diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -3,772 +3,6008 @@
   "best_model_checkpoint": null,
   "epoch": 1.0,
   "eval_steps": 500,
-  "global_step": 536,
+  "global_step": 4275,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.009328358208955223,
-      "grad_norm": 1.454424294525684,
-      "learning_rate": 9.259259259259259e-06,
-      "loss": 0.8491,
+      "epoch": 0.0011695906432748538,
+      "grad_norm": 4.3862214979637395,
+      "learning_rate": 1.1682242990654206e-06,
+      "loss": 1.4545,
       "step": 5
     },
     {
-      "epoch": 0.018656716417910446,
-      "grad_norm": 1.1651263438043256,
-      "learning_rate": 1.8518518518518518e-05,
-      "loss": 0.8006,
+      "epoch": 0.0023391812865497076,
+      "grad_norm": 2.898308786930061,
+      "learning_rate": 2.3364485981308413e-06,
+      "loss": 1.4508,
       "step": 10
     },
     {
-      "epoch": 0.027985074626865673,
-      "grad_norm": 0.7398937799068904,
-      "learning_rate": 2.777777777777778e-05,
-      "loss": 0.7315,
+      "epoch": 0.0035087719298245615,
+      "grad_norm": 3.3436616124637424,
+      "learning_rate": 3.5046728971962617e-06,
+      "loss": 1.3998,
       "step": 15
     },
     {
-      "epoch": 0.03731343283582089,
-      "grad_norm": 0.4574086226332697,
-      "learning_rate": 3.7037037037037037e-05,
-      "loss": 0.6803,
+      "epoch": 0.004678362573099415,
+      "grad_norm": 2.34446456390598,
+      "learning_rate": 4.6728971962616825e-06,
+      "loss": 1.3392,
       "step": 20
     },
     {
-      "epoch": 0.04664179104477612,
-      "grad_norm": 0.44822114984510997,
-      "learning_rate": 4.62962962962963e-05,
-      "loss": 0.6578,
+      "epoch": 0.005847953216374269,
+      "grad_norm": 1.4686088375623854,
+      "learning_rate": 5.841121495327103e-06,
+      "loss": 1.2714,
       "step": 25
     },
     {
-      "epoch": 0.055970149253731345,
-      "grad_norm": 0.37511417859069063,
-      "learning_rate": 4.999614302517356e-05,
-      "loss": 0.6581,
+      "epoch": 0.007017543859649123,
+      "grad_norm": 1.320686406588424,
+      "learning_rate": 7.009345794392523e-06,
+      "loss": 1.2414,
       "step": 30
     },
     {
-      "epoch": 0.06529850746268656,
-      "grad_norm": 0.3849219183366683,
-      "learning_rate": 4.997257741198456e-05,
-      "loss": 0.6419,
+      "epoch": 0.008187134502923977,
+      "grad_norm": 1.2402467459230129,
+      "learning_rate": 8.177570093457943e-06,
+      "loss": 1.1882,
       "step": 35
     },
     {
-      "epoch": 0.07462686567164178,
-      "grad_norm": 0.30443802357759675,
-      "learning_rate": 4.992761136351291e-05,
-      "loss": 0.6271,
+      "epoch": 0.00935672514619883,
+      "grad_norm": 1.2376435378586828,
+      "learning_rate": 9.345794392523365e-06,
+      "loss": 1.1749,
       "step": 40
     },
     {
-      "epoch": 0.08395522388059702,
-      "grad_norm": 0.3217832795991014,
-      "learning_rate": 4.986128770052603e-05,
-      "loss": 0.6084,
+      "epoch": 0.010526315789473684,
+      "grad_norm": 1.1524990455288477,
+      "learning_rate": 1.0514018691588785e-05,
+      "loss": 1.1588,
       "step": 45
     },
     {
-      "epoch": 0.09328358208955224,
-      "grad_norm": 0.28191555588141387,
-      "learning_rate": 4.9773669582457364e-05,
-      "loss": 0.61,
+      "epoch": 0.011695906432748537,
+      "grad_norm": 1.301145432712544,
+      "learning_rate": 1.1682242990654207e-05,
+      "loss": 1.1424,
       "step": 50
     },
     {
-      "epoch": 0.10261194029850747,
-      "grad_norm": 0.3426471273008592,
-      "learning_rate": 4.966484044726024e-05,
-      "loss": 0.6137,
+      "epoch": 0.012865497076023392,
+      "grad_norm": 1.1323491606855578,
+      "learning_rate": 1.2850467289719625e-05,
+      "loss": 1.1172,
       "step": 55
     },
     {
-      "epoch": 0.11194029850746269,
-      "grad_norm": 0.3307025092141417,
-      "learning_rate": 4.953490393195063e-05,
-      "loss": 0.6004,
+      "epoch": 0.014035087719298246,
+      "grad_norm": 1.201565879775835,
+      "learning_rate": 1.4018691588785047e-05,
+      "loss": 1.1029,
       "step": 60
     },
     {
-      "epoch": 0.12126865671641791,
-      "grad_norm": 0.39354087577568503,
-      "learning_rate": 4.938398377391461e-05,
-      "loss": 0.5972,
+      "epoch": 0.0152046783625731,
+      "grad_norm": 1.1433240312787192,
+      "learning_rate": 1.5186915887850467e-05,
+      "loss": 1.086,
       "step": 65
     },
     {
-      "epoch": 0.13059701492537312,
-      "grad_norm": 0.4926227619670805,
-      "learning_rate": 4.921222369307427e-05,
-      "loss": 0.5939,
+      "epoch": 0.016374269005847954,
+      "grad_norm": 1.1855044769686904,
+      "learning_rate": 1.6355140186915887e-05,
+      "loss": 1.1019,
       "step": 70
     },
     {
-      "epoch": 0.13992537313432835,
-      "grad_norm": 0.44012267206553335,
-      "learning_rate": 4.901978725502454e-05,
-      "loss": 0.5895,
+      "epoch": 0.017543859649122806,
+      "grad_norm": 1.2419830025546306,
+      "learning_rate": 1.752336448598131e-05,
+      "loss": 1.0782,
       "step": 75
     },
     {
-      "epoch": 0.14925373134328357,
-      "grad_norm": 0.39694954292298656,
-      "learning_rate": 4.880685771527114e-05,
-      "loss": 0.5895,
+      "epoch": 0.01871345029239766,
+      "grad_norm": 1.1871478349287299,
+      "learning_rate": 1.869158878504673e-05,
+      "loss": 1.0558,
       "step": 80
     },
     {
-      "epoch": 0.15858208955223882,
-      "grad_norm": 0.4202529047479601,
-      "learning_rate": 4.8573637844718e-05,
-      "loss": 0.5945,
+      "epoch": 0.019883040935672516,
+      "grad_norm": 1.4158109594901271,
+      "learning_rate": 1.985981308411215e-05,
+      "loss": 1.0733,
       "step": 85
     },
     {
-      "epoch": 0.16791044776119404,
-      "grad_norm": 0.37246529454641986,
-      "learning_rate": 4.83203497365703e-05,
-      "loss": 0.591,
+      "epoch": 0.021052631578947368,
+      "grad_norm": 1.3659345116154782,
+      "learning_rate": 2.102803738317757e-05,
+      "loss": 1.0608,
       "step": 90
     },
     {
-      "epoch": 0.17723880597014927,
-      "grad_norm": 0.47340274820866607,
-      "learning_rate": 4.8047234594837143e-05,
-      "loss": 0.5783,
+      "epoch": 0.022222222222222223,
+      "grad_norm": 1.1708966657531445,
+      "learning_rate": 2.2196261682242992e-05,
+      "loss": 1.0539,
       "step": 95
     },
     {
-      "epoch": 0.1865671641791045,
-      "grad_norm": 0.4033874126610035,
-      "learning_rate": 4.775455250463507e-05,
-      "loss": 0.5749,
+      "epoch": 0.023391812865497075,
+      "grad_norm": 1.4057049542204068,
+      "learning_rate": 2.3364485981308414e-05,
+      "loss": 1.0491,
       "step": 100
     },
     {
-      "epoch": 0.1958955223880597,
-      "grad_norm": 0.4347921460769215,
-      "learning_rate": 4.744258218451135e-05,
-      "loss": 0.5846,
+      "epoch": 0.02456140350877193,
+      "grad_norm": 1.177573164358867,
+      "learning_rate": 2.4532710280373832e-05,
+      "loss": 1.0382,
       "step": 105
     },
     {
-      "epoch": 0.20522388059701493,
-      "grad_norm": 0.3068655130259012,
-      "learning_rate": 4.71116207210228e-05,
-      "loss": 0.5758,
+      "epoch": 0.025730994152046785,
+      "grad_norm": 1.2987611554079812,
+      "learning_rate": 2.570093457943925e-05,
+      "loss": 1.0269,
       "step": 110
     },
     {
-      "epoch": 0.21455223880597016,
-      "grad_norm": 0.35709478563960256,
-      "learning_rate": 4.676198328582288e-05,
-      "loss": 0.5712,
+      "epoch": 0.026900584795321637,
+      "grad_norm": 1.2805282242501288,
+      "learning_rate": 2.6869158878504675e-05,
+      "loss": 1.0303,
       "step": 115
     },
     {
-      "epoch": 0.22388059701492538,
-      "grad_norm": 0.3273285594289526,
-      "learning_rate": 4.6394002835526535e-05,
-      "loss": 0.582,
+      "epoch": 0.028070175438596492,
+      "grad_norm": 1.2521388436600536,
+      "learning_rate": 2.8037383177570094e-05,
+      "loss": 1.0186,
       "step": 120
     },
     {
-      "epoch": 0.2332089552238806,
-      "grad_norm": 0.3696948340778704,
-      "learning_rate": 4.6008029794638596e-05,
-      "loss": 0.5666,
+      "epoch": 0.029239766081871343,
+      "grad_norm": 1.2503867172495808,
+      "learning_rate": 2.9205607476635515e-05,
+      "loss": 1.0338,
       "step": 125
     },
     {
-      "epoch": 0.24253731343283583,
-      "grad_norm": 0.3484593044458527,
-      "learning_rate": 4.560443172184763e-05,
-      "loss": 0.5692,
+      "epoch": 0.0304093567251462,
+      "grad_norm": 1.2707180378214313,
+      "learning_rate": 3.0373831775700934e-05,
+      "loss": 1.0244,
       "step": 130
     },
     {
-      "epoch": 0.251865671641791,
-      "grad_norm": 0.33156997493993207,
-      "learning_rate": 4.5183592960003104e-05,
-      "loss": 0.5619,
+      "epoch": 0.031578947368421054,
+      "grad_norm": 1.3243403485631375,
+      "learning_rate": 3.1542056074766355e-05,
+      "loss": 0.9971,
       "step": 135
     },
     {
-      "epoch": 0.26119402985074625,
-      "grad_norm": 0.261674612188198,
-      "learning_rate": 4.4745914270109055e-05,
-      "loss": 0.5671,
+      "epoch": 0.03274853801169591,
+      "grad_norm": 1.3071755226415636,
+      "learning_rate": 3.2710280373831774e-05,
+      "loss": 1.0018,
       "step": 140
     },
     {
-      "epoch": 0.27052238805970147,
-      "grad_norm": 0.3635079447291604,
-      "learning_rate": 4.429181244968301e-05,
-      "loss": 0.5663,
+      "epoch": 0.03391812865497076,
+      "grad_norm": 1.30013028710232,
+      "learning_rate": 3.38785046728972e-05,
+      "loss": 1.0152,
       "step": 145
     },
     {
-      "epoch": 0.2798507462686567,
-      "grad_norm": 0.27409631854286964,
-      "learning_rate": 4.38217199358434e-05,
-      "loss": 0.5602,
+      "epoch": 0.03508771929824561,
+      "grad_norm": 1.646624872748112,
+      "learning_rate": 3.504672897196262e-05,
+      "loss": 1.0132,
       "step": 150
     },
     {
-      "epoch": 0.2891791044776119,
-      "grad_norm": 0.35726816948570184,
-      "learning_rate": 4.3336084393503545e-05,
-      "loss": 0.5569,
+      "epoch": 0.03625730994152047,
+      "grad_norm": 1.3045097017626717,
+      "learning_rate": 3.621495327102804e-05,
+      "loss": 1.0012,
       "step": 155
     },
     {
-      "epoch": 0.29850746268656714,
-      "grad_norm": 0.3394262848554844,
-      "learning_rate": 4.283536828906436e-05,
-      "loss": 0.5693,
+      "epoch": 0.03742690058479532,
+      "grad_norm": 1.462682033635244,
+      "learning_rate": 3.738317757009346e-05,
+      "loss": 1.0128,
       "step": 160
     },
     {
-      "epoch": 0.30783582089552236,
-      "grad_norm": 0.3323772279369,
-      "learning_rate": 4.2320048450011684e-05,
-      "loss": 0.5627,
+      "epoch": 0.03859649122807018,
+      "grad_norm": 1.2429985745730512,
+      "learning_rate": 3.855140186915888e-05,
+      "loss": 0.9975,
       "step": 165
     },
     {
-      "epoch": 0.31716417910447764,
-      "grad_norm": 0.2948003685095971,
-      "learning_rate": 4.179061561083777e-05,
-      "loss": 0.5548,
+      "epoch": 0.03976608187134503,
+      "grad_norm": 1.4950336114052705,
+      "learning_rate": 3.97196261682243e-05,
+      "loss": 0.9831,
       "step": 170
     },
     {
-      "epoch": 0.32649253731343286,
-      "grad_norm": 0.3303510942467319,
-      "learning_rate": 4.124757394571914e-05,
-      "loss": 0.561,
+      "epoch": 0.04093567251461988,
+      "grad_norm": 1.3381138350136945,
+      "learning_rate": 4.088785046728972e-05,
+      "loss": 0.9917,
       "step": 175
     },
     {
-      "epoch": 0.3358208955223881,
-      "grad_norm": 0.2532452701988742,
-      "learning_rate": 4.069144058839605e-05,
-      "loss": 0.562,
+      "epoch": 0.042105263157894736,
+      "grad_norm": 1.3781491921537325,
+      "learning_rate": 4.205607476635514e-05,
+      "loss": 0.9728,
       "step": 180
     },
     {
-      "epoch": 0.3451492537313433,
-      "grad_norm": 0.29121513962228895,
-      "learning_rate": 4.012274513971061e-05,
-      "loss": 0.5661,
+      "epoch": 0.04327485380116959,
+      "grad_norm": 1.2707812943966086,
+      "learning_rate": 4.3224299065420565e-05,
+      "loss": 1.0,
       "step": 185
     },
     {
-      "epoch": 0.35447761194029853,
-      "grad_norm": 0.2838831732044427,
-      "learning_rate": 3.954202916327264e-05,
-      "loss": 0.5542,
+      "epoch": 0.044444444444444446,
+      "grad_norm": 1.1629390254978238,
+      "learning_rate": 4.4392523364485984e-05,
+      "loss": 0.9795,
       "step": 190
     },
     {
-      "epoch": 0.36380597014925375,
-      "grad_norm": 0.27387871756079074,
-      "learning_rate": 3.894984566973346e-05,
-      "loss": 0.5542,
+      "epoch": 0.0456140350877193,
+      "grad_norm": 1.380163782528444,
+      "learning_rate": 4.556074766355141e-05,
+      "loss": 0.9728,
       "step": 195
     },
     {
-      "epoch": 0.373134328358209,
-      "grad_norm": 0.28521012157940484,
-      "learning_rate": 3.834675859015876e-05,
-      "loss": 0.5582,
+      "epoch": 0.04678362573099415,
+      "grad_norm": 1.2042238039537414,
+      "learning_rate": 4.672897196261683e-05,
+      "loss": 0.9758,
       "step": 200
     },
     {
-      "epoch": 0.3824626865671642,
-      "grad_norm": 0.2522148272309235,
-      "learning_rate": 3.77333422390021e-05,
-      "loss": 0.549,
+      "epoch": 0.047953216374269005,
+      "grad_norm": 1.279604288513107,
+      "learning_rate": 4.7897196261682245e-05,
+      "loss": 0.9972,
       "step": 205
     },
     {
-      "epoch": 0.3917910447761194,
-      "grad_norm": 0.28536211195515926,
-      "learning_rate": 3.711018076719034e-05,
-      "loss": 0.5561,
+      "epoch": 0.04912280701754386,
+      "grad_norm": 1.2721504709190998,
+      "learning_rate": 4.9065420560747664e-05,
+      "loss": 0.9741,
       "step": 210
     },
     {
-      "epoch": 0.40111940298507465,
-      "grad_norm": 0.27095713532689863,
-      "learning_rate": 3.647786760584194e-05,
-      "loss": 0.5604,
+      "epoch": 0.050292397660818715,
+      "grad_norm": 1.2259731220679402,
+      "learning_rate": 4.9999993267346444e-05,
+      "loss": 0.9679,
       "step": 215
     },
     {
-      "epoch": 0.41044776119402987,
-      "grad_norm": 0.2690157270304413,
-      "learning_rate": 3.583700490114776e-05,
-      "loss": 0.5586,
+      "epoch": 0.05146198830409357,
+      "grad_norm": 1.298654189786969,
+      "learning_rate": 4.999975762489519e-05,
+      "loss": 0.9852,
       "step": 220
     },
     {
-      "epoch": 0.4197761194029851,
-      "grad_norm": 0.30463417382486574,
-      "learning_rate": 3.518820294095267e-05,
-      "loss": 0.5545,
+      "epoch": 0.05263157894736842,
+      "grad_norm": 1.2924316625203613,
+      "learning_rate": 4.9999185353795504e-05,
+      "loss": 0.9673,
       "step": 225
     },
     {
-      "epoch": 0.4291044776119403,
-      "grad_norm": 0.27247515102836756,
-      "learning_rate": 3.453207957358377e-05,
-      "loss": 0.5464,
+      "epoch": 0.05380116959064327,
+      "grad_norm": 1.366376650054188,
+      "learning_rate": 4.99982764626094e-05,
+      "loss": 0.9815,
       "step": 230
     },
     {
-      "epoch": 0.43843283582089554,
-      "grad_norm": 0.2734906878941893,
-      "learning_rate": 3.386925961947906e-05,
-      "loss": 0.5476,
+      "epoch": 0.05497076023391813,
+      "grad_norm": 1.274755614910159,
+      "learning_rate": 4.9997030964935195e-05,
+      "loss": 0.987,
       "step": 235
     },
     {
-      "epoch": 0.44776119402985076,
-      "grad_norm": 0.2830639716162809,
-      "learning_rate": 3.320037427617639e-05,
-      "loss": 0.555,
+      "epoch": 0.056140350877192984,
+      "grad_norm": 1.6362492872363399,
+      "learning_rate": 4.9995448879407316e-05,
+      "loss": 0.961,
       "step": 240
     },
     {
-      "epoch": 0.457089552238806,
-      "grad_norm": 0.2641531917877042,
-      "learning_rate": 3.252606051722972e-05,
-      "loss": 0.5535,
+      "epoch": 0.05730994152046784,
+      "grad_norm": 1.2051858843379946,
+      "learning_rate": 4.999353022969603e-05,
+      "loss": 0.9604,
       "step": 245
     },
     {
-      "epoch": 0.4664179104477612,
-      "grad_norm": 0.29046339988398734,
-      "learning_rate": 3.1846960485624886e-05,
-      "loss": 0.5427,
+      "epoch": 0.05847953216374269,
+      "grad_norm": 1.3793185249466788,
+      "learning_rate": 4.999127504450709e-05,
+      "loss": 0.9591,
       "step": 250
     },
     {
-      "epoch": 0.47574626865671643,
-      "grad_norm": 0.25002394008962564,
-      "learning_rate": 3.1163720882272516e-05,
-      "loss": 0.5505,
+      "epoch": 0.05964912280701754,
+      "grad_norm": 1.2084655281854946,
+      "learning_rate": 4.998868335758132e-05,
+      "loss": 0.9535,
       "step": 255
     },
     {
-      "epoch": 0.48507462686567165,
-      "grad_norm": 0.29039809871972583,
-      "learning_rate": 3.047699235016056e-05,
-      "loss": 0.5428,
+      "epoch": 0.0608187134502924,
+      "grad_norm": 1.1876622728566322,
+      "learning_rate": 4.998575520769404e-05,
+      "loss": 0.9772,
       "step": 260
     },
     {
-      "epoch": 0.4944029850746269,
-      "grad_norm": 0.3096620559575199,
-      "learning_rate": 2.9787428854752736e-05,
-      "loss": 0.5367,
+      "epoch": 0.06198830409356725,
+      "grad_norm": 1.4161498983349252,
+      "learning_rate": 4.99824906386546e-05,
+      "loss": 0.9452,
       "step": 265
     },
     {
-      "epoch": 0.503731343283582,
-      "grad_norm": 0.2519415451164722,
-      "learning_rate": 2.9095687061223058e-05,
-      "loss": 0.5513,
+      "epoch": 0.06315789473684211,
+      "grad_norm": 1.3383887094892624,
+      "learning_rate": 4.997888969930562e-05,
+      "loss": 0.95,
       "step": 270
     },
     {
-      "epoch": 0.5130597014925373,
-      "grad_norm": 0.30354814640323574,
-      "learning_rate": 2.8402425709119435e-05,
-      "loss": 0.5504,
+      "epoch": 0.06432748538011696,
+      "grad_norm": 1.2656564068860054,
+      "learning_rate": 4.997495244352232e-05,
+      "loss": 0.9365,
       "step": 275
     },
     {
-      "epoch": 0.5223880597014925,
-      "grad_norm": 0.27641225616763077,
-      "learning_rate": 2.7708304985051868e-05,
-      "loss": 0.5474,
+      "epoch": 0.06549707602339182,
+      "grad_norm": 1.0862939966775438,
+      "learning_rate": 4.9970678930211704e-05,
+      "loss": 0.9525,
       "step": 280
     },
     {
-      "epoch": 0.5317164179104478,
-      "grad_norm": 0.24522657430707062,
-      "learning_rate": 2.7013985894002623e-05,
-      "loss": 0.5353,
+      "epoch": 0.06666666666666667,
+      "grad_norm": 1.2919679051885777,
+      "learning_rate": 4.996606922331165e-05,
+      "loss": 0.9529,
       "step": 285
     },
     {
-      "epoch": 0.5410447761194029,
-      "grad_norm": 0.23882848280564026,
-      "learning_rate": 2.6320129629857093e-05,
-      "loss": 0.5482,
+      "epoch": 0.06783625730994151,
+      "grad_norm": 1.0647888476826568,
+      "learning_rate": 4.996112339179e-05,
+      "loss": 0.9324,
       "step": 290
     },
     {
-      "epoch": 0.5503731343283582,
-      "grad_norm": 0.24387151760157888,
-      "learning_rate": 2.56273969457547e-05,
-      "loss": 0.537,
+      "epoch": 0.06900584795321638,
+      "grad_norm": 1.305388044138423,
+      "learning_rate": 4.995584150964347e-05,
+      "loss": 0.9259,
       "step": 295
     },
     {
-      "epoch": 0.5597014925373134,
-      "grad_norm": 0.23022777271615255,
-      "learning_rate": 2.4936447524859625e-05,
-      "loss": 0.5414,
+      "epoch": 0.07017543859649122,
+      "grad_norm": 1.231889595549824,
+      "learning_rate": 4.99502236558966e-05,
+      "loss": 0.948,
       "step": 300
     },
     {
-      "epoch": 0.5690298507462687,
-      "grad_norm": 0.29890414385139613,
-      "learning_rate": 2.4247939352150386e-05,
-      "loss": 0.5365,
+      "epoch": 0.07134502923976609,
+      "grad_norm": 1.092939130481041,
+      "learning_rate": 4.994426991460055e-05,
+      "loss": 0.932,
       "step": 305
     },
     {
-      "epoch": 0.5783582089552238,
-      "grad_norm": 0.2500574341333853,
-      "learning_rate": 2.3562528087826573e-05,
-      "loss": 0.5426,
+      "epoch": 0.07251461988304093,
+      "grad_norm": 1.3112640010430932,
+      "learning_rate": 4.993798037483182e-05,
+      "loss": 0.9327,
       "step": 310
     },
     {
-      "epoch": 0.5876865671641791,
-      "grad_norm": 0.23443688919292832,
-      "learning_rate": 2.2880866442929544e-05,
-      "loss": 0.5396,
+      "epoch": 0.07368421052631578,
+      "grad_norm": 1.1476970342552901,
+      "learning_rate": 4.993135513069094e-05,
+      "loss": 0.9482,
       "step": 315
     },
     {
-      "epoch": 0.5970149253731343,
-      "grad_norm": 0.235543729544232,
-      "learning_rate": 2.2203603557771447e-05,
-      "loss": 0.5357,
+      "epoch": 0.07485380116959064,
+      "grad_norm": 1.1633564547545994,
+      "learning_rate": 4.992439428130109e-05,
+      "loss": 0.9217,
       "step": 320
     },
     {
-      "epoch": 0.6063432835820896,
