diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17687 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.999277804525758,
+  "eval_steps": 500,
+  "global_step": 12459,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001203659123736158,
+      "grad_norm": 4.625,
+      "learning_rate": 2.065577232275662e-05,
+      "loss": 1.9428,
+      "step": 5
+    },
+    {
+      "epoch": 0.002407318247472316,
+      "grad_norm": 3.890625,
+      "learning_rate": 4.647548772620239e-05,
+      "loss": 1.8343,
+      "step": 10
+    },
+    {
+      "epoch": 0.0036109773712084737,
+      "grad_norm": 3.09375,
+      "learning_rate": 7.229520312964818e-05,
+      "loss": 1.7166,
+      "step": 15
+    },
+    {
+      "epoch": 0.004814636494944632,
+      "grad_norm": 3.078125,
+      "learning_rate": 9.811491853309394e-05,
+      "loss": 1.5911,
+      "step": 20
+    },
+    {
+      "epoch": 0.00601829561868079,
+      "grad_norm": 3.734375,
+      "learning_rate": 0.00012393463393653973,
+      "loss": 1.5328,
+      "step": 25
+    },
+    {
+      "epoch": 0.007221954742416947,
+      "grad_norm": 3.09375,
+      "learning_rate": 0.0001497543493399855,
+      "loss": 1.4457,
+      "step": 30
+    },
+    {
+      "epoch": 0.008425613866153106,
+      "grad_norm": 2.6875,
+      "learning_rate": 0.00017557406474343128,
+      "loss": 1.3968,
+      "step": 35
+    },
+    {
+      "epoch": 0.009629272989889264,
+      "grad_norm": 2.921875,
+      "learning_rate": 0.0001807379731713583,
+      "loss": 1.3355,
+      "step": 40
+    },
+    {
+      "epoch": 0.010832932113625422,
+      "grad_norm": 3.484375,
+      "learning_rate": 0.00018073783239457288,
+      "loss": 1.3258,
+      "step": 45
+    },
+    {
+      "epoch": 0.01203659123736158,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.00018073758332819127,
+      "loss": 1.2706,
+      "step": 50
+    },
+    {
+      "epoch": 0.013240250361097737,
+      "grad_norm": 2.921875,
+      "learning_rate": 0.00018073722597261146,
+      "loss": 1.2635,
+      "step": 55
+    },
+    {
+      "epoch": 0.014443909484833895,
+      "grad_norm": 2.8125,
+      "learning_rate": 0.00018073676032840438,
+      "loss": 1.2178,
+      "step": 60
+    },
+    {
+      "epoch": 0.015647568608570053,
+      "grad_norm": 2.796875,
+      "learning_rate": 0.00018073618639631402,
+      "loss": 1.2227,
+      "step": 65
+    },
+    {
+      "epoch": 0.016851227732306212,
+      "grad_norm": 2.8125,
+      "learning_rate": 0.00018073550417725735,
+      "loss": 1.1902,
+      "step": 70
+    },
+    {
+      "epoch": 0.018054886856042368,
+      "grad_norm": 2.453125,
+      "learning_rate": 0.0001807347136723244,
+      "loss": 1.1803,
+      "step": 75
+    },
+    {
+      "epoch": 0.019258545979778528,
+      "grad_norm": 2.609375,
+      "learning_rate": 0.00018073381488277823,
+      "loss": 1.1605,
+      "step": 80
+    },
+    {
+      "epoch": 0.020462205103514684,
+      "grad_norm": 2.453125,
+      "learning_rate": 0.00018073280781005481,
+      "loss": 1.1619,
+      "step": 85
+    },
+    {
+      "epoch": 0.021665864227250843,
+      "grad_norm": 2.625,
+      "learning_rate": 0.00018073169245576325,
+      "loss": 1.1545,
+      "step": 90
+    },
+    {
+      "epoch": 0.022869523350987,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00018073046882168553,
+      "loss": 1.13,
+      "step": 95
+    },
+    {
+      "epoch": 0.02407318247472316,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.00018072913690977675,
+      "loss": 1.121,
+      "step": 100
+    },
+    {
+      "epoch": 0.025276841598459315,
+      "grad_norm": 2.5625,
+      "learning_rate": 0.00018072769672216498,
+      "loss": 1.0923,
+      "step": 105
+    },
+    {
+      "epoch": 0.026480500722195474,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.0001807261482611512,
+      "loss": 1.1263,
+      "step": 110
+    },
+    {
+      "epoch": 0.027684159845931634,
+      "grad_norm": 2.84375,
+      "learning_rate": 0.00018072449152920953,
+      "loss": 1.1257,
+      "step": 115
+    },
+    {
+      "epoch": 0.02888781896966779,
+      "grad_norm": 2.875,
+      "learning_rate": 0.00018072272652898695,
+      "loss": 1.0985,
+      "step": 120
+    },
+    {
+      "epoch": 0.03009147809340395,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.0001807208532633035,
+      "loss": 1.0515,
+      "step": 125
+    },
+    {
+      "epoch": 0.031295137217140105,
+      "grad_norm": 2.8125,
+      "learning_rate": 0.0001807188717351522,
+      "loss": 1.1238,
+      "step": 130
+    },
+    {
+      "epoch": 0.03249879634087626,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00018071678194769898,
+      "loss": 1.0551,
+      "step": 135
+    },
+    {
+      "epoch": 0.033702455464612424,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.0001807145839042828,
+      "loss": 1.0581,
+      "step": 140
+    },
+    {
+      "epoch": 0.03490611458834858,
+      "grad_norm": 2.890625,
+      "learning_rate": 0.0001807122776084156,
+      "loss": 1.0491,
+      "step": 145
+    },
+    {
+      "epoch": 0.036109773712084736,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00018070986306378223,
+      "loss": 1.0508,
+      "step": 150
+    },
+    {
+      "epoch": 0.03731343283582089,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00018070734027424048,
+      "loss": 1.053,
+      "step": 155
+    },
+    {
+      "epoch": 0.038517091959557055,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00018070470924382115,
+      "loss": 1.0778,
+      "step": 160
+    },
+    {
+      "epoch": 0.03972075108329321,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00018070196997672797,
+      "loss": 1.0409,
+      "step": 165
+    },
+    {
+      "epoch": 0.04092441020702937,
+      "grad_norm": 2.5,
+      "learning_rate": 0.00018069912247733758,
+      "loss": 1.0466,
+      "step": 170
+    },
+    {
+      "epoch": 0.04212806933076553,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00018069616675019952,
+      "loss": 1.0337,
+      "step": 175
+    },
+    {
+      "epoch": 0.043331728454501686,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.00018069310280003633,
+      "loss": 1.0368,
+      "step": 180
+    },
+    {
+      "epoch": 0.04453538757823784,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.00018068993063174337,
+      "loss": 0.9803,
+      "step": 185
+    },
+    {
+      "epoch": 0.045739046701974,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00018068665025038899,
+      "loss": 1.0449,
+      "step": 190
+    },
+    {
+      "epoch": 0.04694270582571016,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00018068326166121437,
+      "loss": 1.0,
+      "step": 195
+    },
+    {
+      "epoch": 0.04814636494944632,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00018067976486963364,
+      "loss": 0.992,
+      "step": 200
+    },
+    {
+      "epoch": 0.04935002407318247,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.00018067615988123374,
+      "loss": 0.9909,
+      "step": 205
+    },
+    {
+      "epoch": 0.05055368319691863,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00018067244670177452,
+      "loss": 0.9914,
+      "step": 210
+    },
+    {
+      "epoch": 0.05175734232065479,
+      "grad_norm": 2.34375,
+      "learning_rate": 0.00018066862533718873,
+      "loss": 0.9937,
+      "step": 215
+    },
+    {
+      "epoch": 0.05296100144439095,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.0001806646957935819,
+      "loss": 0.9911,
+      "step": 220
+    },
+    {
+      "epoch": 0.054164660568127104,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.00018066065807723243,
+      "loss": 0.9876,
+      "step": 225
+    },
+    {
+      "epoch": 0.05536831969186327,
+      "grad_norm": 2.34375,
+      "learning_rate": 0.00018065651219459158,
+      "loss": 0.9808,
+      "step": 230
+    },
+    {
+      "epoch": 0.05657197881559942,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.00018065225815228335,
+      "loss": 0.9586,
+      "step": 235
+    },
+    {
+      "epoch": 0.05777563793933558,
+      "grad_norm": 2.125,
+      "learning_rate": 0.00018064789595710468,
+      "loss": 0.9513,
+      "step": 240
+    },
+    {
+      "epoch": 0.058979297063071735,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00018064342561602522,
+      "loss": 0.9707,
+      "step": 245
+    },
+    {
+      "epoch": 0.0601829561868079,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00018063884713618737,
+      "loss": 0.9642,
+      "step": 250
+    },
+    {
+      "epoch": 0.061386615310544054,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.00018063416052490648,
+      "loss": 0.9899,
+      "step": 255
+    },
+    {
+      "epoch": 0.06259027443428021,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00018062936578967044,
+      "loss": 0.9258,
+      "step": 260
+    },
+    {
+      "epoch": 0.06379393355801637,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.00018062446293814008,
+      "loss": 0.9583,
+      "step": 265
+    },
+    {
+      "epoch": 0.06499759268175252,
+      "grad_norm": 2.125,
+      "learning_rate": 0.0001806194519781489,
+      "loss": 0.9433,
+      "step": 270
+    },
+    {
+      "epoch": 0.06620125180548869,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.00018061433291770306,
+      "loss": 0.9476,
+      "step": 275
+    },
+    {
+      "epoch": 0.06740491092922485,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00018060910576498158,
+      "loss": 0.9403,
+      "step": 280
+    },
+    {
+      "epoch": 0.068608570052961,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.0001806037705283361,
+      "loss": 0.9219,
+      "step": 285
+    },
+    {
+      "epoch": 0.06981222917669716,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.0001805983272162909,
+      "loss": 0.9263,
+      "step": 290
+    },
+    {
+      "epoch": 0.07101588830043332,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00018059277583754304,
+      "loss": 0.9205,
+      "step": 295
+    },
+    {
+      "epoch": 0.07221954742416947,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00018058711640096223,
+      "loss": 0.8964,
+      "step": 300
+    },
+    {
+      "epoch": 0.07342320654790563,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00018058134891559078,
+      "loss": 0.9138,
+      "step": 305
+    },
+    {
+      "epoch": 0.07462686567164178,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00018057547339064362,
+      "loss": 0.9659,
+      "step": 310
+    },
+    {
+      "epoch": 0.07583052479537795,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00018056948983550834,
+      "loss": 0.8931,
+      "step": 315
+    },
+    {
+      "epoch": 0.07703418391911411,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.00018056339825974518,
+      "loss": 0.9025,
+      "step": 320
+    },
+    {
+      "epoch": 0.07823784304285027,
+      "grad_norm": 1.921875,
+      "learning_rate": 0.00018055719867308685,
+      "loss": 0.9069,
+      "step": 325
+    },
+    {
+      "epoch": 0.07944150216658642,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.00018055089108543872,
+      "loss": 0.9294,
+      "step": 330
+    },
+    {
+      "epoch": 0.08064516129032258,
+      "grad_norm": 2.453125,
+      "learning_rate": 0.00018054447550687873,
+      "loss": 0.9123,
+      "step": 335
+    },
+    {
+      "epoch": 0.08184882041405873,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.00018053795194765732,
+      "loss": 0.9084,
+      "step": 340
+    },
+    {
+      "epoch": 0.08305247953779489,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.00018053132041819745,
+      "loss": 0.8883,
+      "step": 345
+    },
+    {
+      "epoch": 0.08425613866153106,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.00018052458092909456,
+      "loss": 0.9296,
+      "step": 350
+    },
+    {
+      "epoch": 0.08545979778526722,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00018051773349111671,
+      "loss": 0.8994,
+      "step": 355
+    },
+    {
+      "epoch": 0.08666345690900337,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00018051077811520431,
+      "loss": 0.9062,
+      "step": 360
+    },
+    {
+      "epoch": 0.08786711603273953,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.00018050371481247027,
+      "loss": 0.9067,
+      "step": 365
+    },
+    {
+      "epoch": 0.08907077515647568,
+      "grad_norm": 1.828125,
+      "learning_rate": 0.00018049654359419994,
+      "loss": 0.902,
+      "step": 370
+    },
+    {
+      "epoch": 0.09027443428021184,
+      "grad_norm": 2.125,
+      "learning_rate": 0.00018048926447185106,
+      "loss": 0.9134,
+      "step": 375
+    },
+    {
+      "epoch": 0.091478093403948,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.00018048187745705387,
+      "loss": 0.885,
+      "step": 380
+    },
+    {
+      "epoch": 0.09268175252768417,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.00018047438256161086,
+      "loss": 0.9,
+      "step": 385
+    },
+    {
+      "epoch": 0.09388541165142032,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.00018046677979749698,
+      "loss": 0.8892,
+      "step": 390
+    },
+    {
+      "epoch": 0.09508907077515648,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.00018045906917685947,
+      "loss": 0.8873,
+      "step": 395
+    },
+    {
+      "epoch": 0.09629272989889263,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.000180451250712018,
+      "loss": 0.882,
+      "step": 400
+    },
+    {
+      "epoch": 0.09749638902262879,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.00018044332441546437,
+      "loss": 0.8877,
+      "step": 405
+    },
+    {
+      "epoch": 0.09870004814636495,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00018043529029986285,
+      "loss": 0.8968,
+      "step": 410
+    },
+    {
+      "epoch": 0.0999037072701011,
+      "grad_norm": 1.9375,
+      "learning_rate": 0.00018042714837804985,
+      "loss": 0.9009,
+      "step": 415
+    },
+    {
+      "epoch": 0.10110736639383726,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.0001804188986630341,
+      "loss": 0.8936,
+      "step": 420
+    },
+    {
+      "epoch": 0.10231102551757343,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.00018041054116799653,
+      "loss": 0.8864,
+      "step": 425
+    },
+    {
+      "epoch": 0.10351468464130958,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00018040207590629026,
+      "loss": 0.8341,
+      "step": 430
+    },
+    {
+      "epoch": 0.10471834376504574,
+      "grad_norm": 1.8203125,
+      "learning_rate": 0.0001803935028914406,
+      "loss": 0.8809,
+      "step": 435
+    },
+    {
+      "epoch": 0.1059220028887819,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00018038482213714508,
+      "loss": 0.8482,
+      "step": 440
+    },
+    {
+      "epoch": 0.10712566201251805,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00018037603365727323,
+      "loss": 0.8869,
+      "step": 445
+    },
+    {
+      "epoch": 0.10832932113625421,
+      "grad_norm": 1.9765625,
+      "learning_rate": 0.00018036713746586689,
+      "loss": 0.8903,
+      "step": 450
+    },
+    {
+      "epoch": 0.10953298025999036,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.00018035813357713984,
+      "loss": 0.8944,
+      "step": 455
+    },
+    {
+      "epoch": 0.11073663938372653,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.00018034902200547796,
+      "loss": 0.8577,
+      "step": 460
+    },
+    {
+      "epoch": 0.11194029850746269,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.00018033980276543928,
+      "loss": 0.8499,
+      "step": 465
+    },
+    {
+      "epoch": 0.11314395763119885,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.0001803304758717537,
+      "loss": 0.8646,
+      "step": 470
+    },
+    {
+      "epoch": 0.114347616754935,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00018032104133932326,
+      "loss": 0.8994,
+      "step": 475
+    },
+    {
+      "epoch": 0.11555127587867116,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00018031149918322191,
+      "loss": 0.8491,
+      "step": 480
+    },
+    {
+      "epoch": 0.11675493500240731,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.0001803018494186956,
+      "loss": 0.8594,
+      "step": 485
+    },
+    {
+      "epoch": 0.11795859412614347,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.0001802920920611621,
+      "loss": 0.8103,
+      "step": 490
+    },
+    {
+      "epoch": 0.11916225324987964,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00018028222712621126,
+      "loss": 0.8405,
+      "step": 495
+    },
+    {
+      "epoch": 0.1203659123736158,
+      "grad_norm": 1.8984375,
+      "learning_rate": 0.00018027225462960463,
+      "loss": 0.8522,
+      "step": 500
+    },
+    {
+      "epoch": 0.1203659123736158,
+      "eval_loss": 0.7550991773605347,
+      "eval_runtime": 2.3423,
+      "eval_samples_per_second": 85.387,
+      "eval_steps_per_second": 85.387,
+      "step": 500
+    },
+    {
+      "epoch": 0.12156957149735195,
+      "grad_norm": 2.0,
+      "learning_rate": 0.00018026217458727575,
+      "loss": 0.8116,
+      "step": 505
+    },
+    {
+      "epoch": 0.12277323062108811,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00018025198701532993,
+      "loss": 0.8449,
+      "step": 510
+    },
+    {
+      "epoch": 0.12397688974482426,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00018024169193004433,
+      "loss": 0.8823,
+      "step": 515
+    },
+    {
+      "epoch": 0.12518054886856042,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00018023128934786784,
+      "loss": 0.8989,
+      "step": 520
+    },
+    {
+      "epoch": 0.12638420799229658,
+      "grad_norm": 2.0,
+      "learning_rate": 0.0001802207792854211,
+      "loss": 0.8832,
+      "step": 525
+    },
+    {
+      "epoch": 0.12758786711603273,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00018021016175949651,
+      "loss": 0.8406,
+      "step": 530
+    },
+    {
+      "epoch": 0.1287915262397689,
+      "grad_norm": 1.9296875,
+      "learning_rate": 0.00018019943678705816,
+      "loss": 0.8324,
+      "step": 535
+    },
+    {
+      "epoch": 0.12999518536350504,
+      "grad_norm": 1.8203125,
+      "learning_rate": 0.00018018860438524177,
+      "loss": 0.8331,
+      "step": 540
+    },
+    {
+      "epoch": 0.1311988444872412,
+      "grad_norm": 1.9296875,
+      "learning_rate": 0.00018017766457135482,
+      "loss": 0.8508,
+      "step": 545
+    },
+    {
+      "epoch": 0.13240250361097738,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.00018016661736287623,
+      "loss": 0.8103,
+      "step": 550
+    },
+    {
+      "epoch": 0.13360616273471354,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0001801554627774567,
+      "loss": 0.849,
+      "step": 555
+    },
+    {
+      "epoch": 0.1348098218584497,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.0001801442008329183,
+      "loss": 0.8389,
+      "step": 560
+    },
+    {
+      "epoch": 0.13601348098218585,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.00018013283154725486,
+      "loss": 0.8162,
+      "step": 565
