diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.020107784973506633,
+  "epoch": 0.9999547121960056,
   "eval_steps": 500,
-  "global_step": 111,
+  "global_step": 5520,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -136,40 +136,6532 @@
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.8071302737293295e-06,
-      "loss": 1.2382,
+      "learning_rate": 3.804347826086957e-05,
+      "loss": 1.2379,
       "step": 105
     },
     {
       "epoch": 0.02,
-      "learning_rate": 5.0345761681491746e-08,
-      "loss": 1.2627,
+      "learning_rate": 3.985507246376812e-05,
+      "loss": 1.2613,
       "step": 110
     },
     {
       "epoch": 0.02,
-      "eval_loss": 1.2271578311920166,
-      "eval_runtime": 1802.5494,
-      "eval_samples_per_second": 16.27,
-      "eval_steps_per_second": 1.356,
-      "step": 111
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 1.2182,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.347826086956522e-05,
+      "loss": 1.238,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.528985507246377e-05,
+      "loss": 1.2098,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.710144927536232e-05,
+      "loss": 1.2186,
+      "step": 130
     },
     {
       "epoch": 0.02,
-      "step": 111,
-      "total_flos": 79541494087680.0,
-      "train_loss": 1.2791897563246992,
-      "train_runtime": 2444.0756,
-      "train_samples_per_second": 2.168,
-      "train_steps_per_second": 0.045
+      "learning_rate": 4.891304347826087e-05,
+      "loss": 1.2178,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.072463768115943e-05,
+      "loss": 1.2473,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2536231884057975e-05,
+      "loss": 1.2573,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4347826086956524e-05,
+      "loss": 1.2302,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.615942028985508e-05,
+      "loss": 1.2523,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.797101449275363e-05,
+      "loss": 1.3071,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9782608695652175e-05,
+      "loss": 1.2152,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.159420289855072e-05,
+      "loss": 1.2359,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.340579710144928e-05,
+      "loss": 1.2098,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.521739130434783e-05,
+      "loss": 1.2444,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.702898550724638e-05,
+      "loss": 1.2492,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.884057971014493e-05,
+      "loss": 1.1941,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.065217391304349e-05,
+      "loss": 1.2096,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.246376811594203e-05,
+      "loss": 1.2363,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.427536231884058e-05,
+      "loss": 1.2613,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.608695652173914e-05,
+      "loss": 1.2459,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.789855072463769e-05,
+      "loss": 1.2951,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.971014492753623e-05,
+      "loss": 1.2114,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.152173913043478e-05,
+      "loss": 1.2315,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 1.2311,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.514492753623189e-05,
+      "loss": 1.2155,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.695652173913044e-05,
+      "loss": 1.2456,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.876811594202898e-05,
+      "loss": 1.2576,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.057971014492754e-05,
+      "loss": 1.1968,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.239130434782609e-05,
+      "loss": 1.361,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.420289855072463e-05,
+      "loss": 1.2729,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.60144927536232e-05,
+      "loss": 1.2613,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.782608695652174e-05,
+      "loss": 1.2513,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.96376811594203e-05,
+      "loss": 1.2838,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010144927536231885,
+      "loss": 1.216,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010326086956521738,
+      "loss": 1.2206,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010507246376811595,
+      "loss": 1.2322,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001068840579710145,
+      "loss": 1.2467,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010869565217391305,
+      "loss": 1.2177,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001105072463768116,
+      "loss": 1.2339,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011231884057971016,
+      "loss": 1.2207,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001141304347826087,
+      "loss": 1.2125,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011594202898550725,
+      "loss": 1.2331,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001177536231884058,
+      "loss": 1.2279,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011956521739130435,
+      "loss": 1.2772,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001213768115942029,
+      "loss": 1.2357,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012318840579710145,
+      "loss": 1.2611,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000125,
+      "loss": 1.2308,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012681159420289856,
+      "loss": 1.2168,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001286231884057971,
+      "loss": 1.1828,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013043478260869567,
+      "loss": 1.2666,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013224637681159422,
+      "loss": 1.2298,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013405797101449275,
+      "loss": 1.2249,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001358695652173913,
+      "loss": 1.2445,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013768115942028986,
+      "loss": 1.2371,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013949275362318842,
+      "loss": 1.2005,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014130434782608697,
+      "loss": 1.2625,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001431159420289855,
+      "loss": 1.226,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014492753623188405,
+      "loss": 1.2484,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014673913043478264,
+      "loss": 1.2482,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014855072463768116,
+      "loss": 1.209,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015036231884057972,
+      "loss": 1.2102,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015217391304347827,
+      "loss": 1.2347,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001539855072463768,
+      "loss": 1.2134,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015579710144927538,
+      "loss": 1.2436,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001576086956521739,
+      "loss": 1.2616,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015942028985507247,
+      "loss": 1.2392,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016123188405797102,
+      "loss": 1.2507,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016304347826086955,
+      "loss": 1.2315,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016485507246376813,
+      "loss": 1.2151,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 1.168,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016847826086956522,
+      "loss": 1.2123,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017028985507246377,
+      "loss": 1.2129,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017210144927536233,
+      "loss": 1.201,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017391304347826088,
+      "loss": 1.1735,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017572463768115944,
+      "loss": 1.2197,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017753623188405796,
+      "loss": 1.2151,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017934782608695652,
+      "loss": 1.2171,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018115942028985507,
+      "loss": 1.2377,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018297101449275363,
+      "loss": 1.2137,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018478260869565218,
+      "loss": 1.2363,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018659420289855074,
+      "loss": 1.2076,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018840579710144927,
+      "loss": 1.2246,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019021739130434782,
+      "loss": 1.2535,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001920289855072464,
+      "loss": 1.2652,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019384057971014493,
+      "loss": 1.2043,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001956521739130435,
+      "loss": 1.2072,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019746376811594204,
+      "loss": 1.2167,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001992753623188406,
+      "loss": 1.2739,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999982005120014,
+      "loss": 1.2288,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999872036643513,
+      "loss": 1.2974,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999662097944096,
+      "loss": 1.2527,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999352191120556,
+      "loss": 1.2544,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998942319271077,
+      "loss": 1.2211,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998432486493205,
+      "loss": 1.2673,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997822697883822,
+      "loss": 1.2358,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999711295953907,
+      "loss": 1.1941,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999630327855431,
+      "loss": 1.2176,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019995393663024054,
+      "loss": 1.2103,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019994384122041853,
+      "loss": 1.186,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019993274665700244,
+      "loss": 1.2305,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999206530509063,
+      "loss": 1.2734,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019990756052303173,
+      "loss": 1.2792,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019989346920426667,
+      "loss": 1.2577,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001998783792354841,
+      "loss": 1.2192,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998622907675408,
+      "loss": 1.2149,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019984520396127553,
+      "loss": 1.2541,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998271189875077,
+      "loss": 1.1841,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998080360270355,
+      "loss": 1.2418,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997879552706341,
+      "loss": 1.1846,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019976687691905393,
+      "loss": 1.2417,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019974480118301838,
+      "loss": 1.2281,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997217282832219,
+      "loss": 1.2,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019969765845032775,
+      "loss": 1.2256,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996725919249657,
+      "loss": 1.2207,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019964652895772947,
+      "loss": 1.2042,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019961946980917456,
+      "loss": 1.2317,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995914147498153,
+      "loss": 1.2431,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019956236406012232,
+      "loss": 1.2345,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019953231803051974,
+      "loss": 1.2213,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950127696138225,
+      "loss": 1.2243,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019946924116303206,
+      "loss": 1.2061,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019943621095573586,
+      "loss": 1.2563,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019940218666970161,
+      "loss": 1.2556,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993671686450752,
+      "loss": 1.353,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019933115723193707,
+      "loss": 1.4143,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019929415279029873,
+      "loss": 1.2836,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019925615569009916,
+      "loss": 1.3538,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019921716631120107,
+      "loss": 1.3211,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019917718504338714,
+      "loss": 1.2648,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991362122863561,
+      "loss": 1.2779,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019909424844971873,
+      "loss": 1.2609,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990512939529939,
+      "loss": 1.2179,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019900734922560407,
+      "loss": 1.2384,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989624147068713,
+      "loss": 1.2196,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019891649084601278,
+      "loss": 1.2582,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019886957810213619,
+      "loss": 1.2313,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988216769442353,
+      "loss": 1.2451,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019877278785118517,
+      "loss": 1.2274,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019872291131173742,
+      "loss": 1.2455,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001986720478245153,
+      "loss": 1.215,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001986201978980087,
+      "loss": 1.2123,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001985673620505692,
+      "loss": 1.2027,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019851354081040467,
+      "loss": 1.2324,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001984587347155741,
+      "loss": 1.245,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019840294431398226,
+      "loss": 1.2135,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001983461701633742,
+      "loss": 1.1822,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019828841283132964,
+      "loss": 1.2285,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001982296728952573,
+      "loss": 1.237,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019816995094238912,
+      "loss": 1.2098,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019810924756977444,