-      "grad_norm": 0.24168947061983012,
-      "learning_rate": 2.153138438376473e-05,
-      "loss": 0.534,
+      "epoch": 0.07602339181286549,
+      "grad_norm": 1.0735521991747736,
+      "learning_rate": 4.991709793080655e-05,
+      "loss": 0.9124,
       "step": 325
     },
     {
-      "epoch": 0.6156716417910447,
-      "grad_norm": 0.2343970377460005,
-      "learning_rate": 2.0864849069240645e-05,
-      "loss": 0.5387,
+      "epoch": 0.07719298245614035,
+      "grad_norm": 1.0492850297391971,
+      "learning_rate": 4.990946618837117e-05,
+      "loss": 0.9393,
       "step": 330
     },
     {
-      "epoch": 0.625,
-      "grad_norm": 0.22771536353141034,
-      "learning_rate": 2.0204632349841667e-05,
-      "loss": 0.5355,
+      "epoch": 0.0783625730994152,
+      "grad_norm": 1.3559203599333753,
+      "learning_rate": 4.9901499168176786e-05,
+      "loss": 0.9256,
       "step": 335
     },
     {
-      "epoch": 0.6343283582089553,
-      "grad_norm": 0.20829657516995473,
-      "learning_rate": 1.9551362944068462e-05,
-      "loss": 0.5377,
+      "epoch": 0.07953216374269007,
+      "grad_norm": 1.5371317714373305,
+      "learning_rate": 4.989319698942145e-05,
+      "loss": 0.9273,
       "step": 340
     },
     {
-      "epoch": 0.6436567164179104,
-      "grad_norm": 0.2278114191697408,
-      "learning_rate": 1.890566295455678e-05,
-      "loss": 0.531,
+      "epoch": 0.08070175438596491,
+      "grad_norm": 1.9231609823006572,
+      "learning_rate": 4.9884559776317644e-05,
+      "loss": 0.9181,
       "step": 345
     },
     {
-      "epoch": 0.6529850746268657,
-      "grad_norm": 0.21684874667889517,
-      "learning_rate": 1.8268147275654707e-05,
-      "loss": 0.541,
+      "epoch": 0.08187134502923976,
+      "grad_norm": 1.1887874506902512,
+      "learning_rate": 4.987558765809048e-05,
+      "loss": 0.9159,
       "step": 350
     },
     {
-      "epoch": 0.6623134328358209,
-      "grad_norm": 0.20697955028063816,
-      "learning_rate": 1.7639423007864252e-05,
-      "loss": 0.5278,
+      "epoch": 0.08304093567251462,
+      "grad_norm": 1.1113774961997955,
+      "learning_rate": 4.986628076897572e-05,
+      "loss": 0.9255,
       "step": 355
     },
     {
-      "epoch": 0.6716417910447762,
-      "grad_norm": 0.2173024145443538,
-      "learning_rate": 1.702008887970491e-05,
-      "loss": 0.5287,
+      "epoch": 0.08421052631578947,
+      "grad_norm": 1.2529780917385798,
+      "learning_rate": 4.985663924821778e-05,
+      "loss": 0.9134,
       "step": 360
     },
     {
-      "epoch": 0.6809701492537313,
-      "grad_norm": 0.21746637596733098,
-      "learning_rate": 1.6410734677549872e-05,
-      "loss": 0.5319,
+      "epoch": 0.08538011695906433,
+      "grad_norm": 1.1687004195151305,
+      "learning_rate": 4.984666324006763e-05,
+      "loss": 0.939,
       "step": 365
     },
     {
-      "epoch": 0.6902985074626866,
-      "grad_norm": 0.21602212005979432,
-      "learning_rate": 1.58119406839777e-05,
-      "loss": 0.5308,
+      "epoch": 0.08654970760233918,
+      "grad_norm": 1.2034837816155135,
+      "learning_rate": 4.983635289378065e-05,
+      "loss": 0.9166,
       "step": 370
     },
     {
-      "epoch": 0.6996268656716418,
-      "grad_norm": 0.23184190681367317,
-      "learning_rate": 1.5224277125174388e-05,
-      "loss": 0.5338,
+      "epoch": 0.08771929824561403,
+      "grad_norm": 1.1152012751772677,
+      "learning_rate": 4.9825708363614434e-05,
+      "loss": 0.9326,
       "step": 375
     },
     {
-      "epoch": 0.7089552238805971,
-      "grad_norm": 0.21728095589418825,
-      "learning_rate": 1.464830362791204e-05,
-      "loss": 0.536,
+      "epoch": 0.08888888888888889,
+      "grad_norm": 1.282352801615335,
+      "learning_rate": 4.981472980882641e-05,
+      "loss": 0.9097,
       "step": 380
     },
     {
-      "epoch": 0.7182835820895522,
-      "grad_norm": 0.2030760900341267,
-      "learning_rate": 1.4084568686621314e-05,
-      "loss": 0.5383,
+      "epoch": 0.09005847953216374,
+      "grad_norm": 1.3555014384579822,
+      "learning_rate": 4.980341739367151e-05,
+      "loss": 0.9566,
       "step": 385
     },
     {
-      "epoch": 0.7276119402985075,
-      "grad_norm": 0.20591135591474655,
-      "learning_rate": 1.3533609141065008e-05,
-      "loss": 0.5366,
+      "epoch": 0.0912280701754386,
+      "grad_norm": 0.9969319217580619,
+      "learning_rate": 4.979177128739968e-05,
+      "loss": 0.9183,
       "step": 390
     },
     {
-      "epoch": 0.7369402985074627,
-      "grad_norm": 0.2158933827283441,
-      "learning_rate": 1.299594966511038e-05,
-      "loss": 0.5338,
+      "epoch": 0.09239766081871345,
+      "grad_norm": 1.112979405264304,
+      "learning_rate": 4.977979166425339e-05,
+      "loss": 0.9087,
       "step": 395
     },
     {
-      "epoch": 0.746268656716418,
-      "grad_norm": 0.2099802240402189,
-      "learning_rate": 1.2472102267086904e-05,
-      "loss": 0.5296,
+      "epoch": 0.0935672514619883,
+      "grad_norm": 7.571188451889394,
+      "learning_rate": 4.976747870346498e-05,
+      "loss": 0.9075,
       "step": 400
     },
     {
-      "epoch": 0.7555970149253731,
-      "grad_norm": 0.20817868352501673,
-      "learning_rate": 1.1962565802205255e-05,
-      "loss": 0.5362,
+      "epoch": 0.09473684210526316,
+      "grad_norm": 1.2588282458972042,
+      "learning_rate": 4.9754832589254e-05,
+      "loss": 0.9133,
       "step": 405
     },
     {
-      "epoch": 0.7649253731343284,
-      "grad_norm": 0.19443834588933895,
-      "learning_rate": 1.1467825497501954e-05,
-      "loss": 0.5226,
+      "epoch": 0.09590643274853801,
+      "grad_norm": 1.1389033831691695,
+      "learning_rate": 4.974185351082447e-05,
+      "loss": 0.9071,
       "step": 410
     },
     {
-      "epoch": 0.7742537313432836,
-      "grad_norm": 0.21485380548615546,
-      "learning_rate": 1.0988352489762006e-05,
-      "loss": 0.5384,
+      "epoch": 0.09707602339181287,
+      "grad_norm": 1.0464472246368346,
+      "learning_rate": 4.972854166236201e-05,
+      "loss": 0.9033,
       "step": 415
     },
     {
-      "epoch": 0.7835820895522388,
-      "grad_norm": 0.20232217393186297,
-      "learning_rate": 1.052460337685951e-05,
-      "loss": 0.5299,
+      "epoch": 0.09824561403508772,
+      "grad_norm": 1.0609546057929313,
+      "learning_rate": 4.9714897243030984e-05,
+      "loss": 0.9054,
       "step": 420
     },
     {
-      "epoch": 0.792910447761194,
-      "grad_norm": 0.20013419607941366,
-      "learning_rate": 1.0077019782943584e-05,
-      "loss": 0.5282,
+      "epoch": 0.09941520467836257,
+      "grad_norm": 0.9766945821904102,
+      "learning_rate": 4.970092045697146e-05,
+      "loss": 0.892,
       "step": 425
     },
     {
-      "epoch": 0.8022388059701493,
-      "grad_norm": 0.20973920890315037,
-      "learning_rate": 9.646027937883622e-06,
-      "loss": 0.5291,
+      "epoch": 0.10058479532163743,
+      "grad_norm": 1.0824192119502347,
+      "learning_rate": 4.9686611513296216e-05,
+      "loss": 0.9076,
       "step": 430
     },
     {
-      "epoch": 0.8115671641791045,
-      "grad_norm": 0.18017086616755254,
-      "learning_rate": 9.232038271374377e-06,
-      "loss": 0.531,
+      "epoch": 0.10175438596491228,
+      "grad_norm": 1.069699233932774,
+      "learning_rate": 4.9671970626087574e-05,
+      "loss": 0.9131,
       "step": 435
     },
     {
-      "epoch": 0.8208955223880597,
-      "grad_norm": 0.19917505042319333,
-      "learning_rate": 8.835445022087426e-06,
-      "loss": 0.5256,
+      "epoch": 0.10292397660818714,
+      "grad_norm": 1.0973125930518677,
+      "learning_rate": 4.96569980143942e-05,
+      "loss": 0.8959,
       "step": 440
     },
     {
-      "epoch": 0.8302238805970149,
-      "grad_norm": 0.19572367635284946,
-      "learning_rate": 8.456625862241193e-06,
-      "loss": 0.5358,
+      "epoch": 0.10409356725146199,
+      "grad_norm": 1.1776527069272567,
+      "learning_rate": 4.964169390222784e-05,
+      "loss": 0.901,
       "step": 445
     },
     {
-      "epoch": 0.8395522388059702,
-      "grad_norm": 0.18884794801788335,
-      "learning_rate": 8.095941537947057e-06,
-      "loss": 0.5328,
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.0756769524406018,
+      "learning_rate": 4.9626058518559975e-05,
+      "loss": 0.8986,
       "step": 450
     },
     {
-      "epoch": 0.8488805970149254,
-      "grad_norm": 0.20340244217011463,
-      "learning_rate": 7.753735525674059e-06,
-      "loss": 0.5256,
+      "epoch": 0.1064327485380117,
+      "grad_norm": 1.1614986981567177,
+      "learning_rate": 4.961009209731837e-05,
+      "loss": 0.9064,
       "step": 455
     },
     {
-      "epoch": 0.8582089552238806,
-      "grad_norm": 0.19649967832173215,
-      "learning_rate": 7.430333705159286e-06,
-      "loss": 0.536,
+      "epoch": 0.10760233918128655,
+      "grad_norm": 1.1282199066528726,
+      "learning_rate": 4.959379487738359e-05,
+      "loss": 0.8956,
       "step": 460
     },
     {
-      "epoch": 0.8675373134328358,
-      "grad_norm": 0.18387110962359568,
-      "learning_rate": 7.126044049075548e-06,
-      "loss": 0.5408,
+      "epoch": 0.10877192982456141,
+      "grad_norm": 0.9009086073580999,
+      "learning_rate": 4.957716710258543e-05,
+      "loss": 0.893,
       "step": 465
     },
     {
-      "epoch": 0.8768656716417911,
-      "grad_norm": 0.18221197240089465,
-      "learning_rate": 6.8411563297516995e-06,
-      "loss": 0.5209,
+      "epoch": 0.10994152046783626,
+      "grad_norm": 0.9495009535418211,
+      "learning_rate": 4.956020902169924e-05,
+      "loss": 0.8879,
       "step": 470
     },
     {
-      "epoch": 0.8861940298507462,
-      "grad_norm": 0.17412098353043892,
-      "learning_rate": 6.575941843225068e-06,
-      "loss": 0.5246,
+      "epoch": 0.1111111111111111,
+      "grad_norm": 1.1217739879192823,
+      "learning_rate": 4.954292088844223e-05,
+      "loss": 0.8931,
       "step": 475
     },
     {
-      "epoch": 0.8955223880597015,
-      "grad_norm": 0.20332329472740424,
-      "learning_rate": 6.330653150888617e-06,
-      "loss": 0.5293,
+      "epoch": 0.11228070175438597,
+      "grad_norm": 0.9788914879224979,
+      "learning_rate": 4.952530296146969e-05,
+      "loss": 0.893,
       "step": 480
     },
     {
-      "epoch": 0.9048507462686567,
-      "grad_norm": 0.20020565104166105,
-      "learning_rate": 6.105523838979022e-06,
-      "loss": 0.5373,
+      "epoch": 0.11345029239766082,
+      "grad_norm": 0.9997978391833926,
+      "learning_rate": 4.9507355504371064e-05,
+      "loss": 0.8889,
       "step": 485
     },
     {
-      "epoch": 0.914179104477612,
-      "grad_norm": 0.18337411269600562,
-      "learning_rate": 5.900768296134551e-06,
-      "loss": 0.5238,
+      "epoch": 0.11461988304093568,
+      "grad_norm": 0.9268894518304199,
+      "learning_rate": 4.948907878566607e-05,
+      "loss": 0.8903,
       "step": 490
     },
     {
-      "epoch": 0.9235074626865671,
-      "grad_norm": 0.17868528646496779,
-      "learning_rate": 5.7165815092346825e-06,
-      "loss": 0.526,
+      "epoch": 0.11578947368421053,
+      "grad_norm": 72.60231360053689,
+      "learning_rate": 4.947047307880062e-05,
+      "loss": 0.9251,
       "step": 495
     },
     {
-      "epoch": 0.9328358208955224,
-      "grad_norm": 0.18273915963156537,
-      "learning_rate": 5.553138877715833e-06,
-      "loss": 0.5279,
+      "epoch": 0.11695906432748537,
+      "grad_norm": 1.0770947596244338,
+      "learning_rate": 4.945153866214278e-05,
+      "loss": 0.9046,
       "step": 500
     },
     {
-      "epoch": 0.9421641791044776,
-      "grad_norm": 0.18099116978277469,
-      "learning_rate": 5.410596046540051e-06,
-      "loss": 0.5229,
+      "epoch": 0.11812865497076024,
+      "grad_norm": 1.1295561944286985,
+      "learning_rate": 4.9432275818978595e-05,
+      "loss": 0.9037,
       "step": 505
     },
     {
-      "epoch": 0.9514925373134329,
-      "grad_norm": 0.19808608523886856,
-      "learning_rate": 5.28908875797568e-06,
-      "loss": 0.5235,
+      "epoch": 0.11929824561403508,
+      "grad_norm": 1.181923829984784,
+      "learning_rate": 4.941268483750782e-05,
+      "loss": 0.881,
       "step": 510
     },
     {
-      "epoch": 0.960820895522388,
-      "grad_norm": 0.18374809126530406,
-      "learning_rate": 5.1887327223312296e-06,
-      "loss": 0.5246,
+      "epoch": 0.12046783625730995,
+      "grad_norm": 0.9224413133089209,
+      "learning_rate": 4.939276601083965e-05,
+      "loss": 0.887,
       "step": 515
     },
     {
-      "epoch": 0.9701492537313433,
-      "grad_norm": 0.18957964129800453,
-      "learning_rate": 5.109623507765466e-06,
-      "loss": 0.5289,
+      "epoch": 0.1216374269005848,
+      "grad_norm": 0.9833800064012921,
+      "learning_rate": 4.937251963698829e-05,
+      "loss": 0.8942,
       "step": 520
     },
     {
-      "epoch": 0.9794776119402985,
-      "grad_norm": 0.19307156545346743,
-      "learning_rate": 5.051836449278715e-06,
-      "loss": 0.5311,
+      "epoch": 0.12280701754385964,
+      "grad_norm": 1.0462609066040864,
+      "learning_rate": 4.935194601886855e-05,
+      "loss": 0.9002,
       "step": 525
     },
     {
-      "epoch": 0.9888059701492538,
-      "grad_norm": 0.1907323496401664,
-      "learning_rate": 5.015426576972003e-06,
-      "loss": 0.5298,
+      "epoch": 0.1239766081871345,
+      "grad_norm": 0.8684931504628074,
+      "learning_rate": 4.9331045464291246e-05,
+      "loss": 0.8851,
       "step": 530
     },
     {
-      "epoch": 0.9981343283582089,
-      "grad_norm": 0.17828047385506166,
-      "learning_rate": 5.000428563642382e-06,
-      "loss": 0.5299,
+      "epoch": 0.12514619883040937,
+      "grad_norm": 1.1786968981044232,
+      "learning_rate": 4.9309818285958685e-05,
+      "loss": 0.8716,
       "step": 535
     },
+    {
+      "epoch": 0.12631578947368421,
+      "grad_norm": 1.0073700153107397,
+      "learning_rate": 4.928826480145988e-05,
+      "loss": 0.8618,
+      "step": 540
+    },
+    {
+      "epoch": 0.12748538011695906,
+      "grad_norm": 1.238751684745195,
+      "learning_rate": 4.9266385333265884e-05,
+      "loss": 0.932,
+      "step": 545
+    },
+    {
+      "epoch": 0.1286549707602339,
+      "grad_norm": 23.55004868122488,
+      "learning_rate": 4.924418020872493e-05,
+      "loss": 0.9185,
+      "step": 550
+    },
+    {
+      "epoch": 0.12982456140350876,
+      "grad_norm": 1.0626923904298728,
+      "learning_rate": 4.922164976005753e-05,
+      "loss": 0.8921,
+      "step": 555
+    },
+    {
+      "epoch": 0.13099415204678364,
+      "grad_norm": 0.9971583275794336,
+      "learning_rate": 4.91987943243515e-05,
+      "loss": 0.877,
+      "step": 560
+    },
+    {
+      "epoch": 0.13216374269005848,
+      "grad_norm": 2.6820593238577715,
+      "learning_rate": 4.917561424355696e-05,
+      "loss": 0.8934,
+      "step": 565
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 0.8203046983625998,
+      "learning_rate": 4.915210986448117e-05,
+      "loss": 0.8856,
+      "step": 570
+    },
+    {
+      "epoch": 0.13450292397660818,
+      "grad_norm": 0.9484997849093751,
+      "learning_rate": 4.912828153878335e-05,
+      "loss": 0.8754,
+      "step": 575
+    },
+    {
+      "epoch": 0.13567251461988303,
+      "grad_norm": 0.9387568040621501,
+      "learning_rate": 4.910412962296944e-05,
+      "loss": 0.876,
+      "step": 580
+    },
+    {
+      "epoch": 0.1368421052631579,
+      "grad_norm": 1.7811309672532951,
+      "learning_rate": 4.9079654478386724e-05,
+      "loss": 0.8642,
+      "step": 585
+    },
+    {
+      "epoch": 0.13801169590643275,
+      "grad_norm": 0.9033524326022147,
+      "learning_rate": 4.90548564712185e-05,
+      "loss": 0.87,
+      "step": 590
+    },
+    {
+      "epoch": 0.1391812865497076,
+      "grad_norm": 1.0353234405443796,
+      "learning_rate": 4.9029735972478505e-05,
+      "loss": 0.8671,
+      "step": 595
+    },
+    {
+      "epoch": 0.14035087719298245,
+      "grad_norm": 0.9727686082063292,
+      "learning_rate": 4.900429335800545e-05,
+      "loss": 0.8696,
+      "step": 600
+    },
+    {
+      "epoch": 0.1415204678362573,
+      "grad_norm": 1.0701058421505383,
+      "learning_rate": 4.897852900845733e-05,
+      "loss": 0.8763,
+      "step": 605
+    },
+    {
+      "epoch": 0.14269005847953217,
+      "grad_norm": 0.9599394196131743,
+      "learning_rate": 4.8952443309305777e-05,
+      "loss": 0.877,
+      "step": 610
+    },
+    {
+      "epoch": 0.14385964912280702,
+      "grad_norm": 0.9746668570633114,
+      "learning_rate": 4.892603665083027e-05,
+      "loss": 0.8663,
+      "step": 615
+    },
+    {
+      "epoch": 0.14502923976608187,
+      "grad_norm": 1.3661549067902636,
+      "learning_rate": 4.88993094281123e-05,
+      "loss": 0.8626,
+      "step": 620
+    },
+    {
+      "epoch": 0.14619883040935672,
+      "grad_norm": 1.1330888516399584,
+      "learning_rate": 4.887226204102945e-05,
+      "loss": 0.8786,
+      "step": 625
+    },
+    {
+      "epoch": 0.14736842105263157,
+      "grad_norm": 1.0877568219198543,
+      "learning_rate": 4.8844894894249424e-05,
+      "loss": 0.8638,
+      "step": 630
+    },
+    {
+      "epoch": 0.14853801169590644,
+      "grad_norm": 0.8283677070229845,
+      "learning_rate": 4.8817208397224015e-05,
+      "loss": 0.8544,
+      "step": 635
+    },
+    {
+      "epoch": 0.1497076023391813,
+      "grad_norm": 0.9792293752478846,
+      "learning_rate": 4.878920296418292e-05,
+      "loss": 0.8601,
+      "step": 640
+    },
+    {
+      "epoch": 0.15087719298245614,
+      "grad_norm": 1.0042722495236052,
+      "learning_rate": 4.876087901412758e-05,
+      "loss": 0.8611,
+      "step": 645
+    },
+    {
+      "epoch": 0.15204678362573099,
+      "grad_norm": 0.9753390505738546,
+      "learning_rate": 4.873223697082493e-05,
+      "loss": 0.8319,
+      "step": 650
+    },
+    {
+      "epoch": 0.15321637426900586,
+      "grad_norm": 0.8555144137662696,
+      "learning_rate": 4.870327726280103e-05,
+      "loss": 0.8643,
+      "step": 655
+    },
+    {
+      "epoch": 0.1543859649122807,
+      "grad_norm": 0.9466302026901982,
+      "learning_rate": 4.867400032333463e-05,
+      "loss": 0.8618,
+      "step": 660
+    },
+    {
+      "epoch": 0.15555555555555556,
+      "grad_norm": 0.9514870975188799,
+      "learning_rate": 4.8644406590450744e-05,
+      "loss": 0.858,
+      "step": 665
+    },
+    {
+      "epoch": 0.1567251461988304,
+      "grad_norm": 0.9414562779147595,
+      "learning_rate": 4.8614496506914087e-05,
+      "loss": 0.855,
+      "step": 670
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.924267446091397,
+      "learning_rate": 4.85842705202224e-05,
+      "loss": 0.8514,
+      "step": 675
+    },
+    {
+      "epoch": 0.15906432748538013,
+      "grad_norm": 0.8932378986259512,
+      "learning_rate": 4.8553729082599795e-05,
+      "loss": 0.8585,
+      "step": 680
+    },
+    {
+      "epoch": 0.16023391812865498,
+      "grad_norm": 0.9151852425271583,
+      "learning_rate": 4.852287265099e-05,
+      "loss": 0.8616,
+      "step": 685
+    },
+    {
+      "epoch": 0.16140350877192983,
+      "grad_norm": 0.9255147269629532,
+      "learning_rate": 4.849170168704948e-05,
+      "loss": 0.8405,
+      "step": 690
+    },
+    {
+      "epoch": 0.16257309941520467,
+      "grad_norm": 0.9039561346670378,
+      "learning_rate": 4.8460216657140586e-05,
+      "loss": 0.8516,
+      "step": 695
+    },
+    {
+      "epoch": 0.16374269005847952,
+      "grad_norm": 0.8504653507531936,
+      "learning_rate": 4.84284180323245e-05,
+      "loss": 0.8455,
+      "step": 700
+    },
+    {
+      "epoch": 0.1649122807017544,
+      "grad_norm": 0.8953131531650742,
+      "learning_rate": 4.8396306288354294e-05,
+      "loss": 0.8464,
+      "step": 705
+    },
+    {
+      "epoch": 0.16608187134502925,
+      "grad_norm": 0.907830157602171,
+      "learning_rate": 4.83638819056677e-05,
+      "loss": 0.858,
+      "step": 710
+    },
+    {
+      "epoch": 0.1672514619883041,
+      "grad_norm": 0.9199662102205977,
+      "learning_rate": 4.833114536938e-05,
+      "loss": 0.8461,
+      "step": 715
+    },
+    {
+      "epoch": 0.16842105263157894,
+      "grad_norm": 0.9459462552675176,
+      "learning_rate": 4.829809716927674e-05,
+      "loss": 0.8387,
+      "step": 720
+    },
+    {
+      "epoch": 0.1695906432748538,
+      "grad_norm": 0.9000637268586222,
+      "learning_rate": 4.8264737799806395e-05,
+      "loss": 0.8392,
+      "step": 725
+    },
+    {
+      "epoch": 0.17076023391812867,
+      "grad_norm": 0.8933049571672156,
+      "learning_rate": 4.823106776007298e-05,
+      "loss": 0.8467,
+      "step": 730
+    },
+    {
+      "epoch": 0.17192982456140352,
+      "grad_norm": 0.8789216469954654,
+      "learning_rate": 4.819708755382858e-05,
+      "loss": 0.8514,
+      "step": 735
+    },
+    {
+      "epoch": 0.17309941520467836,
+      "grad_norm": 0.9596321090665773,
