_tokens": 159677000.0,
+      "step": 68290
+    },
+    {
+      "epoch": 0.626203355643165,
+      "learning_rate": 7.47611625561566e-05,
+      "loss": 0.6823,
+      "mean_token_accuracy": 0.7935108661651611,
+      "num_tokens": 159702447.0,
+      "step": 68300
+    },
+    {
+      "epoch": 0.6262950398826441,
+      "learning_rate": 7.474282570826075e-05,
+      "loss": 0.6853,
+      "mean_token_accuracy": 0.7863902390003205,
+      "num_tokens": 159727576.0,
+      "step": 68310
+    },
+    {
+      "epoch": 0.6263867241221234,
+      "learning_rate": 7.47244888603649e-05,
+      "loss": 0.6942,
+      "mean_token_accuracy": 0.7879535675048828,
+      "num_tokens": 159752600.0,
+      "step": 68320
+    },
+    {
+      "epoch": 0.6264784083616026,
+      "learning_rate": 7.470615201246906e-05,
+      "loss": 0.6898,
+      "mean_token_accuracy": 0.7876635611057281,
+      "num_tokens": 159777417.0,
+      "step": 68330
+    },
+    {
+      "epoch": 0.6265700926010819,
+      "learning_rate": 7.468781516457321e-05,
+      "loss": 0.6535,
+      "mean_token_accuracy": 0.7975645959377289,
+      "num_tokens": 159802098.0,
+      "step": 68340
+    },
+    {
+      "epoch": 0.6266617768405611,
+      "learning_rate": 7.466947831667736e-05,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.7913213193416595,
+      "num_tokens": 159827170.0,
+      "step": 68350
+    },
+    {
+      "epoch": 0.6267534610800404,
+      "learning_rate": 7.465114146878151e-05,
+      "loss": 0.6605,
+      "mean_token_accuracy": 0.7892809808254242,
+      "num_tokens": 159852126.0,
+      "step": 68360
+    },
+    {
+      "epoch": 0.6268451453195196,
+      "learning_rate": 7.463280462088568e-05,
+      "loss": 0.6893,
+      "mean_token_accuracy": 0.790635347366333,
+      "num_tokens": 159877759.0,
+      "step": 68370
+    },
+    {
+      "epoch": 0.6269368295589988,
+      "learning_rate": 7.461446777298983e-05,
+      "loss": 0.7017,
+      "mean_token_accuracy": 0.7819789409637451,
+      "num_tokens": 159903098.0,
+      "step": 68380
+    },
+    {
+      "epoch": 0.627028513798478,
+      "learning_rate": 7.459613092509398e-05,
+      "loss": 0.6865,
+      "mean_token_accuracy": 0.785049992799759,
+      "num_tokens": 159928235.0,
+      "step": 68390
+    },
+    {
+      "epoch": 0.6271201980379573,
+      "learning_rate": 7.457779407719814e-05,
+      "loss": 0.6836,
+      "mean_token_accuracy": 0.7889967024326324,
+      "num_tokens": 159953100.0,
+      "step": 68400
+    },
+    {
+      "epoch": 0.6272118822774365,
+      "learning_rate": 7.455945722930229e-05,
+      "loss": 0.6901,
+      "mean_token_accuracy": 0.7829911768436432,
+      "num_tokens": 159977772.0,
+      "step": 68410
+    },
+    {
+      "epoch": 0.6273035665169158,
+      "learning_rate": 7.454112038140644e-05,
+      "loss": 0.691,
+      "mean_token_accuracy": 0.7903009057044983,
+      "num_tokens": 160003315.0,
+      "step": 68420
+    },
+    {
+      "epoch": 0.627395250756395,
+      "learning_rate": 7.45227835335106e-05,
+      "loss": 0.696,
+      "mean_token_accuracy": 0.7848869740962983,
+      "num_tokens": 160028126.0,
+      "step": 68430
+    },
+    {
+      "epoch": 0.6274869349958742,
+      "learning_rate": 7.450444668561474e-05,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.7832297563552857,
+      "num_tokens": 160053885.0,
+      "step": 68440
+    },
+    {
+      "epoch": 0.6275786192353534,
+      "learning_rate": 7.44861098377189e-05,
+      "loss": 0.7098,
+      "mean_token_accuracy": 0.7870866298675537,
+      "num_tokens": 160079105.0,
+      "step": 68450
+    },
+    {
+      "epoch": 0.6276703034748327,
+      "learning_rate": 7.446777298982305e-05,
+      "loss": 0.6725,
+      "mean_token_accuracy": 0.79354248046875,
+      "num_tokens": 160103810.0,
+      "step": 68460
+    },
+    {
+      "epoch": 0.6277619877143119,
+      "learning_rate": 7.444943614192721e-05,
+      "loss": 0.7052,
+      "mean_token_accuracy": 0.7852420270442962,
+      "num_tokens": 160129041.0,
+      "step": 68470
+    },
+    {
+      "epoch": 0.6278536719537912,
+      "learning_rate": 7.443109929403136e-05,
+      "loss": 0.7034,
+      "mean_token_accuracy": 0.7870183765888215,
+      "num_tokens": 160154299.0,
+      "step": 68480
+    },
+    {
+      "epoch": 0.6279453561932704,
+      "learning_rate": 7.441276244613552e-05,
+      "loss": 0.6687,
+      "mean_token_accuracy": 0.792228901386261,
+      "num_tokens": 160180271.0,
+      "step": 68490
+    },
+    {
+      "epoch": 0.6280370404327497,
+      "learning_rate": 7.439442559823967e-05,
+      "loss": 0.7165,
+      "mean_token_accuracy": 0.7853408753871918,
+      "num_tokens": 160205271.0,
+      "step": 68500
+    },
+    {
+      "epoch": 0.6281287246722288,
+      "learning_rate": 7.437608875034382e-05,
+      "loss": 0.6822,
+      "mean_token_accuracy": 0.7885018885135651,
+      "num_tokens": 160230505.0,
+      "step": 68510
+    },
+    {
+      "epoch": 0.628220408911708,
+      "learning_rate": 7.435775190244797e-05,
+      "loss": 0.6792,
+      "mean_token_accuracy": 0.7850334107875824,
+      "num_tokens": 160255950.0,
+      "step": 68520
+    },
+    {
+      "epoch": 0.6283120931511873,
+      "learning_rate": 7.433941505455214e-05,
+      "loss": 0.6909,
+      "mean_token_accuracy": 0.78346386551857,
+      "num_tokens": 160280976.0,
+      "step": 68530
+    },
+    {
+      "epoch": 0.6284037773906666,
+      "learning_rate": 7.432107820665627e-05,
+      "loss": 0.6505,
+      "mean_token_accuracy": 0.7980206906795502,
+      "num_tokens": 160305881.0,
+      "step": 68540
+    },
+    {
+      "epoch": 0.6284954616301458,
+      "learning_rate": 7.430274135876043e-05,
+      "loss": 0.6826,
+      "mean_token_accuracy": 0.7878010094165802,
+      "num_tokens": 160331851.0,
+      "step": 68550
+    },
+    {
+      "epoch": 0.628587145869625,
+      "learning_rate": 7.428440451086459e-05,
+      "loss": 0.7262,
+      "mean_token_accuracy": 0.7825631320476532,
+      "num_tokens": 160356091.0,
+      "step": 68560
+    },
+    {
+      "epoch": 0.6286788301091042,
+      "learning_rate": 7.426606766296874e-05,
+      "loss": 0.6708,
+      "mean_token_accuracy": 0.7878499090671539,
+      "num_tokens": 160381118.0,
+      "step": 68570
+    },
+    {
+      "epoch": 0.6287705143485834,
+      "learning_rate": 7.42477308150729e-05,
+      "loss": 0.6771,
+      "mean_token_accuracy": 0.7930330097675323,
+      "num_tokens": 160406181.0,
+      "step": 68580
+    },
+    {
+      "epoch": 0.6288621985880627,
+      "learning_rate": 7.422939396717705e-05,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.7923729062080384,
+      "num_tokens": 160431267.0,
+      "step": 68590
+    },
+    {
+      "epoch": 0.6289538828275419,
+      "learning_rate": 7.42110571192812e-05,
+      "loss": 0.6805,
+      "mean_token_accuracy": 0.7918398797512054,
+      "num_tokens": 160456798.0,
+      "step": 68600
+    },
+    {
+      "epoch": 0.6290455670670212,
+      "learning_rate": 7.419272027138535e-05,
+      "loss": 0.6748,
+      "mean_token_accuracy": 0.7926307797431946,
+      "num_tokens": 160481610.0,
+      "step": 68610
+    },
+    {
+      "epoch": 0.6291372513065004,
+      "learning_rate": 7.41743834234895e-05,
+      "loss": 0.669,
+      "mean_token_accuracy": 0.7898163974285126,
+      "num_tokens": 160506613.0,
+      "step": 68620
+    },
+    {
+      "epoch": 0.6292289355459797,
+      "learning_rate": 7.415604657559367e-05,
+      "loss": 0.6936,
+      "mean_token_accuracy": 0.7894938290119171,
+      "num_tokens": 160531525.0,
+      "step": 68630
+    },
+    {
+      "epoch": 0.6293206197854588,
+      "learning_rate": 7.413770972769781e-05,
+      "loss": 0.6822,
+      "mean_token_accuracy": 0.7915955781936646,
+      "num_tokens": 160557369.0,
+      "step": 68640
+    },
+    {
+      "epoch": 0.6294123040249381,
+      "learning_rate": 7.411937287980196e-05,
+      "loss": 0.6929,
+      "mean_token_accuracy": 0.7915159463882446,
+      "num_tokens": 160582371.0,
+      "step": 68650
+    },
+    {
+      "epoch": 0.6295039882644173,
+      "learning_rate": 7.410103603190612e-05,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.7909412145614624,
+      "num_tokens": 160607686.0,
+      "step": 68660
+    },
+    {
+      "epoch": 0.6295956725038966,
+      "learning_rate": 7.408269918401028e-05,
+      "loss": 0.6877,
+      "mean_token_accuracy": 0.7856991171836853,
+      "num_tokens": 160633502.0,
+      "step": 68670
+    },
+    {
+      "epoch": 0.6296873567433758,
+      "learning_rate": 7.406436233611443e-05,
+      "loss": 0.7075,
+      "mean_token_accuracy": 0.783650916814804,
+      "num_tokens": 160658327.0,
+      "step": 68680
+    },
+    {
+      "epoch": 0.6297790409828551,
+      "learning_rate": 7.404602548821858e-05,
+      "loss": 0.6849,
+      "mean_token_accuracy": 0.7868337094783783,
+      "num_tokens": 160683193.0,
+      "step": 68690
+    },
+    {
+      "epoch": 0.6298707252223342,
+      "learning_rate": 7.402768864032273e-05,
+      "loss": 0.6806,
+      "mean_token_accuracy": 0.7953231811523438,
+      "num_tokens": 160708083.0,
+      "step": 68700
+    },
+    {
+      "epoch": 0.6299624094618135,
+      "learning_rate": 7.400935179242688e-05,
+      "loss": 0.6698,
+      "mean_token_accuracy": 0.7883341610431671,
+      "num_tokens": 160733797.0,
+      "step": 68710
+    },
+    {
+      "epoch": 0.6300540937012927,
+      "learning_rate": 7.399101494453104e-05,
+      "loss": 0.6896,
+      "mean_token_accuracy": 0.7837846398353576,
+      "num_tokens": 160758505.0,
+      "step": 68720
+    },
+    {
+      "epoch": 0.630145777940772,
+      "learning_rate": 7.39726780966352e-05,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.7963154077529907,
+      "num_tokens": 160783497.0,
+      "step": 68730
+    },
+    {
+      "epoch": 0.6302374621802512,
+      "learning_rate": 7.395434124873934e-05,
+      "loss": 0.6693,
+      "mean_token_accuracy": 0.7931505084037781,
+      "num_tokens": 160808186.0,
+      "step": 68740
+    },
+    {
+      "epoch": 0.6303291464197305,
+      "learning_rate": 7.393600440084349e-05,
+      "loss": 0.6868,
+      "mean_token_accuracy": 0.7925294160842895,
+      "num_tokens": 160833366.0,
+      "step": 68750
+    },
+    {
+      "epoch": 0.6304208306592097,
+      "learning_rate": 7.391766755294766e-05,
+      "loss": 0.6862,
+      "mean_token_accuracy": 0.7858851671218872,
+      "num_tokens": 160858612.0,
+      "step": 68760
+    },
+    {
+      "epoch": 0.6305125148986889,
+      "learning_rate": 7.389933070505181e-05,
+      "loss": 0.692,
+      "mean_token_accuracy": 0.7893398821353912,
+      "num_tokens": 160883137.0,
+      "step": 68770
+    },
+    {
+      "epoch": 0.6306041991381681,
+      "learning_rate": 7.388099385715596e-05,
+      "loss": 0.6627,
+      "mean_token_accuracy": 0.7915483295917511,
+      "num_tokens": 160907773.0,
+      "step": 68780
+    },
+    {
+      "epoch": 0.6306958833776474,
+      "learning_rate": 7.386265700926011e-05,
+      "loss": 0.6896,
+      "mean_token_accuracy": 0.7855119585990906,
+      "num_tokens": 160932706.0,
+      "step": 68790
+    },
+    {
+      "epoch": 0.6307875676171266,
+      "learning_rate": 7.384432016136426e-05,
+      "loss": 0.6736,
+      "mean_token_accuracy": 0.789964210987091,
+      "num_tokens": 160958373.0,
+      "step": 68800
+    },
+    {
+      "epoch": 0.6308792518566059,
+      "learning_rate": 7.382598331346842e-05,
+      "loss": 0.7153,
+      "mean_token_accuracy": 0.7786109328269959,
+      "num_tokens": 160983703.0,
+      "step": 68810
+    },
+    {
+      "epoch": 0.6309709360960851,
+      "learning_rate": 7.380764646557257e-05,
+      "loss": 0.679,
+      "mean_token_accuracy": 0.7911896884441376,
+      "num_tokens": 161008482.0,
+      "step": 68820
+    },
+    {
+      "epoch": 0.6310626203355643,
+      "learning_rate": 7.378930961767673e-05,
+      "loss": 0.7013,
+      "mean_token_accuracy": 0.7846454560756684,
+      "num_tokens": 161034256.0,
+      "step": 68830
+    },
+    {
+      "epoch": 0.6311543045750435,
+      "learning_rate": 7.377097276978087e-05,
+      "loss": 0.6884,
+      "mean_token_accuracy": 0.78794926404953,
+      "num_tokens": 161059363.0,
+      "step": 68840
+    },
+    {
+      "epoch": 0.6312459888145228,
+      "learning_rate": 7.375263592188502e-05,
+      "loss": 0.6919,
+      "mean_token_accuracy": 0.7892571866512299,
+      "num_tokens": 161085394.0,
+      "step": 68850
+    },
+    {
+      "epoch": 0.631337673054002,
+      "learning_rate": 7.373429907398919e-05,
+      "loss": 0.7015,
+      "mean_token_accuracy": 0.7867662668228149,
+      "num_tokens": 161111050.0,
+      "step": 68860
+    },
+    {
+      "epoch": 0.6314293572934813,
+      "learning_rate": 7.371596222609334e-05,
+      "loss": 0.6986,
+      "mean_token_accuracy": 0.7836301028728485,
+      "num_tokens": 161136643.0,
+      "step": 68870
+    },
+    {
+      "epoch": 0.6315210415329605,
+      "learning_rate": 7.369762537819749e-05,
+      "loss": 0.6755,
+      "mean_token_accuracy": 0.7927794218063354,
+      "num_tokens": 161161751.0,
+      "step": 68880
+    },
+    {
+      "epoch": 0.6316127257724398,
+      "learning_rate": 7.367928853030164e-05,
+      "loss": 0.6777,
+      "mean_token_accuracy": 0.7892790138721466,
+      "num_tokens": 161186799.0,
+      "step": 68890
+    },
+    {
+      "epoch": 0.6317044100119189,
+      "learning_rate": 7.36609516824058e-05,
+      "loss": 0.7165,
+      "mean_token_accuracy": 0.7814065635204315,
+      "num_tokens": 161211957.0,
+      "step": 68900
+    },
+    {
+      "epoch": 0.6317960942513982,
+      "learning_rate": 7.364261483450995e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.7912361443042755,
+      "num_tokens": 161237330.0,
+      "step": 68910
+    },
+    {
+      "epoch": 0.6318877784908774,
+      "learning_rate": 7.362427798661411e-05,
+      "loss": 0.6828,
+      "mean_token_accuracy": 0.7898914694786072,
+      "num_tokens": 161262585.0,
+      "step": 68920
+    },
+    {
+      "epoch": 0.6319794627303567,
+      "learning_rate": 7.360594113871827e-05,
+      "loss": 0.707,
+      "mean_token_accuracy": 0.7798745572566986,
+      "num_tokens": 161287252.0,
+      "step": 68930
+    },
+    {
+      "epoch": 0.6320711469698359,
+      "learning_rate": 7.35876042908224e-05,
+      "loss": 0.7061,
+      "mean_token_accuracy": 0.783891624212265,
+      "num_tokens": 161312592.0,
+      "step": 68940
+    },
+    {
+      "epoch": 0.6321628312093152,
+      "learning_rate": 7.356926744292656e-05,
+      "loss": 0.6751,
+      "mean_token_accuracy": 0.7935016095638275,
+      "num_tokens": 161338525.0,
+      "step": 68950
+    },
+    {
+      "epoch": 0.6322545154487944,
+      "learning_rate": 7.355093059503072e-05,
+      "loss": 0.6708,
+      "mean_token_accuracy": 0.7926486909389496,
+      "num_tokens": 161364118.0,
+      "step": 68960
+    },
+    {
+      "epoch": 0.6323461996882735,
+      "learning_rate": 7.353259374713487e-05,
+      "loss": 0.6722,
+      "mean_token_accuracy": 0.7905098676681519,
+      "num_tokens": 161388450.0,
+      "step": 68970
+    },
+    {
+      "epoch": 0.6324378839277528,
+      "learning_rate": 7.351425689923902e-05,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.7835709631443024,
+      "num_tokens": 161412983.0,
+      "step": 68980
+    },
+    {
+      "epoch": 0.632529568167232,
+      "learning_rate": 7.349592005134318e-05,
+      "loss": 0.6919,
+      "mean_token_accuracy": 0.7910090208053588,
+      "num_tokens": 161437942.0,
+      "step": 68990
+    },
+    {
+      "epoch": 0.6326212524067113,
+      "learning_rate": 7.347758320344733e-05,
+      "loss": 0.6809,
+      "mean_token_accuracy": 0.7874048054218292,
+      "num_tokens": 161463038.0,
+      "step": 69000
+    },
+    {
+      "epoch": 0.6327129366461905,
+      "learning_rate": 7.345924635555148e-05,
+      "loss": 0.708,
+      "mean_token_accuracy": 0.7828868746757507,
+      "num_tokens": 161488773.0,
+      "step": 69010
+    },
+    {
+      "epoch": 0.6328046208856698,
+      "learning_rate": 7.344090950765565e-05,
+      "loss": 0.7318,
+      "mean_token_accuracy": 0.7716870784759522,
+      "num_tokens": 161513785.0,
+      "step": 69020
+    },
+    {
+      "epoch": 0.6328963051251489,
+      "learning_rate": 7.34225726597598e-05,
+      "loss": 0.6856,
+      "mean_token_accuracy": 0.7896718978881836,
+      "num_tokens": 161538439.0,
+      "step": 69030
+    },
+    {
+      "epoch": 0.6329879893646282,
+      "learning_rate": 7.340423581186394e-05,
+      "loss": 0.6831,
+      "mean_token_accuracy": 0.7908432126045227,
+      "num_tokens": 161562950.0,
+      "step": 69040
+    },
+    {
+      "epoch": 0.6330796736041074,
+      "learning_rate": 7.33858989639681e-05,
+      "loss": 0.6893,
+      "mean_token_accuracy": 0.7905107736587524,
+      "num_tokens": 161588169.0,
+      "step": 69050
+    },
+    {
+      "epoch": 0.6331713578435867,
+      "learning_rate": 7.336756211607225e-05,
+      "loss": 0.7245,
+      "mean_token_accuracy": 0.7878206551074982,
+      "num_tokens": 161613299.0,
+      "step": 69060
+    },
+    {
+      "epoch": 0.6332630420830659,
+      "learning_rate": 7.33492252681764e-05,
+      "loss": 0.717,
+      "mean_token_accuracy": 0.7835732758045196,
+      "num_tokens": 161638649.0,
+      "step": 69070
+    },
+    {
+      "epoch": 0.6333547263225452,
+      "learning_rate": 7.333088842028056e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.78575519323349,
+      "num_tokens": 161663574.0,
+      "step": 69080
+    },
+    {
+      "epoch": 0.6334464105620244,
+      "learning_rate": 7.331255157238471e-05,
+      "loss": 0.6744,
+      "mean_token_accuracy": 0.7893391013145447,
+      "num_tokens": 161689046.0,
+      "step": 69090
+    },
+    {
+      "epoch": 0.6335380948015036,
+      "learning_rate": 7.329421472448886e-05,
+      "loss": 0.6397,
+      "mean_token_accuracy": 0.8034017860889435,
+      "num_tokens": 161714928.0,
+      "step": 69100
+    },
+    {
+      "epoch": 0.6336297790409828,
+      "learning_rate": 7.327587787659301e-05,
+      "loss": 0.7134,
+      "mean_token_accuracy": 0.7835344433784485,
+      "num_tokens": 161740248.0,
+      "step": 69110
+    },
+    {
+      "epoch": 0.6337214632804621,
+      "learning_rate": 7.325754102869718e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.7872528672218323,
+      "num_tokens": 161765673.0,
+      "step": 69120
+    },
+    {
+      "epoch": 0.6338131475199413,
+      "learning_rate": 7.323920418080133e-05,
+      "loss": 0.6927,
+      "mean_token_accuracy": 0.7904843330383301,
+      "num_tokens": 161790754.0,
+      "step": 69130
+    },
+    {
+      "epoch": 0.6339048317594206,
+      "learning_rate": 7.322086733290547e-05,
+      "loss": 0.6893,
+      "mean_token_accuracy": 0.7881133913993835,
+      "num_tokens": 161815418.0,
+      "step": 69140
+    },
+    {
+      "epoch": 0.6339965159988998,
+      "learning_rate": 7.320253048500963e-05,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.7951958656311036,
+      "num_tokens": 161840543.0,
+      "step": 69150
+    },
+    {
+      "epoch": 0.634088200238379,
+      "learning_rate": 7.318419363711379e-05,
+      "loss": 0.698,
+      "mean_token_accuracy": 0.7868954122066498,
+      "num_tokens": 161865723.0,
+      "step": 69160
+    },
+    {
+      "epoch": 0.6341798844778582,
+      "learning_rate": 7.316585678921794e-05,
+      "loss": 0.6851,
+      "mean_token_accuracy": 0.784591406583786,
+      "num_tokens": 161891569.0,
+      "step": 69170
+    },
+    {
+      "epoch": 0.6342715687173375,
+      "learning_rate": 7.314751994132209e-05,
+      "loss": 0.701,
+      "mean_token_accuracy": 0.7829393804073334,
+      "num_tokens": 161916715.0,
+      "step": 69180
+    },
+    {
+      "epoch": 0.6343632529568167,
+      "learning_rate": 7.312918309342624e-05,
+      "loss": 0.6672,
+      "mean_token_accuracy": 0.7902737379074096,
+      "num_tokens": 161942283.0,
+      "step": 69190
+    },
+    {
+      "epoch": 0.634454937196296,
+      "learning_rate": 7.311084624553039e-05,
+      "loss": 0.686,
+      "mean_token_accuracy": 0.7868517696857452,
+      "num_tokens": 161967333.0,
+      "step": 69200
+    },
+    {
+      "epoch": 0.6345466214357752,
+      "learning_rate": 7.309250939763454e-05,
+      "loss": 0.6915,
+      "mean_token_accuracy": 0.7937543034553528,
+      "num_tokens": 161991895.0,
+      "step": 69210
+    },
+    {
+      "epoch": 0.6346383056752545,
+      "learning_rate": 7.307417254973871e-05,
+      "loss": 0.7008,
+      "mean_token_accuracy": 0.7821727871894837,
+      "num_tokens": 162016599.0,
+      "step": 69220
+    },
+    {
+      "epoch": 0.6347299899147336,
+      "learning_rate": 7.305583570184286e-05,
+      "loss": 0.7003,
+      "mean_token_accuracy": 0.7775799095630646,
+      "num_tokens": 162040683.0,
+      "step": 69230
+    },
+    {
+      "epoch": 0.6348216741542129,
+      "learning_rate": 7.3037498853947e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.7876672208309173,
+      "num_tokens": 162065294.0,
+      "step": 69240
+    },
+    {
+      "epoch": 0.6349133583936921,
+      "learning_rate": 7.301916200605117e-05,
+      "loss": 0.705,
+      "mean_token_accuracy": 0.7785340487957001,
+      "num_tokens": 162091109.0,
+      "step": 69250
+    },
+    {
+      "epoch": 0.6350050426331714,
+      "learning_rate": 7.300082515815532e-05,
+      "loss": 0.689,
+      "mean_token_accuracy": 0.7846271574497223,
+      "num_tokens": 162116535.0,
+      "step": 69260
+    },
+    {
+      "epoch": 0.6350967268726506,
+      "learning_rate": 7.298248831025947e-05,
+      "loss": 0.7037,
+      "mean_token_accuracy": 0.7866622090339661,
+      "num_tokens": 162141796.0,
+      "step": 69270
+    },
+    {
+      "epoch": 0.6351884111121299,
+      "learning_rate": 7.296415146236362e-05,
+      "loss": 0.6917,
+      "mean_token_accuracy": 0.7887984454631806,
+      "num_tokens": 162166980.0,
+      "step": 69280
+    },
+    {
+      "epoch": 0.635280095351609,
+      "learning_rate": 7.294581461446777e-05,
+      "loss": 0.7387,
+      "mean_token_accuracy": 0.7799351513385773,
+      "num_tokens": 162191990.0,
+      "step": 69290
+    },
+    {
+      "epoch": 0.6353717795910883,
+      "learning_rate": 7.292747776657193e-05,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.7883966267108917,
+      "num_tokens": 162216548.0,
+      "step": 69300
+    },
+    {
+      "epoch": 0.6354634638305675,
+      "learning_rate": 7.290914091867608e-05,
+      "loss": 0.6486,
+      "mean_token_accuracy": 0.792574554681778,
+      "num_tokens": 162241150.0,
+      "step": 69310
+    },
+    {
+      "epoch": 0.6355551480700468,
+      "learning_rate": 7.289080407078024e-05,
+      "loss": 0.7235,
+      "mean_token_accuracy": 0.7742181360721588,
+      "num_tokens": 162265891.0,
+      "step": 69320
+    },
+    {
+      "epoch": 0.635646832309526,
+      "learning_rate": 7.28724672228844e-05,
+      "loss": 0.6605,
+      "mean_token_accuracy": 0.7921281635761261,
+      "num_tokens": 162291077.0,
+      "step": 69330
+    },
+    {
+      "epoch": 0.6357385165490053,
+      "learning_rate": 7.285413037498853e-05,
+      "loss": 0.6997,
+      "mean_token_accuracy": 0.7839576721191406,
+      "num_tokens": 162316990.0,
+      "step": 69340
+    },
+    {
+      "epoch": 0.6358302007884845,
+      "learning_rate": 7.28357935270927e-05,
+      "loss": 0.7061,
+      "mean_token_accuracy": 0.783429092168808,
+      "num_tokens": 162342798.0,
+      "step": 69350
+    },
+    {
+      "epoch": 0.6359218850279637,
+      "learning_rate": 7.281745667919685e-05,
+      "loss": 0.7112,
+      "mean_token_accuracy": 0.78250093460083,
+      "num_tokens": 162368283.0,
+      "step": 69360
+    },
+    {
+      "epoch": 0.6360135692674429,
+      "learning_rate": 7.2799119831301e-05,
+      "loss": 0.6877,
+      "mean_token_accuracy": 0.7862596631050109,
+      "num_tokens": 162394007.0,
+      "step": 69370
+    },
+    {
+      "epoch": 0.6361052535069222,
+      "learning_rate": 7.278078298340515e-05,
+      "loss": 0.6973,
+      "mean_token_accuracy": 0.7891257047653198,
+      "num_tokens": 162419181.0,
+      "step": 69380
+    },
+    {
+      "epoch": 0.6361969377464014,
+      "learning_rate": 7.27624461355093e-05,
+      "loss": 0.7086,
+      "mean_token_accuracy": 0.781864482164383,
+      "num_tokens": 162443638.0,
+      "step": 69390
+    },
+    {
+      "epoch": 0.6362886219858807,
+      "learning_rate": 7.274410928761346e-05,
+      "loss": 0.7209,
+      "mean_token_accuracy": 0.7780154645442963,
+      "num_tokens": 162468992.0,
+      "step": 69400
+    },
+    {
+      "epoch": 0.6363803062253599,
+      "learning_rate": 7.272577243971762e-05,
+      "loss": 0.6943,
+      "mean_token_accuracy": 0.7876321792602539,
+      "num_tokens": 162494192.0,
+      "step": 69410
+    },
+    {
+      "epoch": 0.636471990464839,
+      "learning_rate": 7.270743559182177e-05,
+      "loss": 0.7028,
+      "mean_token_accuracy": 0.7872909486293793,
+      "num_tokens": 162519425.0,
+      "step": 69420
+    },
+    {
+      "epoch": 0.6365636747043183,
+      "learning_rate": 7.268909874392593e-05,
+      "loss": 0.6848,
+      "mean_token_accuracy": 0.7924340188503265,
+      "num_tokens": 162544628.0,
+      "step": 69430
+    },
+    {
+      "epoch": 0.6366553589437975,
+      "learning_rate": 7.267076189603006e-05,
+      "loss": 0.6849,
+      "mean_token_accuracy": 0.7920091986656189,
+      "num_tokens": 162568395.0,
+      "step": 69440
+    },
+    {
+      "epoch": 0.6367470431832768,
+      "learning_rate": 7.265242504813423e-05,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.7834534525871277,
+      "num_tokens": 162593796.0,
+      "step": 69450
+    },
+    {
+      "epoch": 0.636838727422756,
+      "learning_rate": 7.263408820023838e-05,
+      "loss": 0.6645,
+      "mean_token_accuracy": 0.797721940279007,
+      "num_tokens": 162619053.0,
+      "step": 69460
+    },
+    {
+      "epoch": 0.6369304116622353,
+      "learning_rate": 7.261575135234253e-05,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.7965146541595459,
+      "num_tokens": 162644492.0,
+      "step": 69470
+    },
+    {
+      "epoch": 0.6370220959017145,
+      "learning_rate": 7.259741450444669e-05,
+      "loss": 0.6738,
+      "mean_token_accuracy": 0.7905985891819001,
+      "num_tokens": 162669467.0,
+      "step": 69480
+    },
+    {
+      "epoch": 0.6371137801411937,
+      "learning_rate": 7.257907765655084e-05,
+      "loss": 0.6799,
+      "mean_token_accuracy": 0.7942047357559204,
+      "num_tokens": 162695123.0,
+      "step": 69490
+    },
+    {
+      "epoch": 0.6372054643806729,
+      "learning_rate": 7.256074080865499e-05,
+      "loss": 0.6727,
+      "mean_token_accuracy": 0.7896474719047546,
+      "num_tokens": 162720720.0,
+      "step": 69500
+    },
+    {
+      "epoch": 0.6372971486201522,
+      "learning_rate": 7.254240396075916e-05,
+      "loss": 0.6833,
+      "mean_token_accuracy": 0.7888857007026673,
+      "num_tokens": 162745373.0,
+      "step": 69510
+    },
+    {
+      "epoch": 0.6373888328596314,
+      "learning_rate": 7.252406711286331e-05,
+      "loss": 0.6733,
+      "mean_token_accuracy": 0.7865135848522187,
+      "num_tokens": 162770538.0,
+      "step": 69520
+    },
+    {
+      "epoch": 0.6374805170991107,
+      "learning_rate": 7.250573026496746e-05,
+      "loss": 0.6924,
+      "mean_token_accuracy": 0.7909885168075561,
+      "num_tokens": 162795895.0,
+      "step": 69530
+    },
+    {
+      "epoch": 0.6375722013385899,
+      "learning_rate": 7.248739341707161e-05,
+      "loss": 0.6288,
+      "mean_token_accuracy": 0.8023481726646423,
+      "num_tokens": 162820899.0,
+      "step": 69540
+    },
+    {
+      "epoch": 0.6376638855780691,
+      "learning_rate": 7.246905656917576e-05,
+      "loss": 0.702,
+      "mean_token_accuracy": 0.7822565495967865,
+      "num_tokens": 162845891.0,
+      "step": 69550
+    },
+    {
+      "epoch": 0.6377555698175483,
+      "learning_rate": 7.245071972127991e-05,
+      "loss": 0.704,
+      "mean_token_accuracy": 0.7825054824352264,
+      "num_tokens": 162870929.0,
+      "step": 69560
+    },
+    {
+      "epoch": 0.6378472540570276,
+      "learning_rate": 7.243238287338407e-05,
+      "loss": 0.7278,
+      "mean_token_accuracy": 0.7798559546470643,
+      "num_tokens": 162896226.0,
+      "step": 69570
+    },
+    {
+      "epoch": 0.6379389382965068,
+      "learning_rate": 7.241404602548822e-05,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.7871682465076446,
+      "num_tokens": 162921594.0,
+      "step": 69580
+    },
+    {
+      "epoch": 0.6380306225359861,
+      "learning_rate": 7.239570917759237e-05,
+      "loss": 0.6872,
+      "mean_token_accuracy": 0.7882108747959137,
+      "num_tokens": 162946831.0,
+      "step": 69590
+    },
+    {
+      "epoch": 0.6381223067754653,
+      "learning_rate": 7.237737232969652e-05,
+      "loss": 0.7041,
+      "mean_token_accuracy": 0.7825108349323273,
+      "num_tokens": 162971362.0,
+      "step": 69600
+    },
+    {
+      "epoch": 0.6382139910149446,
+      "learning_rate": 7.235903548180069e-05,
+      "loss": 0.6391,
+      "mean_token_accuracy": 0.8007571935653687,
+      "num_tokens": 162996368.0,
+      "step": 69610
+    },
+    {
+      "epoch": 0.6383056752544237,
+      "learning_rate": 7.234069863390484e-05,
+      "loss": 0.6606,
+      "mean_token_accuracy": 0.7967602014541626,
+      "num_tokens": 163021712.0,
+      "step": 69620
+    },
+    {
+      "epoch": 0.638397359493903,
+      "learning_rate": 7.232236178600899e-05,
+      "loss": 0.7315,
+      "mean_token_accuracy": 0.7771764993667603,
+      "num_tokens": 163047131.0,
+      "step": 69630
+    },
+    {
+      "epoch": 0.6384890437333822,
+      "learning_rate": 7.230402493811314e-05,
+      "loss": 0.6885,
+      "mean_token_accuracy": 0.7899558424949646,
+      "num_tokens": 163072048.0,
+      "step": 69640
+    },
+    {
+      "epoch": 0.6385807279728615,
+      "learning_rate": 7.22856880902173e-05,
+      "loss": 0.6825,
+      "mean_token_accuracy": 0.7926486611366272,
+      "num_tokens": 163098353.0,
+      "step": 69650
+    },
+    {
+      "epoch": 0.6386724122123407,
+      "learning_rate": 7.226735124232145e-05,
+      "loss": 0.6799,
+      "mean_token_accuracy": 0.7868564188480377,
+      "num_tokens": 163123366.0,
+      "step": 69660
+    },
+    {
+      "epoch": 0.63876409645182,
+      "learning_rate": 7.224901439442561e-05,
+      "loss": 0.6733,
+      "mean_token_accuracy": 0.7880558729171753,
+      "num_tokens": 163148302.0,
+      "step": 69670
+    },
+    {
+      "epoch": 0.6388557806912991,
+      "learning_rate": 7.223067754652975e-05,
+      "loss": 0.6666,
+      "mean_token_accuracy": 0.790253871679306,
+      "num_tokens": 163173824.0,
+      "step": 69680
+    },
+    {
+      "epoch": 0.6389474649307784,
+      "learning_rate": 7.22123406986339e-05,
+      "loss": 0.668,
+      "mean_token_accuracy": 0.7905177235603332,
+      "num_tokens": 163198651.0,
+      "step": 69690
+    },
+    {
+      "epoch": 0.6390391491702576,
+      "learning_rate": 7.219400385073805e-05,
+      "loss": 0.667,
+      "mean_token_accuracy": 0.7918867409229279,
+      "num_tokens": 163224308.0,
+      "step": 69700
+    },
+    {
+      "epoch": 0.6391308334097369,
+      "learning_rate": 7.217566700284222e-05,
+      "loss": 0.6939,
+      "mean_token_accuracy": 0.7851086556911469,
+      "num_tokens": 163249267.0,
+      "step": 69710
+    },
+    {
+      "epoch": 0.6392225176492161,
+      "learning_rate": 7.215733015494637e-05,
+      "loss": 0.6415,
+      "mean_token_accuracy": 0.7975740432739258,
+      "num_tokens": 163274410.0,
+      "step": 69720
+    },
+    {
+      "epoch": 0.6393142018886954,
+      "learning_rate": 7.213899330705052e-05,
+      "loss": 0.6817,
+      "mean_token_accuracy": 0.7886723101139068,
+      "num_tokens": 163298932.0,
+      "step": 69730
+    },
+    {
+      "epoch": 0.6394058861281746,
+      "learning_rate": 7.212065645915468e-05,
+      "loss": 0.6715,
+      "mean_token_accuracy": 0.7912049293518066,
+      "num_tokens": 163323643.0,
+      "step": 69740
+    },
+    {
+      "epoch": 0.6394975703676538,
+      "learning_rate": 7.210231961125883e-05,
+      "loss": 0.6816,
+      "mean_token_accuracy": 0.7898310422897339,
+      "num_tokens": 163348119.0,
+      "step": 69750
+    },
+    {
+      "epoch": 0.639589254607133,
+      "learning_rate": 7.208398276336298e-05,
+      "loss": 0.6608,
+      "mean_token_accuracy": 0.7954709529876709,
+      "num_tokens": 163373166.0,
+      "step": 69760
+    },
+    {
+      "epoch": 0.6396809388466123,
+      "learning_rate": 7.206564591546714e-05,
+      "loss": 0.6933,
+      "mean_token_accuracy": 0.7857273757457733,
+      "num_tokens": 163397935.0,
+      "step": 69770
+    },
+    {
+      "epoch": 0.6397726230860915,
+      "learning_rate": 7.204730906757128e-05,
+      "loss": 0.6548,
+      "mean_token_accuracy": 0.7971982300281525,
+      "num_tokens": 163421941.0,
+      "step": 69780
+    },
+    {
+      "epoch": 0.6398643073255708,
+      "learning_rate": 7.202897221967543e-05,
+      "loss": 0.7096,
+      "mean_token_accuracy": 0.7826210558414459,
+      "num_tokens": 163446731.0,
+      "step": 69790
+    },
+    {
+      "epoch": 0.63995599156505,
+      "learning_rate": 7.20106353717796e-05,
+      "loss": 0.6814,
+      "mean_token_accuracy": 0.7860031604766846,
+      "num_tokens": 163472067.0,
+      "step": 69800
+    },
+    {
+      "epoch": 0.6400476758045291,
+      "learning_rate": 7.199229852388375e-05,
+      "loss": 0.6937,
+      "mean_token_accuracy": 0.785373193025589,
+      "num_tokens": 163496986.0,
+      "step": 69810
+    },
+    {
+      "epoch": 0.6401393600440084,
+      "learning_rate": 7.19739616759879e-05,
+      "loss": 0.6741,
+      "mean_token_accuracy": 0.7898578107357025,
+      "num_tokens": 163521793.0,
+      "step": 69820
+    },
+    {
+      "epoch": 0.6402310442834876,
+      "learning_rate": 7.195562482809206e-05,
+      "loss": 0.6911,
+      "mean_token_accuracy": 0.7895965039730072,
+      "num_tokens": 163546456.0,
+      "step": 69830
+    },
+    {
+      "epoch": 0.6403227285229669,
+      "learning_rate": 7.193728798019621e-05,
+      "loss": 0.7088,
+      "mean_token_accuracy": 0.7853328108787536,
+      "num_tokens": 163571253.0,
+      "step": 69840
+    },
+    {
+      "epoch": 0.6404144127624462,
+      "learning_rate": 7.191895113230036e-05,
+      "loss": 0.6655,
+      "mean_token_accuracy": 0.7966778159141541,
+      "num_tokens": 163595905.0,
+      "step": 69850
+    },
+    {
+      "epoch": 0.6405060970019254,
+      "learning_rate": 7.190061428440451e-05,
+      "loss": 0.7084,
+      "mean_token_accuracy": 0.7869901061058044,
+      "num_tokens": 163620350.0,
+      "step": 69860
+    },
+    {
+      "epoch": 0.6405977812414047,
+      "learning_rate": 7.188227743650868e-05,
+      "loss": 0.6574,
+      "mean_token_accuracy": 0.8003265202045441,
+      "num_tokens": 163646068.0,
+      "step": 69870
+    },
+    {
+      "epoch": 0.6406894654808838,
+      "learning_rate": 7.186394058861281e-05,
+      "loss": 0.7159,
+      "mean_token_accuracy": 0.7857119798660278,
+      "num_tokens": 163670436.0,
+      "step": 69880
+    },
+    {
+      "epoch": 0.640781149720363,
+      "learning_rate": 7.184560374071697e-05,
+      "loss": 0.6487,
+      "mean_token_accuracy": 0.7965015709400177,
+      "num_tokens": 163695026.0,
+      "step": 69890
+    },
+    {
+      "epoch": 0.6408728339598423,
+      "learning_rate": 7.182726689282113e-05,
+      "loss": 0.7274,
+      "mean_token_accuracy": 0.7822119891643524,
+      "num_tokens": 163720320.0,
+      "step": 69900
+    },
+    {
+      "epoch": 0.6409645181993215,
+      "learning_rate": 7.180893004492528e-05,
+      "loss": 0.6724,
+      "mean_token_accuracy": 0.7920660018920899,
+      "num_tokens": 163745511.0,
+      "step": 69910
+    },
+    {
+      "epoch": 0.6410562024388008,
+      "learning_rate": 7.179059319702944e-05,
+      "loss": 0.6887,
+      "mean_token_accuracy": 0.7868866443634033,
+      "num_tokens": 163770241.0,
+      "step": 69920
+    },
+    {
+      "epoch": 0.64114788667828,
+      "learning_rate": 7.177225634913359e-05,
+      "loss": 0.7023,
+      "mean_token_accuracy": 0.7846464157104492,
+      "num_tokens": 163795652.0,
+      "step": 69930
+    },
+    {
+      "epoch": 0.6412395709177592,
+      "learning_rate": 7.175391950123774e-05,
+      "loss": 0.6881,
+      "mean_token_accuracy": 0.7887941777706147,
+      "num_tokens": 163821072.0,
+      "step": 69940
+    },
+    {
+      "epoch": 0.6413312551572384,
+      "learning_rate": 7.173558265334189e-05,
+      "loss": 0.7063,
+      "mean_token_accuracy": 0.7892722904682159,
+      "num_tokens": 163845565.0,
+      "step": 69950
+    },
+    {
+      "epoch": 0.6414229393967177,
+      "learning_rate": 7.171724580544604e-05,
+      "loss": 0.6867,
+      "mean_token_accuracy": 0.7910989046096801,
+      "num_tokens": 163870863.0,
+      "step": 69960
+    },
+    {
+      "epoch": 0.6415146236361969,
+      "learning_rate": 7.169890895755021e-05,
+      "loss": 0.7035,
+      "mean_token_accuracy": 0.7852825403213501,
+      "num_tokens": 163896814.0,
+      "step": 69970
+    },
+    {
+      "epoch": 0.6416063078756762,
+      "learning_rate": 7.168057210965435e-05,
+      "loss": 0.7216,
+      "mean_token_accuracy": 0.782462191581726,
+      "num_tokens": 163921146.0,
+      "step": 69980
+    },
+    {
+      "epoch": 0.6416979921151554,
+      "learning_rate": 7.16622352617585e-05,
+      "loss": 0.6836,
+      "mean_token_accuracy": 0.7861079633235931,
+      "num_tokens": 163945924.0,
+      "step": 69990
+    },
+    {
+      "epoch": 0.6417896763546347,
+      "learning_rate": 7.164389841386266e-05,
+      "loss": 0.7098,
+      "mean_token_accuracy": 0.7870415568351745,
+      "num_tokens": 163971537.0,
+      "step": 70000
+    },
+    {
+      "epoch": 0.6418813605941138,
+      "learning_rate": 7.162556156596682e-05,
+      "loss": 0.6903,
+      "mean_token_accuracy": 0.7880664765834808,
+      "num_tokens": 163996861.0,
+      "step": 70010
+    },
+    {
+      "epoch": 0.6419730448335931,
+      "learning_rate": 7.160722471807097e-05,
+      "loss": 0.6748,
+      "mean_token_accuracy": 0.7970955908298493,
+      "num_tokens": 164021869.0,
+      "step": 70020
+    },
+    {
+      "epoch": 0.6420647290730723,
+      "learning_rate": 7.158888787017512e-05,
+      "loss": 0.7076,
+      "mean_token_accuracy": 0.7810359179973603,
+      "num_tokens": 164046334.0,
+      "step": 70030
+    },
+    {
+      "epoch": 0.6421564133125516,
+      "learning_rate": 7.157055102227927e-05,
+      "loss": 0.6875,
+      "mean_token_accuracy": 0.7882570087909698,
+      "num_tokens": 164071630.0,
+      "step": 70040
+    },
+    {
+      "epoch": 0.6422480975520308,
+      "learning_rate": 7.155221417438342e-05,
+      "loss": 0.7018,
+      "mean_token_accuracy": 0.7878653764724731,
+      "num_tokens": 164096558.0,
+      "step": 70050
+    },
+    {
+      "epoch": 0.6423397817915101,
+      "learning_rate": 7.153387732648758e-05,
+      "loss": 0.7,
+      "mean_token_accuracy": 0.789144092798233,
+      "num_tokens": 164121370.0,
+      "step": 70060
+    },
+    {
+      "epoch": 0.6424314660309892,
+      "learning_rate": 7.151554047859174e-05,
+      "loss": 0.7039,
+      "mean_token_accuracy": 0.7815370678901672,
+      "num_tokens": 164146028.0,
+      "step": 70070
+    },
+    {
+      "epoch": 0.6425231502704685,
+      "learning_rate": 7.149720363069588e-05,
+      "loss": 0.6586,
+      "mean_token_accuracy": 0.7944897234439849,
+      "num_tokens": 164171710.0,
+      "step": 70080
+    },
+    {
+      "epoch": 0.6426148345099477,
+      "learning_rate": 7.147886678280003e-05,
+      "loss": 0.686,
+      "mean_token_accuracy": 0.791346150636673,
+      "num_tokens": 164196602.0,
+      "step": 70090
+    },
+    {
+      "epoch": 0.642706518749427,
+      "learning_rate": 7.14605299349042e-05,
+      "loss": 0.6627,
+      "mean_token_accuracy": 0.7907776772975922,
+      "num_tokens": 164221491.0,
+      "step": 70100
+    },
+    {
+      "epoch": 0.6427982029889062,
+      "learning_rate": 7.144219308700835e-05,
+      "loss": 0.6898,
+      "mean_token_accuracy": 0.7849031150341034,
+      "num_tokens": 164247459.0,
+      "step": 70110
+    },
+    {
+      "epoch": 0.6428898872283855,
+      "learning_rate": 7.14238562391125e-05,
+      "loss": 0.6696,
+      "mean_token_accuracy": 0.791460645198822,
+      "num_tokens": 164272549.0,
+      "step": 70120
+    },
+    {
+      "epoch": 0.6429815714678647,
+      "learning_rate": 7.140551939121665e-05,
+      "loss": 0.7107,
+      "mean_token_accuracy": 0.7802798926830292,
+      "num_tokens": 164297250.0,
+      "step": 70130
+    },
+    {
+      "epoch": 0.6430732557073439,
+      "learning_rate": 7.13871825433208e-05,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.7885702729225159,
+      "num_tokens": 164322338.0,
+      "step": 70140
+    },
+    {
+      "epoch": 0.6431649399468231,
+      "learning_rate": 7.136884569542496e-05,
+      "loss": 0.6853,
+      "mean_token_accuracy": 0.7821885645389557,
+      "num_tokens": 164347431.0,
+      "step": 70150
+    },
+    {
+      "epoch": 0.6432566241863024,
+      "learning_rate": 7.135050884752912e-05,
+      "loss": 0.6653,
+      "mean_token_accuracy": 0.8017870903015136,
+      "num_tokens": 164372925.0,
+      "step": 70160
+    },
+    {
+      "epoch": 0.6433483084257816,
+      "learning_rate": 7.133217199963327e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8037239968776703,
+      "num_tokens": 164397533.0,
+      "step": 70170
+    },
+    {
+      "epoch": 0.6434399926652609,
+      "learning_rate": 7.131383515173741e-05,
+      "loss": 0.727,
+      "mean_token_accuracy": 0.783105605840683,
+      "num_tokens": 164422345.0,
+      "step": 70180
+    },
+    {
+      "epoch": 0.6435316769047401,
+      "learning_rate": 7.129549830384156e-05,
+      "loss": 0.6989,
+      "mean_token_accuracy": 0.7893296658992768,
+      "num_tokens": 164446832.0,
+      "step": 70190
+    },
+    {
+      "epoch": 0.6436233611442194,
+      "learning_rate": 7.127716145594573e-05,
+      "loss": 0.6495,
+      "mean_token_accuracy": 0.7958596587181092,
+      "num_tokens": 164471346.0,
+      "step": 70200
+    },
+    {
+      "epoch": 0.6437150453836985,
+      "learning_rate": 7.125882460804988e-05,
+      "loss": 0.702,
+      "mean_token_accuracy": 0.7869407355785369,
+      "num_tokens": 164496920.0,
+      "step": 70210
+    },
+    {
+      "epoch": 0.6438067296231778,
+      "learning_rate": 7.124048776015403e-05,
+      "loss": 0.6865,
+      "mean_token_accuracy": 0.7850927472114563,
+      "num_tokens": 164522293.0,
+      "step": 70220
+    },
+    {
+      "epoch": 0.643898413862657,
+      "learning_rate": 7.122215091225818e-05,
+      "loss": 0.6742,
+      "mean_token_accuracy": 0.7899061143398285,
+      "num_tokens": 164547079.0,
+      "step": 70230
+    },
+    {
+      "epoch": 0.6439900981021363,
+      "learning_rate": 7.120381406436234e-05,
+      "loss": 0.7058,
+      "mean_token_accuracy": 0.784560638666153,
+      "num_tokens": 164572493.0,
+      "step": 70240
+    },
+    {
+      "epoch": 0.6440817823416155,
+      "learning_rate": 7.118547721646649e-05,
+      "loss": 0.7132,
+      "mean_token_accuracy": 0.7789095997810364,
+      "num_tokens": 164597930.0,
+      "step": 70250
+    },
+    {
+      "epoch": 0.6441734665810948,
+      "learning_rate": 7.116714036857065e-05,
+      "loss": 0.6685,
+      "mean_token_accuracy": 0.792503434419632,
+      "num_tokens": 164622996.0,
+      "step": 70260
+    },
+    {
+      "epoch": 0.6442651508205739,
+      "learning_rate": 7.11488035206748e-05,
+      "loss": 0.6744,
+      "mean_token_accuracy": 0.7918026268482208,
+      "num_tokens": 164648448.0,
+      "step": 70270
+    },
+    {
+      "epoch": 0.6443568350600531,
+      "learning_rate": 7.113046667277894e-05,
+      "loss": 0.6697,
+      "mean_token_accuracy": 0.7937646448612213,
+      "num_tokens": 164673196.0,
+      "step": 70280
+    },
+    {
+      "epoch": 0.6444485192995324,
+      "learning_rate": 7.111212982488311e-05,
+      "loss": 0.711,
+      "mean_token_accuracy": 0.7829077959060669,
+      "num_tokens": 164698768.0,
+      "step": 70290
+    },
+    {
+      "epoch": 0.6445402035390116,
+      "learning_rate": 7.109379297698726e-05,
+      "loss": 0.6682,
+      "mean_token_accuracy": 0.7933432757854462,
+      "num_tokens": 164723369.0,
+      "step": 70300
+    },
+    {
+      "epoch": 0.6446318877784909,
+      "learning_rate": 7.107545612909141e-05,
+      "loss": 0.6743,
+      "mean_token_accuracy": 0.7898480892181396,
+      "num_tokens": 164748588.0,
+      "step": 70310
+    },
+    {
+      "epoch": 0.6447235720179701,
+      "learning_rate": 7.105711928119556e-05,
+      "loss": 0.7369,
+      "mean_token_accuracy": 0.7723786234855652,
+      "num_tokens": 164773080.0,
+      "step": 70320
+    },
+    {
+      "epoch": 0.6448152562574494,
+      "learning_rate": 7.103878243329972e-05,
+      "loss": 0.6946,
+      "mean_token_accuracy": 0.7824502050876617,
+      "num_tokens": 164798200.0,
+      "step": 70330
+    },
+    {
+      "epoch": 0.6449069404969285,
+      "learning_rate": 7.102044558540387e-05,
+      "loss": 0.6517,
+      "mean_token_accuracy": 0.8015531122684478,
+      "num_tokens": 164823234.0,
+      "step": 70340
+    },
+    {
+      "epoch": 0.6449986247364078,
+      "learning_rate": 7.100210873750802e-05,
+      "loss": 0.6791,
+      "mean_token_accuracy": 0.7885856449604034,
+      "num_tokens": 164848631.0,
+      "step": 70350
+    },
+    {
+      "epoch": 0.645090308975887,
+      "learning_rate": 7.098377188961219e-05,
+      "loss": 0.682,
+      "mean_token_accuracy": 0.7893922865390778,
+      "num_tokens": 164874434.0,
+      "step": 70360
+    },
+    {
+      "epoch": 0.6451819932153663,
+      "learning_rate": 7.096543504171634e-05,
+      "loss": 0.6924,
+      "mean_token_accuracy": 0.7901115536689758,
+      "num_tokens": 164899874.0,
+      "step": 70370
+    },
+    {
+      "epoch": 0.6452736774548455,
+      "learning_rate": 7.094709819382048e-05,
+      "loss": 0.6939,
+      "mean_token_accuracy": 0.7853074848651886,
+      "num_tokens": 164925257.0,
+      "step": 70380
+    },
+    {
+      "epoch": 0.6453653616943248,
+      "learning_rate": 7.092876134592464e-05,
+      "loss": 0.6707,
+      "mean_token_accuracy": 0.7973959863185882,
+      "num_tokens": 164950537.0,
+      "step": 70390
+    },
+    {
+      "epoch": 0.6454570459338039,
+      "learning_rate": 7.09104244980288e-05,
+      "loss": 0.7193,
+      "mean_token_accuracy": 0.7829456925392151,
+      "num_tokens": 164975172.0,
+      "step": 70400
+    },
+    {
+      "epoch": 0.6455487301732832,
+      "learning_rate": 7.089208765013295e-05,
+      "loss": 0.6875,
+      "mean_token_accuracy": 0.7823552489280701,
+      "num_tokens": 164999525.0,
+      "step": 70410
+    },
+    {
+      "epoch": 0.6456404144127624,
+      "learning_rate": 7.08737508022371e-05,
+      "loss": 0.6499,
+      "mean_token_accuracy": 0.7994419276714325,
+      "num_tokens": 165023937.0,
+      "step": 70420
+    },
+    {
+      "epoch": 0.6457320986522417,
+      "learning_rate": 7.085541395434125e-05,
+      "loss": 0.6969,
+      "mean_token_accuracy": 0.7814277410507202,
+      "num_tokens": 165049014.0,
+      "step": 70430
+    },
+    {
+      "epoch": 0.6458237828917209,
+      "learning_rate": 7.08370771064454e-05,
+      "loss": 0.6853,
+      "mean_token_accuracy": 0.78925821185112,
+      "num_tokens": 165074552.0,
+      "step": 70440
+    },
+    {
+      "epoch": 0.6459154671312002,
+      "learning_rate": 7.081874025854955e-05,
+      "loss": 0.6807,
+      "mean_token_accuracy": 0.7904388070106506,
+      "num_tokens": 165100461.0,
+      "step": 70450
+    },
+    {
+      "epoch": 0.6460071513706794,
+      "learning_rate": 7.080040341065372e-05,
+      "loss": 0.6764,
+      "mean_token_accuracy": 0.7925844073295594,
+      "num_tokens": 165125147.0,
+      "step": 70460
+    },
+    {
+      "epoch": 0.6460988356101586,
+      "learning_rate": 7.078206656275787e-05,
+      "loss": 0.6826,
+      "mean_token_accuracy": 0.7928273975849152,
+      "num_tokens": 165150882.0,
+      "step": 70470
+    },
+    {
+      "epoch": 0.6461905198496378,
+      "learning_rate": 7.076372971486201e-05,
+      "loss": 0.709,
+      "mean_token_accuracy": 0.7848760843276977,
+      "num_tokens": 165175879.0,
+      "step": 70480
+    },
+    {
+      "epoch": 0.6462822040891171,
+      "learning_rate": 7.074539286696617e-05,
+      "loss": 0.683,
+      "mean_token_accuracy": 0.7861987471580505,
+      "num_tokens": 165200599.0,
+      "step": 70490
+    },
+    {
+      "epoch": 0.6463738883285963,
+      "learning_rate": 7.072705601907033e-05,
+      "loss": 0.6658,
+      "mean_token_accuracy": 0.7930368661880494,
+      "num_tokens": 165225446.0,
+      "step": 70500
+    },
+    {
+      "epoch": 0.6464655725680756,
+      "learning_rate": 7.070871917117448e-05,
+      "loss": 0.7041,
+      "mean_token_accuracy": 0.7830127954483033,
+      "num_tokens": 165250007.0,
+      "step": 70510
+    },
+    {
+      "epoch": 0.6465572568075548,
+      "learning_rate": 7.069038232327864e-05,
+      "loss": 0.6522,
+      "mean_token_accuracy": 0.8032989382743836,
+      "num_tokens": 165275395.0,
+      "step": 70520
+    },
+    {
+      "epoch": 0.646648941047034,
+      "learning_rate": 7.067204547538278e-05,
+      "loss": 0.6512,
+      "mean_token_accuracy": 0.7982889413833618,
+      "num_tokens": 165300461.0,
+      "step": 70530
+    },
+    {
+      "epoch": 0.6467406252865132,
+      "learning_rate": 7.065370862748693e-05,
+      "loss": 0.7155,
+      "mean_token_accuracy": 0.7819724440574646,
+      "num_tokens": 165326376.0,
+      "step": 70540
+    },
+    {
+      "epoch": 0.6468323095259925,
+      "learning_rate": 7.063537177959108e-05,
+      "loss": 0.66,
+      "mean_token_accuracy": 0.7927281200885773,
+      "num_tokens": 165350624.0,
+      "step": 70550
+    },
+    {
+      "epoch": 0.6469239937654717,
+      "learning_rate": 7.061703493169525e-05,
+      "loss": 0.724,
+      "mean_token_accuracy": 0.7824005544185638,
+      "num_tokens": 165376460.0,
+      "step": 70560
+    },
+    {
+      "epoch": 0.647015678004951,
+      "learning_rate": 7.05986980837994e-05,
+      "loss": 0.7089,
+      "mean_token_accuracy": 0.7911057472229004,
+      "num_tokens": 165401361.0,
+      "step": 70570
+    },
+    {
+      "epoch": 0.6471073622444302,
+      "learning_rate": 7.058036123590354e-05,
+      "loss": 0.6926,
+      "mean_token_accuracy": 0.7916284382343293,
+      "num_tokens": 165426545.0,
+      "step": 70580
+    },
+    {
+      "epoch": 0.6471990464839095,
+      "learning_rate": 7.05620243880077e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.7930202126502991,
+      "num_tokens": 165452458.0,
+      "step": 70590
+    },
+    {
+      "epoch": 0.6472907307233886,
+      "learning_rate": 7.054368754011186e-05,
+      "loss": 0.6907,
+      "mean_token_accuracy": 0.7872883677482605,
+      "num_tokens": 165477343.0,
+      "step": 70600
+    },
+    {
+      "epoch": 0.6473824149628679,
+      "learning_rate": 7.052535069221601e-05,
+      "loss": 0.6591,
+      "mean_token_accuracy": 0.7991599798202514,
+      "num_tokens": 165502586.0,
+      "step": 70610
+    },
+    {
+      "epoch": 0.6474740992023471,
+      "learning_rate": 7.050701384432018e-05,
+      "loss": 0.6798,
+      "mean_token_accuracy": 0.7952367842197419,
+      "num_tokens": 165528173.0,
+      "step": 70620
+    },
+    {
+      "epoch": 0.6475657834418264,
+      "learning_rate": 7.048867699642431e-05,
+      "loss": 0.7125,
+      "mean_token_accuracy": 0.7836075901985169,
+      "num_tokens": 165553877.0,
+      "step": 70630
+    },
+    {
+      "epoch": 0.6476574676813056,
+      "learning_rate": 7.047034014852847e-05,
+      "loss": 0.6684,
+      "mean_token_accuracy": 0.7939449429512024,
+      "num_tokens": 165578955.0,
+      "step": 70640
+    },
+    {
+      "epoch": 0.6477491519207849,
+      "learning_rate": 7.045200330063263e-05,
+      "loss": 0.6966,
+      "mean_token_accuracy": 0.7831183791160583,
+      "num_tokens": 165603439.0,
+      "step": 70650
+    },
+    {
+      "epoch": 0.647840836160264,
+      "learning_rate": 7.043366645273678e-05,
+      "loss": 0.6888,
+      "mean_token_accuracy": 0.7874828219413758,
+      "num_tokens": 165629096.0,
+      "step": 70660
+    },
+    {
+      "epoch": 0.6479325203997433,
+      "learning_rate": 7.041532960484093e-05,
+      "loss": 0.6739,
+      "mean_token_accuracy": 0.7955126941204071,
+      "num_tokens": 165654003.0,
+      "step": 70670
+    },
+    {
+      "epoch": 0.6480242046392225,
+      "learning_rate": 7.039699275694507e-05,
+      "loss": 0.7204,
+      "mean_token_accuracy": 0.7809175372123718,
+      "num_tokens": 165679006.0,
+      "step": 70680
+    },
+    {
+      "epoch": 0.6481158888787018,
+      "learning_rate": 7.037865590904924e-05,
+      "loss": 0.7067,
+      "mean_token_accuracy": 0.7847817420959473,
+      "num_tokens": 165704266.0,
+      "step": 70690
+    },
+    {
+      "epoch": 0.648207573118181,
+      "learning_rate": 7.036031906115339e-05,
+      "loss": 0.6784,
+      "mean_token_accuracy": 0.7917144119739532,
+      "num_tokens": 165729847.0,
+      "step": 70700
+    },
+    {
+      "epoch": 0.6482992573576603,
+      "learning_rate": 7.034198221325754e-05,
+      "loss": 0.6837,
+      "mean_token_accuracy": 0.7914392411708832,
+      "num_tokens": 165754422.0,
+      "step": 70710
+    },
+    {
+      "epoch": 0.6483909415971395,
+      "learning_rate": 7.032364536536171e-05,
+      "loss": 0.6869,
+      "mean_token_accuracy": 0.7832569360733033,
+      "num_tokens": 165779794.0,
+      "step": 70720
+    },
+    {
+      "epoch": 0.6484826258366186,
+      "learning_rate": 7.030530851746585e-05,
+      "loss": 0.6676,
+      "mean_token_accuracy": 0.7912835299968719,
+      "num_tokens": 165804790.0,
+      "step": 70730
+    },
+    {
+      "epoch": 0.6485743100760979,
+      "learning_rate": 7.028697166957e-05,
+      "loss": 0.6849,
+      "mean_token_accuracy": 0.7877888083457947,
+      "num_tokens": 165830531.0,
+      "step": 70740
+    },
+    {
+      "epoch": 0.6486659943155771,
+      "learning_rate": 7.026863482167416e-05,
+      "loss": 0.6765,
+      "mean_token_accuracy": 0.7906343519687653,
+      "num_tokens": 165854608.0,
+      "step": 70750
+    },
+    {
+      "epoch": 0.6487576785550564,
+      "learning_rate": 7.025029797377831e-05,
+      "loss": 0.644,
+      "mean_token_accuracy": 0.799230980873108,
+      "num_tokens": 165879234.0,
+      "step": 70760
+    },
+    {
+      "epoch": 0.6488493627945356,
+      "learning_rate": 7.023196112588247e-05,
+      "loss": 0.6963,
+      "mean_token_accuracy": 0.7899213194847107,
+      "num_tokens": 165904263.0,
+      "step": 70770
+    },
+    {
+      "epoch": 0.6489410470340149,
+      "learning_rate": 7.021362427798662e-05,
+      "loss": 0.6655,
+      "mean_token_accuracy": 0.7973266541957855,
+      "num_tokens": 165928833.0,
+      "step": 70780
+    },
+    {
+      "epoch": 0.649032731273494,
+      "learning_rate": 7.019528743009077e-05,
+      "loss": 0.6969,
+      "mean_token_accuracy": 0.7848142504692077,
+      "num_tokens": 165953955.0,
+      "step": 70790
+    },
+    {
+      "epoch": 0.6491244155129733,
+      "learning_rate": 7.017695058219492e-05,
+      "loss": 0.6976,
+      "mean_token_accuracy": 0.7819412648677826,
+      "num_tokens": 165980016.0,
+      "step": 70800
+    },
+    {
+      "epoch": 0.6492160997524525,
+      "learning_rate": 7.015861373429907e-05,
+      "loss": 0.6747,
+      "mean_token_accuracy": 0.7877288699150086,
+      "num_tokens": 166004857.0,
+      "step": 70810
+    },
+    {
+      "epoch": 0.6493077839919318,
+      "learning_rate": 7.014027688640324e-05,
+      "loss": 0.6749,
+      "mean_token_accuracy": 0.7964017391204834,
+      "num_tokens": 166030057.0,
+      "step": 70820
+    },
+    {
+      "epoch": 0.649399468231411,
+      "learning_rate": 7.012194003850738e-05,
+      "loss": 0.6931,
+      "mean_token_accuracy": 0.7858947098255158,
+      "num_tokens": 166055103.0,
+      "step": 70830
+    },
+    {
+      "epoch": 0.6494911524708903,
+      "learning_rate": 7.010360319061153e-05,
+      "loss": 0.6935,
+      "mean_token_accuracy": 0.7834270834922791,
+      "num_tokens": 166079202.0,
+      "step": 70840
+    },
+    {
+      "epoch": 0.6495828367103695,
+      "learning_rate": 7.00852663427157e-05,
+      "loss": 0.6896,
+      "mean_token_accuracy": 0.7870459735393525,
+      "num_tokens": 166104363.0,
+      "step": 70850
+    },
+    {
+      "epoch": 0.6496745209498487,
+      "learning_rate": 7.006692949481985e-05,
+      "loss": 0.6816,
+      "mean_token_accuracy": 0.7928813815116882,
+      "num_tokens": 166129539.0,
+      "step": 70860
+    },
+    {
+      "epoch": 0.6497662051893279,
+      "learning_rate": 7.0048592646924e-05,
+      "loss": 0.6732,
+      "mean_token_accuracy": 0.7869620025157928,
+      "num_tokens": 166155118.0,
+      "step": 70870
+    },
+    {
+      "epoch": 0.6498578894288072,
+      "learning_rate": 7.003025579902815e-05,
+      "loss": 0.6416,
+      "mean_token_accuracy": 0.7994863152503967,
+      "num_tokens": 166180039.0,
+      "step": 70880
+    },
+    {
+      "epoch": 0.6499495736682864,
+      "learning_rate": 7.00119189511323e-05,
+      "loss": 0.7179,
+      "mean_token_accuracy": 0.7806203186511993,
+      "num_tokens": 166205355.0,
+      "step": 70890
+    },
+    {
+      "epoch": 0.6500412579077657,
+      "learning_rate": 6.999358210323645e-05,
+      "loss": 0.7008,
+      "mean_token_accuracy": 0.7929344296455383,
+      "num_tokens": 166230529.0,
+      "step": 70900
+    },
+    {
+      "epoch": 0.6501329421472449,
+      "learning_rate": 6.997524525534062e-05,
+      "loss": 0.683,
+      "mean_token_accuracy": 0.787913191318512,
+      "num_tokens": 166255023.0,
+      "step": 70910
+    },
+    {
+      "epoch": 0.6502246263867241,
+      "learning_rate": 6.995690840744477e-05,
+      "loss": 0.6415,
+      "mean_token_accuracy": 0.8023931086063385,
+      "num_tokens": 166279792.0,
+      "step": 70920
+    },
+    {
+      "epoch": 0.6503163106262033,
+      "learning_rate": 6.993857155954891e-05,
+      "loss": 0.6464,
+      "mean_token_accuracy": 0.799000597000122,
+      "num_tokens": 166304000.0,
+      "step": 70930
+    },
+    {
+      "epoch": 0.6504079948656826,
+      "learning_rate": 6.992023471165306e-05,
+      "loss": 0.6863,
+      "mean_token_accuracy": 0.7917823731899262,
+      "num_tokens": 166328112.0,
+      "step": 70940
+    },
+    {
+      "epoch": 0.6504996791051618,
+      "learning_rate": 6.990189786375723e-05,
+      "loss": 0.6656,
+      "mean_token_accuracy": 0.7906832039356232,
+      "num_tokens": 166352677.0,
+      "step": 70950
+    },
+    {
+      "epoch": 0.6505913633446411,
+      "learning_rate": 6.988356101586138e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.7833178222179413,
+      "num_tokens": 166377524.0,
+      "step": 70960
+    },
+    {
+      "epoch": 0.6506830475841203,
+      "learning_rate": 6.986522416796553e-05,
+      "loss": 0.6906,
+      "mean_token_accuracy": 0.788356339931488,
+      "num_tokens": 166402176.0,
+      "step": 70970
+    },
+    {
+      "epoch": 0.6507747318235996,
+      "learning_rate": 6.984688732006968e-05,
+      "loss": 0.6845,
+      "mean_token_accuracy": 0.794219434261322,
+      "num_tokens": 166427401.0,
+      "step": 70980
+    },
+    {
+      "epoch": 0.6508664160630787,
+      "learning_rate": 6.982855047217383e-05,
+      "loss": 0.7068,
+      "mean_token_accuracy": 0.7824733138084412,
+      "num_tokens": 166452391.0,
+      "step": 70990
+    },
+    {
+      "epoch": 0.650958100302558,
+      "learning_rate": 6.981021362427799e-05,
+      "loss": 0.686,
+      "mean_token_accuracy": 0.7882467329502105,
+      "num_tokens": 166477223.0,
+      "step": 71000
+    },
+    {
+      "epoch": 0.6510497845420372,
+      "learning_rate": 6.979187677638215e-05,
+      "loss": 0.6556,
+      "mean_token_accuracy": 0.7950958728790283,
+      "num_tokens": 166503258.0,
+      "step": 71010
+    },
+    {
+      "epoch": 0.6511414687815165,
+      "learning_rate": 6.97735399284863e-05,
+      "loss": 0.6748,
+      "mean_token_accuracy": 0.792359584569931,
+      "num_tokens": 166528085.0,
+      "step": 71020
+    },
+    {
+      "epoch": 0.6512331530209957,
+      "learning_rate": 6.975520308059044e-05,
+      "loss": 0.6988,
+      "mean_token_accuracy": 0.7834649026393891,
+      "num_tokens": 166552873.0,
+      "step": 71030
+    },
+    {
+      "epoch": 0.651324837260475,
+      "learning_rate": 6.97368662326946e-05,
+      "loss": 0.67,
+      "mean_token_accuracy": 0.7926984071731568,
+      "num_tokens": 166579080.0,
+      "step": 71040
+    },
+    {
+      "epoch": 0.6514165214999541,
+      "learning_rate": 6.971852938479876e-05,
+      "loss": 0.6876,
+      "mean_token_accuracy": 0.7916151583194733,
+      "num_tokens": 166603875.0,
+      "step": 71050
+    },
+    {
+      "epoch": 0.6515082057394334,
+      "learning_rate": 6.970019253690291e-05,
+      "loss": 0.6593,
+      "mean_token_accuracy": 0.7899702608585357,
+      "num_tokens": 166628752.0,
+      "step": 71060
+    },
+    {
+      "epoch": 0.6515998899789126,
+      "learning_rate": 6.968185568900706e-05,
+      "loss": 0.6711,
+      "mean_token_accuracy": 0.7901023924350739,
+      "num_tokens": 166653770.0,
+      "step": 71070
+    },
+    {
+      "epoch": 0.6516915742183919,
+      "learning_rate": 6.966351884111122e-05,
+      "loss": 0.6928,
+      "mean_token_accuracy": 0.7889417350292206,
+      "num_tokens": 166678800.0,
+      "step": 71080
+    },
+    {
+      "epoch": 0.6517832584578711,
+      "learning_rate": 6.964518199321537e-05,
+      "loss": 0.6949,
+      "mean_token_accuracy": 0.7840139269828796,
+      "num_tokens": 166703608.0,
+      "step": 71090
+    },
+    {
+      "epoch": 0.6518749426973504,
+      "learning_rate": 6.962684514531952e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.7932865202426911,
+      "num_tokens": 166728886.0,
+      "step": 71100
+    },
+    {
+      "epoch": 0.6519666269368296,
+      "learning_rate": 6.960850829742368e-05,
+      "loss": 0.6754,
+      "mean_token_accuracy": 0.7885345995426178,
+      "num_tokens": 166753171.0,
+      "step": 71110
+    },
+    {
+      "epoch": 0.6520583111763087,
+      "learning_rate": 6.959017144952784e-05,
+      "loss": 0.6949,
+      "mean_token_accuracy": 0.7940915107727051,
+      "num_tokens": 166778699.0,
+      "step": 71120
+    },
+    {
+      "epoch": 0.652149995415788,
+      "learning_rate": 6.957183460163197e-05,
+      "loss": 0.696,
+      "mean_token_accuracy": 0.7829701483249665,
+      "num_tokens": 166803107.0,
+      "step": 71130
+    },
+    {
+      "epoch": 0.6522416796552672,
+      "learning_rate": 6.955349775373614e-05,
+      "loss": 0.6791,
+      "mean_token_accuracy": 0.7943341434001923,
+      "num_tokens": 166828711.0,
+      "step": 71140
+    },
+    {
+      "epoch": 0.6523333638947465,
+      "learning_rate": 6.953516090584029e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.7912250399589539,
+      "num_tokens": 166853598.0,
+      "step": 71150
+    },
+    {
+      "epoch": 0.6524250481342257,
+      "learning_rate": 6.951682405794444e-05,
+      "loss": 0.6956,
+      "mean_token_accuracy": 0.7875594735145569,
+      "num_tokens": 166878884.0,
+      "step": 71160
+    },
+    {
+      "epoch": 0.652516732373705,
+      "learning_rate": 6.94984872100486e-05,
+      "loss": 0.6711,
+      "mean_token_accuracy": 0.7966920077800751,
+      "num_tokens": 166903942.0,
+      "step": 71170
+    },
+    {
+      "epoch": 0.6526084166131841,
+      "learning_rate": 6.948015036215275e-05,
+      "loss": 0.6976,
+      "mean_token_accuracy": 0.7829735338687897,
+      "num_tokens": 166929442.0,
+      "step": 71180
+    },
+    {
+      "epoch": 0.6527001008526634,
+      "learning_rate": 6.94618135142569e-05,
+      "loss": 0.7069,
+      "mean_token_accuracy": 0.7873943626880646,
+      "num_tokens": 166955220.0,
+      "step": 71190
+    },
+    {
+      "epoch": 0.6527917850921426,
+      "learning_rate": 6.944347666636105e-05,
+      "loss": 0.6779,
+      "mean_token_accuracy": 0.7900447189807892,
+      "num_tokens": 166980634.0,
+      "step": 71200
+    },
+    {
+      "epoch": 0.6528834693316219,
+      "learning_rate": 6.942513981846522e-05,
+      "loss": 0.6741,
+      "mean_token_accuracy": 0.7973148584365845,
+      "num_tokens": 167005237.0,
+      "step": 71210
+    },
+    {
+      "epoch": 0.6529751535711011,
+      "learning_rate": 6.940680297056937e-05,
+      "loss": 0.7085,
+      "mean_token_accuracy": 0.7882147192955017,
+      "num_tokens": 167030239.0,
+      "step": 71220
+    },
+    {
+      "epoch": 0.6530668378105804,
+      "learning_rate": 6.93884661226735e-05,
+      "loss": 0.7009,
+      "mean_token_accuracy": 0.780734521150589,
+      "num_tokens": 167055419.0,
+      "step": 71230
+    },
+    {
+      "epoch": 0.6531585220500596,
+      "learning_rate": 6.937012927477767e-05,
+      "loss": 0.6623,
+      "mean_token_accuracy": 0.7908092081546784,
+      "num_tokens": 167080449.0,
+      "step": 71240
+    },
+    {
+      "epoch": 0.6532502062895388,
+      "learning_rate": 6.935179242688182e-05,
+      "loss": 0.7019,
+      "mean_token_accuracy": 0.7876363217830658,
+      "num_tokens": 167105462.0,
+      "step": 71250
+    },
+    {
+      "epoch": 0.653341890529018,
+      "learning_rate": 6.933345557898598e-05,
+      "loss": 0.7132,
+      "mean_token_accuracy": 0.7861493468284607,
+      "num_tokens": 167130952.0,
+      "step": 71260
+    },
+    {
+      "epoch": 0.6534335747684973,
+      "learning_rate": 6.931511873109013e-05,
+      "loss": 0.6802,
+      "mean_token_accuracy": 0.7848678529262543,
+      "num_tokens": 167156303.0,
+      "step": 71270
+    },
+    {
+      "epoch": 0.6535252590079765,
+      "learning_rate": 6.929678188319428e-05,
+      "loss": 0.6602,
+      "mean_token_accuracy": 0.7955906808376312,
+      "num_tokens": 167182702.0,
+      "step": 71280
+    },
+    {
+      "epoch": 0.6536169432474558,
+      "learning_rate": 6.927844503529843e-05,
+      "loss": 0.6819,
+      "mean_token_accuracy": 0.794136357307434,
+      "num_tokens": 167207799.0,
+      "step": 71290
+    },
+    {
+      "epoch": 0.653708627486935,
+      "learning_rate": 6.926010818740258e-05,
+      "loss": 0.7208,
+      "mean_token_accuracy": 0.7786969423294068,
+      "num_tokens": 167233324.0,
+      "step": 71300
+    },
+    {
+      "epoch": 0.6538003117264142,
+      "learning_rate": 6.924177133950675e-05,
+      "loss": 0.726,
+      "mean_token_accuracy": 0.7827441334724426,
+      "num_tokens": 167258748.0,
+      "step": 71310
+    },
+    {
+      "epoch": 0.6538919959658934,
+      "learning_rate": 6.92234344916109e-05,
+      "loss": 0.6867,
+      "mean_token_accuracy": 0.7898482620716095,
+      "num_tokens": 167284325.0,
+      "step": 71320
+    },
+    {
+      "epoch": 0.6539836802053727,
+      "learning_rate": 6.920509764371504e-05,
+      "loss": 0.6657,
+      "mean_token_accuracy": 0.7865634679794311,
+      "num_tokens": 167310076.0,
+      "step": 71330
+    },
+    {
+      "epoch": 0.6540753644448519,
+      "learning_rate": 6.91867607958192e-05,
+      "loss": 0.6725,
+      "mean_token_accuracy": 0.7915717363357544,
+      "num_tokens": 167334811.0,
+      "step": 71340
+    },
+    {
+      "epoch": 0.6541670486843312,
+      "learning_rate": 6.916842394792336e-05,
+      "loss": 0.6903,
+      "mean_token_accuracy": 0.7892451882362366,
+      "num_tokens": 167359807.0,
+      "step": 71350
+    },
+    {
+      "epoch": 0.6542587329238104,
+      "learning_rate": 6.915008710002751e-05,
+      "loss": 0.6944,
+      "mean_token_accuracy": 0.7852569878101349,
+      "num_tokens": 167385181.0,
+      "step": 71360
+    },
+    {
+      "epoch": 0.6543504171632897,
+      "learning_rate": 6.913175025213166e-05,
+      "loss": 0.6771,
+      "mean_token_accuracy": 0.7882788836956024,
+      "num_tokens": 167410414.0,
+      "step": 71370
+    },
+    {
+      "epoch": 0.6544421014027688,
+      "learning_rate": 6.911341340423581e-05,
+      "loss": 0.6762,
+      "mean_token_accuracy": 0.7888165235519409,
+      "num_tokens": 167435843.0,
+      "step": 71380
+    },
+    {
+      "epoch": 0.6545337856422481,
+      "learning_rate": 6.909507655633996e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.7893908262252808,
+      "num_tokens": 167461325.0,
+      "step": 71390
+    },
+    {
+      "epoch": 0.6546254698817273,
+      "learning_rate": 6.907673970844413e-05,
+      "loss": 0.6957,
+      "mean_token_accuracy": 0.7918611347675324,
+      "num_tokens": 167485926.0,
+      "step": 71400
+    },
+    {
+      "epoch": 0.6547171541212066,
+      "learning_rate": 6.905840286054828e-05,
+      "loss": 0.6446,
+      "mean_token_accuracy": 0.7933196544647216,
+      "num_tokens": 167511660.0,
+      "step": 71410
+    },
+    {
+      "epoch": 0.6548088383606858,
+      "learning_rate": 6.904006601265243e-05,
+      "loss": 0.7014,
+      "mean_token_accuracy": 0.7869983971118927,
+      "num_tokens": 167536907.0,
+      "step": 71420
+    },
+    {
+      "epoch": 0.6549005226001651,
+      "learning_rate": 6.902172916475657e-05,
+      "loss": 0.7259,
+      "mean_token_accuracy": 0.7848357141017914,
+      "num_tokens": 167562140.0,
+      "step": 71430
+    },
+    {
+      "epoch": 0.6549922068396443,
+      "learning_rate": 6.900339231686074e-05,
+      "loss": 0.6888,
+      "mean_token_accuracy": 0.7915430128574371,
+      "num_tokens": 167586248.0,
+      "step": 71440
+    },
+    {
+      "epoch": 0.6550838910791235,
+      "learning_rate": 6.898505546896489e-05,
+      "loss": 0.7193,
+      "mean_token_accuracy": 0.7776082277297973,
+      "num_tokens": 167611633.0,
+      "step": 71450
+    },
+    {
+      "epoch": 0.6551755753186027,
+      "learning_rate": 6.896671862106904e-05,
+      "loss": 0.7026,
+      "mean_token_accuracy": 0.783545869588852,
+      "num_tokens": 167636728.0,
+      "step": 71460
+    },
+    {
+      "epoch": 0.655267259558082,
+      "learning_rate": 6.894838177317319e-05,
+      "loss": 0.7239,
+      "mean_token_accuracy": 0.7785880982875824,
+      "num_tokens": 167661386.0,
+      "step": 71470
+    },
+    {
+      "epoch": 0.6553589437975612,
+      "learning_rate": 6.893004492527734e-05,
+      "loss": 0.6606,
+      "mean_token_accuracy": 0.7987613320350647,
+      "num_tokens": 167685801.0,
+      "step": 71480
+    },
+    {
+      "epoch": 0.6554506280370405,
+      "learning_rate": 6.89117080773815e-05,
+      "loss": 0.6685,
+      "mean_token_accuracy": 0.792380154132843,
+      "num_tokens": 167711078.0,
+      "step": 71490
+    },
+    {
+      "epoch": 0.6555423122765197,
+      "learning_rate": 6.889337122948566e-05,
+      "loss": 0.6511,
+      "mean_token_accuracy": 0.7922020494937897,
+      "num_tokens": 167736759.0,
+      "step": 71500
+    },
+    {
+      "epoch": 0.6556339965159989,
+      "learning_rate": 6.887503438158981e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.7862393975257873,
+      "num_tokens": 167761993.0,
+      "step": 71510
+    },
+    {
+      "epoch": 0.6557256807554781,
+      "learning_rate": 6.885669753369397e-05,
+      "loss": 0.6569,
+      "mean_token_accuracy": 0.792601364850998,
+      "num_tokens": 167786582.0,
+      "step": 71520
+    },
+    {
+      "epoch": 0.6558173649949574,
+      "learning_rate": 6.883836068579812e-05,
+      "loss": 0.6876,
+      "mean_token_accuracy": 0.7875175237655639,
+      "num_tokens": 167811083.0,
+      "step": 71530
+    },
+    {
+      "epoch": 0.6559090492344366,
+      "learning_rate": 6.882002383790227e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.7888307452201844,
+      "num_tokens": 167836253.0,
+      "step": 71540
+    },
+    {
+      "epoch": 0.6560007334739159,
+      "learning_rate": 6.880168699000642e-05,
+      "loss": 0.6806,
+      "mean_token_accuracy": 0.790232104063034,
+      "num_tokens": 167860168.0,
+      "step": 71550
+    },
+    {
+      "epoch": 0.6560924177133951,
+      "learning_rate": 6.878335014211057e-05,
+      "loss": 0.7129,
+      "mean_token_accuracy": 0.7792242765426636,
+      "num_tokens": 167885750.0,
+      "step": 71560
+    },
+    {
+      "epoch": 0.6561841019528744,
+      "learning_rate": 6.876501329421472e-05,
+      "loss": 0.6676,
+      "mean_token_accuracy": 0.7946697473526001,
+      "num_tokens": 167911467.0,
+      "step": 71570
+    },
+    {
+      "epoch": 0.6562757861923535,
+      "learning_rate": 6.874667644631888e-05,
+      "loss": 0.6715,
+      "mean_token_accuracy": 0.7902608871459961,
+      "num_tokens": 167936193.0,
+      "step": 71580
+    },
+    {
+      "epoch": 0.6563674704318327,
+      "learning_rate": 6.872833959842303e-05,
+      "loss": 0.6759,
+      "mean_token_accuracy": 0.792341285943985,
+      "num_tokens": 167960758.0,
+      "step": 71590
+    },
+    {
+      "epoch": 0.656459154671312,
+      "learning_rate": 6.87100027505272e-05,
+      "loss": 0.685,
+      "mean_token_accuracy": 0.783710652589798,
+      "num_tokens": 167985686.0,
+      "step": 71600
+    },
+    {
+      "epoch": 0.6565508389107912,
+      "learning_rate": 6.869166590263135e-05,
+      "loss": 0.692,
+      "mean_token_accuracy": 0.8000739336013794,
+      "num_tokens": 168010472.0,
+      "step": 71610
+    },
+    {
+      "epoch": 0.6566425231502705,
+      "learning_rate": 6.86733290547355e-05,
+      "loss": 0.7105,
+      "mean_token_accuracy": 0.7812987685203552,
+      "num_tokens": 168035933.0,
+      "step": 71620
+    },
+    {
+      "epoch": 0.6567342073897497,
+      "learning_rate": 6.865499220683965e-05,
+      "loss": 0.6906,
+      "mean_token_accuracy": 0.7924662590026855,
+      "num_tokens": 168061104.0,
+      "step": 71630
+    },
+    {
+      "epoch": 0.6568258916292289,
+      "learning_rate": 6.86366553589438e-05,
+      "loss": 0.6977,
+      "mean_token_accuracy": 0.7867888748645783,
+      "num_tokens": 168086657.0,
+      "step": 71640
+    },
+    {
+      "epoch": 0.6569175758687081,
+      "learning_rate": 6.861831851104795e-05,
+      "loss": 0.6968,
+      "mean_token_accuracy": 0.7839610815048218,
+      "num_tokens": 168112276.0,
+      "step": 71650
+    },
+    {
+      "epoch": 0.6570092601081874,
+      "learning_rate": 6.85999816631521e-05,
+      "loss": 0.6925,
+      "mean_token_accuracy": 0.7872973442077636,
+      "num_tokens": 168137487.0,
+      "step": 71660
+    },
+    {
+      "epoch": 0.6571009443476666,
+      "learning_rate": 6.858164481525626e-05,
+      "loss": 0.6996,
+      "mean_token_accuracy": 0.7813877284526825,
+      "num_tokens": 168162714.0,
+      "step": 71670
+    },
+    {
+      "epoch": 0.6571926285871459,
+      "learning_rate": 6.856330796736041e-05,
+      "loss": 0.6939,
+      "mean_token_accuracy": 0.7868259906768799,
+      "num_tokens": 168186991.0,
+      "step": 71680
+    },
+    {
+      "epoch": 0.6572843128266251,
+      "learning_rate": 6.854497111946456e-05,
+      "loss": 0.7375,
+      "mean_token_accuracy": 0.779875522851944,
+      "num_tokens": 168212366.0,
+      "step": 71690
+    },
+    {
+      "epoch": 0.6573759970661044,
+      "learning_rate": 6.852663427156873e-05,
+      "loss": 0.6823,
+      "mean_token_accuracy": 0.785092294216156,
+      "num_tokens": 168237763.0,
+      "step": 71700
+    },
+    {
+      "epoch": 0.6574676813055835,
+      "learning_rate": 6.850829742367288e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.7955116391181946,
+      "num_tokens": 168262788.0,
+      "step": 71710
+    },
+    {
+      "epoch": 0.6575593655450628,
+      "learning_rate": 6.848996057577703e-05,
+      "loss": 0.656,
+      "mean_token_accuracy": 0.7953277707099915,
+      "num_tokens": 168287893.0,
+      "step": 71720
+    },
+    {
+      "epoch": 0.657651049784542,
+      "learning_rate": 6.847162372788118e-05,
+      "loss": 0.6541,
+      "mean_token_accuracy": 0.8006919801235199,
+      "num_tokens": 168313793.0,
+      "step": 71730
+    },
+    {
+      "epoch": 0.6577427340240213,
+      "learning_rate": 6.845328687998533e-05,
+      "loss": 0.6694,
+      "mean_token_accuracy": 0.7931058228015899,
+      "num_tokens": 168339033.0,
+      "step": 71740
+    },
+    {
+      "epoch": 0.6578344182635005,
+      "learning_rate": 6.843495003208949e-05,
+      "loss": 0.6912,
+      "mean_token_accuracy": 0.7899622797966004,
+      "num_tokens": 168363352.0,
+      "step": 71750
+    },
+    {
+      "epoch": 0.6579261025029798,
+      "learning_rate": 6.841661318419365e-05,
+      "loss": 0.7138,
+      "mean_token_accuracy": 0.7756168782711029,
+      "num_tokens": 168389113.0,
+      "step": 71760
+    },
+    {
+      "epoch": 0.6580177867424589,
+      "learning_rate": 6.839827633629779e-05,
+      "loss": 0.6981,
+      "mean_token_accuracy": 0.7802573800086975,
+      "num_tokens": 168414002.0,
+      "step": 71770
+    },
+    {
+      "epoch": 0.6581094709819382,
+      "learning_rate": 6.837993948840194e-05,
+      "loss": 0.6902,
+      "mean_token_accuracy": 0.78469517827034,
+      "num_tokens": 168439569.0,
+      "step": 71780
+    },
+    {
+      "epoch": 0.6582011552214174,
+      "learning_rate": 6.836160264050609e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.7881043136119843,
+      "num_tokens": 168464214.0,
+      "step": 71790
+    },
+    {
+      "epoch": 0.6582928394608967,
+      "learning_rate": 6.834326579261026e-05,
+      "loss": 0.7062,
+      "mean_token_accuracy": 0.7843380331993103,
+      "num_tokens": 168489506.0,
+      "step": 71800
+    },
+    {
+      "epoch": 0.6583845237003759,
+      "learning_rate": 6.832492894471441e-05,
+      "loss": 0.6582,
+      "mean_token_accuracy": 0.7941269040107727,
+      "num_tokens": 168514524.0,
+      "step": 71810
+    },
+    {
+      "epoch": 0.6584762079398552,
+      "learning_rate": 6.830659209681856e-05,
+      "loss": 0.6647,
+      "mean_token_accuracy": 0.7903441488742828,
+      "num_tokens": 168539982.0,
+      "step": 71820
+    },
+    {
+      "epoch": 0.6585678921793344,
+      "learning_rate": 6.828825524892271e-05,
+      "loss": 0.7085,
+      "mean_token_accuracy": 0.778989827632904,
+      "num_tokens": 168564877.0,
+      "step": 71830
+    },
+    {
+      "epoch": 0.6586595764188136,
+      "learning_rate": 6.826991840102687e-05,
+      "loss": 0.6232,
+      "mean_token_accuracy": 0.8072134256362915,
+      "num_tokens": 168588968.0,
+      "step": 71840
+    },
+    {
+      "epoch": 0.6587512606582928,
+      "learning_rate": 6.825158155313102e-05,
+      "loss": 0.6653,
+      "mean_token_accuracy": 0.7958389580249786,
+      "num_tokens": 168613877.0,
+      "step": 71850
+    },
+    {
+      "epoch": 0.6588429448977721,
+      "learning_rate": 6.823324470523518e-05,
+      "loss": 0.6722,
+      "mean_token_accuracy": 0.7907417356967926,
+      "num_tokens": 168638901.0,
+      "step": 71860
+    },
+    {
+      "epoch": 0.6589346291372513,
+      "learning_rate": 6.821490785733932e-05,
+      "loss": 0.7133,
+      "mean_token_accuracy": 0.7794717311859131,
+      "num_tokens": 168663687.0,
+      "step": 71870
+    },
+    {
+      "epoch": 0.6590263133767306,
+      "learning_rate": 6.819657100944347e-05,
+      "loss": 0.6796,
+      "mean_token_accuracy": 0.7891310036182404,
+      "num_tokens": 168688512.0,
+      "step": 71880
+    },
+    {
+      "epoch": 0.6591179976162098,
+      "learning_rate": 6.817823416154764e-05,
+      "loss": 0.733,
+      "mean_token_accuracy": 0.7727105498313904,
+      "num_tokens": 168713530.0,
+      "step": 71890
+    },
+    {
+      "epoch": 0.659209681855689,
+      "learning_rate": 6.815989731365179e-05,
+      "loss": 0.6781,
+      "mean_token_accuracy": 0.7864547491073608,
+      "num_tokens": 168738377.0,
+      "step": 71900
+    },
+    {
+      "epoch": 0.6593013660951682,
+      "learning_rate": 6.814156046575594e-05,
+      "loss": 0.6815,
+      "mean_token_accuracy": 0.7860010683536529,
+      "num_tokens": 168762927.0,
+      "step": 71910
+    },
+    {
+      "epoch": 0.6593930503346475,
+      "learning_rate": 6.81232236178601e-05,
+      "loss": 0.6738,
+      "mean_token_accuracy": 0.7960246980190278,
+      "num_tokens": 168787599.0,
+      "step": 71920
+    },
+    {
+      "epoch": 0.6594847345741267,
+      "learning_rate": 6.810488676996425e-05,
+      "loss": 0.7147,
+      "mean_token_accuracy": 0.783091139793396,
+      "num_tokens": 168811616.0,
+      "step": 71930
+    },
+    {
+      "epoch": 0.659576418813606,
+      "learning_rate": 6.80865499220684e-05,
+      "loss": 0.6475,
+      "mean_token_accuracy": 0.798850291967392,
+      "num_tokens": 168836155.0,
+      "step": 71940
+    },
+    {
+      "epoch": 0.6596681030530852,
+      "learning_rate": 6.806821307417255e-05,
+      "loss": 0.694,
+      "mean_token_accuracy": 0.7888308942317963,
+      "num_tokens": 168861533.0,
+      "step": 71950
+    },
+    {
+      "epoch": 0.6597597872925645,
+      "learning_rate": 6.804987622627672e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.7865898370742798,
+      "num_tokens": 168887033.0,
+      "step": 71960
+    },
+    {
+      "epoch": 0.6598514715320436,
+      "learning_rate": 6.803153937838085e-05,
+      "loss": 0.6806,
+      "mean_token_accuracy": 0.790938937664032,
+      "num_tokens": 168912425.0,
+      "step": 71970
+    },
+    {
+      "epoch": 0.6599431557715228,
+      "learning_rate": 6.8013202530485e-05,
+      "loss": 0.673,
+      "mean_token_accuracy": 0.7866240203380584,
+      "num_tokens": 168937504.0,
+      "step": 71980
+    },
+    {
+      "epoch": 0.6600348400110021,
+      "learning_rate": 6.799486568258917e-05,
+      "loss": 0.6705,
+      "mean_token_accuracy": 0.7928176879882812,
+      "num_tokens": 168962648.0,
+      "step": 71990
+    },
+    {
+      "epoch": 0.6601265242504814,
+      "learning_rate": 6.797652883469332e-05,
+      "loss": 0.6746,
+      "mean_token_accuracy": 0.790187019109726,
+      "num_tokens": 168987769.0,
+      "step": 72000
+    },
+    {
+      "epoch": 0.6602182084899606,
+      "learning_rate": 6.795819198679747e-05,
+      "loss": 0.6992,
+      "mean_token_accuracy": 0.786000233888626,
+      "num_tokens": 169012666.0,
+      "step": 72010
+    },
+    {
+      "epoch": 0.6603098927294399,
+      "learning_rate": 6.793985513890163e-05,
+      "loss": 0.6732,
+      "mean_token_accuracy": 0.7908929944038391,
+      "num_tokens": 169038237.0,
+      "step": 72020
+    },
+    {
+      "epoch": 0.660401576968919,
+      "learning_rate": 6.792151829100578e-05,
+      "loss": 0.6584,
+      "mean_token_accuracy": 0.7980163812637329,
+      "num_tokens": 169064039.0,
+      "step": 72030
+    },
+    {
+      "epoch": 0.6604932612083982,
+      "learning_rate": 6.790318144310993e-05,
+      "loss": 0.671,
+      "mean_token_accuracy": 0.7938000857830048,
+      "num_tokens": 169088470.0,
+      "step": 72040
+    },
+    {
+      "epoch": 0.6605849454478775,
+      "learning_rate": 6.788484459521408e-05,
+      "loss": 0.6717,
+      "mean_token_accuracy": 0.7914857923984527,
+      "num_tokens": 169114372.0,
+      "step": 72050
+    },
+    {
+      "epoch": 0.6606766296873567,
+      "learning_rate": 6.786650774731825e-05,
+      "loss": 0.6677,
+      "mean_token_accuracy": 0.7910528302192688,
+      "num_tokens": 169139429.0,
+      "step": 72060
+    },
+    {
+      "epoch": 0.660768313926836,
+      "learning_rate": 6.784817089942239e-05,
+      "loss": 0.7079,
+      "mean_token_accuracy": 0.7883346676826477,
+      "num_tokens": 169164013.0,
+      "step": 72070
+    },
+    {
+      "epoch": 0.6608599981663152,
+      "learning_rate": 6.782983405152654e-05,
+      "loss": 0.6556,
+      "mean_token_accuracy": 0.7989653050899506,
+      "num_tokens": 169188963.0,
+      "step": 72080
+    },
+    {
+      "epoch": 0.6609516824057945,
+      "learning_rate": 6.78114972036307e-05,
+      "loss": 0.735,
+      "mean_token_accuracy": 0.7789423286914825,
+      "num_tokens": 169214278.0,
+      "step": 72090
+    },
+    {
+      "epoch": 0.6610433666452736,
+      "learning_rate": 6.779316035573485e-05,
+      "loss": 0.6858,
+      "mean_token_accuracy": 0.7885903954505921,
+      "num_tokens": 169238598.0,
+      "step": 72100
+    },
+    {
+      "epoch": 0.6611350508847529,
+      "learning_rate": 6.7774823507839e-05,
+      "loss": 0.6708,
+      "mean_token_accuracy": 0.7963763773441315,
+      "num_tokens": 169263385.0,
+      "step": 72110
+    },
+    {
+      "epoch": 0.6612267351242321,
+      "learning_rate": 6.775648665994316e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.7952436149120331,
+      "num_tokens": 169288176.0,
+      "step": 72120
+    },
+    {
+      "epoch": 0.6613184193637114,
+      "learning_rate": 6.773814981204731e-05,
+      "loss": 0.6525,
+      "mean_token_accuracy": 0.7992342174053192,
+      "num_tokens": 169313511.0,
+      "step": 72130
+    },
+    {
+      "epoch": 0.6614101036031906,
+      "learning_rate": 6.771981296415146e-05,
+      "loss": 0.6424,
+      "mean_token_accuracy": 0.8027814626693726,
+      "num_tokens": 169338622.0,
+      "step": 72140
+    },
+    {
+      "epoch": 0.6615017878426699,
+      "learning_rate": 6.770147611625563e-05,
+      "loss": 0.7212,
+      "mean_token_accuracy": 0.7763095557689667,
+      "num_tokens": 169364650.0,
+      "step": 72150
+    },
+    {
+      "epoch": 0.661593472082149,
+      "learning_rate": 6.768313926835978e-05,
+      "loss": 0.6764,
+      "mean_token_accuracy": 0.7919943451881408,
+      "num_tokens": 169390729.0,
+      "step": 72160
+    },
+    {
+      "epoch": 0.6616851563216283,
+      "learning_rate": 6.766480242046392e-05,
+      "loss": 0.7052,
+      "mean_token_accuracy": 0.782336163520813,
+      "num_tokens": 169415761.0,
+      "step": 72170
+    },
+    {
+      "epoch": 0.6617768405611075,
+      "learning_rate": 6.764646557256807e-05,
+      "loss": 0.6943,
+      "mean_token_accuracy": 0.7865187466144562,
+      "num_tokens": 169441300.0,
+      "step": 72180
+    },
+    {
+      "epoch": 0.6618685248005868,
+      "learning_rate": 6.762812872467224e-05,
+      "loss": 0.6811,
+      "mean_token_accuracy": 0.7896221995353698,
+      "num_tokens": 169466401.0,
+      "step": 72190
+    },
+    {
+      "epoch": 0.661960209040066,
+      "learning_rate": 6.760979187677639e-05,
+      "loss": 0.6797,
+      "mean_token_accuracy": 0.7934462070465088,
+      "num_tokens": 169490490.0,
+      "step": 72200
+    },
+    {
+      "epoch": 0.6620518932795453,
+      "learning_rate": 6.759145502888054e-05,
+      "loss": 0.6536,
+      "mean_token_accuracy": 0.7974579572677613,
+      "num_tokens": 169515795.0,
+      "step": 72210
+    },
+    {
+      "epoch": 0.6621435775190245,
+      "learning_rate": 6.757311818098469e-05,
+      "loss": 0.6828,
+      "mean_token_accuracy": 0.7918470799922943,
+      "num_tokens": 169541617.0,
+      "step": 72220
+    },
+    {
+      "epoch": 0.6622352617585037,
+      "learning_rate": 6.755478133308884e-05,
+      "loss": 0.6789,
+      "mean_token_accuracy": 0.79769948720932,
+      "num_tokens": 169566238.0,
+      "step": 72230
+    },
+    {
+      "epoch": 0.6623269459979829,
+      "learning_rate": 6.7536444485193e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.7852485775947571,
+      "num_tokens": 169590963.0,
+      "step": 72240
+    },
+    {
+      "epoch": 0.6624186302374622,
+      "learning_rate": 6.751810763729716e-05,
+      "loss": 0.6587,
+      "mean_token_accuracy": 0.7933691024780274,
+      "num_tokens": 169615570.0,
+      "step": 72250
+    },
+    {
+      "epoch": 0.6625103144769414,
+      "learning_rate": 6.749977078940131e-05,
+      "loss": 0.6882,
+      "mean_token_accuracy": 0.7925223767757416,
+      "num_tokens": 169641329.0,
+      "step": 72260
+    },
+    {
+      "epoch": 0.6626019987164207,
+      "learning_rate": 6.748143394150545e-05,
+      "loss": 0.6557,
+      "mean_token_accuracy": 0.7946510136127471,
+      "num_tokens": 169666528.0,
+      "step": 72270
+    },
+    {
+      "epoch": 0.6626936829558999,
+      "learning_rate": 6.74630970936096e-05,
+      "loss": 0.7183,
+      "mean_token_accuracy": 0.7837998986244201,
+      "num_tokens": 169691669.0,
+      "step": 72280
+    },
+    {
+      "epoch": 0.6627853671953791,
+      "learning_rate": 6.744476024571377e-05,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.7890930533409118,
+      "num_tokens": 169716903.0,
+      "step": 72290
+    },
+    {
+      "epoch": 0.6628770514348583,
+      "learning_rate": 6.742642339781792e-05,
+      "loss": 0.7106,
+      "mean_token_accuracy": 0.782125735282898,
+      "num_tokens": 169741843.0,
+      "step": 72300
+    },
+    {
+      "epoch": 0.6629687356743376,
+      "learning_rate": 6.740808654992207e-05,
+      "loss": 0.6576,
+      "mean_token_accuracy": 0.7944453895092011,
+      "num_tokens": 169766789.0,
+      "step": 72310
+    },
+    {
+      "epoch": 0.6630604199138168,
+      "learning_rate": 6.738974970202622e-05,
+      "loss": 0.6641,
+      "mean_token_accuracy": 0.7893314301967621,
+      "num_tokens": 169792335.0,
+      "step": 72320
+    },
+    {
+      "epoch": 0.6631521041532961,
+      "learning_rate": 6.737141285413037e-05,
+      "loss": 0.6825,
+      "mean_token_accuracy": 0.7878741383552551,
+      "num_tokens": 169816737.0,
+      "step": 72330
+    },
+    {
+      "epoch": 0.6632437883927753,
+      "learning_rate": 6.735307600623453e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.7936313450336456,
+      "num_tokens": 169841544.0,
+      "step": 72340
+    },
+    {
+      "epoch": 0.6633354726322546,
+      "learning_rate": 6.733473915833869e-05,
+      "loss": 0.6719,
+      "mean_token_accuracy": 0.7930039465427399,
+      "num_tokens": 169866904.0,
+      "step": 72350
+    },
+    {
+      "epoch": 0.6634271568717337,
+      "learning_rate": 6.731640231044284e-05,
+      "loss": 0.6814,
+      "mean_token_accuracy": 0.780381029844284,
+      "num_tokens": 169891983.0,
+      "step": 72360
+    },
+    {
+      "epoch": 0.663518841111213,
+      "learning_rate": 6.729806546254698e-05,
+      "loss": 0.7025,
+      "mean_token_accuracy": 0.7849653840065003,
+      "num_tokens": 169916888.0,
+      "step": 72370
+    },
+    {
+      "epoch": 0.6636105253506922,
+      "learning_rate": 6.727972861465115e-05,
+      "loss": 0.671,
+      "mean_token_accuracy": 0.7922280192375183,
+      "num_tokens": 169942786.0,
+      "step": 72380
+    },
+    {
+      "epoch": 0.6637022095901715,
+      "learning_rate": 6.72613917667553e-05,
+      "loss": 0.6817,
+      "mean_token_accuracy": 0.7830262303352356,
+      "num_tokens": 169967689.0,
+      "step": 72390
+    },
+    {
+      "epoch": 0.6637938938296507,
+      "learning_rate": 6.724305491885945e-05,
+      "loss": 0.7013,
+      "mean_token_accuracy": 0.7871264457702637,
+      "num_tokens": 169992642.0,
+      "step": 72400
+    },
+    {
+      "epoch": 0.66388557806913,
+      "learning_rate": 6.72247180709636e-05,
+      "loss": 0.6718,
+      "mean_token_accuracy": 0.7969866991043091,
+      "num_tokens": 170018387.0,
+      "step": 72410
+    },
+    {
+      "epoch": 0.6639772623086091,
+      "learning_rate": 6.720638122306776e-05,
+      "loss": 0.6928,
+      "mean_token_accuracy": 0.786678147315979,
+      "num_tokens": 170043542.0,
+      "step": 72420
+    },
+    {
+      "epoch": 0.6640689465480883,
+      "learning_rate": 6.718804437517191e-05,
+      "loss": 0.684,
+      "mean_token_accuracy": 0.7954052209854126,
+      "num_tokens": 170069024.0,
+      "step": 72430
+    },
+    {
+      "epoch": 0.6641606307875676,
+      "learning_rate": 6.716970752727606e-05,
+      "loss": 0.7049,
+      "mean_token_accuracy": 0.7860666513442993,
+      "num_tokens": 170094224.0,
+      "step": 72440
+    },
+    {
+      "epoch": 0.6642523150270468,
+      "learning_rate": 6.715137067938022e-05,
+      "loss": 0.6566,
+      "mean_token_accuracy": 0.7930137574672699,
+      "num_tokens": 170119561.0,
+      "step": 72450
+    },
+    {
+      "epoch": 0.6643439992665261,
+      "learning_rate": 6.713303383148438e-05,
+      "loss": 0.6375,
+      "mean_token_accuracy": 0.8034508168697357,
+      "num_tokens": 170144527.0,
+      "step": 72460
+    },
+    {
+      "epoch": 0.6644356835060053,
+      "learning_rate": 6.711469698358851e-05,
+      "loss": 0.6943,
+      "mean_token_accuracy": 0.7905678987503052,
+      "num_tokens": 170169450.0,
+      "step": 72470
+    },
+    {
+      "epoch": 0.6645273677454846,
+      "learning_rate": 6.709636013569268e-05,
+      "loss": 0.6599,
+      "mean_token_accuracy": 0.7895504474639893,
+      "num_tokens": 170194966.0,
+      "step": 72480
+    },
+    {
+      "epoch": 0.6646190519849637,
+      "learning_rate": 6.707802328779683e-05,
+      "loss": 0.6562,
+      "mean_token_accuracy": 0.7965170204639435,
+      "num_tokens": 170220289.0,
+      "step": 72490
+    },
+    {
+      "epoch": 0.664710736224443,
+      "learning_rate": 6.705968643990098e-05,
+      "loss": 0.6629,
+      "mean_token_accuracy": 0.7979708194732666,
+      "num_tokens": 170245545.0,
+      "step": 72500
+    },
+    {
+      "epoch": 0.6648024204639222,
+      "learning_rate": 6.704134959200515e-05,
+      "loss": 0.68,
+      "mean_token_accuracy": 0.7887271881103516,
+      "num_tokens": 170270007.0,
+      "step": 72510
+    },
+    {
+      "epoch": 0.6648941047034015,
+      "learning_rate": 6.702301274410929e-05,
+      "loss": 0.6495,
+      "mean_token_accuracy": 0.793360835313797,
+      "num_tokens": 170294936.0,
+      "step": 72520
+    },
+    {
+      "epoch": 0.6649857889428807,
+      "learning_rate": 6.700467589621344e-05,
+      "loss": 0.6705,
+      "mean_token_accuracy": 0.7929691433906555,
+      "num_tokens": 170320028.0,
+      "step": 72530
+    },
+    {
+      "epoch": 0.66507747318236,
+      "learning_rate": 6.698633904831759e-05,
+      "loss": 0.6714,
+      "mean_token_accuracy": 0.7888080418109894,
+      "num_tokens": 170344859.0,
+      "step": 72540
+    },
+    {
+      "epoch": 0.6651691574218392,
+      "learning_rate": 6.696800220042176e-05,
+      "loss": 0.695,
+      "mean_token_accuracy": 0.7836909532546997,
+      "num_tokens": 170369640.0,
+      "step": 72550
+    },
+    {
+      "epoch": 0.6652608416613184,
+      "learning_rate": 6.694966535252591e-05,
+      "loss": 0.6606,
+      "mean_token_accuracy": 0.7975388288497924,
+      "num_tokens": 170394568.0,
+      "step": 72560
+    },
+    {
+      "epoch": 0.6653525259007976,
+      "learning_rate": 6.693132850463005e-05,
+      "loss": 0.707,
+      "mean_token_accuracy": 0.7804718494415284,
+      "num_tokens": 170419414.0,
+      "step": 72570
+    },
+    {
+      "epoch": 0.6654442101402769,
+      "learning_rate": 6.691299165673421e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.7886490046977996,
+      "num_tokens": 170444401.0,
+      "step": 72580
+    },
+    {
+      "epoch": 0.6655358943797561,
+      "learning_rate": 6.689465480883836e-05,
+      "loss": 0.7255,
+      "mean_token_accuracy": 0.7776539385318756,
+      "num_tokens": 170469876.0,
+      "step": 72590
+    },
+    {
+      "epoch": 0.6656275786192354,
+      "learning_rate": 6.687631796094252e-05,
+      "loss": 0.6542,
+      "mean_token_accuracy": 0.7902257859706878,
+      "num_tokens": 170493939.0,
+      "step": 72600
+    },
+    {
+      "epoch": 0.6657192628587146,
+      "learning_rate": 6.685798111304668e-05,
+      "loss": 0.6736,
+      "mean_token_accuracy": 0.7904944241046905,
+      "num_tokens": 170518678.0,
+      "step": 72610
+    },
+    {
+      "epoch": 0.6658109470981938,
+      "learning_rate": 6.683964426515082e-05,
+      "loss": 0.6942,
+      "mean_token_accuracy": 0.7855430006980896,
+      "num_tokens": 170544146.0,
+      "step": 72620
+    },
+    {
+      "epoch": 0.665902631337673,
+      "learning_rate": 6.682130741725497e-05,
+      "loss": 0.702,
+      "mean_token_accuracy": 0.784555172920227,
+      "num_tokens": 170569430.0,
+      "step": 72630
+    },
+    {
+      "epoch": 0.6659943155771523,
+      "learning_rate": 6.680297056935914e-05,
+      "loss": 0.6786,
+      "mean_token_accuracy": 0.7891363143920899,
+      "num_tokens": 170594523.0,
+      "step": 72640
+    },
+    {
+      "epoch": 0.6660859998166315,
+      "learning_rate": 6.678463372146329e-05,
+      "loss": 0.6533,
+      "mean_token_accuracy": 0.7904845714569092,
+      "num_tokens": 170619528.0,
+      "step": 72650
+    },
+    {
+      "epoch": 0.6661776840561108,
+      "learning_rate": 6.676629687356744e-05,
+      "loss": 0.6692,
+      "mean_token_accuracy": 0.7959149599075317,
+      "num_tokens": 170643898.0,
+      "step": 72660
+    },
+    {
+      "epoch": 0.66626936829559,
+      "learning_rate": 6.674796002567158e-05,
+      "loss": 0.6944,
+      "mean_token_accuracy": 0.7878892064094544,
+      "num_tokens": 170669116.0,
+      "step": 72670
+    },
+    {
+      "epoch": 0.6663610525350693,
+      "learning_rate": 6.672962317777574e-05,
+      "loss": 0.6694,
+      "mean_token_accuracy": 0.783225154876709,
+      "num_tokens": 170694052.0,
+      "step": 72680
+    },
+    {
+      "epoch": 0.6664527367745484,
+      "learning_rate": 6.67112863298799e-05,
+      "loss": 0.6842,
+      "mean_token_accuracy": 0.7891632854938507,
+      "num_tokens": 170719295.0,
+      "step": 72690
+    },
+    {
+      "epoch": 0.6665444210140277,
+      "learning_rate": 6.669294948198405e-05,
+      "loss": 0.6794,
+      "mean_token_accuracy": 0.7890183508396149,
+      "num_tokens": 170743973.0,
+      "step": 72700
+    },
+    {
+      "epoch": 0.6666361052535069,
+      "learning_rate": 6.667461263408821e-05,
+      "loss": 0.6688,
+      "mean_token_accuracy": 0.7928049027919769,
+      "num_tokens": 170770263.0,
+      "step": 72710
+    },
+    {
+      "epoch": 0.6667277894929862,
+      "learning_rate": 6.665627578619235e-05,
+      "loss": 0.6981,
+      "mean_token_accuracy": 0.7866098165512085,
+      "num_tokens": 170795734.0,
+      "step": 72720
+    },
+    {
+      "epoch": 0.6668194737324654,
+      "learning_rate": 6.66379389382965e-05,
+      "loss": 0.6872,
+      "mean_token_accuracy": 0.7820593237876892,
+      "num_tokens": 170821340.0,
+      "step": 72730
+    },
+    {
+      "epoch": 0.6669111579719447,
+      "learning_rate": 6.661960209040067e-05,
+      "loss": 0.6711,
+      "mean_token_accuracy": 0.7926416516304016,
+      "num_tokens": 170846356.0,
+      "step": 72740
+    },
+    {
+      "epoch": 0.6670028422114238,
+      "learning_rate": 6.660126524250482e-05,
+      "loss": 0.6996,
+      "mean_token_accuracy": 0.7878145575523376,
+      "num_tokens": 170872345.0,
+      "step": 72750
+    },
+    {
+      "epoch": 0.6670945264509031,
+      "learning_rate": 6.658292839460897e-05,
+      "loss": 0.7001,
+      "mean_token_accuracy": 0.7886099994182587,
+      "num_tokens": 170897693.0,
+      "step": 72760
+    },
+    {
+      "epoch": 0.6671862106903823,
+      "learning_rate": 6.656459154671312e-05,
+      "loss": 0.6685,
+      "mean_token_accuracy": 0.7847908437252045,
+      "num_tokens": 170922159.0,
+      "step": 72770
+    },
+    {
+      "epoch": 0.6672778949298616,
+      "learning_rate": 6.654625469881728e-05,
+      "loss": 0.6428,
+      "mean_token_accuracy": 0.8001977264881134,
+      "num_tokens": 170947580.0,
+      "step": 72780
+    },
+    {
+      "epoch": 0.6673695791693408,
+      "learning_rate": 6.652791785092143e-05,
+      "loss": 0.6993,
+      "mean_token_accuracy": 0.782284551858902,
+      "num_tokens": 170972474.0,
+      "step": 72790
+    },
+    {
+      "epoch": 0.6674612634088201,
+      "learning_rate": 6.650958100302558e-05,
+      "loss": 0.6919,
+      "mean_token_accuracy": 0.7873977839946746,
+      "num_tokens": 170997899.0,
+      "step": 72800
+    },
+    {
+      "epoch": 0.6675529476482993,
+      "learning_rate": 6.649124415512975e-05,
+      "loss": 0.6935,
+      "mean_token_accuracy": 0.7870490491390228,
+      "num_tokens": 171023387.0,
+      "step": 72810
+    },
+    {
+      "epoch": 0.6676446318877785,
+      "learning_rate": 6.647290730723388e-05,
+      "loss": 0.6304,
+      "mean_token_accuracy": 0.7952752888202668,
+      "num_tokens": 171048621.0,
+      "step": 72820
+    },
+    {
+      "epoch": 0.6677363161272577,
+      "learning_rate": 6.645457045933804e-05,
+      "loss": 0.7084,
+      "mean_token_accuracy": 0.7862338602542878,
+      "num_tokens": 171073649.0,
+      "step": 72830
+    },
+    {
+      "epoch": 0.667828000366737,
+      "learning_rate": 6.64362336114422e-05,
+      "loss": 0.7063,
+      "mean_token_accuracy": 0.7843965888023376,
+      "num_tokens": 171098166.0,
+      "step": 72840
+    },
+    {
+      "epoch": 0.6679196846062162,
+      "learning_rate": 6.641789676354635e-05,
+      "loss": 0.6897,
+      "mean_token_accuracy": 0.78678657412529,
+      "num_tokens": 171123799.0,
+      "step": 72850
+    },
+    {
+      "epoch": 0.6680113688456955,
+      "learning_rate": 6.63995599156505e-05,
+      "loss": 0.71,
+      "mean_token_accuracy": 0.7828811645507813,
+      "num_tokens": 171148727.0,
+      "step": 72860
+    },
+    {
+      "epoch": 0.6681030530851747,
+      "learning_rate": 6.638122306775466e-05,
+      "loss": 0.6397,
+      "mean_token_accuracy": 0.7992956578731537,
+      "num_tokens": 171174453.0,
+      "step": 72870
+    },
+    {
+      "epoch": 0.6681947373246538,
+      "learning_rate": 6.636288621985881e-05,
+      "loss": 0.7121,
+      "mean_token_accuracy": 0.7838517189025879,
+      "num_tokens": 171199307.0,
+      "step": 72880
+    },
+    {
+      "epoch": 0.6682864215641331,
+      "learning_rate": 6.634454937196296e-05,
+      "loss": 0.6759,
+      "mean_token_accuracy": 0.7882214546203613,
+      "num_tokens": 171223982.0,
+      "step": 72890
+    },
+    {
+      "epoch": 0.6683781058036123,
+      "learning_rate": 6.632621252406711e-05,
+      "loss": 0.6526,
+      "mean_token_accuracy": 0.7965757131576539,
+      "num_tokens": 171249323.0,
+      "step": 72900
+    },
+    {
+      "epoch": 0.6684697900430916,
+      "learning_rate": 6.630787567617128e-05,
+      "loss": 0.6633,
+      "mean_token_accuracy": 0.799387526512146,
+      "num_tokens": 171275064.0,
+      "step": 72910
+    },
+    {
+      "epoch": 0.6685614742825708,
+      "learning_rate": 6.628953882827542e-05,
+      "loss": 0.7091,
+      "mean_token_accuracy": 0.7831669270992279,
+      "num_tokens": 171300116.0,
+      "step": 72920
+    },
+    {
+      "epoch": 0.6686531585220501,
+      "learning_rate": 6.627120198037957e-05,
+      "loss": 0.71,
+      "mean_token_accuracy": 0.7740885138511657,
+      "num_tokens": 171325405.0,
+      "step": 72930
+    },
+    {
+      "epoch": 0.6687448427615293,
+      "learning_rate": 6.625286513248373e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.7902782261371613,
+      "num_tokens": 171349948.0,
+      "step": 72940
+    },
+    {
+      "epoch": 0.6688365270010085,
+      "learning_rate": 6.623452828458789e-05,
+      "loss": 0.7185,
+      "mean_token_accuracy": 0.7837597966194153,
+      "num_tokens": 171375229.0,
+      "step": 72950
+    },
+    {
+      "epoch": 0.6689282112404877,
+      "learning_rate": 6.621619143669204e-05,
+      "loss": 0.6605,
+      "mean_token_accuracy": 0.790293139219284,
+      "num_tokens": 171400530.0,
+      "step": 72960
+    },
+    {
+      "epoch": 0.669019895479967,
+      "learning_rate": 6.619785458879619e-05,
+      "loss": 0.6556,
+      "mean_token_accuracy": 0.7881445765495301,
+      "num_tokens": 171425526.0,
+      "step": 72970
+    },
+    {
+      "epoch": 0.6691115797194462,
+      "learning_rate": 6.617951774090034e-05,
+      "loss": 0.6922,
+      "mean_token_accuracy": 0.7873681724071503,
+      "num_tokens": 171450600.0,
+      "step": 72980
+    },
+    {
+      "epoch": 0.6692032639589255,
+      "learning_rate": 6.616118089300449e-05,
+      "loss": 0.6467,
+      "mean_token_accuracy": 0.7983907639980317,
+      "num_tokens": 171475655.0,
+      "step": 72990
+    },
+    {
+      "epoch": 0.6692949481984047,
+      "learning_rate": 6.614284404510866e-05,
+      "loss": 0.6668,
+      "mean_token_accuracy": 0.795628833770752,
+      "num_tokens": 171500945.0,
+      "step": 73000
+    },
+    {
+      "epoch": 0.6693866324378839,
+      "learning_rate": 6.612450719721281e-05,
+      "loss": 0.7142,
+      "mean_token_accuracy": 0.7879930019378663,
+      "num_tokens": 171525963.0,
+      "step": 73010
+    },
+    {
+      "epoch": 0.6694783166773631,
+      "learning_rate": 6.610617034931695e-05,
+      "loss": 0.7424,
+      "mean_token_accuracy": 0.778604942560196,
+      "num_tokens": 171551319.0,
+      "step": 73020
+    },
+    {
+      "epoch": 0.6695700009168424,
+      "learning_rate": 6.60878335014211e-05,
+      "loss": 0.7064,
+      "mean_token_accuracy": 0.7856544494628906,
+      "num_tokens": 171577097.0,
+      "step": 73030
+    },
+    {
+      "epoch": 0.6696616851563216,
+      "learning_rate": 6.606949665352527e-05,
+      "loss": 0.6927,
+      "mean_token_accuracy": 0.7791128635406495,
+      "num_tokens": 171601450.0,
+      "step": 73040
+    },
+    {
+      "epoch": 0.6697533693958009,
+      "learning_rate": 6.605115980562942e-05,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.7919116795063019,
+      "num_tokens": 171626694.0,
+      "step": 73050
+    },
+    {
+      "epoch": 0.6698450536352801,
+      "learning_rate": 6.603282295773357e-05,
+      "loss": 0.6766,
+      "mean_token_accuracy": 0.7944331347942353,
+      "num_tokens": 171651962.0,
+      "step": 73060
+    },
+    {
+      "epoch": 0.6699367378747594,
+      "learning_rate": 6.601448610983772e-05,
+      "loss": 0.7054,
+      "mean_token_accuracy": 0.7812461495399475,
+      "num_tokens": 171676244.0,
+      "step": 73070
+    },
+    {
+      "epoch": 0.6700284221142385,
+      "learning_rate": 6.599614926194187e-05,
+      "loss": 0.654,
+      "mean_token_accuracy": 0.7955682694911956,
+      "num_tokens": 171703177.0,
+      "step": 73080
+    },
+    {
+      "epoch": 0.6701201063537178,
+      "learning_rate": 6.597781241404603e-05,
+      "loss": 0.68,
+      "mean_token_accuracy": 0.7879145562648773,
+      "num_tokens": 171728410.0,
+      "step": 73090
+    },
+    {
+      "epoch": 0.670211790593197,
+      "learning_rate": 6.595947556615019e-05,
+      "loss": 0.7112,
+      "mean_token_accuracy": 0.7802545547485351,
+      "num_tokens": 171753339.0,
+      "step": 73100
+    },
+    {
+      "epoch": 0.6703034748326763,
+      "learning_rate": 6.594113871825434e-05,
+      "loss": 0.6859,
+      "mean_token_accuracy": 0.7883873760700226,
+      "num_tokens": 171778954.0,
+      "step": 73110
+    },
+    {
+      "epoch": 0.6703951590721555,
+      "learning_rate": 6.592280187035848e-05,
+      "loss": 0.6746,
+      "mean_token_accuracy": 0.7894088745117187,
+      "num_tokens": 171803285.0,
+      "step": 73120
+    },
+    {
+      "epoch": 0.6704868433116348,
+      "learning_rate": 6.590446502246265e-05,
+      "loss": 0.6615,
+      "mean_token_accuracy": 0.7912877678871155,
+      "num_tokens": 171828691.0,
+      "step": 73130
+    },
+    {
+      "epoch": 0.6705785275511139,
+      "learning_rate": 6.58861281745668e-05,
+      "loss": 0.6771,
+      "mean_token_accuracy": 0.7839479625225068,
+      "num_tokens": 171853681.0,
+      "step": 73140
+    },
+    {
+      "epoch": 0.6706702117905932,
+      "learning_rate": 6.586779132667095e-05,
+      "loss": 0.6995,
+      "mean_token_accuracy": 0.7861020505428314,
+      "num_tokens": 171879410.0,
+      "step": 73150
+    },
+    {
+      "epoch": 0.6707618960300724,
+      "learning_rate": 6.58494544787751e-05,
+      "loss": 0.6972,
+      "mean_token_accuracy": 0.7826953828334808,
+      "num_tokens": 171904417.0,
+      "step": 73160
+    },
+    {
+      "epoch": 0.6708535802695517,
+      "learning_rate": 6.583111763087925e-05,
+      "loss": 0.6528,
+      "mean_token_accuracy": 0.8015499711036682,
+      "num_tokens": 171929734.0,
+      "step": 73170
+    },
+    {
+      "epoch": 0.6709452645090309,
+      "learning_rate": 6.58127807829834e-05,
+      "loss": 0.6749,
+      "mean_token_accuracy": 0.7938455700874328,
+      "num_tokens": 171954603.0,
+      "step": 73180
+    },
+    {
+      "epoch": 0.6710369487485102,
+      "learning_rate": 6.579444393508756e-05,
+      "loss": 0.6744,
+      "mean_token_accuracy": 0.7895184636116028,
+      "num_tokens": 171979922.0,
+      "step": 73190
+    },
+    {
+      "epoch": 0.6711286329879894,
+      "learning_rate": 6.577610708719172e-05,
+      "loss": 0.6739,
+      "mean_token_accuracy": 0.789273327589035,
+      "num_tokens": 172004445.0,
+      "step": 73200
+    },
+    {
+      "epoch": 0.6712203172274686,
+      "learning_rate": 6.575777023929587e-05,
+      "loss": 0.715,
+      "mean_token_accuracy": 0.7817219018936157,
+      "num_tokens": 172029100.0,
+      "step": 73210
+    },
+    {
+      "epoch": 0.6713120014669478,
+      "learning_rate": 6.573943339140001e-05,
+      "loss": 0.6669,
+      "mean_token_accuracy": 0.7843082010746002,
+      "num_tokens": 172054157.0,
+      "step": 73220
+    },
+    {
+      "epoch": 0.6714036857064271,
+      "learning_rate": 6.572109654350418e-05,
+      "loss": 0.7128,
+      "mean_token_accuracy": 0.7877836048603057,
+      "num_tokens": 172080180.0,
+      "step": 73230
+    },
+    {
+      "epoch": 0.6714953699459063,
+      "learning_rate": 6.570275969560833e-05,
+      "loss": 0.6713,
+      "mean_token_accuracy": 0.7891487896442413,
+      "num_tokens": 172105474.0,
+      "step": 73240
+    },
+    {
+      "epoch": 0.6715870541853856,
+      "learning_rate": 6.568442284771248e-05,
+      "loss": 0.6367,
+      "mean_token_accuracy": 0.8021510303020477,
+      "num_tokens": 172130652.0,
+      "step": 73250
+    },
+    {
+      "epoch": 0.6716787384248648,
+      "learning_rate": 6.566608599981663e-05,
+      "loss": 0.6515,
+      "mean_token_accuracy": 0.7935626208782196,
+      "num_tokens": 172155382.0,
+      "step": 73260
+    },
+    {
+      "epoch": 0.671770422664344,
+      "learning_rate": 6.564774915192079e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.7954720377922058,
+      "num_tokens": 172179711.0,
+      "step": 73270
+    },
+    {
+      "epoch": 0.6718621069038232,
+      "learning_rate": 6.562941230402494e-05,
+      "loss": 0.6753,
+      "mean_token_accuracy": 0.7912381947040558,
+      "num_tokens": 172204436.0,
+      "step": 73280
+    },
+    {
+      "epoch": 0.6719537911433024,
+      "learning_rate": 6.561107545612909e-05,
+      "loss": 0.6721,
+      "mean_token_accuracy": 0.7904717266559601,
+      "num_tokens": 172229578.0,
+      "step": 73290
+    },
+    {
+      "epoch": 0.6720454753827817,
+      "learning_rate": 6.559273860823326e-05,
+      "loss": 0.6599,
+      "mean_token_accuracy": 0.7967093229293823,
+      "num_tokens": 172254616.0,
+      "step": 73300
+    },
+    {
+      "epoch": 0.672137159622261,
+      "learning_rate": 6.557440176033741e-05,
+      "loss": 0.7043,
+      "mean_token_accuracy": 0.786264979839325,
+      "num_tokens": 172279921.0,
+      "step": 73310
+    },
+    {
+      "epoch": 0.6722288438617402,
+      "learning_rate": 6.555606491244155e-05,
+      "loss": 0.6793,
+      "mean_token_accuracy": 0.7897971332073211,
+      "num_tokens": 172304945.0,
+      "step": 73320
+    },
+    {
+      "epoch": 0.6723205281012195,
+      "learning_rate": 6.553772806454571e-05,
+      "loss": 0.6589,
+      "mean_token_accuracy": 0.7984269201755524,
+      "num_tokens": 172330363.0,
+      "step": 73330
+    },
+    {
+      "epoch": 0.6724122123406986,
+      "learning_rate": 6.551939121664986e-05,
+      "loss": 0.6807,
+      "mean_token_accuracy": 0.7922760248184204,
+      "num_tokens": 172356502.0,
+      "step": 73340
+    },
+    {
+      "epoch": 0.6725038965801778,
+      "learning_rate": 6.550105436875401e-05,
+      "loss": 0.7083,
+      "mean_token_accuracy": 0.7910124242305756,
+      "num_tokens": 172381830.0,
+      "step": 73350
+    },
+    {
+      "epoch": 0.6725955808196571,
+      "learning_rate": 6.548271752085817e-05,
+      "loss": 0.6998,
+      "mean_token_accuracy": 0.784187787771225,
+      "num_tokens": 172405797.0,
+      "step": 73360
+    },
+    {
+      "epoch": 0.6726872650591363,
+      "learning_rate": 6.546438067296232e-05,
+      "loss": 0.6606,
+      "mean_token_accuracy": 0.7906575322151184,
+      "num_tokens": 172430719.0,
+      "step": 73370
+    },
+    {
+      "epoch": 0.6727789492986156,
+      "learning_rate": 6.544604382506647e-05,
+      "loss": 0.6918,
+      "mean_token_accuracy": 0.7964638113975525,
+      "num_tokens": 172455765.0,
+      "step": 73380
+    },
+    {
+      "epoch": 0.6728706335380948,
+      "learning_rate": 6.542770697717064e-05,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.7809453129768371,
+      "num_tokens": 172481310.0,
+      "step": 73390
+    },
+    {
+      "epoch": 0.672962317777574,
+      "learning_rate": 6.540937012927479e-05,
+      "loss": 0.7024,
+      "mean_token_accuracy": 0.7876185357570649,
+      "num_tokens": 172506628.0,
+      "step": 73400
+    },
+    {
+      "epoch": 0.6730540020170532,
+      "learning_rate": 6.539103328137894e-05,
+      "loss": 0.703,