+      "loss": 1.1901,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000198047563384274,
+      "loss": 1.2362,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019798489900255389,
+      "loss": 1.2439,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019792125505107931,
+      "loss": 1.2188,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019785663216610844,
+      "loss": 1.2794,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019779103099368595,
+      "loss": 1.2214,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001977244521896366,
+      "loss": 1.2287,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001976568964195587,
+      "loss": 1.2682,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019758836435881746,
+      "loss": 1.2253,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019751885669253816,
+      "loss": 1.2026,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019744837411559942,
+      "loss": 1.2285,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001973769173326261,
+      "loss": 1.2431,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019730448705798239,
+      "loss": 1.2365,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019723108401576466,
+      "loss": 1.2019,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019715670893979414,
+      "loss": 1.2133,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019708136257360966,
+      "loss": 1.2687,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019700504567046013,
+      "loss": 1.2272,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019692775899329707,
+      "loss": 1.2396,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019684950331476706,
+      "loss": 1.2365,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019677027941720384,
+      "loss": 1.1987,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019669008809262062,
+      "loss": 1.2087,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019660893014270212,
+      "loss": 1.2339,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019652680637879654,
+      "loss": 1.2094,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001964437176219075,
+      "loss": 1.2445,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019635966470268583,
+      "loss": 1.2229,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019627464846142111,
+      "loss": 1.194,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001961886697480335,
+      "loss": 1.2036,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019610172942206516,
+      "loss": 1.2269,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001960138283526715,
+      "loss": 1.2169,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019592496741861282,
+      "loss": 1.2136,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019583514750824512,
+      "loss": 1.2027,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019574436951951162,
+      "loss": 1.2836,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001956526343599335,
+      "loss": 1.1955,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019555994294660086,
+      "loss": 1.2464,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019546629620616375,
+      "loss": 1.2026,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001953716950748227,
+      "loss": 1.2482,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001952761404983194,
+      "loss": 1.2133,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019517963343192732,
+      "loss": 1.2042,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001950821748404421,
+      "loss": 1.2268,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019498376569817194,
+      "loss": 1.2147,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001948844069889278,
+      "loss": 1.2038,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019478409970601358,
+      "loss": 1.2155,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001946828448522163,
+      "loss": 1.2446,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019458064343979596,
+      "loss": 1.3413,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019447749649047542,
+      "loss": 3.8626,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001943734050354302,
+      "loss": 6.3609,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019426837011527823,
+      "loss": 5.2091,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001941623927800694,
+      "loss": 1.9048,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019405547408927502,
+      "loss": 1.3353,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019394761511177733,
+      "loss": 1.586,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001938388169258587,
+      "loss": 1.6837,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019372908061919097,
+      "loss": 1.4677,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019361840728882447,
+      "loss": 1.4636,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019350679804117711,
+      "loss": 1.8272,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019339425399202327,
+      "loss": 2.1575,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019328077626648278,
+      "loss": 1.8073,
+      "step": 1135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019316636599900946,
+      "loss": 1.5317,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019305102433337998,
+      "loss": 1.4068,
+      "step": 1145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019293475242268223,
+      "loss": 1.9292,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019281755142930407,
+      "loss": 2.3396,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019269942252492133,
+      "loss": 1.7311,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001925803668904865,
+      "loss": 1.5029,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019246038571621657,
+      "loss": 1.5402,
+      "step": 1170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001923394802015814,
+      "loss": 1.518,
+      "step": 1175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019221765155529158,
+      "loss": 1.4632,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019209490099528643,
+      "loss": 1.3406,
+      "step": 1185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019197122974872163,
+      "loss": 1.3142,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001918466390519573,
+      "loss": 1.3173,
+      "step": 1195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019172113015054532,
+      "loss": 1.2899,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019159470429921702,
+      "loss": 1.2821,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019146736276187066,
+      "loss": 1.3015,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019133910681155868,
+      "loss": 1.2785,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019120993773047513,
+      "loss": 1.2912,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019107985680994266,
+      "loss": 1.2846,
+      "step": 1225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019094886535039982,
+      "loss": 1.2638,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001908169646613879,
+      "loss": 1.2445,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019068415606153787,
+      "loss": 1.262,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019055044087855726,
+      "loss": 1.2625,
+      "step": 1245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019041582044921688,
+      "loss": 1.2291,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019028029611933739,
+      "loss": 1.2873,
+      "step": 1255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019014386924377582,
+      "loss": 1.2883,
+      "step": 1260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019000654118641211,
+      "loss": 1.3117,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001898683133201356,
+      "loss": 1.2681,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018972918702683092,
+      "loss": 1.238,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001895891636973646,
+      "loss": 1.2715,
+      "step": 1280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018944824473157086,
+      "loss": 1.2795,
+      "step": 1285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018930643153823777,
+      "loss": 1.255,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018916372553509314,
+      "loss": 1.2555,
+      "step": 1295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001890201281487903,
+      "loss": 1.2326,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018887564081489393,
+      "loss": 1.2387,
+      "step": 1305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001887302649778656,
+      "loss": 1.2303,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018858400209104933,
+      "loss": 1.2334,
+      "step": 1315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018843685361665723,
+      "loss": 1.241,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018828882102575473,
+      "loss": 1.2329,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001881399057982458,
+      "loss": 1.244,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001879901094228584,
+      "loss": 1.1944,
+      "step": 1335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018783943339712938,
+      "loss": 1.2184,
+      "step": 1340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001876878792273896,
+      "loss": 1.2478,
+      "step": 1345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018753544842874887,
+      "loss": 1.2352,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018738214252508073,
+      "loss": 1.2322,
+      "step": 1355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001872279630490074,
+      "loss": 1.2599,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018707291154188425,
+      "loss": 1.2495,
+      "step": 1365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018691698955378445,
+      "loss": 1.2369,
+      "step": 1370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001867601986434836,
+      "loss": 1.2732,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 1.3559,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018644401633479874,
+      "loss": 1.9588,
+      "step": 1385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018628462809733683,
+      "loss": 2.5337,
+      "step": 1390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018612437725948631,
+      "loss": 3.5829,
+      "step": 1395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018596326542329888,
+      "loss": 2.7139,
+      "step": 1400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018580129419943373,
+      "loss": 1.935,
+      "step": 1405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018563846520714154,
+      "loss": 1.5826,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018547478007424823,
+      "loss": 1.4232,
+      "step": 1415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018531024043713868,
+      "loss": 1.4101,
+      "step": 1420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018514484794074026,
+      "loss": 1.4018,
+      "step": 1425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001849786042385067,
+      "loss": 1.3299,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018481151099240123,
+      "loss": 1.306,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018464356987288013,
+      "loss": 1.3258,
+      "step": 1440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018447478255887595,
+      "loss": 1.3129,
+      "step": 1445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018430515073778093,
+      "loss": 1.3296,
+      "step": 1450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001841346761054298,
+      "loss": 1.2667,
+      "step": 1455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018396336036608307,
+      "loss": 1.2641,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001837912052324099,
+      "loss": 1.2361,
+      "step": 1465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001836182124254711,
+      "loss": 1.3251,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018344438367470168,
+      "loss": 1.3017,
+      "step": 1475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001832697207178938,
+      "loss": 1.2847,
+      "step": 1480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018309422530117924,
+      "loss": 1.3278,
+      "step": 1485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001829178991790121,
+      "loss": 1.2803,
+      "step": 1490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018274074411415105,
+      "loss": 1.3346,
+      "step": 1495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018256276187764197,
+      "loss": 1.2782,
+      "step": 1500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018238395424879992,
+      "loss": 1.3485,
+      "step": 1505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018220432301519168,
+      "loss": 1.3187,
+      "step": 1510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001820238699726177,
+      "loss": 1.2878,
+      "step": 1515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018184259692509406,
+      "loss": 1.3199,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018166050568483474,
+      "loss": 1.2897,
+      "step": 1525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001814775980722332,
+      "loss": 1.3277,
+      "step": 1530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001812938759158443,
+      "loss": 1.3135,
+      "step": 1535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000181109341052366,
+      "loss": 1.3059,
+      "step": 1540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018092399532662113,
+      "loss": 1.342,
+      "step": 1545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018073784059153872,
+      "loss": 1.3264,
+      "step": 1550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018055087870813558,
+      "loss": 1.3069,
+      "step": 1555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018036311154549784,
+      "loss": 1.3145,
+      "step": 1560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018017454098076194,
+      "loss": 1.313,
+      "step": 1565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017998516889909614,
+      "loss": 1.2827,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017979499719368168,
+      "loss": 1.2697,
+      "step": 1575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001796040277656936,
+      "loss": 1.3068,
+      "step": 1580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001794122625242819,
+      "loss": 1.2713,
+      "step": 1585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017921970338655266,
+      "loss": 1.2812,
+      "step": 1590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017902635227754838,
+      "loss": 1.2795,
+      "step": 1595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017883221113022916,
+      "loss": 1.2631,