+      "learning_rate": 4.816279768946584e-05,
+      "loss": 0.8413,
+      "step": 740
+    },
+    {
+      "epoch": 0.1742690058479532,
+      "grad_norm": 0.9644057901284268,
+      "learning_rate": 4.8128198680010314e-05,
+      "loss": 0.8516,
+      "step": 745
+    },
+    {
+      "epoch": 0.17543859649122806,
+      "grad_norm": 1.0517939442567537,
+      "learning_rate": 4.8093291043112796e-05,
+      "loss": 0.8624,
+      "step": 750
+    },
+    {
+      "epoch": 0.17660818713450294,
+      "grad_norm": 0.890785305956098,
+      "learning_rate": 4.8058075301041627e-05,
+      "loss": 0.8303,
+      "step": 755
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 0.8003592875071304,
+      "learning_rate": 4.802255198067482e-05,
+      "loss": 0.8363,
+      "step": 760
+    },
+    {
+      "epoch": 0.17894736842105263,
+      "grad_norm": 0.9862618280417637,
+      "learning_rate": 4.7986721613492184e-05,
+      "loss": 0.8293,
+      "step": 765
+    },
+    {
+      "epoch": 0.18011695906432748,
+      "grad_norm": 0.9484996630364185,
+      "learning_rate": 4.795058473556744e-05,
+      "loss": 0.8274,
+      "step": 770
+    },
+    {
+      "epoch": 0.18128654970760233,
+      "grad_norm": 0.8735616854937872,
+      "learning_rate": 4.791414188756009e-05,
+      "loss": 0.8492,
+      "step": 775
+    },
+    {
+      "epoch": 0.1824561403508772,
+      "grad_norm": 0.752039927949095,
+      "learning_rate": 4.787739361470743e-05,
+      "loss": 0.8576,
+      "step": 780
+    },
+    {
+      "epoch": 0.18362573099415205,
+      "grad_norm": 3.062282513190117,
+      "learning_rate": 4.7840340466816316e-05,
+      "loss": 0.8292,
+      "step": 785
+    },
+    {
+      "epoch": 0.1847953216374269,
+      "grad_norm": 0.8316980304126362,
+      "learning_rate": 4.780298299825503e-05,
+      "loss": 0.8351,
+      "step": 790
+    },
+    {
+      "epoch": 0.18596491228070175,
+      "grad_norm": 1.5278125736107897,
+      "learning_rate": 4.776532176794485e-05,
+      "loss": 0.8375,
+      "step": 795
+    },
+    {
+      "epoch": 0.1871345029239766,
+      "grad_norm": 0.8681275682341193,
+      "learning_rate": 4.7727357339351806e-05,
+      "loss": 0.8411,
+      "step": 800
+    },
+    {
+      "epoch": 0.18830409356725147,
+      "grad_norm": 0.9663714659586506,
+      "learning_rate": 4.768909028047823e-05,
+      "loss": 0.8427,
+      "step": 805
+    },
+    {
+      "epoch": 0.18947368421052632,
+      "grad_norm": 0.8699869778471631,
+      "learning_rate": 4.7650521163854205e-05,
+      "loss": 0.8448,
+      "step": 810
+    },
+    {
+      "epoch": 0.19064327485380117,
+      "grad_norm": 0.86171284732001,
+      "learning_rate": 4.761165056652903e-05,
+      "loss": 0.8372,
+      "step": 815
+    },
+    {
+      "epoch": 0.19181286549707602,
+      "grad_norm": 0.8957979278563446,
+      "learning_rate": 4.7572479070062616e-05,
+      "loss": 0.8417,
+      "step": 820
+    },
+    {
+      "epoch": 0.19298245614035087,
+      "grad_norm": 0.8114691833201496,
+      "learning_rate": 4.753300726051671e-05,
+      "loss": 0.8185,
+      "step": 825
+    },
+    {
+      "epoch": 0.19415204678362574,
+      "grad_norm": 0.8454664849785094,
+      "learning_rate": 4.7493235728446244e-05,
+      "loss": 0.8365,
+      "step": 830
+    },
+    {
+      "epoch": 0.1953216374269006,
+      "grad_norm": 0.9905108822443424,
+      "learning_rate": 4.745316506889035e-05,
+      "loss": 0.8457,
+      "step": 835
+    },
+    {
+      "epoch": 0.19649122807017544,
+      "grad_norm": 1.061405796747107,
+      "learning_rate": 4.74127958813636e-05,
+      "loss": 0.8406,
+      "step": 840
+    },
+    {
+      "epoch": 0.1976608187134503,
+      "grad_norm": 0.8473537941451442,
+      "learning_rate": 4.7372128769846924e-05,
+      "loss": 0.8338,
+      "step": 845
+    },
+    {
+      "epoch": 0.19883040935672514,
+      "grad_norm": 1.1466583458521185,
+      "learning_rate": 4.733116434277866e-05,
+      "loss": 0.8258,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7785799226749461,
+      "learning_rate": 4.7289903213045386e-05,
+      "loss": 0.8289,
+      "step": 855
+    },
+    {
+      "epoch": 0.20116959064327486,
+      "grad_norm": 0.8590755241548019,
+      "learning_rate": 4.7248345997972805e-05,
+      "loss": 0.8117,
+      "step": 860
+    },
+    {
+      "epoch": 0.2023391812865497,
+      "grad_norm": 0.7855870003132649,
+      "learning_rate": 4.720649331931645e-05,
+      "loss": 0.8202,
+      "step": 865
+    },
+    {
+      "epoch": 0.20350877192982456,
+      "grad_norm": 1.001202081783778,
+      "learning_rate": 4.716434580325243e-05,
+      "loss": 0.8216,
+      "step": 870
+    },
+    {
+      "epoch": 0.2046783625730994,
+      "grad_norm": 0.827480178465293,
+      "learning_rate": 4.712190408036805e-05,
+      "loss": 0.8262,
+      "step": 875
+    },
+    {
+      "epoch": 0.20584795321637428,
+      "grad_norm": 0.846609703440099,
+      "learning_rate": 4.7079168785652367e-05,
+      "loss": 0.8344,
+      "step": 880
+    },
+    {
+      "epoch": 0.20701754385964913,
+      "grad_norm": 0.9219532959757525,
+      "learning_rate": 4.703614055848668e-05,
+      "loss": 0.8172,
+      "step": 885
+    },
+    {
+      "epoch": 0.20818713450292398,
+      "grad_norm": 0.7637828460875458,
+      "learning_rate": 4.699282004263499e-05,
+      "loss": 0.8388,
+      "step": 890
+    },
+    {
+      "epoch": 0.20935672514619882,
+      "grad_norm": 0.7482566919726855,
+      "learning_rate": 4.6949207886234364e-05,
+      "loss": 0.8186,
+      "step": 895
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.925006863117072,
+      "learning_rate": 4.690530474178522e-05,
+      "loss": 0.8272,
+      "step": 900
+    },
+    {
+      "epoch": 0.21169590643274855,
+      "grad_norm": 0.7993467537605942,
+      "learning_rate": 4.686111126614156e-05,
+      "loss": 0.8197,
+      "step": 905
+    },
+    {
+      "epoch": 0.2128654970760234,
+      "grad_norm": 0.8829285193487382,
+      "learning_rate": 4.681662812050118e-05,
+      "loss": 0.8193,
+      "step": 910
+    },
+    {
+      "epoch": 0.21403508771929824,
+      "grad_norm": 0.9311623970289097,
+      "learning_rate": 4.6771855970395756e-05,
+      "loss": 0.827,
+      "step": 915
+    },
+    {
+      "epoch": 0.2152046783625731,
+      "grad_norm": 0.8106256897838203,
+      "learning_rate": 4.6726795485680866e-05,
+      "loss": 0.8165,
+      "step": 920
+    },
+    {
+      "epoch": 0.21637426900584794,
+      "grad_norm": 1.0733559910224837,
+      "learning_rate": 4.6681447340526e-05,
+      "loss": 0.8253,
+      "step": 925
+    },
+    {
+      "epoch": 0.21754385964912282,
+      "grad_norm": 0.8454361163690003,
+      "learning_rate": 4.663581221340445e-05,
+      "loss": 0.8181,
+      "step": 930
+    },
+    {
+      "epoch": 0.21871345029239767,
+      "grad_norm": 0.9381694065110877,
+      "learning_rate": 4.65898907870832e-05,
+      "loss": 0.8285,
+      "step": 935
+    },
+    {
+      "epoch": 0.2198830409356725,
+      "grad_norm": 0.9095410632213895,
+      "learning_rate": 4.654368374861264e-05,
+      "loss": 0.8294,
+      "step": 940
+    },
+    {
+      "epoch": 0.22105263157894736,
+      "grad_norm": 0.8991983031636669,
+      "learning_rate": 4.649719178931634e-05,
+      "loss": 0.8219,
+      "step": 945
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.8640964269521182,
+      "learning_rate": 4.645041560478073e-05,
+      "loss": 0.8182,
+      "step": 950
+    },
+    {
+      "epoch": 0.22339181286549709,
+      "grad_norm": 0.7717008178376892,
+      "learning_rate": 4.6403355894844603e-05,
+      "loss": 0.828,
+      "step": 955
+    },
+    {
+      "epoch": 0.22456140350877193,
+      "grad_norm": 0.7548344263953722,
+      "learning_rate": 4.635601336358873e-05,
+      "loss": 0.8118,
+      "step": 960
+    },
+    {
+      "epoch": 0.22573099415204678,
+      "grad_norm": 0.8125417926022064,
+      "learning_rate": 4.630838871932529e-05,
+      "loss": 0.8173,
+      "step": 965
+    },
+    {
+      "epoch": 0.22690058479532163,
+      "grad_norm": 0.8562232744379992,
+      "learning_rate": 4.626048267458727e-05,
+      "loss": 0.8018,
+      "step": 970
+    },
+    {
+      "epoch": 0.22807017543859648,
+      "grad_norm": 0.7282902246521552,
+      "learning_rate": 4.621229594611783e-05,
+      "loss": 0.8095,
+      "step": 975
+    },
+    {
+      "epoch": 0.22923976608187135,
+      "grad_norm": 0.8873506288089448,
+      "learning_rate": 4.616382925485953e-05,
+      "loss": 0.8132,
+      "step": 980
+    },
+    {
+      "epoch": 0.2304093567251462,
+      "grad_norm": 0.8104153922565714,
+      "learning_rate": 4.6115083325943606e-05,
+      "loss": 0.8158,
+      "step": 985
+    },
+    {
+      "epoch": 0.23157894736842105,
+      "grad_norm": 0.799851963773728,
+      "learning_rate": 4.606605888867908e-05,
+      "loss": 0.7857,
+      "step": 990
+    },
+    {
+      "epoch": 0.2327485380116959,
+      "grad_norm": 0.7738461797934784,
+      "learning_rate": 4.6016756676541847e-05,
+      "loss": 0.8218,
+      "step": 995
+    },
+    {
+      "epoch": 0.23391812865497075,
+      "grad_norm": 0.8576192649795367,
+      "learning_rate": 4.596717742716372e-05,
+      "loss": 0.8175,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23508771929824562,
+      "grad_norm": 0.923961104746636,
+      "learning_rate": 4.5917321882321396e-05,
+      "loss": 0.8081,
+      "step": 1005
+    },
+    {
+      "epoch": 0.23625730994152047,
+      "grad_norm": 0.8893668795457819,
+      "learning_rate": 4.5867190787925334e-05,
+      "loss": 0.8058,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23742690058479532,
+      "grad_norm": 0.8897258963021679,
+      "learning_rate": 4.5816784894008616e-05,
+      "loss": 0.825,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23859649122807017,
+      "grad_norm": 0.9015776574616987,
+      "learning_rate": 4.576610495471573e-05,
+      "loss": 0.7981,
+      "step": 1020
+    },
+    {
+      "epoch": 0.23976608187134502,
+      "grad_norm": 0.8415340622158813,
+      "learning_rate": 4.571515172829125e-05,
+      "loss": 0.8081,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2409356725146199,
+      "grad_norm": 0.8534832428040439,
+      "learning_rate": 4.5663925977068534e-05,
+      "loss": 0.8052,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24210526315789474,
+      "grad_norm": 0.7671642203239843,
+      "learning_rate": 4.561242846745831e-05,
+      "loss": 0.8083,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2432748538011696,
+      "grad_norm": 0.7947438206908344,
+      "learning_rate": 4.556065996993718e-05,
+      "loss": 0.8094,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24444444444444444,
+      "grad_norm": 0.7944046005545552,
+      "learning_rate": 4.550862125903613e-05,
+      "loss": 0.7997,
+      "step": 1045
+    },
+    {
+      "epoch": 0.24561403508771928,
+      "grad_norm": 0.7962491651355799,
+      "learning_rate": 4.5456313113328925e-05,
+      "loss": 0.8055,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24678362573099416,
+      "grad_norm": 0.7846208889540649,
+      "learning_rate": 4.540373631542045e-05,
+      "loss": 0.8017,
+      "step": 1055
+    },
+    {
+      "epoch": 0.247953216374269,
+      "grad_norm": 0.9255444869098531,
+      "learning_rate": 4.5350891651935024e-05,
+      "loss": 0.7956,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24912280701754386,
+      "grad_norm": 0.8264644211740068,
+      "learning_rate": 4.529777991350462e-05,
+      "loss": 0.787,
+      "step": 1065
+    },
+    {
+      "epoch": 0.25029239766081873,
+      "grad_norm": 0.8612407756813709,
+      "learning_rate": 4.524440189475702e-05,
+      "loss": 0.804,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25146198830409355,
+      "grad_norm": 0.9201193312825702,
+      "learning_rate": 4.519075839430395e-05,
+      "loss": 0.8139,
+      "step": 1075
+    },
+    {
+      "epoch": 0.25263157894736843,
+      "grad_norm": 0.8434760312094466,
+      "learning_rate": 4.513685021472913e-05,
+      "loss": 0.8084,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25380116959064325,
+      "grad_norm": 0.8097744734009323,
+      "learning_rate": 4.5082678162576266e-05,
+      "loss": 0.792,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2549707602339181,
+      "grad_norm": 0.8258410598348814,
+      "learning_rate": 4.502824304833694e-05,
+      "loss": 0.8018,
+      "step": 1090
+    },
+    {
+      "epoch": 0.256140350877193,
+      "grad_norm": 0.8898423105962325,
+      "learning_rate": 4.497354568643856e-05,
+      "loss": 0.801,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2573099415204678,
+      "grad_norm": 0.7682588506389413,
+      "learning_rate": 4.491858689523212e-05,
+      "loss": 0.7997,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2584795321637427,
+      "grad_norm": 0.8436536091119384,
+      "learning_rate": 4.486336749697996e-05,
+      "loss": 0.7958,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2596491228070175,
+      "grad_norm": 0.9134657570403645,
+      "learning_rate": 4.48078883178435e-05,
+      "loss": 0.7979,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2608187134502924,
+      "grad_norm": 0.9063533205659801,
+      "learning_rate": 4.4752150187870835e-05,
+      "loss": 0.8046,
+      "step": 1115
+    },
+    {
+      "epoch": 0.26198830409356727,
+      "grad_norm": 0.9728687505767328,
+      "learning_rate": 4.4696153940984336e-05,
+      "loss": 0.8005,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.8465165107087465,
+      "learning_rate": 4.463990041496819e-05,
+      "loss": 0.7928,
+      "step": 1125
+    },
+    {
+      "epoch": 0.26432748538011697,
+      "grad_norm": 0.8703041775291581,
+      "learning_rate": 4.4583390451455825e-05,
+      "loss": 0.7993,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2654970760233918,
+      "grad_norm": 0.8930384455063077,
+      "learning_rate": 4.4526624895917374e-05,
+      "loss": 0.8187,
+      "step": 1135
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.9533325500126925,
+      "learning_rate": 4.4469604597646955e-05,
+      "loss": 0.7915,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26783625730994154,
+      "grad_norm": 0.849949068893564,
+      "learning_rate": 4.441233040975003e-05,
+      "loss": 0.8084,
+      "step": 1145
+    },
+    {
+      "epoch": 0.26900584795321636,
+      "grad_norm": 0.8296019860693888,
+      "learning_rate": 4.435480318913061e-05,
+      "loss": 0.8099,
+      "step": 1150
+    },
+    {
+      "epoch": 0.27017543859649124,
+      "grad_norm": 0.8897148979616573,
+      "learning_rate": 4.429702379647842e-05,
+      "loss": 0.8076,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27134502923976606,
+      "grad_norm": 1.015942373286734,
+      "learning_rate": 4.4238993096256074e-05,
+      "loss": 0.7987,
+      "step": 1160
+    },
+    {
+      "epoch": 0.27251461988304093,
+      "grad_norm": 0.9170669198858242,
+      "learning_rate": 4.418071195668607e-05,
+      "loss": 0.8026,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2736842105263158,
+      "grad_norm": 0.8544721449457394,
+      "learning_rate": 4.412218124973787e-05,
+      "loss": 0.8037,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27485380116959063,
+      "grad_norm": 0.8109007540099256,
+      "learning_rate": 4.40634018511148e-05,
+      "loss": 0.7853,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2760233918128655,
+      "grad_norm": 0.784688745908909,
+      "learning_rate": 4.4004374640240984e-05,
+      "loss": 0.7829,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2771929824561403,
+      "grad_norm": 0.8426675242380559,
+      "learning_rate": 4.394510050024816e-05,
+      "loss": 0.7865,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2783625730994152,
+      "grad_norm": 0.7417145435611461,
+      "learning_rate": 4.388558031796249e-05,
+      "loss": 0.7912,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2795321637426901,
+      "grad_norm": 0.6946069237835871,
+      "learning_rate": 4.382581498389129e-05,
+      "loss": 0.7889,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2807017543859649,
+      "grad_norm": 0.830664235233533,
+      "learning_rate": 4.376580539220967e-05,
+      "loss": 0.8007,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2818713450292398,
+      "grad_norm": 4.044757818989429,
+      "learning_rate": 4.370555244074721e-05,
+      "loss": 0.803,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2830409356725146,
+      "grad_norm": 0.9700515861829565,
+      "learning_rate": 4.364505703097449e-05,
+      "loss": 0.8076,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28421052631578947,
+      "grad_norm": 0.8624986906785574,
+      "learning_rate": 4.358432006798962e-05,
+      "loss": 0.8062,
+      "step": 1215
+    },
+    {
+      "epoch": 0.28538011695906434,
+      "grad_norm": 0.8400857543523832,
+      "learning_rate": 4.352334246050468e-05,
+      "loss": 0.7977,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28654970760233917,
+      "grad_norm": 0.7507912757298415,
+      "learning_rate": 4.346212512083216e-05,
+      "loss": 0.7961,
+      "step": 1225
+    },
+    {
+      "epoch": 0.28771929824561404,
+      "grad_norm": 0.8662690316314504,
+      "learning_rate": 4.3400668964871255e-05,
+      "loss": 0.8004,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28888888888888886,
+      "grad_norm": 0.7711462814621782,
+      "learning_rate": 4.333897491209424e-05,
+      "loss": 0.7935,
+      "step": 1235
+    },
+    {
+      "epoch": 0.29005847953216374,
+      "grad_norm": 0.814472115106665,
+      "learning_rate": 4.327704388553262e-05,
+      "loss": 0.7875,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2912280701754386,
+      "grad_norm": 0.7777758368706338,
+      "learning_rate": 4.321487681176338e-05,
+      "loss": 0.7767,
+      "step": 1245
+    },
+    {
+      "epoch": 0.29239766081871343,
+      "grad_norm": 3.7426108288257525,
+      "learning_rate": 4.315247462089514e-05,
+      "loss": 0.7869,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2935672514619883,
+      "grad_norm": 0.8796776373301245,
+      "learning_rate": 4.308983824655418e-05,
+      "loss": 0.7885,
+      "step": 1255
+    },
+    {
+      "epoch": 0.29473684210526313,
+      "grad_norm": 0.8152628960370694,
+      "learning_rate": 4.30269686258705e-05,
+      "loss": 0.7793,
+      "step": 1260
+    },
+    {
+      "epoch": 0.295906432748538,
+      "grad_norm": 0.8531873491227141,
+      "learning_rate": 4.296386669946382e-05,
+      "loss": 0.7908,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2970760233918129,
+      "grad_norm": 0.8766898462050617,
+      "learning_rate": 4.290053341142945e-05,
+      "loss": 0.7912,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2982456140350877,
+      "grad_norm": 0.9437011508912876,
+      "learning_rate": 4.283696970932426e-05,
+      "loss": 0.7937,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2994152046783626,
+      "grad_norm": 0.760727957693321,
+      "learning_rate": 4.27731765441524e-05,
+      "loss": 0.7803,
+      "step": 1280
+    },
+    {
+      "epoch": 0.30058479532163745,
+      "grad_norm": 0.8030072219606168,
+      "learning_rate": 4.27091548703511e-05,
+      "loss": 0.7789,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3017543859649123,
+      "grad_norm": 0.7394799267095162,
+      "learning_rate": 4.264490564577647e-05,
+      "loss": 0.7743,
+      "step": 1290
+    },
+    {
+      "epoch": 0.30292397660818715,
+      "grad_norm": 0.8034110876662692,
+      "learning_rate": 4.258042983168906e-05,
+      "loss": 0.7964,
+      "step": 1295
+    },
+    {
+      "epoch": 0.30409356725146197,
+      "grad_norm": 0.8836489925526694,
+      "learning_rate": 4.251572839273953e-05,
+      "loss": 0.7843,
+      "step": 1300
+    },
+    {
+      "epoch": 0.30526315789473685,
+      "grad_norm": 0.8099715904918315,
+      "learning_rate": 4.245080229695422e-05,
+      "loss": 0.7765,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3064327485380117,
+      "grad_norm": 0.7211770553327171,
+      "learning_rate": 4.238565251572065e-05,
+      "loss": 0.7777,
+      "step": 1310
+    },
+    {
+      "epoch": 0.30760233918128654,
+      "grad_norm": 0.8430080190315516,
+      "learning_rate": 4.2320280023773004e-05,
+      "loss": 0.792,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3087719298245614,
+      "grad_norm": 0.7529220584999547,
+      "learning_rate": 4.225468579917755e-05,
+      "loss": 0.772,
+      "step": 1320
+    },
+    {
+      "epoch": 0.30994152046783624,
+      "grad_norm": 0.8246451276524118,
+      "learning_rate": 4.218887082331795e-05,
+      "loss": 0.7953,
+      "step": 1325
+    },
+    {
+      "epoch": 0.3111111111111111,
+      "grad_norm": 0.7572320411115961,
+      "learning_rate": 4.2122836080880656e-05,
+      "loss": 0.782,
+      "step": 1330
+    },
+    {