+      "mean_token_accuracy": 0.7867640376091003,
+      "num_tokens": 172531864.0,
+      "step": 73410
+    },
+    {
+      "epoch": 0.6731456862565325,
+      "learning_rate": 6.537269643348308e-05,
+      "loss": 0.6707,
+      "mean_token_accuracy": 0.789568567276001,
+      "num_tokens": 172556350.0,
+      "step": 73420
+    },
+    {
+      "epoch": 0.6732373704960117,
+      "learning_rate": 6.535435958558724e-05,
+      "loss": 0.6982,
+      "mean_token_accuracy": 0.7860197365283966,
+      "num_tokens": 172580987.0,
+      "step": 73430
+    },
+    {
+      "epoch": 0.673329054735491,
+      "learning_rate": 6.53360227376914e-05,
+      "loss": 0.6548,
+      "mean_token_accuracy": 0.790365743637085,
+      "num_tokens": 172606019.0,
+      "step": 73440
+    },
+    {
+      "epoch": 0.6734207389749702,
+      "learning_rate": 6.531768588979555e-05,
+      "loss": 0.7315,
+      "mean_token_accuracy": 0.7806772828102112,
+      "num_tokens": 172630551.0,
+      "step": 73450
+    },
+    {
+      "epoch": 0.6735124232144495,
+      "learning_rate": 6.52993490418997e-05,
+      "loss": 0.6828,
+      "mean_token_accuracy": 0.7910055220127106,
+      "num_tokens": 172656906.0,
+      "step": 73460
+    },
+    {
+      "epoch": 0.6736041074539286,
+      "learning_rate": 6.528101219400385e-05,
+      "loss": 0.6804,
+      "mean_token_accuracy": 0.7891456961631775,
+      "num_tokens": 172682163.0,
+      "step": 73470
+    },
+    {
+      "epoch": 0.6736957916934079,
+      "learning_rate": 6.5262675346108e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.7979587912559509,
+      "num_tokens": 172707135.0,
+      "step": 73480
+    },
+    {
+      "epoch": 0.6737874759328871,
+      "learning_rate": 6.524433849821217e-05,
+      "loss": 0.68,
+      "mean_token_accuracy": 0.7888090193271637,
+      "num_tokens": 172732381.0,
+      "step": 73490
+    },
+    {
+      "epoch": 0.6738791601723664,
+      "learning_rate": 6.522600165031632e-05,
+      "loss": 0.6538,
+      "mean_token_accuracy": 0.7977280199527741,
+      "num_tokens": 172757946.0,
+      "step": 73500
+    },
+    {
+      "epoch": 0.6739708444118456,
+      "learning_rate": 6.520766480242047e-05,
+      "loss": 0.6804,
+      "mean_token_accuracy": 0.7910053670406342,
+      "num_tokens": 172782827.0,
+      "step": 73510
+    },
+    {
+      "epoch": 0.6740625286513249,
+      "learning_rate": 6.518932795452461e-05,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.7873681724071503,
+      "num_tokens": 172808140.0,
+      "step": 73520
+    },
+    {
+      "epoch": 0.674154212890804,
+      "learning_rate": 6.517099110662878e-05,
+      "loss": 0.6549,
+      "mean_token_accuracy": 0.7948734998703003,
+      "num_tokens": 172833779.0,
+      "step": 73530
+    },
+    {
+      "epoch": 0.6742458971302833,
+      "learning_rate": 6.515265425873293e-05,
+      "loss": 0.6752,
+      "mean_token_accuracy": 0.7912993013858796,
+      "num_tokens": 172858455.0,
+      "step": 73540
+    },
+    {
+      "epoch": 0.6743375813697625,
+      "learning_rate": 6.513431741083708e-05,
+      "loss": 0.6667,
+      "mean_token_accuracy": 0.7883789777755738,
+      "num_tokens": 172883605.0,
+      "step": 73550
+    },
+    {
+      "epoch": 0.6744292656092418,
+      "learning_rate": 6.511598056294123e-05,
+      "loss": 0.6969,
+      "mean_token_accuracy": 0.7820520997047424,
+      "num_tokens": 172908248.0,
+      "step": 73560
+    },
+    {
+      "epoch": 0.674520949848721,
+      "learning_rate": 6.509764371504538e-05,
+      "loss": 0.6855,
+      "mean_token_accuracy": 0.7876793026924134,
+      "num_tokens": 172933553.0,
+      "step": 73570
+    },
+    {
+      "epoch": 0.6746126340882003,
+      "learning_rate": 6.507930686714953e-05,
+      "loss": 0.6937,
+      "mean_token_accuracy": 0.7893419742584229,
+      "num_tokens": 172958479.0,
+      "step": 73580
+    },
+    {
+      "epoch": 0.6747043183276795,
+      "learning_rate": 6.50609700192537e-05,
+      "loss": 0.6733,
+      "mean_token_accuracy": 0.7927237272262573,
+      "num_tokens": 172983148.0,
+      "step": 73590
+    },
+    {
+      "epoch": 0.6747960025671587,
+      "learning_rate": 6.504263317135785e-05,
+      "loss": 0.6691,
+      "mean_token_accuracy": 0.792727929353714,
+      "num_tokens": 173008560.0,
+      "step": 73600
+    },
+    {
+      "epoch": 0.6748876868066379,
+      "learning_rate": 6.5024296323462e-05,
+      "loss": 0.6401,
+      "mean_token_accuracy": 0.8001432240009307,
+      "num_tokens": 173033695.0,
+      "step": 73610
+    },
+    {
+      "epoch": 0.6749793710461172,
+      "learning_rate": 6.500595947556616e-05,
+      "loss": 0.6549,
+      "mean_token_accuracy": 0.792980182170868,
+      "num_tokens": 173059877.0,
+      "step": 73620
+    },
+    {
+      "epoch": 0.6750710552855964,
+      "learning_rate": 6.498762262767031e-05,
+      "loss": 0.6909,
+      "mean_token_accuracy": 0.785872858762741,
+      "num_tokens": 173084537.0,
+      "step": 73630
+    },
+    {
+      "epoch": 0.6751627395250757,
+      "learning_rate": 6.496928577977446e-05,
+      "loss": 0.6566,
+      "mean_token_accuracy": 0.7980120778083801,
+      "num_tokens": 173109454.0,
+      "step": 73640
+    },
+    {
+      "epoch": 0.6752544237645549,
+      "learning_rate": 6.495094893187861e-05,
+      "loss": 0.6648,
+      "mean_token_accuracy": 0.794419401884079,
+      "num_tokens": 173134122.0,
+      "step": 73650
+    },
+    {
+      "epoch": 0.675346108004034,
+      "learning_rate": 6.493261208398276e-05,
+      "loss": 0.6476,
+      "mean_token_accuracy": 0.7986183345317841,
+      "num_tokens": 173159670.0,
+      "step": 73660
+    },
+    {
+      "epoch": 0.6754377922435133,
+      "learning_rate": 6.491427523608691e-05,
+      "loss": 0.6697,
+      "mean_token_accuracy": 0.7906152844429016,
+      "num_tokens": 173184164.0,
+      "step": 73670
+    },
+    {
+      "epoch": 0.6755294764829926,
+      "learning_rate": 6.489593838819107e-05,
+      "loss": 0.6596,
+      "mean_token_accuracy": 0.7963453114032746,
+      "num_tokens": 173208593.0,
+      "step": 73680
+    },
+    {
+      "epoch": 0.6756211607224718,
+      "learning_rate": 6.487760154029523e-05,
+      "loss": 0.6589,
+      "mean_token_accuracy": 0.79412140250206,
+      "num_tokens": 173233750.0,
+      "step": 73690
+    },
+    {
+      "epoch": 0.675712844961951,
+      "learning_rate": 6.485926469239938e-05,
+      "loss": 0.676,
+      "mean_token_accuracy": 0.7839472711086273,
+      "num_tokens": 173258367.0,
+      "step": 73700
+    },
+    {
+      "epoch": 0.6758045292014303,
+      "learning_rate": 6.484092784450354e-05,
+      "loss": 0.7077,
+      "mean_token_accuracy": 0.7804653346538544,
+      "num_tokens": 173283046.0,
+      "step": 73710
+    },
+    {
+      "epoch": 0.6758962134409096,
+      "learning_rate": 6.482259099660769e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.7937946081161499,
+      "num_tokens": 173308416.0,
+      "step": 73720
+    },
+    {
+      "epoch": 0.6759878976803887,
+      "learning_rate": 6.480425414871184e-05,
+      "loss": 0.6814,
+      "mean_token_accuracy": 0.7905770540237427,
+      "num_tokens": 173332804.0,
+      "step": 73730
+    },
+    {
+      "epoch": 0.676079581919868,
+      "learning_rate": 6.478591730081599e-05,
+      "loss": 0.7081,
+      "mean_token_accuracy": 0.7803768277168274,
+      "num_tokens": 173356951.0,
+      "step": 73740
+    },
+    {
+      "epoch": 0.6761712661593472,
+      "learning_rate": 6.476758045292016e-05,
+      "loss": 0.6582,
+      "mean_token_accuracy": 0.7917639195919037,
+      "num_tokens": 173382580.0,
+      "step": 73750
+    },
+    {
+      "epoch": 0.6762629503988264,
+      "learning_rate": 6.47492436050243e-05,
+      "loss": 0.6959,
+      "mean_token_accuracy": 0.7806162357330322,
+      "num_tokens": 173407258.0,
+      "step": 73760
+    },
+    {
+      "epoch": 0.6763546346383057,
+      "learning_rate": 6.473090675712845e-05,
+      "loss": 0.7088,
+      "mean_token_accuracy": 0.788748037815094,
+      "num_tokens": 173432676.0,
+      "step": 73770
+    },
+    {
+      "epoch": 0.676446318877785,
+      "learning_rate": 6.47125699092326e-05,
+      "loss": 0.7046,
+      "mean_token_accuracy": 0.7841828286647796,
+      "num_tokens": 173457184.0,
+      "step": 73780
+    },
+    {
+      "epoch": 0.6765380031172642,
+      "learning_rate": 6.469423306133676e-05,
+      "loss": 0.6905,
+      "mean_token_accuracy": 0.7858833491802215,
+      "num_tokens": 173481866.0,
+      "step": 73790
+    },
+    {
+      "epoch": 0.6766296873567433,
+      "learning_rate": 6.467589621344092e-05,
+      "loss": 0.6471,
+      "mean_token_accuracy": 0.7961485326290131,
+      "num_tokens": 173506678.0,
+      "step": 73800
+    },
+    {
+      "epoch": 0.6767213715962226,
+      "learning_rate": 6.465755936554507e-05,
+      "loss": 0.6613,
+      "mean_token_accuracy": 0.7931310176849365,
+      "num_tokens": 173531666.0,
+      "step": 73810
+    },
+    {
+      "epoch": 0.6768130558357018,
+      "learning_rate": 6.463922251764922e-05,
+      "loss": 0.6936,
+      "mean_token_accuracy": 0.7924591898918152,
+      "num_tokens": 173557300.0,
+      "step": 73820
+    },
+    {
+      "epoch": 0.6769047400751811,
+      "learning_rate": 6.462088566975337e-05,
+      "loss": 0.7154,
+      "mean_token_accuracy": 0.785172563791275,
+      "num_tokens": 173582671.0,
+      "step": 73830
+    },
+    {
+      "epoch": 0.6769964243146603,
+      "learning_rate": 6.460254882185752e-05,
+      "loss": 0.7086,
+      "mean_token_accuracy": 0.7781278610229492,
+      "num_tokens": 173608046.0,
+      "step": 73840
+    },
+    {
+      "epoch": 0.6770881085541396,
+      "learning_rate": 6.458421197396169e-05,
+      "loss": 0.6828,
+      "mean_token_accuracy": 0.7839238464832305,
+      "num_tokens": 173632452.0,
+      "step": 73850
+    },
+    {
+      "epoch": 0.6771797927936187,
+      "learning_rate": 6.456587512606583e-05,
+      "loss": 0.6882,
+      "mean_token_accuracy": 0.7890949189662934,
+      "num_tokens": 173657156.0,
+      "step": 73860
+    },
+    {
+      "epoch": 0.677271477033098,
+      "learning_rate": 6.454753827816998e-05,
+      "loss": 0.7067,
+      "mean_token_accuracy": 0.7835882127285003,
+      "num_tokens": 173681693.0,
+      "step": 73870
+    },
+    {
+      "epoch": 0.6773631612725772,
+      "learning_rate": 6.452920143027414e-05,
+      "loss": 0.6766,
+      "mean_token_accuracy": 0.7939685761928559,
+      "num_tokens": 173706987.0,
+      "step": 73880
+    },
+    {
+      "epoch": 0.6774548455120565,
+      "learning_rate": 6.45108645823783e-05,
+      "loss": 0.6723,
+      "mean_token_accuracy": 0.7910866022109986,
+      "num_tokens": 173732998.0,
+      "step": 73890
+    },
+    {
+      "epoch": 0.6775465297515357,
+      "learning_rate": 6.449252773448245e-05,
+      "loss": 0.6639,
+      "mean_token_accuracy": 0.7986181795597076,
+      "num_tokens": 173757448.0,
+      "step": 73900
+    },
+    {
+      "epoch": 0.677638213991015,
+      "learning_rate": 6.44741908865866e-05,
+      "loss": 0.6625,
+      "mean_token_accuracy": 0.7926622331142426,
+      "num_tokens": 173782658.0,
+      "step": 73910
+    },
+    {
+      "epoch": 0.6777298982304942,
+      "learning_rate": 6.445585403869075e-05,
+      "loss": 0.686,
+      "mean_token_accuracy": 0.7847441971302033,
+      "num_tokens": 173807028.0,
+      "step": 73920
+    },
+    {
+      "epoch": 0.6778215824699734,
+      "learning_rate": 6.44375171907949e-05,
+      "loss": 0.6569,
+      "mean_token_accuracy": 0.7972072541713715,
+      "num_tokens": 173832264.0,
+      "step": 73930
+    },
+    {
+      "epoch": 0.6779132667094526,
+      "learning_rate": 6.441918034289906e-05,
+      "loss": 0.6622,
+      "mean_token_accuracy": 0.7944561958312988,
+      "num_tokens": 173857047.0,
+      "step": 73940
+    },
+    {
+      "epoch": 0.6780049509489319,
+      "learning_rate": 6.440084349500322e-05,
+      "loss": 0.7022,
+      "mean_token_accuracy": 0.7907737970352173,
+      "num_tokens": 173881230.0,
+      "step": 73950
+    },
+    {
+      "epoch": 0.6780966351884111,
+      "learning_rate": 6.438250664710736e-05,
+      "loss": 0.6913,
+      "mean_token_accuracy": 0.7904622495174408,
+      "num_tokens": 173906479.0,
+      "step": 73960
+    },
+    {
+      "epoch": 0.6781883194278904,
+      "learning_rate": 6.436416979921151e-05,
+      "loss": 0.689,
+      "mean_token_accuracy": 0.787735641002655,
+      "num_tokens": 173932016.0,
+      "step": 73970
+    },
+    {
+      "epoch": 0.6782800036673696,
+      "learning_rate": 6.434583295131568e-05,
+      "loss": 0.659,
+      "mean_token_accuracy": 0.7958022475242614,
+      "num_tokens": 173957480.0,
+      "step": 73980
+    },
+    {
+      "epoch": 0.6783716879068488,
+      "learning_rate": 6.432749610341983e-05,
+      "loss": 0.6608,
+      "mean_token_accuracy": 0.79616779088974,
+      "num_tokens": 173982048.0,
+      "step": 73990
+    },
+    {
+      "epoch": 0.678463372146328,
+      "learning_rate": 6.430915925552398e-05,
+      "loss": 0.6986,
+      "mean_token_accuracy": 0.7894332766532898,
+      "num_tokens": 174007560.0,
+      "step": 74000
+    },
+    {
+      "epoch": 0.6785550563858073,
+      "learning_rate": 6.429082240762813e-05,
+      "loss": 0.7092,
+      "mean_token_accuracy": 0.786923611164093,
+      "num_tokens": 174032306.0,
+      "step": 74010
+    },
+    {
+      "epoch": 0.6786467406252865,
+      "learning_rate": 6.427248555973228e-05,
+      "loss": 0.6708,
+      "mean_token_accuracy": 0.7923457264900208,
+      "num_tokens": 174057463.0,
+      "step": 74020
+    },
+    {
+      "epoch": 0.6787384248647658,
+      "learning_rate": 6.425414871183644e-05,
+      "loss": 0.6858,
+      "mean_token_accuracy": 0.7890528321266175,
+      "num_tokens": 174081987.0,
+      "step": 74030
+    },
+    {
+      "epoch": 0.678830109104245,
+      "learning_rate": 6.423581186394059e-05,
+      "loss": 0.6998,
+      "mean_token_accuracy": 0.7877249181270599,
+      "num_tokens": 174106677.0,
+      "step": 74040
+    },
+    {
+      "epoch": 0.6789217933437243,
+      "learning_rate": 6.421747501604475e-05,
+      "loss": 0.6567,
+      "mean_token_accuracy": 0.7965851545333862,
+      "num_tokens": 174131417.0,
+      "step": 74050
+    },
+    {
+      "epoch": 0.6790134775832034,
+      "learning_rate": 6.419913816814889e-05,
+      "loss": 0.6708,
+      "mean_token_accuracy": 0.7861759960651398,
+      "num_tokens": 174155799.0,
+      "step": 74060
+    },
+    {
+      "epoch": 0.6791051618226827,
+      "learning_rate": 6.418080132025304e-05,
+      "loss": 0.6653,
+      "mean_token_accuracy": 0.8005099058151245,
+      "num_tokens": 174180479.0,
+      "step": 74070
+    },
+    {
+      "epoch": 0.6791968460621619,
+      "learning_rate": 6.416246447235721e-05,
+      "loss": 0.6853,
+      "mean_token_accuracy": 0.7853318393230438,
+      "num_tokens": 174205332.0,
+      "step": 74080
+    },
+    {
+      "epoch": 0.6792885303016412,
+      "learning_rate": 6.414412762446136e-05,
+      "loss": 0.6795,
+      "mean_token_accuracy": 0.7856890499591828,
+      "num_tokens": 174230762.0,
+      "step": 74090
+    },
+    {
+      "epoch": 0.6793802145411204,
+      "learning_rate": 6.412579077656551e-05,
+      "loss": 0.6637,
+      "mean_token_accuracy": 0.7959155738353729,
+      "num_tokens": 174256126.0,
+      "step": 74100
+    },
+    {
+      "epoch": 0.6794718987805997,
+      "learning_rate": 6.410745392866966e-05,
+      "loss": 0.6963,
+      "mean_token_accuracy": 0.791889774799347,
+      "num_tokens": 174281286.0,
+      "step": 74110
+    },
+    {
+      "epoch": 0.6795635830200788,
+      "learning_rate": 6.408911708077382e-05,
+      "loss": 0.6779,
+      "mean_token_accuracy": 0.7931702911853791,
+      "num_tokens": 174307231.0,
+      "step": 74120
+    },
+    {
+      "epoch": 0.679655267259558,
+      "learning_rate": 6.407078023287797e-05,
+      "loss": 0.667,
+      "mean_token_accuracy": 0.794391143321991,
+      "num_tokens": 174332261.0,
+      "step": 74130
+    },
+    {
+      "epoch": 0.6797469514990373,
+      "learning_rate": 6.405244338498212e-05,
+      "loss": 0.6895,
+      "mean_token_accuracy": 0.7863744020462036,
+      "num_tokens": 174357108.0,
+      "step": 74140
+    },
+    {
+      "epoch": 0.6798386357385166,
+      "learning_rate": 6.403410653708629e-05,
+      "loss": 0.6753,
+      "mean_token_accuracy": 0.7901894748210907,
+      "num_tokens": 174381106.0,
+      "step": 74150
+    },
+    {
+      "epoch": 0.6799303199779958,
+      "learning_rate": 6.401576968919042e-05,
+      "loss": 0.6859,
+      "mean_token_accuracy": 0.7880065619945527,
+      "num_tokens": 174406090.0,
+      "step": 74160
+    },
+    {
+      "epoch": 0.680022004217475,
+      "learning_rate": 6.399743284129458e-05,
+      "loss": 0.6722,
+      "mean_token_accuracy": 0.7924544394016266,
+      "num_tokens": 174431369.0,
+      "step": 74170
+    },
+    {
+      "epoch": 0.6801136884569543,
+      "learning_rate": 6.397909599339874e-05,
+      "loss": 0.6822,
+      "mean_token_accuracy": 0.7808837294578552,
+      "num_tokens": 174455877.0,
+      "step": 74180
+    },
+    {
+      "epoch": 0.6802053726964334,
+      "learning_rate": 6.396075914550289e-05,
+      "loss": 0.6809,
+      "mean_token_accuracy": 0.7911515176296234,
+      "num_tokens": 174480786.0,
+      "step": 74190
+    },
+    {
+      "epoch": 0.6802970569359127,
+      "learning_rate": 6.394242229760705e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.7959784805774689,
+      "num_tokens": 174506202.0,
+      "step": 74200
+    },
+    {
+      "epoch": 0.6803887411753919,
+      "learning_rate": 6.39240854497112e-05,
+      "loss": 0.6634,
+      "mean_token_accuracy": 0.7917146682739258,
+      "num_tokens": 174531727.0,
+      "step": 74210
+    },
+    {
+      "epoch": 0.6804804254148712,
+      "learning_rate": 6.390574860181535e-05,
+      "loss": 0.6438,
+      "mean_token_accuracy": 0.7946355164051055,
+      "num_tokens": 174556776.0,
+      "step": 74220
+    },
+    {
+      "epoch": 0.6805721096543504,
+      "learning_rate": 6.38874117539195e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.7886763453483582,
+      "num_tokens": 174581341.0,
+      "step": 74230
+    },
+    {
+      "epoch": 0.6806637938938297,
+      "learning_rate": 6.386907490602367e-05,
+      "loss": 0.7162,
+      "mean_token_accuracy": 0.7835057616233826,
+      "num_tokens": 174606350.0,
+      "step": 74240
+    },
+    {
+      "epoch": 0.6807554781333088,
+      "learning_rate": 6.385073805812782e-05,
+      "loss": 0.6958,
+      "mean_token_accuracy": 0.7827196896076203,
+      "num_tokens": 174630726.0,
+      "step": 74250
+    },
+    {
+      "epoch": 0.6808471623727881,
+      "learning_rate": 6.383240121023196e-05,
+      "loss": 0.626,
+      "mean_token_accuracy": 0.7978158891201019,
+      "num_tokens": 174656196.0,
+      "step": 74260
+    },
+    {
+      "epoch": 0.6809388466122673,
+      "learning_rate": 6.381406436233611e-05,
+      "loss": 0.6436,
+      "mean_token_accuracy": 0.7940483272075654,
+      "num_tokens": 174681488.0,
+      "step": 74270
+    },
+    {
+      "epoch": 0.6810305308517466,
+      "learning_rate": 6.379572751444027e-05,
+      "loss": 0.68,
+      "mean_token_accuracy": 0.7875082731246948,
+      "num_tokens": 174706769.0,
+      "step": 74280
+    },
+    {
+      "epoch": 0.6811222150912258,
+      "learning_rate": 6.377739066654443e-05,
+      "loss": 0.7126,
+      "mean_token_accuracy": 0.783999752998352,
+      "num_tokens": 174730689.0,
+      "step": 74290
+    },
+    {
+      "epoch": 0.6812138993307051,
+      "learning_rate": 6.375905381864858e-05,
+      "loss": 0.6652,
+      "mean_token_accuracy": 0.7903046488761902,
+      "num_tokens": 174755620.0,
+      "step": 74300
+    },
+    {
+      "epoch": 0.6813055835701843,
+      "learning_rate": 6.374071697075273e-05,
+      "loss": 0.6661,
+      "mean_token_accuracy": 0.7965183019638061,
+      "num_tokens": 174780860.0,
+      "step": 74310
+    },
+    {
+      "epoch": 0.6813972678096635,
+      "learning_rate": 6.372238012285688e-05,
+      "loss": 0.649,
+      "mean_token_accuracy": 0.7939989566802979,
+      "num_tokens": 174806026.0,
+      "step": 74320
+    },
+    {
+      "epoch": 0.6814889520491427,
+      "learning_rate": 6.370404327496103e-05,
+      "loss": 0.6817,
+      "mean_token_accuracy": 0.7914859235286713,
+      "num_tokens": 174830641.0,
+      "step": 74330
+    },
+    {
+      "epoch": 0.681580636288622,
+      "learning_rate": 6.36857064270652e-05,
+      "loss": 0.6668,
+      "mean_token_accuracy": 0.7908641993999481,
+      "num_tokens": 174855848.0,
+      "step": 74340
+    },
+    {
+      "epoch": 0.6816723205281012,
+      "learning_rate": 6.366736957916935e-05,
+      "loss": 0.6894,
+      "mean_token_accuracy": 0.7872664928436279,
+      "num_tokens": 174880558.0,
+      "step": 74350
+    },
+    {
+      "epoch": 0.6817640047675805,
+      "learning_rate": 6.364903273127349e-05,
+      "loss": 0.6606,
+      "mean_token_accuracy": 0.7983326435089111,
+      "num_tokens": 174904988.0,
+      "step": 74360
+    },
+    {
+      "epoch": 0.6818556890070597,
+      "learning_rate": 6.363069588337765e-05,
+      "loss": 0.7119,
+      "mean_token_accuracy": 0.7824760377407074,
+      "num_tokens": 174930249.0,
+      "step": 74370
+    },
+    {
+      "epoch": 0.6819473732465389,
+      "learning_rate": 6.36123590354818e-05,
+      "loss": 0.675,
+      "mean_token_accuracy": 0.7913299977779389,
+      "num_tokens": 174955480.0,
+      "step": 74380
+    },
+    {
+      "epoch": 0.6820390574860181,
+      "learning_rate": 6.359402218758596e-05,
+      "loss": 0.6812,
+      "mean_token_accuracy": 0.7869184613227844,
+      "num_tokens": 174980423.0,
+      "step": 74390
+    },
+    {
+      "epoch": 0.6821307417254974,
+      "learning_rate": 6.357568533969011e-05,
+      "loss": 0.6836,
+      "mean_token_accuracy": 0.788817697763443,
+      "num_tokens": 175004986.0,
+      "step": 74400
+    },
+    {
+      "epoch": 0.6822224259649766,
+      "learning_rate": 6.355734849179426e-05,
+      "loss": 0.6741,
+      "mean_token_accuracy": 0.7905064821243286,
+      "num_tokens": 175030631.0,
+      "step": 74410
+    },
+    {
+      "epoch": 0.6823141102044559,
+      "learning_rate": 6.353901164389841e-05,
+      "loss": 0.6826,
+      "mean_token_accuracy": 0.7935149848461152,
+      "num_tokens": 175057046.0,
+      "step": 74420
+    },
+    {
+      "epoch": 0.6824057944439351,
+      "learning_rate": 6.352067479600257e-05,
+      "loss": 0.6983,
+      "mean_token_accuracy": 0.7898350715637207,
+      "num_tokens": 175082393.0,
+      "step": 74430
+    },
+    {
+      "epoch": 0.6824974786834144,
+      "learning_rate": 6.350233794810673e-05,
+      "loss": 0.6717,
+      "mean_token_accuracy": 0.7930544734001159,
+      "num_tokens": 175108161.0,
+      "step": 74440
+    },
+    {
+      "epoch": 0.6825891629228935,
+      "learning_rate": 6.348400110021088e-05,
+      "loss": 0.6927,
+      "mean_token_accuracy": 0.7862828433513641,
+      "num_tokens": 175132943.0,
+      "step": 74450
+    },
+    {
+      "epoch": 0.6826808471623728,
+      "learning_rate": 6.346566425231502e-05,
+      "loss": 0.6377,
+      "mean_token_accuracy": 0.7992774963378906,
+      "num_tokens": 175158109.0,
+      "step": 74460
+    },
+    {
+      "epoch": 0.682772531401852,
+      "learning_rate": 6.344732740441919e-05,
+      "loss": 0.6907,
+      "mean_token_accuracy": 0.7890061199665069,
+      "num_tokens": 175182649.0,
+      "step": 74470
+    },
+    {
+      "epoch": 0.6828642156413313,
+      "learning_rate": 6.342899055652334e-05,
+      "loss": 0.7082,
+      "mean_token_accuracy": 0.7810740888118743,
+      "num_tokens": 175208200.0,
+      "step": 74480
+    },
+    {
+      "epoch": 0.6829558998808105,
+      "learning_rate": 6.341065370862749e-05,
+      "loss": 0.7035,
+      "mean_token_accuracy": 0.7838903844356537,
+      "num_tokens": 175232910.0,
+      "step": 74490
+    },
+    {
+      "epoch": 0.6830475841202898,
+      "learning_rate": 6.339231686073166e-05,
+      "loss": 0.652,
+      "mean_token_accuracy": 0.7972717523574829,
+      "num_tokens": 175258232.0,
+      "step": 74500
+    },
+    {
+      "epoch": 0.6831392683597689,
+      "learning_rate": 6.33739800128358e-05,
+      "loss": 0.6479,
+      "mean_token_accuracy": 0.7950194478034973,
+      "num_tokens": 175282824.0,
+      "step": 74510
+    },
+    {
+      "epoch": 0.6832309525992482,
+      "learning_rate": 6.335564316493995e-05,
+      "loss": 0.6951,
+      "mean_token_accuracy": 0.7904062986373901,
+      "num_tokens": 175307586.0,
+      "step": 74520
+    },
+    {
+      "epoch": 0.6833226368387274,
+      "learning_rate": 6.33373063170441e-05,
+      "loss": 0.6757,
+      "mean_token_accuracy": 0.794041782617569,
+      "num_tokens": 175332684.0,
+      "step": 74530
+    },
+    {
+      "epoch": 0.6834143210782067,
+      "learning_rate": 6.331896946914826e-05,
+      "loss": 0.6795,
+      "mean_token_accuracy": 0.7868710994720459,
+      "num_tokens": 175357166.0,
+      "step": 74540
+    },
+    {
+      "epoch": 0.6835060053176859,
+      "learning_rate": 6.330063262125241e-05,
+      "loss": 0.7009,
+      "mean_token_accuracy": 0.787664407491684,
+      "num_tokens": 175381807.0,
+      "step": 74550
+    },
+    {
+      "epoch": 0.6835976895571652,
+      "learning_rate": 6.328229577335655e-05,
+      "loss": 0.6669,
+      "mean_token_accuracy": 0.7960774600505829,
+      "num_tokens": 175407486.0,
+      "step": 74560
+    },
+    {
+      "epoch": 0.6836893737966444,
+      "learning_rate": 6.326395892546072e-05,
+      "loss": 0.6841,
+      "mean_token_accuracy": 0.7878939509391785,
+      "num_tokens": 175431787.0,
+      "step": 74570
+    },
+    {
+      "epoch": 0.6837810580361235,
+      "learning_rate": 6.324562207756487e-05,
+      "loss": 0.6767,
+      "mean_token_accuracy": 0.7889977872371674,
+      "num_tokens": 175456698.0,
+      "step": 74580
+    },
+    {
+      "epoch": 0.6838727422756028,
+      "learning_rate": 6.322728522966902e-05,
+      "loss": 0.7481,
+      "mean_token_accuracy": 0.7701053440570831,
+      "num_tokens": 175480707.0,
+      "step": 74590
+    },
+    {
+      "epoch": 0.683964426515082,
+      "learning_rate": 6.320894838177319e-05,
+      "loss": 0.6867,
+      "mean_token_accuracy": 0.7864112198352814,
+      "num_tokens": 175505215.0,
+      "step": 74600
+    },
+    {
+      "epoch": 0.6840561107545613,
+      "learning_rate": 6.319061153387733e-05,
+      "loss": 0.7053,
+      "mean_token_accuracy": 0.7879898548126221,
+      "num_tokens": 175529383.0,
+      "step": 74610
+    },
+    {
+      "epoch": 0.6841477949940405,
+      "learning_rate": 6.317227468598148e-05,
+      "loss": 0.7052,
+      "mean_token_accuracy": 0.7819807648658752,
+      "num_tokens": 175554601.0,
+      "step": 74620
+    },
+    {
+      "epoch": 0.6842394792335198,
+      "learning_rate": 6.315393783808563e-05,
+      "loss": 0.6865,
+      "mean_token_accuracy": 0.7832239806652069,
+      "num_tokens": 175580848.0,
+      "step": 74630
+    },
+    {
+      "epoch": 0.6843311634729989,
+      "learning_rate": 6.31356009901898e-05,
+      "loss": 0.6976,
+      "mean_token_accuracy": 0.7887556612491607,
+      "num_tokens": 175605583.0,
+      "step": 74640
+    },
+    {
+      "epoch": 0.6844228477124782,
+      "learning_rate": 6.311726414229395e-05,
+      "loss": 0.6877,
+      "mean_token_accuracy": 0.785181850194931,
+      "num_tokens": 175630800.0,
+      "step": 74650
+    },
+    {
+      "epoch": 0.6845145319519574,
+      "learning_rate": 6.309892729439809e-05,
+      "loss": 0.6557,
+      "mean_token_accuracy": 0.7960622072219848,
+      "num_tokens": 175655355.0,
+      "step": 74660
+    },
+    {
+      "epoch": 0.6846062161914367,
+      "learning_rate": 6.308059044650225e-05,
+      "loss": 0.6625,
+      "mean_token_accuracy": 0.7905114889144897,
+      "num_tokens": 175680500.0,
+      "step": 74670
+    },
+    {
+      "epoch": 0.6846979004309159,
+      "learning_rate": 6.30622535986064e-05,
+      "loss": 0.6765,
+      "mean_token_accuracy": 0.7875080823898315,
+      "num_tokens": 175706096.0,
+      "step": 74680
+    },
+    {
+      "epoch": 0.6847895846703952,
+      "learning_rate": 6.304391675071055e-05,
+      "loss": 0.658,
+      "mean_token_accuracy": 0.7931058704853058,
+      "num_tokens": 175731007.0,
+      "step": 74690
+    },
+    {
+      "epoch": 0.6848812689098744,
+      "learning_rate": 6.302557990281472e-05,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.791066724061966,
+      "num_tokens": 175755634.0,
+      "step": 74700
+    },
+    {
+      "epoch": 0.6849729531493536,
+      "learning_rate": 6.300724305491886e-05,
+      "loss": 0.7134,
+      "mean_token_accuracy": 0.7832394957542419,
+      "num_tokens": 175780344.0,
+      "step": 74710
+    },
+    {
+      "epoch": 0.6850646373888328,
+      "learning_rate": 6.298890620702301e-05,
+      "loss": 0.6556,
+      "mean_token_accuracy": 0.7953931987285614,
+      "num_tokens": 175805734.0,
+      "step": 74720
+    },
+    {
+      "epoch": 0.6851563216283121,
+      "learning_rate": 6.297056935912718e-05,
+      "loss": 0.6642,
+      "mean_token_accuracy": 0.7935296833515167,
+      "num_tokens": 175830590.0,
+      "step": 74730
+    },
+    {
+      "epoch": 0.6852480058677913,
+      "learning_rate": 6.295223251123133e-05,
+      "loss": 0.6702,
+      "mean_token_accuracy": 0.7878190457820893,
+      "num_tokens": 175855581.0,
+      "step": 74740
+    },
+    {
+      "epoch": 0.6853396901072706,
+      "learning_rate": 6.293389566333548e-05,
+      "loss": 0.6586,
+      "mean_token_accuracy": 0.7975728690624238,
+      "num_tokens": 175880549.0,
+      "step": 74750
+    },
+    {
+      "epoch": 0.6854313743467498,
+      "learning_rate": 6.291555881543962e-05,
+      "loss": 0.6628,
+      "mean_token_accuracy": 0.7912991404533386,
+      "num_tokens": 175905071.0,
+      "step": 74760
+    },
+    {
+      "epoch": 0.685523058586229,
+      "learning_rate": 6.289722196754378e-05,
+      "loss": 0.6697,
+      "mean_token_accuracy": 0.7938998997211456,
+      "num_tokens": 175930188.0,
+      "step": 74770
+    },
+    {
+      "epoch": 0.6856147428257082,
+      "learning_rate": 6.287888511964793e-05,
+      "loss": 0.6793,
+      "mean_token_accuracy": 0.7936573028564453,
+      "num_tokens": 175955489.0,
+      "step": 74780
+    },
+    {
+      "epoch": 0.6857064270651875,
+      "learning_rate": 6.286054827175209e-05,
+      "loss": 0.6707,
+      "mean_token_accuracy": 0.7974761843681335,
+      "num_tokens": 175980142.0,
+      "step": 74790
+    },
+    {
+      "epoch": 0.6857981113046667,
+      "learning_rate": 6.284221142385625e-05,
+      "loss": 0.6872,
+      "mean_token_accuracy": 0.7910990118980408,
+      "num_tokens": 176005801.0,
+      "step": 74800
+    },
+    {
+      "epoch": 0.685889795544146,
+      "learning_rate": 6.282387457596039e-05,
+      "loss": 0.7063,
+      "mean_token_accuracy": 0.788154023885727,
+      "num_tokens": 176031557.0,
+      "step": 74810
+    },
+    {
+      "epoch": 0.6859814797836252,
+      "learning_rate": 6.280553772806454e-05,
+      "loss": 0.6789,
+      "mean_token_accuracy": 0.7894406259059906,
+      "num_tokens": 176056786.0,
+      "step": 74820
+    },
+    {
+      "epoch": 0.6860731640231045,
+      "learning_rate": 6.278720088016871e-05,
+      "loss": 0.689,
+      "mean_token_accuracy": 0.7851300597190857,
+      "num_tokens": 176081316.0,
+      "step": 74830
+    },
+    {
+      "epoch": 0.6861648482625836,
+      "learning_rate": 6.276886403227286e-05,
+      "loss": 0.6962,
+      "mean_token_accuracy": 0.7852903544902802,
+      "num_tokens": 176106431.0,
+      "step": 74840
+    },
+    {
+      "epoch": 0.6862565325020629,
+      "learning_rate": 6.275052718437701e-05,
+      "loss": 0.6942,
+      "mean_token_accuracy": 0.7856343805789947,
+      "num_tokens": 176132456.0,
+      "step": 74850
+    },
+    {
+      "epoch": 0.6863482167415421,
+      "learning_rate": 6.273219033648116e-05,
+      "loss": 0.6819,
+      "mean_token_accuracy": 0.7907595276832581,
+      "num_tokens": 176157247.0,
+      "step": 74860
+    },
+    {
+      "epoch": 0.6864399009810214,
+      "learning_rate": 6.271385348858532e-05,
+      "loss": 0.6958,
+      "mean_token_accuracy": 0.7894757807254791,
+      "num_tokens": 176182554.0,
+      "step": 74870
+    },
+    {
+      "epoch": 0.6865315852205006,
+      "learning_rate": 6.269551664068947e-05,
+      "loss": 0.6695,
+      "mean_token_accuracy": 0.789708924293518,
+      "num_tokens": 176207647.0,
+      "step": 74880
+    },
+    {
+      "epoch": 0.6866232694599799,
+      "learning_rate": 6.267717979279362e-05,
+      "loss": 0.6944,
+      "mean_token_accuracy": 0.788169902563095,
+      "num_tokens": 176232333.0,
+      "step": 74890
+    },
+    {
+      "epoch": 0.686714953699459,
+      "learning_rate": 6.265884294489778e-05,
+      "loss": 0.6958,
+      "mean_token_accuracy": 0.7871954560279846,
+      "num_tokens": 176256999.0,
+      "step": 74900
+    },
+    {
+      "epoch": 0.6868066379389383,
+      "learning_rate": 6.264050609700192e-05,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.7925026893615723,
+      "num_tokens": 176282468.0,
+      "step": 74910
+    },
+    {
+      "epoch": 0.6868983221784175,
+      "learning_rate": 6.262216924910607e-05,
+      "loss": 0.684,
+      "mean_token_accuracy": 0.7899601221084595,
+      "num_tokens": 176307350.0,
+      "step": 74920
+    },
+    {
+      "epoch": 0.6869900064178968,
+      "learning_rate": 6.260383240121024e-05,
+      "loss": 0.7292,
+      "mean_token_accuracy": 0.7792982399463654,
+      "num_tokens": 176332281.0,
+      "step": 74930
+    },
+    {
+      "epoch": 0.687081690657376,
+      "learning_rate": 6.258549555331439e-05,
+      "loss": 0.6984,
+      "mean_token_accuracy": 0.7857298851013184,
+      "num_tokens": 176356415.0,
+      "step": 74940
+    },
+    {
+      "epoch": 0.6871733748968553,
+      "learning_rate": 6.256715870541854e-05,
+      "loss": 0.6982,
+      "mean_token_accuracy": 0.7876420080661773,
+      "num_tokens": 176381918.0,
+      "step": 74950
+    },
+    {
+      "epoch": 0.6872650591363345,
+      "learning_rate": 6.25488218575227e-05,
+      "loss": 0.7004,
+      "mean_token_accuracy": 0.7868048846721649,
+      "num_tokens": 176408042.0,
+      "step": 74960
+    },
+    {
+      "epoch": 0.6873567433758137,
+      "learning_rate": 6.253048500962685e-05,
+      "loss": 0.6993,
+      "mean_token_accuracy": 0.7900464177131653,
+      "num_tokens": 176433361.0,
+      "step": 74970
+    },
+    {
+      "epoch": 0.6874484276152929,
+      "learning_rate": 6.2512148161731e-05,
+      "loss": 0.6782,
+      "mean_token_accuracy": 0.791328901052475,
+      "num_tokens": 176458697.0,
+      "step": 74980
+    },
+    {
+      "epoch": 0.6875401118547722,
+      "learning_rate": 6.249381131383516e-05,
+      "loss": 0.6399,
+      "mean_token_accuracy": 0.8019151329994202,
+      "num_tokens": 176484658.0,
+      "step": 74990
+    },
+    {
+      "epoch": 0.6876317960942514,
+      "learning_rate": 6.247547446593932e-05,
+      "loss": 0.6879,
+      "mean_token_accuracy": 0.7877026379108429,
+      "num_tokens": 176509669.0,
+      "step": 75000
+    },
+    {
+      "epoch": 0.6877234803337307,
+      "learning_rate": 6.245713761804345e-05,
+      "loss": 0.6717,
+      "mean_token_accuracy": 0.793970137834549,
+      "num_tokens": 176534034.0,
+      "step": 75010
+    },
+    {
+      "epoch": 0.6878151645732099,
+      "learning_rate": 6.24388007701476e-05,
+      "loss": 0.6699,
+      "mean_token_accuracy": 0.7973458468914032,
+      "num_tokens": 176559260.0,
+      "step": 75020
+    },
+    {
+      "epoch": 0.6879068488126892,
+      "learning_rate": 6.242046392225177e-05,
+      "loss": 0.6774,
+      "mean_token_accuracy": 0.7890937805175782,
+      "num_tokens": 176584684.0,
+      "step": 75030
+    },
+    {
+      "epoch": 0.6879985330521683,
+      "learning_rate": 6.240212707435592e-05,
+      "loss": 0.6848,
+      "mean_token_accuracy": 0.7890117824077606,
+      "num_tokens": 176610226.0,
+      "step": 75040
+    },
+    {
+      "epoch": 0.6880902172916475,
+      "learning_rate": 6.238379022646008e-05,
+      "loss": 0.7119,
+      "mean_token_accuracy": 0.7836406707763672,
+      "num_tokens": 176635726.0,
+      "step": 75050
+    },
+    {
+      "epoch": 0.6881819015311268,
+      "learning_rate": 6.236545337856423e-05,
+      "loss": 0.6726,
+      "mean_token_accuracy": 0.793917179107666,
+      "num_tokens": 176660949.0,
+      "step": 75060
+    },
+    {
+      "epoch": 0.688273585770606,
+      "learning_rate": 6.234711653066838e-05,
+      "loss": 0.6581,
+      "mean_token_accuracy": 0.7956817448139191,
+      "num_tokens": 176686556.0,
+      "step": 75070
+    },
+    {
+      "epoch": 0.6883652700100853,
+      "learning_rate": 6.232877968277253e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.7914393424987793,
+      "num_tokens": 176712147.0,
+      "step": 75080
+    },
+    {
+      "epoch": 0.6884569542495645,
+      "learning_rate": 6.23104428348767e-05,
+      "loss": 0.6749,
+      "mean_token_accuracy": 0.7879520237445832,
+      "num_tokens": 176738115.0,
+      "step": 75090
+    },
+    {
+      "epoch": 0.6885486384890437,
+      "learning_rate": 6.229210598698085e-05,
+      "loss": 0.6593,
+      "mean_token_accuracy": 0.7963025748729706,
+      "num_tokens": 176763003.0,
+      "step": 75100
+    },
+    {
+      "epoch": 0.6886403227285229,
+      "learning_rate": 6.227376913908499e-05,
+      "loss": 0.6585,
+      "mean_token_accuracy": 0.7946280837059021,
+      "num_tokens": 176788072.0,
+      "step": 75110
+    },
+    {
+      "epoch": 0.6887320069680022,
+      "learning_rate": 6.225543229118915e-05,
+      "loss": 0.67,
+      "mean_token_accuracy": 0.7864981591701508,
+      "num_tokens": 176812355.0,
+      "step": 75120
+    },
+    {
+      "epoch": 0.6888236912074814,
+      "learning_rate": 6.22370954432933e-05,
+      "loss": 0.6627,
+      "mean_token_accuracy": 0.7990666508674622,
+      "num_tokens": 176837879.0,
+      "step": 75130
+    },
+    {
+      "epoch": 0.6889153754469607,
+      "learning_rate": 6.221875859539746e-05,
+      "loss": 0.6846,
+      "mean_token_accuracy": 0.7910236716270447,
+      "num_tokens": 176863173.0,
+      "step": 75140
+    },
+    {
+      "epoch": 0.6890070596864399,
+      "learning_rate": 6.220042174750161e-05,
+      "loss": 0.7191,
+      "mean_token_accuracy": 0.7803408920764923,
+      "num_tokens": 176887485.0,
+      "step": 75150
+    },
+    {
+      "epoch": 0.6890987439259192,
+      "learning_rate": 6.218208489960576e-05,
+      "loss": 0.6582,
+      "mean_token_accuracy": 0.7980068564414978,
+      "num_tokens": 176912921.0,
+      "step": 75160
+    },
+    {
+      "epoch": 0.6891904281653983,
+      "learning_rate": 6.216374805170991e-05,
+      "loss": 0.7019,
+      "mean_token_accuracy": 0.7854265809059143,
+      "num_tokens": 176937210.0,
+      "step": 75170
+    },
+    {
+      "epoch": 0.6892821124048776,
+      "learning_rate": 6.214541120381406e-05,
+      "loss": 0.6882,
+      "mean_token_accuracy": 0.7832460999488831,
+      "num_tokens": 176961269.0,
+      "step": 75180
+    },
+    {
+      "epoch": 0.6893737966443568,
+      "learning_rate": 6.212707435591823e-05,
+      "loss": 0.6857,
+      "mean_token_accuracy": 0.7883375585079193,
+      "num_tokens": 176986388.0,
+      "step": 75190
+    },
+    {
+      "epoch": 0.6894654808838361,
+      "learning_rate": 6.210873750802238e-05,
+      "loss": 0.6956,
+      "mean_token_accuracy": 0.7831620633602142,
+      "num_tokens": 177011397.0,
+      "step": 75200
+    },
+    {
+      "epoch": 0.6895571651233153,
+      "learning_rate": 6.209040066012652e-05,
+      "loss": 0.6854,
+      "mean_token_accuracy": 0.7810591638088227,
+      "num_tokens": 177036885.0,
+      "step": 75210
+    },
+    {
+      "epoch": 0.6896488493627946,
+      "learning_rate": 6.207206381223068e-05,
+      "loss": 0.6611,
+      "mean_token_accuracy": 0.7925475060939788,
+      "num_tokens": 177062510.0,
+      "step": 75220
+    },
+    {
+      "epoch": 0.6897405336022737,
+      "learning_rate": 6.205372696433484e-05,
+      "loss": 0.6533,
+      "mean_token_accuracy": 0.7994494497776031,
+      "num_tokens": 177086825.0,
+      "step": 75230
+    },
+    {
+      "epoch": 0.689832217841753,
+      "learning_rate": 6.203539011643899e-05,
+      "loss": 0.6987,
+      "mean_token_accuracy": 0.7893237888813018,
+      "num_tokens": 177111978.0,
+      "step": 75240
+    },
+    {
+      "epoch": 0.6899239020812322,
+      "learning_rate": 6.201705326854314e-05,
+      "loss": 0.6868,
+      "mean_token_accuracy": 0.7898162841796875,
+      "num_tokens": 177137586.0,
+      "step": 75250
+    },
+    {
+      "epoch": 0.6900155863207115,
+      "learning_rate": 6.199871642064729e-05,
+      "loss": 0.6605,
+      "mean_token_accuracy": 0.7909402132034302,
+      "num_tokens": 177163012.0,
+      "step": 75260
+    },
+    {
+      "epoch": 0.6901072705601907,
+      "learning_rate": 6.198037957275144e-05,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.7860115885734558,
+      "num_tokens": 177188443.0,
+      "step": 75270
+    },
+    {
+      "epoch": 0.69019895479967,
+      "learning_rate": 6.19620427248556e-05,
+      "loss": 0.6926,
+      "mean_token_accuracy": 0.7854578197002411,
+      "num_tokens": 177212813.0,
+      "step": 75280
+    },
+    {
+      "epoch": 0.6902906390391492,
+      "learning_rate": 6.194370587695976e-05,
+      "loss": 0.6944,
+      "mean_token_accuracy": 0.7880738377571106,
+      "num_tokens": 177238263.0,
+      "step": 75290
+    },
+    {
+      "epoch": 0.6903823232786284,
+      "learning_rate": 6.192536902906391e-05,
+      "loss": 0.7251,
+      "mean_token_accuracy": 0.7831411898136139,
+      "num_tokens": 177263489.0,
+      "step": 75300
+    },
+    {
+      "epoch": 0.6904740075181076,
+      "learning_rate": 6.190703218116805e-05,
+      "loss": 0.6338,
+      "mean_token_accuracy": 0.801439744234085,
+      "num_tokens": 177288428.0,
+      "step": 75310
+    },
+    {
+      "epoch": 0.6905656917575869,
+      "learning_rate": 6.188869533327222e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.7955764472484589,
+      "num_tokens": 177313945.0,
+      "step": 75320
+    },
+    {
+      "epoch": 0.6906573759970661,
+      "learning_rate": 6.187035848537637e-05,
+      "loss": 0.6987,
+      "mean_token_accuracy": 0.7869789004325867,
+      "num_tokens": 177339227.0,
+      "step": 75330
+    },
+    {
+      "epoch": 0.6907490602365454,
+      "learning_rate": 6.185202163748052e-05,
+      "loss": 0.6667,
+      "mean_token_accuracy": 0.7907529294490814,
+      "num_tokens": 177363411.0,
+      "step": 75340
+    },
+    {
+      "epoch": 0.6908407444760246,
+      "learning_rate": 6.183368478958467e-05,
+      "loss": 0.6942,
+      "mean_token_accuracy": 0.7806631922721863,
+      "num_tokens": 177387858.0,
+      "step": 75350
+    },
+    {
+      "epoch": 0.6909324287155038,
+      "learning_rate": 6.181534794168882e-05,
+      "loss": 0.6747,
+      "mean_token_accuracy": 0.792816162109375,
+      "num_tokens": 177414118.0,
+      "step": 75360
+    },
+    {
+      "epoch": 0.691024112954983,
+      "learning_rate": 6.179701109379298e-05,
+      "loss": 0.6949,
+      "mean_token_accuracy": 0.7874690413475036,
+      "num_tokens": 177438751.0,
+      "step": 75370
+    },
+    {
+      "epoch": 0.6911157971944623,
+      "learning_rate": 6.177867424589713e-05,
+      "loss": 0.6898,
+      "mean_token_accuracy": 0.7809600532054901,
+      "num_tokens": 177463691.0,
+      "step": 75380
+    },
+    {
+      "epoch": 0.6912074814339415,
+      "learning_rate": 6.17603373980013e-05,
+      "loss": 0.6795,
+      "mean_token_accuracy": 0.7870772123336792,
+      "num_tokens": 177488303.0,
+      "step": 75390
+    },
+    {
+      "epoch": 0.6912991656734208,
+      "learning_rate": 6.174200055010545e-05,
+      "loss": 0.6601,
+      "mean_token_accuracy": 0.7902093172073364,
+      "num_tokens": 177513019.0,
+      "step": 75400
+    },
+    {
+      "epoch": 0.6913908499129,
+      "learning_rate": 6.172366370220958e-05,
+      "loss": 0.6664,
+      "mean_token_accuracy": 0.7936054110527039,
+      "num_tokens": 177537750.0,
+      "step": 75410
+    },
+    {
+      "epoch": 0.6914825341523793,
+      "learning_rate": 6.170532685431375e-05,
+      "loss": 0.6764,
+      "mean_token_accuracy": 0.7926254570484161,
+      "num_tokens": 177563617.0,
+      "step": 75420
+    },
+    {
+      "epoch": 0.6915742183918584,
+      "learning_rate": 6.16869900064179e-05,
+      "loss": 0.6905,
+      "mean_token_accuracy": 0.7925673961639405,
+      "num_tokens": 177590017.0,
+      "step": 75430
+    },
+    {
+      "epoch": 0.6916659026313376,
+      "learning_rate": 6.166865315852205e-05,
+      "loss": 0.6711,
+      "mean_token_accuracy": 0.7915888249874115,
+      "num_tokens": 177615609.0,
+      "step": 75440
+    },
+    {
+      "epoch": 0.6917575868708169,
+      "learning_rate": 6.16503163106262e-05,
+      "loss": 0.6932,
+      "mean_token_accuracy": 0.7848492503166199,
+      "num_tokens": 177641094.0,
+      "step": 75450
+    },
+    {
+      "epoch": 0.6918492711102961,
+      "learning_rate": 6.163197946273036e-05,
+      "loss": 0.691,
+      "mean_token_accuracy": 0.7870258510112762,
+      "num_tokens": 177665845.0,
+      "step": 75460
+    },
+    {
+      "epoch": 0.6919409553497754,
+      "learning_rate": 6.161364261483451e-05,
+      "loss": 0.6693,
+      "mean_token_accuracy": 0.7906091928482055,
+      "num_tokens": 177691782.0,
+      "step": 75470
+    },
+    {
+      "epoch": 0.6920326395892547,
+      "learning_rate": 6.159530576693867e-05,
+      "loss": 0.6525,
+      "mean_token_accuracy": 0.7949664413928985,
+      "num_tokens": 177716605.0,
+      "step": 75480
+    },
+    {
+      "epoch": 0.6921243238287338,
+      "learning_rate": 6.157696891904283e-05,
+      "loss": 0.676,
+      "mean_token_accuracy": 0.7979806184768676,
+      "num_tokens": 177741785.0,
+      "step": 75490
+    },
+    {
+      "epoch": 0.692216008068213,
+      "learning_rate": 6.155863207114698e-05,
+      "loss": 0.6735,
+      "mean_token_accuracy": 0.7911450088024139,
+      "num_tokens": 177767958.0,
+      "step": 75500
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "learning_rate": 6.154029522325112e-05,
+      "loss": 0.6486,
+      "mean_token_accuracy": 0.8005157113075256,
+      "num_tokens": 177792220.0,
+      "step": 75510
+    },
+    {
+      "epoch": 0.6923993765471715,
+      "learning_rate": 6.152195837535528e-05,
+      "loss": 0.6788,
+      "mean_token_accuracy": 0.7899494409561157,
+      "num_tokens": 177817232.0,
+      "step": 75520
+    },
+    {
+      "epoch": 0.6924910607866508,
+      "learning_rate": 6.150362152745943e-05,
+      "loss": 0.6877,
+      "mean_token_accuracy": 0.7852452516555786,
+      "num_tokens": 177842140.0,
+      "step": 75530
+    },
+    {
+      "epoch": 0.69258274502613,
+      "learning_rate": 6.148528467956359e-05,
+      "loss": 0.6645,
+      "mean_token_accuracy": 0.7939651489257813,
+      "num_tokens": 177867930.0,
+      "step": 75540
+    },
+    {
+      "epoch": 0.6926744292656093,
+      "learning_rate": 6.146694783166774e-05,
+      "loss": 0.7164,
+      "mean_token_accuracy": 0.7811005592346192,
+      "num_tokens": 177892720.0,
+      "step": 75550
+    },
+    {
+      "epoch": 0.6927661135050884,
+      "learning_rate": 6.144861098377189e-05,
+      "loss": 0.7102,
+      "mean_token_accuracy": 0.7842880129814148,
+      "num_tokens": 177917877.0,
+      "step": 75560
+    },
+    {
+      "epoch": 0.6928577977445677,
+      "learning_rate": 6.143027413587604e-05,
+      "loss": 0.6923,
+      "mean_token_accuracy": 0.782748019695282,
+      "num_tokens": 177943076.0,
+      "step": 75570
+    },
+    {
+      "epoch": 0.6929494819840469,
+      "learning_rate": 6.14119372879802e-05,
+      "loss": 0.6618,
+      "mean_token_accuracy": 0.7946770131587982,
+      "num_tokens": 177968703.0,
+      "step": 75580
+    },
+    {
+      "epoch": 0.6930411662235262,
+      "learning_rate": 6.139360044008436e-05,
+      "loss": 0.6828,
+      "mean_token_accuracy": 0.7888063311576843,
+      "num_tokens": 177993985.0,
+      "step": 75590
+    },
+    {
+      "epoch": 0.6931328504630054,
+      "learning_rate": 6.137526359218851e-05,
+      "loss": 0.6891,
+      "mean_token_accuracy": 0.779136347770691,
+      "num_tokens": 178018934.0,
+      "step": 75600
+    },
+    {
+      "epoch": 0.6932245347024847,
+      "learning_rate": 6.135692674429266e-05,
+      "loss": 0.6589,
+      "mean_token_accuracy": 0.7927046358585358,
+      "num_tokens": 178044845.0,
+      "step": 75610
+    },
+    {
+      "epoch": 0.6933162189419638,
+      "learning_rate": 6.133858989639681e-05,
+      "loss": 0.6808,
+      "mean_token_accuracy": 0.788493013381958,
+      "num_tokens": 178070021.0,
+      "step": 75620
+    },
+    {
+      "epoch": 0.6934079031814431,
+      "learning_rate": 6.132025304850097e-05,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.7846109211444855,
+      "num_tokens": 178095090.0,
+      "step": 75630
+    },
+    {
+      "epoch": 0.6934995874209223,
+      "learning_rate": 6.130191620060512e-05,
+      "loss": 0.6734,
+      "mean_token_accuracy": 0.7920516550540924,
+      "num_tokens": 178120327.0,
+      "step": 75640
+    },
+    {
+      "epoch": 0.6935912716604016,
+      "learning_rate": 6.128357935270927e-05,
+      "loss": 0.6597,
+      "mean_token_accuracy": 0.7969824314117432,
+      "num_tokens": 178145092.0,
+      "step": 75650
+    },
+    {
+      "epoch": 0.6936829558998808,
+      "learning_rate": 6.126524250481342e-05,
+      "loss": 0.7192,
+      "mean_token_accuracy": 0.7807035684585572,
+      "num_tokens": 178169803.0,
+      "step": 75660
+    },
+    {
+      "epoch": 0.6937746401393601,
+      "learning_rate": 6.124690565691757e-05,
+      "loss": 0.6721,
+      "mean_token_accuracy": 0.7994394123554229,
+      "num_tokens": 178195806.0,
+      "step": 75670
+    },
+    {
+      "epoch": 0.6938663243788393,
+      "learning_rate": 6.122856880902174e-05,
+      "loss": 0.7059,
+      "mean_token_accuracy": 0.7798124194145203,
+      "num_tokens": 178220866.0,
+      "step": 75680
+    },
+    {
+      "epoch": 0.6939580086183185,
+      "learning_rate": 6.121023196112589e-05,
+      "loss": 0.6599,
+      "mean_token_accuracy": 0.7958055436611176,
+      "num_tokens": 178245969.0,
+      "step": 75690
+    },
+    {
+      "epoch": 0.6940496928577977,
+      "learning_rate": 6.119189511323004e-05,
+      "loss": 0.6593,
+      "mean_token_accuracy": 0.7950552761554718,
+      "num_tokens": 178271298.0,
+      "step": 75700
+    },
+    {
+      "epoch": 0.694141377097277,
+      "learning_rate": 6.11735582653342e-05,
+      "loss": 0.6692,
+      "mean_token_accuracy": 0.7914415419101715,
+      "num_tokens": 178296247.0,
+      "step": 75710
+    },
+    {
+      "epoch": 0.6942330613367562,
+      "learning_rate": 6.115522141743835e-05,
+      "loss": 0.6735,
+      "mean_token_accuracy": 0.7871362745761872,
+      "num_tokens": 178321619.0,
+      "step": 75720
+    },
+    {
+      "epoch": 0.6943247455762355,
+      "learning_rate": 6.11368845695425e-05,
+      "loss": 0.7062,
+      "mean_token_accuracy": 0.7866231739521027,
+      "num_tokens": 178346554.0,
+      "step": 75730
+    },
+    {
+      "epoch": 0.6944164298157147,
+      "learning_rate": 6.111854772164666e-05,
+      "loss": 0.6786,
+      "mean_token_accuracy": 0.7940635979175568,
+      "num_tokens": 178371551.0,
+      "step": 75740
+    },
+    {
+      "epoch": 0.6945081140551939,
+      "learning_rate": 6.11002108737508e-05,
+      "loss": 0.6367,
+      "mean_token_accuracy": 0.8015643656253815,
+      "num_tokens": 178396997.0,
+      "step": 75750
+    },
+    {
+      "epoch": 0.6945997982946731,
+      "learning_rate": 6.108187402585495e-05,
+      "loss": 0.6738,
+      "mean_token_accuracy": 0.7952834844589234,
+      "num_tokens": 178421864.0,
+      "step": 75760
+    },
+    {
+      "epoch": 0.6946914825341524,
+      "learning_rate": 6.10635371779591e-05,
+      "loss": 0.6645,
+      "mean_token_accuracy": 0.7930844902992249,
+      "num_tokens": 178447351.0,
+      "step": 75770
+    },
+    {
+      "epoch": 0.6947831667736316,
+      "learning_rate": 6.104520033006327e-05,
+      "loss": 0.7042,
+      "mean_token_accuracy": 0.7845900595188141,
+      "num_tokens": 178472558.0,
+      "step": 75780
+    },
+    {
+      "epoch": 0.6948748510131109,
+      "learning_rate": 6.1026863482167416e-05,
+      "loss": 0.6677,
+      "mean_token_accuracy": 0.7874324023723602,
+      "num_tokens": 178497672.0,
+      "step": 75790
+    },
+    {
+      "epoch": 0.6949665352525901,
+      "learning_rate": 6.100852663427157e-05,
+      "loss": 0.7032,
+      "mean_token_accuracy": 0.7855457067489624,
+      "num_tokens": 178521761.0,
+      "step": 75800
+    },
+    {
+      "epoch": 0.6950582194920694,
+      "learning_rate": 6.0990189786375726e-05,
+      "loss": 0.7079,
+      "mean_token_accuracy": 0.7808915078639984,
+      "num_tokens": 178546512.0,
+      "step": 75810
+    },
+    {
+      "epoch": 0.6951499037315485,
+      "learning_rate": 6.097185293847988e-05,
+      "loss": 0.6691,
+      "mean_token_accuracy": 0.7944140315055848,
+      "num_tokens": 178572627.0,
+      "step": 75820
+    },
+    {
+      "epoch": 0.6952415879710278,
+      "learning_rate": 6.095351609058403e-05,
+      "loss": 0.706,
+      "mean_token_accuracy": 0.7832296431064606,
+      "num_tokens": 178598214.0,
+      "step": 75830
+    },
+    {
+      "epoch": 0.695333272210507,
+      "learning_rate": 6.093517924268819e-05,
+      "loss": 0.7282,
+      "mean_token_accuracy": 0.7790860772132874,
+      "num_tokens": 178623376.0,
+      "step": 75840
+    },
+    {
+      "epoch": 0.6954249564499863,
+      "learning_rate": 6.091684239479234e-05,
+      "loss": 0.6928,
+      "mean_token_accuracy": 0.7807729482650757,
+      "num_tokens": 178648911.0,
+      "step": 75850
+    },
+    {
+      "epoch": 0.6955166406894655,
+      "learning_rate": 6.089850554689649e-05,
+      "loss": 0.7165,
+      "mean_token_accuracy": 0.7829519927501678,
+      "num_tokens": 178674426.0,
+      "step": 75860
+    },
+    {
+      "epoch": 0.6956083249289448,
+      "learning_rate": 6.088016869900064e-05,
+      "loss": 0.6753,
+      "mean_token_accuracy": 0.7965818285942078,
+      "num_tokens": 178699367.0,
+      "step": 75870
+    },
+    {
+      "epoch": 0.6957000091684239,
+      "learning_rate": 6.08618318511048e-05,
+      "loss": 0.6931,
+      "mean_token_accuracy": 0.7931291162967682,
+      "num_tokens": 178723765.0,
+      "step": 75880
+    },
+    {
+      "epoch": 0.6957916934079031,
+      "learning_rate": 6.084349500320895e-05,
+      "loss": 0.682,
+      "mean_token_accuracy": 0.7902142465114593,
+      "num_tokens": 178748686.0,
+      "step": 75890
+    },
+    {
+      "epoch": 0.6958833776473824,
+      "learning_rate": 6.08251581553131e-05,
+      "loss": 0.6778,
+      "mean_token_accuracy": 0.7876346409320831,
+      "num_tokens": 178772815.0,
+      "step": 75900
+    },
+    {
+      "epoch": 0.6959750618868616,
+      "learning_rate": 6.080682130741726e-05,
+      "loss": 0.6933,
+      "mean_token_accuracy": 0.787034273147583,
+      "num_tokens": 178797692.0,
+      "step": 75910
+    },
+    {
+      "epoch": 0.6960667461263409,
+      "learning_rate": 6.078848445952141e-05,
+      "loss": 0.6792,
+      "mean_token_accuracy": 0.7882838785648346,
+      "num_tokens": 178823018.0,
+      "step": 75920
+    },
+    {
+      "epoch": 0.6961584303658201,
+      "learning_rate": 6.077014761162556e-05,
+      "loss": 0.6752,
+      "mean_token_accuracy": 0.7874985039234161,
+      "num_tokens": 178848157.0,
+      "step": 75930
+    },
+    {
+      "epoch": 0.6962501146052994,
+      "learning_rate": 6.075181076372972e-05,
+      "loss": 0.6875,
+      "mean_token_accuracy": 0.7919121146202087,
+      "num_tokens": 178873190.0,
+      "step": 75940
+    },
+    {
+      "epoch": 0.6963417988447785,
+      "learning_rate": 6.073347391583387e-05,
+      "loss": 0.6829,
+      "mean_token_accuracy": 0.792927211523056,
+      "num_tokens": 178899419.0,
+      "step": 75950
+    },
+    {
+      "epoch": 0.6964334830842578,
+      "learning_rate": 6.0715137067938024e-05,
+      "loss": 0.6725,
+      "mean_token_accuracy": 0.7921659231185914,
+      "num_tokens": 178924798.0,
+      "step": 75960
+    },
+    {
+      "epoch": 0.696525167323737,
+      "learning_rate": 6.069680022004218e-05,
+      "loss": 0.6586,
+      "mean_token_accuracy": 0.7956588447093964,
+      "num_tokens": 178950818.0,
+      "step": 75970
+    },
+    {
+      "epoch": 0.6966168515632163,
+      "learning_rate": 6.0678463372146335e-05,
+      "loss": 0.6736,
+      "mean_token_accuracy": 0.7838387608528137,
+      "num_tokens": 178976757.0,
+      "step": 75980
+    },
+    {
+      "epoch": 0.6967085358026955,
+      "learning_rate": 6.066012652425048e-05,
+      "loss": 0.7003,
+      "mean_token_accuracy": 0.7866319477558136,
+      "num_tokens": 179001693.0,
+      "step": 75990
+    },
+    {
+      "epoch": 0.6968002200421748,
+      "learning_rate": 6.064178967635463e-05,
+      "loss": 0.6691,
+      "mean_token_accuracy": 0.7886835098266601,
+      "num_tokens": 179026674.0,
+      "step": 76000
+    },
+    {
+      "epoch": 0.6968919042816539,
+      "learning_rate": 6.062345282845879e-05,
+      "loss": 0.6592,
+      "mean_token_accuracy": 0.7934810280799866,
+      "num_tokens": 179051807.0,
+      "step": 76010
+    },
+    {
+      "epoch": 0.6969835885211332,
+      "learning_rate": 6.060511598056294e-05,
+      "loss": 0.6687,
+      "mean_token_accuracy": 0.7910148620605468,
+      "num_tokens": 179077149.0,
+      "step": 76020
+    },
+    {
+      "epoch": 0.6970752727606124,
+      "learning_rate": 6.0586779132667094e-05,
+      "loss": 0.6614,
+      "mean_token_accuracy": 0.7928116798400879,
+      "num_tokens": 179102955.0,
+      "step": 76030
+    },
+    {
+      "epoch": 0.6971669570000917,
+      "learning_rate": 6.056844228477125e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.7920890152454376,
+      "num_tokens": 179128374.0,
+      "step": 76040
+    },
+    {
+      "epoch": 0.6972586412395709,
+      "learning_rate": 6.0550105436875405e-05,
+      "loss": 0.6918,
+      "mean_token_accuracy": 0.7865221977233887,
+      "num_tokens": 179153390.0,
+      "step": 76050
+    },
+    {
+      "epoch": 0.6973503254790502,
+      "learning_rate": 6.053176858897956e-05,
+      "loss": 0.6735,
+      "mean_token_accuracy": 0.7937344133853912,
+      "num_tokens": 179178350.0,
+      "step": 76060
+    },
+    {
+      "epoch": 0.6974420097185294,
+      "learning_rate": 6.0513431741083715e-05,
+      "loss": 0.664,
+      "mean_token_accuracy": 0.7925346910953521,
+      "num_tokens": 179202990.0,
+      "step": 76070
+    },
+    {
+      "epoch": 0.6975336939580086,
+      "learning_rate": 6.049509489318787e-05,
+      "loss": 0.681,
+      "mean_token_accuracy": 0.7910955011844635,
+      "num_tokens": 179228708.0,
+      "step": 76080
+    },
+    {
+      "epoch": 0.6976253781974878,
+      "learning_rate": 6.047675804529201e-05,
+      "loss": 0.7205,
+      "mean_token_accuracy": 0.7769014358520507,
+      "num_tokens": 179254154.0,
+      "step": 76090
+    },
+    {
+      "epoch": 0.6977170624369671,
+      "learning_rate": 6.045842119739618e-05,
+      "loss": 0.6712,
+      "mean_token_accuracy": 0.7906412899494171,
+      "num_tokens": 179279894.0,
+      "step": 76100
+    },
+    {
+      "epoch": 0.6978087466764463,
+      "learning_rate": 6.044008434950032e-05,
+      "loss": 0.6632,
+      "mean_token_accuracy": 0.7903281688690186,
+      "num_tokens": 179304958.0,
+      "step": 76110
+    },
+    {
+      "epoch": 0.6979004309159256,
+      "learning_rate": 6.0421747501604475e-05,
+      "loss": 0.6502,
+      "mean_token_accuracy": 0.7969510674476623,
+      "num_tokens": 179329896.0,
+      "step": 76120
+    },
+    {
+      "epoch": 0.6979921151554048,
+      "learning_rate": 6.0403410653708627e-05,
+      "loss": 0.6948,
+      "mean_token_accuracy": 0.7847719728946686,
+      "num_tokens": 179354591.0,
+      "step": 76130
+    },
+    {
+      "epoch": 0.6980837993948841,
+      "learning_rate": 6.0385073805812785e-05,
+      "loss": 0.7005,
+      "mean_token_accuracy": 0.7879184305667877,
+      "num_tokens": 179380583.0,
+      "step": 76140
+    },
+    {
+      "epoch": 0.6981754836343632,
+      "learning_rate": 6.036673695791694e-05,
+      "loss": 0.6927,
+      "mean_token_accuracy": 0.7896237909793854,
+      "num_tokens": 179405606.0,
+      "step": 76150
+    },
+    {
+      "epoch": 0.6982671678738425,
+      "learning_rate": 6.034840011002109e-05,
+      "loss": 0.6773,
+      "mean_token_accuracy": 0.7949445068836212,
+      "num_tokens": 179431105.0,
+      "step": 76160
+    },
+    {
+      "epoch": 0.6983588521133217,
+      "learning_rate": 6.033006326212525e-05,
+      "loss": 0.67,
+      "mean_token_accuracy": 0.7901997804641724,
+      "num_tokens": 179456593.0,
+      "step": 76170
+    },
+    {
+      "epoch": 0.698450536352801,
+      "learning_rate": 6.03117264142294e-05,
+      "loss": 0.6815,
+      "mean_token_accuracy": 0.7900418698787689,
+      "num_tokens": 179481639.0,
+      "step": 76180
+    },
+    {
+      "epoch": 0.6985422205922802,
+      "learning_rate": 6.0293389566333545e-05,
+      "loss": 0.6631,
+      "mean_token_accuracy": 0.7938229978084564,
+      "num_tokens": 179507111.0,
+      "step": 76190
+    },
+    {
+      "epoch": 0.6986339048317595,
+      "learning_rate": 6.027505271843771e-05,
+      "loss": 0.6655,
+      "mean_token_accuracy": 0.791388314962387,
+      "num_tokens": 179532400.0,
+      "step": 76200
+    },
+    {
+      "epoch": 0.6987255890712386,
+      "learning_rate": 6.0256715870541855e-05,
+      "loss": 0.6608,
+      "mean_token_accuracy": 0.7964693009853363,
+      "num_tokens": 179557888.0,
+      "step": 76210
+    },
+    {
+      "epoch": 0.6988172733107179,
+      "learning_rate": 6.023837902264601e-05,
+      "loss": 0.6949,
+      "mean_token_accuracy": 0.7877890586853027,
+      "num_tokens": 179583210.0,
+      "step": 76220
+    },
+    {
+      "epoch": 0.6989089575501971,
+      "learning_rate": 6.0220042174750166e-05,
+      "loss": 0.706,
+      "mean_token_accuracy": 0.7825822949409484,
+      "num_tokens": 179609231.0,
+      "step": 76230
+    },
+    {
+      "epoch": 0.6990006417896764,
+      "learning_rate": 6.020170532685432e-05,
+      "loss": 0.6495,
+      "mean_token_accuracy": 0.8009159028530121,
+      "num_tokens": 179634299.0,
+      "step": 76240
+    },
+    {
+      "epoch": 0.6990923260291556,
+      "learning_rate": 6.018336847895847e-05,
+      "loss": 0.6879,
+      "mean_token_accuracy": 0.7836112380027771,
+      "num_tokens": 179659224.0,
+      "step": 76250
+    },
+    {
+      "epoch": 0.6991840102686349,
+      "learning_rate": 6.016503163106262e-05,
+      "loss": 0.7075,
+      "mean_token_accuracy": 0.7844682037830353,
+      "num_tokens": 179684487.0,
+      "step": 76260
+    },
+    {
+      "epoch": 0.6992756945081141,
+      "learning_rate": 6.014669478316678e-05,
+      "loss": 0.7047,
+      "mean_token_accuracy": 0.7841387391090393,
+      "num_tokens": 179709986.0,
+      "step": 76270
+    },
+    {
+      "epoch": 0.6993673787475932,
+      "learning_rate": 6.012835793527093e-05,
+      "loss": 0.6919,
+      "mean_token_accuracy": 0.7843936502933502,
+      "num_tokens": 179735307.0,
+      "step": 76280
+    },
+    {
+      "epoch": 0.6994590629870725,
+      "learning_rate": 6.011002108737508e-05,
+      "loss": 0.6909,
+      "mean_token_accuracy": 0.7878920257091522,
+      "num_tokens": 179761106.0,
+      "step": 76290
+    },
+    {
+      "epoch": 0.6995507472265518,
+      "learning_rate": 6.009168423947924e-05,
+      "loss": 0.6882,
+      "mean_token_accuracy": 0.7916716039180756,
+      "num_tokens": 179786240.0,
+      "step": 76300
+    },
+    {
+      "epoch": 0.699642431466031,
+      "learning_rate": 6.007334739158339e-05,
+      "loss": 0.6609,
+      "mean_token_accuracy": 0.7954664885997772,
+      "num_tokens": 179811970.0,
+      "step": 76310
+    },
+    {
+      "epoch": 0.6997341157055103,
+      "learning_rate": 6.005501054368754e-05,
+      "loss": 0.7142,
+      "mean_token_accuracy": 0.7827772855758667,
+      "num_tokens": 179836667.0,
+      "step": 76320
+    },
+    {
+      "epoch": 0.6998257999449895,
+      "learning_rate": 6.00366736957917e-05,
+      "loss": 0.6555,
+      "mean_token_accuracy": 0.7956376850605011,
+      "num_tokens": 179861968.0,
+      "step": 76330
+    },
+    {
+      "epoch": 0.6999174841844686,
+      "learning_rate": 6.001833684789585e-05,
+      "loss": 0.6638,
+      "mean_token_accuracy": 0.7971653342247009,
+      "num_tokens": 179887636.0,
+      "step": 76340
+    },
+    {
+      "epoch": 0.7000091684239479,
+      "learning_rate": 6e-05,
+      "loss": 0.688,
+      "mean_token_accuracy": 0.785324776172638,
+      "num_tokens": 179912698.0,
+      "step": 76350
+    },
+    {
+      "epoch": 0.7001008526634271,
+      "learning_rate": 5.998166315210416e-05,
+      "loss": 0.6523,
+      "mean_token_accuracy": 0.7965922951698303,
+      "num_tokens": 179938140.0,
+      "step": 76360
+    },
+    {
+      "epoch": 0.7001925369029064,
+      "learning_rate": 5.996332630420831e-05,
+      "loss": 0.6551,
+      "mean_token_accuracy": 0.7932762920856475,
+      "num_tokens": 179963079.0,
+      "step": 76370
+    },
+    {
+      "epoch": 0.7002842211423856,
+      "learning_rate": 5.9944989456312464e-05,
+      "loss": 0.6949,
+      "mean_token_accuracy": 0.7842654764652253,
+      "num_tokens": 179988743.0,
+      "step": 76380
+    },
+    {
+      "epoch": 0.7003759053818649,
+      "learning_rate": 5.992665260841661e-05,
+      "loss": 0.6589,
+      "mean_token_accuracy": 0.7966576039791107,
+      "num_tokens": 180014016.0,
+      "step": 76390
+    },
+    {
+      "epoch": 0.7004675896213441,
+      "learning_rate": 5.9908315760520774e-05,
+      "loss": 0.6792,
+      "mean_token_accuracy": 0.7953386723995208,
+      "num_tokens": 180039818.0,
+      "step": 76400
+    },
+    {
+      "epoch": 0.7005592738608233,
+      "learning_rate": 5.988997891262492e-05,
+      "loss": 0.6695,
+      "mean_token_accuracy": 0.79258491396904,
+      "num_tokens": 180065228.0,
+      "step": 76410
+    },
+    {
+      "epoch": 0.7006509581003025,
+      "learning_rate": 5.987164206472907e-05,
+      "loss": 0.6875,
+      "mean_token_accuracy": 0.7920036256313324,
+      "num_tokens": 180090602.0,
+      "step": 76420
+    },
+    {
+      "epoch": 0.7007426423397818,
+      "learning_rate": 5.985330521683323e-05,
+      "loss": 0.6319,
+      "mean_token_accuracy": 0.8041131377220154,
+      "num_tokens": 180115623.0,
+      "step": 76430
+    },
+    {
+      "epoch": 0.700834326579261,
+      "learning_rate": 5.983496836893738e-05,
+      "loss": 0.6781,
+      "mean_token_accuracy": 0.7869494497776032,
+      "num_tokens": 180140359.0,
+      "step": 76440
+    },
+    {
+      "epoch": 0.7009260108187403,
+      "learning_rate": 5.9816631521041534e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.7869329512119293,
+      "num_tokens": 180164972.0,
+      "step": 76450
+    },
+    {
+      "epoch": 0.7010176950582195,
+      "learning_rate": 5.979829467314569e-05,
+      "loss": 0.6601,
+      "mean_token_accuracy": 0.796118974685669,
+      "num_tokens": 180189791.0,
+      "step": 76460
+    },
+    {
+      "epoch": 0.7011093792976987,
+      "learning_rate": 5.9779957825249844e-05,
+      "loss": 0.6826,
+      "mean_token_accuracy": 0.7869082808494567,
+      "num_tokens": 180215349.0,
+      "step": 76470
+    },
+    {
+      "epoch": 0.7012010635371779,
+      "learning_rate": 5.9761620977353996e-05,
+      "loss": 0.6826,
+      "mean_token_accuracy": 0.7972529768943787,
+      "num_tokens": 180240882.0,
+      "step": 76480
+    },
+    {
+      "epoch": 0.7012927477766572,
+      "learning_rate": 5.974328412945814e-05,
+      "loss": 0.6973,
+      "mean_token_accuracy": 0.7850015044212342,
+      "num_tokens": 180266371.0,
+      "step": 76490
+    },
+    {
+      "epoch": 0.7013844320161364,
+      "learning_rate": 5.972494728156231e-05,
+      "loss": 0.657,
+      "mean_token_accuracy": 0.7914840102195739,
+      "num_tokens": 180291410.0,
+      "step": 76500
+    },
+    {
+      "epoch": 0.7014761162556157,
+      "learning_rate": 5.970661043366645e-05,
+      "loss": 0.6154,
+      "mean_token_accuracy": 0.8100368440151214,
+      "num_tokens": 180317090.0,
+      "step": 76510
+    },
+    {
+      "epoch": 0.7015678004950949,
+      "learning_rate": 5.9688273585770604e-05,
+      "loss": 0.6448,
+      "mean_token_accuracy": 0.8013284623622894,
+      "num_tokens": 180342333.0,
+      "step": 76520
+    },
+    {
+      "epoch": 0.7016594847345742,
+      "learning_rate": 5.966993673787476e-05,
+      "loss": 0.6946,
+      "mean_token_accuracy": 0.7870315790176392,
+      "num_tokens": 180367314.0,
+      "step": 76530
+    },
+    {
+      "epoch": 0.7017511689740533,
+      "learning_rate": 5.9651599889978914e-05,
+      "loss": 0.6896,
+      "mean_token_accuracy": 0.7894434034824371,
+      "num_tokens": 180392770.0,
+      "step": 76540
+    },
+    {
+      "epoch": 0.7018428532135326,
+      "learning_rate": 5.9633263042083066e-05,
+      "loss": 0.6748,
+      "mean_token_accuracy": 0.7907483339309692,
+      "num_tokens": 180418879.0,
+      "step": 76550
+    },
+    {
+      "epoch": 0.7019345374530118,
+      "learning_rate": 5.9614926194187225e-05,
+      "loss": 0.6828,
+      "mean_token_accuracy": 0.7872988939285278,
+      "num_tokens": 180444107.0,
+      "step": 76560
+    },
+    {
+      "epoch": 0.7020262216924911,
+      "learning_rate": 5.9596589346291376e-05,
+      "loss": 0.684,
+      "mean_token_accuracy": 0.791980254650116,
+      "num_tokens": 180469720.0,
+      "step": 76570
+    },
+    {
+      "epoch": 0.7021179059319703,
+      "learning_rate": 5.957825249839553e-05,
+      "loss": 0.6655,
+      "mean_token_accuracy": 0.7936712384223938,
+      "num_tokens": 180495426.0,
+      "step": 76580
+    },
+    {
+      "epoch": 0.7022095901714496,
+      "learning_rate": 5.955991565049969e-05,
+      "loss": 0.6735,
+      "mean_token_accuracy": 0.794761061668396,
+      "num_tokens": 180520531.0,
+      "step": 76590
+    },
+    {
+      "epoch": 0.7023012744109287,
+      "learning_rate": 5.954157880260384e-05,
+      "loss": 0.6687,
+      "mean_token_accuracy": 0.7890596449375152,
+      "num_tokens": 180545814.0,
+      "step": 76600
+    },
+    {
+      "epoch": 0.702392958650408,
+      "learning_rate": 5.9523241954707984e-05,
+      "loss": 0.6626,
+      "mean_token_accuracy": 0.7914606809616089,
+      "num_tokens": 180570631.0,
+      "step": 76610
+    },
+    {
+      "epoch": 0.7024846428898872,
+      "learning_rate": 5.9504905106812136e-05,
+      "loss": 0.6723,
+      "mean_token_accuracy": 0.7984126508235931,
+      "num_tokens": 180596049.0,
+      "step": 76620
+    },
+    {
+      "epoch": 0.7025763271293665,
+      "learning_rate": 5.9486568258916294e-05,
+      "loss": 0.6636,
+      "mean_token_accuracy": 0.7984091997146606,
+      "num_tokens": 180622258.0,
+      "step": 76630
+    },
+    {
+      "epoch": 0.7026680113688457,
+      "learning_rate": 5.9468231411020446e-05,
+      "loss": 0.6718,
+      "mean_token_accuracy": 0.7853006243705749,
+      "num_tokens": 180646728.0,
+      "step": 76640
+    },
+    {
+      "epoch": 0.702759695608325,
+      "learning_rate": 5.94498945631246e-05,
+      "loss": 0.6951,
+      "mean_token_accuracy": 0.7850263833999633,
+      "num_tokens": 180671574.0,
+      "step": 76650
+    },
+    {
+      "epoch": 0.7028513798478042,
+      "learning_rate": 5.943155771522876e-05,
+      "loss": 0.6799,
+      "mean_token_accuracy": 0.7914558529853821,
+      "num_tokens": 180696395.0,
+      "step": 76660
+    },
+    {
+      "epoch": 0.7029430640872834,
+      "learning_rate": 5.941322086733291e-05,
+      "loss": 0.7179,
+      "mean_token_accuracy": 0.781195193529129,
+      "num_tokens": 180721002.0,
+      "step": 76670
+    },
+    {
+      "epoch": 0.7030347483267626,
+      "learning_rate": 5.939488401943706e-05,
+      "loss": 0.6632,
+      "mean_token_accuracy": 0.7917946457862854,
+      "num_tokens": 180746339.0,
+      "step": 76680
+    },
+    {
+      "epoch": 0.7031264325662419,
+      "learning_rate": 5.937654717154122e-05,
+      "loss": 0.6678,
+      "mean_token_accuracy": 0.792183643579483,
+      "num_tokens": 180770717.0,
+      "step": 76690
+    },
+    {
+      "epoch": 0.7032181168057211,
+      "learning_rate": 5.935821032364537e-05,
+      "loss": 0.7112,
+      "mean_token_accuracy": 0.784171599149704,
+      "num_tokens": 180795691.0,
+      "step": 76700
+    },
+    {
+      "epoch": 0.7033098010452004,
+      "learning_rate": 5.9339873475749516e-05,
+      "loss": 0.6619,
+      "mean_token_accuracy": 0.7923989355564117,
+      "num_tokens": 180821094.0,
+      "step": 76710
+    },
+    {
+      "epoch": 0.7034014852846796,
+      "learning_rate": 5.932153662785368e-05,
+      "loss": 0.6319,
+      "mean_token_accuracy": 0.8048372387886047,
+      "num_tokens": 180846700.0,
+      "step": 76720
+    },
+    {
+      "epoch": 0.7034931695241587,
+      "learning_rate": 5.930319977995783e-05,
+      "loss": 0.6769,
+      "mean_token_accuracy": 0.7893557667732238,
+      "num_tokens": 180872282.0,
+      "step": 76730
+    },
+    {
+      "epoch": 0.703584853763638,
+      "learning_rate": 5.928486293206198e-05,
+      "loss": 0.7249,
+      "mean_token_accuracy": 0.7826806366443634,
+      "num_tokens": 180897235.0,
+      "step": 76740
+    },
+    {
+      "epoch": 0.7036765380031172,
+      "learning_rate": 5.926652608416613e-05,
+      "loss": 0.6778,
+      "mean_token_accuracy": 0.7904228389263153,
+      "num_tokens": 180922529.0,
+      "step": 76750
+    },
+    {
+      "epoch": 0.7037682222425965,
+      "learning_rate": 5.924818923627029e-05,
+      "loss": 0.6492,
+      "mean_token_accuracy": 0.7943959832191467,
+      "num_tokens": 180948432.0,
+      "step": 76760
+    },
+    {
+      "epoch": 0.7038599064820757,
+      "learning_rate": 5.922985238837444e-05,
+      "loss": 0.6322,
+      "mean_token_accuracy": 0.7964687287807465,
+      "num_tokens": 180973094.0,
+      "step": 76770
+    },
+    {
+      "epoch": 0.703951590721555,
+      "learning_rate": 5.921151554047859e-05,
+      "loss": 0.6803,
+      "mean_token_accuracy": 0.7894576191902161,
+      "num_tokens": 180997814.0,
+      "step": 76780
+    },
+    {
+      "epoch": 0.7040432749610342,
+      "learning_rate": 5.919317869258275e-05,
+      "loss": 0.6803,
+      "mean_token_accuracy": 0.7830595135688782,
+      "num_tokens": 181021971.0,
+      "step": 76790
+    },
+    {
+      "epoch": 0.7041349592005134,
+      "learning_rate": 5.91748418446869e-05,
+      "loss": 0.6886,
+      "mean_token_accuracy": 0.7863920331001282,
+      "num_tokens": 181047177.0,
+      "step": 76800
+    },
+    {
+      "epoch": 0.7042266434399926,
+      "learning_rate": 5.915650499679105e-05,
+      "loss": 0.6879,
+      "mean_token_accuracy": 0.7897522926330567,
+      "num_tokens": 181072613.0,
+      "step": 76810
+    },
+    {
+      "epoch": 0.7043183276794719,
+      "learning_rate": 5.9138168148895214e-05,
+      "loss": 0.7155,
+      "mean_token_accuracy": 0.7848186373710633,
+      "num_tokens": 181098545.0,
+      "step": 76820
+    },
+    {
+      "epoch": 0.7044100119189511,
+      "learning_rate": 5.911983130099936e-05,
+      "loss": 0.6931,
+      "mean_token_accuracy": 0.7870139658451081,
+      "num_tokens": 181123650.0,
+      "step": 76830
+    },
+    {
+      "epoch": 0.7045016961584304,
+      "learning_rate": 5.910149445310351e-05,
+      "loss": 0.679,
+      "mean_token_accuracy": 0.7967150986194611,
+      "num_tokens": 181148046.0,
+      "step": 76840
+    },
+    {
+      "epoch": 0.7045933803979096,
+      "learning_rate": 5.9083157605207676e-05,
+      "loss": 0.6688,
+      "mean_token_accuracy": 0.793145477771759,
+      "num_tokens": 181173276.0,
+      "step": 76850
+    },
+    {
+      "epoch": 0.7046850646373888,
+      "learning_rate": 5.906482075731182e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.7901999592781067,
+      "num_tokens": 181198838.0,
+      "step": 76860
+    },
+    {
+      "epoch": 0.704776748876868,
+      "learning_rate": 5.904648390941597e-05,
+      "loss": 0.6761,
+      "mean_token_accuracy": 0.7846197009086608,
+      "num_tokens": 181224476.0,
+      "step": 76870
+    },
+    {
+      "epoch": 0.7048684331163473,
+      "learning_rate": 5.9028147061520125e-05,
+      "loss": 0.6622,
+      "mean_token_accuracy": 0.7943405210971832,
+      "num_tokens": 181248987.0,
+      "step": 76880
+    },
+    {
+      "epoch": 0.7049601173558265,
+      "learning_rate": 5.9009810213624284e-05,
+      "loss": 0.7147,
+      "mean_token_accuracy": 0.7793645620346069,
+      "num_tokens": 181274286.0,
+      "step": 76890
+    },
+    {
+      "epoch": 0.7050518015953058,
+      "learning_rate": 5.8991473365728436e-05,
+      "loss": 0.6614,
+      "mean_token_accuracy": 0.7946879148483277,
+      "num_tokens": 181299549.0,
+      "step": 76900
+    },
+    {
+      "epoch": 0.705143485834785,
+      "learning_rate": 5.897313651783258e-05,
+      "loss": 0.7211,
+      "mean_token_accuracy": 0.7797550737857819,
+      "num_tokens": 181324757.0,
+      "step": 76910
+    },
+    {
+      "epoch": 0.7052351700742643,
+      "learning_rate": 5.8954799669936746e-05,
+      "loss": 0.6767,
+      "mean_token_accuracy": 0.7889573633670807,
+      "num_tokens": 181349457.0,
+      "step": 76920
+    },
+    {
+      "epoch": 0.7053268543137434,
+      "learning_rate": 5.893646282204089e-05,
+      "loss": 0.6817,
+      "mean_token_accuracy": 0.7939691424369812,
+      "num_tokens": 181374695.0,
+      "step": 76930
+    },
+    {
+      "epoch": 0.7054185385532227,
+      "learning_rate": 5.891812597414504e-05,
+      "loss": 0.6669,
+      "mean_token_accuracy": 0.7937139928340912,
+      "num_tokens": 181400364.0,
+      "step": 76940
+    },
+    {
+      "epoch": 0.7055102227927019,
+      "learning_rate": 5.889978912624921e-05,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.7926955759525299,
+      "num_tokens": 181425822.0,
+      "step": 76950
+    },
+    {
+      "epoch": 0.7056019070321812,
+      "learning_rate": 5.8881452278353354e-05,
+      "loss": 0.6693,
+      "mean_token_accuracy": 0.7917500972747803,
+      "num_tokens": 181451487.0,
+      "step": 76960
+    },
+    {
+      "epoch": 0.7056935912716604,
+      "learning_rate": 5.8863115430457505e-05,
+      "loss": 0.7228,
+      "mean_token_accuracy": 0.7817276835441589,
+      "num_tokens": 181477328.0,
+      "step": 76970
+    },
+    {
+      "epoch": 0.7057852755111397,
+      "learning_rate": 5.8844778582561664e-05,
+      "loss": 0.6799,
+      "mean_token_accuracy": 0.7889807641506195,
+      "num_tokens": 181502502.0,
+      "step": 76980
+    },
+    {
+      "epoch": 0.7058769597506188,
+      "learning_rate": 5.8826441734665816e-05,
+      "loss": 0.6712,
+      "mean_token_accuracy": 0.7955949425697326,
+      "num_tokens": 181528497.0,
+      "step": 76990
+    },
+    {
+      "epoch": 0.7059686439900981,
+      "learning_rate": 5.880810488676997e-05,
+      "loss": 0.6782,
+      "mean_token_accuracy": 0.7896027624607086,
+      "num_tokens": 181552879.0,
+      "step": 77000
+    },
+    {
+      "epoch": 0.7060603282295773,
+      "learning_rate": 5.878976803887411e-05,
+      "loss": 0.6735,
+      "mean_token_accuracy": 0.7945159912109375,
+      "num_tokens": 181578229.0,
+      "step": 77010
+    },
+    {
+      "epoch": 0.7061520124690566,
+      "learning_rate": 5.877143119097828e-05,
+      "loss": 0.689,
+      "mean_token_accuracy": 0.791348934173584,
+      "num_tokens": 181603749.0,
+      "step": 77020
+    },
+    {
+      "epoch": 0.7062436967085358,
+      "learning_rate": 5.875309434308242e-05,
+      "loss": 0.6732,
+      "mean_token_accuracy": 0.7935809016227722,
+      "num_tokens": 181629104.0,
+      "step": 77030
+    },
+    {
+      "epoch": 0.7063353809480151,
+      "learning_rate": 5.8734757495186575e-05,
+      "loss": 0.6819,
+      "mean_token_accuracy": 0.7947469830513001,
+      "num_tokens": 181653839.0,
+      "step": 77040
+    },
+    {
+      "epoch": 0.7064270651874943,
+      "learning_rate": 5.871642064729074e-05,
+      "loss": 0.7116,
+      "mean_token_accuracy": 0.7810812950134277,
+      "num_tokens": 181679173.0,
+      "step": 77050
+    },
+    {
+      "epoch": 0.7065187494269735,
+      "learning_rate": 5.8698083799394886e-05,
+      "loss": 0.6704,
+      "mean_token_accuracy": 0.7936798632144928,
+      "num_tokens": 181704820.0,
+      "step": 77060
+    },
+    {
+      "epoch": 0.7066104336664527,
+      "learning_rate": 5.867974695149904e-05,
+      "loss": 0.6719,
+      "mean_token_accuracy": 0.7947412610054017,
+      "num_tokens": 181729585.0,
+      "step": 77070
+    },
+    {
+      "epoch": 0.706702117905932,
+      "learning_rate": 5.8661410103603196e-05,
+      "loss": 0.6822,
+      "mean_token_accuracy": 0.789141708612442,
+      "num_tokens": 181754007.0,
+      "step": 77080
+    },
+    {
+      "epoch": 0.7067938021454112,
+      "learning_rate": 5.864307325570735e-05,
+      "loss": 0.6848,
+      "mean_token_accuracy": 0.7885053396224976,
+      "num_tokens": 181779081.0,
+      "step": 77090
+    },
+    {
+      "epoch": 0.7068854863848905,
+      "learning_rate": 5.86247364078115e-05,
+      "loss": 0.6794,
+      "mean_token_accuracy": 0.7914442479610443,
+      "num_tokens": 181804143.0,
+      "step": 77100
+    },
+    {
+      "epoch": 0.7069771706243697,
+      "learning_rate": 5.8606399559915645e-05,
+      "loss": 0.6618,
+      "mean_token_accuracy": 0.7903600037097931,
+      "num_tokens": 181829360.0,
+      "step": 77110
+    },
+    {
+      "epoch": 0.7070688548638489,
+      "learning_rate": 5.858806271201981e-05,
+      "loss": 0.6787,
+      "mean_token_accuracy": 0.7898567736148834,
+      "num_tokens": 181854744.0,
+      "step": 77120
+    },
+    {
+      "epoch": 0.7071605391033281,
+      "learning_rate": 5.8569725864123956e-05,
+      "loss": 0.6817,
+      "mean_token_accuracy": 0.789820522069931,
+      "num_tokens": 181878917.0,
+      "step": 77130
+    },
+    {
+      "epoch": 0.7072522233428074,
+      "learning_rate": 5.855138901622811e-05,
+      "loss": 0.6598,
+      "mean_token_accuracy": 0.796710067987442,
+      "num_tokens": 181904273.0,
+      "step": 77140
+    },
+    {
+      "epoch": 0.7073439075822866,
+      "learning_rate": 5.853305216833227e-05,
+      "loss": 0.663,
+      "mean_token_accuracy": 0.7935916244983673,
+      "num_tokens": 181929692.0,
+      "step": 77150
+    },
+    {
+      "epoch": 0.7074355918217659,
+      "learning_rate": 5.851471532043642e-05,
+      "loss": 0.6619,
+      "mean_token_accuracy": 0.7959802508354187,
+      "num_tokens": 181954993.0,
+      "step": 77160
+    },
+    {
+      "epoch": 0.7075272760612451,
+      "learning_rate": 5.849637847254057e-05,
+      "loss": 0.6688,
+      "mean_token_accuracy": 0.7960519790649414,
+      "num_tokens": 181979921.0,
+      "step": 77170
+    },
+    {
+      "epoch": 0.7076189603007244,
+      "learning_rate": 5.847804162464473e-05,
+      "loss": 0.6837,
+      "mean_token_accuracy": 0.794447660446167,
+      "num_tokens": 182004918.0,
+      "step": 77180
+    },
+    {
+      "epoch": 0.7077106445402035,
+      "learning_rate": 5.845970477674888e-05,
+      "loss": 0.6876,
+      "mean_token_accuracy": 0.784761905670166,
+      "num_tokens": 182029610.0,
+      "step": 77190
+    },
+    {
+      "epoch": 0.7078023287796827,
+      "learning_rate": 5.844136792885303e-05,
+      "loss": 0.6725,
+      "mean_token_accuracy": 0.7931979894638062,
+      "num_tokens": 182054775.0,
+      "step": 77200
+    },
+    {
+      "epoch": 0.707894013019162,
+      "learning_rate": 5.842303108095719e-05,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7726520478725434,
+      "num_tokens": 182079881.0,
+      "step": 77210
+    },
+    {
+      "epoch": 0.7079856972586412,
+      "learning_rate": 5.840469423306134e-05,
+      "loss": 0.6818,
+      "mean_token_accuracy": 0.7893115043640136,
+      "num_tokens": 182105885.0,
+      "step": 77220
+    },
+    {
+      "epoch": 0.7080773814981205,
+      "learning_rate": 5.838635738516549e-05,
+      "loss": 0.7009,
+      "mean_token_accuracy": 0.7863038897514343,
+      "num_tokens": 182130750.0,
+      "step": 77230
+    },
+    {
+      "epoch": 0.7081690657375997,
+      "learning_rate": 5.836802053726964e-05,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.7787288665771485,
+      "num_tokens": 182155639.0,
+      "step": 77240
+    },
+    {
+      "epoch": 0.7082607499770789,
+      "learning_rate": 5.8349683689373805e-05,
+      "loss": 0.664,
+      "mean_token_accuracy": 0.7939791321754456,
+      "num_tokens": 182179991.0,
+      "step": 77250
+    },
+    {
+      "epoch": 0.7083524342165581,
+      "learning_rate": 5.833134684147795e-05,
+      "loss": 0.6617,
+      "mean_token_accuracy": 0.7917009890079498,
+      "num_tokens": 182204313.0,
+      "step": 77260
+    },
+    {
+      "epoch": 0.7084441184560374,
+      "learning_rate": 5.83130099935821e-05,
+      "loss": 0.6856,
+      "mean_token_accuracy": 0.7876560926437378,
+      "num_tokens": 182228841.0,
+      "step": 77270
+    },
+    {
+      "epoch": 0.7085358026955166,
+      "learning_rate": 5.829467314568626e-05,
+      "loss": 0.7318,
+      "mean_token_accuracy": 0.7744341731071472,
+      "num_tokens": 182253508.0,
+      "step": 77280
+    },
+    {
+      "epoch": 0.7086274869349959,
+      "learning_rate": 5.827633629779041e-05,
+      "loss": 0.7052,
+      "mean_token_accuracy": 0.7833770155906677,
+      "num_tokens": 182278515.0,
+      "step": 77290
+    },
+    {
+      "epoch": 0.7087191711744751,
+      "learning_rate": 5.8257999449894564e-05,
+      "loss": 0.6478,
+      "mean_token_accuracy": 0.7999937832355499,
+      "num_tokens": 182304055.0,
+      "step": 77300
+    },
+    {
+      "epoch": 0.7088108554139544,
+      "learning_rate": 5.823966260199872e-05,
+      "loss": 0.6582,
+      "mean_token_accuracy": 0.7930777072906494,
+      "num_tokens": 182329142.0,
+      "step": 77310
+    },
+    {
+      "epoch": 0.7089025396534335,
+      "learning_rate": 5.8221325754102875e-05,
+      "loss": 0.6816,
+      "mean_token_accuracy": 0.7900965869426727,
+      "num_tokens": 182353892.0,
+      "step": 77320
+    },
+    {
+      "epoch": 0.7089942238929128,
+      "learning_rate": 5.820298890620702e-05,
+      "loss": 0.6786,
+      "mean_token_accuracy": 0.7906683802604675,
+      "num_tokens": 182378853.0,
+      "step": 77330
+    },
+    {
+      "epoch": 0.709085908132392,
+      "learning_rate": 5.8184652058311185e-05,
+      "loss": 0.6946,
+      "mean_token_accuracy": 0.7840655148029327,
+      "num_tokens": 182404074.0,
+      "step": 77340
+    },
+    {
+      "epoch": 0.7091775923718713,
+      "learning_rate": 5.816631521041534e-05,
+      "loss": 0.6715,
+      "mean_token_accuracy": 0.7909890592098237,
+      "num_tokens": 182429726.0,
+      "step": 77350
+    },
+    {
+      "epoch": 0.7092692766113505,
+      "learning_rate": 5.814797836251948e-05,
+      "loss": 0.6804,
+      "mean_token_accuracy": 0.7889102041721344,
+      "num_tokens": 182455672.0,
+      "step": 77360
+    },
+    {
+      "epoch": 0.7093609608508298,
+      "learning_rate": 5.8129641514623634e-05,
+      "loss": 0.6953,
+      "mean_token_accuracy": 0.7895851194858551,
+      "num_tokens": 182480786.0,
+      "step": 77370
+    },
+    {
+      "epoch": 0.709452645090309,
+      "learning_rate": 5.811130466672779e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.7942793130874634,
+      "num_tokens": 182506078.0,
+      "step": 77380
+    },
+    {
+      "epoch": 0.7095443293297882,
+      "learning_rate": 5.8092967818831945e-05,
+      "loss": 0.6555,
+      "mean_token_accuracy": 0.7945311725139618,
+      "num_tokens": 182531261.0,
+      "step": 77390
+    },
+    {
+      "epoch": 0.7096360135692674,
+      "learning_rate": 5.80746309709361e-05,
+      "loss": 0.651,
+      "mean_token_accuracy": 0.7952611863613128,
+      "num_tokens": 182556398.0,
+      "step": 77400
+    },
+    {
+      "epoch": 0.7097276978087467,
+      "learning_rate": 5.8056294123040255e-05,
+      "loss": 0.6794,
+      "mean_token_accuracy": 0.7877298772335053,
+      "num_tokens": 182581226.0,
+      "step": 77410
+    },
+    {
+      "epoch": 0.7098193820482259,
+      "learning_rate": 5.803795727514441e-05,
+      "loss": 0.6868,
+      "mean_token_accuracy": 0.7852868854999542,
+      "num_tokens": 182606563.0,
+      "step": 77420
+    },
+    {
+      "epoch": 0.7099110662877052,
+      "learning_rate": 5.801962042724855e-05,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.7888477265834808,
+      "num_tokens": 182631515.0,
+      "step": 77430
+    },
+    {
+      "epoch": 0.7100027505271844,
+      "learning_rate": 5.800128357935272e-05,
+      "loss": 0.7187,
+      "mean_token_accuracy": 0.787151426076889,
+      "num_tokens": 182656070.0,
+      "step": 77440
+    },
+    {
+      "epoch": 0.7100944347666636,
+      "learning_rate": 5.798294673145687e-05,
+      "loss": 0.6444,
+      "mean_token_accuracy": 0.7948240578174591,
+      "num_tokens": 182681575.0,
+      "step": 77450
+    },
+    {
+      "epoch": 0.7101861190061428,
+      "learning_rate": 5.7964609883561015e-05,
+      "loss": 0.6702,
+      "mean_token_accuracy": 0.7921246230602265,
+      "num_tokens": 182707205.0,
+      "step": 77460
+    },
+    {
+      "epoch": 0.7102778032456221,
+      "learning_rate": 5.794627303566518e-05,
+      "loss": 0.6597,
+      "mean_token_accuracy": 0.7928208768367767,
+      "num_tokens": 182732143.0,
+      "step": 77470
+    },
+    {
+      "epoch": 0.7103694874851013,
+      "learning_rate": 5.7927936187769325e-05,
+      "loss": 0.6765,
+      "mean_token_accuracy": 0.7914460361003876,
+      "num_tokens": 182757778.0,
+      "step": 77480
+    },
+    {
+      "epoch": 0.7104611717245806,
+      "learning_rate": 5.790959933987348e-05,
+      "loss": 0.6648,
+      "mean_token_accuracy": 0.7961377739906311,
+      "num_tokens": 182782644.0,
+      "step": 77490
+    },
+    {
+      "epoch": 0.7105528559640598,
+      "learning_rate": 5.789126249197763e-05,
+      "loss": 0.6614,
+      "mean_token_accuracy": 0.7947397768497467,
+      "num_tokens": 182807805.0,
+      "step": 77500
+    },
+    {
+      "epoch": 0.7106445402035391,
+      "learning_rate": 5.787292564408179e-05,
+      "loss": 0.6889,
+      "mean_token_accuracy": 0.7911662220954895,
+      "num_tokens": 182833562.0,
+      "step": 77510
+    },
+    {
+      "epoch": 0.7107362244430182,
+      "learning_rate": 5.785458879618594e-05,
+      "loss": 0.6902,
+      "mean_token_accuracy": 0.7831215023994446,
+      "num_tokens": 182859458.0,
+      "step": 77520
+    },
+    {
+      "epoch": 0.7108279086824975,
+      "learning_rate": 5.7836251948290085e-05,
+      "loss": 0.6976,
+      "mean_token_accuracy": 0.7894893825054169,
+      "num_tokens": 182884385.0,
+      "step": 77530
+    },
+    {
+      "epoch": 0.7109195929219767,
+      "learning_rate": 5.781791510039425e-05,
+      "loss": 0.6462,
+      "mean_token_accuracy": 0.7979496419429779,
+      "num_tokens": 182909871.0,
+      "step": 77540
+    },
+    {
+      "epoch": 0.711011277161456,
+      "learning_rate": 5.77995782524984e-05,
+      "loss": 0.7004,
+      "mean_token_accuracy": 0.7869316816329956,
+      "num_tokens": 182935038.0,
+      "step": 77550
+    },
+    {
+      "epoch": 0.7111029614009352,
+      "learning_rate": 5.778124140460255e-05,
+      "loss": 0.6854,
+      "mean_token_accuracy": 0.7915327310562134,
+      "num_tokens": 182960594.0,
+      "step": 77560
+    },
+    {
+      "epoch": 0.7111946456404145,
+      "learning_rate": 5.776290455670671e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8074375748634338,
+      "num_tokens": 182985612.0,
+      "step": 77570
+    },
+    {
+      "epoch": 0.7112863298798936,
+      "learning_rate": 5.774456770881086e-05,
+      "loss": 0.6528,
+      "mean_token_accuracy": 0.7955219149589539,
+      "num_tokens": 183010303.0,
+      "step": 77580
+    },
+    {
+      "epoch": 0.7113780141193728,
+      "learning_rate": 5.772623086091501e-05,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.789950567483902,
+      "num_tokens": 183035037.0,
+      "step": 77590
+    },
+    {
+      "epoch": 0.7114696983588521,
+      "learning_rate": 5.770789401301917e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.7843886494636536,
+      "num_tokens": 183059573.0,
+      "step": 77600
+    },
+    {
+      "epoch": 0.7115613825983313,
+      "learning_rate": 5.768955716512332e-05,
+      "loss": 0.6424,
+      "mean_token_accuracy": 0.8002339243888855,
+      "num_tokens": 183084243.0,
+      "step": 77610
+    },
+    {
+      "epoch": 0.7116530668378106,
+      "learning_rate": 5.767122031722747e-05,
+      "loss": 0.712,
+      "mean_token_accuracy": 0.7875123262405396,
+      "num_tokens": 183110146.0,
+      "step": 77620
+    },
+    {
+      "epoch": 0.7117447510772899,
+      "learning_rate": 5.765288346933162e-05,
+      "loss": 0.6586,
+      "mean_token_accuracy": 0.7953888654708863,
+      "num_tokens": 183135299.0,
+      "step": 77630
+    },
+    {
+      "epoch": 0.7118364353167691,
+      "learning_rate": 5.763454662143578e-05,
+      "loss": 0.6636,
+      "mean_token_accuracy": 0.7930649042129516,
+      "num_tokens": 183160040.0,
+      "step": 77640
+    },
+    {
+      "epoch": 0.7119281195562482,
+      "learning_rate": 5.7616209773539934e-05,
+      "loss": 0.6884,
+      "mean_token_accuracy": 0.7891636252403259,
+      "num_tokens": 183184777.0,
+      "step": 77650
+    },
+    {
+      "epoch": 0.7120198037957275,
+      "learning_rate": 5.759787292564408e-05,
+      "loss": 0.6578,
+      "mean_token_accuracy": 0.7926559567451477,
+      "num_tokens": 183210093.0,
+      "step": 77660
+    },
+    {
+      "epoch": 0.7121114880352067,
+      "learning_rate": 5.7579536077748245e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.7948283314704895,
+      "num_tokens": 183235845.0,
+      "step": 77670
+    },
+    {
+      "epoch": 0.712203172274686,
+      "learning_rate": 5.756119922985239e-05,
+      "loss": 0.6649,
+      "mean_token_accuracy": 0.7924795091152191,
+      "num_tokens": 183261107.0,
+      "step": 77680
+    },
+    {
+      "epoch": 0.7122948565141652,
+      "learning_rate": 5.754286238195654e-05,
+      "loss": 0.6692,
+      "mean_token_accuracy": 0.7879079580307007,
+      "num_tokens": 183286263.0,
+      "step": 77690
+    },
+    {
+      "epoch": 0.7123865407536445,
+      "learning_rate": 5.75245255340607e-05,
+      "loss": 0.6687,
+      "mean_token_accuracy": 0.7926979243755341,
+      "num_tokens": 183311998.0,
+      "step": 77700
+    },
+    {
+      "epoch": 0.7124782249931236,
+      "learning_rate": 5.750618868616485e-05,
+      "loss": 0.7008,
+      "mean_token_accuracy": 0.7857622802257538,
+      "num_tokens": 183336922.0,
+      "step": 77710
+    },
+    {
+      "epoch": 0.7125699092326029,
+      "learning_rate": 5.7487851838269004e-05,
+      "loss": 0.7156,
+      "mean_token_accuracy": 0.7862915635108948,
+      "num_tokens": 183362223.0,
+      "step": 77720
+    },
+    {
+      "epoch": 0.7126615934720821,
+      "learning_rate": 5.746951499037315e-05,
+      "loss": 0.71,
+      "mean_token_accuracy": 0.7855953216552735,
+      "num_tokens": 183386996.0,
+      "step": 77730
+    },
+    {
+      "epoch": 0.7127532777115614,
+      "learning_rate": 5.7451178142477314e-05,
+      "loss": 0.6739,
+      "mean_token_accuracy": 0.7907623469829559,
+      "num_tokens": 183411714.0,
+      "step": 77740
+    },
+    {
+      "epoch": 0.7128449619510406,
+      "learning_rate": 5.7432841294581466e-05,
+      "loss": 0.6828,
+      "mean_token_accuracy": 0.786265766620636,
+      "num_tokens": 183436531.0,
+      "step": 77750
+    },
+    {
+      "epoch": 0.7129366461905199,
+      "learning_rate": 5.741450444668561e-05,
+      "loss": 0.6642,
+      "mean_token_accuracy": 0.7964664816856384,
+      "num_tokens": 183461573.0,
+      "step": 77760
+    },
+    {
+      "epoch": 0.7130283304299991,
+      "learning_rate": 5.739616759878978e-05,
+      "loss": 0.6899,
+      "mean_token_accuracy": 0.7901533544063568,
+      "num_tokens": 183487012.0,
+      "step": 77770
+    },
+    {
+      "epoch": 0.7131200146694783,
+      "learning_rate": 5.737783075089392e-05,
+      "loss": 0.6765,
+      "mean_token_accuracy": 0.7908352613449097,
+      "num_tokens": 183512918.0,
+      "step": 77780
+    },
+    {
+      "epoch": 0.7132116989089575,
+      "learning_rate": 5.7359493902998074e-05,
+      "loss": 0.6814,
+      "mean_token_accuracy": 0.7902816355228424,
+      "num_tokens": 183538668.0,
+      "step": 77790
+    },
+    {
+      "epoch": 0.7133033831484368,
+      "learning_rate": 5.734115705510223e-05,
+      "loss": 0.6658,
+      "mean_token_accuracy": 0.7939606606960297,
+      "num_tokens": 183563237.0,
+      "step": 77800
+    },
+    {
+      "epoch": 0.713395067387916,
+      "learning_rate": 5.7322820207206384e-05,
+      "loss": 0.6865,
+      "mean_token_accuracy": 0.7885273933410645,
+      "num_tokens": 183588836.0,
+      "step": 77810
+    },
+    {
+      "epoch": 0.7134867516273953,
+      "learning_rate": 5.7304483359310536e-05,
+      "loss": 0.7152,
+      "mean_token_accuracy": 0.7769215404987335,
+      "num_tokens": 183613841.0,
+      "step": 77820
+    },
+    {
+      "epoch": 0.7135784358668745,
+      "learning_rate": 5.7286146511414695e-05,
+      "loss": 0.6771,
+      "mean_token_accuracy": 0.7876941502094269,
+      "num_tokens": 183638946.0,
+      "step": 77830
+    },
+    {
+      "epoch": 0.7136701201063537,
+      "learning_rate": 5.726780966351885e-05,
+      "loss": 0.6726,
+      "mean_token_accuracy": 0.7927168607711792,
+      "num_tokens": 183664240.0,
+      "step": 77840
+    },
+    {
+      "epoch": 0.7137618043458329,
+      "learning_rate": 5.7249472815623e-05,
+      "loss": 0.6796,
+      "mean_token_accuracy": 0.7934220433235168,
+      "num_tokens": 183689042.0,
+      "step": 77850
+    },
+    {
+      "epoch": 0.7138534885853122,
+      "learning_rate": 5.7231135967727144e-05,
+      "loss": 0.6972,
+      "mean_token_accuracy": 0.7839462220668793,
+      "num_tokens": 183713691.0,
+      "step": 77860
+    },
+    {
+      "epoch": 0.7139451728247914,
+      "learning_rate": 5.721279911983131e-05,
+      "loss": 0.6963,
+      "mean_token_accuracy": 0.7842570841312408,
+      "num_tokens": 183739332.0,
+      "step": 77870
+    },
+    {
+      "epoch": 0.7140368570642707,
+      "learning_rate": 5.7194462271935454e-05,
+      "loss": 0.6864,
+      "mean_token_accuracy": 0.7891451597213746,
+      "num_tokens": 183763411.0,
+      "step": 77880
+    },
+    {
+      "epoch": 0.7141285413037499,
+      "learning_rate": 5.7176125424039606e-05,
+      "loss": 0.6821,
+      "mean_token_accuracy": 0.7885421097278595,
+      "num_tokens": 183789127.0,
+      "step": 77890
+    },
+    {
+      "epoch": 0.7142202255432292,
+      "learning_rate": 5.7157788576143765e-05,
+      "loss": 0.6711,
+      "mean_token_accuracy": 0.7992440283298492,
+      "num_tokens": 183813609.0,
+      "step": 77900
+    },
+    {
+      "epoch": 0.7143119097827083,
+      "learning_rate": 5.7139451728247916e-05,
+      "loss": 0.6896,
+      "mean_token_accuracy": 0.792653614282608,
+      "num_tokens": 183839326.0,
+      "step": 77910
+    },
+    {
+      "epoch": 0.7144035940221876,
+      "learning_rate": 5.712111488035207e-05,
+      "loss": 0.6584,
+      "mean_token_accuracy": 0.7985167622566223,
+      "num_tokens": 183864306.0,
+      "step": 77920
+    },
+    {
+      "epoch": 0.7144952782616668,
+      "learning_rate": 5.710277803245623e-05,
+      "loss": 0.6647,
+      "mean_token_accuracy": 0.7934771120548249,
+      "num_tokens": 183888814.0,
+      "step": 77930
+    },
+    {
+      "epoch": 0.7145869625011461,
+      "learning_rate": 5.708444118456038e-05,
+      "loss": 0.6897,
+      "mean_token_accuracy": 0.7820202767848968,
+      "num_tokens": 183913876.0,
+      "step": 77940
+    },
+    {
+      "epoch": 0.7146786467406253,
+      "learning_rate": 5.706610433666453e-05,
+      "loss": 0.6741,
+      "mean_token_accuracy": 0.7928056299686432,
+      "num_tokens": 183940439.0,
+      "step": 77950
+    },
+    {
+      "epoch": 0.7147703309801046,
+      "learning_rate": 5.704776748876869e-05,
+      "loss": 0.6822,
+      "mean_token_accuracy": 0.7908152997493744,
+      "num_tokens": 183965069.0,
+      "step": 77960
+    },
+    {
+      "epoch": 0.7148620152195837,
+      "learning_rate": 5.702943064087284e-05,
+      "loss": 0.6795,
+      "mean_token_accuracy": 0.788107693195343,
+      "num_tokens": 183990839.0,
+      "step": 77970
+    },
+    {
+      "epoch": 0.714953699459063,
+      "learning_rate": 5.7011093792976986e-05,
+      "loss": 0.6618,
+      "mean_token_accuracy": 0.7950737059116364,
+      "num_tokens": 184016216.0,
+      "step": 77980
+    },
+    {
+      "epoch": 0.7150453836985422,
+      "learning_rate": 5.699275694508114e-05,
+      "loss": 0.6588,
+      "mean_token_accuracy": 0.7969727396965027,
+      "num_tokens": 184041948.0,
+      "step": 77990
+    },
+    {
+      "epoch": 0.7151370679380215,
+      "learning_rate": 5.69744200971853e-05,
+      "loss": 0.6799,
+      "mean_token_accuracy": 0.7882644653320312,
+      "num_tokens": 184067725.0,
+      "step": 78000
+    },
+    {
+      "epoch": 0.7152287521775007,
+      "learning_rate": 5.695608324928945e-05,
+      "loss": 0.6748,
+      "mean_token_accuracy": 0.7901932060718536,
+      "num_tokens": 184093110.0,
+      "step": 78010
+    },
+    {
+      "epoch": 0.71532043641698,
+      "learning_rate": 5.69377464013936e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.7816961944103241,
+      "num_tokens": 184118664.0,
+      "step": 78020
+    },
+    {
+      "epoch": 0.7154121206564592,
+      "learning_rate": 5.691940955349776e-05,
+      "loss": 0.6699,
+      "mean_token_accuracy": 0.7908800780773163,
+      "num_tokens": 184144459.0,
+      "step": 78030
+    },
+    {
+      "epoch": 0.7155038048959383,
+      "learning_rate": 5.690107270560191e-05,
+      "loss": 0.6561,
+      "mean_token_accuracy": 0.7967715203762055,
+      "num_tokens": 184170559.0,
+      "step": 78040
+    },
+    {
+      "epoch": 0.7155954891354176,
+      "learning_rate": 5.688273585770606e-05,
+      "loss": 0.6547,
+      "mean_token_accuracy": 0.7976432323455811,
+      "num_tokens": 184195970.0,
+      "step": 78050
+    },
+    {
+      "epoch": 0.7156871733748968,
+      "learning_rate": 5.686439900981022e-05,
+      "loss": 0.6971,
+      "mean_token_accuracy": 0.7896312355995179,
+      "num_tokens": 184221762.0,
+      "step": 78060
+    },
+    {
+      "epoch": 0.7157788576143761,
+      "learning_rate": 5.6846062161914373e-05,
+      "loss": 0.6806,
+      "mean_token_accuracy": 0.7902872085571289,
+      "num_tokens": 184246845.0,
+      "step": 78070
+    },
+    {
+      "epoch": 0.7158705418538553,
+      "learning_rate": 5.682772531401852e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.7874055624008178,
+      "num_tokens": 184271935.0,
+      "step": 78080
+    },
+    {
+      "epoch": 0.7159622260933346,
+      "learning_rate": 5.6809388466122684e-05,
+      "loss": 0.6831,
+      "mean_token_accuracy": 0.7964799761772156,
+      "num_tokens": 184296852.0,
+      "step": 78090
+    },
+    {
+      "epoch": 0.7160539103328137,
+      "learning_rate": 5.679105161822683e-05,
+      "loss": 0.6495,
+      "mean_token_accuracy": 0.7958796322345734,
+      "num_tokens": 184322161.0,
+      "step": 78100
+    },
+    {
+      "epoch": 0.716145594572293,
+      "learning_rate": 5.677271477033098e-05,
+      "loss": 0.6727,
+      "mean_token_accuracy": 0.7924707174301148,
+      "num_tokens": 184346686.0,
+      "step": 78110
+    },
+    {
+      "epoch": 0.7162372788117722,
+      "learning_rate": 5.675437792243513e-05,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.7871766507625579,
+      "num_tokens": 184371236.0,
+      "step": 78120
+    },
+    {
+      "epoch": 0.7163289630512515,
+      "learning_rate": 5.673604107453929e-05,
+      "loss": 0.678,
+      "mean_token_accuracy": 0.7908380270004273,
+      "num_tokens": 184396549.0,
+      "step": 78130
+    },
+    {
+      "epoch": 0.7164206472907307,
+      "learning_rate": 5.671770422664344e-05,
+      "loss": 0.739,
+      "mean_token_accuracy": 0.7799354910850524,
+      "num_tokens": 184422153.0,
+      "step": 78140
+    },
+    {
+      "epoch": 0.71651233153021,
+      "learning_rate": 5.6699367378747595e-05,
+      "loss": 0.6654,
+      "mean_token_accuracy": 0.7889452934265136,
+      "num_tokens": 184447079.0,
+      "step": 78150
+    },
+    {
+      "epoch": 0.7166040157696892,
+      "learning_rate": 5.6681030530851754e-05,
+      "loss": 0.6541,
+      "mean_token_accuracy": 0.8021519839763641,
+      "num_tokens": 184471782.0,
+      "step": 78160
+    },
+    {
+      "epoch": 0.7166957000091684,
+      "learning_rate": 5.6662693682955906e-05,
+      "loss": 0.6971,
+      "mean_token_accuracy": 0.790956997871399,
+      "num_tokens": 184496885.0,
+      "step": 78170
+    },
+    {
+      "epoch": 0.7167873842486476,
+      "learning_rate": 5.664435683506005e-05,
+      "loss": 0.7147,
+      "mean_token_accuracy": 0.7888283967971802,
+      "num_tokens": 184521899.0,
+      "step": 78180
+    },
+    {
+      "epoch": 0.7168790684881269,
+      "learning_rate": 5.6626019987164216e-05,
+      "loss": 0.6686,
+      "mean_token_accuracy": 0.7869354009628295,
+      "num_tokens": 184546847.0,
+      "step": 78190
+    },
+    {
+      "epoch": 0.7169707527276061,
+      "learning_rate": 5.660768313926836e-05,
+      "loss": 0.6797,
+      "mean_token_accuracy": 0.7971076250076294,
+      "num_tokens": 184571856.0,
+      "step": 78200
+    },
+    {
+      "epoch": 0.7170624369670854,
+      "learning_rate": 5.658934629137251e-05,
+      "loss": 0.654,
+      "mean_token_accuracy": 0.7961293399333954,
+      "num_tokens": 184596631.0,
+      "step": 78210
+    },
+    {
+      "epoch": 0.7171541212065646,
+      "learning_rate": 5.6571009443476665e-05,
+      "loss": 0.6728,
+      "mean_token_accuracy": 0.7975492119789124,
+      "num_tokens": 184621829.0,
+      "step": 78220
+    },
+    {
+      "epoch": 0.7172458054460438,
+      "learning_rate": 5.6552672595580824e-05,
+      "loss": 0.6894,
+      "mean_token_accuracy": 0.7944309294223786,
+      "num_tokens": 184646998.0,
+      "step": 78230
+    },
+    {
+      "epoch": 0.717337489685523,
+      "learning_rate": 5.6534335747684976e-05,
+      "loss": 0.6811,
+      "mean_token_accuracy": 0.794684249162674,
+      "num_tokens": 184671729.0,
+      "step": 78240
+    },
+    {
+      "epoch": 0.7174291739250023,
+      "learning_rate": 5.651599889978913e-05,
+      "loss": 0.6638,
+      "mean_token_accuracy": 0.7963333606719971,
+      "num_tokens": 184696035.0,
+      "step": 78250
+    },
+    {
+      "epoch": 0.7175208581644815,
+      "learning_rate": 5.6497662051893286e-05,
+      "loss": 0.6787,
+      "mean_token_accuracy": 0.7910999178886413,
+      "num_tokens": 184721316.0,
+      "step": 78260
+    },
+    {
+      "epoch": 0.7176125424039608,
+      "learning_rate": 5.647932520399744e-05,
+      "loss": 0.6586,
+      "mean_token_accuracy": 0.797629326581955,
+      "num_tokens": 184745371.0,
+      "step": 78270
+    },
+    {
+      "epoch": 0.71770422664344,
+      "learning_rate": 5.646098835610158e-05,
+      "loss": 0.6686,
+      "mean_token_accuracy": 0.7965006291866302,
+      "num_tokens": 184770817.0,
+      "step": 78280
+    },
+    {
+      "epoch": 0.7177959108829193,
+      "learning_rate": 5.644265150820575e-05,
+      "loss": 0.6872,
+      "mean_token_accuracy": 0.784556770324707,
+      "num_tokens": 184796011.0,
+      "step": 78290
+    },
+    {
+      "epoch": 0.7178875951223984,
+      "learning_rate": 5.6424314660309894e-05,
+      "loss": 0.6748,
+      "mean_token_accuracy": 0.7958771646022796,
+      "num_tokens": 184821089.0,
+      "step": 78300
+    },
+    {
+      "epoch": 0.7179792793618777,
+      "learning_rate": 5.6405977812414045e-05,
+      "loss": 0.7017,
+      "mean_token_accuracy": 0.7866156160831451,
+      "num_tokens": 184845931.0,
+      "step": 78310
+    },
+    {
+      "epoch": 0.7180709636013569,
+      "learning_rate": 5.6387640964518204e-05,
+      "loss": 0.6559,
+      "mean_token_accuracy": 0.7917865931987762,
+      "num_tokens": 184871315.0,
+      "step": 78320
+    },
+    {
+      "epoch": 0.7181626478408362,
+      "learning_rate": 5.6369304116622356e-05,
+      "loss": 0.6518,
+      "mean_token_accuracy": 0.7963665962219239,
+      "num_tokens": 184896598.0,
+      "step": 78330
+    },
+    {
+      "epoch": 0.7182543320803154,
+      "learning_rate": 5.635096726872651e-05,
+      "loss": 0.6603,
+      "mean_token_accuracy": 0.7953488826751709,
+      "num_tokens": 184922270.0,
+      "step": 78340
+    },
+    {
+      "epoch": 0.7183460163197947,
+      "learning_rate": 5.633263042083066e-05,
+      "loss": 0.684,
+      "mean_token_accuracy": 0.7936093211174011,
+      "num_tokens": 184947628.0,
+      "step": 78350
+    },
+    {
+      "epoch": 0.7184377005592738,
+      "learning_rate": 5.631429357293482e-05,
+      "loss": 0.691,
+      "mean_token_accuracy": 0.7842152416706085,
+      "num_tokens": 184972995.0,
+      "step": 78360
+    },
+    {
+      "epoch": 0.7185293847987531,
+      "learning_rate": 5.629595672503897e-05,
+      "loss": 0.6555,
+      "mean_token_accuracy": 0.7920524537563324,
+      "num_tokens": 184998223.0,
+      "step": 78370
+    },
+    {
+      "epoch": 0.7186210690382323,
+      "learning_rate": 5.6277619877143115e-05,
+      "loss": 0.6673,
+      "mean_token_accuracy": 0.7891258299350739,
+      "num_tokens": 185023009.0,
+      "step": 78380
+    },
+    {
+      "epoch": 0.7187127532777116,
+      "learning_rate": 5.625928302924728e-05,
+      "loss": 0.7255,
+      "mean_token_accuracy": 0.7829273462295532,
+      "num_tokens": 185048043.0,
+      "step": 78390
+    },
+    {
+      "epoch": 0.7188044375171908,
+      "learning_rate": 5.6240946181351426e-05,
+      "loss": 0.7108,
+      "mean_token_accuracy": 0.7833603918552399,
+      "num_tokens": 185072602.0,
+      "step": 78400
+    },
+    {
+      "epoch": 0.7188961217566701,
+      "learning_rate": 5.622260933345558e-05,
+      "loss": 0.6909,
+      "mean_token_accuracy": 0.7880420744419098,
+      "num_tokens": 185097881.0,
+      "step": 78410
+    },
+    {
+      "epoch": 0.7189878059961493,
+      "learning_rate": 5.6204272485559736e-05,
+      "loss": 0.6747,
+      "mean_token_accuracy": 0.7878067791461945,
+      "num_tokens": 185122978.0,
+      "step": 78420
+    },
+    {
+      "epoch": 0.7190794902356284,
+      "learning_rate": 5.618593563766389e-05,
+      "loss": 0.6386,
+      "mean_token_accuracy": 0.8035356938838959,
+      "num_tokens": 185148717.0,
+      "step": 78430
+    },
+    {
+      "epoch": 0.7191711744751077,
+      "learning_rate": 5.616759878976804e-05,
+      "loss": 0.6618,
+      "mean_token_accuracy": 0.7942996501922608,
+      "num_tokens": 185174113.0,
+      "step": 78440
+    },
+    {
+      "epoch": 0.719262858714587,
+      "learning_rate": 5.61492619418722e-05,
+      "loss": 0.6626,
+      "mean_token_accuracy": 0.7976067721843719,
+      "num_tokens": 185199948.0,
+      "step": 78450
+    },
+    {
+      "epoch": 0.7193545429540662,
+      "learning_rate": 5.613092509397635e-05,
+      "loss": 0.6354,
+      "mean_token_accuracy": 0.8027196645736694,
+      "num_tokens": 185225238.0,
+      "step": 78460
+    },
+    {
+      "epoch": 0.7194462271935455,
+      "learning_rate": 5.61125882460805e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.7983404517173767,
+      "num_tokens": 185249699.0,
+      "step": 78470
+    },
+    {
+      "epoch": 0.7195379114330247,
+      "learning_rate": 5.609425139818465e-05,
+      "loss": 0.6722,
+      "mean_token_accuracy": 0.7945112943649292,
+      "num_tokens": 185274338.0,
+      "step": 78480
+    },
+    {
+      "epoch": 0.7196295956725038,
+      "learning_rate": 5.607591455028881e-05,
+      "loss": 0.6664,
+      "mean_token_accuracy": 0.7904851019382477,
+      "num_tokens": 185300034.0,
+      "step": 78490
+    },
+    {
+      "epoch": 0.7197212799119831,
+      "learning_rate": 5.605757770239296e-05,
+      "loss": 0.6707,
+      "mean_token_accuracy": 0.7888002574443818,
+      "num_tokens": 185324876.0,
+      "step": 78500
+    },
+    {
+      "epoch": 0.7198129641514623,
+      "learning_rate": 5.603924085449711e-05,
+      "loss": 0.6407,
+      "mean_token_accuracy": 0.8010428786277771,
+      "num_tokens": 185349660.0,
+      "step": 78510
+    },
+    {
+      "epoch": 0.7199046483909416,
+      "learning_rate": 5.602090400660127e-05,
+      "loss": 0.7143,
+      "mean_token_accuracy": 0.786019253730774,