+      "step": 1600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017863728188545326,
+      "loss": 1.2359,
+      "step": 1605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017844156649195759,
+      "loss": 1.272,
+      "step": 1610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017824506690633832,
+      "loss": 1.2845,
+      "step": 1615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017804778509303138,
+      "loss": 1.2616,
+      "step": 1620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017784972302429264,
+      "loss": 1.2332,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001776508826801784,
+      "loss": 1.2521,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017745126604852538,
+      "loss": 1.2938,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001772508751249311,
+      "loss": 1.2933,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017704971191273368,
+      "loss": 1.2702,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017684777842299205,
+      "loss": 1.2851,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017664507667446554,
+      "loss": 1.2907,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017644160869359404,
+      "loss": 1.2637,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001762373765144775,
+      "loss": 1.2571,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017603238217885569,
+      "loss": 1.2194,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017582662773608777,
+      "loss": 1.2755,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017562011524313185,
+      "loss": 1.2936,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001754128467645243,
+      "loss": 1.2564,
+      "step": 1685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001752048243723593,
+      "loss": 1.2177,
+      "step": 1690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017499605014626788,
+      "loss": 1.3018,
+      "step": 1695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017478652617339738,
+      "loss": 1.217,
+      "step": 1700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017457625454839039,
+      "loss": 1.2899,
+      "step": 1705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017436523737336402,
+      "loss": 1.2608,
+      "step": 1710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017415347675788856,
+      "loss": 1.2849,
+      "step": 1715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017394097481896676,
+      "loss": 1.3029,
+      "step": 1720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001737277336810124,
+      "loss": 1.2907,
+      "step": 1725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001735137554758292,
+      "loss": 1.308,
+      "step": 1730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001732990423425894,
+      "loss": 1.3432,
+      "step": 1735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017308359642781242,
+      "loss": 1.3239,
+      "step": 1740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017286741988534348,
+      "loss": 1.2793,
+      "step": 1745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001726505148763319,
+      "loss": 1.342,
+      "step": 1750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001724328835692097,
+      "loss": 1.2451,
+      "step": 1755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001722145281396697,
+      "loss": 1.2804,
+      "step": 1760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017199545077064394,
+      "loss": 1.3129,
+      "step": 1765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017177565365228178,
+      "loss": 1.2582,
+      "step": 1770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017155513898192806,
+      "loss": 1.2644,
+      "step": 1775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017133390896410106,
+      "loss": 1.2875,
+      "step": 1780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001711119658104705,
+      "loss": 1.3237,
+      "step": 1785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017088931173983539,
+      "loss": 1.3011,
+      "step": 1790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017066594897810196,
+      "loss": 1.3125,
+      "step": 1795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017044187975826124,
+      "loss": 1.35,
+      "step": 1800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017021710632036694,
+      "loss": 1.4059,
+      "step": 1805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016999163091151287,
+      "loss": 1.4471,
+      "step": 1810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016976545578581057,
+      "loss": 1.3584,
+      "step": 1815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016953858320436672,
+      "loss": 1.3308,
+      "step": 1820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001693110154352606,
+      "loss": 1.3613,
+      "step": 1825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001690827547535214,
+      "loss": 1.3879,
+      "step": 1830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016885380344110545,
+      "loss": 1.3412,
+      "step": 1835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001686241637868734,
+      "loss": 1.3521,
+      "step": 1840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016839383808656732,
+      "loss": 1.3658,
+      "step": 1845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016816282864278793,
+      "loss": 1.3324,
+      "step": 1850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016793113776497127,
+      "loss": 1.3101,
+      "step": 1855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001676987677693659,
+      "loss": 1.3923,
+      "step": 1860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001674657209790095,
+      "loss": 1.3177,
+      "step": 1865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016723199972370594,
+      "loss": 1.2952,
+      "step": 1870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016699760634000165,
+      "loss": 1.3095,
+      "step": 1875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016676254317116252,
+      "loss": 1.3412,
+      "step": 1880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016652681256715032,
+      "loss": 1.2945,
+      "step": 1885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016629041688459941,
+      "loss": 1.3498,
+      "step": 1890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001660533584867928,
+      "loss": 1.38,
+      "step": 1895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016581563974363902,
+      "loss": 1.3754,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016557726303164803,
+      "loss": 1.357,
+      "step": 1905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016533823073390757,
+      "loss": 1.3054,
+      "step": 1910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016509854524005948,
+      "loss": 1.3338,
+      "step": 1915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001648582089462756,
+      "loss": 1.2822,
+      "step": 1920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016461722425523402,
+      "loss": 1.3557,
+      "step": 1925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016437559357609488,
+      "loss": 1.3282,
+      "step": 1930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016413331932447638,
+      "loss": 1.3206,
+      "step": 1935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016389040392243056,
+      "loss": 1.2873,
+      "step": 1940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016364684979841925,
+      "loss": 1.3082,
+      "step": 1945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016340265938728958,
+      "loss": 1.3155,
+      "step": 1950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016315783513024977,
+      "loss": 1.3231,
+      "step": 1955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001629123794748447,
+      "loss": 1.3236,
+      "step": 1960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016266629487493144,
+      "loss": 1.3079,
+      "step": 1965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001624195837906547,
+      "loss": 1.3281,
+      "step": 1970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001621722486884222,
+      "loss": 1.3313,
+      "step": 1975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001619242920408802,
+      "loss": 1.3004,
+      "step": 1980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001616757163268885,
+      "loss": 1.3087,
+      "step": 1985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016142652403149582,
+      "loss": 1.3004,
+      "step": 1990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016117671764591504,
+      "loss": 1.3005,
+      "step": 1995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001609262996674981,
+      "loss": 1.3026,
+      "step": 2000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016067527259971113,
+      "loss": 1.3331,
+      "step": 2005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016042363895210946,
+      "loss": 1.2832,
+      "step": 2010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016017140124031245,
+      "loss": 1.3191,
+      "step": 2015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001599185619859784,
+      "loss": 1.3237,
+      "step": 2020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015966512371677928,
+      "loss": 1.2876,
+      "step": 2025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001594110889663756,
+      "loss": 1.307,
+      "step": 2030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015915646027439087,
+      "loss": 1.3195,
+      "step": 2035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015890124018638638,
+      "loss": 1.2279,
+      "step": 2040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015864543125383574,
+      "loss": 1.2658,
+      "step": 2045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015838903603409925,
+      "loss": 1.3215,
+      "step": 2050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001581320570903984,
+      "loss": 1.2436,
+      "step": 2055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015787449699179035,
+      "loss": 1.3196,
+      "step": 2060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000157616358313142,
+      "loss": 1.3022,
+      "step": 2065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001573576436351046,
+      "loss": 1.3023,
+      "step": 2070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015709835554408765,
+      "loss": 1.2972,
+      "step": 2075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015683849663223308,
+      "loss": 1.2841,
+      "step": 2080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015657806949738947,
+      "loss": 1.2875,
+      "step": 2085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015631707674308606,
+      "loss": 1.2808,
+      "step": 2090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001560555209785066,
+      "loss": 1.2274,
+      "step": 2095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015579340481846336,
+      "loss": 1.2724,
+      "step": 2100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015553073088337094,
+      "loss": 1.2879,
+      "step": 2105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015526750179922013,
+      "loss": 1.3452,
+      "step": 2110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015500372019755168,
+      "loss": 1.3314,
+      "step": 2115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015473938871542986,
+      "loss": 1.3538,
+      "step": 2120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015447450999541616,
+      "loss": 1.2747,
+      "step": 2125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015420908668554298,
+      "loss": 1.289,
+      "step": 2130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000153943121439287,
+      "loss": 1.2945,
+      "step": 2135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001536766169155428,
+      "loss": 1.2736,
+      "step": 2140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015340957577859605,
+      "loss": 1.3336,
+      "step": 2145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015314200069809712,
+      "loss": 1.2652,
+      "step": 2150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015287389434903435,
+      "loss": 1.2656,
+      "step": 2155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015260525941170712,
+      "loss": 1.3415,
+      "step": 2160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001523360985716993,
+      "loss": 1.3093,
+      "step": 2165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015206641451985222,
+      "loss": 1.3094,
+      "step": 2170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015179620995223783,
+      "loss": 1.3243,
+      "step": 2175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015152548757013182,
+      "loss": 1.3247,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015125425007998653,
+      "loss": 1.3668,
+      "step": 2185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015098250019340387,
+      "loss": 1.3491,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015071024062710824,
+      "loss": 1.3402,
+      "step": 2195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015043747410291945,
+      "loss": 1.358,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015016420334772543,
+      "loss": 1.32,
+      "step": 2205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014989043109345498,
+      "loss": 1.3746,
+      "step": 2210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014961616007705042,
+      "loss": 1.3256,
+      "step": 2215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014934139304044033,
+      "loss": 1.355,
+      "step": 2220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014906613273051202,
+      "loss": 1.2912,
+      "step": 2225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014879038189908415,
+      "loss": 1.3153,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014851414330287928,
+      "loss": 1.3717,
+      "step": 2235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014823741970349606,
+      "loss": 1.4614,
+      "step": 2240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014796021386738193,
+      "loss": 1.3926,
+      "step": 2245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001476825285658053,
+      "loss": 1.3294,
+      "step": 2250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014740436657482777,
+      "loss": 1.3613,
+      "step": 2255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014712573067527664,
+      "loss": 1.2937,
+      "step": 2260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014684662365271675,
+      "loss": 1.2975,
+      "step": 2265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000146567048297423,
+      "loss": 1.3285,
+      "step": 2270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014628700740435221,
+      "loss": 1.3075,