+      "epoch": 0.312280701754386,
+      "grad_norm": 0.8474595360345947,
+      "learning_rate": 4.2056582559840156e-05,
+      "loss": 0.7904,
+      "step": 1335
+    },
+    {
+      "epoch": 0.3134502923976608,
+      "grad_norm": 0.939458085785471,
+      "learning_rate": 4.199011125144414e-05,
+      "loss": 0.7763,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3146198830409357,
+      "grad_norm": 0.9781432479632555,
+      "learning_rate": 4.192342315019875e-05,
+      "loss": 0.7824,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.7908263498688014,
+      "learning_rate": 4.185651925385361e-05,
+      "loss": 0.78,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3169590643274854,
+      "grad_norm": 0.8281872357852522,
+      "learning_rate": 4.1789400563387014e-05,
+      "loss": 0.7786,
+      "step": 1355
+    },
+    {
+      "epoch": 0.31812865497076026,
+      "grad_norm": 0.7780377594605055,
+      "learning_rate": 4.172206808299082e-05,
+      "loss": 0.7729,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3192982456140351,
+      "grad_norm": 0.994616363239153,
+      "learning_rate": 4.1654522820055543e-05,
+      "loss": 0.7804,
+      "step": 1365
+    },
+    {
+      "epoch": 0.32046783625730996,
+      "grad_norm": 0.7754504675615862,
+      "learning_rate": 4.158676578515518e-05,
+      "loss": 0.7753,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3216374269005848,
+      "grad_norm": 0.7599612940476631,
+      "learning_rate": 4.1518797992032186e-05,
+      "loss": 0.773,
+      "step": 1375
+    },
+    {
+      "epoch": 0.32280701754385965,
+      "grad_norm": 0.7752784188686452,
+      "learning_rate": 4.145062045758223e-05,
+      "loss": 0.7763,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32397660818713453,
+      "grad_norm": 0.9495320531498634,
+      "learning_rate": 4.138223420183902e-05,
+      "loss": 0.7654,
+      "step": 1385
+    },
+    {
+      "epoch": 0.32514619883040935,
+      "grad_norm": 0.7827188590394395,
+      "learning_rate": 4.1313640247959056e-05,
+      "loss": 0.7718,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3263157894736842,
+      "grad_norm": 0.7318405468594709,
+      "learning_rate": 4.124483962220627e-05,
+      "loss": 0.77,
+      "step": 1395
+    },
+    {
+      "epoch": 0.32748538011695905,
+      "grad_norm": 0.763847628926599,
+      "learning_rate": 4.11758333539367e-05,
+      "loss": 0.7821,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3286549707602339,
+      "grad_norm": 0.7009721668324501,
+      "learning_rate": 4.1106622475583125e-05,
+      "loss": 0.7696,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3298245614035088,
+      "grad_norm": 0.7615443102851462,
+      "learning_rate": 4.1037208022639553e-05,
+      "loss": 0.7652,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3309941520467836,
+      "grad_norm": 0.7973051424443336,
+      "learning_rate": 4.0967591033645774e-05,
+      "loss": 0.7804,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3321637426900585,
+      "grad_norm": 0.7265512805457267,
+      "learning_rate": 4.08977725501718e-05,
+      "loss": 0.7744,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.7572181228671228,
+      "learning_rate": 4.08277536168023e-05,
+      "loss": 0.7692,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3345029239766082,
+      "grad_norm": 0.750546140392922,
+      "learning_rate": 4.075753528112095e-05,
+      "loss": 0.7731,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33567251461988307,
+      "grad_norm": 0.707104320903681,
+      "learning_rate": 4.068711859369478e-05,
+      "loss": 0.7584,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3368421052631579,
+      "grad_norm": 0.8018923237653571,
+      "learning_rate": 4.061650460805843e-05,
+      "loss": 0.7801,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33801169590643276,
+      "grad_norm": 0.7830790535393684,
+      "learning_rate": 4.054569438069843e-05,
+      "loss": 0.7665,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3391812865497076,
+      "grad_norm": 0.8337132040033033,
+      "learning_rate": 4.047468897103734e-05,
+      "loss": 0.7676,
+      "step": 1450
+    },
+    {
+      "epoch": 0.34035087719298246,
+      "grad_norm": 0.7352284392082878,
+      "learning_rate": 4.040348944141795e-05,
+      "loss": 0.7859,
+      "step": 1455
+    },
+    {
+      "epoch": 0.34152046783625734,
+      "grad_norm": 0.7819042171394409,
+      "learning_rate": 4.0332096857087346e-05,
+      "loss": 0.7802,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34269005847953216,
+      "grad_norm": 0.6839132320063686,
+      "learning_rate": 4.026051228618101e-05,
+      "loss": 0.7556,
+      "step": 1465
+    },
+    {
+      "epoch": 0.34385964912280703,
+      "grad_norm": 0.7721614295992741,
+      "learning_rate": 4.018873679970679e-05,
+      "loss": 0.7572,
+      "step": 1470
+    },
+    {
+      "epoch": 0.34502923976608185,
+      "grad_norm": 0.7881351916456097,
+      "learning_rate": 4.0116771471528946e-05,
+      "loss": 0.7836,
+      "step": 1475
+    },
+    {
+      "epoch": 0.34619883040935673,
+      "grad_norm": 17.824492025260795,
+      "learning_rate": 4.004461737835199e-05,
+      "loss": 0.8288,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3473684210526316,
+      "grad_norm": 0.9522727006560597,
+      "learning_rate": 3.9972275599704675e-05,
+      "loss": 0.7831,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3485380116959064,
+      "grad_norm": 1.0123330953920058,
+      "learning_rate": 3.989974721792376e-05,
+      "loss": 0.7834,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3497076023391813,
+      "grad_norm": 0.9256023414663554,
+      "learning_rate": 3.982703331813789e-05,
+      "loss": 0.7629,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3508771929824561,
+      "grad_norm": 0.8885054731766725,
+      "learning_rate": 3.97541349882513e-05,
+      "loss": 0.7872,
+      "step": 1500
+    },
+    {
+      "epoch": 0.352046783625731,
+      "grad_norm": 0.895363232481551,
+      "learning_rate": 3.9681053318927576e-05,
+      "loss": 0.7787,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3532163742690059,
+      "grad_norm": 0.7774622671008159,
+      "learning_rate": 3.960778940357332e-05,
+      "loss": 0.7595,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3543859649122807,
+      "grad_norm": 0.8521171304891902,
+      "learning_rate": 3.9534344338321804e-05,
+      "loss": 0.7752,
+      "step": 1515
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 0.7186359807168464,
+      "learning_rate": 3.946071922201654e-05,
+      "loss": 0.766,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3567251461988304,
+      "grad_norm": 0.7748704605835699,
+      "learning_rate": 3.9386915156194896e-05,
+      "loss": 0.7729,
+      "step": 1525
+    },
+    {
+      "epoch": 0.35789473684210527,
+      "grad_norm": 0.7519175958314207,
+      "learning_rate": 3.931293324507157e-05,
+      "loss": 0.7622,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35906432748538014,
+      "grad_norm": 0.9436517074298926,
+      "learning_rate": 3.9238774595522035e-05,
+      "loss": 0.7733,
+      "step": 1535
+    },
+    {
+      "epoch": 0.36023391812865496,
+      "grad_norm": 0.8056589850419347,
+      "learning_rate": 3.9164440317066106e-05,
+      "loss": 0.7508,
+      "step": 1540
+    },
+    {
+      "epoch": 0.36140350877192984,
+      "grad_norm": 0.7582238115959192,
+      "learning_rate": 3.9089931521851196e-05,
+      "loss": 0.7731,
+      "step": 1545
+    },
+    {
+      "epoch": 0.36257309941520466,
+      "grad_norm": 0.8651560934781966,
+      "learning_rate": 3.9015249324635765e-05,
+      "loss": 0.7841,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36374269005847953,
+      "grad_norm": 0.7138257240856716,
+      "learning_rate": 3.89403948427726e-05,
+      "loss": 0.7738,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3649122807017544,
+      "grad_norm": 0.7982217557908895,
+      "learning_rate": 3.8865369196192134e-05,
+      "loss": 0.7571,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36608187134502923,
+      "grad_norm": 0.8705834592275034,
+      "learning_rate": 3.8790173507385664e-05,
+      "loss": 0.7634,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3672514619883041,
+      "grad_norm": 0.7921869201643883,
+      "learning_rate": 3.871480890138854e-05,
+      "loss": 0.7665,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.7999595873888212,
+      "learning_rate": 3.863927650576339e-05,
+      "loss": 0.7547,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3695906432748538,
+      "grad_norm": 0.7509863075089731,
+      "learning_rate": 3.856357745058318e-05,
+      "loss": 0.759,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3707602339181287,
+      "grad_norm": 0.7307808913797633,
+      "learning_rate": 3.848771286841439e-05,
+      "loss": 0.7758,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3719298245614035,
+      "grad_norm": 0.7909239920658141,
+      "learning_rate": 3.841168389429996e-05,
+      "loss": 0.7675,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3730994152046784,
+      "grad_norm": 0.7482298719782716,
+      "learning_rate": 3.8335491665742405e-05,
+      "loss": 0.756,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3742690058479532,
+      "grad_norm": 0.67253397627471,
+      "learning_rate": 3.825913732268677e-05,
+      "loss": 0.7717,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37543859649122807,
+      "grad_norm": 0.7953984210357787,
+      "learning_rate": 3.818262200750356e-05,
+      "loss": 0.7666,
+      "step": 1605
+    },
+    {
+      "epoch": 0.37660818713450295,
+      "grad_norm": 0.7822921314337719,
+      "learning_rate": 3.810594686497163e-05,
+      "loss": 0.7709,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37777777777777777,
+      "grad_norm": 0.6928025407807362,
+      "learning_rate": 3.8029113042261097e-05,
+      "loss": 0.7351,
+      "step": 1615
+    },
+    {
+      "epoch": 0.37894736842105264,
+      "grad_norm": 0.8574680022140106,
+      "learning_rate": 3.795212168891618e-05,
+      "loss": 0.7742,
+      "step": 1620
+    },
+    {
+      "epoch": 0.38011695906432746,
+      "grad_norm": 0.835331112988705,
+      "learning_rate": 3.787497395683794e-05,
+      "loss": 0.7625,
+      "step": 1625
+    },
+    {
+      "epoch": 0.38128654970760234,
+      "grad_norm": 0.8993766649941991,
+      "learning_rate": 3.779767100026711e-05,
+      "loss": 0.7649,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3824561403508772,
+      "grad_norm": 0.8136822440130091,
+      "learning_rate": 3.772021397576683e-05,
+      "loss": 0.7564,
+      "step": 1635
+    },
+    {
+      "epoch": 0.38362573099415204,
+      "grad_norm": 0.8295895321913241,
+      "learning_rate": 3.764260404220529e-05,
+      "loss": 0.7745,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3847953216374269,
+      "grad_norm": 0.772989242303164,
+      "learning_rate": 3.75648423607384e-05,
+      "loss": 0.7668,
+      "step": 1645
+    },
+    {
+      "epoch": 0.38596491228070173,
+      "grad_norm": 0.7401867656024123,
+      "learning_rate": 3.748693009479248e-05,
+      "loss": 0.7564,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3871345029239766,
+      "grad_norm": 0.7597052025519834,
+      "learning_rate": 3.740886841004678e-05,
+      "loss": 0.7544,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3883040935672515,
+      "grad_norm": 0.7177753699525617,
+      "learning_rate": 3.7330658474416076e-05,
+      "loss": 0.7442,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3894736842105263,
+      "grad_norm": 0.7417872144963117,
+      "learning_rate": 3.725230145803319e-05,
+      "loss": 0.7633,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3906432748538012,
+      "grad_norm": 0.6613723653550772,
+      "learning_rate": 3.7173798533231493e-05,
+      "loss": 0.7633,
+      "step": 1670
+    },
+    {
+      "epoch": 0.391812865497076,
+      "grad_norm": 0.6732226054962314,
+      "learning_rate": 3.709515087452734e-05,
+      "loss": 0.767,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3929824561403509,
+      "grad_norm": 0.7377795343279899,
+      "learning_rate": 3.701635965860252e-05,
+      "loss": 0.7639,
+      "step": 1680
+    },
+    {
+      "epoch": 0.39415204678362575,
+      "grad_norm": 0.6914484317276971,
+      "learning_rate": 3.693742606428666e-05,
+      "loss": 0.748,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3953216374269006,
+      "grad_norm": 0.7827509721068225,
+      "learning_rate": 3.6858351272539524e-05,
+      "loss": 0.7716,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39649122807017545,
+      "grad_norm": 0.7243740975471005,
+      "learning_rate": 3.677913646643346e-05,
+      "loss": 0.7461,
+      "step": 1695
+    },
+    {
+      "epoch": 0.39766081871345027,
+      "grad_norm": 0.7271096610255222,
+      "learning_rate": 3.669978283113557e-05,
+      "loss": 0.7499,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39883040935672515,
+      "grad_norm": 0.8171882683333956,
+      "learning_rate": 3.662029155389007e-05,
+      "loss": 0.7592,
+      "step": 1705
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.729897299282628,
+      "learning_rate": 3.65406638240005e-05,
+      "loss": 0.7563,
+      "step": 1710
+    },
+    {
+      "epoch": 0.40116959064327484,
+      "grad_norm": 0.7843875646778868,
+      "learning_rate": 3.646090083281191e-05,
+      "loss": 0.7465,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4023391812865497,
+      "grad_norm": 0.8337858126407102,
+      "learning_rate": 3.638100377369308e-05,
+      "loss": 0.7525,
+      "step": 1720
+    },
+    {
+      "epoch": 0.40350877192982454,
+      "grad_norm": 0.7746287703227326,
+      "learning_rate": 3.630097384201859e-05,
+      "loss": 0.7474,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4046783625730994,
+      "grad_norm": 0.752728483401707,
+      "learning_rate": 3.6220812235151054e-05,
+      "loss": 0.7671,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4058479532163743,
+      "grad_norm": 0.7384298994892262,
+      "learning_rate": 3.614052015242307e-05,
+      "loss": 0.7569,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4070175438596491,
+      "grad_norm": 0.7301640418761658,
+      "learning_rate": 3.606009879511937e-05,
+      "loss": 0.7517,
+      "step": 1740
+    },
+    {
+      "epoch": 0.408187134502924,
+      "grad_norm": 0.7606719895102059,
+      "learning_rate": 3.597954936645883e-05,
+      "loss": 0.7627,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4093567251461988,
+      "grad_norm": 0.7993038146039282,
+      "learning_rate": 3.589887307157644e-05,
+      "loss": 0.762,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4105263157894737,
+      "grad_norm": 0.8077703903929386,
+      "learning_rate": 3.5818071117505285e-05,
+      "loss": 0.7449,
+      "step": 1755
+    },
+    {
+      "epoch": 0.41169590643274856,
+      "grad_norm": 0.7099492412004633,
+      "learning_rate": 3.573714471315852e-05,
+      "loss": 0.7631,
+      "step": 1760
+    },
+    {
+      "epoch": 0.4128654970760234,
+      "grad_norm": 0.7784601423237745,
+      "learning_rate": 3.565609506931124e-05,
+      "loss": 0.7557,
+      "step": 1765
+    },
+    {
+      "epoch": 0.41403508771929826,
+      "grad_norm": 0.7159896176820426,
+      "learning_rate": 3.557492339858236e-05,
+      "loss": 0.7536,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4152046783625731,
+      "grad_norm": 0.7469578404845566,
+      "learning_rate": 3.549363091541652e-05,
+      "loss": 0.7393,
+      "step": 1775
+    },
+    {
+      "epoch": 0.41637426900584795,
+      "grad_norm": 0.6837755812631964,
+      "learning_rate": 3.541221883606587e-05,
+      "loss": 0.7673,
+      "step": 1780
+    },
+    {
+      "epoch": 0.41754385964912283,
+      "grad_norm": 0.7526930864574983,
+      "learning_rate": 3.533068837857191e-05,
+      "loss": 0.7502,
+      "step": 1785
+    },
+    {
+      "epoch": 0.41871345029239765,
+      "grad_norm": 0.6964791661553479,
+      "learning_rate": 3.5249040762747216e-05,
+      "loss": 0.7643,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4198830409356725,
+      "grad_norm": 0.7176587460315915,
+      "learning_rate": 3.516727721015725e-05,
+      "loss": 0.7445,
+      "step": 1795
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.6949913252110833,
+      "learning_rate": 3.508539894410204e-05,
+      "loss": 0.7384,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4222222222222222,
+      "grad_norm": 0.6879985495866497,
+      "learning_rate": 3.500340718959789e-05,
+      "loss": 0.7474,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4233918128654971,
+      "grad_norm": 0.7255729290182649,
+      "learning_rate": 3.492130317335908e-05,
+      "loss": 0.7642,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4245614035087719,
+      "grad_norm": 0.7320558645901049,
+      "learning_rate": 3.483908812377944e-05,
+      "loss": 0.7485,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4257309941520468,
+      "grad_norm": 0.7458875901771815,
+      "learning_rate": 3.475676327091405e-05,
+      "loss": 0.7486,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4269005847953216,
+      "grad_norm": 0.8578208276395972,
+      "learning_rate": 3.46743298464608e-05,
+      "loss": 0.7495,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4280701754385965,
+      "grad_norm": 0.7859910962069061,
+      "learning_rate": 3.459178908374198e-05,
+      "loss": 0.7513,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42923976608187137,
+      "grad_norm": 0.6911557004560248,
+      "learning_rate": 3.450914221768577e-05,
+      "loss": 0.7367,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4304093567251462,
+      "grad_norm": 0.7456771712242715,
+      "learning_rate": 3.442639048480786e-05,
+      "loss": 0.7529,
+      "step": 1840
+    },
+    {
+      "epoch": 0.43157894736842106,
+      "grad_norm": 0.6893752794828176,
+      "learning_rate": 3.434353512319287e-05,
+      "loss": 0.7534,
+      "step": 1845
+    },
+    {
+      "epoch": 0.4327485380116959,
+      "grad_norm": 0.715785646529028,
+      "learning_rate": 3.426057737247585e-05,
+      "loss": 0.7528,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43391812865497076,
+      "grad_norm": 0.8076534821677026,
+      "learning_rate": 3.4177518473823765e-05,
+      "loss": 0.7629,
+      "step": 1855
+    },
+    {
+      "epoch": 0.43508771929824563,
+      "grad_norm": 0.7529189699100864,
+      "learning_rate": 3.409435966991687e-05,
+      "loss": 0.7558,
+      "step": 1860
+    },
+    {
+      "epoch": 0.43625730994152045,
+      "grad_norm": 0.7513403291393522,
+      "learning_rate": 3.4011102204930164e-05,
+      "loss": 0.7406,
+      "step": 1865
+    },
+    {
+      "epoch": 0.43742690058479533,
+      "grad_norm": 0.7777527987715089,
+      "learning_rate": 3.392774732451474e-05,
+      "loss": 0.7491,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43859649122807015,
+      "grad_norm": 0.743482516351825,
+      "learning_rate": 3.384429627577919e-05,
+      "loss": 0.73,
+      "step": 1875
+    },
+    {
+      "epoch": 0.439766081871345,
+      "grad_norm": 0.7648732013556975,
+      "learning_rate": 3.3760750307270885e-05,
+      "loss": 0.7582,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4409356725146199,
+      "grad_norm": 3.3134636103773465,
+      "learning_rate": 3.367711066895737e-05,
+      "loss": 0.7523,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4421052631578947,
+      "grad_norm": 0.7590057881358742,
+      "learning_rate": 3.359337861220762e-05,
+      "loss": 0.7581,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4432748538011696,
+      "grad_norm": 0.9364115069402698,
+      "learning_rate": 3.3509555389773295e-05,
+      "loss": 0.7489,
+      "step": 1895
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.7556042031825977,
+      "learning_rate": 3.3425642255770044e-05,
+      "loss": 0.7386,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4456140350877193,
+      "grad_norm": 0.7613584930553752,
+      "learning_rate": 3.334164046565873e-05,
+      "loss": 0.7366,
+      "step": 1905
+    },
+    {
+      "epoch": 0.44678362573099417,
+      "grad_norm": 0.7369071648445169,
+      "learning_rate": 3.3257551276226617e-05,
+      "loss": 0.733,
+      "step": 1910
+    },
+    {
+      "epoch": 0.447953216374269,
+      "grad_norm": 0.801402699988636,
+      "learning_rate": 3.31733759455686e-05,
+      "loss": 0.7411,
+      "step": 1915
+    },
+    {
+      "epoch": 0.44912280701754387,
+      "grad_norm": 0.7238254574287333,
+      "learning_rate": 3.308911573306837e-05,
+      "loss": 0.7387,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4502923976608187,
+      "grad_norm": 0.6952586214256408,
+      "learning_rate": 3.300477189937958e-05,
+      "loss": 0.7366,