+      "num_tokens": 185374978.0,
+      "step": 78520
+    },
+    {
+      "epoch": 0.7199963326304208,
+      "learning_rate": 5.600256715870542e-05,
+      "loss": 0.6712,
+      "mean_token_accuracy": 0.7957077205181122,
+      "num_tokens": 185398963.0,
+      "step": 78530
+    },
+    {
+      "epoch": 0.7200880168699001,
+      "learning_rate": 5.598423031080957e-05,
+      "loss": 0.6747,
+      "mean_token_accuracy": 0.7959294199943543,
+      "num_tokens": 185424009.0,
+      "step": 78540
+    },
+    {
+      "epoch": 0.7201797011093793,
+      "learning_rate": 5.596589346291373e-05,
+      "loss": 0.6583,
+      "mean_token_accuracy": 0.7982489109039307,
+      "num_tokens": 185448764.0,
+      "step": 78550
+    },
+    {
+      "epoch": 0.7202713853488585,
+      "learning_rate": 5.594755661501788e-05,
+      "loss": 0.6693,
+      "mean_token_accuracy": 0.7920622050762176,
+      "num_tokens": 185473983.0,
+      "step": 78560
+    },
+    {
+      "epoch": 0.7203630695883377,
+      "learning_rate": 5.5929219767122035e-05,
+      "loss": 0.6775,
+      "mean_token_accuracy": 0.7899169385433197,
+      "num_tokens": 185499079.0,
+      "step": 78570
+    },
+    {
+      "epoch": 0.720454753827817,
+      "learning_rate": 5.591088291922619e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.7954796433448792,
+      "num_tokens": 185524004.0,
+      "step": 78580
+    },
+    {
+      "epoch": 0.7205464380672962,
+      "learning_rate": 5.5892546071330345e-05,
+      "loss": 0.6807,
+      "mean_token_accuracy": 0.7902134656906128,
+      "num_tokens": 185549231.0,
+      "step": 78590
+    },
+    {
+      "epoch": 0.7206381223067755,
+      "learning_rate": 5.587420922343449e-05,
+      "loss": 0.6637,
+      "mean_token_accuracy": 0.7953526675701141,
+      "num_tokens": 185574857.0,
+      "step": 78600
+    },
+    {
+      "epoch": 0.7207298065462547,
+      "learning_rate": 5.585587237553864e-05,
+      "loss": 0.6884,
+      "mean_token_accuracy": 0.7841347217559814,
+      "num_tokens": 185600589.0,
+      "step": 78610
+    },
+    {
+      "epoch": 0.720821490785734,
+      "learning_rate": 5.58375355276428e-05,
+      "loss": 0.6725,
+      "mean_token_accuracy": 0.7924810409545898,
+      "num_tokens": 185626131.0,
+      "step": 78620
+    },
+    {
+      "epoch": 0.7209131750252131,
+      "learning_rate": 5.581919867974695e-05,
+      "loss": 0.739,
+      "mean_token_accuracy": 0.7724352061748505,
+      "num_tokens": 185650834.0,
+      "step": 78630
+    },
+    {
+      "epoch": 0.7210048592646924,
+      "learning_rate": 5.5800861831851104e-05,
+      "loss": 0.7,
+      "mean_token_accuracy": 0.7873457729816437,
+      "num_tokens": 185675546.0,
+      "step": 78640
+    },
+    {
+      "epoch": 0.7210965435041716,
+      "learning_rate": 5.578252498395526e-05,
+      "loss": 0.7074,
+      "mean_token_accuracy": 0.7801327109336853,
+      "num_tokens": 185700499.0,
+      "step": 78650
+    },
+    {
+      "epoch": 0.7211882277436509,
+      "learning_rate": 5.5764188136059415e-05,
+      "loss": 0.6962,
+      "mean_token_accuracy": 0.7899401545524597,
+      "num_tokens": 185725749.0,
+      "step": 78660
+    },
+    {
+      "epoch": 0.7212799119831301,
+      "learning_rate": 5.574585128816357e-05,
+      "loss": 0.6895,
+      "mean_token_accuracy": 0.7878952562808991,
+      "num_tokens": 185751311.0,
+      "step": 78670
+    },
+    {
+      "epoch": 0.7213715962226094,
+      "learning_rate": 5.5727514440267725e-05,
+      "loss": 0.6718,
+      "mean_token_accuracy": 0.7948323488235474,
+      "num_tokens": 185776398.0,
+      "step": 78680
+    },
+    {
+      "epoch": 0.7214632804620885,
+      "learning_rate": 5.570917759237188e-05,
+      "loss": 0.6829,
+      "mean_token_accuracy": 0.788807487487793,
+      "num_tokens": 185801653.0,
+      "step": 78690
+    },
+    {
+      "epoch": 0.7215549647015678,
+      "learning_rate": 5.569084074447602e-05,
+      "loss": 0.6704,
+      "mean_token_accuracy": 0.7911393821239472,
+      "num_tokens": 185826353.0,
+      "step": 78700
+    },
+    {
+      "epoch": 0.721646648941047,
+      "learning_rate": 5.567250389658019e-05,
+      "loss": 0.6506,
+      "mean_token_accuracy": 0.7973729252815247,
+      "num_tokens": 185851793.0,
+      "step": 78710
+    },
+    {
+      "epoch": 0.7217383331805263,
+      "learning_rate": 5.565416704868433e-05,
+      "loss": 0.6958,
+      "mean_token_accuracy": 0.7847602307796478,
+      "num_tokens": 185876706.0,
+      "step": 78720
+    },
+    {
+      "epoch": 0.7218300174200055,
+      "learning_rate": 5.5635830200788485e-05,
+      "loss": 0.679,
+      "mean_token_accuracy": 0.7909967482089997,
+      "num_tokens": 185901364.0,
+      "step": 78730
+    },
+    {
+      "epoch": 0.7219217016594848,
+      "learning_rate": 5.561749335289264e-05,
+      "loss": 0.6357,
+      "mean_token_accuracy": 0.8032341659069061,
+      "num_tokens": 185926419.0,
+      "step": 78740
+    },
+    {
+      "epoch": 0.722013385898964,
+      "learning_rate": 5.5599156504996795e-05,
+      "loss": 0.6583,
+      "mean_token_accuracy": 0.7890056610107422,
+      "num_tokens": 185952069.0,
+      "step": 78750
+    },
+    {
+      "epoch": 0.7221050701384432,
+      "learning_rate": 5.558081965710095e-05,
+      "loss": 0.6903,
+      "mean_token_accuracy": 0.7933500289916993,
+      "num_tokens": 185976944.0,
+      "step": 78760
+    },
+    {
+      "epoch": 0.7221967543779224,
+      "learning_rate": 5.55624828092051e-05,
+      "loss": 0.6541,
+      "mean_token_accuracy": 0.7989000022411347,
+      "num_tokens": 186002291.0,
+      "step": 78770
+    },
+    {
+      "epoch": 0.7222884386174017,
+      "learning_rate": 5.554414596130926e-05,
+      "loss": 0.6555,
+      "mean_token_accuracy": 0.793702495098114,
+      "num_tokens": 186026919.0,
+      "step": 78780
+    },
+    {
+      "epoch": 0.7223801228568809,
+      "learning_rate": 5.552580911341341e-05,
+      "loss": 0.6872,
+      "mean_token_accuracy": 0.789022421836853,
+      "num_tokens": 186051340.0,
+      "step": 78790
+    },
+    {
+      "epoch": 0.7224718070963602,
+      "learning_rate": 5.5507472265517555e-05,
+      "loss": 0.6527,
+      "mean_token_accuracy": 0.7955657958984375,
+      "num_tokens": 186076872.0,
+      "step": 78800
+    },
+    {
+      "epoch": 0.7225634913358394,
+      "learning_rate": 5.548913541762172e-05,
+      "loss": 0.6916,
+      "mean_token_accuracy": 0.7900915682315827,
+      "num_tokens": 186101860.0,
+      "step": 78810
+    },
+    {
+      "epoch": 0.7226551755753186,
+      "learning_rate": 5.5470798569725865e-05,
+      "loss": 0.6668,
+      "mean_token_accuracy": 0.7958865940570832,
+      "num_tokens": 186127330.0,
+      "step": 78820
+    },
+    {
+      "epoch": 0.7227468598147978,
+      "learning_rate": 5.545246172183002e-05,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.7916256606578826,
+      "num_tokens": 186152052.0,
+      "step": 78830
+    },
+    {
+      "epoch": 0.722838544054277,
+      "learning_rate": 5.543412487393417e-05,
+      "loss": 0.7049,
+      "mean_token_accuracy": 0.7891476035118103,
+      "num_tokens": 186177014.0,
+      "step": 78840
+    },
+    {
+      "epoch": 0.7229302282937563,
+      "learning_rate": 5.541578802603833e-05,
+      "loss": 0.6732,
+      "mean_token_accuracy": 0.7945551216602326,
+      "num_tokens": 186201813.0,
+      "step": 78850
+    },
+    {
+      "epoch": 0.7230219125332356,
+      "learning_rate": 5.539745117814248e-05,
+      "loss": 0.7291,
+      "mean_token_accuracy": 0.7831566095352173,
+      "num_tokens": 186226559.0,
+      "step": 78860
+    },
+    {
+      "epoch": 0.7231135967727148,
+      "learning_rate": 5.537911433024663e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.7860622882843018,
+      "num_tokens": 186252464.0,
+      "step": 78870
+    },
+    {
+      "epoch": 0.7232052810121941,
+      "learning_rate": 5.536077748235079e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.791435694694519,
+      "num_tokens": 186277649.0,
+      "step": 78880
+    },
+    {
+      "epoch": 0.7232969652516732,
+      "learning_rate": 5.534244063445494e-05,
+      "loss": 0.6624,
+      "mean_token_accuracy": 0.7957864880561829,
+      "num_tokens": 186303861.0,
+      "step": 78890
+    },
+    {
+      "epoch": 0.7233886494911524,
+      "learning_rate": 5.532410378655909e-05,
+      "loss": 0.6504,
+      "mean_token_accuracy": 0.7934846758842469,
+      "num_tokens": 186329435.0,
+      "step": 78900
+    },
+    {
+      "epoch": 0.7234803337306317,
+      "learning_rate": 5.530576693866325e-05,
+      "loss": 0.6774,
+      "mean_token_accuracy": 0.7886178255081177,
+      "num_tokens": 186355071.0,
+      "step": 78910
+    },
+    {
+      "epoch": 0.723572017970111,
+      "learning_rate": 5.52874300907674e-05,
+      "loss": 0.7046,
+      "mean_token_accuracy": 0.7809918045997619,
+      "num_tokens": 186380489.0,
+      "step": 78920
+    },
+    {
+      "epoch": 0.7236637022095902,
+      "learning_rate": 5.526909324287155e-05,
+      "loss": 0.6755,
+      "mean_token_accuracy": 0.7858296811580658,
+      "num_tokens": 186405377.0,
+      "step": 78930
+    },
+    {
+      "epoch": 0.7237553864490694,
+      "learning_rate": 5.525075639497571e-05,
+      "loss": 0.6941,
+      "mean_token_accuracy": 0.7824526309967041,
+      "num_tokens": 186430471.0,
+      "step": 78940
+    },
+    {
+      "epoch": 0.7238470706885486,
+      "learning_rate": 5.523241954707986e-05,
+      "loss": 0.667,
+      "mean_token_accuracy": 0.7918890476226806,
+      "num_tokens": 186456585.0,
+      "step": 78950
+    },
+    {
+      "epoch": 0.7239387549280278,
+      "learning_rate": 5.521408269918401e-05,
+      "loss": 0.6809,
+      "mean_token_accuracy": 0.7902523100376129,
+      "num_tokens": 186481110.0,
+      "step": 78960
+    },
+    {
+      "epoch": 0.7240304391675071,
+      "learning_rate": 5.5195745851288163e-05,
+      "loss": 0.6793,
+      "mean_token_accuracy": 0.7880862593650818,
+      "num_tokens": 186505364.0,
+      "step": 78970
+    },
+    {
+      "epoch": 0.7241221234069863,
+      "learning_rate": 5.517740900339232e-05,
+      "loss": 0.6652,
+      "mean_token_accuracy": 0.7870647728443145,
+      "num_tokens": 186530464.0,
+      "step": 78980
+    },
+    {
+      "epoch": 0.7242138076464656,
+      "learning_rate": 5.5159072155496474e-05,
+      "loss": 0.6807,
+      "mean_token_accuracy": 0.788888442516327,
+      "num_tokens": 186555351.0,
+      "step": 78990
+    },
+    {
+      "epoch": 0.7243054918859448,
+      "learning_rate": 5.514073530760062e-05,
+      "loss": 0.6564,
+      "mean_token_accuracy": 0.7987925112247467,
+      "num_tokens": 186580817.0,
+      "step": 79000
+    },
+    {
+      "epoch": 0.7243971761254241,
+      "learning_rate": 5.5122398459704785e-05,
+      "loss": 0.6549,
+      "mean_token_accuracy": 0.7931228756904602,
+      "num_tokens": 186605941.0,
+      "step": 79010
+    },
+    {
+      "epoch": 0.7244888603649032,
+      "learning_rate": 5.510406161180893e-05,
+      "loss": 0.6836,
+      "mean_token_accuracy": 0.7867121160030365,
+      "num_tokens": 186630863.0,
+      "step": 79020
+    },
+    {
+      "epoch": 0.7245805446043825,
+      "learning_rate": 5.508572476391308e-05,
+      "loss": 0.7233,
+      "mean_token_accuracy": 0.7841231107711792,
+      "num_tokens": 186656597.0,
+      "step": 79030
+    },
+    {
+      "epoch": 0.7246722288438617,
+      "learning_rate": 5.506738791601724e-05,
+      "loss": 0.6919,
+      "mean_token_accuracy": 0.7882936596870422,
+      "num_tokens": 186682475.0,
+      "step": 79040
+    },
+    {
+      "epoch": 0.724763913083341,
+      "learning_rate": 5.504905106812139e-05,
+      "loss": 0.6513,
+      "mean_token_accuracy": 0.7978208243846894,
+      "num_tokens": 186707856.0,
+      "step": 79050
+    },
+    {
+      "epoch": 0.7248555973228202,
+      "learning_rate": 5.5030714220225544e-05,
+      "loss": 0.6796,
+      "mean_token_accuracy": 0.7892354786396026,
+      "num_tokens": 186732330.0,
+      "step": 79060
+    },
+    {
+      "epoch": 0.7249472815622995,
+      "learning_rate": 5.50123773723297e-05,
+      "loss": 0.683,
+      "mean_token_accuracy": 0.7898118734359741,
+      "num_tokens": 186757324.0,
+      "step": 79070
+    },
+    {
+      "epoch": 0.7250389658017786,
+      "learning_rate": 5.4994040524433854e-05,
+      "loss": 0.6786,
+      "mean_token_accuracy": 0.7866646826267243,
+      "num_tokens": 186782426.0,
+      "step": 79080
+    },
+    {
+      "epoch": 0.7251306500412579,
+      "learning_rate": 5.4975703676538006e-05,
+      "loss": 0.6582,
+      "mean_token_accuracy": 0.7960130989551544,
+      "num_tokens": 186808127.0,
+      "step": 79090
+    },
+    {
+      "epoch": 0.7252223342807371,
+      "learning_rate": 5.495736682864215e-05,
+      "loss": 0.7164,
+      "mean_token_accuracy": 0.7812751412391663,
+      "num_tokens": 186833079.0,
+      "step": 79100
+    },
+    {
+      "epoch": 0.7253140185202164,
+      "learning_rate": 5.493902998074632e-05,
+      "loss": 0.6752,
+      "mean_token_accuracy": 0.7962970733642578,
+      "num_tokens": 186858565.0,
+      "step": 79110
+    },
+    {
+      "epoch": 0.7254057027596956,
+      "learning_rate": 5.492069313285046e-05,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.7864541113376617,
+      "num_tokens": 186883852.0,
+      "step": 79120
+    },
+    {
+      "epoch": 0.7254973869991749,
+      "learning_rate": 5.4902356284954614e-05,
+      "loss": 0.6894,
+      "mean_token_accuracy": 0.7863449811935425,
+      "num_tokens": 186909679.0,
+      "step": 79130
+    },
+    {
+      "epoch": 0.7255890712386541,
+      "learning_rate": 5.488401943705877e-05,
+      "loss": 0.6794,
+      "mean_token_accuracy": 0.7928593873977661,
+      "num_tokens": 186934040.0,
+      "step": 79140
+    },
+    {
+      "epoch": 0.7256807554781333,
+      "learning_rate": 5.4865682589162924e-05,
+      "loss": 0.6579,
+      "mean_token_accuracy": 0.793781453371048,
+      "num_tokens": 186958758.0,
+      "step": 79150
+    },
+    {
+      "epoch": 0.7257724397176125,
+      "learning_rate": 5.4847345741267076e-05,
+      "loss": 0.6897,
+      "mean_token_accuracy": 0.7864092826843262,
+      "num_tokens": 186983248.0,
+      "step": 79160
+    },
+    {
+      "epoch": 0.7258641239570918,
+      "learning_rate": 5.4829008893371235e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.7873319625854492,
+      "num_tokens": 187008200.0,
+      "step": 79170
+    },
+    {
+      "epoch": 0.725955808196571,
+      "learning_rate": 5.4810672045475387e-05,
+      "loss": 0.7107,
+      "mean_token_accuracy": 0.7863698482513428,
+      "num_tokens": 187033568.0,
+      "step": 79180
+    },
+    {
+      "epoch": 0.7260474924360503,
+      "learning_rate": 5.479233519757954e-05,
+      "loss": 0.6627,
+      "mean_token_accuracy": 0.7947986125946045,
+      "num_tokens": 187058239.0,
+      "step": 79190
+    },
+    {
+      "epoch": 0.7261391766755295,
+      "learning_rate": 5.47739983496837e-05,
+      "loss": 0.6536,
+      "mean_token_accuracy": 0.7928597748279571,
+      "num_tokens": 187083095.0,
+      "step": 79200
+    },
+    {
+      "epoch": 0.7262308609150087,
+      "learning_rate": 5.475566150178785e-05,
+      "loss": 0.6224,
+      "mean_token_accuracy": 0.8028665602207183,
+      "num_tokens": 187108524.0,
+      "step": 79210
+    },
+    {
+      "epoch": 0.7263225451544879,
+      "learning_rate": 5.4737324653891994e-05,
+      "loss": 0.653,
+      "mean_token_accuracy": 0.8016027867794037,
+      "num_tokens": 187133647.0,
+      "step": 79220
+    },
+    {
+      "epoch": 0.7264142293939672,
+      "learning_rate": 5.4718987805996146e-05,
+      "loss": 0.7373,
+      "mean_token_accuracy": 0.7745027780532837,
+      "num_tokens": 187158796.0,
+      "step": 79230
+    },
+    {
+      "epoch": 0.7265059136334464,
+      "learning_rate": 5.4700650958100305e-05,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.7769539535045624,
+      "num_tokens": 187183073.0,
+      "step": 79240
+    },
+    {
+      "epoch": 0.7265975978729257,
+      "learning_rate": 5.4682314110204456e-05,
+      "loss": 0.686,
+      "mean_token_accuracy": 0.7857530534267425,
+      "num_tokens": 187208311.0,
+      "step": 79250
+    },
+    {
+      "epoch": 0.7266892821124049,
+      "learning_rate": 5.466397726230861e-05,
+      "loss": 0.6749,
+      "mean_token_accuracy": 0.7894758582115173,
+      "num_tokens": 187233143.0,
+      "step": 79260
+    },
+    {
+      "epoch": 0.7267809663518842,
+      "learning_rate": 5.464564041441277e-05,
+      "loss": 0.6843,
+      "mean_token_accuracy": 0.7927349925041198,
+      "num_tokens": 187257810.0,
+      "step": 79270
+    },
+    {
+      "epoch": 0.7268726505913633,
+      "learning_rate": 5.462730356651692e-05,
+      "loss": 0.6607,
+      "mean_token_accuracy": 0.8018940210342407,
+      "num_tokens": 187283158.0,
+      "step": 79280
+    },
+    {
+      "epoch": 0.7269643348308426,
+      "learning_rate": 5.460896671862107e-05,
+      "loss": 0.6781,
+      "mean_token_accuracy": 0.7852358162403107,
+      "num_tokens": 187308925.0,
+      "step": 79290
+    },
+    {
+      "epoch": 0.7270560190703218,
+      "learning_rate": 5.459062987072523e-05,
+      "loss": 0.6915,
+      "mean_token_accuracy": 0.7866684913635253,
+      "num_tokens": 187333988.0,
+      "step": 79300
+    },
+    {
+      "epoch": 0.727147703309801,
+      "learning_rate": 5.457229302282938e-05,
+      "loss": 0.6616,
+      "mean_token_accuracy": 0.7948778569698334,
+      "num_tokens": 187358668.0,
+      "step": 79310
+    },
+    {
+      "epoch": 0.7272393875492803,
+      "learning_rate": 5.4553956174933526e-05,
+      "loss": 0.6912,
+      "mean_token_accuracy": 0.7864884972572327,
+      "num_tokens": 187383966.0,
+      "step": 79320
+    },
+    {
+      "epoch": 0.7273310717887596,
+      "learning_rate": 5.453561932703769e-05,
+      "loss": 0.6864,
+      "mean_token_accuracy": 0.7897108733654022,
+      "num_tokens": 187409724.0,
+      "step": 79330
+    },
+    {
+      "epoch": 0.7274227560282387,
+      "learning_rate": 5.451728247914184e-05,
+      "loss": 0.6758,
+      "mean_token_accuracy": 0.7939182937145233,
+      "num_tokens": 187434764.0,
+      "step": 79340
+    },
+    {
+      "epoch": 0.7275144402677179,
+      "learning_rate": 5.449894563124599e-05,
+      "loss": 0.6694,
+      "mean_token_accuracy": 0.7932112634181976,
+      "num_tokens": 187459904.0,
+      "step": 79350
+    },
+    {
+      "epoch": 0.7276061245071972,
+      "learning_rate": 5.448060878335014e-05,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.7892306923866272,
+      "num_tokens": 187484285.0,
+      "step": 79360
+    },
+    {
+      "epoch": 0.7276978087466764,
+      "learning_rate": 5.44622719354543e-05,
+      "loss": 0.6938,
+      "mean_token_accuracy": 0.7868093073368072,
+      "num_tokens": 187508917.0,
+      "step": 79370
+    },
+    {
+      "epoch": 0.7277894929861557,
+      "learning_rate": 5.444393508755845e-05,
+      "loss": 0.6741,
+      "mean_token_accuracy": 0.7889336228370667,
+      "num_tokens": 187533158.0,
+      "step": 79380
+    },
+    {
+      "epoch": 0.727881177225635,
+      "learning_rate": 5.44255982396626e-05,
+      "loss": 0.6848,
+      "mean_token_accuracy": 0.7920760989189148,
+      "num_tokens": 187558298.0,
+      "step": 79390
+    },
+    {
+      "epoch": 0.7279728614651142,
+      "learning_rate": 5.440726139176676e-05,
+      "loss": 0.699,
+      "mean_token_accuracy": 0.786843329668045,
+      "num_tokens": 187582384.0,
+      "step": 79400
+    },
+    {
+      "epoch": 0.7280645457045933,
+      "learning_rate": 5.4388924543870913e-05,
+      "loss": 0.6537,
+      "mean_token_accuracy": 0.7969779133796692,
+      "num_tokens": 187606863.0,
+      "step": 79410
+    },
+    {
+      "epoch": 0.7281562299440726,
+      "learning_rate": 5.437058769597506e-05,
+      "loss": 0.6615,
+      "mean_token_accuracy": 0.7937743186950683,
+      "num_tokens": 187631978.0,
+      "step": 79420
+    },
+    {
+      "epoch": 0.7282479141835518,
+      "learning_rate": 5.4352250848079224e-05,
+      "loss": 0.6863,
+      "mean_token_accuracy": 0.788418835401535,
+      "num_tokens": 187657222.0,
+      "step": 79430
+    },
+    {
+      "epoch": 0.7283395984230311,
+      "learning_rate": 5.433391400018337e-05,
+      "loss": 0.6521,
+      "mean_token_accuracy": 0.7989540219306945,
+      "num_tokens": 187682955.0,
+      "step": 79440
+    },
+    {
+      "epoch": 0.7284312826625103,
+      "learning_rate": 5.431557715228752e-05,
+      "loss": 0.6683,
+      "mean_token_accuracy": 0.7878883302211761,
+      "num_tokens": 187708001.0,
+      "step": 79450
+    },
+    {
+      "epoch": 0.7285229669019896,
+      "learning_rate": 5.429724030439167e-05,
+      "loss": 0.6394,
+      "mean_token_accuracy": 0.8020074605941773,
+      "num_tokens": 187733155.0,
+      "step": 79460
+    },
+    {
+      "epoch": 0.7286146511414687,
+      "learning_rate": 5.427890345649583e-05,
+      "loss": 0.675,
+      "mean_token_accuracy": 0.7911485075950623,
+      "num_tokens": 187757810.0,
+      "step": 79470
+    },
+    {
+      "epoch": 0.728706335380948,
+      "learning_rate": 5.426056660859998e-05,
+      "loss": 0.6423,
+      "mean_token_accuracy": 0.8018137454986572,
+      "num_tokens": 187783479.0,
+      "step": 79480
+    },
+    {
+      "epoch": 0.7287980196204272,
+      "learning_rate": 5.4242229760704135e-05,
+      "loss": 0.7191,
+      "mean_token_accuracy": 0.7751696288585663,
+      "num_tokens": 187808545.0,
+      "step": 79490
+    },
+    {
+      "epoch": 0.7288897038599065,
+      "learning_rate": 5.4223892912808294e-05,
+      "loss": 0.678,
+      "mean_token_accuracy": 0.7919573366641999,
+      "num_tokens": 187832865.0,
+      "step": 79500
+    },
+    {
+      "epoch": 0.7289813880993857,
+      "learning_rate": 5.4205556064912446e-05,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.7857295870780945,
+      "num_tokens": 187858257.0,
+      "step": 79510
+    },
+    {
+      "epoch": 0.729073072338865,
+      "learning_rate": 5.418721921701659e-05,
+      "loss": 0.7078,
+      "mean_token_accuracy": 0.7824679374694824,
+      "num_tokens": 187883762.0,
+      "step": 79520
+    },
+    {
+      "epoch": 0.7291647565783442,
+      "learning_rate": 5.4168882369120756e-05,
+      "loss": 0.6654,
+      "mean_token_accuracy": 0.7948396861553192,
+      "num_tokens": 187909411.0,
+      "step": 79530
+    },
+    {
+      "epoch": 0.7292564408178234,
+      "learning_rate": 5.415054552122491e-05,
+      "loss": 0.6835,
+      "mean_token_accuracy": 0.7936994552612304,
+      "num_tokens": 187934935.0,
+      "step": 79540
+    },
+    {
+      "epoch": 0.7293481250573026,
+      "learning_rate": 5.413220867332905e-05,
+      "loss": 0.6583,
+      "mean_token_accuracy": 0.7967337548732758,
+      "num_tokens": 187960807.0,
+      "step": 79550
+    },
+    {
+      "epoch": 0.7294398092967819,
+      "learning_rate": 5.411387182543322e-05,
+      "loss": 0.7154,
+      "mean_token_accuracy": 0.7844775915145874,
+      "num_tokens": 187985519.0,
+      "step": 79560
+    },
+    {
+      "epoch": 0.7295314935362611,
+      "learning_rate": 5.4095534977537364e-05,
+      "loss": 0.6693,
+      "mean_token_accuracy": 0.7947754561901093,
+      "num_tokens": 188010595.0,
+      "step": 79570
+    },
+    {
+      "epoch": 0.7296231777757404,
+      "learning_rate": 5.4077198129641516e-05,
+      "loss": 0.6715,
+      "mean_token_accuracy": 0.7909493327140809,
+      "num_tokens": 188035749.0,
+      "step": 79580
+    },
+    {
+      "epoch": 0.7297148620152196,
+      "learning_rate": 5.405886128174567e-05,
+      "loss": 0.6849,
+      "mean_token_accuracy": 0.7898952901363373,
+      "num_tokens": 188061378.0,
+      "step": 79590
+    },
+    {
+      "epoch": 0.7298065462546988,
+      "learning_rate": 5.4040524433849826e-05,
+      "loss": 0.6771,
+      "mean_token_accuracy": 0.7913882613182068,
+      "num_tokens": 188085486.0,
+      "step": 79600
+    },
+    {
+      "epoch": 0.729898230494178,
+      "learning_rate": 5.402218758595398e-05,
+      "loss": 0.6478,
+      "mean_token_accuracy": 0.7929193317890167,
+      "num_tokens": 188110385.0,
+      "step": 79610
+    },
+    {
+      "epoch": 0.7299899147336573,
+      "learning_rate": 5.400385073805812e-05,
+      "loss": 0.6686,
+      "mean_token_accuracy": 0.7889566123485565,
+      "num_tokens": 188136078.0,
+      "step": 79620
+    },
+    {
+      "epoch": 0.7300815989731365,
+      "learning_rate": 5.398551389016229e-05,
+      "loss": 0.6699,
+      "mean_token_accuracy": 0.7939992249011993,
+      "num_tokens": 188162378.0,
+      "step": 79630
+    },
+    {
+      "epoch": 0.7301732832126158,
+      "learning_rate": 5.396717704226644e-05,
+      "loss": 0.6702,
+      "mean_token_accuracy": 0.7892761111259461,
+      "num_tokens": 188187023.0,
+      "step": 79640
+    },
+    {
+      "epoch": 0.730264967452095,
+      "learning_rate": 5.3948840194370585e-05,
+      "loss": 0.6937,
+      "mean_token_accuracy": 0.7890453279018402,
+      "num_tokens": 188211766.0,
+      "step": 79650
+    },
+    {
+      "epoch": 0.7303566516915743,
+      "learning_rate": 5.393050334647475e-05,
+      "loss": 0.6807,
+      "mean_token_accuracy": 0.7881542801856994,
+      "num_tokens": 188237641.0,
+      "step": 79660
+    },
+    {
+      "epoch": 0.7304483359310534,
+      "learning_rate": 5.3912166498578896e-05,
+      "loss": 0.7006,
+      "mean_token_accuracy": 0.7916547179222106,
+      "num_tokens": 188262338.0,
+      "step": 79670
+    },
+    {
+      "epoch": 0.7305400201705327,
+      "learning_rate": 5.389382965068305e-05,
+      "loss": 0.6598,
+      "mean_token_accuracy": 0.795462167263031,
+      "num_tokens": 188288257.0,
+      "step": 79680
+    },
+    {
+      "epoch": 0.7306317044100119,
+      "learning_rate": 5.3875492802787206e-05,
+      "loss": 0.6559,
+      "mean_token_accuracy": 0.7956127762794495,
+      "num_tokens": 188313085.0,
+      "step": 79690
+    },
+    {
+      "epoch": 0.7307233886494912,
+      "learning_rate": 5.385715595489136e-05,
+      "loss": 0.7098,
+      "mean_token_accuracy": 0.7878253340721131,
+      "num_tokens": 188337592.0,
+      "step": 79700
+    },
+    {
+      "epoch": 0.7308150728889704,
+      "learning_rate": 5.383881910699551e-05,
+      "loss": 0.6787,
+      "mean_token_accuracy": 0.7885228514671325,
+      "num_tokens": 188363528.0,
+      "step": 79710
+    },
+    {
+      "epoch": 0.7309067571284497,
+      "learning_rate": 5.3820482259099655e-05,
+      "loss": 0.6931,
+      "mean_token_accuracy": 0.7864165544509888,
+      "num_tokens": 188389037.0,
+      "step": 79720
+    },
+    {
+      "epoch": 0.7309984413679289,
+      "learning_rate": 5.380214541120382e-05,
+      "loss": 0.6835,
+      "mean_token_accuracy": 0.7906298160552978,
+      "num_tokens": 188413023.0,
+      "step": 79730
+    },
+    {
+      "epoch": 0.731090125607408,
+      "learning_rate": 5.378380856330797e-05,
+      "loss": 0.7135,
+      "mean_token_accuracy": 0.7800632536411285,
+      "num_tokens": 188438215.0,
+      "step": 79740
+    },
+    {
+      "epoch": 0.7311818098468873,
+      "learning_rate": 5.376547171541212e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.7926500022411347,
+      "num_tokens": 188463977.0,
+      "step": 79750
+    },
+    {
+      "epoch": 0.7312734940863665,
+      "learning_rate": 5.374713486751628e-05,
+      "loss": 0.6831,
+      "mean_token_accuracy": 0.7888989210128784,
+      "num_tokens": 188488445.0,
+      "step": 79760
+    },
+    {
+      "epoch": 0.7313651783258458,
+      "learning_rate": 5.372879801962043e-05,
+      "loss": 0.6517,
+      "mean_token_accuracy": 0.7938899755477905,
+      "num_tokens": 188513727.0,
+      "step": 79770
+    },
+    {
+      "epoch": 0.731456862565325,
+      "learning_rate": 5.371046117172458e-05,
+      "loss": 0.6637,
+      "mean_token_accuracy": 0.7909485876560212,
+      "num_tokens": 188539201.0,
+      "step": 79780
+    },
+    {
+      "epoch": 0.7315485468048043,
+      "learning_rate": 5.369212432382874e-05,
+      "loss": 0.6688,
+      "mean_token_accuracy": 0.7948606193065644,
+      "num_tokens": 188563571.0,
+      "step": 79790
+    },
+    {
+      "epoch": 0.7316402310442834,
+      "learning_rate": 5.367378747593289e-05,
+      "loss": 0.6606,
+      "mean_token_accuracy": 0.7913779854774475,
+      "num_tokens": 188588312.0,
+      "step": 79800
+    },
+    {
+      "epoch": 0.7317319152837627,
+      "learning_rate": 5.365545062803704e-05,
+      "loss": 0.6624,
+      "mean_token_accuracy": 0.7969732344150543,
+      "num_tokens": 188613882.0,
+      "step": 79810
+    },
+    {
+      "epoch": 0.7318235995232419,
+      "learning_rate": 5.36371137801412e-05,
+      "loss": 0.6948,
+      "mean_token_accuracy": 0.7828238844871521,
+      "num_tokens": 188639571.0,
+      "step": 79820
+    },
+    {
+      "epoch": 0.7319152837627212,
+      "learning_rate": 5.361877693224535e-05,
+      "loss": 0.6649,
+      "mean_token_accuracy": 0.7942383110523223,
+      "num_tokens": 188664226.0,
+      "step": 79830
+    },
+    {
+      "epoch": 0.7320069680022004,
+      "learning_rate": 5.3600440084349505e-05,
+      "loss": 0.6544,
+      "mean_token_accuracy": 0.7965122163295746,
+      "num_tokens": 188689153.0,
+      "step": 79840
+    },
+    {
+      "epoch": 0.7320986522416797,
+      "learning_rate": 5.358210323645365e-05,
+      "loss": 0.677,
+      "mean_token_accuracy": 0.7880686104297638,
+      "num_tokens": 188714115.0,
+      "step": 79850
+    },
+    {
+      "epoch": 0.7321903364811589,
+      "learning_rate": 5.3563766388557815e-05,
+      "loss": 0.7405,
+      "mean_token_accuracy": 0.7743090927600861,
+      "num_tokens": 188739708.0,
+      "step": 79860
+    },
+    {
+      "epoch": 0.7322820207206381,
+      "learning_rate": 5.354542954066196e-05,
+      "loss": 0.6461,
+      "mean_token_accuracy": 0.7960841357707977,
+      "num_tokens": 188765488.0,
+      "step": 79870
+    },
+    {
+      "epoch": 0.7323737049601173,
+      "learning_rate": 5.352709269276611e-05,
+      "loss": 0.6588,
+      "mean_token_accuracy": 0.7978433609008789,
+      "num_tokens": 188790052.0,
+      "step": 79880
+    },
+    {
+      "epoch": 0.7324653891995966,
+      "learning_rate": 5.350875584487027e-05,
+      "loss": 0.7014,
+      "mean_token_accuracy": 0.7835223853588105,
+      "num_tokens": 188815253.0,
+      "step": 79890
+    },
+    {
+      "epoch": 0.7325570734390758,
+      "learning_rate": 5.349041899697442e-05,
+      "loss": 0.6699,
+      "mean_token_accuracy": 0.7923356831073761,
+      "num_tokens": 188840657.0,
+      "step": 79900
+    },
+    {
+      "epoch": 0.7326487576785551,
+      "learning_rate": 5.3472082149078575e-05,
+      "loss": 0.6786,
+      "mean_token_accuracy": 0.7936592340469361,
+      "num_tokens": 188866276.0,
+      "step": 79910
+    },
+    {
+      "epoch": 0.7327404419180343,
+      "learning_rate": 5.345374530118273e-05,
+      "loss": 0.6856,
+      "mean_token_accuracy": 0.7864917993545533,
+      "num_tokens": 188891612.0,
+      "step": 79920
+    },
+    {
+      "epoch": 0.7328321261575135,
+      "learning_rate": 5.3435408453286885e-05,
+      "loss": 0.636,
+      "mean_token_accuracy": 0.8033297240734101,
+      "num_tokens": 188917104.0,
+      "step": 79930
+    },
+    {
+      "epoch": 0.7329238103969927,
+      "learning_rate": 5.341707160539104e-05,
+      "loss": 0.6569,
+      "mean_token_accuracy": 0.7944091916084289,
+      "num_tokens": 188941870.0,
+      "step": 79940
+    },
+    {
+      "epoch": 0.733015494636472,
+      "learning_rate": 5.3398734757495196e-05,
+      "loss": 0.6729,
+      "mean_token_accuracy": 0.7910976231098175,
+      "num_tokens": 188966674.0,
+      "step": 79950
+    },
+    {
+      "epoch": 0.7331071788759512,
+      "learning_rate": 5.338039790959935e-05,
+      "loss": 0.6593,
+      "mean_token_accuracy": 0.7933946967124939,
+      "num_tokens": 188992114.0,
+      "step": 79960
+    },
+    {
+      "epoch": 0.7331988631154305,
+      "learning_rate": 5.336206106170349e-05,
+      "loss": 0.6451,
+      "mean_token_accuracy": 0.8009393155574799,
+      "num_tokens": 189017131.0,
+      "step": 79970
+    },
+    {
+      "epoch": 0.7332905473549097,
+      "learning_rate": 5.3343724213807644e-05,
+      "loss": 0.6532,
+      "mean_token_accuracy": 0.7954998254776001,
+      "num_tokens": 189041900.0,
+      "step": 79980
+    },
+    {
+      "epoch": 0.733382231594389,
+      "learning_rate": 5.33253873659118e-05,
+      "loss": 0.7007,
+      "mean_token_accuracy": 0.7831595361232757,
+      "num_tokens": 189066439.0,
+      "step": 79990
+    },
+    {
+      "epoch": 0.7334739158338681,
+      "learning_rate": 5.3307050518015955e-05,
+      "loss": 0.675,
+      "mean_token_accuracy": 0.7911353826522827,
+      "num_tokens": 189091733.0,
+      "step": 80000
+    },
+    {
+      "epoch": 0.7335656000733474,
+      "learning_rate": 5.328871367012011e-05,
+      "loss": 0.6725,
+      "mean_token_accuracy": 0.7923921167850494,
+      "num_tokens": 189117742.0,
+      "step": 80010
+    },
+    {
+      "epoch": 0.7336572843128266,
+      "learning_rate": 5.3270376822224265e-05,
+      "loss": 0.6875,
+      "mean_token_accuracy": 0.7831376135349274,
+      "num_tokens": 189143435.0,
+      "step": 80020
+    },
+    {
+      "epoch": 0.7337489685523059,
+      "learning_rate": 5.325203997432842e-05,
+      "loss": 0.7023,
+      "mean_token_accuracy": 0.7888618469238281,
+      "num_tokens": 189168928.0,
+      "step": 80030
+    },
+    {
+      "epoch": 0.7338406527917851,
+      "learning_rate": 5.323370312643257e-05,
+      "loss": 0.6345,
+      "mean_token_accuracy": 0.801346880197525,
+      "num_tokens": 189194610.0,
+      "step": 80040
+    },
+    {
+      "epoch": 0.7339323370312644,
+      "learning_rate": 5.321536627853673e-05,
+      "loss": 0.7002,
+      "mean_token_accuracy": 0.7882979094982148,
+      "num_tokens": 189219652.0,
+      "step": 80050
+    },
+    {
+      "epoch": 0.7340240212707435,
+      "learning_rate": 5.319702943064088e-05,
+      "loss": 0.6795,
+      "mean_token_accuracy": 0.7905156254768372,
+      "num_tokens": 189244274.0,
+      "step": 80060
+    },
+    {
+      "epoch": 0.7341157055102228,
+      "learning_rate": 5.3178692582745025e-05,
+      "loss": 0.6471,
+      "mean_token_accuracy": 0.8073345959186554,
+      "num_tokens": 189269626.0,
+      "step": 80070
+    },
+    {
+      "epoch": 0.734207389749702,
+      "learning_rate": 5.316035573484918e-05,
+      "loss": 0.6737,
+      "mean_token_accuracy": 0.792496246099472,
+      "num_tokens": 189295147.0,
+      "step": 80080
+    },
+    {
+      "epoch": 0.7342990739891813,
+      "learning_rate": 5.3142018886953335e-05,
+      "loss": 0.6483,
+      "mean_token_accuracy": 0.7981096088886261,
+      "num_tokens": 189320059.0,
+      "step": 80090
+    },
+    {
+      "epoch": 0.7343907582286605,
+      "learning_rate": 5.312368203905749e-05,
+      "loss": 0.6817,
+      "mean_token_accuracy": 0.7924897193908691,
+      "num_tokens": 189346014.0,
+      "step": 80100
+    },
+    {
+      "epoch": 0.7344824424681398,
+      "learning_rate": 5.310534519116164e-05,
+      "loss": 0.6588,
+      "mean_token_accuracy": 0.7879400074481964,
+      "num_tokens": 189370992.0,
+      "step": 80110
+    },
+    {
+      "epoch": 0.734574126707619,
+      "learning_rate": 5.30870083432658e-05,
+      "loss": 0.6527,
+      "mean_token_accuracy": 0.7944809854030609,
+      "num_tokens": 189395670.0,
+      "step": 80120
+    },
+    {
+      "epoch": 0.7346658109470982,
+      "learning_rate": 5.306867149536995e-05,
+      "loss": 0.677,
+      "mean_token_accuracy": 0.7863703429698944,
+      "num_tokens": 189420540.0,
+      "step": 80130
+    },
+    {
+      "epoch": 0.7347574951865774,
+      "learning_rate": 5.30503346474741e-05,
+      "loss": 0.6632,
+      "mean_token_accuracy": 0.7973880887031555,
+      "num_tokens": 189445936.0,
+      "step": 80140
+    },
+    {
+      "epoch": 0.7348491794260567,
+      "learning_rate": 5.303199779957826e-05,
+      "loss": 0.7363,
+      "mean_token_accuracy": 0.7791323900222779,
+      "num_tokens": 189470074.0,
+      "step": 80150
+    },
+    {
+      "epoch": 0.7349408636655359,
+      "learning_rate": 5.301366095168241e-05,
+      "loss": 0.6967,
+      "mean_token_accuracy": 0.780820369720459,
+      "num_tokens": 189495225.0,
+      "step": 80160
+    },
+    {
+      "epoch": 0.7350325479050152,
+      "learning_rate": 5.299532410378656e-05,
+      "loss": 0.6859,
+      "mean_token_accuracy": 0.7911755859851837,
+      "num_tokens": 189520377.0,
+      "step": 80170
+    },
+    {
+      "epoch": 0.7351242321444944,
+      "learning_rate": 5.297698725589072e-05,
+      "loss": 0.6969,
+      "mean_token_accuracy": 0.7861814081668854,
+      "num_tokens": 189545211.0,
+      "step": 80180
+    },
+    {
+      "epoch": 0.7352159163839735,
+      "learning_rate": 5.295865040799487e-05,
+      "loss": 0.7017,
+      "mean_token_accuracy": 0.7866884708404541,
+      "num_tokens": 189569528.0,
+      "step": 80190
+    },
+    {
+      "epoch": 0.7353076006234528,
+      "learning_rate": 5.294031356009902e-05,
+      "loss": 0.656,
+      "mean_token_accuracy": 0.7965950012207031,
+      "num_tokens": 189593828.0,
+      "step": 80200
+    },
+    {
+      "epoch": 0.735399284862932,
+      "learning_rate": 5.292197671220317e-05,
+      "loss": 0.6496,
+      "mean_token_accuracy": 0.7972586572170257,
+      "num_tokens": 189619342.0,
+      "step": 80210
+    },
+    {
+      "epoch": 0.7354909691024113,
+      "learning_rate": 5.290363986430733e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.7861582815647126,
+      "num_tokens": 189644553.0,
+      "step": 80220
+    },
+    {
+      "epoch": 0.7355826533418905,
+      "learning_rate": 5.288530301641148e-05,
+      "loss": 0.6955,
+      "mean_token_accuracy": 0.789175671339035,
+      "num_tokens": 189669080.0,
+      "step": 80230
+    },
+    {
+      "epoch": 0.7356743375813698,
+      "learning_rate": 5.2866966168515634e-05,
+      "loss": 0.7158,
+      "mean_token_accuracy": 0.7790573120117188,
+      "num_tokens": 189693329.0,
+      "step": 80240
+    },
+    {
+      "epoch": 0.735766021820849,
+      "learning_rate": 5.284862932061979e-05,
+      "loss": 0.6777,
+      "mean_token_accuracy": 0.7848819613456726,
+      "num_tokens": 189718376.0,
+      "step": 80250
+    },
+    {
+      "epoch": 0.7358577060603282,
+      "learning_rate": 5.2830292472723944e-05,
+      "loss": 0.685,
+      "mean_token_accuracy": 0.7896092176437378,
+      "num_tokens": 189743345.0,
+      "step": 80260
+    },
+    {
+      "epoch": 0.7359493902998074,
+      "learning_rate": 5.281195562482809e-05,
+      "loss": 0.7118,
+      "mean_token_accuracy": 0.7788365304470062,
+      "num_tokens": 189768513.0,
+      "step": 80270
+    },
+    {
+      "epoch": 0.7360410745392867,
+      "learning_rate": 5.2793618776932255e-05,
+      "loss": 0.6942,
+      "mean_token_accuracy": 0.7790515959262848,
+      "num_tokens": 189792843.0,
+      "step": 80280
+    },
+    {
+      "epoch": 0.7361327587787659,
+      "learning_rate": 5.27752819290364e-05,
+      "loss": 0.6719,
+      "mean_token_accuracy": 0.7966277360916137,
+      "num_tokens": 189817526.0,
+      "step": 80290
+    },
+    {
+      "epoch": 0.7362244430182452,
+      "learning_rate": 5.275694508114055e-05,
+      "loss": 0.6863,
+      "mean_token_accuracy": 0.7920669257640839,
+      "num_tokens": 189842854.0,
+      "step": 80300
+    },
+    {
+      "epoch": 0.7363161272577244,
+      "learning_rate": 5.273860823324471e-05,
+      "loss": 0.6438,
+      "mean_token_accuracy": 0.7960306406021118,
+      "num_tokens": 189867998.0,
+      "step": 80310
+    },
+    {
+      "epoch": 0.7364078114972036,
+      "learning_rate": 5.272027138534886e-05,
+      "loss": 0.675,
+      "mean_token_accuracy": 0.7901750206947327,
+      "num_tokens": 189893032.0,
+      "step": 80320
+    },
+    {
+      "epoch": 0.7364994957366828,
+      "learning_rate": 5.2701934537453014e-05,
+      "loss": 0.6655,
+      "mean_token_accuracy": 0.7903243482112885,
+      "num_tokens": 189917930.0,
+      "step": 80330
+    },
+    {
+      "epoch": 0.7365911799761621,
+      "learning_rate": 5.2683597689557166e-05,
+      "loss": 0.6882,
+      "mean_token_accuracy": 0.7891256392002106,
+      "num_tokens": 189942763.0,
+      "step": 80340
+    },
+    {
+      "epoch": 0.7366828642156413,
+      "learning_rate": 5.2665260841661324e-05,
+      "loss": 0.6675,
+      "mean_token_accuracy": 0.7998230636119843,
+      "num_tokens": 189968362.0,
+      "step": 80350
+    },
+    {
+      "epoch": 0.7367745484551206,
+      "learning_rate": 5.2646923993765476e-05,
+      "loss": 0.6689,
+      "mean_token_accuracy": 0.7921317398548127,
+      "num_tokens": 189994107.0,
+      "step": 80360
+    },
+    {
+      "epoch": 0.7368662326945998,
+      "learning_rate": 5.262858714586962e-05,
+      "loss": 0.6829,
+      "mean_token_accuracy": 0.7936067163944245,
+      "num_tokens": 190019056.0,
+      "step": 80370
+    },
+    {
+      "epoch": 0.7369579169340791,
+      "learning_rate": 5.261025029797379e-05,
+      "loss": 0.6836,
+      "mean_token_accuracy": 0.7922172784805298,
+      "num_tokens": 190044149.0,
+      "step": 80380
+    },
+    {
+      "epoch": 0.7370496011735582,
+      "learning_rate": 5.259191345007793e-05,
+      "loss": 0.726,
+      "mean_token_accuracy": 0.783193153142929,
+      "num_tokens": 190069332.0,
+      "step": 80390
+    },
+    {
+      "epoch": 0.7371412854130375,
+      "learning_rate": 5.2573576602182084e-05,
+      "loss": 0.6944,
+      "mean_token_accuracy": 0.7866541266441345,
+      "num_tokens": 190095058.0,
+      "step": 80400
+    },
+    {
+      "epoch": 0.7372329696525167,
+      "learning_rate": 5.255523975428624e-05,
+      "loss": 0.7162,
+      "mean_token_accuracy": 0.7798259079456329,
+      "num_tokens": 190119728.0,
+      "step": 80410
+    },
+    {
+      "epoch": 0.737324653891996,
+      "learning_rate": 5.2536902906390394e-05,
+      "loss": 0.6717,
+      "mean_token_accuracy": 0.7873007714748382,
+      "num_tokens": 190145020.0,
+      "step": 80420
+    },
+    {
+      "epoch": 0.7374163381314752,
+      "learning_rate": 5.2518566058494546e-05,
+      "loss": 0.6984,
+      "mean_token_accuracy": 0.7886401295661927,
+      "num_tokens": 190169938.0,
+      "step": 80430
+    },
+    {
+      "epoch": 0.7375080223709545,
+      "learning_rate": 5.2500229210598705e-05,
+      "loss": 0.6665,
+      "mean_token_accuracy": 0.7915529727935791,
+      "num_tokens": 190195735.0,
+      "step": 80440
+    },
+    {
+      "epoch": 0.7375997066104336,
+      "learning_rate": 5.248189236270286e-05,
+      "loss": 0.6613,
+      "mean_token_accuracy": 0.7929513573646545,
+      "num_tokens": 190221335.0,
+      "step": 80450
+    },
+    {
+      "epoch": 0.7376913908499129,
+      "learning_rate": 5.246355551480701e-05,
+      "loss": 0.6916,
+      "mean_token_accuracy": 0.7865171432495117,
+      "num_tokens": 190247419.0,
+      "step": 80460
+    },
+    {
+      "epoch": 0.7377830750893921,
+      "learning_rate": 5.2445218666911154e-05,
+      "loss": 0.6701,
+      "mean_token_accuracy": 0.7929346561431885,
+      "num_tokens": 190272489.0,
+      "step": 80470
+    },
+    {
+      "epoch": 0.7378747593288714,
+      "learning_rate": 5.242688181901532e-05,
+      "loss": 0.6675,
+      "mean_token_accuracy": 0.7921228289604187,
+      "num_tokens": 190297713.0,
+      "step": 80480
+    },
+    {
+      "epoch": 0.7379664435683506,
+      "learning_rate": 5.2408544971119464e-05,
+      "loss": 0.7003,
+      "mean_token_accuracy": 0.7888290166854859,
+      "num_tokens": 190322807.0,
+      "step": 80490
+    },
+    {
+      "epoch": 0.7380581278078299,
+      "learning_rate": 5.2390208123223616e-05,
+      "loss": 0.6652,
+      "mean_token_accuracy": 0.794892543554306,
+      "num_tokens": 190347267.0,
+      "step": 80500
+    },
+    {
+      "epoch": 0.7381498120473091,
+      "learning_rate": 5.2371871275327775e-05,
+      "loss": 0.702,
+      "mean_token_accuracy": 0.7819541871547699,
+      "num_tokens": 190372358.0,
+      "step": 80510
+    },
+    {
+      "epoch": 0.7382414962867883,
+      "learning_rate": 5.2353534427431927e-05,
+      "loss": 0.6504,
+      "mean_token_accuracy": 0.7943321585655212,
+      "num_tokens": 190397426.0,
+      "step": 80520
+    },
+    {
+      "epoch": 0.7383331805262675,
+      "learning_rate": 5.233519757953608e-05,
+      "loss": 0.6846,
+      "mean_token_accuracy": 0.7873305916786194,
+      "num_tokens": 190422588.0,
+      "step": 80530
+    },
+    {
+      "epoch": 0.7384248647657468,
+      "learning_rate": 5.231686073164024e-05,
+      "loss": 0.634,
+      "mean_token_accuracy": 0.7992035388946533,
+      "num_tokens": 190448233.0,
+      "step": 80540
+    },
+    {
+      "epoch": 0.738516549005226,
+      "learning_rate": 5.229852388374439e-05,
+      "loss": 0.6922,
+      "mean_token_accuracy": 0.7892242312431336,
+      "num_tokens": 190473498.0,
+      "step": 80550
+    },
+    {
+      "epoch": 0.7386082332447053,
+      "learning_rate": 5.228018703584854e-05,
+      "loss": 0.7053,
+      "mean_token_accuracy": 0.7846129417419434,
+      "num_tokens": 190499721.0,
+      "step": 80560
+    },
+    {
+      "epoch": 0.7386999174841845,
+      "learning_rate": 5.22618501879527e-05,
+      "loss": 0.6695,
+      "mean_token_accuracy": 0.7938514113426208,
+      "num_tokens": 190524438.0,
+      "step": 80570
+    },
+    {
+      "epoch": 0.7387916017236636,
+      "learning_rate": 5.224351334005685e-05,
+      "loss": 0.6859,
+      "mean_token_accuracy": 0.7834884047508239,
+      "num_tokens": 190549560.0,
+      "step": 80580
+    },
+    {
+      "epoch": 0.7388832859631429,
+      "learning_rate": 5.2225176492160996e-05,
+      "loss": 0.6764,
+      "mean_token_accuracy": 0.79319748878479,
+      "num_tokens": 190574271.0,
+      "step": 80590
+    },
+    {
+      "epoch": 0.7389749702026222,
+      "learning_rate": 5.220683964426515e-05,
+      "loss": 0.6765,
+      "mean_token_accuracy": 0.7932425737380981,
+      "num_tokens": 190599548.0,
+      "step": 80600
+    },
+    {
+      "epoch": 0.7390666544421014,
+      "learning_rate": 5.218850279636931e-05,
+      "loss": 0.6879,
+      "mean_token_accuracy": 0.7921907782554627,
+      "num_tokens": 190625290.0,
+      "step": 80610
+    },
+    {
+      "epoch": 0.7391583386815807,
+      "learning_rate": 5.217016594847346e-05,
+      "loss": 0.6744,
+      "mean_token_accuracy": 0.789624297618866,
+      "num_tokens": 190649458.0,
+      "step": 80620
+    },
+    {
+      "epoch": 0.7392500229210599,
+      "learning_rate": 5.215182910057761e-05,
+      "loss": 0.6833,
+      "mean_token_accuracy": 0.7908877253532409,
+      "num_tokens": 190674600.0,
+      "step": 80630
+    },
+    {
+      "epoch": 0.7393417071605392,
+      "learning_rate": 5.213349225268177e-05,
+      "loss": 0.6692,
+      "mean_token_accuracy": 0.7917872250080109,
+      "num_tokens": 190699891.0,
+      "step": 80640
+    },
+    {
+      "epoch": 0.7394333914000183,
+      "learning_rate": 5.211515540478592e-05,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.7962104916572571,
+      "num_tokens": 190724105.0,
+      "step": 80650
+    },
+    {
+      "epoch": 0.7395250756394975,
+      "learning_rate": 5.209681855689007e-05,
+      "loss": 0.6819,
+      "mean_token_accuracy": 0.7885459780693054,
+      "num_tokens": 190749063.0,
+      "step": 80660
+    },
+    {
+      "epoch": 0.7396167598789768,
+      "learning_rate": 5.207848170899423e-05,
+      "loss": 0.6524,
+      "mean_token_accuracy": 0.8003826260566711,
+      "num_tokens": 190773615.0,
+      "step": 80670
+    },
+    {
+      "epoch": 0.739708444118456,
+      "learning_rate": 5.2060144861098384e-05,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.7876509904861451,
+      "num_tokens": 190799197.0,
+      "step": 80680
+    },
+    {
+      "epoch": 0.7398001283579353,
+      "learning_rate": 5.204180801320253e-05,
+      "loss": 0.6871,
+      "mean_token_accuracy": 0.7913267910480499,
+      "num_tokens": 190824244.0,
+      "step": 80690
+    },
+    {
+      "epoch": 0.7398918125974145,
+      "learning_rate": 5.202347116530668e-05,
+      "loss": 0.7032,
+      "mean_token_accuracy": 0.7855749905109406,
+      "num_tokens": 190848853.0,
+      "step": 80700
+    },
+    {
+      "epoch": 0.7399834968368937,
+      "learning_rate": 5.200513431741084e-05,
+      "loss": 0.6792,
+      "mean_token_accuracy": 0.7900627136230469,
+      "num_tokens": 190874001.0,
+      "step": 80710
+    },
+    {
+      "epoch": 0.7400751810763729,
+      "learning_rate": 5.198679746951499e-05,
+      "loss": 0.6745,
+      "mean_token_accuracy": 0.7865762948989868,
+      "num_tokens": 190899139.0,
+      "step": 80720
+    },
+    {
+      "epoch": 0.7401668653158522,
+      "learning_rate": 5.196846062161914e-05,
+      "loss": 0.6584,
+      "mean_token_accuracy": 0.7981809079647064,
+      "num_tokens": 190924074.0,
+      "step": 80730
+    },
+    {
+      "epoch": 0.7402585495553314,
+      "learning_rate": 5.19501237737233e-05,
+      "loss": 0.6772,
+      "mean_token_accuracy": 0.7952097833156586,
+      "num_tokens": 190949958.0,
+      "step": 80740
+    },
+    {
+      "epoch": 0.7403502337948107,
+      "learning_rate": 5.1931786925827453e-05,
+      "loss": 0.7061,
+      "mean_token_accuracy": 0.7916275203227997,
+      "num_tokens": 190975420.0,
+      "step": 80750
+    },
+    {
+      "epoch": 0.7404419180342899,
+      "learning_rate": 5.1913450077931605e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.7918824315071106,
+      "num_tokens": 191000167.0,
+      "step": 80760
+    },
+    {
+      "epoch": 0.7405336022737692,
+      "learning_rate": 5.1895113230035764e-05,
+      "loss": 0.6711,
+      "mean_token_accuracy": 0.7908523559570313,
+      "num_tokens": 191024939.0,
+      "step": 80770
+    },
+    {
+      "epoch": 0.7406252865132483,
+      "learning_rate": 5.1876776382139916e-05,
+      "loss": 0.667,
+      "mean_token_accuracy": 0.8000496864318848,
+      "num_tokens": 191050622.0,
+      "step": 80780
+    },
+    {
+      "epoch": 0.7407169707527276,
+      "learning_rate": 5.185843953424406e-05,
+      "loss": 0.626,
+      "mean_token_accuracy": 0.8028790414333343,
+      "num_tokens": 191076075.0,
+      "step": 80790
+    },
+    {
+      "epoch": 0.7408086549922068,
+      "learning_rate": 5.1840102686348226e-05,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.7822464406490326,
+      "num_tokens": 191101571.0,
+      "step": 80800
+    },
+    {
+      "epoch": 0.7409003392316861,
+      "learning_rate": 5.182176583845237e-05,
+      "loss": 0.6393,
+      "mean_token_accuracy": 0.800845468044281,
+      "num_tokens": 191126630.0,
+      "step": 80810
+    },
+    {
+      "epoch": 0.7409920234711653,
+      "learning_rate": 5.180342899055652e-05,
+      "loss": 0.6678,
+      "mean_token_accuracy": 0.7918784499168396,
+      "num_tokens": 191151715.0,
+      "step": 80820
+    },
+    {
+      "epoch": 0.7410837077106446,
+      "learning_rate": 5.1785092142660675e-05,
+      "loss": 0.6951,
+      "mean_token_accuracy": 0.7841989398002625,
+      "num_tokens": 191176017.0,
+      "step": 80830
+    },
+    {
+      "epoch": 0.7411753919501237,
+      "learning_rate": 5.1766755294764834e-05,
+      "loss": 0.6798,
+      "mean_token_accuracy": 0.7880084216594696,
+      "num_tokens": 191201586.0,
+      "step": 80840
+    },
+    {
+      "epoch": 0.741267076189603,
+      "learning_rate": 5.1748418446868986e-05,
+      "loss": 0.6921,
+      "mean_token_accuracy": 0.7903082966804504,
+      "num_tokens": 191225840.0,
+      "step": 80850
+    },
+    {
+      "epoch": 0.7413587604290822,
+      "learning_rate": 5.173008159897314e-05,
+      "loss": 0.7165,
+      "mean_token_accuracy": 0.7847334921360016,
+      "num_tokens": 191251739.0,
+      "step": 80860
+    },
+    {
+      "epoch": 0.7414504446685615,
+      "learning_rate": 5.1711744751077296e-05,
+      "loss": 0.6817,
+      "mean_token_accuracy": 0.787997943162918,
+      "num_tokens": 191276880.0,
+      "step": 80870
+    },
+    {
+      "epoch": 0.7415421289080407,
+      "learning_rate": 5.169340790318145e-05,
+      "loss": 0.6506,
+      "mean_token_accuracy": 0.7952578723430633,
+      "num_tokens": 191302208.0,
+      "step": 80880
+    },
+    {
+      "epoch": 0.74163381314752,
+      "learning_rate": 5.167507105528559e-05,
+      "loss": 0.6663,
+      "mean_token_accuracy": 0.7942912399768829,
+      "num_tokens": 191327867.0,
+      "step": 80890
+    },
+    {
+      "epoch": 0.7417254973869992,
+      "learning_rate": 5.165673420738976e-05,
+      "loss": 0.6645,
+      "mean_token_accuracy": 0.7928799092769623,
+      "num_tokens": 191352428.0,
+      "step": 80900
+    },
+    {
+      "epoch": 0.7418171816264784,
+      "learning_rate": 5.1638397359493904e-05,
+      "loss": 0.7115,
+      "mean_token_accuracy": 0.7827491044998169,
+      "num_tokens": 191378634.0,
+      "step": 80910
+    },
+    {
+      "epoch": 0.7419088658659576,
+      "learning_rate": 5.1620060511598055e-05,
+      "loss": 0.718,
+      "mean_token_accuracy": 0.7823315680027008,
+      "num_tokens": 191403429.0,
+      "step": 80920
+    },
+    {
+      "epoch": 0.7420005501054369,
+      "learning_rate": 5.1601723663702214e-05,
+      "loss": 0.6668,
+      "mean_token_accuracy": 0.7925114870071411,
+      "num_tokens": 191428721.0,
+      "step": 80930
+    },
+    {
+      "epoch": 0.7420922343449161,
+      "learning_rate": 5.1583386815806366e-05,
+      "loss": 0.6859,
+      "mean_token_accuracy": 0.78622407913208,
+      "num_tokens": 191453870.0,
+      "step": 80940
+    },
+    {
+      "epoch": 0.7421839185843954,
+      "learning_rate": 5.156504996791052e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.7923462331295014,
+      "num_tokens": 191479112.0,
+      "step": 80950
+    },
+    {
+      "epoch": 0.7422756028238746,
+      "learning_rate": 5.154671312001467e-05,
+      "loss": 0.7193,
+      "mean_token_accuracy": 0.7837464988231659,
+      "num_tokens": 191505489.0,
+      "step": 80960
+    },
+    {
+      "epoch": 0.7423672870633539,
+      "learning_rate": 5.152837627211883e-05,
+      "loss": 0.7645,
+      "mean_token_accuracy": 0.7756374478340149,
+      "num_tokens": 191531165.0,
+      "step": 80970
+    },
+    {
+      "epoch": 0.742458971302833,
+      "learning_rate": 5.151003942422298e-05,
+      "loss": 0.6538,
+      "mean_token_accuracy": 0.800909286737442,
+      "num_tokens": 191556349.0,
+      "step": 80980
+    },
+    {
+      "epoch": 0.7425506555423123,
+      "learning_rate": 5.1491702576327125e-05,
+      "loss": 0.7393,
+      "mean_token_accuracy": 0.7746599197387696,
+      "num_tokens": 191582371.0,
+      "step": 80990
+    },
+    {
+      "epoch": 0.7426423397817915,
+      "learning_rate": 5.147336572843129e-05,
+      "loss": 0.738,
+      "mean_token_accuracy": 0.7749620497226715,
+      "num_tokens": 191608035.0,
+      "step": 81000
+    },
+    {
+      "epoch": 0.7427340240212708,
+      "learning_rate": 5.1455028880535436e-05,
+      "loss": 0.699,
+      "mean_token_accuracy": 0.7854167759418488,
+      "num_tokens": 191632882.0,
+      "step": 81010
+    },
+    {
+      "epoch": 0.74282570826075,
+      "learning_rate": 5.143669203263959e-05,
+      "loss": 0.6982,
+      "mean_token_accuracy": 0.7897755622863769,
+      "num_tokens": 191657882.0,
+      "step": 81020
+    },
+    {
+      "epoch": 0.7429173925002293,
+      "learning_rate": 5.1418355184743746e-05,
+      "loss": 0.7065,
+      "mean_token_accuracy": 0.7861283838748931,
+      "num_tokens": 191683462.0,
+      "step": 81030
+    },
+    {
+      "epoch": 0.7430090767397084,
+      "learning_rate": 5.14000183368479e-05,
+      "loss": 0.732,
+      "mean_token_accuracy": 0.7773496985435486,
+      "num_tokens": 191709163.0,
+      "step": 81040
+    },
+    {
+      "epoch": 0.7431007609791876,
+      "learning_rate": 5.138168148895205e-05,
+      "loss": 0.711,
+      "mean_token_accuracy": 0.7879119396209717,
+      "num_tokens": 191734696.0,
+      "step": 81050
+    },
+    {
+      "epoch": 0.7431924452186669,
+      "learning_rate": 5.136334464105621e-05,
+      "loss": 0.715,
+      "mean_token_accuracy": 0.7803106546401978,
+      "num_tokens": 191760517.0,
+      "step": 81060
+    },
+    {
+      "epoch": 0.7432841294581461,
+      "learning_rate": 5.134500779316036e-05,
+      "loss": 0.7087,
+      "mean_token_accuracy": 0.781575471162796,
+      "num_tokens": 191786137.0,
+      "step": 81070
+    },
+    {
+      "epoch": 0.7433758136976254,
+      "learning_rate": 5.132667094526451e-05,
+      "loss": 0.6951,
+      "mean_token_accuracy": 0.7844762027263641,
+      "num_tokens": 191812243.0,
+      "step": 81080
+    },
+    {
+      "epoch": 0.7434674979371046,
+      "learning_rate": 5.130833409736866e-05,
+      "loss": 0.7159,
+      "mean_token_accuracy": 0.7803437769412994,
+      "num_tokens": 191837797.0,
+      "step": 81090
+    },
+    {
+      "epoch": 0.7435591821765839,
+      "learning_rate": 5.128999724947282e-05,
+      "loss": 0.7113,
+      "mean_token_accuracy": 0.7871942818164825,
+      "num_tokens": 191862973.0,
+      "step": 81100
+    },
+    {
+      "epoch": 0.743650866416063,
+      "learning_rate": 5.127166040157697e-05,
+      "loss": 0.6935,
+      "mean_token_accuracy": 0.784513133764267,
+      "num_tokens": 191888434.0,
+      "step": 81110
+    },
+    {
+      "epoch": 0.7437425506555423,
+      "learning_rate": 5.125332355368112e-05,
+      "loss": 0.6924,
+      "mean_token_accuracy": 0.7868725597858429,
+      "num_tokens": 191912876.0,
+      "step": 81120
+    },
+    {
+      "epoch": 0.7438342348950215,
+      "learning_rate": 5.123498670578528e-05,
+      "loss": 0.6631,
+      "mean_token_accuracy": 0.7957233726978302,
+      "num_tokens": 191938253.0,
+      "step": 81130
+    },
+    {
+      "epoch": 0.7439259191345008,
+      "learning_rate": 5.121664985788943e-05,
+      "loss": 0.709,
+      "mean_token_accuracy": 0.787634426355362,
+      "num_tokens": 191963475.0,
+      "step": 81140
+    },
+    {
+      "epoch": 0.74401760337398,
+      "learning_rate": 5.119831300999358e-05,
+      "loss": 0.7082,
+      "mean_token_accuracy": 0.7793751299381256,
+      "num_tokens": 191989423.0,
+      "step": 81150
+    },
+    {
+      "epoch": 0.7441092876134593,
+      "learning_rate": 5.117997616209774e-05,
+      "loss": 0.6721,
+      "mean_token_accuracy": 0.7872091174125672,
+      "num_tokens": 192015225.0,
+      "step": 81160
+    },
+    {
+      "epoch": 0.7442009718529384,
+      "learning_rate": 5.116163931420189e-05,
+      "loss": 0.6879,
+      "mean_token_accuracy": 0.7934359252452851,
+      "num_tokens": 192040218.0,
+      "step": 81170
+    },
+    {
+      "epoch": 0.7442926560924177,
+      "learning_rate": 5.1143302466306045e-05,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7663623631000519,
+      "num_tokens": 192066391.0,
+      "step": 81180
+    },
+    {
+      "epoch": 0.7443843403318969,
+      "learning_rate": 5.11249656184102e-05,
+      "loss": 0.7058,
+      "mean_token_accuracy": 0.7837217688560486,
+      "num_tokens": 192091329.0,
+      "step": 81190
+    },
+    {
+      "epoch": 0.7444760245713762,
+      "learning_rate": 5.1106628770514355e-05,
+      "loss": 0.6478,
+      "mean_token_accuracy": 0.7939362585544586,
+      "num_tokens": 192116864.0,
+      "step": 81200
+    },
+    {
+      "epoch": 0.7445677088108554,
+      "learning_rate": 5.10882919226185e-05,
+      "loss": 0.7065,
+      "mean_token_accuracy": 0.7797066390514373,
+      "num_tokens": 192142193.0,
+      "step": 81210
+    },
+    {
+      "epoch": 0.7446593930503347,
+      "learning_rate": 5.106995507472265e-05,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.7788159012794494,
+      "num_tokens": 192166842.0,
+      "step": 81220
+    },
+    {
+      "epoch": 0.7447510772898139,
+      "learning_rate": 5.105161822682681e-05,
+      "loss": 0.7078,
+      "mean_token_accuracy": 0.7824737727642059,
+      "num_tokens": 192192793.0,
+      "step": 81230
+    },
+    {
+      "epoch": 0.7448427615292931,
+      "learning_rate": 5.103328137893096e-05,
+      "loss": 0.7076,
+      "mean_token_accuracy": 0.7827406227588654,
+      "num_tokens": 192217991.0,
+      "step": 81240
+    },
+    {
+      "epoch": 0.7449344457687723,
+      "learning_rate": 5.1014944531035115e-05,
+      "loss": 0.6649,
+      "mean_token_accuracy": 0.7950560629367829,
+      "num_tokens": 192243524.0,
+      "step": 81250
+    },
+    {
+      "epoch": 0.7450261300082516,
+      "learning_rate": 5.099660768313927e-05,
+      "loss": 0.6779,
+      "mean_token_accuracy": 0.7937249064445495,
+      "num_tokens": 192269617.0,
+      "step": 81260
+    },
+    {
+      "epoch": 0.7451178142477308,
+      "learning_rate": 5.0978270835243425e-05,
+      "loss": 0.6857,
+      "mean_token_accuracy": 0.7913187861442565,
+      "num_tokens": 192294719.0,
+      "step": 81270
+    },
+    {
+      "epoch": 0.7452094984872101,
+      "learning_rate": 5.095993398734758e-05,
+      "loss": 0.74,
+      "mean_token_accuracy": 0.7766092956066132,
+      "num_tokens": 192320843.0,
+      "step": 81280
+    },
+    {
+      "epoch": 0.7453011827266893,
+      "learning_rate": 5.0941597139451736e-05,
+      "loss": 0.687,
+      "mean_token_accuracy": 0.7891350746154785,
+      "num_tokens": 192346091.0,
+      "step": 81290
+    },
+    {
+      "epoch": 0.7453928669661685,
+      "learning_rate": 5.092326029155589e-05,
+      "loss": 0.7044,
+      "mean_token_accuracy": 0.7799795985221862,
+      "num_tokens": 192371537.0,
+      "step": 81300
+    },
+    {
+      "epoch": 0.7454845512056477,
+      "learning_rate": 5.090492344366003e-05,
+      "loss": 0.6818,
+      "mean_token_accuracy": 0.7914417624473572,
+      "num_tokens": 192397037.0,
+      "step": 81310
+    },
+    {
+      "epoch": 0.745576235445127,
+      "learning_rate": 5.0886586595764184e-05,
+      "loss": 0.7279,
+      "mean_token_accuracy": 0.7714911818504333,
+      "num_tokens": 192423175.0,
+      "step": 81320
+    },
+    {
+      "epoch": 0.7456679196846062,
+      "learning_rate": 5.086824974786834e-05,
+      "loss": 0.7194,
+      "mean_token_accuracy": 0.7792998969554901,
+      "num_tokens": 192448252.0,
+      "step": 81330
+    },
+    {
+      "epoch": 0.7457596039240855,
+      "learning_rate": 5.0849912899972495e-05,
+      "loss": 0.7337,
+      "mean_token_accuracy": 0.7705634891986847,
+      "num_tokens": 192474216.0,
+      "step": 81340
+    },
+    {
+      "epoch": 0.7458512881635647,
+      "learning_rate": 5.083157605207665e-05,
+      "loss": 0.7424,
+      "mean_token_accuracy": 0.7710492253303528,
+      "num_tokens": 192500074.0,
+      "step": 81350
+    },
+    {
+      "epoch": 0.745942972403044,
+      "learning_rate": 5.0813239204180805e-05,
+      "loss": 0.6846,
+      "mean_token_accuracy": 0.7872656166553498,
+      "num_tokens": 192524546.0,
+      "step": 81360
+    },
+    {
+      "epoch": 0.7460346566425231,
+      "learning_rate": 5.079490235628496e-05,
+      "loss": 0.6398,
+      "mean_token_accuracy": 0.7989014267921448,
+      "num_tokens": 192548585.0,
+      "step": 81370
+    },
+    {
+      "epoch": 0.7461263408820024,
+      "learning_rate": 5.077656550838911e-05,
+      "loss": 0.6677,
+      "mean_token_accuracy": 0.7930905401706696,
+      "num_tokens": 192573340.0,
+      "step": 81380
+    },
+    {
+      "epoch": 0.7462180251214816,
+      "learning_rate": 5.075822866049327e-05,
+      "loss": 0.7075,
+      "mean_token_accuracy": 0.7884686052799225,
+      "num_tokens": 192598768.0,
+      "step": 81390
+    },
+    {
+      "epoch": 0.7463097093609609,
+      "learning_rate": 5.073989181259742e-05,
+      "loss": 0.6599,
+      "mean_token_accuracy": 0.796476936340332,
+      "num_tokens": 192624715.0,
+      "step": 81400
+    },
+    {
+      "epoch": 0.7464013936004401,
+      "learning_rate": 5.0721554964701565e-05,
+      "loss": 0.7241,
+      "mean_token_accuracy": 0.7784188091754913,
+      "num_tokens": 192649733.0,
+      "step": 81410
+    },
+    {
+      "epoch": 0.7464930778399194,
+      "learning_rate": 5.070321811680573e-05,
+      "loss": 0.6986,
+      "mean_token_accuracy": 0.7878309786319733,
+      "num_tokens": 192675901.0,
+      "step": 81420
+    },
+    {
+      "epoch": 0.7465847620793985,
+      "learning_rate": 5.0684881268909875e-05,
+      "loss": 0.6701,
+      "mean_token_accuracy": 0.7902319371700287,
+      "num_tokens": 192700966.0,
+      "step": 81430
+    },
+    {
+      "epoch": 0.7466764463188778,
+      "learning_rate": 5.066654442101403e-05,
+      "loss": 0.7343,
+      "mean_token_accuracy": 0.7827136933803558,
+      "num_tokens": 192725850.0,
+      "step": 81440
+    },
+    {
+      "epoch": 0.746768130558357,
+      "learning_rate": 5.064820757311818e-05,
+      "loss": 0.6962,
+      "mean_token_accuracy": 0.7829465329647064,
+      "num_tokens": 192751309.0,
+      "step": 81450
+    },
+    {
+      "epoch": 0.7468598147978363,
+      "learning_rate": 5.062987072522234e-05,
+      "loss": 0.7434,
+      "mean_token_accuracy": 0.7767478287220001,
+      "num_tokens": 192776640.0,
+      "step": 81460
+    },
+    {
+      "epoch": 0.7469514990373155,
+      "learning_rate": 5.061153387732649e-05,
+      "loss": 0.7085,
+      "mean_token_accuracy": 0.7855776369571685,
+      "num_tokens": 192802231.0,
+      "step": 81470
+    },
+    {
+      "epoch": 0.7470431832767948,
+      "learning_rate": 5.059319702943064e-05,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.764584481716156,
+      "num_tokens": 192828243.0,
+      "step": 81480
+    },
+    {
+      "epoch": 0.747134867516274,
+      "learning_rate": 5.05748601815348e-05,
+      "loss": 0.6605,
+      "mean_token_accuracy": 0.7890223801136017,
+      "num_tokens": 192853669.0,
+      "step": 81490
+    },
+    {
+      "epoch": 0.7472265517557531,
+      "learning_rate": 5.055652333363895e-05,
+      "loss": 0.7202,
+      "mean_token_accuracy": 0.7801950812339783,
+      "num_tokens": 192879309.0,
+      "step": 81500
+    },
+    {
+      "epoch": 0.7473182359952324,
+      "learning_rate": 5.05381864857431e-05,
+      "loss": 0.7504,
+      "mean_token_accuracy": 0.7756495952606202,
+      "num_tokens": 192905080.0,
+      "step": 81510
+    },
+    {
+      "epoch": 0.7474099202347116,
+      "learning_rate": 5.051984963784726e-05,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.7540276765823364,
+      "num_tokens": 192931322.0,
+      "step": 81520
+    },
+    {
+      "epoch": 0.7475016044741909,
+      "learning_rate": 5.050151278995141e-05,
+      "loss": 0.7431,
+      "mean_token_accuracy": 0.7807523667812347,
+      "num_tokens": 192956899.0,
+      "step": 81530
+    },
+    {
+      "epoch": 0.7475932887136701,
+      "learning_rate": 5.048317594205556e-05,
+      "loss": 0.7985,
+      "mean_token_accuracy": 0.7603351235389709,
+      "num_tokens": 192983812.0,
+      "step": 81540
+    },
+    {
+      "epoch": 0.7476849729531494,
+      "learning_rate": 5.0464839094159725e-05,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.7857788622379303,
+      "num_tokens": 193008604.0,
+      "step": 81550
+    },
+    {
+      "epoch": 0.7477766571926285,
+      "learning_rate": 5.044650224626387e-05,
+      "loss": 0.7176,
+      "mean_token_accuracy": 0.7779941558837891,
+      "num_tokens": 193034820.0,
+      "step": 81560
+    },
+    {
+      "epoch": 0.7478683414321078,
+      "learning_rate": 5.042816539836802e-05,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.772507619857788,
+      "num_tokens": 193061170.0,
+      "step": 81570
+    },
+    {
+      "epoch": 0.747960025671587,
+      "learning_rate": 5.0409828550472174e-05,
+      "loss": 0.6938,
+      "mean_token_accuracy": 0.785056871175766,
+      "num_tokens": 193086556.0,
+      "step": 81580
+    },
+    {
+      "epoch": 0.7480517099110663,
+      "learning_rate": 5.039149170257633e-05,
+      "loss": 0.8707,
+      "mean_token_accuracy": 0.7458867251873016,
+      "num_tokens": 193113779.0,
+      "step": 81590
+    },
+    {
+      "epoch": 0.7481433941505455,
+      "learning_rate": 5.0373154854680484e-05,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.7856736123561859,
+      "num_tokens": 193140171.0,
+      "step": 81600
+    },
+    {
+      "epoch": 0.7482350783900248,
+      "learning_rate": 5.035481800678463e-05,
+      "loss": 0.6818,
+      "mean_token_accuracy": 0.7914603292942047,
+      "num_tokens": 193165390.0,
+      "step": 81610
+    },
+    {
+      "epoch": 0.748326762629504,
+      "learning_rate": 5.0336481158888795e-05,
+      "loss": 0.6895,
+      "mean_token_accuracy": 0.7843019604682923,
+      "num_tokens": 193192098.0,
+      "step": 81620
+    },
+    {
+      "epoch": 0.7484184468689832,
+      "learning_rate": 5.031814431099294e-05,
+      "loss": 0.7608,
+      "mean_token_accuracy": 0.7733224153518676,
+      "num_tokens": 193217305.0,
+      "step": 81630
+    },
+    {
+      "epoch": 0.7485101311084624,
+      "learning_rate": 5.029980746309709e-05,
+      "loss": 0.8569,
+      "mean_token_accuracy": 0.74853515625,
+      "num_tokens": 193245930.0,
+      "step": 81640
+    },
+    {
+      "epoch": 0.7486018153479417,
+      "learning_rate": 5.028147061520126e-05,
+      "loss": 0.7385,
+      "mean_token_accuracy": 0.7779390633106231,
+      "num_tokens": 193272335.0,
+      "step": 81650
+    },
+    {
+      "epoch": 0.7486934995874209,
+      "learning_rate": 5.02631337673054e-05,
+      "loss": 0.7516,
+      "mean_token_accuracy": 0.7716957569122315,
+      "num_tokens": 193298482.0,
+      "step": 81660
+    },
+    {
+      "epoch": 0.7487851838269002,
+      "learning_rate": 5.0244796919409554e-05,
+      "loss": 0.6871,
+      "mean_token_accuracy": 0.7878741502761841,
+      "num_tokens": 193323393.0,
+      "step": 81670
+    },
+    {
+      "epoch": 0.7488768680663794,
+      "learning_rate": 5.022646007151371e-05,
+      "loss": 0.7245,
+      "mean_token_accuracy": 0.7755094230175018,
+      "num_tokens": 193348487.0,
+      "step": 81680
+    },
+    {
+      "epoch": 0.7489685523058586,
+      "learning_rate": 5.0208123223617864e-05,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7713774621486664,
+      "num_tokens": 193373605.0,
+      "step": 81690
+    },
+    {
+      "epoch": 0.7490602365453378,
+      "learning_rate": 5.0189786375722016e-05,
+      "loss": 0.8449,
+      "mean_token_accuracy": 0.750771677494049,
+      "num_tokens": 193402804.0,
+      "step": 81700
+    },
+    {
+      "epoch": 0.7491519207848171,
+      "learning_rate": 5.017144952782616e-05,
+      "loss": 0.7251,
+      "mean_token_accuracy": 0.7795206308364868,
+      "num_tokens": 193428268.0,
+      "step": 81710
+    },
+    {
+      "epoch": 0.7492436050242963,
+      "learning_rate": 5.015311267993033e-05,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.7697794437408447,
+      "num_tokens": 193454181.0,
+      "step": 81720
+    },
+    {
+      "epoch": 0.7493352892637756,
+      "learning_rate": 5.013477583203447e-05,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7685450136661529,
+      "num_tokens": 193480553.0,
+      "step": 81730
+    },
+    {
+      "epoch": 0.7494269735032548,
+      "learning_rate": 5.0116438984138624e-05,
+      "loss": 0.7514,
+      "mean_token_accuracy": 0.7715674877166748,
+      "num_tokens": 193507507.0,
+      "step": 81740
+    },
+    {
+      "epoch": 0.7495186577427341,
+      "learning_rate": 5.009810213624279e-05,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7732964396476746,
+      "num_tokens": 193534341.0,
+      "step": 81750
+    },
+    {
+      "epoch": 0.7496103419822132,
+      "learning_rate": 5.0079765288346934e-05,
+      "loss": 0.7842,
+      "mean_token_accuracy": 0.7643706262111664,
+      "num_tokens": 193561440.0,
+      "step": 81760
+    },
+    {
+      "epoch": 0.7497020262216925,
+      "learning_rate": 5.0061428440451086e-05,
+      "loss": 0.8439,
+      "mean_token_accuracy": 0.7495325326919555,
+      "num_tokens": 193590705.0,
+      "step": 81770
+    },
+    {
+      "epoch": 0.7497937104611717,
+      "learning_rate": 5.0043091592555245e-05,
+      "loss": 0.6869,
+      "mean_token_accuracy": 0.7873552143573761,
+      "num_tokens": 193616380.0,
+      "step": 81780
+    },
+    {
+      "epoch": 0.749885394700651,
+      "learning_rate": 5.00247547446594e-05,
+      "loss": 0.7524,
+      "mean_token_accuracy": 0.7729937136173248,
+      "num_tokens": 193640836.0,
+      "step": 81790
+    },
+    {
+      "epoch": 0.7499770789401302,
+      "learning_rate": 5.000641789676355e-05,
+      "loss": 0.793,
+      "mean_token_accuracy": 0.7657868981361389,
+      "num_tokens": 193667721.0,
+      "step": 81800
+    },
+    {
+      "epoch": 0.7500687631796095,
+      "learning_rate": 4.99880810488677e-05,
+      "loss": 0.806,
+      "mean_token_accuracy": 0.7578528225421906,
+      "num_tokens": 193694718.0,
+      "step": 81810
+    },
+    {
+      "epoch": 0.7501604474190886,
+      "learning_rate": 4.996974420097186e-05,
+      "loss": 0.7636,
+      "mean_token_accuracy": 0.7735159158706665,
+      "num_tokens": 193720375.0,
+      "step": 81820
+    },
+    {
+      "epoch": 0.7502521316585679,
+      "learning_rate": 4.9951407353076004e-05,
+      "loss": 0.6742,
+      "mean_token_accuracy": 0.7935140013694764,
+      "num_tokens": 193746126.0,
+      "step": 81830
+    },
+    {
+      "epoch": 0.7503438158980471,
+      "learning_rate": 4.993307050518016e-05,
+      "loss": 0.8474,
+      "mean_token_accuracy": 0.7509146213531495,
+      "num_tokens": 193774702.0,
+      "step": 81840
+    },
+    {
+      "epoch": 0.7504355001375264,
+      "learning_rate": 4.9914733657284315e-05,
+      "loss": 0.7746,
+      "mean_token_accuracy": 0.7741905748844147,
+      "num_tokens": 193800746.0,
+      "step": 81850
+    },
+    {
+      "epoch": 0.7505271843770056,
+      "learning_rate": 4.9896396809388467e-05,
+      "loss": 0.7429,
+      "mean_token_accuracy": 0.7753175258636474,
+      "num_tokens": 193827327.0,
+      "step": 81860
+    },
+    {
+      "epoch": 0.7506188686164849,
+      "learning_rate": 4.9878059961492625e-05,
+      "loss": 0.7318,
+      "mean_token_accuracy": 0.7754132926464081,
+      "num_tokens": 193853189.0,
+      "step": 81870
+    },
+    {
+      "epoch": 0.7507105528559641,
+      "learning_rate": 4.985972311359677e-05,
+      "loss": 0.6884,
+      "mean_token_accuracy": 0.7916352927684784,
+      "num_tokens": 193879486.0,
+      "step": 81880
+    },
+    {
+      "epoch": 0.7508022370954432,
+      "learning_rate": 4.984138626570093e-05,
+      "loss": 0.8695,
+      "mean_token_accuracy": 0.7477820873260498,
+      "num_tokens": 193908094.0,
+      "step": 81890
+    },
+    {
+      "epoch": 0.7508939213349225,
+      "learning_rate": 4.982304941780509e-05,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.749259614944458,
+      "num_tokens": 193936986.0,
+      "step": 81900
+    },
+    {
+      "epoch": 0.7509856055744017,
+      "learning_rate": 4.980471256990923e-05,
+      "loss": 0.7081,
+      "mean_token_accuracy": 0.7845314800739288,
+      "num_tokens": 193963648.0,
+      "step": 81910
+    },
+    {
+      "epoch": 0.751077289813881,
+      "learning_rate": 4.978637572201339e-05,
+      "loss": 0.7839,
+      "mean_token_accuracy": 0.7613317847251893,
+      "num_tokens": 193989812.0,
+      "step": 81920
+    },
+    {
+      "epoch": 0.7511689740533602,
+      "learning_rate": 4.9768038874117536e-05,
+      "loss": 0.7988,
+      "mean_token_accuracy": 0.7601967811584472,
+      "num_tokens": 194017550.0,
+      "step": 81930
+    },
+    {
+      "epoch": 0.7512606582928395,
+      "learning_rate": 4.9749702026221695e-05,
+      "loss": 0.7725,
+      "mean_token_accuracy": 0.7678104877471924,
+      "num_tokens": 194044106.0,
+      "step": 81940
+    },
+    {
+      "epoch": 0.7513523425323186,
+      "learning_rate": 4.9731365178325854e-05,
+      "loss": 0.7516,
+      "mean_token_accuracy": 0.7737136840820312,
+      "num_tokens": 194070051.0,
+      "step": 81950
+    },
+    {
+      "epoch": 0.7514440267717979,
+      "learning_rate": 4.971302833043e-05,
+      "loss": 0.8265,
+      "mean_token_accuracy": 0.7505439996719361,
+      "num_tokens": 194098289.0,
+      "step": 81960
+    },
+    {
+      "epoch": 0.7515357110112771,
+      "learning_rate": 4.969469148253416e-05,
+      "loss": 0.7109,
+      "mean_token_accuracy": 0.7753967463970184,
+      "num_tokens": 194124068.0,
+      "step": 81970
+    },
+    {
+      "epoch": 0.7516273952507564,
+      "learning_rate": 4.96763546346383e-05,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.769416093826294,
+      "num_tokens": 194150676.0,
+      "step": 81980
+    },
+    {
+      "epoch": 0.7517190794902356,
+      "learning_rate": 4.965801778674246e-05,
+      "loss": 0.8358,
+      "mean_token_accuracy": 0.747289651632309,
+      "num_tokens": 194179280.0,
+      "step": 81990
+    },
+    {
+      "epoch": 0.7518107637297149,
+      "learning_rate": 4.963968093884662e-05,
+      "loss": 0.8271,
+      "mean_token_accuracy": 0.7535883486270905,
+      "num_tokens": 194207018.0,
+      "step": 82000
+    },
+    {
+      "epoch": 0.7519024479691941,
+      "learning_rate": 4.9621344090950765e-05,
+      "loss": 0.9148,
+      "mean_token_accuracy": 0.7365788459777832,
+      "num_tokens": 194236736.0,
+      "step": 82010
+    },
+    {
+      "epoch": 0.7519941322086733,
+      "learning_rate": 4.9603007243054924e-05,
+      "loss": 0.8665,
+      "mean_token_accuracy": 0.7488987386226654,
+      "num_tokens": 194264334.0,
+      "step": 82020
+    },
+    {
+      "epoch": 0.7520858164481525,
+      "learning_rate": 4.958467039515907e-05,
+      "loss": 0.7868,
+      "mean_token_accuracy": 0.7618037223815918,
+      "num_tokens": 194291772.0,
+      "step": 82030
+    },
+    {
+      "epoch": 0.7521775006876318,
+      "learning_rate": 4.956633354726323e-05,
+      "loss": 0.7589,
+      "mean_token_accuracy": 0.7747104287147522,
+      "num_tokens": 194318381.0,
+      "step": 82040
+    },
+    {
+      "epoch": 0.752269184927111,
+      "learning_rate": 4.9547996699367386e-05,
+      "loss": 0.7958,
+      "mean_token_accuracy": 0.760222977399826,
+      "num_tokens": 194346059.0,
+      "step": 82050
+    },
+    {
+      "epoch": 0.7523608691665903,
+      "learning_rate": 4.952965985147153e-05,
+      "loss": 0.8587,
+      "mean_token_accuracy": 0.7498927354812622,
+      "num_tokens": 194373753.0,
+      "step": 82060
+    },
+    {
+      "epoch": 0.7524525534060695,
+      "learning_rate": 4.951132300357569e-05,
+      "loss": 0.6909,
+      "mean_token_accuracy": 0.7929769337177277,
+      "num_tokens": 194399024.0,
+      "step": 82070
+    },
+    {
+      "epoch": 0.7525442376455487,
+      "learning_rate": 4.949298615567984e-05,
+      "loss": 0.7802,
+      "mean_token_accuracy": 0.7730043709278107,
+      "num_tokens": 194425365.0,
+      "step": 82080
+    },
+    {
+      "epoch": 0.7526359218850279,
+      "learning_rate": 4.947464930778399e-05,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7677356362342834,
+      "num_tokens": 194451118.0,
+      "step": 82090
+    },
+    {
+      "epoch": 0.7527276061245072,
+      "learning_rate": 4.945631245988815e-05,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.7705836236476898,
+      "num_tokens": 194477681.0,
+      "step": 82100
+    },
+    {
+      "epoch": 0.7528192903639864,
+      "learning_rate": 4.94379756119923e-05,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7777624130249023,
+      "num_tokens": 194504189.0,
+      "step": 82110
+    },
+    {
+      "epoch": 0.7529109746034657,
+      "learning_rate": 4.9419638764096456e-05,
+      "loss": 0.884,
+      "mean_token_accuracy": 0.7406049251556397,
+      "num_tokens": 194532981.0,
+      "step": 82120
+    },
+    {
+      "epoch": 0.7530026588429449,
+      "learning_rate": 4.940130191620061e-05,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.7547918617725372,
+      "num_tokens": 194560608.0,
+      "step": 82130
+    },
+    {
+      "epoch": 0.7530943430824242,
+      "learning_rate": 4.938296506830476e-05,
+      "loss": 0.8782,
+      "mean_token_accuracy": 0.7428805947303772,
+      "num_tokens": 194588926.0,
+      "step": 82140
+    },
+    {
+      "epoch": 0.7531860273219033,
+      "learning_rate": 4.936462822040892e-05,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7802633821964264,
+      "num_tokens": 194613475.0,
+      "step": 82150
+    },
+    {
+      "epoch": 0.7532777115613826,
+      "learning_rate": 4.934629137251306e-05,
+      "loss": 0.7837,
+      "mean_token_accuracy": 0.7672086238861084,
+      "num_tokens": 194640374.0,
+      "step": 82160
+    },
+    {
+      "epoch": 0.7533693958008618,
+      "learning_rate": 4.932795452461722e-05,
+      "loss": 0.8208,
+      "mean_token_accuracy": 0.7582300066947937,
+      "num_tokens": 194668213.0,
+      "step": 82170
+    },
+    {
+      "epoch": 0.7534610800403411,
+      "learning_rate": 4.9309617676721374e-05,
+      "loss": 0.937,
+      "mean_token_accuracy": 0.7293002188205719,
+      "num_tokens": 194698994.0,
+      "step": 82180
+    },
+    {
+      "epoch": 0.7535527642798203,
+      "learning_rate": 4.9291280828825526e-05,
+      "loss": 0.8646,
+      "mean_token_accuracy": 0.7472356975078582,
+      "num_tokens": 194727596.0,
+      "step": 82190
+    },
+    {
+      "epoch": 0.7536444485192996,
+      "learning_rate": 4.9272943980929684e-05,
+      "loss": 0.8644,
+      "mean_token_accuracy": 0.7457275450229645,
+      "num_tokens": 194755504.0,
+      "step": 82200
+    },
+    {
+      "epoch": 0.7537361327587788,
+      "learning_rate": 4.9254607133033836e-05,
+      "loss": 0.7907,
+      "mean_token_accuracy": 0.7620010018348694,
+      "num_tokens": 194781799.0,
+      "step": 82210
+    },
+    {
+      "epoch": 0.753827816998258,
+      "learning_rate": 4.923627028513799e-05,
+      "loss": 0.8128,
+      "mean_token_accuracy": 0.7590906023979187,
+      "num_tokens": 194808353.0,
+      "step": 82220
+    },
+    {
+      "epoch": 0.7539195012377372,
+      "learning_rate": 4.921793343724214e-05,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7560511231422424,
+      "num_tokens": 194836071.0,
+      "step": 82230
+    },
+    {
+      "epoch": 0.7540111854772165,
+      "learning_rate": 4.919959658934629e-05,
+      "loss": 0.831,
+      "mean_token_accuracy": 0.7527174770832061,
+      "num_tokens": 194864236.0,
+      "step": 82240
+    },
+    {
+      "epoch": 0.7541028697166957,
+      "learning_rate": 4.918125974145045e-05,
+      "loss": 0.8575,
+      "mean_token_accuracy": 0.7455669045448303,
+      "num_tokens": 194893911.0,
+      "step": 82250
+    },
+    {
+      "epoch": 0.754194553956175,
+      "learning_rate": 4.91629228935546e-05,
+      "loss": 0.8665,
+      "mean_token_accuracy": 0.7482231497764588,
+      "num_tokens": 194922090.0,
+      "step": 82260
+    },
+    {
+      "epoch": 0.7542862381956542,
+      "learning_rate": 4.9144586045658754e-05,
+      "loss": 0.7877,
+      "mean_token_accuracy": 0.7638835310935974,
+      "num_tokens": 194950326.0,
+      "step": 82270
+    },
+    {
+      "epoch": 0.7543779224351334,
+      "learning_rate": 4.9126249197762906e-05,
+      "loss": 0.7719,
+      "mean_token_accuracy": 0.7675490319728852,
+      "num_tokens": 194978589.0,
+      "step": 82280
+    },
+    {
+      "epoch": 0.7544696066746126,
+      "learning_rate": 4.910791234986706e-05,
+      "loss": 0.772,
+      "mean_token_accuracy": 0.773044204711914,
+      "num_tokens": 195005342.0,
+      "step": 82290
+    },
+    {
+      "epoch": 0.7545612909140919,
+      "learning_rate": 4.9089575501971216e-05,
+      "loss": 0.8271,
+      "mean_token_accuracy": 0.7553254723548889,
+      "num_tokens": 195033339.0,
+      "step": 82300
+    },
+    {
+      "epoch": 0.7546529751535711,
+      "learning_rate": 4.907123865407537e-05,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.7502288699150086,
+      "num_tokens": 195062836.0,
+      "step": 82310
+    },
+    {
+      "epoch": 0.7547446593930504,
+      "learning_rate": 4.905290180617952e-05,
+      "loss": 0.7888,
+      "mean_token_accuracy": 0.7607177436351776,
+      "num_tokens": 195089469.0,
+      "step": 82320
+    },
+    {
+      "epoch": 0.7548363436325296,
+      "learning_rate": 4.903456495828367e-05,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7668349742889404,
+      "num_tokens": 195115880.0,
+      "step": 82330
+    },
+    {
+      "epoch": 0.7549280278720089,
+      "learning_rate": 4.9016228110387824e-05,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.768760347366333,
+      "num_tokens": 195142804.0,
+      "step": 82340
+    },
+    {
+      "epoch": 0.755019712111488,
+      "learning_rate": 4.899789126249198e-05,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.7467792809009552,
+      "num_tokens": 195170419.0,
+      "step": 82350
+    },
+    {
+      "epoch": 0.7551113963509672,
+      "learning_rate": 4.8979554414596134e-05,
+      "loss": 0.7737,
+      "mean_token_accuracy": 0.7684077203273774,
+      "num_tokens": 195197127.0,
+      "step": 82360
+    },
+    {
+      "epoch": 0.7552030805904465,
+      "learning_rate": 4.8961217566700286e-05,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7499734938144684,
+      "num_tokens": 195225039.0,
+      "step": 82370
+    },
+    {
+      "epoch": 0.7552947648299257,
+      "learning_rate": 4.894288071880444e-05,
+      "loss": 0.8044,
+      "mean_token_accuracy": 0.7629131078720093,
+      "num_tokens": 195252697.0,
+      "step": 82380
+    },
+    {
+      "epoch": 0.755386449069405,
+      "learning_rate": 4.89245438709086e-05,
+      "loss": 0.8057,
+      "mean_token_accuracy": 0.7554194688796997,
+      "num_tokens": 195280706.0,
+      "step": 82390
+    },
+    {
+      "epoch": 0.7554781333088842,
+      "learning_rate": 4.890620702301275e-05,
+      "loss": 0.8122,
+      "mean_token_accuracy": 0.7619563937187195,
+      "num_tokens": 195309976.0,
+      "step": 82400
+    },
+    {
+      "epoch": 0.7555698175483634,
+      "learning_rate": 4.88878701751169e-05,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7478345215320588,
+      "num_tokens": 195340195.0,
+      "step": 82410
+    },
+    {
+      "epoch": 0.7556615017878426,
+      "learning_rate": 4.886953332722105e-05,
+      "loss": 0.8824,
+      "mean_token_accuracy": 0.7405252516269684,
+      "num_tokens": 195369556.0,
+      "step": 82420
+    },
+    {
+      "epoch": 0.7557531860273219,
+      "learning_rate": 4.8851196479325204e-05,
+      "loss": 0.959,
+      "mean_token_accuracy": 0.7258198380470275,
+      "num_tokens": 195400518.0,
+      "step": 82430
+    },
+    {
+      "epoch": 0.7558448702668011,
+      "learning_rate": 4.883285963142936e-05,
+      "loss": 0.8132,
+      "mean_token_accuracy": 0.7587051033973694,
+      "num_tokens": 195427781.0,
+      "step": 82440
+    },
+    {
+      "epoch": 0.7559365545062804,
+      "learning_rate": 4.8814522783533515e-05,
+      "loss": 0.8026,
+      "mean_token_accuracy": 0.7610012650489807,
+      "num_tokens": 195455949.0,
+      "step": 82450
+    },
+    {
+      "epoch": 0.7560282387457596,
+      "learning_rate": 4.879618593563767e-05,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.763341772556305,
+      "num_tokens": 195482634.0,
+      "step": 82460
+    },
+    {
+      "epoch": 0.7561199229852389,
+      "learning_rate": 4.877784908774182e-05,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7670092403888702,
+      "num_tokens": 195508898.0,
+      "step": 82470
+    },
+    {
+      "epoch": 0.756211607224718,
+      "learning_rate": 4.875951223984597e-05,
+      "loss": 0.8222,
+      "mean_token_accuracy": 0.7569530308246613,
+      "num_tokens": 195536749.0,
+      "step": 82480
+    },
+    {
+      "epoch": 0.7563032914641973,
+      "learning_rate": 4.874117539195013e-05,
+      "loss": 0.879,
+      "mean_token_accuracy": 0.7417121946811676,
+      "num_tokens": 195568229.0,
+      "step": 82490
+    },
+    {
+      "epoch": 0.7563949757036765,
+      "learning_rate": 4.872283854405428e-05,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7439973294734955,
+      "num_tokens": 195596713.0,
+      "step": 82500
+    },
+    {
+      "epoch": 0.7564866599431558,
+      "learning_rate": 4.870450169615843e-05,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7655384838581085,
+      "num_tokens": 195624296.0,
+      "step": 82510
+    },
+    {
+      "epoch": 0.756578344182635,
+      "learning_rate": 4.868616484826259e-05,
+      "loss": 0.8227,
+      "mean_token_accuracy": 0.7539513170719147,
+      "num_tokens": 195652740.0,
+      "step": 82520
+    },
+    {
+      "epoch": 0.7566700284221143,
+      "learning_rate": 4.8667828000366737e-05,
+      "loss": 0.8639,
+      "mean_token_accuracy": 0.7521308600902558,
+      "num_tokens": 195681474.0,
+      "step": 82530
+    },
+    {
+      "epoch": 0.7567617126615934,
+      "learning_rate": 4.8649491152470895e-05,
+      "loss": 0.8572,
+      "mean_token_accuracy": 0.7475384712219239,
+      "num_tokens": 195711310.0,
+      "step": 82540
+    },
+    {
+      "epoch": 0.7568533969010727,
+      "learning_rate": 4.863115430457505e-05,
+      "loss": 0.879,
+      "mean_token_accuracy": 0.7408627092838287,
+      "num_tokens": 195740761.0,
+      "step": 82550
+    },
+    {
+      "epoch": 0.7569450811405519,
+      "learning_rate": 4.86128174566792e-05,
+      "loss": 0.8709,
+      "mean_token_accuracy": 0.7468157231807708,
+      "num_tokens": 195769948.0,
+      "step": 82560
+    },
+    {
+      "epoch": 0.7570367653800312,
+      "learning_rate": 4.859448060878336e-05,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7560142934322357,
+      "num_tokens": 195797197.0,
+      "step": 82570
+    },
+    {
+      "epoch": 0.7571284496195104,
+      "learning_rate": 4.85761437608875e-05,
+      "loss": 0.7946,
+      "mean_token_accuracy": 0.7610917687416077,
+      "num_tokens": 195824521.0,
+      "step": 82580
+    },
+    {
+      "epoch": 0.7572201338589897,
+      "learning_rate": 4.855780691299166e-05,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7522114157676697,
+      "num_tokens": 195852487.0,
+      "step": 82590
+    },
+    {
+      "epoch": 0.7573118180984689,
+      "learning_rate": 4.853947006509581e-05,
+      "loss": 0.8967,
+      "mean_token_accuracy": 0.732669985294342,
+      "num_tokens": 195880618.0,
+      "step": 82600
+    },
+    {
+      "epoch": 0.7574035023379481,
+      "learning_rate": 4.8521133217199965e-05,
+      "loss": 0.7925,
+      "mean_token_accuracy": 0.7619107902050019,
+      "num_tokens": 195908327.0,
+      "step": 82610
+    },
+    {
+      "epoch": 0.7574951865774273,
+      "learning_rate": 4.8502796369304124e-05,
+      "loss": 0.8886,
+      "mean_token_accuracy": 0.7358214437961579,
+      "num_tokens": 195937783.0,
+      "step": 82620
+    },
+    {
+      "epoch": 0.7575868708169066,
+      "learning_rate": 4.848445952140827e-05,
+      "loss": 0.7455,
+      "mean_token_accuracy": 0.7729065895080567,
+      "num_tokens": 195963372.0,
+      "step": 82630
+    },
+    {
+      "epoch": 0.7576785550563858,
+      "learning_rate": 4.846612267351243e-05,
+      "loss": 0.7431,
+      "mean_token_accuracy": 0.7816286683082581,
+      "num_tokens": 195989696.0,
+      "step": 82640
+    },
+    {
+      "epoch": 0.7577702392958651,
+      "learning_rate": 4.844778582561658e-05,
+      "loss": 0.822,
+      "mean_token_accuracy": 0.7576343536376953,
+      "num_tokens": 196018147.0,
+      "step": 82650
+    },
+    {
+      "epoch": 0.7578619235353443,
+      "learning_rate": 4.842944897772073e-05,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7488945424556732,
+      "num_tokens": 196048614.0,
+      "step": 82660
+    },
+    {
+      "epoch": 0.7579536077748235,
+      "learning_rate": 4.841111212982489e-05,
+      "loss": 0.8822,
+      "mean_token_accuracy": 0.7415592491626739,
+      "num_tokens": 196080408.0,
+      "step": 82670
+    },
+    {
+      "epoch": 0.7580452920143027,
+      "learning_rate": 4.8392775281929035e-05,
+      "loss": 0.8703,
+      "mean_token_accuracy": 0.7413014054298401,
+      "num_tokens": 196109770.0,
+      "step": 82680
+    },
+    {
+      "epoch": 0.758136976253782,
+      "learning_rate": 4.8374438434033194e-05,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.7770328938961029,
+      "num_tokens": 196136569.0,
+      "step": 82690
+    },
+    {
+      "epoch": 0.7582286604932612,
+      "learning_rate": 4.8356101586137345e-05,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.7713229894638062,
+      "num_tokens": 196164399.0,
+      "step": 82700
+    },
+    {
+      "epoch": 0.7583203447327405,
+      "learning_rate": 4.83377647382415e-05,
+      "loss": 0.8432,
+      "mean_token_accuracy": 0.7497777819633484,
+      "num_tokens": 196196164.0,
+      "step": 82710
+    },
+    {
+      "epoch": 0.7584120289722197,
+      "learning_rate": 4.8319427890345656e-05,
+      "loss": 0.9047,
+      "mean_token_accuracy": 0.7338837504386901,
+      "num_tokens": 196225746.0,
+      "step": 82720
+    },
+    {
+      "epoch": 0.758503713211699,
+      "learning_rate": 4.83010910424498e-05,
+      "loss": 0.8781,
+      "mean_token_accuracy": 0.7373611271381378,
+      "num_tokens": 196254408.0,
+      "step": 82730
+    },
+    {
+      "epoch": 0.7585953974511781,
+      "learning_rate": 4.828275419455396e-05,
+      "loss": 0.8818,
+      "mean_token_accuracy": 0.7404162049293518,
+      "num_tokens": 196284528.0,
+      "step": 82740
+    },
+    {
+      "epoch": 0.7586870816906574,
+      "learning_rate": 4.826441734665811e-05,
+      "loss": 0.7751,
+      "mean_token_accuracy": 0.7641188085079194,
+      "num_tokens": 196312235.0,
+      "step": 82750
+    },
+    {
+      "epoch": 0.7587787659301366,
+      "learning_rate": 4.824608049876226e-05,
+      "loss": 0.8109,
+      "mean_token_accuracy": 0.756779420375824,
+      "num_tokens": 196340891.0,
+      "step": 82760
+    },
+    {
+      "epoch": 0.7588704501696159,
+      "learning_rate": 4.822774365086642e-05,
+      "loss": 0.8108,
+      "mean_token_accuracy": 0.7533910214900971,
+      "num_tokens": 196369247.0,
+      "step": 82770
+    },
+    {
+      "epoch": 0.7589621344090951,
+      "learning_rate": 4.820940680297057e-05,
+      "loss": 0.8806,
+      "mean_token_accuracy": 0.7383510410785675,
+      "num_tokens": 196401289.0,
+      "step": 82780
+    },
+    {
+      "epoch": 0.7590538186485744,
+      "learning_rate": 4.8191069955074726e-05,
+      "loss": 0.8864,
+      "mean_token_accuracy": 0.7434684038162231,
+      "num_tokens": 196430309.0,
+      "step": 82790
+    },
+    {
+      "epoch": 0.7591455028880535,
+      "learning_rate": 4.817273310717888e-05,
+      "loss": 0.9156,
+      "mean_token_accuracy": 0.7362958908081054,
+      "num_tokens": 196461997.0,
+      "step": 82800
+    },
+    {
+      "epoch": 0.7592371871275327,
+      "learning_rate": 4.815439625928303e-05,
+      "loss": 0.7971,
+      "mean_token_accuracy": 0.7583592057228088,
+      "num_tokens": 196489372.0,
+      "step": 82810
+    },
+    {
+      "epoch": 0.759328871367012,
+      "learning_rate": 4.813605941138719e-05,
+      "loss": 0.8019,
+      "mean_token_accuracy": 0.759622585773468,
+      "num_tokens": 196516849.0,
+      "step": 82820
+    },
+    {
+      "epoch": 0.7594205556064912,
+      "learning_rate": 4.811772256349134e-05,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7465933322906494,
+      "num_tokens": 196545889.0,
+      "step": 82830
+    },
+    {
+      "epoch": 0.7595122398459705,
+      "learning_rate": 4.809938571559549e-05,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7508033812046051,
+      "num_tokens": 196575364.0,
+      "step": 82840
+    },
+    {
+      "epoch": 0.7596039240854497,
+      "learning_rate": 4.8081048867699644e-05,
+      "loss": 0.8958,
+      "mean_token_accuracy": 0.7373965382575989,
+      "num_tokens": 196604712.0,
+      "step": 82850
+    },
+    {
+      "epoch": 0.759695608324929,
+      "learning_rate": 4.8062712019803796e-05,
+      "loss": 0.9364,
+      "mean_token_accuracy": 0.7314100861549377,
+      "num_tokens": 196636162.0,
+      "step": 82860
+    },
+    {
+      "epoch": 0.7597872925644081,
+      "learning_rate": 4.8044375171907954e-05,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.754850035905838,
+      "num_tokens": 196665750.0,
+      "step": 82870
+    },
+    {
+      "epoch": 0.7598789768038874,
+      "learning_rate": 4.8026038324012106e-05,
+      "loss": 0.8319,
+      "mean_token_accuracy": 0.7528897941112518,
+      "num_tokens": 196694928.0,
+      "step": 82880
+    },
+    {
+      "epoch": 0.7599706610433666,
+      "learning_rate": 4.800770147611626e-05,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7441156387329102,
+      "num_tokens": 196725611.0,
+      "step": 82890
+    },
+    {
+      "epoch": 0.7600623452828459,
+      "learning_rate": 4.798936462822041e-05,
+      "loss": 0.9098,
+      "mean_token_accuracy": 0.7359290599822998,
+      "num_tokens": 196756044.0,
+      "step": 82900
+    },
+    {
+      "epoch": 0.7601540295223251,
+      "learning_rate": 4.797102778032456e-05,
+      "loss": 0.881,
+      "mean_token_accuracy": 0.7421162307262421,
+      "num_tokens": 196786604.0,
+      "step": 82910
+    },
+    {
+      "epoch": 0.7602457137618044,
+      "learning_rate": 4.795269093242872e-05,
+      "loss": 0.8285,
+      "mean_token_accuracy": 0.7487018823623657,
+      "num_tokens": 196816354.0,
+      "step": 82920
+    },
+    {
+      "epoch": 0.7603373980012835,
+      "learning_rate": 4.793435408453287e-05,
+      "loss": 0.8021,
+      "mean_token_accuracy": 0.7550116896629333,
+      "num_tokens": 196844974.0,
+      "step": 82930
+    },
+    {
+      "epoch": 0.7604290822407628,
+      "learning_rate": 4.7916017236637024e-05,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.7594165742397309,
+      "num_tokens": 196873330.0,
+      "step": 82940
+    },
+    {
+      "epoch": 0.760520766480242,
+      "learning_rate": 4.7897680388741176e-05,
+      "loss": 0.8343,
+      "mean_token_accuracy": 0.7532778263092041,
+      "num_tokens": 196904073.0,
+      "step": 82950
+    },
+    {
+      "epoch": 0.7606124507197213,
+      "learning_rate": 4.787934354084533e-05,
+      "loss": 0.8622,
+      "mean_token_accuracy": 0.7470668494701386,
+      "num_tokens": 196934111.0,
+      "step": 82960
+    },
+    {
+      "epoch": 0.7607041349592005,
+      "learning_rate": 4.7861006692949486e-05,
+      "loss": 0.8852,
+      "mean_token_accuracy": 0.7404623091220855,
+      "num_tokens": 196965605.0,
+      "step": 82970
+    },
+    {
+      "epoch": 0.7607958191986798,
+      "learning_rate": 4.784266984505364e-05,
+      "loss": 0.8767,
+      "mean_token_accuracy": 0.7419257044792176,
+      "num_tokens": 196996708.0,
+      "step": 82980
+    },
+    {
+      "epoch": 0.760887503438159,
+      "learning_rate": 4.782433299715779e-05,
+      "loss": 0.7645,
+      "mean_token_accuracy": 0.773752760887146,
+      "num_tokens": 197023533.0,
+      "step": 82990
+    },
+    {
+      "epoch": 0.7609791876776382,
+      "learning_rate": 4.780599614926194e-05,
+      "loss": 0.8064,
+      "mean_token_accuracy": 0.7623594880104065,
+      "num_tokens": 197050875.0,
+      "step": 83000
+    },
+    {
+      "epoch": 0.7610708719171174,
+      "learning_rate": 4.77876593013661e-05,
+      "loss": 0.8016,
+      "mean_token_accuracy": 0.7565568685531616,
+      "num_tokens": 197079546.0,
+      "step": 83010
+    },
+    {
+      "epoch": 0.7611625561565967,
+      "learning_rate": 4.776932245347025e-05,
+      "loss": 0.8875,
+      "mean_token_accuracy": 0.737514728307724,
+      "num_tokens": 197109931.0,
+      "step": 83020
+    },
+    {
+      "epoch": 0.7612542403960759,
+      "learning_rate": 4.7750985605574404e-05,
+      "loss": 0.8826,
+      "mean_token_accuracy": 0.74748295545578,
+      "num_tokens": 197138034.0,
+      "step": 83030
+    },
+    {
+      "epoch": 0.7613459246355552,
+      "learning_rate": 4.7732648757678556e-05,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.7447394669055939,
+      "num_tokens": 197168245.0,
+      "step": 83040
+    },
+    {
+      "epoch": 0.7614376088750344,
+      "learning_rate": 4.771431190978271e-05,
+      "loss": 0.7924,
+      "mean_token_accuracy": 0.7621159255504608,
+      "num_tokens": 197195422.0,
+      "step": 83050
+    },
+    {
+      "epoch": 0.7615292931145136,
+      "learning_rate": 4.769597506188687e-05,
+      "loss": 0.8145,
+      "mean_token_accuracy": 0.7605322599411011,
+      "num_tokens": 197223696.0,
+      "step": 83060
+    },
+    {
+      "epoch": 0.7616209773539928,
+      "learning_rate": 4.767763821399102e-05,
+      "loss": 0.8269,
+      "mean_token_accuracy": 0.753108811378479,
+      "num_tokens": 197252412.0,
+      "step": 83070
+    },
+    {
+      "epoch": 0.7617126615934721,
+      "learning_rate": 4.765930136609517e-05,
+      "loss": 0.8927,
+      "mean_token_accuracy": 0.744899469614029,
+      "num_tokens": 197283719.0,
+      "step": 83080
+    },
+    {
+      "epoch": 0.7618043458329513,
+      "learning_rate": 4.764096451819932e-05,
+      "loss": 0.8821,
+      "mean_token_accuracy": 0.7437337219715119,
+      "num_tokens": 197313851.0,
+      "step": 83090
+    },
+    {
+      "epoch": 0.7618960300724306,
+      "learning_rate": 4.7622627670303474e-05,
+      "loss": 0.8824,
+      "mean_token_accuracy": 0.73881094455719,
+      "num_tokens": 197344418.0,
+      "step": 83100
+    },
+    {
+      "epoch": 0.7619877143119098,
+      "learning_rate": 4.760429082240763e-05,
+      "loss": 0.8003,
+      "mean_token_accuracy": 0.7583040416240692,
+      "num_tokens": 197373297.0,
+      "step": 83110
+    },
+    {
+      "epoch": 0.7620793985513891,
+      "learning_rate": 4.7585953974511785e-05,
+      "loss": 0.826,
+      "mean_token_accuracy": 0.7492796063423157,
+      "num_tokens": 197402050.0,
+      "step": 83120
+    },
+    {
+      "epoch": 0.7621710827908682,
+      "learning_rate": 4.756761712661594e-05,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.7554567039012909,
+      "num_tokens": 197431839.0,
+      "step": 83130
+    },
+    {
+      "epoch": 0.7622627670303475,
+      "learning_rate": 4.7549280278720095e-05,
+      "loss": 0.8877,
+      "mean_token_accuracy": 0.735871410369873,
+      "num_tokens": 197461976.0,
+      "step": 83140
+    },
+    {
+      "epoch": 0.7623544512698267,
+      "learning_rate": 4.753094343082424e-05,
+      "loss": 0.8506,
+      "mean_token_accuracy": 0.7509628176689148,
+      "num_tokens": 197491842.0,
+      "step": 83150
+    },
+    {
+      "epoch": 0.762446135509306,
+      "learning_rate": 4.75126065829284e-05,
+      "loss": 0.8941,
+      "mean_token_accuracy": 0.7376525521278381,
+      "num_tokens": 197523068.0,
+      "step": 83160
+    },
+    {
+      "epoch": 0.7625378197487852,
+      "learning_rate": 4.749426973503255e-05,
+      "loss": 0.8546,
+      "mean_token_accuracy": 0.7550153732299805,
+      "num_tokens": 197550315.0,
+      "step": 83170
+    },
+    {
+      "epoch": 0.7626295039882645,
+      "learning_rate": 4.74759328871367e-05,
+      "loss": 0.7762,
+      "mean_token_accuracy": 0.763424015045166,
+      "num_tokens": 197578711.0,
+      "step": 83180
+    },
+    {
+      "epoch": 0.7627211882277436,
+      "learning_rate": 4.745759603924086e-05,
+      "loss": 0.8624,
+      "mean_token_accuracy": 0.7480132102966308,
+      "num_tokens": 197608068.0,
+      "step": 83190
+    },
+    {
+      "epoch": 0.7628128724672228,
+      "learning_rate": 4.7439259191345007e-05,
+      "loss": 0.8742,
+      "mean_token_accuracy": 0.7452224850654602,
+      "num_tokens": 197639765.0,
+      "step": 83200
+    },
+    {
+      "epoch": 0.7629045567067021,
+      "learning_rate": 4.7420922343449165e-05,
+      "loss": 0.8813,
+      "mean_token_accuracy": 0.7406799018383026,
+      "num_tokens": 197671467.0,
+      "step": 83210
+    },
+    {
+      "epoch": 0.7629962409461813,
+      "learning_rate": 4.740258549555332e-05,
+      "loss": 0.8808,
+      "mean_token_accuracy": 0.7393268704414367,
+      "num_tokens": 197701847.0,
+      "step": 83220
+    },
+    {
+      "epoch": 0.7630879251856606,
+      "learning_rate": 4.738424864765747e-05,
+      "loss": 0.8206,
+      "mean_token_accuracy": 0.7561595857143402,
+      "num_tokens": 197730915.0,
+      "step": 83230
+    },
+    {
+      "epoch": 0.7631796094251398,
+      "learning_rate": 4.736591179976163e-05,
+      "loss": 0.7857,
+      "mean_token_accuracy": 0.764354532957077,
+      "num_tokens": 197759327.0,
+      "step": 83240
+    },
+    {
+      "epoch": 0.7632712936646191,
+      "learning_rate": 4.734757495186577e-05,
+      "loss": 0.8239,
+      "mean_token_accuracy": 0.7565077424049378,
+      "num_tokens": 197790398.0,
+      "step": 83250
+    },
+    {
+      "epoch": 0.7633629779040982,
+      "learning_rate": 4.732923810396993e-05,
+      "loss": 0.9003,
+      "mean_token_accuracy": 0.7374303638935089,
+      "num_tokens": 197822397.0,
+      "step": 83260
+    },
+    {
+      "epoch": 0.7634546621435775,
+      "learning_rate": 4.731090125607408e-05,
+      "loss": 0.8814,
+      "mean_token_accuracy": 0.7407468259334564,
+      "num_tokens": 197853152.0,
+      "step": 83270
+    },
+    {
+      "epoch": 0.7635463463830567,
+      "learning_rate": 4.7292564408178235e-05,
+      "loss": 0.9331,
+      "mean_token_accuracy": 0.7341228485107422,
+      "num_tokens": 197885302.0,
+      "step": 83280
+    },
+    {
+      "epoch": 0.763638030622536,
+      "learning_rate": 4.7274227560282394e-05,
+      "loss": 0.7912,
+      "mean_token_accuracy": 0.768936812877655,
+      "num_tokens": 197914041.0,
+      "step": 83290
+    },
+    {
+      "epoch": 0.7637297148620152,
+      "learning_rate": 4.725589071238654e-05,
+      "loss": 0.847,
+      "mean_token_accuracy": 0.7490604758262634,
+      "num_tokens": 197942657.0,
+      "step": 83300
+    },
+    {
+      "epoch": 0.7638213991014945,
+      "learning_rate": 4.72375538644907e-05,
+      "loss": 0.8241,
+      "mean_token_accuracy": 0.7578671872615814,
+      "num_tokens": 197971980.0,
+      "step": 83310
+    },
+    {
+      "epoch": 0.7639130833409737,
+      "learning_rate": 4.721921701659485e-05,
+      "loss": 0.877,
+      "mean_token_accuracy": 0.7484039485454559,
+      "num_tokens": 198002428.0,
+      "step": 83320
+    },
+    {
+      "epoch": 0.7640047675804529,
+      "learning_rate": 4.7200880168699e-05,
+      "loss": 0.9238,
+      "mean_token_accuracy": 0.7342865407466889,
+      "num_tokens": 198034811.0,
+      "step": 83330
+    },
+    {
+      "epoch": 0.7640964518199321,
+      "learning_rate": 4.718254332080316e-05,
+      "loss": 0.8576,
+      "mean_token_accuracy": 0.7481640815734864,
+      "num_tokens": 198066354.0,
+      "step": 83340
+    },
+    {
+      "epoch": 0.7641881360594114,
+      "learning_rate": 4.7164206472907305e-05,
+      "loss": 0.7919,
+      "mean_token_accuracy": 0.76490917801857,
+      "num_tokens": 198094464.0,
+      "step": 83350
+    },
+    {
+      "epoch": 0.7642798202988906,
+      "learning_rate": 4.7145869625011464e-05,
+      "loss": 0.8135,
+      "mean_token_accuracy": 0.7590794086456298,
+      "num_tokens": 198123125.0,
+      "step": 83360
+    },
+    {
+      "epoch": 0.7643715045383699,
+      "learning_rate": 4.7127532777115615e-05,
+      "loss": 0.87,
+      "mean_token_accuracy": 0.7405919671058655,
+      "num_tokens": 198154745.0,
+      "step": 83370
+    },
+    {
+      "epoch": 0.7644631887778491,
+      "learning_rate": 4.710919592921977e-05,
+      "loss": 0.9065,
+      "mean_token_accuracy": 0.7394024491310119,
+      "num_tokens": 198186066.0,
+      "step": 83380
+    },
+    {
+      "epoch": 0.7645548730173283,
+      "learning_rate": 4.7090859081323926e-05,
+      "loss": 0.8664,
+      "mean_token_accuracy": 0.7451089084148407,
+      "num_tokens": 198215268.0,
+      "step": 83390
+    },
+    {
+      "epoch": 0.7646465572568075,
+      "learning_rate": 4.707252223342807e-05,
+      "loss": 0.9282,
+      "mean_token_accuracy": 0.7341436266899108,
+      "num_tokens": 198245601.0,
+      "step": 83400
+    },
+    {
+      "epoch": 0.7647382414962868,
+      "learning_rate": 4.705418538553223e-05,
+      "loss": 0.8043,
+      "mean_token_accuracy": 0.7628137052059174,
+      "num_tokens": 198273978.0,
+      "step": 83410
+    },
+    {
+      "epoch": 0.764829925735766,
+      "learning_rate": 4.703584853763638e-05,
+      "loss": 0.8209,
+      "mean_token_accuracy": 0.7553904771804809,
+      "num_tokens": 198302357.0,
+      "step": 83420
+    },
+    {
+      "epoch": 0.7649216099752453,
+      "learning_rate": 4.701751168974053e-05,
+      "loss": 0.8497,
+      "mean_token_accuracy": 0.7504721581935883,
+      "num_tokens": 198332530.0,
+      "step": 83430
+    },
+    {
+      "epoch": 0.7650132942147245,
+      "learning_rate": 4.699917484184469e-05,
+      "loss": 0.8985,
+      "mean_token_accuracy": 0.7399794936180115,
+      "num_tokens": 198363573.0,
+      "step": 83440
+    },
+    {
+      "epoch": 0.7651049784542038,
+      "learning_rate": 4.6980837993948844e-05,
+      "loss": 0.8401,
+      "mean_token_accuracy": 0.7502639472484589,
+      "num_tokens": 198393768.0,
+      "step": 83450
+    },
+    {
+      "epoch": 0.7651966626936829,
+      "learning_rate": 4.6962501146052996e-05,
+      "loss": 0.876,
+      "mean_token_accuracy": 0.7438281536102295,
+      "num_tokens": 198424956.0,
+      "step": 83460
+    },
+    {
+      "epoch": 0.7652883469331622,
+      "learning_rate": 4.694416429815715e-05,
+      "loss": 0.8343,
+      "mean_token_accuracy": 0.7502440989017487,
+      "num_tokens": 198452901.0,
+      "step": 83470
+    },
+    {
+      "epoch": 0.7653800311726414,
+      "learning_rate": 4.69258274502613e-05,
+      "loss": 0.7875,
+      "mean_token_accuracy": 0.761256217956543,
+      "num_tokens": 198481584.0,
+      "step": 83480
+    },
+    {
+      "epoch": 0.7654717154121207,
+      "learning_rate": 4.690749060236546e-05,
+      "loss": 0.8424,
+      "mean_token_accuracy": 0.7500324428081513,
+      "num_tokens": 198510949.0,
+      "step": 83490
+    },
+    {
+      "epoch": 0.7655633996515999,
+      "learning_rate": 4.688915375446961e-05,
+      "loss": 0.8741,
+      "mean_token_accuracy": 0.7372589588165284,
+      "num_tokens": 198541344.0,
+      "step": 83500
+    },
+    {
+      "epoch": 0.7656550838910792,
+      "learning_rate": 4.687081690657376e-05,
+      "loss": 0.8291,
+      "mean_token_accuracy": 0.7572034418582916,
+      "num_tokens": 198570505.0,
+      "step": 83510
+    },
+    {
+      "epoch": 0.7657467681305583,
+      "learning_rate": 4.6852480058677914e-05,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.7366003334522248,
+      "num_tokens": 198601033.0,
+      "step": 83520
+    },
+    {
+      "epoch": 0.7658384523700376,
+      "learning_rate": 4.6834143210782066e-05,
+      "loss": 0.7692,
+      "mean_token_accuracy": 0.7724420845508575,
+      "num_tokens": 198628800.0,
+      "step": 83530
+    },
+    {
+      "epoch": 0.7659301366095168,
+      "learning_rate": 4.6815806362886224e-05,
+      "loss": 0.8236,
+      "mean_token_accuracy": 0.7546740949153901,
+      "num_tokens": 198656600.0,
+      "step": 83540
+    },
+    {
+      "epoch": 0.7660218208489961,
+      "learning_rate": 4.6797469514990376e-05,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7460628628730774,
+      "num_tokens": 198685408.0,
+      "step": 83550
+    },
+    {
+      "epoch": 0.7661135050884753,
+      "learning_rate": 4.677913266709453e-05,
+      "loss": 0.8539,
+      "mean_token_accuracy": 0.7493011832237244,
+      "num_tokens": 198715341.0,
+      "step": 83560
+    },
+    {
+      "epoch": 0.7662051893279546,
+      "learning_rate": 4.676079581919868e-05,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7442546784877777,
+      "num_tokens": 198745625.0,
+      "step": 83570
+    },
+    {
+      "epoch": 0.7662968735674338,
+      "learning_rate": 4.674245897130283e-05,
+      "loss": 0.8705,
+      "mean_token_accuracy": 0.7369942724704742,
+      "num_tokens": 198777699.0,
+      "step": 83580
+    },
+    {
+      "epoch": 0.766388557806913,
+      "learning_rate": 4.672412212340699e-05,
+      "loss": 0.8153,
+      "mean_token_accuracy": 0.7576233804225921,
+      "num_tokens": 198807217.0,
+      "step": 83590
+    },
+    {
+      "epoch": 0.7664802420463922,
+      "learning_rate": 4.670578527551114e-05,
+      "loss": 0.7992,
+      "mean_token_accuracy": 0.7615856409072876,
+      "num_tokens": 198835789.0,
+      "step": 83600
+    },
+    {
+      "epoch": 0.7665719262858715,
+      "learning_rate": 4.6687448427615294e-05,
+      "loss": 0.8994,
+      "mean_token_accuracy": 0.7378689825534821,
+      "num_tokens": 198866279.0,
+      "step": 83610
+    },
+    {
+      "epoch": 0.7666636105253507,
+      "learning_rate": 4.6669111579719446e-05,
+      "loss": 0.838,
+      "mean_token_accuracy": 0.753299742937088,
+      "num_tokens": 198897040.0,
+      "step": 83620
+    },
+    {
+      "epoch": 0.76675529476483,
+      "learning_rate": 4.6650774731823605e-05,
+      "loss": 0.8763,