+      "step": 2275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014600650377311522,
+      "loss": 1.3492,
+      "step": 2280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000145725540207949,
+      "loss": 1.2762,
+      "step": 2285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014544411951768852,
+      "loss": 1.2766,
+      "step": 2290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001451622445157387,
+      "loss": 1.2881,
+      "step": 2295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014487991802004623,
+      "loss": 1.3359,
+      "step": 2300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014459714285307152,
+      "loss": 1.3266,
+      "step": 2305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014431392184176042,
+      "loss": 1.237,
+      "step": 2310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014403025781751594,
+      "loss": 1.3363,
+      "step": 2315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014374615361616985,
+      "loss": 1.3084,
+      "step": 2320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014346161207795462,
+      "loss": 1.3463,
+      "step": 2325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014317663604747477,
+      "loss": 1.3328,
+      "step": 2330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014289122837367834,
+      "loss": 1.3092,
+      "step": 2335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014260539190982886,
+      "loss": 1.3619,
+      "step": 2340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014231912951347632,
+      "loss": 1.7222,
+      "step": 2345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014203244404642894,
+      "loss": 2.15,
+      "step": 2350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001417453383747244,
+      "loss": 1.6064,
+      "step": 2355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014145781536860122,
+      "loss": 1.5031,
+      "step": 2360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014116987790247003,
+      "loss": 1.409,
+      "step": 2365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014088152885488502,
+      "loss": 1.4162,
+      "step": 2370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001405927711085149,
+      "loss": 1.3969,
+      "step": 2375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014030360755011424,
+      "loss": 1.4638,
+      "step": 2380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014001404107049454,
+      "loss": 1.4287,
+      "step": 2385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001397240745644954,
+      "loss": 1.3563,
+      "step": 2390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013943371093095558,
+      "loss": 1.5156,
+      "step": 2395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013914295307268396,
+      "loss": 1.4148,
+      "step": 2400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001388518038964304,
+      "loss": 1.4221,
+      "step": 2405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001385602663128571,
+      "loss": 1.3772,
+      "step": 2410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000138268343236509,
+      "loss": 1.3444,
+      "step": 2415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013797603758578496,
+      "loss": 1.3119,
+      "step": 2420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013768335228290845,
+      "loss": 1.3686,
+      "step": 2425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013739029025389846,
+      "loss": 1.3505,
+      "step": 2430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013709685442854012,
+      "loss": 1.3769,
+      "step": 2435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013680304774035538,
+      "loss": 1.3505,
+      "step": 2440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013650887312657392,
+      "loss": 1.362,
+      "step": 2445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013621433352810353,
+      "loss": 1.3773,
+      "step": 2450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001359194318895008,
+      "loss": 1.325,
+      "step": 2455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013562417115894172,
+      "loss": 1.3583,
+      "step": 2460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013532855428819213,
+      "loss": 1.3345,
+      "step": 2465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013503258423257835,
+      "loss": 1.3346,
+      "step": 2470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001347362639509574,
+      "loss": 1.2946,
+      "step": 2475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001344395964056878,
+      "loss": 1.3796,
+      "step": 2480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013414258456259943,
+      "loss": 1.3362,
+      "step": 2485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001338452313909644,
+      "loss": 1.2915,
+      "step": 2490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013354753986346692,
+      "loss": 1.3044,
+      "step": 2495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013324951295617398,
+      "loss": 1.3056,
+      "step": 2500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013295115364850534,
+      "loss": 1.3265,
+      "step": 2505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013265246492320383,
+      "loss": 1.3577,
+      "step": 2510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013235344976630546,
+      "loss": 1.3357,
+      "step": 2515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013205411116710972,
+      "loss": 1.293,
+      "step": 2520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013175445211814952,
+      "loss": 1.3595,
+      "step": 2525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013145447561516138,
+      "loss": 1.3376,
+      "step": 2530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001311541846570555,
+      "loss": 1.2918,
+      "step": 2535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013085358224588565,
+      "loss": 1.3296,
+      "step": 2540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013055267138681936,
+      "loss": 1.2756,
+      "step": 2545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001302514550881076,
+      "loss": 1.3264,
+      "step": 2550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000129949936361055,
+      "loss": 1.2966,
+      "step": 2555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001296481182199896,
+      "loss": 1.2693,
+      "step": 2560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012934600368223265,
+      "loss": 1.3288,
+      "step": 2565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012904359576806858,
+      "loss": 1.258,
+      "step": 2570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012874089750071477,
+      "loss": 1.2904,
+      "step": 2575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001284379119062912,
+      "loss": 1.2678,
+      "step": 2580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012813464201379043,
+      "loss": 1.3021,
+      "step": 2585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001278310908550471,
+      "loss": 1.2584,
+      "step": 2590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012752726146470774,
+      "loss": 1.2773,
+      "step": 2595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012722315688020047,
+      "loss": 1.2542,
+      "step": 2600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001269187801417044,
+      "loss": 1.2914,
+      "step": 2605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012661413429211957,
+      "loss": 1.2748,
+      "step": 2610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001263092223770363,
+      "loss": 1.2961,
+      "step": 2615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001260040474447048,
+      "loss": 1.2932,
+      "step": 2620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001256986125460047,
+      "loss": 1.2617,
+      "step": 2625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001253929207344146,
+      "loss": 1.3035,
+      "step": 2630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012508697506598144,
+      "loss": 1.3183,
+      "step": 2635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012478077859929,
+      "loss": 1.2684,
+      "step": 2640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001244743343954324,
+      "loss": 1.3013,
+      "step": 2645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012416764551797732,
+      "loss": 1.2886,
+      "step": 2650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012386071503293962,
+      "loss": 1.297,
+      "step": 2655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001235535460087494,
+      "loss": 1.2574,
+      "step": 2660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012324614151622154,
+      "loss": 1.259,
+      "step": 2665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012293850462852496,
+      "loss": 1.2833,
+      "step": 2670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012263063842115184,
+      "loss": 1.3061,
+      "step": 2675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012232254597188688,
+      "loss": 1.2917,
+      "step": 2680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012201423036077657,
+      "loss": 1.2163,
+      "step": 2685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001217056946700984,
+      "loss": 1.2623,
+      "step": 2690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012139694198433004,
+      "loss": 1.3024,
+      "step": 2695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012108797539011847,
+      "loss": 1.3067,
+      "step": 2700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012077879797624909,
+      "loss": 1.2481,
+      "step": 2705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012046941283361502,
+      "loss": 1.2721,
+      "step": 2710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012015982305518594,
+      "loss": 1.237,
+      "step": 2715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001198500317359774,
+      "loss": 1.2805,
+      "step": 2720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011954004197301978,
+      "loss": 1.2614,
+      "step": 2725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011922985686532726,
+      "loss": 1.3051,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011891947951386701,
+      "loss": 1.2543,
+      "step": 2735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011860891302152798,
+      "loss": 1.3179,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011829816049309009,
+      "loss": 1.2996,
+      "step": 2745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011798722503519304,
+      "loss": 1.2597,
+      "step": 2750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001176761097563053,
+      "loss": 1.2888,
+      "step": 2755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 1.3059,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011705335217838909,
+      "loss": 1.2971,
+      "step": 2765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011674171610516165,
+      "loss": 1.3073,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011642991266248338,
+      "loss": 1.2934,
+      "step": 2775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011611794496750019,
+      "loss": 1.2918,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011580581613899992,
+      "loss": 1.2676,
+      "step": 2785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011549352929738142,
+      "loss": 1.2998,
+      "step": 2790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001151810875646231,
+      "loss": 1.2429,
+      "step": 2795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011486849406425188,
+      "loss": 1.2739,
+      "step": 2800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011455575192131204,
+      "loss": 1.2622,
+      "step": 2805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011424286426233368,
+      "loss": 1.2449,
+      "step": 2810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011392983421530175,
+      "loss": 1.2877,
+      "step": 2815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011361666490962468,
+      "loss": 1.3152,
+      "step": 2820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000113303359476103,
+      "loss": 1.2471,
+      "step": 2825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011298992104689825,
+      "loss": 1.2751,
+      "step": 2830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011267635275550148,
+      "loss": 1.3161,
+      "step": 2835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011236265773670196,
+      "loss": 1.2622,
+      "step": 2840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011204883912655597,
+      "loss": 1.2995,
+      "step": 2845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011173490006235528,
+      "loss": 1.2573,
+      "step": 2850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011142084368259585,
+      "loss": 1.2543,
+      "step": 2855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011110667312694653,
+      "loss": 1.2807,
+      "step": 2860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011079239153621752,
+      "loss": 1.2458,
+      "step": 2865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001104780020523291,
+      "loss": 1.2905,
+      "step": 2870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011016350781828019,
+      "loss": 1.2601,
+      "step": 2875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010984891197811687,
+      "loss": 1.2885,
+      "step": 2880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010953421767690104,
+      "loss": 1.2934,
+      "step": 2885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010921942806067886,
+      "loss": 1.2722,
+      "step": 2890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010890454627644944,
+      "loss": 1.2924,
+      "step": 2895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010858957547213327,
+      "loss": 1.3087,
+      "step": 2900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010827451879654076,
+      "loss": 1.2663,
+      "step": 2905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010795937939934088,
+      "loss": 1.2771,
+      "step": 2910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010764416043102952,
+      "loss": 1.2424,
+      "step": 2915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000107328865042898,
+      "loss": 1.2307,
+      "step": 2920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010701349638700173,
+      "loss": 1.2394,
+      "step": 2925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010669805761612854,
+      "loss": 1.2664,
+      "step": 2930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010638255188376717,
+      "loss": 1.2833,
+      "step": 2935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010606698234407586,
+      "loss": 1.2873,
+      "step": 2940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001057513521518507,