+      "step": 1925
+    },
+    {
+      "epoch": 0.45146198830409356,
+      "grad_norm": 0.6960451113389217,
+      "learning_rate": 3.292034570640695e-05,
+      "loss": 0.7514,
+      "step": 1930
+    },
+    {
+      "epoch": 0.45263157894736844,
+      "grad_norm": 0.6934571085329695,
+      "learning_rate": 3.2835838417287446e-05,
+      "loss": 0.7409,
+      "step": 1935
+    },
+    {
+      "epoch": 0.45380116959064326,
+      "grad_norm": 0.7794450187246763,
+      "learning_rate": 3.2751251296371325e-05,
+      "loss": 0.7309,
+      "step": 1940
+    },
+    {
+      "epoch": 0.45497076023391814,
+      "grad_norm": 0.7783530977017811,
+      "learning_rate": 3.266658560920326e-05,
+      "loss": 0.7363,
+      "step": 1945
+    },
+    {
+      "epoch": 0.45614035087719296,
+      "grad_norm": 0.7530850469408922,
+      "learning_rate": 3.2581842622503366e-05,
+      "loss": 0.7474,
+      "step": 1950
+    },
+    {
+      "epoch": 0.45730994152046783,
+      "grad_norm": 0.7293168415552727,
+      "learning_rate": 3.249702360414829e-05,
+      "loss": 0.742,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4584795321637427,
+      "grad_norm": 0.7976185776852677,
+      "learning_rate": 3.24121298231522e-05,
+      "loss": 0.7252,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45964912280701753,
+      "grad_norm": 0.7634063554800602,
+      "learning_rate": 3.232716254964785e-05,
+      "loss": 0.7505,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4608187134502924,
+      "grad_norm": 0.7865100028320574,
+      "learning_rate": 3.224212305486753e-05,
+      "loss": 0.7541,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4619883040935672,
+      "grad_norm": 0.7166973372102381,
+      "learning_rate": 3.215701261112406e-05,
+      "loss": 0.7503,
+      "step": 1975
+    },
+    {
+      "epoch": 0.4631578947368421,
+      "grad_norm": 0.7887863496852786,
+      "learning_rate": 3.207183249179177e-05,
+      "loss": 0.7402,
+      "step": 1980
+    },
+    {
+      "epoch": 0.464327485380117,
+      "grad_norm": 0.6861005458337223,
+      "learning_rate": 3.198658397128742e-05,
+      "loss": 0.7403,
+      "step": 1985
+    },
+    {
+      "epoch": 0.4654970760233918,
+      "grad_norm": 0.713511921412075,
+      "learning_rate": 3.190126832505116e-05,
+      "loss": 0.7448,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 0.7395465637381757,
+      "learning_rate": 3.181588682952745e-05,
+      "loss": 0.742,
+      "step": 1995
+    },
+    {
+      "epoch": 0.4678362573099415,
+      "grad_norm": 0.716843196936708,
+      "learning_rate": 3.173044076214592e-05,
+      "loss": 0.7215,
+      "step": 2000
+    },
+    {
+      "epoch": 0.46900584795321637,
+      "grad_norm": 0.7143308102409406,
+      "learning_rate": 3.164493140130232e-05,
+      "loss": 0.7358,
+      "step": 2005
+    },
+    {
+      "epoch": 0.47017543859649125,
+      "grad_norm": 0.7158629240615204,
+      "learning_rate": 3.1559360026339335e-05,
+      "loss": 0.7454,
+      "step": 2010
+    },
+    {
+      "epoch": 0.47134502923976607,
+      "grad_norm": 0.8189451275414514,
+      "learning_rate": 3.1473727917527485e-05,
+      "loss": 0.7352,
+      "step": 2015
+    },
+    {
+      "epoch": 0.47251461988304094,
+      "grad_norm": 0.7565575619571483,
+      "learning_rate": 3.138803635604596e-05,
+      "loss": 0.7237,
+      "step": 2020
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.7072878072619535,
+      "learning_rate": 3.1302286623963414e-05,
+      "loss": 0.7476,
+      "step": 2025
+    },
+    {
+      "epoch": 0.47485380116959064,
+      "grad_norm": 0.7391386470759366,
+      "learning_rate": 3.121648000421886e-05,
+      "loss": 0.7454,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4760233918128655,
+      "grad_norm": 0.754752084236822,
+      "learning_rate": 3.113061778060241e-05,
+      "loss": 0.7392,
+      "step": 2035
+    },
+    {
+      "epoch": 0.47719298245614034,
+      "grad_norm": 0.7258136873037546,
+      "learning_rate": 3.10447012377361e-05,
+      "loss": 0.7485,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4783625730994152,
+      "grad_norm": 0.740460400541068,
+      "learning_rate": 3.0958731661054636e-05,
+      "loss": 0.7345,
+      "step": 2045
+    },
+    {
+      "epoch": 0.47953216374269003,
+      "grad_norm": 0.6894454665827823,
+      "learning_rate": 3.08727103367862e-05,
+      "loss": 0.7436,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4807017543859649,
+      "grad_norm": 0.7276684254100515,
+      "learning_rate": 3.078663855193322e-05,
+      "loss": 0.7316,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4818713450292398,
+      "grad_norm": 0.6805155330798511,
+      "learning_rate": 3.070051759425305e-05,
+      "loss": 0.7305,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4830409356725146,
+      "grad_norm": 0.7546120968812617,
+      "learning_rate": 3.0614348752238746e-05,
+      "loss": 0.739,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4842105263157895,
+      "grad_norm": 0.7229476827464224,
+      "learning_rate": 3.052813331509978e-05,
+      "loss": 0.7353,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4853801169590643,
+      "grad_norm": 0.6897066828114895,
+      "learning_rate": 3.0441872572742785e-05,
+      "loss": 0.7428,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4865497076023392,
+      "grad_norm": 0.6803167588334946,
+      "learning_rate": 3.035556781575219e-05,
+      "loss": 0.7377,
+      "step": 2080
+    },
+    {
+      "epoch": 0.48771929824561405,
+      "grad_norm": 0.7154975306867315,
+      "learning_rate": 3.0269220335370945e-05,
+      "loss": 0.751,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4888888888888889,
+      "grad_norm": 0.7036402820624741,
+      "learning_rate": 3.0182831423481227e-05,
+      "loss": 0.7372,
+      "step": 2090
+    },
+    {
+      "epoch": 0.49005847953216375,
+      "grad_norm": 0.6928895572678453,
+      "learning_rate": 3.0096402372585075e-05,
+      "loss": 0.7324,
+      "step": 2095
+    },
+    {
+      "epoch": 0.49122807017543857,
+      "grad_norm": 0.7467539933803762,
+      "learning_rate": 3.0009934475785083e-05,
+      "loss": 0.7292,
+      "step": 2100
+    },
+    {
+      "epoch": 0.49239766081871345,
+      "grad_norm": 0.7420950776012448,
+      "learning_rate": 2.9923429026765003e-05,
+      "loss": 0.7322,
+      "step": 2105
+    },
+    {
+      "epoch": 0.4935672514619883,
+      "grad_norm": 0.7072542140173895,
+      "learning_rate": 2.983688731977044e-05,
+      "loss": 0.7339,
+      "step": 2110
+    },
+    {
+      "epoch": 0.49473684210526314,
+      "grad_norm": 0.6860501654248469,
+      "learning_rate": 2.9750310649589465e-05,
+      "loss": 0.7218,
+      "step": 2115
+    },
+    {
+      "epoch": 0.495906432748538,
+      "grad_norm": 0.8043839479752058,
+      "learning_rate": 2.966370031153326e-05,
+      "loss": 0.7347,
+      "step": 2120
+    },
+    {
+      "epoch": 0.49707602339181284,
+      "grad_norm": 0.7213757854384905,
+      "learning_rate": 2.9577057601416717e-05,
+      "loss": 0.7378,
+      "step": 2125
+    },
+    {
+      "epoch": 0.4982456140350877,
+      "grad_norm": 0.6834101797567773,
+      "learning_rate": 2.9490383815539058e-05,
+      "loss": 0.7298,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4994152046783626,
+      "grad_norm": 0.7213077762540968,
+      "learning_rate": 2.9403680250664445e-05,
+      "loss": 0.7431,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5005847953216375,
+      "grad_norm": 0.7289094379202761,
+      "learning_rate": 2.931694820400259e-05,
+      "loss": 0.7347,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5017543859649123,
+      "grad_norm": 0.8099640413051555,
+      "learning_rate": 2.923018897318932e-05,
+      "loss": 0.7303,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5029239766081871,
+      "grad_norm": 0.7449546312217549,
+      "learning_rate": 2.914340385626717e-05,
+      "loss": 0.7347,
+      "step": 2150
+    },
+    {
+      "epoch": 0.504093567251462,
+      "grad_norm": 0.8184676100727405,
+      "learning_rate": 2.9056594151665985e-05,
+      "loss": 0.7081,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5052631578947369,
+      "grad_norm": 0.8611040013444063,
+      "learning_rate": 2.8969761158183466e-05,
+      "loss": 0.7311,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5064327485380117,
+      "grad_norm": 0.7244547799769302,
+      "learning_rate": 2.8882906174965742e-05,
+      "loss": 0.7468,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5076023391812865,
+      "grad_norm": 0.7499135951738628,
+      "learning_rate": 2.879603050148796e-05,
+      "loss": 0.7457,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5087719298245614,
+      "grad_norm": 0.6938064677300961,
+      "learning_rate": 2.8709135437534806e-05,
+      "loss": 0.7354,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5099415204678363,
+      "grad_norm": 0.6650288090095613,
+      "learning_rate": 2.8622222283181087e-05,
+      "loss": 0.7414,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5111111111111111,
+      "grad_norm": 0.7133364365712555,
+      "learning_rate": 2.853529233877227e-05,
+      "loss": 0.73,
+      "step": 2185
+    },
+    {
+      "epoch": 0.512280701754386,
+      "grad_norm": 0.6540426370572834,
+      "learning_rate": 2.8448346904905e-05,
+      "loss": 0.7294,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5134502923976608,
+      "grad_norm": 0.6939489433904377,
+      "learning_rate": 2.8361387282407704e-05,
+      "loss": 0.7114,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5146198830409356,
+      "grad_norm": 0.732811254127614,
+      "learning_rate": 2.827441477232105e-05,
+      "loss": 0.7296,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5157894736842106,
+      "grad_norm": 0.6531853113547864,
+      "learning_rate": 2.818743067587857e-05,
+      "loss": 0.72,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5169590643274854,
+      "grad_norm": 0.6766250203305809,
+      "learning_rate": 2.8100436294487092e-05,
+      "loss": 0.746,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5181286549707602,
+      "grad_norm": 0.6629062039445471,
+      "learning_rate": 2.8013432929707374e-05,
+      "loss": 0.7245,
+      "step": 2215
+    },
+    {
+      "epoch": 0.519298245614035,
+      "grad_norm": 0.6740973772958735,
+      "learning_rate": 2.7926421883234544e-05,
+      "loss": 0.7156,
+      "step": 2220
+    },
+    {
+      "epoch": 0.52046783625731,
+      "grad_norm": 0.7106665581583453,
+      "learning_rate": 2.7839404456878666e-05,
+      "loss": 0.7324,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5216374269005848,
+      "grad_norm": 0.6367158771455951,
+      "learning_rate": 2.775238195254526e-05,
+      "loss": 0.7169,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5228070175438596,
+      "grad_norm": 0.690435470611476,
+      "learning_rate": 2.7665355672215824e-05,
+      "loss": 0.7402,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5239766081871345,
+      "grad_norm": 0.7138742813319091,
+      "learning_rate": 2.757832691792834e-05,
+      "loss": 0.7374,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5251461988304094,
+      "grad_norm": 0.9702115835714958,
+      "learning_rate": 2.7491296991757804e-05,
+      "loss": 0.7251,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.5558673963991443,
+      "learning_rate": 2.7404267195796752e-05,
+      "loss": 0.7293,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5274853801169591,
+      "grad_norm": 0.852336833207038,
+      "learning_rate": 2.7317238832135783e-05,
+      "loss": 0.7434,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5286549707602339,
+      "grad_norm": 0.7594137186423496,
+      "learning_rate": 2.723021320284404e-05,
+      "loss": 0.7215,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5298245614035088,
+      "grad_norm": 0.6842736411453925,
+      "learning_rate": 2.7143191609949764e-05,
+      "loss": 0.7221,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5309941520467836,
+      "grad_norm": 0.7448450500927407,
+      "learning_rate": 2.705617535542083e-05,
+      "loss": 0.7122,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5321637426900585,
+      "grad_norm": 0.7440720641697212,
+      "learning_rate": 2.6969165741145213e-05,
+      "loss": 0.7248,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.7031441607842945,
+      "learning_rate": 2.6882164068911554e-05,
+      "loss": 0.7308,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5345029239766081,
+      "grad_norm": 0.7425827882634033,
+      "learning_rate": 2.6795171640389673e-05,
+      "loss": 0.724,
+      "step": 2285
+    },
+    {
+      "epoch": 0.5356725146198831,
+      "grad_norm": 0.8104245018702781,
+      "learning_rate": 2.670818975711107e-05,
+      "loss": 0.7185,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5368421052631579,
+      "grad_norm": 0.7033113831122282,
+      "learning_rate": 2.66212197204495e-05,
+      "loss": 0.731,
+      "step": 2295
+    },
+    {
+      "epoch": 0.5380116959064327,
+      "grad_norm": 0.6838589814643976,
+      "learning_rate": 2.6534262831601464e-05,
+      "loss": 0.721,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5391812865497077,
+      "grad_norm": 0.712100800360047,
+      "learning_rate": 2.6447320391566738e-05,
+      "loss": 0.7245,
+      "step": 2305
+    },
+    {
+      "epoch": 0.5403508771929825,
+      "grad_norm": 0.6723092101148208,
+      "learning_rate": 2.6360393701128968e-05,
+      "loss": 0.7247,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5415204678362573,
+      "grad_norm": 0.6805561969651415,
+      "learning_rate": 2.6273484060836113e-05,
+      "loss": 0.7222,
+      "step": 2315
+    },
+    {
+      "epoch": 0.5426900584795321,
+      "grad_norm": 0.6306098328964164,
+      "learning_rate": 2.618659277098105e-05,
+      "loss": 0.7102,
+      "step": 2320
+    },
+    {
+      "epoch": 0.543859649122807,
+      "grad_norm": 0.8205802441393052,
+      "learning_rate": 2.6099721131582134e-05,
+      "loss": 0.7242,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5450292397660819,
+      "grad_norm": 0.6949882040032072,
+      "learning_rate": 2.6012870442363686e-05,
+      "loss": 0.7311,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5461988304093567,
+      "grad_norm": 0.6981026562322318,
+      "learning_rate": 2.592604200273661e-05,
+      "loss": 0.7251,
+      "step": 2335
+    },
+    {
+      "epoch": 0.5473684210526316,
+      "grad_norm": 0.7024702109216225,
+      "learning_rate": 2.583923711177891e-05,
+      "loss": 0.7246,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5485380116959064,
+      "grad_norm": 0.6977023494940424,
+      "learning_rate": 2.5752457068216256e-05,
+      "loss": 0.7219,
+      "step": 2345
+    },
+    {
+      "epoch": 0.5497076023391813,
+      "grad_norm": 0.6577178559620495,
+      "learning_rate": 2.56657031704026e-05,
+      "loss": 0.716,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5508771929824562,
+      "grad_norm": 0.6876417494610477,
+      "learning_rate": 2.557897671630069e-05,
+      "loss": 0.7305,
+      "step": 2355
+    },
+    {
+      "epoch": 0.552046783625731,
+      "grad_norm": 0.705663481497826,
+      "learning_rate": 2.549227900346267e-05,
+      "loss": 0.7264,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5532163742690058,
+      "grad_norm": 0.6381128045940445,
+      "learning_rate": 2.5405611329010703e-05,
+      "loss": 0.7194,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5543859649122806,
+      "grad_norm": 0.7563891767825265,
+      "learning_rate": 2.53189749896175e-05,
+      "loss": 0.7194,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.773436863349914,
+      "learning_rate": 2.5232371281487e-05,
+      "loss": 0.7328,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5567251461988304,
+      "grad_norm": 0.7188605088844063,
+      "learning_rate": 2.514580150033487e-05,
+      "loss": 0.7084,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5578947368421052,
+      "grad_norm": 0.7060166679772149,
+      "learning_rate": 2.5059266941369235e-05,
+      "loss": 0.7298,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5590643274853802,
+      "grad_norm": 0.7085718046130549,
+      "learning_rate": 2.4972768899271216e-05,
+      "loss": 0.739,
+      "step": 2390
+    },
+    {
+      "epoch": 0.560233918128655,
+      "grad_norm": 0.7549743099712579,
+      "learning_rate": 2.4886308668175613e-05,
+      "loss": 0.7265,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5614035087719298,
+      "grad_norm": 0.6538332655299417,
+      "learning_rate": 2.479988754165148e-05,
+      "loss": 0.7206,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5625730994152047,
+      "grad_norm": 0.7162742971948052,
+      "learning_rate": 2.4713506812682864e-05,
+      "loss": 0.7241,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5637426900584795,
+      "grad_norm": 0.644979711634962,
+      "learning_rate": 2.4627167773649347e-05,
+      "loss": 0.7037,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5649122807017544,
+      "grad_norm": 0.6988424837088039,
+      "learning_rate": 2.454087171630683e-05,
+      "loss": 0.7311,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5660818713450292,
+      "grad_norm": 0.6592590124650294,
+      "learning_rate": 2.445461993176809e-05,
+      "loss": 0.7136,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5672514619883041,
+      "grad_norm": 0.7241992097338749,
+      "learning_rate": 2.4368413710483563e-05,
+      "loss": 0.7085,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5684210526315789,
+      "grad_norm": 0.7177458102026997,
+      "learning_rate": 2.4282254342221972e-05,
+      "loss": 0.716,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5695906432748538,
+      "grad_norm": 0.7014993276429681,
+      "learning_rate": 2.419614311605106e-05,
+      "loss": 0.7235,
+      "step": 2435
+    },
+    {
+      "epoch": 0.5707602339181287,
+      "grad_norm": 0.6974967208078983,
+      "learning_rate": 2.411008132031827e-05,
+      "loss": 0.7176,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5719298245614035,
+      "grad_norm": 0.6428857393584856,
+      "learning_rate": 2.402407024263155e-05,
+      "loss": 0.7207,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5730994152046783,
+      "grad_norm": 0.6587656996704534,
+      "learning_rate": 2.3938111169839983e-05,
+      "loss": 0.7218,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5742690058479533,
+      "grad_norm": 0.6511884593126578,
+      "learning_rate": 2.3852205388014587e-05,
+      "loss": 0.7215,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5754385964912281,
+      "grad_norm": 0.7305618449614039,
+      "learning_rate": 2.3766354182429102e-05,
+      "loss": 0.7289,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5766081871345029,
+      "grad_norm": 0.6734495634477441,
+      "learning_rate": 2.3680558837540696e-05,
+      "loss": 0.7209,
+      "step": 2465
+    },
+    {
+      "epoch": 0.5777777777777777,
+      "grad_norm": 0.693329829649744,
+      "learning_rate": 2.359482063697081e-05,
+      "loss": 0.7085,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.6572826509123701,
+      "learning_rate": 2.3509140863485913e-05,
+      "loss": 0.7072,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5801169590643275,
+      "grad_norm": 0.7132176217171703,
+      "learning_rate": 2.34235207989783e-05,
+      "loss": 0.7197,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5812865497076023,
+      "grad_norm": 0.7471511475477799,
+      "learning_rate": 2.3337961724446967e-05,
+      "loss": 0.7143,
+      "step": 2485
+    },
+    {
+      "epoch": 0.5824561403508772,
+      "grad_norm": 0.7569514713011818,
+      "learning_rate": 2.3252464919978394e-05,
+      "loss": 0.7197,
+      "step": 2490
+    },
+    {
+      "epoch": 0.583625730994152,
+      "grad_norm": 0.672678541624009,
+      "learning_rate": 2.3167031664727406e-05,
+      "loss": 0.7131,
+      "step": 2495
+    },
+    {
+      "epoch": 0.5847953216374269,
+      "grad_norm": 0.7138771698681281,
+      "learning_rate": 2.3081663236898065e-05,
+      "loss": 0.715,
+      "step": 2500
+    },
+    {
+      "epoch": 0.5859649122807018,
+      "grad_norm": 0.6632737853478047,
+      "learning_rate": 2.299636091372449e-05,
+      "loss": 0.716,
+      "step": 2505
+    },
+    {
+      "epoch": 0.5871345029239766,
+      "grad_norm": 0.646368439045504,
+      "learning_rate": 2.2911125971451814e-05,
+      "loss": 0.7105,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5883040935672514,
+      "grad_norm": 0.7502430275200059,
+      "learning_rate": 2.2825959685317026e-05,
+      "loss": 0.7018,
+      "step": 2515
+    },
+    {
+      "epoch": 0.5894736842105263,
+      "grad_norm": 0.6686250670322178,
+      "learning_rate": 2.274086332952993e-05,