+    },
+    {
+      "epoch": 0.137217140105922,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00018012135493863146,
+      "loss": 0.8182,
+      "step": 570
+    },
+    {
+      "epoch": 0.13842079922965816,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0001801097710253848,
+      "loss": 0.7734,
+      "step": 575
+    },
+    {
+      "epoch": 0.13962445835339432,
+      "grad_norm": 1.8671875,
+      "learning_rate": 0.00018009807982602308,
+      "loss": 0.8289,
+      "step": 580
+    },
+    {
+      "epoch": 0.14082811747713048,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.00018008628135922578,
+      "loss": 0.8231,
+      "step": 585
+    },
+    {
+      "epoch": 0.14203177660086663,
+      "grad_norm": 1.921875,
+      "learning_rate": 0.00018007437564384381,
+      "loss": 0.8323,
+      "step": 590
+    },
+    {
+      "epoch": 0.1432354357246028,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.00018006236269889952,
+      "loss": 0.8316,
+      "step": 595
+    },
+    {
+      "epoch": 0.14443909484833894,
+      "grad_norm": 1.9140625,
+      "learning_rate": 0.00018005024254358646,
+      "loss": 0.8063,
+      "step": 600
+    },
+    {
+      "epoch": 0.1456427539720751,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00018003801519726954,
+      "loss": 0.8243,
+      "step": 605
+    },
+    {
+      "epoch": 0.14684641309581126,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.00018002568067948494,
+      "loss": 0.8155,
+      "step": 610
+    },
+    {
+      "epoch": 0.1480500722195474,
+      "grad_norm": 1.8046875,
+      "learning_rate": 0.00018001323900994,
+      "loss": 0.8342,
+      "step": 615
+    },
+    {
+      "epoch": 0.14925373134328357,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.00018000069020851343,
+      "loss": 0.8569,
+      "step": 620
+    },
+    {
+      "epoch": 0.15045739046701975,
+      "grad_norm": 1.96875,
+      "learning_rate": 0.00017998803429525488,
+      "loss": 0.8212,
+      "step": 625
+    },
+    {
+      "epoch": 0.1516610495907559,
+      "grad_norm": 1.828125,
+      "learning_rate": 0.00017997527129038534,
+      "loss": 0.8478,
+      "step": 630
+    },
+    {
+      "epoch": 0.15286470871449206,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.00017996240121429677,
+      "loss": 0.8347,
+      "step": 635
+    },
+    {
+      "epoch": 0.15406836783822822,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0001799494240875523,
+      "loss": 0.8067,
+      "step": 640
+    },
+    {
+      "epoch": 0.15527202696196438,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.000179936339930886,
+      "loss": 0.8296,
+      "step": 645
+    },
+    {
+      "epoch": 0.15647568608570053,
+      "grad_norm": 1.6640625,
+      "learning_rate": 0.000179923148765203,
+      "loss": 0.8329,
+      "step": 650
+    },
+    {
+      "epoch": 0.1576793452094367,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00017990985061157948,
+      "loss": 0.8557,
+      "step": 655
+    },
+    {
+      "epoch": 0.15888300433317284,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00017989644549126234,
+      "loss": 0.8177,
+      "step": 660
+    },
+    {
+      "epoch": 0.160086663456909,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00017988293342566965,
+      "loss": 0.8396,
+      "step": 665
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.0001798693144363902,
+      "loss": 0.8214,
+      "step": 670
+    },
+    {
+      "epoch": 0.1624939817043813,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.00017985558854518362,
+      "loss": 0.789,
+      "step": 675
+    },
+    {
+      "epoch": 0.16369764082811747,
+      "grad_norm": 1.921875,
+      "learning_rate": 0.00017984175577398037,
+      "loss": 0.8046,
+      "step": 680
+    },
+    {
+      "epoch": 0.16490129995185362,
+      "grad_norm": 1.828125,
+      "learning_rate": 0.0001798278161448817,
+      "loss": 0.8014,
+      "step": 685
+    },
+    {
+      "epoch": 0.16610495907558978,
+      "grad_norm": 1.8203125,
+      "learning_rate": 0.0001798137696801595,
+      "loss": 0.8043,
+      "step": 690
+    },
+    {
+      "epoch": 0.16730861819932596,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.00017979961640225648,
+      "loss": 0.7911,
+      "step": 695
+    },
+    {
+      "epoch": 0.16851227732306212,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00017978535633378595,
+      "loss": 0.7907,
+      "step": 700
+    },
+    {
+      "epoch": 0.16971593644679828,
+      "grad_norm": 1.96875,
+      "learning_rate": 0.00017977098949753179,
+      "loss": 0.8338,
+      "step": 705
+    },
+    {
+      "epoch": 0.17091959557053443,
+      "grad_norm": 1.9375,
+      "learning_rate": 0.00017975651591644855,
+      "loss": 0.8179,
+      "step": 710
+    },
+    {
+      "epoch": 0.1721232546942706,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00017974193561366133,
+      "loss": 0.8143,
+      "step": 715
+    },
+    {
+      "epoch": 0.17332691381800674,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.0001797272486124657,
+      "loss": 0.7901,
+      "step": 720
+    },
+    {
+      "epoch": 0.1745305729417429,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.0001797124549363277,
+      "loss": 0.8494,
+      "step": 725
+    },
+    {
+      "epoch": 0.17573423206547906,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00017969755460888387,
+      "loss": 0.8067,
+      "step": 730
+    },
+    {
+      "epoch": 0.1769378911892152,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.00017968254765394107,
+      "loss": 0.8148,
+      "step": 735
+    },
+    {
+      "epoch": 0.17814155031295137,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0001796674340954766,
+      "loss": 0.7815,
+      "step": 740
+    },
+    {
+      "epoch": 0.17934520943668752,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.00017965221395763802,
+      "loss": 0.7736,
+      "step": 745
+    },
+    {
+      "epoch": 0.18054886856042368,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00017963688726474326,
+      "loss": 0.7575,
+      "step": 750
+    },
+    {
+      "epoch": 0.18175252768415984,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00017962145404128038,
+      "loss": 0.7874,
+      "step": 755
+    },
+    {
+      "epoch": 0.182956186807896,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.0001796059143119078,
+      "loss": 0.7536,
+      "step": 760
+    },
+    {
+      "epoch": 0.18415984593163215,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0001795902681014539,
+      "loss": 0.7967,
+      "step": 765
+    },
+    {
+      "epoch": 0.18536350505536833,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00017957451543491746,
+      "loss": 0.7517,
+      "step": 770
+    },
+    {
+      "epoch": 0.1865671641791045,
+      "grad_norm": 1.6640625,
+      "learning_rate": 0.00017955865633746711,
+      "loss": 0.8332,
+      "step": 775
+    },
+    {
+      "epoch": 0.18777082330284064,
+      "grad_norm": 1.9921875,
+      "learning_rate": 0.00017954269083444164,
+      "loss": 0.7853,
+      "step": 780
+    },
+    {
+      "epoch": 0.1889744824265768,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0001795266189513498,
+      "loss": 0.8002,
+      "step": 785
+    },
+    {
+      "epoch": 0.19017814155031296,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.00017951044071387042,
+      "loss": 0.779,
+      "step": 790
+    },
+    {
+      "epoch": 0.1913818006740491,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00017949415614785212,
+      "loss": 0.7482,
+      "step": 795
+    },
+    {
+      "epoch": 0.19258545979778527,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00017947776527931346,
+      "loss": 0.7858,
+      "step": 800
+    },
+    {
+      "epoch": 0.19378911892152142,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00017946126813444285,
+      "loss": 0.7877,
+      "step": 805
+    },
+    {
+      "epoch": 0.19499277804525758,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00017944466473959853,
+      "loss": 0.7832,
+      "step": 810
+    },
+    {
+      "epoch": 0.19619643716899374,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.00017942795512130845,
+      "loss": 0.7673,
+      "step": 815
+    },
+    {
+      "epoch": 0.1974000962927299,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00017941113930627025,
+      "loss": 0.7841,
+      "step": 820
+    },
+    {
+      "epoch": 0.19860375541646605,
+      "grad_norm": 1.8203125,
+      "learning_rate": 0.00017939421732135135,
+      "loss": 0.778,
+      "step": 825
+    },
+    {
+      "epoch": 0.1998074145402022,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00017937718919358873,
+      "loss": 0.7848,
+      "step": 830
+    },
+    {
+      "epoch": 0.20101107366393836,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.00017936005495018897,
+      "loss": 0.7925,
+      "step": 835
+    },
+    {
+      "epoch": 0.20221473278767452,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.0001793428146185282,
+      "loss": 0.7661,
+      "step": 840
+    },
+    {
+      "epoch": 0.2034183919114107,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00017932546822615206,
+      "loss": 0.7855,
+      "step": 845
+    },
+    {
+      "epoch": 0.20462205103514686,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00017930801580077563,
+      "loss": 0.7898,
+      "step": 850
+    },
+    {
+      "epoch": 0.205825710158883,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00017929045737028336,
+      "loss": 0.8075,
+      "step": 855
+    },
+    {
+      "epoch": 0.20702936928261917,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.0001792727929627292,
+      "loss": 0.7859,
+      "step": 860
+    },
+    {
+      "epoch": 0.20823302840635532,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.0001792550226063363,
+      "loss": 0.8242,
+      "step": 865
+    },
+    {
+      "epoch": 0.20943668753009148,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00017923714632949716,
+      "loss": 0.7973,
+      "step": 870
+    },
+    {
+      "epoch": 0.21064034665382764,
+      "grad_norm": 1.921875,
+      "learning_rate": 0.00017921916416077348,
+      "loss": 0.7716,
+      "step": 875
+    },
+    {
+      "epoch": 0.2118440057775638,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00017920107612889616,
+      "loss": 0.7995,
+      "step": 880
+    },
+    {
+      "epoch": 0.21304766490129995,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.0001791828822627652,
+      "loss": 0.7775,
+      "step": 885
+    },
+    {
+      "epoch": 0.2142513240250361,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0001791645825914498,
+      "loss": 0.7624,
+      "step": 890
+    },
+    {
+      "epoch": 0.21545498314877226,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00017914617714418808,
+      "loss": 0.7662,
+      "step": 895
+    },
+    {
+      "epoch": 0.21665864227250842,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00017912766595038726,
+      "loss": 0.7519,
+      "step": 900
+    },
+    {
+      "epoch": 0.21786230139624457,
+      "grad_norm": 1.796875,
+      "learning_rate": 0.00017910904903962349,
+      "loss": 0.7878,
+      "step": 905
+    },
+    {
+      "epoch": 0.21906596051998073,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00017909032644164178,
+      "loss": 0.7837,
+      "step": 910
+    },
+    {
+      "epoch": 0.2202696196437169,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.0001790714981863561,
+      "loss": 0.7594,
+      "step": 915
+    },
+    {
+      "epoch": 0.22147327876745307,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.0001790525643038491,
+      "loss": 0.7812,
+      "step": 920
+    },
+    {
+      "epoch": 0.22267693789118922,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00017903352482437235,
+      "loss": 0.7765,
+      "step": 925
+    },
+    {
+      "epoch": 0.22388059701492538,
+      "grad_norm": 1.921875,
+      "learning_rate": 0.00017901437977834605,
+      "loss": 0.7673,
+      "step": 930
+    },
+    {
+      "epoch": 0.22508425613866154,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.00017899512919635898,
+      "loss": 0.7732,
+      "step": 935
+    },
+    {
+      "epoch": 0.2262879152623977,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.00017897577310916875,
+      "loss": 0.7819,
+      "step": 940
+    },
+    {
+      "epoch": 0.22749157438613385,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00017895631154770135,
+      "loss": 0.7743,
+      "step": 945
+    },
+    {
+      "epoch": 0.22869523350987,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00017893674454305143,
+      "loss": 0.7987,
+      "step": 950
+    },
+    {
+      "epoch": 0.22989889263360616,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.000178917072126482,
+      "loss": 0.7453,
+      "step": 955
+    },
+    {
+      "epoch": 0.23110255175734232,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0001788972943294245,
+      "loss": 0.7587,
+      "step": 960
+    },
+    {
+      "epoch": 0.23230621088107847,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.0001788774111834789,
+      "loss": 0.7719,
+      "step": 965
+    },
+    {
+      "epoch": 0.23350987000481463,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.00017885742272041326,
+      "loss": 0.7624,
+      "step": 970
+    },
+    {
+      "epoch": 0.23471352912855079,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.00017883732897216407,
+      "loss": 0.7362,
+      "step": 975
+    },
+    {
+      "epoch": 0.23591718825228694,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.000178817129970836,
+      "loss": 0.7371,
+      "step": 980
+    },
+    {
+      "epoch": 0.2371208473760231,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00017879682574870184,
+      "loss": 0.7411,
+      "step": 985
+    },
+    {
+      "epoch": 0.23832450649975928,
+      "grad_norm": 1.796875,
+      "learning_rate": 0.00017877641633820253,
+      "loss": 0.746,
+      "step": 990
+    },
+    {
+      "epoch": 0.23952816562349544,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.0001787559017719471,
+      "loss": 0.7595,
+      "step": 995
+    },
+    {
+      "epoch": 0.2407318247472316,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00017873528208271254,
+      "loss": 0.7449,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2407318247472316,
+      "eval_loss": 0.655630350112915,
+      "eval_runtime": 2.3436,
+      "eval_samples_per_second": 85.338,
+      "eval_steps_per_second": 85.338,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24193548387096775,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00017871455730344388,
+      "loss": 0.7731,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2431391429947039,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00017869372746725398,
+      "loss": 0.7532,
+      "step": 1010
+    },
+    {
+      "epoch": 0.24434280211844006,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00017867279260742354,
+      "loss": 0.7577,
+      "step": 1015
+    },
+    {
+      "epoch": 0.24554646124217622,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.00017865175275740117,
+      "loss": 0.7327,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24675012036591237,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00017863060795080308,
+      "loss": 0.7278,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24795377948964853,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.0001786093582214133,
+      "loss": 0.7203,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24915743861338469,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00017858800360318343,
+      "loss": 0.7666,
+      "step": 1035
+    },
+    {
+      "epoch": 0.25036109773712084,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0001785665441302327,
+      "loss": 0.7579,
+      "step": 1040
+    },
+    {
+      "epoch": 0.251564756860857,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00017854497983684782,
+      "loss": 0.7048,
+      "step": 1045
+    },
+    {
+      "epoch": 0.25276841598459315,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.000178523310757483,
+      "loss": 0.743,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2539720751083293,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.0001785015369267599,
+      "loss": 0.7023,
+      "step": 1055
+    },
+    {
+      "epoch": 0.25517573423206547,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.0001784796583794675,
+      "loss": 0.7709,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2563793933558016,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00017845767515056214,
+      "loss": 0.7612,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2575830524795378,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00017843558727516733,
+      "loss": 0.7406,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25878671160327393,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.0001784133947885739,
+      "loss": 0.7741,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2599903707270101,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00017839109772623966,
+      "loss": 0.8016,
+      "step": 1080
+    },
+    {
+      "epoch": 0.26119402985074625,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0001783686961237897,
+      "loss": 0.712,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2623976889744824,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00017834619001701597,
+      "loss": 0.7446,
+      "step": 1090
+    },
+    {
+      "epoch": 0.26360134809821856,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0001783235794418775,
+      "loss": 0.7314,
+      "step": 1095
+    },
+    {
+      "epoch": 0.26480500722195477,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0001783008644345002,
+      "loss": 0.743,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2660086663456909,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00017827804503117676,
+      "loss": 0.7261,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2672123254694271,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.0001782551212683668,
+      "loss": 0.6977,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26841598459316324,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00017823209318269662,
+      "loss": 0.7521,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2696196437168994,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00017820896081095918,
+      "loss": 0.7073,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27082330284063555,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00017818572419011402,
+      "loss": 0.7562,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2720269619643717,
+      "grad_norm": 1.625,
+      "learning_rate": 0.0001781623833572874,
+      "loss": 0.7414,
+      "step": 1130
+    },
+    {
+      "epoch": 0.27323062108810786,
+      "grad_norm": 1.6640625,
+      "learning_rate": 0.00017813893834977196,
+      "loss": 0.7684,
+      "step": 1135
+    },
+    {
+      "epoch": 0.274434280211844,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00017811538920502678,