+      "loss": 1.2672,
+      "step": 2945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010543566446249408,
+      "loss": 1.252,
+      "step": 2950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010511992243198334,
+      "loss": 1.2806,
+      "step": 2955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010480412921683888,
+      "loss": 1.3012,
+      "step": 2960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000104488287974093,
+      "loss": 1.2897,
+      "step": 2965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010417240186125805,
+      "loss": 1.2662,
+      "step": 2970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010385647403629488,
+      "loss": 1.2983,
+      "step": 2975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010354050765758147,
+      "loss": 1.3074,
+      "step": 2980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010322450588388117,
+      "loss": 1.2433,
+      "step": 2985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010290847187431113,
+      "loss": 1.2837,
+      "step": 2990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010259240878831091,
+      "loss": 1.2503,
+      "step": 2995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010227631978561056,
+      "loss": 1.2471,
+      "step": 3000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010196020802619941,
+      "loss": 1.2549,
+      "step": 3005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010164407667029417,
+      "loss": 1.3157,
+      "step": 3010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010132792887830744,
+      "loss": 1.2674,
+      "step": 3015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010101176781081625,
+      "loss": 1.2781,
+      "step": 3020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010069559662853027,
+      "loss": 1.2948,
+      "step": 3025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010037941849226032,
+      "loss": 1.3088,
+      "step": 3030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010006323656288669,
+      "loss": 1.277,
+      "step": 3035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.974705400132764e-05,
+      "loss": 1.2629,
+      "step": 3040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.943087396850773e-05,
+      "loss": 1.2973,
+      "step": 3045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.911469962532627e-05,
+      "loss": 1.3478,
+      "step": 3050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.879853413262563e-05,
+      "loss": 1.2582,
+      "step": 3055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.848238065115975e-05,
+      "loss": 1.247,
+      "step": 3060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.816624234156249e-05,
+      "loss": 1.2527,
+      "step": 3065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.785012236431598e-05,
+      "loss": 1.211,
+      "step": 3070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.753402387971917e-05,
+      "loss": 1.2773,
+      "step": 3075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.721795004785605e-05,
+      "loss": 1.2478,
+      "step": 3080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.69019040285642e-05,
+      "loss": 1.2596,
+      "step": 3085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.658588898140322e-05,
+      "loss": 1.2535,
+      "step": 3090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.626990806562291e-05,
+      "loss": 1.2332,
+      "step": 3095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.595396444013205e-05,
+      "loss": 1.2232,
+      "step": 3100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.563806126346642e-05,
+      "loss": 1.2822,
+      "step": 3105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.532220169375761e-05,
+      "loss": 1.2272,
+      "step": 3110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.500638888870113e-05,
+      "loss": 1.2345,
+      "step": 3115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.469062600552509e-05,
+      "loss": 1.3178,
+      "step": 3120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.43749162009584e-05,
+      "loss": 1.248,
+      "step": 3125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.405926263119945e-05,
+      "loss": 1.2272,
+      "step": 3130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.374366845188442e-05,
+      "loss": 1.265,
+      "step": 3135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.342813681805565e-05,
+      "loss": 1.2711,
+      "step": 3140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.311267088413035e-05,
+      "loss": 1.2777,
+      "step": 3145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.27972738038688e-05,
+      "loss": 1.3019,
+      "step": 3150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.248194873034301e-05,
+      "loss": 1.2909,
+      "step": 3155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.216669881590515e-05,
+      "loss": 1.2529,
+      "step": 3160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.18515272121559e-05,
+      "loss": 1.3024,
+      "step": 3165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.153643706991318e-05,
+      "loss": 1.2295,
+      "step": 3170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.122143153918045e-05,
+      "loss": 1.2647,
+      "step": 3175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.09065137691153e-05,
+      "loss": 1.2777,
+      "step": 3180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.059168690799804e-05,
+      "loss": 1.2287,
+      "step": 3185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.027695410320004e-05,
+      "loss": 1.2618,
+      "step": 3190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.996231850115246e-05,
+      "loss": 1.2821,
+      "step": 3195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.964778324731467e-05,
+      "loss": 1.2888,
+      "step": 3200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.933335148614284e-05,
+      "loss": 1.2946,
+      "step": 3205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.901902636105854e-05,
+      "loss": 1.3121,
+      "step": 3210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.870481101441723e-05,
+      "loss": 1.3313,
+      "step": 3215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.839070858747697e-05,
+      "loss": 1.2501,
+      "step": 3220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.807672222036691e-05,
+      "loss": 1.3022,
+      "step": 3225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.77628550520559e-05,
+      "loss": 1.3413,
+      "step": 3230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.744911022032115e-05,
+      "loss": 1.2728,
+      "step": 3235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.713549086171691e-05,
+      "loss": 1.2878,
+      "step": 3240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.682200011154302e-05,
+      "loss": 1.3301,
+      "step": 3245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.650864110381357e-05,
+      "loss": 1.2421,
+      "step": 3250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.619541697122568e-05,
+      "loss": 1.3447,
+      "step": 3255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.5882330845128e-05,
+      "loss": 1.2773,
+      "step": 3260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.55693858554896e-05,
+      "loss": 1.2936,
+      "step": 3265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.525658513086857e-05,
+      "loss": 1.3005,
+      "step": 3270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.49439317983807e-05,
+      "loss": 1.2903,
+      "step": 3275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.463142898366834e-05,
+      "loss": 1.2755,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.431907981086906e-05,
+      "loss": 1.2755,
+      "step": 3285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.400688740258447e-05,
+      "loss": 1.2669,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.3694854879849e-05,
+      "loss": 1.2689,
+      "step": 3295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.33829853620986e-05,
+      "loss": 1.2619,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.307128196713972e-05,
+      "loss": 1.287,
+      "step": 3305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.275974781111804e-05,
+      "loss": 1.2392,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.244838600848727e-05,
+      "loss": 1.3307,
+      "step": 3315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.213719967197817e-05,
+      "loss": 1.281,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.182619191256724e-05,
+      "loss": 1.3068,
+      "step": 3325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.15153658394458e-05,
+      "loss": 1.2864,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.120472455998882e-05,
+      "loss": 1.2889,
+      "step": 3335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.089427117972378e-05,
+      "loss": 1.2591,
+      "step": 3340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.058400880229978e-05,
+      "loss": 1.3001,
+      "step": 3345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.027394052945648e-05,
+      "loss": 1.2743,
+      "step": 3350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.996406946099289e-05,
+      "loss": 1.271,
+      "step": 3355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.965439869473664e-05,
+      "loss": 1.2902,
+      "step": 3360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.934493132651294e-05,
+      "loss": 1.3117,
+      "step": 3365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.903567045011352e-05,
+      "loss": 1.2478,
+      "step": 3370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.872661915726584e-05,
+      "loss": 1.2871,
+      "step": 3375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.841778053760211e-05,
+      "loss": 1.2891,
+      "step": 3380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.810915767862837e-05,
+      "loss": 1.2877,
+      "step": 3385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.780075366569374e-05,
+      "loss": 1.2385,
+      "step": 3390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.749257158195943e-05,
+      "loss": 1.2439,
+      "step": 3395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.718461450836804e-05,
+      "loss": 1.331,
+      "step": 3400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.687688552361272e-05,
+      "loss": 1.2249,
+      "step": 3405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.65693877041063e-05,
+      "loss": 1.2736,
+      "step": 3410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.626212412395072e-05,
+      "loss": 1.2991,
+      "step": 3415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.595509785490617e-05,
+      "loss": 1.2505,
+      "step": 3420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.564831196636032e-05,
+      "loss": 1.3018,
+      "step": 3425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.534176952529782e-05,
+      "loss": 1.2937,
+      "step": 3430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.503547359626948e-05,
+      "loss": 1.2756,
+      "step": 3435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.472942724136174e-05,
+      "loss": 1.306,
+      "step": 3440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.442363352016598e-05,
+      "loss": 1.2915,
+      "step": 3445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.411809548974792e-05,
+      "loss": 1.3106,
+      "step": 3450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.381281620461722e-05,
+      "loss": 1.2993,
+      "step": 3455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.350779871669669e-05,
+      "loss": 1.2557,
+      "step": 3460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.3203046075292e-05,
+      "loss": 1.2596,
+      "step": 3465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.289856132706112e-05,
+      "loss": 1.2693,
+      "step": 3470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.25943475159838e-05,
+      "loss": 1.245,
+      "step": 3475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.229040768333115e-05,
+      "loss": 1.2455,
+      "step": 3480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.198674486763537e-05,
+      "loss": 1.2628,
+      "step": 3485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.168336210465928e-05,
+      "loss": 1.2572,
+      "step": 3490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.138026242736589e-05,
+      "loss": 1.232,
+      "step": 3495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.107744886588824e-05,
+      "loss": 1.3451,
+      "step": 3500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.077492444749895e-05,
+      "loss": 1.3035,
+      "step": 3505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.04726921965801e-05,
+      "loss": 1.2449,
+      "step": 3510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.017075513459292e-05,
+      "loss": 1.239,
+      "step": 3515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.986911628004753e-05,
+      "loss": 1.2471,
+      "step": 3520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.956777864847291e-05,
+      "loss": 1.3258,
+      "step": 3525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.926674525238663e-05,
+      "loss": 1.2594,
+      "step": 3530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.896601910126475e-05,
+      "loss": 1.2699,
+      "step": 3535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.866560320151179e-05,
+      "loss": 1.2516,
+      "step": 3540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.83655005564306e-05,
+      "loss": 1.2536,
+      "step": 3545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.806571416619246e-05,
+      "loss": 1.2782,
+      "step": 3550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.77662470278069e-05,
+      "loss": 1.2205,
+      "step": 3555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.74671021350919e-05,
+      "loss": 1.2888,
+      "step": 3560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.71682824786439e-05,
+      "loss": 1.2725,
+      "step": 3565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.686979104580788e-05,
+      "loss": 1.2165,
+      "step": 3570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.657163082064752e-05,
+      "loss": 1.2787,
+      "step": 3575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.627380478391543e-05,
+      "loss": 1.2569,
+      "step": 3580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.597631591302319e-05,
+      "loss": 1.2977,
+      "step": 3585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.567916718201174e-05,
+      "loss": 1.2661,
+      "step": 3590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.538236156152163e-05,
+      "loss": 1.3089,
+      "step": 3595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.508590201876317e-05,