+      "loss": 0.7142,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5906432748538012,
+      "grad_norm": 0.6713488705252936,
+      "learning_rate": 2.2655838177254084e-05,
+      "loss": 0.7096,
+      "step": 2525
+    },
+    {
+      "epoch": 0.591812865497076,
+      "grad_norm": 0.74066155087446,
+      "learning_rate": 2.2570885500587724e-05,
+      "loss": 0.7161,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5929824561403508,
+      "grad_norm": 0.7203926540322918,
+      "learning_rate": 2.248600657054474e-05,
+      "loss": 0.7169,
+      "step": 2535
+    },
+    {
+      "epoch": 0.5941520467836258,
+      "grad_norm": 0.7338466015320976,
+      "learning_rate": 2.2401202657035695e-05,
+      "loss": 0.7333,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5953216374269006,
+      "grad_norm": 0.6914730478070041,
+      "learning_rate": 2.231647502884877e-05,
+      "loss": 0.7093,
+      "step": 2545
+    },
+    {
+      "epoch": 0.5964912280701754,
+      "grad_norm": 0.7190763244458849,
+      "learning_rate": 2.2231824953630826e-05,
+      "loss": 0.7194,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5976608187134503,
+      "grad_norm": 0.7443168685764103,
+      "learning_rate": 2.2147253697868404e-05,
+      "loss": 0.7148,
+      "step": 2555
+    },
+    {
+      "epoch": 0.5988304093567252,
+      "grad_norm": 0.6671658986596564,
+      "learning_rate": 2.2062762526868802e-05,
+      "loss": 0.7106,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9259171966978712,
+      "learning_rate": 2.1978352704741144e-05,
+      "loss": 0.7091,
+      "step": 2565
+    },
+    {
+      "epoch": 0.6011695906432749,
+      "grad_norm": 0.647797375674224,
+      "learning_rate": 2.189402549437745e-05,
+      "loss": 0.6978,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6023391812865497,
+      "grad_norm": 0.6738560804966286,
+      "learning_rate": 2.1809782157433738e-05,
+      "loss": 0.7093,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6035087719298246,
+      "grad_norm": 0.6388985088846765,
+      "learning_rate": 2.172562395431118e-05,
+      "loss": 0.7045,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6046783625730994,
+      "grad_norm": 0.628701446161192,
+      "learning_rate": 2.1641552144137206e-05,
+      "loss": 0.7085,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6058479532163743,
+      "grad_norm": 0.671782052313191,
+      "learning_rate": 2.1557567984746696e-05,
+      "loss": 0.7209,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6070175438596491,
+      "grad_norm": 0.6644467772573169,
+      "learning_rate": 2.147367273266314e-05,
+      "loss": 0.7205,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6081871345029239,
+      "grad_norm": 0.6954352839860809,
+      "learning_rate": 2.1389867643079848e-05,
+      "loss": 0.7204,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6093567251461989,
+      "grad_norm": 0.6959121822967036,
+      "learning_rate": 2.1306153969841192e-05,
+      "loss": 0.7214,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6105263157894737,
+      "grad_norm": 0.6785383454163235,
+      "learning_rate": 2.1222532965423792e-05,
+      "loss": 0.7071,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6116959064327485,
+      "grad_norm": 0.7766968741058371,
+      "learning_rate": 2.1139005880917805e-05,
+      "loss": 0.7117,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6128654970760234,
+      "grad_norm": 0.6514576309987281,
+      "learning_rate": 2.1055573966008264e-05,
+      "loss": 0.7241,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6140350877192983,
+      "grad_norm": 0.7024878214812843,
+      "learning_rate": 2.0972238468956267e-05,
+      "loss": 0.7149,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6152046783625731,
+      "grad_norm": 0.7189809009653396,
+      "learning_rate": 2.0889000636580398e-05,
+      "loss": 0.698,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6163742690058479,
+      "grad_norm": 0.681435970631106,
+      "learning_rate": 2.080586171423803e-05,
+      "loss": 0.7188,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6175438596491228,
+      "grad_norm": 0.6766943134449012,
+      "learning_rate": 2.0722822945806697e-05,
+      "loss": 0.7073,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6187134502923977,
+      "grad_norm": 0.6479985354365313,
+      "learning_rate": 2.063988557366548e-05,
+      "loss": 0.7145,
+      "step": 2645
+    },
+    {
+      "epoch": 0.6198830409356725,
+      "grad_norm": 0.749913572914423,
+      "learning_rate": 2.0557050838676445e-05,
+      "loss": 0.7132,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6210526315789474,
+      "grad_norm": 0.6054029968957726,
+      "learning_rate": 2.047431998016604e-05,
+      "loss": 0.7085,
+      "step": 2655
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 0.6625793556031936,
+      "learning_rate": 2.0391694235906594e-05,
+      "loss": 0.7207,
+      "step": 2660
+    },
+    {
+      "epoch": 0.623391812865497,
+      "grad_norm": 0.6265379430575465,
+      "learning_rate": 2.0309174842097755e-05,
+      "loss": 0.7193,
+      "step": 2665
+    },
+    {
+      "epoch": 0.624561403508772,
+      "grad_norm": 0.7859616347118907,
+      "learning_rate": 2.0226763033348005e-05,
+      "loss": 0.7123,
+      "step": 2670
+    },
+    {
+      "epoch": 0.6257309941520468,
+      "grad_norm": 0.6856709481316304,
+      "learning_rate": 2.0144460042656244e-05,
+      "loss": 0.7056,
+      "step": 2675
+    },
+    {
+      "epoch": 0.6269005847953216,
+      "grad_norm": 0.6640584443200321,
+      "learning_rate": 2.0062267101393255e-05,
+      "loss": 0.7096,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6280701754385964,
+      "grad_norm": 0.7024631624759632,
+      "learning_rate": 1.9980185439283343e-05,
+      "loss": 0.7144,
+      "step": 2685
+    },
+    {
+      "epoch": 0.6292397660818714,
+      "grad_norm": 0.6513218786027092,
+      "learning_rate": 1.9898216284385924e-05,
+      "loss": 0.7207,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6304093567251462,
+      "grad_norm": 0.6736080868317267,
+      "learning_rate": 1.9816360863077106e-05,
+      "loss": 0.7134,
+      "step": 2695
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.7136228459371672,
+      "learning_rate": 1.973462040003144e-05,
+      "loss": 0.6967,
+      "step": 2700
+    },
+    {
+      "epoch": 0.632748538011696,
+      "grad_norm": 0.6503816533129796,
+      "learning_rate": 1.9652996118203487e-05,
+      "loss": 0.7028,
+      "step": 2705
+    },
+    {
+      "epoch": 0.6339181286549708,
+      "grad_norm": 0.6716498502463782,
+      "learning_rate": 1.9571489238809586e-05,
+      "loss": 0.6952,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6350877192982456,
+      "grad_norm": 0.702194650319827,
+      "learning_rate": 1.949010098130958e-05,
+      "loss": 0.7117,
+      "step": 2715
+    },
+    {
+      "epoch": 0.6362573099415205,
+      "grad_norm": 0.7035074020264389,
+      "learning_rate": 1.940883256338854e-05,
+      "loss": 0.7086,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6374269005847953,
+      "grad_norm": 0.6724554904596103,
+      "learning_rate": 1.9327685200938567e-05,
+      "loss": 0.6959,
+      "step": 2725
+    },
+    {
+      "epoch": 0.6385964912280702,
+      "grad_norm": 0.6703506374383876,
+      "learning_rate": 1.9246660108040615e-05,
+      "loss": 0.7074,
+      "step": 2730
+    },
+    {
+      "epoch": 0.639766081871345,
+      "grad_norm": 0.7136435424759008,
+      "learning_rate": 1.9165758496946296e-05,
+      "loss": 0.6949,
+      "step": 2735
+    },
+    {
+      "epoch": 0.6409356725146199,
+      "grad_norm": 0.6811443394345809,
+      "learning_rate": 1.9084981578059745e-05,
+      "loss": 0.7249,
+      "step": 2740
+    },
+    {
+      "epoch": 0.6421052631578947,
+      "grad_norm": 0.7047496550502742,
+      "learning_rate": 1.900433055991956e-05,
+      "loss": 0.6977,
+      "step": 2745
+    },
+    {
+      "epoch": 0.6432748538011696,
+      "grad_norm": 0.6693712668333812,
+      "learning_rate": 1.8923806649180636e-05,
+      "loss": 0.7042,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6444444444444445,
+      "grad_norm": 0.7062577866647437,
+      "learning_rate": 1.8843411050596215e-05,
+      "loss": 0.7255,
+      "step": 2755
+    },
+    {
+      "epoch": 0.6456140350877193,
+      "grad_norm": 0.7457329699470406,
+      "learning_rate": 1.8763144966999742e-05,
+      "loss": 0.7131,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6467836257309941,
+      "grad_norm": 0.6949977616539524,
+      "learning_rate": 1.8683009599286976e-05,
+      "loss": 0.7032,
+      "step": 2765
+    },
+    {
+      "epoch": 0.6479532163742691,
+      "grad_norm": 0.6446049989679632,
+      "learning_rate": 1.8603006146397984e-05,
+      "loss": 0.7045,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6491228070175439,
+      "grad_norm": 0.6868201452976932,
+      "learning_rate": 1.8523135805299164e-05,
+      "loss": 0.7078,
+      "step": 2775
+    },
+    {
+      "epoch": 0.6502923976608187,
+      "grad_norm": 0.6549013173661192,
+      "learning_rate": 1.8443399770965368e-05,
+      "loss": 0.7034,
+      "step": 2780
+    },
+    {
+      "epoch": 0.6514619883040935,
+      "grad_norm": 0.6747132830101538,
+      "learning_rate": 1.836379923636209e-05,
+      "loss": 0.7047,
+      "step": 2785
+    },
+    {
+      "epoch": 0.6526315789473685,
+      "grad_norm": 0.6189077937045445,
+      "learning_rate": 1.8284335392427464e-05,
+      "loss": 0.7065,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6538011695906433,
+      "grad_norm": 0.6712899258383389,
+      "learning_rate": 1.8205009428054616e-05,
+      "loss": 0.7226,
+      "step": 2795
+    },
+    {
+      "epoch": 0.6549707602339181,
+      "grad_norm": 0.6476379586448503,
+      "learning_rate": 1.812582253007375e-05,
+      "loss": 0.7098,
+      "step": 2800
+    },
+    {
+      "epoch": 0.656140350877193,
+      "grad_norm": 0.6632887980494508,
+      "learning_rate": 1.804677588323443e-05,
+      "loss": 0.6931,
+      "step": 2805
+    },
+    {
+      "epoch": 0.6573099415204678,
+      "grad_norm": 0.6676152770497438,
+      "learning_rate": 1.7967870670187903e-05,
+      "loss": 0.7026,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6584795321637427,
+      "grad_norm": 0.6249710453695605,
+      "learning_rate": 1.7889108071469323e-05,
+      "loss": 0.7122,
+      "step": 2815
+    },
+    {
+      "epoch": 0.6596491228070176,
+      "grad_norm": 0.6910085759118031,
+      "learning_rate": 1.781048926548016e-05,
+      "loss": 0.6948,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6608187134502924,
+      "grad_norm": 0.7039245820842112,
+      "learning_rate": 1.7732015428470522e-05,
+      "loss": 0.7057,
+      "step": 2825
+    },
+    {
+      "epoch": 0.6619883040935672,
+      "grad_norm": 0.6560416507502473,
+      "learning_rate": 1.7653687734521572e-05,
+      "loss": 0.7162,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6631578947368421,
+      "grad_norm": 0.6369401839904856,
+      "learning_rate": 1.7575507355527965e-05,
+      "loss": 0.7022,
+      "step": 2835
+    },
+    {
+      "epoch": 0.664327485380117,
+      "grad_norm": 0.7618791478324256,
+      "learning_rate": 1.7497475461180324e-05,
+      "loss": 0.6871,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6654970760233918,
+      "grad_norm": 0.7035877403668513,
+      "learning_rate": 1.7419593218947706e-05,
+      "loss": 0.7037,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 3.171993920451794,
+      "learning_rate": 1.734186179406019e-05,
+      "loss": 0.7086,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6678362573099416,
+      "grad_norm": 0.7752177765101442,
+      "learning_rate": 1.7264282349491382e-05,
+      "loss": 0.7163,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6690058479532164,
+      "grad_norm": 0.6730042085183996,
+      "learning_rate": 1.7186856045941044e-05,
+      "loss": 0.7016,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6701754385964912,
+      "grad_norm": 0.6538520154513303,
+      "learning_rate": 1.7109584041817765e-05,
+      "loss": 0.6934,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6713450292397661,
+      "grad_norm": 0.702951285614432,
+      "learning_rate": 1.7032467493221556e-05,
+      "loss": 0.704,
+      "step": 2870
+    },
+    {
+      "epoch": 0.672514619883041,
+      "grad_norm": 0.7222284007051577,
+      "learning_rate": 1.6955507553926584e-05,
+      "loss": 0.6891,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6736842105263158,
+      "grad_norm": 0.655031348765028,
+      "learning_rate": 1.6878705375363964e-05,
+      "loss": 0.6959,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6748538011695906,
+      "grad_norm": 0.650372228862769,
+      "learning_rate": 1.6802062106604435e-05,
+      "loss": 0.7037,
+      "step": 2885
+    },
+    {
+      "epoch": 0.6760233918128655,
+      "grad_norm": 0.6648549632114491,
+      "learning_rate": 1.6725578894341253e-05,
+      "loss": 0.6955,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6771929824561403,
+      "grad_norm": 0.6717090275475817,
+      "learning_rate": 1.664925688287297e-05,
+      "loss": 0.7103,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6783625730994152,
+      "grad_norm": 0.6499960801937501,
+      "learning_rate": 1.657309721408636e-05,
+      "loss": 0.7011,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6795321637426901,
+      "grad_norm": 0.6839419015653185,
+      "learning_rate": 1.649710102743931e-05,
+      "loss": 0.6903,
+      "step": 2905
+    },
+    {
+      "epoch": 0.6807017543859649,
+      "grad_norm": 0.6488668540739176,
+      "learning_rate": 1.64212694599438e-05,
+      "loss": 0.705,
+      "step": 2910
+    },
+    {
+      "epoch": 0.6818713450292397,
+      "grad_norm": 0.6715390084742834,
+      "learning_rate": 1.634560364614883e-05,
+      "loss": 0.6983,
+      "step": 2915
+    },
+    {
+      "epoch": 0.6830409356725147,
+      "grad_norm": 0.7229792154843714,
+      "learning_rate": 1.6270104718123535e-05,
+      "loss": 0.7059,
+      "step": 2920
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.7217490768423835,
+      "learning_rate": 1.6194773805440166e-05,
+      "loss": 0.709,
+      "step": 2925
+    },
+    {
+      "epoch": 0.6853801169590643,
+      "grad_norm": 0.7323151054068626,
+      "learning_rate": 1.6119612035157227e-05,
+      "loss": 0.7249,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6865497076023391,
+      "grad_norm": 0.693099764226788,
+      "learning_rate": 1.604462053180263e-05,
+      "loss": 0.6886,
+      "step": 2935
+    },
+    {
+      "epoch": 0.6877192982456141,
+      "grad_norm": 0.7447053266775585,
+      "learning_rate": 1.5969800417356817e-05,
+      "loss": 0.6943,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6888888888888889,
+      "grad_norm": 0.7056733168103325,
+      "learning_rate": 1.5895152811236046e-05,
+      "loss": 0.6987,
+      "step": 2945
+    },
+    {
+      "epoch": 0.6900584795321637,
+      "grad_norm": 0.6333643279860763,
+      "learning_rate": 1.582067883027557e-05,
+      "loss": 0.6834,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6912280701754386,
+      "grad_norm": 0.6657715044830338,
+      "learning_rate": 1.574637958871297e-05,
+      "loss": 0.6901,
+      "step": 2955
+    },
+    {
+      "epoch": 0.6923976608187135,
+      "grad_norm": 0.6305769577184307,
+      "learning_rate": 1.567225619817148e-05,
+      "loss": 0.6985,
+      "step": 2960
+    },
+    {
+      "epoch": 0.6935672514619883,
+      "grad_norm": 0.6912315845824749,
+      "learning_rate": 1.5598309767643355e-05,
+      "loss": 0.6909,
+      "step": 2965
+    },
+    {
+      "epoch": 0.6947368421052632,
+      "grad_norm": 0.6874615844912404,
+      "learning_rate": 1.5524541403473244e-05,
+      "loss": 0.7061,
+      "step": 2970
+    },
+    {
+      "epoch": 0.695906432748538,
+      "grad_norm": 0.6300180071584462,
+      "learning_rate": 1.5450952209341717e-05,
+      "loss": 0.6993,
+      "step": 2975
+    },
+    {
+      "epoch": 0.6970760233918128,
+      "grad_norm": 0.6805832683256726,
+      "learning_rate": 1.5377543286248653e-05,
+      "loss": 0.7066,
+      "step": 2980
+    },
+    {
+      "epoch": 0.6982456140350877,
+      "grad_norm": 0.6492844517838456,
+      "learning_rate": 1.5304315732496867e-05,
+      "loss": 0.7011,
+      "step": 2985
+    },
+    {
+      "epoch": 0.6994152046783626,
+      "grad_norm": 0.6412533888420374,
+      "learning_rate": 1.5231270643675577e-05,
+      "loss": 0.7033,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7005847953216374,
+      "grad_norm": 0.6316222267104177,
+      "learning_rate": 1.5158409112644103e-05,
+      "loss": 0.7052,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7017543859649122,
+      "grad_norm": 0.7274584018785283,
+      "learning_rate": 1.5085732229515476e-05,
+      "loss": 0.7037,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7029239766081872,
+      "grad_norm": 0.6553959454076399,
+      "learning_rate": 1.5013241081640101e-05,
+      "loss": 0.6985,
+      "step": 3005
+    },
+    {
+      "epoch": 0.704093567251462,
+      "grad_norm": 0.6640325261209403,
+      "learning_rate": 1.4940936753589533e-05,
+      "loss": 0.7042,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7052631578947368,
+      "grad_norm": 0.7105270721109682,
+      "learning_rate": 1.4868820327140249e-05,
+      "loss": 0.691,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7064327485380117,
+      "grad_norm": 0.703378689555471,
+      "learning_rate": 1.479689288125742e-05,
+      "loss": 0.6955,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7076023391812866,
+      "grad_norm": 0.692679701776449,
+      "learning_rate": 1.4725155492078813e-05,
+      "loss": 0.6876,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7087719298245614,
+      "grad_norm": 0.7203262346843042,
+      "learning_rate": 1.4653609232898684e-05,
+      "loss": 0.6972,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7099415204678362,
+      "grad_norm": 0.7433252096654863,
+      "learning_rate": 1.4582255174151683e-05,
+      "loss": 0.6978,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 0.6965363631959524,
+      "learning_rate": 1.45110943833969e-05,
+      "loss": 0.705,
+      "step": 3040
+    },
+    {
+      "epoch": 0.712280701754386,
+      "grad_norm": 0.6282817213921031,
+      "learning_rate": 1.4440127925301827e-05,
+      "loss": 0.6826,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7134502923976608,
+      "grad_norm": 0.6974830082717185,
+      "learning_rate": 1.4369356861626467e-05,
+      "loss": 0.6962,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7146198830409357,
+      "grad_norm": 0.6601146384075475,
+      "learning_rate": 1.4298782251207468e-05,
+      "loss": 0.6906,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7157894736842105,
+      "grad_norm": 0.6814736421646006,
+      "learning_rate": 1.4228405149942226e-05,
+      "loss": 0.6911,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7169590643274854,
+      "grad_norm": 2.666703566903681,
+      "learning_rate": 1.4158226610773117e-05,
+      "loss": 0.7079,
+      "step": 3065
+    },
+    {
+      "epoch": 0.7181286549707603,
+      "grad_norm": 0.7144980012945954,
+      "learning_rate": 1.4088247683671768e-05,
+      "loss": 0.6949,
+      "step": 3070
+    },
+    {
+      "epoch": 0.7192982456140351,
+      "grad_norm": 0.66028336916713,
+      "learning_rate": 1.4018469415623309e-05,
+      "loss": 0.6844,
+      "step": 3075
+    },
+    {
+      "epoch": 0.7204678362573099,
+      "grad_norm": 0.7116011959517159,
+      "learning_rate": 1.3948892850610709e-05,
+      "loss": 0.6944,
+      "step": 3080
+    },
+    {
+      "epoch": 0.7216374269005847,
+      "grad_norm": 0.719826020172388,
+      "learning_rate": 1.3879519029599197e-05,
+      "loss": 0.6855,
+      "step": 3085
+    },
+    {
+      "epoch": 0.7228070175438597,
+      "grad_norm": 0.646119503804467,
+      "learning_rate": 1.3810348990520635e-05,
+      "loss": 0.698,
+      "step": 3090
+    },
+    {
+      "epoch": 0.7239766081871345,
+      "grad_norm": 0.6590521039258358,
+      "learning_rate": 1.3741383768258043e-05,
+      "loss": 0.7111,
+      "step": 3095
+    },
+    {
+      "epoch": 0.7251461988304093,
+      "grad_norm": 0.6618851155256175,
+      "learning_rate": 1.3672624394630062e-05,
+      "loss": 0.694,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7263157894736842,
+      "grad_norm": 0.6259226479372295,
+      "learning_rate": 1.3604071898375548e-05,
+      "loss": 0.6899,
+      "step": 3105
+    },
+    {
+      "epoch": 0.7274853801169591,
+      "grad_norm": 0.6544836178913599,
+      "learning_rate": 1.3535727305138185e-05,
+      "loss": 0.6802,
+      "step": 3110
+    },
+    {
+      "epoch": 0.7286549707602339,