+      "loss": 0.7436,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2756379393355802,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.0001780917359606774,
+      "loss": 0.7491,
+      "step": 1145
+    },
+    {
+      "epoch": 0.27684159845931633,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00017806797865451557,
+      "loss": 0.7647,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2780452575830525,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.00017804411732449946,
+      "loss": 0.7486,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27924891670678864,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.0001780201520087533,
+      "loss": 0.7637,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2804525758305248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.00017799608274556757,
+      "loss": 0.703,
+      "step": 1165
+    },
+    {
+      "epoch": 0.28165623495426095,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.00017797190957339872,
+      "loss": 0.7377,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2828598940779971,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00017794763253086934,
+      "loss": 0.7271,
+      "step": 1175
+    },
+    {
+      "epoch": 0.28406355320173327,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00017792325165676788,
+      "loss": 0.7541,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2852672123254694,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00017789876699004874,
+      "loss": 0.7331,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2864708714492056,
+      "grad_norm": 1.6640625,
+      "learning_rate": 0.00017787417856983216,
+      "loss": 0.7607,
+      "step": 1190
+    },
+    {
+      "epoch": 0.28767453057294173,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.0001778494864354041,
+      "loss": 0.7371,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2888781896966779,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00017782469062621627,
+      "loss": 0.744,
+      "step": 1200
+    },
+    {
+      "epoch": 0.29008184882041405,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00017779979118188603,
+      "loss": 0.7065,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2912855079441502,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00017777478814219632,
+      "loss": 0.7438,
+      "step": 1210
+    },
+    {
+      "epoch": 0.29248916706788636,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00017774968154709558,
+      "loss": 0.7451,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2936928261916225,
+      "grad_norm": 1.6640625,
+      "learning_rate": 0.00017772447143669766,
+      "loss": 0.7435,
+      "step": 1220
+    },
+    {
+      "epoch": 0.29489648531535867,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00017769915785128193,
+      "loss": 0.7337,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2961001444390948,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00017767374083129295,
+      "loss": 0.7463,
+      "step": 1230
+    },
+    {
+      "epoch": 0.297303803562831,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00017764822041734066,
+      "loss": 0.7128,
+      "step": 1235
+    },
+    {
+      "epoch": 0.29850746268656714,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.0001776225966502001,
+      "loss": 0.7143,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29971112181030335,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0001775968695708115,
+      "loss": 0.7268,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3009147809340395,
+      "grad_norm": 1.6640625,
+      "learning_rate": 0.00017757103922028008,
+      "loss": 0.7318,
+      "step": 1250
+    },
+    {
+      "epoch": 0.30211844005777566,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.0001775451056398762,
+      "loss": 0.7249,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3033220991815118,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00017751906887103502,
+      "loss": 0.7291,
+      "step": 1260
+    },
+    {
+      "epoch": 0.304525758305248,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.0001774929289553567,
+      "loss": 0.7048,
+      "step": 1265
+    },
+    {
+      "epoch": 0.30572941742898413,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.000177466685934606,
+      "loss": 0.7292,
+      "step": 1270
+    },
+    {
+      "epoch": 0.3069330765527203,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00017744033985071263,
+      "loss": 0.7248,
+      "step": 1275
+    },
+    {
+      "epoch": 0.30813673567645644,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00017741389074577086,
+      "loss": 0.7061,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3093403948001926,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00017738733866203956,
+      "loss": 0.7451,
+      "step": 1285
+    },
+    {
+      "epoch": 0.31054405392392875,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.00017736068364194218,
+      "loss": 0.7291,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3117477130476649,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00017733392572806658,
+      "loss": 0.7464,
+      "step": 1295
+    },
+    {
+      "epoch": 0.31295137217140107,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00017730706496316506,
+      "loss": 0.7026,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3141550312951372,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00017728010139015426,
+      "loss": 0.7097,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3153586904188734,
+      "grad_norm": 1.828125,
+      "learning_rate": 0.000177253035052115,
+      "loss": 0.727,
+      "step": 1310
+    },
+    {
+      "epoch": 0.31656234954260953,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.0001772258659922924,
+      "loss": 0.7524,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3177660086663457,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00017719859425409566,
+      "loss": 0.7328,
+      "step": 1320
+    },
+    {
+      "epoch": 0.31896966779008185,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.000177171219881098,
+      "loss": 0.7341,
+      "step": 1325
+    },
+    {
+      "epoch": 0.320173326913818,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.0001771437429170366,
+      "loss": 0.7141,
+      "step": 1330
+    },
+    {
+      "epoch": 0.32137698603755416,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.00017711616340581273,
+      "loss": 0.718,
+      "step": 1335
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00017708848139149128,
+      "loss": 0.6941,
+      "step": 1340
+    },
+    {
+      "epoch": 0.32378430428502647,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00017706069691830105,
+      "loss": 0.6899,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3249879634087626,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00017703281003063448,
+      "loss": 0.6839,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3261916225324988,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.0001770048207730477,
+      "loss": 0.7249,
+      "step": 1355
+    },
+    {
+      "epoch": 0.32739528165623494,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.00017697672919026036,
+      "loss": 0.71,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3285989407799711,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00017694853532715558,
+      "loss": 0.7557,
+      "step": 1365
+    },
+    {
+      "epoch": 0.32980259990370725,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00017692023922877997,
+      "loss": 0.6683,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3310062590274434,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.0001768918409403434,
+      "loss": 0.7334,
+      "step": 1375
+    },
+    {
+      "epoch": 0.33220991815117956,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.0001768633405072191,
+      "loss": 0.684,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3334135772749157,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.0001768347379749434,
+      "loss": 0.6837,
+      "step": 1385
+    },
+    {
+      "epoch": 0.33461723639865193,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00017680603338921587,
+      "loss": 0.7148,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3358208955223881,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00017677722679589901,
+      "loss": 0.7273,
+      "step": 1395
+    },
+    {
+      "epoch": 0.33702455464612424,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00017674831824101845,
+      "loss": 0.7363,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3382282137698604,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.0001767193077707626,
+      "loss": 0.68,
+      "step": 1405
+    },
+    {
+      "epoch": 0.33943187289359655,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00017669019543148278,
+      "loss": 0.7127,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3406355320173327,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00017666098126969303,
+      "loss": 0.72,
+      "step": 1415
+    },
+    {
+      "epoch": 0.34183919114106887,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.0001766316653320701,
+      "loss": 0.7284,
+      "step": 1420
+    },
+    {
+      "epoch": 0.343042850264805,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00017660224766545332,
+      "loss": 0.729,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3442465093885412,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00017657272831684458,
+      "loss": 0.7067,
+      "step": 1430
+    },
+    {
+      "epoch": 0.34545016851227733,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0001765431073334083,
+      "loss": 0.7222,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3466538276360135,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00017651338476247117,
+      "loss": 0.7052,
+      "step": 1440
+    },
+    {
+      "epoch": 0.34785748675974965,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0001764835606515222,
+      "loss": 0.6729,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3490611458834858,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.00017645363504821275,
+      "loss": 0.6867,
+      "step": 1450
+    },
+    {
+      "epoch": 0.35026480500722196,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00017642360800035619,
+      "loss": 0.6774,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3514684641309581,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.0001763934795559281,
+      "loss": 0.6915,
+      "step": 1460
+    },
+    {
+      "epoch": 0.35267212325469427,
+      "grad_norm": 1.5,
+      "learning_rate": 0.000176363249763066,
+      "loss": 0.7101,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3538757823784304,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.0001763329186700693,
+      "loss": 0.7403,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3550794415021666,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00017630248632539937,
+      "loss": 0.7208,
+      "step": 1475
+    },
+    {
+      "epoch": 0.35628310062590274,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.00017627195277767924,
+      "loss": 0.714,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3574867597496389,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00017624131807569376,
+      "loss": 0.6912,
+      "step": 1485
+    },
+    {
+      "epoch": 0.35869041887337505,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00017621058226838932,
+      "loss": 0.6765,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3598940779971112,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.0001761797454048738,
+      "loss": 0.6883,
+      "step": 1495
+    },
+    {
+      "epoch": 0.36109773712084736,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00017614880753441666,
+      "loss": 0.6799,
+      "step": 1500
+    },
+    {
+      "epoch": 0.36109773712084736,
+      "eval_loss": 0.603824257850647,
+      "eval_runtime": 2.3494,
+      "eval_samples_per_second": 85.128,
+      "eval_steps_per_second": 85.128,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3623013962445835,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00017611776870644867,
+      "loss": 0.7172,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3635050553683197,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00017608662897056193,
+      "loss": 0.6978,
+      "step": 1510
+    },
+    {
+      "epoch": 0.36470871449205583,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.0001760553883765097,
+      "loss": 0.706,
+      "step": 1515
+    },
+    {
+      "epoch": 0.365912373615792,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00017602404697420653,
+      "loss": 0.6777,
+      "step": 1520
+    },
+    {
+      "epoch": 0.36711603273952814,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0001759926048137279,
+      "loss": 0.6695,
+      "step": 1525
+    },
+    {
+      "epoch": 0.3683196918632643,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.0001759610619453103,
+      "loss": 0.7313,
+      "step": 1530
+    },
+    {
+      "epoch": 0.36952335098700045,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00017592941841935118,
+      "loss": 0.722,
+      "step": 1535
+    },
+    {
+      "epoch": 0.37072701011073667,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00017589767428640874,
+      "loss": 0.7167,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3719306692344728,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.00017586582959720203,
+      "loss": 0.69,
+      "step": 1545
+    },
+    {
+      "epoch": 0.373134328358209,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00017583388440261066,
+      "loss": 0.691,
+      "step": 1550
+    },
+    {
+      "epoch": 0.37433798748194513,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00017580183875367486,
+      "loss": 0.7092,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3755416466056813,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00017576969270159543,
+      "loss": 0.7057,
+      "step": 1560
+    },
+    {
+      "epoch": 0.37674530572941745,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00017573744629773342,
+      "loss": 0.6886,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3779489648531536,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00017570509959361034,
+      "loss": 0.7254,
+      "step": 1570
+    },
+    {
+      "epoch": 0.37915262397688976,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.000175672652640908,
+      "loss": 0.7132,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3803562831006259,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00017564010549146823,
+      "loss": 0.687,
+      "step": 1580
+    },
+    {
+      "epoch": 0.38155994222436207,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00017560745819729306,
+      "loss": 0.6896,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3827636013480982,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00017557471081054452,
+      "loss": 0.7031,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3839672604718344,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00017554186338354451,
+      "loss": 0.6904,
+      "step": 1595
+    },
+    {
+      "epoch": 0.38517091959557054,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00017550891596877484,
+      "loss": 0.6994,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3863745787193067,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00017547586861887696,
+      "loss": 0.6923,
+      "step": 1605
+    },
+    {
+      "epoch": 0.38757823784304285,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00017544272138665212,
+      "loss": 0.6936,
+      "step": 1610
+    },
+    {
+      "epoch": 0.388781896966779,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.0001754094743250611,
+      "loss": 0.7102,
+      "step": 1615
+    },
+    {
+      "epoch": 0.38998555609051516,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00017537612748722416,
+      "loss": 0.6807,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3911892152142513,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.00017534268092642098,
+      "loss": 0.7496,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3923928743379875,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00017530913469609066,
+      "loss": 0.6886,
+      "step": 1630
+    },
+    {
+      "epoch": 0.39359653346172363,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00017527548884983137,
+      "loss": 0.7033,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3948001925854598,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00017524174344140067,
+      "loss": 0.7094,
+      "step": 1640
+    },
+    {
+      "epoch": 0.39600385170919594,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00017520789852471498,
+      "loss": 0.6735,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3972075108329321,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00017517395415384983,
+      "loss": 0.6832,
+      "step": 1650
+    },
+    {
+      "epoch": 0.39841116995666825,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00017513991038303964,
+      "loss": 0.6997,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3996148290804044,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001751057672666776,
+      "loss": 0.6992,
+      "step": 1660
+    },
+    {
+      "epoch": 0.40081848820414057,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00017507152485931565,
+      "loss": 0.674,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4020221473278767,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00017503718321566444,
+      "loss": 0.67,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4032258064516129,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00017500274239059304,
+      "loss": 0.6838,
+      "step": 1675
+    },
+    {
+      "epoch": 0.40442946557534903,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.0001749682024391291,
+      "loss": 0.716,
+      "step": 1680
+    },
+    {
+      "epoch": 0.40563312469908525,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00017493356341645862,
+      "loss": 0.6694,
+      "step": 1685
+    },
+    {
+      "epoch": 0.4068367838228214,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00017489882537792583,
+      "loss": 0.7015,
+      "step": 1690
+    },
+    {
+      "epoch": 0.40804044294655756,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00017486398837903325,
+      "loss": 0.6801,
+      "step": 1695
+    },
+    {
+      "epoch": 0.4092441020702937,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00017482905247544146,
+      "loss": 0.6746,
+      "step": 1700
+    },
+    {
+      "epoch": 0.41044776119402987,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00017479401772296907,
+      "loss": 0.7102,
+      "step": 1705
+    },
+    {
+      "epoch": 0.411651420317766,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00017475888417759264,
+      "loss": 0.6631,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4128550794415022,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0001747236518954466,
+      "loss": 0.6714,
+      "step": 1715