+      "loss": 1.2441,
+      "step": 3600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.478979151748694e-05,
+      "loss": 1.2697,
+      "step": 3605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.449403301795416e-05,
+      "loss": 1.2843,
+      "step": 3610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.419862947690692e-05,
+      "loss": 1.2846,
+      "step": 3615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.390358384753881e-05,
+      "loss": 1.2265,
+      "step": 3620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.360889907946534e-05,
+      "loss": 1.2765,
+      "step": 3625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.331457811869437e-05,
+      "loss": 1.2698,
+      "step": 3630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.302062390759677e-05,
+      "loss": 1.2683,
+      "step": 3635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.272703938487694e-05,
+      "loss": 1.3128,
+      "step": 3640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.243382748554346e-05,
+      "loss": 1.2875,
+      "step": 3645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.214099114087975e-05,
+      "loss": 1.2476,
+      "step": 3650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.18485332784147e-05,
+      "loss": 1.2495,
+      "step": 3655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.155645682189351e-05,
+      "loss": 1.2578,
+      "step": 3660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.126476469124842e-05,
+      "loss": 1.2243,
+      "step": 3665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.097345980256942e-05,
+      "loss": 1.2923,
+      "step": 3670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0682545068075317e-05,
+      "loss": 1.2724,
+      "step": 3675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.039202339608432e-05,
+      "loss": 1.2479,
+      "step": 3680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.010189769098529e-05,
+      "loss": 1.2146,
+      "step": 3685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9812170853208496e-05,
+      "loss": 1.2868,
+      "step": 3690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.952284577919659e-05,
+      "loss": 1.2859,
+      "step": 3695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9233925361375864e-05,
+      "loss": 1.2669,
+      "step": 3700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8945412488127096e-05,
+      "loss": 1.2299,
+      "step": 3705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.865731004375683e-05,
+      "loss": 1.2655,
+      "step": 3710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8369620908468503e-05,
+      "loss": 1.2446,
+      "step": 3715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8082347958333625e-05,
+      "loss": 1.2017,
+      "step": 3720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7795494065262956e-05,
+      "loss": 1.2805,
+      "step": 3725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.750906209697802e-05,
+      "loss": 1.2901,
+      "step": 3730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.722305491698219e-05,
+      "loss": 1.266,
+      "step": 3735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.693747538453229e-05,
+      "loss": 1.2995,
+      "step": 3740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.665232635460971e-05,
+      "loss": 1.2705,
+      "step": 3745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6367610677892177e-05,
+      "loss": 1.2111,
+      "step": 3750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6083331200725074e-05,
+      "loss": 1.2395,
+      "step": 3755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.579949076509305e-05,
+      "loss": 1.2338,
+      "step": 3760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5516092208591594e-05,
+      "loss": 1.2674,
+      "step": 3765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5233138364398604e-05,
+      "loss": 1.2674,
+      "step": 3770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.495063206124619e-05,
+      "loss": 1.2451,
+      "step": 3775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.466857612339229e-05,
+      "loss": 1.2155,
+      "step": 3780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4386973370592485e-05,
+      "loss": 1.2747,
+      "step": 3785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.410582661807182e-05,
+      "loss": 1.2164,
+      "step": 3790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.382513867649663e-05,
+      "loss": 1.2605,
+      "step": 3795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.354491235194635e-05,
+      "loss": 1.2679,
+      "step": 3800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.32651504458857e-05,
+      "loss": 1.2327,
+      "step": 3805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.298585575513648e-05,
+      "loss": 1.287,
+      "step": 3810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2707031071849644e-05,
+      "loss": 1.2592,
+      "step": 3815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2428679183477505e-05,
+      "loss": 1.2454,
+      "step": 3820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.215080287274561e-05,
+      "loss": 1.2447,
+      "step": 3825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.18734049176252e-05,
+      "loss": 1.2778,
+      "step": 3830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.159648809130534e-05,
+      "loss": 1.2305,
+      "step": 3835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1320055162165115e-05,
+      "loss": 1.2759,
+      "step": 3840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.104410889374611e-05,
+      "loss": 1.2292,
+      "step": 3845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.076865204472454e-05,
+      "loss": 1.282,
+      "step": 3850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0493687368883904e-05,
+      "loss": 1.2767,
+      "step": 3855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.021921761508739e-05,
+      "loss": 1.2544,
+      "step": 3860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.994524552725036e-05,
+      "loss": 1.2261,
+      "step": 3865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.967177384431293e-05,
+      "loss": 1.2189,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.939880530021263e-05,
+      "loss": 1.2411,
+      "step": 3875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.912634262385695e-05,
+      "loss": 1.235,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8854388539096205e-05,
+      "loss": 1.2501,
+      "step": 3885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8582945764696244e-05,
+      "loss": 1.2903,
+      "step": 3890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.831201701431124e-05,
+      "loss": 1.2874,
+      "step": 3895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.804160499645667e-05,
+      "loss": 1.2588,
+      "step": 3900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7771712414482015e-05,
+      "loss": 1.34,
+      "step": 3905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7502341966544e-05,
+      "loss": 1.2294,
+      "step": 3910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7233496345579444e-05,
+      "loss": 1.2135,
+      "step": 3915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.696517823927842e-05,
+      "loss": 1.2082,
+      "step": 3920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6697390330057335e-05,
+      "loss": 1.2583,
+      "step": 3925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6430135295032184e-05,
+      "loss": 1.2368,
+      "step": 3930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6163415805991626e-05,
+      "loss": 1.2219,
+      "step": 3935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.589723452937049e-05,
+      "loss": 1.2668,
+      "step": 3940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5631594126222995e-05,
+      "loss": 1.2654,
+      "step": 3945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.536649725219615e-05,
+      "loss": 1.2436,
+      "step": 3950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.510194655750326e-05,
+      "loss": 1.2278,
+      "step": 3955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.483794468689728e-05,
+      "loss": 1.2904,
+      "step": 3960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.457449427964463e-05,
+      "loss": 1.277,
+      "step": 3965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.431159796949862e-05,
+      "loss": 1.2656,
+      "step": 3970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4049258384673085e-05,
+      "loss": 1.2845,
+      "step": 3975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3787478147816296e-05,
+      "loss": 1.2586,
+      "step": 3980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.352625987598467e-05,
+      "loss": 1.2354,
+      "step": 3985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.326560618061639e-05,
+      "loss": 1.2829,
+      "step": 3990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3005519667505675e-05,
+      "loss": 1.2248,
+      "step": 3995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.274600293677647e-05,
+      "loss": 1.2534,
+      "step": 4000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.248705858285649e-05,
+      "loss": 1.2016,
+      "step": 4005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.222868919445139e-05,
+      "loss": 1.2276,
+      "step": 4010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.197089735451868e-05,
+      "loss": 1.2504,
+      "step": 4015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1713685640242165e-05,
+      "loss": 1.2469,
+      "step": 4020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.145705662300595e-05,
+      "loss": 1.2582,
+      "step": 4025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1201012868368915e-05,
+      "loss": 1.2125,
+      "step": 4030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.094555693603891e-05,
+      "loss": 1.2324,
+      "step": 4035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.069069137984731e-05,
+      "loss": 1.3027,
+      "step": 4040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0436418747723295e-05,
+      "loss": 1.2772,
+      "step": 4045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0182741581668593e-05,
+      "loss": 1.2269,
+      "step": 4050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.992966241773194e-05,
+      "loss": 1.2746,
+      "step": 4055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.967718378598376e-05,
+      "loss": 1.2536,
+      "step": 4060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9425308210490905e-05,
+      "loss": 1.2511,
+      "step": 4065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.917403820929126e-05,
+      "loss": 1.2734,
+      "step": 4070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8923376294368806e-05,
+      "loss": 1.2622,
+      "step": 4075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8673324971628357e-05,
+      "loss": 1.3099,
+      "step": 4080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8423886740870566e-05,
+      "loss": 1.2349,
+      "step": 4085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.817506409576691e-05,
+      "loss": 1.2152,
+      "step": 4090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7926859523834725e-05,
+      "loss": 1.2543,
+      "step": 4095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767927550641237e-05,
+      "loss": 1.2365,
+      "step": 4100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.743231451863448e-05,
+      "loss": 1.2476,
+      "step": 4105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.718597902940717e-05,
+      "loss": 1.2462,
+      "step": 4110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.694027150138331e-05,
+      "loss": 1.2745,
+      "step": 4115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.669519439093801e-05,
+      "loss": 1.2563,
+      "step": 4120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6450750148143884e-05,
+      "loss": 1.2764,
+      "step": 4125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.620694121674679e-05,
+      "loss": 1.2162,
+      "step": 4130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.596377003414124e-05,
+      "loss": 1.2474,
+      "step": 4135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5721239031346066e-05,
+      "loss": 1.2109,
+      "step": 4140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.547935063298018e-05,
+      "loss": 1.2179,
+      "step": 4145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.523810725723816e-05,
+      "loss": 1.244,
+      "step": 4150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4997511315866306e-05,
+      "loss": 1.2279,
+      "step": 4155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.475756521413839e-05,
+      "loss": 1.2368,
+      "step": 4160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4518271350831647e-05,
+      "loss": 1.3006,
+      "step": 4165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.427963211820274e-05,
+      "loss": 1.2726,
+      "step": 4170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4041649901964e-05,
+      "loss": 1.3108,
+      "step": 4175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3804327081259304e-05,
+      "loss": 1.2491,
+      "step": 4180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.356766602864056e-05,
+      "loss": 1.2916,
+      "step": 4185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.33316691100439e-05,
+      "loss": 1.2198,
+      "step": 4190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.309633868476594e-05,
+      "loss": 1.2532,
+      "step": 4195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2861677105440336e-05,
+      "loss": 1.2575,
+      "step": 4200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.262768671801407e-05,
+      "loss": 1.2602,
+      "step": 4205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.239436986172425e-05,
+      "loss": 1.2675,
+      "step": 4210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.216172886907452e-05,
+      "loss": 1.2613,
+      "step": 4215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.192976606581186e-05,
+      "loss": 1.2542,
+      "step": 4220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1698483770903207e-05,
+      "loss": 1.2262,
+      "step": 4225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.146788429651246e-05,
+      "loss": 1.2854,
+      "step": 4230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1237969947977153e-05,
+      "loss": 1.2374,
+      "step": 4235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.100874302378559e-05,
+      "loss": 1.2252,
+      "step": 4240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.078020581555376e-05,
+      "loss": 1.2981,
+      "step": 4245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.055236060800247e-05,
+      "loss": 1.268,