+      "mean_token_accuracy": 0.7379263043403625,
+      "num_tokens": 198926864.0,
+      "step": 83630
+    },
+    {
+      "epoch": 0.7668469790043092,
+      "learning_rate": 4.6632437883927756e-05,
+      "loss": 0.8474,
+      "mean_token_accuracy": 0.7527245163917542,
+      "num_tokens": 198956688.0,
+      "step": 83640
+    },
+    {
+      "epoch": 0.7669386632437883,
+      "learning_rate": 4.661410103603191e-05,
+      "loss": 0.812,
+      "mean_token_accuracy": 0.7580152869224548,
+      "num_tokens": 198984155.0,
+      "step": 83650
+    },
+    {
+      "epoch": 0.7670303474832676,
+      "learning_rate": 4.659576418813606e-05,
+      "loss": 0.8323,
+      "mean_token_accuracy": 0.7555800259113312,
+      "num_tokens": 199012874.0,
+      "step": 83660
+    },
+    {
+      "epoch": 0.7671220317227468,
+      "learning_rate": 4.657742734024021e-05,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7447729170322418,
+      "num_tokens": 199045041.0,
+      "step": 83670
+    },
+    {
+      "epoch": 0.7672137159622261,
+      "learning_rate": 4.655909049234437e-05,
+      "loss": 0.8939,
+      "mean_token_accuracy": 0.7398266553878784,
+      "num_tokens": 199075229.0,
+      "step": 83680
+    },
+    {
+      "epoch": 0.7673054002017053,
+      "learning_rate": 4.654075364444852e-05,
+      "loss": 0.896,
+      "mean_token_accuracy": 0.7374585330486297,
+      "num_tokens": 199106935.0,
+      "step": 83690
+    },
+    {
+      "epoch": 0.7673970844411846,
+      "learning_rate": 4.6522416796552674e-05,
+      "loss": 0.908,
+      "mean_token_accuracy": 0.734346729516983,
+      "num_tokens": 199139301.0,
+      "step": 83700
+    },
+    {
+      "epoch": 0.7674887686806638,
+      "learning_rate": 4.6504079948656826e-05,
+      "loss": 0.7976,
+      "mean_token_accuracy": 0.7622849762439727,
+      "num_tokens": 199168604.0,
+      "step": 83710
+    },
+    {
+      "epoch": 0.767580452920143,
+      "learning_rate": 4.648574310076098e-05,
+      "loss": 0.8412,
+      "mean_token_accuracy": 0.7505931615829468,
+      "num_tokens": 199198406.0,
+      "step": 83720
+    },
+    {
+      "epoch": 0.7676721371596222,
+      "learning_rate": 4.646740625286514e-05,
+      "loss": 0.8734,
+      "mean_token_accuracy": 0.7455028116703033,
+      "num_tokens": 199228350.0,
+      "step": 83730
+    },
+    {
+      "epoch": 0.7677638213991015,
+      "learning_rate": 4.644906940496929e-05,
+      "loss": 0.914,
+      "mean_token_accuracy": 0.7364421486854553,
+      "num_tokens": 199260398.0,
+      "step": 83740
+    },
+    {
+      "epoch": 0.7678555056385807,
+      "learning_rate": 4.643073255707344e-05,
+      "loss": 0.8972,
+      "mean_token_accuracy": 0.7416184782981873,
+      "num_tokens": 199291315.0,
+      "step": 83750
+    },
+    {
+      "epoch": 0.76794718987806,
+      "learning_rate": 4.641239570917759e-05,
+      "loss": 0.8853,
+      "mean_token_accuracy": 0.7424000024795532,
+      "num_tokens": 199321979.0,
+      "step": 83760
+    },
+    {
+      "epoch": 0.7680388741175392,
+      "learning_rate": 4.6394058861281744e-05,
+      "loss": 0.7973,
+      "mean_token_accuracy": 0.7612855434417725,
+      "num_tokens": 199350143.0,
+      "step": 83770
+    },
+    {
+      "epoch": 0.7681305583570184,
+      "learning_rate": 4.63757220133859e-05,
+      "loss": 0.7887,
+      "mean_token_accuracy": 0.7664248049259186,
+      "num_tokens": 199378249.0,
+      "step": 83780
+    },
+    {
+      "epoch": 0.7682222425964976,
+      "learning_rate": 4.6357385165490055e-05,
+      "loss": 0.856,
+      "mean_token_accuracy": 0.74514821767807,
+      "num_tokens": 199406682.0,
+      "step": 83790
+    },
+    {
+      "epoch": 0.7683139268359769,
+      "learning_rate": 4.633904831759421e-05,
+      "loss": 0.8775,
+      "mean_token_accuracy": 0.7419812917709351,
+      "num_tokens": 199437232.0,
+      "step": 83800
+    },
+    {
+      "epoch": 0.7684056110754561,
+      "learning_rate": 4.6320711469698365e-05,
+      "loss": 0.8988,
+      "mean_token_accuracy": 0.7366764843463898,
+      "num_tokens": 199467606.0,
+      "step": 83810
+    },
+    {
+      "epoch": 0.7684972953149354,
+      "learning_rate": 4.630237462180251e-05,
+      "loss": 0.9094,
+      "mean_token_accuracy": 0.734112274646759,
+      "num_tokens": 199498892.0,
+      "step": 83820
+    },
+    {
+      "epoch": 0.7685889795544146,
+      "learning_rate": 4.628403777390667e-05,
+      "loss": 0.8255,
+      "mean_token_accuracy": 0.7513435065746308,
+      "num_tokens": 199528224.0,
+      "step": 83830
+    },
+    {
+      "epoch": 0.7686806637938939,
+      "learning_rate": 4.626570092601082e-05,
+      "loss": 0.8432,
+      "mean_token_accuracy": 0.7515334784984589,
+      "num_tokens": 199557846.0,
+      "step": 83840
+    },
+    {
+      "epoch": 0.768772348033373,
+      "learning_rate": 4.624736407811497e-05,
+      "loss": 0.8959,
+      "mean_token_accuracy": 0.7378568053245544,
+      "num_tokens": 199589559.0,
+      "step": 83850
+    },
+    {
+      "epoch": 0.7688640322728523,
+      "learning_rate": 4.622902723021913e-05,
+      "loss": 0.9457,
+      "mean_token_accuracy": 0.7292453050613403,
+      "num_tokens": 199622291.0,
+      "step": 83860
+    },
+    {
+      "epoch": 0.7689557165123315,
+      "learning_rate": 4.6210690382323277e-05,
+      "loss": 0.8829,
+      "mean_token_accuracy": 0.7407571256160737,
+      "num_tokens": 199652353.0,
+      "step": 83870
+    },
+    {
+      "epoch": 0.7690474007518108,
+      "learning_rate": 4.6192353534427435e-05,
+      "loss": 0.9178,
+      "mean_token_accuracy": 0.7299648880958557,
+      "num_tokens": 199683926.0,
+      "step": 83880
+    },
+    {
+      "epoch": 0.76913908499129,
+      "learning_rate": 4.617401668653159e-05,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7676116824150085,
+      "num_tokens": 199711655.0,
+      "step": 83890
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "learning_rate": 4.615567983863574e-05,
+      "loss": 0.8324,
+      "mean_token_accuracy": 0.7521481812000275,
+      "num_tokens": 199740477.0,
+      "step": 83900
+    },
+    {
+      "epoch": 0.7693224534702484,
+      "learning_rate": 4.61373429907399e-05,
+      "loss": 0.8689,
+      "mean_token_accuracy": 0.7458119213581085,
+      "num_tokens": 199769862.0,
+      "step": 83910
+    },
+    {
+      "epoch": 0.7694141377097277,
+      "learning_rate": 4.611900614284404e-05,
+      "loss": 0.8911,
+      "mean_token_accuracy": 0.7416059374809265,
+      "num_tokens": 199802392.0,
+      "step": 83920
+    },
+    {
+      "epoch": 0.7695058219492069,
+      "learning_rate": 4.61006692949482e-05,
+      "loss": 0.9199,
+      "mean_token_accuracy": 0.736228609085083,
+      "num_tokens": 199834387.0,
+      "step": 83930
+    },
+    {
+      "epoch": 0.7695975061886862,
+      "learning_rate": 4.608233244705236e-05,
+      "loss": 0.8846,
+      "mean_token_accuracy": 0.7385479390621186,
+      "num_tokens": 199866001.0,
+      "step": 83940
+    },
+    {
+      "epoch": 0.7696891904281654,
+      "learning_rate": 4.6063995599156505e-05,
+      "loss": 0.8315,
+      "mean_token_accuracy": 0.7526941895484924,
+      "num_tokens": 199896559.0,
+      "step": 83950
+    },
+    {
+      "epoch": 0.7697808746676447,
+      "learning_rate": 4.6045658751260664e-05,
+      "loss": 0.7973,
+      "mean_token_accuracy": 0.758766508102417,
+      "num_tokens": 199925437.0,
+      "step": 83960
+    },
+    {
+      "epoch": 0.7698725589071239,
+      "learning_rate": 4.602732190336481e-05,
+      "loss": 0.8696,
+      "mean_token_accuracy": 0.7400624334812165,
+      "num_tokens": 199954560.0,
+      "step": 83970
+    },
+    {
+      "epoch": 0.7699642431466031,
+      "learning_rate": 4.600898505546897e-05,
+      "loss": 0.92,
+      "mean_token_accuracy": 0.7286685883998871,
+      "num_tokens": 199987075.0,
+      "step": 83980
+    },
+    {
+      "epoch": 0.7700559273860823,
+      "learning_rate": 4.5990648207573126e-05,
+      "loss": 0.8926,
+      "mean_token_accuracy": 0.7395737707614899,
+      "num_tokens": 200016068.0,
+      "step": 83990
+    },
+    {
+      "epoch": 0.7701476116255616,
+      "learning_rate": 4.597231135967727e-05,
+      "loss": 0.9438,
+      "mean_token_accuracy": 0.7261269629001618,
+      "num_tokens": 200048478.0,
+      "step": 84000
+    },
+    {
+      "epoch": 0.7702392958650408,
+      "learning_rate": 4.595397451178143e-05,
+      "loss": 0.8153,
+      "mean_token_accuracy": 0.7573321878910064,
+      "num_tokens": 200075808.0,
+      "step": 84010
+    },
+    {
+      "epoch": 0.7703309801045201,
+      "learning_rate": 4.5935637663885575e-05,
+      "loss": 0.8183,
+      "mean_token_accuracy": 0.7568739056587219,
+      "num_tokens": 200104732.0,
+      "step": 84020
+    },
+    {
+      "epoch": 0.7704226643439993,
+      "learning_rate": 4.5917300815989733e-05,
+      "loss": 0.8919,
+      "mean_token_accuracy": 0.7371907711029053,
+      "num_tokens": 200135263.0,
+      "step": 84030
+    },
+    {
+      "epoch": 0.7705143485834784,
+      "learning_rate": 4.589896396809389e-05,
+      "loss": 0.892,
+      "mean_token_accuracy": 0.7385146617889404,
+      "num_tokens": 200167528.0,
+      "step": 84040
+    },
+    {
+      "epoch": 0.7706060328229577,
+      "learning_rate": 4.588062712019804e-05,
+      "loss": 0.8677,
+      "mean_token_accuracy": 0.7420639574527741,
+      "num_tokens": 200198130.0,
+      "step": 84050
+    },
+    {
+      "epoch": 0.770697717062437,
+      "learning_rate": 4.5862290272302196e-05,
+      "loss": 0.9424,
+      "mean_token_accuracy": 0.7321244180202484,
+      "num_tokens": 200230737.0,
+      "step": 84060
+    },
+    {
+      "epoch": 0.7707894013019162,
+      "learning_rate": 4.584395342440634e-05,
+      "loss": 0.8257,
+      "mean_token_accuracy": 0.7498086869716645,
+      "num_tokens": 200259700.0,
+      "step": 84070
+    },
+    {
+      "epoch": 0.7708810855413954,
+      "learning_rate": 4.58256165765105e-05,
+      "loss": 0.7886,
+      "mean_token_accuracy": 0.7610655963420868,
+      "num_tokens": 200289453.0,
+      "step": 84080
+    },
+    {
+      "epoch": 0.7709727697808747,
+      "learning_rate": 4.580727972861466e-05,
+      "loss": 0.8737,
+      "mean_token_accuracy": 0.7464800536632538,
+      "num_tokens": 200319562.0,
+      "step": 84090
+    },
+    {
+      "epoch": 0.771064454020354,
+      "learning_rate": 4.57889428807188e-05,
+      "loss": 0.8658,
+      "mean_token_accuracy": 0.743821769952774,
+      "num_tokens": 200350979.0,
+      "step": 84100
+    },
+    {
+      "epoch": 0.7711561382598331,
+      "learning_rate": 4.577060603282296e-05,
+      "loss": 0.8971,
+      "mean_token_accuracy": 0.7383689880371094,
+      "num_tokens": 200382264.0,
+      "step": 84110
+    },
+    {
+      "epoch": 0.7712478224993123,
+      "learning_rate": 4.5752269184927114e-05,
+      "loss": 0.9323,
+      "mean_token_accuracy": 0.725857412815094,
+      "num_tokens": 200415235.0,
+      "step": 84120
+    },
+    {
+      "epoch": 0.7713395067387916,
+      "learning_rate": 4.5733932337031266e-05,
+      "loss": 0.8418,
+      "mean_token_accuracy": 0.747672826051712,
+      "num_tokens": 200445720.0,
+      "step": 84130
+    },
+    {
+      "epoch": 0.7714311909782708,
+      "learning_rate": 4.5715595489135424e-05,
+      "loss": 0.803,
+      "mean_token_accuracy": 0.7635540187358856,
+      "num_tokens": 200476026.0,
+      "step": 84140
+    },
+    {
+      "epoch": 0.7715228752177501,
+      "learning_rate": 4.569725864123957e-05,
+      "loss": 0.8746,
+      "mean_token_accuracy": 0.7424876809120178,
+      "num_tokens": 200506677.0,
+      "step": 84150
+    },
+    {
+      "epoch": 0.7716145594572293,
+      "learning_rate": 4.567892179334373e-05,
+      "loss": 0.89,
+      "mean_token_accuracy": 0.7448381662368775,
+      "num_tokens": 200538444.0,
+      "step": 84160
+    },
+    {
+      "epoch": 0.7717062436967085,
+      "learning_rate": 4.566058494544788e-05,
+      "loss": 0.9157,
+      "mean_token_accuracy": 0.7318846702575683,
+      "num_tokens": 200570153.0,
+      "step": 84170
+    },
+    {
+      "epoch": 0.7717979279361877,
+      "learning_rate": 4.564224809755203e-05,
+      "loss": 0.9456,
+      "mean_token_accuracy": 0.7250260949134827,
+      "num_tokens": 200602444.0,
+      "step": 84180
+    },
+    {
+      "epoch": 0.771889612175667,
+      "learning_rate": 4.562391124965619e-05,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7524682223796845,
+      "num_tokens": 200631794.0,
+      "step": 84190
+    },
+    {
+      "epoch": 0.7719812964151462,
+      "learning_rate": 4.5605574401760336e-05,
+      "loss": 0.8398,
+      "mean_token_accuracy": 0.7515843987464905,
+      "num_tokens": 200660885.0,
+      "step": 84200
+    },
+    {
+      "epoch": 0.7720729806546255,
+      "learning_rate": 4.5587237553864494e-05,
+      "loss": 0.8494,
+      "mean_token_accuracy": 0.7495680034160614,
+      "num_tokens": 200689887.0,
+      "step": 84210
+    },
+    {
+      "epoch": 0.7721646648941047,
+      "learning_rate": 4.5568900705968646e-05,
+      "loss": 0.8933,
+      "mean_token_accuracy": 0.74183189868927,
+      "num_tokens": 200721427.0,
+      "step": 84220
+    },
+    {
+      "epoch": 0.772256349133584,
+      "learning_rate": 4.55505638580728e-05,
+      "loss": 0.9562,
+      "mean_token_accuracy": 0.7235827565193176,
+      "num_tokens": 200754076.0,
+      "step": 84230
+    },
+    {
+      "epoch": 0.7723480333730631,
+      "learning_rate": 4.5532227010176957e-05,
+      "loss": 0.9456,
+      "mean_token_accuracy": 0.7272350370883942,
+      "num_tokens": 200786233.0,
+      "step": 84240
+    },
+    {
+      "epoch": 0.7724397176125424,
+      "learning_rate": 4.551389016228111e-05,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.751090008020401,
+      "num_tokens": 200816630.0,
+      "step": 84250
+    },
+    {
+      "epoch": 0.7725314018520216,
+      "learning_rate": 4.549555331438526e-05,
+      "loss": 0.8299,
+      "mean_token_accuracy": 0.7541809618473053,
+      "num_tokens": 200846453.0,
+      "step": 84260
+    },
+    {
+      "epoch": 0.7726230860915009,
+      "learning_rate": 4.547721646648941e-05,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7469097554683686,
+      "num_tokens": 200878278.0,
+      "step": 84270
+    },
+    {
+      "epoch": 0.7727147703309801,
+      "learning_rate": 4.5458879618593564e-05,
+      "loss": 0.9055,
+      "mean_token_accuracy": 0.7331558406352997,
+      "num_tokens": 200910438.0,
+      "step": 84280
+    },
+    {
+      "epoch": 0.7728064545704594,
+      "learning_rate": 4.544054277069772e-05,
+      "loss": 0.9232,
+      "mean_token_accuracy": 0.7295502841472625,
+      "num_tokens": 200942229.0,
+      "step": 84290
+    },
+    {
+      "epoch": 0.7728981388099385,
+      "learning_rate": 4.5422205922801875e-05,
+      "loss": 0.9289,
+      "mean_token_accuracy": 0.7294548392295838,
+      "num_tokens": 200973791.0,
+      "step": 84300
+    },
+    {
+      "epoch": 0.7729898230494178,
+      "learning_rate": 4.5403869074906026e-05,
+      "loss": 0.8163,
+      "mean_token_accuracy": 0.7611810564994812,
+      "num_tokens": 201002468.0,
+      "step": 84310
+    },
+    {
+      "epoch": 0.773081507288897,
+      "learning_rate": 4.538553222701018e-05,
+      "loss": 0.8038,
+      "mean_token_accuracy": 0.7570303320884705,
+      "num_tokens": 201030933.0,
+      "step": 84320
+    },
+    {
+      "epoch": 0.7731731915283763,
+      "learning_rate": 4.536719537911433e-05,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7499175071716309,
+      "num_tokens": 201060148.0,
+      "step": 84330
+    },
+    {
+      "epoch": 0.7732648757678555,
+      "learning_rate": 4.534885853121849e-05,
+      "loss": 0.9224,
+      "mean_token_accuracy": 0.7309393227100373,
+      "num_tokens": 201092407.0,
+      "step": 84340
+    },
+    {
+      "epoch": 0.7733565600073348,
+      "learning_rate": 4.533052168332264e-05,
+      "loss": 0.9403,
+      "mean_token_accuracy": 0.7300654292106629,
+      "num_tokens": 201124023.0,
+      "step": 84350
+    },
+    {
+      "epoch": 0.773448244246814,
+      "learning_rate": 4.531218483542679e-05,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.7361896634101868,
+      "num_tokens": 201157314.0,
+      "step": 84360
+    },
+    {
+      "epoch": 0.7735399284862932,
+      "learning_rate": 4.5293847987530944e-05,
+      "loss": 0.8118,
+      "mean_token_accuracy": 0.7588523328304291,
+      "num_tokens": 201187175.0,
+      "step": 84370
+    },
+    {
+      "epoch": 0.7736316127257724,
+      "learning_rate": 4.5275511139635096e-05,
+      "loss": 0.8386,
+      "mean_token_accuracy": 0.7514602661132812,
+      "num_tokens": 201215879.0,
+      "step": 84380
+    },
+    {
+      "epoch": 0.7737232969652517,
+      "learning_rate": 4.5257174291739255e-05,
+      "loss": 0.8733,
+      "mean_token_accuracy": 0.7433182656764984,
+      "num_tokens": 201246132.0,
+      "step": 84390
+    },
+    {
+      "epoch": 0.7738149812047309,
+      "learning_rate": 4.523883744384341e-05,
+      "loss": 0.8868,
+      "mean_token_accuracy": 0.7378167867660522,
+      "num_tokens": 201277352.0,
+      "step": 84400
+    },
+    {
+      "epoch": 0.7739066654442102,
+      "learning_rate": 4.522050059594756e-05,
+      "loss": 0.9115,
+      "mean_token_accuracy": 0.7333837032318116,
+      "num_tokens": 201307674.0,
+      "step": 84410
+    },
+    {
+      "epoch": 0.7739983496836894,
+      "learning_rate": 4.520216374805171e-05,
+      "loss": 0.9436,
+      "mean_token_accuracy": 0.7246712803840637,
+      "num_tokens": 201338505.0,
+      "step": 84420
+    },
+    {
+      "epoch": 0.7740900339231686,
+      "learning_rate": 4.518382690015587e-05,
+      "loss": 0.8314,
+      "mean_token_accuracy": 0.7528190314769745,
+      "num_tokens": 201368586.0,
+      "step": 84430
+    },
+    {
+      "epoch": 0.7741817181626478,
+      "learning_rate": 4.516549005226002e-05,
+      "loss": 0.8547,
+      "mean_token_accuracy": 0.7471593379974365,
+      "num_tokens": 201398791.0,
+      "step": 84440
+    },
+    {
+      "epoch": 0.774273402402127,
+      "learning_rate": 4.514715320436417e-05,
+      "loss": 0.8734,
+      "mean_token_accuracy": 0.7454115688800812,
+      "num_tokens": 201428615.0,
+      "step": 84450
+    },
+    {
+      "epoch": 0.7743650866416063,
+      "learning_rate": 4.5128816356468325e-05,
+      "loss": 0.8787,
+      "mean_token_accuracy": 0.7385600090026856,
+      "num_tokens": 201457486.0,
+      "step": 84460
+    },
+    {
+      "epoch": 0.7744567708810856,
+      "learning_rate": 4.511047950857248e-05,
+      "loss": 0.9413,
+      "mean_token_accuracy": 0.7282081961631774,
+      "num_tokens": 201490066.0,
+      "step": 84470
+    },
+    {
+      "epoch": 0.7745484551205648,
+      "learning_rate": 4.5092142660676635e-05,
+      "loss": 0.9275,
+      "mean_token_accuracy": 0.7271670460700989,
+      "num_tokens": 201521823.0,
+      "step": 84480
+    },
+    {
+      "epoch": 0.774640139360044,
+      "learning_rate": 4.507380581278079e-05,
+      "loss": 0.816,
+      "mean_token_accuracy": 0.7563360810279847,
+      "num_tokens": 201551414.0,
+      "step": 84490
+    },
+    {
+      "epoch": 0.7747318235995232,
+      "learning_rate": 4.505546896488494e-05,
+      "loss": 0.828,
+      "mean_token_accuracy": 0.751528549194336,
+      "num_tokens": 201581548.0,
+      "step": 84500
+    },
+    {
+      "epoch": 0.7748235078390024,
+      "learning_rate": 4.503713211698909e-05,
+      "loss": 0.8881,
+      "mean_token_accuracy": 0.7387636959552765,
+      "num_tokens": 201612070.0,
+      "step": 84510
+    },
+    {
+      "epoch": 0.7749151920784817,
+      "learning_rate": 4.501879526909324e-05,
+      "loss": 0.9027,
+      "mean_token_accuracy": 0.7346662044525146,
+      "num_tokens": 201643590.0,
+      "step": 84520
+    },
+    {
+      "epoch": 0.775006876317961,
+      "learning_rate": 4.50004584211974e-05,
+      "loss": 0.8896,
+      "mean_token_accuracy": 0.73323375582695,
+      "num_tokens": 201676507.0,
+      "step": 84530
+    },
+    {
+      "epoch": 0.7750985605574402,
+      "learning_rate": 4.498212157330155e-05,
+      "loss": 0.8294,
+      "mean_token_accuracy": 0.7489039301872253,
+      "num_tokens": 201706025.0,
+      "step": 84540
+    },
+    {
+      "epoch": 0.7751902447969194,
+      "learning_rate": 4.4963784725405705e-05,
+      "loss": 0.8175,
+      "mean_token_accuracy": 0.7632289946079254,
+      "num_tokens": 201736084.0,
+      "step": 84550
+    },
+    {
+      "epoch": 0.7752819290363987,
+      "learning_rate": 4.4945447877509864e-05,
+      "loss": 0.8187,
+      "mean_token_accuracy": 0.7637658715248108,
+      "num_tokens": 201765245.0,
+      "step": 84560
+    },
+    {
+      "epoch": 0.7753736132758778,
+      "learning_rate": 4.492711102961401e-05,
+      "loss": 0.8779,
+      "mean_token_accuracy": 0.7411769688129425,
+      "num_tokens": 201796342.0,
+      "step": 84570
+    },
+    {
+      "epoch": 0.7754652975153571,
+      "learning_rate": 4.490877418171817e-05,
+      "loss": 0.8723,
+      "mean_token_accuracy": 0.7425032436847687,
+      "num_tokens": 201827445.0,
+      "step": 84580
+    },
+    {
+      "epoch": 0.7755569817548363,
+      "learning_rate": 4.489043733382232e-05,
+      "loss": 0.919,
+      "mean_token_accuracy": 0.733311927318573,
+      "num_tokens": 201860193.0,
+      "step": 84590
+    },
+    {
+      "epoch": 0.7756486659943156,
+      "learning_rate": 4.487210048592647e-05,
+      "loss": 0.8842,
+      "mean_token_accuracy": 0.7423205018043518,
+      "num_tokens": 201891888.0,
+      "step": 84600
+    },
+    {
+      "epoch": 0.7757403502337948,
+      "learning_rate": 4.485376363803063e-05,
+      "loss": 0.8954,
+      "mean_token_accuracy": 0.7401693165302277,
+      "num_tokens": 201922830.0,
+      "step": 84610
+    },
+    {
+      "epoch": 0.7758320344732741,
+      "learning_rate": 4.4835426790134775e-05,
+      "loss": 0.8415,
+      "mean_token_accuracy": 0.7449678301811218,
+      "num_tokens": 201953230.0,
+      "step": 84620
+    },
+    {
+      "epoch": 0.7759237187127532,
+      "learning_rate": 4.4817089942238934e-05,
+      "loss": 0.8791,
+      "mean_token_accuracy": 0.7392831981182099,
+      "num_tokens": 201984164.0,
+      "step": 84630
+    },
+    {
+      "epoch": 0.7760154029522325,
+      "learning_rate": 4.4798753094343086e-05,
+      "loss": 0.8799,
+      "mean_token_accuracy": 0.7358763754367829,
+      "num_tokens": 202016124.0,
+      "step": 84640
+    },
+    {
+      "epoch": 0.7761070871917117,
+      "learning_rate": 4.478041624644724e-05,
+      "loss": 0.9292,
+      "mean_token_accuracy": 0.728614866733551,
+      "num_tokens": 202047239.0,
+      "step": 84650
+    },
+    {
+      "epoch": 0.776198771431191,
+      "learning_rate": 4.4762079398551396e-05,
+      "loss": 0.9635,
+      "mean_token_accuracy": 0.7204731047153473,
+      "num_tokens": 202080247.0,
+      "step": 84660
+    },
+    {
+      "epoch": 0.7762904556706702,
+      "learning_rate": 4.474374255065554e-05,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.7442466080188751,
+      "num_tokens": 202110467.0,
+      "step": 84670
+    },
+    {
+      "epoch": 0.7763821399101495,
+      "learning_rate": 4.47254057027597e-05,
+      "loss": 0.9123,
+      "mean_token_accuracy": 0.7306386232376099,
+      "num_tokens": 202140448.0,
+      "step": 84680
+    },
+    {
+      "epoch": 0.7764738241496287,
+      "learning_rate": 4.470706885486385e-05,
+      "loss": 0.9171,
+      "mean_token_accuracy": 0.7329727530479431,
+      "num_tokens": 202171500.0,
+      "step": 84690
+    },
+    {
+      "epoch": 0.7765655083891079,
+      "learning_rate": 4.4688732006968003e-05,
+      "loss": 0.9455,
+      "mean_token_accuracy": 0.7198031067848205,
+      "num_tokens": 202202563.0,
+      "step": 84700
+    },
+    {
+      "epoch": 0.7766571926285871,
+      "learning_rate": 4.467039515907216e-05,
+      "loss": 0.9206,
+      "mean_token_accuracy": 0.7255331695079803,
+      "num_tokens": 202235172.0,
+      "step": 84710
+    },
+    {
+      "epoch": 0.7767488768680664,
+      "learning_rate": 4.465205831117631e-05,
+      "loss": 0.9466,
+      "mean_token_accuracy": 0.7252993047237396,
+      "num_tokens": 202266689.0,
+      "step": 84720
+    },
+    {
+      "epoch": 0.7768405611075456,
+      "learning_rate": 4.4633721463280466e-05,
+      "loss": 0.8708,
+      "mean_token_accuracy": 0.7456136643886566,
+      "num_tokens": 202296034.0,
+      "step": 84730
+    },
+    {
+      "epoch": 0.7769322453470249,
+      "learning_rate": 4.461538461538462e-05,
+      "loss": 0.9051,
+      "mean_token_accuracy": 0.7344099283218384,
+      "num_tokens": 202325764.0,
+      "step": 84740
+    },
+    {
+      "epoch": 0.7770239295865041,
+      "learning_rate": 4.459704776748877e-05,
+      "loss": 0.9168,
+      "mean_token_accuracy": 0.7299425065517425,
+      "num_tokens": 202357366.0,
+      "step": 84750
+    },
+    {
+      "epoch": 0.7771156138259833,
+      "learning_rate": 4.457871091959293e-05,
+      "loss": 0.9272,
+      "mean_token_accuracy": 0.7272787809371948,
+      "num_tokens": 202388766.0,
+      "step": 84760
+    },
+    {
+      "epoch": 0.7772072980654625,
+      "learning_rate": 4.456037407169707e-05,
+      "loss": 0.9505,
+      "mean_token_accuracy": 0.7295663893222809,
+      "num_tokens": 202420890.0,
+      "step": 84770
+    },
+    {
+      "epoch": 0.7772989823049418,
+      "learning_rate": 4.454203722380123e-05,
+      "loss": 0.9288,
+      "mean_token_accuracy": 0.7241439163684845,
+      "num_tokens": 202453652.0,
+      "step": 84780
+    },
+    {
+      "epoch": 0.777390666544421,
+      "learning_rate": 4.4523700375905384e-05,
+      "loss": 0.7904,
+      "mean_token_accuracy": 0.764022845029831,
+      "num_tokens": 202483012.0,
+      "step": 84790
+    },
+    {
+      "epoch": 0.7774823507839003,
+      "learning_rate": 4.4505363528009536e-05,
+      "loss": 0.876,
+      "mean_token_accuracy": 0.74369495511055,
+      "num_tokens": 202512468.0,
+      "step": 84800
+    },
+    {
+      "epoch": 0.7775740350233795,
+      "learning_rate": 4.4487026680113694e-05,
+      "loss": 0.8888,
+      "mean_token_accuracy": 0.7436029970645904,
+      "num_tokens": 202543085.0,
+      "step": 84810
+    },
+    {
+      "epoch": 0.7776657192628588,
+      "learning_rate": 4.446868983221784e-05,
+      "loss": 0.8977,
+      "mean_token_accuracy": 0.7374972462654114,
+      "num_tokens": 202574931.0,
+      "step": 84820
+    },
+    {
+      "epoch": 0.7777574035023379,
+      "learning_rate": 4.4450352984322e-05,
+      "loss": 0.9219,
+      "mean_token_accuracy": 0.7311857998371124,
+      "num_tokens": 202605767.0,
+      "step": 84830
+    },
+    {
+      "epoch": 0.7778490877418172,
+      "learning_rate": 4.443201613642615e-05,
+      "loss": 0.9917,
+      "mean_token_accuracy": 0.708983552455902,
+      "num_tokens": 202637227.0,
+      "step": 84840
+    },
+    {
+      "epoch": 0.7779407719812964,
+      "learning_rate": 4.44136792885303e-05,
+      "loss": 0.8176,
+      "mean_token_accuracy": 0.7531644344329834,
+      "num_tokens": 202666585.0,
+      "step": 84850
+    },
+    {
+      "epoch": 0.7780324562207757,
+      "learning_rate": 4.439534244063446e-05,
+      "loss": 0.8822,
+      "mean_token_accuracy": 0.7419060647487641,
+      "num_tokens": 202695105.0,
+      "step": 84860
+    },
+    {
+      "epoch": 0.7781241404602549,
+      "learning_rate": 4.437700559273861e-05,
+      "loss": 0.9061,
+      "mean_token_accuracy": 0.7396051228046417,
+      "num_tokens": 202725029.0,
+      "step": 84870
+    },
+    {
+      "epoch": 0.7782158246997342,
+      "learning_rate": 4.4358668744842764e-05,
+      "loss": 0.9203,
+      "mean_token_accuracy": 0.7300542056560516,
+      "num_tokens": 202756201.0,
+      "step": 84880
+    },
+    {
+      "epoch": 0.7783075089392133,
+      "learning_rate": 4.4340331896946916e-05,
+      "loss": 0.9299,
+      "mean_token_accuracy": 0.7266810297966003,
+      "num_tokens": 202787496.0,
+      "step": 84890
+    },
+    {
+      "epoch": 0.7783991931786925,
+      "learning_rate": 4.432199504905107e-05,
+      "loss": 0.9405,
+      "mean_token_accuracy": 0.7293095231056214,
+      "num_tokens": 202818788.0,
+      "step": 84900
+    },
+    {
+      "epoch": 0.7784908774181718,
+      "learning_rate": 4.4303658201155227e-05,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7482227504253387,
+      "num_tokens": 202848512.0,
+      "step": 84910
+    },
+    {
+      "epoch": 0.778582561657651,
+      "learning_rate": 4.428532135325938e-05,
+      "loss": 0.9322,
+      "mean_token_accuracy": 0.7277007639408112,
+      "num_tokens": 202880024.0,
+      "step": 84920
+    },
+    {
+      "epoch": 0.7786742458971303,
+      "learning_rate": 4.426698450536353e-05,
+      "loss": 0.8641,
+      "mean_token_accuracy": 0.7422133386135101,
+      "num_tokens": 202908196.0,
+      "step": 84930
+    },
+    {
+      "epoch": 0.7787659301366096,
+      "learning_rate": 4.424864765746768e-05,
+      "loss": 0.9212,
+      "mean_token_accuracy": 0.7326892197132111,
+      "num_tokens": 202939300.0,
+      "step": 84940
+    },
+    {
+      "epoch": 0.7788576143760888,
+      "learning_rate": 4.4230310809571834e-05,
+      "loss": 0.9712,
+      "mean_token_accuracy": 0.7154947459697724,
+      "num_tokens": 202970845.0,
+      "step": 84950
+    },
+    {
+      "epoch": 0.7789492986155679,
+      "learning_rate": 4.421197396167599e-05,
+      "loss": 1.0009,
+      "mean_token_accuracy": 0.709311020374298,
+      "num_tokens": 203003510.0,
+      "step": 84960
+    },
+    {
+      "epoch": 0.7790409828550472,
+      "learning_rate": 4.4193637113780145e-05,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7473345458507538,
+      "num_tokens": 203033017.0,
+      "step": 84970
+    },
+    {
+      "epoch": 0.7791326670945264,
+      "learning_rate": 4.4175300265884296e-05,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7498608231544495,
+      "num_tokens": 203063345.0,
+      "step": 84980
+    },
+    {
+      "epoch": 0.7792243513340057,
+      "learning_rate": 4.415696341798845e-05,
+      "loss": 0.9385,
+      "mean_token_accuracy": 0.7267225861549378,
+      "num_tokens": 203095697.0,
+      "step": 84990
+    },
+    {
+      "epoch": 0.7793160355734849,
+      "learning_rate": 4.41386265700926e-05,
+      "loss": 0.9433,
+      "mean_token_accuracy": 0.7269436657428742,
+      "num_tokens": 203124455.0,
+      "step": 85000
+    },
+    {
+      "epoch": 0.7794077198129642,
+      "learning_rate": 4.412028972219676e-05,
+      "loss": 0.9273,
+      "mean_token_accuracy": 0.728283804655075,
+      "num_tokens": 203154919.0,
+      "step": 85010
+    },
+    {
+      "epoch": 0.7794994040524433,
+      "learning_rate": 4.410195287430091e-05,
+      "loss": 0.9829,
+      "mean_token_accuracy": 0.7103117644786835,
+      "num_tokens": 203187800.0,
+      "step": 85020
+    },
+    {
+      "epoch": 0.7795910882919226,
+      "learning_rate": 4.408361602640506e-05,
+      "loss": 0.9565,
+      "mean_token_accuracy": 0.7182743668556213,
+      "num_tokens": 203218880.0,
+      "step": 85030
+    },
+    {
+      "epoch": 0.7796827725314018,
+      "learning_rate": 4.4065279178509214e-05,
+      "loss": 0.8866,
+      "mean_token_accuracy": 0.7394263863563537,
+      "num_tokens": 203248934.0,
+      "step": 85040
+    },
+    {
+      "epoch": 0.7797744567708811,
+      "learning_rate": 4.404694233061337e-05,
+      "loss": 0.9379,
+      "mean_token_accuracy": 0.7228012382984161,
+      "num_tokens": 203280843.0,
+      "step": 85050
+    },
+    {
+      "epoch": 0.7798661410103603,
+      "learning_rate": 4.4028605482717525e-05,
+      "loss": 0.9425,
+      "mean_token_accuracy": 0.7212895274162292,
+      "num_tokens": 203311333.0,
+      "step": 85060
+    },
+    {
+      "epoch": 0.7799578252498396,
+      "learning_rate": 4.401026863482168e-05,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7190118849277496,
+      "num_tokens": 203340792.0,
+      "step": 85070
+    },
+    {
+      "epoch": 0.7800495094893188,
+      "learning_rate": 4.399193178692583e-05,
+      "loss": 1.0117,
+      "mean_token_accuracy": 0.7057639479637146,
+      "num_tokens": 203373247.0,
+      "step": 85080
+    },
+    {
+      "epoch": 0.780141193728798,
+      "learning_rate": 4.397359493902998e-05,
+      "loss": 0.9147,
+      "mean_token_accuracy": 0.7353187441825867,
+      "num_tokens": 203403771.0,
+      "step": 85090
+    },
+    {
+      "epoch": 0.7802328779682772,
+      "learning_rate": 4.395525809113414e-05,
+      "loss": 0.8419,
+      "mean_token_accuracy": 0.7530937850475311,
+      "num_tokens": 203432165.0,
+      "step": 85100
+    },
+    {
+      "epoch": 0.7803245622077565,
+      "learning_rate": 4.393692124323829e-05,
+      "loss": 0.9221,
+      "mean_token_accuracy": 0.7253668546676636,
+      "num_tokens": 203463622.0,
+      "step": 85110
+    },
+    {
+      "epoch": 0.7804162464472357,
+      "learning_rate": 4.391858439534244e-05,
+      "loss": 0.9305,
+      "mean_token_accuracy": 0.7277561604976654,
+      "num_tokens": 203493975.0,
+      "step": 85120
+    },
+    {
+      "epoch": 0.780507930686715,
+      "learning_rate": 4.3900247547446595e-05,
+      "loss": 0.9299,
+      "mean_token_accuracy": 0.7265092074871063,
+      "num_tokens": 203526252.0,
+      "step": 85130
+    },
+    {
+      "epoch": 0.7805996149261942,
+      "learning_rate": 4.388191069955075e-05,
+      "loss": 0.9744,
+      "mean_token_accuracy": 0.7142821848392487,
+      "num_tokens": 203558265.0,
+      "step": 85140
+    },
+    {
+      "epoch": 0.7806912991656734,
+      "learning_rate": 4.3863573851654905e-05,
+      "loss": 0.9289,
+      "mean_token_accuracy": 0.7281145811080932,
+      "num_tokens": 203588600.0,
+      "step": 85150
+    },
+    {
+      "epoch": 0.7807829834051526,
+      "learning_rate": 4.384523700375906e-05,
+      "loss": 0.8988,
+      "mean_token_accuracy": 0.7322739660739899,
+      "num_tokens": 203618580.0,
+      "step": 85160
+    },
+    {
+      "epoch": 0.7808746676446319,
+      "learning_rate": 4.382690015586321e-05,
+      "loss": 0.9139,
+      "mean_token_accuracy": 0.732365733385086,
+      "num_tokens": 203648771.0,
+      "step": 85170
+    },
+    {
+      "epoch": 0.7809663518841111,
+      "learning_rate": 4.380856330796737e-05,
+      "loss": 0.8923,
+      "mean_token_accuracy": 0.7362757682800293,
+      "num_tokens": 203679033.0,
+      "step": 85180
+    },
+    {
+      "epoch": 0.7810580361235904,
+      "learning_rate": 4.379022646007151e-05,
+      "loss": 0.9599,
+      "mean_token_accuracy": 0.7195428729057312,
+      "num_tokens": 203709799.0,
+      "step": 85190
+    },
+    {
+      "epoch": 0.7811497203630696,
+      "learning_rate": 4.377188961217567e-05,
+      "loss": 0.9826,
+      "mean_token_accuracy": 0.7132642209529877,
+      "num_tokens": 203742326.0,
+      "step": 85200
+    },
+    {
+      "epoch": 0.7812414046025489,
+      "learning_rate": 4.375355276427982e-05,
+      "loss": 0.9505,
+      "mean_token_accuracy": 0.7204755544662476,
+      "num_tokens": 203772828.0,
+      "step": 85210
+    },
+    {
+      "epoch": 0.781333088842028,
+      "learning_rate": 4.3735215916383975e-05,
+      "loss": 0.925,
+      "mean_token_accuracy": 0.7311704814434051,
+      "num_tokens": 203803662.0,
+      "step": 85220
+    },
+    {
+      "epoch": 0.7814247730815073,
+      "learning_rate": 4.3716879068488134e-05,
+      "loss": 0.908,
+      "mean_token_accuracy": 0.730689388513565,
+      "num_tokens": 203835342.0,
+      "step": 85230
+    },
+    {
+      "epoch": 0.7815164573209865,
+      "learning_rate": 4.369854222059228e-05,
+      "loss": 0.9193,
+      "mean_token_accuracy": 0.7279924213886261,
+      "num_tokens": 203866841.0,
+      "step": 85240
+    },
+    {
+      "epoch": 0.7816081415604658,
+      "learning_rate": 4.368020537269644e-05,
+      "loss": 0.9362,
+      "mean_token_accuracy": 0.7220112979412079,
+      "num_tokens": 203898902.0,
+      "step": 85250
+    },
+    {
+      "epoch": 0.781699825799945,
+      "learning_rate": 4.366186852480059e-05,
+      "loss": 0.982,
+      "mean_token_accuracy": 0.7149511456489563,
+      "num_tokens": 203929119.0,
+      "step": 85260
+    },
+    {
+      "epoch": 0.7817915100394243,
+      "learning_rate": 4.364353167690474e-05,
+      "loss": 0.8408,
+      "mean_token_accuracy": 0.7517880320549011,
+      "num_tokens": 203957857.0,
+      "step": 85270
+    },
+    {
+      "epoch": 0.7818831942789034,
+      "learning_rate": 4.36251948290089e-05,
+      "loss": 0.8561,
+      "mean_token_accuracy": 0.7452545762062073,
+      "num_tokens": 203987995.0,
+      "step": 85280
+    },
+    {
+      "epoch": 0.7819748785183827,
+      "learning_rate": 4.3606857981113045e-05,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.7230924427509308,
+      "num_tokens": 204019828.0,
+      "step": 85290
+    },
+    {
+      "epoch": 0.7820665627578619,
+      "learning_rate": 4.3588521133217204e-05,
+      "loss": 0.9771,
+      "mean_token_accuracy": 0.7162556827068329,
+      "num_tokens": 204050518.0,
+      "step": 85300
+    },
+    {
+      "epoch": 0.7821582469973412,
+      "learning_rate": 4.3570184285321355e-05,
+      "loss": 0.9974,
+      "mean_token_accuracy": 0.7094563007354736,
+      "num_tokens": 204081752.0,
+      "step": 85310
+    },
+    {
+      "epoch": 0.7822499312368204,
+      "learning_rate": 4.355184743742551e-05,
+      "loss": 0.9511,
+      "mean_token_accuracy": 0.7223043501377105,
+      "num_tokens": 204113843.0,
+      "step": 85320
+    },
+    {
+      "epoch": 0.7823416154762997,
+      "learning_rate": 4.3533510589529666e-05,
+      "loss": 0.918,
+      "mean_token_accuracy": 0.7277221381664276,
+      "num_tokens": 204144549.0,
+      "step": 85330
+    },
+    {
+      "epoch": 0.7824332997157789,
+      "learning_rate": 4.351517374163381e-05,
+      "loss": 0.9252,
+      "mean_token_accuracy": 0.7253208816051483,
+      "num_tokens": 204174183.0,
+      "step": 85340
+    },
+    {
+      "epoch": 0.782524983955258,
+      "learning_rate": 4.349683689373797e-05,
+      "loss": 0.9265,
+      "mean_token_accuracy": 0.7277453958988189,
+      "num_tokens": 204203736.0,
+      "step": 85350
+    },
+    {
+      "epoch": 0.7826166681947373,
+      "learning_rate": 4.347850004584212e-05,
+      "loss": 0.9236,
+      "mean_token_accuracy": 0.7261824786663056,
+      "num_tokens": 204234407.0,
+      "step": 85360
+    },
+    {
+      "epoch": 0.7827083524342165,
+      "learning_rate": 4.3460163197946273e-05,
+      "loss": 1.0011,
+      "mean_token_accuracy": 0.7111550986766815,
+      "num_tokens": 204265568.0,
+      "step": 85370
+    },
+    {
+      "epoch": 0.7828000366736958,
+      "learning_rate": 4.344182635005043e-05,
+      "loss": 1.0047,
+      "mean_token_accuracy": 0.7094777584075928,
+      "num_tokens": 204297246.0,
+      "step": 85380
+    },
+    {
+      "epoch": 0.782891720913175,
+      "learning_rate": 4.342348950215458e-05,
+      "loss": 0.9529,
+      "mean_token_accuracy": 0.7222318828105927,
+      "num_tokens": 204328172.0,
+      "step": 85390
+    },
+    {
+      "epoch": 0.7829834051526543,
+      "learning_rate": 4.3405152654258736e-05,
+      "loss": 0.9411,
+      "mean_token_accuracy": 0.724736875295639,
+      "num_tokens": 204359965.0,
+      "step": 85400
+    },
+    {
+      "epoch": 0.7830750893921334,
+      "learning_rate": 4.338681580636289e-05,
+      "loss": 0.9049,
+      "mean_token_accuracy": 0.7343593955039978,
+      "num_tokens": 204388535.0,
+      "step": 85410
+    },
+    {
+      "epoch": 0.7831667736316127,
+      "learning_rate": 4.336847895846704e-05,
+      "loss": 0.9364,
+      "mean_token_accuracy": 0.7264476478099823,
+      "num_tokens": 204419646.0,
+      "step": 85420
+    },
+    {
+      "epoch": 0.7832584578710919,
+      "learning_rate": 4.33501421105712e-05,
+      "loss": 0.9928,
+      "mean_token_accuracy": 0.7078803420066834,
+      "num_tokens": 204451427.0,
+      "step": 85430
+    },
+    {
+      "epoch": 0.7833501421105712,
+      "learning_rate": 4.333180526267534e-05,
+      "loss": 0.9807,
+      "mean_token_accuracy": 0.7140648603439331,
+      "num_tokens": 204483315.0,
+      "step": 85440
+    },
+    {
+      "epoch": 0.7834418263500504,
+      "learning_rate": 4.33134684147795e-05,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7227704703807831,
+      "num_tokens": 204514192.0,
+      "step": 85450
+    },
+    {
+      "epoch": 0.7835335105895297,
+      "learning_rate": 4.3295131566883654e-05,
+      "loss": 0.9696,
+      "mean_token_accuracy": 0.7156088948249817,
+      "num_tokens": 204546443.0,
+      "step": 85460
+    },
+    {
+      "epoch": 0.7836251948290089,
+      "learning_rate": 4.3276794718987806e-05,
+      "loss": 0.9062,
+      "mean_token_accuracy": 0.7343481063842774,
+      "num_tokens": 204575418.0,
+      "step": 85470
+    },
+    {
+      "epoch": 0.7837168790684881,
+      "learning_rate": 4.3258457871091964e-05,
+      "loss": 0.9247,
+      "mean_token_accuracy": 0.7347521424293518,
+      "num_tokens": 204604497.0,
+      "step": 85480
+    },
+    {
+      "epoch": 0.7838085633079673,
+      "learning_rate": 4.3240121023196116e-05,
+      "loss": 0.9379,
+      "mean_token_accuracy": 0.7259818851947785,
+      "num_tokens": 204636027.0,
+      "step": 85490
+    },
+    {
+      "epoch": 0.7839002475474466,
+      "learning_rate": 4.322178417530027e-05,
+      "loss": 0.9598,
+      "mean_token_accuracy": 0.7194625914096833,
+      "num_tokens": 204665751.0,
+      "step": 85500
+    },
+    {
+      "epoch": 0.7839919317869258,
+      "learning_rate": 4.320344732740442e-05,
+      "loss": 0.9427,
+      "mean_token_accuracy": 0.7234639585018158,
+      "num_tokens": 204696619.0,
+      "step": 85510
+    },
+    {
+      "epoch": 0.7840836160264051,
+      "learning_rate": 4.318511047950857e-05,
+      "loss": 0.9258,
+      "mean_token_accuracy": 0.72845818400383,
+      "num_tokens": 204727234.0,
+      "step": 85520
+    },
+    {
+      "epoch": 0.7841753002658843,
+      "learning_rate": 4.316677363161273e-05,
+      "loss": 0.9738,
+      "mean_token_accuracy": 0.7165741324424744,
+      "num_tokens": 204758365.0,
+      "step": 85530
+    },
+    {
+      "epoch": 0.7842669845053635,
+      "learning_rate": 4.314843678371688e-05,
+      "loss": 0.9308,
+      "mean_token_accuracy": 0.7249511480331421,
+      "num_tokens": 204788072.0,
+      "step": 85540
+    },
+    {
+      "epoch": 0.7843586687448427,
+      "learning_rate": 4.3130099935821034e-05,
+      "loss": 0.9726,
+      "mean_token_accuracy": 0.7123758435249329,
+      "num_tokens": 204820049.0,
+      "step": 85550
+    },
+    {
+      "epoch": 0.784450352984322,
+      "learning_rate": 4.3111763087925186e-05,
+      "loss": 0.9901,
+      "mean_token_accuracy": 0.7127663850784302,
+      "num_tokens": 204852210.0,
+      "step": 85560
+    },
+    {
+      "epoch": 0.7845420372238012,
+      "learning_rate": 4.309342624002934e-05,
+      "loss": 0.9583,
+      "mean_token_accuracy": 0.7225412964820862,
+      "num_tokens": 204883826.0,
+      "step": 85570
+    },
+    {
+      "epoch": 0.7846337214632805,
+      "learning_rate": 4.3075089392133497e-05,
+      "loss": 0.9376,
+      "mean_token_accuracy": 0.7293244659900665,
+      "num_tokens": 204913908.0,
+      "step": 85580
+    },
+    {
+      "epoch": 0.7847254057027597,
+      "learning_rate": 4.305675254423765e-05,
+      "loss": 0.9075,
+      "mean_token_accuracy": 0.7326092720031738,
+      "num_tokens": 204943379.0,
+      "step": 85590
+    },
+    {
+      "epoch": 0.784817089942239,
+      "learning_rate": 4.30384156963418e-05,
+      "loss": 0.963,
+      "mean_token_accuracy": 0.7229960322380066,
+      "num_tokens": 204973014.0,
+      "step": 85600
+    },
+    {
+      "epoch": 0.7849087741817181,
+      "learning_rate": 4.302007884844595e-05,
+      "loss": 0.9614,
+      "mean_token_accuracy": 0.7152164518833161,
+      "num_tokens": 205003916.0,
+      "step": 85610
+    },
+    {
+      "epoch": 0.7850004584211974,
+      "learning_rate": 4.3001742000550104e-05,
+      "loss": 0.9744,
+      "mean_token_accuracy": 0.7146295487880707,
+      "num_tokens": 205035936.0,
+      "step": 85620
+    },
+    {
+      "epoch": 0.7850921426606766,
+      "learning_rate": 4.298340515265426e-05,
+      "loss": 0.8923,
+      "mean_token_accuracy": 0.7326450228691102,
+      "num_tokens": 205065906.0,
+      "step": 85630
+    },
+    {
+      "epoch": 0.7851838269001559,
+      "learning_rate": 4.2965068304758415e-05,
+      "loss": 0.9256,
+      "mean_token_accuracy": 0.7261049807071686,
+      "num_tokens": 205096946.0,
+      "step": 85640
+    },
+    {
+      "epoch": 0.7852755111396351,
+      "learning_rate": 4.2946731456862566e-05,
+      "loss": 0.9383,
+      "mean_token_accuracy": 0.7194655418395997,
+      "num_tokens": 205127826.0,
+      "step": 85650
+    },
+    {
+      "epoch": 0.7853671953791144,
+      "learning_rate": 4.292839460896672e-05,
+      "loss": 0.9784,
+      "mean_token_accuracy": 0.717509651184082,
+      "num_tokens": 205158790.0,
+      "step": 85660
+    },
+    {
+      "epoch": 0.7854588796185935,
+      "learning_rate": 4.291005776107088e-05,
+      "loss": 0.9821,
+      "mean_token_accuracy": 0.7088741362094879,
+      "num_tokens": 205190469.0,
+      "step": 85670
+    },
+    {
+      "epoch": 0.7855505638580728,
+      "learning_rate": 4.289172091317503e-05,
+      "loss": 0.9785,
+      "mean_token_accuracy": 0.7216758668422699,
+      "num_tokens": 205222866.0,
+      "step": 85680
+    },
+    {
+      "epoch": 0.785642248097552,
+      "learning_rate": 4.287338406527918e-05,
+      "loss": 0.9381,
+      "mean_token_accuracy": 0.7183633208274841,
+      "num_tokens": 205254301.0,
+      "step": 85690
+    },
+    {
+      "epoch": 0.7857339323370313,
+      "learning_rate": 4.285504721738333e-05,
+      "loss": 0.9383,
+      "mean_token_accuracy": 0.7200577795505524,
+      "num_tokens": 205285066.0,
+      "step": 85700
+    },
+    {
+      "epoch": 0.7858256165765105,
+      "learning_rate": 4.2836710369487484e-05,
+      "loss": 0.9244,
+      "mean_token_accuracy": 0.7277416586875916,
+      "num_tokens": 205315437.0,
+      "step": 85710
+    },
+    {
+      "epoch": 0.7859173008159898,
+      "learning_rate": 4.281837352159164e-05,
+      "loss": 0.9436,
+      "mean_token_accuracy": 0.7256674766540527,
+      "num_tokens": 205345801.0,
+      "step": 85720
+    },
+    {
+      "epoch": 0.786008985055469,
+      "learning_rate": 4.2800036673695795e-05,
+      "loss": 0.9864,
+      "mean_token_accuracy": 0.7110663473606109,
+      "num_tokens": 205376755.0,
+      "step": 85730
+    },
+    {
+      "epoch": 0.7861006692949482,
+      "learning_rate": 4.278169982579995e-05,
+      "loss": 0.94,
+      "mean_token_accuracy": 0.7245600104331971,
+      "num_tokens": 205408368.0,
+      "step": 85740
+    },
+    {
+      "epoch": 0.7861923535344274,
+      "learning_rate": 4.27633629779041e-05,
+      "loss": 0.9223,
+      "mean_token_accuracy": 0.7295901298522949,
+      "num_tokens": 205439240.0,
+      "step": 85750
+    },
+    {
+      "epoch": 0.7862840377739067,
+      "learning_rate": 4.274502613000825e-05,
+      "loss": 0.9529,
+      "mean_token_accuracy": 0.7168604910373688,
+      "num_tokens": 205470574.0,
+      "step": 85760
+    },
+    {
+      "epoch": 0.7863757220133859,
+      "learning_rate": 4.272668928211241e-05,
+      "loss": 0.9254,
+      "mean_token_accuracy": 0.7269407451152802,
+      "num_tokens": 205500238.0,
+      "step": 85770
+    },
+    {
+      "epoch": 0.7864674062528652,
+      "learning_rate": 4.270835243421656e-05,
+      "loss": 0.9354,
+      "mean_token_accuracy": 0.723615849018097,
+      "num_tokens": 205531274.0,
+      "step": 85780
+    },
+    {
+      "epoch": 0.7865590904923444,
+      "learning_rate": 4.269001558632071e-05,
+      "loss": 0.9792,
+      "mean_token_accuracy": 0.7170191884040833,
+      "num_tokens": 205561885.0,
+      "step": 85790
+    },
+    {
+      "epoch": 0.7866507747318237,
+      "learning_rate": 4.267167873842487e-05,
+      "loss": 1.0023,
+      "mean_token_accuracy": 0.7014797627925873,
+      "num_tokens": 205594519.0,
+      "step": 85800
+    },
+    {
+      "epoch": 0.7867424589713028,
+      "learning_rate": 4.265334189052902e-05,
+      "loss": 0.9237,
+      "mean_token_accuracy": 0.7250907301902771,
+      "num_tokens": 205624925.0,
+      "step": 85810
+    },
+    {
+      "epoch": 0.786834143210782,
+      "learning_rate": 4.2635005042633175e-05,
+      "loss": 0.9818,
+      "mean_token_accuracy": 0.718622499704361,
+      "num_tokens": 205655815.0,
+      "step": 85820
+    },
+    {
+      "epoch": 0.7869258274502613,
+      "learning_rate": 4.261666819473733e-05,
+      "loss": 0.9674,
+      "mean_token_accuracy": 0.7219990491867065,
+      "num_tokens": 205686521.0,
+      "step": 85830
+    },
+    {
+      "epoch": 0.7870175116897405,
+      "learning_rate": 4.259833134684148e-05,
+      "loss": 0.9835,
+      "mean_token_accuracy": 0.711795425415039,
+      "num_tokens": 205717221.0,
+      "step": 85840
+    },
+    {
+      "epoch": 0.7871091959292198,
+      "learning_rate": 4.257999449894564e-05,
+      "loss": 0.9802,
+      "mean_token_accuracy": 0.7074107944965362,
+      "num_tokens": 205748332.0,
+      "step": 85850
+    },
+    {
+      "epoch": 0.787200880168699,
+      "learning_rate": 4.256165765104978e-05,
+      "loss": 1.0081,
+      "mean_token_accuracy": 0.7061544060707092,
+      "num_tokens": 205779804.0,
+      "step": 85860
+    },
+    {
+      "epoch": 0.7872925644081782,
+      "learning_rate": 4.254332080315394e-05,
+      "loss": 0.9367,
+      "mean_token_accuracy": 0.7253042876720428,
+      "num_tokens": 205810867.0,
+      "step": 85870
+    },
+    {
+      "epoch": 0.7873842486476574,
+      "learning_rate": 4.252498395525809e-05,
+      "loss": 0.9277,
+      "mean_token_accuracy": 0.7289303541183472,
+      "num_tokens": 205841162.0,
+      "step": 85880
+    },
+    {
+      "epoch": 0.7874759328871367,
+      "learning_rate": 4.2506647107362245e-05,
+      "loss": 0.9194,
+      "mean_token_accuracy": 0.7317416131496429,
+      "num_tokens": 205871116.0,
+      "step": 85890
+    },
+    {
+      "epoch": 0.7875676171266159,
+      "learning_rate": 4.2488310259466404e-05,
+      "loss": 0.9376,
+      "mean_token_accuracy": 0.7267865717411042,
+      "num_tokens": 205902306.0,
+      "step": 85900
+    },
+    {
+      "epoch": 0.7876593013660952,
+      "learning_rate": 4.246997341157055e-05,
+      "loss": 1.0033,
+      "mean_token_accuracy": 0.710684472322464,
+      "num_tokens": 205933430.0,
+      "step": 85910
+    },
+    {
+      "epoch": 0.7877509856055744,
+      "learning_rate": 4.245163656367471e-05,
+      "loss": 0.9965,
+      "mean_token_accuracy": 0.7057823538780212,
+      "num_tokens": 205965521.0,
+      "step": 85920
+    },
+    {
+      "epoch": 0.7878426698450537,
+      "learning_rate": 4.243329971577886e-05,
+      "loss": 0.9651,
+      "mean_token_accuracy": 0.7150196135044098,
+      "num_tokens": 205996241.0,
+      "step": 85930
+    },
+    {
+      "epoch": 0.7879343540845328,
+      "learning_rate": 4.241496286788301e-05,
+      "loss": 0.913,
+      "mean_token_accuracy": 0.7292705357074738,
+      "num_tokens": 206028691.0,
+      "step": 85940
+    },
+    {
+      "epoch": 0.7880260383240121,
+      "learning_rate": 4.239662601998717e-05,
+      "loss": 0.9167,
+      "mean_token_accuracy": 0.727678781747818,
+      "num_tokens": 206058956.0,
+      "step": 85950
+    },
+    {
+      "epoch": 0.7881177225634913,
+      "learning_rate": 4.2378289172091315e-05,
+      "loss": 0.9074,
+      "mean_token_accuracy": 0.731557309627533,
+      "num_tokens": 206088534.0,
+      "step": 85960
+    },
+    {
+      "epoch": 0.7882094068029706,
+      "learning_rate": 4.2359952324195474e-05,
+      "loss": 0.9231,
+      "mean_token_accuracy": 0.7272307634353637,
+      "num_tokens": 206118561.0,
+      "step": 85970
+    },
+    {
+      "epoch": 0.7883010910424498,
+      "learning_rate": 4.2341615476299625e-05,
+      "loss": 0.9724,
+      "mean_token_accuracy": 0.7200353980064392,
+      "num_tokens": 206149601.0,
+      "step": 85980
+    },
+    {
+      "epoch": 0.7883927752819291,
+      "learning_rate": 4.232327862840378e-05,
+      "loss": 0.941,
+      "mean_token_accuracy": 0.7229769468307495,
+      "num_tokens": 206181182.0,
+      "step": 85990
+    },
+    {
+      "epoch": 0.7884844595214082,
+      "learning_rate": 4.2304941780507936e-05,
+      "loss": 0.9742,
+      "mean_token_accuracy": 0.7207611382007599,
+      "num_tokens": 206212657.0,
+      "step": 86000
+    },
+    {
+      "epoch": 0.7885761437608875,
+      "learning_rate": 4.228660493261208e-05,
+      "loss": 0.9226,
+      "mean_token_accuracy": 0.7291601181030274,
+      "num_tokens": 206242838.0,
+      "step": 86010
+    },
+    {
+      "epoch": 0.7886678280003667,
+      "learning_rate": 4.226826808471624e-05,
+      "loss": 0.9221,
+      "mean_token_accuracy": 0.7228618323802948,
+      "num_tokens": 206273413.0,
+      "step": 86020
+    },
+    {
+      "epoch": 0.788759512239846,
+      "learning_rate": 4.224993123682039e-05,
+      "loss": 0.9624,
+      "mean_token_accuracy": 0.7189518511295319,
+      "num_tokens": 206303787.0,
+      "step": 86030
+    },
+    {
+      "epoch": 0.7888511964793252,
+      "learning_rate": 4.2231594388924543e-05,
+      "loss": 1.0058,
+      "mean_token_accuracy": 0.7108011722564698,
+      "num_tokens": 206335862.0,
+      "step": 86040
+    },
+    {
+      "epoch": 0.7889428807188045,
+      "learning_rate": 4.22132575410287e-05,
+      "loss": 0.9642,
+      "mean_token_accuracy": 0.7218930304050446,
+      "num_tokens": 206367383.0,
+      "step": 86050
+    },
+    {
+      "epoch": 0.7890345649582837,
+      "learning_rate": 4.219492069313285e-05,
+      "loss": 0.9627,
+      "mean_token_accuracy": 0.7193548262119294,
+      "num_tokens": 206399338.0,
+      "step": 86060
+    },
+    {
+      "epoch": 0.7891262491977629,
+      "learning_rate": 4.2176583845237006e-05,
+      "loss": 0.9422,
+      "mean_token_accuracy": 0.7254627764225006,
+      "num_tokens": 206428629.0,
+      "step": 86070
+    },
+    {
+      "epoch": 0.7892179334372421,
+      "learning_rate": 4.215824699734116e-05,
+      "loss": 0.9064,
+      "mean_token_accuracy": 0.7301589667797088,
+      "num_tokens": 206459320.0,
+      "step": 86080
+    },
+    {
+      "epoch": 0.7893096176767214,
+      "learning_rate": 4.213991014944531e-05,
+      "loss": 1.0068,
+      "mean_token_accuracy": 0.7062439560890198,
+      "num_tokens": 206490337.0,
+      "step": 86090
+    },
+    {
+      "epoch": 0.7894013019162006,
+      "learning_rate": 4.212157330154947e-05,
+      "loss": 1.0443,
+      "mean_token_accuracy": 0.701622599363327,
+      "num_tokens": 206523471.0,
+      "step": 86100
+    },
+    {
+      "epoch": 0.7894929861556799,
+      "learning_rate": 4.210323645365362e-05,
+      "loss": 0.938,
+      "mean_token_accuracy": 0.7256350219249725,
+      "num_tokens": 206555000.0,
+      "step": 86110
+    },
+    {
+      "epoch": 0.7895846703951591,
+      "learning_rate": 4.208489960575777e-05,
+      "loss": 0.9443,
+      "mean_token_accuracy": 0.7241138577461242,
+      "num_tokens": 206586403.0,
+      "step": 86120
+    },
+    {
+      "epoch": 0.7896763546346383,
+      "learning_rate": 4.2066562757861924e-05,
+      "loss": 0.9196,
+      "mean_token_accuracy": 0.7292301297187805,
+      "num_tokens": 206617668.0,
+      "step": 86130
+    },
+    {
+      "epoch": 0.7897680388741175,
+      "learning_rate": 4.2048225909966076e-05,
+      "loss": 0.9188,
+      "mean_token_accuracy": 0.7290456116199493,
+      "num_tokens": 206647003.0,
+      "step": 86140
+    },
+    {
+      "epoch": 0.7898597231135968,
+      "learning_rate": 4.2029889062070234e-05,
+      "loss": 0.9503,
+      "mean_token_accuracy": 0.7188356459140778,
+      "num_tokens": 206677488.0,
+      "step": 86150
+    },
+    {
+      "epoch": 0.789951407353076,
+      "learning_rate": 4.2011552214174386e-05,
+      "loss": 1.0146,
+      "mean_token_accuracy": 0.7011561453342438,
+      "num_tokens": 206710237.0,
+      "step": 86160
+    },
+    {
+      "epoch": 0.7900430915925553,
+      "learning_rate": 4.199321536627854e-05,
+      "loss": 0.9456,
+      "mean_token_accuracy": 0.7251852512359619,
+      "num_tokens": 206741244.0,
+      "step": 86170
+    },
+    {
+      "epoch": 0.7901347758320345,
+      "learning_rate": 4.197487851838269e-05,
+      "loss": 0.9304,
+      "mean_token_accuracy": 0.726626968383789,
+      "num_tokens": 206771750.0,
+      "step": 86180
+    },
+    {
+      "epoch": 0.7902264600715138,
+      "learning_rate": 4.195654167048684e-05,
+      "loss": 0.9277,
+      "mean_token_accuracy": 0.7317323207855224,
+      "num_tokens": 206801113.0,
+      "step": 86190
+    },
+    {
+      "epoch": 0.7903181443109929,
+      "learning_rate": 4.1938204822591e-05,
+      "loss": 0.9484,
+      "mean_token_accuracy": 0.7219361066818237,
+      "num_tokens": 206831264.0,
+      "step": 86200
+    },
+    {
+      "epoch": 0.7904098285504721,
+      "learning_rate": 4.191986797469515e-05,
+      "loss": 0.985,
+      "mean_token_accuracy": 0.7095797717571258,
+      "num_tokens": 206861152.0,
+      "step": 86210
+    },
+    {
+      "epoch": 0.7905015127899514,
+      "learning_rate": 4.1901531126799304e-05,
+      "loss": 1.0338,
+      "mean_token_accuracy": 0.7008227586746216,
+      "num_tokens": 206894524.0,
+      "step": 86220
+    },
+    {
+      "epoch": 0.7905931970294306,
+      "learning_rate": 4.1883194278903456e-05,
+      "loss": 0.947,
+      "mean_token_accuracy": 0.7251901149749755,
+      "num_tokens": 206926614.0,
+      "step": 86230
+    },
+    {
+      "epoch": 0.7906848812689099,
+      "learning_rate": 4.186485743100761e-05,
+      "loss": 0.8958,
+      "mean_token_accuracy": 0.7344070792198181,
+      "num_tokens": 206956324.0,
+      "step": 86240
+    },
+    {
+      "epoch": 0.7907765655083892,
+      "learning_rate": 4.1846520583111767e-05,
+      "loss": 0.918,
+      "mean_token_accuracy": 0.7315007209777832,
+      "num_tokens": 206986553.0,
+      "step": 86250
+    },
+    {
+      "epoch": 0.7908682497478683,
+      "learning_rate": 4.182818373521592e-05,
+      "loss": 0.9576,
+      "mean_token_accuracy": 0.7177666246891021,
+      "num_tokens": 207015442.0,
+      "step": 86260
+    },
+    {
+      "epoch": 0.7909599339873475,
+      "learning_rate": 4.180984688732007e-05,
+      "loss": 0.9983,
+      "mean_token_accuracy": 0.7080096542835236,
+      "num_tokens": 207046148.0,
+      "step": 86270
+    },
+    {
+      "epoch": 0.7910516182268268,
+      "learning_rate": 4.179151003942422e-05,
+      "loss": 0.993,
+      "mean_token_accuracy": 0.7078933179378509,
+      "num_tokens": 207078683.0,
+      "step": 86280
+    },
+    {
+      "epoch": 0.791143302466306,
+      "learning_rate": 4.177317319152838e-05,
+      "loss": 0.9512,
+      "mean_token_accuracy": 0.7187549829483032,
+      "num_tokens": 207110629.0,
+      "step": 86290
+    },
+    {
+      "epoch": 0.7912349867057853,
+      "learning_rate": 4.175483634363253e-05,
+      "loss": 0.9383,
+      "mean_token_accuracy": 0.7207545340061188,
+      "num_tokens": 207142785.0,
+      "step": 86300
+    },
+    {
+      "epoch": 0.7913266709452645,
+      "learning_rate": 4.1736499495736685e-05,
+      "loss": 0.9432,
+      "mean_token_accuracy": 0.7170807123184204,
+      "num_tokens": 207174222.0,
+      "step": 86310
+    },
+    {
+      "epoch": 0.7914183551847438,
+      "learning_rate": 4.1718162647840836e-05,
+      "loss": 0.9318,
+      "mean_token_accuracy": 0.7256855189800262,
+      "num_tokens": 207203828.0,
+      "step": 86320
+    },
+    {
+      "epoch": 0.7915100394242229,
+      "learning_rate": 4.169982579994499e-05,
+      "loss": 0.9996,
+      "mean_token_accuracy": 0.7121340036392212,
+      "num_tokens": 207235423.0,
+      "step": 86330
+    },
+    {
+      "epoch": 0.7916017236637022,
+      "learning_rate": 4.168148895204915e-05,
+      "loss": 1.025,
+      "mean_token_accuracy": 0.7054444551467896,
+      "num_tokens": 207267939.0,
+      "step": 86340
+    },
+    {
+      "epoch": 0.7916934079031814,
+      "learning_rate": 4.16631521041533e-05,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.7115237712860107,
+      "num_tokens": 207299252.0,
+      "step": 86350
+    },
+    {
+      "epoch": 0.7917850921426607,
+      "learning_rate": 4.164481525625745e-05,
+      "loss": 0.9356,
+      "mean_token_accuracy": 0.7275644659996032,
+      "num_tokens": 207330027.0,
+      "step": 86360
+    },
+    {
+      "epoch": 0.7918767763821399,
+      "learning_rate": 4.16264784083616e-05,
+      "loss": 0.9645,
+      "mean_token_accuracy": 0.7202916145324707,
+      "num_tokens": 207360485.0,
+      "step": 86370
+    },
+    {
+      "epoch": 0.7919684606216192,
+      "learning_rate": 4.1608141560465754e-05,
+      "loss": 0.9338,
+      "mean_token_accuracy": 0.7204845726490021,
+      "num_tokens": 207390182.0,
+      "step": 86380
+    },
+    {
+      "epoch": 0.7920601448610983,
+      "learning_rate": 4.158980471256991e-05,
+      "loss": 0.9812,
+      "mean_token_accuracy": 0.7126319408416748,
+      "num_tokens": 207421621.0,
+      "step": 86390
+    },
+    {
+      "epoch": 0.7921518291005776,
+      "learning_rate": 4.1571467864674065e-05,
+      "loss": 0.982,
+      "mean_token_accuracy": 0.7128823220729827,
+      "num_tokens": 207452915.0,
+      "step": 86400
+    },
+    {
+      "epoch": 0.7922435133400568,
+      "learning_rate": 4.155313101677822e-05,
+      "loss": 0.9098,
+      "mean_token_accuracy": 0.7332341194152832,
+      "num_tokens": 207485097.0,
+      "step": 86410
+    },
+    {
+      "epoch": 0.7923351975795361,
+      "learning_rate": 4.1534794168882375e-05,
+      "loss": 0.9255,
+      "mean_token_accuracy": 0.726281863451004,
+      "num_tokens": 207516399.0,
+      "step": 86420
+    },
+    {
+      "epoch": 0.7924268818190153,
+      "learning_rate": 4.151645732098652e-05,
+      "loss": 0.9435,
+      "mean_token_accuracy": 0.7231832683086395,
+      "num_tokens": 207545822.0,
+      "step": 86430
+    },
+    {
+      "epoch": 0.7925185660584946,
+      "learning_rate": 4.149812047309068e-05,
+      "loss": 0.9137,
+      "mean_token_accuracy": 0.7281588971614837,
+      "num_tokens": 207577365.0,
+      "step": 86440
+    },
+    {
+      "epoch": 0.7926102502979738,
+      "learning_rate": 4.147978362519483e-05,
+      "loss": 0.9902,
+      "mean_token_accuracy": 0.7107552766799927,
+      "num_tokens": 207609490.0,
+      "step": 86450
+    },
+    {
+      "epoch": 0.792701934537453,
+      "learning_rate": 4.146144677729898e-05,
+      "loss": 1.0435,
+      "mean_token_accuracy": 0.6976045489311218,
+      "num_tokens": 207641595.0,
+      "step": 86460
+    },
+    {
+      "epoch": 0.7927936187769322,
+      "learning_rate": 4.144310992940314e-05,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7304158270359039,
+      "num_tokens": 207672793.0,
+      "step": 86470
+    },
+    {
+      "epoch": 0.7928853030164115,
+      "learning_rate": 4.142477308150729e-05,
+      "loss": 0.9302,
+      "mean_token_accuracy": 0.7257893681526184,
+      "num_tokens": 207704947.0,
+      "step": 86480
+    },
+    {
+      "epoch": 0.7929769872558907,
+      "learning_rate": 4.1406436233611445e-05,
+      "loss": 0.8942,
+      "mean_token_accuracy": 0.7341605126857758,
+      "num_tokens": 207732974.0,
+      "step": 86490
+    },
+    {
+      "epoch": 0.79306867149537,
+      "learning_rate": 4.13880993857156e-05,
+      "loss": 0.9159,
+      "mean_token_accuracy": 0.7367580354213714,
+      "num_tokens": 207764858.0,
+      "step": 86500
+    },
+    {
+      "epoch": 0.7931603557348492,
+      "learning_rate": 4.136976253781975e-05,
+      "loss": 0.9533,
+      "mean_token_accuracy": 0.7214776396751403,
+      "num_tokens": 207796150.0,
+      "step": 86510
+    },
+    {
+      "epoch": 0.7932520399743284,
+      "learning_rate": 4.135142568992391e-05,
+      "loss": 1.0098,
+      "mean_token_accuracy": 0.7000498592853546,
+      "num_tokens": 207828544.0,
+      "step": 86520
+    },
+    {
+      "epoch": 0.7933437242138076,
+      "learning_rate": 4.133308884202805e-05,
+      "loss": 0.9567,
+      "mean_token_accuracy": 0.7159042656421661,
+      "num_tokens": 207860620.0,
+      "step": 86530
+    },
+    {
+      "epoch": 0.7934354084532869,
+      "learning_rate": 4.131475199413221e-05,
+      "loss": 0.9484,
+      "mean_token_accuracy": 0.7242303311824798,
+      "num_tokens": 207891856.0,
+      "step": 86540
+    },
+    {
+      "epoch": 0.7935270926927661,
+      "learning_rate": 4.129641514623636e-05,
+      "loss": 0.9261,
+      "mean_token_accuracy": 0.7282198309898377,
+      "num_tokens": 207923777.0,
+      "step": 86550
+    },
+    {
+      "epoch": 0.7936187769322454,
+      "learning_rate": 4.1278078298340515e-05,
+      "loss": 0.9429,
+      "mean_token_accuracy": 0.7287190020084381,
+      "num_tokens": 207952121.0,
+      "step": 86560
+    },
+    {
+      "epoch": 0.7937104611717246,
+      "learning_rate": 4.1259741450444674e-05,
+      "loss": 0.9804,
+      "mean_token_accuracy": 0.7127610504627228,
+      "num_tokens": 207983744.0,
+      "step": 86570
+    },
+    {
+      "epoch": 0.7938021454112039,
+      "learning_rate": 4.124140460254882e-05,
+      "loss": 1.0303,
+      "mean_token_accuracy": 0.7046155571937561,
+      "num_tokens": 208016574.0,
+      "step": 86580
+    },
+    {
+      "epoch": 0.793893829650683,
+      "learning_rate": 4.122306775465298e-05,
+      "loss": 0.9377,
+      "mean_token_accuracy": 0.7199788987636566,
+      "num_tokens": 208048233.0,
+      "step": 86590
+    },
+    {
+      "epoch": 0.7939855138901623,
+      "learning_rate": 4.1204730906757136e-05,
+      "loss": 0.9388,
+      "mean_token_accuracy": 0.7236958265304565,
+      "num_tokens": 208078638.0,
+      "step": 86600
+    },
+    {
+      "epoch": 0.7940771981296415,
+      "learning_rate": 4.118639405886128e-05,
+      "loss": 0.9066,
+      "mean_token_accuracy": 0.7310633838176728,
+      "num_tokens": 208109993.0,
+      "step": 86610
+    },
+    {
+      "epoch": 0.7941688823691208,
+      "learning_rate": 4.116805721096544e-05,
+      "loss": 0.9464,
+      "mean_token_accuracy": 0.7216531753540039,
+      "num_tokens": 208140842.0,
+      "step": 86620
+    },
+    {
+      "epoch": 0.7942605666086,
+      "learning_rate": 4.1149720363069585e-05,
+      "loss": 1.0227,
+      "mean_token_accuracy": 0.7035574436187744,
+      "num_tokens": 208171321.0,
+      "step": 86630
+    },
+    {
+      "epoch": 0.7943522508480793,
+      "learning_rate": 4.1131383515173744e-05,
+      "loss": 0.958,
+      "mean_token_accuracy": 0.7149340093135834,
+      "num_tokens": 208202668.0,
+      "step": 86640
+    },
+    {
+      "epoch": 0.7944439350875584,
+      "learning_rate": 4.11130466672779e-05,
+      "loss": 0.9444,
+      "mean_token_accuracy": 0.7256955981254578,
+      "num_tokens": 208234024.0,
+      "step": 86650
+    },
+    {
+      "epoch": 0.7945356193270376,
+      "learning_rate": 4.109470981938205e-05,
+      "loss": 0.9552,
+      "mean_token_accuracy": 0.721126115322113,
+      "num_tokens": 208265638.0,
+      "step": 86660
+    },
+    {
+      "epoch": 0.7946273035665169,
+      "learning_rate": 4.1076372971486206e-05,
+      "loss": 0.9349,
+      "mean_token_accuracy": 0.7214091718196869,
+      "num_tokens": 208295256.0,
+      "step": 86670
+    },
+    {
+      "epoch": 0.7947189878059961,
+      "learning_rate": 4.105803612359035e-05,
+      "loss": 0.9303,
+      "mean_token_accuracy": 0.7256463825702667,
+      "num_tokens": 208324026.0,
+      "step": 86680
+    },
+    {
+      "epoch": 0.7948106720454754,
+      "learning_rate": 4.103969927569451e-05,
+      "loss": 0.9361,
+      "mean_token_accuracy": 0.7245767474174499,
+      "num_tokens": 208354238.0,
+      "step": 86690
+    },
+    {
+      "epoch": 0.7949023562849546,
+      "learning_rate": 4.102136242779867e-05,
+      "loss": 1.0113,
+      "mean_token_accuracy": 0.7028233170509338,
+      "num_tokens": 208387109.0,
+      "step": 86700
+    },
+    {
+      "epoch": 0.7949940405244339,
+      "learning_rate": 4.1003025579902813e-05,
+      "loss": 0.9399,
+      "mean_token_accuracy": 0.7222343683242798,
+      "num_tokens": 208419858.0,
+      "step": 86710
+    },
+    {
+      "epoch": 0.795085724763913,
+      "learning_rate": 4.098468873200697e-05,
+      "loss": 0.9351,
+      "mean_token_accuracy": 0.7292609870433807,
+      "num_tokens": 208450896.0,
+      "step": 86720
+    },
+    {
+      "epoch": 0.7951774090033923,
+      "learning_rate": 4.0966351884111124e-05,
+      "loss": 0.9541,
+      "mean_token_accuracy": 0.7215969681739807,
+      "num_tokens": 208481716.0,
+      "step": 86730
+    },
+    {
+      "epoch": 0.7952690932428715,
+      "learning_rate": 4.0948015036215276e-05,
+      "loss": 0.9366,
+      "mean_token_accuracy": 0.7248545944690704,
+      "num_tokens": 208511158.0,
+      "step": 86740
+    },
+    {
+      "epoch": 0.7953607774823508,
+      "learning_rate": 4.0929678188319434e-05,
+      "loss": 0.9853,
+      "mean_token_accuracy": 0.7170604169368744,
+      "num_tokens": 208542521.0,
+      "step": 86750
+    },
+    {
+      "epoch": 0.79545246172183,
+      "learning_rate": 4.091134134042358e-05,
+      "loss": 0.9998,
+      "mean_token_accuracy": 0.7087662935256958,
+      "num_tokens": 208574677.0,
+      "step": 86760
+    },
+    {
+      "epoch": 0.7955441459613093,
+      "learning_rate": 4.089300449252774e-05,
+      "loss": 0.945,
+      "mean_token_accuracy": 0.7175221562385559,
+      "num_tokens": 208606521.0,
+      "step": 86770
+    },
+    {
+      "epoch": 0.7956358302007884,
+      "learning_rate": 4.087466764463189e-05,
+      "loss": 0.9327,
+      "mean_token_accuracy": 0.7264738619327545,
+      "num_tokens": 208637324.0,
+      "step": 86780
+    },
+    {
+      "epoch": 0.7957275144402677,
+      "learning_rate": 4.085633079673604e-05,
+      "loss": 0.9405,
+      "mean_token_accuracy": 0.7284891366958618,
+      "num_tokens": 208667218.0,
+      "step": 86790
+    },
+    {
+      "epoch": 0.7958191986797469,
+      "learning_rate": 4.08379939488402e-05,
+      "loss": 0.9122,
+      "mean_token_accuracy": 0.7302503824234009,
+      "num_tokens": 208695456.0,
+      "step": 86800
+    },
+    {
+      "epoch": 0.7959108829192262,
+      "learning_rate": 4.0819657100944346e-05,
+      "loss": 0.964,
+      "mean_token_accuracy": 0.7192913830280304,
+      "num_tokens": 208727231.0,
+      "step": 86810
+    },
+    {
+      "epoch": 0.7960025671587054,
+      "learning_rate": 4.0801320253048504e-05,
+      "loss": 1.0259,
+      "mean_token_accuracy": 0.6997593402862549,
+      "num_tokens": 208759898.0,
+      "step": 86820
+    },
+    {
+      "epoch": 0.7960942513981847,
+      "learning_rate": 4.0782983405152656e-05,
+      "loss": 0.9517,
+      "mean_token_accuracy": 0.7208040475845336,
+      "num_tokens": 208790664.0,
+      "step": 86830
+    },
+    {
+      "epoch": 0.7961859356376639,
+      "learning_rate": 4.076464655725681e-05,
+      "loss": 0.9211,
+      "mean_token_accuracy": 0.7288584172725677,
+      "num_tokens": 208821993.0,
+      "step": 86840
+    },
+    {
+      "epoch": 0.7962776198771431,
+      "learning_rate": 4.074630970936097e-05,
+      "loss": 0.9069,
+      "mean_token_accuracy": 0.7319348752498627,
+      "num_tokens": 208853251.0,
+      "step": 86850
+    },
+    {
+      "epoch": 0.7963693041166223,
+      "learning_rate": 4.072797286146511e-05,
+      "loss": 0.9621,
+      "mean_token_accuracy": 0.7171362936496735,
+      "num_tokens": 208884437.0,
+      "step": 86860
+    },
+    {
+      "epoch": 0.7964609883561016,
+      "learning_rate": 4.070963601356927e-05,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7199141204357147,
+      "num_tokens": 208915128.0,
+      "step": 86870
+    },
+    {
+      "epoch": 0.7965526725955808,
+      "learning_rate": 4.069129916567342e-05,
+      "loss": 1.0149,
+      "mean_token_accuracy": 0.7024094223976135,
+      "num_tokens": 208947548.0,
+      "step": 86880
+    },
+    {
+      "epoch": 0.7966443568350601,
+      "learning_rate": 4.0672962317777574e-05,
+      "loss": 0.9352,
+      "mean_token_accuracy": 0.723261296749115,
+      "num_tokens": 208979031.0,
+      "step": 86890
+    },
+    {
+      "epoch": 0.7967360410745393,
+      "learning_rate": 4.065462546988173e-05,
+      "loss": 0.9509,
+      "mean_token_accuracy": 0.7243140757083892,
+      "num_tokens": 209009123.0,
+      "step": 86900
+    },
+    {
+      "epoch": 0.7968277253140186,
+      "learning_rate": 4.0636288621985885e-05,
+      "loss": 0.963,
+      "mean_token_accuracy": 0.72268648147583,
+      "num_tokens": 209039210.0,
+      "step": 86910
+    },
+    {
+      "epoch": 0.7969194095534977,
+      "learning_rate": 4.0617951774090037e-05,
+      "loss": 0.9351,
+      "mean_token_accuracy": 0.7223830938339233,
+      "num_tokens": 209069237.0,
+      "step": 86920
+    },
+    {
+      "epoch": 0.797011093792977,
+      "learning_rate": 4.059961492619419e-05,
+      "loss": 0.9747,
+      "mean_token_accuracy": 0.7122148394584655,
+      "num_tokens": 209100369.0,
+      "step": 86930
+    },
+    {
+      "epoch": 0.7971027780324562,
+      "learning_rate": 4.058127807829834e-05,
+      "loss": 0.9923,
+      "mean_token_accuracy": 0.7069045722484588,
+      "num_tokens": 209133037.0,
+      "step": 86940
+    },
+    {
+      "epoch": 0.7971944622719355,
+      "learning_rate": 4.05629412304025e-05,
+      "loss": 0.902,
+      "mean_token_accuracy": 0.7337145149707794,
+      "num_tokens": 209165006.0,
+      "step": 86950
+    },
+    {
+      "epoch": 0.7972861465114147,
+      "learning_rate": 4.054460438250665e-05,
+      "loss": 0.9339,
+      "mean_token_accuracy": 0.7205852448940278,
+      "num_tokens": 209196496.0,
+      "step": 86960
+    },
+    {
+      "epoch": 0.797377830750894,
+      "learning_rate": 4.05262675346108e-05,
+      "loss": 0.9287,
+      "mean_token_accuracy": 0.7206258594989776,
+      "num_tokens": 209226131.0,
+      "step": 86970
+    },
+    {
+      "epoch": 0.7974695149903731,
+      "learning_rate": 4.0507930686714955e-05,
+      "loss": 0.9249,
+      "mean_token_accuracy": 0.7250255346298218,
+      "num_tokens": 209255177.0,
+      "step": 86980
+    },
+    {
+      "epoch": 0.7975611992298524,
+      "learning_rate": 4.0489593838819106e-05,
+      "loss": 0.9618,
+      "mean_token_accuracy": 0.7173052787780761,
+      "num_tokens": 209285005.0,
+      "step": 86990
+    },
+    {
+      "epoch": 0.7976528834693316,
+      "learning_rate": 4.0471256990923265e-05,
+      "loss": 1.0192,
+      "mean_token_accuracy": 0.7025117516517639,
+      "num_tokens": 209316683.0,
+      "step": 87000
+    },
+    {
+      "epoch": 0.7977445677088109,
+      "learning_rate": 4.045292014302742e-05,
+      "loss": 0.9388,
+      "mean_token_accuracy": 0.7230490922927857,
+      "num_tokens": 209348067.0,
+      "step": 87010
+    },
+    {
+      "epoch": 0.7978362519482901,
+      "learning_rate": 4.043458329513157e-05,
+      "loss": 0.9596,
+      "mean_token_accuracy": 0.7197664737701416,
+      "num_tokens": 209380138.0,
+      "step": 87020
+    },
+    {
+      "epoch": 0.7979279361877694,
+      "learning_rate": 4.041624644723572e-05,
+      "loss": 0.9374,
+      "mean_token_accuracy": 0.7165299355983734,
+      "num_tokens": 209410731.0,
+      "step": 87030
+    },
+    {
+      "epoch": 0.7980196204272486,
+      "learning_rate": 4.039790959933988e-05,
+      "loss": 0.9274,
+      "mean_token_accuracy": 0.7212114334106445,
+      "num_tokens": 209440554.0,
+      "step": 87040
+    },
+    {
+      "epoch": 0.7981113046667277,
+      "learning_rate": 4.037957275144403e-05,
+      "loss": 0.9596,
+      "mean_token_accuracy": 0.7149106860160828,
+      "num_tokens": 209470542.0,
+      "step": 87050
+    },
+    {
+      "epoch": 0.798202988906207,
+      "learning_rate": 4.036123590354818e-05,
+      "loss": 1.0066,
+      "mean_token_accuracy": 0.7041074275970459,
+      "num_tokens": 209502895.0,
+      "step": 87060
+    },
+    {
+      "epoch": 0.7982946731456863,
+      "learning_rate": 4.0342899055652335e-05,
+      "loss": 0.9236,
+      "mean_token_accuracy": 0.7260433256626129,
+      "num_tokens": 209534663.0,
+      "step": 87070
+    },
+    {
+      "epoch": 0.7983863573851655,
+      "learning_rate": 4.032456220775649e-05,
+      "loss": 0.9451,
+      "mean_token_accuracy": 0.7250676095485687,
+      "num_tokens": 209565168.0,
+      "step": 87080
+    },
+    {
+      "epoch": 0.7984780416246448,
+      "learning_rate": 4.0306225359860645e-05,
+      "loss": 0.9323,
+      "mean_token_accuracy": 0.7251429498195648,
+      "num_tokens": 209595188.0,
+      "step": 87090
+    },
+    {
+      "epoch": 0.798569725864124,
+      "learning_rate": 4.02878885119648e-05,
+      "loss": 0.9534,
+      "mean_token_accuracy": 0.7167445003986359,
+      "num_tokens": 209626060.0,
+      "step": 87100
+    },
+    {
+      "epoch": 0.7986614101036031,
+      "learning_rate": 4.026955166406895e-05,
+      "loss": 0.9482,
+      "mean_token_accuracy": 0.7180288314819336,
+      "num_tokens": 209657260.0,
+      "step": 87110
+    },
+    {
+      "epoch": 0.7987530943430824,
+      "learning_rate": 4.02512148161731e-05,
+      "loss": 0.9954,
+      "mean_token_accuracy": 0.708745276927948,
+      "num_tokens": 209689358.0,
+      "step": 87120
+    },
+    {
+      "epoch": 0.7988447785825616,
+      "learning_rate": 4.023287796827725e-05,
+      "loss": 0.916,
+      "mean_token_accuracy": 0.7291933119297027,
+      "num_tokens": 209720348.0,
+      "step": 87130
+    },
+    {
+      "epoch": 0.7989364628220409,
+      "learning_rate": 4.021454112038141e-05,
+      "loss": 0.9235,
+      "mean_token_accuracy": 0.7270193874835968,
+      "num_tokens": 209749593.0,
+      "step": 87140
+    },
+    {
+      "epoch": 0.7990281470615201,
+      "learning_rate": 4.019620427248556e-05,
+      "loss": 0.9296,
+      "mean_token_accuracy": 0.7201617002487183,
+      "num_tokens": 209782138.0,
+      "step": 87150
+    },
+    {
+      "epoch": 0.7991198313009994,
+      "learning_rate": 4.0177867424589715e-05,
+      "loss": 0.8967,
+      "mean_token_accuracy": 0.7318171620368957,
+      "num_tokens": 209811050.0,
+      "step": 87160
+    },
+    {
+      "epoch": 0.7992115155404786,
+      "learning_rate": 4.015953057669387e-05,
+      "loss": 0.9692,
+      "mean_token_accuracy": 0.7147955060005188,
+      "num_tokens": 209842641.0,
+      "step": 87170
+    },
+    {
+      "epoch": 0.7993031997799578,
+      "learning_rate": 4.014119372879802e-05,
+      "loss": 1.0144,
+      "mean_token_accuracy": 0.7072067379951477,
+      "num_tokens": 209875507.0,
+      "step": 87180
+    },
+    {
+      "epoch": 0.799394884019437,
+      "learning_rate": 4.012285688090218e-05,
+      "loss": 0.9469,
+      "mean_token_accuracy": 0.7181179940700531,
+      "num_tokens": 209907302.0,
+      "step": 87190
+    },
+    {
+      "epoch": 0.7994865682589163,
+      "learning_rate": 4.010452003300633e-05,
+      "loss": 0.9515,
+      "mean_token_accuracy": 0.7170276045799255,
+      "num_tokens": 209938370.0,
+      "step": 87200
+    },
+    {
+      "epoch": 0.7995782524983955,
+      "learning_rate": 4.008618318511048e-05,
+      "loss": 0.9073,
+      "mean_token_accuracy": 0.7291929602622986,
+      "num_tokens": 209967429.0,
+      "step": 87210
+    },
+    {
+      "epoch": 0.7996699367378748,
+      "learning_rate": 4.006784633721464e-05,
+      "loss": 0.9496,
+      "mean_token_accuracy": 0.7195300698280335,
+      "num_tokens": 209996925.0,
+      "step": 87220
+    },
+    {
+      "epoch": 0.799761620977354,
+      "learning_rate": 4.0049509489318785e-05,
+      "loss": 0.9654,
+      "mean_token_accuracy": 0.7161861419677734,
+      "num_tokens": 210028490.0,
+      "step": 87230
+    },
+    {
+      "epoch": 0.7998533052168332,
+      "learning_rate": 4.0031172641422944e-05,
+      "loss": 1.0231,
+      "mean_token_accuracy": 0.6989761531352997,
+      "num_tokens": 210060393.0,
+      "step": 87240
+    },
+    {
+      "epoch": 0.7999449894563124,
+      "learning_rate": 4.0012835793527096e-05,
+      "loss": 0.9493,
+      "mean_token_accuracy": 0.7177074491977692,
+      "num_tokens": 210091680.0,
+      "step": 87250
+    },
+    {
+      "epoch": 0.8000366736957917,
+      "learning_rate": 3.999449894563125e-05,
+      "loss": 0.9206,
+      "mean_token_accuracy": 0.7272840738296509,
+      "num_tokens": 210123428.0,
+      "step": 87260
+    },
+    {
+      "epoch": 0.8001283579352709,
+      "learning_rate": 3.9976162097735406e-05,
+      "loss": 0.9429,
+      "mean_token_accuracy": 0.7239596605300903,
+      "num_tokens": 210153951.0,
+      "step": 87270
+    },
+    {
+      "epoch": 0.8002200421747502,
+      "learning_rate": 3.995782524983955e-05,
+      "loss": 0.9122,
+      "mean_token_accuracy": 0.7328599035739899,
+      "num_tokens": 210183418.0,
+      "step": 87280
+    },
+    {
+      "epoch": 0.8003117264142294,
+      "learning_rate": 3.993948840194371e-05,
+      "loss": 0.9814,
+      "mean_token_accuracy": 0.7061148047447204,
+      "num_tokens": 210212876.0,
+      "step": 87290
+    },
+    {
+      "epoch": 0.8004034106537087,
+      "learning_rate": 3.992115155404786e-05,
+      "loss": 0.9947,
+      "mean_token_accuracy": 0.7061928927898407,
+      "num_tokens": 210244168.0,
+      "step": 87300
+    },
+    {
+      "epoch": 0.8004950948931878,
+      "learning_rate": 3.9902814706152014e-05,
+      "loss": 0.9538,
+      "mean_token_accuracy": 0.7256619215011597,
+      "num_tokens": 210275609.0,
+      "step": 87310
+    },
+    {
+      "epoch": 0.8005867791326671,
+      "learning_rate": 3.988447785825617e-05,
+      "loss": 0.9519,
+      "mean_token_accuracy": 0.7208737552165985,
+      "num_tokens": 210306174.0,
+      "step": 87320
+    },
+    {
+      "epoch": 0.8006784633721463,
+      "learning_rate": 3.986614101036032e-05,
+      "loss": 0.9393,
+      "mean_token_accuracy": 0.7272839725017548,
+      "num_tokens": 210337464.0,
+      "step": 87330
+    },
+    {
+      "epoch": 0.8007701476116256,
+      "learning_rate": 3.9847804162464476e-05,
+      "loss": 0.9365,
+      "mean_token_accuracy": 0.7243816077709198,
+      "num_tokens": 210365979.0,
+      "step": 87340
+    },
+    {
+      "epoch": 0.8008618318511048,
+      "learning_rate": 3.982946731456863e-05,
+      "loss": 0.9699,
+      "mean_token_accuracy": 0.7135676264762878,
+      "num_tokens": 210396758.0,
+      "step": 87350
+    },
+    {
+      "epoch": 0.8009535160905841,
+      "learning_rate": 3.981113046667278e-05,
+      "loss": 1.0267,
+      "mean_token_accuracy": 0.6961562693119049,
+      "num_tokens": 210428116.0,
+      "step": 87360
+    },
+    {
+      "epoch": 0.8010452003300632,
+      "learning_rate": 3.979279361877694e-05,
+      "loss": 0.9412,
+      "mean_token_accuracy": 0.7198286354541779,
+      "num_tokens": 210459124.0,
+      "step": 87370
+    },
+    {
+      "epoch": 0.8011368845695425,
+      "learning_rate": 3.9774456770881083e-05,
+      "loss": 0.9239,
+      "mean_token_accuracy": 0.7228979408740998,
+      "num_tokens": 210489292.0,
+      "step": 87380
+    },
+    {
+      "epoch": 0.8012285688090217,
+      "learning_rate": 3.975611992298524e-05,
+      "loss": 0.9269,
+      "mean_token_accuracy": 0.7259806156158447,
+      "num_tokens": 210518533.0,
+      "step": 87390
+    },
+    {
+      "epoch": 0.801320253048501,
+      "learning_rate": 3.9737783075089394e-05,
+      "loss": 0.948,
+      "mean_token_accuracy": 0.7187861204147339,
+      "num_tokens": 210548130.0,
+      "step": 87400
+    },
+    {
+      "epoch": 0.8014119372879802,
+      "learning_rate": 3.9719446227193546e-05,
+      "loss": 0.976,
+      "mean_token_accuracy": 0.7104677975177764,
+      "num_tokens": 210578522.0,
+      "step": 87410
+    },
+    {
+      "epoch": 0.8015036215274595,
+      "learning_rate": 3.9701109379297704e-05,
+      "loss": 1.0142,
+      "mean_token_accuracy": 0.7038933455944061,
+      "num_tokens": 210610676.0,
+      "step": 87420
+    },
+    {
+      "epoch": 0.8015953057669387,
+      "learning_rate": 3.968277253140185e-05,
+      "loss": 0.9731,
+      "mean_token_accuracy": 0.7116360366344452,
+      "num_tokens": 210641489.0,
+      "step": 87430
+    },
+    {
+      "epoch": 0.8016869900064179,
+      "learning_rate": 3.966443568350601e-05,
+      "loss": 0.9531,
+      "mean_token_accuracy": 0.7212050378322601,
+      "num_tokens": 210672789.0,
+      "step": 87440
+    },
+    {
+      "epoch": 0.8017786742458971,
+      "learning_rate": 3.964609883561016e-05,
+      "loss": 0.9566,
+      "mean_token_accuracy": 0.7202536642551423,
+      "num_tokens": 210702757.0,
+      "step": 87450
+    },
+    {
+      "epoch": 0.8018703584853764,
+      "learning_rate": 3.962776198771431e-05,
+      "loss": 0.9312,
+      "mean_token_accuracy": 0.7226182639598846,
+      "num_tokens": 210733530.0,
+      "step": 87460
+    },
+    {
+      "epoch": 0.8019620427248556,
+      "learning_rate": 3.960942513981847e-05,
+      "loss": 0.9909,
+      "mean_token_accuracy": 0.7056439638137817,
+      "num_tokens": 210764589.0,
+      "step": 87470
+    },
+    {
+      "epoch": 0.8020537269643349,
+      "learning_rate": 3.9591088291922616e-05,
+      "loss": 0.9983,
+      "mean_token_accuracy": 0.707426768541336,
+      "num_tokens": 210796035.0,
+      "step": 87480
+    },
+    {
+      "epoch": 0.8021454112038141,
+      "learning_rate": 3.9572751444026774e-05,
+      "loss": 0.9138,
+      "mean_token_accuracy": 0.7283466637134552,
+      "num_tokens": 210827490.0,
+      "step": 87490
+    },
+    {
+      "epoch": 0.8022370954432932,
+      "learning_rate": 3.9554414596130926e-05,
+      "loss": 0.9406,
+      "mean_token_accuracy": 0.7218400359153747,
+      "num_tokens": 210859404.0,
+      "step": 87500
+    },
+    {
+      "epoch": 0.8023287796827725,
+      "learning_rate": 3.953607774823508e-05,
+      "loss": 0.9397,
+      "mean_token_accuracy": 0.7297419905662537,
+      "num_tokens": 210888658.0,
+      "step": 87510
+    },
+    {
+      "epoch": 0.8024204639222517,
+      "learning_rate": 3.951774090033924e-05,
+      "loss": 0.9209,
+      "mean_token_accuracy": 0.72423837184906,
+      "num_tokens": 210919037.0,
+      "step": 87520
+    },
+    {
+      "epoch": 0.802512148161731,
+      "learning_rate": 3.949940405244339e-05,
+      "loss": 0.9985,
+      "mean_token_accuracy": 0.7135941743850708,
+      "num_tokens": 210949611.0,
+      "step": 87530
+    },
+    {
+      "epoch": 0.8026038324012102,
+      "learning_rate": 3.948106720454754e-05,
+      "loss": 1.0144,
+      "mean_token_accuracy": 0.7023302793502808,
+      "num_tokens": 210981073.0,
+      "step": 87540
+    },
+    {
+      "epoch": 0.8026955166406895,
+      "learning_rate": 3.946273035665169e-05,
+      "loss": 0.9583,
+      "mean_token_accuracy": 0.7169875621795654,
+      "num_tokens": 211012249.0,
+      "step": 87550
+    },
+    {
+      "epoch": 0.8027872008801687,
+      "learning_rate": 3.9444393508755844e-05,
+      "loss": 0.9546,
+      "mean_token_accuracy": 0.7173968315124511,
+      "num_tokens": 211042131.0,
+      "step": 87560
+    },
+    {
+      "epoch": 0.8028788851196479,
+      "learning_rate": 3.942605666086e-05,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.7187240719795227,
+      "num_tokens": 211072035.0,
+      "step": 87570
+    },
+    {
+      "epoch": 0.8029705693591271,
+      "learning_rate": 3.9407719812964155e-05,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.7188745498657226,
+      "num_tokens": 211102200.0,
+      "step": 87580
+    },
+    {
+      "epoch": 0.8030622535986064,
+      "learning_rate": 3.9389382965068307e-05,
+      "loss": 1.0042,
+      "mean_token_accuracy": 0.7035655081272125,
+      "num_tokens": 211132455.0,
+      "step": 87590
+    },
+    {
+      "epoch": 0.8031539378380856,
+      "learning_rate": 3.937104611717246e-05,
+      "loss": 1.0106,
+      "mean_token_accuracy": 0.7055139124393464,
+      "num_tokens": 211164525.0,
+      "step": 87600
+    },
+    {
+      "epoch": 0.8032456220775649,
+      "learning_rate": 3.935270926927661e-05,
+      "loss": 0.919,
+      "mean_token_accuracy": 0.7271938025951385,
+      "num_tokens": 211194734.0,
+      "step": 87610
+    },
+    {
+      "epoch": 0.8033373063170441,
+      "learning_rate": 3.933437242138077e-05,
+      "loss": 0.9212,
+      "mean_token_accuracy": 0.7250953376293182,
+      "num_tokens": 211225827.0,
+      "step": 87620
+    },
+    {
+      "epoch": 0.8034289905565233,
+      "learning_rate": 3.931603557348492e-05,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.719471538066864,
+      "num_tokens": 211255154.0,
+      "step": 87630
+    },
+    {
+      "epoch": 0.8035206747960025,
+      "learning_rate": 3.929769872558907e-05,
+      "loss": 0.9396,
+      "mean_token_accuracy": 0.7259688198566436,
+      "num_tokens": 211284628.0,
+      "step": 87640
+    },
+    {
+      "epoch": 0.8036123590354818,
+      "learning_rate": 3.9279361877693225e-05,
+      "loss": 0.9453,
+      "mean_token_accuracy": 0.7172414362430573,
+      "num_tokens": 211315715.0,
+      "step": 87650
+    },
+    {
+      "epoch": 0.803704043274961,
+      "learning_rate": 3.9261025029797376e-05,
+      "loss": 1.0265,
+      "mean_token_accuracy": 0.696022343635559,
+      "num_tokens": 211346988.0,
+      "step": 87660
+    },
+    {
+      "epoch": 0.8037957275144403,
+      "learning_rate": 3.9242688181901535e-05,
+      "loss": 0.9324,
+      "mean_token_accuracy": 0.7222636938095093,
+      "num_tokens": 211378233.0,
+      "step": 87670
+    },
+    {
+      "epoch": 0.8038874117539195,
+      "learning_rate": 3.922435133400569e-05,
+      "loss": 0.9872,
+      "mean_token_accuracy": 0.7116622984409332,
+      "num_tokens": 211408954.0,
+      "step": 87680
+    },
+    {
+      "epoch": 0.8039790959933988,
+      "learning_rate": 3.920601448610984e-05,
+      "loss": 0.9102,
+      "mean_token_accuracy": 0.7304248034954071,
+      "num_tokens": 211437232.0,
+      "step": 87690
+    },
+    {
+      "epoch": 0.8040707802328779,
+      "learning_rate": 3.918767763821399e-05,
+      "loss": 0.9743,
+      "mean_token_accuracy": 0.7167141556739807,
+      "num_tokens": 211469487.0,
+      "step": 87700
+    },
+    {
+      "epoch": 0.8041624644723572,
+      "learning_rate": 3.916934079031815e-05,
+      "loss": 1.0197,
+      "mean_token_accuracy": 0.7021766901016235,
+      "num_tokens": 211499026.0,
+      "step": 87710
+    },
+    {
+      "epoch": 0.8042541487118364,
+      "learning_rate": 3.91510039424223e-05,
+      "loss": 1.0043,
+      "mean_token_accuracy": 0.7064680993556977,
+      "num_tokens": 211531472.0,
+      "step": 87720
+    },
+    {
+      "epoch": 0.8043458329513157,
+      "learning_rate": 3.913266709452645e-05,
+      "loss": 0.9211,
+      "mean_token_accuracy": 0.7292511284351348,
+      "num_tokens": 211562880.0,
+      "step": 87730
+    },
+    {
+      "epoch": 0.8044375171907949,
+      "learning_rate": 3.9114330246630605e-05,
+      "loss": 0.9782,
+      "mean_token_accuracy": 0.7146872460842133,
+      "num_tokens": 211594193.0,
+      "step": 87740
+    },
+    {
+      "epoch": 0.8045292014302742,
+      "learning_rate": 3.909599339873476e-05,
+      "loss": 0.918,
+      "mean_token_accuracy": 0.7276806056499481,
+      "num_tokens": 211624063.0,
+      "step": 87750
+    },
+    {
+      "epoch": 0.8046208856697533,
+      "learning_rate": 3.9077656550838915e-05,
+      "loss": 0.994,
+      "mean_token_accuracy": 0.7086891889572143,
+      "num_tokens": 211655049.0,
+      "step": 87760
+    },
+    {
+      "epoch": 0.8047125699092326,
+      "learning_rate": 3.905931970294307e-05,
+      "loss": 1.0015,
+      "mean_token_accuracy": 0.7083421170711517,
+      "num_tokens": 211685571.0,
+      "step": 87770
+    },
+    {
+      "epoch": 0.8048042541487118,
+      "learning_rate": 3.904098285504722e-05,
+      "loss": 1.0112,
+      "mean_token_accuracy": 0.704827344417572,
+      "num_tokens": 211717536.0,
+      "step": 87780
+    },
+    {
+      "epoch": 0.8048959383881911,
+      "learning_rate": 3.902264600715137e-05,
+      "loss": 0.988,
+      "mean_token_accuracy": 0.7118992805480957,
+      "num_tokens": 211748833.0,
+      "step": 87790
+    },
+    {
+      "epoch": 0.8049876226276703,
+      "learning_rate": 3.900430915925552e-05,
+      "loss": 0.9308,
+      "mean_token_accuracy": 0.723883455991745,
+      "num_tokens": 211779141.0,
+      "step": 87800
+    },
+    {
+      "epoch": 0.8050793068671496,
+      "learning_rate": 3.898597231135968e-05,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7299412369728089,
+      "num_tokens": 211809033.0,
+      "step": 87810
+    },
+    {
+      "epoch": 0.8051709911066288,
+      "learning_rate": 3.896763546346383e-05,
+      "loss": 0.9067,
+      "mean_token_accuracy": 0.7267274975776672,
+      "num_tokens": 211837328.0,
+      "step": 87820
+    },
+    {
+      "epoch": 0.805262675346108,
+      "learning_rate": 3.8949298615567985e-05,
+      "loss": 0.9796,
+      "mean_token_accuracy": 0.7142065763473511,
+      "num_tokens": 211869504.0,
+      "step": 87830
+    },
+    {
+      "epoch": 0.8053543595855872,
+      "learning_rate": 3.8930961767672144e-05,
+      "loss": 0.9873,
+      "mean_token_accuracy": 0.7059406995773315,
+      "num_tokens": 211900504.0,
+      "step": 87840
+    },
+    {
+      "epoch": 0.8054460438250665,
+      "learning_rate": 3.891262491977629e-05,
+      "loss": 1.0178,
+      "mean_token_accuracy": 0.6987842381000519,
+      "num_tokens": 211932474.0,
+      "step": 87850
+    },
+    {
+      "epoch": 0.8055377280645457,
+      "learning_rate": 3.889428807188045e-05,
+      "loss": 0.9856,
+      "mean_token_accuracy": 0.7069795370101929,
+      "num_tokens": 211962924.0,
+      "step": 87860
+    },
+    {
+      "epoch": 0.805629412304025,
+      "learning_rate": 3.88759512239846e-05,
+      "loss": 1.0099,
+      "mean_token_accuracy": 0.7000271320343018,
+      "num_tokens": 211993217.0,
+      "step": 87870
+    },
+    {
+      "epoch": 0.8057210965435042,
+      "learning_rate": 3.885761437608875e-05,
+      "loss": 0.915,
+      "mean_token_accuracy": 0.7269558608531952,
+      "num_tokens": 212021789.0,
+      "step": 87880
+    },
+    {
+      "epoch": 0.8058127807829834,
+      "learning_rate": 3.883927752819291e-05,
+      "loss": 0.9503,
+      "mean_token_accuracy": 0.7160397291183471,
+      "num_tokens": 212052730.0,
+      "step": 87890
+    },
+    {
+      "epoch": 0.8059044650224626,
+      "learning_rate": 3.8820940680297055e-05,
+      "loss": 0.9924,
+      "mean_token_accuracy": 0.7088140308856964,
+      "num_tokens": 212084561.0,
+      "step": 87900
+    },
+    {
+      "epoch": 0.8059961492619419,
+      "learning_rate": 3.8802603832401214e-05,
+      "loss": 0.9428,
+      "mean_token_accuracy": 0.7223440527915954,
+      "num_tokens": 212115883.0,
+      "step": 87910
+    },
+    {
+      "epoch": 0.8060878335014211,
+      "learning_rate": 3.8784266984505366e-05,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.7104128360748291,
+      "num_tokens": 212147656.0,
+      "step": 87920
+    },
+    {
+      "epoch": 0.8061795177409004,
+      "learning_rate": 3.876593013660952e-05,
+      "loss": 0.992,
+      "mean_token_accuracy": 0.7046168982982636,
+      "num_tokens": 212177902.0,
+      "step": 87930
+    },
+    {
+      "epoch": 0.8062712019803796,
+      "learning_rate": 3.8747593288713676e-05,
+      "loss": 0.9981,
+      "mean_token_accuracy": 0.7021388173103332,
+      "num_tokens": 212208062.0,
+      "step": 87940
+    },
+    {
+      "epoch": 0.8063628862198589,
+      "learning_rate": 3.872925644081782e-05,
+      "loss": 0.9651,
+      "mean_token_accuracy": 0.7151989221572876,
+      "num_tokens": 212236622.0,
+      "step": 87950
+    },
+    {
+      "epoch": 0.806454570459338,
+      "learning_rate": 3.871091959292198e-05,
+      "loss": 1.0139,
+      "mean_token_accuracy": 0.702283525466919,
+      "num_tokens": 212267305.0,
+      "step": 87960
+    },
+    {
+      "epoch": 0.8065462546988172,
+      "learning_rate": 3.869258274502613e-05,
+      "loss": 0.9738,
+      "mean_token_accuracy": 0.7131257295608521,
+      "num_tokens": 212298509.0,
+      "step": 87970
+    },
+    {
+      "epoch": 0.8066379389382965,
+      "learning_rate": 3.8674245897130284e-05,
+      "loss": 0.9946,
+      "mean_token_accuracy": 0.7092668652534485,
+      "num_tokens": 212329561.0,
+      "step": 87980
+    },
+    {
+      "epoch": 0.8067296231777757,
+      "learning_rate": 3.865590904923444e-05,
+      "loss": 0.9438,
+      "mean_token_accuracy": 0.7176726818084717,
+      "num_tokens": 212359206.0,
+      "step": 87990
+    },
+    {
+      "epoch": 0.806821307417255,
+      "learning_rate": 3.863757220133859e-05,
+      "loss": 1.015,
+      "mean_token_accuracy": 0.7055783033370971,
+      "num_tokens": 212390480.0,
+      "step": 88000
+    },
+    {
+      "epoch": 0.8069129916567342,
+      "learning_rate": 3.8619235353442746e-05,
+      "loss": 0.9686,
+      "mean_token_accuracy": 0.7132801949977875,
+      "num_tokens": 212421124.0,
+      "step": 88010
+    },
+    {
+      "epoch": 0.8070046758962134,
+      "learning_rate": 3.86008985055469e-05,
+      "loss": 1.0125,
+      "mean_token_accuracy": 0.6998761415481567,
+      "num_tokens": 212451158.0,
+      "step": 88020
+    },
+    {
+      "epoch": 0.8070963601356926,
+      "learning_rate": 3.858256165765105e-05,
+      "loss": 0.9471,
+      "mean_token_accuracy": 0.7199070811271667,
+      "num_tokens": 212480837.0,
+      "step": 88030
+    },
+    {
+      "epoch": 0.8071880443751719,
+      "learning_rate": 3.856422480975521e-05,
+      "loss": 1.0048,
+      "mean_token_accuracy": 0.707367992401123,
+      "num_tokens": 212511966.0,
+      "step": 88040
+    },
+    {
+      "epoch": 0.8072797286146511,
+      "learning_rate": 3.8545887961859353e-05,
+      "loss": 0.9472,
+      "mean_token_accuracy": 0.7156034588813782,
+      "num_tokens": 212541582.0,
+      "step": 88050
+    },
+    {
+      "epoch": 0.8073714128541304,
+      "learning_rate": 3.852755111396351e-05,
+      "loss": 1.0048,
+      "mean_token_accuracy": 0.7002568304538727,
+      "num_tokens": 212572851.0,
+      "step": 88060
+    },
+    {
+      "epoch": 0.8074630970936096,
+      "learning_rate": 3.8509214266067664e-05,
+      "loss": 0.9805,
+      "mean_token_accuracy": 0.711814159154892,
+      "num_tokens": 212601763.0,
+      "step": 88070
+    },
+    {
+      "epoch": 0.8075547813330889,
+      "learning_rate": 3.8490877418171816e-05,
+      "loss": 1.0326,
+      "mean_token_accuracy": 0.6988707661628724,
+      "num_tokens": 212633133.0,
+      "step": 88080
+    },
+    {
+      "epoch": 0.807646465572568,
+      "learning_rate": 3.8472540570275974e-05,
+      "loss": 1.012,
+      "mean_token_accuracy": 0.7042496144771576,
+      "num_tokens": 212665040.0,
+      "step": 88090
+    },
+    {
+      "epoch": 0.8077381498120473,
+      "learning_rate": 3.845420372238012e-05,
+      "loss": 0.9293,
+      "mean_token_accuracy": 0.7194316208362579,
+      "num_tokens": 212694268.0,
+      "step": 88100
+    },
+    {
+      "epoch": 0.8078298340515265,
+      "learning_rate": 3.843586687448428e-05,
+      "loss": 1.0078,
+      "mean_token_accuracy": 0.7069931268692017,
+      "num_tokens": 212725343.0,
+      "step": 88110
+    },
+    {
+      "epoch": 0.8079215182910058,
+      "learning_rate": 3.841753002658843e-05,
+      "loss": 0.9689,
+      "mean_token_accuracy": 0.708724421262741,
+      "num_tokens": 212757430.0,
+      "step": 88120
+    },
+    {
+      "epoch": 0.808013202530485,
+      "learning_rate": 3.839919317869258e-05,
+      "loss": 1.0366,
+      "mean_token_accuracy": 0.6936649978160858,
+      "num_tokens": 212785740.0,
+      "step": 88130
+    },
+    {
+      "epoch": 0.8081048867699643,
+      "learning_rate": 3.838085633079674e-05,
+      "loss": 1.0409,
+      "mean_token_accuracy": 0.6952789068222046,
+      "num_tokens": 212815872.0,
+      "step": 88140
+    },
+    {
+      "epoch": 0.8081965710094435,
+      "learning_rate": 3.836251948290089e-05,
+      "loss": 1.0242,
+      "mean_token_accuracy": 0.7023537755012512,
+      "num_tokens": 212847005.0,
+      "step": 88150
+    },
+    {
+      "epoch": 0.8082882552489227,
+      "learning_rate": 3.8344182635005044e-05,
+      "loss": 1.0176,
+      "mean_token_accuracy": 0.7029186666011811,
+      "num_tokens": 212879746.0,
+      "step": 88160
+    },
+    {
+      "epoch": 0.8083799394884019,
+      "learning_rate": 3.8325845787109196e-05,
+      "loss": 1.0404,
+      "mean_token_accuracy": 0.7016336560249329,
+      "num_tokens": 212910525.0,
+      "step": 88170
+    },
+    {
+      "epoch": 0.8084716237278812,
+      "learning_rate": 3.830750893921335e-05,
+      "loss": 0.9744,
+      "mean_token_accuracy": 0.7135962665081024,
+      "num_tokens": 212940403.0,
+      "step": 88180
+    },
+    {
+      "epoch": 0.8085633079673604,
+      "learning_rate": 3.828917209131751e-05,
+      "loss": 0.9793,
+      "mean_token_accuracy": 0.710008054971695,
+      "num_tokens": 212971219.0,
+      "step": 88190
+    },
+    {
+      "epoch": 0.8086549922068397,
+      "learning_rate": 3.827083524342166e-05,
+      "loss": 1.0374,
+      "mean_token_accuracy": 0.693979698419571,
+      "num_tokens": 213001472.0,
+      "step": 88200
+    },
+    {
+      "epoch": 0.8087466764463189,
+      "learning_rate": 3.825249839552581e-05,
+      "loss": 1.0347,
+      "mean_token_accuracy": 0.6930490970611572,
+      "num_tokens": 213032740.0,
+      "step": 88210
+    },
+    {
+      "epoch": 0.8088383606857981,
+      "learning_rate": 3.823416154762996e-05,
+      "loss": 0.9726,
+      "mean_token_accuracy": 0.7108381807804107,
+      "num_tokens": 213064307.0,
+      "step": 88220
+    },
+    {
+      "epoch": 0.8089300449252773,
+      "learning_rate": 3.8215824699734114e-05,
+      "loss": 1.0181,
+      "mean_token_accuracy": 0.7040467023849487,
+      "num_tokens": 213095133.0,
+      "step": 88230
+    },
+    {
+      "epoch": 0.8090217291647566,
+      "learning_rate": 3.819748785183827e-05,
+      "loss": 0.99,
+      "mean_token_accuracy": 0.7062864422798156,
+      "num_tokens": 213125845.0,
+      "step": 88240
+    },
+    {
+      "epoch": 0.8091134134042358,
+      "learning_rate": 3.8179151003942425e-05,
+      "loss": 1.0044,
+      "mean_token_accuracy": 0.7024176120758057,
+      "num_tokens": 213155881.0,
+      "step": 88250
+    },
+    {
+      "epoch": 0.8092050976437151,
+      "learning_rate": 3.8160814156046577e-05,
+      "loss": 1.0466,
+      "mean_token_accuracy": 0.6926968991756439,
+      "num_tokens": 213187092.0,
+      "step": 88260
+    },
+    {
+      "epoch": 0.8092967818831943,
+      "learning_rate": 3.814247730815073e-05,
+      "loss": 1.0012,
+      "mean_token_accuracy": 0.7008063197135925,
+      "num_tokens": 213218664.0,
+      "step": 88270
+    },
+    {
+      "epoch": 0.8093884661226736,
+      "learning_rate": 3.812414046025488e-05,
+      "loss": 1.0249,
+      "mean_token_accuracy": 0.7034690320491791,
+      "num_tokens": 213250393.0,
+      "step": 88280
+    },
+    {
+      "epoch": 0.8094801503621527,
+      "learning_rate": 3.810580361235904e-05,
+      "loss": 0.9598,
+      "mean_token_accuracy": 0.7189905226230622,
+      "num_tokens": 213281184.0,
+      "step": 88290
+    },
+    {
+      "epoch": 0.809571834601632,
+      "learning_rate": 3.808746676446319e-05,
+      "loss": 0.9845,
+      "mean_token_accuracy": 0.7038617312908173,
+      "num_tokens": 213312259.0,
+      "step": 88300
+    },
+    {
+      "epoch": 0.8096635188411112,
+      "learning_rate": 3.806912991656734e-05,
+      "loss": 0.9706,
+      "mean_token_accuracy": 0.7138518333435059,
+      "num_tokens": 213341191.0,
+      "step": 88310
+    },
+    {
+      "epoch": 0.8097552030805905,
+      "learning_rate": 3.8050793068671495e-05,
+      "loss": 1.0368,
+      "mean_token_accuracy": 0.6929193556308746,
+      "num_tokens": 213372392.0,
+      "step": 88320
+    },
+    {
+      "epoch": 0.8098468873200697,
+      "learning_rate": 3.803245622077565e-05,
+      "loss": 0.9991,
+      "mean_token_accuracy": 0.7051734685897827,
+      "num_tokens": 213403814.0,
+      "step": 88330
+    },
+    {
+      "epoch": 0.809938571559549,
+      "learning_rate": 3.8014119372879805e-05,
+      "loss": 1.0016,
+      "mean_token_accuracy": 0.7056295692920684,
+      "num_tokens": 213436675.0,
+      "step": 88340
+    },
+    {
+      "epoch": 0.8100302557990281,
+      "learning_rate": 3.799578252498396e-05,
+      "loss": 0.9922,
+      "mean_token_accuracy": 0.710987263917923,
+      "num_tokens": 213466825.0,
+      "step": 88350
+    },
+    {
+      "epoch": 0.8101219400385073,
+      "learning_rate": 3.797744567708811e-05,
+      "loss": 0.9356,
+      "mean_token_accuracy": 0.7210259854793548,
+      "num_tokens": 213496287.0,
+      "step": 88360
+    },
+    {
+      "epoch": 0.8102136242779866,
+      "learning_rate": 3.795910882919226e-05,
+      "loss": 1.0192,
+      "mean_token_accuracy": 0.7016069829463959,
+      "num_tokens": 213528028.0,
+      "step": 88370
+    },
+    {
+      "epoch": 0.8103053085174658,
+      "learning_rate": 3.794077198129642e-05,
+      "loss": 1.0424,
+      "mean_token_accuracy": 0.6924838542938232,
+      "num_tokens": 213558841.0,
+      "step": 88380
+    },
+    {
+      "epoch": 0.8103969927569451,
+      "learning_rate": 3.792243513340057e-05,
+      "loss": 1.028,
+      "mean_token_accuracy": 0.6980836987495422,
+      "num_tokens": 213589731.0,
+      "step": 88390
+    },
+    {
+      "epoch": 0.8104886769964244,
+      "learning_rate": 3.790409828550472e-05,
+      "loss": 1.0422,
+      "mean_token_accuracy": 0.6992207884788513,
+      "num_tokens": 213622145.0,
+      "step": 88400
+    },
+    {
+      "epoch": 0.8105803612359036,
+      "learning_rate": 3.7885761437608875e-05,
+      "loss": 1.0096,
+      "mean_token_accuracy": 0.7027938604354859,
+      "num_tokens": 213653104.0,
+      "step": 88410
+    },
+    {
+      "epoch": 0.8106720454753827,
+      "learning_rate": 3.786742458971303e-05,
+      "loss": 0.9938,
+      "mean_token_accuracy": 0.7057611823081971,
+      "num_tokens": 213683900.0,
+      "step": 88420
+    },
+    {
+      "epoch": 0.810763729714862,
+      "learning_rate": 3.7849087741817185e-05,
+      "loss": 1.0263,
+      "mean_token_accuracy": 0.7007830560207366,
+      "num_tokens": 213713769.0,
+      "step": 88430
+    },
+    {
+      "epoch": 0.8108554139543412,
+      "learning_rate": 3.783075089392134e-05,
+      "loss": 1.0262,
+      "mean_token_accuracy": 0.6971380650997162,
+      "num_tokens": 213744230.0,
+      "step": 88440
+    },
+    {
+      "epoch": 0.8109470981938205,
+      "learning_rate": 3.781241404602549e-05,
+      "loss": 1.0232,
+      "mean_token_accuracy": 0.7045233488082886,
+      "num_tokens": 213776054.0,
+      "step": 88450
+    },
+    {
+      "epoch": 0.8110387824332997,
+      "learning_rate": 3.779407719812965e-05,
+      "loss": 1.026,
+      "mean_token_accuracy": 0.7000523269176483,
+      "num_tokens": 213807639.0,
+      "step": 88460
+    },
+    {
+      "epoch": 0.811130466672779,
+      "learning_rate": 3.777574035023379e-05,
+      "loss": 0.9708,
+      "mean_token_accuracy": 0.7135145664215088,
+      "num_tokens": 213838473.0,
+      "step": 88470
+    },
+    {
+      "epoch": 0.8112221509122581,
+      "learning_rate": 3.775740350233795e-05,
+      "loss": 1.0188,
+      "mean_token_accuracy": 0.70450279712677,
+      "num_tokens": 213869075.0,
+      "step": 88480
+    },
+    {
+      "epoch": 0.8113138351517374,
+      "learning_rate": 3.77390666544421e-05,
+      "loss": 1.0109,
+      "mean_token_accuracy": 0.7031781196594238,
+      "num_tokens": 213899235.0,
+      "step": 88490
+    },
+    {
+      "epoch": 0.8114055193912166,
+      "learning_rate": 3.7720729806546255e-05,
+      "loss": 1.0088,
+      "mean_token_accuracy": 0.699275553226471,
+      "num_tokens": 213930406.0,
+      "step": 88500
+    },
+    {
+      "epoch": 0.8114972036306959,
+      "learning_rate": 3.7702392958650414e-05,
+      "loss": 1.0298,
+      "mean_token_accuracy": 0.7035545229911804,
+      "num_tokens": 213961438.0,
+      "step": 88510
+    },
+    {
+      "epoch": 0.8115888878701751,
+      "learning_rate": 3.768405611075456e-05,
+      "loss": 1.0406,
+      "mean_token_accuracy": 0.7002984166145325,
+      "num_tokens": 213994024.0,
+      "step": 88520
+    },
+    {
+      "epoch": 0.8116805721096544,
+      "learning_rate": 3.766571926285872e-05,
+      "loss": 1.0164,
+      "mean_token_accuracy": 0.6988596498966217,
+      "num_tokens": 214025254.0,
+      "step": 88530
+    },
+    {
+      "epoch": 0.8117722563491336,
+      "learning_rate": 3.764738241496287e-05,
+      "loss": 0.9976,
+      "mean_token_accuracy": 0.7057909548282624,
+      "num_tokens": 214054777.0,
+      "step": 88540
+    },
+    {
+      "epoch": 0.8118639405886128,
+      "learning_rate": 3.762904556706702e-05,
+      "loss": 1.0154,
+      "mean_token_accuracy": 0.6962915360927582,
+      "num_tokens": 214085391.0,
+      "step": 88550
+    },
+    {
+      "epoch": 0.811955624828092,
+      "learning_rate": 3.761070871917118e-05,
+      "loss": 1.0146,
+      "mean_token_accuracy": 0.7051073789596558,
+      "num_tokens": 214115470.0,
+      "step": 88560
+    },
+    {
+      "epoch": 0.8120473090675713,
+      "learning_rate": 3.7592371871275325e-05,
+      "loss": 0.9843,
+      "mean_token_accuracy": 0.7093298554420471,
+      "num_tokens": 214147881.0,
+      "step": 88570
+    },
+    {
+      "epoch": 0.8121389933070505,
+      "learning_rate": 3.7574035023379484e-05,
+      "loss": 1.0514,
+      "mean_token_accuracy": 0.6973120331764221,
+      "num_tokens": 214178899.0,
+      "step": 88580
+    },
+    {
+      "epoch": 0.8122306775465298,
+      "learning_rate": 3.7555698175483636e-05,
+      "loss": 1.0301,
+      "mean_token_accuracy": 0.6959855020046234,
+      "num_tokens": 214210317.0,
+      "step": 88590
+    },
+    {
+      "epoch": 0.812322361786009,
+      "learning_rate": 3.753736132758779e-05,
+      "loss": 1.0216,
+      "mean_token_accuracy": 0.6968937814235687,
+      "num_tokens": 214240586.0,
+      "step": 88600
+    },
+    {
+      "epoch": 0.8124140460254882,
+      "learning_rate": 3.7519024479691946e-05,
+      "loss": 1.0107,
+      "mean_token_accuracy": 0.7076817810535431,
+      "num_tokens": 214270481.0,
+      "step": 88610
+    },
+    {
+      "epoch": 0.8125057302649674,
+      "learning_rate": 3.750068763179609e-05,
+      "loss": 1.0213,
+      "mean_token_accuracy": 0.7024851560592651,
+      "num_tokens": 214301220.0,
+      "step": 88620
+    },
+    {
+      "epoch": 0.8125974145044467,
+      "learning_rate": 3.748235078390025e-05,
+      "loss": 0.9704,
+      "mean_token_accuracy": 0.7130574345588684,
+      "num_tokens": 214332446.0,
+      "step": 88630
+    },
+    {
+      "epoch": 0.8126890987439259,
+      "learning_rate": 3.746401393600441e-05,
+      "loss": 1.011,
+      "mean_token_accuracy": 0.7014073669910431,
+      "num_tokens": 214363295.0,
+      "step": 88640
+    },
+    {
+      "epoch": 0.8127807829834052,
+      "learning_rate": 3.7445677088108554e-05,
+      "loss": 1.0504,
+      "mean_token_accuracy": 0.6962876319885254,
+      "num_tokens": 214396536.0,
+      "step": 88650
+    },
+    {
+      "epoch": 0.8128724672228844,
+      "learning_rate": 3.742734024021271e-05,
+      "loss": 1.0115,
+      "mean_token_accuracy": 0.6957520246505737,
+      "num_tokens": 214427161.0,
+      "step": 88660
+    },
+    {
+      "epoch": 0.8129641514623637,
+      "learning_rate": 3.740900339231686e-05,
+      "loss": 0.9785,
+      "mean_token_accuracy": 0.7070945560932159,
+      "num_tokens": 214456377.0,
+      "step": 88670
+    },
+    {
+      "epoch": 0.8130558357018428,
+      "learning_rate": 3.7390666544421016e-05,
+      "loss": 1.0255,
+      "mean_token_accuracy": 0.6996080458164216,
+      "num_tokens": 214488332.0,
+      "step": 88680
+    },
+    {
+      "epoch": 0.8131475199413221,
+      "learning_rate": 3.7372329696525175e-05,
+      "loss": 1.0398,
+      "mean_token_accuracy": 0.6954891562461853,
+      "num_tokens": 214520267.0,
+      "step": 88690
+    },
+    {
+      "epoch": 0.8132392041808013,
+      "learning_rate": 3.735399284862932e-05,
+      "loss": 1.0084,
+      "mean_token_accuracy": 0.6975082814693451,
+      "num_tokens": 214551572.0,
+      "step": 88700
+    },
+    {
+      "epoch": 0.8133308884202806,
+      "learning_rate": 3.733565600073348e-05,
+      "loss": 0.9864,
+      "mean_token_accuracy": 0.7096741735935211,
+      "num_tokens": 214581318.0,
+      "step": 88710
+    },
+    {
+      "epoch": 0.8134225726597598,
+      "learning_rate": 3.7317319152837623e-05,
+      "loss": 0.9964,
+      "mean_token_accuracy": 0.7107142269611358,
+      "num_tokens": 214612112.0,
+      "step": 88720
+    },
+    {
+      "epoch": 0.8135142568992391,
+      "learning_rate": 3.729898230494178e-05,
+      "loss": 1.0236,
+      "mean_token_accuracy": 0.7034564018249512,
+      "num_tokens": 214643725.0,