+    },
+    {
+      "epoch": 0.41405873856523834,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00017468832093282303,
+      "loss": 0.6976,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4152623976889745,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00017465289134617183,
+      "loss": 0.6858,
+      "step": 1725
+    },
+    {
+      "epoch": 0.41646605681271065,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00017461736319210038,
+      "loss": 0.6579,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4176697159364468,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00017458173652737353,
+      "loss": 0.6895,
+      "step": 1735
+    },
+    {
+      "epoch": 0.41887337506018296,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.00017454601140891353,
+      "loss": 0.6781,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4200770341839191,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00017451018789380008,
+      "loss": 0.6734,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4212806933076553,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.0001744742660392699,
+      "loss": 0.687,
+      "step": 1750
+    },
+    {
+      "epoch": 0.42248435243139143,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00017443824590271685,
+      "loss": 0.6909,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4236880115551276,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00017440212754169193,
+      "loss": 0.6658,
+      "step": 1760
+    },
+    {
+      "epoch": 0.42489167067886374,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00017436591101390304,
+      "loss": 0.6879,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4260953298025999,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00017432959637721492,
+      "loss": 0.6712,
+      "step": 1770
+    },
+    {
+      "epoch": 0.42729898892633605,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00017429318368964897,
+      "loss": 0.6484,
+      "step": 1775
+    },
+    {
+      "epoch": 0.4285026480500722,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0001742566730093834,
+      "loss": 0.6784,
+      "step": 1780
+    },
+    {
+      "epoch": 0.42970630717380837,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0001742200643947529,
+      "loss": 0.699,
+      "step": 1785
+    },
+    {
+      "epoch": 0.4309099662975445,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00017418335790424862,
+      "loss": 0.7127,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4321136254212807,
+      "grad_norm": 1.625,
+      "learning_rate": 0.00017414655359651818,
+      "loss": 0.6998,
+      "step": 1795
+    },
+    {
+      "epoch": 0.43331728454501683,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00017410965153036537,
+      "loss": 0.6865,
+      "step": 1800
+    },
+    {
+      "epoch": 0.434520943668753,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0001740726517647503,
+      "loss": 0.7226,
+      "step": 1805
+    },
+    {
+      "epoch": 0.43572460279248915,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00017403555435878903,
+      "loss": 0.6809,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4369282619162253,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00017399835937175376,
+      "loss": 0.6601,
+      "step": 1815
+    },
+    {
+      "epoch": 0.43813192103996146,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00017396106686307252,
+      "loss": 0.6662,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4393355801636976,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00017392367689232922,
+      "loss": 0.6769,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4405392392874338,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00017388618951926337,
+      "loss": 0.6615,
+      "step": 1830
+    },
+    {
+      "epoch": 0.44174289841117,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00017384860480377026,
+      "loss": 0.6853,
+      "step": 1835
+    },
+    {
+      "epoch": 0.44294655753490614,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00017381092280590061,
+      "loss": 0.7043,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4441502166586423,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00017377314358586055,
+      "loss": 0.662,
+      "step": 1845
+    },
+    {
+      "epoch": 0.44535387578237845,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00017373526720401163,
+      "loss": 0.6932,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4465575349061146,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001736972937208706,
+      "loss": 0.6811,
+      "step": 1855
+    },
+    {
+      "epoch": 0.44776119402985076,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00017365922319710934,
+      "loss": 0.6901,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4489648531535869,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00017362105569355477,
+      "loss": 0.6312,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4501685122773231,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.0001735827912711888,
+      "loss": 0.6867,
+      "step": 1870
+    },
+    {
+      "epoch": 0.45137217140105923,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00017354442999114816,
+      "loss": 0.6822,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4525758305247954,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00017350597191472436,
+      "loss": 0.672,
+      "step": 1880
+    },
+    {
+      "epoch": 0.45377948964853154,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00017346741710336352,
+      "loss": 0.6874,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4549831487722677,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00017342876561866636,
+      "loss": 0.6388,
+      "step": 1890
+    },
+    {
+      "epoch": 0.45618680789600385,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00017339001752238805,
+      "loss": 0.6422,
+      "step": 1895
+    },
+    {
+      "epoch": 0.45739046701974,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00017335117287643807,
+      "loss": 0.7195,
+      "step": 1900
+    },
+    {
+      "epoch": 0.45859412614347617,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00017331223174288027,
+      "loss": 0.6871,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4597977852672123,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00017327319418393257,
+      "loss": 0.6816,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4610014443909485,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.000173234060261967,
+      "loss": 0.6728,
+      "step": 1915
+    },
+    {
+      "epoch": 0.46220510351468463,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00017319483003950948,
+      "loss": 0.6706,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4634087626384208,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.0001731555035792399,
+      "loss": 0.707,
+      "step": 1925
+    },
+    {
+      "epoch": 0.46461242176215695,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.0001731160809439918,
+      "loss": 0.6717,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4658160808858931,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00017307656219675257,
+      "loss": 0.6665,
+      "step": 1935
+    },
+    {
+      "epoch": 0.46701974000962926,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00017303694740066294,
+      "loss": 0.6663,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4682233991333654,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00017299723661901717,
+      "loss": 0.663,
+      "step": 1945
+    },
+    {
+      "epoch": 0.46942705825710157,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.000172957429915263,
+      "loss": 0.6483,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4706307173808377,
+      "grad_norm": 1.375,
+      "learning_rate": 0.0001729175273530013,
+      "loss": 0.6526,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4718343765045739,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.0001728775289959861,
+      "loss": 0.6336,
+      "step": 1960
+    },
+    {
+      "epoch": 0.47303803562831004,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.0001728374349081246,
+      "loss": 0.6845,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4742416947520462,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00017279724515347682,
+      "loss": 0.6943,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4754453538757824,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00017275695979625564,
+      "loss": 0.678,
+      "step": 1975
+    },
+    {
+      "epoch": 0.47664901299951856,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00017271657890082683,
+      "loss": 0.6725,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4778526721232547,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00017267610253170865,
+      "loss": 0.6427,
+      "step": 1985
+    },
+    {
+      "epoch": 0.4790563312469909,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00017263553075357195,
+      "loss": 0.66,
+      "step": 1990
+    },
+    {
+      "epoch": 0.48025999037072703,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00017259486363124013,
+      "loss": 0.6455,
+      "step": 1995
+    },
+    {
+      "epoch": 0.4814636494944632,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001725541012296887,
+      "loss": 0.6376,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4814636494944632,
+      "eval_loss": 0.5736396312713623,
+      "eval_runtime": 2.3497,
+      "eval_samples_per_second": 85.118,
+      "eval_steps_per_second": 85.118,
+      "step": 2000
+    },
+    {
+      "epoch": 0.48266730861819934,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00017251324361404558,
+      "loss": 0.6659,
+      "step": 2005
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.0001724722908495908,
+      "loss": 0.6464,
+      "step": 2010
+    },
+    {
+      "epoch": 0.48507462686567165,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0001724312430017563,
+      "loss": 0.6679,
+      "step": 2015
+    },
+    {
+      "epoch": 0.4862782859894078,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0001723901001361261,
+      "loss": 0.6484,
+      "step": 2020
+    },
+    {
+      "epoch": 0.48748194511314397,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.0001723488623184359,
+      "loss": 0.6647,
+      "step": 2025
+    },
+    {
+      "epoch": 0.4886856042368801,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0001723075296145732,
+      "loss": 0.6346,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4898892633606163,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00017226610209057698,
+      "loss": 0.6896,
+      "step": 2035
+    },
+    {
+      "epoch": 0.49109292248435243,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00017222457981263793,
+      "loss": 0.6608,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4922965816080886,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00017218296284709795,
+      "loss": 0.6392,
+      "step": 2045
+    },
+    {
+      "epoch": 0.49350024073182475,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00017214125126045022,
+      "loss": 0.6745,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4947038998555609,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00017209944511933925,
+      "loss": 0.6423,
+      "step": 2055
+    },
+    {
+      "epoch": 0.49590755897929706,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0001720575444905605,
+      "loss": 0.6368,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4971112181030332,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00017201554944106044,
+      "loss": 0.6609,
+      "step": 2065
+    },
+    {
+      "epoch": 0.49831487722676937,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001719734600379364,
+      "loss": 0.6643,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4995185363505055,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00017193127634843643,
+      "loss": 0.6324,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5007221954742417,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00017188899843995927,
+      "loss": 0.6465,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5019258545979779,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00017184662638005418,
+      "loss": 0.6856,
+      "step": 2085
+    },
+    {
+      "epoch": 0.503129513721714,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00017180416023642085,
+      "loss": 0.6859,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5043331728454502,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00017176160007690926,
+      "loss": 0.6251,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5055368319691863,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00017171894596951974,
+      "loss": 0.6829,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5067404910929225,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00017167619798240247,
+      "loss": 0.6331,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5079441502166586,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00017163335618385788,
+      "loss": 0.6639,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5091478093403948,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00017159042064233615,
+      "loss": 0.673,
+      "step": 2115
+    },
+    {
+      "epoch": 0.5103514684641309,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0001715473914264373,
+      "loss": 0.6561,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5115551275878671,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00017150426860491089,
+      "loss": 0.6932,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5127587867116032,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00017146105224665622,
+      "loss": 0.604,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5139624458353395,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00017141774242072195,
+      "loss": 0.6275,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5151661049590756,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00017137433919630604,
+      "loss": 0.6817,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5163697640828118,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00017133084264275576,
+      "loss": 0.6489,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5175734232065479,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001712872528295674,
+      "loss": 0.6387,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5187770823302841,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00017124356982638637,
+      "loss": 0.6671,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5199807414540202,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00017119979370300683,
+      "loss": 0.6469,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5211844005777564,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00017115592452937188,
+      "loss": 0.6686,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5223880597014925,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00017111196237557316,
+      "loss": 0.6767,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5235917188252287,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00017106790731185096,
+      "loss": 0.6494,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5247953779489648,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00017102375940859397,
+      "loss": 0.6521,
+      "step": 2180
+    },
+    {
+      "epoch": 0.525999037072701,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00017097951873633918,
+      "loss": 0.6689,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5272026961964371,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00017093518536577193,
+      "loss": 0.65,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5284063553201733,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001708907593677255,
+      "loss": 0.6565,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5296100144439095,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001708462408131813,
+      "loss": 0.6561,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5308136735676456,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0001708016297732685,
+      "loss": 0.6656,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5320173326913819,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00017075692631926418,
+      "loss": 0.6479,
+      "step": 2210
+    },
+    {
+      "epoch": 0.533220991815118,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00017071213052259295,
+      "loss": 0.6401,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5344246509388542,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00017066724245482702,
+      "loss": 0.6677,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5356283100625903,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00017062226218768608,
+      "loss": 0.6617,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5368319691863265,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00017057718979303696,
+      "loss": 0.6479,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5380356283100626,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00017053202534289384,
+      "loss": 0.6772,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5392392874337988,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00017048676890941796,
+      "loss": 0.6362,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5404429465575349,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00017044142056491746,
+      "loss": 0.629,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5416466056812711,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00017039598038184737,
+      "loss": 0.6257,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5428502648050072,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001703504484328095,
+      "loss": 0.6523,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5440539239287434,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00017030482479055216,
+      "loss": 0.6657,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5452575830524795,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.0001702591095279703,
+      "loss": 0.6787,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5464612421762157,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00017021330271810515,
+      "loss": 0.6605,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5476649012999518,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00017016740443414428,
+      "loss": 0.6444,
+      "step": 2275
+    },
+    {
+      "epoch": 0.548868560423688,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00017012141474942135,
+      "loss": 0.6792,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5500722195474241,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0001700753337374161,
+      "loss": 0.6539,
+      "step": 2285
+    },
+    {
+      "epoch": 0.5512758786711603,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0001700291614717542,
+      "loss": 0.6465,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5524795377948964,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00016998289802620705,