+      "step": 4250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.032520967893453e-05,
+      "loss": 1.2497,
+      "step": 4255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.009875529921181e-05,
+      "loss": 1.3,
+      "step": 4260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.987299973273282e-05,
+      "loss": 1.2536,
+      "step": 4265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9647945236409848e-05,
+      "loss": 1.2717,
+      "step": 4270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.942359406014652e-05,
+      "loss": 1.2778,
+      "step": 4275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.919994844681524e-05,
+      "loss": 1.247,
+      "step": 4280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8977010632234826e-05,
+      "loss": 1.2549,
+      "step": 4285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8754782845148043e-05,
+      "loss": 1.257,
+      "step": 4290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8533267307199497e-05,
+      "loss": 1.2337,
+      "step": 4295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8312466232913282e-05,
+      "loss": 1.2896,
+      "step": 4300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.809238182967092e-05,
+      "loss": 1.264,
+      "step": 4305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7873016297689268e-05,
+      "loss": 1.2336,
+      "step": 4310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.765437182999846e-05,
+      "loss": 1.205,
+      "step": 4315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7436450612420095e-05,
+      "loss": 1.2147,
+      "step": 4320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7219254823545336e-05,
+      "loss": 1.2887,
+      "step": 4325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7002786634713094e-05,
+      "loss": 1.2237,
+      "step": 4330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.678704820998842e-05,
+      "loss": 1.2635,
+      "step": 4335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6572041706140683e-05,
+      "loss": 1.2738,
+      "step": 4340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.635776927262227e-05,
+      "loss": 1.24,
+      "step": 4345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6144233051546796e-05,
+      "loss": 1.2806,
+      "step": 4350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5931435177668006e-05,
+      "loss": 1.2598,
+      "step": 4355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.57193777783582e-05,
+      "loss": 1.1954,
+      "step": 4360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5508062973587076e-05,
+      "loss": 1.2031,
+      "step": 4365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.529749287590042e-05,
+      "loss": 1.2506,
+      "step": 4370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5087669590399178e-05,
+      "loss": 1.225,
+      "step": 4375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4878595214718236e-05,
+      "loss": 1.2695,
+      "step": 4380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4670271839005542e-05,
+      "loss": 1.2714,
+      "step": 4385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.446270154590117e-05,
+      "loss": 1.2735,
+      "step": 4390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.425588641051656e-05,
+      "loss": 1.2415,
+      "step": 4395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.404982850041363e-05,
+      "loss": 1.271,
+      "step": 4400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3844529875584278e-05,
+      "loss": 1.2445,
+      "step": 4405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3639992588429705e-05,
+      "loss": 1.2269,
+      "step": 4410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3436218683739896e-05,
+      "loss": 1.2577,
+      "step": 4415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3233210198673218e-05,
+      "loss": 1.2917,
+      "step": 4420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3030969162735926e-05,
+      "loss": 1.2436,
+      "step": 4425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.282949759776206e-05,
+      "loss": 1.2601,
+      "step": 4430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.262879751789314e-05,
+      "loss": 1.308,
+      "step": 4435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.242887092955801e-05,
+      "loss": 1.2468,
+      "step": 4440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2229719831452823e-05,
+      "loss": 1.2763,
+      "step": 4445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2031346214520966e-05,
+      "loss": 1.2405,
+      "step": 4450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.183375206193331e-05,
+      "loss": 1.2632,
+      "step": 4455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1636939349068308e-05,
+      "loss": 1.2975,
+      "step": 4460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1440910043492212e-05,
+      "loss": 1.2662,
+      "step": 4465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.12456661049394e-05,
+      "loss": 1.2363,
+      "step": 4470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.105120948529291e-05,
+      "loss": 1.2217,
+      "step": 4475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.085754212856471e-05,
+      "loss": 1.2094,
+      "step": 4480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0664665970876496e-05,
+      "loss": 1.2848,
+      "step": 4485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.04725829404402e-05,
+      "loss": 1.2706,
+      "step": 4490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.028129495753871e-05,
+      "loss": 1.2804,
+      "step": 4495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0090803934506764e-05,
+      "loss": 1.2642,
+      "step": 4500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9901111775711677e-05,
+      "loss": 1.2477,
+      "step": 4505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9712220377534496e-05,
+      "loss": 1.2972,
+      "step": 4510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9524131628350883e-05,
+      "loss": 1.2812,
+      "step": 4515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9336847408512328e-05,
+      "loss": 1.2165,
+      "step": 4520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.915036959032732e-05,
+      "loss": 1.2503,
+      "step": 4525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8964700038042626e-05,
+      "loss": 1.2767,
+      "step": 4530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8779840607824618e-05,
+      "loss": 1.2897,
+      "step": 4535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.859579314774079e-05,
+      "loss": 1.1977,
+      "step": 4540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8412559497741278e-05,
+      "loss": 1.2283,
+      "step": 4545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8230141489640394e-05,
+      "loss": 1.2615,
+      "step": 4550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.804854094709838e-05,
+      "loss": 1.2232,
+      "step": 4555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7867759685603114e-05,
+      "loss": 1.2341,
+      "step": 4560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.768779951245202e-05,
+      "loss": 1.2484,
+      "step": 4565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7508662226734006e-05,
+      "loss": 1.255,
+      "step": 4570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7330349619311415e-05,
+      "loss": 1.2868,
+      "step": 4575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7152863472802195e-05,
+      "loss": 1.2346,
+      "step": 4580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976205561561975e-05,
+      "loss": 1.2512,
+      "step": 4585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6800377651666465e-05,
+      "loss": 1.2173,
+      "step": 4590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6625381500893655e-05,
+      "loss": 1.2773,
+      "step": 4595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6451218858706374e-05,
+      "loss": 1.2677,
+      "step": 4600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6277891466234708e-05,
+      "loss": 1.2503,
+      "step": 4605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6105401056258674e-05,
+      "loss": 1.2477,
+      "step": 4610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5933749353190764e-05,
+      "loss": 1.2885,
+      "step": 4615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5762938073058853e-05,
+      "loss": 1.2545,
+      "step": 4620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.559296892348897e-05,
+      "loss": 1.3032,
+      "step": 4625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5423843603688236e-05,
+      "loss": 1.2375,
+      "step": 4630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5255563804427885e-05,
+      "loss": 1.2736,
+      "step": 4635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5088131208026367e-05,
+      "loss": 1.2613,
+      "step": 4640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4921547488332454e-05,
+      "loss": 1.2215,
+      "step": 4645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.475581431070865e-05,
+      "loss": 1.2022,
+      "step": 4650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4590933332014401e-05,
+      "loss": 1.242,
+      "step": 4655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.442690620058964e-05,
+      "loss": 1.2596,
+      "step": 4660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4263734556238263e-05,
+      "loss": 1.2308,
+      "step": 4665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4101420030211654e-05,
+      "loss": 1.2559,
+      "step": 4670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3939964245192538e-05,
+      "loss": 1.2474,
+      "step": 4675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3779368815278647e-05,
+      "loss": 1.2696,
+      "step": 4680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3619635345966641e-05,
+      "loss": 1.2504,
+      "step": 4685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3460765434136003e-05,
+      "loss": 1.2421,
+      "step": 4690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3302760668033076e-05,
+      "loss": 1.2001,
+      "step": 4695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.314562262725526e-05,
+      "loss": 1.2147,
+      "step": 4700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.298935288273515e-05,
+      "loss": 1.2805,
+      "step": 4705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2833952996724863e-05,
+      "loss": 1.257,
+      "step": 4710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2679424522780426e-05,
+      "loss": 1.2333,
+      "step": 4715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.252576900574618e-05,
+      "loss": 1.2537,
+      "step": 4720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2372987981739393e-05,
+      "loss": 1.2611,
+      "step": 4725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2221082978134935e-05,
+      "loss": 1.2067,
+      "step": 4730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2070055513549938e-05,
+      "loss": 1.2493,
+      "step": 4735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1919907097828653e-05,
+      "loss": 1.2574,
+      "step": 4740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1770639232027358e-05,
+      "loss": 1.2496,
+      "step": 4745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1622253408399341e-05,
+      "loss": 1.2225,
+      "step": 4750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1474751110379933e-05,
+      "loss": 1.2766,
+      "step": 4755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1328133812571784e-05,
+      "loss": 1.256,
+      "step": 4760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1182402980730044e-05,
+      "loss": 1.2526,
+      "step": 4765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1037560071747732e-05,
+      "loss": 1.218,
+      "step": 4770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0893606533641187e-05,
+      "loss": 1.2296,
+      "step": 4775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0750543805535518e-05,
+      "loss": 1.2528,
+      "step": 4780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0608373317650323e-05,
+      "loss": 1.2522,
+      "step": 4785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0467096491285333e-05,
+      "loss": 1.2501,
+      "step": 4790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0326714738806198e-05,
+      "loss": 1.2425,
+      "step": 4795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.01872294636304e-05,
+      "loss": 1.2714,
+      "step": 4800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0048642060213154e-05,
+      "loss": 1.2768,
+      "step": 4805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.910953914033572e-06,
+      "loss": 1.2464,
+      "step": 4810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.774166401580732e-06,
+      "loss": 1.2267,
+      "step": 4815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.638280890339945e-06,
+      "loss": 1.2463,
+      "step": 4820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.503298738779098e-06,
+      "loss": 1.2651,
+      "step": 4825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.369221296335006e-06,
+      "loss": 1.2172,
+      "step": 4830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.236049903400012e-06,
+      "loss": 1.2365,
+      "step": 4835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.103785891308547e-06,
+      "loss": 1.2667,
+      "step": 4840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.972430582323787e-06,
+      "loss": 1.2578,
+      "step": 4845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.84198528962451e-06,
+      "loss": 1.1834,
+      "step": 4850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.712451317291902e-06,
+      "loss": 1.2584,
+      "step": 4855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.58382996029652e-06,
+      "loss": 1.284,
+      "step": 4860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.456122504485397e-06,
+      "loss": 1.2217,
+      "step": 4865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.329330226569166e-06,
+      "loss": 1.227,
+      "step": 4870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.203454394109266e-06,
+      "loss": 1.2694,
+      "step": 4875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.07849626550531e-06,
+      "loss": 1.2169,
+      "step": 4880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.954457089982492e-06,
+      "loss": 1.2436,
+      "step": 4885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.831338107579056e-06,
+      "loss": 1.2592,
+      "step": 4890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.709140549133975e-06,
+      "loss": 1.2316,
+      "step": 4895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.587865636274594e-06,
+      "loss": 1.2511,