+      "step": 88730
+    },
+    {
+      "epoch": 0.8136059411387182,
+      "learning_rate": 3.728064545704594e-05,
+      "loss": 1.0077,
+      "mean_token_accuracy": 0.6969237089157104,
+      "num_tokens": 214673828.0,
+      "step": 88740
+    },
+    {
+      "epoch": 0.8136976253781975,
+      "learning_rate": 3.7262308609150086e-05,
+      "loss": 0.9537,
+      "mean_token_accuracy": 0.7197325646877288,
+      "num_tokens": 214704474.0,
+      "step": 88750
+    },
+    {
+      "epoch": 0.8137893096176767,
+      "learning_rate": 3.7243971761254244e-05,
+      "loss": 0.9941,
+      "mean_token_accuracy": 0.7051733493804931,
+      "num_tokens": 214736266.0,
+      "step": 88760
+    },
+    {
+      "epoch": 0.813880993857156,
+      "learning_rate": 3.7225634913358396e-05,
+      "loss": 1.0042,
+      "mean_token_accuracy": 0.7060157775878906,
+      "num_tokens": 214768187.0,
+      "step": 88770
+    },
+    {
+      "epoch": 0.8139726780966352,
+      "learning_rate": 3.720729806546255e-05,
+      "loss": 1.043,
+      "mean_token_accuracy": 0.6956783473491669,
+      "num_tokens": 214798631.0,
+      "step": 88780
+    },
+    {
+      "epoch": 0.8140643623361145,
+      "learning_rate": 3.718896121756671e-05,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.7153402268886566,
+      "num_tokens": 214827238.0,
+      "step": 88790
+    },
+    {
+      "epoch": 0.8141560465755937,
+      "learning_rate": 3.717062436967085e-05,
+      "loss": 1.0014,
+      "mean_token_accuracy": 0.7031769633293152,
+      "num_tokens": 214857982.0,
+      "step": 88800
+    },
+    {
+      "epoch": 0.8142477308150728,
+      "learning_rate": 3.715228752177501e-05,
+      "loss": 1.0093,
+      "mean_token_accuracy": 0.7037087738513946,
+      "num_tokens": 214888607.0,
+      "step": 88810
+    },
+    {
+      "epoch": 0.8143394150545521,
+      "learning_rate": 3.713395067387916e-05,
+      "loss": 0.9918,
+      "mean_token_accuracy": 0.7101965010166168,
+      "num_tokens": 214919555.0,
+      "step": 88820
+    },
+    {
+      "epoch": 0.8144310992940313,
+      "learning_rate": 3.7115613825983314e-05,
+      "loss": 1.0026,
+      "mean_token_accuracy": 0.7070383369922638,
+      "num_tokens": 214950616.0,
+      "step": 88830
+    },
+    {
+      "epoch": 0.8145227835335106,
+      "learning_rate": 3.709727697808747e-05,
+      "loss": 1.0025,
+      "mean_token_accuracy": 0.7008816838264466,
+      "num_tokens": 214981063.0,
+      "step": 88840
+    },
+    {
+      "epoch": 0.8146144677729898,
+      "learning_rate": 3.707894013019162e-05,
+      "loss": 1.0292,
+      "mean_token_accuracy": 0.6949926257133484,
+      "num_tokens": 215011461.0,
+      "step": 88850
+    },
+    {
+      "epoch": 0.8147061520124691,
+      "learning_rate": 3.706060328229578e-05,
+      "loss": 1.0562,
+      "mean_token_accuracy": 0.6896575272083283,
+      "num_tokens": 215042214.0,
+      "step": 88860
+    },
+    {
+      "epoch": 0.8147978362519482,
+      "learning_rate": 3.704226643439993e-05,
+      "loss": 1.0016,
+      "mean_token_accuracy": 0.7054397463798523,
+      "num_tokens": 215072638.0,
+      "step": 88870
+    },
+    {
+      "epoch": 0.8148895204914275,
+      "learning_rate": 3.702392958650408e-05,
+      "loss": 0.9907,
+      "mean_token_accuracy": 0.7066368162631989,
+      "num_tokens": 215102781.0,
+      "step": 88880
+    },
+    {
+      "epoch": 0.8149812047309067,
+      "learning_rate": 3.700559273860824e-05,
+      "loss": 1.0176,
+      "mean_token_accuracy": 0.6987467110157013,
+      "num_tokens": 215134156.0,
+      "step": 88890
+    },
+    {
+      "epoch": 0.815072888970386,
+      "learning_rate": 3.6987255890712384e-05,
+      "loss": 0.9487,
+      "mean_token_accuracy": 0.7208304941654206,
+      "num_tokens": 215163529.0,
+      "step": 88900
+    },
+    {
+      "epoch": 0.8151645732098652,
+      "learning_rate": 3.696891904281654e-05,
+      "loss": 1.0128,
+      "mean_token_accuracy": 0.696195912361145,
+      "num_tokens": 215193480.0,
+      "step": 88910
+    },
+    {
+      "epoch": 0.8152562574493445,
+      "learning_rate": 3.6950582194920695e-05,
+      "loss": 1.0287,
+      "mean_token_accuracy": 0.6967327237129212,
+      "num_tokens": 215223621.0,
+      "step": 88920
+    },
+    {
+      "epoch": 0.8153479416888237,
+      "learning_rate": 3.6932245347024847e-05,
+      "loss": 1.0166,
+      "mean_token_accuracy": 0.7003809750080109,
+      "num_tokens": 215255584.0,
+      "step": 88930
+    },
+    {
+      "epoch": 0.8154396259283029,
+      "learning_rate": 3.6913908499129005e-05,
+      "loss": 1.0066,
+      "mean_token_accuracy": 0.7073103427886963,
+      "num_tokens": 215288761.0,
+      "step": 88940
+    },
+    {
+      "epoch": 0.8155313101677821,
+      "learning_rate": 3.689557165123316e-05,
+      "loss": 1.0326,
+      "mean_token_accuracy": 0.6977166712284089,
+      "num_tokens": 215320455.0,
+      "step": 88950
+    },
+    {
+      "epoch": 0.8156229944072614,
+      "learning_rate": 3.687723480333731e-05,
+      "loss": 0.9917,
+      "mean_token_accuracy": 0.7111059904098511,
+      "num_tokens": 215350223.0,
+      "step": 88960
+    },
+    {
+      "epoch": 0.8157146786467406,
+      "learning_rate": 3.685889795544146e-05,
+      "loss": 0.9994,
+      "mean_token_accuracy": 0.7083847284317016,
+      "num_tokens": 215379750.0,
+      "step": 88970
+    },
+    {
+      "epoch": 0.8158063628862199,
+      "learning_rate": 3.684056110754561e-05,
+      "loss": 1.0133,
+      "mean_token_accuracy": 0.704751193523407,
+      "num_tokens": 215409774.0,
+      "step": 88980
+    },
+    {
+      "epoch": 0.8158980471256991,
+      "learning_rate": 3.682222425964977e-05,
+      "loss": 1.0352,
+      "mean_token_accuracy": 0.6992090284824372,
+      "num_tokens": 215441878.0,
+      "step": 88990
+    },
+    {
+      "epoch": 0.8159897313651783,
+      "learning_rate": 3.680388741175392e-05,
+      "loss": 1.0072,
+      "mean_token_accuracy": 0.7007029592990875,
+      "num_tokens": 215473840.0,
+      "step": 89000
+    },
+    {
+      "epoch": 0.8160814156046575,
+      "learning_rate": 3.6785550563858075e-05,
+      "loss": 1.0101,
+      "mean_token_accuracy": 0.7076938152313232,
+      "num_tokens": 215505306.0,
+      "step": 89010
+    },
+    {
+      "epoch": 0.8161730998441368,
+      "learning_rate": 3.676721371596223e-05,
+      "loss": 0.9989,
+      "mean_token_accuracy": 0.7039797902107239,
+      "num_tokens": 215536470.0,
+      "step": 89020
+    },
+    {
+      "epoch": 0.816264784083616,
+      "learning_rate": 3.674887686806638e-05,
+      "loss": 0.9962,
+      "mean_token_accuracy": 0.7050273239612579,
+      "num_tokens": 215568089.0,
+      "step": 89030
+    },
+    {
+      "epoch": 0.8163564683230953,
+      "learning_rate": 3.673054002017054e-05,
+      "loss": 1.0111,
+      "mean_token_accuracy": 0.705344295501709,
+      "num_tokens": 215597927.0,
+      "step": 89040
+    },
+    {
+      "epoch": 0.8164481525625745,
+      "learning_rate": 3.671220317227469e-05,
+      "loss": 1.0453,
+      "mean_token_accuracy": 0.6925242125988007,
+      "num_tokens": 215628923.0,
+      "step": 89050
+    },
+    {
+      "epoch": 0.8165398368020538,
+      "learning_rate": 3.669386632437884e-05,
+      "loss": 1.0501,
+      "mean_token_accuracy": 0.6919217348098755,
+      "num_tokens": 215660744.0,
+      "step": 89060
+    },
+    {
+      "epoch": 0.8166315210415329,
+      "learning_rate": 3.667552947648299e-05,
+      "loss": 0.9929,
+      "mean_token_accuracy": 0.7100502789020539,
+      "num_tokens": 215691789.0,
+      "step": 89070
+    },
+    {
+      "epoch": 0.8167232052810122,
+      "learning_rate": 3.665719262858715e-05,
+      "loss": 0.9895,
+      "mean_token_accuracy": 0.705121123790741,
+      "num_tokens": 215721524.0,
+      "step": 89080
+    },
+    {
+      "epoch": 0.8168148895204914,
+      "learning_rate": 3.6638855780691303e-05,
+      "loss": 1.0218,
+      "mean_token_accuracy": 0.6980223894119263,
+      "num_tokens": 215752607.0,
+      "step": 89090
+    },
+    {
+      "epoch": 0.8169065737599707,
+      "learning_rate": 3.6620518932795455e-05,
+      "loss": 1.0035,
+      "mean_token_accuracy": 0.7023726284503937,
+      "num_tokens": 215782477.0,
+      "step": 89100
+    },
+    {
+      "epoch": 0.8169982579994499,
+      "learning_rate": 3.660218208489961e-05,
+      "loss": 1.0128,
+      "mean_token_accuracy": 0.7043716669082641,
+      "num_tokens": 215814852.0,
+      "step": 89110
+    },
+    {
+      "epoch": 0.8170899422389292,
+      "learning_rate": 3.658384523700376e-05,
+      "loss": 1.0231,
+      "mean_token_accuracy": 0.6985966920852661,
+      "num_tokens": 215846160.0,
+      "step": 89120
+    },
+    {
+      "epoch": 0.8171816264784083,
+      "learning_rate": 3.656550838910792e-05,
+      "loss": 0.9957,
+      "mean_token_accuracy": 0.7034257829189301,
+      "num_tokens": 215877751.0,
+      "step": 89130
+    },
+    {
+      "epoch": 0.8172733107178876,
+      "learning_rate": 3.654717154121207e-05,
+      "loss": 0.9372,
+      "mean_token_accuracy": 0.7221430957317352,
+      "num_tokens": 215907995.0,
+      "step": 89140
+    },
+    {
+      "epoch": 0.8173649949573668,
+      "learning_rate": 3.652883469331622e-05,
+      "loss": 1.025,
+      "mean_token_accuracy": 0.6958905518054962,
+      "num_tokens": 215939823.0,
+      "step": 89150
+    },
+    {
+      "epoch": 0.8174566791968461,
+      "learning_rate": 3.651049784542037e-05,
+      "loss": 1.0073,
+      "mean_token_accuracy": 0.6971342086791992,
+      "num_tokens": 215970094.0,
+      "step": 89160
+    },
+    {
+      "epoch": 0.8175483634363253,
+      "learning_rate": 3.6492160997524525e-05,
+      "loss": 1.0489,
+      "mean_token_accuracy": 0.6948488593101502,
+      "num_tokens": 216001819.0,
+      "step": 89170
+    },
+    {
+      "epoch": 0.8176400476758046,
+      "learning_rate": 3.6473824149628684e-05,
+      "loss": 1.0069,
+      "mean_token_accuracy": 0.7015217125415802,
+      "num_tokens": 216031591.0,
+      "step": 89180
+    },
+    {
+      "epoch": 0.8177317319152838,
+      "learning_rate": 3.6455487301732836e-05,
+      "loss": 1.0493,
+      "mean_token_accuracy": 0.695898962020874,
+      "num_tokens": 216063653.0,
+      "step": 89190
+    },
+    {
+      "epoch": 0.817823416154763,
+      "learning_rate": 3.643715045383699e-05,
+      "loss": 1.0123,
+      "mean_token_accuracy": 0.6982617616653443,
+      "num_tokens": 216094424.0,
+      "step": 89200
+    },
+    {
+      "epoch": 0.8179151003942422,
+      "learning_rate": 3.641881360594114e-05,
+      "loss": 1.005,
+      "mean_token_accuracy": 0.7010263442993164,
+      "num_tokens": 216124983.0,
+      "step": 89210
+    },
+    {
+      "epoch": 0.8180067846337215,
+      "learning_rate": 3.640047675804529e-05,
+      "loss": 1.0066,
+      "mean_token_accuracy": 0.704931628704071,
+      "num_tokens": 216155769.0,
+      "step": 89220
+    },
+    {
+      "epoch": 0.8180984688732007,
+      "learning_rate": 3.638213991014945e-05,
+      "loss": 1.0059,
+      "mean_token_accuracy": 0.7023326694965363,
+      "num_tokens": 216186840.0,
+      "step": 89230
+    },
+    {
+      "epoch": 0.81819015311268,
+      "learning_rate": 3.63638030622536e-05,
+      "loss": 1.0034,
+      "mean_token_accuracy": 0.7027499079704285,
+      "num_tokens": 216217092.0,
+      "step": 89240
+    },
+    {
+      "epoch": 0.8182818373521592,
+      "learning_rate": 3.6345466214357754e-05,
+      "loss": 0.9911,
+      "mean_token_accuracy": 0.706219208240509,
+      "num_tokens": 216248515.0,
+      "step": 89250
+    },
+    {
+      "epoch": 0.8183735215916383,
+      "learning_rate": 3.632712936646191e-05,
+      "loss": 0.989,
+      "mean_token_accuracy": 0.709886884689331,
+      "num_tokens": 216279085.0,
+      "step": 89260
+    },
+    {
+      "epoch": 0.8184652058311176,
+      "learning_rate": 3.630879251856606e-05,
+      "loss": 0.997,
+      "mean_token_accuracy": 0.7053792595863342,
+      "num_tokens": 216311495.0,
+      "step": 89270
+    },
+    {
+      "epoch": 0.8185568900705968,
+      "learning_rate": 3.6290455670670216e-05,
+      "loss": 1.0224,
+      "mean_token_accuracy": 0.6974895119667053,
+      "num_tokens": 216340883.0,
+      "step": 89280
+    },
+    {
+      "epoch": 0.8186485743100761,
+      "learning_rate": 3.627211882277437e-05,
+      "loss": 0.9947,
+      "mean_token_accuracy": 0.7037676692008972,
+      "num_tokens": 216370882.0,
+      "step": 89290
+    },
+    {
+      "epoch": 0.8187402585495553,
+      "learning_rate": 3.625378197487852e-05,
+      "loss": 1.0227,
+      "mean_token_accuracy": 0.6979583740234375,
+      "num_tokens": 216401683.0,
+      "step": 89300
+    },
+    {
+      "epoch": 0.8188319427890346,
+      "learning_rate": 3.623544512698268e-05,
+      "loss": 1.0257,
+      "mean_token_accuracy": 0.6965280830860138,
+      "num_tokens": 216433933.0,
+      "step": 89310
+    },
+    {
+      "epoch": 0.8189236270285138,
+      "learning_rate": 3.6217108279086824e-05,
+      "loss": 1.0254,
+      "mean_token_accuracy": 0.6979999899864197,
+      "num_tokens": 216464153.0,
+      "step": 89320
+    },
+    {
+      "epoch": 0.819015311267993,
+      "learning_rate": 3.619877143119098e-05,
+      "loss": 1.0098,
+      "mean_token_accuracy": 0.7058261871337891,
+      "num_tokens": 216494895.0,
+      "step": 89330
+    },
+    {
+      "epoch": 0.8191069955074722,
+      "learning_rate": 3.6180434583295134e-05,
+      "loss": 0.9862,
+      "mean_token_accuracy": 0.7088732123374939,
+      "num_tokens": 216525433.0,
+      "step": 89340
+    },
+    {
+      "epoch": 0.8191986797469515,
+      "learning_rate": 3.6162097735399286e-05,
+      "loss": 1.0223,
+      "mean_token_accuracy": 0.697758823633194,
+      "num_tokens": 216556873.0,
+      "step": 89350
+    },
+    {
+      "epoch": 0.8192903639864307,
+      "learning_rate": 3.6143760887503445e-05,
+      "loss": 1.0144,
+      "mean_token_accuracy": 0.6988425016403198,
+      "num_tokens": 216588096.0,
+      "step": 89360
+    },
+    {
+      "epoch": 0.81938204822591,
+      "learning_rate": 3.612542403960759e-05,
+      "loss": 0.9913,
+      "mean_token_accuracy": 0.7108243405818939,
+      "num_tokens": 216619745.0,
+      "step": 89370
+    },
+    {
+      "epoch": 0.8194737324653892,
+      "learning_rate": 3.610708719171175e-05,
+      "loss": 1.0102,
+      "mean_token_accuracy": 0.7040644705295562,
+      "num_tokens": 216651167.0,
+      "step": 89380
+    },
+    {
+      "epoch": 0.8195654167048685,
+      "learning_rate": 3.60887503438159e-05,
+      "loss": 1.0176,
+      "mean_token_accuracy": 0.6944675505161285,
+      "num_tokens": 216681878.0,
+      "step": 89390
+    },
+    {
+      "epoch": 0.8196571009443476,
+      "learning_rate": 3.607041349592005e-05,
+      "loss": 0.9937,
+      "mean_token_accuracy": 0.7025558471679687,
+      "num_tokens": 216711604.0,
+      "step": 89400
+    },
+    {
+      "epoch": 0.8197487851838269,
+      "learning_rate": 3.605207664802421e-05,
+      "loss": 1.0134,
+      "mean_token_accuracy": 0.7018278241157532,
+      "num_tokens": 216744331.0,
+      "step": 89410
+    },
+    {
+      "epoch": 0.8198404694233061,
+      "learning_rate": 3.6033739800128356e-05,
+      "loss": 0.9921,
+      "mean_token_accuracy": 0.7037769675254821,
+      "num_tokens": 216775567.0,
+      "step": 89420
+    },
+    {
+      "epoch": 0.8199321536627854,
+      "learning_rate": 3.6015402952232514e-05,
+      "loss": 1.027,
+      "mean_token_accuracy": 0.6993145942687988,
+      "num_tokens": 216806757.0,
+      "step": 89430
+    },
+    {
+      "epoch": 0.8200238379022646,
+      "learning_rate": 3.5997066104336666e-05,
+      "loss": 0.9929,
+      "mean_token_accuracy": 0.704660850763321,
+      "num_tokens": 216837817.0,
+      "step": 89440
+    },
+    {
+      "epoch": 0.8201155221417439,
+      "learning_rate": 3.597872925644082e-05,
+      "loss": 1.0033,
+      "mean_token_accuracy": 0.7048147082328796,
+      "num_tokens": 216867635.0,
+      "step": 89450
+    },
+    {
+      "epoch": 0.820207206381223,
+      "learning_rate": 3.596039240854498e-05,
+      "loss": 0.9987,
+      "mean_token_accuracy": 0.7068140983581543,
+      "num_tokens": 216899603.0,
+      "step": 89460
+    },
+    {
+      "epoch": 0.8202988906207023,
+      "learning_rate": 3.594205556064912e-05,
+      "loss": 1.0177,
+      "mean_token_accuracy": 0.7009139716625213,
+      "num_tokens": 216930405.0,
+      "step": 89470
+    },
+    {
+      "epoch": 0.8203905748601815,
+      "learning_rate": 3.592371871275328e-05,
+      "loss": 0.9835,
+      "mean_token_accuracy": 0.7070664763450623,
+      "num_tokens": 216961820.0,
+      "step": 89480
+    },
+    {
+      "epoch": 0.8204822590996608,
+      "learning_rate": 3.590538186485743e-05,
+      "loss": 1.0187,
+      "mean_token_accuracy": 0.6993075489997864,
+      "num_tokens": 216992694.0,
+      "step": 89490
+    },
+    {
+      "epoch": 0.82057394333914,
+      "learning_rate": 3.5887045016961584e-05,
+      "loss": 1.0052,
+      "mean_token_accuracy": 0.6994746148586273,
+      "num_tokens": 217024756.0,
+      "step": 89500
+    },
+    {
+      "epoch": 0.8206656275786193,
+      "learning_rate": 3.586870816906574e-05,
+      "loss": 0.9744,
+      "mean_token_accuracy": 0.7065632164478302,
+      "num_tokens": 217052681.0,
+      "step": 89510
+    },
+    {
+      "epoch": 0.8207573118180985,
+      "learning_rate": 3.585037132116989e-05,
+      "loss": 0.9998,
+      "mean_token_accuracy": 0.7040503263473511,
+      "num_tokens": 217082496.0,
+      "step": 89520
+    },
+    {
+      "epoch": 0.8208489960575777,
+      "learning_rate": 3.583203447327405e-05,
+      "loss": 1.0255,
+      "mean_token_accuracy": 0.7011173784732818,
+      "num_tokens": 217113957.0,
+      "step": 89530
+    },
+    {
+      "epoch": 0.8209406802970569,
+      "learning_rate": 3.58136976253782e-05,
+      "loss": 0.9932,
+      "mean_token_accuracy": 0.7075741112232208,
+      "num_tokens": 217145042.0,
+      "step": 89540
+    },
+    {
+      "epoch": 0.8210323645365362,
+      "learning_rate": 3.579536077748235e-05,
+      "loss": 0.9658,
+      "mean_token_accuracy": 0.7145200908184052,
+      "num_tokens": 217176799.0,
+      "step": 89550
+    },
+    {
+      "epoch": 0.8211240487760154,
+      "learning_rate": 3.577702392958651e-05,
+      "loss": 0.9976,
+      "mean_token_accuracy": 0.7022757530212402,
+      "num_tokens": 217207580.0,
+      "step": 89560
+    },
+    {
+      "epoch": 0.8212157330154947,
+      "learning_rate": 3.575868708169066e-05,
+      "loss": 0.9734,
+      "mean_token_accuracy": 0.7052068829536438,
+      "num_tokens": 217238051.0,
+      "step": 89570
+    },
+    {
+      "epoch": 0.8213074172549739,
+      "learning_rate": 3.574035023379481e-05,
+      "loss": 1.0238,
+      "mean_token_accuracy": 0.6927455723285675,
+      "num_tokens": 217267887.0,
+      "step": 89580
+    },
+    {
+      "epoch": 0.821399101494453,
+      "learning_rate": 3.5722013385898965e-05,
+      "loss": 0.9652,
+      "mean_token_accuracy": 0.7125248312950134,
+      "num_tokens": 217299104.0,
+      "step": 89590
+    },
+    {
+      "epoch": 0.8214907857339323,
+      "learning_rate": 3.5703676538003117e-05,
+      "loss": 1.0341,
+      "mean_token_accuracy": 0.697223562002182,
+      "num_tokens": 217329923.0,
+      "step": 89600
+    },
+    {
+      "epoch": 0.8215824699734116,
+      "learning_rate": 3.5685339690107275e-05,
+      "loss": 1.0219,
+      "mean_token_accuracy": 0.6976146042346955,
+      "num_tokens": 217361147.0,
+      "step": 89610
+    },
+    {
+      "epoch": 0.8216741542128908,
+      "learning_rate": 3.566700284221143e-05,
+      "loss": 1.0214,
+      "mean_token_accuracy": 0.7045253276824951,
+      "num_tokens": 217392845.0,
+      "step": 89620
+    },
+    {
+      "epoch": 0.8217658384523701,
+      "learning_rate": 3.564866599431558e-05,
+      "loss": 1.0016,
+      "mean_token_accuracy": 0.7000697910785675,
+      "num_tokens": 217422580.0,
+      "step": 89630
+    },
+    {
+      "epoch": 0.8218575226918493,
+      "learning_rate": 3.563032914641973e-05,
+      "loss": 1.0179,
+      "mean_token_accuracy": 0.6992241501808166,
+      "num_tokens": 217450942.0,
+      "step": 89640
+    },
+    {
+      "epoch": 0.8219492069313286,
+      "learning_rate": 3.561199229852388e-05,
+      "loss": 1.0205,
+      "mean_token_accuracy": 0.6992136180400849,
+      "num_tokens": 217482403.0,
+      "step": 89650
+    },
+    {
+      "epoch": 0.8220408911708077,
+      "learning_rate": 3.559365545062804e-05,
+      "loss": 1.0334,
+      "mean_token_accuracy": 0.6970073759555817,
+      "num_tokens": 217514582.0,
+      "step": 89660
+    },
+    {
+      "epoch": 0.822132575410287,
+      "learning_rate": 3.557531860273219e-05,
+      "loss": 0.9955,
+      "mean_token_accuracy": 0.7020188570022583,
+      "num_tokens": 217546495.0,
+      "step": 89670
+    },
+    {
+      "epoch": 0.8222242596497662,
+      "learning_rate": 3.5556981754836345e-05,
+      "loss": 1.0184,
+      "mean_token_accuracy": 0.6983431875705719,
+      "num_tokens": 217578598.0,
+      "step": 89680
+    },
+    {
+      "epoch": 0.8223159438892454,
+      "learning_rate": 3.55386449069405e-05,
+      "loss": 1.0252,
+      "mean_token_accuracy": 0.6960439920425415,
+      "num_tokens": 217609151.0,
+      "step": 89690
+    },
+    {
+      "epoch": 0.8224076281287247,
+      "learning_rate": 3.5520308059044656e-05,
+      "loss": 0.9948,
+      "mean_token_accuracy": 0.6960030615329742,
+      "num_tokens": 217638871.0,
+      "step": 89700
+    },
+    {
+      "epoch": 0.822499312368204,
+      "learning_rate": 3.550197121114881e-05,
+      "loss": 1.0476,
+      "mean_token_accuracy": 0.6921974420547485,
+      "num_tokens": 217670232.0,
+      "step": 89710
+    },
+    {
+      "epoch": 0.8225909966076831,
+      "learning_rate": 3.548363436325296e-05,
+      "loss": 1.0083,
+      "mean_token_accuracy": 0.7024901568889618,
+      "num_tokens": 217701328.0,
+      "step": 89720
+    },
+    {
+      "epoch": 0.8226826808471623,
+      "learning_rate": 3.546529751535711e-05,
+      "loss": 1.0081,
+      "mean_token_accuracy": 0.7058868885040284,
+      "num_tokens": 217732749.0,
+      "step": 89730
+    },
+    {
+      "epoch": 0.8227743650866416,
+      "learning_rate": 3.544696066746126e-05,
+      "loss": 1.0074,
+      "mean_token_accuracy": 0.6999070465564727,
+      "num_tokens": 217764441.0,
+      "step": 89740
+    },
+    {
+      "epoch": 0.8228660493261208,
+      "learning_rate": 3.542862381956542e-05,
+      "loss": 1.0087,
+      "mean_token_accuracy": 0.6974696278572082,
+      "num_tokens": 217794611.0,
+      "step": 89750
+    },
+    {
+      "epoch": 0.8229577335656001,
+      "learning_rate": 3.5410286971669573e-05,
+      "loss": 0.9934,
+      "mean_token_accuracy": 0.7009398221969605,
+      "num_tokens": 217824505.0,
+      "step": 89760
+    },
+    {
+      "epoch": 0.8230494178050793,
+      "learning_rate": 3.5391950123773725e-05,
+      "loss": 1.0075,
+      "mean_token_accuracy": 0.7024871706962585,
+      "num_tokens": 217856092.0,
+      "step": 89770
+    },
+    {
+      "epoch": 0.8231411020445586,
+      "learning_rate": 3.537361327587788e-05,
+      "loss": 1.0062,
+      "mean_token_accuracy": 0.7027764260768891,
+      "num_tokens": 217888141.0,
+      "step": 89780
+    },
+    {
+      "epoch": 0.8232327862840377,
+      "learning_rate": 3.535527642798203e-05,
+      "loss": 1.0308,
+      "mean_token_accuracy": 0.6936926186084748,
+      "num_tokens": 217919838.0,
+      "step": 89790
+    },
+    {
+      "epoch": 0.823324470523517,
+      "learning_rate": 3.533693958008619e-05,
+      "loss": 0.9868,
+      "mean_token_accuracy": 0.7085601270198822,
+      "num_tokens": 217948838.0,
+      "step": 89800
+    },
+    {
+      "epoch": 0.8234161547629962,
+      "learning_rate": 3.531860273219034e-05,
+      "loss": 0.9783,
+      "mean_token_accuracy": 0.708295714855194,
+      "num_tokens": 217980138.0,
+      "step": 89810
+    },
+    {
+      "epoch": 0.8235078390024755,
+      "learning_rate": 3.530026588429449e-05,
+      "loss": 0.9814,
+      "mean_token_accuracy": 0.7044698834419251,
+      "num_tokens": 218010471.0,
+      "step": 89820
+    },
+    {
+      "epoch": 0.8235995232419547,
+      "learning_rate": 3.528192903639864e-05,
+      "loss": 1.0154,
+      "mean_token_accuracy": 0.7017762362957001,
+      "num_tokens": 218042498.0,
+      "step": 89830
+    },
+    {
+      "epoch": 0.823691207481434,
+      "learning_rate": 3.5263592188502795e-05,
+      "loss": 1.0158,
+      "mean_token_accuracy": 0.6979267001152039,
+      "num_tokens": 218074215.0,
+      "step": 89840
+    },
+    {
+      "epoch": 0.8237828917209131,
+      "learning_rate": 3.5245255340606954e-05,
+      "loss": 1.0274,
+      "mean_token_accuracy": 0.7039165735244751,
+      "num_tokens": 218105056.0,
+      "step": 89850
+    },
+    {
+      "epoch": 0.8238745759603924,
+      "learning_rate": 3.5226918492711106e-05,
+      "loss": 1.0199,
+      "mean_token_accuracy": 0.6995380878448486,
+      "num_tokens": 218136579.0,
+      "step": 89860
+    },
+    {
+      "epoch": 0.8239662601998716,
+      "learning_rate": 3.520858164481526e-05,
+      "loss": 1.0113,
+      "mean_token_accuracy": 0.6981495440006256,
+      "num_tokens": 218167125.0,
+      "step": 89870
+    },
+    {
+      "epoch": 0.8240579444393509,
+      "learning_rate": 3.5190244796919416e-05,
+      "loss": 1.0159,
+      "mean_token_accuracy": 0.6977125823497772,
+      "num_tokens": 218196264.0,
+      "step": 89880
+    },
+    {
+      "epoch": 0.8241496286788301,
+      "learning_rate": 3.517190794902356e-05,
+      "loss": 1.0179,
+      "mean_token_accuracy": 0.7013065934181213,
+      "num_tokens": 218227752.0,
+      "step": 89890
+    },
+    {
+      "epoch": 0.8242413129183094,
+      "learning_rate": 3.515357110112772e-05,
+      "loss": 1.0255,
+      "mean_token_accuracy": 0.6967093110084533,
+      "num_tokens": 218259509.0,
+      "step": 89900
+    },
+    {
+      "epoch": 0.8243329971577886,
+      "learning_rate": 3.513523425323187e-05,
+      "loss": 0.9907,
+      "mean_token_accuracy": 0.7059114515781403,
+      "num_tokens": 218290287.0,
+      "step": 89910
+    },
+    {
+      "epoch": 0.8244246813972678,
+      "learning_rate": 3.5116897405336024e-05,
+      "loss": 1.0226,
+      "mean_token_accuracy": 0.7021823167800904,
+      "num_tokens": 218321876.0,
+      "step": 89920
+    },
+    {
+      "epoch": 0.824516365636747,
+      "learning_rate": 3.509856055744018e-05,
+      "loss": 1.0221,
+      "mean_token_accuracy": 0.7003953516483307,
+      "num_tokens": 218352743.0,
+      "step": 89930
+    },
+    {
+      "epoch": 0.8246080498762263,
+      "learning_rate": 3.508022370954433e-05,
+      "loss": 0.9892,
+      "mean_token_accuracy": 0.7012200355529785,
+      "num_tokens": 218382613.0,
+      "step": 89940
+    },
+    {
+      "epoch": 0.8246997341157055,
+      "learning_rate": 3.5061886861648486e-05,
+      "loss": 1.0293,
+      "mean_token_accuracy": 0.6971102952957153,
+      "num_tokens": 218413926.0,
+      "step": 89950
+    },
+    {
+      "epoch": 0.8247914183551848,
+      "learning_rate": 3.504355001375264e-05,
+      "loss": 1.0004,
+      "mean_token_accuracy": 0.7024656474590302,
+      "num_tokens": 218445904.0,
+      "step": 89960
+    },
+    {
+      "epoch": 0.824883102594664,
+      "learning_rate": 3.502521316585679e-05,
+      "loss": 0.9816,
+      "mean_token_accuracy": 0.7075865745544434,
+      "num_tokens": 218477458.0,
+      "step": 89970
+    },
+    {
+      "epoch": 0.8249747868341432,
+      "learning_rate": 3.500687631796095e-05,
+      "loss": 1.0043,
+      "mean_token_accuracy": 0.7051225125789642,
+      "num_tokens": 218507963.0,
+      "step": 89980
+    },
+    {
+      "epoch": 0.8250664710736224,
+      "learning_rate": 3.4988539470065094e-05,
+      "loss": 1.0097,
+      "mean_token_accuracy": 0.7018232405185699,
+      "num_tokens": 218538219.0,
+      "step": 89990
+    },
+    {
+      "epoch": 0.8251581553131017,
+      "learning_rate": 3.497020262216925e-05,
+      "loss": 1.0053,
+      "mean_token_accuracy": 0.7008326649665833,
+      "num_tokens": 218567533.0,
+      "step": 90000
+    },
+    {
+      "epoch": 0.8252498395525809,
+      "learning_rate": 3.4951865774273404e-05,
+      "loss": 1.0281,
+      "mean_token_accuracy": 0.7011227130889892,
+      "num_tokens": 218599491.0,
+      "step": 90010
+    },
+    {
+      "epoch": 0.8253415237920602,
+      "learning_rate": 3.4933528926377556e-05,
+      "loss": 1.0019,
+      "mean_token_accuracy": 0.7045192062854767,
+      "num_tokens": 218631049.0,
+      "step": 90020
+    },
+    {
+      "epoch": 0.8254332080315394,
+      "learning_rate": 3.4915192078481715e-05,
+      "loss": 1.0167,
+      "mean_token_accuracy": 0.700628525018692,
+      "num_tokens": 218661502.0,
+      "step": 90030
+    },
+    {
+      "epoch": 0.8255248922710187,
+      "learning_rate": 3.489685523058586e-05,
+      "loss": 0.997,
+      "mean_token_accuracy": 0.7090557515621185,
+      "num_tokens": 218692564.0,
+      "step": 90040
+    },
+    {
+      "epoch": 0.8256165765104978,
+      "learning_rate": 3.487851838269002e-05,
+      "loss": 1.011,
+      "mean_token_accuracy": 0.6990772426128388,
+      "num_tokens": 218722354.0,
+      "step": 90050
+    },
+    {
+      "epoch": 0.825708260749977,
+      "learning_rate": 3.486018153479417e-05,
+      "loss": 1.0096,
+      "mean_token_accuracy": 0.700513380765915,
+      "num_tokens": 218752899.0,
+      "step": 90060
+    },
+    {
+      "epoch": 0.8257999449894563,
+      "learning_rate": 3.484184468689832e-05,
+      "loss": 1.0326,
+      "mean_token_accuracy": 0.7009813487529755,
+      "num_tokens": 218784013.0,
+      "step": 90070
+    },
+    {
+      "epoch": 0.8258916292289356,
+      "learning_rate": 3.482350783900248e-05,
+      "loss": 1.0228,
+      "mean_token_accuracy": 0.6992367684841156,
+      "num_tokens": 218815934.0,
+      "step": 90080
+    },
+    {
+      "epoch": 0.8259833134684148,
+      "learning_rate": 3.4805170991106626e-05,
+      "loss": 1.0141,
+      "mean_token_accuracy": 0.7031391203403473,
+      "num_tokens": 218846950.0,
+      "step": 90090
+    },
+    {
+      "epoch": 0.826074997707894,
+      "learning_rate": 3.4786834143210784e-05,
+      "loss": 1.0186,
+      "mean_token_accuracy": 0.6979718327522277,
+      "num_tokens": 218879495.0,
+      "step": 90100
+    },
+    {
+      "epoch": 0.8261666819473732,
+      "learning_rate": 3.4768497295314936e-05,
+      "loss": 1.0066,
+      "mean_token_accuracy": 0.697433739900589,
+      "num_tokens": 218909582.0,
+      "step": 90110
+    },
+    {
+      "epoch": 0.8262583661868524,
+      "learning_rate": 3.475016044741909e-05,
+      "loss": 0.9725,
+      "mean_token_accuracy": 0.7097954154014587,
+      "num_tokens": 218938605.0,
+      "step": 90120
+    },
+    {
+      "epoch": 0.8263500504263317,
+      "learning_rate": 3.473182359952325e-05,
+      "loss": 1.0176,
+      "mean_token_accuracy": 0.6963508069515228,
+      "num_tokens": 218970352.0,
+      "step": 90130
+    },
+    {
+      "epoch": 0.826441734665811,
+      "learning_rate": 3.471348675162739e-05,
+      "loss": 1.0314,
+      "mean_token_accuracy": 0.6978467702865601,
+      "num_tokens": 219001659.0,
+      "step": 90140
+    },
+    {
+      "epoch": 0.8265334189052902,
+      "learning_rate": 3.469514990373155e-05,
+      "loss": 1.0079,
+      "mean_token_accuracy": 0.7022634148597717,
+      "num_tokens": 219032760.0,
+      "step": 90150
+    },
+    {
+      "epoch": 0.8266251031447694,
+      "learning_rate": 3.46768130558357e-05,
+      "loss": 1.004,
+      "mean_token_accuracy": 0.6996465802192688,
+      "num_tokens": 219063722.0,
+      "step": 90160
+    },
+    {
+      "epoch": 0.8267167873842487,
+      "learning_rate": 3.4658476207939854e-05,
+      "loss": 1.0202,
+      "mean_token_accuracy": 0.7048012614250183,
+      "num_tokens": 219093961.0,
+      "step": 90170
+    },
+    {
+      "epoch": 0.8268084716237278,
+      "learning_rate": 3.464013936004401e-05,
+      "loss": 0.9807,
+      "mean_token_accuracy": 0.7043280661106109,
+      "num_tokens": 219121756.0,
+      "step": 90180
+    },
+    {
+      "epoch": 0.8269001558632071,
+      "learning_rate": 3.4621802512148165e-05,
+      "loss": 0.9924,
+      "mean_token_accuracy": 0.7078893721103668,
+      "num_tokens": 219153052.0,
+      "step": 90190
+    },
+    {
+      "epoch": 0.8269918401026863,
+      "learning_rate": 3.460346566425232e-05,
+      "loss": 1.0259,
+      "mean_token_accuracy": 0.6989540219306946,
+      "num_tokens": 219184459.0,
+      "step": 90200
+    },
+    {
+      "epoch": 0.8270835243421656,
+      "learning_rate": 3.458512881635647e-05,
+      "loss": 1.0017,
+      "mean_token_accuracy": 0.7045412361621857,
+      "num_tokens": 219215437.0,
+      "step": 90210
+    },
+    {
+      "epoch": 0.8271752085816448,
+      "learning_rate": 3.456679196846062e-05,
+      "loss": 1.0165,
+      "mean_token_accuracy": 0.7025941193103791,
+      "num_tokens": 219247036.0,
+      "step": 90220
+    },
+    {
+      "epoch": 0.8272668928211241,
+      "learning_rate": 3.454845512056478e-05,
+      "loss": 1.0215,
+      "mean_token_accuracy": 0.698662930727005,
+      "num_tokens": 219277685.0,
+      "step": 90230
+    },
+    {
+      "epoch": 0.8273585770606032,
+      "learning_rate": 3.453011827266893e-05,
+      "loss": 1.0025,
+      "mean_token_accuracy": 0.7031780660152436,
+      "num_tokens": 219307850.0,
+      "step": 90240
+    },
+    {
+      "epoch": 0.8274502613000825,
+      "learning_rate": 3.451178142477308e-05,
+      "loss": 1.008,
+      "mean_token_accuracy": 0.7078184187412262,
+      "num_tokens": 219338499.0,
+      "step": 90250
+    },
+    {
+      "epoch": 0.8275419455395617,
+      "learning_rate": 3.4493444576877235e-05,
+      "loss": 0.9852,
+      "mean_token_accuracy": 0.7059654235839844,
+      "num_tokens": 219370701.0,
+      "step": 90260
+    },
+    {
+      "epoch": 0.827633629779041,
+      "learning_rate": 3.4475107728981386e-05,
+      "loss": 1.017,
+      "mean_token_accuracy": 0.6988462030887603,
+      "num_tokens": 219401369.0,
+      "step": 90270
+    },
+    {
+      "epoch": 0.8277253140185202,
+      "learning_rate": 3.4456770881085545e-05,
+      "loss": 1.0309,
+      "mean_token_accuracy": 0.6990213990211487,
+      "num_tokens": 219434393.0,
+      "step": 90280
+    },
+    {
+      "epoch": 0.8278169982579995,
+      "learning_rate": 3.44384340331897e-05,
+      "loss": 0.9756,
+      "mean_token_accuracy": 0.706069940328598,
+      "num_tokens": 219463910.0,
+      "step": 90290
+    },
+    {
+      "epoch": 0.8279086824974787,
+      "learning_rate": 3.442009718529385e-05,
+      "loss": 0.9971,
+      "mean_token_accuracy": 0.7024085760116577,
+      "num_tokens": 219492780.0,
+      "step": 90300
+    },
+    {
+      "epoch": 0.8280003667369579,
+      "learning_rate": 3.4401760337398e-05,
+      "loss": 1.0256,
+      "mean_token_accuracy": 0.6982463598251343,
+      "num_tokens": 219524639.0,
+      "step": 90310
+    },
+    {
+      "epoch": 0.8280920509764371,
+      "learning_rate": 3.438342348950216e-05,
+      "loss": 1.0181,
+      "mean_token_accuracy": 0.6987064480781555,
+      "num_tokens": 219556189.0,
+      "step": 90320
+    },
+    {
+      "epoch": 0.8281837352159164,
+      "learning_rate": 3.436508664160631e-05,
+      "loss": 1.0209,
+      "mean_token_accuracy": 0.700279176235199,
+      "num_tokens": 219586879.0,
+      "step": 90330
+    },
+    {
+      "epoch": 0.8282754194553956,
+      "learning_rate": 3.434674979371046e-05,
+      "loss": 1.0371,
+      "mean_token_accuracy": 0.6897796988487244,
+      "num_tokens": 219618058.0,
+      "step": 90340
+    },
+    {
+      "epoch": 0.8283671036948749,
+      "learning_rate": 3.4328412945814615e-05,
+      "loss": 1.034,
+      "mean_token_accuracy": 0.6963342905044556,
+      "num_tokens": 219647893.0,
+      "step": 90350
+    },
+    {
+      "epoch": 0.8284587879343541,
+      "learning_rate": 3.431007609791877e-05,
+      "loss": 0.9868,
+      "mean_token_accuracy": 0.7044245958328247,
+      "num_tokens": 219678596.0,
+      "step": 90360
+    },
+    {
+      "epoch": 0.8285504721738333,
+      "learning_rate": 3.4291739250022925e-05,
+      "loss": 1.0159,
+      "mean_token_accuracy": 0.7038574755191803,
+      "num_tokens": 219709534.0,
+      "step": 90370
+    },
+    {
+      "epoch": 0.8286421564133125,
+      "learning_rate": 3.427340240212708e-05,
+      "loss": 1.0055,
+      "mean_token_accuracy": 0.7060499966144562,
+      "num_tokens": 219739381.0,
+      "step": 90380
+    },
+    {
+      "epoch": 0.8287338406527918,
+      "learning_rate": 3.425506555423123e-05,
+      "loss": 0.9914,
+      "mean_token_accuracy": 0.7050133585929871,
+      "num_tokens": 219770230.0,
+      "step": 90390
+    },
+    {
+      "epoch": 0.828825524892271,
+      "learning_rate": 3.423672870633538e-05,
+      "loss": 1.0049,
+      "mean_token_accuracy": 0.7040414988994599,
+      "num_tokens": 219800660.0,
+      "step": 90400
+    },
+    {
+      "epoch": 0.8289172091317503,
+      "learning_rate": 3.421839185843953e-05,
+      "loss": 1.0395,
+      "mean_token_accuracy": 0.6939877152442933,
+      "num_tokens": 219831733.0,
+      "step": 90410
+    },
+    {
+      "epoch": 0.8290088933712295,
+      "learning_rate": 3.420005501054369e-05,
+      "loss": 1.0159,
+      "mean_token_accuracy": 0.698121440410614,
+      "num_tokens": 219861302.0,
+      "step": 90420
+    },
+    {
+      "epoch": 0.8291005776107088,
+      "learning_rate": 3.4181718162647843e-05,
+      "loss": 1.003,
+      "mean_token_accuracy": 0.6998128890991211,
+      "num_tokens": 219891627.0,
+      "step": 90430
+    },
+    {
+      "epoch": 0.8291922618501879,
+      "learning_rate": 3.4163381314751995e-05,
+      "loss": 1.0458,
+      "mean_token_accuracy": 0.6945520102977752,
+      "num_tokens": 219923847.0,
+      "step": 90440
+    },
+    {
+      "epoch": 0.8292839460896672,
+      "learning_rate": 3.414504446685615e-05,
+      "loss": 0.9994,
+      "mean_token_accuracy": 0.7000109255313873,
+      "num_tokens": 219955720.0,
+      "step": 90450
+    },
+    {
+      "epoch": 0.8293756303291464,
+      "learning_rate": 3.41267076189603e-05,
+      "loss": 1.0181,
+      "mean_token_accuracy": 0.6990287780761719,
+      "num_tokens": 219987438.0,
+      "step": 90460
+    },
+    {
+      "epoch": 0.8294673145686257,
+      "learning_rate": 3.410837077106446e-05,
+      "loss": 1.0016,
+      "mean_token_accuracy": 0.7014613449573517,
+      "num_tokens": 220016924.0,
+      "step": 90470
+    },
+    {
+      "epoch": 0.8295589988081049,
+      "learning_rate": 3.409003392316861e-05,
+      "loss": 1.0117,
+      "mean_token_accuracy": 0.6967711329460144,
+      "num_tokens": 220046028.0,
+      "step": 90480
+    },
+    {
+      "epoch": 0.8296506830475842,
+      "learning_rate": 3.407169707527276e-05,
+      "loss": 1.0282,
+      "mean_token_accuracy": 0.7010338187217713,
+      "num_tokens": 220077468.0,
+      "step": 90490
+    },
+    {
+      "epoch": 0.8297423672870634,
+      "learning_rate": 3.405336022737692e-05,
+      "loss": 1.0289,
+      "mean_token_accuracy": 0.6963151395320892,
+      "num_tokens": 220109071.0,
+      "step": 90500
+    },
+    {
+      "epoch": 0.8298340515265425,
+      "learning_rate": 3.4035023379481065e-05,
+      "loss": 0.9816,
+      "mean_token_accuracy": 0.7081152498722076,
+      "num_tokens": 220140515.0,
+      "step": 90510
+    },
+    {
+      "epoch": 0.8299257357660218,
+      "learning_rate": 3.4016686531585224e-05,
+      "loss": 1.0178,
+      "mean_token_accuracy": 0.6993720293045044,
+      "num_tokens": 220170837.0,
+      "step": 90520
+    },
+    {
+      "epoch": 0.830017420005501,
+      "learning_rate": 3.3998349683689376e-05,
+      "loss": 1.0202,
+      "mean_token_accuracy": 0.7022325396537781,
+      "num_tokens": 220201937.0,
+      "step": 90530
+    },
+    {
+      "epoch": 0.8301091042449803,
+      "learning_rate": 3.398001283579353e-05,
+      "loss": 1.0012,
+      "mean_token_accuracy": 0.6982310712337494,
+      "num_tokens": 220231434.0,
+      "step": 90540
+    },
+    {
+      "epoch": 0.8302007884844596,
+      "learning_rate": 3.3961675987897686e-05,
+      "loss": 1.0367,
+      "mean_token_accuracy": 0.699015998840332,
+      "num_tokens": 220263165.0,
+      "step": 90550
+    },
+    {
+      "epoch": 0.8302924727239388,
+      "learning_rate": 3.394333914000183e-05,
+      "loss": 1.0574,
+      "mean_token_accuracy": 0.6866241574287415,
+      "num_tokens": 220295248.0,
+      "step": 90560
+    },
+    {
+      "epoch": 0.8303841569634179,
+      "learning_rate": 3.392500229210599e-05,
+      "loss": 1.0033,
+      "mean_token_accuracy": 0.7033745348453522,
+      "num_tokens": 220326770.0,
+      "step": 90570
+    },
+    {
+      "epoch": 0.8304758412028972,
+      "learning_rate": 3.390666544421014e-05,
+      "loss": 0.984,
+      "mean_token_accuracy": 0.7064521014690399,
+      "num_tokens": 220357751.0,
+      "step": 90580
+    },
+    {
+      "epoch": 0.8305675254423764,
+      "learning_rate": 3.3888328596314294e-05,
+      "loss": 1.0167,
+      "mean_token_accuracy": 0.6974581778049469,
+      "num_tokens": 220388449.0,
+      "step": 90590
+    },
+    {
+      "epoch": 0.8306592096818557,
+      "learning_rate": 3.386999174841845e-05,
+      "loss": 1.0096,
+      "mean_token_accuracy": 0.6993870735168457,
+      "num_tokens": 220418448.0,
+      "step": 90600
+    },
+    {
+      "epoch": 0.8307508939213349,
+      "learning_rate": 3.38516549005226e-05,
+      "loss": 1.0177,
+      "mean_token_accuracy": 0.7009082913398743,
+      "num_tokens": 220450316.0,
+      "step": 90610
+    },
+    {
+      "epoch": 0.8308425781608142,
+      "learning_rate": 3.3833318052626756e-05,
+      "loss": 1.0182,
+      "mean_token_accuracy": 0.6961587607860565,
+      "num_tokens": 220482135.0,
+      "step": 90620
+    },
+    {
+      "epoch": 0.8309342624002934,
+      "learning_rate": 3.381498120473091e-05,
+      "loss": 0.9942,
+      "mean_token_accuracy": 0.7034704446792602,
+      "num_tokens": 220514095.0,
+      "step": 90630
+    },
+    {
+      "epoch": 0.8310259466397726,
+      "learning_rate": 3.379664435683506e-05,
+      "loss": 1.0255,
+      "mean_token_accuracy": 0.6975188851356506,
+      "num_tokens": 220545431.0,
+      "step": 90640
+    },
+    {
+      "epoch": 0.8311176308792518,
+      "learning_rate": 3.377830750893922e-05,
+      "loss": 1.0203,
+      "mean_token_accuracy": 0.6978191614151001,
+      "num_tokens": 220575769.0,
+      "step": 90650
+    },
+    {
+      "epoch": 0.8312093151187311,
+      "learning_rate": 3.3759970661043364e-05,
+      "loss": 0.9952,
+      "mean_token_accuracy": 0.6992539465427399,
+      "num_tokens": 220604764.0,
+      "step": 90660
+    },
+    {
+      "epoch": 0.8313009993582103,
+      "learning_rate": 3.374163381314752e-05,
+      "loss": 1.0284,
+      "mean_token_accuracy": 0.6933148920536041,
+      "num_tokens": 220636782.0,
+      "step": 90670
+    },
+    {
+      "epoch": 0.8313926835976896,
+      "learning_rate": 3.3723296965251674e-05,
+      "loss": 1.0167,
+      "mean_token_accuracy": 0.6988054394721985,
+      "num_tokens": 220668817.0,
+      "step": 90680
+    },
+    {
+      "epoch": 0.8314843678371688,
+      "learning_rate": 3.3704960117355826e-05,
+      "loss": 1.0148,
+      "mean_token_accuracy": 0.6956416130065918,
+      "num_tokens": 220699556.0,
+      "step": 90690
+    },
+    {
+      "epoch": 0.831576052076648,
+      "learning_rate": 3.3686623269459985e-05,
+      "loss": 1.0118,
+      "mean_token_accuracy": 0.6987706780433655,
+      "num_tokens": 220730887.0,
+      "step": 90700
+    },
+    {
+      "epoch": 0.8316677363161272,
+      "learning_rate": 3.366828642156413e-05,
+      "loss": 1.0083,
+      "mean_token_accuracy": 0.7023152768611908,
+      "num_tokens": 220761398.0,
+      "step": 90710
+    },
+    {
+      "epoch": 0.8317594205556065,
+      "learning_rate": 3.364994957366829e-05,
+      "loss": 1.0054,
+      "mean_token_accuracy": 0.7016844093799591,
+      "num_tokens": 220791129.0,
+      "step": 90720
+    },
+    {
+      "epoch": 0.8318511047950857,
+      "learning_rate": 3.363161272577244e-05,
+      "loss": 0.9968,
+      "mean_token_accuracy": 0.7054231762886047,
+      "num_tokens": 220822540.0,
+      "step": 90730
+    },
+    {
+      "epoch": 0.831942789034565,
+      "learning_rate": 3.361327587787659e-05,
+      "loss": 1.0025,
+      "mean_token_accuracy": 0.7013937592506408,
+      "num_tokens": 220852936.0,
+      "step": 90740
+    },
+    {
+      "epoch": 0.8320344732740442,
+      "learning_rate": 3.359493902998075e-05,
+      "loss": 1.0188,
+      "mean_token_accuracy": 0.7003185212612152,
+      "num_tokens": 220884434.0,
+      "step": 90750
+    },
+    {
+      "epoch": 0.8321261575135235,
+      "learning_rate": 3.3576602182084896e-05,
+      "loss": 1.0149,
+      "mean_token_accuracy": 0.6970611453056336,
+      "num_tokens": 220916046.0,
+      "step": 90760
+    },
+    {
+      "epoch": 0.8322178417530026,
+      "learning_rate": 3.3558265334189054e-05,
+      "loss": 1.0136,
+      "mean_token_accuracy": 0.6984663009643555,
+      "num_tokens": 220946779.0,
+      "step": 90770
+    },
+    {
+      "epoch": 0.8323095259924819,
+      "learning_rate": 3.3539928486293206e-05,
+      "loss": 0.9984,
+      "mean_token_accuracy": 0.7013566851615906,
+      "num_tokens": 220976094.0,
+      "step": 90780
+    },
+    {
+      "epoch": 0.8324012102319611,
+      "learning_rate": 3.352159163839736e-05,
+      "loss": 1.0262,
+      "mean_token_accuracy": 0.6944224417209626,
+      "num_tokens": 221008597.0,
+      "step": 90790
+    },
+    {
+      "epoch": 0.8324928944714404,
+      "learning_rate": 3.350325479050152e-05,
+      "loss": 1.009,
+      "mean_token_accuracy": 0.7017030000686646,
+      "num_tokens": 221039625.0,
+      "step": 90800
+    },
+    {
+      "epoch": 0.8325845787109196,
+      "learning_rate": 3.348491794260567e-05,
+      "loss": 1.0247,
+      "mean_token_accuracy": 0.6982905983924865,
+      "num_tokens": 221071119.0,
+      "step": 90810
+    },
+    {
+      "epoch": 0.8326762629503989,
+      "learning_rate": 3.346658109470982e-05,
+      "loss": 1.0039,
+      "mean_token_accuracy": 0.695676851272583,
+      "num_tokens": 221101358.0,
+      "step": 90820
+    },
+    {
+      "epoch": 0.832767947189878,
+      "learning_rate": 3.344824424681397e-05,
+      "loss": 1.0041,
+      "mean_token_accuracy": 0.7050034761428833,
+      "num_tokens": 221131862.0,
+      "step": 90830
+    },
+    {
+      "epoch": 0.8328596314293573,
+      "learning_rate": 3.3429907398918124e-05,
+      "loss": 1.01,
+      "mean_token_accuracy": 0.6967243671417236,
+      "num_tokens": 221161063.0,
+      "step": 90840
+    },
+    {
+      "epoch": 0.8329513156688365,
+      "learning_rate": 3.341157055102228e-05,
+      "loss": 0.9924,
+      "mean_token_accuracy": 0.701437908411026,
+      "num_tokens": 221192622.0,
+      "step": 90850
+    },
+    {
+      "epoch": 0.8330429999083158,
+      "learning_rate": 3.3393233703126435e-05,
+      "loss": 1.0119,
+      "mean_token_accuracy": 0.7018767178058625,
+      "num_tokens": 221223913.0,
+      "step": 90860
+    },
+    {
+      "epoch": 0.833134684147795,
+      "learning_rate": 3.337489685523059e-05,
+      "loss": 1.036,
+      "mean_token_accuracy": 0.6921814620494843,
+      "num_tokens": 221255890.0,
+      "step": 90870
+    },
+    {
+      "epoch": 0.8332263683872743,
+      "learning_rate": 3.335656000733474e-05,
+      "loss": 1.0031,
+      "mean_token_accuracy": 0.7054585635662078,
+      "num_tokens": 221286108.0,
+      "step": 90880
+    },
+    {
+      "epoch": 0.8333180526267535,
+      "learning_rate": 3.333822315943889e-05,
+      "loss": 0.9852,
+      "mean_token_accuracy": 0.7051441967487335,
+      "num_tokens": 221316435.0,
+      "step": 90890
+    },
+    {
+      "epoch": 0.8334097368662327,
+      "learning_rate": 3.331988631154305e-05,
+      "loss": 1.0213,
+      "mean_token_accuracy": 0.700000774860382,
+      "num_tokens": 221345928.0,
+      "step": 90900
+    },
+    {
+      "epoch": 0.8335014211057119,
+      "learning_rate": 3.33015494636472e-05,
+      "loss": 1.0119,
+      "mean_token_accuracy": 0.6981248319149017,
+      "num_tokens": 221377151.0,
+      "step": 90910
+    },
+    {
+      "epoch": 0.8335931053451912,
+      "learning_rate": 3.328321261575135e-05,
+      "loss": 1.0044,
+      "mean_token_accuracy": 0.6982454180717468,
+      "num_tokens": 221407959.0,
+      "step": 90920
+    },
+    {
+      "epoch": 0.8336847895846704,
+      "learning_rate": 3.3264875767855505e-05,
+      "loss": 1.0177,
+      "mean_token_accuracy": 0.6998045980930329,
+      "num_tokens": 221439794.0,
+      "step": 90930
+    },
+    {
+      "epoch": 0.8337764738241497,
+      "learning_rate": 3.324653891995966e-05,
+      "loss": 1.0388,
+      "mean_token_accuracy": 0.6901152074337006,
+      "num_tokens": 221472015.0,
+      "step": 90940
+    },
+    {
+      "epoch": 0.8338681580636289,
+      "learning_rate": 3.3228202072063815e-05,
+      "loss": 0.9849,
+      "mean_token_accuracy": 0.6987499892711639,
+      "num_tokens": 221502175.0,
+      "step": 90950
+    },
+    {
+      "epoch": 0.833959842303108,
+      "learning_rate": 3.320986522416797e-05,
+      "loss": 1.0011,
+      "mean_token_accuracy": 0.7005788147449493,
+      "num_tokens": 221531535.0,
+      "step": 90960
+    },
+    {
+      "epoch": 0.8340515265425873,
+      "learning_rate": 3.319152837627212e-05,
+      "loss": 1.0067,
+      "mean_token_accuracy": 0.6989462614059448,
+      "num_tokens": 221562774.0,
+      "step": 90970
+    },
+    {
+      "epoch": 0.8341432107820665,
+      "learning_rate": 3.317319152837627e-05,
+      "loss": 1.0042,
+      "mean_token_accuracy": 0.7047168850898743,
+      "num_tokens": 221593380.0,
+      "step": 90980
+    },
+    {
+      "epoch": 0.8342348950215458,
+      "learning_rate": 3.315485468048043e-05,
+      "loss": 1.003,
+      "mean_token_accuracy": 0.7033378303050994,
+      "num_tokens": 221624592.0,
+      "step": 90990
+    },
+    {
+      "epoch": 0.834326579261025,
+      "learning_rate": 3.313651783258458e-05,
+      "loss": 1.038,
+      "mean_token_accuracy": 0.6974779486656189,
+      "num_tokens": 221656119.0,
+      "step": 91000
+    },
+    {
+      "epoch": 0.8344182635005043,
+      "learning_rate": 3.311818098468873e-05,
+      "loss": 0.9984,
+      "mean_token_accuracy": 0.7083793520927429,
+      "num_tokens": 221686031.0,
+      "step": 91010
+    },
+    {
+      "epoch": 0.8345099477399835,
+      "learning_rate": 3.3099844136792885e-05,
+      "loss": 1.0157,
+      "mean_token_accuracy": 0.7002842545509338,
+      "num_tokens": 221717223.0,
+      "step": 91020
+    },
+    {
+      "epoch": 0.8346016319794627,
+      "learning_rate": 3.308150728889704e-05,
+      "loss": 1.0262,
+      "mean_token_accuracy": 0.6953075408935547,
+      "num_tokens": 221749389.0,
+      "step": 91030
+    },
+    {
+      "epoch": 0.8346933162189419,
+      "learning_rate": 3.3063170441001195e-05,
+      "loss": 1.0022,
+      "mean_token_accuracy": 0.700938618183136,
+      "num_tokens": 221779349.0,
+      "step": 91040
+    },
+    {
+      "epoch": 0.8347850004584212,
+      "learning_rate": 3.304483359310535e-05,
+      "loss": 1.0029,
+      "mean_token_accuracy": 0.703021252155304,
+      "num_tokens": 221810248.0,
+      "step": 91050
+    },
+    {
+      "epoch": 0.8348766846979004,
+      "learning_rate": 3.30264967452095e-05,
+      "loss": 0.9991,
+      "mean_token_accuracy": 0.7039738297462463,
+      "num_tokens": 221842744.0,
+      "step": 91060
+    },
+    {
+      "epoch": 0.8349683689373797,
+      "learning_rate": 3.300815989731365e-05,
+      "loss": 1.0108,
+      "mean_token_accuracy": 0.7010632276535034,
+      "num_tokens": 221871445.0,
+      "step": 91070
+    },
+    {
+      "epoch": 0.8350600531768589,
+      "learning_rate": 3.29898230494178e-05,
+      "loss": 0.9787,
+      "mean_token_accuracy": 0.7070703208446503,
+      "num_tokens": 221900249.0,
+      "step": 91080
+    },
+    {
+      "epoch": 0.8351517374163381,
+      "learning_rate": 3.297148620152196e-05,
+      "loss": 1.0075,
+      "mean_token_accuracy": 0.6969705879688263,
+      "num_tokens": 221932060.0,
+      "step": 91090
+    },
+    {
+      "epoch": 0.8352434216558173,
+      "learning_rate": 3.2953149353626113e-05,
+      "loss": 0.9713,
+      "mean_token_accuracy": 0.7090980887413025,
+      "num_tokens": 221962548.0,
+      "step": 91100
+    },
+    {
+      "epoch": 0.8353351058952966,
+      "learning_rate": 3.2934812505730265e-05,
+      "loss": 1.0061,
+      "mean_token_accuracy": 0.7025028169155121,
+      "num_tokens": 221992851.0,
+      "step": 91110
+    },
+    {
+      "epoch": 0.8354267901347758,
+      "learning_rate": 3.2916475657834424e-05,
+      "loss": 0.9818,
+      "mean_token_accuracy": 0.7086591601371766,
+      "num_tokens": 222023824.0,
+      "step": 91120
+    },
+    {
+      "epoch": 0.8355184743742551,
+      "learning_rate": 3.289813880993857e-05,
+      "loss": 0.9995,
+      "mean_token_accuracy": 0.7047827005386352,
+      "num_tokens": 222054188.0,
+      "step": 91130
+    },
+    {
+      "epoch": 0.8356101586137343,
+      "learning_rate": 3.287980196204273e-05,
+      "loss": 0.9851,
+      "mean_token_accuracy": 0.701361620426178,
+      "num_tokens": 222081929.0,
+      "step": 91140
+    },
+    {
+      "epoch": 0.8357018428532136,
+      "learning_rate": 3.286146511414688e-05,
+      "loss": 0.9817,
+      "mean_token_accuracy": 0.7063040316104889,
+      "num_tokens": 222112090.0,
+      "step": 91150
+    },
+    {
+      "epoch": 0.8357935270926927,
+      "learning_rate": 3.284312826625103e-05,
+      "loss": 1.0185,
+      "mean_token_accuracy": 0.6978872060775757,
+      "num_tokens": 222143478.0,
+      "step": 91160
+    },
+    {
+      "epoch": 0.835885211332172,
+      "learning_rate": 3.282479141835519e-05,
+      "loss": 1.0516,
+      "mean_token_accuracy": 0.6882432162761688,
+      "num_tokens": 222174845.0,
+      "step": 91170
+    },
+    {
+      "epoch": 0.8359768955716512,
+      "learning_rate": 3.2806454570459335e-05,
+      "loss": 0.9762,
+      "mean_token_accuracy": 0.7065334260463715,
+      "num_tokens": 222205944.0,
+      "step": 91180
+    },
+    {
+      "epoch": 0.8360685798111305,
+      "learning_rate": 3.2788117722563494e-05,
+      "loss": 0.9541,
+      "mean_token_accuracy": 0.7162926018238067,
+      "num_tokens": 222236445.0,
+      "step": 91190
+    },
+    {
+      "epoch": 0.8361602640506097,
+      "learning_rate": 3.2769780874667646e-05,
+      "loss": 0.9723,
+      "mean_token_accuracy": 0.7066868841648102,
+      "num_tokens": 222265041.0,
+      "step": 91200
+    },
+    {
+      "epoch": 0.836251948290089,
+      "learning_rate": 3.27514440267718e-05,
+      "loss": 1.0265,
+      "mean_token_accuracy": 0.6991253197193146,
+      "num_tokens": 222296667.0,
+      "step": 91210
+    },
+    {
+      "epoch": 0.8363436325295681,
+      "learning_rate": 3.2733107178875956e-05,
+      "loss": 0.9812,
+      "mean_token_accuracy": 0.7063963234424591,
+      "num_tokens": 222327158.0,
+      "step": 91220
+    },
+    {
+      "epoch": 0.8364353167690474,
+      "learning_rate": 3.27147703309801e-05,
+      "loss": 1.0213,
+      "mean_token_accuracy": 0.6984895050525666,
+      "num_tokens": 222357989.0,
+      "step": 91230
+    },
+    {
+      "epoch": 0.8365270010085266,
+      "learning_rate": 3.269643348308426e-05,
+      "loss": 1.0321,
+      "mean_token_accuracy": 0.7004836976528168,
+      "num_tokens": 222388577.0,
+      "step": 91240
+    },
+    {
+      "epoch": 0.8366186852480059,
+      "learning_rate": 3.267809663518841e-05,
+      "loss": 0.9905,
+      "mean_token_accuracy": 0.7060478389263153,
+      "num_tokens": 222418326.0,
+      "step": 91250
+    },
+    {
+      "epoch": 0.8367103694874851,
+      "learning_rate": 3.2659759787292564e-05,
+      "loss": 0.9547,
+      "mean_token_accuracy": 0.7119473099708558,
+      "num_tokens": 222446805.0,
+      "step": 91260
+    },
+    {
+      "epoch": 0.8368020537269644,
+      "learning_rate": 3.264142293939672e-05,
+      "loss": 0.986,
+      "mean_token_accuracy": 0.7030215203762055,
+      "num_tokens": 222477988.0,
+      "step": 91270
+    },
+    {
+      "epoch": 0.8368937379664436,
+      "learning_rate": 3.262308609150087e-05,
+      "loss": 1.0069,
+      "mean_token_accuracy": 0.698031347990036,
+      "num_tokens": 222509889.0,
+      "step": 91280
+    },
+    {
+      "epoch": 0.8369854222059228,
+      "learning_rate": 3.2604749243605026e-05,
+      "loss": 0.9937,
+      "mean_token_accuracy": 0.699799257516861,
+      "num_tokens": 222541293.0,
+      "step": 91290
+    },
+    {
+      "epoch": 0.837077106445402,
+      "learning_rate": 3.2586412395709185e-05,
+      "loss": 1.0211,
+      "mean_token_accuracy": 0.6946413099765778,
+      "num_tokens": 222572968.0,
+      "step": 91300
+    },
+    {
+      "epoch": 0.8371687906848813,
+      "learning_rate": 3.256807554781333e-05,
+      "loss": 0.9897,
+      "mean_token_accuracy": 0.7046654760837555,
+      "num_tokens": 222604491.0,
+      "step": 91310
+    },
+    {
+      "epoch": 0.8372604749243605,
+      "learning_rate": 3.254973869991749e-05,
+      "loss": 0.9729,
+      "mean_token_accuracy": 0.7089137971401215,
+      "num_tokens": 222632305.0,
+      "step": 91320
+    },
+    {
+      "epoch": 0.8373521591638398,
+      "learning_rate": 3.2531401852021634e-05,
+      "loss": 1.0027,
+      "mean_token_accuracy": 0.7008799970149994,
+      "num_tokens": 222663773.0,
+      "step": 91330
+    },
+    {
+      "epoch": 0.837443843403319,
+      "learning_rate": 3.251306500412579e-05,
+      "loss": 0.9815,
+      "mean_token_accuracy": 0.7048599421977997,
+      "num_tokens": 222694460.0,
+      "step": 91340
+    },
+    {
+      "epoch": 0.8375355276427981,
+      "learning_rate": 3.249472815622995e-05,
+      "loss": 1.0105,
+      "mean_token_accuracy": 0.7016370117664337,
+      "num_tokens": 222725008.0,
+      "step": 91350
+    },
+    {
+      "epoch": 0.8376272118822774,
+      "learning_rate": 3.2476391308334096e-05,
+      "loss": 0.9618,
+      "mean_token_accuracy": 0.7082127451896667,
+      "num_tokens": 222755161.0,
+      "step": 91360
+    },
+    {
+      "epoch": 0.8377188961217567,
+      "learning_rate": 3.2458054460438255e-05,
+      "loss": 1.0034,
+      "mean_token_accuracy": 0.6991991460323334,
+      "num_tokens": 222786384.0,
+      "step": 91370
+    },
+    {
+      "epoch": 0.8378105803612359,
+      "learning_rate": 3.24397176125424e-05,
+      "loss": 0.9636,
+      "mean_token_accuracy": 0.7071750700473786,
+      "num_tokens": 222815206.0,
+      "step": 91380
+    },
+    {
+      "epoch": 0.8379022646007152,
+      "learning_rate": 3.242138076464656e-05,
+      "loss": 0.9969,
+      "mean_token_accuracy": 0.7010862648487091,
+      "num_tokens": 222846301.0,
+      "step": 91390
+    },
+    {
+      "epoch": 0.8379939488401944,
+      "learning_rate": 3.240304391675072e-05,
+      "loss": 1.0288,
+      "mean_token_accuracy": 0.7008266806602478,
+      "num_tokens": 222877510.0,
+      "step": 91400
+    },
+    {
+      "epoch": 0.8380856330796737,
+      "learning_rate": 3.238470706885486e-05,
+      "loss": 1.0092,
+      "mean_token_accuracy": 0.7055033504962921,
+      "num_tokens": 222909382.0,
+      "step": 91410
+    },
+    {
+      "epoch": 0.8381773173191528,
+      "learning_rate": 3.236637022095902e-05,
+      "loss": 0.9609,
+      "mean_token_accuracy": 0.7142740547657013,
+      "num_tokens": 222939790.0,
+      "step": 91420
+    },
+    {
+      "epoch": 0.838269001558632,
+      "learning_rate": 3.234803337306317e-05,
+      "loss": 0.9964,
+      "mean_token_accuracy": 0.7005717277526855,
+      "num_tokens": 222971254.0,
+      "step": 91430
+    },
+    {
+      "epoch": 0.8383606857981113,
+      "learning_rate": 3.2329696525167324e-05,
+      "loss": 1.0312,
+      "mean_token_accuracy": 0.6940638840198516,
+      "num_tokens": 223000601.0,
+      "step": 91440
+    },
+    {
+      "epoch": 0.8384523700375905,
+      "learning_rate": 3.231135967727148e-05,
+      "loss": 1.0222,
+      "mean_token_accuracy": 0.6965833485126496,
+      "num_tokens": 223031669.0,
+      "step": 91450
+    },
+    {
+      "epoch": 0.8385440542770698,
+      "learning_rate": 3.229302282937563e-05,
+      "loss": 1.0148,
+      "mean_token_accuracy": 0.6938436567783356,
+      "num_tokens": 223062222.0,
+      "step": 91460
+    },
+    {
+      "epoch": 0.838635738516549,
+      "learning_rate": 3.227468598147979e-05,
+      "loss": 1.0141,
+      "mean_token_accuracy": 0.6961119532585144,
+      "num_tokens": 223093582.0,
+      "step": 91470
+    },
+    {
+      "epoch": 0.8387274227560282,
+      "learning_rate": 3.225634913358394e-05,
+      "loss": 1.0159,
+      "mean_token_accuracy": 0.7005566835403443,
+      "num_tokens": 223124729.0,
+      "step": 91480
+    },
+    {
+      "epoch": 0.8388191069955074,
+      "learning_rate": 3.223801228568809e-05,
+      "loss": 0.9972,
+      "mean_token_accuracy": 0.7011977195739746,
+      "num_tokens": 223155276.0,
+      "step": 91490
+    },
+    {
+      "epoch": 0.8389107912349867,
+      "learning_rate": 3.221967543779225e-05,
+      "loss": 0.9902,
+      "mean_token_accuracy": 0.7020883321762085,
+      "num_tokens": 223183719.0,
+      "step": 91500
+    },
+    {
+      "epoch": 0.8390024754744659,
+      "learning_rate": 3.2201338589896394e-05,
+      "loss": 0.9973,
+      "mean_token_accuracy": 0.7041495442390442,
+      "num_tokens": 223214502.0,
+      "step": 91510
+    },
+    {
+      "epoch": 0.8390941597139452,
+      "learning_rate": 3.218300174200055e-05,
+      "loss": 1.0081,
+      "mean_token_accuracy": 0.6985054612159729,
+      "num_tokens": 223246196.0,
+      "step": 91520
+    },
+    {
+      "epoch": 0.8391858439534244,
+      "learning_rate": 3.2164664894104705e-05,
+      "loss": 0.9957,
+      "mean_token_accuracy": 0.7056102693080902,
+      "num_tokens": 223277272.0,
+      "step": 91530
+    },
+    {
+      "epoch": 0.8392775281929037,
+      "learning_rate": 3.214632804620886e-05,
+      "loss": 1.0193,
+      "mean_token_accuracy": 0.6959621608257294,
+      "num_tokens": 223307640.0,
+      "step": 91540
+    },
+    {
+      "epoch": 0.8393692124323828,
+      "learning_rate": 3.2127991198313015e-05,
+      "loss": 1.0015,
+      "mean_token_accuracy": 0.7042932510375977,
+      "num_tokens": 223337961.0,
+      "step": 91550
+    },
+    {
+      "epoch": 0.8394608966718621,
+      "learning_rate": 3.210965435041716e-05,
+      "loss": 0.9959,
+      "mean_token_accuracy": 0.7025859355926514,
+      "num_tokens": 223367569.0,
+      "step": 91560
+    },
+    {
+      "epoch": 0.8395525809113413,
+      "learning_rate": 3.209131750252132e-05,
+      "loss": 1.021,
+      "mean_token_accuracy": 0.6990630745887756,
+      "num_tokens": 223398545.0,
+      "step": 91570
+    },
+    {
+      "epoch": 0.8396442651508206,
+      "learning_rate": 3.207298065462547e-05,
+      "loss": 1.0004,
+      "mean_token_accuracy": 0.7005080997943878,
+      "num_tokens": 223428995.0,
+      "step": 91580
+    },
+    {
+      "epoch": 0.8397359493902998,
+      "learning_rate": 3.205464380672962e-05,
+      "loss": 1.0029,
+      "mean_token_accuracy": 0.7037331581115722,
+      "num_tokens": 223459170.0,
+      "step": 91590
+    },
+    {
+      "epoch": 0.8398276336297791,
+      "learning_rate": 3.203630695883378e-05,
+      "loss": 1.0277,
+      "mean_token_accuracy": 0.7013119220733642,
+      "num_tokens": 223489915.0,
+      "step": 91600
+    },
+    {
+      "epoch": 0.8399193178692582,
+      "learning_rate": 3.201797011093793e-05,
+      "loss": 0.9914,
+      "mean_token_accuracy": 0.7053504765033722,
+      "num_tokens": 223520451.0,
+      "step": 91610
+    },
+    {
+      "epoch": 0.8400110021087375,
+      "learning_rate": 3.1999633263042085e-05,
+      "loss": 0.9655,
+      "mean_token_accuracy": 0.7083058834075928,
+      "num_tokens": 223548955.0,
+      "step": 91620
+    },
+    {
+      "epoch": 0.8401026863482167,
+      "learning_rate": 3.198129641514624e-05,
+      "loss": 1.0169,
+      "mean_token_accuracy": 0.7000997841358185,
+      "num_tokens": 223580163.0,
+      "step": 91630
+    },
+    {
+      "epoch": 0.840194370587696,
+      "learning_rate": 3.196295956725039e-05,
+      "loss": 0.9834,
+      "mean_token_accuracy": 0.6998641967773438,
+      "num_tokens": 223611275.0,
+      "step": 91640
+    },
+    {
+      "epoch": 0.8402860548271752,
+      "learning_rate": 3.194462271935455e-05,
+      "loss": 0.9652,
+      "mean_token_accuracy": 0.7117608606815338,
+      "num_tokens": 223641395.0,
+      "step": 91650
+    },
+    {
+      "epoch": 0.8403777390666545,
+      "learning_rate": 3.19262858714587e-05,
+      "loss": 1.0022,
+      "mean_token_accuracy": 0.7010492026805878,
+      "num_tokens": 223671826.0,
+      "step": 91660
+    },
+    {
+      "epoch": 0.8404694233061337,
+      "learning_rate": 3.190794902356285e-05,
+      "loss": 0.9946,
+      "mean_token_accuracy": 0.7014322578907013,
+      "num_tokens": 223702931.0,
+      "step": 91670
+    },
+    {
+      "epoch": 0.8405611075456129,
+      "learning_rate": 3.1889612175667e-05,
+      "loss": 0.9705,
+      "mean_token_accuracy": 0.7093854248523712,
+      "num_tokens": 223731713.0,
+      "step": 91680
+    },
+    {
+      "epoch": 0.8406527917850921,
+      "learning_rate": 3.1871275327771155e-05,
+      "loss": 1.0141,
+      "mean_token_accuracy": 0.7003601014614105,
+      "num_tokens": 223763117.0,
+      "step": 91690
+    },
+    {
+      "epoch": 0.8407444760245714,
+      "learning_rate": 3.1852938479875314e-05,
+      "loss": 1.003,
+      "mean_token_accuracy": 0.7010448932647705,
+      "num_tokens": 223793505.0,
+      "step": 91700
+    },
+    {
+      "epoch": 0.8408361602640506,
+      "learning_rate": 3.1834601631979465e-05,
+      "loss": 1.0087,
+      "mean_token_accuracy": 0.7001794517040253,
+      "num_tokens": 223824394.0,
+      "step": 91710
+    },
+    {
+      "epoch": 0.8409278445035299,
+      "learning_rate": 3.181626478408362e-05,
+      "loss": 0.9893,
+      "mean_token_accuracy": 0.7095170080661773,
+      "num_tokens": 223856529.0,
+      "step": 91720
+    },
+    {
+      "epoch": 0.8410195287430091,
+      "learning_rate": 3.179792793618777e-05,
+      "loss": 1.0034,
+      "mean_token_accuracy": 0.7024901330471038,
+      "num_tokens": 223886958.0,
+      "step": 91730
+    },
+    {
+      "epoch": 0.8411112129824884,
+      "learning_rate": 3.177959108829193e-05,
+      "loss": 0.9706,
+      "mean_token_accuracy": 0.7048344671726227,
+      "num_tokens": 223914994.0,
+      "step": 91740
+    },
+    {
+      "epoch": 0.8412028972219675,
+      "learning_rate": 3.176125424039608e-05,
+      "loss": 1.0093,
+      "mean_token_accuracy": 0.7007959365844727,
+      "num_tokens": 223946012.0,
+      "step": 91750
+    },
+    {
+      "epoch": 0.8412945814614468,
+      "learning_rate": 3.174291739250023e-05,
+      "loss": 0.9975,
+      "mean_token_accuracy": 0.7080795049667359,
+      "num_tokens": 223976892.0,
+      "step": 91760
+    },
+    {
+      "epoch": 0.841386265700926,
+      "learning_rate": 3.1724580544604383e-05,
+      "loss": 1.0113,
+      "mean_token_accuracy": 0.7008758664131165,
+      "num_tokens": 224008730.0,
+      "step": 91770
+    },
+    {
+      "epoch": 0.8414779499404053,
+      "learning_rate": 3.1706243696708535e-05,
+      "loss": 1.0137,
+      "mean_token_accuracy": 0.6989102780818939,
+      "num_tokens": 224040108.0,
+      "step": 91780
+    },
+    {
+      "epoch": 0.8415696341798845,
+      "learning_rate": 3.1687906848812694e-05,
+      "loss": 1.0018,
+      "mean_token_accuracy": 0.7042802751064301,
+      "num_tokens": 224070662.0,
+      "step": 91790
+    },
+    {
+      "epoch": 0.8416613184193638,
+      "learning_rate": 3.1669570000916846e-05,
+      "loss": 0.9769,
+      "mean_token_accuracy": 0.7085938453674316,
+      "num_tokens": 224098487.0,
+      "step": 91800
+    },
+    {
+      "epoch": 0.8417530026588429,
+      "learning_rate": 3.1651233153021e-05,
+      "loss": 0.9952,
+      "mean_token_accuracy": 0.7060381650924683,
+      "num_tokens": 224128721.0,
+      "step": 91810
+    },
+    {
+      "epoch": 0.8418446868983221,
+      "learning_rate": 3.163289630512515e-05,
+      "loss": 0.9723,
+      "mean_token_accuracy": 0.7097331464290619,
+      "num_tokens": 224158648.0,
+      "step": 91820
+    },
+    {
+      "epoch": 0.8419363711378014,
+      "learning_rate": 3.16145594572293e-05,
+      "loss": 1.0277,
+      "mean_token_accuracy": 0.699338573217392,
+      "num_tokens": 224189975.0,
+      "step": 91830
+    },
+    {
+      "epoch": 0.8420280553772806,
+      "learning_rate": 3.159622260933346e-05,
+      "loss": 0.9778,
+      "mean_token_accuracy": 0.7073618948459626,
+      "num_tokens": 224220671.0,
+      "step": 91840
+    },
+    {
+      "epoch": 0.8421197396167599,
+      "learning_rate": 3.157788576143761e-05,
+      "loss": 0.9942,
+      "mean_token_accuracy": 0.7037886440753937,
+      "num_tokens": 224251242.0,
+      "step": 91850
+    },
+    {
+      "epoch": 0.8422114238562391,
+      "learning_rate": 3.1559548913541764e-05,
+      "loss": 0.9671,
+      "mean_token_accuracy": 0.7061097323894501,
+      "num_tokens": 224279067.0,
+      "step": 91860
+    },
+    {
+      "epoch": 0.8423031080957184,
+      "learning_rate": 3.1541212065645916e-05,
+      "loss": 1.021,
+      "mean_token_accuracy": 0.6982515513896942,
+      "num_tokens": 224310698.0,
+      "step": 91870
+    },
+    {
+      "epoch": 0.8423947923351975,
+      "learning_rate": 3.152287521775007e-05,
+      "loss": 1.0091,
+      "mean_token_accuracy": 0.7030785799026489,
+      "num_tokens": 224341772.0,
+      "step": 91880
+    },
+    {
+      "epoch": 0.8424864765746768,
+      "learning_rate": 3.1504538369854226e-05,
+      "loss": 1.0125,
+      "mean_token_accuracy": 0.6984831392765045,
+      "num_tokens": 224371944.0,
+      "step": 91890
+    },
+    {
+      "epoch": 0.842578160814156,
+      "learning_rate": 3.148620152195838e-05,
+      "loss": 0.9942,
+      "mean_token_accuracy": 0.7027344167232513,
+      "num_tokens": 224403172.0,
+      "step": 91900
+    },
+    {
+      "epoch": 0.8426698450536353,
+      "learning_rate": 3.146786467406253e-05,
+      "loss": 0.9834,
+      "mean_token_accuracy": 0.7057047128677368,
+      "num_tokens": 224432314.0,
+      "step": 91910
+    },
+    {
+      "epoch": 0.8427615292931145,
+      "learning_rate": 3.144952782616669e-05,
+      "loss": 0.987,
+      "mean_token_accuracy": 0.7073067247867584,
+      "num_tokens": 224461293.0,
+      "step": 91920
+    },
+    {
+      "epoch": 0.8428532135325938,
+      "learning_rate": 3.1431190978270834e-05,
+      "loss": 1.0092,
+      "mean_token_accuracy": 0.6986635565757752,
+      "num_tokens": 224491704.0,
+      "step": 91930
+    },
+    {
+      "epoch": 0.8429448977720729,
+      "learning_rate": 3.141285413037499e-05,
+      "loss": 0.9689,
+      "mean_token_accuracy": 0.7099559724330902,
+      "num_tokens": 224522159.0,
+      "step": 91940
+    },
+    {
+      "epoch": 0.8430365820115522,
+      "learning_rate": 3.1394517282479144e-05,
+      "loss": 1.0004,
+      "mean_token_accuracy": 0.7031835913658142,
+      "num_tokens": 224553742.0,
+      "step": 91950
+    },
+    {
+      "epoch": 0.8431282662510314,
+      "learning_rate": 3.1376180434583296e-05,
+      "loss": 0.9605,
+      "mean_token_accuracy": 0.7088021457195282,
+      "num_tokens": 224584837.0,
+      "step": 91960
+    },
+    {
+      "epoch": 0.8432199504905107,
+      "learning_rate": 3.1357843586687455e-05,
+      "loss": 0.9923,
+      "mean_token_accuracy": 0.7099215149879455,
+      "num_tokens": 224614366.0,
+      "step": 91970
+    },
+    {
+      "epoch": 0.8433116347299899,
+      "learning_rate": 3.13395067387916e-05,
+      "loss": 1.0107,
+      "mean_token_accuracy": 0.6980745732784271,
+      "num_tokens": 224643211.0,
+      "step": 91980
+    },
+    {
+      "epoch": 0.8434033189694692,
+      "learning_rate": 3.132116989089576e-05,
+      "loss": 1.0051,
+      "mean_token_accuracy": 0.7066904246807099,
+      "num_tokens": 224674876.0,
+      "step": 91990
+    },
+    {
+      "epoch": 0.8434950032089484,
+      "learning_rate": 3.130283304299991e-05,
+      "loss": 0.992,
+      "mean_token_accuracy": 0.7022707223892212,
+      "num_tokens": 224706286.0,
+      "step": 92000
+    },
+    {
+      "epoch": 0.8435866874484276,
+      "learning_rate": 3.128449619510406e-05,
+      "loss": 1.0358,
+      "mean_token_accuracy": 0.6943967401981354,
+      "num_tokens": 224735673.0,
+      "step": 92010
+    },
+    {
+      "epoch": 0.8436783716879068,
+      "learning_rate": 3.126615934720822e-05,
+      "loss": 0.984,
+      "mean_token_accuracy": 0.7018739461898804,
+      "num_tokens": 224766405.0,
+      "step": 92020
+    },
+    {
+      "epoch": 0.8437700559273861,
+      "learning_rate": 3.1247822499312366e-05,
+      "loss": 0.9948,
+      "mean_token_accuracy": 0.700128835439682,
+      "num_tokens": 224796389.0,
+      "step": 92030
+    },
+    {
+      "epoch": 0.8438617401668653,
+      "learning_rate": 3.1229485651416525e-05,
+      "loss": 1.0073,
+      "mean_token_accuracy": 0.7004628121852875,
+      "num_tokens": 224823526.0,
+      "step": 92040
+    },
+    {
+      "epoch": 0.8439534244063446,
+      "learning_rate": 3.1211148803520676e-05,
+      "loss": 1.0031,
+      "mean_token_accuracy": 0.6994020581245423,
+      "num_tokens": 224854352.0,
+      "step": 92050
+    },
+    {
+      "epoch": 0.8440451086458238,
+      "learning_rate": 3.119281195562483e-05,
+      "loss": 1.0303,
+      "mean_token_accuracy": 0.6925749242305755,
+      "num_tokens": 224884760.0,
+      "step": 92060
+    },
+    {
+      "epoch": 0.844136792885303,
+      "learning_rate": 3.117447510772899e-05,
+      "loss": 0.9964,
+      "mean_token_accuracy": 0.7054547429084778,
+      "num_tokens": 224915431.0,
+      "step": 92070
+    },
+    {
+      "epoch": 0.8442284771247822,
+      "learning_rate": 3.115613825983313e-05,
+      "loss": 0.9917,
+      "mean_token_accuracy": 0.7053548216819763,
+      "num_tokens": 224947166.0,
+      "step": 92080
+    },
+    {
+      "epoch": 0.8443201613642615,
+      "learning_rate": 3.113780141193729e-05,
+      "loss": 0.9829,
+      "mean_token_accuracy": 0.7075178861618042,
+      "num_tokens": 224977578.0,
+      "step": 92090
+    },
+    {
+      "epoch": 0.8444118456037407,
+      "learning_rate": 3.111946456404144e-05,
+      "loss": 0.9719,
+      "mean_token_accuracy": 0.7024227261543274,
+      "num_tokens": 225005512.0,
+      "step": 92100
+    },
+    {
+      "epoch": 0.84450352984322,
+      "learning_rate": 3.1101127716145594e-05,
+      "loss": 0.9747,
+      "mean_token_accuracy": 0.7060999393463134,
+      "num_tokens": 225036673.0,
+      "step": 92110
+    },
+    {
+      "epoch": 0.8445952140826992,
+      "learning_rate": 3.108279086824975e-05,
+      "loss": 1.0147,
+      "mean_token_accuracy": 0.6976794540882111,
+      "num_tokens": 225068821.0,
+      "step": 92120
+    },
+    {
+      "epoch": 0.8446868983221785,
+      "learning_rate": 3.10644540203539e-05,
+      "loss": 1.0105,
+      "mean_token_accuracy": 0.7012093305587769,
+      "num_tokens": 225099406.0,
+      "step": 92130
+    },
+    {
+      "epoch": 0.8447785825616576,
+      "learning_rate": 3.104611717245806e-05,
+      "loss": 1.0045,
+      "mean_token_accuracy": 0.7001675188541412,
+      "num_tokens": 225129722.0,
+      "step": 92140
+    },
+    {
+      "epoch": 0.8448702668011369,
+      "learning_rate": 3.102778032456221e-05,
+      "loss": 0.9859,
+      "mean_token_accuracy": 0.7040951013565063,
+      "num_tokens": 225160334.0,
+      "step": 92150
+    },
+    {
+      "epoch": 0.8449619510406161,
+      "learning_rate": 3.100944347666636e-05,
+      "loss": 0.9732,
+      "mean_token_accuracy": 0.7056873083114624,
+      "num_tokens": 225188221.0,
+      "step": 92160
+    },
+    {
+      "epoch": 0.8450536352800954,
+      "learning_rate": 3.099110662877052e-05,
+      "loss": 0.9988,
+      "mean_token_accuracy": 0.7046827554702759,
+      "num_tokens": 225218756.0,
+      "step": 92170
+    },
+    {
+      "epoch": 0.8451453195195746,
+      "learning_rate": 3.0972769780874664e-05,
+      "loss": 0.9929,
+      "mean_token_accuracy": 0.7025894045829773,
+      "num_tokens": 225248959.0,
+      "step": 92180
+    },
+    {
+      "epoch": 0.8452370037590539,
+      "learning_rate": 3.095443293297882e-05,
+      "loss": 1.0225,
+      "mean_token_accuracy": 0.6961394786834717,
+      "num_tokens": 225279958.0,
+      "step": 92190
+    },
+    {
+      "epoch": 0.845328687998533,
+      "learning_rate": 3.0936096085082975e-05,
+      "loss": 0.9839,
+      "mean_token_accuracy": 0.70698561668396,
+      "num_tokens": 225310359.0,
+      "step": 92200
+    },
+    {
+      "epoch": 0.8454203722380123,
+      "learning_rate": 3.0917759237187127e-05,
+      "loss": 1.0208,
+      "mean_token_accuracy": 0.696915727853775,
+      "num_tokens": 225341492.0,
+      "step": 92210
+    },
+    {
+      "epoch": 0.8455120564774915,
+      "learning_rate": 3.0899422389291285e-05,
+      "loss": 0.9691,
+      "mean_token_accuracy": 0.7075960218906403,
+      "num_tokens": 225369895.0,
+      "step": 92220
+    },
+    {
+      "epoch": 0.8456037407169708,
+      "learning_rate": 3.088108554139544e-05,
+      "loss": 1.0033,
+      "mean_token_accuracy": 0.7020844161510468,
+      "num_tokens": 225400552.0,
+      "step": 92230
+    },
+    {
+      "epoch": 0.84569542495645,
+      "learning_rate": 3.086274869349959e-05,
+      "loss": 0.9857,
+      "mean_token_accuracy": 0.7039596199989319,
+      "num_tokens": 225431494.0,
+      "step": 92240
+    },
+    {
+      "epoch": 0.8457871091959293,
+      "learning_rate": 3.084441184560374e-05,
+      "loss": 0.9952,
+      "mean_token_accuracy": 0.7030720889568329,
+      "num_tokens": 225462254.0,
+      "step": 92250
+    },
+    {
+      "epoch": 0.8458787934354085,
+      "learning_rate": 3.082607499770789e-05,
+      "loss": 1.0347,
+      "mean_token_accuracy": 0.6975192964076996,
+      "num_tokens": 225492772.0,
+      "step": 92260
+    },
+    {
+      "epoch": 0.8459704776748876,
+      "learning_rate": 3.080773814981205e-05,
+      "loss": 0.9845,
+      "mean_token_accuracy": 0.7059388339519501,
+      "num_tokens": 225522832.0,
+      "step": 92270
+    },
+    {
+      "epoch": 0.8460621619143669,
+      "learning_rate": 3.07894013019162e-05,
+      "loss": 0.9733,
+      "mean_token_accuracy": 0.7059954524040222,
+      "num_tokens": 225552321.0,
+      "step": 92280
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "learning_rate": 3.0771064454020355e-05,
+      "loss": 1.0148,
+      "mean_token_accuracy": 0.6935909867286683,
+      "num_tokens": 225582357.0,
+      "step": 92290
+    },
+    {
+      "epoch": 0.8462455303933254,
+      "learning_rate": 3.075272760612451e-05,
+      "loss": 0.9941,
+      "mean_token_accuracy": 0.7026721596717834,
+      "num_tokens": 225613377.0,
+      "step": 92300
+    },
+    {
+      "epoch": 0.8463372146328046,
+      "learning_rate": 3.073439075822866e-05,
+      "loss": 1.0023,
+      "mean_token_accuracy": 0.703583025932312,
+      "num_tokens": 225643992.0,
+      "step": 92310
+    },
+    {
+      "epoch": 0.8464288988722839,
+      "learning_rate": 3.071605391033282e-05,
+      "loss": 0.9809,
+      "mean_token_accuracy": 0.70487340092659,
+      "num_tokens": 225674099.0,
+      "step": 92320
+    },
+    {
+      "epoch": 0.846520583111763,
+      "learning_rate": 3.069771706243697e-05,
+      "loss": 0.9966,
+      "mean_token_accuracy": 0.6940248966217041,
+      "num_tokens": 225704104.0,
+      "step": 92330
+    },
+    {
+      "epoch": 0.8466122673512423,
+      "learning_rate": 3.067938021454112e-05,
+      "loss": 0.9931,
+      "mean_token_accuracy": 0.7026736319065094,
+      "num_tokens": 225733018.0,
+      "step": 92340
+    },
+    {
+      "epoch": 0.8467039515907215,
+      "learning_rate": 3.066104336664527e-05,
+      "loss": 0.9937,
+      "mean_token_accuracy": 0.7105529487133027,
+      "num_tokens": 225763917.0,
+      "step": 92350
+    },
+    {
+      "epoch": 0.8467956358302008,
+      "learning_rate": 3.064270651874943e-05,
+      "loss": 0.999,
+      "mean_token_accuracy": 0.7028055250644684,
+      "num_tokens": 225795207.0,
+      "step": 92360
+    },
+    {
+      "epoch": 0.84688732006968,
+      "learning_rate": 3.0624369670853584e-05,
+      "loss": 1.003,
+      "mean_token_accuracy": 0.7021788954734802,
+      "num_tokens": 225825180.0,
+      "step": 92370
+    },
+    {
+      "epoch": 0.8469790043091593,
+      "learning_rate": 3.0606032822957735e-05,
+      "loss": 0.9792,
+      "mean_token_accuracy": 0.7024985313415527,
+      "num_tokens": 225855340.0,
+      "step": 92380
+    },
+    {
+      "epoch": 0.8470706885486385,
+      "learning_rate": 3.058769597506189e-05,
+      "loss": 1.0036,
+      "mean_token_accuracy": 0.7014451801776886,
+      "num_tokens": 225885618.0,
+      "step": 92390
+    },
+    {
+      "epoch": 0.8471623727881177,
+      "learning_rate": 3.056935912716604e-05,
+      "loss": 0.9674,
+      "mean_token_accuracy": 0.7104009866714478,
+      "num_tokens": 225914346.0,
+      "step": 92400
+    },
+    {
+      "epoch": 0.8472540570275969,
+      "learning_rate": 3.05510222792702e-05,
+      "loss": 0.9999,
+      "mean_token_accuracy": 0.7026920735836029,
+      "num_tokens": 225945234.0,
+      "step": 92410
+    },
+    {
+      "epoch": 0.8473457412670762,
+      "learning_rate": 3.053268543137435e-05,
+      "loss": 0.996,
+      "mean_token_accuracy": 0.7054856359958649,
+      "num_tokens": 225976466.0,
+      "step": 92420
+    },
+    {
+      "epoch": 0.8474374255065554,
+      "learning_rate": 3.05143485834785e-05,
+      "loss": 0.9868,
+      "mean_token_accuracy": 0.7047296702861786,
+      "num_tokens": 226007524.0,
+      "step": 92430
+    },
+    {
+      "epoch": 0.8475291097460347,
+      "learning_rate": 3.0496011735582653e-05,
+      "loss": 1.0165,
+      "mean_token_accuracy": 0.696868097782135,
+      "num_tokens": 226037824.0,
+      "step": 92440
+    },
+    {
+      "epoch": 0.8476207939855139,
+      "learning_rate": 3.047767488768681e-05,
+      "loss": 0.9866,
+      "mean_token_accuracy": 0.7058529853820801,
+      "num_tokens": 226067719.0,
+      "step": 92450
+    },
+    {
+      "epoch": 0.8477124782249931,
+      "learning_rate": 3.0459338039790964e-05,
+      "loss": 0.9797,
+      "mean_token_accuracy": 0.7003554463386535,
+      "num_tokens": 226096110.0,
+      "step": 92460
+    },
+    {
+      "epoch": 0.8478041624644723,
+      "learning_rate": 3.0441001191895112e-05,
+      "loss": 0.973,
+      "mean_token_accuracy": 0.7082064628601075,
+      "num_tokens": 226127058.0,
+      "step": 92470
+    },
+    {
+      "epoch": 0.8478958467039516,
+      "learning_rate": 3.0422664343999268e-05,
+      "loss": 1.0354,
+      "mean_token_accuracy": 0.6950591146945954,
+      "num_tokens": 226158699.0,
+      "step": 92480
+    },
+    {
+      "epoch": 0.8479875309434308,
+      "learning_rate": 3.040432749610342e-05,
+      "loss": 1.0177,
+      "mean_token_accuracy": 0.700346302986145,
+      "num_tokens": 226190250.0,
+      "step": 92490
+    },
+    {
+      "epoch": 0.8480792151829101,
+      "learning_rate": 3.0385990648207575e-05,
+      "loss": 0.9955,
+      "mean_token_accuracy": 0.7033508360385895,
+      "num_tokens": 226219946.0,
+      "step": 92500
+    },
+    {
+      "epoch": 0.8481708994223893,
+      "learning_rate": 3.036765380031173e-05,
+      "loss": 0.9986,
+      "mean_token_accuracy": 0.6954771339893341,
+      "num_tokens": 226250130.0,
+      "step": 92510
+    },
+    {
+      "epoch": 0.8482625836618686,
+      "learning_rate": 3.034931695241588e-05,
+      "loss": 0.9591,
+      "mean_token_accuracy": 0.7096432268619537,
+      "num_tokens": 226277856.0,
+      "step": 92520
+    },
+    {
+      "epoch": 0.8483542679013477,
+      "learning_rate": 3.0330980104520034e-05,
+      "loss": 0.9792,
+      "mean_token_accuracy": 0.7000100553035736,
+      "num_tokens": 226308809.0,
+      "step": 92530
+    },
+    {
+      "epoch": 0.848445952140827,
+      "learning_rate": 3.031264325662419e-05,
+      "loss": 0.9898,
+      "mean_token_accuracy": 0.7065817296504975,
+      "num_tokens": 226338960.0,
+      "step": 92540
+    },
+    {
+      "epoch": 0.8485376363803062,
+      "learning_rate": 3.029430640872834e-05,
+      "loss": 1.0085,
+      "mean_token_accuracy": 0.7020676970481873,
+      "num_tokens": 226369586.0,
+      "step": 92550
+    },
+    {
+      "epoch": 0.8486293206197855,
+      "learning_rate": 3.0275969560832496e-05,
+      "loss": 1.0288,
+      "mean_token_accuracy": 0.7000457465648651,
+      "num_tokens": 226400590.0,
+      "step": 92560
+    },
+    {
+      "epoch": 0.8487210048592647,
+      "learning_rate": 3.0257632712936645e-05,
+      "loss": 0.9855,
+      "mean_token_accuracy": 0.7003734707832336,
+      "num_tokens": 226429608.0,
+      "step": 92570
+    },
+    {
+      "epoch": 0.848812689098744,
+      "learning_rate": 3.02392958650408e-05,
+      "loss": 1.0009,
+      "mean_token_accuracy": 0.7032172381877899,
+      "num_tokens": 226458672.0,
+      "step": 92580
+    },
+    {
+      "epoch": 0.8489043733382231,
+      "learning_rate": 3.0220959017144955e-05,
+      "loss": 0.9864,
+      "mean_token_accuracy": 0.7043384611606598,
+      "num_tokens": 226490294.0,
+      "step": 92590
+    },
+    {
+      "epoch": 0.8489960575777024,
+      "learning_rate": 3.0202622169249107e-05,
+      "loss": 0.9927,
+      "mean_token_accuracy": 0.7044949352741241,
+      "num_tokens": 226521305.0,
+      "step": 92600
+    },
+    {
+      "epoch": 0.8490877418171816,
+      "learning_rate": 3.0184285321353262e-05,
+      "loss": 0.9948,
+      "mean_token_accuracy": 0.7017820060253144,
+      "num_tokens": 226551759.0,
+      "step": 92610
+    },
+    {
+      "epoch": 0.8491794260566609,
+      "learning_rate": 3.016594847345741e-05,
+      "loss": 0.9957,
+      "mean_token_accuracy": 0.70472252368927,
+      "num_tokens": 226582424.0,
+      "step": 92620
+    },
+    {
+      "epoch": 0.8492711102961401,
+      "learning_rate": 3.0147611625561566e-05,
+      "loss": 0.9818,
+      "mean_token_accuracy": 0.7079247176647187,
+      "num_tokens": 226611851.0,
+      "step": 92630
+    },
+    {
+      "epoch": 0.8493627945356194,
+      "learning_rate": 3.012927477766572e-05,
+      "loss": 1.0019,
+      "mean_token_accuracy": 0.6940906465053558,
+      "num_tokens": 226639158.0,
+      "step": 92640
+    },
+    {
+      "epoch": 0.8494544787750986,
+      "learning_rate": 3.0110937929769873e-05,
+      "loss": 0.9949,
+      "mean_token_accuracy": 0.7073172271251679,
+      "num_tokens": 226669578.0,
+      "step": 92650
+    },
+    {
+      "epoch": 0.8495461630145777,
+      "learning_rate": 3.009260108187403e-05,
+      "loss": 0.9883,
+      "mean_token_accuracy": 0.7053599774837493,
+      "num_tokens": 226700559.0,
+      "step": 92660
+    },
+    {
+      "epoch": 0.849637847254057,
+      "learning_rate": 3.0074264233978184e-05,
+      "loss": 1.0105,
+      "mean_token_accuracy": 0.7003489255905151,
+      "num_tokens": 226731054.0,
+      "step": 92670
+    },
+    {
+      "epoch": 0.8497295314935362,
+      "learning_rate": 3.0055927386082332e-05,
+      "loss": 1.0094,
+      "mean_token_accuracy": 0.7078044176101684,
+      "num_tokens": 226762562.0,
+      "step": 92680
+    },
+    {
+      "epoch": 0.8498212157330155,
+      "learning_rate": 3.0037590538186487e-05,
+      "loss": 0.9634,
+      "mean_token_accuracy": 0.7059079289436341,
+      "num_tokens": 226793398.0,
+      "step": 92690
+    },
+    {
+      "epoch": 0.8499128999724948,
+      "learning_rate": 3.001925369029064e-05,
+      "loss": 0.9902,
+      "mean_token_accuracy": 0.697595739364624,
+      "num_tokens": 226821303.0,
+      "step": 92700
+    },
+    {
+      "epoch": 0.850004584211974,
+      "learning_rate": 3.0000916842394795e-05,
+      "loss": 1.0235,
+      "mean_token_accuracy": 0.6974358201026917,
+      "num_tokens": 226852263.0,
+      "step": 92710
+    },
+    {
+      "epoch": 0.8500962684514531,
+      "learning_rate": 2.998257999449895e-05,
+      "loss": 0.9852,
+      "mean_token_accuracy": 0.7041719734668732,
+      "num_tokens": 226884205.0,
+      "step": 92720
+    },
+    {
+      "epoch": 0.8501879526909324,
+      "learning_rate": 2.9964243146603098e-05,
+      "loss": 0.9983,
+      "mean_token_accuracy": 0.7009266495704651,
+      "num_tokens": 226914671.0,
+      "step": 92730
+    },
+    {
+      "epoch": 0.8502796369304116,
+      "learning_rate": 2.9945906298707254e-05,
+      "loss": 0.9967,
+      "mean_token_accuracy": 0.70213183760643,
+      "num_tokens": 226946000.0,
+      "step": 92740
+    },
+    {
+      "epoch": 0.8503713211698909,
+      "learning_rate": 2.9927569450811405e-05,
+      "loss": 0.9985,
+      "mean_token_accuracy": 0.7022172152996063,
+      "num_tokens": 226977389.0,
+      "step": 92750
+    },
+    {
+      "epoch": 0.8504630054093701,
+      "learning_rate": 2.990923260291556e-05,
+      "loss": 0.9854,
+      "mean_token_accuracy": 0.701825100183487,
+      "num_tokens": 227005215.0,
+      "step": 92760
+    },
+    {
+      "epoch": 0.8505546896488494,
+      "learning_rate": 2.9890895755019716e-05,
+      "loss": 0.993,
+      "mean_token_accuracy": 0.7050695061683655,
+      "num_tokens": 227035909.0,
+      "step": 92770
+    },
+    {
+      "epoch": 0.8506463738883286,
+      "learning_rate": 2.9872558907123864e-05,
+      "loss": 1.024,
+      "mean_token_accuracy": 0.6951681792736053,
+      "num_tokens": 227066372.0,
+      "step": 92780
+    },
+    {
+      "epoch": 0.8507380581278078,
+      "learning_rate": 2.985422205922802e-05,
+      "loss": 0.9706,
+      "mean_token_accuracy": 0.7126036167144776,
+      "num_tokens": 227096199.0,
+      "step": 92790
+    },
+    {
+      "epoch": 0.850829742367287,
+      "learning_rate": 2.983588521133217e-05,
+      "loss": 0.988,
+      "mean_token_accuracy": 0.7058151662349701,
+      "num_tokens": 227126512.0,
+      "step": 92800
+    },
+    {
+      "epoch": 0.8509214266067663,
+      "learning_rate": 2.9817548363436327e-05,
+      "loss": 1.0053,
+      "mean_token_accuracy": 0.7011158227920532,
+      "num_tokens": 227157255.0,
+      "step": 92810
+    },
+    {
+      "epoch": 0.8510131108462455,
+      "learning_rate": 2.9799211515540482e-05,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.7099586308002472,
+      "num_tokens": 227185865.0,
+      "step": 92820
+    },
+    {
+      "epoch": 0.8511047950857248,
+      "learning_rate": 2.978087466764463e-05,
+      "loss": 1.0201,
+      "mean_token_accuracy": 0.694517332315445,
+      "num_tokens": 227216234.0,
+      "step": 92830
+    },
+    {
+      "epoch": 0.851196479325204,
+      "learning_rate": 2.9762537819748786e-05,
+      "loss": 1.0242,
+      "mean_token_accuracy": 0.6950756072998047,
+      "num_tokens": 227248508.0,
+      "step": 92840
+    },
+    {
+      "epoch": 0.8512881635646832,
+      "learning_rate": 2.974420097185294e-05,
+      "loss": 1.0285,
+      "mean_token_accuracy": 0.6953354895114898,
+      "num_tokens": 227279847.0,
+      "step": 92850
+    },
+    {
+      "epoch": 0.8513798478041624,
+      "learning_rate": 2.9725864123957093e-05,
+      "loss": 1.0041,
+      "mean_token_accuracy": 0.7030609607696533,
+      "num_tokens": 227309341.0,
+      "step": 92860
+    },
+    {
+      "epoch": 0.8514715320436417,
+      "learning_rate": 2.9707527276061248e-05,
+      "loss": 1.0174,
+      "mean_token_accuracy": 0.6984331011772156,
+      "num_tokens": 227339388.0,
+      "step": 92870
+    },
+    {
+      "epoch": 0.8515632162831209,
+      "learning_rate": 2.9689190428165397e-05,
+      "loss": 0.9989,
+      "mean_token_accuracy": 0.7009221613407135,
+      "num_tokens": 227367283.0,
+      "step": 92880
+    },
+    {
+      "epoch": 0.8516549005226002,
+      "learning_rate": 2.9670853580269552e-05,
+      "loss": 0.9628,
+      "mean_token_accuracy": 0.7050508320331573,
+      "num_tokens": 227396838.0,
+      "step": 92890
+    },
+    {
+      "epoch": 0.8517465847620794,
+      "learning_rate": 2.9652516732373707e-05,
+      "loss": 0.9862,
+      "mean_token_accuracy": 0.7043337821960449,
+      "num_tokens": 227427306.0,
+      "step": 92900
+    },
+    {
+      "epoch": 0.8518382690015587,
+      "learning_rate": 2.963417988447786e-05,
+      "loss": 1.0065,
+      "mean_token_accuracy": 0.6985112607479096,
+      "num_tokens": 227458049.0,
+      "step": 92910
+    },
+    {
+      "epoch": 0.8519299532410378,
+      "learning_rate": 2.9615843036582014e-05,
+      "loss": 0.9808,
+      "mean_token_accuracy": 0.7056490421295166,
+      "num_tokens": 227488684.0,
+      "step": 92920
+    },
+    {
+      "epoch": 0.8520216374805171,
+      "learning_rate": 2.9597506188686163e-05,
+      "loss": 0.9879,
+      "mean_token_accuracy": 0.6996640086174011,
+      "num_tokens": 227519462.0,
+      "step": 92930
+    },
+    {
+      "epoch": 0.8521133217199963,
+      "learning_rate": 2.9579169340790318e-05,
+      "loss": 0.9627,
+      "mean_token_accuracy": 0.7048620879650116,
+      "num_tokens": 227547906.0,
+      "step": 92940
+    },
+    {
+      "epoch": 0.8522050059594756,
+      "learning_rate": 2.9560832492894473e-05,
+      "loss": 0.9794,
+      "mean_token_accuracy": 0.7041147887706757,
+      "num_tokens": 227578381.0,
+      "step": 92950
+    },
+    {
+      "epoch": 0.8522966901989548,
+      "learning_rate": 2.9542495644998625e-05,
+      "loss": 1.0082,
+      "mean_token_accuracy": 0.6972006142139435,
+      "num_tokens": 227608937.0,
+      "step": 92960
+    },
+    {
+      "epoch": 0.8523883744384341,
+      "learning_rate": 2.952415879710278e-05,
+      "loss": 0.9965,
+      "mean_token_accuracy": 0.7008302807807922,
+      "num_tokens": 227637957.0,
+      "step": 92970
+    },
+    {
+      "epoch": 0.8524800586779133,
+      "learning_rate": 2.9505821949206936e-05,
+      "loss": 0.9912,
+      "mean_token_accuracy": 0.7032665610313416,
+      "num_tokens": 227668663.0,
+      "step": 92980
+    },
+    {
+      "epoch": 0.8525717429173925,
+      "learning_rate": 2.9487485101311084e-05,
+      "loss": 0.9642,
+      "mean_token_accuracy": 0.7147470891475678,
+      "num_tokens": 227699322.0,
+      "step": 92990
+    },
+    {
+      "epoch": 0.8526634271568717,
+      "learning_rate": 2.946914825341524e-05,
+      "loss": 0.9716,
+      "mean_token_accuracy": 0.7087487041950226,
+      "num_tokens": 227726986.0,
+      "step": 93000
+    },
+    {
+      "epoch": 0.852755111396351,
+      "learning_rate": 2.945081140551939e-05,
+      "loss": 0.9902,
+      "mean_token_accuracy": 0.7011621952056885,
+      "num_tokens": 227757175.0,
+      "step": 93010
+    },
+    {
+      "epoch": 0.8528467956358302,
+      "learning_rate": 2.9432474557623546e-05,
+      "loss": 0.9922,
+      "mean_token_accuracy": 0.7050215721130371,
+      "num_tokens": 227788463.0,
+      "step": 93020
+    },
+    {
+      "epoch": 0.8529384798753095,
+      "learning_rate": 2.9414137709727702e-05,
+      "loss": 1.0041,
+      "mean_token_accuracy": 0.7073045194149017,
+      "num_tokens": 227818333.0,
+      "step": 93030
+    },
+    {
+      "epoch": 0.8530301641147887,
+      "learning_rate": 2.939580086183185e-05,
+      "loss": 0.987,
+      "mean_token_accuracy": 0.7010485291481018,
+      "num_tokens": 227848632.0,
+      "step": 93040
+    },
+    {
+      "epoch": 0.8531218483542679,
+      "learning_rate": 2.9377464013936005e-05,
+      "loss": 0.9979,
+      "mean_token_accuracy": 0.7053632616996766,
+      "num_tokens": 227879608.0,
+      "step": 93050
+    },
+    {
+      "epoch": 0.8532135325937471,
+      "learning_rate": 2.9359127166040157e-05,
+      "loss": 0.9789,
+      "mean_token_accuracy": 0.7041413962841034,
+      "num_tokens": 227907252.0,
+      "step": 93060
+    },
+    {
+      "epoch": 0.8533052168332264,
+      "learning_rate": 2.9340790318144313e-05,
+      "loss": 0.9767,
+      "mean_token_accuracy": 0.7085775375366211,
+      "num_tokens": 227936995.0,
+      "step": 93070
+    },
+    {
+      "epoch": 0.8533969010727056,
+      "learning_rate": 2.9322453470248468e-05,
+      "loss": 0.9912,
+      "mean_token_accuracy": 0.7036928236484528,
+      "num_tokens": 227967171.0,
+      "step": 93080
+    },
+    {
+      "epoch": 0.8534885853121849,
+      "learning_rate": 2.9304116622352616e-05,
+      "loss": 1.0076,
+      "mean_token_accuracy": 0.7038045585155487,
+      "num_tokens": 227997434.0,
+      "step": 93090
+    },
+    {
+      "epoch": 0.8535802695516641,
+      "learning_rate": 2.928577977445677e-05,
+      "loss": 0.9962,
+      "mean_token_accuracy": 0.7029118537902832,
+      "num_tokens": 228028749.0,
+      "step": 93100
+    },
+    {
+      "epoch": 0.8536719537911434,
+      "learning_rate": 2.9267442926560923e-05,
+      "loss": 0.9875,
+      "mean_token_accuracy": 0.7051879584789276,
+      "num_tokens": 228059345.0,
+      "step": 93110
+    },
+    {
+      "epoch": 0.8537636380306225,
+      "learning_rate": 2.924910607866508e-05,
+      "loss": 0.9984,
+      "mean_token_accuracy": 0.6979428648948669,
+      "num_tokens": 228087885.0,
+      "step": 93120
+    },
+    {
+      "epoch": 0.8538553222701017,
+      "learning_rate": 2.9230769230769234e-05,
+      "loss": 0.9948,
+      "mean_token_accuracy": 0.6957296907901764,
+      "num_tokens": 228117280.0,
+      "step": 93130
+    },
+    {
+      "epoch": 0.853947006509581,
+      "learning_rate": 2.9212432382873382e-05,
+      "loss": 1.021,
+      "mean_token_accuracy": 0.6990654051303864,
+      "num_tokens": 228148102.0,
+      "step": 93140
+    },
+    {
+      "epoch": 0.8540386907490602,
+      "learning_rate": 2.9194095534977538e-05,
+      "loss": 1.0296,
+      "mean_token_accuracy": 0.6973524630069733,
+      "num_tokens": 228178163.0,
+      "step": 93150
+    },
+    {
+      "epoch": 0.8541303749885395,
+      "learning_rate": 2.9175758687081693e-05,
+      "loss": 1.0202,
+      "mean_token_accuracy": 0.6996358692646026,
+      "num_tokens": 228209516.0,
+      "step": 93160
+    },
+    {
+      "epoch": 0.8542220592280187,
+      "learning_rate": 2.9157421839185845e-05,
+      "loss": 1.0032,
+      "mean_token_accuracy": 0.7008218824863434,
+      "num_tokens": 228240313.0,
+      "step": 93170
+    },
+    {
+      "epoch": 0.8543137434674979,
+      "learning_rate": 2.913908499129e-05,
+      "loss": 0.9588,
+      "mean_token_accuracy": 0.7090696334838867,
+      "num_tokens": 228267800.0,
+      "step": 93180
+    },
+    {
+      "epoch": 0.8544054277069771,
+      "learning_rate": 2.912074814339415e-05,
+      "loss": 1.0105,
+      "mean_token_accuracy": 0.6994145393371582,
+      "num_tokens": 228297318.0,
+      "step": 93190
+    },
+    {
+      "epoch": 0.8544971119464564,
+      "learning_rate": 2.9102411295498304e-05,
+      "loss": 0.9708,
+      "mean_token_accuracy": 0.706597113609314,
+      "num_tokens": 228328054.0,
+      "step": 93200
+    },
+    {
+      "epoch": 0.8545887961859356,
+      "learning_rate": 2.908407444760246e-05,
+      "loss": 0.9846,
+      "mean_token_accuracy": 0.7094608187675476,
+      "num_tokens": 228358528.0,
+      "step": 93210
+    },
+    {
+      "epoch": 0.8546804804254149,
+      "learning_rate": 2.906573759970661e-05,
+      "loss": 0.9662,
+      "mean_token_accuracy": 0.7111461758613586,
+      "num_tokens": 228388798.0,
+      "step": 93220
+    },
+    {
+      "epoch": 0.8547721646648941,
+      "learning_rate": 2.9047400751810766e-05,
+      "loss": 0.9762,
+      "mean_token_accuracy": 0.7065777361392975,
+      "num_tokens": 228418669.0,
+      "step": 93230
+    },
+    {
+      "epoch": 0.8548638489043734,
+      "learning_rate": 2.9029063903914915e-05,
+      "loss": 0.9534,
+      "mean_token_accuracy": 0.707000195980072,
+      "num_tokens": 228446004.0,
+      "step": 93240
+    },
+    {
+      "epoch": 0.8549555331438525,
+      "learning_rate": 2.901072705601907e-05,
+      "loss": 1.0229,
+      "mean_token_accuracy": 0.6997511804103851,
+      "num_tokens": 228476661.0,
+      "step": 93250
+    },
+    {
+      "epoch": 0.8550472173833318,
+      "learning_rate": 2.8992390208123225e-05,
+      "loss": 0.9833,
+      "mean_token_accuracy": 0.7088791608810425,
+      "num_tokens": 228506208.0,
+      "step": 93260
+    },
+    {
+      "epoch": 0.855138901622811,
+      "learning_rate": 2.8974053360227377e-05,
+      "loss": 1.0133,
+      "mean_token_accuracy": 0.7014099001884461,
+      "num_tokens": 228537490.0,
+      "step": 93270
+    },
+    {
+      "epoch": 0.8552305858622903,
+      "learning_rate": 2.8955716512331532e-05,
+      "loss": 0.9899,
+      "mean_token_accuracy": 0.7015993654727936,
+      "num_tokens": 228568692.0,
+      "step": 93280
+    },
+    {
+      "epoch": 0.8553222701017695,
+      "learning_rate": 2.8937379664435688e-05,
+      "loss": 1.002,
+      "mean_token_accuracy": 0.7001181602478027,
+      "num_tokens": 228598813.0,
+      "step": 93290
+    },
+    {
+      "epoch": 0.8554139543412488,
+      "learning_rate": 2.8919042816539836e-05,
+      "loss": 0.9799,
+      "mean_token_accuracy": 0.7049507439136505,
+      "num_tokens": 228627304.0,
+      "step": 93300
+    },
+    {
+      "epoch": 0.8555056385807279,
+      "learning_rate": 2.890070596864399e-05,
+      "loss": 1.0223,
+      "mean_token_accuracy": 0.696961623430252,
+      "num_tokens": 228657602.0,
+      "step": 93310
+    },
+    {
+      "epoch": 0.8555973228202072,
+      "learning_rate": 2.8882369120748143e-05,
+      "loss": 1.0106,
+      "mean_token_accuracy": 0.701234781742096,
+      "num_tokens": 228689627.0,
+      "step": 93320
+    },
+    {
+      "epoch": 0.8556890070596864,
+      "learning_rate": 2.88640322728523e-05,
+      "loss": 0.9927,
+      "mean_token_accuracy": 0.7064007699489594,
+      "num_tokens": 228720524.0,
+      "step": 93330
+    },
+    {
+      "epoch": 0.8557806912991657,
+      "learning_rate": 2.8845695424956454e-05,
+      "loss": 0.9997,
+      "mean_token_accuracy": 0.7018190264701843,
+      "num_tokens": 228750989.0,
+      "step": 93340
+    },
+    {
+      "epoch": 0.8558723755386449,
+      "learning_rate": 2.8827358577060602e-05,
+      "loss": 1.0075,
+      "mean_token_accuracy": 0.7025477647781372,
+      "num_tokens": 228781701.0,
+      "step": 93350
+    },
+    {
+      "epoch": 0.8559640597781242,
+      "learning_rate": 2.8809021729164757e-05,
+      "loss": 0.9801,
+      "mean_token_accuracy": 0.7058349132537842,
+      "num_tokens": 228811126.0,
+      "step": 93360
+    },
+    {
+      "epoch": 0.8560557440176034,
+      "learning_rate": 2.879068488126891e-05,
+      "loss": 0.9824,
+      "mean_token_accuracy": 0.7093196511268616,
+      "num_tokens": 228841124.0,
+      "step": 93370
+    },
+    {
+      "epoch": 0.8561474282570826,
+      "learning_rate": 2.8772348033373065e-05,
+      "loss": 1.0145,
+      "mean_token_accuracy": 0.6966848611831665,
+      "num_tokens": 228872263.0,
+      "step": 93380
+    },
+    {
+      "epoch": 0.8562391124965618,
+      "learning_rate": 2.875401118547722e-05,
+      "loss": 0.9932,
+      "mean_token_accuracy": 0.7081319630146027,
+      "num_tokens": 228902807.0,
+      "step": 93390
+    },
+    {
+      "epoch": 0.8563307967360411,
+      "learning_rate": 2.8735674337581368e-05,
+      "loss": 1.0117,
+      "mean_token_accuracy": 0.698861300945282,
+      "num_tokens": 228933482.0,
+      "step": 93400
+    },
+    {
+      "epoch": 0.8564224809755203,
+      "learning_rate": 2.8717337489685524e-05,
+      "loss": 0.9978,
+      "mean_token_accuracy": 0.704002457857132,
+      "num_tokens": 228963433.0,
+      "step": 93410
+    },
+    {
+      "epoch": 0.8565141652149996,
+      "learning_rate": 2.8699000641789675e-05,
+      "loss": 0.9639,
+      "mean_token_accuracy": 0.7072536885738373,
+      "num_tokens": 228991320.0,
+      "step": 93420
+    },
+    {
+      "epoch": 0.8566058494544788,
+      "learning_rate": 2.868066379389383e-05,
+      "loss": 0.9952,
+      "mean_token_accuracy": 0.7035923779010773,
+      "num_tokens": 229022233.0,
+      "step": 93430
+    },
+    {
+      "epoch": 0.856697533693958,
+      "learning_rate": 2.8662326945997986e-05,
+      "loss": 0.9896,
+      "mean_token_accuracy": 0.7070930659770965,
+      "num_tokens": 229053338.0,
+      "step": 93440
+    },
+    {
+      "epoch": 0.8567892179334372,
+      "learning_rate": 2.8643990098102134e-05,
+      "loss": 0.986,
+      "mean_token_accuracy": 0.7019206285476685,
+      "num_tokens": 229084415.0,
+      "step": 93450
+    },
+    {
+      "epoch": 0.8568809021729165,
+      "learning_rate": 2.862565325020629e-05,
+      "loss": 0.9855,
+      "mean_token_accuracy": 0.7055654048919677,
+      "num_tokens": 229114439.0,
+      "step": 93460
+    },
+    {
+      "epoch": 0.8569725864123957,
+      "learning_rate": 2.8607316402310448e-05,
+      "loss": 0.9565,
+      "mean_token_accuracy": 0.7126664459705353,
+      "num_tokens": 229145282.0,
+      "step": 93470
+    },
+    {
+      "epoch": 0.857064270651875,
+      "learning_rate": 2.8588979554414597e-05,
+      "loss": 0.9589,
+      "mean_token_accuracy": 0.7095115303993225,
+      "num_tokens": 229174284.0,
+      "step": 93480
+    },
+    {
+      "epoch": 0.8571559548913542,
+      "learning_rate": 2.8570642706518752e-05,
+      "loss": 0.9664,
+      "mean_token_accuracy": 0.7075452685356141,
+      "num_tokens": 229205198.0,
+      "step": 93490
+    },
+    {
+      "epoch": 0.8572476391308335,
+      "learning_rate": 2.85523058586229e-05,
+      "loss": 1.0149,
+      "mean_token_accuracy": 0.6994210004806518,
+      "num_tokens": 229235339.0,
+      "step": 93500
+    },
+    {
+      "epoch": 0.8573393233703126,
+      "learning_rate": 2.8533969010727056e-05,
+      "loss": 0.9983,
+      "mean_token_accuracy": 0.7030344009399414,
+      "num_tokens": 229265819.0,
+      "step": 93510
+    },
+    {
+      "epoch": 0.8574310076097919,
+      "learning_rate": 2.8515632162831214e-05,
+      "loss": 0.9721,
+      "mean_token_accuracy": 0.7081470489501953,
+      "num_tokens": 229296826.0,
+      "step": 93520
+    },
+    {
+      "epoch": 0.8575226918492711,
+      "learning_rate": 2.8497295314935363e-05,
+      "loss": 0.9868,
+      "mean_token_accuracy": 0.7028627216815948,
+      "num_tokens": 229327198.0,
+      "step": 93530
+    },
+    {
+      "epoch": 0.8576143760887504,
+      "learning_rate": 2.8478958467039518e-05,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.7105356931686402,
+      "num_tokens": 229355379.0,
+      "step": 93540
+    },
+    {
+      "epoch": 0.8577060603282296,
+      "learning_rate": 2.8460621619143667e-05,
+      "loss": 1.0173,
+      "mean_token_accuracy": 0.7038970828056336,
+      "num_tokens": 229386855.0,
+      "step": 93550
+    },
+    {
+      "epoch": 0.8577977445677089,
+      "learning_rate": 2.8442284771247822e-05,
+      "loss": 0.97,
+      "mean_token_accuracy": 0.7084010183811188,
+      "num_tokens": 229417133.0,
+      "step": 93560
+    },
+    {
+      "epoch": 0.857889428807188,
+      "learning_rate": 2.842394792335198e-05,
+      "loss": 1.0135,
+      "mean_token_accuracy": 0.6986473262310028,
+      "num_tokens": 229448517.0,
+      "step": 93570
+    },
+    {
+      "epoch": 0.8579811130466672,
+      "learning_rate": 2.840561107545613e-05,
+      "loss": 0.9992,
+      "mean_token_accuracy": 0.6998164474964141,
+      "num_tokens": 229479214.0,
+      "step": 93580
+    },
+    {
+      "epoch": 0.8580727972861465,
+      "learning_rate": 2.8387274227560284e-05,
+      "loss": 0.9878,
+      "mean_token_accuracy": 0.7021778881549835,
+      "num_tokens": 229509169.0,
+      "step": 93590
+    },
+    {
+      "epoch": 0.8581644815256257,
+      "learning_rate": 2.836893737966444e-05,
+      "loss": 0.9576,
+      "mean_token_accuracy": 0.7058420538902282,
+      "num_tokens": 229537172.0,
+      "step": 93600
+    },
+    {
+      "epoch": 0.858256165765105,
+      "learning_rate": 2.835060053176859e-05,
+      "loss": 0.9875,
+      "mean_token_accuracy": 0.7006553530693054,
+      "num_tokens": 229567771.0,
+      "step": 93610
+    },
+    {
+      "epoch": 0.8583478500045842,
+      "learning_rate": 2.8332263683872747e-05,
+      "loss": 1.0289,
+      "mean_token_accuracy": 0.6926831424236297,
+      "num_tokens": 229598995.0,
+      "step": 93620
+    },
+    {
+      "epoch": 0.8584395342440635,
+      "learning_rate": 2.8313926835976895e-05,
+      "loss": 1.0102,
+      "mean_token_accuracy": 0.6992372453212738,
+      "num_tokens": 229629697.0,
+      "step": 93630
+    },
+    {
+      "epoch": 0.8585312184835426,
+      "learning_rate": 2.829558998808105e-05,
+      "loss": 1.0076,
+      "mean_token_accuracy": 0.7017565727233886,
+      "num_tokens": 229659414.0,
+      "step": 93640
+    },
+    {
+      "epoch": 0.8586229027230219,
+      "learning_rate": 2.8277253140185206e-05,
+      "loss": 1.0009,
+      "mean_token_accuracy": 0.7049071252346039,
+      "num_tokens": 229690851.0,
+      "step": 93650
+    },
+    {
+      "epoch": 0.8587145869625011,
+      "learning_rate": 2.8258916292289357e-05,
+      "loss": 0.9649,
+      "mean_token_accuracy": 0.7046798527240753,
+      "num_tokens": 229717969.0,
+      "step": 93660
+    },
+    {
+      "epoch": 0.8588062712019804,
+      "learning_rate": 2.8240579444393513e-05,
+      "loss": 1.0143,
+      "mean_token_accuracy": 0.699289733171463,
+      "num_tokens": 229749126.0,
+      "step": 93670
+    },
+    {
+      "epoch": 0.8588979554414596,
+      "learning_rate": 2.822224259649766e-05,
+      "loss": 1.0323,
+      "mean_token_accuracy": 0.6974205195903778,
+      "num_tokens": 229779925.0,
+      "step": 93680
+    },
+    {
+      "epoch": 0.8589896396809389,
+      "learning_rate": 2.8203905748601816e-05,
+      "loss": 0.9913,
+      "mean_token_accuracy": 0.7032601416110993,
+      "num_tokens": 229811482.0,
+      "step": 93690
+    },
+    {
+      "epoch": 0.859081323920418,
+      "learning_rate": 2.8185568900705972e-05,
+      "loss": 0.9861,
+      "mean_token_accuracy": 0.7054169058799744,
+      "num_tokens": 229842304.0,
+      "step": 93700
+    },
+    {
+      "epoch": 0.8591730081598973,
+      "learning_rate": 2.8167232052810124e-05,
+      "loss": 1.0035,
+      "mean_token_accuracy": 0.7035200297832489,
+      "num_tokens": 229872102.0,
+      "step": 93710
+    },
+    {
+      "epoch": 0.8592646923993765,
+      "learning_rate": 2.814889520491428e-05,
+      "loss": 0.9992,
+      "mean_token_accuracy": 0.7006593167781829,
+      "num_tokens": 229900293.0,
+      "step": 93720
+    },
+    {
+      "epoch": 0.8593563766388558,
+      "learning_rate": 2.8130558357018427e-05,
+      "loss": 0.9557,
+      "mean_token_accuracy": 0.710857379436493,
+      "num_tokens": 229931341.0,
+      "step": 93730
+    },
+    {
+      "epoch": 0.859448060878335,
+      "learning_rate": 2.8112221509122583e-05,
+      "loss": 0.9887,
+      "mean_token_accuracy": 0.7041126251220703,
+      "num_tokens": 229961720.0,
+      "step": 93740
+    },
+    {
+      "epoch": 0.8595397451178143,
+      "learning_rate": 2.8093884661226738e-05,
+      "loss": 0.9628,
+      "mean_token_accuracy": 0.707321721315384,
+      "num_tokens": 229991875.0,
+      "step": 93750
+    },
+    {
+      "epoch": 0.8596314293572935,
+      "learning_rate": 2.807554781333089e-05,
+      "loss": 1.0085,
+      "mean_token_accuracy": 0.6946305811405182,
+      "num_tokens": 230022718.0,
+      "step": 93760
+    },
+    {
+      "epoch": 0.8597231135967727,
+      "learning_rate": 2.8057210965435045e-05,
+      "loss": 1.0083,
+      "mean_token_accuracy": 0.7017239272594452,
+      "num_tokens": 230052335.0,
+      "step": 93770
+    },
+    {
+      "epoch": 0.8598147978362519,
+      "learning_rate": 2.80388741175392e-05,
+      "loss": 1.0019,
+      "mean_token_accuracy": 0.6990306317806244,
+      "num_tokens": 230080329.0,
+      "step": 93780
+    },
+    {
+      "epoch": 0.8599064820757312,
+      "learning_rate": 2.802053726964335e-05,
+      "loss": 0.9721,
+      "mean_token_accuracy": 0.7119702637195587,
+      "num_tokens": 230109908.0,
+      "step": 93790
+    },
+    {
+      "epoch": 0.8599981663152104,
+      "learning_rate": 2.8002200421747504e-05,
+      "loss": 0.9931,
+      "mean_token_accuracy": 0.7025001227855683,
+      "num_tokens": 230140443.0,
+      "step": 93800
+    },
+    {
+      "epoch": 0.8600898505546897,
+      "learning_rate": 2.7983863573851656e-05,
+      "loss": 1.0093,
+      "mean_token_accuracy": 0.7040551781654358,
+      "num_tokens": 230170316.0,
+      "step": 93810
+    },
+    {
+      "epoch": 0.8601815347941689,
+      "learning_rate": 2.796552672595581e-05,
+      "loss": 0.9965,
+      "mean_token_accuracy": 0.7011925458908081,
+      "num_tokens": 230200380.0,
+      "step": 93820
+    },
+    {
+      "epoch": 0.8602732190336481,
+      "learning_rate": 2.7947189878059966e-05,
+      "loss": 0.992,
+      "mean_token_accuracy": 0.7017391383647918,