+      "loss": 0.6388,
+      "step": 2295
+    },
+    {
+      "epoch": 0.5536831969186327,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00016993654347469183,
+      "loss": 0.6497,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5548868560423688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.00016989009789127114,
+      "loss": 0.6617,
+      "step": 2305
+    },
+    {
+      "epoch": 0.556090515166105,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00016984356135015322,
+      "loss": 0.6336,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5572941742898411,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00016979693392569145,
+      "loss": 0.628,
+      "step": 2315
+    },
+    {
+      "epoch": 0.5584978334135773,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00016975021569238456,
+      "loss": 0.6733,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5597014925373134,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00016970340672487627,
+      "loss": 0.6282,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5609051516610496,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00016965650709795536,
+      "loss": 0.64,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5621088107847857,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00016960951688655535,
+      "loss": 0.6105,
+      "step": 2335
+    },
+    {
+      "epoch": 0.5633124699085219,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00016956243616575458,
+      "loss": 0.6637,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5645161290322581,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00016951526501077598,
+      "loss": 0.6459,
+      "step": 2345
+    },
+    {
+      "epoch": 0.5657197881559942,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00016946800349698697,
+      "loss": 0.6501,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5669234472797304,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00016942065169989933,
+      "loss": 0.6138,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5681271064034665,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00016937320969516906,
+      "loss": 0.6259,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5693307655272027,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016932567755859634,
+      "loss": 0.6153,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5705344246509388,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00016927805536612534,
+      "loss": 0.6106,
+      "step": 2370
+    },
+    {
+      "epoch": 0.571738083774675,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00016923034319384414,
+      "loss": 0.6342,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5729417428984112,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00016918254111798455,
+      "loss": 0.627,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5741454020221474,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00016913464921492199,
+      "loss": 0.6348,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5753490611458835,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0001690866675611755,
+      "loss": 0.6824,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5765527202696197,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00016903859623340742,
+      "loss": 0.6375,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5777563793933558,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00016899043530842344,
+      "loss": 0.6473,
+      "step": 2400
+    },
+    {
+      "epoch": 0.578960038517092,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00016894218486317235,
+      "loss": 0.6524,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5801636976408281,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.000168893844974746,
+      "loss": 0.6084,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5813673567645643,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00016884541572037917,
+      "loss": 0.612,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5825710158883004,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00016879689717744936,
+      "loss": 0.6657,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5837746750120366,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00016874828942347673,
+      "loss": 0.629,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5849783341357727,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00016869959253612408,
+      "loss": 0.6586,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5861819932595089,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.0001686508065931965,
+      "loss": 0.6274,
+      "step": 2435
+    },
+    {
+      "epoch": 0.587385652383245,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00016860193167264148,
+      "loss": 0.6241,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5885893115069812,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00016855296785254852,
+      "loss": 0.6426,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5897929706307173,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016850391521114937,
+      "loss": 0.6492,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5909966297544536,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0001684547738268175,
+      "loss": 0.6299,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5922002888781897,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00016840554377806823,
+      "loss": 0.6246,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5934039480019259,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.0001683562251435586,
+      "loss": 0.6388,
+      "step": 2465
+    },
+    {
+      "epoch": 0.594607607125662,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00016830681800208714,
+      "loss": 0.6423,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5958112662493982,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001682573224325938,
+      "loss": 0.6324,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5970149253731343,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00016820773851415976,
+      "loss": 0.6533,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5982185844968705,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0001681580663260075,
+      "loss": 0.7084,
+      "step": 2485
+    },
+    {
+      "epoch": 0.5994222436206067,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00016810830594750034,
+      "loss": 0.618,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6006259027443428,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0001680584574581427,
+      "loss": 0.6244,
+      "step": 2495
+    },
+    {
+      "epoch": 0.601829561868079,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00016800852093757965,
+      "loss": 0.6373,
+      "step": 2500
+    },
+    {
+      "epoch": 0.601829561868079,
+      "eval_loss": 0.5356536507606506,
+      "eval_runtime": 2.3478,
+      "eval_samples_per_second": 85.185,
+      "eval_steps_per_second": 85.185,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6030332209918151,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00016795849646559694,
+      "loss": 0.6164,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6042368801155513,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00016790838412212086,
+      "loss": 0.6377,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6054405392392874,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00016785818398721807,
+      "loss": 0.6171,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6066441983630236,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016780789614109556,
+      "loss": 0.649,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6078478574867597,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0001677575206641004,
+      "loss": 0.644,
+      "step": 2525
+    },
+    {
+      "epoch": 0.609051516610496,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00016770705763671966,
+      "loss": 0.6246,
+      "step": 2530
+    },
+    {
+      "epoch": 0.610255175734232,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016765650713958035,
+      "loss": 0.6154,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6114588348579683,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00016760586925344923,
+      "loss": 0.6131,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6126624939817044,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00016755514405923256,
+      "loss": 0.6289,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6138661531054406,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00016750433163797632,
+      "loss": 0.6327,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6150698122291767,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00016745343207086566,
+      "loss": 0.6132,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6162734713529129,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00016740244543922504,
+      "loss": 0.613,
+      "step": 2560
+    },
+    {
+      "epoch": 0.617477130476649,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00016735137182451808,
+      "loss": 0.6422,
+      "step": 2565
+    },
+    {
+      "epoch": 0.6186807896003852,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00016730021130834729,
+      "loss": 0.6259,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6198844487241213,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016724896397245402,
+      "loss": 0.6017,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6210881078478575,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00016719762989871845,
+      "loss": 0.5948,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6222917669715936,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00016714620916915922,
+      "loss": 0.6597,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6234954260953298,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.0001670947018659335,
+      "loss": 0.6211,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6246990852190659,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00016704310807133673,
+      "loss": 0.6293,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6259027443428021,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0001669914278678026,
+      "loss": 0.6254,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6271064034665382,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.0001669396613379028,
+      "loss": 0.633,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6283100625902744,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00016688780856434696,
+      "loss": 0.6533,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6295137217140105,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00016683586962998258,
+      "loss": 0.641,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6307173808377468,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00016678384461779472,
+      "loss": 0.6204,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6319210399614829,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.000166731733610906,
+      "loss": 0.6634,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6331246990852191,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00016667953669257648,
+      "loss": 0.6318,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6343283582089553,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00016662725394620345,
+      "loss": 0.6196,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6355320173326914,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016657488545532132,
+      "loss": 0.6651,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6367356764564276,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00016652243130360153,
+      "loss": 0.6361,
+      "step": 2645
+    },
+    {
+      "epoch": 0.6379393355801637,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00016646989157485237,
+      "loss": 0.6251,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6391429947038999,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00016641726635301883,
+      "loss": 0.6421,
+      "step": 2655
+    },
+    {
+      "epoch": 0.640346653827636,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00016636455572218255,
+      "loss": 0.6256,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6415503129513722,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00016631175976656158,
+      "loss": 0.6211,
+      "step": 2665
+    },
+    {
+      "epoch": 0.6427539720751083,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00016625887857051036,
+      "loss": 0.6005,
+      "step": 2670
+    },
+    {
+      "epoch": 0.6439576311988445,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00016620591221851947,
+      "loss": 0.6192,
+      "step": 2675
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00016615286079521555,
+      "loss": 0.63,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6463649494463168,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001660997243853612,
+      "loss": 0.6259,
+      "step": 2685
+    },
+    {
+      "epoch": 0.6475686085700529,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00016604650307385475,
+      "loss": 0.6238,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6487722676937892,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0001659931969457302,
+      "loss": 0.6168,
+      "step": 2695
+    },
+    {
+      "epoch": 0.6499759268175253,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00016593980608615717,
+      "loss": 0.6238,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6511795859412615,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00016588633058044045,
+      "loss": 0.6764,
+      "step": 2705
+    },
+    {
+      "epoch": 0.6523832450649976,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00016583277051402027,
+      "loss": 0.6082,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6535869041887338,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00016577912597247183,
+      "loss": 0.6378,
+      "step": 2715
+    },
+    {
+      "epoch": 0.6547905633124699,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0001657253970415054,
+      "loss": 0.62,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6559942224362061,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.000165671583806966,
+      "loss": 0.6159,
+      "step": 2725
+    },
+    {
+      "epoch": 0.6571978815599422,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0001656176863548334,
+      "loss": 0.6557,
+      "step": 2730
+    },
+    {
+      "epoch": 0.6584015406836784,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00016556370477122196,
+      "loss": 0.6616,
+      "step": 2735
+    },
+    {
+      "epoch": 0.6596051998074145,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00016550963914238031,
+      "loss": 0.6135,
+      "step": 2740
+    },
+    {
+      "epoch": 0.6608088589311507,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016545548955469158,
+      "loss": 0.5916,
+      "step": 2745
+    },
+    {
+      "epoch": 0.6620125180548868,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.0001654012560946728,
+      "loss": 0.6324,
+      "step": 2750
+    },
+    {
+      "epoch": 0.663216177178623,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0001653469388489753,
+      "loss": 0.6404,
+      "step": 2755
+    },
+    {
+      "epoch": 0.6644198363023591,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00016529253790438395,
+      "loss": 0.6192,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6656234954260953,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00016523805334781762,
+      "loss": 0.6326,
+      "step": 2765
+    },
+    {
+      "epoch": 0.6668271545498314,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00016518348526632865,
+      "loss": 0.635,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6680308136735676,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0001651288337471028,
+      "loss": 0.6192,
+      "step": 2775
+    },
+    {
+      "epoch": 0.6692344727973039,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00016507409887745922,
+      "loss": 0.6423,
+      "step": 2780
+    },
+    {
+      "epoch": 0.67043813192104,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00016501928074485016,
+      "loss": 0.6264,
+      "step": 2785
+    },
+    {
+      "epoch": 0.6716417910447762,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00016496437943686103,
+      "loss": 0.6028,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6728454501685123,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016490939504120998,
+      "loss": 0.6289,
+      "step": 2795
+    },
+    {
+      "epoch": 0.6740491092922485,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00016485432764574797,
+      "loss": 0.6272,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6752527684159846,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00016479917733845862,
+      "loss": 0.6095,
+      "step": 2805
+    },
+    {
+      "epoch": 0.6764564275397208,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00016474394420745798,
+      "loss": 0.6311,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6776600866634569,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0001646886283409944,
+      "loss": 0.6475,
+      "step": 2815
+    },
+    {
+      "epoch": 0.6788637457871931,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.0001646332298274485,
+      "loss": 0.63,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6800674049109292,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00016457774875533283,
+      "loss": 0.6144,
+      "step": 2825
+    },
+    {
+      "epoch": 0.6812710640346654,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.000164522185213292,
+      "loss": 0.6001,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6824747231584015,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001644665392901023,
+      "loss": 0.621,
+      "step": 2835
+    },
+    {
+      "epoch": 0.6836783822821377,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00016441081107467157,
+      "loss": 0.639,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6848820414058738,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00016435500065603933,
+      "loss": 0.6184,
+      "step": 2845
+    },
+    {
+      "epoch": 0.68608570052961,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001642991081233762,
+      "loss": 0.6164,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6872893596533461,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00016424313356598423,
+      "loss": 0.5994,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6884930187770824,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00016418707707329636,
+      "loss": 0.6175,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6896966779008185,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001641309387348765,
+      "loss": 0.6209,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6909003370245547,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00016407471864041926,
+      "loss": 0.6126,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6921039961482908,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00016401841687975005,