+      "grad_norm": 0.6291569958373459,
+      "learning_rate": 1.3467591637451126e-05,
+      "loss": 0.6929,
+      "step": 3115
+    },
+    {
+      "epoch": 0.7298245614035088,
+      "grad_norm": 0.6512874151441448,
+      "learning_rate": 1.3399665914721682e-05,
+      "loss": 0.6873,
+      "step": 3120
+    },
+    {
+      "epoch": 0.7309941520467836,
+      "grad_norm": 0.6448721183547438,
+      "learning_rate": 1.3331951153216115e-05,
+      "loss": 0.6929,
+      "step": 3125
+    },
+    {
+      "epoch": 0.7321637426900585,
+      "grad_norm": 0.6512033452820375,
+      "learning_rate": 1.326444836604438e-05,
+      "loss": 0.6978,
+      "step": 3130
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 0.6699223613337226,
+      "learning_rate": 1.3197158563145013e-05,
+      "loss": 0.6878,
+      "step": 3135
+    },
+    {
+      "epoch": 0.7345029239766082,
+      "grad_norm": 0.6785655173965759,
+      "learning_rate": 1.3130082751269973e-05,
+      "loss": 0.7074,
+      "step": 3140
+    },
+    {
+      "epoch": 0.735672514619883,
+      "grad_norm": 0.6893282702509805,
+      "learning_rate": 1.3063221933969627e-05,
+      "loss": 0.6884,
+      "step": 3145
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.6801941856058638,
+      "learning_rate": 1.2996577111577714e-05,
+      "loss": 0.6956,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7380116959064328,
+      "grad_norm": 0.7429147439910412,
+      "learning_rate": 1.2930149281196366e-05,
+      "loss": 0.6935,
+      "step": 3155
+    },
+    {
+      "epoch": 0.7391812865497076,
+      "grad_norm": 0.7584692264067732,
+      "learning_rate": 1.2863939436681211e-05,
+      "loss": 0.7084,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7403508771929824,
+      "grad_norm": 0.6496501154726846,
+      "learning_rate": 1.2797948568626514e-05,
+      "loss": 0.6811,
+      "step": 3165
+    },
+    {
+      "epoch": 0.7415204678362574,
+      "grad_norm": 0.6947826629815981,
+      "learning_rate": 1.2732177664350297e-05,
+      "loss": 0.7068,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7426900584795322,
+      "grad_norm": 0.6322459875062904,
+      "learning_rate": 1.266662770787965e-05,
+      "loss": 0.6771,
+      "step": 3175
+    },
+    {
+      "epoch": 0.743859649122807,
+      "grad_norm": 0.6610295189441215,
+      "learning_rate": 1.2601299679935944e-05,
+      "loss": 0.6976,
+      "step": 3180
+    },
+    {
+      "epoch": 0.7450292397660818,
+      "grad_norm": 0.6090430443039708,
+      "learning_rate": 1.2536194557920173e-05,
+      "loss": 0.6877,
+      "step": 3185
+    },
+    {
+      "epoch": 0.7461988304093568,
+      "grad_norm": 0.657566522164305,
+      "learning_rate": 1.2471313315898369e-05,
+      "loss": 0.6919,
+      "step": 3190
+    },
+    {
+      "epoch": 0.7473684210526316,
+      "grad_norm": 0.6442799866282577,
+      "learning_rate": 1.2406656924586971e-05,
+      "loss": 0.7097,
+      "step": 3195
+    },
+    {
+      "epoch": 0.7485380116959064,
+      "grad_norm": 0.6517637658675814,
+      "learning_rate": 1.2342226351338333e-05,
+      "loss": 0.6964,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7497076023391813,
+      "grad_norm": 0.6298193875835328,
+      "learning_rate": 1.227802256012627e-05,
+      "loss": 0.7067,
+      "step": 3205
+    },
+    {
+      "epoch": 0.7508771929824561,
+      "grad_norm": 0.6400435584951044,
+      "learning_rate": 1.2214046511531579e-05,
+      "loss": 0.6996,
+      "step": 3210
+    },
+    {
+      "epoch": 0.752046783625731,
+      "grad_norm": 0.6682447845425866,
+      "learning_rate": 1.215029916272771e-05,
+      "loss": 0.7052,
+      "step": 3215
+    },
+    {
+      "epoch": 0.7532163742690059,
+      "grad_norm": 0.6345008179067555,
+      "learning_rate": 1.2086781467466466e-05,
+      "loss": 0.6815,
+      "step": 3220
+    },
+    {
+      "epoch": 0.7543859649122807,
+      "grad_norm": 0.6269106487695227,
+      "learning_rate": 1.2023494376063655e-05,
+      "loss": 0.6889,
+      "step": 3225
+    },
+    {
+      "epoch": 0.7555555555555555,
+      "grad_norm": 0.6766539926521694,
+      "learning_rate": 1.196043883538496e-05,
+      "loss": 0.6896,
+      "step": 3230
+    },
+    {
+      "epoch": 0.7567251461988304,
+      "grad_norm": 0.657654184164444,
+      "learning_rate": 1.1897615788831715e-05,
+      "loss": 0.697,
+      "step": 3235
+    },
+    {
+      "epoch": 0.7578947368421053,
+      "grad_norm": 0.6479759415667077,
+      "learning_rate": 1.1835026176326817e-05,
+      "loss": 0.7061,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7590643274853801,
+      "grad_norm": 0.6454453155585288,
+      "learning_rate": 1.1772670934300637e-05,
+      "loss": 0.6944,
+      "step": 3245
+    },
+    {
+      "epoch": 0.7602339181286549,
+      "grad_norm": 0.6717289664011613,
+      "learning_rate": 1.171055099567705e-05,
+      "loss": 0.6975,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7614035087719299,
+      "grad_norm": 0.6446562340482254,
+      "learning_rate": 1.164866728985944e-05,
+      "loss": 0.6868,
+      "step": 3255
+    },
+    {
+      "epoch": 0.7625730994152047,
+      "grad_norm": 0.6842764077833369,
+      "learning_rate": 1.1587020742716822e-05,
+      "loss": 0.6902,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7637426900584795,
+      "grad_norm": 0.6529175563359729,
+      "learning_rate": 1.1525612276569954e-05,
+      "loss": 0.686,
+      "step": 3265
+    },
+    {
+      "epoch": 0.7649122807017544,
+      "grad_norm": 0.6878863831926055,
+      "learning_rate": 1.1464442810177591e-05,
+      "loss": 0.6847,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7660818713450293,
+      "grad_norm": 0.634988623157636,
+      "learning_rate": 1.1403513258722689e-05,
+      "loss": 0.689,
+      "step": 3275
+    },
+    {
+      "epoch": 0.7672514619883041,
+      "grad_norm": 0.6427892453571996,
+      "learning_rate": 1.134282453379873e-05,
+      "loss": 0.6855,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7684210526315789,
+      "grad_norm": 0.6571927205325351,
+      "learning_rate": 1.12823775433961e-05,
+      "loss": 0.689,
+      "step": 3285
+    },
+    {
+      "epoch": 0.7695906432748538,
+      "grad_norm": 0.6553286498653648,
+      "learning_rate": 1.1222173191888482e-05,
+      "loss": 0.701,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7707602339181286,
+      "grad_norm": 0.6477793773584922,
+      "learning_rate": 1.1162212380019327e-05,
+      "loss": 0.6864,
+      "step": 3295
+    },
+    {
+      "epoch": 0.7719298245614035,
+      "grad_norm": 0.677952780272042,
+      "learning_rate": 1.11024960048884e-05,
+      "loss": 0.6795,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7730994152046784,
+      "grad_norm": 0.6183556845058249,
+      "learning_rate": 1.1043024959938327e-05,
+      "loss": 0.6839,
+      "step": 3305
+    },
+    {
+      "epoch": 0.7742690058479532,
+      "grad_norm": 0.6685159928157001,
+      "learning_rate": 1.098380013494124e-05,
+      "loss": 0.6852,
+      "step": 3310
+    },
+    {
+      "epoch": 0.775438596491228,
+      "grad_norm": 0.6119355012203586,
+      "learning_rate": 1.0924822415985483e-05,
+      "loss": 0.6804,
+      "step": 3315
+    },
+    {
+      "epoch": 0.776608187134503,
+      "grad_norm": 0.6311449859816257,
+      "learning_rate": 1.086609268546234e-05,
+      "loss": 0.6725,
+      "step": 3320
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 0.6980417098074275,
+      "learning_rate": 1.0807611822052802e-05,
+      "loss": 0.678,
+      "step": 3325
+    },
+    {
+      "epoch": 0.7789473684210526,
+      "grad_norm": 0.5801631810377051,
+      "learning_rate": 1.0749380700714495e-05,
+      "loss": 0.6763,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7801169590643274,
+      "grad_norm": 0.6589353523991406,
+      "learning_rate": 1.0691400192668502e-05,
+      "loss": 0.7009,
+      "step": 3335
+    },
+    {
+      "epoch": 0.7812865497076024,
+      "grad_norm": 0.6130234769951651,
+      "learning_rate": 1.063367116538641e-05,
+      "loss": 0.6876,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7824561403508772,
+      "grad_norm": 0.609038354087779,
+      "learning_rate": 1.0576194482577268e-05,
+      "loss": 0.6821,
+      "step": 3345
+    },
+    {
+      "epoch": 0.783625730994152,
+      "grad_norm": 0.6075902184182757,
+      "learning_rate": 1.0518971004174691e-05,
+      "loss": 0.6971,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7847953216374269,
+      "grad_norm": 0.6404325058010116,
+      "learning_rate": 1.0462001586324009e-05,
+      "loss": 0.6743,
+      "step": 3355
+    },
+    {
+      "epoch": 0.7859649122807018,
+      "grad_norm": 0.6755723087064721,
+      "learning_rate": 1.0405287081369425e-05,
+      "loss": 0.6946,
+      "step": 3360
+    },
+    {
+      "epoch": 0.7871345029239766,
+      "grad_norm": 0.64889270230153,
+      "learning_rate": 1.034882833784129e-05,
+      "loss": 0.6976,
+      "step": 3365
+    },
+    {
+      "epoch": 0.7883040935672515,
+      "grad_norm": 0.6371962111123631,
+      "learning_rate": 1.0292626200443396e-05,
+      "loss": 0.6785,
+      "step": 3370
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.6338899877498075,
+      "learning_rate": 1.0236681510040328e-05,
+      "loss": 0.6846,
+      "step": 3375
+    },
+    {
+      "epoch": 0.7906432748538011,
+      "grad_norm": 0.6073429500137454,
+      "learning_rate": 1.018099510364491e-05,
+      "loss": 0.6806,
+      "step": 3380
+    },
+    {
+      "epoch": 0.791812865497076,
+      "grad_norm": 0.6901673440168534,
+      "learning_rate": 1.0125567814405661e-05,
+      "loss": 0.6718,
+      "step": 3385
+    },
+    {
+      "epoch": 0.7929824561403509,
+      "grad_norm": 0.6557491619211937,
+      "learning_rate": 1.0070400471594324e-05,
+      "loss": 0.6714,
+      "step": 3390
+    },
+    {
+      "epoch": 0.7941520467836257,
+      "grad_norm": 0.6610369238302171,
+      "learning_rate": 1.0015493900593495e-05,
+      "loss": 0.6859,
+      "step": 3395
+    },
+    {
+      "epoch": 0.7953216374269005,
+      "grad_norm": 0.6249323170970883,
+      "learning_rate": 9.960848922884225e-06,
+      "loss": 0.681,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7964912280701755,
+      "grad_norm": 0.6801586512490546,
+      "learning_rate": 9.906466356033766e-06,
+      "loss": 0.6912,
+      "step": 3405
+    },
+    {
+      "epoch": 0.7976608187134503,
+      "grad_norm": 0.6227084687338438,
+      "learning_rate": 9.852347013683339e-06,
+      "loss": 0.6985,
+      "step": 3410
+    },
+    {
+      "epoch": 0.7988304093567251,
+      "grad_norm": 0.6623667229073018,
+      "learning_rate": 9.79849170553592e-06,
+      "loss": 0.6674,
+      "step": 3415
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.65595305832523,
+      "learning_rate": 9.744901237344183e-06,
+      "loss": 0.709,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8011695906432749,
+      "grad_norm": 0.6675026240893706,
+      "learning_rate": 9.691576410898398e-06,
+      "loss": 0.6693,
+      "step": 3425
+    },
+    {
+      "epoch": 0.8023391812865497,
+      "grad_norm": 0.6930655599750254,
+      "learning_rate": 9.638518024014453e-06,
+      "loss": 0.6883,
+      "step": 3430
+    },
+    {
+      "epoch": 0.8035087719298246,
+      "grad_norm": 0.6733979146956586,
+      "learning_rate": 9.585726870521938e-06,
+      "loss": 0.6892,
+      "step": 3435
+    },
+    {
+      "epoch": 0.8046783625730994,
+      "grad_norm": 0.6450346538264267,
+      "learning_rate": 9.53320374025223e-06,
+      "loss": 0.6862,
+      "step": 3440
+    },
+    {
+      "epoch": 0.8058479532163743,
+      "grad_norm": 0.6637412669868189,
+      "learning_rate": 9.480949419026689e-06,
+      "loss": 0.6829,
+      "step": 3445
+    },
+    {
+      "epoch": 0.8070175438596491,
+      "grad_norm": 0.6846438436543126,
+      "learning_rate": 9.428964688644927e-06,
+      "loss": 0.6852,
+      "step": 3450
+    },
+    {
+      "epoch": 0.808187134502924,
+      "grad_norm": 0.6575226232119401,
+      "learning_rate": 9.377250326873071e-06,
+      "loss": 0.6781,
+      "step": 3455
+    },
+    {
+      "epoch": 0.8093567251461988,
+      "grad_norm": 0.629796602059127,
+      "learning_rate": 9.325807107432164e-06,
+      "loss": 0.6771,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8105263157894737,
+      "grad_norm": 0.5978452605861834,
+      "learning_rate": 9.274635799986554e-06,
+      "loss": 0.6798,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8116959064327486,
+      "grad_norm": 0.6129382768821403,
+      "learning_rate": 9.223737170132394e-06,
+      "loss": 0.6845,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8128654970760234,
+      "grad_norm": 0.5979463163900788,
+      "learning_rate": 9.173111979386215e-06,
+      "loss": 0.6961,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8140350877192982,
+      "grad_norm": 0.6342223790039097,
+      "learning_rate": 9.122760985173471e-06,
+      "loss": 0.6935,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8152046783625732,
+      "grad_norm": 0.6718113091931592,
+      "learning_rate": 9.072684940817275e-06,
+      "loss": 0.6849,
+      "step": 3485
+    },
+    {
+      "epoch": 0.816374269005848,
+      "grad_norm": 0.6081957293761066,
+      "learning_rate": 9.022884595527074e-06,
+      "loss": 0.6813,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8175438596491228,
+      "grad_norm": 0.6766450913418733,
+      "learning_rate": 8.97336069438747e-06,
+      "loss": 0.695,
+      "step": 3495
+    },
+    {
+      "epoch": 0.8187134502923976,
+      "grad_norm": 0.6127087903695808,
+      "learning_rate": 8.92411397834706e-06,
+      "loss": 0.6841,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8198830409356725,
+      "grad_norm": 0.637488440649515,
+      "learning_rate": 8.875145184207363e-06,
+      "loss": 0.6755,
+      "step": 3505
+    },
+    {
+      "epoch": 0.8210526315789474,
+      "grad_norm": 0.6558356891747749,
+      "learning_rate": 8.826455044611775e-06,
+      "loss": 0.6864,
+      "step": 3510
+    },
+    {
+      "epoch": 0.8222222222222222,
+      "grad_norm": 0.6909295666801539,
+      "learning_rate": 8.778044288034635e-06,
+      "loss": 0.6932,
+      "step": 3515
+    },
+    {
+      "epoch": 0.8233918128654971,
+      "grad_norm": 0.6216064629061299,
+      "learning_rate": 8.729913638770295e-06,
+      "loss": 0.6708,
+      "step": 3520
+    },
+    {
+      "epoch": 0.8245614035087719,
+      "grad_norm": 0.6470556870374983,
+      "learning_rate": 8.682063816922312e-06,
+      "loss": 0.6805,
+      "step": 3525
+    },
+    {
+      "epoch": 0.8257309941520468,
+      "grad_norm": 0.6441674769859189,
+      "learning_rate": 8.634495538392659e-06,
+      "loss": 0.6723,
+      "step": 3530
+    },
+    {
+      "epoch": 0.8269005847953217,
+      "grad_norm": 0.6294239666284314,
+      "learning_rate": 8.587209514871018e-06,
+      "loss": 0.6833,
+      "step": 3535
+    },
+    {
+      "epoch": 0.8280701754385965,
+      "grad_norm": 0.649777714204493,
+      "learning_rate": 8.540206453824119e-06,
+      "loss": 0.6818,
+      "step": 3540
+    },
+    {
+      "epoch": 0.8292397660818713,
+      "grad_norm": 0.6581271790880595,
+      "learning_rate": 8.493487058485191e-06,
+      "loss": 0.6907,
+      "step": 3545
+    },
+    {
+      "epoch": 0.8304093567251462,
+      "grad_norm": 0.6316641622404449,
+      "learning_rate": 8.44705202784339e-06,
+      "loss": 0.6752,
+      "step": 3550
+    },
+    {
+      "epoch": 0.8315789473684211,
+      "grad_norm": 0.6299397681956188,
+      "learning_rate": 8.4009020566334e-06,
+      "loss": 0.6763,
+      "step": 3555
+    },
+    {
+      "epoch": 0.8327485380116959,
+      "grad_norm": 0.6428132759357637,
+      "learning_rate": 8.355037835324978e-06,
+      "loss": 0.6911,
+      "step": 3560
+    },
+    {
+      "epoch": 0.8339181286549707,
+      "grad_norm": 0.6492937756009263,
+      "learning_rate": 8.30946005011266e-06,
+      "loss": 0.6768,
+      "step": 3565
+    },
+    {
+      "epoch": 0.8350877192982457,
+      "grad_norm": 0.630015343422525,
+      "learning_rate": 8.264169382905499e-06,
+      "loss": 0.6807,
+      "step": 3570
+    },
+    {
+      "epoch": 0.8362573099415205,
+      "grad_norm": 0.6721711166095163,
+      "learning_rate": 8.219166511316844e-06,
+      "loss": 0.6856,
+      "step": 3575
+    },
+    {
+      "epoch": 0.8374269005847953,
+      "grad_norm": 0.6845672226540863,
+      "learning_rate": 8.174452108654198e-06,
+      "loss": 0.676,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8385964912280702,
+      "grad_norm": 0.610978989043903,
+      "learning_rate": 8.130026843909174e-06,
+      "loss": 0.675,
+      "step": 3585
+    },
+    {
+      "epoch": 0.839766081871345,
+      "grad_norm": 0.6285639064741724,
+      "learning_rate": 8.08589138174746e-06,
+      "loss": 0.67,
+      "step": 3590
+    },
+    {
+      "epoch": 0.8409356725146199,
+      "grad_norm": 0.6696016237475442,
+      "learning_rate": 8.042046382498862e-06,
+      "loss": 0.6896,
+      "step": 3595
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.6489876686398364,
+      "learning_rate": 7.998492502147478e-06,
+      "loss": 0.6765,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8432748538011696,
+      "grad_norm": 0.642937395823205,
+      "learning_rate": 7.955230392321826e-06,
+      "loss": 0.6797,
+      "step": 3605
+    },
+    {
+      "epoch": 0.8444444444444444,
+      "grad_norm": 0.6564324117611929,
+      "learning_rate": 7.91226070028513e-06,
+      "loss": 0.6744,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8456140350877193,
+      "grad_norm": 0.646811905002716,
+      "learning_rate": 7.869584068925617e-06,
+      "loss": 0.6752,
+      "step": 3615
+    },
+    {
+      "epoch": 0.8467836257309942,
+      "grad_norm": 0.6675140856614825,
+      "learning_rate": 7.827201136746903e-06,
+      "loss": 0.6868,
+      "step": 3620
+    },
+    {
+      "epoch": 0.847953216374269,
+      "grad_norm": 0.6822010982710452,
+      "learning_rate": 7.78511253785846e-06,
+      "loss": 0.6715,
+      "step": 3625
+    },
+    {
+      "epoch": 0.8491228070175438,
+      "grad_norm": 0.6392666243584267,
+      "learning_rate": 7.743318901966097e-06,
+      "loss": 0.6868,
+      "step": 3630
+    },
+    {
+      "epoch": 0.8502923976608188,
+      "grad_norm": 0.6113947733753667,
+      "learning_rate": 7.701820854362548e-06,
+      "loss": 0.6785,
+      "step": 3635
+    },
+    {
+      "epoch": 0.8514619883040936,
+      "grad_norm": 0.6747821078646472,
+      "learning_rate": 7.660619015918146e-06,
+      "loss": 0.6718,
+      "step": 3640
+    },
+    {
+      "epoch": 0.8526315789473684,
+      "grad_norm": 0.63975289517103,
+      "learning_rate": 7.6197140030714796e-06,
+      "loss": 0.6694,
+      "step": 3645
+    },
+    {
+      "epoch": 0.8538011695906432,
+      "grad_norm": 0.626650114047046,
+      "learning_rate": 7.579106427820232e-06,
+      "loss": 0.6805,
+      "step": 3650
+    },
+    {
+      "epoch": 0.8549707602339182,
+      "grad_norm": 0.6409849046483511,
+      "learning_rate": 7.538796897711965e-06,
+      "loss": 0.6761,
+      "step": 3655
+    },
+    {
+      "epoch": 0.856140350877193,
+      "grad_norm": 0.6471022035342423,
+      "learning_rate": 7.498786015835073e-06,
+      "loss": 0.673,
+      "step": 3660
+    },
+    {
+      "epoch": 0.8573099415204678,
+      "grad_norm": 0.6407581110519617,
+      "learning_rate": 7.459074380809753e-06,
+      "loss": 0.6877,
+      "step": 3665
+    },
+    {
+      "epoch": 0.8584795321637427,
+      "grad_norm": 0.6272955753937088,
+      "learning_rate": 7.419662586779016e-06,
+      "loss": 0.6798,
+      "step": 3670
+    },
+    {
+      "epoch": 0.8596491228070176,
+      "grad_norm": 0.6367532485529698,
+      "learning_rate": 7.380551223399836e-06,
+      "loss": 0.6817,
+      "step": 3675
+    },
+    {
+      "epoch": 0.8608187134502924,
+      "grad_norm": 0.6005127501375619,
+      "learning_rate": 7.341740875834319e-06,
+      "loss": 0.6746,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8619883040935673,
+      "grad_norm": 0.648068901146122,
+      "learning_rate": 7.303232124740925e-06,
+      "loss": 0.6702,
+      "step": 3685
+    },
+    {
+      "epoch": 0.8631578947368421,
+      "grad_norm": 0.6960242714907998,
+      "learning_rate": 7.265025546265813e-06,
+      "loss": 0.6734,
+      "step": 3690
+    },
+    {
+      "epoch": 0.8643274853801169,
+      "grad_norm": 0.6594167870334117,
+      "learning_rate": 7.227121712034209e-06,
+      "loss": 0.6655,
+      "step": 3695
+    },
+    {
+      "epoch": 0.8654970760233918,
+      "grad_norm": 0.6910368581395679,
+      "learning_rate": 7.189521189141829e-06,
+      "loss": 0.6782,
+      "step": 3700
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 0.6174471063653484,
+      "learning_rate": 7.152224540146443e-06,
+      "loss": 0.6675,