+      "num_tokens": 230230488.0,
+      "step": 93830
+    },
+    {
+      "epoch": 0.8603649032731273,
+      "learning_rate": 2.7928853030164115e-05,
+      "loss": 0.978,
+      "mean_token_accuracy": 0.7033119022846221,
+      "num_tokens": 230258354.0,
+      "step": 93840
+    },
+    {
+      "epoch": 0.8604565875126066,
+      "learning_rate": 2.791051618226827e-05,
+      "loss": 0.9922,
+      "mean_token_accuracy": 0.7034036040306091,
+      "num_tokens": 230288543.0,
+      "step": 93850
+    },
+    {
+      "epoch": 0.8605482717520858,
+      "learning_rate": 2.7892179334372422e-05,
+      "loss": 0.9878,
+      "mean_token_accuracy": 0.7018364071846008,
+      "num_tokens": 230319075.0,
+      "step": 93860
+    },
+    {
+      "epoch": 0.8606399559915651,
+      "learning_rate": 2.7873842486476577e-05,
+      "loss": 0.9814,
+      "mean_token_accuracy": 0.7068473875522614,
+      "num_tokens": 230349285.0,
+      "step": 93870
+    },
+    {
+      "epoch": 0.8607316402310443,
+      "learning_rate": 2.7855505638580732e-05,
+      "loss": 1.0028,
+      "mean_token_accuracy": 0.6989137411117554,
+      "num_tokens": 230379656.0,
+      "step": 93880
+    },
+    {
+      "epoch": 0.8608233244705236,
+      "learning_rate": 2.783716879068488e-05,
+      "loss": 0.9928,
+      "mean_token_accuracy": 0.698908656835556,
+      "num_tokens": 230410080.0,
+      "step": 93890
+    },
+    {
+      "epoch": 0.8609150087100027,
+      "learning_rate": 2.7818831942789036e-05,
+      "loss": 0.9962,
+      "mean_token_accuracy": 0.7035494387149811,
+      "num_tokens": 230438196.0,
+      "step": 93900
+    },
+    {
+      "epoch": 0.861006692949482,
+      "learning_rate": 2.780049509489319e-05,
+      "loss": 1.0011,
+      "mean_token_accuracy": 0.7058464050292969,
+      "num_tokens": 230468924.0,
+      "step": 93910
+    },
+    {
+      "epoch": 0.8610983771889612,
+      "learning_rate": 2.7782158246997343e-05,
+      "loss": 0.998,
+      "mean_token_accuracy": 0.702565610408783,
+      "num_tokens": 230499205.0,
+      "step": 93920
+    },
+    {
+      "epoch": 0.8611900614284405,
+      "learning_rate": 2.77638213991015e-05,
+      "loss": 0.973,
+      "mean_token_accuracy": 0.706601870059967,
+      "num_tokens": 230528770.0,
+      "step": 93930
+    },
+    {
+      "epoch": 0.8612817456679197,
+      "learning_rate": 2.7745484551205647e-05,
+      "loss": 1.0087,
+      "mean_token_accuracy": 0.6979085981845856,
+      "num_tokens": 230559888.0,
+      "step": 93940
+    },
+    {
+      "epoch": 0.861373429907399,
+      "learning_rate": 2.7727147703309802e-05,
+      "loss": 1.0179,
+      "mean_token_accuracy": 0.6932749927043915,
+      "num_tokens": 230589187.0,
+      "step": 93950
+    },
+    {
+      "epoch": 0.8614651141468781,
+      "learning_rate": 2.7708810855413958e-05,
+      "loss": 0.9798,
+      "mean_token_accuracy": 0.7050952792167664,
+      "num_tokens": 230616591.0,
+      "step": 93960
+    },
+    {
+      "epoch": 0.8615567983863573,
+      "learning_rate": 2.769047400751811e-05,
+      "loss": 0.967,
+      "mean_token_accuracy": 0.7060283780097961,
+      "num_tokens": 230646084.0,
+      "step": 93970
+    },
+    {
+      "epoch": 0.8616484826258366,
+      "learning_rate": 2.7672137159622265e-05,
+      "loss": 0.969,
+      "mean_token_accuracy": 0.7095028698444367,
+      "num_tokens": 230674694.0,
+      "step": 93980
+    },
+    {
+      "epoch": 0.8617401668653158,
+      "learning_rate": 2.7653800311726413e-05,
+      "loss": 0.9702,
+      "mean_token_accuracy": 0.705646425485611,
+      "num_tokens": 230705039.0,
+      "step": 93990
+    },
+    {
+      "epoch": 0.8618318511047951,
+      "learning_rate": 2.763546346383057e-05,
+      "loss": 1.002,
+      "mean_token_accuracy": 0.7054389953613281,
+      "num_tokens": 230735426.0,
+      "step": 94000
+    },
+    {
+      "epoch": 0.8619235353442743,
+      "learning_rate": 2.7617126615934724e-05,
+      "loss": 1.0114,
+      "mean_token_accuracy": 0.6922630369663239,
+      "num_tokens": 230764606.0,
+      "step": 94010
+    },
+    {
+      "epoch": 0.8620152195837536,
+      "learning_rate": 2.7598789768038876e-05,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.7118881821632386,
+      "num_tokens": 230792909.0,
+      "step": 94020
+    },
+    {
+      "epoch": 0.8621069038232327,
+      "learning_rate": 2.758045292014303e-05,
+      "loss": 0.9866,
+      "mean_token_accuracy": 0.7054672062397003,
+      "num_tokens": 230822743.0,
+      "step": 94030
+    },
+    {
+      "epoch": 0.862198588062712,
+      "learning_rate": 2.756211607224718e-05,
+      "loss": 0.9656,
+      "mean_token_accuracy": 0.7076707184314728,
+      "num_tokens": 230852177.0,
+      "step": 94040
+    },
+    {
+      "epoch": 0.8622902723021912,
+      "learning_rate": 2.7543779224351334e-05,
+      "loss": 0.9814,
+      "mean_token_accuracy": 0.7012471199035645,
+      "num_tokens": 230881719.0,
+      "step": 94050
+    },
+    {
+      "epoch": 0.8623819565416705,
+      "learning_rate": 2.752544237645549e-05,
+      "loss": 1.0074,
+      "mean_token_accuracy": 0.7048568487167358,
+      "num_tokens": 230912558.0,
+      "step": 94060
+    },
+    {
+      "epoch": 0.8624736407811497,
+      "learning_rate": 2.750710552855964e-05,
+      "loss": 1.0192,
+      "mean_token_accuracy": 0.6959864974021912,
+      "num_tokens": 230942152.0,
+      "step": 94070
+    },
+    {
+      "epoch": 0.862565325020629,
+      "learning_rate": 2.7488768680663797e-05,
+      "loss": 1.0034,
+      "mean_token_accuracy": 0.6967722237110138,
+      "num_tokens": 230970808.0,
+      "step": 94080
+    },
+    {
+      "epoch": 0.8626570092601082,
+      "learning_rate": 2.7470431832767952e-05,
+      "loss": 0.9748,
+      "mean_token_accuracy": 0.707591074705124,
+      "num_tokens": 231000175.0,
+      "step": 94090
+    },
+    {
+      "epoch": 0.8627486934995874,
+      "learning_rate": 2.74520949848721e-05,
+      "loss": 0.9897,
+      "mean_token_accuracy": 0.7098795533180237,
+      "num_tokens": 231029227.0,
+      "step": 94100
+    },
+    {
+      "epoch": 0.8628403777390666,
+      "learning_rate": 2.7433758136976256e-05,
+      "loss": 1.006,
+      "mean_token_accuracy": 0.7012426614761352,
+      "num_tokens": 231059404.0,
+      "step": 94110
+    },
+    {
+      "epoch": 0.8629320619785459,
+      "learning_rate": 2.7415421289080408e-05,
+      "loss": 1.0141,
+      "mean_token_accuracy": 0.6978681802749633,
+      "num_tokens": 231089996.0,
+      "step": 94120
+    },
+    {
+      "epoch": 0.8630237462180251,
+      "learning_rate": 2.7397084441184563e-05,
+      "loss": 0.9544,
+      "mean_token_accuracy": 0.7131826937198639,
+      "num_tokens": 231119730.0,
+      "step": 94130
+    },
+    {
+      "epoch": 0.8631154304575044,
+      "learning_rate": 2.7378747593288718e-05,
+      "loss": 0.984,
+      "mean_token_accuracy": 0.7006629467010498,
+      "num_tokens": 231146825.0,
+      "step": 94140
+    },
+    {
+      "epoch": 0.8632071146969836,
+      "learning_rate": 2.7360410745392867e-05,
+      "loss": 0.9961,
+      "mean_token_accuracy": 0.6985842466354371,
+      "num_tokens": 231177431.0,
+      "step": 94150
+    },
+    {
+      "epoch": 0.8632987989364628,
+      "learning_rate": 2.7342073897497022e-05,
+      "loss": 1.0247,
+      "mean_token_accuracy": 0.6996906638145447,
+      "num_tokens": 231207506.0,
+      "step": 94160
+    },
+    {
+      "epoch": 0.863390483175942,
+      "learning_rate": 2.7323737049601174e-05,
+      "loss": 0.9896,
+      "mean_token_accuracy": 0.7020743370056153,
+      "num_tokens": 231236608.0,
+      "step": 94170
+    },
+    {
+      "epoch": 0.8634821674154213,
+      "learning_rate": 2.730540020170533e-05,
+      "loss": 1.0192,
+      "mean_token_accuracy": 0.699338161945343,
+      "num_tokens": 231266722.0,
+      "step": 94180
+    },
+    {
+      "epoch": 0.8635738516549005,
+      "learning_rate": 2.7287063353809484e-05,
+      "loss": 0.9814,
+      "mean_token_accuracy": 0.6987640619277954,
+      "num_tokens": 231296180.0,
+      "step": 94190
+    },
+    {
+      "epoch": 0.8636655358943798,
+      "learning_rate": 2.7268726505913633e-05,
+      "loss": 0.9644,
+      "mean_token_accuracy": 0.710695606470108,
+      "num_tokens": 231324156.0,
+      "step": 94200
+    },
+    {
+      "epoch": 0.863757220133859,
+      "learning_rate": 2.7250389658017788e-05,
+      "loss": 0.977,
+      "mean_token_accuracy": 0.7077572464942932,
+      "num_tokens": 231355039.0,
+      "step": 94210
+    },
+    {
+      "epoch": 0.8638489043733383,
+      "learning_rate": 2.7232052810121943e-05,
+      "loss": 0.9703,
+      "mean_token_accuracy": 0.7070972979068756,
+      "num_tokens": 231384375.0,
+      "step": 94220
+    },
+    {
+      "epoch": 0.8639405886128174,
+      "learning_rate": 2.7213715962226095e-05,
+      "loss": 0.9995,
+      "mean_token_accuracy": 0.7043955206871033,
+      "num_tokens": 231413825.0,
+      "step": 94230
+    },
+    {
+      "epoch": 0.8640322728522967,
+      "learning_rate": 2.719537911433025e-05,
+      "loss": 0.9708,
+      "mean_token_accuracy": 0.7065513074398041,
+      "num_tokens": 231443164.0,
+      "step": 94240
+    },
+    {
+      "epoch": 0.8641239570917759,
+      "learning_rate": 2.71770422664344e-05,
+      "loss": 1.0008,
+      "mean_token_accuracy": 0.7002853393554688,
+      "num_tokens": 231473461.0,
+      "step": 94250
+    },
+    {
+      "epoch": 0.8642156413312552,
+      "learning_rate": 2.7158705418538554e-05,
+      "loss": 0.9558,
+      "mean_token_accuracy": 0.7069169521331787,
+      "num_tokens": 231501040.0,
+      "step": 94260
+    },
+    {
+      "epoch": 0.8643073255707344,
+      "learning_rate": 2.714036857064271e-05,
+      "loss": 0.9944,
+      "mean_token_accuracy": 0.7037984371185303,
+      "num_tokens": 231531327.0,
+      "step": 94270
+    },
+    {
+      "epoch": 0.8643990098102137,
+      "learning_rate": 2.712203172274686e-05,
+      "loss": 1.0082,
+      "mean_token_accuracy": 0.7028061032295227,
+      "num_tokens": 231560377.0,
+      "step": 94280
+    },
+    {
+      "epoch": 0.8644906940496928,
+      "learning_rate": 2.7103694874851017e-05,
+      "loss": 0.9888,
+      "mean_token_accuracy": 0.7046666860580444,
+      "num_tokens": 231589614.0,
+      "step": 94290
+    },
+    {
+      "epoch": 0.8645823782891721,
+      "learning_rate": 2.7085358026955165e-05,
+      "loss": 0.9998,
+      "mean_token_accuracy": 0.700808972120285,
+      "num_tokens": 231618458.0,
+      "step": 94300
+    },
+    {
+      "epoch": 0.8646740625286513,
+      "learning_rate": 2.706702117905932e-05,
+      "loss": 0.99,
+      "mean_token_accuracy": 0.7057264983654022,
+      "num_tokens": 231649163.0,
+      "step": 94310
+    },
+    {
+      "epoch": 0.8647657467681306,
+      "learning_rate": 2.7048684331163476e-05,
+      "loss": 0.9561,
+      "mean_token_accuracy": 0.7108542919158936,
+      "num_tokens": 231676889.0,
+      "step": 94320
+    },
+    {
+      "epoch": 0.8648574310076098,
+      "learning_rate": 2.7030347483267627e-05,
+      "loss": 1.0102,
+      "mean_token_accuracy": 0.6991156876087189,
+      "num_tokens": 231706959.0,
+      "step": 94330
+    },
+    {
+      "epoch": 0.8649491152470891,
+      "learning_rate": 2.7012010635371783e-05,
+      "loss": 0.9949,
+      "mean_token_accuracy": 0.7035431027412414,
+      "num_tokens": 231736139.0,
+      "step": 94340
+    },
+    {
+      "epoch": 0.8650407994865683,
+      "learning_rate": 2.699367378747593e-05,
+      "loss": 0.966,
+      "mean_token_accuracy": 0.7110806465148926,
+      "num_tokens": 231765629.0,
+      "step": 94350
+    },
+    {
+      "epoch": 0.8651324837260475,
+      "learning_rate": 2.6975336939580086e-05,
+      "loss": 0.99,
+      "mean_token_accuracy": 0.7026050984859467,
+      "num_tokens": 231795445.0,
+      "step": 94360
+    },
+    {
+      "epoch": 0.8652241679655267,
+      "learning_rate": 2.695700009168424e-05,
+      "loss": 0.9915,
+      "mean_token_accuracy": 0.7057330727577209,
+      "num_tokens": 231824756.0,
+      "step": 94370
+    },
+    {
+      "epoch": 0.865315852205006,
+      "learning_rate": 2.6938663243788394e-05,
+      "loss": 0.9787,
+      "mean_token_accuracy": 0.7037960708141326,
+      "num_tokens": 231852294.0,
+      "step": 94380
+    },
+    {
+      "epoch": 0.8654075364444852,
+      "learning_rate": 2.692032639589255e-05,
+      "loss": 0.9734,
+      "mean_token_accuracy": 0.7058798670768738,
+      "num_tokens": 231881665.0,
+      "step": 94390
+    },
+    {
+      "epoch": 0.8654992206839645,
+      "learning_rate": 2.6901989547996704e-05,
+      "loss": 1.0152,
+      "mean_token_accuracy": 0.6974703252315522,
+      "num_tokens": 231911070.0,
+      "step": 94400
+    },
+    {
+      "epoch": 0.8655909049234437,
+      "learning_rate": 2.6883652700100853e-05,
+      "loss": 0.984,
+      "mean_token_accuracy": 0.7004394233226776,
+      "num_tokens": 231940680.0,
+      "step": 94410
+    },
+    {
+      "epoch": 0.8656825891629228,
+      "learning_rate": 2.6865315852205008e-05,
+      "loss": 0.9946,
+      "mean_token_accuracy": 0.7020886182785034,
+      "num_tokens": 231969937.0,
+      "step": 94420
+    },
+    {
+      "epoch": 0.8657742734024021,
+      "learning_rate": 2.684697900430916e-05,
+      "loss": 1.0296,
+      "mean_token_accuracy": 0.694058620929718,
+      "num_tokens": 231998670.0,
+      "step": 94430
+    },
+    {
+      "epoch": 0.8658659576418813,
+      "learning_rate": 2.6828642156413315e-05,
+      "loss": 1.0034,
+      "mean_token_accuracy": 0.6922867357730865,
+      "num_tokens": 232026165.0,
+      "step": 94440
+    },
+    {
+      "epoch": 0.8659576418813606,
+      "learning_rate": 2.681030530851747e-05,
+      "loss": 1.0265,
+      "mean_token_accuracy": 0.6965956032276154,
+      "num_tokens": 232056410.0,
+      "step": 94450
+    },
+    {
+      "epoch": 0.8660493261208398,
+      "learning_rate": 2.679196846062162e-05,
+      "loss": 0.999,
+      "mean_token_accuracy": 0.7027850270271301,
+      "num_tokens": 232086808.0,
+      "step": 94460
+    },
+    {
+      "epoch": 0.8661410103603191,
+      "learning_rate": 2.6773631612725774e-05,
+      "loss": 0.978,
+      "mean_token_accuracy": 0.7054557502269745,
+      "num_tokens": 232115545.0,
+      "step": 94470
+    },
+    {
+      "epoch": 0.8662326945997983,
+      "learning_rate": 2.6755294764829926e-05,
+      "loss": 0.9514,
+      "mean_token_accuracy": 0.7109884262084961,
+      "num_tokens": 232145307.0,
+      "step": 94480
+    },
+    {
+      "epoch": 0.8663243788392775,
+      "learning_rate": 2.673695791693408e-05,
+      "loss": 0.9716,
+      "mean_token_accuracy": 0.7080773949623108,
+      "num_tokens": 232174173.0,
+      "step": 94490
+    },
+    {
+      "epoch": 0.8664160630787567,
+      "learning_rate": 2.6718621069038236e-05,
+      "loss": 0.9585,
+      "mean_token_accuracy": 0.7067557334899902,
+      "num_tokens": 232202027.0,
+      "step": 94500
+    },
+    {
+      "epoch": 0.866507747318236,
+      "learning_rate": 2.6700284221142385e-05,
+      "loss": 0.9805,
+      "mean_token_accuracy": 0.7113572180271148,
+      "num_tokens": 232232278.0,
+      "step": 94510
+    },
+    {
+      "epoch": 0.8665994315577152,
+      "learning_rate": 2.668194737324654e-05,
+      "loss": 0.9828,
+      "mean_token_accuracy": 0.7049754381179809,
+      "num_tokens": 232261493.0,
+      "step": 94520
+    },
+    {
+      "epoch": 0.8666911157971945,
+      "learning_rate": 2.6663610525350695e-05,
+      "loss": 0.9949,
+      "mean_token_accuracy": 0.7011536180973053,
+      "num_tokens": 232290512.0,
+      "step": 94530
+    },
+    {
+      "epoch": 0.8667828000366737,
+      "learning_rate": 2.6645273677454847e-05,
+      "loss": 1.0086,
+      "mean_token_accuracy": 0.697703379392624,
+      "num_tokens": 232320930.0,
+      "step": 94540
+    },
+    {
+      "epoch": 0.8668744842761529,
+      "learning_rate": 2.6626936829559002e-05,
+      "loss": 0.9921,
+      "mean_token_accuracy": 0.7040180504322052,
+      "num_tokens": 232350121.0,
+      "step": 94550
+    },
+    {
+      "epoch": 0.8669661685156321,
+      "learning_rate": 2.660859998166315e-05,
+      "loss": 0.9816,
+      "mean_token_accuracy": 0.7010047674179077,
+      "num_tokens": 232378606.0,
+      "step": 94560
+    },
+    {
+      "epoch": 0.8670578527551114,
+      "learning_rate": 2.6590263133767306e-05,
+      "loss": 1.0194,
+      "mean_token_accuracy": 0.6984214723110199,
+      "num_tokens": 232408145.0,
+      "step": 94570
+    },
+    {
+      "epoch": 0.8671495369945906,
+      "learning_rate": 2.657192628587146e-05,
+      "loss": 0.9622,
+      "mean_token_accuracy": 0.7074166297912597,
+      "num_tokens": 232437784.0,
+      "step": 94580
+    },
+    {
+      "epoch": 0.8672412212340699,
+      "learning_rate": 2.6553589437975613e-05,
+      "loss": 0.9427,
+      "mean_token_accuracy": 0.7127412855625153,
+      "num_tokens": 232467373.0,
+      "step": 94590
+    },
+    {
+      "epoch": 0.8673329054735491,
+      "learning_rate": 2.653525259007977e-05,
+      "loss": 1.0081,
+      "mean_token_accuracy": 0.6999133467674256,
+      "num_tokens": 232496390.0,
+      "step": 94600
+    },
+    {
+      "epoch": 0.8674245897130284,
+      "learning_rate": 2.6516915742183917e-05,
+      "loss": 0.998,
+      "mean_token_accuracy": 0.7039475798606872,
+      "num_tokens": 232525541.0,
+      "step": 94610
+    },
+    {
+      "epoch": 0.8675162739525075,
+      "learning_rate": 2.6498578894288072e-05,
+      "loss": 0.9571,
+      "mean_token_accuracy": 0.7070056378841401,
+      "num_tokens": 232554456.0,
+      "step": 94620
+    },
+    {
+      "epoch": 0.8676079581919868,
+      "learning_rate": 2.6480242046392228e-05,
+      "loss": 0.9749,
+      "mean_token_accuracy": 0.7079532027244568,
+      "num_tokens": 232583904.0,
+      "step": 94630
+    },
+    {
+      "epoch": 0.867699642431466,
+      "learning_rate": 2.646190519849638e-05,
+      "loss": 0.956,
+      "mean_token_accuracy": 0.7131348252296448,
+      "num_tokens": 232613638.0,
+      "step": 94640
+    },
+    {
+      "epoch": 0.8677913266709453,
+      "learning_rate": 2.6443568350600535e-05,
+      "loss": 0.9597,
+      "mean_token_accuracy": 0.7060459673404693,
+      "num_tokens": 232642199.0,
+      "step": 94650
+    },
+    {
+      "epoch": 0.8678830109104245,
+      "learning_rate": 2.6425231502704683e-05,
+      "loss": 0.9639,
+      "mean_token_accuracy": 0.7111473381519318,
+      "num_tokens": 232670925.0,
+      "step": 94660
+    },
+    {
+      "epoch": 0.8679746951499038,
+      "learning_rate": 2.640689465480884e-05,
+      "loss": 0.9892,
+      "mean_token_accuracy": 0.7017959654331207,
+      "num_tokens": 232699820.0,
+      "step": 94670
+    },
+    {
+      "epoch": 0.8680663793893829,
+      "learning_rate": 2.6388557806912994e-05,
+      "loss": 0.9897,
+      "mean_token_accuracy": 0.7052757978439331,
+      "num_tokens": 232728175.0,
+      "step": 94680
+    },
+    {
+      "epoch": 0.8681580636288622,
+      "learning_rate": 2.6370220959017145e-05,
+      "loss": 0.9394,
+      "mean_token_accuracy": 0.7146795988082886,
+      "num_tokens": 232756588.0,
+      "step": 94690
+    },
+    {
+      "epoch": 0.8682497478683414,
+      "learning_rate": 2.63518841111213e-05,
+      "loss": 0.9949,
+      "mean_token_accuracy": 0.6988874673843384,
+      "num_tokens": 232785942.0,
+      "step": 94700
+    },
+    {
+      "epoch": 0.8683414321078207,
+      "learning_rate": 2.6333547263225456e-05,
+      "loss": 0.9749,
+      "mean_token_accuracy": 0.7053543865680695,
+      "num_tokens": 232814163.0,
+      "step": 94710
+    },
+    {
+      "epoch": 0.8684331163472999,
+      "learning_rate": 2.6315210415329604e-05,
+      "loss": 1.0181,
+      "mean_token_accuracy": 0.6953714549541473,
+      "num_tokens": 232843536.0,
+      "step": 94720
+    },
+    {
+      "epoch": 0.8685248005867792,
+      "learning_rate": 2.629687356743376e-05,
+      "loss": 0.9833,
+      "mean_token_accuracy": 0.7034175992012024,
+      "num_tokens": 232872725.0,
+      "step": 94730
+    },
+    {
+      "epoch": 0.8686164848262584,
+      "learning_rate": 2.627853671953791e-05,
+      "loss": 0.9623,
+      "mean_token_accuracy": 0.7084377586841584,
+      "num_tokens": 232899619.0,
+      "step": 94740
+    },
+    {
+      "epoch": 0.8687081690657376,
+      "learning_rate": 2.6260199871642067e-05,
+      "loss": 0.9837,
+      "mean_token_accuracy": 0.7027204811573029,
+      "num_tokens": 232929152.0,
+      "step": 94750
+    },
+    {
+      "epoch": 0.8687998533052168,
+      "learning_rate": 2.6241863023746222e-05,
+      "loss": 1.0264,
+      "mean_token_accuracy": 0.6943724989891052,
+      "num_tokens": 232958774.0,
+      "step": 94760
+    },
+    {
+      "epoch": 0.8688915375446961,
+      "learning_rate": 2.622352617585037e-05,
+      "loss": 0.9549,
+      "mean_token_accuracy": 0.7107896268367767,
+      "num_tokens": 232987364.0,
+      "step": 94770
+    },
+    {
+      "epoch": 0.8689832217841753,
+      "learning_rate": 2.6205189327954526e-05,
+      "loss": 0.9701,
+      "mean_token_accuracy": 0.7117876529693603,
+      "num_tokens": 233015145.0,
+      "step": 94780
+    },
+    {
+      "epoch": 0.8690749060236546,
+      "learning_rate": 2.6186852480058678e-05,
+      "loss": 0.9652,
+      "mean_token_accuracy": 0.7065112113952636,
+      "num_tokens": 233043891.0,
+      "step": 94790
+    },
+    {
+      "epoch": 0.8691665902631338,
+      "learning_rate": 2.6168515632162833e-05,
+      "loss": 0.9754,
+      "mean_token_accuracy": 0.7085714340209961,
+      "num_tokens": 233071931.0,
+      "step": 94800
+    },
+    {
+      "epoch": 0.869258274502613,
+      "learning_rate": 2.6150178784266988e-05,
+      "loss": 0.996,
+      "mean_token_accuracy": 0.702934467792511,
+      "num_tokens": 233101663.0,
+      "step": 94810
+    },
+    {
+      "epoch": 0.8693499587420922,
+      "learning_rate": 2.6131841936371137e-05,
+      "loss": 0.9667,
+      "mean_token_accuracy": 0.704881340265274,
+      "num_tokens": 233130016.0,
+      "step": 94820
+    },
+    {
+      "epoch": 0.8694416429815714,
+      "learning_rate": 2.6113505088475292e-05,
+      "loss": 0.99,
+      "mean_token_accuracy": 0.7008226096630097,
+      "num_tokens": 233158300.0,
+      "step": 94830
+    },
+    {
+      "epoch": 0.8695333272210507,
+      "learning_rate": 2.6095168240579444e-05,
+      "loss": 1.0013,
+      "mean_token_accuracy": 0.701530122756958,
+      "num_tokens": 233188440.0,
+      "step": 94840
+    },
+    {
+      "epoch": 0.86962501146053,
+      "learning_rate": 2.60768313926836e-05,
+      "loss": 0.9663,
+      "mean_token_accuracy": 0.7130723536014557,
+      "num_tokens": 233218179.0,
+      "step": 94850
+    },
+    {
+      "epoch": 0.8697166957000092,
+      "learning_rate": 2.6058494544787754e-05,
+      "loss": 1.0023,
+      "mean_token_accuracy": 0.6981267809867859,
+      "num_tokens": 233246321.0,
+      "step": 94860
+    },
+    {
+      "epoch": 0.8698083799394885,
+      "learning_rate": 2.6040157696891903e-05,
+      "loss": 0.9526,
+      "mean_token_accuracy": 0.7105185151100158,
+      "num_tokens": 233275232.0,
+      "step": 94870
+    },
+    {
+      "epoch": 0.8699000641789676,
+      "learning_rate": 2.6021820848996058e-05,
+      "loss": 0.9756,
+      "mean_token_accuracy": 0.7113110899925232,
+      "num_tokens": 233302680.0,
+      "step": 94880
+    },
+    {
+      "epoch": 0.8699917484184468,
+      "learning_rate": 2.6003484001100213e-05,
+      "loss": 0.9531,
+      "mean_token_accuracy": 0.7159253478050231,
+      "num_tokens": 233331749.0,
+      "step": 94890
+    },
+    {
+      "epoch": 0.8700834326579261,
+      "learning_rate": 2.5985147153204365e-05,
+      "loss": 0.9449,
+      "mean_token_accuracy": 0.7144491136074066,
+      "num_tokens": 233360578.0,
+      "step": 94900
+    },
+    {
+      "epoch": 0.8701751168974053,
+      "learning_rate": 2.596681030530852e-05,
+      "loss": 0.9573,
+      "mean_token_accuracy": 0.7086441040039062,
+      "num_tokens": 233389492.0,
+      "step": 94910
+    },
+    {
+      "epoch": 0.8702668011368846,
+      "learning_rate": 2.594847345741267e-05,
+      "loss": 0.9643,
+      "mean_token_accuracy": 0.7050327181816101,
+      "num_tokens": 233417646.0,
+      "step": 94920
+    },
+    {
+      "epoch": 0.8703584853763638,
+      "learning_rate": 2.5930136609516824e-05,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.7074222266674042,
+      "num_tokens": 233445439.0,
+      "step": 94930
+    },
+    {
+      "epoch": 0.870450169615843,
+      "learning_rate": 2.591179976162098e-05,
+      "loss": 0.9242,
+      "mean_token_accuracy": 0.7211675465106964,
+      "num_tokens": 233473966.0,
+      "step": 94940
+    },
+    {
+      "epoch": 0.8705418538553222,
+      "learning_rate": 2.589346291372513e-05,
+      "loss": 0.9854,
+      "mean_token_accuracy": 0.7037066578865051,
+      "num_tokens": 233502820.0,
+      "step": 94950
+    },
+    {
+      "epoch": 0.8706335380948015,
+      "learning_rate": 2.5875126065829287e-05,
+      "loss": 0.9972,
+      "mean_token_accuracy": 0.7012357950210572,
+      "num_tokens": 233530257.0,
+      "step": 94960
+    },
+    {
+      "epoch": 0.8707252223342807,
+      "learning_rate": 2.5856789217933435e-05,
+      "loss": 0.9663,
+      "mean_token_accuracy": 0.7035387933254242,
+      "num_tokens": 233560472.0,
+      "step": 94970
+    },
+    {
+      "epoch": 0.87081690657376,
+      "learning_rate": 2.583845237003759e-05,
+      "loss": 0.9637,
+      "mean_token_accuracy": 0.708971232175827,
+      "num_tokens": 233588403.0,
+      "step": 94980
+    },
+    {
+      "epoch": 0.8709085908132392,
+      "learning_rate": 2.5820115522141746e-05,
+      "loss": 1.0025,
+      "mean_token_accuracy": 0.7040158629417419,
+      "num_tokens": 233615823.0,
+      "step": 94990
+    },
+    {
+      "epoch": 0.8710002750527185,
+      "learning_rate": 2.5801778674245897e-05,
+      "loss": 0.9713,
+      "mean_token_accuracy": 0.7070195496082305,
+      "num_tokens": 233645321.0,
+      "step": 95000
+    },
+    {
+      "epoch": 0.8710919592921976,
+      "learning_rate": 2.5783441826350053e-05,
+      "loss": 0.9863,
+      "mean_token_accuracy": 0.7015394985675811,
+      "num_tokens": 233675199.0,
+      "step": 95010
+    },
+    {
+      "epoch": 0.8711836435316769,
+      "learning_rate": 2.5765104978454208e-05,
+      "loss": 0.9805,
+      "mean_token_accuracy": 0.7036999762058258,
+      "num_tokens": 233703256.0,
+      "step": 95020
+    },
+    {
+      "epoch": 0.8712753277711561,
+      "learning_rate": 2.5746768130558356e-05,
+      "loss": 0.9551,
+      "mean_token_accuracy": 0.7102802157402038,
+      "num_tokens": 233732170.0,
+      "step": 95030
+    },
+    {
+      "epoch": 0.8713670120106354,
+      "learning_rate": 2.572843128266251e-05,
+      "loss": 0.9677,
+      "mean_token_accuracy": 0.7078596889972687,
+      "num_tokens": 233760230.0,
+      "step": 95040
+    },
+    {
+      "epoch": 0.8714586962501146,
+      "learning_rate": 2.5710094434766664e-05,
+      "loss": 0.9759,
+      "mean_token_accuracy": 0.7082966566085815,
+      "num_tokens": 233789705.0,
+      "step": 95050
+    },
+    {
+      "epoch": 0.8715503804895939,
+      "learning_rate": 2.569175758687082e-05,
+      "loss": 0.9796,
+      "mean_token_accuracy": 0.703370201587677,
+      "num_tokens": 233817490.0,
+      "step": 95060
+    },
+    {
+      "epoch": 0.871642064729073,
+      "learning_rate": 2.5673420738974974e-05,
+      "loss": 0.9765,
+      "mean_token_accuracy": 0.7046668887138366,
+      "num_tokens": 233845745.0,
+      "step": 95070
+    },
+    {
+      "epoch": 0.8717337489685523,
+      "learning_rate": 2.5655083891079123e-05,
+      "loss": 0.9799,
+      "mean_token_accuracy": 0.7063077211380004,
+      "num_tokens": 233874955.0,
+      "step": 95080
+    },
+    {
+      "epoch": 0.8718254332080315,
+      "learning_rate": 2.5636747043183278e-05,
+      "loss": 0.9647,
+      "mean_token_accuracy": 0.7077544212341309,
+      "num_tokens": 233902908.0,
+      "step": 95090
+    },
+    {
+      "epoch": 0.8719171174475108,
+      "learning_rate": 2.561841019528743e-05,
+      "loss": 0.9876,
+      "mean_token_accuracy": 0.6998224079608917,
+      "num_tokens": 233931375.0,
+      "step": 95100
+    },
+    {
+      "epoch": 0.87200880168699,
+      "learning_rate": 2.5600073347391585e-05,
+      "loss": 0.9622,
+      "mean_token_accuracy": 0.7097045123577118,
+      "num_tokens": 233961956.0,
+      "step": 95110
+    },
+    {
+      "epoch": 0.8721004859264693,
+      "learning_rate": 2.558173649949574e-05,
+      "loss": 0.9701,
+      "mean_token_accuracy": 0.7076535224914551,
+      "num_tokens": 233990360.0,
+      "step": 95120
+    },
+    {
+      "epoch": 0.8721921701659485,
+      "learning_rate": 2.556339965159989e-05,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.7034613966941834,
+      "num_tokens": 234020465.0,
+      "step": 95130
+    },
+    {
+      "epoch": 0.8722838544054277,
+      "learning_rate": 2.5545062803704044e-05,
+      "loss": 0.9643,
+      "mean_token_accuracy": 0.708831250667572,
+      "num_tokens": 234048471.0,
+      "step": 95140
+    },
+    {
+      "epoch": 0.8723755386449069,
+      "learning_rate": 2.5526725955808196e-05,
+      "loss": 1.0,
+      "mean_token_accuracy": 0.7020787358283996,
+      "num_tokens": 234077948.0,
+      "step": 95150
+    },
+    {
+      "epoch": 0.8724672228843862,
+      "learning_rate": 2.550838910791235e-05,
+      "loss": 0.9822,
+      "mean_token_accuracy": 0.7070713996887207,
+      "num_tokens": 234106271.0,
+      "step": 95160
+    },
+    {
+      "epoch": 0.8725589071238654,
+      "learning_rate": 2.5490052260016506e-05,
+      "loss": 0.9846,
+      "mean_token_accuracy": 0.7102813124656677,
+      "num_tokens": 234134655.0,
+      "step": 95170
+    },
+    {
+      "epoch": 0.8726505913633447,
+      "learning_rate": 2.5471715412120655e-05,
+      "loss": 0.9719,
+      "mean_token_accuracy": 0.7100964784622192,
+      "num_tokens": 234162795.0,
+      "step": 95180
+    },
+    {
+      "epoch": 0.8727422756028239,
+      "learning_rate": 2.545337856422481e-05,
+      "loss": 0.9509,
+      "mean_token_accuracy": 0.7117873907089234,
+      "num_tokens": 234192416.0,
+      "step": 95190
+    },
+    {
+      "epoch": 0.872833959842303,
+      "learning_rate": 2.5435041716328965e-05,
+      "loss": 0.972,
+      "mean_token_accuracy": 0.7072008550167084,
+      "num_tokens": 234221732.0,
+      "step": 95200
+    },
+    {
+      "epoch": 0.8729256440817823,
+      "learning_rate": 2.5416704868433117e-05,
+      "loss": 0.9356,
+      "mean_token_accuracy": 0.7139463067054749,
+      "num_tokens": 234250512.0,
+      "step": 95210
+    },
+    {
+      "epoch": 0.8730173283212616,
+      "learning_rate": 2.5398368020537272e-05,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.7087880313396454,
+      "num_tokens": 234278950.0,
+      "step": 95220
+    },
+    {
+      "epoch": 0.8731090125607408,
+      "learning_rate": 2.538003117264142e-05,
+      "loss": 0.9947,
+      "mean_token_accuracy": 0.6958837747573853,
+      "num_tokens": 234308344.0,
+      "step": 95230
+    },
+    {
+      "epoch": 0.87320069680022,
+      "learning_rate": 2.5361694324745576e-05,
+      "loss": 0.9845,
+      "mean_token_accuracy": 0.7043018460273742,
+      "num_tokens": 234337370.0,
+      "step": 95240
+    },
+    {
+      "epoch": 0.8732923810396993,
+      "learning_rate": 2.534335747684973e-05,
+      "loss": 0.9475,
+      "mean_token_accuracy": 0.7157825469970703,
+      "num_tokens": 234367341.0,
+      "step": 95250
+    },
+    {
+      "epoch": 0.8733840652791786,
+      "learning_rate": 2.5325020628953883e-05,
+      "loss": 0.9805,
+      "mean_token_accuracy": 0.7008367896080017,
+      "num_tokens": 234396779.0,
+      "step": 95260
+    },
+    {
+      "epoch": 0.8734757495186577,
+      "learning_rate": 2.530668378105804e-05,
+      "loss": 0.9728,
+      "mean_token_accuracy": 0.7053045749664306,
+      "num_tokens": 234425469.0,
+      "step": 95270
+    },
+    {
+      "epoch": 0.873567433758137,
+      "learning_rate": 2.5288346933162187e-05,
+      "loss": 0.9709,
+      "mean_token_accuracy": 0.7075625956058502,
+      "num_tokens": 234453451.0,
+      "step": 95280
+    },
+    {
+      "epoch": 0.8736591179976162,
+      "learning_rate": 2.5270010085266342e-05,
+      "loss": 0.9498,
+      "mean_token_accuracy": 0.711888062953949,
+      "num_tokens": 234480748.0,
+      "step": 95290
+    },
+    {
+      "epoch": 0.8737508022370954,
+      "learning_rate": 2.5251673237370498e-05,
+      "loss": 0.9419,
+      "mean_token_accuracy": 0.7127594470977783,
+      "num_tokens": 234509115.0,
+      "step": 95300
+    },
+    {
+      "epoch": 0.8738424864765747,
+      "learning_rate": 2.523333638947465e-05,
+      "loss": 0.9791,
+      "mean_token_accuracy": 0.7048508524894714,
+      "num_tokens": 234537613.0,
+      "step": 95310
+    },
+    {
+      "epoch": 0.873934170716054,
+      "learning_rate": 2.5214999541578805e-05,
+      "loss": 0.9787,
+      "mean_token_accuracy": 0.7064248204231263,
+      "num_tokens": 234566964.0,
+      "step": 95320
+    },
+    {
+      "epoch": 0.8740258549555332,
+      "learning_rate": 2.519666269368296e-05,
+      "loss": 0.9684,
+      "mean_token_accuracy": 0.7089863121509552,
+      "num_tokens": 234595953.0,
+      "step": 95330
+    },
+    {
+      "epoch": 0.8741175391950123,
+      "learning_rate": 2.517832584578711e-05,
+      "loss": 0.9586,
+      "mean_token_accuracy": 0.7119853258132934,
+      "num_tokens": 234624611.0,
+      "step": 95340
+    },
+    {
+      "epoch": 0.8742092234344916,
+      "learning_rate": 2.5159988997891264e-05,
+      "loss": 0.9696,
+      "mean_token_accuracy": 0.7046167969703674,
+      "num_tokens": 234652585.0,
+      "step": 95350
+    },
+    {
+      "epoch": 0.8743009076739708,
+      "learning_rate": 2.5141652149995415e-05,
+      "loss": 0.9918,
+      "mean_token_accuracy": 0.7043960213661193,
+      "num_tokens": 234682260.0,
+      "step": 95360
+    },
+    {
+      "epoch": 0.8743925919134501,
+      "learning_rate": 2.512331530209957e-05,
+      "loss": 1.0129,
+      "mean_token_accuracy": 0.6981764316558838,
+      "num_tokens": 234711641.0,
+      "step": 95370
+    },
+    {
+      "epoch": 0.8744842761529293,
+      "learning_rate": 2.5104978454203726e-05,
+      "loss": 0.9632,
+      "mean_token_accuracy": 0.7097373843193054,
+      "num_tokens": 234740476.0,
+      "step": 95380
+    },
+    {
+      "epoch": 0.8745759603924086,
+      "learning_rate": 2.5086641606307874e-05,
+      "loss": 0.955,
+      "mean_token_accuracy": 0.7070928752422333,
+      "num_tokens": 234768921.0,
+      "step": 95390
+    },
+    {
+      "epoch": 0.8746676446318877,
+      "learning_rate": 2.506830475841203e-05,
+      "loss": 0.9468,
+      "mean_token_accuracy": 0.7109356701374054,
+      "num_tokens": 234796761.0,
+      "step": 95400
+    },
+    {
+      "epoch": 0.874759328871367,
+      "learning_rate": 2.504996791051618e-05,
+      "loss": 0.9568,
+      "mean_token_accuracy": 0.7114959239959717,
+      "num_tokens": 234825311.0,
+      "step": 95410
+    },
+    {
+      "epoch": 0.8748510131108462,
+      "learning_rate": 2.5031631062620337e-05,
+      "loss": 0.9503,
+      "mean_token_accuracy": 0.7101741313934327,
+      "num_tokens": 234854479.0,
+      "step": 95420
+    },
+    {
+      "epoch": 0.8749426973503255,
+      "learning_rate": 2.5013294214724492e-05,
+      "loss": 1.017,
+      "mean_token_accuracy": 0.6984456002712249,
+      "num_tokens": 234884382.0,
+      "step": 95430
+    },
+    {
+      "epoch": 0.8750343815898047,
+      "learning_rate": 2.4994957366828644e-05,
+      "loss": 0.9699,
+      "mean_token_accuracy": 0.7042559802532196,
+      "num_tokens": 234913271.0,
+      "step": 95440
+    },
+    {
+      "epoch": 0.875126065829284,
+      "learning_rate": 2.4976620518932796e-05,
+      "loss": 0.99,
+      "mean_token_accuracy": 0.7045483469963074,
+      "num_tokens": 234940914.0,
+      "step": 95450
+    },
+    {
+      "epoch": 0.8752177500687632,
+      "learning_rate": 2.495828367103695e-05,
+      "loss": 0.9818,
+      "mean_token_accuracy": 0.7026433408260345,
+      "num_tokens": 234968617.0,
+      "step": 95460
+    },
+    {
+      "epoch": 0.8753094343082424,
+      "learning_rate": 2.4939946823141103e-05,
+      "loss": 0.9761,
+      "mean_token_accuracy": 0.7051691889762879,
+      "num_tokens": 234997132.0,
+      "step": 95470
+    },
+    {
+      "epoch": 0.8754011185477216,
+      "learning_rate": 2.4921609975245255e-05,
+      "loss": 1.0013,
+      "mean_token_accuracy": 0.7004251718521118,
+      "num_tokens": 235025374.0,
+      "step": 95480
+    },
+    {
+      "epoch": 0.8754928027872009,
+      "learning_rate": 2.490327312734941e-05,
+      "loss": 0.9646,
+      "mean_token_accuracy": 0.7087277770042419,
+      "num_tokens": 235054127.0,
+      "step": 95490
+    },
+    {
+      "epoch": 0.8755844870266801,
+      "learning_rate": 2.4884936279453562e-05,
+      "loss": 0.9766,
+      "mean_token_accuracy": 0.7062869131565094,
+      "num_tokens": 235082647.0,
+      "step": 95500
+    },
+    {
+      "epoch": 0.8756761712661594,
+      "learning_rate": 2.4866599431557717e-05,
+      "loss": 0.9695,
+      "mean_token_accuracy": 0.7079085886478425,
+      "num_tokens": 235110432.0,
+      "step": 95510
+    },
+    {
+      "epoch": 0.8757678555056386,
+      "learning_rate": 2.484826258366187e-05,
+      "loss": 0.9696,
+      "mean_token_accuracy": 0.7054381608963013,
+      "num_tokens": 235138219.0,
+      "step": 95520
+    },
+    {
+      "epoch": 0.8758595397451178,
+      "learning_rate": 2.4829925735766024e-05,
+      "loss": 0.9694,
+      "mean_token_accuracy": 0.7025360882282257,
+      "num_tokens": 235166725.0,
+      "step": 95530
+    },
+    {
+      "epoch": 0.875951223984597,
+      "learning_rate": 2.4811588887870176e-05,
+      "loss": 0.9833,
+      "mean_token_accuracy": 0.7031918048858643,
+      "num_tokens": 235195793.0,
+      "step": 95540
+    },
+    {
+      "epoch": 0.8760429082240763,
+      "learning_rate": 2.4793252039974328e-05,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7089531123638153,
+      "num_tokens": 235225290.0,
+      "step": 95550
+    },
+    {
+      "epoch": 0.8761345924635555,
+      "learning_rate": 2.4774915192078483e-05,
+      "loss": 0.9587,
+      "mean_token_accuracy": 0.7031606495380401,
+      "num_tokens": 235253050.0,
+      "step": 95560
+    },
+    {
+      "epoch": 0.8762262767030348,
+      "learning_rate": 2.4756578344182635e-05,
+      "loss": 0.9654,
+      "mean_token_accuracy": 0.7063882887363434,
+      "num_tokens": 235281109.0,
+      "step": 95570
+    },
+    {
+      "epoch": 0.876317960942514,
+      "learning_rate": 2.473824149628679e-05,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.7035144686698913,
+      "num_tokens": 235309797.0,
+      "step": 95580
+    },
+    {
+      "epoch": 0.8764096451819933,
+      "learning_rate": 2.4719904648390942e-05,
+      "loss": 0.9657,
+      "mean_token_accuracy": 0.7078938066959382,
+      "num_tokens": 235338386.0,
+      "step": 95590
+    },
+    {
+      "epoch": 0.8765013294214724,
+      "learning_rate": 2.4701567800495094e-05,
+      "loss": 0.9946,
+      "mean_token_accuracy": 0.6995946943759919,
+      "num_tokens": 235367347.0,
+      "step": 95600
+    },
+    {
+      "epoch": 0.8765930136609517,
+      "learning_rate": 2.468323095259925e-05,
+      "loss": 0.9675,
+      "mean_token_accuracy": 0.7101439595222473,
+      "num_tokens": 235396414.0,
+      "step": 95610
+    },
+    {
+      "epoch": 0.8766846979004309,
+      "learning_rate": 2.4664894104703405e-05,
+      "loss": 0.9913,
+      "mean_token_accuracy": 0.6996130347251892,
+      "num_tokens": 235425276.0,
+      "step": 95620
+    },
+    {
+      "epoch": 0.8767763821399102,
+      "learning_rate": 2.4646557256807557e-05,
+      "loss": 0.9742,
+      "mean_token_accuracy": 0.7068098068237305,
+      "num_tokens": 235452571.0,
+      "step": 95630
+    },
+    {
+      "epoch": 0.8768680663793894,
+      "learning_rate": 2.462822040891171e-05,
+      "loss": 1.0056,
+      "mean_token_accuracy": 0.700124728679657,
+      "num_tokens": 235482425.0,
+      "step": 95640
+    },
+    {
+      "epoch": 0.8769597506188687,
+      "learning_rate": 2.460988356101586e-05,
+      "loss": 0.9326,
+      "mean_token_accuracy": 0.7112212598323822,
+      "num_tokens": 235510113.0,
+      "step": 95650
+    },
+    {
+      "epoch": 0.8770514348583478,
+      "learning_rate": 2.4591546713120016e-05,
+      "loss": 0.9737,
+      "mean_token_accuracy": 0.7041312277317047,
+      "num_tokens": 235538601.0,
+      "step": 95660
+    },
+    {
+      "epoch": 0.877143119097827,
+      "learning_rate": 2.457320986522417e-05,
+      "loss": 0.9726,
+      "mean_token_accuracy": 0.7038525938987732,
+      "num_tokens": 235567940.0,
+      "step": 95670
+    },
+    {
+      "epoch": 0.8772348033373063,
+      "learning_rate": 2.4554873017328323e-05,
+      "loss": 0.9739,
+      "mean_token_accuracy": 0.7062692642211914,
+      "num_tokens": 235597253.0,
+      "step": 95680
+    },
+    {
+      "epoch": 0.8773264875767856,
+      "learning_rate": 2.4536536169432475e-05,
+      "loss": 0.9766,
+      "mean_token_accuracy": 0.7078110098838806,
+      "num_tokens": 235625708.0,
+      "step": 95690
+    },
+    {
+      "epoch": 0.8774181718162648,
+      "learning_rate": 2.4518199321536626e-05,
+      "loss": 0.9599,
+      "mean_token_accuracy": 0.7113600492477417,
+      "num_tokens": 235653706.0,
+      "step": 95700
+    },
+    {
+      "epoch": 0.877509856055744,
+      "learning_rate": 2.449986247364078e-05,
+      "loss": 0.964,
+      "mean_token_accuracy": 0.7109102964401245,
+      "num_tokens": 235681221.0,
+      "step": 95710
+    },
+    {
+      "epoch": 0.8776015402952233,
+      "learning_rate": 2.4481525625744937e-05,
+      "loss": 0.9534,
+      "mean_token_accuracy": 0.7113423347473145,
+      "num_tokens": 235708927.0,
+      "step": 95720
+    },
+    {
+      "epoch": 0.8776932245347024,
+      "learning_rate": 2.446318877784909e-05,
+      "loss": 0.9683,
+      "mean_token_accuracy": 0.7058891594409943,
+      "num_tokens": 235737474.0,
+      "step": 95730
+    },
+    {
+      "epoch": 0.8777849087741817,
+      "learning_rate": 2.444485192995324e-05,
+      "loss": 0.9834,
+      "mean_token_accuracy": 0.7096074402332306,
+      "num_tokens": 235765542.0,
+      "step": 95740
+    },
+    {
+      "epoch": 0.8778765930136609,
+      "learning_rate": 2.4426515082057396e-05,
+      "loss": 0.9643,
+      "mean_token_accuracy": 0.7085775971412659,
+      "num_tokens": 235794028.0,
+      "step": 95750
+    },
+    {
+      "epoch": 0.8779682772531402,
+      "learning_rate": 2.4408178234161548e-05,
+      "loss": 0.9723,
+      "mean_token_accuracy": 0.7010929107666015,
+      "num_tokens": 235822112.0,
+      "step": 95760
+    },
+    {
+      "epoch": 0.8780599614926194,
+      "learning_rate": 2.4389841386265703e-05,
+      "loss": 0.9543,
+      "mean_token_accuracy": 0.7064978361129761,
+      "num_tokens": 235850543.0,
+      "step": 95770
+    },
+    {
+      "epoch": 0.8781516457320987,
+      "learning_rate": 2.4371504538369855e-05,
+      "loss": 0.9773,
+      "mean_token_accuracy": 0.7082438051700592,
+      "num_tokens": 235879312.0,
+      "step": 95780
+    },
+    {
+      "epoch": 0.8782433299715778,
+      "learning_rate": 2.4353167690474007e-05,
+      "loss": 0.9623,
+      "mean_token_accuracy": 0.710709398984909,
+      "num_tokens": 235908264.0,
+      "step": 95790
+    },
+    {
+      "epoch": 0.8783350142110571,
+      "learning_rate": 2.4334830842578162e-05,
+      "loss": 0.9693,
+      "mean_token_accuracy": 0.7027275800704956,
+      "num_tokens": 235937186.0,
+      "step": 95800
+    },
+    {
+      "epoch": 0.8784266984505363,
+      "learning_rate": 2.4316493994682317e-05,
+      "loss": 1.0035,
+      "mean_token_accuracy": 0.6995549619197845,
+      "num_tokens": 235966037.0,
+      "step": 95810
+    },
+    {
+      "epoch": 0.8785183826900156,
+      "learning_rate": 2.429815714678647e-05,
+      "loss": 0.9826,
+      "mean_token_accuracy": 0.7057016968727112,
+      "num_tokens": 235993270.0,
+      "step": 95820
+    },
+    {
+      "epoch": 0.8786100669294948,
+      "learning_rate": 2.427982029889062e-05,
+      "loss": 0.9874,
+      "mean_token_accuracy": 0.7001482546329498,
+      "num_tokens": 236022678.0,
+      "step": 95830
+    },
+    {
+      "epoch": 0.8787017511689741,
+      "learning_rate": 2.4261483450994776e-05,
+      "loss": 0.9667,
+      "mean_token_accuracy": 0.7085700333118439,
+      "num_tokens": 236050301.0,
+      "step": 95840
+    },
+    {
+      "epoch": 0.8787934354084533,
+      "learning_rate": 2.4243146603098928e-05,
+      "loss": 0.9667,
+      "mean_token_accuracy": 0.7068552613258362,
+      "num_tokens": 236077861.0,
+      "step": 95850
+    },
+    {
+      "epoch": 0.8788851196479325,
+      "learning_rate": 2.4224809755203083e-05,
+      "loss": 0.9695,
+      "mean_token_accuracy": 0.7068959414958954,
+      "num_tokens": 236105802.0,
+      "step": 95860
+    },
+    {
+      "epoch": 0.8789768038874117,
+      "learning_rate": 2.4206472907307235e-05,
+      "loss": 0.9593,
+      "mean_token_accuracy": 0.7121960997581482,
+      "num_tokens": 236134564.0,
+      "step": 95870
+    },
+    {
+      "epoch": 0.879068488126891,
+      "learning_rate": 2.4188136059411387e-05,
+      "loss": 0.9654,
+      "mean_token_accuracy": 0.7062169075012207,
+      "num_tokens": 236162176.0,
+      "step": 95880
+    },
+    {
+      "epoch": 0.8791601723663702,
+      "learning_rate": 2.4169799211515542e-05,
+      "loss": 0.96,
+      "mean_token_accuracy": 0.7049622416496277,
+      "num_tokens": 236191137.0,
+      "step": 95890
+    },
+    {
+      "epoch": 0.8792518566058495,
+      "learning_rate": 2.4151462363619694e-05,
+      "loss": 0.9617,
+      "mean_token_accuracy": 0.7068059146404266,
+      "num_tokens": 236220024.0,
+      "step": 95900
+    },
+    {
+      "epoch": 0.8793435408453287,
+      "learning_rate": 2.413312551572385e-05,
+      "loss": 0.9872,
+      "mean_token_accuracy": 0.6995436012744903,
+      "num_tokens": 236248522.0,
+      "step": 95910
+    },
+    {
+      "epoch": 0.8794352250848079,
+      "learning_rate": 2.4114788667828e-05,
+      "loss": 0.9547,
+      "mean_token_accuracy": 0.7080778479576111,
+      "num_tokens": 236277254.0,
+      "step": 95920
+    },
+    {
+      "epoch": 0.8795269093242871,
+      "learning_rate": 2.4096451819932157e-05,
+      "loss": 0.9917,
+      "mean_token_accuracy": 0.6987809240818024,
+      "num_tokens": 236305947.0,
+      "step": 95930
+    },
+    {
+      "epoch": 0.8796185935637664,
+      "learning_rate": 2.407811497203631e-05,
+      "loss": 0.9543,
+      "mean_token_accuracy": 0.7094142258167266,
+      "num_tokens": 236334529.0,
+      "step": 95940
+    },
+    {
+      "epoch": 0.8797102778032456,
+      "learning_rate": 2.405977812414046e-05,
+      "loss": 0.9831,
+      "mean_token_accuracy": 0.6997954726219178,
+      "num_tokens": 236362266.0,
+      "step": 95950
+    },
+    {
+      "epoch": 0.8798019620427249,
+      "learning_rate": 2.4041441276244616e-05,
+      "loss": 1.0043,
+      "mean_token_accuracy": 0.7003909707069397,
+      "num_tokens": 236390276.0,
+      "step": 95960
+    },
+    {
+      "epoch": 0.8798936462822041,
+      "learning_rate": 2.4023104428348767e-05,
+      "loss": 0.9631,
+      "mean_token_accuracy": 0.7065606951713562,
+      "num_tokens": 236418401.0,
+      "step": 95970
+    },
+    {
+      "epoch": 0.8799853305216834,
+      "learning_rate": 2.4004767580452923e-05,
+      "loss": 0.9725,
+      "mean_token_accuracy": 0.7057132840156555,
+      "num_tokens": 236446703.0,
+      "step": 95980
+    },
+    {
+      "epoch": 0.8800770147611625,
+      "learning_rate": 2.3986430732557075e-05,
+      "loss": 0.9855,
+      "mean_token_accuracy": 0.706520426273346,
+      "num_tokens": 236475330.0,
+      "step": 95990
+    },
+    {
+      "epoch": 0.8801686990006418,
+      "learning_rate": 2.3968093884661226e-05,
+      "loss": 0.9699,
+      "mean_token_accuracy": 0.7042117118835449,
+      "num_tokens": 236504253.0,
+      "step": 96000
+    },
+    {
+      "epoch": 0.880260383240121,
+      "learning_rate": 2.3949757036765382e-05,
+      "loss": 0.9846,
+      "mean_token_accuracy": 0.7061336159706115,
+      "num_tokens": 236532302.0,
+      "step": 96010
+    },
+    {
+      "epoch": 0.8803520674796003,
+      "learning_rate": 2.3931420188869537e-05,
+      "loss": 0.9829,
+      "mean_token_accuracy": 0.7032469272613525,
+      "num_tokens": 236561433.0,
+      "step": 96020
+    },
+    {
+      "epoch": 0.8804437517190795,
+      "learning_rate": 2.391308334097369e-05,
+      "loss": 0.9989,
+      "mean_token_accuracy": 0.6975636303424835,
+      "num_tokens": 236589528.0,
+      "step": 96030
+    },
+    {
+      "epoch": 0.8805354359585588,
+      "learning_rate": 2.389474649307784e-05,
+      "loss": 0.9552,
+      "mean_token_accuracy": 0.712497740983963,
+      "num_tokens": 236617946.0,
+      "step": 96040
+    },
+    {
+      "epoch": 0.8806271201980379,
+      "learning_rate": 2.3876409645181993e-05,
+      "loss": 0.9513,
+      "mean_token_accuracy": 0.7090511202812195,
+      "num_tokens": 236647618.0,
+      "step": 96050
+    },
+    {
+      "epoch": 0.8807188044375172,
+      "learning_rate": 2.3858072797286148e-05,
+      "loss": 0.9822,
+      "mean_token_accuracy": 0.7052377939224244,
+      "num_tokens": 236675720.0,
+      "step": 96060
+    },
+    {
+      "epoch": 0.8808104886769964,
+      "learning_rate": 2.3839735949390303e-05,
+      "loss": 0.9717,
+      "mean_token_accuracy": 0.7043699324131012,
+      "num_tokens": 236704286.0,
+      "step": 96070
+    },
+    {
+      "epoch": 0.8809021729164757,
+      "learning_rate": 2.3821399101494455e-05,
+      "loss": 0.9718,
+      "mean_token_accuracy": 0.706047248840332,
+      "num_tokens": 236732401.0,
+      "step": 96080
+    },
+    {
+      "epoch": 0.8809938571559549,
+      "learning_rate": 2.3803062253598607e-05,
+      "loss": 0.9767,
+      "mean_token_accuracy": 0.7095676183700561,
+      "num_tokens": 236760944.0,
+      "step": 96090
+    },
+    {
+      "epoch": 0.8810855413954342,
+      "learning_rate": 2.378472540570276e-05,
+      "loss": 0.9803,
+      "mean_token_accuracy": 0.7032656311988831,
+      "num_tokens": 236789216.0,
+      "step": 96100
+    },
+    {
+      "epoch": 0.8811772256349134,
+      "learning_rate": 2.3766388557806914e-05,
+      "loss": 0.9917,
+      "mean_token_accuracy": 0.7007404029369354,
+      "num_tokens": 236818347.0,
+      "step": 96110
+    },
+    {
+      "epoch": 0.8812689098743925,
+      "learning_rate": 2.374805170991107e-05,
+      "loss": 0.9521,
+      "mean_token_accuracy": 0.7090234696865082,
+      "num_tokens": 236846810.0,
+      "step": 96120
+    },
+    {
+      "epoch": 0.8813605941138718,
+      "learning_rate": 2.372971486201522e-05,
+      "loss": 0.962,
+      "mean_token_accuracy": 0.7073679208755493,
+      "num_tokens": 236875071.0,
+      "step": 96130
+    },
+    {
+      "epoch": 0.881452278353351,
+      "learning_rate": 2.3711378014119373e-05,
+      "loss": 0.9451,
+      "mean_token_accuracy": 0.7103154122829437,
+      "num_tokens": 236903111.0,
+      "step": 96140
+    },
+    {
+      "epoch": 0.8815439625928303,
+      "learning_rate": 2.3693041166223528e-05,
+      "loss": 0.9731,
+      "mean_token_accuracy": 0.7064365804195404,
+      "num_tokens": 236931871.0,
+      "step": 96150
+    },
+    {
+      "epoch": 0.8816356468323095,
+      "learning_rate": 2.367470431832768e-05,
+      "loss": 0.9943,
+      "mean_token_accuracy": 0.7025649309158325,
+      "num_tokens": 236961731.0,
+      "step": 96160
+    },
+    {
+      "epoch": 0.8817273310717888,
+      "learning_rate": 2.3656367470431835e-05,
+      "loss": 0.973,
+      "mean_token_accuracy": 0.7142005324363708,
+      "num_tokens": 236990347.0,
+      "step": 96170
+    },
+    {
+      "epoch": 0.8818190153112679,
+      "learning_rate": 2.3638030622535987e-05,
+      "loss": 0.9471,
+      "mean_token_accuracy": 0.7102129638195038,
+      "num_tokens": 237017421.0,
+      "step": 96180
+    },
+    {
+      "epoch": 0.8819106995507472,
+      "learning_rate": 2.361969377464014e-05,
+      "loss": 0.9288,
+      "mean_token_accuracy": 0.721216493844986,
+      "num_tokens": 237045388.0,
+      "step": 96190
+    },
+    {
+      "epoch": 0.8820023837902264,
+      "learning_rate": 2.3601356926744294e-05,
+      "loss": 0.9711,
+      "mean_token_accuracy": 0.707420414686203,
+      "num_tokens": 237073012.0,
+      "step": 96200
+    },
+    {
+      "epoch": 0.8820940680297057,
+      "learning_rate": 2.3583020078848446e-05,
+      "loss": 0.9903,
+      "mean_token_accuracy": 0.7036702990531921,
+      "num_tokens": 237102049.0,
+      "step": 96210
+    },
+    {
+      "epoch": 0.8821857522691849,
+      "learning_rate": 2.35646832309526e-05,
+      "loss": 0.9678,
+      "mean_token_accuracy": 0.7045400202274322,
+      "num_tokens": 237130406.0,
+      "step": 96220
+    },
+    {
+      "epoch": 0.8822774365086642,
+      "learning_rate": 2.3546346383056753e-05,
+      "loss": 0.9807,
+      "mean_token_accuracy": 0.7066488981246948,
+      "num_tokens": 237158095.0,
+      "step": 96230
+    },
+    {
+      "epoch": 0.8823691207481434,
+      "learning_rate": 2.352800953516091e-05,
+      "loss": 0.9539,
+      "mean_token_accuracy": 0.7105531454086303,
+      "num_tokens": 237185653.0,
+      "step": 96240
+    },
+    {
+      "epoch": 0.8824608049876226,
+      "learning_rate": 2.350967268726506e-05,
+      "loss": 0.9699,
+      "mean_token_accuracy": 0.7080231070518493,
+      "num_tokens": 237214270.0,
+      "step": 96250
+    },
+    {
+      "epoch": 0.8825524892271018,
+      "learning_rate": 2.3491335839369212e-05,
+      "loss": 0.9723,
+      "mean_token_accuracy": 0.7070541739463806,
+      "num_tokens": 237243048.0,
+      "step": 96260
+    },
+    {
+      "epoch": 0.8826441734665811,
+      "learning_rate": 2.3472998991473368e-05,
+      "loss": 0.983,
+      "mean_token_accuracy": 0.7049121856689453,
+      "num_tokens": 237271063.0,
+      "step": 96270
+    },
+    {
+      "epoch": 0.8827358577060603,
+      "learning_rate": 2.345466214357752e-05,
+      "loss": 0.9831,
+      "mean_token_accuracy": 0.7061281263828277,
+      "num_tokens": 237299560.0,
+      "step": 96280
+    },
+    {
+      "epoch": 0.8828275419455396,
+      "learning_rate": 2.3436325295681675e-05,
+      "loss": 0.9732,
+      "mean_token_accuracy": 0.7040581047534943,
+      "num_tokens": 237328309.0,
+      "step": 96290
+    },
+    {
+      "epoch": 0.8829192261850188,
+      "learning_rate": 2.3417988447785827e-05,
+      "loss": 0.9618,
+      "mean_token_accuracy": 0.7013525426387787,
+      "num_tokens": 237356287.0,
+      "step": 96300
+    },
+    {
+      "epoch": 0.883010910424498,
+      "learning_rate": 2.339965159988998e-05,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.7045096695423126,
+      "num_tokens": 237385418.0,
+      "step": 96310
+    },
+    {
+      "epoch": 0.8831025946639772,
+      "learning_rate": 2.3381314751994134e-05,
+      "loss": 0.9748,
+      "mean_token_accuracy": 0.702563327550888,
+      "num_tokens": 237413924.0,
+      "step": 96320
+    },
+    {
+      "epoch": 0.8831942789034565,
+      "learning_rate": 2.336297790409829e-05,
+      "loss": 0.9521,
+      "mean_token_accuracy": 0.7084671258926392,
+      "num_tokens": 237441379.0,
+      "step": 96330
+    },
+    {
+      "epoch": 0.8832859631429357,
+      "learning_rate": 2.334464105620244e-05,
+      "loss": 0.9728,
+      "mean_token_accuracy": 0.7088446319103241,
+      "num_tokens": 237469444.0,
+      "step": 96340
+    },
+    {
+      "epoch": 0.883377647382415,
+      "learning_rate": 2.3326304208306593e-05,
+      "loss": 0.9826,
+      "mean_token_accuracy": 0.7064197480678558,
+      "num_tokens": 237498444.0,
+      "step": 96350
+    },
+    {
+      "epoch": 0.8834693316218942,
+      "learning_rate": 2.3307967360410745e-05,
+      "loss": 0.9819,
+      "mean_token_accuracy": 0.7058658063411712,
+      "num_tokens": 237526235.0,
+      "step": 96360
+    },
+    {
+      "epoch": 0.8835610158613735,
+      "learning_rate": 2.32896305125149e-05,
+      "loss": 0.9749,
+      "mean_token_accuracy": 0.7051710247993469,
+      "num_tokens": 237554660.0,
+      "step": 96370
+    },
+    {
+      "epoch": 0.8836527001008526,
+      "learning_rate": 2.3271293664619055e-05,
+      "loss": 0.9958,
+      "mean_token_accuracy": 0.7004640519618988,
+      "num_tokens": 237583348.0,
+      "step": 96380
+    },
+    {
+      "epoch": 0.8837443843403319,
+      "learning_rate": 2.3252956816723207e-05,
+      "loss": 0.9699,
+      "mean_token_accuracy": 0.7076722145080566,
+      "num_tokens": 237611774.0,
+      "step": 96390
+    },
+    {
+      "epoch": 0.8838360685798111,
+      "learning_rate": 2.323461996882736e-05,
+      "loss": 0.9714,
+      "mean_token_accuracy": 0.7071040511131287,
+      "num_tokens": 237640235.0,
+      "step": 96400
+    },
+    {
+      "epoch": 0.8839277528192904,
+      "learning_rate": 2.321628312093151e-05,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7125555276870728,
+      "num_tokens": 237669710.0,
+      "step": 96410
+    },
+    {
+      "epoch": 0.8840194370587696,
+      "learning_rate": 2.3197946273035666e-05,
+      "loss": 0.9396,
+      "mean_token_accuracy": 0.7129753589630127,
+      "num_tokens": 237697698.0,
+      "step": 96420
+    },
+    {
+      "epoch": 0.8841111212982489,
+      "learning_rate": 2.317960942513982e-05,
+      "loss": 1.0042,
+      "mean_token_accuracy": 0.7032662093639374,
+      "num_tokens": 237726235.0,
+      "step": 96430
+    },
+    {
+      "epoch": 0.884202805537728,
+      "learning_rate": 2.3161272577243973e-05,
+      "loss": 0.9517,
+      "mean_token_accuracy": 0.7114958107471466,
+      "num_tokens": 237754077.0,
+      "step": 96440
+    },
+    {
+      "epoch": 0.8842944897772073,
+      "learning_rate": 2.3142935729348125e-05,
+      "loss": 0.9631,
+      "mean_token_accuracy": 0.7093418061733245,
+      "num_tokens": 237782454.0,
+      "step": 96450
+    },
+    {
+      "epoch": 0.8843861740166865,
+      "learning_rate": 2.312459888145228e-05,
+      "loss": 0.9837,
+      "mean_token_accuracy": 0.7035586893558502,
+      "num_tokens": 237810900.0,
+      "step": 96460
+    },
+    {
+      "epoch": 0.8844778582561658,
+      "learning_rate": 2.3106262033556432e-05,
+      "loss": 0.9725,
+      "mean_token_accuracy": 0.7076469600200653,
+      "num_tokens": 237839366.0,
+      "step": 96470
+    },
+    {
+      "epoch": 0.884569542495645,
+      "learning_rate": 2.3087925185660587e-05,
+      "loss": 0.9594,
+      "mean_token_accuracy": 0.7062601745128632,
+      "num_tokens": 237867267.0,
+      "step": 96480
+    },
+    {
+      "epoch": 0.8846612267351243,
+      "learning_rate": 2.306958833776474e-05,
+      "loss": 0.9869,
+      "mean_token_accuracy": 0.704833859205246,
+      "num_tokens": 237896233.0,
+      "step": 96490
+    },
+    {
+      "epoch": 0.8847529109746035,
+      "learning_rate": 2.305125148986889e-05,
+      "loss": 0.9797,
+      "mean_token_accuracy": 0.7050725817680359,
+      "num_tokens": 237924634.0,
+      "step": 96500
+    },
+    {
+      "epoch": 0.8848445952140827,
+      "learning_rate": 2.3032914641973046e-05,
+      "loss": 0.9679,
+      "mean_token_accuracy": 0.7066472947597504,
+      "num_tokens": 237953567.0,
+      "step": 96510
+    },
+    {
+      "epoch": 0.8849362794535619,
+      "learning_rate": 2.3014577794077198e-05,
+      "loss": 0.9672,
+      "mean_token_accuracy": 0.7009207725524902,
+      "num_tokens": 237981852.0,
+      "step": 96520
+    },
+    {
+      "epoch": 0.8850279636930412,
+      "learning_rate": 2.2996240946181353e-05,
+      "loss": 0.9553,
+      "mean_token_accuracy": 0.7090889692306519,
+      "num_tokens": 238009713.0,
+      "step": 96530
+    },
+    {
+      "epoch": 0.8851196479325204,
+      "learning_rate": 2.2977904098285505e-05,
+      "loss": 1.0145,
+      "mean_token_accuracy": 0.6924037039279938,
+      "num_tokens": 238037679.0,
+      "step": 96540
+    },
+    {
+      "epoch": 0.8852113321719997,
+      "learning_rate": 2.295956725038966e-05,
+      "loss": 0.9701,
+      "mean_token_accuracy": 0.7095661163330078,
+      "num_tokens": 238065350.0,
+      "step": 96550
+    },
+    {
+      "epoch": 0.8853030164114789,
+      "learning_rate": 2.2941230402493812e-05,
+      "loss": 0.9667,
+      "mean_token_accuracy": 0.7087648987770081,
+      "num_tokens": 238094915.0,
+      "step": 96560
+    },
+    {
+      "epoch": 0.8853947006509582,
+      "learning_rate": 2.2922893554597964e-05,
+      "loss": 0.9943,
+      "mean_token_accuracy": 0.7007356286048889,
+      "num_tokens": 238123245.0,
+      "step": 96570
+    },
+    {
+      "epoch": 0.8854863848904373,
+      "learning_rate": 2.290455670670212e-05,
+      "loss": 0.9883,
+      "mean_token_accuracy": 0.707362300157547,
+      "num_tokens": 238151699.0,
+      "step": 96580
+    },
+    {
+      "epoch": 0.8855780691299165,
+      "learning_rate": 2.288621985880627e-05,
+      "loss": 0.9578,
+      "mean_token_accuracy": 0.7084274351596832,
+      "num_tokens": 238179378.0,
+      "step": 96590
+    },
+    {
+      "epoch": 0.8856697533693958,
+      "learning_rate": 2.2867883010910427e-05,
+      "loss": 0.9346,
+      "mean_token_accuracy": 0.7091085612773895,
+      "num_tokens": 238206805.0,
+      "step": 96600
+    },
+    {
+      "epoch": 0.885761437608875,
+      "learning_rate": 2.284954616301458e-05,
+      "loss": 0.9826,
+      "mean_token_accuracy": 0.7047915995121002,
+      "num_tokens": 238235084.0,
+      "step": 96610
+    },
+    {
+      "epoch": 0.8858531218483543,
+      "learning_rate": 2.283120931511873e-05,
+      "loss": 0.9923,
+      "mean_token_accuracy": 0.7035689234733582,
+      "num_tokens": 238263579.0,
+      "step": 96620
+    },
+    {
+      "epoch": 0.8859448060878335,
+      "learning_rate": 2.2812872467222886e-05,
+      "loss": 0.97,
+      "mean_token_accuracy": 0.7095637738704681,
+      "num_tokens": 238291103.0,
+      "step": 96630
+    },
+    {
+      "epoch": 0.8860364903273127,
+      "learning_rate": 2.279453561932704e-05,
+      "loss": 0.9359,
+      "mean_token_accuracy": 0.7145771861076355,
+      "num_tokens": 238318154.0,
+      "step": 96640
+    },
+    {
+      "epoch": 0.8861281745667919,
+      "learning_rate": 2.2776198771431193e-05,
+      "loss": 0.9982,
+      "mean_token_accuracy": 0.6963150262832641,
+      "num_tokens": 238345981.0,
+      "step": 96650
+    },
+    {
+      "epoch": 0.8862198588062712,
+      "learning_rate": 2.2757861923535345e-05,
+      "loss": 0.924,
+      "mean_token_accuracy": 0.7200303912162781,
+      "num_tokens": 238374039.0,
+      "step": 96660
+    },
+    {
+      "epoch": 0.8863115430457504,
+      "learning_rate": 2.2739525075639496e-05,
+      "loss": 0.9494,
+      "mean_token_accuracy": 0.7126326620578766,
+      "num_tokens": 238401339.0,
+      "step": 96670
+    },
+    {
+      "epoch": 0.8864032272852297,
+      "learning_rate": 2.2721188227743652e-05,
+      "loss": 0.9689,
+      "mean_token_accuracy": 0.7026102662086486,
+      "num_tokens": 238429213.0,
+      "step": 96680
+    },
+    {
+      "epoch": 0.8864949115247089,
+      "learning_rate": 2.2702851379847807e-05,
+      "loss": 0.9732,
+      "mean_token_accuracy": 0.7079392790794372,
+      "num_tokens": 238458508.0,
+      "step": 96690
+    },
+    {
+      "epoch": 0.8865865957641882,
+      "learning_rate": 2.268451453195196e-05,
+      "loss": 0.9415,
+      "mean_token_accuracy": 0.7129162967205047,
+      "num_tokens": 238486420.0,
+      "step": 96700
+    },
+    {
+      "epoch": 0.8866782800036673,
+      "learning_rate": 2.266617768405611e-05,
+      "loss": 0.9727,
+      "mean_token_accuracy": 0.7034138381481171,
+      "num_tokens": 238514631.0,
+      "step": 96710
+    },
+    {
+      "epoch": 0.8867699642431466,
+      "learning_rate": 2.2647840836160263e-05,
+      "loss": 0.9675,
+      "mean_token_accuracy": 0.7059114933013916,
+      "num_tokens": 238543042.0,
+      "step": 96720
+    },
+    {
+      "epoch": 0.8868616484826258,
+      "learning_rate": 2.2629503988264418e-05,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.7054970860481262,
+      "num_tokens": 238571414.0,
+      "step": 96730
+    },
+    {
+      "epoch": 0.8869533327221051,
+      "learning_rate": 2.2611167140368573e-05,
+      "loss": 0.9637,
+      "mean_token_accuracy": 0.7084570229053497,
+      "num_tokens": 238599175.0,
+      "step": 96740
+    },
+    {
+      "epoch": 0.8870450169615843,
+      "learning_rate": 2.2592830292472725e-05,
+      "loss": 0.9901,
+      "mean_token_accuracy": 0.7020326614379883,
+      "num_tokens": 238626897.0,
+      "step": 96750
+    },
+    {
+      "epoch": 0.8871367012010636,
+      "learning_rate": 2.2574493444576877e-05,
+      "loss": 0.9427,
+      "mean_token_accuracy": 0.7104795098304748,
+      "num_tokens": 238655498.0,
+      "step": 96760
+    },
+    {
+      "epoch": 0.8872283854405427,
+      "learning_rate": 2.2556156596681032e-05,
+      "loss": 0.9625,
+      "mean_token_accuracy": 0.710882818698883,
+      "num_tokens": 238684165.0,
+      "step": 96770
+    },
+    {
+      "epoch": 0.887320069680022,
+      "learning_rate": 2.2537819748785184e-05,
+      "loss": 0.9437,
+      "mean_token_accuracy": 0.7077939689159394,
+      "num_tokens": 238712287.0,
+      "step": 96780
+    },
+    {
+      "epoch": 0.8874117539195012,
+      "learning_rate": 2.251948290088934e-05,
+      "loss": 0.9751,
+      "mean_token_accuracy": 0.7079537868499756,
+      "num_tokens": 238741552.0,
+      "step": 96790
+    },
+    {
+      "epoch": 0.8875034381589805,
+      "learning_rate": 2.250114605299349e-05,
+      "loss": 0.9492,
+      "mean_token_accuracy": 0.7128849506378174,
+      "num_tokens": 238769951.0,
+      "step": 96800
+    },
+    {
+      "epoch": 0.8875951223984597,
+      "learning_rate": 2.2482809205097643e-05,
+      "loss": 0.9676,
+      "mean_token_accuracy": 0.7078661918640137,
+      "num_tokens": 238798496.0,
+      "step": 96810
+    },
+    {
+      "epoch": 0.887686806637939,
+      "learning_rate": 2.2464472357201798e-05,
+      "loss": 0.9743,
+      "mean_token_accuracy": 0.705557507276535,
+      "num_tokens": 238827458.0,
+      "step": 96820
+    },
+    {
+      "epoch": 0.8877784908774182,
+      "learning_rate": 2.244613550930595e-05,
+      "loss": 0.9806,
+      "mean_token_accuracy": 0.7007512450218201,
+      "num_tokens": 238856477.0,
+      "step": 96830
+    },
+    {
+      "epoch": 0.8878701751168974,
+      "learning_rate": 2.2427798661410105e-05,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.7021511018276214,
+      "num_tokens": 238884243.0,
+      "step": 96840
+    },
+    {
+      "epoch": 0.8879618593563766,
+      "learning_rate": 2.2409461813514257e-05,
+      "loss": 0.9804,
+      "mean_token_accuracy": 0.703189742565155,
+      "num_tokens": 238912612.0,
+      "step": 96850
+    },
+    {
+      "epoch": 0.8880535435958559,
+      "learning_rate": 2.2391124965618412e-05,
+      "loss": 0.9405,
+      "mean_token_accuracy": 0.7142318785190582,
+      "num_tokens": 238940807.0,
+      "step": 96860
+    },
+    {
+      "epoch": 0.8881452278353351,
+      "learning_rate": 2.2372788117722564e-05,
+      "loss": 0.9313,
+      "mean_token_accuracy": 0.7137115716934204,
+      "num_tokens": 238968612.0,
+      "step": 96870
+    },
+    {
+      "epoch": 0.8882369120748144,
+      "learning_rate": 2.2354451269826716e-05,
+      "loss": 0.9161,
+      "mean_token_accuracy": 0.7169967651367187,
+      "num_tokens": 238996205.0,
+      "step": 96880
+    },
+    {
+      "epoch": 0.8883285963142936,
+      "learning_rate": 2.233611442193087e-05,
+      "loss": 0.9827,
+      "mean_token_accuracy": 0.7005254030227661,
+      "num_tokens": 239024865.0,
+      "step": 96890
+    },
+    {
+      "epoch": 0.8884202805537728,
+      "learning_rate": 2.2317777574035023e-05,
+      "loss": 0.9743,
+      "mean_token_accuracy": 0.7017959892749787,
+      "num_tokens": 239053388.0,
+      "step": 96900
+    },
+    {
+      "epoch": 0.888511964793252,
+      "learning_rate": 2.229944072613918e-05,
+      "loss": 0.9964,
+      "mean_token_accuracy": 0.7043148577213287,
+      "num_tokens": 239082342.0,
+      "step": 96910
+    },
+    {
+      "epoch": 0.8886036490327313,
+      "learning_rate": 2.228110387824333e-05,
+      "loss": 0.9737,
+      "mean_token_accuracy": 0.702905124425888,
+      "num_tokens": 239111159.0,
+      "step": 96920
+    },
+    {
+      "epoch": 0.8886953332722105,
+      "learning_rate": 2.2262767030347482e-05,
+      "loss": 0.9568,
+      "mean_token_accuracy": 0.706221753358841,
+      "num_tokens": 239140197.0,
+      "step": 96930
+    },
+    {
+      "epoch": 0.8887870175116898,
+      "learning_rate": 2.2244430182451638e-05,
+      "loss": 0.9881,
+      "mean_token_accuracy": 0.7017104625701904,
+      "num_tokens": 239168682.0,
+      "step": 96940
+    },
+    {
+      "epoch": 0.888878701751169,
+      "learning_rate": 2.2226093334555793e-05,
+      "loss": 0.9631,
+      "mean_token_accuracy": 0.7079795181751252,
+      "num_tokens": 239197190.0,
+      "step": 96950
+    },
+    {
+      "epoch": 0.8889703859906483,
+      "learning_rate": 2.2207756486659945e-05,
+      "loss": 0.9418,
+      "mean_token_accuracy": 0.7149965763092041,
+      "num_tokens": 239224099.0,
+      "step": 96960
+    },
+    {
+      "epoch": 0.8890620702301274,
+      "learning_rate": 2.2189419638764097e-05,
+      "loss": 0.9304,
+      "mean_token_accuracy": 0.7135801434516906,
+      "num_tokens": 239252220.0,
+      "step": 96970
+    },
+    {
+      "epoch": 0.8891537544696066,
+      "learning_rate": 2.217108279086825e-05,
+      "loss": 0.9858,
+      "mean_token_accuracy": 0.7067170798778534,
+      "num_tokens": 239280284.0,
+      "step": 96980
+    },
+    {
+      "epoch": 0.8892454387090859,
+      "learning_rate": 2.2152745942972407e-05,
+      "loss": 0.9569,
+      "mean_token_accuracy": 0.709110152721405,
+      "num_tokens": 239307859.0,
+      "step": 96990
+    },
+    {
+      "epoch": 0.8893371229485652,
+      "learning_rate": 2.213440909507656e-05,
+      "loss": 0.9806,
+      "mean_token_accuracy": 0.7043768763542175,
+      "num_tokens": 239336514.0,
+      "step": 97000
+    },
+    {
+      "epoch": 0.8894288071880444,
+      "learning_rate": 2.211607224718071e-05,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.7031792461872101,
+      "num_tokens": 239365349.0,
+      "step": 97010
+    },
+    {
+      "epoch": 0.8895204914275237,
+      "learning_rate": 2.2097735399284863e-05,
+      "loss": 0.9405,
+      "mean_token_accuracy": 0.7163421332836151,
+      "num_tokens": 239393319.0,
+      "step": 97020
+    },
+    {
+      "epoch": 0.8896121756670028,
+      "learning_rate": 2.2079398551389015e-05,
+      "loss": 0.9584,
+      "mean_token_accuracy": 0.7097893059253693,
+      "num_tokens": 239420833.0,
+      "step": 97030
+    },
+    {
+      "epoch": 0.889703859906482,
+      "learning_rate": 2.2061061703493173e-05,
+      "loss": 0.963,
+      "mean_token_accuracy": 0.7110153555870056,
+      "num_tokens": 239448339.0,
+      "step": 97040
+    },
+    {
+      "epoch": 0.8897955441459613,
+      "learning_rate": 2.2042724855597325e-05,
+      "loss": 0.9649,
+      "mean_token_accuracy": 0.7093183338642121,
+      "num_tokens": 239476061.0,
+      "step": 97050
+    },
+    {
+      "epoch": 0.8898872283854405,
+      "learning_rate": 2.2024388007701477e-05,
+      "loss": 0.9976,
+      "mean_token_accuracy": 0.7013116657733918,
+      "num_tokens": 239503886.0,
+      "step": 97060
+    },
+    {
+      "epoch": 0.8899789126249198,
+      "learning_rate": 2.200605115980563e-05,
+      "loss": 0.9559,
+      "mean_token_accuracy": 0.7107174158096313,
+      "num_tokens": 239531527.0,
+      "step": 97070
+    },
+    {
+      "epoch": 0.890070596864399,
+      "learning_rate": 2.1987714311909784e-05,
+      "loss": 0.9431,
+      "mean_token_accuracy": 0.713870745897293,
+      "num_tokens": 239558029.0,
+      "step": 97080
+    },
+    {
+      "epoch": 0.8901622811038783,
+      "learning_rate": 2.196937746401394e-05,
+      "loss": 0.9742,
+      "mean_token_accuracy": 0.7022699773311615,
+      "num_tokens": 239586449.0,
+      "step": 97090
+    },
+    {
+      "epoch": 0.8902539653433574,
+      "learning_rate": 2.195104061611809e-05,
+      "loss": 0.9516,
+      "mean_token_accuracy": 0.7131372094154358,
+      "num_tokens": 239614337.0,
+      "step": 97100
+    },
+    {
+      "epoch": 0.8903456495828367,
+      "learning_rate": 2.1932703768222243e-05,
+      "loss": 0.9785,
+      "mean_token_accuracy": 0.7015217185020447,
+      "num_tokens": 239640820.0,
+      "step": 97110
+    },
+    {
+      "epoch": 0.8904373338223159,
+      "learning_rate": 2.1914366920326395e-05,
+      "loss": 0.9467,
+      "mean_token_accuracy": 0.7167421400547027,
+      "num_tokens": 239669059.0,
+      "step": 97120
+    },
+    {
+      "epoch": 0.8905290180617952,
+      "learning_rate": 2.189603007243055e-05,
+      "loss": 0.9918,
+      "mean_token_accuracy": 0.7034554660320282,
+      "num_tokens": 239697143.0,
+      "step": 97130
+    },
+    {
+      "epoch": 0.8906207023012744,
+      "learning_rate": 2.1877693224534705e-05,
+      "loss": 0.9442,
+      "mean_token_accuracy": 0.7145998239517212,
+      "num_tokens": 239725341.0,
+      "step": 97140
+    },
+    {
+      "epoch": 0.8907123865407537,
+      "learning_rate": 2.1859356376638857e-05,
+      "loss": 0.9939,
+      "mean_token_accuracy": 0.7020884215831756,
+      "num_tokens": 239753977.0,
+      "step": 97150
+    },
+    {
+      "epoch": 0.8908040707802328,
+      "learning_rate": 2.184101952874301e-05,
+      "loss": 0.934,
+      "mean_token_accuracy": 0.718560802936554,
+      "num_tokens": 239780901.0,
+      "step": 97160
+    },
+    {
+      "epoch": 0.8908957550197121,
+      "learning_rate": 2.1822682680847164e-05,
+      "loss": 0.9922,
+      "mean_token_accuracy": 0.7035746157169342,
+      "num_tokens": 239808979.0,
+      "step": 97170
+    },
+    {
+      "epoch": 0.8909874392591913,
+      "learning_rate": 2.1804345832951316e-05,
+      "loss": 0.9862,
+      "mean_token_accuracy": 0.702910327911377,
+      "num_tokens": 239837834.0,
+      "step": 97180
+    },
+    {
+      "epoch": 0.8910791234986706,
+      "learning_rate": 2.178600898505547e-05,
+      "loss": 0.9433,
+      "mean_token_accuracy": 0.7108165144920349,
+      "num_tokens": 239865397.0,
+      "step": 97190
+    },
+    {
+      "epoch": 0.8911708077381498,
+      "learning_rate": 2.1767672137159623e-05,
+      "loss": 0.9445,
+      "mean_token_accuracy": 0.7101834297180176,
+      "num_tokens": 239893456.0,
+      "step": 97200
+    },
+    {
+      "epoch": 0.8912624919776291,
+      "learning_rate": 2.1749335289263775e-05,
+      "loss": 0.9576,
+      "mean_token_accuracy": 0.7073729932308197,
+      "num_tokens": 239921202.0,
+      "step": 97210
+    },
+    {
+      "epoch": 0.8913541762171083,
+      "learning_rate": 2.173099844136793e-05,
+      "loss": 0.9374,
+      "mean_token_accuracy": 0.7145473480224609,
+      "num_tokens": 239949475.0,
+      "step": 97220
+    },
+    {
+      "epoch": 0.8914458604565875,
+      "learning_rate": 2.1712661593472082e-05,
+      "loss": 0.9846,
+      "mean_token_accuracy": 0.7072529911994934,
+      "num_tokens": 239977541.0,
+      "step": 97230
+    },
+    {
+      "epoch": 0.8915375446960667,
+      "learning_rate": 2.1694324745576238e-05,
+      "loss": 0.9614,
+      "mean_token_accuracy": 0.70999316573143,
+      "num_tokens": 240004967.0,
+      "step": 97240
+    },
+    {
+      "epoch": 0.891629228935546,
+      "learning_rate": 2.167598789768039e-05,
+      "loss": 0.9391,
+      "mean_token_accuracy": 0.7105208516120911,
+      "num_tokens": 240033137.0,
+      "step": 97250
+    },
+    {
+      "epoch": 0.8917209131750252,
+      "learning_rate": 2.1657651049784545e-05,
+      "loss": 0.9743,
+      "mean_token_accuracy": 0.7062634706497193,
+      "num_tokens": 240060370.0,
+      "step": 97260
+    },
+    {
+      "epoch": 0.8918125974145045,
+      "learning_rate": 2.1639314201888697e-05,
+      "loss": 0.9827,
+      "mean_token_accuracy": 0.7027591407299042,
+      "num_tokens": 240088910.0,
+      "step": 97270
+    },
+    {
+      "epoch": 0.8919042816539837,
+      "learning_rate": 2.162097735399285e-05,
+      "loss": 0.951,
+      "mean_token_accuracy": 0.7139334440231323,
+      "num_tokens": 240116855.0,
+      "step": 97280
+    },
+    {
+      "epoch": 0.8919959658934629,
+      "learning_rate": 2.1602640506097004e-05,
+      "loss": 0.9578,
+      "mean_token_accuracy": 0.711462140083313,
+      "num_tokens": 240144392.0,
+      "step": 97290
+    },
+    {
+      "epoch": 0.8920876501329421,
+      "learning_rate": 2.158430365820116e-05,
+      "loss": 0.9811,
+      "mean_token_accuracy": 0.7059013426303864,
+      "num_tokens": 240171591.0,
+      "step": 97300
+    },
+    {
+      "epoch": 0.8921793343724214,
+      "learning_rate": 2.156596681030531e-05,
+      "loss": 0.9846,
+      "mean_token_accuracy": 0.70493643283844,
+      "num_tokens": 240199999.0,
+      "step": 97310
+    },
+    {
+      "epoch": 0.8922710186119006,
+      "learning_rate": 2.1547629962409463e-05,
+      "loss": 0.9725,
+      "mean_token_accuracy": 0.7052574157714844,
+      "num_tokens": 240228436.0,
+      "step": 97320
+    },
+    {
+      "epoch": 0.8923627028513799,
+      "learning_rate": 2.1529293114513615e-05,
+      "loss": 0.9705,
+      "mean_token_accuracy": 0.710043752193451,
+      "num_tokens": 240255741.0,
+      "step": 97330
+    },
+    {
+      "epoch": 0.8924543870908591,
+      "learning_rate": 2.151095626661777e-05,
+      "loss": 0.9806,
+      "mean_token_accuracy": 0.7028496623039245,
+      "num_tokens": 240284504.0,
+      "step": 97340
+    },
+    {
+      "epoch": 0.8925460713303384,
+      "learning_rate": 2.1492619418721925e-05,
+      "loss": 0.9851,
+      "mean_token_accuracy": 0.7029170513153076,
+      "num_tokens": 240313451.0,
+      "step": 97350
+    },
+    {
+      "epoch": 0.8926377555698175,
+      "learning_rate": 2.1474282570826077e-05,
+      "loss": 0.9343,
+      "mean_token_accuracy": 0.71030895113945,
+      "num_tokens": 240341868.0,
+      "step": 97360
+    },
+    {
+      "epoch": 0.8927294398092968,
+      "learning_rate": 2.145594572293023e-05,
+      "loss": 0.9954,
+      "mean_token_accuracy": 0.7004880666732788,
+      "num_tokens": 240369515.0,
+      "step": 97370
+    },
+    {
+      "epoch": 0.892821124048776,
+      "learning_rate": 2.143760887503438e-05,
+      "loss": 0.9662,
+      "mean_token_accuracy": 0.7119122266769409,
+      "num_tokens": 240398227.0,
+      "step": 97380
+    },
+    {
+      "epoch": 0.8929128082882553,
+      "learning_rate": 2.1419272027138536e-05,
+      "loss": 0.9702,
+      "mean_token_accuracy": 0.7040501415729523,
+      "num_tokens": 240426108.0,
+      "step": 97390
+    },
+    {
+      "epoch": 0.8930044925277345,
+      "learning_rate": 2.140093517924269e-05,
+      "loss": 0.9702,
+      "mean_token_accuracy": 0.7058905005455017,
+      "num_tokens": 240454373.0,
+      "step": 97400
+    },
+    {
+      "epoch": 0.8930961767672138,
+      "learning_rate": 2.1382598331346843e-05,
+      "loss": 0.9967,
+      "mean_token_accuracy": 0.6971623063087463,
+      "num_tokens": 240483144.0,
+      "step": 97410
+    },
+    {
+      "epoch": 0.8931878610066929,
+      "learning_rate": 2.1364261483450995e-05,
+      "loss": 0.9838,
+      "mean_token_accuracy": 0.7018255293369293,
+      "num_tokens": 240510562.0,
+      "step": 97420
+    },
+    {
+      "epoch": 0.8932795452461721,
+      "learning_rate": 2.1345924635555147e-05,
+      "loss": 0.9558,
+      "mean_token_accuracy": 0.7107691287994384,
+      "num_tokens": 240539387.0,
+      "step": 97430
+    },
+    {
+      "epoch": 0.8933712294856514,
+      "learning_rate": 2.1327587787659302e-05,
+      "loss": 0.9911,
+      "mean_token_accuracy": 0.7008166253566742,
+      "num_tokens": 240566876.0,
+      "step": 97440
+    },
+    {
+      "epoch": 0.8934629137251306,
+      "learning_rate": 2.1309250939763457e-05,
+      "loss": 0.9847,
+      "mean_token_accuracy": 0.7042464196681977,
+      "num_tokens": 240594001.0,
+      "step": 97450
+    },
+    {
+      "epoch": 0.8935545979646099,
+      "learning_rate": 2.129091409186761e-05,
+      "loss": 0.9485,
+      "mean_token_accuracy": 0.7077670335769654,
+      "num_tokens": 240621941.0,
+      "step": 97460
+    },
+    {
+      "epoch": 0.8936462822040891,
+      "learning_rate": 2.127257724397176e-05,
+      "loss": 0.9822,
+      "mean_token_accuracy": 0.7049382746219635,
+      "num_tokens": 240649252.0,
+      "step": 97470
+    },
+    {
+      "epoch": 0.8937379664435684,
+      "learning_rate": 2.1254240396075916e-05,
+      "loss": 0.9271,
+      "mean_token_accuracy": 0.7190112054347992,
+      "num_tokens": 240676809.0,
+      "step": 97480
+    },
+    {
+      "epoch": 0.8938296506830475,
+      "learning_rate": 2.1235903548180068e-05,
+      "loss": 0.9588,
+      "mean_token_accuracy": 0.7135937631130218,
+      "num_tokens": 240705868.0,
+      "step": 97490
+    },
+    {
+      "epoch": 0.8939213349225268,
+      "learning_rate": 2.1217566700284223e-05,
+      "loss": 0.976,
+      "mean_token_accuracy": 0.7038445949554444,
+      "num_tokens": 240733608.0,
+      "step": 97500
+    },
+    {
+      "epoch": 0.894013019162006,
+      "learning_rate": 2.1199229852388375e-05,
+      "loss": 0.9703,
+      "mean_token_accuracy": 0.7072181105613708,
+      "num_tokens": 240762169.0,
+      "step": 97510
+    },
+    {
+      "epoch": 0.8941047034014853,
+      "learning_rate": 2.1180893004492527e-05,
+      "loss": 0.9749,
+      "mean_token_accuracy": 0.7093082249164582,
+      "num_tokens": 240790284.0,
+      "step": 97520
+    },
+    {
+      "epoch": 0.8941963876409645,
+      "learning_rate": 2.1162556156596682e-05,
+      "loss": 0.9496,
+      "mean_token_accuracy": 0.7182524740695954,
+      "num_tokens": 240818279.0,
+      "step": 97530
+    },
+    {
+      "epoch": 0.8942880718804438,
+      "learning_rate": 2.1144219308700834e-05,
+      "loss": 0.9908,
+      "mean_token_accuracy": 0.7036314368247986,
+      "num_tokens": 240846334.0,
+      "step": 97540
+    },
+    {
+      "epoch": 0.8943797561199229,
+      "learning_rate": 2.112588246080499e-05,
+      "loss": 0.9266,
+      "mean_token_accuracy": 0.7167088270187378,
+      "num_tokens": 240874300.0,
+      "step": 97550
+    },
+    {
+      "epoch": 0.8944714403594022,
+      "learning_rate": 2.110754561290914e-05,
+      "loss": 0.9477,
+      "mean_token_accuracy": 0.7165841042995453,
+      "num_tokens": 240901484.0,
+      "step": 97560
+    },
+    {
+      "epoch": 0.8945631245988814,
+      "learning_rate": 2.1089208765013297e-05,
+      "loss": 0.9992,
+      "mean_token_accuracy": 0.704637223482132,
+      "num_tokens": 240930147.0,
+      "step": 97570
+    },
+    {
+      "epoch": 0.8946548088383607,
+      "learning_rate": 2.107087191711745e-05,
+      "loss": 0.9664,
+      "mean_token_accuracy": 0.7051290452480317,
+      "num_tokens": 240958434.0,
+      "step": 97580
+    },
+    {
+      "epoch": 0.8947464930778399,
+      "learning_rate": 2.10525350692216e-05,
+      "loss": 0.9558,
+      "mean_token_accuracy": 0.7086008548736572,
+      "num_tokens": 240986338.0,
+      "step": 97590
+    },
+    {
+      "epoch": 0.8948381773173192,
+      "learning_rate": 2.1034198221325756e-05,
+      "loss": 0.9603,
+      "mean_token_accuracy": 0.705905967950821,
+      "num_tokens": 241014936.0,
+      "step": 97600
+    },
+    {
+      "epoch": 0.8949298615567984,
+      "learning_rate": 2.1015861373429908e-05,
+      "loss": 0.9688,
+      "mean_token_accuracy": 0.7053589344024658,
+      "num_tokens": 241041784.0,
+      "step": 97610
+    },
+    {
+      "epoch": 0.8950215457962776,
+      "learning_rate": 2.0997524525534063e-05,
+      "loss": 0.9545,
+      "mean_token_accuracy": 0.707834267616272,
+      "num_tokens": 241069063.0,
+      "step": 97620
+    },
+    {
+      "epoch": 0.8951132300357568,
+      "learning_rate": 2.0979187677638215e-05,
+      "loss": 1.0178,
+      "mean_token_accuracy": 0.6941853821277618,
+      "num_tokens": 241097606.0,
+      "step": 97630
+    },
+    {
+      "epoch": 0.8952049142752361,
+      "learning_rate": 2.0960850829742367e-05,
+      "loss": 0.9389,
+      "mean_token_accuracy": 0.7142586946487427,
+      "num_tokens": 241125382.0,
+      "step": 97640
+    },
+    {
+      "epoch": 0.8952965985147153,
+      "learning_rate": 2.0942513981846522e-05,
+      "loss": 0.9715,
+      "mean_token_accuracy": 0.704793119430542,
+      "num_tokens": 241152699.0,
+      "step": 97650
+    },
+    {
+      "epoch": 0.8953882827541946,
+      "learning_rate": 2.0924177133950677e-05,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7088997960090637,
+      "num_tokens": 241181195.0,
+      "step": 97660
+    },
+    {
+      "epoch": 0.8954799669936738,
+      "learning_rate": 2.090584028605483e-05,
+      "loss": 0.9943,
+      "mean_token_accuracy": 0.7021056711673737,
+      "num_tokens": 241209766.0,
+      "step": 97670
+    },
+    {
+      "epoch": 0.895571651233153,
+      "learning_rate": 2.088750343815898e-05,
+      "loss": 0.9612,
+      "mean_token_accuracy": 0.7067963480949402,
+      "num_tokens": 241237038.0,
+      "step": 97680
+    },
+    {
+      "epoch": 0.8956633354726322,
+      "learning_rate": 2.0869166590263133e-05,
+      "loss": 0.9515,
+      "mean_token_accuracy": 0.7101170957088471,
+      "num_tokens": 241265491.0,
+      "step": 97690
+    },
+    {
+      "epoch": 0.8957550197121115,
+      "learning_rate": 2.0850829742367288e-05,
+      "loss": 0.9481,
+      "mean_token_accuracy": 0.7074155569076538,
+      "num_tokens": 241292131.0,
+      "step": 97700
+    },
+    {
+      "epoch": 0.8958467039515907,
+      "learning_rate": 2.0832492894471443e-05,
+      "loss": 0.9605,
+      "mean_token_accuracy": 0.7086700677871705,
+      "num_tokens": 241319995.0,
+      "step": 97710
+    },
+    {
+      "epoch": 0.89593838819107,
+      "learning_rate": 2.0814156046575595e-05,
+      "loss": 0.9693,
+      "mean_token_accuracy": 0.7046679139137269,
+      "num_tokens": 241347349.0,
+      "step": 97720
+    },
+    {
+      "epoch": 0.8960300724305492,
+      "learning_rate": 2.0795819198679747e-05,
+      "loss": 0.9346,
+      "mean_token_accuracy": 0.7158856809139251,
+      "num_tokens": 241374999.0,
+      "step": 97730
+    },
+    {
+      "epoch": 0.8961217566700285,
+      "learning_rate": 2.07774823507839e-05,
+      "loss": 0.9811,
+      "mean_token_accuracy": 0.7040074944496155,
+      "num_tokens": 241402937.0,
+      "step": 97740
+    },
+    {
+      "epoch": 0.8962134409095076,
+      "learning_rate": 2.0759145502888054e-05,
+      "loss": 0.961,
+      "mean_token_accuracy": 0.7127393186092377,
+      "num_tokens": 241431821.0,
+      "step": 97750
+    },
+    {
+      "epoch": 0.8963051251489869,
+      "learning_rate": 2.074080865499221e-05,
+      "loss": 0.9859,
+      "mean_token_accuracy": 0.702030336856842,
+      "num_tokens": 241460322.0,
+      "step": 97760
+    },
+    {
+      "epoch": 0.8963968093884661,
+      "learning_rate": 2.072247180709636e-05,
+      "loss": 0.9521,
+      "mean_token_accuracy": 0.7114170610904693,
+      "num_tokens": 241487720.0,
+      "step": 97770
+    },
+    {
+      "epoch": 0.8964884936279454,
+      "learning_rate": 2.0704134959200513e-05,
+      "loss": 0.9699,
+      "mean_token_accuracy": 0.7097540855407715,
+      "num_tokens": 241515097.0,
+      "step": 97780
+    },
+    {
+      "epoch": 0.8965801778674246,
+      "learning_rate": 2.0685798111304668e-05,
+      "loss": 0.9837,
+      "mean_token_accuracy": 0.7047342956066132,
+      "num_tokens": 241543472.0,
+      "step": 97790
+    },
+    {
+      "epoch": 0.8966718621069039,
+      "learning_rate": 2.066746126340882e-05,
+      "loss": 0.9605,
+      "mean_token_accuracy": 0.7147351503372192,
+      "num_tokens": 241571616.0,
+      "step": 97800
+    },
+    {
+      "epoch": 0.8967635463463831,
+      "learning_rate": 2.0649124415512975e-05,
+      "loss": 0.9441,
+      "mean_token_accuracy": 0.7077078938484191,
+      "num_tokens": 241600183.0,
+      "step": 97810
+    },
+    {
+      "epoch": 0.8968552305858623,
+      "learning_rate": 2.0630787567617127e-05,
+      "loss": 0.9806,
+      "mean_token_accuracy": 0.7040668785572052,
+      "num_tokens": 241627875.0,
+      "step": 97820
+    },
+    {
+      "epoch": 0.8969469148253415,
+      "learning_rate": 2.061245071972128e-05,
+      "loss": 0.9481,
+      "mean_token_accuracy": 0.7153401970863342,
+      "num_tokens": 241655914.0,
+      "step": 97830
+    },
+    {
+      "epoch": 0.8970385990648208,
+      "learning_rate": 2.0594113871825434e-05,
+      "loss": 0.9467,
+      "mean_token_accuracy": 0.7099112391471862,
+      "num_tokens": 241683316.0,
+      "step": 97840
+    },
+    {
+      "epoch": 0.8971302833043,
+      "learning_rate": 2.0575777023929586e-05,
+      "loss": 0.9543,
+      "mean_token_accuracy": 0.7093935608863831,
+      "num_tokens": 241712258.0,
+      "step": 97850
+    },
+    {
+      "epoch": 0.8972219675437793,
+      "learning_rate": 2.055744017603374e-05,
+      "loss": 0.9511,
+      "mean_token_accuracy": 0.7118758618831634,
+      "num_tokens": 241741411.0,
+      "step": 97860
+    },
+    {
+      "epoch": 0.8973136517832585,
+      "learning_rate": 2.0539103328137893e-05,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.7053055465221405,
+      "num_tokens": 241768167.0,
+      "step": 97870
+    },
+    {
+      "epoch": 0.8974053360227376,
+      "learning_rate": 2.052076648024205e-05,
+      "loss": 0.9862,
+      "mean_token_accuracy": 0.7043285131454468,
+      "num_tokens": 241795756.0,
+      "step": 97880
+    },
+    {
+      "epoch": 0.8974970202622169,
+      "learning_rate": 2.05024296323462e-05,
+      "loss": 0.9676,
+      "mean_token_accuracy": 0.7091193974018097,
+      "num_tokens": 241824214.0,
+      "step": 97890
+    },
+    {
+      "epoch": 0.8975887045016961,
+      "learning_rate": 2.0484092784450352e-05,
+      "loss": 0.9547,
+      "mean_token_accuracy": 0.7121727526187897,
+      "num_tokens": 241851507.0,
+      "step": 97900
+    },
+    {
+      "epoch": 0.8976803887411754,
+      "learning_rate": 2.0465755936554508e-05,
+      "loss": 0.9444,
+      "mean_token_accuracy": 0.7150976479053497,
+      "num_tokens": 241878894.0,
+      "step": 97910
+    },
+    {
+      "epoch": 0.8977720729806546,
+      "learning_rate": 2.044741908865866e-05,
+      "loss": 0.939,
+      "mean_token_accuracy": 0.7140742897987366,
+      "num_tokens": 241908163.0,
+      "step": 97920
+    },
+    {
+      "epoch": 0.8978637572201339,
+      "learning_rate": 2.0429082240762815e-05,
+      "loss": 0.9576,
+      "mean_token_accuracy": 0.7055696487426758,
+      "num_tokens": 241937421.0,
+      "step": 97930
+    },
+    {
+      "epoch": 0.8979554414596131,
+      "learning_rate": 2.0410745392866967e-05,
+      "loss": 0.9352,
+      "mean_token_accuracy": 0.7135234117507935,
+      "num_tokens": 241965751.0,
+      "step": 97940
+    },
+    {
+      "epoch": 0.8980471256990923,
+      "learning_rate": 2.039240854497112e-05,
+      "loss": 0.9555,
+      "mean_token_accuracy": 0.7085389316082,
+      "num_tokens": 241994856.0,
+      "step": 97950
+    },
+    {
+      "epoch": 0.8981388099385715,
+      "learning_rate": 2.0374071697075274e-05,
+      "loss": 0.9718,
+      "mean_token_accuracy": 0.7018843948841095,
+      "num_tokens": 242022793.0,
+      "step": 97960
+    },
+    {
+      "epoch": 0.8982304941780508,
+      "learning_rate": 2.035573484917943e-05,
+      "loss": 0.9844,
+      "mean_token_accuracy": 0.700729638338089,
+      "num_tokens": 242050774.0,
+      "step": 97970
+    },
+    {
+      "epoch": 0.89832217841753,
+      "learning_rate": 2.033739800128358e-05,
+      "loss": 0.9406,
+      "mean_token_accuracy": 0.7121575176715851,
+      "num_tokens": 242078689.0,
+      "step": 97980
+    },
+    {
+      "epoch": 0.8984138626570093,
+      "learning_rate": 2.0319061153387733e-05,
+      "loss": 0.9529,
+      "mean_token_accuracy": 0.716164642572403,
+      "num_tokens": 242106528.0,
+      "step": 97990
+    },
+    {
+      "epoch": 0.8985055468964885,
+      "learning_rate": 2.0300724305491885e-05,
+      "loss": 0.9923,
+      "mean_token_accuracy": 0.6991463661193847,
+      "num_tokens": 242134110.0,
+      "step": 98000
+    },
+    {
+      "epoch": 0.8985972311359677,
+      "learning_rate": 2.028238745759604e-05,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7144601285457611,
+      "num_tokens": 242161617.0,
+      "step": 98010
+    },
+    {
+      "epoch": 0.8986889153754469,
+      "learning_rate": 2.0264050609700195e-05,
+      "loss": 0.9762,
+      "mean_token_accuracy": 0.7069809973239899,
+      "num_tokens": 242190524.0,
+      "step": 98020
+    },
+    {
+      "epoch": 0.8987805996149262,
+      "learning_rate": 2.0245713761804347e-05,
+      "loss": 0.9535,
+      "mean_token_accuracy": 0.7110555589199066,
+      "num_tokens": 242218902.0,
+      "step": 98030
+    },
+    {
+      "epoch": 0.8988722838544054,
+      "learning_rate": 2.02273769139085e-05,
+      "loss": 0.9715,
+      "mean_token_accuracy": 0.7055911839008331,
+      "num_tokens": 242246071.0,
+      "step": 98040
+    },
+    {
+      "epoch": 0.8989639680938847,
+      "learning_rate": 2.020904006601265e-05,
+      "loss": 0.9596,
+      "mean_token_accuracy": 0.7030866265296936,
+      "num_tokens": 242274414.0,
+      "step": 98050
+    },
+    {
+      "epoch": 0.8990556523333639,
+      "learning_rate": 2.019070321811681e-05,
+      "loss": 0.9666,
+      "mean_token_accuracy": 0.7040348529815674,
+      "num_tokens": 242302623.0,
+      "step": 98060
+    },
+    {
+      "epoch": 0.8991473365728432,
+      "learning_rate": 2.017236637022096e-05,
+      "loss": 0.9532,
+      "mean_token_accuracy": 0.7138877093791962,
+      "num_tokens": 242331298.0,
+      "step": 98070
+    },
+    {
+      "epoch": 0.8992390208123223,
+      "learning_rate": 2.0154029522325113e-05,
+      "loss": 1.0112,
+      "mean_token_accuracy": 0.6944770693778992,
+      "num_tokens": 242359653.0,
+      "step": 98080
+    },
+    {
+      "epoch": 0.8993307050518016,
+      "learning_rate": 2.0135692674429265e-05,
+      "loss": 0.9643,
+      "mean_token_accuracy": 0.7098870873451233,
+      "num_tokens": 242388824.0,
+      "step": 98090
+    },
+    {
+      "epoch": 0.8994223892912808,
+      "learning_rate": 2.011735582653342e-05,
+      "loss": 0.9725,
+      "mean_token_accuracy": 0.7049516499042511,
+      "num_tokens": 242416601.0,
+      "step": 98100
+    },
+    {
+      "epoch": 0.8995140735307601,
+      "learning_rate": 2.0099018978637575e-05,
+      "loss": 0.9294,
+      "mean_token_accuracy": 0.7129597067832947,
+      "num_tokens": 242445464.0,
+      "step": 98110
+    },
+    {
+      "epoch": 0.8996057577702393,
+      "learning_rate": 2.0080682130741727e-05,
+      "loss": 0.9686,
+      "mean_token_accuracy": 0.7078495383262634,
+      "num_tokens": 242473337.0,
+      "step": 98120
+    },
+    {
+      "epoch": 0.8996974420097186,
+      "learning_rate": 2.006234528284588e-05,
+      "loss": 0.9496,
+      "mean_token_accuracy": 0.7084646999835968,
+      "num_tokens": 242502275.0,
+      "step": 98130
+    },
+    {
+      "epoch": 0.8997891262491977,
+      "learning_rate": 2.004400843495003e-05,
+      "loss": 1.0012,
+      "mean_token_accuracy": 0.703138554096222,
+      "num_tokens": 242530146.0,
+      "step": 98140
+    },
+    {
+      "epoch": 0.899880810488677,
+      "learning_rate": 2.0025671587054186e-05,
+      "loss": 0.982,
+      "mean_token_accuracy": 0.7043869853019714,
+      "num_tokens": 242557847.0,
+      "step": 98150
+    },
+    {
+      "epoch": 0.8999724947281562,
+      "learning_rate": 2.000733473915834e-05,
+      "loss": 0.972,
+      "mean_token_accuracy": 0.7044876873493194,
+      "num_tokens": 242586696.0,
+      "step": 98160
+    },
+    {
+      "epoch": 0.9000641789676355,
+      "learning_rate": 1.9988997891262493e-05,
+      "loss": 1.0118,
+      "mean_token_accuracy": 0.7073430299758912,
+      "num_tokens": 242614961.0,
+      "step": 98170
+    },
+    {
+      "epoch": 0.9001558632071147,
+      "learning_rate": 1.9970661043366645e-05,
+      "loss": 0.9528,
+      "mean_token_accuracy": 0.7149098515510559,
+      "num_tokens": 242643463.0,
+      "step": 98180
+    },
+    {
+      "epoch": 0.900247547446594,
+      "learning_rate": 1.99523241954708e-05,
+      "loss": 0.9449,
+      "mean_token_accuracy": 0.7118351042270661,
+      "num_tokens": 242672199.0,
+      "step": 98190
+    },
+    {
+      "epoch": 0.9003392316860732,
+      "learning_rate": 1.9933987347574952e-05,
+      "loss": 0.9504,
+      "mean_token_accuracy": 0.7106627702713013,
+      "num_tokens": 242700144.0,
+      "step": 98200
+    },
+    {
+      "epoch": 0.9004309159255524,
+      "learning_rate": 1.9915650499679108e-05,
+      "loss": 0.9515,
+      "mean_token_accuracy": 0.7086313486099243,
+      "num_tokens": 242728642.0,
+      "step": 98210
+    },
+    {
+      "epoch": 0.9005226001650316,
+      "learning_rate": 1.989731365178326e-05,
+      "loss": 0.9935,
+      "mean_token_accuracy": 0.7040349960327148,
+      "num_tokens": 242757097.0,
+      "step": 98220
+    },
+    {
+      "epoch": 0.9006142844045109,
+      "learning_rate": 1.987897680388741e-05,
+      "loss": 0.9826,
+      "mean_token_accuracy": 0.7031432092189789,
+      "num_tokens": 242785799.0,
+      "step": 98230
+    },
+    {
+      "epoch": 0.9007059686439901,
+      "learning_rate": 1.9860639955991567e-05,
+      "loss": 0.9377,
+      "mean_token_accuracy": 0.7090740144252777,
+      "num_tokens": 242813630.0,
+      "step": 98240
+    },
+    {
+      "epoch": 0.9007976528834694,
+      "learning_rate": 1.984230310809572e-05,
+      "loss": 0.9676,
+      "mean_token_accuracy": 0.7060315907001495,
+      "num_tokens": 242840394.0,
+      "step": 98250
+    },
+    {
+      "epoch": 0.9008893371229486,
+      "learning_rate": 1.9823966260199874e-05,
+      "loss": 0.973,
+      "mean_token_accuracy": 0.710243034362793,
+      "num_tokens": 242868857.0,
+      "step": 98260
+    },
+    {
+      "epoch": 0.9009810213624277,
+      "learning_rate": 1.9805629412304026e-05,
+      "loss": 0.9689,
+      "mean_token_accuracy": 0.7057923972606659,
+      "num_tokens": 242897116.0,
+      "step": 98270
+    },
+    {
+      "epoch": 0.901072705601907,
+      "learning_rate": 1.978729256440818e-05,
+      "loss": 0.9887,
+      "mean_token_accuracy": 0.7076674342155457,
+      "num_tokens": 242923929.0,
+      "step": 98280
+    },
+    {
+      "epoch": 0.9011643898413862,
+      "learning_rate": 1.9768955716512333e-05,
+      "loss": 0.929,
+      "mean_token_accuracy": 0.7210639715194702,
+      "num_tokens": 242951170.0,
+      "step": 98290
+    },
+    {
+      "epoch": 0.9012560740808655,
+      "learning_rate": 1.9750618868616485e-05,
+      "loss": 0.9481,
+      "mean_token_accuracy": 0.7090279400348664,
+      "num_tokens": 242978839.0,
+      "step": 98300
+    },
+    {
+      "epoch": 0.9013477583203447,
+      "learning_rate": 1.973228202072064e-05,
+      "loss": 0.947,
+      "mean_token_accuracy": 0.7095263004302979,
+      "num_tokens": 243006189.0,
+      "step": 98310
+    },
+    {
+      "epoch": 0.901439442559824,
+      "learning_rate": 1.9713945172824795e-05,
+      "loss": 0.9627,
+      "mean_token_accuracy": 0.7099762558937073,
+      "num_tokens": 243034368.0,
+      "step": 98320
+    },
+    {
+      "epoch": 0.9015311267993033,
+      "learning_rate": 1.9695608324928947e-05,
+      "loss": 0.9838,
+      "mean_token_accuracy": 0.7022603631019593,
+      "num_tokens": 243062477.0,
+      "step": 98330
+    },
+    {
+      "epoch": 0.9016228110387824,
+      "learning_rate": 1.96772714770331e-05,
+      "loss": 0.9514,
+      "mean_token_accuracy": 0.7108042061328887,
+      "num_tokens": 243090238.0,
+      "step": 98340
+    },
+    {
+      "epoch": 0.9017144952782616,
+      "learning_rate": 1.965893462913725e-05,
+      "loss": 0.9602,
+      "mean_token_accuracy": 0.708228987455368,
+      "num_tokens": 243118273.0,
+      "step": 98350
+    },
+    {
+      "epoch": 0.9018061795177409,
+      "learning_rate": 1.9640597781241406e-05,
+      "loss": 0.9663,
+      "mean_token_accuracy": 0.7075661361217499,
+      "num_tokens": 243146550.0,
+      "step": 98360
+    },
+    {
+      "epoch": 0.9018978637572201,
+      "learning_rate": 1.962226093334556e-05,
+      "loss": 0.9335,
+      "mean_token_accuracy": 0.7181289672851563,
+      "num_tokens": 243174904.0,
+      "step": 98370
+    },
+    {
+      "epoch": 0.9019895479966994,
+      "learning_rate": 1.9603924085449713e-05,
+      "loss": 0.9808,
+      "mean_token_accuracy": 0.7027455568313599,
+      "num_tokens": 243203644.0,
+      "step": 98380
+    },
+    {
+      "epoch": 0.9020812322361786,
+      "learning_rate": 1.9585587237553865e-05,
+      "loss": 0.9631,
+      "mean_token_accuracy": 0.7073743343353271,
+      "num_tokens": 243230956.0,
+      "step": 98390
+    },
+    {
+      "epoch": 0.9021729164756578,
+      "learning_rate": 1.9567250389658017e-05,
+      "loss": 0.9781,
+      "mean_token_accuracy": 0.7030749142169952,
+      "num_tokens": 243258979.0,
+      "step": 98400
+    },
+    {
+      "epoch": 0.902264600715137,
+      "learning_rate": 1.9548913541762172e-05,
+      "loss": 0.9357,
+      "mean_token_accuracy": 0.713792484998703,
+      "num_tokens": 243287241.0,
+      "step": 98410
+    },
+    {
+      "epoch": 0.9023562849546163,
+      "learning_rate": 1.9530576693866327e-05,
+      "loss": 0.944,
+      "mean_token_accuracy": 0.7122910678386688,
+      "num_tokens": 243315310.0,
+      "step": 98420
+    },
+    {
+      "epoch": 0.9024479691940955,
+      "learning_rate": 1.951223984597048e-05,
+      "loss": 0.9686,
+      "mean_token_accuracy": 0.7054218947887421,
+      "num_tokens": 243343271.0,
+      "step": 98430
+    },
+    {
+      "epoch": 0.9025396534335748,
+      "learning_rate": 1.949390299807463e-05,
+      "loss": 0.9403,
+      "mean_token_accuracy": 0.7059735238552094,
+      "num_tokens": 243371525.0,
+      "step": 98440
+    },
+    {
+      "epoch": 0.902631337673054,
+      "learning_rate": 1.9475566150178783e-05,
+      "loss": 0.9718,
+      "mean_token_accuracy": 0.70748171210289,
+      "num_tokens": 243399455.0,
+      "step": 98450
+    },
+    {
+      "epoch": 0.9027230219125333,
+      "learning_rate": 1.9457229302282938e-05,
+      "loss": 0.9467,
+      "mean_token_accuracy": 0.7119232892990113,
+      "num_tokens": 243426714.0,
+      "step": 98460
+    },
+    {
+      "epoch": 0.9028147061520124,
+      "learning_rate": 1.9438892454387094e-05,
+      "loss": 0.9364,
+      "mean_token_accuracy": 0.7129967391490937,
+      "num_tokens": 243455255.0,
+      "step": 98470
+    },
+    {
+      "epoch": 0.9029063903914917,
+      "learning_rate": 1.9420555606491245e-05,
+      "loss": 0.9252,
+      "mean_token_accuracy": 0.718877625465393,
+      "num_tokens": 243483096.0,
+      "step": 98480
+    },
+    {
+      "epoch": 0.9029980746309709,
+      "learning_rate": 1.9402218758595397e-05,
+      "loss": 0.9585,
+      "mean_token_accuracy": 0.7124249160289764,
+      "num_tokens": 243511074.0,
+      "step": 98490
+    },
+    {
+      "epoch": 0.9030897588704502,
+      "learning_rate": 1.9383881910699552e-05,
+      "loss": 0.9555,
+      "mean_token_accuracy": 0.7075623571872711,
+      "num_tokens": 243539196.0,
+      "step": 98500
+    },
+    {
+      "epoch": 0.9031814431099294,
+      "learning_rate": 1.9365545062803704e-05,
+      "loss": 0.9611,
+      "mean_token_accuracy": 0.7051093578338623,
+      "num_tokens": 243566995.0,
+      "step": 98510
+    },
+    {
+      "epoch": 0.9032731273494087,
+      "learning_rate": 1.934720821490786e-05,
+      "loss": 0.977,
+      "mean_token_accuracy": 0.706699115037918,
+      "num_tokens": 243594851.0,
+      "step": 98520
+    },
+    {
+      "epoch": 0.9033648115888878,
+      "learning_rate": 1.932887136701201e-05,
+      "loss": 0.9476,
+      "mean_token_accuracy": 0.7158394098281861,
+      "num_tokens": 243623453.0,
+      "step": 98530
+    },
+    {
+      "epoch": 0.9034564958283671,
+      "learning_rate": 1.9310534519116163e-05,
+      "loss": 0.9466,
+      "mean_token_accuracy": 0.7080992043018342,
+      "num_tokens": 243650900.0,
+      "step": 98540
+    },
+    {
+      "epoch": 0.9035481800678463,
+      "learning_rate": 1.929219767122032e-05,
+      "loss": 0.967,
+      "mean_token_accuracy": 0.7112585842609406,
+      "num_tokens": 243678387.0,
+      "step": 98550
+    },
+    {
+      "epoch": 0.9036398643073256,
+      "learning_rate": 1.927386082332447e-05,
+      "loss": 0.9418,
+      "mean_token_accuracy": 0.7078060448169708,
+      "num_tokens": 243706298.0,
+      "step": 98560
+    },
+    {
+      "epoch": 0.9037315485468048,
+      "learning_rate": 1.9255523975428626e-05,
+      "loss": 0.968,
+      "mean_token_accuracy": 0.7058828592300415,
+      "num_tokens": 243734181.0,
+      "step": 98570
+    },
+    {
+      "epoch": 0.9038232327862841,
+      "learning_rate": 1.9237187127532778e-05,
+      "loss": 0.9831,
+      "mean_token_accuracy": 0.7044422030448914,
+      "num_tokens": 243761708.0,
+      "step": 98580
+    },
+    {
+      "epoch": 0.9039149170257633,
+      "learning_rate": 1.9218850279636933e-05,
+      "loss": 0.948,
+      "mean_token_accuracy": 0.7123528182506561,
+      "num_tokens": 243789810.0,
+      "step": 98590
+    },
+    {
+      "epoch": 0.9040066012652425,
+      "learning_rate": 1.9200513431741085e-05,
+      "loss": 0.9378,
+      "mean_token_accuracy": 0.7146504878997803,
+      "num_tokens": 243817994.0,
+      "step": 98600
+    },
+    {
+      "epoch": 0.9040982855047217,
+      "learning_rate": 1.9182176583845237e-05,
+      "loss": 0.9505,
+      "mean_token_accuracy": 0.7088361203670501,
+      "num_tokens": 243845439.0,
+      "step": 98610
+    },
+    {
+      "epoch": 0.904189969744201,
+      "learning_rate": 1.9163839735949392e-05,
+      "loss": 0.9522,
+      "mean_token_accuracy": 0.7110304951667785,
+      "num_tokens": 243873228.0,
+      "step": 98620
+    },
+    {
+      "epoch": 0.9042816539836802,
+      "learning_rate": 1.9145502888053547e-05,
+      "loss": 0.9687,
+      "mean_token_accuracy": 0.7034121513366699,
+      "num_tokens": 243900947.0,
+      "step": 98630
+    },
+    {
+      "epoch": 0.9043733382231595,
+      "learning_rate": 1.91271660401577e-05,
+      "loss": 0.9489,
+      "mean_token_accuracy": 0.7118391335010529,
+      "num_tokens": 243929521.0,
+      "step": 98640
+    },
+    {
+      "epoch": 0.9044650224626387,
+      "learning_rate": 1.910882919226185e-05,
+      "loss": 0.9541,
+      "mean_token_accuracy": 0.7096317946910858,
+      "num_tokens": 243957669.0,
+      "step": 98650
+    },
+    {
+      "epoch": 0.9045567067021179,
+      "learning_rate": 1.9090492344366003e-05,
+      "loss": 0.9445,
+      "mean_token_accuracy": 0.7117459297180175,
+      "num_tokens": 243985484.0,
+      "step": 98660
+    },
+    {
+      "epoch": 0.9046483909415971,
+      "learning_rate": 1.9072155496470158e-05,
+      "loss": 0.9604,
+      "mean_token_accuracy": 0.7134973287582398,
+      "num_tokens": 244012831.0,
+      "step": 98670
+    },
+    {
+      "epoch": 0.9047400751810764,
+      "learning_rate": 1.9053818648574313e-05,
+      "loss": 0.9697,
+      "mean_token_accuracy": 0.7078468203544617,
+      "num_tokens": 244041318.0,
+      "step": 98680
+    },
+    {
+      "epoch": 0.9048317594205556,
+      "learning_rate": 1.9035481800678465e-05,
+      "loss": 0.9749,
+      "mean_token_accuracy": 0.6981931924819946,
+      "num_tokens": 244069948.0,
+      "step": 98690
+    },
+    {
+      "epoch": 0.9049234436600349,
+      "learning_rate": 1.9017144952782617e-05,
+      "loss": 0.9545,
+      "mean_token_accuracy": 0.7123080134391785,
+      "num_tokens": 244097819.0,
+      "step": 98700
+    },
+    {
+      "epoch": 0.9050151278995141,
+      "learning_rate": 1.899880810488677e-05,
+      "loss": 0.9694,
+      "mean_token_accuracy": 0.7063113808631897,
+      "num_tokens": 244125848.0,
+      "step": 98710
+    },
+    {
+      "epoch": 0.9051068121389934,
+      "learning_rate": 1.8980471256990924e-05,
+      "loss": 0.9544,
+      "mean_token_accuracy": 0.7099994361400604,
+      "num_tokens": 244154158.0,
+      "step": 98720
+    },
+    {
+      "epoch": 0.9051984963784725,
+      "learning_rate": 1.896213440909508e-05,
+      "loss": 0.9469,
+      "mean_token_accuracy": 0.7134117543697357,
+      "num_tokens": 244181368.0,
+      "step": 98730
+    },
+    {
+      "epoch": 0.9052901806179517,
+      "learning_rate": 1.894379756119923e-05,
+      "loss": 0.9588,
+      "mean_token_accuracy": 0.7087450921535492,
+      "num_tokens": 244209801.0,
+      "step": 98740
+    },
+    {
+      "epoch": 0.905381864857431,
+      "learning_rate": 1.8925460713303383e-05,
+      "loss": 0.9643,
+      "mean_token_accuracy": 0.7068106710910798,
+      "num_tokens": 244237529.0,
+      "step": 98750
+    },
+    {
+      "epoch": 0.9054735490969102,
+      "learning_rate": 1.8907123865407535e-05,
+      "loss": 0.9769,
+      "mean_token_accuracy": 0.7031436681747436,
+      "num_tokens": 244264568.0,
+      "step": 98760
+    },
+    {
+      "epoch": 0.9055652333363895,
+      "learning_rate": 1.888878701751169e-05,
+      "loss": 0.9577,
+      "mean_token_accuracy": 0.7108761012554169,
+      "num_tokens": 244292725.0,
+      "step": 98770
+    },
+    {
+      "epoch": 0.9056569175758687,
+      "learning_rate": 1.8870450169615845e-05,
+      "loss": 0.9686,
+      "mean_token_accuracy": 0.706019914150238,
+      "num_tokens": 244321019.0,
+      "step": 98780
+    },
+    {
+      "epoch": 0.9057486018153479,
+      "learning_rate": 1.8852113321719997e-05,
+      "loss": 0.9717,
+      "mean_token_accuracy": 0.7070391476154327,
+      "num_tokens": 244348792.0,
+      "step": 98790
+    },
+    {
+      "epoch": 0.9058402860548271,
+      "learning_rate": 1.883377647382415e-05,
+      "loss": 0.9616,
+      "mean_token_accuracy": 0.7171567678451538,
+      "num_tokens": 244376666.0,
+      "step": 98800
+    },
+    {
+      "epoch": 0.9059319702943064,
+      "learning_rate": 1.8815439625928304e-05,
+      "loss": 0.9533,
+      "mean_token_accuracy": 0.7125464618206024,
+      "num_tokens": 244404454.0,
+      "step": 98810
+    },
+    {
+      "epoch": 0.9060236545337856,
+      "learning_rate": 1.8797102778032456e-05,
+      "loss": 0.9868,
+      "mean_token_accuracy": 0.7024095237255097,
+      "num_tokens": 244432600.0,
+      "step": 98820
+    },
+    {
+      "epoch": 0.9061153387732649,
+      "learning_rate": 1.877876593013661e-05,
+      "loss": 0.9966,
+      "mean_token_accuracy": 0.703163868188858,
+      "num_tokens": 244460141.0,
+      "step": 98830
+    },
+    {
+      "epoch": 0.9062070230127441,
+      "learning_rate": 1.8760429082240763e-05,
+      "loss": 0.9545,
+      "mean_token_accuracy": 0.7115595698356628,
+      "num_tokens": 244488764.0,
+      "step": 98840
+    },
+    {
+      "epoch": 0.9062987072522234,
+      "learning_rate": 1.8742092234344915e-05,
+      "loss": 0.9634,
+      "mean_token_accuracy": 0.7091646432876587,
+      "num_tokens": 244516920.0,
+      "step": 98850
+    },
+    {
+      "epoch": 0.9063903914917025,
+      "learning_rate": 1.872375538644907e-05,
+      "loss": 0.9697,
+      "mean_token_accuracy": 0.7015791535377502,
+      "num_tokens": 244544901.0,
+      "step": 98860
+    },
+    {
+      "epoch": 0.9064820757311818,
+      "learning_rate": 1.8705418538553222e-05,
+      "loss": 0.9769,
+      "mean_token_accuracy": 0.7043940842151641,
+      "num_tokens": 244573250.0,
+      "step": 98870
+    },
+    {
+      "epoch": 0.906573759970661,
+      "learning_rate": 1.8687081690657378e-05,
+      "loss": 0.9802,
+      "mean_token_accuracy": 0.7076828479766846,
+      "num_tokens": 244600931.0,
+      "step": 98880
+    },
+    {
+      "epoch": 0.9066654442101403,
+      "learning_rate": 1.866874484276153e-05,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7101810514926911,
+      "num_tokens": 244629485.0,
+      "step": 98890
+    },
+    {
+      "epoch": 0.9067571284496195,
+      "learning_rate": 1.8650407994865685e-05,
+      "loss": 0.9765,
+      "mean_token_accuracy": 0.705142331123352,
+      "num_tokens": 244657446.0,
+      "step": 98900
+    },
+    {
+      "epoch": 0.9068488126890988,
+      "learning_rate": 1.8632071146969837e-05,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.711612057685852,
+      "num_tokens": 244685629.0,
+      "step": 98910
+    },
+    {
+      "epoch": 0.906940496928578,
+      "learning_rate": 1.861373429907399e-05,
+      "loss": 0.9303,
+      "mean_token_accuracy": 0.7186287760734558,
+      "num_tokens": 244713457.0,
+      "step": 98920
+    },
+    {
+      "epoch": 0.9070321811680572,