+      "loss": 0.605,
+      "step": 2875
+    },
+    {
+      "epoch": 0.693307655272027,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00016396203354282458,
+      "loss": 0.6187,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6945113143957631,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00016390556871972902,
+      "loss": 0.6374,
+      "step": 2885
+    },
+    {
+      "epoch": 0.6957149735194993,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00016384902250067963,
+      "loss": 0.6246,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6969186326432354,
+      "grad_norm": 1.5,
+      "learning_rate": 0.0001637923949760228,
+      "loss": 0.6356,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6981222917669716,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00016373568623623478,
+      "loss": 0.5897,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6993259508907077,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016367889637192162,
+      "loss": 0.63,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7005296100144439,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00016362202547381892,
+      "loss": 0.5854,
+      "step": 2910
+    },
+    {
+      "epoch": 0.70173326913818,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00016356507363279188,
+      "loss": 0.6286,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7029369282619162,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001635080409398349,
+      "loss": 0.5537,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7041405873856523,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00016345092748607158,
+      "loss": 0.6272,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7053442465093885,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00016339373336275457,
+      "loss": 0.5848,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7065479056331248,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00016333645866126544,
+      "loss": 0.5941,
+      "step": 2935
+    },
+    {
+      "epoch": 0.7077515647568609,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00016327910347311452,
+      "loss": 0.583,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7089552238805971,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00016322166788994063,
+      "loss": 0.6296,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7101588830043332,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00016316415200351116,
+      "loss": 0.6183,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7113625421280694,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00016310655590572173,
+      "loss": 0.604,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7125662012518055,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001630488796885961,
+      "loss": 0.5714,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7137698603755417,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00016299112344428615,
+      "loss": 0.5934,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7149735194992778,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00016293328726507152,
+      "loss": 0.612,
+      "step": 2970
+    },
+    {
+      "epoch": 0.716177178623014,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00016287537124335958,
+      "loss": 0.5918,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7173808377467501,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00016281737547168534,
+      "loss": 0.6047,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7185844968704863,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00016275930004271114,
+      "loss": 0.6189,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7197881559942224,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001627011450492266,
+      "loss": 0.6126,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7209918151179586,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.0001626429105841485,
+      "loss": 0.5967,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7221954742416947,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00016258459674052067,
+      "loss": 0.5948,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7221954742416947,
+      "eval_loss": 0.5183995962142944,
+      "eval_runtime": 2.351,
+      "eval_samples_per_second": 85.07,
+      "eval_steps_per_second": 85.07,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7233991333654309,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00016252620361151358,
+      "loss": 0.6386,
+      "step": 3005
+    },
+    {
+      "epoch": 0.724602792489167,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00016246773129042453,
+      "loss": 0.6711,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7258064516129032,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0001624091798706773,
+      "loss": 0.6083,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7270101107366393,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00016235054944582203,
+      "loss": 0.5965,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7282137698603756,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.0001622918401095351,
+      "loss": 0.593,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7294174289841117,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.000162233051955619,
+      "loss": 0.6226,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7306210881078479,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00016217418507800214,
+      "loss": 0.5963,
+      "step": 3035
+    },
+    {
+      "epoch": 0.731824747231584,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0001621152395707387,
+      "loss": 0.5788,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7330284063553202,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00016205621552800847,
+      "loss": 0.5831,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7342320654790563,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00016199711304411673,
+      "loss": 0.6308,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7354357246027925,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00016193793221349413,
+      "loss": 0.5763,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7366393837265286,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00016187867313069643,
+      "loss": 0.622,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7378430428502648,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00016181933589040447,
+      "loss": 0.6127,
+      "step": 3065
+    },
+    {
+      "epoch": 0.7390467019740009,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00016175992058742398,
+      "loss": 0.6363,
+      "step": 3070
+    },
+    {
+      "epoch": 0.7402503610977371,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00016170042731668537,
+      "loss": 0.6344,
+      "step": 3075
+    },
+    {
+      "epoch": 0.7414540202214733,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00016164085617324358,
+      "loss": 0.6138,
+      "step": 3080
+    },
+    {
+      "epoch": 0.7426576793452094,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00016158120725227808,
+      "loss": 0.5386,
+      "step": 3085
+    },
+    {
+      "epoch": 0.7438613384689456,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00016152148064909255,
+      "loss": 0.6276,
+      "step": 3090
+    },
+    {
+      "epoch": 0.7450649975926817,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00016146167645911478,
+      "loss": 0.5944,
+      "step": 3095
+    },
+    {
+      "epoch": 0.746268656716418,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00016140179477789653,
+      "loss": 0.6163,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7474723158401541,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00016134183570111336,
+      "loss": 0.6032,
+      "step": 3105
+    },
+    {
+      "epoch": 0.7486759749638903,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00016128179932456456,
+      "loss": 0.569,
+      "step": 3110
+    },
+    {
+      "epoch": 0.7498796340876264,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001612216857441728,
+      "loss": 0.6101,
+      "step": 3115
+    },
+    {
+      "epoch": 0.7510832932113626,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00016116149505598424,
+      "loss": 0.5896,
+      "step": 3120
+    },
+    {
+      "epoch": 0.7522869523350987,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00016110122735616815,
+      "loss": 0.5981,
+      "step": 3125
+    },
+    {
+      "epoch": 0.7534906114588349,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001610408827410168,
+      "loss": 0.6139,
+      "step": 3130
+    },
+    {
+      "epoch": 0.754694270582571,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.0001609804613069455,
+      "loss": 0.5773,
+      "step": 3135
+    },
+    {
+      "epoch": 0.7558979297063072,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00016091996315049215,
+      "loss": 0.6342,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7571015888300433,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00016085938836831736,
+      "loss": 0.5896,
+      "step": 3145
+    },
+    {
+      "epoch": 0.7583052479537795,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00016079873705720404,
+      "loss": 0.6124,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7595089070775156,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00016073800931405748,
+      "loss": 0.5761,
+      "step": 3155
+    },
+    {
+      "epoch": 0.7607125662012518,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00016067720523590503,
+      "loss": 0.6186,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7619162253249879,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00016061632491989604,
+      "loss": 0.604,
+      "step": 3165
+    },
+    {
+      "epoch": 0.7631198844487241,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00016055536846330161,
+      "loss": 0.5822,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7643235435724602,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00016049433596351453,
+      "loss": 0.6034,
+      "step": 3175
+    },
+    {
+      "epoch": 0.7655272026961965,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00016043322751804913,
+      "loss": 0.5935,
+      "step": 3180
+    },
+    {
+      "epoch": 0.7667308618199326,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00016037204322454101,
+      "loss": 0.6009,
+      "step": 3185
+    },
+    {
+      "epoch": 0.7679345209436688,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.000160310783180747,
+      "loss": 0.6069,
+      "step": 3190
+    },
+    {
+      "epoch": 0.7691381800674049,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00016024944748454495,
+      "loss": 0.6134,
+      "step": 3195
+    },
+    {
+      "epoch": 0.7703418391911411,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00016018803623393354,
+      "loss": 0.6333,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7715454983148772,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00016012654952703227,
+      "loss": 0.5912,
+      "step": 3205
+    },
+    {
+      "epoch": 0.7727491574386134,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00016006498746208106,
+      "loss": 0.6249,
+      "step": 3210
+    },
+    {
+      "epoch": 0.7739528165623495,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001600033501374404,
+      "loss": 0.6048,
+      "step": 3215
+    },
+    {
+      "epoch": 0.7751564756860857,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00015994163765159085,
+      "loss": 0.5762,
+      "step": 3220
+    },
+    {
+      "epoch": 0.7763601348098219,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015987985010313318,
+      "loss": 0.6221,
+      "step": 3225
+    },
+    {
+      "epoch": 0.777563793933558,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00015981798759078807,
+      "loss": 0.6514,
+      "step": 3230
+    },
+    {
+      "epoch": 0.7787674530572942,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00015975605021339588,
+      "loss": 0.6116,
+      "step": 3235
+    },
+    {
+      "epoch": 0.7799711121810303,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00015969403806991679,
+      "loss": 0.5857,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7811747713047665,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001596319512594302,
+      "loss": 0.5562,
+      "step": 3245
+    },
+    {
+      "epoch": 0.7823784304285026,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00015956978988113492,
+      "loss": 0.6025,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7835820895522388,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00015950755403434897,
+      "loss": 0.5709,
+      "step": 3255
+    },
+    {
+      "epoch": 0.784785748675975,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00015944524381850922,
+      "loss": 0.6152,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7859894077997112,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00015938285933317145,
+      "loss": 0.5825,
+      "step": 3265
+    },
+    {
+      "epoch": 0.7871930669234473,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00015932040067801007,
+      "loss": 0.5696,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7883967260471835,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00015925786795281799,
+      "loss": 0.6222,
+      "step": 3275
+    },
+    {
+      "epoch": 0.7896003851709196,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00015919526125750646,
+      "loss": 0.5668,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7908040442946558,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00015913258069210495,
+      "loss": 0.6167,
+      "step": 3285
+    },
+    {
+      "epoch": 0.7920077034183919,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00015906982635676092,
+      "loss": 0.5837,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7932113625421281,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00015900699835173977,
+      "loss": 0.6018,
+      "step": 3295
+    },
+    {
+      "epoch": 0.7944150216658642,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00015894409677742445,
+      "loss": 0.5993,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7956186807896004,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00015888112173431565,
+      "loss": 0.5965,
+      "step": 3305
+    },
+    {
+      "epoch": 0.7968223399133365,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00015881807332303126,
+      "loss": 0.5797,
+      "step": 3310
+    },
+    {
+      "epoch": 0.7980259990370727,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00015875495164430653,
+      "loss": 0.5991,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7992296581608088,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001586917567989937,
+      "loss": 0.6081,
+      "step": 3320
+    },
+    {
+      "epoch": 0.800433317284545,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00015862848888806194,
+      "loss": 0.6085,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8016369764082811,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015856514801259713,
+      "loss": 0.5954,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8028406355320173,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00015850173427380177,
+      "loss": 0.5854,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8040442946557534,
+      "grad_norm": 1.5,
+      "learning_rate": 0.0001584382477729948,
+      "loss": 0.6125,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8052479537794897,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00015837468861161133,
+      "loss": 0.5994,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001583110568912026,
+      "loss": 0.5539,
+      "step": 3350
+    },
+    {
+      "epoch": 0.807655272026962,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00015824735271343577,
+      "loss": 0.5923,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8088589311506981,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00015818357618009384,
+      "loss": 0.5316,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8100625902744343,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00015811972739307532,
+      "loss": 0.574,
+      "step": 3365
+    },
+    {
+      "epoch": 0.8112662493981705,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015805580645439422,
+      "loss": 0.591,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8124699085219066,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00015799181346617985,
+      "loss": 0.595,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8136735676456428,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00015792774853067652,
+      "loss": 0.6029,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8148772267693789,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00015786361175024362,
+      "loss": 0.5517,
+      "step": 3385
+    },
+    {
+      "epoch": 0.8160808858931151,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00015779940322735529,
+      "loss": 0.6423,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8172845450168512,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00015773512306460025,
+      "loss": 0.5942,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8184882041405874,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001576707713646818,
+      "loss": 0.5914,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8196918632643235,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00015760634823041737,
+      "loss": 0.6314,
+      "step": 3405
+    },
+    {
+      "epoch": 0.8208955223880597,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00015754185376473865,
+      "loss": 0.5919,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8220991815117958,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0001574772880706913,
+      "loss": 0.5966,
+      "step": 3415
+    },
+    {
+      "epoch": 0.823302840635532,
+      "grad_norm": 1.375,
+      "learning_rate": 0.0001574126512514347,
+      "loss": 0.5682,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8245064997592682,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00015734794341024192,
+      "loss": 0.6198,
+      "step": 3425
+    },
+    {
+      "epoch": 0.8257101588830044,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00015728316465049954,
+      "loss": 0.6307,
+      "step": 3430
+    },
+    {
+      "epoch": 0.8269138180067405,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00015721831507570734,
+      "loss": 0.5518,
+      "step": 3435
+    },
+    {
+      "epoch": 0.8281174771304767,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0001571533947894784,
+      "loss": 0.561,
+      "step": 3440
+    },
+    {
+      "epoch": 0.8293211362542128,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00015708840389553862,
+      "loss": 0.5705,
+      "step": 3445
+    },
+    {
+      "epoch": 0.830524795377949,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00015702334249772688,
+      "loss": 0.5753,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8317284545016851,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00015695821069999452,