+      "step": 3705
+    },
+    {
+      "epoch": 0.8678362573099415,
+      "grad_norm": 0.6567058868642893,
+      "learning_rate": 7.115232323059417e-06,
+      "loss": 0.6704,
+      "step": 3710
+    },
+    {
+      "epoch": 0.8690058479532163,
+      "grad_norm": 0.6424216241729048,
+      "learning_rate": 7.078545091337374e-06,
+      "loss": 0.6748,
+      "step": 3715
+    },
+    {
+      "epoch": 0.8701754385964913,
+      "grad_norm": 0.6487145669681874,
+      "learning_rate": 7.042163393873935e-06,
+      "loss": 0.687,
+      "step": 3720
+    },
+    {
+      "epoch": 0.8713450292397661,
+      "grad_norm": 0.647202081451824,
+      "learning_rate": 7.006087774991478e-06,
+      "loss": 0.6773,
+      "step": 3725
+    },
+    {
+      "epoch": 0.8725146198830409,
+      "grad_norm": 0.6436919512408469,
+      "learning_rate": 6.970318774433005e-06,
+      "loss": 0.6618,
+      "step": 3730
+    },
+    {
+      "epoch": 0.8736842105263158,
+      "grad_norm": 0.6534618918546282,
+      "learning_rate": 6.934856927354077e-06,
+      "loss": 0.6869,
+      "step": 3735
+    },
+    {
+      "epoch": 0.8748538011695907,
+      "grad_norm": 0.6766619943500544,
+      "learning_rate": 6.899702764314796e-06,
+      "loss": 0.6872,
+      "step": 3740
+    },
+    {
+      "epoch": 0.8760233918128655,
+      "grad_norm": 0.609566488590824,
+      "learning_rate": 6.8648568112718606e-06,
+      "loss": 0.6729,
+      "step": 3745
+    },
+    {
+      "epoch": 0.8771929824561403,
+      "grad_norm": 0.6371626902320939,
+      "learning_rate": 6.830319589570722e-06,
+      "loss": 0.6718,
+      "step": 3750
+    },
+    {
+      "epoch": 0.8783625730994152,
+      "grad_norm": 0.6378883436857535,
+      "learning_rate": 6.796091615937747e-06,
+      "loss": 0.6601,
+      "step": 3755
+    },
+    {
+      "epoch": 0.87953216374269,
+      "grad_norm": 0.6586370697757771,
+      "learning_rate": 6.76217340247253e-06,
+      "loss": 0.6709,
+      "step": 3760
+    },
+    {
+      "epoch": 0.8807017543859649,
+      "grad_norm": 0.6137769168492903,
+      "learning_rate": 6.728565456640189e-06,
+      "loss": 0.6776,
+      "step": 3765
+    },
+    {
+      "epoch": 0.8818713450292398,
+      "grad_norm": 0.6101836408040017,
+      "learning_rate": 6.695268281263803e-06,
+      "loss": 0.6901,
+      "step": 3770
+    },
+    {
+      "epoch": 0.8830409356725146,
+      "grad_norm": 0.6329628323094153,
+      "learning_rate": 6.6622823745168844e-06,
+      "loss": 0.6897,
+      "step": 3775
+    },
+    {
+      "epoch": 0.8842105263157894,
+      "grad_norm": 0.6462015161934214,
+      "learning_rate": 6.629608229915907e-06,
+      "loss": 0.6808,
+      "step": 3780
+    },
+    {
+      "epoch": 0.8853801169590644,
+      "grad_norm": 0.6218725318180922,
+      "learning_rate": 6.597246336312947e-06,
+      "loss": 0.6934,
+      "step": 3785
+    },
+    {
+      "epoch": 0.8865497076023392,
+      "grad_norm": 0.6325806500322131,
+      "learning_rate": 6.565197177888353e-06,
+      "loss": 0.6797,
+      "step": 3790
+    },
+    {
+      "epoch": 0.887719298245614,
+      "grad_norm": 0.6680853833728703,
+      "learning_rate": 6.533461234143503e-06,
+      "loss": 0.6805,
+      "step": 3795
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.6327716959909178,
+      "learning_rate": 6.502038979893646e-06,
+      "loss": 0.6805,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8900584795321638,
+      "grad_norm": 0.6591393204504788,
+      "learning_rate": 6.4709308852607755e-06,
+      "loss": 0.6671,
+      "step": 3805
+    },
+    {
+      "epoch": 0.8912280701754386,
+      "grad_norm": 0.6952150402441917,
+      "learning_rate": 6.440137415666606e-06,
+      "loss": 0.6844,
+      "step": 3810
+    },
+    {
+      "epoch": 0.8923976608187134,
+      "grad_norm": 0.6353053040972749,
+      "learning_rate": 6.409659031825618e-06,
+      "loss": 0.6858,
+      "step": 3815
+    },
+    {
+      "epoch": 0.8935672514619883,
+      "grad_norm": 0.6049168982376912,
+      "learning_rate": 6.379496189738146e-06,
+      "loss": 0.6706,
+      "step": 3820
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.606413815250955,
+      "learning_rate": 6.3496493406835835e-06,
+      "loss": 0.664,
+      "step": 3825
+    },
+    {
+      "epoch": 0.895906432748538,
+      "grad_norm": 0.6784185660908973,
+      "learning_rate": 6.320118931213605e-06,
+      "loss": 0.6732,
+      "step": 3830
+    },
+    {
+      "epoch": 0.8970760233918129,
+      "grad_norm": 0.6731290698360186,
+      "learning_rate": 6.290905403145488e-06,
+      "loss": 0.6769,
+      "step": 3835
+    },
+    {
+      "epoch": 0.8982456140350877,
+      "grad_norm": 0.6804372100085168,
+      "learning_rate": 6.262009193555523e-06,
+      "loss": 0.6741,
+      "step": 3840
+    },
+    {
+      "epoch": 0.8994152046783626,
+      "grad_norm": 0.6284090122583083,
+      "learning_rate": 6.233430734772457e-06,
+      "loss": 0.6779,
+      "step": 3845
+    },
+    {
+      "epoch": 0.9005847953216374,
+      "grad_norm": 0.6821827933733495,
+      "learning_rate": 6.205170454371017e-06,
+      "loss": 0.6665,
+      "step": 3850
+    },
+    {
+      "epoch": 0.9017543859649123,
+      "grad_norm": 0.6189927650661127,
+      "learning_rate": 6.1772287751655465e-06,
+      "loss": 0.6837,
+      "step": 3855
+    },
+    {
+      "epoch": 0.9029239766081871,
+      "grad_norm": 0.6632044649328058,
+      "learning_rate": 6.149606115203644e-06,
+      "loss": 0.6943,
+      "step": 3860
+    },
+    {
+      "epoch": 0.904093567251462,
+      "grad_norm": 0.620002257673005,
+      "learning_rate": 6.122302887759918e-06,
+      "loss": 0.68,
+      "step": 3865
+    },
+    {
+      "epoch": 0.9052631578947369,
+      "grad_norm": 0.6441357917286101,
+      "learning_rate": 6.0953195013298255e-06,
+      "loss": 0.6825,
+      "step": 3870
+    },
+    {
+      "epoch": 0.9064327485380117,
+      "grad_norm": 2.236180657709731,
+      "learning_rate": 6.068656359623525e-06,
+      "loss": 0.6714,
+      "step": 3875
+    },
+    {
+      "epoch": 0.9076023391812865,
+      "grad_norm": 0.6171661564106794,
+      "learning_rate": 6.042313861559872e-06,
+      "loss": 0.6791,
+      "step": 3880
+    },
+    {
+      "epoch": 0.9087719298245615,
+      "grad_norm": 0.6121228446850321,
+      "learning_rate": 6.016292401260419e-06,
+      "loss": 0.6866,
+      "step": 3885
+    },
+    {
+      "epoch": 0.9099415204678363,
+      "grad_norm": 0.6768230538525649,
+      "learning_rate": 5.990592368043533e-06,
+      "loss": 0.6844,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9111111111111111,
+      "grad_norm": 0.6713551675375381,
+      "learning_rate": 5.965214146418583e-06,
+      "loss": 0.6865,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9122807017543859,
+      "grad_norm": 0.629363611094676,
+      "learning_rate": 5.9401581160801645e-06,
+      "loss": 0.6657,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9134502923976608,
+      "grad_norm": 0.6612627229951298,
+      "learning_rate": 5.915424651902437e-06,
+      "loss": 0.6911,
+      "step": 3905
+    },
+    {
+      "epoch": 0.9146198830409357,
+      "grad_norm": 0.6167590167685036,
+      "learning_rate": 5.891014123933495e-06,
+      "loss": 0.6764,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9157894736842105,
+      "grad_norm": 0.6280604076161018,
+      "learning_rate": 5.866926897389862e-06,
+      "loss": 0.6725,
+      "step": 3915
+    },
+    {
+      "epoch": 0.9169590643274854,
+      "grad_norm": 0.6363040677586755,
+      "learning_rate": 5.8431633326509895e-06,
+      "loss": 0.6839,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9181286549707602,
+      "grad_norm": 0.6233483061515912,
+      "learning_rate": 5.819723785253901e-06,
+      "loss": 0.67,
+      "step": 3925
+    },
+    {
+      "epoch": 0.9192982456140351,
+      "grad_norm": 0.6151583619535353,
+      "learning_rate": 5.796608605887838e-06,
+      "loss": 0.6695,
+      "step": 3930
+    },
+    {
+      "epoch": 0.92046783625731,
+      "grad_norm": 0.6038641602208659,
+      "learning_rate": 5.773818140389052e-06,
+      "loss": 0.6741,
+      "step": 3935
+    },
+    {
+      "epoch": 0.9216374269005848,
+      "grad_norm": 0.5940675664971363,
+      "learning_rate": 5.751352729735594e-06,
+      "loss": 0.6845,
+      "step": 3940
+    },
+    {
+      "epoch": 0.9228070175438596,
+      "grad_norm": 0.619934570260161,
+      "learning_rate": 5.729212710042228e-06,
+      "loss": 0.6749,
+      "step": 3945
+    },
+    {
+      "epoch": 0.9239766081871345,
+      "grad_norm": 0.6403366226216299,
+      "learning_rate": 5.707398412555415e-06,
+      "loss": 0.6787,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9251461988304094,
+      "grad_norm": 0.6392263250591611,
+      "learning_rate": 5.685910163648331e-06,
+      "loss": 0.6751,
+      "step": 3955
+    },
+    {
+      "epoch": 0.9263157894736842,
+      "grad_norm": 0.6037302371230343,
+      "learning_rate": 5.664748284815999e-06,
+      "loss": 0.6745,
+      "step": 3960
+    },
+    {
+      "epoch": 0.927485380116959,
+      "grad_norm": 0.6454567383685781,
+      "learning_rate": 5.6439130926704926e-06,
+      "loss": 0.6875,
+      "step": 3965
+    },
+    {
+      "epoch": 0.928654970760234,
+      "grad_norm": 0.6680699529149163,
+      "learning_rate": 5.623404898936162e-06,
+      "loss": 0.6682,
+      "step": 3970
+    },
+    {
+      "epoch": 0.9298245614035088,
+      "grad_norm": 0.6464856994915182,
+      "learning_rate": 5.603224010445013e-06,
+      "loss": 0.6853,
+      "step": 3975
+    },
+    {
+      "epoch": 0.9309941520467836,
+      "grad_norm": 0.6425767818922832,
+      "learning_rate": 5.5833707291320785e-06,
+      "loss": 0.68,
+      "step": 3980
+    },
+    {
+      "epoch": 0.9321637426900585,
+      "grad_norm": 0.662544717871915,
+      "learning_rate": 5.563845352030928e-06,
+      "loss": 0.6757,
+      "step": 3985
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 0.6297971690649151,
+      "learning_rate": 5.544648171269207e-06,
+      "loss": 0.6907,
+      "step": 3990
+    },
+    {
+      "epoch": 0.9345029239766082,
+      "grad_norm": 0.6337592219063408,
+      "learning_rate": 5.525779474064284e-06,
+      "loss": 0.6799,
+      "step": 3995
+    },
+    {
+      "epoch": 0.935672514619883,
+      "grad_norm": 0.6059097225565415,
+      "learning_rate": 5.507239542718928e-06,
+      "loss": 0.6764,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9368421052631579,
+      "grad_norm": 0.644532218298727,
+      "learning_rate": 5.48902865461711e-06,
+      "loss": 0.6902,
+      "step": 4005
+    },
+    {
+      "epoch": 0.9380116959064327,
+      "grad_norm": 0.6775880658086887,
+      "learning_rate": 5.471147082219839e-06,
+      "loss": 0.66,
+      "step": 4010
+    },
+    {
+      "epoch": 0.9391812865497076,
+      "grad_norm": 0.6140932767025524,
+      "learning_rate": 5.453595093061084e-06,
+      "loss": 0.6827,
+      "step": 4015
+    },
+    {
+      "epoch": 0.9403508771929825,
+      "grad_norm": 0.6442340334722829,
+      "learning_rate": 5.436372949743784e-06,
+      "loss": 0.6721,
+      "step": 4020
+    },
+    {
+      "epoch": 0.9415204678362573,
+      "grad_norm": 0.6188202885625294,
+      "learning_rate": 5.4194809099359016e-06,
+      "loss": 0.6723,
+      "step": 4025
+    },
+    {
+      "epoch": 0.9426900584795321,
+      "grad_norm": 0.6015661608114689,
+      "learning_rate": 5.402919226366589e-06,
+      "loss": 0.6683,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9438596491228071,
+      "grad_norm": 0.6298388121183706,
+      "learning_rate": 5.386688146822386e-06,
+      "loss": 0.6669,
+      "step": 4035
+    },
+    {
+      "epoch": 0.9450292397660819,
+      "grad_norm": 0.6643089494356405,
+      "learning_rate": 5.370787914143523e-06,
+      "loss": 0.6758,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9461988304093567,
+      "grad_norm": 0.7117885834075908,
+      "learning_rate": 5.355218766220295e-06,
+      "loss": 0.6747,
+      "step": 4045
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 1.069267992100781,
+      "learning_rate": 5.3399809359894815e-06,
+      "loss": 0.6923,
+      "step": 4050
+    },
+    {
+      "epoch": 0.9485380116959065,
+      "grad_norm": 0.6155712114839152,
+      "learning_rate": 5.325074651430884e-06,
+      "loss": 0.6778,
+      "step": 4055
+    },
+    {
+      "epoch": 0.9497076023391813,
+      "grad_norm": 0.6171532400917544,
+      "learning_rate": 5.3105001355638965e-06,
+      "loss": 0.6702,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9508771929824561,
+      "grad_norm": 0.6746041528422164,
+      "learning_rate": 5.296257606444188e-06,
+      "loss": 0.6903,
+      "step": 4065
+    },
+    {
+      "epoch": 0.952046783625731,
+      "grad_norm": 0.6261575844875003,
+      "learning_rate": 5.2823472771604235e-06,
+      "loss": 0.6792,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9532163742690059,
+      "grad_norm": 0.6213266374401143,
+      "learning_rate": 5.268769355831078e-06,
+      "loss": 0.6756,
+      "step": 4075
+    },
+    {
+      "epoch": 0.9543859649122807,
+      "grad_norm": 0.6235253860423335,
+      "learning_rate": 5.255524045601336e-06,
+      "loss": 0.6695,
+      "step": 4080
+    },
+    {
+      "epoch": 0.9555555555555556,
+      "grad_norm": 0.6199005302450762,
+      "learning_rate": 5.2426115446400375e-06,
+      "loss": 0.674,
+      "step": 4085
+    },
+    {
+      "epoch": 0.9567251461988304,
+      "grad_norm": 0.6472698916651094,
+      "learning_rate": 5.230032046136718e-06,
+      "loss": 0.677,
+      "step": 4090
+    },
+    {
+      "epoch": 0.9578947368421052,
+      "grad_norm": 0.6508423399986626,
+      "learning_rate": 5.217785738298716e-06,
+      "loss": 0.6818,
+      "step": 4095
+    },
+    {
+      "epoch": 0.9590643274853801,
+      "grad_norm": 0.6272752070860681,
+      "learning_rate": 5.20587280434837e-06,
+      "loss": 0.6655,
+      "step": 4100
+    },
+    {
+      "epoch": 0.960233918128655,
+      "grad_norm": 0.6249813913743417,
+      "learning_rate": 5.1942934225202566e-06,
+      "loss": 0.6649,
+      "step": 4105
+    },
+    {
+      "epoch": 0.9614035087719298,
+      "grad_norm": 0.6088008404693435,
+      "learning_rate": 5.183047766058537e-06,
+      "loss": 0.6673,
+      "step": 4110
+    },
+    {
+      "epoch": 0.9625730994152046,
+      "grad_norm": 0.6157262304244052,
+      "learning_rate": 5.172136003214364e-06,
+      "loss": 0.661,
+      "step": 4115
+    },
+    {
+      "epoch": 0.9637426900584796,
+      "grad_norm": 0.6202554098999495,
+      "learning_rate": 5.16155829724336e-06,
+      "loss": 0.6788,
+      "step": 4120
+    },
+    {
+      "epoch": 0.9649122807017544,
+      "grad_norm": 0.6460466051369329,
+      "learning_rate": 5.151314806403183e-06,
+      "loss": 0.6722,
+      "step": 4125
+    },
+    {
+      "epoch": 0.9660818713450292,
+      "grad_norm": 0.6197826502012264,
+      "learning_rate": 5.141405683951148e-06,
+      "loss": 0.6757,
+      "step": 4130
+    },
+    {
+      "epoch": 0.9672514619883041,
+      "grad_norm": 0.6430775828131704,
+      "learning_rate": 5.13183107814194e-06,
+      "loss": 0.6714,
+      "step": 4135
+    },
+    {
+      "epoch": 0.968421052631579,
+      "grad_norm": 0.6716734720725397,
+      "learning_rate": 5.1225911322253975e-06,
+      "loss": 0.6689,
+      "step": 4140
+    },
+    {
+      "epoch": 0.9695906432748538,
+      "grad_norm": 0.7005363493700107,
+      "learning_rate": 5.113685984444362e-06,
+      "loss": 0.6775,
+      "step": 4145
+    },
+    {
+      "epoch": 0.9707602339181286,
+      "grad_norm": 0.6210358032380061,
+      "learning_rate": 5.105115768032622e-06,
+      "loss": 0.6705,
+      "step": 4150
+    },
+    {
+      "epoch": 0.9719298245614035,
+      "grad_norm": 0.6552414802641572,
+      "learning_rate": 5.096880611212908e-06,
+      "loss": 0.6747,
+      "step": 4155
+    },
+    {
+      "epoch": 0.9730994152046784,
+      "grad_norm": 0.61884873350533,
+      "learning_rate": 5.0889806371949775e-06,
+      "loss": 0.6682,
+      "step": 4160
+    },
+    {
+      "epoch": 0.9742690058479532,
+      "grad_norm": 0.611707842587258,
+      "learning_rate": 5.081415964173772e-06,
+      "loss": 0.6815,
+      "step": 4165
+    },
+    {
+      "epoch": 0.9754385964912281,
+      "grad_norm": 0.6309765692328979,
+      "learning_rate": 5.074186705327656e-06,
+      "loss": 0.6673,
+      "step": 4170
+    },
+    {
+      "epoch": 0.9766081871345029,
+      "grad_norm": 0.6488169707206964,
+      "learning_rate": 5.067292968816706e-06,
+      "loss": 0.6785,
+      "step": 4175
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 0.6039253712245951,
+      "learning_rate": 5.060734857781115e-06,
+      "loss": 0.6771,
+      "step": 4180
+    },
+    {
+      "epoch": 0.9789473684210527,
+      "grad_norm": 0.6120032767383402,
+      "learning_rate": 5.0545124703396265e-06,
+      "loss": 0.6723,
+      "step": 4185
+    },
+    {
+      "epoch": 0.9801169590643275,
+      "grad_norm": 0.6868334998745559,
+      "learning_rate": 5.048625899588081e-06,
+      "loss": 0.6613,
+      "step": 4190
+    },
+    {
+      "epoch": 0.9812865497076023,
+      "grad_norm": 0.6296139609807546,
+      "learning_rate": 5.043075233598026e-06,
+      "loss": 0.6633,
+      "step": 4195
+    },
+    {
+      "epoch": 0.9824561403508771,
+      "grad_norm": 0.6349010729713203,
+      "learning_rate": 5.037860555415383e-06,
+      "loss": 0.6718,
+      "step": 4200
+    },
+    {
+      "epoch": 0.9836257309941521,
+      "grad_norm": 0.6739022750199818,
+      "learning_rate": 5.032981943059216e-06,
+      "loss": 0.6772,
+      "step": 4205
+    },
+    {
+      "epoch": 0.9847953216374269,
+      "grad_norm": 0.650862024747413,
+      "learning_rate": 5.028439469520571e-06,
+      "loss": 0.6711,
+      "step": 4210
+    },
+    {
+      "epoch": 0.9859649122807017,
+      "grad_norm": 0.6323528177012202,
+      "learning_rate": 5.024233202761362e-06,
+      "loss": 0.6672,
+      "step": 4215
+    },
+    {
+      "epoch": 0.9871345029239766,
+      "grad_norm": 0.6237317952427651,
+      "learning_rate": 5.020363205713377e-06,
+      "loss": 0.661,
+      "step": 4220
+    },
+    {
+      "epoch": 0.9883040935672515,
+      "grad_norm": 0.6005441441484929,
+      "learning_rate": 5.016829536277317e-06,
+      "loss": 0.6823,
+      "step": 4225
+    },
+    {
+      "epoch": 0.9894736842105263,
+      "grad_norm": 0.6095441780661974,
+      "learning_rate": 5.0136322473219525e-06,
+      "loss": 0.6764,
+      "step": 4230
+    },
+    {
+      "epoch": 0.9906432748538012,
+      "grad_norm": 0.6162120041059115,
+      "learning_rate": 5.010771386683308e-06,
+      "loss": 0.6639,
+      "step": 4235
+    },
+    {
+      "epoch": 0.991812865497076,
+      "grad_norm": 0.6852116289986068,
+      "learning_rate": 5.008246997163965e-06,
+      "loss": 0.6782,
+      "step": 4240
+    },
+    {
+      "epoch": 0.9929824561403509,
+      "grad_norm": 0.6310290602971912,
+      "learning_rate": 5.006059116532412e-06,
+      "loss": 0.6858,
+      "step": 4245
+    },
+    {
+      "epoch": 0.9941520467836257,
+      "grad_norm": 0.6277161545850843,
+      "learning_rate": 5.00420777752248e-06,
+      "loss": 0.6857,
+      "step": 4250
+    },
+    {
+      "epoch": 0.9953216374269006,
+      "grad_norm": 0.627301841338485,
+      "learning_rate": 5.002693007832853e-06,
+      "loss": 0.6743,
+      "step": 4255
+    },
+    {
+      "epoch": 0.9964912280701754,
+      "grad_norm": 0.6332268702872285,
+      "learning_rate": 5.0015148301266646e-06,
+      "loss": 0.6542,
+      "step": 4260
+    },
+    {
+      "epoch": 0.9976608187134502,
+      "grad_norm": 0.6299167135136249,
+      "learning_rate": 5.000673262031141e-06,
+      "loss": 0.6787,
+      "step": 4265
+    },
+    {
+      "epoch": 0.9988304093567252,
+      "grad_norm": 0.6180934069153444,
+      "learning_rate": 5.000168316137349e-06,
+      "loss": 0.6723,
+      "step": 4270
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6381738406824355,
+      "learning_rate": 5e-06,
+      "loss": 0.6605,
+      "step": 4275
+    },
     {
       "epoch": 1.0,
-      "step": 536,
-      "total_flos": 488621249396736.0,
-      "train_loss": 0.5610741035484555,
-      "train_runtime": 6486.6391,
-      "train_samples_per_second": 5.288,
-      "train_steps_per_second": 0.083
+      "step": 4275,
+      "total_flos": 588699087667200.0,
+      "train_loss": 0.7670444376984535,
+      "train_runtime": 29567.5701,
+      "train_samples_per_second": 18.507,
+      "train_steps_per_second": 0.145
     }
   ],
   "logging_steps": 5,
-  "max_steps": 536,
+  "max_steps": 4275,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,
@@ -784,7 +6020,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 488621249396736.0,
+  "total_flos": 588699087667200.0,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null