+      "learning_rate": 1.8595397451178144e-05,
+      "loss": 0.9377,
+      "mean_token_accuracy": 0.7127879858016968,
+      "num_tokens": 244740850.0,
+      "step": 98930
+    },
+    {
+      "epoch": 0.9071238654075364,
+      "learning_rate": 1.85770606032823e-05,
+      "loss": 0.9446,
+      "mean_token_accuracy": 0.713395619392395,
+      "num_tokens": 244767394.0,
+      "step": 98940
+    },
+    {
+      "epoch": 0.9072155496470157,
+      "learning_rate": 1.855872375538645e-05,
+      "loss": 0.9687,
+      "mean_token_accuracy": 0.7102888941764831,
+      "num_tokens": 244795492.0,
+      "step": 98950
+    },
+    {
+      "epoch": 0.9073072338864949,
+      "learning_rate": 1.8540386907490603e-05,
+      "loss": 0.9541,
+      "mean_token_accuracy": 0.7060974717140198,
+      "num_tokens": 244823023.0,
+      "step": 98960
+    },
+    {
+      "epoch": 0.9073989181259742,
+      "learning_rate": 1.8522050059594755e-05,
+      "loss": 0.9743,
+      "mean_token_accuracy": 0.7049482047557831,
+      "num_tokens": 244851393.0,
+      "step": 98970
+    },
+    {
+      "epoch": 0.9074906023654534,
+      "learning_rate": 1.850371321169891e-05,
+      "loss": 0.9382,
+      "mean_token_accuracy": 0.7122291028499603,
+      "num_tokens": 244879127.0,
+      "step": 98980
+    },
+    {
+      "epoch": 0.9075822866049326,
+      "learning_rate": 1.8485376363803065e-05,
+      "loss": 0.9678,
+      "mean_token_accuracy": 0.7073399364948273,
+      "num_tokens": 244907138.0,
+      "step": 98990
+    },
+    {
+      "epoch": 0.9076739708444118,
+      "learning_rate": 1.8467039515907217e-05,
+      "loss": 0.9475,
+      "mean_token_accuracy": 0.7094966411590576,
+      "num_tokens": 244934511.0,
+      "step": 99000
+    },
+    {
+      "epoch": 0.9077656550838911,
+      "learning_rate": 1.844870266801137e-05,
+      "loss": 0.946,
+      "mean_token_accuracy": 0.7111939191818237,
+      "num_tokens": 244962022.0,
+      "step": 99010
+    },
+    {
+      "epoch": 0.9078573393233703,
+      "learning_rate": 1.843036582011552e-05,
+      "loss": 1.0062,
+      "mean_token_accuracy": 0.6980007529258728,
+      "num_tokens": 244989987.0,
+      "step": 99020
+    },
+    {
+      "epoch": 0.9079490235628496,
+      "learning_rate": 1.8412028972219676e-05,
+      "loss": 0.9577,
+      "mean_token_accuracy": 0.7091847062110901,
+      "num_tokens": 245018560.0,
+      "step": 99030
+    },
+    {
+      "epoch": 0.9080407078023288,
+      "learning_rate": 1.839369212432383e-05,
+      "loss": 0.9409,
+      "mean_token_accuracy": 0.7124989271163941,
+      "num_tokens": 245046548.0,
+      "step": 99040
+    },
+    {
+      "epoch": 0.9081323920418081,
+      "learning_rate": 1.8375355276427983e-05,
+      "loss": 0.9533,
+      "mean_token_accuracy": 0.710144329071045,
+      "num_tokens": 245074974.0,
+      "step": 99050
+    },
+    {
+      "epoch": 0.9082240762812872,
+      "learning_rate": 1.8357018428532135e-05,
+      "loss": 0.9767,
+      "mean_token_accuracy": 0.7003639161586761,
+      "num_tokens": 245102673.0,
+      "step": 99060
+    },
+    {
+      "epoch": 0.9083157605207665,
+      "learning_rate": 1.8338681580636287e-05,
+      "loss": 0.9644,
+      "mean_token_accuracy": 0.7086762249469757,
+      "num_tokens": 245131267.0,
+      "step": 99070
+    },
+    {
+      "epoch": 0.9084074447602457,
+      "learning_rate": 1.8320344732740442e-05,
+      "loss": 0.984,
+      "mean_token_accuracy": 0.7076332032680511,
+      "num_tokens": 245160271.0,
+      "step": 99080
+    },
+    {
+      "epoch": 0.908499128999725,
+      "learning_rate": 1.8302007884844597e-05,
+      "loss": 0.9315,
+      "mean_token_accuracy": 0.7142459034919739,
+      "num_tokens": 245188732.0,
+      "step": 99090
+    },
+    {
+      "epoch": 0.9085908132392042,
+      "learning_rate": 1.828367103694875e-05,
+      "loss": 0.9599,
+      "mean_token_accuracy": 0.7091251969337463,
+      "num_tokens": 245216631.0,
+      "step": 99100
+    },
+    {
+      "epoch": 0.9086824974786835,
+      "learning_rate": 1.82653341890529e-05,
+      "loss": 0.9498,
+      "mean_token_accuracy": 0.7134461283683777,
+      "num_tokens": 245244130.0,
+      "step": 99110
+    },
+    {
+      "epoch": 0.9087741817181626,
+      "learning_rate": 1.8246997341157056e-05,
+      "loss": 0.9777,
+      "mean_token_accuracy": 0.702265340089798,
+      "num_tokens": 245271220.0,
+      "step": 99120
+    },
+    {
+      "epoch": 0.9088658659576418,
+      "learning_rate": 1.8228660493261208e-05,
+      "loss": 0.9438,
+      "mean_token_accuracy": 0.7100129187107086,
+      "num_tokens": 245299301.0,
+      "step": 99130
+    },
+    {
+      "epoch": 0.9089575501971211,
+      "learning_rate": 1.8210323645365363e-05,
+      "loss": 0.9909,
+      "mean_token_accuracy": 0.699597829580307,
+      "num_tokens": 245327480.0,
+      "step": 99140
+    },
+    {
+      "epoch": 0.9090492344366004,
+      "learning_rate": 1.8191986797469515e-05,
+      "loss": 0.9541,
+      "mean_token_accuracy": 0.7078088283538818,
+      "num_tokens": 245356376.0,
+      "step": 99150
+    },
+    {
+      "epoch": 0.9091409186760796,
+      "learning_rate": 1.8173649949573667e-05,
+      "loss": 0.9648,
+      "mean_token_accuracy": 0.7084352314472199,
+      "num_tokens": 245383818.0,
+      "step": 99160
+    },
+    {
+      "epoch": 0.9092326029155589,
+      "learning_rate": 1.8155313101677822e-05,
+      "loss": 0.9596,
+      "mean_token_accuracy": 0.7130893707275391,
+      "num_tokens": 245411880.0,
+      "step": 99170
+    },
+    {
+      "epoch": 0.9093242871550381,
+      "learning_rate": 1.8136976253781974e-05,
+      "loss": 0.9689,
+      "mean_token_accuracy": 0.7035489439964294,
+      "num_tokens": 245438299.0,
+      "step": 99180
+    },
+    {
+      "epoch": 0.9094159713945172,
+      "learning_rate": 1.811863940588613e-05,
+      "loss": 0.9411,
+      "mean_token_accuracy": 0.7125300884246826,
+      "num_tokens": 245465862.0,
+      "step": 99190
+    },
+    {
+      "epoch": 0.9095076556339965,
+      "learning_rate": 1.810030255799028e-05,
+      "loss": 0.9604,
+      "mean_token_accuracy": 0.7077223598957062,
+      "num_tokens": 245495043.0,
+      "step": 99200
+    },
+    {
+      "epoch": 0.9095993398734757,
+      "learning_rate": 1.8081965710094437e-05,
+      "loss": 0.9366,
+      "mean_token_accuracy": 0.7149075448513031,
+      "num_tokens": 245522478.0,
+      "step": 99210
+    },
+    {
+      "epoch": 0.909691024112955,
+      "learning_rate": 1.806362886219859e-05,
+      "loss": 0.9709,
+      "mean_token_accuracy": 0.7035350382328034,
+      "num_tokens": 245549466.0,
+      "step": 99220
+    },
+    {
+      "epoch": 0.9097827083524342,
+      "learning_rate": 1.804529201430274e-05,
+      "loss": 0.9765,
+      "mean_token_accuracy": 0.6995810747146607,
+      "num_tokens": 245578705.0,
+      "step": 99230
+    },
+    {
+      "epoch": 0.9098743925919135,
+      "learning_rate": 1.8026955166406896e-05,
+      "loss": 0.9627,
+      "mean_token_accuracy": 0.7103352427482605,
+      "num_tokens": 245605976.0,
+      "step": 99240
+    },
+    {
+      "epoch": 0.9099660768313926,
+      "learning_rate": 1.800861831851105e-05,
+      "loss": 0.9563,
+      "mean_token_accuracy": 0.7094795227050781,
+      "num_tokens": 245633499.0,
+      "step": 99250
+    },
+    {
+      "epoch": 0.9100577610708719,
+      "learning_rate": 1.7990281470615203e-05,
+      "loss": 0.9544,
+      "mean_token_accuracy": 0.7120623230934143,
+      "num_tokens": 245662205.0,
+      "step": 99260
+    },
+    {
+      "epoch": 0.9101494453103511,
+      "learning_rate": 1.7971944622719355e-05,
+      "loss": 0.9688,
+      "mean_token_accuracy": 0.7137180268764496,
+      "num_tokens": 245689488.0,
+      "step": 99270
+    },
+    {
+      "epoch": 0.9102411295498304,
+      "learning_rate": 1.7953607774823507e-05,
+      "loss": 0.964,
+      "mean_token_accuracy": 0.708096569776535,
+      "num_tokens": 245716912.0,
+      "step": 99280
+    },
+    {
+      "epoch": 0.9103328137893096,
+      "learning_rate": 1.7935270926927662e-05,
+      "loss": 0.9566,
+      "mean_token_accuracy": 0.7079792618751526,
+      "num_tokens": 245745035.0,
+      "step": 99290
+    },
+    {
+      "epoch": 0.9104244980287889,
+      "learning_rate": 1.7916934079031817e-05,
+      "loss": 0.9825,
+      "mean_token_accuracy": 0.7026017665863037,
+      "num_tokens": 245773141.0,
+      "step": 99300
+    },
+    {
+      "epoch": 0.9105161822682681,
+      "learning_rate": 1.789859723113597e-05,
+      "loss": 0.9718,
+      "mean_token_accuracy": 0.7071141004562378,
+      "num_tokens": 245801463.0,
+      "step": 99310
+    },
+    {
+      "epoch": 0.9106078665077473,
+      "learning_rate": 1.788026038324012e-05,
+      "loss": 0.9458,
+      "mean_token_accuracy": 0.7118683815002441,
+      "num_tokens": 245830678.0,
+      "step": 99320
+    },
+    {
+      "epoch": 0.9106995507472265,
+      "learning_rate": 1.7861923535344273e-05,
+      "loss": 0.9877,
+      "mean_token_accuracy": 0.7052164196968078,
+      "num_tokens": 245858415.0,
+      "step": 99330
+    },
+    {
+      "epoch": 0.9107912349867058,
+      "learning_rate": 1.784358668744843e-05,
+      "loss": 0.9943,
+      "mean_token_accuracy": 0.695398610830307,
+      "num_tokens": 245886988.0,
+      "step": 99340
+    },
+    {
+      "epoch": 0.910882919226185,
+      "learning_rate": 1.7825249839552583e-05,
+      "loss": 0.9179,
+      "mean_token_accuracy": 0.7180028676986694,
+      "num_tokens": 245914904.0,
+      "step": 99350
+    },
+    {
+      "epoch": 0.9109746034656643,
+      "learning_rate": 1.7806912991656735e-05,
+      "loss": 0.9511,
+      "mean_token_accuracy": 0.7151628613471985,
+      "num_tokens": 245942862.0,
+      "step": 99360
+    },
+    {
+      "epoch": 0.9110662877051435,
+      "learning_rate": 1.7788576143760887e-05,
+      "loss": 0.9493,
+      "mean_token_accuracy": 0.7169644296169281,
+      "num_tokens": 245969861.0,
+      "step": 99370
+    },
+    {
+      "epoch": 0.9111579719446227,
+      "learning_rate": 1.777023929586504e-05,
+      "loss": 0.9589,
+      "mean_token_accuracy": 0.7156728386878968,
+      "num_tokens": 245998071.0,
+      "step": 99380
+    },
+    {
+      "epoch": 0.9112496561841019,
+      "learning_rate": 1.7751902447969197e-05,
+      "loss": 0.9643,
+      "mean_token_accuracy": 0.7130199491977691,
+      "num_tokens": 246026673.0,
+      "step": 99390
+    },
+    {
+      "epoch": 0.9113413404235812,
+      "learning_rate": 1.773356560007335e-05,
+      "loss": 0.9391,
+      "mean_token_accuracy": 0.7124180436134339,
+      "num_tokens": 246055196.0,
+      "step": 99400
+    },
+    {
+      "epoch": 0.9114330246630604,
+      "learning_rate": 1.77152287521775e-05,
+      "loss": 0.9396,
+      "mean_token_accuracy": 0.7146018385887146,
+      "num_tokens": 246083376.0,
+      "step": 99410
+    },
+    {
+      "epoch": 0.9115247089025397,
+      "learning_rate": 1.7696891904281653e-05,
+      "loss": 0.9725,
+      "mean_token_accuracy": 0.7048784852027893,
+      "num_tokens": 246111594.0,
+      "step": 99420
+    },
+    {
+      "epoch": 0.9116163931420189,
+      "learning_rate": 1.767855505638581e-05,
+      "loss": 0.9431,
+      "mean_token_accuracy": 0.7131402134895325,
+      "num_tokens": 246139183.0,
+      "step": 99430
+    },
+    {
+      "epoch": 0.9117080773814982,
+      "learning_rate": 1.7660218208489964e-05,
+      "loss": 0.9725,
+      "mean_token_accuracy": 0.7039356589317322,
+      "num_tokens": 246167209.0,
+      "step": 99440
+    },
+    {
+      "epoch": 0.9117997616209773,
+      "learning_rate": 1.7641881360594115e-05,
+      "loss": 0.955,
+      "mean_token_accuracy": 0.7087647438049316,
+      "num_tokens": 246194224.0,
+      "step": 99450
+    },
+    {
+      "epoch": 0.9118914458604566,
+      "learning_rate": 1.7623544512698267e-05,
+      "loss": 0.9464,
+      "mean_token_accuracy": 0.7096237003803253,
+      "num_tokens": 246221570.0,
+      "step": 99460
+    },
+    {
+      "epoch": 0.9119831300999358,
+      "learning_rate": 1.760520766480242e-05,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.70811527967453,
+      "num_tokens": 246250606.0,
+      "step": 99470
+    },
+    {
+      "epoch": 0.9120748143394151,
+      "learning_rate": 1.7586870816906574e-05,
+      "loss": 0.9406,
+      "mean_token_accuracy": 0.7127262830734253,
+      "num_tokens": 246278513.0,
+      "step": 99480
+    },
+    {
+      "epoch": 0.9121664985788943,
+      "learning_rate": 1.756853396901073e-05,
+      "loss": 0.9319,
+      "mean_token_accuracy": 0.7115866303443908,
+      "num_tokens": 246307389.0,
+      "step": 99490
+    },
+    {
+      "epoch": 0.9122581828183736,
+      "learning_rate": 1.755019712111488e-05,
+      "loss": 0.9756,
+      "mean_token_accuracy": 0.705430406332016,
+      "num_tokens": 246334845.0,
+      "step": 99500
+    },
+    {
+      "epoch": 0.9123498670578527,
+      "learning_rate": 1.7531860273219033e-05,
+      "loss": 0.9308,
+      "mean_token_accuracy": 0.7134301126003265,
+      "num_tokens": 246362959.0,
+      "step": 99510
+    },
+    {
+      "epoch": 0.912441551297332,
+      "learning_rate": 1.751352342532319e-05,
+      "loss": 0.9867,
+      "mean_token_accuracy": 0.703470128774643,
+      "num_tokens": 246391278.0,
+      "step": 99520
+    },
+    {
+      "epoch": 0.9125332355368112,
+      "learning_rate": 1.749518657742734e-05,
+      "loss": 0.9511,
+      "mean_token_accuracy": 0.7146863520145417,
+      "num_tokens": 246418411.0,
+      "step": 99530
+    },
+    {
+      "epoch": 0.9126249197762905,
+      "learning_rate": 1.7476849729531496e-05,
+      "loss": 0.9828,
+      "mean_token_accuracy": 0.7031401574611664,
+      "num_tokens": 246446501.0,
+      "step": 99540
+    },
+    {
+      "epoch": 0.9127166040157697,
+      "learning_rate": 1.7458512881635648e-05,
+      "loss": 0.9813,
+      "mean_token_accuracy": 0.7034123122692109,
+      "num_tokens": 246474522.0,
+      "step": 99550
+    },
+    {
+      "epoch": 0.912808288255249,
+      "learning_rate": 1.74401760337398e-05,
+      "loss": 0.9408,
+      "mean_token_accuracy": 0.7101796746253968,
+      "num_tokens": 246503042.0,
+      "step": 99560
+    },
+    {
+      "epoch": 0.9128999724947282,
+      "learning_rate": 1.7421839185843955e-05,
+      "loss": 0.9274,
+      "mean_token_accuracy": 0.7152628660202026,
+      "num_tokens": 246531643.0,
+      "step": 99570
+    },
+    {
+      "epoch": 0.9129916567342073,
+      "learning_rate": 1.7403502337948107e-05,
+      "loss": 0.96,
+      "mean_token_accuracy": 0.7102968633174896,
+      "num_tokens": 246560115.0,
+      "step": 99580
+    },
+    {
+      "epoch": 0.9130833409736866,
+      "learning_rate": 1.7385165490052262e-05,
+      "loss": 0.9941,
+      "mean_token_accuracy": 0.6982282519340515,
+      "num_tokens": 246588271.0,
+      "step": 99590
+    },
+    {
+      "epoch": 0.9131750252131658,
+      "learning_rate": 1.7366828642156414e-05,
+      "loss": 0.9507,
+      "mean_token_accuracy": 0.7108680367469787,
+      "num_tokens": 246616955.0,
+      "step": 99600
+    },
+    {
+      "epoch": 0.9132667094526451,
+      "learning_rate": 1.734849179426057e-05,
+      "loss": 0.9148,
+      "mean_token_accuracy": 0.7212695956230164,
+      "num_tokens": 246645517.0,
+      "step": 99610
+    },
+    {
+      "epoch": 0.9133583936921243,
+      "learning_rate": 1.733015494636472e-05,
+      "loss": 0.9604,
+      "mean_token_accuracy": 0.7041295111179352,
+      "num_tokens": 246673947.0,
+      "step": 99620
+    },
+    {
+      "epoch": 0.9134500779316036,
+      "learning_rate": 1.7311818098468873e-05,
+      "loss": 0.9508,
+      "mean_token_accuracy": 0.7109302341938019,
+      "num_tokens": 246701153.0,
+      "step": 99630
+    },
+    {
+      "epoch": 0.9135417621710827,
+      "learning_rate": 1.7293481250573028e-05,
+      "loss": 0.9308,
+      "mean_token_accuracy": 0.7114158809185028,
+      "num_tokens": 246728338.0,
+      "step": 99640
+    },
+    {
+      "epoch": 0.913633446410562,
+      "learning_rate": 1.7275144402677183e-05,
+      "loss": 0.9586,
+      "mean_token_accuracy": 0.7120016038417816,
+      "num_tokens": 246756496.0,
+      "step": 99650
+    },
+    {
+      "epoch": 0.9137251306500412,
+      "learning_rate": 1.7256807554781335e-05,
+      "loss": 0.9286,
+      "mean_token_accuracy": 0.7170893132686615,
+      "num_tokens": 246783803.0,
+      "step": 99660
+    },
+    {
+      "epoch": 0.9138168148895205,
+      "learning_rate": 1.7238470706885487e-05,
+      "loss": 0.9398,
+      "mean_token_accuracy": 0.7143398344516754,
+      "num_tokens": 246810903.0,
+      "step": 99670
+    },
+    {
+      "epoch": 0.9139084991289997,
+      "learning_rate": 1.722013385898964e-05,
+      "loss": 0.9505,
+      "mean_token_accuracy": 0.7043926000595093,
+      "num_tokens": 246838738.0,
+      "step": 99680
+    },
+    {
+      "epoch": 0.914000183368479,
+      "learning_rate": 1.7201797011093794e-05,
+      "loss": 0.9494,
+      "mean_token_accuracy": 0.7065604150295257,
+      "num_tokens": 246868248.0,
+      "step": 99690
+    },
+    {
+      "epoch": 0.9140918676079582,
+      "learning_rate": 1.718346016319795e-05,
+      "loss": 0.9558,
+      "mean_token_accuracy": 0.7034044682979583,
+      "num_tokens": 246896457.0,
+      "step": 99700
+    },
+    {
+      "epoch": 0.9141835518474374,
+      "learning_rate": 1.71651233153021e-05,
+      "loss": 0.942,
+      "mean_token_accuracy": 0.714598786830902,
+      "num_tokens": 246923841.0,
+      "step": 99710
+    },
+    {
+      "epoch": 0.9142752360869166,
+      "learning_rate": 1.7146786467406253e-05,
+      "loss": 0.9386,
+      "mean_token_accuracy": 0.7097809672355652,
+      "num_tokens": 246951572.0,
+      "step": 99720
+    },
+    {
+      "epoch": 0.9143669203263959,
+      "learning_rate": 1.7128449619510405e-05,
+      "loss": 0.9442,
+      "mean_token_accuracy": 0.7133966445922851,
+      "num_tokens": 246980629.0,
+      "step": 99730
+    },
+    {
+      "epoch": 0.9144586045658751,
+      "learning_rate": 1.711011277161456e-05,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7073602676391602,
+      "num_tokens": 247007679.0,
+      "step": 99740
+    },
+    {
+      "epoch": 0.9145502888053544,
+      "learning_rate": 1.7091775923718716e-05,
+      "loss": 0.961,
+      "mean_token_accuracy": 0.7088682174682617,
+      "num_tokens": 247035494.0,
+      "step": 99750
+    },
+    {
+      "epoch": 0.9146419730448336,
+      "learning_rate": 1.7073439075822867e-05,
+      "loss": 0.9605,
+      "mean_token_accuracy": 0.7071173548698425,
+      "num_tokens": 247062410.0,
+      "step": 99760
+    },
+    {
+      "epoch": 0.9147336572843128,
+      "learning_rate": 1.705510222792702e-05,
+      "loss": 0.9393,
+      "mean_token_accuracy": 0.7137604415416717,
+      "num_tokens": 247090463.0,
+      "step": 99770
+    },
+    {
+      "epoch": 0.914825341523792,
+      "learning_rate": 1.703676538003117e-05,
+      "loss": 0.9298,
+      "mean_token_accuracy": 0.718628978729248,
+      "num_tokens": 247118705.0,
+      "step": 99780
+    },
+    {
+      "epoch": 0.9149170257632713,
+      "learning_rate": 1.7018428532135326e-05,
+      "loss": 0.9632,
+      "mean_token_accuracy": 0.7087247788906097,
+      "num_tokens": 247146866.0,
+      "step": 99790
+    },
+    {
+      "epoch": 0.9150087100027505,
+      "learning_rate": 1.700009168423948e-05,
+      "loss": 0.9252,
+      "mean_token_accuracy": 0.7173439919948578,
+      "num_tokens": 247174348.0,
+      "step": 99800
+    },
+    {
+      "epoch": 0.9151003942422298,
+      "learning_rate": 1.6981754836343633e-05,
+      "loss": 0.945,
+      "mean_token_accuracy": 0.7133069396018982,
+      "num_tokens": 247201025.0,
+      "step": 99810
+    },
+    {
+      "epoch": 0.915192078481709,
+      "learning_rate": 1.6963417988447785e-05,
+      "loss": 0.9693,
+      "mean_token_accuracy": 0.7048952162265778,
+      "num_tokens": 247227932.0,
+      "step": 99820
+    },
+    {
+      "epoch": 0.9152837627211883,
+      "learning_rate": 1.694508114055194e-05,
+      "loss": 0.9652,
+      "mean_token_accuracy": 0.703224265575409,
+      "num_tokens": 247256143.0,
+      "step": 99830
+    },
+    {
+      "epoch": 0.9153754469606674,
+      "learning_rate": 1.6926744292656092e-05,
+      "loss": 0.9858,
+      "mean_token_accuracy": 0.7009982287883758,
+      "num_tokens": 247284651.0,
+      "step": 99840
+    },
+    {
+      "epoch": 0.9154671312001467,
+      "learning_rate": 1.6908407444760248e-05,
+      "loss": 0.9349,
+      "mean_token_accuracy": 0.719301962852478,
+      "num_tokens": 247312168.0,
+      "step": 99850
+    },
+    {
+      "epoch": 0.9155588154396259,
+      "learning_rate": 1.68900705968644e-05,
+      "loss": 0.9543,
+      "mean_token_accuracy": 0.7073159754276276,
+      "num_tokens": 247339562.0,
+      "step": 99860
+    },
+    {
+      "epoch": 0.9156504996791052,
+      "learning_rate": 1.687173374896855e-05,
+      "loss": 0.9736,
+      "mean_token_accuracy": 0.7037499010562897,
+      "num_tokens": 247367591.0,
+      "step": 99870
+    },
+    {
+      "epoch": 0.9157421839185844,
+      "learning_rate": 1.6853396901072707e-05,
+      "loss": 0.9568,
+      "mean_token_accuracy": 0.710649061203003,
+      "num_tokens": 247394895.0,
+      "step": 99880
+    },
+    {
+      "epoch": 0.9158338681580637,
+      "learning_rate": 1.683506005317686e-05,
+      "loss": 0.9364,
+      "mean_token_accuracy": 0.716763186454773,
+      "num_tokens": 247421838.0,
+      "step": 99890
+    },
+    {
+      "epoch": 0.9159255523975428,
+      "learning_rate": 1.6816723205281014e-05,
+      "loss": 0.9384,
+      "mean_token_accuracy": 0.7132778227329254,
+      "num_tokens": 247449576.0,
+      "step": 99900
+    },
+    {
+      "epoch": 0.9160172366370221,
+      "learning_rate": 1.6798386357385166e-05,
+      "loss": 0.9398,
+      "mean_token_accuracy": 0.7177938222885132,
+      "num_tokens": 247476638.0,
+      "step": 99910
+    },
+    {
+      "epoch": 0.9161089208765013,
+      "learning_rate": 1.678004950948932e-05,
+      "loss": 0.9669,
+      "mean_token_accuracy": 0.7109892904758454,
+      "num_tokens": 247504955.0,
+      "step": 99920
+    },
+    {
+      "epoch": 0.9162006051159806,
+      "learning_rate": 1.6761712661593473e-05,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.7131851136684417,
+      "num_tokens": 247532364.0,
+      "step": 99930
+    },
+    {
+      "epoch": 0.9162922893554598,
+      "learning_rate": 1.6743375813697625e-05,
+      "loss": 0.9645,
+      "mean_token_accuracy": 0.7099905014038086,
+      "num_tokens": 247560532.0,
+      "step": 99940
+    },
+    {
+      "epoch": 0.9163839735949391,
+      "learning_rate": 1.672503896580178e-05,
+      "loss": 0.9669,
+      "mean_token_accuracy": 0.7084422290325165,
+      "num_tokens": 247588427.0,
+      "step": 99950
+    },
+    {
+      "epoch": 0.9164756578344183,
+      "learning_rate": 1.6706702117905935e-05,
+      "loss": 0.9606,
+      "mean_token_accuracy": 0.7048634946346283,
+      "num_tokens": 247616600.0,
+      "step": 99960
+    },
+    {
+      "epoch": 0.9165673420738975,
+      "learning_rate": 1.6688365270010087e-05,
+      "loss": 0.9683,
+      "mean_token_accuracy": 0.7047381281852723,
+      "num_tokens": 247644520.0,
+      "step": 99970
+    },
+    {
+      "epoch": 0.9166590263133767,
+      "learning_rate": 1.667002842211424e-05,
+      "loss": 0.9759,
+      "mean_token_accuracy": 0.7052361905574799,
+      "num_tokens": 247672647.0,
+      "step": 99980
+    },
+    {
+      "epoch": 0.916750710552856,
+      "learning_rate": 1.665169157421839e-05,
+      "loss": 0.9489,
+      "mean_token_accuracy": 0.7082134187221527,
+      "num_tokens": 247700084.0,
+      "step": 99990
+    },
+    {
+      "epoch": 0.9168423947923352,
+      "learning_rate": 1.6633354726322546e-05,
+      "loss": 0.9397,
+      "mean_token_accuracy": 0.7144883096218109,
+      "num_tokens": 247727674.0,
+      "step": 100000
+    },
+    {
+      "epoch": 0.9169340790318145,
+      "learning_rate": 1.66150178784267e-05,
+      "loss": 0.9367,
+      "mean_token_accuracy": 0.7193460941314698,
+      "num_tokens": 247755286.0,
+      "step": 100010
+    },
+    {
+      "epoch": 0.9170257632712937,
+      "learning_rate": 1.6596681030530853e-05,
+      "loss": 0.9658,
+      "mean_token_accuracy": 0.7092611610889434,
+      "num_tokens": 247783728.0,
+      "step": 100020
+    },
+    {
+      "epoch": 0.9171174475107728,
+      "learning_rate": 1.6578344182635005e-05,
+      "loss": 0.976,
+      "mean_token_accuracy": 0.704130619764328,
+      "num_tokens": 247811777.0,
+      "step": 100030
+    },
+    {
+      "epoch": 0.9172091317502521,
+      "learning_rate": 1.6560007334739157e-05,
+      "loss": 0.9499,
+      "mean_token_accuracy": 0.715839284658432,
+      "num_tokens": 247840103.0,
+      "step": 100040
+    },
+    {
+      "epoch": 0.9173008159897313,
+      "learning_rate": 1.6541670486843312e-05,
+      "loss": 0.9467,
+      "mean_token_accuracy": 0.7104815423488617,
+      "num_tokens": 247867476.0,
+      "step": 100050
+    },
+    {
+      "epoch": 0.9173925002292106,
+      "learning_rate": 1.6523333638947467e-05,
+      "loss": 0.9482,
+      "mean_token_accuracy": 0.7036938011646271,
+      "num_tokens": 247894941.0,
+      "step": 100060
+    },
+    {
+      "epoch": 0.9174841844686898,
+      "learning_rate": 1.650499679105162e-05,
+      "loss": 0.941,
+      "mean_token_accuracy": 0.711452579498291,
+      "num_tokens": 247923325.0,
+      "step": 100070
+    },
+    {
+      "epoch": 0.9175758687081691,
+      "learning_rate": 1.648665994315577e-05,
+      "loss": 0.9682,
+      "mean_token_accuracy": 0.7044674515724182,
+      "num_tokens": 247951961.0,
+      "step": 100080
+    },
+    {
+      "epoch": 0.9176675529476483,
+      "learning_rate": 1.6468323095259923e-05,
+      "loss": 0.9615,
+      "mean_token_accuracy": 0.7075078308582305,
+      "num_tokens": 247981138.0,
+      "step": 100090
+    },
+    {
+      "epoch": 0.9177592371871275,
+      "learning_rate": 1.6449986247364078e-05,
+      "loss": 0.9681,
+      "mean_token_accuracy": 0.7054571211338043,
+      "num_tokens": 248009302.0,
+      "step": 100100
+    },
+    {
+      "epoch": 0.9178509214266067,
+      "learning_rate": 1.6431649399468234e-05,
+      "loss": 0.9577,
+      "mean_token_accuracy": 0.7064182341098786,
+      "num_tokens": 248037454.0,
+      "step": 100110
+    },
+    {
+      "epoch": 0.917942605666086,
+      "learning_rate": 1.6413312551572385e-05,
+      "loss": 0.9564,
+      "mean_token_accuracy": 0.707996416091919,
+      "num_tokens": 248064480.0,
+      "step": 100120
+    },
+    {
+      "epoch": 0.9180342899055652,
+      "learning_rate": 1.6394975703676537e-05,
+      "loss": 0.9554,
+      "mean_token_accuracy": 0.7101505339145661,
+      "num_tokens": 248092366.0,
+      "step": 100130
+    },
+    {
+      "epoch": 0.9181259741450445,
+      "learning_rate": 1.6376638855780693e-05,
+      "loss": 0.95,
+      "mean_token_accuracy": 0.7134077310562134,
+      "num_tokens": 248119625.0,
+      "step": 100140
+    },
+    {
+      "epoch": 0.9182176583845237,
+      "learning_rate": 1.6358302007884844e-05,
+      "loss": 0.9784,
+      "mean_token_accuracy": 0.7073571622371674,
+      "num_tokens": 248147521.0,
+      "step": 100150
+    },
+    {
+      "epoch": 0.918309342624003,
+      "learning_rate": 1.6339965159989e-05,
+      "loss": 0.9642,
+      "mean_token_accuracy": 0.7048455774784088,
+      "num_tokens": 248174983.0,
+      "step": 100160
+    },
+    {
+      "epoch": 0.9184010268634821,
+      "learning_rate": 1.632162831209315e-05,
+      "loss": 0.9473,
+      "mean_token_accuracy": 0.713927811384201,
+      "num_tokens": 248203396.0,
+      "step": 100170
+    },
+    {
+      "epoch": 0.9184927111029614,
+      "learning_rate": 1.6303291464197303e-05,
+      "loss": 0.9453,
+      "mean_token_accuracy": 0.7139868080615998,
+      "num_tokens": 248230782.0,
+      "step": 100180
+    },
+    {
+      "epoch": 0.9185843953424406,
+      "learning_rate": 1.628495461630146e-05,
+      "loss": 0.9569,
+      "mean_token_accuracy": 0.7076826691627502,
+      "num_tokens": 248258475.0,
+      "step": 100190
+    },
+    {
+      "epoch": 0.9186760795819199,
+      "learning_rate": 1.626661776840561e-05,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.708940851688385,
+      "num_tokens": 248286546.0,
+      "step": 100200
+    },
+    {
+      "epoch": 0.9187677638213991,
+      "learning_rate": 1.6248280920509766e-05,
+      "loss": 0.9509,
+      "mean_token_accuracy": 0.7117028594017029,
+      "num_tokens": 248314924.0,
+      "step": 100210
+    },
+    {
+      "epoch": 0.9188594480608784,
+      "learning_rate": 1.6229944072613918e-05,
+      "loss": 0.9577,
+      "mean_token_accuracy": 0.7107714116573334,
+      "num_tokens": 248343404.0,
+      "step": 100220
+    },
+    {
+      "epoch": 0.9189511323003575,
+      "learning_rate": 1.6211607224718073e-05,
+      "loss": 0.9226,
+      "mean_token_accuracy": 0.7155018031597138,
+      "num_tokens": 248371074.0,
+      "step": 100230
+    },
+    {
+      "epoch": 0.9190428165398368,
+      "learning_rate": 1.6193270376822225e-05,
+      "loss": 0.9863,
+      "mean_token_accuracy": 0.7039390444755554,
+      "num_tokens": 248398758.0,
+      "step": 100240
+    },
+    {
+      "epoch": 0.919134500779316,
+      "learning_rate": 1.6174933528926377e-05,
+      "loss": 0.9752,
+      "mean_token_accuracy": 0.7047863602638245,
+      "num_tokens": 248426748.0,
+      "step": 100250
+    },
+    {
+      "epoch": 0.9192261850187953,
+      "learning_rate": 1.6156596681030532e-05,
+      "loss": 0.9636,
+      "mean_token_accuracy": 0.7020634472370147,
+      "num_tokens": 248454586.0,
+      "step": 100260
+    },
+    {
+      "epoch": 0.9193178692582745,
+      "learning_rate": 1.6138259833134687e-05,
+      "loss": 0.978,
+      "mean_token_accuracy": 0.7049915015697479,
+      "num_tokens": 248482809.0,
+      "step": 100270
+    },
+    {
+      "epoch": 0.9194095534977538,
+      "learning_rate": 1.611992298523884e-05,
+      "loss": 0.96,
+      "mean_token_accuracy": 0.7088077008724213,
+      "num_tokens": 248511108.0,
+      "step": 100280
+    },
+    {
+      "epoch": 0.919501237737233,
+      "learning_rate": 1.610158613734299e-05,
+      "loss": 0.9377,
+      "mean_token_accuracy": 0.7143801927566529,
+      "num_tokens": 248538791.0,
+      "step": 100290
+    },
+    {
+      "epoch": 0.9195929219767122,
+      "learning_rate": 1.6083249289447143e-05,
+      "loss": 1.0013,
+      "mean_token_accuracy": 0.702047199010849,
+      "num_tokens": 248567153.0,
+      "step": 100300
+    },
+    {
+      "epoch": 0.9196846062161914,
+      "learning_rate": 1.6064912441551298e-05,
+      "loss": 0.9588,
+      "mean_token_accuracy": 0.7086064577102661,
+      "num_tokens": 248594810.0,
+      "step": 100310
+    },
+    {
+      "epoch": 0.9197762904556707,
+      "learning_rate": 1.6046575593655453e-05,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7123761355876923,
+      "num_tokens": 248622865.0,
+      "step": 100320
+    },
+    {
+      "epoch": 0.9198679746951499,
+      "learning_rate": 1.6028238745759605e-05,
+      "loss": 0.9697,
+      "mean_token_accuracy": 0.7054692029953002,
+      "num_tokens": 248649871.0,
+      "step": 100330
+    },
+    {
+      "epoch": 0.9199596589346292,
+      "learning_rate": 1.6009901897863757e-05,
+      "loss": 0.9767,
+      "mean_token_accuracy": 0.7065792143344879,
+      "num_tokens": 248678583.0,
+      "step": 100340
+    },
+    {
+      "epoch": 0.9200513431741084,
+      "learning_rate": 1.599156504996791e-05,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7048300802707672,
+      "num_tokens": 248706543.0,
+      "step": 100350
+    },
+    {
+      "epoch": 0.9201430274135876,
+      "learning_rate": 1.5973228202072068e-05,
+      "loss": 0.9153,
+      "mean_token_accuracy": 0.7165494084358215,
+      "num_tokens": 248734240.0,
+      "step": 100360
+    },
+    {
+      "epoch": 0.9202347116530668,
+      "learning_rate": 1.595489135417622e-05,
+      "loss": 0.9482,
+      "mean_token_accuracy": 0.7105397284030914,
+      "num_tokens": 248762583.0,
+      "step": 100370
+    },
+    {
+      "epoch": 0.9203263958925461,
+      "learning_rate": 1.593655450628037e-05,
+      "loss": 0.9528,
+      "mean_token_accuracy": 0.7131660044193268,
+      "num_tokens": 248790061.0,
+      "step": 100380
+    },
+    {
+      "epoch": 0.9204180801320253,
+      "learning_rate": 1.5918217658384523e-05,
+      "loss": 0.9914,
+      "mean_token_accuracy": 0.7015332520008087,
+      "num_tokens": 248817815.0,
+      "step": 100390
+    },
+    {
+      "epoch": 0.9205097643715046,
+      "learning_rate": 1.5899880810488675e-05,
+      "loss": 0.9368,
+      "mean_token_accuracy": 0.7151429891586304,
+      "num_tokens": 248846376.0,
+      "step": 100400
+    },
+    {
+      "epoch": 0.9206014486109838,
+      "learning_rate": 1.5881543962592834e-05,
+      "loss": 0.9636,
+      "mean_token_accuracy": 0.7101771891117096,
+      "num_tokens": 248874429.0,
+      "step": 100410
+    },
+    {
+      "epoch": 0.9206931328504631,
+      "learning_rate": 1.5863207114696985e-05,
+      "loss": 0.9803,
+      "mean_token_accuracy": 0.7020486354827881,
+      "num_tokens": 248902431.0,
+      "step": 100420
+    },
+    {
+      "epoch": 0.9207848170899422,
+      "learning_rate": 1.5844870266801137e-05,
+      "loss": 0.9452,
+      "mean_token_accuracy": 0.717402845621109,
+      "num_tokens": 248930897.0,
+      "step": 100430
+    },
+    {
+      "epoch": 0.9208765013294214,
+      "learning_rate": 1.582653341890529e-05,
+      "loss": 0.9368,
+      "mean_token_accuracy": 0.7192105889320374,
+      "num_tokens": 248958107.0,
+      "step": 100440
+    },
+    {
+      "epoch": 0.9209681855689007,
+      "learning_rate": 1.5808196571009444e-05,
+      "loss": 0.9564,
+      "mean_token_accuracy": 0.7145209670066833,
+      "num_tokens": 248986624.0,
+      "step": 100450
+    },
+    {
+      "epoch": 0.92105986980838,
+      "learning_rate": 1.57898597231136e-05,
+      "loss": 0.9729,
+      "mean_token_accuracy": 0.7003249943256378,
+      "num_tokens": 249015154.0,
+      "step": 100460
+    },
+    {
+      "epoch": 0.9211515540478592,
+      "learning_rate": 1.577152287521775e-05,
+      "loss": 0.9657,
+      "mean_token_accuracy": 0.7100578725337983,
+      "num_tokens": 249043422.0,
+      "step": 100470
+    },
+    {
+      "epoch": 0.9212432382873385,
+      "learning_rate": 1.5753186027321903e-05,
+      "loss": 0.9685,
+      "mean_token_accuracy": 0.705563622713089,
+      "num_tokens": 249070893.0,
+      "step": 100480
+    },
+    {
+      "epoch": 0.9213349225268176,
+      "learning_rate": 1.5734849179426055e-05,
+      "loss": 0.9303,
+      "mean_token_accuracy": 0.7130422115325927,
+      "num_tokens": 249099313.0,
+      "step": 100490
+    },
+    {
+      "epoch": 0.9214266067662968,
+      "learning_rate": 1.571651233153021e-05,
+      "loss": 0.945,
+      "mean_token_accuracy": 0.7152499079704284,
+      "num_tokens": 249127599.0,
+      "step": 100500
+    },
+    {
+      "epoch": 0.9215182910057761,
+      "learning_rate": 1.5698175483634366e-05,
+      "loss": 0.9557,
+      "mean_token_accuracy": 0.7070723831653595,
+      "num_tokens": 249155063.0,
+      "step": 100510
+    },
+    {
+      "epoch": 0.9216099752452553,
+      "learning_rate": 1.5679838635738518e-05,
+      "loss": 0.9533,
+      "mean_token_accuracy": 0.7138455748558045,
+      "num_tokens": 249182490.0,
+      "step": 100520
+    },
+    {
+      "epoch": 0.9217016594847346,
+      "learning_rate": 1.566150178784267e-05,
+      "loss": 0.961,
+      "mean_token_accuracy": 0.705112737417221,
+      "num_tokens": 249211041.0,
+      "step": 100530
+    },
+    {
+      "epoch": 0.9217933437242138,
+      "learning_rate": 1.5643164939946825e-05,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.7117710888385773,
+      "num_tokens": 249239833.0,
+      "step": 100540
+    },
+    {
+      "epoch": 0.9218850279636931,
+      "learning_rate": 1.5624828092050977e-05,
+      "loss": 0.9996,
+      "mean_token_accuracy": 0.6976643860340118,
+      "num_tokens": 249268502.0,
+      "step": 100550
+    },
+    {
+      "epoch": 0.9219767122031722,
+      "learning_rate": 1.5606491244155132e-05,
+      "loss": 0.9607,
+      "mean_token_accuracy": 0.7141707420349122,
+      "num_tokens": 249296653.0,
+      "step": 100560
+    },
+    {
+      "epoch": 0.9220683964426515,
+      "learning_rate": 1.5588154396259284e-05,
+      "loss": 0.9231,
+      "mean_token_accuracy": 0.7176756381988525,
+      "num_tokens": 249323642.0,
+      "step": 100570
+    },
+    {
+      "epoch": 0.9221600806821307,
+      "learning_rate": 1.556981754836344e-05,
+      "loss": 0.9516,
+      "mean_token_accuracy": 0.7097282767295837,
+      "num_tokens": 249352347.0,
+      "step": 100580
+    },
+    {
+      "epoch": 0.92225176492161,
+      "learning_rate": 1.555148070046759e-05,
+      "loss": 0.9485,
+      "mean_token_accuracy": 0.7109363079071045,
+      "num_tokens": 249381443.0,
+      "step": 100590
+    },
+    {
+      "epoch": 0.9223434491610892,
+      "learning_rate": 1.5533143852571743e-05,
+      "loss": 0.9792,
+      "mean_token_accuracy": 0.6996920287609101,
+      "num_tokens": 249409473.0,
+      "step": 100600
+    },
+    {
+      "epoch": 0.9224351334005685,
+      "learning_rate": 1.5514807004675898e-05,
+      "loss": 0.968,
+      "mean_token_accuracy": 0.7075730323791504,
+      "num_tokens": 249437197.0,
+      "step": 100610
+    },
+    {
+      "epoch": 0.9225268176400476,
+      "learning_rate": 1.549647015678005e-05,
+      "loss": 0.9445,
+      "mean_token_accuracy": 0.7086631119251251,
+      "num_tokens": 249464874.0,
+      "step": 100620
+    },
+    {
+      "epoch": 0.9226185018795269,
+      "learning_rate": 1.5478133308884205e-05,
+      "loss": 0.9476,
+      "mean_token_accuracy": 0.713670402765274,
+      "num_tokens": 249492726.0,
+      "step": 100630
+    },
+    {
+      "epoch": 0.9227101861190061,
+      "learning_rate": 1.5459796460988357e-05,
+      "loss": 0.9602,
+      "mean_token_accuracy": 0.7114905118942261,
+      "num_tokens": 249519502.0,
+      "step": 100640
+    },
+    {
+      "epoch": 0.9228018703584854,
+      "learning_rate": 1.544145961309251e-05,
+      "loss": 0.9772,
+      "mean_token_accuracy": 0.7055188179016113,
+      "num_tokens": 249546660.0,
+      "step": 100650
+    },
+    {
+      "epoch": 0.9228935545979646,
+      "learning_rate": 1.5423122765196664e-05,
+      "loss": 0.9126,
+      "mean_token_accuracy": 0.7191270232200623,
+      "num_tokens": 249574683.0,
+      "step": 100660
+    },
+    {
+      "epoch": 0.9229852388374439,
+      "learning_rate": 1.540478591730082e-05,
+      "loss": 0.9692,
+      "mean_token_accuracy": 0.7035593748092651,
+      "num_tokens": 249603450.0,
+      "step": 100670
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "learning_rate": 1.538644906940497e-05,
+      "loss": 0.9328,
+      "mean_token_accuracy": 0.7143905282020568,
+      "num_tokens": 249630873.0,
+      "step": 100680
+    },
+    {
+      "epoch": 0.9231686073164023,
+      "learning_rate": 1.5368112221509123e-05,
+      "loss": 0.9709,
+      "mean_token_accuracy": 0.7070143282413482,
+      "num_tokens": 249659458.0,
+      "step": 100690
+    },
+    {
+      "epoch": 0.9232602915558815,
+      "learning_rate": 1.5349775373613275e-05,
+      "loss": 0.9417,
+      "mean_token_accuracy": 0.7143168449401855,
+      "num_tokens": 249686948.0,
+      "step": 100700
+    },
+    {
+      "epoch": 0.9233519757953608,
+      "learning_rate": 1.533143852571743e-05,
+      "loss": 0.9728,
+      "mean_token_accuracy": 0.7064504802227021,
+      "num_tokens": 249714505.0,
+      "step": 100710
+    },
+    {
+      "epoch": 0.92344366003484,
+      "learning_rate": 1.5313101677821586e-05,
+      "loss": 0.9597,
+      "mean_token_accuracy": 0.7091737151145935,
+      "num_tokens": 249742298.0,
+      "step": 100720
+    },
+    {
+      "epoch": 0.9235353442743193,
+      "learning_rate": 1.5294764829925737e-05,
+      "loss": 0.9315,
+      "mean_token_accuracy": 0.7164255201816558,
+      "num_tokens": 249769413.0,
+      "step": 100730
+    },
+    {
+      "epoch": 0.9236270285137985,
+      "learning_rate": 1.527642798202989e-05,
+      "loss": 0.9574,
+      "mean_token_accuracy": 0.7113082051277161,
+      "num_tokens": 249797908.0,
+      "step": 100740
+    },
+    {
+      "epoch": 0.9237187127532777,
+      "learning_rate": 1.5258091134134043e-05,
+      "loss": 0.9512,
+      "mean_token_accuracy": 0.7109612941741943,
+      "num_tokens": 249825695.0,
+      "step": 100750
+    },
+    {
+      "epoch": 0.9238103969927569,
+      "learning_rate": 1.5239754286238198e-05,
+      "loss": 0.9481,
+      "mean_token_accuracy": 0.708267605304718,
+      "num_tokens": 249853543.0,
+      "step": 100760
+    },
+    {
+      "epoch": 0.9239020812322362,
+      "learning_rate": 1.522141743834235e-05,
+      "loss": 0.928,
+      "mean_token_accuracy": 0.7176929891109467,
+      "num_tokens": 249881404.0,
+      "step": 100770
+    },
+    {
+      "epoch": 0.9239937654717154,
+      "learning_rate": 1.5203080590446504e-05,
+      "loss": 0.9938,
+      "mean_token_accuracy": 0.6983058214187622,
+      "num_tokens": 249909254.0,
+      "step": 100780
+    },
+    {
+      "epoch": 0.9240854497111947,
+      "learning_rate": 1.5184743742550655e-05,
+      "loss": 0.9379,
+      "mean_token_accuracy": 0.7137408256530762,
+      "num_tokens": 249936898.0,
+      "step": 100790
+    },
+    {
+      "epoch": 0.9241771339506739,
+      "learning_rate": 1.5166406894654809e-05,
+      "loss": 0.9425,
+      "mean_token_accuracy": 0.716493946313858,
+      "num_tokens": 249964352.0,
+      "step": 100800
+    },
+    {
+      "epoch": 0.9242688181901532,
+      "learning_rate": 1.5148070046758964e-05,
+      "loss": 0.9716,
+      "mean_token_accuracy": 0.7079392313957215,
+      "num_tokens": 249992129.0,
+      "step": 100810
+    },
+    {
+      "epoch": 0.9243605024296323,
+      "learning_rate": 1.5129733198863116e-05,
+      "loss": 0.9452,
+      "mean_token_accuracy": 0.7150243043899536,
+      "num_tokens": 250020345.0,
+      "step": 100820
+    },
+    {
+      "epoch": 0.9244521866691116,
+      "learning_rate": 1.511139635096727e-05,
+      "loss": 0.9529,
+      "mean_token_accuracy": 0.709873503446579,
+      "num_tokens": 250048073.0,
+      "step": 100830
+    },
+    {
+      "epoch": 0.9245438709085908,
+      "learning_rate": 1.5093059503071422e-05,
+      "loss": 0.9575,
+      "mean_token_accuracy": 0.7116186380386352,
+      "num_tokens": 250076032.0,
+      "step": 100840
+    },
+    {
+      "epoch": 0.92463555514807,
+      "learning_rate": 1.5074722655175577e-05,
+      "loss": 0.9589,
+      "mean_token_accuracy": 0.7067624449729919,
+      "num_tokens": 250104165.0,
+      "step": 100850
+    },
+    {
+      "epoch": 0.9247272393875493,
+      "learning_rate": 1.505638580727973e-05,
+      "loss": 0.9384,
+      "mean_token_accuracy": 0.7108546733856201,
+      "num_tokens": 250132564.0,
+      "step": 100860
+    },
+    {
+      "epoch": 0.9248189236270286,
+      "learning_rate": 1.5038048959383882e-05,
+      "loss": 0.9858,
+      "mean_token_accuracy": 0.7066552817821503,
+      "num_tokens": 250160719.0,
+      "step": 100870
+    },
+    {
+      "epoch": 0.9249106078665077,
+      "learning_rate": 1.5019712111488036e-05,
+      "loss": 0.9681,
+      "mean_token_accuracy": 0.7042514204978942,
+      "num_tokens": 250188000.0,
+      "step": 100880
+    },
+    {
+      "epoch": 0.925002292105987,
+      "learning_rate": 1.5001375263592191e-05,
+      "loss": 0.9345,
+      "mean_token_accuracy": 0.7192918658256531,
+      "num_tokens": 250216120.0,
+      "step": 100890
+    },
+    {
+      "epoch": 0.9250939763454662,
+      "learning_rate": 1.4983038415696343e-05,
+      "loss": 0.9579,
+      "mean_token_accuracy": 0.7102409124374389,
+      "num_tokens": 250243980.0,
+      "step": 100900
+    },
+    {
+      "epoch": 0.9251856605849454,
+      "learning_rate": 1.4964701567800496e-05,
+      "loss": 0.9587,
+      "mean_token_accuracy": 0.7072044909000397,
+      "num_tokens": 250271061.0,
+      "step": 100910
+    },
+    {
+      "epoch": 0.9252773448244247,
+      "learning_rate": 1.4946364719904648e-05,
+      "loss": 0.9673,
+      "mean_token_accuracy": 0.7044660568237304,
+      "num_tokens": 250298980.0,
+      "step": 100920
+    },
+    {
+      "epoch": 0.925369029063904,
+      "learning_rate": 1.4928027872008802e-05,
+      "loss": 0.9743,
+      "mean_token_accuracy": 0.7075785040855408,
+      "num_tokens": 250327253.0,
+      "step": 100930
+    },
+    {
+      "epoch": 0.9254607133033832,
+      "learning_rate": 1.4909691024112957e-05,
+      "loss": 0.9603,
+      "mean_token_accuracy": 0.7101104617118835,
+      "num_tokens": 250355145.0,
+      "step": 100940
+    },
+    {
+      "epoch": 0.9255523975428623,
+      "learning_rate": 1.4891354176217109e-05,
+      "loss": 0.9607,
+      "mean_token_accuracy": 0.7078048586845398,
+      "num_tokens": 250382940.0,
+      "step": 100950
+    },
+    {
+      "epoch": 0.9256440817823416,
+      "learning_rate": 1.4873017328321263e-05,
+      "loss": 0.9457,
+      "mean_token_accuracy": 0.7161997199058533,
+      "num_tokens": 250410482.0,
+      "step": 100960
+    },
+    {
+      "epoch": 0.9257357660218208,
+      "learning_rate": 1.4854680480425414e-05,
+      "loss": 0.9501,
+      "mean_token_accuracy": 0.709575068950653,
+      "num_tokens": 250439187.0,
+      "step": 100970
+    },
+    {
+      "epoch": 0.9258274502613001,
+      "learning_rate": 1.483634363252957e-05,
+      "loss": 0.9398,
+      "mean_token_accuracy": 0.7123695969581604,
+      "num_tokens": 250467911.0,
+      "step": 100980
+    },
+    {
+      "epoch": 0.9259191345007793,
+      "learning_rate": 1.4818006784633723e-05,
+      "loss": 0.9891,
+      "mean_token_accuracy": 0.6986816644668579,
+      "num_tokens": 250496091.0,
+      "step": 100990
+    },
+    {
+      "epoch": 0.9260108187402586,
+      "learning_rate": 1.4799669936737875e-05,
+      "loss": 0.9414,
+      "mean_token_accuracy": 0.7116996347904205,
+      "num_tokens": 250524262.0,
+      "step": 101000
+    },
+    {
+      "epoch": 0.9261025029797377,
+      "learning_rate": 1.4781333088842029e-05,
+      "loss": 0.9509,
+      "mean_token_accuracy": 0.7090371906757355,
+      "num_tokens": 250551630.0,
+      "step": 101010
+    },
+    {
+      "epoch": 0.926194187219217,
+      "learning_rate": 1.476299624094618e-05,
+      "loss": 0.9518,
+      "mean_token_accuracy": 0.7161370575428009,
+      "num_tokens": 250579426.0,
+      "step": 101020
+    },
+    {
+      "epoch": 0.9262858714586962,
+      "learning_rate": 1.4744659393050336e-05,
+      "loss": 0.9847,
+      "mean_token_accuracy": 0.6985947608947753,
+      "num_tokens": 250607037.0,
+      "step": 101030
+    },
+    {
+      "epoch": 0.9263775556981755,
+      "learning_rate": 1.472632254515449e-05,
+      "loss": 0.9642,
+      "mean_token_accuracy": 0.7043582618236541,
+      "num_tokens": 250635378.0,
+      "step": 101040
+    },
+    {
+      "epoch": 0.9264692399376547,
+      "learning_rate": 1.4707985697258641e-05,
+      "loss": 0.9693,
+      "mean_token_accuracy": 0.7046302676200866,
+      "num_tokens": 250663272.0,
+      "step": 101050
+    },
+    {
+      "epoch": 0.926560924177134,
+      "learning_rate": 1.4689648849362795e-05,
+      "loss": 0.9678,
+      "mean_token_accuracy": 0.7033112704753876,
+      "num_tokens": 250690883.0,
+      "step": 101060
+    },
+    {
+      "epoch": 0.9266526084166132,
+      "learning_rate": 1.467131200146695e-05,
+      "loss": 0.9315,
+      "mean_token_accuracy": 0.7134780585765839,
+      "num_tokens": 250719225.0,
+      "step": 101070
+    },
+    {
+      "epoch": 0.9267442926560924,
+      "learning_rate": 1.4652975153571102e-05,
+      "loss": 0.9579,
+      "mean_token_accuracy": 0.7079670011997223,
+      "num_tokens": 250747249.0,
+      "step": 101080
+    },
+    {
+      "epoch": 0.9268359768955716,
+      "learning_rate": 1.4634638305675255e-05,
+      "loss": 0.9559,
+      "mean_token_accuracy": 0.706309849023819,
+      "num_tokens": 250775586.0,
+      "step": 101090
+    },
+    {
+      "epoch": 0.9269276611350509,
+      "learning_rate": 1.4616301457779407e-05,
+      "loss": 0.9361,
+      "mean_token_accuracy": 0.7131204187870026,
+      "num_tokens": 250803850.0,
+      "step": 101100
+    },
+    {
+      "epoch": 0.9270193453745301,
+      "learning_rate": 1.4597964609883561e-05,
+      "loss": 0.9526,
+      "mean_token_accuracy": 0.7098627209663391,
+      "num_tokens": 250832286.0,
+      "step": 101110
+    },
+    {
+      "epoch": 0.9271110296140094,
+      "learning_rate": 1.4579627761987716e-05,
+      "loss": 0.9641,
+      "mean_token_accuracy": 0.7075021147727967,
+      "num_tokens": 250861050.0,
+      "step": 101120
+    },
+    {
+      "epoch": 0.9272027138534886,
+      "learning_rate": 1.4561290914091868e-05,
+      "loss": 0.9672,
+      "mean_token_accuracy": 0.7055325388908387,
+      "num_tokens": 250889481.0,
+      "step": 101130
+    },
+    {
+      "epoch": 0.9272943980929678,
+      "learning_rate": 1.4542954066196022e-05,
+      "loss": 0.9304,
+      "mean_token_accuracy": 0.7094467341899872,
+      "num_tokens": 250916663.0,
+      "step": 101140
+    },
+    {
+      "epoch": 0.927386082332447,
+      "learning_rate": 1.4524617218300173e-05,
+      "loss": 0.9571,
+      "mean_token_accuracy": 0.7061554670333863,
+      "num_tokens": 250944766.0,
+      "step": 101150
+    },
+    {
+      "epoch": 0.9274777665719263,
+      "learning_rate": 1.4506280370404329e-05,
+      "loss": 0.9345,
+      "mean_token_accuracy": 0.7128243148326874,
+      "num_tokens": 250972471.0,
+      "step": 101160
+    },
+    {
+      "epoch": 0.9275694508114055,
+      "learning_rate": 1.4487943522508482e-05,
+      "loss": 0.9532,
+      "mean_token_accuracy": 0.7105517029762268,
+      "num_tokens": 250999623.0,
+      "step": 101170
+    },
+    {
+      "epoch": 0.9276611350508848,
+      "learning_rate": 1.4469606674612634e-05,
+      "loss": 0.9549,
+      "mean_token_accuracy": 0.7110942423343658,
+      "num_tokens": 251026727.0,
+      "step": 101180
+    },
+    {
+      "epoch": 0.927752819290364,
+      "learning_rate": 1.4451269826716788e-05,
+      "loss": 0.9394,
+      "mean_token_accuracy": 0.7118489742279053,
+      "num_tokens": 251054410.0,
+      "step": 101190
+    },
+    {
+      "epoch": 0.9278445035298433,
+      "learning_rate": 1.4432932978820943e-05,
+      "loss": 0.9458,
+      "mean_token_accuracy": 0.7079277634620667,
+      "num_tokens": 251082805.0,
+      "step": 101200
+    },
+    {
+      "epoch": 0.9279361877693224,
+      "learning_rate": 1.4414596130925095e-05,
+      "loss": 0.9575,
+      "mean_token_accuracy": 0.7132249712944031,
+      "num_tokens": 251112012.0,
+      "step": 101210
+    },
+    {
+      "epoch": 0.9280278720088017,
+      "learning_rate": 1.4396259283029248e-05,
+      "loss": 0.9675,
+      "mean_token_accuracy": 0.7074493169784546,
+      "num_tokens": 251139131.0,
+      "step": 101220
+    },
+    {
+      "epoch": 0.9281195562482809,
+      "learning_rate": 1.43779224351334e-05,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.7088360667228699,
+      "num_tokens": 251167332.0,
+      "step": 101230
+    },
+    {
+      "epoch": 0.9282112404877602,
+      "learning_rate": 1.4359585587237554e-05,
+      "loss": 0.9377,
+      "mean_token_accuracy": 0.7143978297710418,
+      "num_tokens": 251194765.0,
+      "step": 101240
+    },
+    {
+      "epoch": 0.9283029247272394,
+      "learning_rate": 1.4341248739341709e-05,
+      "loss": 0.9561,
+      "mean_token_accuracy": 0.7093397915363312,
+      "num_tokens": 251222159.0,
+      "step": 101250
+    },
+    {
+      "epoch": 0.9283946089667187,
+      "learning_rate": 1.4322911891445861e-05,
+      "loss": 0.9245,
+      "mean_token_accuracy": 0.7178255259990692,
+      "num_tokens": 251249107.0,
+      "step": 101260
+    },
+    {
+      "epoch": 0.9284862932061978,
+      "learning_rate": 1.4304575043550015e-05,
+      "loss": 0.9749,
+      "mean_token_accuracy": 0.7051239430904388,
+      "num_tokens": 251277101.0,
+      "step": 101270
+    },
+    {
+      "epoch": 0.928577977445677,
+      "learning_rate": 1.4286238195654166e-05,
+      "loss": 0.9612,
+      "mean_token_accuracy": 0.7061001837253571,
+      "num_tokens": 251304552.0,
+      "step": 101280
+    },
+    {
+      "epoch": 0.9286696616851563,
+      "learning_rate": 1.4267901347758322e-05,
+      "loss": 0.9647,
+      "mean_token_accuracy": 0.7067552268505096,
+      "num_tokens": 251332198.0,
+      "step": 101290
+    },
+    {
+      "epoch": 0.9287613459246356,
+      "learning_rate": 1.4249564499862475e-05,
+      "loss": 0.9469,
+      "mean_token_accuracy": 0.7085455417633056,
+      "num_tokens": 251360424.0,
+      "step": 101300
+    },
+    {
+      "epoch": 0.9288530301641148,
+      "learning_rate": 1.4231227651966627e-05,
+      "loss": 0.9734,
+      "mean_token_accuracy": 0.7017104625701904,
+      "num_tokens": 251388581.0,
+      "step": 101310
+    },
+    {
+      "epoch": 0.928944714403594,
+      "learning_rate": 1.421289080407078e-05,
+      "loss": 0.9634,
+      "mean_token_accuracy": 0.7036072313785553,
+      "num_tokens": 251416086.0,
+      "step": 101320
+    },
+    {
+      "epoch": 0.9290363986430733,
+      "learning_rate": 1.4194553956174932e-05,
+      "loss": 0.9925,
+      "mean_token_accuracy": 0.7028466284275054,
+      "num_tokens": 251444125.0,
+      "step": 101330
+    },
+    {
+      "epoch": 0.9291280828825524,
+      "learning_rate": 1.4176217108279088e-05,
+      "loss": 0.9431,
+      "mean_token_accuracy": 0.7126602232456207,
+      "num_tokens": 251472825.0,
+      "step": 101340
+    },
+    {
+      "epoch": 0.9292197671220317,
+      "learning_rate": 1.4157880260383241e-05,
+      "loss": 0.9588,
+      "mean_token_accuracy": 0.7097715973854065,
+      "num_tokens": 251500351.0,
+      "step": 101350
+    },
+    {
+      "epoch": 0.9293114513615109,
+      "learning_rate": 1.4139543412487393e-05,
+      "loss": 0.963,
+      "mean_token_accuracy": 0.7072455525398255,
+      "num_tokens": 251528803.0,
+      "step": 101360
+    },
+    {
+      "epoch": 0.9294031356009902,
+      "learning_rate": 1.4121206564591547e-05,
+      "loss": 0.9476,
+      "mean_token_accuracy": 0.7109055340290069,
+      "num_tokens": 251557303.0,
+      "step": 101370
+    },
+    {
+      "epoch": 0.9294948198404694,
+      "learning_rate": 1.4102869716695702e-05,
+      "loss": 0.9617,
+      "mean_token_accuracy": 0.7036132216453552,
+      "num_tokens": 251585660.0,
+      "step": 101380
+    },
+    {
+      "epoch": 0.9295865040799487,
+      "learning_rate": 1.4084532868799854e-05,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.7148278295993805,
+      "num_tokens": 251614408.0,
+      "step": 101390
+    },
+    {
+      "epoch": 0.929678188319428,
+      "learning_rate": 1.4066196020904007e-05,
+      "loss": 0.9454,
+      "mean_token_accuracy": 0.7096316576004028,
+      "num_tokens": 251642152.0,
+      "step": 101400
+    },
+    {
+      "epoch": 0.9297698725589071,
+      "learning_rate": 1.404785917300816e-05,
+      "loss": 0.9912,
+      "mean_token_accuracy": 0.7031850814819336,
+      "num_tokens": 251670052.0,
+      "step": 101410
+    },
+    {
+      "epoch": 0.9298615567983863,
+      "learning_rate": 1.4029522325112313e-05,
+      "loss": 0.9436,
+      "mean_token_accuracy": 0.7077620148658752,
+      "num_tokens": 251698642.0,
+      "step": 101420
+    },
+    {
+      "epoch": 0.9299532410378656,
+      "learning_rate": 1.4011185477216468e-05,
+      "loss": 0.97,
+      "mean_token_accuracy": 0.701517003774643,
+      "num_tokens": 251727569.0,
+      "step": 101430
+    },
+    {
+      "epoch": 0.9300449252773448,
+      "learning_rate": 1.399284862932062e-05,
+      "loss": 0.9476,
+      "mean_token_accuracy": 0.7125502228736877,
+      "num_tokens": 251755531.0,
+      "step": 101440
+    },
+    {
+      "epoch": 0.9301366095168241,
+      "learning_rate": 1.3974511781424774e-05,
+      "loss": 0.9625,
+      "mean_token_accuracy": 0.708078944683075,
+      "num_tokens": 251782304.0,
+      "step": 101450
+    },
+    {
+      "epoch": 0.9302282937563033,
+      "learning_rate": 1.3956174933528925e-05,
+      "loss": 0.9653,
+      "mean_token_accuracy": 0.7106375336647034,
+      "num_tokens": 251809774.0,
+      "step": 101460
+    },
+    {
+      "epoch": 0.9303199779957825,
+      "learning_rate": 1.393783808563308e-05,
+      "loss": 0.9236,
+      "mean_token_accuracy": 0.7176149249076843,
+      "num_tokens": 251837524.0,
+      "step": 101470
+    },
+    {
+      "epoch": 0.9304116622352617,
+      "learning_rate": 1.3919501237737234e-05,
+      "loss": 0.9498,
+      "mean_token_accuracy": 0.7041159570217133,
+      "num_tokens": 251865288.0,
+      "step": 101480
+    },
+    {
+      "epoch": 0.930503346474741,
+      "learning_rate": 1.3901164389841386e-05,
+      "loss": 0.9685,
+      "mean_token_accuracy": 0.709986287355423,
+      "num_tokens": 251893657.0,
+      "step": 101490
+    },
+    {
+      "epoch": 0.9305950307142202,
+      "learning_rate": 1.388282754194554e-05,
+      "loss": 0.9724,
+      "mean_token_accuracy": 0.7070633471012115,
+      "num_tokens": 251922286.0,
+      "step": 101500
+    },
+    {
+      "epoch": 0.9306867149536995,
+      "learning_rate": 1.3864490694049692e-05,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.7094040215015411,
+      "num_tokens": 251949615.0,
+      "step": 101510
+    },
+    {
+      "epoch": 0.9307783991931787,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.9574,
+      "mean_token_accuracy": 0.7079304873943328,
+      "num_tokens": 251977913.0,
+      "step": 101520
+    },
+    {
+      "epoch": 0.930870083432658,
+      "learning_rate": 1.3827816998258e-05,
+      "loss": 0.9481,
+      "mean_token_accuracy": 0.7147959291934967,
+      "num_tokens": 252006111.0,
+      "step": 101530
+    },
+    {
+      "epoch": 0.9309617676721371,
+      "learning_rate": 1.3809480150362152e-05,
+      "loss": 0.9494,
+      "mean_token_accuracy": 0.7095328748226166,
+      "num_tokens": 252033972.0,
+      "step": 101540
+    },
+    {
+      "epoch": 0.9310534519116164,
+      "learning_rate": 1.3791143302466306e-05,
+      "loss": 0.9193,
+      "mean_token_accuracy": 0.7209644436836242,
+      "num_tokens": 252061731.0,
+      "step": 101550
+    },
+    {
+      "epoch": 0.9311451361510956,
+      "learning_rate": 1.3772806454570461e-05,
+      "loss": 0.9397,
+      "mean_token_accuracy": 0.7114929676055908,
+      "num_tokens": 252089921.0,
+      "step": 101560
+    },
+    {
+      "epoch": 0.9312368203905749,
+      "learning_rate": 1.3754469606674613e-05,
+      "loss": 0.9446,
+      "mean_token_accuracy": 0.7108826100826263,
+      "num_tokens": 252117592.0,
+      "step": 101570
+    },
+    {
+      "epoch": 0.9313285046300541,
+      "learning_rate": 1.3736132758778766e-05,
+      "loss": 0.9357,
+      "mean_token_accuracy": 0.714998471736908,
+      "num_tokens": 252145036.0,
+      "step": 101580
+    },
+    {
+      "epoch": 0.9314201888695334,
+      "learning_rate": 1.3717795910882918e-05,
+      "loss": 0.9509,
+      "mean_token_accuracy": 0.7109091997146606,
+      "num_tokens": 252173369.0,
+      "step": 101590
+    },
+    {
+      "epoch": 0.9315118731090125,
+      "learning_rate": 1.3699459062987075e-05,
+      "loss": 0.9446,
+      "mean_token_accuracy": 0.7140315294265747,
+      "num_tokens": 252201094.0,
+      "step": 101600
+    },
+    {
+      "epoch": 0.9316035573484918,
+      "learning_rate": 1.3681122215091227e-05,
+      "loss": 0.9568,
+      "mean_token_accuracy": 0.707772308588028,
+      "num_tokens": 252228076.0,
+      "step": 101610
+    },
+    {
+      "epoch": 0.931695241587971,
+      "learning_rate": 1.3662785367195379e-05,
+      "loss": 0.9364,
+      "mean_token_accuracy": 0.7147785305976868,
+      "num_tokens": 252255759.0,
+      "step": 101620
+    },
+    {
+      "epoch": 0.9317869258274503,
+      "learning_rate": 1.3644448519299533e-05,
+      "loss": 0.9598,
+      "mean_token_accuracy": 0.7093723058700562,
+      "num_tokens": 252283682.0,
+      "step": 101630
+    },
+    {
+      "epoch": 0.9318786100669295,
+      "learning_rate": 1.3626111671403684e-05,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.7094361782073975,
+      "num_tokens": 252311610.0,
+      "step": 101640
+    },
+    {
+      "epoch": 0.9319702943064088,
+      "learning_rate": 1.3607774823507841e-05,
+      "loss": 0.9531,
+      "mean_token_accuracy": 0.7079195976257324,
+      "num_tokens": 252339223.0,
+      "step": 101650
+    },
+    {
+      "epoch": 0.932061978545888,
+      "learning_rate": 1.3589437975611993e-05,
+      "loss": 0.9299,
+      "mean_token_accuracy": 0.7151159167289733,
+      "num_tokens": 252367776.0,
+      "step": 101660
+    },
+    {
+      "epoch": 0.9321536627853672,
+      "learning_rate": 1.3571101127716145e-05,
+      "loss": 0.9566,
+      "mean_token_accuracy": 0.7099621832370758,
+      "num_tokens": 252395613.0,
+      "step": 101670
+    },
+    {
+      "epoch": 0.9322453470248464,
+      "learning_rate": 1.3552764279820299e-05,
+      "loss": 0.987,
+      "mean_token_accuracy": 0.7034406185150146,
+      "num_tokens": 252423526.0,
+      "step": 101680
+    },
+    {
+      "epoch": 0.9323370312643257,
+      "learning_rate": 1.3534427431924454e-05,
+      "loss": 0.9181,
+      "mean_token_accuracy": 0.7155993700027465,
+      "num_tokens": 252451375.0,
+      "step": 101690
+    },
+    {
+      "epoch": 0.9324287155038049,
+      "learning_rate": 1.3516090584028607e-05,
+      "loss": 0.9643,
+      "mean_token_accuracy": 0.712075161933899,
+      "num_tokens": 252479086.0,
+      "step": 101700
+    },
+    {
+      "epoch": 0.9325203997432842,
+      "learning_rate": 1.349775373613276e-05,
+      "loss": 0.9352,
+      "mean_token_accuracy": 0.7157521069049835,
+      "num_tokens": 252507028.0,
+      "step": 101710
+    },
+    {
+      "epoch": 0.9326120839827634,
+      "learning_rate": 1.3479416888236911e-05,
+      "loss": 0.961,
+      "mean_token_accuracy": 0.7072499692440033,
+      "num_tokens": 252534821.0,
+      "step": 101720
+    },
+    {
+      "epoch": 0.9327037682222425,
+      "learning_rate": 1.3461080040341065e-05,
+      "loss": 0.9684,
+      "mean_token_accuracy": 0.7069429457187653,
+      "num_tokens": 252563456.0,
+      "step": 101730
+    },
+    {
+      "epoch": 0.9327954524617218,
+      "learning_rate": 1.344274319244522e-05,
+      "loss": 0.9268,
+      "mean_token_accuracy": 0.7155016183853149,
+      "num_tokens": 252590277.0,
+      "step": 101740
+    },
+    {
+      "epoch": 0.932887136701201,
+      "learning_rate": 1.3424406344549374e-05,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.710516220331192,
+      "num_tokens": 252617635.0,
+      "step": 101750
+    },
+    {
+      "epoch": 0.9329788209406803,
+      "learning_rate": 1.3406069496653525e-05,
+      "loss": 0.9457,
+      "mean_token_accuracy": 0.7113356769084931,
+      "num_tokens": 252645654.0,
+      "step": 101760
+    },
+    {
+      "epoch": 0.9330705051801595,
+      "learning_rate": 1.3387732648757679e-05,
+      "loss": 0.9324,
+      "mean_token_accuracy": 0.7156642973423004,
+      "num_tokens": 252674099.0,
+      "step": 101770
+    },
+    {
+      "epoch": 0.9331621894196388,
+      "learning_rate": 1.3369395800861834e-05,
+      "loss": 0.9484,
+      "mean_token_accuracy": 0.7076420068740845,
+      "num_tokens": 252702306.0,
+      "step": 101780
+    },
+    {
+      "epoch": 0.933253873659118,
+      "learning_rate": 1.3351058952965986e-05,
+      "loss": 0.9529,
+      "mean_token_accuracy": 0.7138663232326508,
+      "num_tokens": 252729853.0,
+      "step": 101790
+    },
+    {
+      "epoch": 0.9333455578985972,
+      "learning_rate": 1.333272210507014e-05,
+      "loss": 0.9299,
+      "mean_token_accuracy": 0.7191498577594757,
+      "num_tokens": 252756698.0,
+      "step": 101800
+    },
+    {
+      "epoch": 0.9334372421380764,
+      "learning_rate": 1.3314385257174292e-05,
+      "loss": 0.9384,
+      "mean_token_accuracy": 0.7146835505962372,
+      "num_tokens": 252783938.0,
+      "step": 101810
+    },
+    {
+      "epoch": 0.9335289263775557,
+      "learning_rate": 1.3296048409278445e-05,
+      "loss": 0.9572,
+      "mean_token_accuracy": 0.7084337413311005,
+      "num_tokens": 252811611.0,
+      "step": 101820
+    },
+    {
+      "epoch": 0.9336206106170349,
+      "learning_rate": 1.32777115613826e-05,
+      "loss": 0.9605,
+      "mean_token_accuracy": 0.7086575090885162,
+      "num_tokens": 252839895.0,
+      "step": 101830
+    },
+    {
+      "epoch": 0.9337122948565142,
+      "learning_rate": 1.3259374713486752e-05,
+      "loss": 0.9679,
+      "mean_token_accuracy": 0.7077599704265595,
+      "num_tokens": 252867833.0,
+      "step": 101840
+    },
+    {
+      "epoch": 0.9338039790959934,
+      "learning_rate": 1.3241037865590906e-05,
+      "loss": 0.9604,
+      "mean_token_accuracy": 0.71173597574234,
+      "num_tokens": 252895954.0,
+      "step": 101850
+    },
+    {
+      "epoch": 0.9338956633354726,
+      "learning_rate": 1.3222701017695058e-05,
+      "loss": 0.9773,
+      "mean_token_accuracy": 0.7013697922229767,
+      "num_tokens": 252924627.0,
+      "step": 101860
+    },
+    {
+      "epoch": 0.9339873475749518,
+      "learning_rate": 1.3204364169799213e-05,
+      "loss": 0.9359,
+      "mean_token_accuracy": 0.7167365610599518,
+      "num_tokens": 252953616.0,
+      "step": 101870
+    },
+    {
+      "epoch": 0.9340790318144311,
+      "learning_rate": 1.3186027321903367e-05,
+      "loss": 0.9209,
+      "mean_token_accuracy": 0.7197913587093353,
+      "num_tokens": 252981384.0,
+      "step": 101880
+    },
+    {
+      "epoch": 0.9341707160539103,
+      "learning_rate": 1.3167690474007518e-05,
+      "loss": 0.9075,
+      "mean_token_accuracy": 0.7227952063083649,
+      "num_tokens": 253009639.0,
+      "step": 101890
+    },
+    {
+      "epoch": 0.9342624002933896,
+      "learning_rate": 1.3149353626111672e-05,
+      "loss": 0.9629,
+      "mean_token_accuracy": 0.7085484147071839,
+      "num_tokens": 253036932.0,
+      "step": 101900
+    },
+    {
+      "epoch": 0.9343540845328688,
+      "learning_rate": 1.3131016778215827e-05,
+      "loss": 0.934,
+      "mean_token_accuracy": 0.7167655944824218,
+      "num_tokens": 253064856.0,
+      "step": 101910
+    },
+    {
+      "epoch": 0.9344457687723481,
+      "learning_rate": 1.3112679930319979e-05,
+      "loss": 0.9563,
+      "mean_token_accuracy": 0.7140512526035309,
+      "num_tokens": 253092877.0,
+      "step": 101920
+    },
+    {
+      "epoch": 0.9345374530118272,
+      "learning_rate": 1.3094343082424133e-05,
+      "loss": 0.9385,
+      "mean_token_accuracy": 0.711508572101593,
+      "num_tokens": 253120696.0,
+      "step": 101930
+    },
+    {
+      "epoch": 0.9346291372513065,
+      "learning_rate": 1.3076006234528284e-05,
+      "loss": 0.9281,
+      "mean_token_accuracy": 0.7137085556983948,
+      "num_tokens": 253148616.0,
+      "step": 101940
+    },
+    {
+      "epoch": 0.9347208214907857,
+      "learning_rate": 1.3057669386632438e-05,
+      "loss": 0.9541,
+      "mean_token_accuracy": 0.7092041730880737,
+      "num_tokens": 253176685.0,
+      "step": 101950
+    },
+    {
+      "epoch": 0.934812505730265,
+      "learning_rate": 1.3039332538736593e-05,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.7092980742454529,
+      "num_tokens": 253204508.0,
+      "step": 101960
+    },
+    {
+      "epoch": 0.9349041899697442,
+      "learning_rate": 1.3020995690840745e-05,
+      "loss": 0.9491,
+      "mean_token_accuracy": 0.7087125301361084,
+      "num_tokens": 253232864.0,
+      "step": 101970
+    },
+    {
+      "epoch": 0.9349958742092235,
+      "learning_rate": 1.3002658842944899e-05,
+      "loss": 0.9567,
+      "mean_token_accuracy": 0.708165442943573,
+      "num_tokens": 253260119.0,
+      "step": 101980
+    },
+    {
+      "epoch": 0.9350875584487026,
+      "learning_rate": 1.298432199504905e-05,
+      "loss": 0.9617,
+      "mean_token_accuracy": 0.7049558699131012,
+      "num_tokens": 253287693.0,
+      "step": 101990
+    },
+    {
+      "epoch": 0.9351792426881819,
+      "learning_rate": 1.2965985147153206e-05,
+      "loss": 0.9632,
+      "mean_token_accuracy": 0.7053045332431793,
+      "num_tokens": 253316909.0,
+      "step": 102000
+    },
+    {
+      "epoch": 0.9352709269276611,
+      "learning_rate": 1.294764829925736e-05,
+      "loss": 0.9534,
+      "mean_token_accuracy": 0.7133051753044128,
+      "num_tokens": 253344304.0,
+      "step": 102010
+    },
+    {
+      "epoch": 0.9353626111671404,
+      "learning_rate": 1.2929311451361511e-05,
+      "loss": 0.9526,
+      "mean_token_accuracy": 0.7138311803340912,
+      "num_tokens": 253372145.0,
+      "step": 102020
+    },
+    {
+      "epoch": 0.9354542954066196,
+      "learning_rate": 1.2910974603465665e-05,
+      "loss": 0.9742,
+      "mean_token_accuracy": 0.7077650666236878,
+      "num_tokens": 253400686.0,
+      "step": 102030
+    },
+    {
+      "epoch": 0.9355459796460989,
+      "learning_rate": 1.2892637755569817e-05,
+      "loss": 0.9608,
+      "mean_token_accuracy": 0.7084186375141144,
+      "num_tokens": 253429197.0,
+      "step": 102040
+    },
+    {
+      "epoch": 0.9356376638855781,
+      "learning_rate": 1.2874300907673972e-05,
+      "loss": 0.9211,
+      "mean_token_accuracy": 0.7175590336322785,
+      "num_tokens": 253457161.0,
+      "step": 102050
+    },
+    {
+      "epoch": 0.9357293481250573,
+      "learning_rate": 1.2855964059778126e-05,
+      "loss": 0.9631,
+      "mean_token_accuracy": 0.7109842479228974,
+      "num_tokens": 253485237.0,
+      "step": 102060
+    },
+    {
+      "epoch": 0.9358210323645365,
+      "learning_rate": 1.2837627211882277e-05,
+      "loss": 0.9629,
+      "mean_token_accuracy": 0.7063953459262848,
+      "num_tokens": 253514390.0,
+      "step": 102070
+    },
+    {
+      "epoch": 0.9359127166040158,
+      "learning_rate": 1.2819290363986431e-05,
+      "loss": 0.9619,
+      "mean_token_accuracy": 0.7067668616771698,
+      "num_tokens": 253542531.0,
+      "step": 102080
+    },
+    {
+      "epoch": 0.936004400843495,
+      "learning_rate": 1.2800953516090586e-05,
+      "loss": 0.9593,
+      "mean_token_accuracy": 0.7030080020427704,
+      "num_tokens": 253569624.0,
+      "step": 102090
+    },
+    {
+      "epoch": 0.9360960850829743,
+      "learning_rate": 1.2782616668194738e-05,
+      "loss": 0.9374,
+      "mean_token_accuracy": 0.7127783238887787,
+      "num_tokens": 253596874.0,
+      "step": 102100
+    },
+    {
+      "epoch": 0.9361877693224535,
+      "learning_rate": 1.2764279820298892e-05,
+      "loss": 0.9839,
+      "mean_token_accuracy": 0.6990141153335572,
+      "num_tokens": 253624378.0,
+      "step": 102110
+    },
+    {
+      "epoch": 0.9362794535619327,
+      "learning_rate": 1.2745942972403044e-05,
+      "loss": 0.9601,
+      "mean_token_accuracy": 0.7078054070472717,
+      "num_tokens": 253651947.0,
+      "step": 102120
+    },
+    {
+      "epoch": 0.9363711378014119,
+      "learning_rate": 1.2727606124507197e-05,
+      "loss": 0.9349,
+      "mean_token_accuracy": 0.7184051156044007,
+      "num_tokens": 253680296.0,
+      "step": 102130
+    },
+    {
+      "epoch": 0.9364628220408912,
+      "learning_rate": 1.2709269276611352e-05,
+      "loss": 0.9666,
+      "mean_token_accuracy": 0.7019192457199097,
+      "num_tokens": 253708585.0,
+      "step": 102140
+    },
+    {
+      "epoch": 0.9365545062803704,
+      "learning_rate": 1.2690932428715504e-05,
+      "loss": 0.9409,
+      "mean_token_accuracy": 0.713731199502945,
+      "num_tokens": 253735826.0,
+      "step": 102150
+    },
+    {
+      "epoch": 0.9366461905198497,
+      "learning_rate": 1.2672595580819658e-05,
+      "loss": 0.9334,
+      "mean_token_accuracy": 0.7147872805595398,
+      "num_tokens": 253762513.0,
+      "step": 102160
+    },
+    {
+      "epoch": 0.9367378747593289,
+      "learning_rate": 1.265425873292381e-05,
+      "loss": 0.9461,
+      "mean_token_accuracy": 0.7095882177352906,
+      "num_tokens": 253790226.0,
+      "step": 102170
+    },
+    {
+      "epoch": 0.9368295589988082,
+      "learning_rate": 1.2635921885027965e-05,
+      "loss": 0.9557,
+      "mean_token_accuracy": 0.7096666812896728,
+      "num_tokens": 253818231.0,
+      "step": 102180
+    },
+    {
+      "epoch": 0.9369212432382873,
+      "learning_rate": 1.2617585037132118e-05,
+      "loss": 0.9768,
+      "mean_token_accuracy": 0.706231164932251,
+      "num_tokens": 253846390.0,
+      "step": 102190
+    },
+    {
+      "epoch": 0.9370129274777665,
+      "learning_rate": 1.259924818923627e-05,
+      "loss": 0.9545,
+      "mean_token_accuracy": 0.7078743278980255,
+      "num_tokens": 253874150.0,
+      "step": 102200
+    },
+    {
+      "epoch": 0.9371046117172458,
+      "learning_rate": 1.2580911341340424e-05,
+      "loss": 0.9565,
+      "mean_token_accuracy": 0.7112432301044465,
+      "num_tokens": 253902431.0,
+      "step": 102210
+    },
+    {
+      "epoch": 0.937196295956725,
+      "learning_rate": 1.2562574493444579e-05,
+      "loss": 0.93,
+      "mean_token_accuracy": 0.7090595901012421,
+      "num_tokens": 253930473.0,
+      "step": 102220
+    },
+    {
+      "epoch": 0.9372879801962043,
+      "learning_rate": 1.2544237645548731e-05,
+      "loss": 0.9554,
+      "mean_token_accuracy": 0.7088334858417511,
+      "num_tokens": 253958093.0,
+      "step": 102230
+    },
+    {
+      "epoch": 0.9373796644356835,
+      "learning_rate": 1.2525900797652885e-05,
+      "loss": 0.9422,
+      "mean_token_accuracy": 0.7113230943679809,
+      "num_tokens": 253985350.0,
+      "step": 102240
+    },
+    {
+      "epoch": 0.9374713486751627,
+      "learning_rate": 1.2507563949757036e-05,
+      "loss": 0.9302,
+      "mean_token_accuracy": 0.7184086322784424,
+      "num_tokens": 254013143.0,
+      "step": 102250
+    },
+    {
+      "epoch": 0.9375630329146419,
+      "learning_rate": 1.2489227101861192e-05,
+      "loss": 0.9352,
+      "mean_token_accuracy": 0.7166255116462708,
+      "num_tokens": 254040989.0,
+      "step": 102260
+    },
+    {
+      "epoch": 0.9376547171541212,
+      "learning_rate": 1.2470890253965344e-05,
+      "loss": 0.9744,
+      "mean_token_accuracy": 0.7072222352027893,
+      "num_tokens": 254069361.0,
+      "step": 102270
+    },
+    {
+      "epoch": 0.9377464013936004,
+      "learning_rate": 1.2452553406069497e-05,
+      "loss": 0.9537,
+      "mean_token_accuracy": 0.7117165207862854,
+      "num_tokens": 254096868.0,
+      "step": 102280
+    },
+    {
+      "epoch": 0.9378380856330797,
+      "learning_rate": 1.243421655817365e-05,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.7056435406208038,
+      "num_tokens": 254124369.0,
+      "step": 102290
+    },
+    {
+      "epoch": 0.9379297698725589,
+      "learning_rate": 1.2415879710277804e-05,
+      "loss": 0.9713,
+      "mean_token_accuracy": 0.7107076525688172,
+      "num_tokens": 254152127.0,
+      "step": 102300
+    },
+    {
+      "epoch": 0.9380214541120382,
+      "learning_rate": 1.2397542862381958e-05,
+      "loss": 0.9778,
+      "mean_token_accuracy": 0.7051454901695251,
+      "num_tokens": 254180863.0,
+      "step": 102310
+    },
+    {
+      "epoch": 0.9381131383515173,
+      "learning_rate": 1.2379206014486111e-05,
+      "loss": 0.9682,
+      "mean_token_accuracy": 0.7083398342132569,
+      "num_tokens": 254209016.0,
+      "step": 102320
+    },
+    {
+      "epoch": 0.9382048225909966,
+      "learning_rate": 1.2360869166590263e-05,
+      "loss": 0.9294,
+      "mean_token_accuracy": 0.7196935713291168,
+      "num_tokens": 254236176.0,
+      "step": 102330
+    },
+    {
+      "epoch": 0.9382965068304758,
+      "learning_rate": 1.2342532318694417e-05,
+      "loss": 0.9597,
+      "mean_token_accuracy": 0.7071610927581787,
+      "num_tokens": 254263966.0,
+      "step": 102340
+    },
+    {
+      "epoch": 0.9383881910699551,
+      "learning_rate": 1.232419547079857e-05,
+      "loss": 0.9621,
+      "mean_token_accuracy": 0.7054771661758423,
+      "num_tokens": 254291939.0,
+      "step": 102350
+    },
+    {
+      "epoch": 0.9384798753094343,
+      "learning_rate": 1.2305858622902724e-05,
+      "loss": 0.9584,
+      "mean_token_accuracy": 0.7100796222686767,
+      "num_tokens": 254319702.0,
+      "step": 102360
+    },
+    {
+      "epoch": 0.9385715595489136,
+      "learning_rate": 1.2287521775006877e-05,
+      "loss": 0.9913,
+      "mean_token_accuracy": 0.7060858964920044,
+      "num_tokens": 254347511.0,
+      "step": 102370
+    },
+    {
+      "epoch": 0.9386632437883927,
+      "learning_rate": 1.226918492711103e-05,
+      "loss": 0.9714,
+      "mean_token_accuracy": 0.7083201467990875,
+      "num_tokens": 254376053.0,
+      "step": 102380
+    },
+    {
+      "epoch": 0.938754928027872,
+      "learning_rate": 1.2250848079215185e-05,
+      "loss": 0.973,
+      "mean_token_accuracy": 0.7036384522914887,
+      "num_tokens": 254404048.0,
+      "step": 102390
+    },
+    {
+      "epoch": 0.9388466122673512,
+      "learning_rate": 1.2232511231319336e-05,
+      "loss": 0.9503,
+      "mean_token_accuracy": 0.7131141543388366,
+      "num_tokens": 254431605.0,
+      "step": 102400
+    },
+    {
+      "epoch": 0.9389382965068305,
+      "learning_rate": 1.221417438342349e-05,
+      "loss": 0.9388,
+      "mean_token_accuracy": 0.7128082752227783,
+      "num_tokens": 254459806.0,
+      "step": 102410
+    },
+    {
+      "epoch": 0.9390299807463097,
+      "learning_rate": 1.2195837535527644e-05,
+      "loss": 0.9712,
+      "mean_token_accuracy": 0.7045019030570984,
+      "num_tokens": 254487273.0,
+      "step": 102420
+    },
+    {
+      "epoch": 0.939121664985789,
+      "learning_rate": 1.2177500687631797e-05,
+      "loss": 0.9457,
+      "mean_token_accuracy": 0.7112976431846618,
+      "num_tokens": 254515332.0,
+      "step": 102430
+    },
+    {
+      "epoch": 0.9392133492252682,
+      "learning_rate": 1.215916383973595e-05,
+      "loss": 0.93,
+      "mean_token_accuracy": 0.7165105700492859,
+      "num_tokens": 254543712.0,
+      "step": 102440
+    },
+    {
+      "epoch": 0.9393050334647474,
+      "learning_rate": 1.2140826991840103e-05,
+      "loss": 0.9363,
+      "mean_token_accuracy": 0.7132966458797455,
+      "num_tokens": 254572780.0,
+      "step": 102450
+    },
+    {
+      "epoch": 0.9393967177042266,
+      "learning_rate": 1.2122490143944256e-05,
+      "loss": 0.9426,
+      "mean_token_accuracy": 0.7113722503185272,
+      "num_tokens": 254600234.0,
+      "step": 102460
+    },
+    {
+      "epoch": 0.9394884019437059,
+      "learning_rate": 1.210415329604841e-05,
+      "loss": 0.9723,
+      "mean_token_accuracy": 0.706432455778122,
+      "num_tokens": 254628822.0,
+      "step": 102470
+    },
+    {
+      "epoch": 0.9395800861831851,
+      "learning_rate": 1.2085816448152563e-05,
+      "loss": 0.9545,
+      "mean_token_accuracy": 0.7113306283950805,
+      "num_tokens": 254657481.0,
+      "step": 102480
+    },
+    {
+      "epoch": 0.9396717704226644,
+      "learning_rate": 1.2067479600256717e-05,
+      "loss": 0.9716,
+      "mean_token_accuracy": 0.7082769691944122,
+      "num_tokens": 254684772.0,
+      "step": 102490
+    },
+    {
+      "epoch": 0.9397634546621436,
+      "learning_rate": 1.204914275236087e-05,
+      "loss": 0.9264,
+      "mean_token_accuracy": 0.718409514427185,
+      "num_tokens": 254711876.0,
+      "step": 102500
+    },
+    {
+      "epoch": 0.9398551389016229,
+      "learning_rate": 1.2030805904465022e-05,
+      "loss": 0.9726,
+      "mean_token_accuracy": 0.6997957825660706,
+      "num_tokens": 254740231.0,
+      "step": 102510
+    },
+    {
+      "epoch": 0.939946823141102,
+      "learning_rate": 1.2012469056569178e-05,
+      "loss": 0.9299,
+      "mean_token_accuracy": 0.7119948267936707,
+      "num_tokens": 254768375.0,
+      "step": 102520
+    },
+    {
+      "epoch": 0.9400385073805813,
+      "learning_rate": 1.199413220867333e-05,
+      "loss": 0.9598,
+      "mean_token_accuracy": 0.708594286441803,
+      "num_tokens": 254795435.0,
+      "step": 102530
+    },
+    {
+      "epoch": 0.9401301916200605,
+      "learning_rate": 1.1975795360777483e-05,
+      "loss": 0.9742,
+      "mean_token_accuracy": 0.7056930780410766,
+      "num_tokens": 254823645.0,
+      "step": 102540
+    },
+    {
+      "epoch": 0.9402218758595398,
+      "learning_rate": 1.1957458512881636e-05,
+      "loss": 0.9364,
+      "mean_token_accuracy": 0.7146631479263306,
+      "num_tokens": 254850616.0,
+      "step": 102550
+    },
+    {
+      "epoch": 0.940313560099019,
+      "learning_rate": 1.1939121664985788e-05,
+      "loss": 0.9171,
+      "mean_token_accuracy": 0.7172474384307861,
+      "num_tokens": 254878318.0,
+      "step": 102560
+    },
+    {
+      "epoch": 0.9404052443384983,
+      "learning_rate": 1.1920784817089944e-05,
+      "loss": 0.9188,
+      "mean_token_accuracy": 0.7150501072406769,
+      "num_tokens": 254906065.0,
+      "step": 102570
+    },
+    {
+      "epoch": 0.9404969285779774,
+      "learning_rate": 1.1902447969194095e-05,
+      "loss": 0.9454,
+      "mean_token_accuracy": 0.7106794238090515,
+      "num_tokens": 254933663.0,
+      "step": 102580
+    },
+    {
+      "epoch": 0.9405886128174566,
+      "learning_rate": 1.1884111121298249e-05,
+      "loss": 0.9727,
+      "mean_token_accuracy": 0.7085217356681823,
+      "num_tokens": 254961576.0,
+      "step": 102590
+    },
+    {
+      "epoch": 0.9406802970569359,
+      "learning_rate": 1.1865774273402403e-05,
+      "loss": 0.9957,
+      "mean_token_accuracy": 0.7010318994522095,
+      "num_tokens": 254989926.0,
+      "step": 102600
+    },
+    {
+      "epoch": 0.9407719812964151,
+      "learning_rate": 1.1847437425506556e-05,
+      "loss": 0.9675,
+      "mean_token_accuracy": 0.7058118462562561,
+      "num_tokens": 255018011.0,
+      "step": 102610
+    },
+    {
+      "epoch": 0.9408636655358944,
+      "learning_rate": 1.182910057761071e-05,
+      "loss": 0.9388,
+      "mean_token_accuracy": 0.7153486907482147,
+      "num_tokens": 255046107.0,
+      "step": 102620
+    },
+    {
+      "epoch": 0.9409553497753737,
+      "learning_rate": 1.1810763729714863e-05,
+      "loss": 0.9451,
+      "mean_token_accuracy": 0.708610600233078,
+      "num_tokens": 255074703.0,
+      "step": 102630
+    },
+    {
+      "epoch": 0.9410470340148529,
+      "learning_rate": 1.1792426881819015e-05,
+      "loss": 0.9487,
+      "mean_token_accuracy": 0.7160908460617066,
+      "num_tokens": 255102673.0,
+      "step": 102640
+    },
+    {
+      "epoch": 0.941138718254332,
+      "learning_rate": 1.1774090033923169e-05,
+      "loss": 0.9296,
+      "mean_token_accuracy": 0.7148681581020355,
+      "num_tokens": 255130748.0,
+      "step": 102650
+    },
+    {
+      "epoch": 0.9412304024938113,
+      "learning_rate": 1.1755753186027322e-05,
+      "loss": 0.9211,
+      "mean_token_accuracy": 0.7172384858131409,
+      "num_tokens": 255159178.0,
+      "step": 102660
+    },
+    {
+      "epoch": 0.9413220867332905,
+      "learning_rate": 1.1737416338131476e-05,
+      "loss": 0.9517,
+      "mean_token_accuracy": 0.7140165328979492,
+      "num_tokens": 255186210.0,
+      "step": 102670
+    },
+    {
+      "epoch": 0.9414137709727698,
+      "learning_rate": 1.171907949023563e-05,
+      "loss": 0.9305,
+      "mean_token_accuracy": 0.7180571615695953,
+      "num_tokens": 255213123.0,
+      "step": 102680
+    },
+    {
+      "epoch": 0.941505455212249,
+      "learning_rate": 1.1700742642339781e-05,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.7052584767341614,
+      "num_tokens": 255241689.0,
+      "step": 102690
+    },
+    {
+      "epoch": 0.9415971394517283,
+      "learning_rate": 1.1682405794443937e-05,
+      "loss": 0.9402,
+      "mean_token_accuracy": 0.7112890183925629,
+      "num_tokens": 255270439.0,
+      "step": 102700
+    },
+    {
+      "epoch": 0.9416888236912074,
+      "learning_rate": 1.1664068946548088e-05,
+      "loss": 0.9901,
+      "mean_token_accuracy": 0.7062026262283325,
+      "num_tokens": 255298352.0,
+      "step": 102710
+    },
+    {
+      "epoch": 0.9417805079306867,
+      "learning_rate": 1.1645732098652242e-05,
+      "loss": 0.9602,
+      "mean_token_accuracy": 0.7041107833385467,
+      "num_tokens": 255327198.0,
+      "step": 102720
+    },
+    {
+      "epoch": 0.9418721921701659,
+      "learning_rate": 1.1627395250756396e-05,
+      "loss": 0.9666,
+      "mean_token_accuracy": 0.7073990762233734,
+      "num_tokens": 255355082.0,
+      "step": 102730
+    },
+    {
+      "epoch": 0.9419638764096452,
+      "learning_rate": 1.1609058402860549e-05,
+      "loss": 0.9532,
+      "mean_token_accuracy": 0.7126502692699432,
+      "num_tokens": 255382685.0,
+      "step": 102740
+    },
+    {
+      "epoch": 0.9420555606491244,
+      "learning_rate": 1.1590721554964703e-05,
+      "loss": 0.9282,
+      "mean_token_accuracy": 0.7141442239284516,
+      "num_tokens": 255409901.0,
+      "step": 102750
+    },
+    {
+      "epoch": 0.9421472448886037,
+      "learning_rate": 1.1572384707068855e-05,
+      "loss": 0.972,
+      "mean_token_accuracy": 0.70137899518013,
+      "num_tokens": 255438524.0,
+      "step": 102760
+    },
+    {
+      "epoch": 0.9422389291280829,
+      "learning_rate": 1.1554047859173008e-05,
+      "loss": 0.9561,
+      "mean_token_accuracy": 0.710118705034256,
+      "num_tokens": 255467764.0,
+      "step": 102770
+    },
+    {
+      "epoch": 0.9423306133675621,
+      "learning_rate": 1.1535711011277162e-05,
+      "loss": 0.9489,
+      "mean_token_accuracy": 0.7068646609783172,
+      "num_tokens": 255496428.0,
+      "step": 102780
+    },
+    {
+      "epoch": 0.9424222976070413,
+      "learning_rate": 1.1517374163381315e-05,
+      "loss": 0.9485,
+      "mean_token_accuracy": 0.7078204989433289,
+      "num_tokens": 255524565.0,
+      "step": 102790
+    },
+    {
+      "epoch": 0.9425139818465206,
+      "learning_rate": 1.1499037315485469e-05,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.7103520512580872,
+      "num_tokens": 255551758.0,
+      "step": 102800
+    },
+    {
+      "epoch": 0.9426056660859998,
+      "learning_rate": 1.1480700467589622e-05,
+      "loss": 0.9442,
+      "mean_token_accuracy": 0.7119756937026978,
+      "num_tokens": 255579314.0,
+      "step": 102810
+    },
+    {
+      "epoch": 0.9426973503254791,
+      "learning_rate": 1.1462363619693774e-05,
+      "loss": 0.9523,
+      "mean_token_accuracy": 0.7119636416435242,
+      "num_tokens": 255607939.0,
+      "step": 102820
+    },
+    {
+      "epoch": 0.9427890345649583,
+      "learning_rate": 1.144402677179793e-05,
+      "loss": 0.9103,
+      "mean_token_accuracy": 0.7157583832740784,
+      "num_tokens": 255636202.0,
+      "step": 102830
+    },
+    {
+      "epoch": 0.9428807188044375,
+      "learning_rate": 1.1425689923902081e-05,
+      "loss": 0.9288,
+      "mean_token_accuracy": 0.7167107999324799,
+      "num_tokens": 255662473.0,
+      "step": 102840
+    },
+    {
+      "epoch": 0.9429724030439167,
+      "learning_rate": 1.1407353076006235e-05,
+      "loss": 0.9504,
+      "mean_token_accuracy": 0.713322114944458,
+      "num_tokens": 255691494.0,
+      "step": 102850
+    },
+    {
+      "epoch": 0.943064087283396,
+      "learning_rate": 1.1389016228110388e-05,
+      "loss": 0.9028,
+      "mean_token_accuracy": 0.7244246780872345,
+      "num_tokens": 255718669.0,
+      "step": 102860
+    },
+    {
+      "epoch": 0.9431557715228752,
+      "learning_rate": 1.1370679380214542e-05,
+      "loss": 0.944,
+      "mean_token_accuracy": 0.7100665211677551,
+      "num_tokens": 255746572.0,
+      "step": 102870
+    },
+    {
+      "epoch": 0.9432474557623545,
+      "learning_rate": 1.1352342532318696e-05,
+      "loss": 0.9769,
+      "mean_token_accuracy": 0.7085854828357696,
+      "num_tokens": 255773817.0,
+      "step": 102880
+    },
+    {
+      "epoch": 0.9433391400018337,
+      "learning_rate": 1.1334005684422847e-05,
+      "loss": 0.9391,
+      "mean_token_accuracy": 0.7132640063762665,
+      "num_tokens": 255801392.0,
+      "step": 102890
+    },
+    {
+      "epoch": 0.943430824241313,
+      "learning_rate": 1.1315668836527003e-05,
+      "loss": 0.9796,
+      "mean_token_accuracy": 0.7016677379608154,
+      "num_tokens": 255829170.0,
+      "step": 102900
+    },
+    {
+      "epoch": 0.9435225084807921,
+      "learning_rate": 1.1297331988631155e-05,
+      "loss": 0.9782,
+      "mean_token_accuracy": 0.7059294164180756,
+      "num_tokens": 255856735.0,
+      "step": 102910
+    },
+    {
+      "epoch": 0.9436141927202714,
+      "learning_rate": 1.1278995140735308e-05,
+      "loss": 0.9449,
+      "mean_token_accuracy": 0.7062544524669647,
+      "num_tokens": 255885068.0,
+      "step": 102920
+    },
+    {
+      "epoch": 0.9437058769597506,
+      "learning_rate": 1.1260658292839462e-05,
+      "loss": 0.9609,
+      "mean_token_accuracy": 0.7042105615139007,
+      "num_tokens": 255913407.0,
+      "step": 102930
+    },
+    {
+      "epoch": 0.9437975611992299,
+      "learning_rate": 1.1242321444943615e-05,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.706026577949524,
+      "num_tokens": 255940903.0,
+      "step": 102940
+    },
+    {
+      "epoch": 0.9438892454387091,
+      "learning_rate": 1.1223984597047769e-05,
+      "loss": 0.919,
+      "mean_token_accuracy": 0.7188919186592102,
+      "num_tokens": 255969553.0,
+      "step": 102950
+    },
+    {
+      "epoch": 0.9439809296781884,
+      "learning_rate": 1.120564774915192e-05,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.7064568102359772,
+      "num_tokens": 255996700.0,
+      "step": 102960
+    },
+    {
+      "epoch": 0.9440726139176675,
+      "learning_rate": 1.1187310901256074e-05,
+      "loss": 0.9773,
+      "mean_token_accuracy": 0.7042794942855835,
+      "num_tokens": 256026068.0,
+      "step": 102970
+    },
+    {
+      "epoch": 0.9441642981571468,
+      "learning_rate": 1.1168974053360228e-05,
+      "loss": 0.9851,
+      "mean_token_accuracy": 0.7081849694252014,
+      "num_tokens": 256054448.0,
+      "step": 102980
+    },
+    {
+      "epoch": 0.944255982396626,
+      "learning_rate": 1.1150637205464381e-05,
+      "loss": 0.9317,
+      "mean_token_accuracy": 0.7123378932476043,
+      "num_tokens": 256082585.0,
+      "step": 102990
+    },
+    {
+      "epoch": 0.9443476666361053,
+      "learning_rate": 1.1132300357568535e-05,
+      "loss": 0.9662,
+      "mean_token_accuracy": 0.7105158627033233,
+      "num_tokens": 256110622.0,
+      "step": 103000
+    },
+    {
+      "epoch": 0.9444393508755845,
+      "learning_rate": 1.1113963509672688e-05,
+      "loss": 0.9478,
+      "mean_token_accuracy": 0.707587081193924,
+      "num_tokens": 256139438.0,
+      "step": 103010
+    },
+    {
+      "epoch": 0.9445310351150638,
+      "learning_rate": 1.109562666177684e-05,
+      "loss": 0.9507,
+      "mean_token_accuracy": 0.7156376719474793,
+      "num_tokens": 256166230.0,
+      "step": 103020
+    },
+    {
+      "epoch": 0.944622719354543,
+      "learning_rate": 1.1077289813880996e-05,
+      "loss": 0.9391,
+      "mean_token_accuracy": 0.7084162771701813,
+      "num_tokens": 256195537.0,
+      "step": 103030
+    },
+    {
+      "epoch": 0.9447144035940221,
+      "learning_rate": 1.1058952965985147e-05,
+      "loss": 0.9257,
+      "mean_token_accuracy": 0.7145484566688538,
+      "num_tokens": 256223271.0,
+      "step": 103040
+    },
+    {
+      "epoch": 0.9448060878335014,
+      "learning_rate": 1.1040616118089301e-05,
+      "loss": 0.9578,
+      "mean_token_accuracy": 0.708631718158722,
+      "num_tokens": 256250945.0,
+      "step": 103050
+    },
+    {
+      "epoch": 0.9448977720729806,
+      "learning_rate": 1.1022279270193455e-05,
+      "loss": 0.9471,
+      "mean_token_accuracy": 0.7148377418518066,
+      "num_tokens": 256278424.0,
+      "step": 103060
+    },
+    {
+      "epoch": 0.9449894563124599,
+      "learning_rate": 1.1003942422297606e-05,
+      "loss": 0.948,
+      "mean_token_accuracy": 0.712054181098938,
+      "num_tokens": 256306208.0,
+      "step": 103070
+    },
+    {
+      "epoch": 0.9450811405519391,
+      "learning_rate": 1.0985605574401762e-05,
+      "loss": 0.9452,
+      "mean_token_accuracy": 0.7049896419048309,
+      "num_tokens": 256334342.0,
+      "step": 103080
+    },
+    {
+      "epoch": 0.9451728247914184,
+      "learning_rate": 1.0967268726505914e-05,
+      "loss": 0.9564,
+      "mean_token_accuracy": 0.7121441900730133,
+      "num_tokens": 256362039.0,
+      "step": 103090
+    },
+    {
+      "epoch": 0.9452645090308975,
+      "learning_rate": 1.0948931878610067e-05,
+      "loss": 0.9535,
+      "mean_token_accuracy": 0.7116735219955445,
+      "num_tokens": 256390405.0,
+      "step": 103100
+    },
+    {
+      "epoch": 0.9453561932703768,
+      "learning_rate": 1.093059503071422e-05,
+      "loss": 0.9444,
+      "mean_token_accuracy": 0.7108181834220886,
+      "num_tokens": 256417494.0,
+      "step": 103110
+    },
+    {
+      "epoch": 0.945447877509856,
+      "learning_rate": 1.0912258182818374e-05,
+      "loss": 0.9486,
+      "mean_token_accuracy": 0.714609807729721,
+      "num_tokens": 256445639.0,
+      "step": 103120
+    },
+    {
+      "epoch": 0.9455395617493353,
+      "learning_rate": 1.0893921334922528e-05,
+      "loss": 0.9675,
+      "mean_token_accuracy": 0.7062018990516663,
+      "num_tokens": 256474701.0,
+      "step": 103130
+    },
+    {
+      "epoch": 0.9456312459888145,
+      "learning_rate": 1.0875584487026681e-05,
+      "loss": 0.9765,
+      "mean_token_accuracy": 0.7075938940048218,
+      "num_tokens": 256502867.0,
+      "step": 103140
+    },
+    {
+      "epoch": 0.9457229302282938,
+      "learning_rate": 1.0857247639130833e-05,
+      "loss": 0.9737,
+      "mean_token_accuracy": 0.7039482593536377,
+      "num_tokens": 256530942.0,
+      "step": 103150
+    },
+    {
+      "epoch": 0.945814614467773,
+      "learning_rate": 1.0838910791234987e-05,
+      "loss": 0.9747,
+      "mean_token_accuracy": 0.7098252058029175,
+      "num_tokens": 256559324.0,
+      "step": 103160
+    },
+    {
+      "epoch": 0.9459062987072522,
+      "learning_rate": 1.082057394333914e-05,
+      "loss": 0.9916,
+      "mean_token_accuracy": 0.6999709665775299,
+      "num_tokens": 256587536.0,
+      "step": 103170
+    },
+    {
+      "epoch": 0.9459979829467314,
+      "learning_rate": 1.0802237095443294e-05,
+      "loss": 0.9436,
+      "mean_token_accuracy": 0.7075395107269287,
+      "num_tokens": 256615740.0,
+      "step": 103180
+    },
+    {
+      "epoch": 0.9460896671862107,
+      "learning_rate": 1.0783900247547447e-05,
+      "loss": 0.9468,
+      "mean_token_accuracy": 0.7163018643856048,
+      "num_tokens": 256643101.0,
+      "step": 103190
+    },
+    {
+      "epoch": 0.9461813514256899,
+      "learning_rate": 1.07655633996516e-05,
+      "loss": 0.9785,
+      "mean_token_accuracy": 0.7065538823604584,
+      "num_tokens": 256671159.0,
+      "step": 103200
+    },
+    {
+      "epoch": 0.9462730356651692,
+      "learning_rate": 1.0747226551755755e-05,
+      "loss": 0.9701,
+      "mean_token_accuracy": 0.7056053638458252,
+      "num_tokens": 256698939.0,
+      "step": 103210
+    },
+    {
+      "epoch": 0.9463647199046484,
+      "learning_rate": 1.0728889703859906e-05,
+      "loss": 0.9811,
+      "mean_token_accuracy": 0.7037706851959229,
+      "num_tokens": 256727759.0,
+      "step": 103220
+    },
+    {
+      "epoch": 0.9464564041441276,
+      "learning_rate": 1.071055285596406e-05,
+      "loss": 0.9543,
+      "mean_token_accuracy": 0.7111855804920196,
+      "num_tokens": 256754783.0,
+      "step": 103230
+    },
+    {
+      "epoch": 0.9465480883836068,
+      "learning_rate": 1.0692216008068214e-05,
+      "loss": 0.9585,
+      "mean_token_accuracy": 0.7117890298366547,
+      "num_tokens": 256782722.0,
+      "step": 103240
+    },
+    {
+      "epoch": 0.9466397726230861,
+      "learning_rate": 1.0673879160172367e-05,
+      "loss": 0.9238,
+      "mean_token_accuracy": 0.7099150657653809,
+      "num_tokens": 256810530.0,
+      "step": 103250
+    },
+    {
+      "epoch": 0.9467314568625653,
+      "learning_rate": 1.065554231227652e-05,
+      "loss": 0.9325,
+      "mean_token_accuracy": 0.7152824342250824,
+      "num_tokens": 256838939.0,
+      "step": 103260
+    },
+    {
+      "epoch": 0.9468231411020446,
+      "learning_rate": 1.0637205464380673e-05,
+      "loss": 0.9652,
+      "mean_token_accuracy": 0.7128736257553101,
+      "num_tokens": 256866720.0,
+      "step": 103270
+    },
+    {
+      "epoch": 0.9469148253415238,
+      "learning_rate": 1.0618868616484826e-05,
+      "loss": 0.9292,
+      "mean_token_accuracy": 0.7194776773452759,
+      "num_tokens": 256894357.0,
+      "step": 103280
+    },
+    {
+      "epoch": 0.9470065095810031,
+      "learning_rate": 1.060053176858898e-05,
+      "loss": 0.9609,
+      "mean_token_accuracy": 0.7091909885406494,
+      "num_tokens": 256921804.0,
+      "step": 103290
+    },
+    {
+      "epoch": 0.9470981938204822,
+      "learning_rate": 1.0582194920693133e-05,
+      "loss": 0.9423,
+      "mean_token_accuracy": 0.7118369340896606,
+      "num_tokens": 256949732.0,
+      "step": 103300
+    },
+    {
+      "epoch": 0.9471898780599615,
+      "learning_rate": 1.0563858072797287e-05,
+      "loss": 0.9768,
+      "mean_token_accuracy": 0.7053621292114258,
+      "num_tokens": 256977435.0,
+      "step": 103310
+    },
+    {
+      "epoch": 0.9472815622994407,
+      "learning_rate": 1.054552122490144e-05,
+      "loss": 0.9483,
+      "mean_token_accuracy": 0.7116852819919586,
+      "num_tokens": 257004798.0,
+      "step": 103320
+    },
+    {
+      "epoch": 0.94737324653892,
+      "learning_rate": 1.0527184377005592e-05,
+      "loss": 0.9697,
+      "mean_token_accuracy": 0.7076071083545685,
+      "num_tokens": 257032795.0,
+      "step": 103330
+    },
+    {
+      "epoch": 0.9474649307783992,
+      "learning_rate": 1.0508847529109748e-05,
+      "loss": 0.9503,
+      "mean_token_accuracy": 0.7118573665618897,
+      "num_tokens": 257060989.0,
+      "step": 103340
+    },
+    {
+      "epoch": 0.9475566150178785,
+      "learning_rate": 1.04905106812139e-05,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.7086583495140075,
+      "num_tokens": 257089396.0,
+      "step": 103350
+    },
+    {
+      "epoch": 0.9476482992573576,
+      "learning_rate": 1.0472173833318053e-05,
+      "loss": 0.9412,
+      "mean_token_accuracy": 0.7133853733539581,
+      "num_tokens": 257117707.0,
+      "step": 103360
+    },
+    {
+      "epoch": 0.9477399834968369,
+      "learning_rate": 1.0453836985422207e-05,
+      "loss": 0.9491,
+      "mean_token_accuracy": 0.7088606655597687,
+      "num_tokens": 257145689.0,
+      "step": 103370
+    },
+    {
+      "epoch": 0.9478316677363161,
+      "learning_rate": 1.0435500137526358e-05,
+      "loss": 0.9344,
+      "mean_token_accuracy": 0.7158215522766114,
+      "num_tokens": 257174306.0,
+      "step": 103380
+    },
+    {
+      "epoch": 0.9479233519757954,
+      "learning_rate": 1.0417163289630514e-05,
+      "loss": 0.9179,
+      "mean_token_accuracy": 0.7221521854400634,
+      "num_tokens": 257202398.0,
+      "step": 103390
+    },
+    {
+      "epoch": 0.9480150362152746,
+      "learning_rate": 1.0398826441734666e-05,
+      "loss": 0.9549,
+      "mean_token_accuracy": 0.7112609088420868,
+      "num_tokens": 257230126.0,
+      "step": 103400
+    },
+    {
+      "epoch": 0.9481067204547539,
+      "learning_rate": 1.038048959383882e-05,
+      "loss": 0.9357,
+      "mean_token_accuracy": 0.7162086427211761,
+      "num_tokens": 257257389.0,
+      "step": 103410
+    },
+    {
+      "epoch": 0.9481984046942331,
+      "learning_rate": 1.0362152745942973e-05,
+      "loss": 0.9562,
+      "mean_token_accuracy": 0.7039444208145141,
+      "num_tokens": 257284823.0,
+      "step": 103420
+    },
+    {
+      "epoch": 0.9482900889337122,
+      "learning_rate": 1.0343815898047126e-05,
+      "loss": 0.9677,
+      "mean_token_accuracy": 0.7077718138694763,
+      "num_tokens": 257312928.0,
+      "step": 103430
+    },
+    {
+      "epoch": 0.9483817731731915,
+      "learning_rate": 1.032547905015128e-05,
+      "loss": 0.943,
+      "mean_token_accuracy": 0.7146332323551178,
+      "num_tokens": 257340869.0,
+      "step": 103440
+    },
+    {
+      "epoch": 0.9484734574126708,
+      "learning_rate": 1.0307142202255433e-05,
+      "loss": 0.9522,
+      "mean_token_accuracy": 0.7099611103534699,
+      "num_tokens": 257369527.0,
+      "step": 103450
+    },
+    {
+      "epoch": 0.94856514165215,
+      "learning_rate": 1.0288805354359587e-05,
+      "loss": 0.9468,
+      "mean_token_accuracy": 0.7089724659919738,
+      "num_tokens": 257397140.0,
+      "step": 103460
+    },
+    {
+      "epoch": 0.9486568258916293,
+      "learning_rate": 1.0270468506463739e-05,
+      "loss": 0.9767,
+      "mean_token_accuracy": 0.7073847889900208,
+      "num_tokens": 257425107.0,
+      "step": 103470
+    },
+    {
+      "epoch": 0.9487485101311085,
+      "learning_rate": 1.0252131658567892e-05,
+      "loss": 0.9364,
+      "mean_token_accuracy": 0.712783819437027,
+      "num_tokens": 257453099.0,
+      "step": 103480
+    },
+    {
+      "epoch": 0.9488401943705876,
+      "learning_rate": 1.0233794810672046e-05,
+      "loss": 0.9565,
+      "mean_token_accuracy": 0.7080893158912659,
+      "num_tokens": 257480863.0,
+      "step": 103490
+    },
+    {
+      "epoch": 0.9489318786100669,
+      "learning_rate": 1.02154579627762e-05,
+      "loss": 0.945,
+      "mean_token_accuracy": 0.7114560902118683,
+      "num_tokens": 257509350.0,
+      "step": 103500
+    },
+    {
+      "epoch": 0.9490235628495461,
+      "learning_rate": 1.0197121114880353e-05,
+      "loss": 0.9153,
+      "mean_token_accuracy": 0.7201562821865082,
+      "num_tokens": 257538072.0,
+      "step": 103510
+    },
+    {
+      "epoch": 0.9491152470890254,
+      "learning_rate": 1.0178784266984507e-05,
+      "loss": 0.9306,
+      "mean_token_accuracy": 0.7134028077125549,
+      "num_tokens": 257566075.0,
+      "step": 103520
+    },
+    {
+      "epoch": 0.9492069313285046,
+      "learning_rate": 1.0160447419088658e-05,
+      "loss": 0.9646,
+      "mean_token_accuracy": 0.711558175086975,
+      "num_tokens": 257593780.0,
+      "step": 103530
+    },
+    {
+      "epoch": 0.9492986155679839,
+      "learning_rate": 1.0142110571192814e-05,
+      "loss": 0.9591,
+      "mean_token_accuracy": 0.711041659116745,
+      "num_tokens": 257621778.0,
+      "step": 103540
+    },
+    {
+      "epoch": 0.9493902998074631,
+      "learning_rate": 1.0123773723296966e-05,
+      "loss": 0.9593,
+      "mean_token_accuracy": 0.7107082903385162,
+      "num_tokens": 257649639.0,
+      "step": 103550
+    },
+    {
+      "epoch": 0.9494819840469423,
+      "learning_rate": 1.0105436875401119e-05,
+      "loss": 0.9894,
+      "mean_token_accuracy": 0.7032026469707489,
+      "num_tokens": 257677953.0,
+      "step": 103560
+    },
+    {
+      "epoch": 0.9495736682864215,
+      "learning_rate": 1.0087100027505273e-05,
+      "loss": 0.9577,
+      "mean_token_accuracy": 0.7094796299934387,
+      "num_tokens": 257706411.0,
+      "step": 103570
+    },
+    {
+      "epoch": 0.9496653525259008,
+      "learning_rate": 1.0068763179609425e-05,
+      "loss": 0.9493,
+      "mean_token_accuracy": 0.7116182744503021,
+      "num_tokens": 257735209.0,
+      "step": 103580
+    },
+    {
+      "epoch": 0.94975703676538,
+      "learning_rate": 1.005042633171358e-05,
+      "loss": 0.9616,
+      "mean_token_accuracy": 0.7038978099822998,
+      "num_tokens": 257763694.0,
+      "step": 103590
+    },
+    {
+      "epoch": 0.9498487210048593,
+      "learning_rate": 1.0032089483817732e-05,
+      "loss": 0.943,
+      "mean_token_accuracy": 0.7120588600635529,
+      "num_tokens": 257791616.0,
+      "step": 103600
+    },
+    {
+      "epoch": 0.9499404052443385,
+      "learning_rate": 1.0013752635921885e-05,
+      "loss": 0.9589,
+      "mean_token_accuracy": 0.7063844621181488,
+      "num_tokens": 257819375.0,
+      "step": 103610
+    },
+    {
+      "epoch": 0.9500320894838177,
+      "learning_rate": 9.995415788026039e-06,
+      "loss": 0.929,
+      "mean_token_accuracy": 0.7149990439414978,
+      "num_tokens": 257846476.0,
+      "step": 103620
+    },
+    {
+      "epoch": 0.9501237737232969,
+      "learning_rate": 9.977078940130192e-06,
+      "loss": 0.9526,
+      "mean_token_accuracy": 0.7108969807624816,
+      "num_tokens": 257874735.0,
+      "step": 103630
+    },
+    {
+      "epoch": 0.9502154579627762,
+      "learning_rate": 9.958742092234346e-06,
+      "loss": 0.9784,
+      "mean_token_accuracy": 0.7079266607761383,
+      "num_tokens": 257902304.0,
+      "step": 103640
+    },
+    {
+      "epoch": 0.9503071422022554,
+      "learning_rate": 9.9404052443385e-06,
+      "loss": 0.9779,
+      "mean_token_accuracy": 0.7078679800033569,
+      "num_tokens": 257931179.0,
+      "step": 103650
+    },
+    {
+      "epoch": 0.9503988264417347,
+      "learning_rate": 9.922068396442651e-06,
+      "loss": 0.962,
+      "mean_token_accuracy": 0.7135915398597718,
+      "num_tokens": 257959103.0,
+      "step": 103660
+    },
+    {
+      "epoch": 0.9504905106812139,
+      "learning_rate": 9.903731548546805e-06,
+      "loss": 0.9377,
+      "mean_token_accuracy": 0.7142771303653717,
+      "num_tokens": 257986710.0,
+      "step": 103670
+    },
+    {
+      "epoch": 0.9505821949206932,
+      "learning_rate": 9.885394700650958e-06,
+      "loss": 0.9879,
+      "mean_token_accuracy": 0.7014545917510986,
+      "num_tokens": 258014745.0,
+      "step": 103680
+    },
+    {
+      "epoch": 0.9506738791601723,
+      "learning_rate": 9.867057852755112e-06,
+      "loss": 0.9633,
+      "mean_token_accuracy": 0.7079756379127502,
+      "num_tokens": 258042307.0,
+      "step": 103690
+    },
+    {
+      "epoch": 0.9507655633996516,
+      "learning_rate": 9.848721004859266e-06,
+      "loss": 0.9506,
+      "mean_token_accuracy": 0.7097310304641724,
+      "num_tokens": 258071402.0,
+      "step": 103700
+    },
+    {
+      "epoch": 0.9508572476391308,
+      "learning_rate": 9.830384156963417e-06,
+      "loss": 0.9473,
+      "mean_token_accuracy": 0.7122018277645111,
+      "num_tokens": 258099658.0,
+      "step": 103710
+    },
+    {
+      "epoch": 0.9509489318786101,
+      "learning_rate": 9.812047309067573e-06,
+      "loss": 0.9514,
+      "mean_token_accuracy": 0.7135707855224609,
+      "num_tokens": 258127085.0,
+      "step": 103720
+    },
+    {
+      "epoch": 0.9510406161180893,
+      "learning_rate": 9.793710461171725e-06,
+      "loss": 0.9441,
+      "mean_token_accuracy": 0.7074814796447754,
+      "num_tokens": 258155995.0,
+      "step": 103730
+    },
+    {
+      "epoch": 0.9511323003575686,
+      "learning_rate": 9.775373613275878e-06,
+      "loss": 0.9696,
+      "mean_token_accuracy": 0.7047323524951935,
+      "num_tokens": 258183360.0,
+      "step": 103740
+    },
+    {
+      "epoch": 0.9512239845970478,
+      "learning_rate": 9.757036765380032e-06,
+      "loss": 0.9555,
+      "mean_token_accuracy": 0.7139915108680726,
+      "num_tokens": 258210942.0,
+      "step": 103750
+    },
+    {
+      "epoch": 0.951315668836527,
+      "learning_rate": 9.738699917484185e-06,
+      "loss": 0.9254,
+      "mean_token_accuracy": 0.717748486995697,
+      "num_tokens": 258239633.0,
+      "step": 103760
+    },
+    {
+      "epoch": 0.9514073530760062,
+      "learning_rate": 9.720363069588339e-06,
+      "loss": 0.9686,
+      "mean_token_accuracy": 0.7114035665988923,
+      "num_tokens": 258266974.0,
+      "step": 103770
+    },
+    {
+      "epoch": 0.9514990373154855,
+      "learning_rate": 9.70202622169249e-06,
+      "loss": 0.9671,
+      "mean_token_accuracy": 0.7076778531074523,
+      "num_tokens": 258294246.0,
+      "step": 103780
+    },
+    {
+      "epoch": 0.9515907215549647,
+      "learning_rate": 9.683689373796644e-06,
+      "loss": 0.9433,
+      "mean_token_accuracy": 0.7101082801818848,
+      "num_tokens": 258323185.0,
+      "step": 103790
+    },
+    {
+      "epoch": 0.951682405794444,
+      "learning_rate": 9.665352525900798e-06,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.7070024132728576,
+      "num_tokens": 258351205.0,
+      "step": 103800
+    },
+    {
+      "epoch": 0.9517740900339232,
+      "learning_rate": 9.647015678004951e-06,
+      "loss": 0.9463,
+      "mean_token_accuracy": 0.715484869480133,
+      "num_tokens": 258378655.0,
+      "step": 103810
+    },
+    {
+      "epoch": 0.9518657742734024,
+      "learning_rate": 9.628678830109105e-06,
+      "loss": 0.9428,
+      "mean_token_accuracy": 0.7091742873191833,
+      "num_tokens": 258406924.0,
+      "step": 103820
+    },
+    {
+      "epoch": 0.9519574585128816,
+      "learning_rate": 9.610341982213258e-06,
+      "loss": 0.9439,
+      "mean_token_accuracy": 0.7174428999423981,
+      "num_tokens": 258434029.0,
+      "step": 103830
+    },
+    {
+      "epoch": 0.9520491427523609,
+      "learning_rate": 9.59200513431741e-06,
+      "loss": 0.9685,
+      "mean_token_accuracy": 0.7059837639331817,
+      "num_tokens": 258461863.0,
+      "step": 103840
+    },
+    {
+      "epoch": 0.9521408269918401,
+      "learning_rate": 9.573668286421566e-06,
+      "loss": 0.9689,
+      "mean_token_accuracy": 0.7063271284103394,
+      "num_tokens": 258490273.0,
+      "step": 103850
+    },
+    {
+      "epoch": 0.9522325112313194,
+      "learning_rate": 9.555331438525717e-06,
+      "loss": 0.9458,
+      "mean_token_accuracy": 0.7109232485294342,
+      "num_tokens": 258517609.0,
+      "step": 103860
+    },
+    {
+      "epoch": 0.9523241954707986,
+      "learning_rate": 9.536994590629871e-06,
+      "loss": 0.9717,
+      "mean_token_accuracy": 0.7070327520370483,
+      "num_tokens": 258545184.0,
+      "step": 103870
+    },
+    {
+      "epoch": 0.9524158797102779,
+      "learning_rate": 9.518657742734025e-06,
+      "loss": 0.9157,
+      "mean_token_accuracy": 0.7196493923664093,
+      "num_tokens": 258572310.0,
+      "step": 103880
+    },
+    {
+      "epoch": 0.952507563949757,
+      "learning_rate": 9.500320894838176e-06,
+      "loss": 0.9544,
+      "mean_token_accuracy": 0.7129814445972442,
+      "num_tokens": 258601242.0,
+      "step": 103890
+    },
+    {
+      "epoch": 0.9525992481892362,
+      "learning_rate": 9.481984046942332e-06,
+      "loss": 0.9386,
+      "mean_token_accuracy": 0.7104416191577911,
+      "num_tokens": 258628772.0,
+      "step": 103900
+    },
+    {
+      "epoch": 0.9526909324287155,
+      "learning_rate": 9.463647199046484e-06,
+      "loss": 0.9517,
+      "mean_token_accuracy": 0.7113486289978027,
+      "num_tokens": 258656102.0,
+      "step": 103910
+    },
+    {
+      "epoch": 0.9527826166681947,
+      "learning_rate": 9.445310351150637e-06,
+      "loss": 0.9578,
+      "mean_token_accuracy": 0.7102086842060089,
+      "num_tokens": 258683737.0,
+      "step": 103920
+    },
+    {
+      "epoch": 0.952874300907674,
+      "learning_rate": 9.42697350325479e-06,
+      "loss": 0.9755,
+      "mean_token_accuracy": 0.7076316356658936,
+      "num_tokens": 258712208.0,
+      "step": 103930
+    },
+    {
+      "epoch": 0.9529659851471532,
+      "learning_rate": 9.408636655358944e-06,
+      "loss": 0.941,
+      "mean_token_accuracy": 0.7127200841903687,
+      "num_tokens": 258740392.0,
+      "step": 103940
+    },
+    {
+      "epoch": 0.9530576693866324,
+      "learning_rate": 9.390299807463098e-06,
+      "loss": 0.9295,
+      "mean_token_accuracy": 0.7164123833179474,
+      "num_tokens": 258768852.0,
+      "step": 103950
+    },
+    {
+      "epoch": 0.9531493536261116,
+      "learning_rate": 9.371962959567251e-06,
+      "loss": 0.9206,
+      "mean_token_accuracy": 0.7147091567516327,
+      "num_tokens": 258796594.0,
+      "step": 103960
+    },
+    {
+      "epoch": 0.9532410378655909,
+      "learning_rate": 9.353626111671405e-06,
+      "loss": 0.9584,
+      "mean_token_accuracy": 0.7077657818794251,
+      "num_tokens": 258824947.0,
+      "step": 103970
+    },
+    {
+      "epoch": 0.9533327221050701,
+      "learning_rate": 9.335289263775557e-06,
+      "loss": 0.9542,
+      "mean_token_accuracy": 0.7099952220916748,
+      "num_tokens": 258852795.0,
+      "step": 103980
+    },
+    {
+      "epoch": 0.9534244063445494,
+      "learning_rate": 9.31695241587971e-06,
+      "loss": 0.9696,
+      "mean_token_accuracy": 0.7050184786319733,
+      "num_tokens": 258880797.0,
+      "step": 103990
+    },
+    {
+      "epoch": 0.9535160905840286,
+      "learning_rate": 9.298615567983864e-06,
+      "loss": 0.9578,
+      "mean_token_accuracy": 0.7135156214237213,
+      "num_tokens": 258907358.0,
+      "step": 104000