+      "loss": 0.6035,
+      "step": 3455
+    },
+    {
+      "epoch": 0.8329321136254213,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0001568930086064055,
+      "loss": 0.6083,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8341357727491574,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00015682773632113604,
+      "loss": 0.5731,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8353394318728936,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00015676239394847458,
+      "loss": 0.5819,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8365430909966297,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00015669698159282135,
+      "loss": 0.5685,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8377467501203659,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00015663149935868866,
+      "loss": 0.5876,
+      "step": 3480
+    },
+    {
+      "epoch": 0.838950409244102,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015656594735070025,
+      "loss": 0.5996,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8401540683678382,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00015650032567359146,
+      "loss": 0.5421,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8413577274915743,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00015643463443220886,
+      "loss": 0.6113,
+      "step": 3495
+    },
+    {
+      "epoch": 0.8425613866153105,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.0001563688737315103,
+      "loss": 0.6228,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8425613866153105,
+      "eval_loss": 0.49500542879104614,
+      "eval_runtime": 2.3464,
+      "eval_samples_per_second": 85.235,
+      "eval_steps_per_second": 85.235,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8437650457390466,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00015630304367656445,
+      "loss": 0.5976,
+      "step": 3505
+    },
+    {
+      "epoch": 0.8449687048627829,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001562371443725509,
+      "loss": 0.5939,
+      "step": 3510
+    },
+    {
+      "epoch": 0.8461723639865191,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00015617117592475982,
+      "loss": 0.5936,
+      "step": 3515
+    },
+    {
+      "epoch": 0.8473760231102552,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00015610513843859189,
+      "loss": 0.5781,
+      "step": 3520
+    },
+    {
+      "epoch": 0.8485796822339914,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00015603903201955808,
+      "loss": 0.6042,
+      "step": 3525
+    },
+    {
+      "epoch": 0.8497833413577275,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001559728567732795,
+      "loss": 0.588,
+      "step": 3530
+    },
+    {
+      "epoch": 0.8509870004814637,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00015590661280548724,
+      "loss": 0.6451,
+      "step": 3535
+    },
+    {
+      "epoch": 0.8521906596051998,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00015584030022202218,
+      "loss": 0.5958,
+      "step": 3540
+    },
+    {
+      "epoch": 0.853394318728936,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001557739191288348,
+      "loss": 0.5927,
+      "step": 3545
+    },
+    {
+      "epoch": 0.8545979778526721,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00015570746963198512,
+      "loss": 0.5637,
+      "step": 3550
+    },
+    {
+      "epoch": 0.8558016369764083,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00015564095183764234,
+      "loss": 0.6069,
+      "step": 3555
+    },
+    {
+      "epoch": 0.8570052961001444,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00015557436585208489,
+      "loss": 0.5895,
+      "step": 3560
+    },
+    {
+      "epoch": 0.8582089552238806,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001555077117817001,
+      "loss": 0.6186,
+      "step": 3565
+    },
+    {
+      "epoch": 0.8594126143476167,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00015544098973298406,
+      "loss": 0.5915,
+      "step": 3570
+    },
+    {
+      "epoch": 0.8606162734713529,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00015537419981254152,
+      "loss": 0.595,
+      "step": 3575
+    },
+    {
+      "epoch": 0.861819932595089,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00015530734212708568,
+      "loss": 0.5774,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8630235917188253,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00015524041678343793,
+      "loss": 0.5634,
+      "step": 3585
+    },
+    {
+      "epoch": 0.8642272508425614,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001551734238885278,
+      "loss": 0.5802,
+      "step": 3590
+    },
+    {
+      "epoch": 0.8654309099662976,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00015510636354939284,
+      "loss": 0.6019,
+      "step": 3595
+    },
+    {
+      "epoch": 0.8666345690900337,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0001550392358731782,
+      "loss": 0.5786,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8678382282137699,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001549720409671368,
+      "loss": 0.5844,
+      "step": 3605
+    },
+    {
+      "epoch": 0.869041887337506,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00015490477893862875,
+      "loss": 0.5678,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8702455464612422,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001548374498951216,
+      "loss": 0.6117,
+      "step": 3615
+    },
+    {
+      "epoch": 0.8714492055849783,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00015477005394418992,
+      "loss": 0.5746,
+      "step": 3620
+    },
+    {
+      "epoch": 0.8726528647087145,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015470259119351512,
+      "loss": 0.5975,
+      "step": 3625
+    },
+    {
+      "epoch": 0.8738565238324506,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00015463506175088536,
+      "loss": 0.6064,
+      "step": 3630
+    },
+    {
+      "epoch": 0.8750601829561868,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00015456746572419544,
+      "loss": 0.5696,
+      "step": 3635
+    },
+    {
+      "epoch": 0.8762638420799229,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00015449980322144644,
+      "loss": 0.6189,
+      "step": 3640
+    },
+    {
+      "epoch": 0.8774675012036591,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00015443207435074572,
+      "loss": 0.5896,
+      "step": 3645
+    },
+    {
+      "epoch": 0.8786711603273952,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00015436427922030667,
+      "loss": 0.5845,
+      "step": 3650
+    },
+    {
+      "epoch": 0.8798748194511314,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00015429641793844844,
+      "loss": 0.5737,
+      "step": 3655
+    },
+    {
+      "epoch": 0.8810784785748677,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00015422849061359607,
+      "loss": 0.5527,
+      "step": 3660
+    },
+    {
+      "epoch": 0.8822821376986038,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00015416049735427994,
+      "loss": 0.5435,
+      "step": 3665
+    },
+    {
+      "epoch": 0.88348579682234,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001540924382691359,
+      "loss": 0.614,
+      "step": 3670
+    },
+    {
+      "epoch": 0.8846894559460761,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00015402431346690483,
+      "loss": 0.5945,
+      "step": 3675
+    },
+    {
+      "epoch": 0.8858931150698123,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00015395612305643282,
+      "loss": 0.5986,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8870967741935484,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001538878671466706,
+      "loss": 0.5526,
+      "step": 3685
+    },
+    {
+      "epoch": 0.8883004333172846,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001538195458466736,
+      "loss": 0.5799,
+      "step": 3690
+    },
+    {
+      "epoch": 0.8895040924410207,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00015375115926560178,
+      "loss": 0.586,
+      "step": 3695
+    },
+    {
+      "epoch": 0.8907077515647569,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00015368270751271937,
+      "loss": 0.5983,
+      "step": 3700
+    },
+    {
+      "epoch": 0.891911410688493,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001536141906973947,
+      "loss": 0.5521,
+      "step": 3705
+    },
+    {
+      "epoch": 0.8931150698122292,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00015354560892910006,
+      "loss": 0.581,
+      "step": 3710
+    },
+    {
+      "epoch": 0.8943187289359653,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00015347696231741157,
+      "loss": 0.584,
+      "step": 3715
+    },
+    {
+      "epoch": 0.8955223880597015,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00015340825097200891,
+      "loss": 0.5708,
+      "step": 3720
+    },
+    {
+      "epoch": 0.8967260471834376,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001533394750026752,
+      "loss": 0.5845,
+      "step": 3725
+    },
+    {
+      "epoch": 0.8979297063071738,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00015327063451929675,
+      "loss": 0.5956,
+      "step": 3730
+    },
+    {
+      "epoch": 0.8991333654309099,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00015320172963186312,
+      "loss": 0.5435,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9003370245546461,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00015313276045046656,
+      "loss": 0.5731,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9015406836783822,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00015306372708530217,
+      "loss": 0.5524,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9027443428021185,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00015299462964666756,
+      "loss": 0.578,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9039480019258546,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00015292546824496276,
+      "loss": 0.5885,
+      "step": 3755
+    },
+    {
+      "epoch": 0.9051516610495908,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00015285624299068992,
+      "loss": 0.5916,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9063553201733269,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00015278695399445328,
+      "loss": 0.6174,
+      "step": 3765
+    },
+    {
+      "epoch": 0.9075589792970631,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00015271760136695888,
+      "loss": 0.5735,
+      "step": 3770
+    },
+    {
+      "epoch": 0.9087626384207992,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001526481852190144,
+      "loss": 0.583,
+      "step": 3775
+    },
+    {
+      "epoch": 0.9099662975445354,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00015257870566152916,
+      "loss": 0.5883,
+      "step": 3780
+    },
+    {
+      "epoch": 0.9111699566682715,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00015250916280551356,
+      "loss": 0.579,
+      "step": 3785
+    },
+    {
+      "epoch": 0.9123736157920077,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015243955676207932,
+      "loss": 0.5754,
+      "step": 3790
+    },
+    {
+      "epoch": 0.9135772749157438,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00015236988764243906,
+      "loss": 0.561,
+      "step": 3795
+    },
+    {
+      "epoch": 0.91478093403948,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00015230015555790614,
+      "loss": 0.596,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9159845931632162,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00015223036061989454,
+      "loss": 0.5806,
+      "step": 3805
+    },
+    {
+      "epoch": 0.9171882522869523,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00015216050293991875,
+      "loss": 0.6107,
+      "step": 3810
+    },
+    {
+      "epoch": 0.9183919114106885,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00015209058262959338,
+      "loss": 0.5624,
+      "step": 3815
+    },
+    {
+      "epoch": 0.9195955705344246,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00015202059980063315,
+      "loss": 0.6023,
+      "step": 3820
+    },
+    {
+      "epoch": 0.9207992296581609,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00015195055456485276,
+      "loss": 0.5684,
+      "step": 3825
+    },
+    {
+      "epoch": 0.922002888781897,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00015188044703416646,
+      "loss": 0.589,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9232065479056332,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015181027732058812,
+      "loss": 0.5965,
+      "step": 3835
+    },
+    {
+      "epoch": 0.9244102070293693,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00015174004553623096,
+      "loss": 0.5394,
+      "step": 3840
+    },
+    {
+      "epoch": 0.9256138661531055,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0001516697517933074,
+      "loss": 0.5685,
+      "step": 3845
+    },
+    {
+      "epoch": 0.9268175252768416,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00015159939620412881,
+      "loss": 0.5824,
+      "step": 3850
+    },
+    {
+      "epoch": 0.9280211844005778,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00015152897888110538,
+      "loss": 0.6205,
+      "step": 3855
+    },
+    {
+      "epoch": 0.9292248435243139,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00015145849993674596,
+      "loss": 0.5724,
+      "step": 3860
+    },
+    {
+      "epoch": 0.9304285026480501,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00015138795948365773,
+      "loss": 0.593,
+      "step": 3865
+    },
+    {
+      "epoch": 0.9316321617717862,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0001513173576345464,
+      "loss": 0.605,
+      "step": 3870
+    },
+    {
+      "epoch": 0.9328358208955224,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015124669450221554,
+      "loss": 0.5699,
+      "step": 3875
+    },
+    {
+      "epoch": 0.9340394800192585,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00015117597019956674,
+      "loss": 0.5892,
+      "step": 3880
+    },
+    {
+      "epoch": 0.9352431391429947,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00015110518483959929,
+      "loss": 0.5434,
+      "step": 3885
+    },
+    {
+      "epoch": 0.9364467982667308,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015103433853541008,
+      "loss": 0.5682,
+      "step": 3890
+    },
+    {
+      "epoch": 0.937650457390467,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00015096343140019332,
+      "loss": 0.565,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9388541165142031,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001508924635472404,
+      "loss": 0.5885,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9400577756379394,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001508214350899399,
+      "loss": 0.5608,
+      "step": 3905
+    },
+    {
+      "epoch": 0.9412614347616755,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001507503461417769,
+      "loss": 0.6009,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9424650938854117,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001506791968163334,
+      "loss": 0.5904,
+      "step": 3915
+    },
+    {
+      "epoch": 0.9436687530091478,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00015060798722728782,
+      "loss": 0.5937,
+      "step": 3920
+    },
+    {
+      "epoch": 0.944872412132884,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001505367174884148,
+      "loss": 0.5483,
+      "step": 3925
+    },
+    {
+      "epoch": 0.9460760712566201,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00015046538771358508,
+      "loss": 0.5637,
+      "step": 3930
+    },
+    {
+      "epoch": 0.9472797303803563,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00015039399801676536,
+      "loss": 0.5642,
+      "step": 3935
+    },
+    {
+      "epoch": 0.9484833895040924,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001503225485120181,
+      "loss": 0.5515,
+      "step": 3940
+    },
+    {
+      "epoch": 0.9496870486278286,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001502510393135013,
+      "loss": 0.5478,
+      "step": 3945
+    },
+    {
+      "epoch": 0.9508907077515648,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00015017947053546835,
+      "loss": 0.5527,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9520943668753009,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00015010784229226772,
+      "loss": 0.5404,
+      "step": 3955
+    },
+    {
+      "epoch": 0.9532980259990371,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00015003615469834301,
+      "loss": 0.5384,
+      "step": 3960
+    },
+    {
+      "epoch": 0.9545016851227732,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00014996440786823272,
+      "loss": 0.5844,
+      "step": 3965
+    },
+    {
+      "epoch": 0.9557053442465094,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00014989260191656975,
+      "loss": 0.5613,
+      "step": 3970
+    },
+    {
+      "epoch": 0.9569090033702455,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001498207369580817,
+      "loss": 0.5627,
+      "step": 3975
+    },
+    {
+      "epoch": 0.9581126624939817,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0001497488131075903,
+      "loss": 0.5782,
+      "step": 3980
+    },
+    {
+      "epoch": 0.9593163216177178,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00014967683048001146,
+      "loss": 0.5931,
+      "step": 3985
+    },
+    {
+      "epoch": 0.9605199807414541,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00014960478919035495,
+      "loss": 0.5425,
+      "step": 3990
+    },
+    {
+      "epoch": 0.9617236398651902,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00014953268935372427,
+      "loss": 0.5878,
+      "step": 3995
+    },
+    {
+      "epoch": 0.9629272989889264,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00014946053108531648,
+      "loss": 0.5478,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9629272989889264,
+      "eval_loss": 0.47590532898902893,
+      "eval_runtime": 2.3469,
+      "eval_samples_per_second": 85.22,
+      "eval_steps_per_second": 85.22,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9641309581126625,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00014938831450042202,
+      "loss": 0.6152,
+      "step": 4005
+    },
+    {
+      "epoch": 0.9653346172363987,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00014931603971442444,
+      "loss": 0.5709,
+      "step": 4010
+    },
+    {
+      "epoch": 0.9665382763601348,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00014924370684280035,
+      "loss": 0.5675,
+      "step": 4015
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00014917131600111912,
+      "loss": 0.5769,
+      "step": 4020
+    },
+    {
+      "epoch": 0.9689455946076071,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001490988673050428,
+      "loss": 0.5611,
+      "step": 4025