+      "step": 4900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.467514581404444e-06,
+      "loss": 1.2475,
+      "step": 4905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3480885876911156e-06,
+      "loss": 1.2627,
+      "step": 4910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.229588849054158e-06,
+      "loss": 1.2263,
+      "step": 4915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.1120165501533e-06,
+      "loss": 1.259,
+      "step": 4920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.995372866376459e-06,
+      "loss": 1.222,
+      "step": 4925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.879658963828062e-06,
+      "loss": 1.2719,
+      "step": 4930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7648759993174104e-06,
+      "loss": 1.2065,
+      "step": 4935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.651025120346988e-06,
+      "loss": 1.2355,
+      "step": 4940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.538107465101162e-06,
+      "loss": 1.2374,
+      "step": 4945
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.426124162434688e-06,
+      "loss": 1.2555,
+      "step": 4950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.3150763318614695e-06,
+      "loss": 1.2187,
+      "step": 4955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.204965083543368e-06,
+      "loss": 1.2505,
+      "step": 4960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.095791518279059e-06,
+      "loss": 1.2313,
+      "step": 4965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.987556727493049e-06,
+      "loss": 1.216,
+      "step": 4970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.880261793224828e-06,
+      "loss": 1.2232,
+      "step": 4975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.77390778811796e-06,
+      "loss": 1.2839,
+      "step": 4980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6684957754094105e-06,
+      "loss": 1.2096,
+      "step": 4985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.564026808918921e-06,
+      "loss": 1.2564,
+      "step": 4990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.460501933038442e-06,
+      "loss": 1.2488,
+      "step": 4995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.357922182721687e-06,
+      "loss": 1.2381,
+      "step": 5000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.256288583473834e-06,
+      "loss": 1.2258,
+      "step": 5005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.1556021513412544e-06,
+      "loss": 1.1936,
+      "step": 5010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.0558638929013354e-06,
+      "loss": 1.2509,
+      "step": 5015
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.957074805252437e-06,
+      "loss": 1.2375,
+      "step": 5020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.859235876003876e-06,
+      "loss": 1.2556,
+      "step": 5025
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.762348083266144e-06,
+      "loss": 1.2136,
+      "step": 5030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.666412395641062e-06,
+      "loss": 1.2863,
+      "step": 5035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5714297722121106e-06,
+      "loss": 1.243,
+      "step": 5040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.477401162534856e-06,
+      "loss": 1.2579,
+      "step": 5045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.384327506627395e-06,
+      "loss": 1.2646,
+      "step": 5050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2922097349610835e-06,
+      "loss": 1.2615,
+      "step": 5055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.20104876845111e-06,
+      "loss": 1.2481,
+      "step": 5060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.110845518447348e-06,
+      "loss": 1.2763,
+      "step": 5065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.021600886725263e-06,
+      "loss": 1.2249,
+      "step": 5070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.933315765476808e-06,
+      "loss": 1.2437,
+      "step": 5075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.84599103730161e-06,
+      "loss": 1.2199,
+      "step": 5080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.75962757519811e-06,
+      "loss": 1.2349,
+      "step": 5085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6742262425548125e-06,
+      "loss": 1.2319,
+      "step": 5090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5897878931416497e-06,
+      "loss": 1.2168,
+      "step": 5095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5063133711014882e-06,
+      "loss": 1.2151,
+      "step": 5100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4238035109416922e-06,
+      "loss": 1.2785,
+      "step": 5105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.342259137525694e-06,
+      "loss": 1.2178,
+      "step": 5110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2616810660648588e-06,
+      "loss": 1.2742,
+      "step": 5115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.182070102110257e-06,
+      "loss": 1.2287,
+      "step": 5120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.103427041544682e-06,
+      "loss": 1.2327,
+      "step": 5125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.025752670574622e-06,
+      "loss": 1.2371,
+      "step": 5130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9490477657224014e-06,
+      "loss": 1.2718,
+      "step": 5135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.873313093818486e-06,
+      "loss": 1.2082,
+      "step": 5140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.798549411993789e-06,
+      "loss": 1.1962,
+      "step": 5145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7247574676720454e-06,
+      "loss": 1.2537,
+      "step": 5150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.651937998562437e-06,
+      "loss": 1.2331,
+      "step": 5155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.580091732652101e-06,
+      "loss": 1.2437,
+      "step": 5160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.509219388198958e-06,
+      "loss": 1.2562,
+      "step": 5165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.439321673724504e-06,
+      "loss": 1.2432,
+      "step": 5170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3703992880066638e-06,
+      "loss": 1.2832,
+      "step": 5175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.302452920072895e-06,
+      "loss": 1.2662,
+      "step": 5180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2354832491932486e-06,
+      "loss": 1.2605,
+      "step": 5185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1694909448735645e-06,
+      "loss": 1.2612,
+      "step": 5190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1044766668488424e-06,
+      "loss": 1.2727,
+      "step": 5195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0404410650765817e-06,
+      "loss": 1.2572,
+      "step": 5200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.977384779730307e-06,
+      "loss": 1.2633,
+      "step": 5205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9153084411931863e-06,
+      "loss": 1.229,
+      "step": 5210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8542126700516804e-06,
+      "loss": 1.2002,
+      "step": 5215
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7940980770894122e-06,
+      "loss": 1.2471,
+      "step": 5220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7349652632809744e-06,
+      "loss": 1.257,
+      "step": 5225
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6768148197860212e-06,
+      "loss": 1.2562,
+      "step": 5230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6196473279432412e-06,
+      "loss": 1.2322,
+      "step": 5235
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5634633592646609e-06,
+      "loss": 1.225,
+      "step": 5240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5082634754298385e-06,
+      "loss": 1.2315,
+      "step": 5245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4540482282803137e-06,
+      "loss": 1.2741,
+      "step": 5250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4008181598140547e-06,
+      "loss": 1.2291,
+      "step": 5255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.348573802180053e-06,
+      "loss": 1.2508,
+      "step": 5260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2973156776729944e-06,
+      "loss": 1.2438,
+      "step": 5265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2470442987280617e-06,
+      "loss": 1.2337,
+      "step": 5270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1977601679157625e-06,
+      "loss": 1.2373,
+      "step": 5275
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1494637779369766e-06,
+      "loss": 1.2654,
+      "step": 5280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.10215561161795e-06,
+      "loss": 1.2273,
+      "step": 5285
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.055836141905553e-06,
+      "loss": 1.2503,
+      "step": 5290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0105058318624517e-06,
+      "loss": 1.2419,
+      "step": 5295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.661651346625889e-07,
+      "loss": 1.2124,
+      "step": 5300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.228144935865657e-07,
+      "loss": 1.276,
+      "step": 5305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.804543420172562e-07,
+      "loss": 1.1957,
+      "step": 5310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.390851034354552e-07,
+      "loss": 1.2669,
+      "step": 5315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.987071914156596e-07,
+      "loss": 1.2554,
+      "step": 5320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.593210096219161e-07,
+      "loss": 1.2942,
+      "step": 5325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.20926951803802e-07,
+      "loss": 1.2459,
+      "step": 5330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.835254017924953e-07,
+      "loss": 1.213,
+      "step": 5335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.471167334968886e-07,
+      "loss": 1.2581,
+      "step": 5340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.117013108999037e-07,
+      "loss": 1.2228,
+      "step": 5345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.772794880548715e-07,
+      "loss": 1.2393,
+      "step": 5350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.438516090819024e-07,
+      "loss": 1.2756,
+      "step": 5355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.114180081645214e-07,
+      "loss": 1.2465,
+      "step": 5360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.799790095463164e-07,
+      "loss": 1.2646,
+      "step": 5365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.495349275276839e-07,
+      "loss": 1.254,
+      "step": 5370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.200860664626882e-07,
+      "loss": 1.2278,
+      "step": 5375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9163272075599664e-07,
+      "loss": 1.2163,
+      "step": 5380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.641751748600042e-07,
+      "loss": 1.2457,
+      "step": 5385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3771370327190246e-07,
+      "loss": 1.2151,
+      "step": 5390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.122485705310041e-07,
+      "loss": 1.1919,
+      "step": 5395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.877800312160783e-07,
+      "loss": 1.2623,
+      "step": 5400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.643083299427751e-07,
+      "loss": 1.2373,
+      "step": 5405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4183370136121595e-07,
+      "loss": 1.2287,
+      "step": 5410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2035637015365152e-07,
+      "loss": 1.2602,
+      "step": 5415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9987655103217428e-07,
+      "loss": 1.231,
+      "step": 5420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8039444873663158e-07,
+      "loss": 1.1957,
+      "step": 5425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6191025803250492e-07,
+      "loss": 1.2533,
+      "step": 5430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.444241637090338e-07,
+      "loss": 1.2552,
+      "step": 5435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2793634057732818e-07,
+      "loss": 1.2871,
+      "step": 5440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1244695346864786e-07,
+      "loss": 1.2085,
+      "step": 5445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.795615723270369e-08,
+      "loss": 1.2469,
+      "step": 5450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.446409673615874e-08,
+      "loss": 1.2277,
+      "step": 5455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.197090686119623e-08,
+      "loss": 1.2988,
+      "step": 5460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.047671250408726e-08,
+      "loss": 1.2706,
+      "step": 5465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.998162857402511e-08,
+      "loss": 1.2576,
+      "step": 5470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0485759991937264e-08,
+      "loss": 1.2001,
+      "step": 5475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1989201689452967e-08,
+      "loss": 1.2299,
+      "step": 5480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4492038607948353e-08,
+      "loss": 1.2341,
+      "step": 5485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7994345697680547e-08,
+      "loss": 1.2108,
+      "step": 5490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2496187917065972e-08,
+      "loss": 1.2237,
+      "step": 5495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.997620232014225e-09,
+      "loss": 1.3129,
+      "step": 5500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.498687615372976e-09,
+      "loss": 1.1938,
+      "step": 5505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999425046506076e-09,
+      "loss": 1.224,
+      "step": 5510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.998575109160797e-10,
+      "loss": 1.254,
+      "step": 5515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.2513,
+      "step": 5520
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.2233707904815674,
+      "eval_runtime": 1795.568,
+      "eval_samples_per_second": 16.333,
+      "eval_steps_per_second": 1.361,
+      "step": 5520
+    },
+    {
+      "epoch": 1.0,
+      "step": 5520,
+      "total_flos": 3958946251407360.0,
+      "train_loss": 1.269419441966043,
+      "train_runtime": 32984.8618,
+      "train_samples_per_second": 8.033,
+      "train_steps_per_second": 0.167
     }
   ],
   "logging_steps": 5,
-  "max_steps": 111,
+  "max_steps": 5520,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,
-  "total_flos": 79541494087680.0,
+  "total_flos": 3958946251407360.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null