+    },
+    {
+      "epoch": 0.9536077748235079,
+      "learning_rate": 9.280278720088018e-06,
+      "loss": 0.9527,
+      "mean_token_accuracy": 0.7083232760429382,
+      "num_tokens": 258936386.0,
+      "step": 104010
+    },
+    {
+      "epoch": 0.953699459062987,
+      "learning_rate": 9.261941872192171e-06,
+      "loss": 0.9294,
+      "mean_token_accuracy": 0.7145907461643219,
+      "num_tokens": 258963604.0,
+      "step": 104020
+    },
+    {
+      "epoch": 0.9537911433024663,
+      "learning_rate": 9.243605024296325e-06,
+      "loss": 0.9783,
+      "mean_token_accuracy": 0.7060829818248748,
+      "num_tokens": 258990818.0,
+      "step": 104030
+    },
+    {
+      "epoch": 0.9538828275419455,
+      "learning_rate": 9.225268176400477e-06,
+      "loss": 0.9476,
+      "mean_token_accuracy": 0.7087313115596772,
+      "num_tokens": 259018300.0,
+      "step": 104040
+    },
+    {
+      "epoch": 0.9539745117814248,
+      "learning_rate": 9.206931328504632e-06,
+      "loss": 0.9471,
+      "mean_token_accuracy": 0.7067500948905945,
+      "num_tokens": 259046198.0,
+      "step": 104050
+    },
+    {
+      "epoch": 0.954066196020904,
+      "learning_rate": 9.188594480608784e-06,
+      "loss": 0.9529,
+      "mean_token_accuracy": 0.7083405673503875,
+      "num_tokens": 259074383.0,
+      "step": 104060
+    },
+    {
+      "epoch": 0.9541578802603833,
+      "learning_rate": 9.170257632712937e-06,
+      "loss": 0.9646,
+      "mean_token_accuracy": 0.7116964161396027,
+      "num_tokens": 259101842.0,
+      "step": 104070
+    },
+    {
+      "epoch": 0.9542495644998624,
+      "learning_rate": 9.15192078481709e-06,
+      "loss": 0.9598,
+      "mean_token_accuracy": 0.709268081188202,
+      "num_tokens": 259131262.0,
+      "step": 104080
+    },
+    {
+      "epoch": 0.9543412487393417,
+      "learning_rate": 9.133583936921243e-06,
+      "loss": 0.9474,
+      "mean_token_accuracy": 0.711880898475647,
+      "num_tokens": 259159972.0,
+      "step": 104090
+    },
+    {
+      "epoch": 0.9544329329788209,
+      "learning_rate": 9.115247089025398e-06,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.7128210484981536,
+      "num_tokens": 259186961.0,
+      "step": 104100
+    },
+    {
+      "epoch": 0.9545246172183002,
+      "learning_rate": 9.09691024112955e-06,
+      "loss": 0.9587,
+      "mean_token_accuracy": 0.7122468411922455,
+      "num_tokens": 259215286.0,
+      "step": 104110
+    },
+    {
+      "epoch": 0.9546163014577794,
+      "learning_rate": 9.078573393233703e-06,
+      "loss": 0.9072,
+      "mean_token_accuracy": 0.7194814205169677,
+      "num_tokens": 259242399.0,
+      "step": 104120
+    },
+    {
+      "epoch": 0.9547079856972587,
+      "learning_rate": 9.060236545337857e-06,
+      "loss": 0.9426,
+      "mean_token_accuracy": 0.7142708241939545,
+      "num_tokens": 259269298.0,
+      "step": 104130
+    },
+    {
+      "epoch": 0.9547996699367379,
+      "learning_rate": 9.04189969744201e-06,
+      "loss": 0.9368,
+      "mean_token_accuracy": 0.7143619596958161,
+      "num_tokens": 259296549.0,
+      "step": 104140
+    },
+    {
+      "epoch": 0.9548913541762171,
+      "learning_rate": 9.023562849546164e-06,
+      "loss": 0.9814,
+      "mean_token_accuracy": 0.7055559635162354,
+      "num_tokens": 259325831.0,
+      "step": 104150
+    },
+    {
+      "epoch": 0.9549830384156963,
+      "learning_rate": 9.005226001650318e-06,
+      "loss": 0.9528,
+      "mean_token_accuracy": 0.7074734926223755,
+      "num_tokens": 259353912.0,
+      "step": 104160
+    },
+    {
+      "epoch": 0.9550747226551756,
+      "learning_rate": 8.98688915375447e-06,
+      "loss": 0.9498,
+      "mean_token_accuracy": 0.7051776826381684,
+      "num_tokens": 259382346.0,
+      "step": 104170
+    },
+    {
+      "epoch": 0.9551664068946548,
+      "learning_rate": 8.968552305858623e-06,
+      "loss": 0.9475,
+      "mean_token_accuracy": 0.7156764984130859,
+      "num_tokens": 259411168.0,
+      "step": 104180
+    },
+    {
+      "epoch": 0.9552580911341341,
+      "learning_rate": 8.950215457962777e-06,
+      "loss": 0.945,
+      "mean_token_accuracy": 0.7124345541000366,
+      "num_tokens": 259439682.0,
+      "step": 104190
+    },
+    {
+      "epoch": 0.9553497753736133,
+      "learning_rate": 8.93187861006693e-06,
+      "loss": 0.9625,
+      "mean_token_accuracy": 0.7090458929538727,
+      "num_tokens": 259467899.0,
+      "step": 104200
+    },
+    {
+      "epoch": 0.9554414596130925,
+      "learning_rate": 8.913541762171084e-06,
+      "loss": 0.9432,
+      "mean_token_accuracy": 0.7154362559318542,
+      "num_tokens": 259496247.0,
+      "step": 104210
+    },
+    {
+      "epoch": 0.9555331438525717,
+      "learning_rate": 8.895204914275236e-06,
+      "loss": 0.986,
+      "mean_token_accuracy": 0.7031678676605224,
+      "num_tokens": 259524253.0,
+      "step": 104220
+    },
+    {
+      "epoch": 0.955624828092051,
+      "learning_rate": 8.87686806637939e-06,
+      "loss": 0.9524,
+      "mean_token_accuracy": 0.7027429163455963,
+      "num_tokens": 259552243.0,
+      "step": 104230
+    },
+    {
+      "epoch": 0.9557165123315302,
+      "learning_rate": 8.858531218483543e-06,
+      "loss": 0.9664,
+      "mean_token_accuracy": 0.71009481549263,
+      "num_tokens": 259579815.0,
+      "step": 104240
+    },
+    {
+      "epoch": 0.9558081965710095,
+      "learning_rate": 8.840194370587696e-06,
+      "loss": 0.9498,
+      "mean_token_accuracy": 0.7081762790679932,
+      "num_tokens": 259607855.0,
+      "step": 104250
+    },
+    {
+      "epoch": 0.9558998808104887,
+      "learning_rate": 8.82185752269185e-06,
+      "loss": 0.9328,
+      "mean_token_accuracy": 0.7151672303676605,
+      "num_tokens": 259635978.0,
+      "step": 104260
+    },
+    {
+      "epoch": 0.955991565049968,
+      "learning_rate": 8.803520674796003e-06,
+      "loss": 0.9514,
+      "mean_token_accuracy": 0.7089860737323761,
+      "num_tokens": 259663824.0,
+      "step": 104270
+    },
+    {
+      "epoch": 0.9560832492894471,
+      "learning_rate": 8.785183826900157e-06,
+      "loss": 0.9286,
+      "mean_token_accuracy": 0.7175964951515198,
+      "num_tokens": 259691232.0,
+      "step": 104280
+    },
+    {
+      "epoch": 0.9561749335289264,
+      "learning_rate": 8.766846979004309e-06,
+      "loss": 0.97,
+      "mean_token_accuracy": 0.7069647789001465,
+      "num_tokens": 259719031.0,
+      "step": 104290
+    },
+    {
+      "epoch": 0.9562666177684056,
+      "learning_rate": 8.748510131108462e-06,
+      "loss": 0.9523,
+      "mean_token_accuracy": 0.7097716927528381,
+      "num_tokens": 259746249.0,
+      "step": 104300
+    },
+    {
+      "epoch": 0.9563583020078849,
+      "learning_rate": 8.730173283212616e-06,
+      "loss": 0.955,
+      "mean_token_accuracy": 0.7088623106479645,
+      "num_tokens": 259767504.0,
+      "step": 104310
+    },
+    {
+      "epoch": 0.9564499862473641,
+      "learning_rate": 8.71183643531677e-06,
+      "loss": 0.9427,
+      "mean_token_accuracy": 0.711846262216568,
+      "num_tokens": 259794682.0,
+      "step": 104320
+    },
+    {
+      "epoch": 0.9565416704868434,
+      "learning_rate": 8.693499587420923e-06,
+      "loss": 0.937,
+      "mean_token_accuracy": 0.7208527982234955,
+      "num_tokens": 259822354.0,
+      "step": 104330
+    },
+    {
+      "epoch": 0.9566333547263225,
+      "learning_rate": 8.675162739525077e-06,
+      "loss": 1.0061,
+      "mean_token_accuracy": 0.700165057182312,
+      "num_tokens": 259850180.0,
+      "step": 104340
+    },
+    {
+      "epoch": 0.9567250389658017,
+      "learning_rate": 8.656825891629228e-06,
+      "loss": 0.9694,
+      "mean_token_accuracy": 0.7124208092689515,
+      "num_tokens": 259877507.0,
+      "step": 104350
+    },
+    {
+      "epoch": 0.956816723205281,
+      "learning_rate": 8.638489043733384e-06,
+      "loss": 0.9335,
+      "mean_token_accuracy": 0.7144210040569305,
+      "num_tokens": 259905768.0,
+      "step": 104360
+    },
+    {
+      "epoch": 0.9569084074447602,
+      "learning_rate": 8.620152195837536e-06,
+      "loss": 0.9622,
+      "mean_token_accuracy": 0.7091310858726502,
+      "num_tokens": 259933653.0,
+      "step": 104370
+    },
+    {
+      "epoch": 0.9570000916842395,
+      "learning_rate": 8.601815347941689e-06,
+      "loss": 0.9641,
+      "mean_token_accuracy": 0.7072034657001496,
+      "num_tokens": 259961990.0,
+      "step": 104380
+    },
+    {
+      "epoch": 0.9570917759237187,
+      "learning_rate": 8.583478500045843e-06,
+      "loss": 0.9313,
+      "mean_token_accuracy": 0.7128926217556,
+      "num_tokens": 259988734.0,
+      "step": 104390
+    },
+    {
+      "epoch": 0.957183460163198,
+      "learning_rate": 8.565141652149995e-06,
+      "loss": 0.9676,
+      "mean_token_accuracy": 0.7056384682655334,
+      "num_tokens": 260017025.0,
+      "step": 104400
+    },
+    {
+      "epoch": 0.9572751444026771,
+      "learning_rate": 8.54680480425415e-06,
+      "loss": 0.9444,
+      "mean_token_accuracy": 0.7095074772834777,
+      "num_tokens": 260044921.0,
+      "step": 104410
+    },
+    {
+      "epoch": 0.9573668286421564,
+      "learning_rate": 8.528467956358302e-06,
+      "loss": 0.95,
+      "mean_token_accuracy": 0.7101873934268952,
+      "num_tokens": 260072620.0,
+      "step": 104420
+    },
+    {
+      "epoch": 0.9574585128816356,
+      "learning_rate": 8.510131108462455e-06,
+      "loss": 0.9375,
+      "mean_token_accuracy": 0.7134177446365356,
+      "num_tokens": 260100351.0,
+      "step": 104430
+    },
+    {
+      "epoch": 0.9575501971211149,
+      "learning_rate": 8.491794260566609e-06,
+      "loss": 0.9624,
+      "mean_token_accuracy": 0.7066910684108734,
+      "num_tokens": 260128389.0,
+      "step": 104440
+    },
+    {
+      "epoch": 0.9576418813605941,
+      "learning_rate": 8.473457412670762e-06,
+      "loss": 0.933,
+      "mean_token_accuracy": 0.7151216864585876,
+      "num_tokens": 260156546.0,
+      "step": 104450
+    },
+    {
+      "epoch": 0.9577335656000734,
+      "learning_rate": 8.455120564774916e-06,
+      "loss": 0.9883,
+      "mean_token_accuracy": 0.704510647058487,
+      "num_tokens": 260185019.0,
+      "step": 104460
+    },
+    {
+      "epoch": 0.9578252498395525,
+      "learning_rate": 8.43678371687907e-06,
+      "loss": 0.9482,
+      "mean_token_accuracy": 0.7077221691608429,
+      "num_tokens": 260213293.0,
+      "step": 104470
+    },
+    {
+      "epoch": 0.9579169340790318,
+      "learning_rate": 8.418446868983221e-06,
+      "loss": 0.9332,
+      "mean_token_accuracy": 0.7108700573444366,
+      "num_tokens": 260241811.0,
+      "step": 104480
+    },
+    {
+      "epoch": 0.958008618318511,
+      "learning_rate": 8.400110021087375e-06,
+      "loss": 0.9649,
+      "mean_token_accuracy": 0.7052967786788941,
+      "num_tokens": 260269445.0,
+      "step": 104490
+    },
+    {
+      "epoch": 0.9581003025579903,
+      "learning_rate": 8.381773173191528e-06,
+      "loss": 0.9279,
+      "mean_token_accuracy": 0.716957277059555,
+      "num_tokens": 260297199.0,
+      "step": 104500
+    },
+    {
+      "epoch": 0.9581919867974695,
+      "learning_rate": 8.363436325295682e-06,
+      "loss": 0.9542,
+      "mean_token_accuracy": 0.71466823220253,
+      "num_tokens": 260325191.0,
+      "step": 104510
+    },
+    {
+      "epoch": 0.9582836710369488,
+      "learning_rate": 8.345099477399836e-06,
+      "loss": 0.9847,
+      "mean_token_accuracy": 0.6977778315544129,
+      "num_tokens": 260353127.0,
+      "step": 104520
+    },
+    {
+      "epoch": 0.958375355276428,
+      "learning_rate": 8.326762629503987e-06,
+      "loss": 0.9426,
+      "mean_token_accuracy": 0.7164582431316375,
+      "num_tokens": 260380751.0,
+      "step": 104530
+    },
+    {
+      "epoch": 0.9584670395159072,
+      "learning_rate": 8.308425781608143e-06,
+      "loss": 0.9342,
+      "mean_token_accuracy": 0.717404443025589,
+      "num_tokens": 260407954.0,
+      "step": 104540
+    },
+    {
+      "epoch": 0.9585587237553864,
+      "learning_rate": 8.290088933712295e-06,
+      "loss": 0.9328,
+      "mean_token_accuracy": 0.7158634006977082,
+      "num_tokens": 260435911.0,
+      "step": 104550
+    },
+    {
+      "epoch": 0.9586504079948657,
+      "learning_rate": 8.27175208581645e-06,
+      "loss": 0.9257,
+      "mean_token_accuracy": 0.7167222380638123,
+      "num_tokens": 260463609.0,
+      "step": 104560
+    },
+    {
+      "epoch": 0.9587420922343449,
+      "learning_rate": 8.253415237920602e-06,
+      "loss": 0.9509,
+      "mean_token_accuracy": 0.7157267689704895,
+      "num_tokens": 260491648.0,
+      "step": 104570
+    },
+    {
+      "epoch": 0.9588337764738242,
+      "learning_rate": 8.235078390024755e-06,
+      "loss": 0.9423,
+      "mean_token_accuracy": 0.7102207064628601,
+      "num_tokens": 260519458.0,
+      "step": 104580
+    },
+    {
+      "epoch": 0.9589254607133034,
+      "learning_rate": 8.216741542128909e-06,
+      "loss": 0.9458,
+      "mean_token_accuracy": 0.7071184813976288,
+      "num_tokens": 260547608.0,
+      "step": 104590
+    },
+    {
+      "epoch": 0.9590171449527826,
+      "learning_rate": 8.19840469423306e-06,
+      "loss": 0.9482,
+      "mean_token_accuracy": 0.7091720163822174,
+      "num_tokens": 260575430.0,
+      "step": 104600
+    },
+    {
+      "epoch": 0.9591088291922618,
+      "learning_rate": 8.180067846337216e-06,
+      "loss": 0.9689,
+      "mean_token_accuracy": 0.7091881632804871,
+      "num_tokens": 260603092.0,
+      "step": 104610
+    },
+    {
+      "epoch": 0.9592005134317411,
+      "learning_rate": 8.161730998441368e-06,
+      "loss": 0.9404,
+      "mean_token_accuracy": 0.7148373484611511,
+      "num_tokens": 260630793.0,
+      "step": 104620
+    },
+    {
+      "epoch": 0.9592921976712203,
+      "learning_rate": 8.143394150545521e-06,
+      "loss": 0.9567,
+      "mean_token_accuracy": 0.7130004167556763,
+      "num_tokens": 260658365.0,
+      "step": 104630
+    },
+    {
+      "epoch": 0.9593838819106996,
+      "learning_rate": 8.125057302649675e-06,
+      "loss": 0.9447,
+      "mean_token_accuracy": 0.7086138665676117,
+      "num_tokens": 260685867.0,
+      "step": 104640
+    },
+    {
+      "epoch": 0.9594755661501788,
+      "learning_rate": 8.106720454753829e-06,
+      "loss": 0.9401,
+      "mean_token_accuracy": 0.7134073853492737,
+      "num_tokens": 260713414.0,
+      "step": 104650
+    },
+    {
+      "epoch": 0.9595672503896581,
+      "learning_rate": 8.088383606857982e-06,
+      "loss": 0.9656,
+      "mean_token_accuracy": 0.7081871271133423,
+      "num_tokens": 260740940.0,
+      "step": 104660
+    },
+    {
+      "epoch": 0.9596589346291372,
+      "learning_rate": 8.070046758962136e-06,
+      "loss": 0.96,
+      "mean_token_accuracy": 0.7073591411113739,
+      "num_tokens": 260769025.0,
+      "step": 104670
+    },
+    {
+      "epoch": 0.9597506188686165,
+      "learning_rate": 8.051709911066288e-06,
+      "loss": 0.9348,
+      "mean_token_accuracy": 0.7133885443210601,
+      "num_tokens": 260796858.0,
+      "step": 104680
+    },
+    {
+      "epoch": 0.9598423031080957,
+      "learning_rate": 8.033373063170441e-06,
+      "loss": 0.9303,
+      "mean_token_accuracy": 0.7136583507061005,
+      "num_tokens": 260824880.0,
+      "step": 104690
+    },
+    {
+      "epoch": 0.959933987347575,
+      "learning_rate": 8.015036215274595e-06,
+      "loss": 0.9415,
+      "mean_token_accuracy": 0.7110433340072632,
+      "num_tokens": 260853167.0,
+      "step": 104700
+    },
+    {
+      "epoch": 0.9600256715870542,
+      "learning_rate": 7.996699367378748e-06,
+      "loss": 0.9731,
+      "mean_token_accuracy": 0.7005956888198852,
+      "num_tokens": 260881834.0,
+      "step": 104710
+    },
+    {
+      "epoch": 0.9601173558265335,
+      "learning_rate": 7.978362519482902e-06,
+      "loss": 0.9598,
+      "mean_token_accuracy": 0.710364180803299,
+      "num_tokens": 260909873.0,
+      "step": 104720
+    },
+    {
+      "epoch": 0.9602090400660126,
+      "learning_rate": 7.960025671587054e-06,
+      "loss": 0.9535,
+      "mean_token_accuracy": 0.708672821521759,
+      "num_tokens": 260938189.0,
+      "step": 104730
+    },
+    {
+      "epoch": 0.9603007243054918,
+      "learning_rate": 7.941688823691209e-06,
+      "loss": 0.9752,
+      "mean_token_accuracy": 0.7049290001392364,
+      "num_tokens": 260966413.0,
+      "step": 104740
+    },
+    {
+      "epoch": 0.9603924085449711,
+      "learning_rate": 7.92335197579536e-06,
+      "loss": 1.0027,
+      "mean_token_accuracy": 0.6991028249263763,
+      "num_tokens": 260995689.0,
+      "step": 104750
+    },
+    {
+      "epoch": 0.9604840927844503,
+      "learning_rate": 7.905015127899514e-06,
+      "loss": 0.9393,
+      "mean_token_accuracy": 0.717213100194931,
+      "num_tokens": 261023409.0,
+      "step": 104760
+    },
+    {
+      "epoch": 0.9605757770239296,
+      "learning_rate": 7.886678280003668e-06,
+      "loss": 0.973,
+      "mean_token_accuracy": 0.7059387028217315,
+      "num_tokens": 261051734.0,
+      "step": 104770
+    },
+    {
+      "epoch": 0.9606674612634089,
+      "learning_rate": 7.868341432107821e-06,
+      "loss": 0.9709,
+      "mean_token_accuracy": 0.7021915435791015,
+      "num_tokens": 261080014.0,
+      "step": 104780
+    },
+    {
+      "epoch": 0.9607591455028881,
+      "learning_rate": 7.850004584211975e-06,
+      "loss": 0.9611,
+      "mean_token_accuracy": 0.7105751633644104,
+      "num_tokens": 261106989.0,
+      "step": 104790
+    },
+    {
+      "epoch": 0.9608508297423672,
+      "learning_rate": 7.831667736316127e-06,
+      "loss": 0.9339,
+      "mean_token_accuracy": 0.7151186645030976,
+      "num_tokens": 261135728.0,
+      "step": 104800
+    },
+    {
+      "epoch": 0.9609425139818465,
+      "learning_rate": 7.81333088842028e-06,
+      "loss": 0.9809,
+      "mean_token_accuracy": 0.7078758299350738,
+      "num_tokens": 261164127.0,
+      "step": 104810
+    },
+    {
+      "epoch": 0.9610341982213257,
+      "learning_rate": 7.794994040524434e-06,
+      "loss": 0.9533,
+      "mean_token_accuracy": 0.7144318103790284,
+      "num_tokens": 261192070.0,
+      "step": 104820
+    },
+    {
+      "epoch": 0.961125882460805,
+      "learning_rate": 7.776657192628588e-06,
+      "loss": 0.9378,
+      "mean_token_accuracy": 0.7188721597194672,
+      "num_tokens": 261219868.0,
+      "step": 104830
+    },
+    {
+      "epoch": 0.9612175667002842,
+      "learning_rate": 7.758320344732741e-06,
+      "loss": 0.9592,
+      "mean_token_accuracy": 0.714866328239441,
+      "num_tokens": 261247594.0,
+      "step": 104840
+    },
+    {
+      "epoch": 0.9613092509397635,
+      "learning_rate": 7.739983496836895e-06,
+      "loss": 0.9672,
+      "mean_token_accuracy": 0.7044365048408509,
+      "num_tokens": 261275505.0,
+      "step": 104850
+    },
+    {
+      "epoch": 0.9614009351792426,
+      "learning_rate": 7.721646648941047e-06,
+      "loss": 0.9351,
+      "mean_token_accuracy": 0.710863733291626,
+      "num_tokens": 261304391.0,
+      "step": 104860
+    },
+    {
+      "epoch": 0.9614926194187219,
+      "learning_rate": 7.703309801045202e-06,
+      "loss": 0.9576,
+      "mean_token_accuracy": 0.7064379572868347,
+      "num_tokens": 261331377.0,
+      "step": 104870
+    },
+    {
+      "epoch": 0.9615843036582011,
+      "learning_rate": 7.684972953149354e-06,
+      "loss": 0.9644,
+      "mean_token_accuracy": 0.7093068182468414,
+      "num_tokens": 261358620.0,
+      "step": 104880
+    },
+    {
+      "epoch": 0.9616759878976804,
+      "learning_rate": 7.666636105253507e-06,
+      "loss": 0.9472,
+      "mean_token_accuracy": 0.7095429003238678,
+      "num_tokens": 261387075.0,
+      "step": 104890
+    },
+    {
+      "epoch": 0.9617676721371596,
+      "learning_rate": 7.64829925735766e-06,
+      "loss": 0.9816,
+      "mean_token_accuracy": 0.7029368340969085,
+      "num_tokens": 261415523.0,
+      "step": 104900
+    },
+    {
+      "epoch": 0.9618593563766389,
+      "learning_rate": 7.629962409461813e-06,
+      "loss": 0.9659,
+      "mean_token_accuracy": 0.7069908022880554,
+      "num_tokens": 261443811.0,
+      "step": 104910
+    },
+    {
+      "epoch": 0.9619510406161181,
+      "learning_rate": 7.611625561565967e-06,
+      "loss": 0.941,
+      "mean_token_accuracy": 0.7135602593421936,
+      "num_tokens": 261471413.0,
+      "step": 104920
+    },
+    {
+      "epoch": 0.9620427248555973,
+      "learning_rate": 7.59328871367012e-06,
+      "loss": 0.9446,
+      "mean_token_accuracy": 0.7059894621372222,
+      "num_tokens": 261499435.0,
+      "step": 104930
+    },
+    {
+      "epoch": 0.9621344090950765,
+      "learning_rate": 7.574951865774274e-06,
+      "loss": 0.9864,
+      "mean_token_accuracy": 0.702091658115387,
+      "num_tokens": 261527461.0,
+      "step": 104940
+    },
+    {
+      "epoch": 0.9622260933345558,
+      "learning_rate": 7.556615017878427e-06,
+      "loss": 0.9681,
+      "mean_token_accuracy": 0.7039337754249573,
+      "num_tokens": 261555566.0,
+      "step": 104950
+    },
+    {
+      "epoch": 0.962317777574035,
+      "learning_rate": 7.5382781699825805e-06,
+      "loss": 0.9608,
+      "mean_token_accuracy": 0.7046442866325379,
+      "num_tokens": 261584279.0,
+      "step": 104960
+    },
+    {
+      "epoch": 0.9624094618135143,
+      "learning_rate": 7.519941322086733e-06,
+      "loss": 0.9809,
+      "mean_token_accuracy": 0.7080289125442505,
+      "num_tokens": 261612065.0,
+      "step": 104970
+    },
+    {
+      "epoch": 0.9625011460529935,
+      "learning_rate": 7.501604474190888e-06,
+      "loss": 0.9794,
+      "mean_token_accuracy": 0.7039623856544495,
+      "num_tokens": 261639937.0,
+      "step": 104980
+    },
+    {
+      "epoch": 0.9625928302924728,
+      "learning_rate": 7.48326762629504e-06,
+      "loss": 0.952,
+      "mean_token_accuracy": 0.7151970744132996,
+      "num_tokens": 261667681.0,
+      "step": 104990
+    },
+    {
+      "epoch": 0.9626845145319519,
+      "learning_rate": 7.464930778399193e-06,
+      "loss": 0.9684,
+      "mean_token_accuracy": 0.7099954128265381,
+      "num_tokens": 261695653.0,
+      "step": 105000
+    },
+    {
+      "epoch": 0.9627761987714312,
+      "learning_rate": 7.4465939305033466e-06,
+      "loss": 0.9435,
+      "mean_token_accuracy": 0.7093705058097839,
+      "num_tokens": 261724006.0,
+      "step": 105010
+    },
+    {
+      "epoch": 0.9628678830109104,
+      "learning_rate": 7.428257082607499e-06,
+      "loss": 0.9607,
+      "mean_token_accuracy": 0.7109884142875671,
+      "num_tokens": 261752799.0,
+      "step": 105020
+    },
+    {
+      "epoch": 0.9629595672503897,
+      "learning_rate": 7.409920234711654e-06,
+      "loss": 0.9432,
+      "mean_token_accuracy": 0.7140044867992401,
+      "num_tokens": 261780743.0,
+      "step": 105030
+    },
+    {
+      "epoch": 0.9630512514898689,
+      "learning_rate": 7.391583386815806e-06,
+      "loss": 0.9481,
+      "mean_token_accuracy": 0.7099042296409607,
+      "num_tokens": 261808303.0,
+      "step": 105040
+    },
+    {
+      "epoch": 0.9631429357293482,
+      "learning_rate": 7.37324653891996e-06,
+      "loss": 0.9544,
+      "mean_token_accuracy": 0.7053620159626007,
+      "num_tokens": 261836709.0,
+      "step": 105050
+    },
+    {
+      "epoch": 0.9632346199688273,
+      "learning_rate": 7.354909691024113e-06,
+      "loss": 0.927,
+      "mean_token_accuracy": 0.7229625821113587,
+      "num_tokens": 261863968.0,
+      "step": 105060
+    },
+    {
+      "epoch": 0.9633263042083066,
+      "learning_rate": 7.336572843128267e-06,
+      "loss": 0.9511,
+      "mean_token_accuracy": 0.7075356125831604,
+      "num_tokens": 261891478.0,
+      "step": 105070
+    },
+    {
+      "epoch": 0.9634179884477858,
+      "learning_rate": 7.31823599523242e-06,
+      "loss": 0.9783,
+      "mean_token_accuracy": 0.7071390867233276,
+      "num_tokens": 261919917.0,
+      "step": 105080
+    },
+    {
+      "epoch": 0.9635096726872651,
+      "learning_rate": 7.299899147336574e-06,
+      "loss": 0.9461,
+      "mean_token_accuracy": 0.7109041452407837,
+      "num_tokens": 261948322.0,
+      "step": 105090
+    },
+    {
+      "epoch": 0.9636013569267443,
+      "learning_rate": 7.281562299440726e-06,
+      "loss": 0.9438,
+      "mean_token_accuracy": 0.7165783166885376,
+      "num_tokens": 261975145.0,
+      "step": 105100
+    },
+    {
+      "epoch": 0.9636930411662236,
+      "learning_rate": 7.263225451544879e-06,
+      "loss": 0.9446,
+      "mean_token_accuracy": 0.7081939458847046,
+      "num_tokens": 262003039.0,
+      "step": 105110
+    },
+    {
+      "epoch": 0.9637847254057028,
+      "learning_rate": 7.244888603649033e-06,
+      "loss": 0.9325,
+      "mean_token_accuracy": 0.7147288918495178,
+      "num_tokens": 262030580.0,
+      "step": 105120
+    },
+    {
+      "epoch": 0.963876409645182,
+      "learning_rate": 7.226551755753186e-06,
+      "loss": 0.9625,
+      "mean_token_accuracy": 0.7078769147396088,
+      "num_tokens": 262058695.0,
+      "step": 105130
+    },
+    {
+      "epoch": 0.9639680938846612,
+      "learning_rate": 7.20821490785734e-06,
+      "loss": 0.9357,
+      "mean_token_accuracy": 0.7082073032855988,
+      "num_tokens": 262086304.0,
+      "step": 105140
+    },
+    {
+      "epoch": 0.9640597781241405,
+      "learning_rate": 7.189878059961492e-06,
+      "loss": 0.9531,
+      "mean_token_accuracy": 0.7161878883838654,
+      "num_tokens": 262114218.0,
+      "step": 105150
+    },
+    {
+      "epoch": 0.9641514623636197,
+      "learning_rate": 7.171541212065647e-06,
+      "loss": 0.9298,
+      "mean_token_accuracy": 0.7189928710460662,
+      "num_tokens": 262142632.0,
+      "step": 105160
+    },
+    {
+      "epoch": 0.964243146603099,
+      "learning_rate": 7.153204364169799e-06,
+      "loss": 0.9334,
+      "mean_token_accuracy": 0.7166423559188843,
+      "num_tokens": 262171540.0,
+      "step": 105170
+    },
+    {
+      "epoch": 0.9643348308425782,
+      "learning_rate": 7.134867516273954e-06,
+      "loss": 0.9293,
+      "mean_token_accuracy": 0.7156631767749786,
+      "num_tokens": 262198739.0,
+      "step": 105180
+    },
+    {
+      "epoch": 0.9644265150820573,
+      "learning_rate": 7.1165306683781064e-06,
+      "loss": 0.9675,
+      "mean_token_accuracy": 0.7043153405189514,
+      "num_tokens": 262226147.0,
+      "step": 105190
+    },
+    {
+      "epoch": 0.9645181993215366,
+      "learning_rate": 7.09819382048226e-06,
+      "loss": 0.9573,
+      "mean_token_accuracy": 0.7088045179843903,
+      "num_tokens": 262254458.0,
+      "step": 105200
+    },
+    {
+      "epoch": 0.9646098835610158,
+      "learning_rate": 7.079856972586413e-06,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7146477222442627,
+      "num_tokens": 262283339.0,
+      "step": 105210
+    },
+    {
+      "epoch": 0.9647015678004951,
+      "learning_rate": 7.0615201246905654e-06,
+      "loss": 0.9447,
+      "mean_token_accuracy": 0.7116816759109497,
+      "num_tokens": 262311967.0,
+      "step": 105220
+    },
+    {
+      "epoch": 0.9647932520399743,
+      "learning_rate": 7.04318327679472e-06,
+      "loss": 0.9443,
+      "mean_token_accuracy": 0.7132414698600769,
+      "num_tokens": 262339962.0,
+      "step": 105230
+    },
+    {
+      "epoch": 0.9648849362794536,
+      "learning_rate": 7.0248464288988726e-06,
+      "loss": 0.9504,
+      "mean_token_accuracy": 0.7133159816265107,
+      "num_tokens": 262368672.0,
+      "step": 105240
+    },
+    {
+      "epoch": 0.9649766205189328,
+      "learning_rate": 7.006509581003026e-06,
+      "loss": 0.9547,
+      "mean_token_accuracy": 0.7095539152622223,
+      "num_tokens": 262396747.0,
+      "step": 105250
+    },
+    {
+      "epoch": 0.965068304758412,
+      "learning_rate": 6.988172733107179e-06,
+      "loss": 0.9724,
+      "mean_token_accuracy": 0.7134536623954773,
+      "num_tokens": 262424982.0,
+      "step": 105260
+    },
+    {
+      "epoch": 0.9651599889978912,
+      "learning_rate": 6.969835885211333e-06,
+      "loss": 0.9258,
+      "mean_token_accuracy": 0.7163802027702332,
+      "num_tokens": 262452977.0,
+      "step": 105270
+    },
+    {
+      "epoch": 0.9652516732373705,
+      "learning_rate": 6.951499037315486e-06,
+      "loss": 0.9448,
+      "mean_token_accuracy": 0.7127069056034088,
+      "num_tokens": 262481087.0,
+      "step": 105280
+    },
+    {
+      "epoch": 0.9653433574768497,
+      "learning_rate": 6.9331621894196395e-06,
+      "loss": 0.957,
+      "mean_token_accuracy": 0.7055570185184479,
+      "num_tokens": 262508553.0,
+      "step": 105290
+    },
+    {
+      "epoch": 0.965435041716329,
+      "learning_rate": 6.914825341523792e-06,
+      "loss": 0.9393,
+      "mean_token_accuracy": 0.710353285074234,
+      "num_tokens": 262535151.0,
+      "step": 105300
+    },
+    {
+      "epoch": 0.9655267259558082,
+      "learning_rate": 6.896488493627945e-06,
+      "loss": 0.9435,
+      "mean_token_accuracy": 0.7103923082351684,
+      "num_tokens": 262562783.0,
+      "step": 105310
+    },
+    {
+      "epoch": 0.9656184101952874,
+      "learning_rate": 6.878151645732099e-06,
+      "loss": 0.9524,
+      "mean_token_accuracy": 0.7078456401824951,
+      "num_tokens": 262591205.0,
+      "step": 105320
+    },
+    {
+      "epoch": 0.9657100944347666,
+      "learning_rate": 6.859814797836252e-06,
+      "loss": 0.9518,
+      "mean_token_accuracy": 0.7144934654235839,
+      "num_tokens": 262619282.0,
+      "step": 105330
+    },
+    {
+      "epoch": 0.9658017786742459,
+      "learning_rate": 6.841477949940406e-06,
+      "loss": 0.9788,
+      "mean_token_accuracy": 0.7054666340351105,
+      "num_tokens": 262647481.0,
+      "step": 105340
+    },
+    {
+      "epoch": 0.9658934629137251,
+      "learning_rate": 6.823141102044558e-06,
+      "loss": 0.9646,
+      "mean_token_accuracy": 0.7075164198875428,
+      "num_tokens": 262675147.0,
+      "step": 105350
+    },
+    {
+      "epoch": 0.9659851471532044,
+      "learning_rate": 6.804804254148713e-06,
+      "loss": 0.9408,
+      "mean_token_accuracy": 0.7170015037059784,
+      "num_tokens": 262702677.0,
+      "step": 105360
+    },
+    {
+      "epoch": 0.9660768313926836,
+      "learning_rate": 6.7864674062528655e-06,
+      "loss": 0.95,
+      "mean_token_accuracy": 0.7101018011569977,
+      "num_tokens": 262731098.0,
+      "step": 105370
+    },
+    {
+      "epoch": 0.9661685156321629,
+      "learning_rate": 6.768130558357019e-06,
+      "loss": 0.9564,
+      "mean_token_accuracy": 0.7104446828365326,
+      "num_tokens": 262759224.0,
+      "step": 105380
+    },
+    {
+      "epoch": 0.966260199871642,
+      "learning_rate": 6.749793710461172e-06,
+      "loss": 0.9706,
+      "mean_token_accuracy": 0.7109690129756927,
+      "num_tokens": 262786028.0,
+      "step": 105390
+    },
+    {
+      "epoch": 0.9663518841111213,
+      "learning_rate": 6.731456862565326e-06,
+      "loss": 0.9449,
+      "mean_token_accuracy": 0.7134011149406433,
+      "num_tokens": 262813623.0,
+      "step": 105400
+    },
+    {
+      "epoch": 0.9664435683506005,
+      "learning_rate": 6.713120014669479e-06,
+      "loss": 0.9664,
+      "mean_token_accuracy": 0.7101104557514191,
+      "num_tokens": 262841230.0,
+      "step": 105410
+    },
+    {
+      "epoch": 0.9665352525900798,
+      "learning_rate": 6.694783166773632e-06,
+      "loss": 0.9366,
+      "mean_token_accuracy": 0.7136660933494567,
+      "num_tokens": 262869144.0,
+      "step": 105420
+    },
+    {
+      "epoch": 0.966626936829559,
+      "learning_rate": 6.676446318877785e-06,
+      "loss": 0.9469,
+      "mean_token_accuracy": 0.7105951428413391,
+      "num_tokens": 262897349.0,
+      "step": 105430
+    },
+    {
+      "epoch": 0.9667186210690383,
+      "learning_rate": 6.658109470981938e-06,
+      "loss": 0.9895,
+      "mean_token_accuracy": 0.7003681600093842,
+      "num_tokens": 262925732.0,
+      "step": 105440
+    },
+    {
+      "epoch": 0.9668103053085174,
+      "learning_rate": 6.639772623086092e-06,
+      "loss": 0.9481,
+      "mean_token_accuracy": 0.7137551486492157,
+      "num_tokens": 262953066.0,
+      "step": 105450
+    },
+    {
+      "epoch": 0.9669019895479967,
+      "learning_rate": 6.621435775190245e-06,
+      "loss": 0.9664,
+      "mean_token_accuracy": 0.7069268226623535,
+      "num_tokens": 262980482.0,
+      "step": 105460
+    },
+    {
+      "epoch": 0.9669936737874759,
+      "learning_rate": 6.6030989272943985e-06,
+      "loss": 0.9718,
+      "mean_token_accuracy": 0.7009297668933868,
+      "num_tokens": 263008269.0,
+      "step": 105470
+    },
+    {
+      "epoch": 0.9670853580269552,
+      "learning_rate": 6.584762079398551e-06,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.7061017990112305,
+      "num_tokens": 263037066.0,
+      "step": 105480
+    },
+    {
+      "epoch": 0.9671770422664344,
+      "learning_rate": 6.566425231502706e-06,
+      "loss": 0.9384,
+      "mean_token_accuracy": 0.7098778247833252,
+      "num_tokens": 263064869.0,
+      "step": 105490
+    },
+    {
+      "epoch": 0.9672687265059137,
+      "learning_rate": 6.548088383606858e-06,
+      "loss": 0.9452,
+      "mean_token_accuracy": 0.7145616352558136,
+      "num_tokens": 263092062.0,
+      "step": 105500
+    },
+    {
+      "epoch": 0.9673604107453929,
+      "learning_rate": 6.529751535711012e-06,
+      "loss": 0.9376,
+      "mean_token_accuracy": 0.7121304333209991,
+      "num_tokens": 263120349.0,
+      "step": 105510
+    },
+    {
+      "epoch": 0.9674520949848721,
+      "learning_rate": 6.511414687815165e-06,
+      "loss": 0.9398,
+      "mean_token_accuracy": 0.7116935789585114,
+      "num_tokens": 263148407.0,
+      "step": 105520
+    },
+    {
+      "epoch": 0.9675437792243513,
+      "learning_rate": 6.493077839919317e-06,
+      "loss": 0.9463,
+      "mean_token_accuracy": 0.7124622941017151,
+      "num_tokens": 263176388.0,
+      "step": 105530
+    },
+    {
+      "epoch": 0.9676354634638306,
+      "learning_rate": 6.474740992023472e-06,
+      "loss": 0.9404,
+      "mean_token_accuracy": 0.708881276845932,
+      "num_tokens": 263204048.0,
+      "step": 105540
+    },
+    {
+      "epoch": 0.9677271477033098,
+      "learning_rate": 6.4564041441276245e-06,
+      "loss": 0.9435,
+      "mean_token_accuracy": 0.7130149960517883,
+      "num_tokens": 263232383.0,
+      "step": 105550
+    },
+    {
+      "epoch": 0.9678188319427891,
+      "learning_rate": 6.438067296231778e-06,
+      "loss": 0.9235,
+      "mean_token_accuracy": 0.718024605512619,
+      "num_tokens": 263260435.0,
+      "step": 105560
+    },
+    {
+      "epoch": 0.9679105161822683,
+      "learning_rate": 6.419730448335931e-06,
+      "loss": 0.9416,
+      "mean_token_accuracy": 0.7133037686347962,
+      "num_tokens": 263286886.0,
+      "step": 105570
+    },
+    {
+      "epoch": 0.9680022004217474,
+      "learning_rate": 6.401393600440085e-06,
+      "loss": 0.9785,
+      "mean_token_accuracy": 0.7028957009315491,
+      "num_tokens": 263314053.0,
+      "step": 105580
+    },
+    {
+      "epoch": 0.9680938846612267,
+      "learning_rate": 6.383056752544238e-06,
+      "loss": 0.9432,
+      "mean_token_accuracy": 0.7095845997333526,
+      "num_tokens": 263341797.0,
+      "step": 105590
+    },
+    {
+      "epoch": 0.968185568900706,
+      "learning_rate": 6.3647199046483915e-06,
+      "loss": 0.9373,
+      "mean_token_accuracy": 0.7112450778484345,
+      "num_tokens": 263369146.0,
+      "step": 105600
+    },
+    {
+      "epoch": 0.9682772531401852,
+      "learning_rate": 6.346383056752544e-06,
+      "loss": 0.9494,
+      "mean_token_accuracy": 0.7115423262119294,
+      "num_tokens": 263397012.0,
+      "step": 105610
+    },
+    {
+      "epoch": 0.9683689373796645,
+      "learning_rate": 6.328046208856697e-06,
+      "loss": 0.9983,
+      "mean_token_accuracy": 0.7018362998962402,
+      "num_tokens": 263424803.0,
+      "step": 105620
+    },
+    {
+      "epoch": 0.9684606216191437,
+      "learning_rate": 6.309709360960851e-06,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7116221487522125,
+      "num_tokens": 263451470.0,
+      "step": 105630
+    },
+    {
+      "epoch": 0.968552305858623,
+      "learning_rate": 6.291372513065004e-06,
+      "loss": 0.9268,
+      "mean_token_accuracy": 0.7161617338657379,
+      "num_tokens": 263479191.0,
+      "step": 105640
+    },
+    {
+      "epoch": 0.9686439900981021,
+      "learning_rate": 6.2730356651691576e-06,
+      "loss": 0.9652,
+      "mean_token_accuracy": 0.7056613743305207,
+      "num_tokens": 263507135.0,
+      "step": 105650
+    },
+    {
+      "epoch": 0.9687356743375813,
+      "learning_rate": 6.25469881727331e-06,
+      "loss": 0.9407,
+      "mean_token_accuracy": 0.7149433672428132,
+      "num_tokens": 263534794.0,
+      "step": 105660
+    },
+    {
+      "epoch": 0.9688273585770606,
+      "learning_rate": 6.236361969377464e-06,
+      "loss": 0.962,
+      "mean_token_accuracy": 0.7107852518558502,
+      "num_tokens": 263561540.0,
+      "step": 105670
+    },
+    {
+      "epoch": 0.9689190428165398,
+      "learning_rate": 6.218025121481617e-06,
+      "loss": 0.9334,
+      "mean_token_accuracy": 0.7102938771247864,
+      "num_tokens": 263589535.0,
+      "step": 105680
+    },
+    {
+      "epoch": 0.9690107270560191,
+      "learning_rate": 6.199688273585771e-06,
+      "loss": 0.9606,
+      "mean_token_accuracy": 0.7076380312442779,
+      "num_tokens": 263617009.0,
+      "step": 105690
+    },
+    {
+      "epoch": 0.9691024112954983,
+      "learning_rate": 6.181351425689924e-06,
+      "loss": 0.9445,
+      "mean_token_accuracy": 0.7133013665676117,
+      "num_tokens": 263645206.0,
+      "step": 105700
+    },
+    {
+      "epoch": 0.9691940955349775,
+      "learning_rate": 6.163014577794077e-06,
+      "loss": 0.9477,
+      "mean_token_accuracy": 0.7096609771251678,
+      "num_tokens": 263672933.0,
+      "step": 105710
+    },
+    {
+      "epoch": 0.9692857797744567,
+      "learning_rate": 6.144677729898231e-06,
+      "loss": 0.9296,
+      "mean_token_accuracy": 0.7168624103069305,
+      "num_tokens": 263701127.0,
+      "step": 105720
+    },
+    {
+      "epoch": 0.969377464013936,
+      "learning_rate": 6.126340882002384e-06,
+      "loss": 0.9661,
+      "mean_token_accuracy": 0.7041217029094696,
+      "num_tokens": 263729579.0,
+      "step": 105730
+    },
+    {
+      "epoch": 0.9694691482534152,
+      "learning_rate": 6.108004034106538e-06,
+      "loss": 0.9826,
+      "mean_token_accuracy": 0.7042882442474365,
+      "num_tokens": 263757512.0,
+      "step": 105740
+    },
+    {
+      "epoch": 0.9695608324928945,
+      "learning_rate": 6.089667186210691e-06,
+      "loss": 0.9596,
+      "mean_token_accuracy": 0.7068013250827789,
+      "num_tokens": 263785390.0,
+      "step": 105750
+    },
+    {
+      "epoch": 0.9696525167323737,
+      "learning_rate": 6.071330338314844e-06,
+      "loss": 0.9658,
+      "mean_token_accuracy": 0.7056807041168213,
+      "num_tokens": 263813407.0,
+      "step": 105760
+    },
+    {
+      "epoch": 0.969744200971853,
+      "learning_rate": 6.052993490418997e-06,
+      "loss": 0.933,
+      "mean_token_accuracy": 0.7154004454612732,
+      "num_tokens": 263839629.0,
+      "step": 105770
+    },
+    {
+      "epoch": 0.9698358852113321,
+      "learning_rate": 6.0346566425231505e-06,
+      "loss": 0.9253,
+      "mean_token_accuracy": 0.7129565477371216,
+      "num_tokens": 263867682.0,
+      "step": 105780
+    },
+    {
+      "epoch": 0.9699275694508114,
+      "learning_rate": 6.016319794627304e-06,
+      "loss": 0.9732,
+      "mean_token_accuracy": 0.7071649730205536,
+      "num_tokens": 263895853.0,
+      "step": 105790
+    },
+    {
+      "epoch": 0.9700192536902906,
+      "learning_rate": 5.997982946731457e-06,
+      "loss": 0.9445,
+      "mean_token_accuracy": 0.7088455379009246,
+      "num_tokens": 263924163.0,
+      "step": 105800
+    },
+    {
+      "epoch": 0.9701109379297699,
+      "learning_rate": 5.97964609883561e-06,
+      "loss": 0.9721,
+      "mean_token_accuracy": 0.7049594759941101,
+      "num_tokens": 263951542.0,
+      "step": 105810
+    },
+    {
+      "epoch": 0.9702026221692491,
+      "learning_rate": 5.961309250939764e-06,
+      "loss": 0.968,
+      "mean_token_accuracy": 0.7079796195030212,
+      "num_tokens": 263979628.0,
+      "step": 105820
+    },
+    {
+      "epoch": 0.9702943064087284,
+      "learning_rate": 5.9429724030439174e-06,
+      "loss": 0.959,
+      "mean_token_accuracy": 0.7132765114307403,
+      "num_tokens": 264007484.0,
+      "step": 105830
+    },
+    {
+      "epoch": 0.9703859906482075,
+      "learning_rate": 5.92463555514807e-06,
+      "loss": 0.9537,
+      "mean_token_accuracy": 0.7099129557609558,
+      "num_tokens": 264036186.0,
+      "step": 105840
+    },
+    {
+      "epoch": 0.9704776748876868,
+      "learning_rate": 5.906298707252224e-06,
+      "loss": 0.948,
+      "mean_token_accuracy": 0.7068579435348511,
+      "num_tokens": 264063993.0,
+      "step": 105850
+    },
+    {
+      "epoch": 0.970569359127166,
+      "learning_rate": 5.887961859356377e-06,
+      "loss": 0.9788,
+      "mean_token_accuracy": 0.7074842393398285,
+      "num_tokens": 264091842.0,
+      "step": 105860
+    },
+    {
+      "epoch": 0.9706610433666453,
+      "learning_rate": 5.86962501146053e-06,
+      "loss": 0.9667,
+      "mean_token_accuracy": 0.7055523753166199,
+      "num_tokens": 264119708.0,
+      "step": 105870
+    },
+    {
+      "epoch": 0.9707527276061245,
+      "learning_rate": 5.8512881635646836e-06,
+      "loss": 0.9637,
+      "mean_token_accuracy": 0.7075406670570373,
+      "num_tokens": 264147846.0,
+      "step": 105880
+    },
+    {
+      "epoch": 0.9708444118456038,
+      "learning_rate": 5.832951315668836e-06,
+      "loss": 0.942,
+      "mean_token_accuracy": 0.7142965495586395,
+      "num_tokens": 264175416.0,
+      "step": 105890
+    },
+    {
+      "epoch": 0.970936096085083,
+      "learning_rate": 5.81461446777299e-06,
+      "loss": 0.9655,
+      "mean_token_accuracy": 0.7091979920864105,
+      "num_tokens": 264203049.0,
+      "step": 105900
+    },
+    {
+      "epoch": 0.9710277803245622,
+      "learning_rate": 5.796277619877143e-06,
+      "loss": 0.9635,
+      "mean_token_accuracy": 0.7098026514053345,
+      "num_tokens": 264229934.0,
+      "step": 105910
+    },
+    {
+      "epoch": 0.9711194645640414,
+      "learning_rate": 5.777940771981297e-06,
+      "loss": 0.9288,
+      "mean_token_accuracy": 0.7123159825801849,
+      "num_tokens": 264257309.0,
+      "step": 105920
+    },
+    {
+      "epoch": 0.9712111488035207,
+      "learning_rate": 5.75960392408545e-06,
+      "loss": 0.9748,
+      "mean_token_accuracy": 0.7069181799888611,
+      "num_tokens": 264284430.0,
+      "step": 105930
+    },
+    {
+      "epoch": 0.9713028330429999,
+      "learning_rate": 5.741267076189603e-06,
+      "loss": 0.9513,
+      "mean_token_accuracy": 0.7107536494731903,
+      "num_tokens": 264312628.0,
+      "step": 105940
+    },
+    {
+      "epoch": 0.9713945172824792,
+      "learning_rate": 5.722930228293757e-06,
+      "loss": 0.9587,
+      "mean_token_accuracy": 0.7086152553558349,
+      "num_tokens": 264341729.0,
+      "step": 105950
+    },
+    {
+      "epoch": 0.9714862015219584,
+      "learning_rate": 5.70459338039791e-06,
+      "loss": 0.9469,
+      "mean_token_accuracy": 0.7136449694633484,
+      "num_tokens": 264369762.0,
+      "step": 105960
+    },
+    {
+      "epoch": 0.9715778857614376,
+      "learning_rate": 5.686256532502064e-06,
+      "loss": 0.9527,
+      "mean_token_accuracy": 0.7097286641597748,
+      "num_tokens": 264396983.0,
+      "step": 105970
+    },
+    {
+      "epoch": 0.9716695700009168,
+      "learning_rate": 5.667919684606216e-06,
+      "loss": 0.9564,
+      "mean_token_accuracy": 0.7044010639190674,
+      "num_tokens": 264424606.0,
+      "step": 105980
+    },
+    {
+      "epoch": 0.971761254240396,
+      "learning_rate": 5.649582836710369e-06,
+      "loss": 0.9512,
+      "mean_token_accuracy": 0.7110239386558532,
+      "num_tokens": 264453567.0,
+      "step": 105990
+    },
+    {
+      "epoch": 0.9718529384798753,
+      "learning_rate": 5.631245988814523e-06,
+      "loss": 0.939,
+      "mean_token_accuracy": 0.7180724620819092,
+      "num_tokens": 264481233.0,
+      "step": 106000
+    },
+    {
+      "epoch": 0.9719446227193546,
+      "learning_rate": 5.6129091409186765e-06,
+      "loss": 0.9823,
+      "mean_token_accuracy": 0.7070731580257416,
+      "num_tokens": 264508448.0,
+      "step": 106010
+    },
+    {
+      "epoch": 0.9720363069588338,
+      "learning_rate": 5.59457229302283e-06,
+      "loss": 0.9581,
+      "mean_token_accuracy": 0.7078408002853394,
+      "num_tokens": 264535958.0,
+      "step": 106020
+    },
+    {
+      "epoch": 0.9721279911983131,
+      "learning_rate": 5.576235445126983e-06,
+      "loss": 0.9699,
+      "mean_token_accuracy": 0.7089482188224793,
+      "num_tokens": 264563452.0,
+      "step": 106030
+    },
+    {
+      "epoch": 0.9722196754377922,
+      "learning_rate": 5.557898597231136e-06,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.7037977755069733,
+      "num_tokens": 264591147.0,
+      "step": 106040
+    },
+    {
+      "epoch": 0.9723113596772714,
+      "learning_rate": 5.53956174933529e-06,
+      "loss": 0.9537,
+      "mean_token_accuracy": 0.7138411521911621,
+      "num_tokens": 264619344.0,
+      "step": 106050
+    },
+    {
+      "epoch": 0.9724030439167507,
+      "learning_rate": 5.5212249014394434e-06,
+      "loss": 0.9313,
+      "mean_token_accuracy": 0.7180680215358735,
+      "num_tokens": 264647475.0,
+      "step": 106060
+    },
+    {
+      "epoch": 0.97249472815623,
+      "learning_rate": 5.502888053543596e-06,
+      "loss": 0.9589,
+      "mean_token_accuracy": 0.7090674996376037,
+      "num_tokens": 264674309.0,
+      "step": 106070
+    },
+    {
+      "epoch": 0.9725864123957092,
+      "learning_rate": 5.484551205647749e-06,
+      "loss": 0.9751,
+      "mean_token_accuracy": 0.7034432351589203,
+      "num_tokens": 264702248.0,
+      "step": 106080
+    },
+    {
+      "epoch": 0.9726780966351884,
+      "learning_rate": 5.466214357751902e-06,
+      "loss": 0.9138,
+      "mean_token_accuracy": 0.7175294697284699,
+      "num_tokens": 264729778.0,
+      "step": 106090
+    },
+    {
+      "epoch": 0.9727697808746677,
+      "learning_rate": 5.447877509856056e-06,
+      "loss": 0.96,
+      "mean_token_accuracy": 0.7127042353153229,
+      "num_tokens": 264757489.0,
+      "step": 106100
+    },
+    {
+      "epoch": 0.9728614651141468,
+      "learning_rate": 5.4295406619602095e-06,
+      "loss": 0.9158,
+      "mean_token_accuracy": 0.7221392869949341,
+      "num_tokens": 264785683.0,
+      "step": 106110
+    },
+    {
+      "epoch": 0.9729531493536261,
+      "learning_rate": 5.411203814064362e-06,
+      "loss": 0.9778,
+      "mean_token_accuracy": 0.7075231373310089,
+      "num_tokens": 264813270.0,
+      "step": 106120
+    },
+    {
+      "epoch": 0.9730448335931053,
+      "learning_rate": 5.392866966168516e-06,
+      "loss": 0.9207,
+      "mean_token_accuracy": 0.7147023320198059,
+      "num_tokens": 264840288.0,
+      "step": 106130
+    },
+    {
+      "epoch": 0.9731365178325846,
+      "learning_rate": 5.374530118272669e-06,
+      "loss": 0.9514,
+      "mean_token_accuracy": 0.7081270277500152,
+      "num_tokens": 264868023.0,
+      "step": 106140
+    },
+    {
+      "epoch": 0.9732282020720638,
+      "learning_rate": 5.356193270376823e-06,
+      "loss": 0.9371,
+      "mean_token_accuracy": 0.7109180867671967,
+      "num_tokens": 264895408.0,
+      "step": 106150
+    },
+    {
+      "epoch": 0.9733198863115431,
+      "learning_rate": 5.337856422480976e-06,
+      "loss": 0.9578,
+      "mean_token_accuracy": 0.7068678259849548,
+      "num_tokens": 264922737.0,
+      "step": 106160
+    },
+    {
+      "epoch": 0.9734115705510222,
+      "learning_rate": 5.319519574585129e-06,
+      "loss": 0.9543,
+      "mean_token_accuracy": 0.7071869909763336,
+      "num_tokens": 264951109.0,
+      "step": 106170
+    },
+    {
+      "epoch": 0.9735032547905015,
+      "learning_rate": 5.301182726689282e-06,
+      "loss": 0.9513,
+      "mean_token_accuracy": 0.711639142036438,
+      "num_tokens": 264978504.0,
+      "step": 106180
+    },
+    {
+      "epoch": 0.9735949390299807,
+      "learning_rate": 5.2828458787934355e-06,
+      "loss": 0.9569,
+      "mean_token_accuracy": 0.7065433025360107,
+      "num_tokens": 265006590.0,
+      "step": 106190
+    },
+    {
+      "epoch": 0.97368662326946,
+      "learning_rate": 5.264509030897589e-06,
+      "loss": 0.9273,
+      "mean_token_accuracy": 0.7189335882663727,
+      "num_tokens": 265035287.0,
+      "step": 106200
+    },
+    {
+      "epoch": 0.9737783075089392,
+      "learning_rate": 5.246172183001742e-06,
+      "loss": 0.9557,
+      "mean_token_accuracy": 0.7066478848457336,
+      "num_tokens": 265062813.0,
+      "step": 106210
+    },
+    {
+      "epoch": 0.9738699917484185,
+      "learning_rate": 5.227835335105895e-06,
+      "loss": 0.9384,
+      "mean_token_accuracy": 0.7121494829654693,
+      "num_tokens": 265090979.0,
+      "step": 106220
+    },
+    {
+      "epoch": 0.9739616759878977,
+      "learning_rate": 5.209498487210049e-06,
+      "loss": 0.9563,
+      "mean_token_accuracy": 0.7068563342094422,
+      "num_tokens": 265118358.0,
+      "step": 106230
+    },
+    {
+      "epoch": 0.9740533602273769,
+      "learning_rate": 5.1911616393142025e-06,
+      "loss": 0.9583,
+      "mean_token_accuracy": 0.7092251658439637,
+      "num_tokens": 265146472.0,
+      "step": 106240
+    },
+    {
+      "epoch": 0.9741450444668561,
+      "learning_rate": 5.172824791418355e-06,
+      "loss": 0.9762,
+      "mean_token_accuracy": 0.7026823461055756,
+      "num_tokens": 265174138.0,
+      "step": 106250
+    },
+    {
+      "epoch": 0.9742367287063354,
+      "learning_rate": 5.154487943522509e-06,
+      "loss": 0.9571,
+      "mean_token_accuracy": 0.7108917355537414,
+      "num_tokens": 265201558.0,
+      "step": 106260
+    },
+    {
+      "epoch": 0.9743284129458146,
+      "learning_rate": 5.136151095626662e-06,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.7082030355930329,
+      "num_tokens": 265230750.0,
+      "step": 106270
+    },
+    {
+      "epoch": 0.9744200971852939,
+      "learning_rate": 5.117814247730816e-06,
+      "loss": 0.9732,
+      "mean_token_accuracy": 0.7060872912406921,
+      "num_tokens": 265257747.0,
+      "step": 106280
+    },
+    {
+      "epoch": 0.9745117814247731,
+      "learning_rate": 5.0994773998349686e-06,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.7112521946430206,
+      "num_tokens": 265285523.0,
+      "step": 106290
+    },
+    {
+      "epoch": 0.9746034656642523,
+      "learning_rate": 5.081140551939121e-06,
+      "loss": 0.9535,
+      "mean_token_accuracy": 0.7129338502883911,
+      "num_tokens": 265313178.0,
+      "step": 106300
+    },
+    {
+      "epoch": 0.9746951499037315,
+      "learning_rate": 5.062803704043275e-06,
+      "loss": 0.9557,
+      "mean_token_accuracy": 0.7092982530593872,
+      "num_tokens": 265340824.0,
+      "step": 106310
+    },
+    {
+      "epoch": 0.9747868341432108,
+      "learning_rate": 5.044466856147428e-06,
+      "loss": 0.9406,
+      "mean_token_accuracy": 0.7075697779655457,
+      "num_tokens": 265368493.0,
+      "step": 106320
+    },
+    {
+      "epoch": 0.97487851838269,
+      "learning_rate": 5.026130008251582e-06,
+      "loss": 0.9494,
+      "mean_token_accuracy": 0.7144442856311798,
+      "num_tokens": 265396552.0,
+      "step": 106330
+    },
+    {
+      "epoch": 0.9749702026221693,
+      "learning_rate": 5.0077931603557355e-06,
+      "loss": 0.9342,
+      "mean_token_accuracy": 0.715404623746872,
+      "num_tokens": 265424256.0,
+      "step": 106340
+    },
+    {
+      "epoch": 0.9750618868616485,
+      "learning_rate": 4.989456312459888e-06,
+      "loss": 0.9462,
+      "mean_token_accuracy": 0.7162149667739868,
+      "num_tokens": 265451815.0,
+      "step": 106350
+    },
+    {
+      "epoch": 0.9751535711011278,
+      "learning_rate": 4.971119464564042e-06,
+      "loss": 0.979,
+      "mean_token_accuracy": 0.7023785650730133,
+      "num_tokens": 265480157.0,
+      "step": 106360
+    },
+    {
+      "epoch": 0.9752452553406069,
+      "learning_rate": 4.952782616668195e-06,
+      "loss": 0.9623,
+      "mean_token_accuracy": 0.7010857045650483,
+      "num_tokens": 265508075.0,
+      "step": 106370
+    },
+    {
+      "epoch": 0.9753369395800862,
+      "learning_rate": 4.934445768772349e-06,
+      "loss": 0.9822,
+      "mean_token_accuracy": 0.7071395516395569,
+      "num_tokens": 265535857.0,
+      "step": 106380
+    },
+    {
+      "epoch": 0.9754286238195654,
+      "learning_rate": 4.916108920876502e-06,
+      "loss": 0.9742,
+      "mean_token_accuracy": 0.7060118317604065,
+      "num_tokens": 265563541.0,
+      "step": 106390
+    },
+    {
+      "epoch": 0.9755203080590447,
+      "learning_rate": 4.897772072980654e-06,
+      "loss": 0.9035,
+      "mean_token_accuracy": 0.7206015467643738,
+      "num_tokens": 265591285.0,
+      "step": 106400
+    },
+    {
+      "epoch": 0.9756119922985239,
+      "learning_rate": 4.879435225084808e-06,
+      "loss": 0.9435,
+      "mean_token_accuracy": 0.7146306753158569,
+      "num_tokens": 265618805.0,
+      "step": 106410
+    },
+    {
+      "epoch": 0.9757036765380032,
+      "learning_rate": 4.8610983771889615e-06,
+      "loss": 0.972,
+      "mean_token_accuracy": 0.7071013391017914,
+      "num_tokens": 265647800.0,
+      "step": 106420
+    },
+    {
+      "epoch": 0.9757953607774823,
+      "learning_rate": 4.842761529293115e-06,
+      "loss": 0.9456,
+      "mean_token_accuracy": 0.7121579349040985,
+      "num_tokens": 265675256.0,
+      "step": 106430
+    },
+    {
+      "epoch": 0.9758870450169616,
+      "learning_rate": 4.824424681397268e-06,
+      "loss": 0.9694,
+      "mean_token_accuracy": 0.7056652307510376,
+      "num_tokens": 265703943.0,
+      "step": 106440
+    },
+    {
+      "epoch": 0.9759787292564408,
+      "learning_rate": 4.806087833501421e-06,
+      "loss": 0.9331,
+      "mean_token_accuracy": 0.7131913483142853,
+      "num_tokens": 265731873.0,
+      "step": 106450
+    },
+    {
+      "epoch": 0.97607041349592,
+      "learning_rate": 4.787750985605575e-06,
+      "loss": 0.9562,
+      "mean_token_accuracy": 0.7071906805038453,
+      "num_tokens": 265759759.0,
+      "step": 106460
+    },
+    {
+      "epoch": 0.9761620977353993,
+      "learning_rate": 4.7694141377097284e-06,
+      "loss": 0.9534,
+      "mean_token_accuracy": 0.7083316087722779,
+      "num_tokens": 265787275.0,
+      "step": 106470
+    },
+    {
+      "epoch": 0.9762537819748786,
+      "learning_rate": 4.751077289813881e-06,
+      "loss": 0.9486,
+      "mean_token_accuracy": 0.7126296639442444,
+      "num_tokens": 265815670.0,
+      "step": 106480
+    },
+    {
+      "epoch": 0.9763454662143578,
+      "learning_rate": 4.732740441918034e-06,
+      "loss": 0.9639,
+      "mean_token_accuracy": 0.7070540189743042,
+      "num_tokens": 265843969.0,
+      "step": 106490
+    },
+    {
+      "epoch": 0.9764371504538369,
+      "learning_rate": 4.7144035940221874e-06,
+      "loss": 0.9686,
+      "mean_token_accuracy": 0.710118442773819,
+      "num_tokens": 265872055.0,
+      "step": 106500
+    },
+    {
+      "epoch": 0.9765288346933162,
+      "learning_rate": 4.696066746126341e-06,
+      "loss": 0.979,
+      "mean_token_accuracy": 0.7022459447383881,
+      "num_tokens": 265900485.0,
+      "step": 106510
+    },
+    {
+      "epoch": 0.9766205189327954,
+      "learning_rate": 4.6777298982304946e-06,
+      "loss": 0.9562,
+      "mean_token_accuracy": 0.7102384209632874,
+      "num_tokens": 265927929.0,
+      "step": 106520
+    },
+    {
+      "epoch": 0.9767122031722747,
+      "learning_rate": 4.659393050334647e-06,
+      "loss": 0.9668,
+      "mean_token_accuracy": 0.7090049982070923,
+      "num_tokens": 265956047.0,
+      "step": 106530
+    },
+    {
+      "epoch": 0.976803887411754,
+      "learning_rate": 4.641056202438801e-06,
+      "loss": 0.9515,
+      "mean_token_accuracy": 0.709384435415268,
+      "num_tokens": 265984890.0,
+      "step": 106540
+    },
+    {
+      "epoch": 0.9768955716512332,
+      "learning_rate": 4.622719354542954e-06,
+      "loss": 0.9064,
+      "mean_token_accuracy": 0.7182036519050599,
+      "num_tokens": 266012722.0,
+      "step": 106550
+    },
+    {
+      "epoch": 0.9769872558907123,
+      "learning_rate": 4.604382506647108e-06,
+      "loss": 0.9884,
+      "mean_token_accuracy": 0.7036374330520629,
+      "num_tokens": 266041019.0,
+      "step": 106560
+    },
+    {
+      "epoch": 0.9770789401301916,
+      "learning_rate": 4.5860456587512615e-06,
+      "loss": 0.9733,
+      "mean_token_accuracy": 0.7008701801300049,
+      "num_tokens": 266068470.0,
+      "step": 106570
+    },
+    {
+      "epoch": 0.9771706243696708,
+      "learning_rate": 4.567708810855414e-06,
+      "loss": 0.9794,
+      "mean_token_accuracy": 0.7040487945079803,
+      "num_tokens": 266096213.0,
+      "step": 106580
+    },
+    {
+      "epoch": 0.9772623086091501,
+      "learning_rate": 4.549371962959567e-06,
+      "loss": 0.9669,
+      "mean_token_accuracy": 0.7039185881614685,
+      "num_tokens": 266124188.0,
+      "step": 106590
+    },
+    {
+      "epoch": 0.9773539928486293,
+      "learning_rate": 4.5310351150637205e-06,
+      "loss": 0.961,
+      "mean_token_accuracy": 0.7082109749317169,
+      "num_tokens": 266151873.0,
+      "step": 106600
+    },
+    {
+      "epoch": 0.9774456770881086,
+      "learning_rate": 4.512698267167874e-06,
+      "loss": 0.9339,
+      "mean_token_accuracy": 0.7134939074516297,
+      "num_tokens": 266179822.0,
+      "step": 106610
+    },
+    {
+      "epoch": 0.9775373613275878,
+      "learning_rate": 4.494361419272028e-06,
+      "loss": 0.9337,
+      "mean_token_accuracy": 0.718496972322464,
+      "num_tokens": 266208001.0,
+      "step": 106620
+    },
+    {
+      "epoch": 0.977629045567067,
+      "learning_rate": 4.47602457137618e-06,
+      "loss": 0.9424,
+      "mean_token_accuracy": 0.7149182856082916,
+      "num_tokens": 266236315.0,
+      "step": 106630
+    },
+    {
+      "epoch": 0.9777207298065462,
+      "learning_rate": 4.457687723480334e-06,
+      "loss": 0.9821,
+      "mean_token_accuracy": 0.7090541541576385,
+      "num_tokens": 266265058.0,
+      "step": 106640
+    },
+    {
+      "epoch": 0.9778124140460255,
+      "learning_rate": 4.4393508755844875e-06,
+      "loss": 0.9424,
+      "mean_token_accuracy": 0.7143906235694886,
+      "num_tokens": 266292251.0,
+      "step": 106650
+    },
+    {
+      "epoch": 0.9779040982855047,
+      "learning_rate": 4.421014027688641e-06,
+      "loss": 0.9228,
+      "mean_token_accuracy": 0.7171306908130646,
+      "num_tokens": 266319962.0,
+      "step": 106660
+    },
+    {
+      "epoch": 0.977995782524984,
+      "learning_rate": 4.402677179792794e-06,
+      "loss": 0.9609,
+      "mean_token_accuracy": 0.7086678266525268,
+      "num_tokens": 266347973.0,
+      "step": 106670
+    },
+    {
+      "epoch": 0.9780874667644632,
+      "learning_rate": 4.384340331896947e-06,
+      "loss": 0.9419,
+      "mean_token_accuracy": 0.7154306769371033,
+      "num_tokens": 266376650.0,
+      "step": 106680
+    },
+    {
+      "epoch": 0.9781791510039424,
+      "learning_rate": 4.366003484001101e-06,
+      "loss": 0.9431,
+      "mean_token_accuracy": 0.7204812347888947,
+      "num_tokens": 266404202.0,
+      "step": 106690
+    },
+    {
+      "epoch": 0.9782708352434216,
+      "learning_rate": 4.347666636105254e-06,
+      "loss": 0.9713,
+      "mean_token_accuracy": 0.7069407284259797,
+      "num_tokens": 266432172.0,
+      "step": 106700
+    },
+    {
+      "epoch": 0.9783625194829009,
+      "learning_rate": 4.329329788209407e-06,
+      "loss": 0.9628,
+      "mean_token_accuracy": 0.7112584054470062,
+      "num_tokens": 266459803.0,
+      "step": 106710
+    },
+    {
+      "epoch": 0.9784542037223801,
+      "learning_rate": 4.31099294031356e-06,
+      "loss": 0.9202,
+      "mean_token_accuracy": 0.7176357507705688,
+      "num_tokens": 266488015.0,
+      "step": 106720
+    },
+    {
+      "epoch": 0.9785458879618594,
+      "learning_rate": 4.292656092417713e-06,
+      "loss": 0.9538,
+      "mean_token_accuracy": 0.7114960134029389,
+      "num_tokens": 266515594.0,
+      "step": 106730
+    },
+    {
+      "epoch": 0.9786375722013386,
+      "learning_rate": 4.274319244521867e-06,
+      "loss": 0.9458,
+      "mean_token_accuracy": 0.7117897689342498,
+      "num_tokens": 266543314.0,
+      "step": 106740
+    },
+    {
+      "epoch": 0.9787292564408179,
+      "learning_rate": 4.2559823966260205e-06,
+      "loss": 0.968,
+      "mean_token_accuracy": 0.7120352983474731,
+      "num_tokens": 266571211.0,
+      "step": 106750
+    },
+    {
+      "epoch": 0.978820940680297,
+      "learning_rate": 4.237645548730173e-06,
+      "loss": 0.9492,
+      "mean_token_accuracy": 0.7072612285614014,
+      "num_tokens": 266599142.0,
+      "step": 106760
+    },
+    {
+      "epoch": 0.9789126249197763,
+      "learning_rate": 4.219308700834327e-06,
+      "loss": 0.9631,
+      "mean_token_accuracy": 0.7073183059692383,
+      "num_tokens": 266627729.0,
+      "step": 106770
+    },
+    {
+      "epoch": 0.9790043091592555,
+      "learning_rate": 4.20097185293848e-06,
+      "loss": 0.9625,
+      "mean_token_accuracy": 0.7120635509490967,
+      "num_tokens": 266656536.0,
+      "step": 106780
+    },
+    {
+      "epoch": 0.9790959933987348,
+      "learning_rate": 4.182635005042634e-06,
+      "loss": 0.9282,
+      "mean_token_accuracy": 0.7160631656646729,
+      "num_tokens": 266684094.0,
+      "step": 106790
+    },
+    {
+      "epoch": 0.979187677638214,
+      "learning_rate": 4.164298157146787e-06,
+      "loss": 0.9467,
+      "mean_token_accuracy": 0.7072149753570557,
+      "num_tokens": 266712354.0,
+      "step": 106800
+    },
+    {
+      "epoch": 0.9792793618776933,
+      "learning_rate": 4.145961309250939e-06,
+      "loss": 0.9457,
+      "mean_token_accuracy": 0.7111077785491944,
+      "num_tokens": 266740969.0,
+      "step": 106810
+    },
+    {
+      "epoch": 0.9793710461171724,
+      "learning_rate": 4.127624461355093e-06,
+      "loss": 0.9679,
+      "mean_token_accuracy": 0.7086487650871277,
+      "num_tokens": 266768822.0,
+      "step": 106820
+    },
+    {
+      "epoch": 0.9794627303566517,
+      "learning_rate": 4.1092876134592465e-06,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.713647437095642,
+      "num_tokens": 266795807.0,
+      "step": 106830
+    },
+    {
+      "epoch": 0.9795544145961309,
+      "learning_rate": 4.0909507655634e-06,
+      "loss": 0.9546,
+      "mean_token_accuracy": 0.7055732071399688,
+      "num_tokens": 266825252.0,
+      "step": 106840
+    },
+    {
+      "epoch": 0.9796460988356102,
+      "learning_rate": 4.072613917667553e-06,
+      "loss": 0.936,
+      "mean_token_accuracy": 0.7111010015010834,
+      "num_tokens": 266854538.0,
+      "step": 106850
+    },
+    {
+      "epoch": 0.9797377830750894,
+      "learning_rate": 4.054277069771706e-06,
+      "loss": 0.9588,
+      "mean_token_accuracy": 0.7043581068515777,
+      "num_tokens": 266882684.0,
+      "step": 106860
+    },
+    {
+      "epoch": 0.9798294673145687,
+      "learning_rate": 4.03594022187586e-06,
+      "loss": 0.9505,
+      "mean_token_accuracy": 0.7151164412498474,
+      "num_tokens": 266911360.0,
+      "step": 106870
+    },
+    {
+      "epoch": 0.9799211515540479,
+      "learning_rate": 4.0176033739800135e-06,
+      "loss": 0.9509,
+      "mean_token_accuracy": 0.7084515392780304,
+      "num_tokens": 266939420.0,
+      "step": 106880
+    },
+    {
+      "epoch": 0.980012835793527,
+      "learning_rate": 3.999266526084167e-06,
+      "loss": 0.9493,
+      "mean_token_accuracy": 0.7123298645019531,
+      "num_tokens": 266967245.0,
+      "step": 106890
+    },
+    {
+      "epoch": 0.9801045200330063,
+      "learning_rate": 3.980929678188319e-06,
+      "loss": 0.9588,
+      "mean_token_accuracy": 0.7104962050914765,
+      "num_tokens": 266994626.0,
+      "step": 106900
+    },
+    {
+      "epoch": 0.9801962042724855,
+      "learning_rate": 3.9625928302924724e-06,
+      "loss": 0.9501,
+      "mean_token_accuracy": 0.7113950729370118,
+      "num_tokens": 267022198.0,
+      "step": 106910
+    },
+    {
+      "epoch": 0.9802878885119648,
+      "learning_rate": 3.944255982396626e-06,
+      "loss": 0.9536,
+      "mean_token_accuracy": 0.7085495054721832,
+      "num_tokens": 267050493.0,
+      "step": 106920
+    },
+    {
+      "epoch": 0.980379572751444,
+      "learning_rate": 3.9259191345007796e-06,
+      "loss": 0.9789,
+      "mean_token_accuracy": 0.7024223268032074,
+      "num_tokens": 267078180.0,
+      "step": 106930
+    },
+    {
+      "epoch": 0.9804712569909233,
+      "learning_rate": 3.907582286604933e-06,
+      "loss": 0.9732,
+      "mean_token_accuracy": 0.7009832084178924,
+      "num_tokens": 267106692.0,
+      "step": 106940
+    },
+    {
+      "epoch": 0.9805629412304024,
+      "learning_rate": 3.889245438709086e-06,
+      "loss": 0.941,
+      "mean_token_accuracy": 0.7072395741939544,
+      "num_tokens": 267135277.0,
+      "step": 106950
+    },
+    {
+      "epoch": 0.9806546254698817,
+      "learning_rate": 3.870908590813239e-06,
+      "loss": 0.9739,
+      "mean_token_accuracy": 0.7103749692440033,
+      "num_tokens": 267163665.0,
+      "step": 106960
+    },
+    {
+      "epoch": 0.9807463097093609,
+      "learning_rate": 3.852571742917393e-06,
+      "loss": 0.9434,
+      "mean_token_accuracy": 0.7175757586956024,
+      "num_tokens": 267191513.0,
+      "step": 106970
+    },
+    {
+      "epoch": 0.9808379939488402,
+      "learning_rate": 3.8342348950215465e-06,
+      "loss": 0.9247,
+      "mean_token_accuracy": 0.7166262805461884,
+      "num_tokens": 267219257.0,
+      "step": 106980
+    },
+    {
+      "epoch": 0.9809296781883194,
+      "learning_rate": 3.815898047125699e-06,
+      "loss": 0.9671,
+      "mean_token_accuracy": 0.7055789470672608,
+      "num_tokens": 267247251.0,
+      "step": 106990
+    },
+    {
+      "epoch": 0.9810213624277987,
+      "learning_rate": 3.7975611992298524e-06,
+      "loss": 0.942,
+      "mean_token_accuracy": 0.7121406674385071,
+      "num_tokens": 267275112.0,
+      "step": 107000
+    },
+    {
+      "epoch": 0.9811130466672779,
+      "learning_rate": 3.7792243513340055e-06,
+      "loss": 0.9329,
+      "mean_token_accuracy": 0.7172952950000763,
+      "num_tokens": 267302819.0,
+      "step": 107010
+    },
+    {
+      "epoch": 0.9812047309067571,
+      "learning_rate": 3.760887503438159e-06,
+      "loss": 0.9454,
+      "mean_token_accuracy": 0.7095207214355469,
+      "num_tokens": 267330985.0,
+      "step": 107020
+    },
+    {
+      "epoch": 0.9812964151462363,
+      "learning_rate": 3.7425506555423122e-06,
+      "loss": 0.9647,
+      "mean_token_accuracy": 0.7062365531921386,
+      "num_tokens": 267358692.0,
+      "step": 107030
+    },
+    {
+      "epoch": 0.9813880993857156,
+      "learning_rate": 3.7242138076464658e-06,
+      "loss": 0.982,
+      "mean_token_accuracy": 0.7057001054286957,
+      "num_tokens": 267387786.0,
+      "step": 107040
+    },
+    {
+      "epoch": 0.9814797836251948,
+      "learning_rate": 3.705876959750619e-06,
+      "loss": 0.9314,
+      "mean_token_accuracy": 0.7117950439453125,
+      "num_tokens": 267416620.0,
+      "step": 107050
+    },
+    {
+      "epoch": 0.9815714678646741,
+      "learning_rate": 3.6875401118547725e-06,
+      "loss": 0.9199,
+      "mean_token_accuracy": 0.7160579800605774,
+      "num_tokens": 267444554.0,
+      "step": 107060
+    },
+    {
+      "epoch": 0.9816631521041533,
+      "learning_rate": 3.6692032639589256e-06,
+      "loss": 0.929,
+      "mean_token_accuracy": 0.7167559683322906,
+      "num_tokens": 267472585.0,
+      "step": 107070
+    },
+    {
+      "epoch": 0.9817548363436325,
+      "learning_rate": 3.650866416063079e-06,
+      "loss": 0.9428,
+      "mean_token_accuracy": 0.7125068247318268,
+      "num_tokens": 267499728.0,
+      "step": 107080
+    },
+    {
+      "epoch": 0.9818465205831117,
+      "learning_rate": 3.6325295681672323e-06,
+      "loss": 0.9567,
+      "mean_token_accuracy": 0.7096069157123566,
+      "num_tokens": 267527845.0,
+      "step": 107090
+    },
+    {
+      "epoch": 0.981938204822591,
+      "learning_rate": 3.614192720271386e-06,
+      "loss": 0.9611,
+      "mean_token_accuracy": 0.7039044857025146,
+      "num_tokens": 267555763.0,
+      "step": 107100
+    },
+    {
+      "epoch": 0.9820298890620702,
+      "learning_rate": 3.5958558723755386e-06,
+      "loss": 0.9461,
+      "mean_token_accuracy": 0.7165573835372925,
+      "num_tokens": 267583758.0,
+      "step": 107110
+    },
+    {
+      "epoch": 0.9821215733015495,
+      "learning_rate": 3.5775190244796917e-06,
+      "loss": 0.9668,
+      "mean_token_accuracy": 0.7083516955375672,
+      "num_tokens": 267611223.0,
+      "step": 107120
+    },
+    {
+      "epoch": 0.9822132575410287,
+      "learning_rate": 3.5591821765838453e-06,
+      "loss": 0.9754,
+      "mean_token_accuracy": 0.7050501585006714,
+      "num_tokens": 267639537.0,
+      "step": 107130
+    },
+    {
+      "epoch": 0.982304941780508,
+      "learning_rate": 3.5408453286879984e-06,
+      "loss": 0.9538,
+      "mean_token_accuracy": 0.7140101850032806,
+      "num_tokens": 267667864.0,
+      "step": 107140
+    },
+    {
+      "epoch": 0.9823966260199871,
+      "learning_rate": 3.522508480792152e-06,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7126223623752594,
+      "num_tokens": 267696208.0,
+      "step": 107150
+    },
+    {
+      "epoch": 0.9824883102594664,
+      "learning_rate": 3.5041716328963056e-06,
+      "loss": 0.9257,
+      "mean_token_accuracy": 0.7188680231571197,
+      "num_tokens": 267723540.0,
+      "step": 107160
+    },
+    {
+      "epoch": 0.9825799944989456,
+      "learning_rate": 3.4858347850004587e-06,
+      "loss": 0.9418,
+      "mean_token_accuracy": 0.7131708323955536,
+      "num_tokens": 267750849.0,
+      "step": 107170
+    },
+    {
+      "epoch": 0.9826716787384249,
+      "learning_rate": 3.4674979371046122e-06,
+      "loss": 0.9241,
+      "mean_token_accuracy": 0.715410715341568,
+      "num_tokens": 267778672.0,
+      "step": 107180
+    },
+    {
+      "epoch": 0.9827633629779041,
+      "learning_rate": 3.4491610892087654e-06,
+      "loss": 0.9635,
+      "mean_token_accuracy": 0.7099226772785187,
+      "num_tokens": 267806503.0,
+      "step": 107190
+    },
+    {
+      "epoch": 0.9828550472173834,
+      "learning_rate": 3.430824241312919e-06,
+      "loss": 0.9219,
+      "mean_token_accuracy": 0.7198168754577636,
+      "num_tokens": 267833691.0,
+      "step": 107200
+    },
+    {
+      "epoch": 0.9829467314568625,
+      "learning_rate": 3.4124873934170717e-06,
+      "loss": 0.9576,
+      "mean_token_accuracy": 0.7134953022003174,
+      "num_tokens": 267862815.0,
+      "step": 107210
+    },
+    {
+      "epoch": 0.9830384156963418,
+      "learning_rate": 3.394150545521225e-06,
+      "loss": 0.9451,
+      "mean_token_accuracy": 0.7121977984905243,
+      "num_tokens": 267891105.0,
+      "step": 107220
+    },
+    {
+      "epoch": 0.983130099935821,
+      "learning_rate": 3.3758136976253784e-06,
+      "loss": 0.9592,
+      "mean_token_accuracy": 0.7109839439392089,
+      "num_tokens": 267918486.0,
+      "step": 107230
+    },
+    {
+      "epoch": 0.9832217841753003,
+      "learning_rate": 3.3574768497295315e-06,
+      "loss": 0.968,
+      "mean_token_accuracy": 0.7080407559871673,
+      "num_tokens": 267946117.0,
+      "step": 107240
+    },
+    {
+      "epoch": 0.9833134684147795,
+      "learning_rate": 3.339140001833685e-06,
+      "loss": 0.951,
+      "mean_token_accuracy": 0.7099920630455017,
+      "num_tokens": 267973868.0,
+      "step": 107250
+    },
+    {
+      "epoch": 0.9834051526542588,
+      "learning_rate": 3.320803153937838e-06,
+      "loss": 0.9451,
+      "mean_token_accuracy": 0.7140229761600494,
+      "num_tokens": 268002045.0,
+      "step": 107260
+    },
+    {
+      "epoch": 0.983496836893738,
+      "learning_rate": 3.3024663060419918e-06,
+      "loss": 0.9356,
+      "mean_token_accuracy": 0.7128445625305175,
+      "num_tokens": 268029878.0,
+      "step": 107270
+    },
+    {
+      "epoch": 0.9835885211332172,
+      "learning_rate": 3.284129458146145e-06,
+      "loss": 0.9735,
+      "mean_token_accuracy": 0.7094855904579163,
+      "num_tokens": 268058681.0,
+      "step": 107280
+    },
+    {
+      "epoch": 0.9836802053726964,
+      "learning_rate": 3.2657926102502985e-06,
+      "loss": 0.9508,
+      "mean_token_accuracy": 0.7092310309410095,
+      "num_tokens": 268086448.0,
+      "step": 107290
+    },
+    {
+      "epoch": 0.9837718896121757,
+      "learning_rate": 3.2474557623544516e-06,
+      "loss": 0.9507,
+      "mean_token_accuracy": 0.7072018980979919,
+      "num_tokens": 268115374.0,
+      "step": 107300
+    },
+    {
+      "epoch": 0.9838635738516549,
+      "learning_rate": 3.2291189144586043e-06,
+      "loss": 0.9685,
+      "mean_token_accuracy": 0.7047235190868377,
+      "num_tokens": 268142870.0,
+      "step": 107310
+    },
+    {
+      "epoch": 0.9839552580911342,
+      "learning_rate": 3.210782066562758e-06,
+      "loss": 0.9689,
+      "mean_token_accuracy": 0.7041001141071319,
+      "num_tokens": 268168195.0,
+      "step": 107320
+    },
+    {
+      "epoch": 0.9840469423306134,
+      "learning_rate": 3.192445218666911e-06,
+      "loss": 0.9562,
+      "mean_token_accuracy": 0.712040513753891,
+      "num_tokens": 268195680.0,
+      "step": 107330
+    },
+    {
+      "epoch": 0.9841386265700927,
+      "learning_rate": 3.1741083707710646e-06,
+      "loss": 0.9688,
+      "mean_token_accuracy": 0.7109183728694916,
+      "num_tokens": 268223300.0,
+      "step": 107340
+    },
+    {
+      "epoch": 0.9842303108095718,
+      "learning_rate": 3.1557715228752177e-06,
+      "loss": 0.9466,
+      "mean_token_accuracy": 0.7137140512466431,
+      "num_tokens": 268251991.0,
+      "step": 107350
+    },
+    {
+      "epoch": 0.984321995049051,
+      "learning_rate": 3.1374346749793713e-06,
+      "loss": 0.9799,
+      "mean_token_accuracy": 0.7029484987258912,
+      "num_tokens": 268279720.0,
+      "step": 107360
+    },
+    {
+      "epoch": 0.9844136792885303,
+      "learning_rate": 3.1190978270835244e-06,
+      "loss": 0.9707,
+      "mean_token_accuracy": 0.7057084023952485,
+      "num_tokens": 268308437.0,
+      "step": 107370
+    },
+    {
+      "epoch": 0.9845053635280095,
+      "learning_rate": 3.100760979187678e-06,
+      "loss": 0.926,
+      "mean_token_accuracy": 0.7118915379047394,
+      "num_tokens": 268337074.0,
+      "step": 107380
+    },
+    {
+      "epoch": 0.9845970477674888,
+      "learning_rate": 3.082424131291831e-06,
+      "loss": 0.981,
+      "mean_token_accuracy": 0.7060979902744293,
+      "num_tokens": 268365336.0,
+      "step": 107390
+    },
+    {
+      "epoch": 0.984688732006968,
+      "learning_rate": 3.0640872833959843e-06,
+      "loss": 0.9706,
+      "mean_token_accuracy": 0.7084679424762725,
+      "num_tokens": 268392832.0,
+      "step": 107400
+    },
+    {
+      "epoch": 0.9847804162464472,
+      "learning_rate": 3.045750435500138e-06,
+      "loss": 0.9551,
+      "mean_token_accuracy": 0.7108550250530243,
+      "num_tokens": 268420757.0,
+      "step": 107410
+    },
+    {
+      "epoch": 0.9848721004859264,
+      "learning_rate": 3.027413587604291e-06,
+      "loss": 0.9593,
+      "mean_token_accuracy": 0.70405193567276,
+      "num_tokens": 268448810.0,
+      "step": 107420
+    },
+    {
+      "epoch": 0.9849637847254057,
+      "learning_rate": 3.0090767397084445e-06,
+      "loss": 0.9585,
+      "mean_token_accuracy": 0.7112325847148895,
+      "num_tokens": 268476505.0,
+      "step": 107430
+    },
+    {
+      "epoch": 0.9850554689648849,
+      "learning_rate": 2.9907398918125972e-06,
+      "loss": 0.9498,
+      "mean_token_accuracy": 0.7112827479839325,
+      "num_tokens": 268503902.0,
+      "step": 107440
+    },
+    {
+      "epoch": 0.9851471532043642,
+      "learning_rate": 2.9724030439167508e-06,
+      "loss": 0.9677,
+      "mean_token_accuracy": 0.7116233706474304,
+      "num_tokens": 268531374.0,
+      "step": 107450
+    },
+    {
+      "epoch": 0.9852388374438434,
+      "learning_rate": 2.9540661960209043e-06,
+      "loss": 0.9695,
+      "mean_token_accuracy": 0.7061333119869232,
+      "num_tokens": 268559105.0,
+      "step": 107460
+    },
+    {
+      "epoch": 0.9853305216833227,
+      "learning_rate": 2.9357293481250575e-06,
+      "loss": 0.959,
+      "mean_token_accuracy": 0.7100592672824859,
+      "num_tokens": 268586478.0,
+      "step": 107470
+    },
+    {
+      "epoch": 0.9854222059228018,
+      "learning_rate": 2.917392500229211e-06,
+      "loss": 0.9365,
+      "mean_token_accuracy": 0.7193555891513824,
+      "num_tokens": 268614547.0,
+      "step": 107480
+    },
+    {
+      "epoch": 0.9855138901622811,
+      "learning_rate": 2.8990556523333638e-06,
+      "loss": 0.9314,
+      "mean_token_accuracy": 0.7129702806472779,
+      "num_tokens": 268643275.0,
+      "step": 107490
+    },
+    {
+      "epoch": 0.9856055744017603,
+      "learning_rate": 2.8807188044375173e-06,
+      "loss": 0.9417,
+      "mean_token_accuracy": 0.7112998008728028,
+      "num_tokens": 268671856.0,
+      "step": 107500
+    },
+    {
+      "epoch": 0.9856972586412396,
+      "learning_rate": 2.8623819565416705e-06,
+      "loss": 0.9339,
+      "mean_token_accuracy": 0.7164718568325043,
+      "num_tokens": 268699381.0,
+      "step": 107510
+    },
+    {
+      "epoch": 0.9857889428807188,
+      "learning_rate": 2.844045108645824e-06,
+      "loss": 0.9438,
+      "mean_token_accuracy": 0.7116261959075928,
+      "num_tokens": 268727217.0,
+      "step": 107520
+    },
+    {
+      "epoch": 0.9858806271201981,
+      "learning_rate": 2.825708260749977e-06,
+      "loss": 0.9288,
+      "mean_token_accuracy": 0.715890783071518,
+      "num_tokens": 268754171.0,
+      "step": 107530
+    },
+    {
+      "epoch": 0.9859723113596772,
+      "learning_rate": 2.8073714128541303e-06,
+      "loss": 0.9608,
+      "mean_token_accuracy": 0.7092635333538055,
+      "num_tokens": 268783298.0,
+      "step": 107540
+    },
+    {
+      "epoch": 0.9860639955991565,
+      "learning_rate": 2.789034564958284e-06,
+      "loss": 0.9457,
+      "mean_token_accuracy": 0.7123358964920044,
+      "num_tokens": 268811658.0,
+      "step": 107550
+    },
+    {
+      "epoch": 0.9861556798386357,
+      "learning_rate": 2.770697717062437e-06,
+      "loss": 0.9387,
+      "mean_token_accuracy": 0.7104895353317261,
+      "num_tokens": 268839747.0,
+      "step": 107560
+    },
+    {
+      "epoch": 0.986247364078115,
+      "learning_rate": 2.7523608691665906e-06,
+      "loss": 0.9708,
+      "mean_token_accuracy": 0.7031299591064453,
+      "num_tokens": 268868105.0,
+      "step": 107570
+    },
+    {
+      "epoch": 0.9863390483175942,
+      "learning_rate": 2.7340240212707437e-06,
+      "loss": 0.9189,
+      "mean_token_accuracy": 0.7171532988548279,
+      "num_tokens": 268896025.0,
+      "step": 107580
+    },
+    {
+      "epoch": 0.9864307325570735,
+      "learning_rate": 2.715687173374897e-06,
+      "loss": 0.9343,
+      "mean_token_accuracy": 0.7195134341716767,
+      "num_tokens": 268923445.0,
+      "step": 107590
+    },
+    {
+      "epoch": 0.9865224167965527,
+      "learning_rate": 2.69735032547905e-06,
+      "loss": 0.9575,
+      "mean_token_accuracy": 0.7066919088363648,
+      "num_tokens": 268950796.0,
+      "step": 107600
+    },
+    {
+      "epoch": 0.9866141010360319,
+      "learning_rate": 2.6790134775832035e-06,
+      "loss": 0.9166,
+      "mean_token_accuracy": 0.7156722962856292,
+      "num_tokens": 268978990.0,
+      "step": 107610
+    },
+    {
+      "epoch": 0.9867057852755111,
+      "learning_rate": 2.660676629687357e-06,
+      "loss": 0.9409,
+      "mean_token_accuracy": 0.7169091105461121,
+      "num_tokens": 269007019.0,
+      "step": 107620
+    },
+    {
+      "epoch": 0.9867974695149904,
+      "learning_rate": 2.6423397817915102e-06,
+      "loss": 0.9657,
+      "mean_token_accuracy": 0.7096025884151459,
+      "num_tokens": 269035085.0,
+      "step": 107630
+    },
+    {
+      "epoch": 0.9868891537544696,
+      "learning_rate": 2.6240029338956634e-06,
+      "loss": 0.955,
+      "mean_token_accuracy": 0.7107065618038177,
+      "num_tokens": 269063429.0,
+      "step": 107640
+    },
+    {
+      "epoch": 0.9869808379939489,
+      "learning_rate": 2.6056660859998165e-06,
+      "loss": 0.9836,
+      "mean_token_accuracy": 0.7031045734882355,
+      "num_tokens": 269088747.0,
+      "step": 107650
+    },
+    {
+      "epoch": 0.9870725222334281,
+      "learning_rate": 2.58732923810397e-06,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.7059447705745697,
+      "num_tokens": 269116551.0,
+      "step": 107660
+    },
+    {
+      "epoch": 0.9871642064729073,
+      "learning_rate": 2.5689923902081232e-06,
+      "loss": 0.9283,
+      "mean_token_accuracy": 0.7176011502742767,
+      "num_tokens": 269144389.0,
+      "step": 107670
+    },
+    {
+      "epoch": 0.9872558907123865,
+      "learning_rate": 2.5506555423122768e-06,
+      "loss": 0.9474,
+      "mean_token_accuracy": 0.7118645429611206,
+      "num_tokens": 269172289.0,
+      "step": 107680
+    },
+    {
+      "epoch": 0.9873475749518658,
+      "learning_rate": 2.53231869441643e-06,
+      "loss": 0.9482,
+      "mean_token_accuracy": 0.7113101065158844,
+      "num_tokens": 269200845.0,
+      "step": 107690
+    },
+    {
+      "epoch": 0.987439259191345,
+      "learning_rate": 2.513981846520583e-06,
+      "loss": 0.9556,
+      "mean_token_accuracy": 0.7083694040775299,
+      "num_tokens": 269229350.0,
+      "step": 107700
+    },
+    {
+      "epoch": 0.9875309434308243,
+      "learning_rate": 2.4956449986247366e-06,
+      "loss": 0.9351,
+      "mean_token_accuracy": 0.714234220981598,
+      "num_tokens": 269257317.0,
+      "step": 107710
+    },
+    {
+      "epoch": 0.9876226276703035,
+      "learning_rate": 2.4773081507288897e-06,
+      "loss": 0.9331,
+      "mean_token_accuracy": 0.7184779822826386,
+      "num_tokens": 269285084.0,
+      "step": 107720
+    },
+    {
+      "epoch": 0.9877143119097828,
+      "learning_rate": 2.4589713028330433e-06,
+      "loss": 0.9617,
+      "mean_token_accuracy": 0.707352077960968,
+      "num_tokens": 269313336.0,
+      "step": 107730
+    },
+    {
+      "epoch": 0.9878059961492619,
+      "learning_rate": 2.4406344549371964e-06,
+      "loss": 0.9859,
+      "mean_token_accuracy": 0.6971431612968445,
+      "num_tokens": 269341414.0,
+      "step": 107740
+    },
+    {
+      "epoch": 0.9878976803887412,
+      "learning_rate": 2.4222976070413496e-06,
+      "loss": 0.9717,
+      "mean_token_accuracy": 0.705696564912796,
+      "num_tokens": 269369551.0,
+      "step": 107750
+    },
+    {
+      "epoch": 0.9879893646282204,
+      "learning_rate": 2.403960759145503e-06,
+      "loss": 0.9548,
+      "mean_token_accuracy": 0.7121150851249695,
+      "num_tokens": 269396822.0,
+      "step": 107760
+    },
+    {
+      "epoch": 0.9880810488676997,
+      "learning_rate": 2.3856239112496563e-06,
+      "loss": 0.9632,
+      "mean_token_accuracy": 0.7125303685665131,
+      "num_tokens": 269424623.0,
+      "step": 107770
+    },
+    {
+      "epoch": 0.9881727331071789,
+      "learning_rate": 2.36728706335381e-06,
+      "loss": 0.9601,
+      "mean_token_accuracy": 0.7132599592208863,
+      "num_tokens": 269452041.0,
+      "step": 107780
+    },
+    {
+      "epoch": 0.9882644173466582,
+      "learning_rate": 2.348950215457963e-06,
+      "loss": 0.9479,
+      "mean_token_accuracy": 0.7090761542320252,
+      "num_tokens": 269477961.0,
+      "step": 107790
+    },
+    {
+      "epoch": 0.9883561015861373,
+      "learning_rate": 2.330613367562116e-06,
+      "loss": 0.9456,
+      "mean_token_accuracy": 0.708621370792389,
+      "num_tokens": 269504979.0,
+      "step": 107800
+    },
+    {
+      "epoch": 0.9884477858256165,
+      "learning_rate": 2.3122765196662693e-06,
+      "loss": 0.9555,
+      "mean_token_accuracy": 0.7117623150348663,
+      "num_tokens": 269533066.0,
+      "step": 107810
+    },
+    {
+      "epoch": 0.9885394700650958,
+      "learning_rate": 2.293939671770423e-06,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.710222738981247,
+      "num_tokens": 269560809.0,
+      "step": 107820
+    },
+    {
+      "epoch": 0.988631154304575,
+      "learning_rate": 2.275602823874576e-06,
+      "loss": 0.952,
+      "mean_token_accuracy": 0.7048126757144928,
+      "num_tokens": 269589536.0,
+      "step": 107830
+    },
+    {
+      "epoch": 0.9887228385440543,
+      "learning_rate": 2.2572659759787295e-06,
+      "loss": 0.9393,
+      "mean_token_accuracy": 0.711496913433075,
+      "num_tokens": 269618091.0,
+      "step": 107840
+    },
+    {
+      "epoch": 0.9888145227835335,
+      "learning_rate": 2.2389291280828827e-06,
+      "loss": 0.9684,
+      "mean_token_accuracy": 0.706806880235672,
+      "num_tokens": 269646262.0,
+      "step": 107850
+    },
+    {
+      "epoch": 0.9889062070230128,
+      "learning_rate": 2.220592280187036e-06,
+      "loss": 0.9522,
+      "mean_token_accuracy": 0.7087924242019653,
+      "num_tokens": 269674407.0,
+      "step": 107860
+    },
+    {
+      "epoch": 0.9889978912624919,
+      "learning_rate": 2.2022554322911894e-06,
+      "loss": 0.9667,
+      "mean_token_accuracy": 0.7084844768047333,
+      "num_tokens": 269703500.0,
+      "step": 107870
+    },
+    {
+      "epoch": 0.9890895755019712,
+      "learning_rate": 2.1839185843953425e-06,
+      "loss": 0.933,
+      "mean_token_accuracy": 0.7124791145324707,
+      "num_tokens": 269730973.0,
+      "step": 107880
+    },
+    {
+      "epoch": 0.9891812597414504,
+      "learning_rate": 2.165581736499496e-06,
+      "loss": 0.9329,
+      "mean_token_accuracy": 0.7177101850509644,
+      "num_tokens": 269759446.0,
+      "step": 107890
+    },
+    {
+      "epoch": 0.9892729439809297,
+      "learning_rate": 2.147244888603649e-06,
+      "loss": 0.948,
+      "mean_token_accuracy": 0.7117880403995513,
+      "num_tokens": 269787020.0,
+      "step": 107900
+    },
+    {
+      "epoch": 0.9893646282204089,
+      "learning_rate": 2.1289080407078023e-06,
+      "loss": 0.9597,
+      "mean_token_accuracy": 0.7092513144016266,
+      "num_tokens": 269814669.0,
+      "step": 107910
+    },
+    {
+      "epoch": 0.9894563124598882,
+      "learning_rate": 2.110571192811956e-06,
+      "loss": 0.9611,
+      "mean_token_accuracy": 0.7070608317852021,
+      "num_tokens": 269842618.0,
+      "step": 107920
+    },
+    {
+      "epoch": 0.9895479966993673,
+      "learning_rate": 2.092234344916109e-06,
+      "loss": 0.9428,
+      "mean_token_accuracy": 0.714831417798996,
+      "num_tokens": 269871870.0,
+      "step": 107930
+    },
+    {
+      "epoch": 0.9896396809388466,
+      "learning_rate": 2.0738974970202626e-06,
+      "loss": 0.9395,
+      "mean_token_accuracy": 0.7137292504310608,
+      "num_tokens": 269899786.0,
+      "step": 107940
+    },
+    {
+      "epoch": 0.9897313651783258,
+      "learning_rate": 2.0555606491244153e-06,
+      "loss": 0.933,
+      "mean_token_accuracy": 0.7155964374542236,
+      "num_tokens": 269927187.0,
+      "step": 107950
+    },
+    {
+      "epoch": 0.9898230494178051,
+      "learning_rate": 2.037223801228569e-06,
+      "loss": 0.9082,
+      "mean_token_accuracy": 0.7195849716663361,
+      "num_tokens": 269955587.0,
+      "step": 107960
+    },
+    {
+      "epoch": 0.9899147336572843,
+      "learning_rate": 2.018886953332722e-06,
+      "loss": 0.9385,
+      "mean_token_accuracy": 0.7136310696601867,
+      "num_tokens": 269982937.0,
+      "step": 107970
+    },
+    {
+      "epoch": 0.9900064178967636,
+      "learning_rate": 2.0005501054368756e-06,
+      "loss": 0.9442,
+      "mean_token_accuracy": 0.7146109282970429,
+      "num_tokens": 270010417.0,
+      "step": 107980
+    },
+    {
+      "epoch": 0.9900981021362428,
+      "learning_rate": 1.9822132575410287e-06,
+      "loss": 0.9608,
+      "mean_token_accuracy": 0.7055068433284759,
+      "num_tokens": 270038197.0,
+      "step": 107990
+    },
+    {
+      "epoch": 0.990189786375722,
+      "learning_rate": 1.963876409645182e-06,
+      "loss": 0.9737,
+      "mean_token_accuracy": 0.7058661997318267,
+      "num_tokens": 270066540.0,
+      "step": 108000
+    },
+    {
+      "epoch": 0.9902814706152012,
+      "learning_rate": 1.9455395617493354e-06,
+      "loss": 0.9617,
+      "mean_token_accuracy": 0.7059551119804383,
+      "num_tokens": 270094791.0,
+      "step": 108010
+    },
+    {
+      "epoch": 0.9903731548546805,
+      "learning_rate": 1.9272027138534885e-06,
+      "loss": 0.9634,
+      "mean_token_accuracy": 0.7042786717414856,
+      "num_tokens": 270123404.0,
+      "step": 108020
+    },
+    {
+      "epoch": 0.9904648390941597,
+      "learning_rate": 1.908865865957642e-06,
+      "loss": 0.9701,
+      "mean_token_accuracy": 0.7064362466335297,
+      "num_tokens": 270151470.0,
+      "step": 108030
+    },
+    {
+      "epoch": 0.990556523333639,
+      "learning_rate": 1.8905290180617955e-06,
+      "loss": 0.9676,
+      "mean_token_accuracy": 0.7073017716407776,
+      "num_tokens": 270179636.0,
+      "step": 108040
+    },
+    {
+      "epoch": 0.9906482075731182,
+      "learning_rate": 1.8721921701659488e-06,
+      "loss": 0.9514,
+      "mean_token_accuracy": 0.7094836533069611,
+      "num_tokens": 270207493.0,
+      "step": 108050
+    },
+    {
+      "epoch": 0.9907398918125974,
+      "learning_rate": 1.8538553222701017e-06,
+      "loss": 0.9347,
+      "mean_token_accuracy": 0.7093855202198028,
+      "num_tokens": 270235131.0,
+      "step": 108060
+    },
+    {
+      "epoch": 0.9908315760520766,
+      "learning_rate": 1.835518474374255e-06,
+      "loss": 0.9652,
+      "mean_token_accuracy": 0.7078740477561951,
+      "num_tokens": 270263598.0,
+      "step": 108070
+    },
+    {
+      "epoch": 0.9909232602915559,
+      "learning_rate": 1.8171816264784084e-06,
+      "loss": 0.9318,
+      "mean_token_accuracy": 0.7160435378551483,
+      "num_tokens": 270291199.0,
+      "step": 108080
+    },
+    {
+      "epoch": 0.9910149445310351,
+      "learning_rate": 1.7988447785825618e-06,
+      "loss": 0.9584,
+      "mean_token_accuracy": 0.7111521482467651,
+      "num_tokens": 270318981.0,
+      "step": 108090
+    },
+    {
+      "epoch": 0.9911066287705144,
+      "learning_rate": 1.7805079306867151e-06,
+      "loss": 0.9377,
+      "mean_token_accuracy": 0.7162557542324066,
+      "num_tokens": 270347137.0,
+      "step": 108100
+    },
+    {
+      "epoch": 0.9911983130099936,
+      "learning_rate": 1.7621710827908683e-06,
+      "loss": 0.9648,
+      "mean_token_accuracy": 0.7065443277359009,
+      "num_tokens": 270374958.0,
+      "step": 108110
+    },
+    {
+      "epoch": 0.9912899972494729,
+      "learning_rate": 1.7438342348950216e-06,
+      "loss": 0.9354,
+      "mean_token_accuracy": 0.7149212837219239,
+      "num_tokens": 270404129.0,
+      "step": 108120
+    },
+    {
+      "epoch": 0.991381681488952,
+      "learning_rate": 1.725497386999175e-06,
+      "loss": 0.9375,
+      "mean_token_accuracy": 0.7164669811725617,
+      "num_tokens": 270432238.0,
+      "step": 108130
+    },
+    {
+      "epoch": 0.9914733657284313,
+      "learning_rate": 1.7071605391033283e-06,
+      "loss": 0.968,
+      "mean_token_accuracy": 0.7033616840839386,
+      "num_tokens": 270461170.0,
+      "step": 108140
+    },
+    {
+      "epoch": 0.9915650499679105,
+      "learning_rate": 1.6888236912074817e-06,
+      "loss": 0.9381,
+      "mean_token_accuracy": 0.7152381062507629,
+      "num_tokens": 270488544.0,
+      "step": 108150
+    },
+    {
+      "epoch": 0.9916567342073898,
+      "learning_rate": 1.6704868433116346e-06,
+      "loss": 0.9797,
+      "mean_token_accuracy": 0.7032285392284393,
+      "num_tokens": 270515714.0,
+      "step": 108160
+    },
+    {
+      "epoch": 0.991748418446869,
+      "learning_rate": 1.652149995415788e-06,
+      "loss": 0.9434,
+      "mean_token_accuracy": 0.71032155752182,
+      "num_tokens": 270543296.0,
+      "step": 108170
+    },
+    {
+      "epoch": 0.9918401026863483,
+      "learning_rate": 1.6338131475199415e-06,
+      "loss": 0.9267,
+      "mean_token_accuracy": 0.7193844795227051,
+      "num_tokens": 270571215.0,
+      "step": 108180
+    },
+    {
+      "epoch": 0.9919317869258274,
+      "learning_rate": 1.6154762996240949e-06,
+      "loss": 0.9513,
+      "mean_token_accuracy": 0.7087843775749206,
+      "num_tokens": 270599346.0,
+      "step": 108190
+    },
+    {
+      "epoch": 0.9920234711653066,
+      "learning_rate": 1.5971394517282482e-06,
+      "loss": 0.9619,
+      "mean_token_accuracy": 0.7091204702854157,
+      "num_tokens": 270627760.0,
+      "step": 108200
+    },
+    {
+      "epoch": 0.9921151554047859,
+      "learning_rate": 1.5788026038324011e-06,
+      "loss": 0.9077,
+      "mean_token_accuracy": 0.7241861045360565,
+      "num_tokens": 270655493.0,
+      "step": 108210
+    },
+    {
+      "epoch": 0.9922068396442651,
+      "learning_rate": 1.5604657559365545e-06,
+      "loss": 0.9518,
+      "mean_token_accuracy": 0.7072457790374755,
+      "num_tokens": 270683485.0,
+      "step": 108220
+    },
+    {
+      "epoch": 0.9922985238837444,
+      "learning_rate": 1.5421289080407078e-06,
+      "loss": 0.9406,
+      "mean_token_accuracy": 0.7132770359516144,
+      "num_tokens": 270711649.0,
+      "step": 108230
+    },
+    {
+      "epoch": 0.9923902081232236,
+      "learning_rate": 1.5237920601448612e-06,
+      "loss": 0.9824,
+      "mean_token_accuracy": 0.7037883758544922,
+      "num_tokens": 270739935.0,
+      "step": 108240
+    },
+    {
+      "epoch": 0.9924818923627029,
+      "learning_rate": 1.5054552122490145e-06,
+      "loss": 0.9676,
+      "mean_token_accuracy": 0.7086636006832123,
+      "num_tokens": 270768643.0,
+      "step": 108250
+    },
+    {
+      "epoch": 0.992573576602182,
+      "learning_rate": 1.4871183643531679e-06,
+      "loss": 0.9596,
+      "mean_token_accuracy": 0.7082400918006897,
+      "num_tokens": 270796222.0,
+      "step": 108260
+    },
+    {
+      "epoch": 0.9926652608416613,
+      "learning_rate": 1.468781516457321e-06,
+      "loss": 0.9236,
+      "mean_token_accuracy": 0.7113487184047699,
+      "num_tokens": 270824348.0,
+      "step": 108270
+    },
+    {
+      "epoch": 0.9927569450811405,
+      "learning_rate": 1.4504446685614744e-06,
+      "loss": 0.9335,
+      "mean_token_accuracy": 0.7143835961818695,
+      "num_tokens": 270851632.0,
+      "step": 108280
+    },
+    {
+      "epoch": 0.9928486293206198,
+      "learning_rate": 1.4321078206656277e-06,
+      "loss": 0.9691,
+      "mean_token_accuracy": 0.7031129777431488,
+      "num_tokens": 270879073.0,
+      "step": 108290
+    },
+    {
+      "epoch": 0.992940313560099,
+      "learning_rate": 1.4137709727697809e-06,
+      "loss": 0.9621,
+      "mean_token_accuracy": 0.70713369846344,
+      "num_tokens": 270907964.0,
+      "step": 108300
+    },
+    {
+      "epoch": 0.9930319977995783,
+      "learning_rate": 1.3954341248739342e-06,
+      "loss": 0.9544,
+      "mean_token_accuracy": 0.7095125436782836,
+      "num_tokens": 270936402.0,
+      "step": 108310
+    },
+    {
+      "epoch": 0.9931236820390574,
+      "learning_rate": 1.3770972769780873e-06,
+      "loss": 0.9407,
+      "mean_token_accuracy": 0.7117841303348541,
+      "num_tokens": 270964626.0,
+      "step": 108320
+    },
+    {
+      "epoch": 0.9932153662785367,
+      "learning_rate": 1.358760429082241e-06,
+      "loss": 0.9675,
+      "mean_token_accuracy": 0.7025663137435914,
+      "num_tokens": 270993097.0,
+      "step": 108330
+    },
+    {
+      "epoch": 0.9933070505180159,
+      "learning_rate": 1.3404235811863943e-06,
+      "loss": 0.9263,
+      "mean_token_accuracy": 0.7160776257514954,
+      "num_tokens": 271021091.0,
+      "step": 108340
+    },
+    {
+      "epoch": 0.9933987347574952,
+      "learning_rate": 1.3220867332905474e-06,
+      "loss": 0.9456,
+      "mean_token_accuracy": 0.7149143099784852,
+      "num_tokens": 271048572.0,
+      "step": 108350
+    },
+    {
+      "epoch": 0.9934904189969744,
+      "learning_rate": 1.3037498853947007e-06,
+      "loss": 0.9721,
+      "mean_token_accuracy": 0.7054606318473816,
+      "num_tokens": 271075801.0,
+      "step": 108360
+    },
+    {
+      "epoch": 0.9935821032364537,
+      "learning_rate": 1.285413037498854e-06,
+      "loss": 0.9746,
+      "mean_token_accuracy": 0.707843953371048,
+      "num_tokens": 271103833.0,
+      "step": 108370
+    },
+    {
+      "epoch": 0.9936737874759329,
+      "learning_rate": 1.2670761896030072e-06,
+      "loss": 0.9489,
+      "mean_token_accuracy": 0.7131265759468078,
+      "num_tokens": 271130967.0,
+      "step": 108380
+    },
+    {
+      "epoch": 0.9937654717154121,
+      "learning_rate": 1.2487393417071606e-06,
+      "loss": 0.974,
+      "mean_token_accuracy": 0.7050150334835052,
+      "num_tokens": 271158191.0,
+      "step": 108390
+    },
+    {
+      "epoch": 0.9938571559548913,
+      "learning_rate": 1.230402493811314e-06,
+      "loss": 0.9432,
+      "mean_token_accuracy": 0.710757851600647,
+      "num_tokens": 271185860.0,
+      "step": 108400
+    },
+    {
+      "epoch": 0.9939488401943706,
+      "learning_rate": 1.2120656459154673e-06,
+      "loss": 0.9674,
+      "mean_token_accuracy": 0.7087086319923401,
+      "num_tokens": 271214253.0,
+      "step": 108410
+    },
+    {
+      "epoch": 0.9940405244338498,
+      "learning_rate": 1.1937287980196206e-06,
+      "loss": 0.9621,
+      "mean_token_accuracy": 0.7093576073646546,
+      "num_tokens": 271242404.0,
+      "step": 108420
+    },
+    {
+      "epoch": 0.9941322086733291,
+      "learning_rate": 1.1753919501237738e-06,
+      "loss": 0.9294,
+      "mean_token_accuracy": 0.7140588998794556,
+      "num_tokens": 271269773.0,
+      "step": 108430
+    },
+    {
+      "epoch": 0.9942238929128083,
+      "learning_rate": 1.1570551022279271e-06,
+      "loss": 1.0008,
+      "mean_token_accuracy": 0.7005892097949982,
+      "num_tokens": 271298028.0,
+      "step": 108440
+    },
+    {
+      "epoch": 0.9943155771522875,
+      "learning_rate": 1.1387182543320803e-06,
+      "loss": 0.92,
+      "mean_token_accuracy": 0.7150518417358398,
+      "num_tokens": 271325650.0,
+      "step": 108450
+    },
+    {
+      "epoch": 0.9944072613917667,
+      "learning_rate": 1.1203814064362336e-06,
+      "loss": 0.9756,
+      "mean_token_accuracy": 0.7039842367172241,
+      "num_tokens": 271354299.0,
+      "step": 108460
+    },
+    {
+      "epoch": 0.994498945631246,
+      "learning_rate": 1.102044558540387e-06,
+      "loss": 0.9615,
+      "mean_token_accuracy": 0.706362110376358,
+      "num_tokens": 271382096.0,
+      "step": 108470
+    },
+    {
+      "epoch": 0.9945906298707252,
+      "learning_rate": 1.0837077106445403e-06,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.7117473840713501,
+      "num_tokens": 271410248.0,
+      "step": 108480
+    },
+    {
+      "epoch": 0.9946823141102045,
+      "learning_rate": 1.0653708627486937e-06,
+      "loss": 0.9504,
+      "mean_token_accuracy": 0.7121224582195282,
+      "num_tokens": 271438094.0,
+      "step": 108490
+    },
+    {
+      "epoch": 0.9947739983496837,
+      "learning_rate": 1.0470340148528468e-06,
+      "loss": 0.9419,
+      "mean_token_accuracy": 0.7153161048889161,
+      "num_tokens": 271466215.0,
+      "step": 108500
+    },
+    {
+      "epoch": 0.994865682589163,
+      "learning_rate": 1.0286971669570001e-06,
+      "loss": 0.9281,
+      "mean_token_accuracy": 0.7177415311336517,
+      "num_tokens": 271494690.0,
+      "step": 108510
+    },
+    {
+      "epoch": 0.9949573668286421,
+      "learning_rate": 1.0103603190611535e-06,
+      "loss": 0.9482,
+      "mean_token_accuracy": 0.7120618402957917,
+      "num_tokens": 271523304.0,
+      "step": 108520
+    },
+    {
+      "epoch": 0.9950490510681214,
+      "learning_rate": 9.920234711653066e-07,
+      "loss": 0.9501,
+      "mean_token_accuracy": 0.7138809025287628,
+      "num_tokens": 271552187.0,
+      "step": 108530
+    },
+    {
+      "epoch": 0.9951407353076006,
+      "learning_rate": 9.7368662326946e-07,
+      "loss": 0.9444,
+      "mean_token_accuracy": 0.7139100313186646,
+      "num_tokens": 271580179.0,
+      "step": 108540
+    },
+    {
+      "epoch": 0.9952324195470799,
+      "learning_rate": 9.553497753736133e-07,
+      "loss": 0.9949,
+      "mean_token_accuracy": 0.6991130471229553,
+      "num_tokens": 271605081.0,
+      "step": 108550
+    },
+    {
+      "epoch": 0.9953241037865591,
+      "learning_rate": 9.370129274777666e-07,
+      "loss": 0.9518,
+      "mean_token_accuracy": 0.7130214869976044,
+      "num_tokens": 271633152.0,
+      "step": 108560
+    },
+    {
+      "epoch": 0.9954157880260384,
+      "learning_rate": 9.186760795819199e-07,
+      "loss": 0.9566,
+      "mean_token_accuracy": 0.7135081589221954,
+      "num_tokens": 271660525.0,
+      "step": 108570
+    },
+    {
+      "epoch": 0.9955074722655176,
+      "learning_rate": 9.003392316860732e-07,
+      "loss": 0.9443,
+      "mean_token_accuracy": 0.7129908621311187,
+      "num_tokens": 271688678.0,
+      "step": 108580
+    },
+    {
+      "epoch": 0.9955991565049968,
+      "learning_rate": 8.820023837902265e-07,
+      "loss": 0.9484,
+      "mean_token_accuracy": 0.711221432685852,
+      "num_tokens": 271716407.0,
+      "step": 108590
+    },
+    {
+      "epoch": 0.995690840744476,
+      "learning_rate": 8.636655358943799e-07,
+      "loss": 0.9509,
+      "mean_token_accuracy": 0.7100203394889831,
+      "num_tokens": 271743725.0,
+      "step": 108600
+    },
+    {
+      "epoch": 0.9957825249839553,
+      "learning_rate": 8.45328687998533e-07,
+      "loss": 0.9633,
+      "mean_token_accuracy": 0.706583422422409,
+      "num_tokens": 271772720.0,
+      "step": 108610
+    },
+    {
+      "epoch": 0.9958742092234345,
+      "learning_rate": 8.269918401026865e-07,
+      "loss": 0.9531,
+      "mean_token_accuracy": 0.713868010044098,
+      "num_tokens": 271800659.0,
+      "step": 108620
+    },
+    {
+      "epoch": 0.9959658934629138,
+      "learning_rate": 8.086549922068396e-07,
+      "loss": 0.9472,
+      "mean_token_accuracy": 0.7135761678218842,
+      "num_tokens": 271829073.0,
+      "step": 108630
+    },
+    {
+      "epoch": 0.996057577702393,
+      "learning_rate": 7.90318144310993e-07,
+      "loss": 0.9296,
+      "mean_token_accuracy": 0.7176508367061615,
+      "num_tokens": 271857684.0,
+      "step": 108640
+    },
+    {
+      "epoch": 0.9961492619418721,
+      "learning_rate": 7.719812964151463e-07,
+      "loss": 0.9427,
+      "mean_token_accuracy": 0.7142350792884826,
+      "num_tokens": 271885236.0,
+      "step": 108650
+    },
+    {
+      "epoch": 0.9962409461813514,
+      "learning_rate": 7.536444485192995e-07,
+      "loss": 0.9592,
+      "mean_token_accuracy": 0.7084473371505737,
+      "num_tokens": 271913274.0,
+      "step": 108660
+    },
+    {
+      "epoch": 0.9963326304208306,
+      "learning_rate": 7.353076006234529e-07,
+      "loss": 0.9529,
+      "mean_token_accuracy": 0.7091939210891723,
+      "num_tokens": 271940681.0,
+      "step": 108670
+    },
+    {
+      "epoch": 0.9964243146603099,
+      "learning_rate": 7.169707527276061e-07,
+      "loss": 0.9558,
+      "mean_token_accuracy": 0.7062992513179779,
+      "num_tokens": 271968803.0,
+      "step": 108680
+    },
+    {
+      "epoch": 0.9965159988997891,
+      "learning_rate": 6.986339048317595e-07,
+      "loss": 0.9429,
+      "mean_token_accuracy": 0.7165329098701477,
+      "num_tokens": 271996730.0,
+      "step": 108690
+    },
+    {
+      "epoch": 0.9966076831392684,
+      "learning_rate": 6.802970569359127e-07,
+      "loss": 0.9452,
+      "mean_token_accuracy": 0.7138762652873993,
+      "num_tokens": 272024490.0,
+      "step": 108700
+    },
+    {
+      "epoch": 0.9966993673787476,
+      "learning_rate": 6.619602090400661e-07,
+      "loss": 0.9569,
+      "mean_token_accuracy": 0.7109337151050568,
+      "num_tokens": 272050972.0,
+      "step": 108710
+    },
+    {
+      "epoch": 0.9967910516182268,
+      "learning_rate": 6.436233611442193e-07,
+      "loss": 0.9528,
+      "mean_token_accuracy": 0.7097187757492065,
+      "num_tokens": 272078689.0,
+      "step": 108720
+    },
+    {
+      "epoch": 0.996882735857706,
+      "learning_rate": 6.252865132483727e-07,
+      "loss": 0.9488,
+      "mean_token_accuracy": 0.710624486207962,
+      "num_tokens": 272106417.0,
+      "step": 108730
+    },
+    {
+      "epoch": 0.9969744200971853,
+      "learning_rate": 6.069496653525259e-07,
+      "loss": 0.9253,
+      "mean_token_accuracy": 0.7121653497219086,
+      "num_tokens": 272134030.0,
+      "step": 108740
+    },
+    {
+      "epoch": 0.9970661043366645,
+      "learning_rate": 5.886128174566792e-07,
+      "loss": 0.9305,
+      "mean_token_accuracy": 0.7163199841976166,
+      "num_tokens": 272161855.0,
+      "step": 108750
+    },
+    {
+      "epoch": 0.9971577885761438,
+      "learning_rate": 5.702759695608325e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.7092844486236572,
+      "num_tokens": 272189414.0,
+      "step": 108760
+    },
+    {
+      "epoch": 0.997249472815623,
+      "learning_rate": 5.519391216649859e-07,
+      "loss": 0.9543,
+      "mean_token_accuracy": 0.7108266532421113,
+      "num_tokens": 272217548.0,
+      "step": 108770
+    },
+    {
+      "epoch": 0.9973411570551022,
+      "learning_rate": 5.336022737691391e-07,
+      "loss": 0.9519,
+      "mean_token_accuracy": 0.71286461353302,
+      "num_tokens": 272245748.0,
+      "step": 108780
+    },
+    {
+      "epoch": 0.9974328412945814,
+      "learning_rate": 5.152654258732924e-07,
+      "loss": 0.9713,
+      "mean_token_accuracy": 0.7049710333347321,
+      "num_tokens": 272273014.0,
+      "step": 108790
+    },
+    {
+      "epoch": 0.9975245255340607,
+      "learning_rate": 4.969285779774457e-07,
+      "loss": 0.9808,
+      "mean_token_accuracy": 0.7022957563400268,
+      "num_tokens": 272301065.0,
+      "step": 108800
+    },
+    {
+      "epoch": 0.9976162097735399,
+      "learning_rate": 4.785917300815991e-07,
+      "loss": 0.9772,
+      "mean_token_accuracy": 0.7063236474990845,
+      "num_tokens": 272328845.0,
+      "step": 108810
+    },
+    {
+      "epoch": 0.9977078940130192,
+      "learning_rate": 4.602548821857523e-07,
+      "loss": 0.946,
+      "mean_token_accuracy": 0.7124539732933044,
+      "num_tokens": 272356524.0,
+      "step": 108820
+    },
+    {
+      "epoch": 0.9977995782524984,
+      "learning_rate": 4.419180342899056e-07,
+      "loss": 0.9525,
+      "mean_token_accuracy": 0.7134114861488342,
+      "num_tokens": 272384910.0,
+      "step": 108830
+    },
+    {
+      "epoch": 0.9978912624919777,
+      "learning_rate": 4.2358118639405884e-07,
+      "loss": 0.9209,
+      "mean_token_accuracy": 0.7175331294536591,
+      "num_tokens": 272413154.0,
+      "step": 108840
+    },
+    {
+      "epoch": 0.9979829467314568,
+      "learning_rate": 4.052443384982122e-07,
+      "loss": 0.963,
+      "mean_token_accuracy": 0.7068632543087006,
+      "num_tokens": 272441661.0,
+      "step": 108850
+    },
+    {
+      "epoch": 0.9980746309709361,
+      "learning_rate": 3.869074906023655e-07,
+      "loss": 0.9593,
+      "mean_token_accuracy": 0.7033331573009491,
+      "num_tokens": 272470052.0,
+      "step": 108860
+    },
+    {
+      "epoch": 0.9981663152104153,
+      "learning_rate": 3.685706427065188e-07,
+      "loss": 0.969,
+      "mean_token_accuracy": 0.7020613729953766,
+      "num_tokens": 272498173.0,
+      "step": 108870
+    },
+    {
+      "epoch": 0.9982579994498946,
+      "learning_rate": 3.502337948106721e-07,
+      "loss": 0.9765,
+      "mean_token_accuracy": 0.7058265626430511,
+      "num_tokens": 272526028.0,
+      "step": 108880
+    },
+    {
+      "epoch": 0.9983496836893738,
+      "learning_rate": 3.318969469148254e-07,
+      "loss": 0.927,
+      "mean_token_accuracy": 0.7112305581569671,
+      "num_tokens": 272554507.0,
+      "step": 108890
+    },
+    {
+      "epoch": 0.9984413679288531,
+      "learning_rate": 3.135600990189786e-07,
+      "loss": 0.9531,
+      "mean_token_accuracy": 0.7085698962211608,
+      "num_tokens": 272583042.0,
+      "step": 108900
+    },
+    {
+      "epoch": 0.9985330521683322,
+      "learning_rate": 2.9522325112313197e-07,
+      "loss": 0.951,
+      "mean_token_accuracy": 0.7063367486000061,
+      "num_tokens": 272610733.0,
+      "step": 108910
+    },
+    {
+      "epoch": 0.9986247364078115,
+      "learning_rate": 2.768864032272852e-07,
+      "loss": 0.9817,
+      "mean_token_accuracy": 0.705414742231369,
+      "num_tokens": 272638792.0,
+      "step": 108920
+    },
+    {
+      "epoch": 0.9987164206472907,
+      "learning_rate": 2.5854955533143856e-07,
+      "loss": 0.9997,
+      "mean_token_accuracy": 0.7029419541358948,
+      "num_tokens": 272666918.0,
+      "step": 108930
+    },
+    {
+      "epoch": 0.99880810488677,
+      "learning_rate": 2.402127074355918e-07,
+      "loss": 0.9656,
+      "mean_token_accuracy": 0.7078922510147094,
+      "num_tokens": 272694387.0,
+      "step": 108940
+    },
+    {
+      "epoch": 0.9988997891262492,
+      "learning_rate": 2.2187585953974516e-07,
+      "loss": 0.9329,
+      "mean_token_accuracy": 0.7136320233345032,
+      "num_tokens": 272720667.0,
+      "step": 108950
+    },
+    {
+      "epoch": 0.9989914733657285,
+      "learning_rate": 2.0353901164389843e-07,
+      "loss": 0.9608,
+      "mean_token_accuracy": 0.7071583569049835,
+      "num_tokens": 272748319.0,
+      "step": 108960
+    },
+    {
+      "epoch": 0.9990831576052077,
+      "learning_rate": 1.8520216374805172e-07,
+      "loss": 0.966,
+      "mean_token_accuracy": 0.7109744012355804,
+      "num_tokens": 272776982.0,
+      "step": 108970
+    },
+    {
+      "epoch": 0.9991748418446869,
+      "learning_rate": 1.6686531585220502e-07,
+      "loss": 0.9342,
+      "mean_token_accuracy": 0.7147205471992493,
+      "num_tokens": 272804956.0,
+      "step": 108980
+    },
+    {
+      "epoch": 0.9992665260841661,
+      "learning_rate": 1.4852846795635832e-07,
+      "loss": 0.9635,
+      "mean_token_accuracy": 0.7084476709365845,
+      "num_tokens": 272833433.0,
+      "step": 108990
+    },
+    {
+      "epoch": 0.9993582103236454,
+      "learning_rate": 1.3019162006051161e-07,
+      "loss": 0.9791,
+      "mean_token_accuracy": 0.7073094964027404,
+      "num_tokens": 272861260.0,
+      "step": 109000
+    },
+    {
+      "epoch": 0.9994498945631246,
+      "learning_rate": 1.1185477216466491e-07,
+      "loss": 0.9614,
+      "mean_token_accuracy": 0.7112366139888764,
+      "num_tokens": 272889446.0,
+      "step": 109010
+    },
+    {
+      "epoch": 0.9995415788026039,
+      "learning_rate": 9.35179242688182e-08,
+      "loss": 0.9253,
+      "mean_token_accuracy": 0.7157534658908844,
+      "num_tokens": 272917134.0,
+      "step": 109020
+    },
+    {
+      "epoch": 0.9996332630420831,
+      "learning_rate": 7.518107637297149e-08,
+      "loss": 0.9747,
+      "mean_token_accuracy": 0.7045438110828399,
+      "num_tokens": 272945009.0,
+      "step": 109030
+    },
+    {
+      "epoch": 0.9997249472815622,
+      "learning_rate": 5.684422847712479e-08,
+      "loss": 0.9436,
+      "mean_token_accuracy": 0.7104832291603088,
+      "num_tokens": 272973145.0,
+      "step": 109040
+    },
+    {
+      "epoch": 0.9998166315210415,
+      "learning_rate": 3.8507380581278085e-08,
+      "loss": 0.9714,
+      "mean_token_accuracy": 0.7046187996864319,
+      "num_tokens": 273000995.0,
+      "step": 109050
+    },
+    {
+      "epoch": 1.0000366736957917,
+      "learning_rate": 2.0170532685431375e-08,
+      "loss": 0.9708,
+      "mean_token_accuracy": 0.705879271030426,
+      "num_tokens": 273021976.0,
+      "step": 109060
+    },
+    {
+      "epoch": 1.0001283579352709,
+      "learning_rate": 1.8336847895846704e-09,
+      "loss": 0.9834,
+      "mean_token_accuracy": 0.7151719331741333,
+      "num_tokens": 273045815.0,
+      "step": 109070
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 109070,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 6816,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.7755502837531136e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}