+    },
+    {
+      "epoch": 0.9701492537313433,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001490263608703258,
+      "loss": 0.5722,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9713529128550794,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00014895379681281477,
+      "loss": 0.5755,
+      "step": 4035
+    },
+    {
+      "epoch": 0.9725565719788156,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00014888117524844855,
+      "loss": 0.5695,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9737602311025517,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001488084962932577,
+      "loss": 0.561,
+      "step": 4045
+    },
+    {
+      "epoch": 0.9749638902262879,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00014873576006336472,
+      "loss": 0.5846,
+      "step": 4050
+    },
+    {
+      "epoch": 0.976167549350024,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00014866296667498329,
+      "loss": 0.5544,
+      "step": 4055
+    },
+    {
+      "epoch": 0.9773712084737602,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0001485901162444186,
+      "loss": 0.5687,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9785748675974963,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00014851720888806706,
+      "loss": 0.5482,
+      "step": 4065
+    },
+    {
+      "epoch": 0.9797785267212326,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00014844424472241582,
+      "loss": 0.5589,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9809821858449687,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00014837122386404303,
+      "loss": 0.6161,
+      "step": 4075
+    },
+    {
+      "epoch": 0.9821858449687049,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001482981464296173,
+      "loss": 0.5831,
+      "step": 4080
+    },
+    {
+      "epoch": 0.983389504092441,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00014822501253589746,
+      "loss": 0.5635,
+      "step": 4085
+    },
+    {
+      "epoch": 0.9845931632161772,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001481518222997329,
+      "loss": 0.5789,
+      "step": 4090
+    },
+    {
+      "epoch": 0.9857968223399133,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00014807857583806283,
+      "loss": 0.5644,
+      "step": 4095
+    },
+    {
+      "epoch": 0.9870004814636495,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00014800527326791626,
+      "loss": 0.5779,
+      "step": 4100
+    },
+    {
+      "epoch": 0.9882041405873857,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00014793191470641192,
+      "loss": 0.5736,
+      "step": 4105
+    },
+    {
+      "epoch": 0.9894077997111218,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00014785850027075804,
+      "loss": 0.5663,
+      "step": 4110
+    },
+    {
+      "epoch": 0.990611458834858,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00014778503007825203,
+      "loss": 0.5569,
+      "step": 4115
+    },
+    {
+      "epoch": 0.9918151179585941,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001477115042462804,
+      "loss": 0.5779,
+      "step": 4120
+    },
+    {
+      "epoch": 0.9930187770823303,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00014763792289231858,
+      "loss": 0.5641,
+      "step": 4125
+    },
+    {
+      "epoch": 0.9942224362060664,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001475642861339308,
+      "loss": 0.5579,
+      "step": 4130
+    },
+    {
+      "epoch": 0.9954260953298026,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001474905940887696,
+      "loss": 0.5679,
+      "step": 4135
+    },
+    {
+      "epoch": 0.9966297544535387,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00014741684687457607,
+      "loss": 0.5502,
+      "step": 4140
+    },
+    {
+      "epoch": 0.997833413577275,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0001473430446091793,
+      "loss": 0.5893,
+      "step": 4145
+    },
+    {
+      "epoch": 0.999037072701011,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.0001472691874104964,
+      "loss": 0.5679,
+      "step": 4150
+    },
+    {
+      "epoch": 0.9997592681752527,
+      "eval_loss": 0.472748339176178,
+      "eval_runtime": 2.335,
+      "eval_samples_per_second": 85.653,
+      "eval_steps_per_second": 85.653,
+      "step": 4153
+    },
+    {
+      "epoch": 1.0002407318247473,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001471952753965323,
+      "loss": 0.5774,
+      "step": 4155
+    },
+    {
+      "epoch": 1.0014443909484834,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00014712130868537934,
+      "loss": 0.5232,
+      "step": 4160
+    },
+    {
+      "epoch": 1.0026480500722195,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00014704728739521744,
+      "loss": 0.5089,
+      "step": 4165
+    },
+    {
+      "epoch": 1.0038517091959558,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00014697321164431365,
+      "loss": 0.5132,
+      "step": 4170
+    },
+    {
+      "epoch": 1.0050553683196919,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.000146899081551022,
+      "loss": 0.5083,
+      "step": 4175
+    },
+    {
+      "epoch": 1.006259027443428,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00014682489723378335,
+      "loss": 0.5214,
+      "step": 4180
+    },
+    {
+      "epoch": 1.007462686567164,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00014675065881112533,
+      "loss": 0.5504,
+      "step": 4185
+    },
+    {
+      "epoch": 1.0086663456909004,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00014667636640166177,
+      "loss": 0.5096,
+      "step": 4190
+    },
+    {
+      "epoch": 1.0098700048146365,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.000146602020124093,
+      "loss": 0.5299,
+      "step": 4195
+    },
+    {
+      "epoch": 1.0110736639383726,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00014652762009720527,
+      "loss": 0.5343,
+      "step": 4200
+    },
+    {
+      "epoch": 1.0122773230621087,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00014645316643987073,
+      "loss": 0.5527,
+      "step": 4205
+    },
+    {
+      "epoch": 1.013480982185845,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00014637865927104727,
+      "loss": 0.5428,
+      "step": 4210
+    },
+    {
+      "epoch": 1.0146846413095811,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00014630409870977824,
+      "loss": 0.5018,
+      "step": 4215
+    },
+    {
+      "epoch": 1.0158883004333172,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001462294848751923,
+      "loss": 0.5233,
+      "step": 4220
+    },
+    {
+      "epoch": 1.0170919595570533,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00014615481788650322,
+      "loss": 0.5147,
+      "step": 4225
+    },
+    {
+      "epoch": 1.0182956186807897,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00014608009786300973,
+      "loss": 0.533,
+      "step": 4230
+    },
+    {
+      "epoch": 1.0194992778045258,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00014600532492409525,
+      "loss": 0.5167,
+      "step": 4235
+    },
+    {
+      "epoch": 1.0207029369282619,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00014593049918922778,
+      "loss": 0.522,
+      "step": 4240
+    },
+    {
+      "epoch": 1.021906596051998,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00014585562077795968,
+      "loss": 0.5444,
+      "step": 4245
+    },
+    {
+      "epoch": 1.0231102551757343,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001457806898099274,
+      "loss": 0.5221,
+      "step": 4250
+    },
+    {
+      "epoch": 1.0243139142994704,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00014570570640485149,
+      "loss": 0.5383,
+      "step": 4255
+    },
+    {
+      "epoch": 1.0255175734232065,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00014563067068253617,
+      "loss": 0.5406,
+      "step": 4260
+    },
+    {
+      "epoch": 1.0267212325469428,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00014555558276286928,
+      "loss": 0.5441,
+      "step": 4265
+    },
+    {
+      "epoch": 1.027924891670679,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001454804427658221,
+      "loss": 0.5236,
+      "step": 4270
+    },
+    {
+      "epoch": 1.029128550794415,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00014540525081144904,
+      "loss": 0.5415,
+      "step": 4275
+    },
+    {
+      "epoch": 1.030332209918151,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001453300070198876,
+      "loss": 0.5187,
+      "step": 4280
+    },
+    {
+      "epoch": 1.0315358690418874,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00014525471151135804,
+      "loss": 0.5211,
+      "step": 4285
+    },
+    {
+      "epoch": 1.0327395281656235,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0001451793644061633,
+      "loss": 0.4976,
+      "step": 4290
+    },
+    {
+      "epoch": 1.0339431872893596,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00014510396582468878,
+      "loss": 0.4808,
+      "step": 4295
+    },
+    {
+      "epoch": 1.0351468464130957,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00014502851588740203,
+      "loss": 0.4813,
+      "step": 4300
+    },
+    {
+      "epoch": 1.036350505536832,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001449530147148527,
+      "loss": 0.5307,
+      "step": 4305
+    },
+    {
+      "epoch": 1.0375541646605682,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00014487746242767231,
+      "loss": 0.5252,
+      "step": 4310
+    },
+    {
+      "epoch": 1.0387578237843043,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00014480185914657413,
+      "loss": 0.5193,
+      "step": 4315
+    },
+    {
+      "epoch": 1.0399614829080404,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00014472620499235276,
+      "loss": 0.5147,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0411651420317767,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001446505000858841,
+      "loss": 0.539,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0423688011555128,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00014457474454812524,
+      "loss": 0.5113,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0435724602792489,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00014449893850011406,
+      "loss": 0.4988,
+      "step": 4335
+    },
+    {
+      "epoch": 1.044776119402985,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001444230820629693,
+      "loss": 0.5076,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0459797785267213,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00014434717535788998,
+      "loss": 0.5036,
+      "step": 4345
+    },
+    {
+      "epoch": 1.0471834376504574,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00014427121850615562,
+      "loss": 0.5293,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0483870967741935,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00014419521162912577,
+      "loss": 0.495,
+      "step": 4355
+    },
+    {
+      "epoch": 1.0495907558979296,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00014411915484823998,
+      "loss": 0.4997,
+      "step": 4360
+    },
+    {
+      "epoch": 1.050794415021666,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0001440430482850174,
+      "loss": 0.5098,
+      "step": 4365
+    },
+    {
+      "epoch": 1.051998074145402,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00014396689206105692,
+      "loss": 0.5052,
+      "step": 4370
+    },
+    {
+      "epoch": 1.0532017332691381,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00014389068629803658,
+      "loss": 0.5218,
+      "step": 4375
+    },
+    {
+      "epoch": 1.0544053923928742,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00014381443111771372,
+      "loss": 0.4885,
+      "step": 4380
+    },
+    {
+      "epoch": 1.0556090515166106,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00014373812664192446,
+      "loss": 0.5334,
+      "step": 4385
+    },
+    {
+      "epoch": 1.0568127106403467,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001436617729925839,
+      "loss": 0.5438,
+      "step": 4390
+    },
+    {
+      "epoch": 1.0580163697640828,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00014358537029168555,
+      "loss": 0.5318,
+      "step": 4395
+    },
+    {
+      "epoch": 1.059220028887819,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001435089186613014,
+      "loss": 0.5104,
+      "step": 4400
+    },
+    {
+      "epoch": 1.0604236880115552,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00014343241822358145,
+      "loss": 0.5239,
+      "step": 4405
+    },
+    {
+      "epoch": 1.0616273471352913,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00014335586910075383,
+      "loss": 0.5177,
+      "step": 4410
+    },
+    {
+      "epoch": 1.0628310062590274,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00014327927141512446,
+      "loss": 0.5196,
+      "step": 4415
+    },
+    {
+      "epoch": 1.0640346653827637,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00014320262528907677,
+      "loss": 0.5273,
+      "step": 4420
+    },
+    {
+      "epoch": 1.0652383245064998,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00014312593084507167,
+      "loss": 0.5357,
+      "step": 4425
+    },
+    {
+      "epoch": 1.066441983630236,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00014304918820564713,
+      "loss": 0.5211,
+      "step": 4430
+    },
+    {
+      "epoch": 1.067645642753972,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00014297239749341828,
+      "loss": 0.561,
+      "step": 4435
+    },
+    {
+      "epoch": 1.0688493018777083,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00014289555883107696,
+      "loss": 0.5179,
+      "step": 4440
+    },
+    {
+      "epoch": 1.0700529610014444,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00014281867234139164,
+      "loss": 0.5178,
+      "step": 4445
+    },
+    {
+      "epoch": 1.0712566201251805,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00014274173814720733,
+      "loss": 0.5149,
+      "step": 4450
+    },
+    {
+      "epoch": 1.0724602792489166,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00014266475637144502,
+      "loss": 0.5222,
+      "step": 4455
+    },
+    {
+      "epoch": 1.073663938372653,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001425877271371019,
+      "loss": 0.4933,
+      "step": 4460
+    },
+    {
+      "epoch": 1.074867597496389,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00014251065056725094,
+      "loss": 0.5129,
+      "step": 4465
+    },
+    {
+      "epoch": 1.0760712566201251,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00014243352678504074,
+      "loss": 0.5195,
+      "step": 4470
+    },
+    {
+      "epoch": 1.0772749157438612,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00014235635591369536,
+      "loss": 0.51,
+      "step": 4475
+    },
+    {
+      "epoch": 1.0784785748675976,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00014227913807651402,
+      "loss": 0.4842,
+      "step": 4480
+    },
+    {
+      "epoch": 1.0796822339913337,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00014220187339687108,
+      "loss": 0.5143,
+      "step": 4485
+    },
+    {
+      "epoch": 1.0808858931150698,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00014212456199821566,
+      "loss": 0.5289,
+      "step": 4490
+    },
+    {
+      "epoch": 1.0820895522388059,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00014204720400407155,
+      "loss": 0.5034,
+      "step": 4495
+    },
+    {
+      "epoch": 1.0832932113625422,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.000141969799538037,
+      "loss": 0.5247,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0832932113625422,
+      "eval_loss": 0.4665265679359436,
+      "eval_runtime": 2.3497,
+      "eval_samples_per_second": 85.118,
+      "eval_steps_per_second": 85.118,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0844968704862783,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00014189234872378457,
+      "loss": 0.5059,
+      "step": 4505
+    },
+    {
+      "epoch": 1.0857005296100144,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00014181485168506067,
+      "loss": 0.5525,
+      "step": 4510
+    },
+    {
+      "epoch": 1.0869041887337505,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00014173730854568582,
+      "loss": 0.5133,
+      "step": 4515
+    },
+    {
+      "epoch": 1.0881078478574868,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.000141659719429554,
+      "loss": 0.5797,
+      "step": 4520
+    },
+    {
+      "epoch": 1.089311506981223,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00014158208446063278,
+      "loss": 0.5077,
+      "step": 4525
+    },
+    {
+      "epoch": 1.090515166104959,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001415044037629629,
+      "loss": 0.5252,
+      "step": 4530
+    },
+    {
+      "epoch": 1.0917188252286953,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00014142667746065826,
+      "loss": 0.5603,
+      "step": 4535
+    },
+    {
+      "epoch": 1.0929224843524314,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00014134890567790552,
+      "loss": 0.5004,
+      "step": 4540
+    },
+    {
+      "epoch": 1.0941261434761675,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00014127108853896405,
+      "loss": 0.5274,
+      "step": 4545
+    },
+    {
+      "epoch": 1.0953298025999036,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.0001411932261681657,
+      "loss": 0.4984,
+      "step": 4550
+    },
+    {
+      "epoch": 1.09653346172364,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00014111531868991458,
+      "loss": 0.5327,
+      "step": 4555
+    },
+    {
+      "epoch": 1.097737120847376,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0001410373662286869,
+      "loss": 0.5464,
+      "step": 4560
+    },
+    {
+      "epoch": 1.0989407799711122,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00014095936890903062,
+      "loss": 0.5222,
+      "step": 4565
+    },
+    {
+      "epoch": 1.1001444390948483,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00014088132685556554,
+      "loss": 0.5265,
+      "step": 4570
+    },
+    {
+      "epoch": 1.1013480982185846,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00014080324019298283,
+      "loss": 0.5023,
+      "step": 4575
+    },
+    {
+      "epoch": 1.1025517573423207,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00014072510904604497,
+      "loss": 0.4935,
+      "step": 4580
+    },
+    {
+      "epoch": 1.1037554164660568,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001406469335395855,
+      "loss": 0.5368,
+      "step": 4585
+    },
+    {
+      "epoch": 1.104959075589793,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00014056871379850884,
+      "loss": 0.522,
+      "step": 4590
+    },
+    {
+      "epoch": 1.1061627347135292,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00014049044994779005,
+      "loss": 0.491,
+      "step": 4595
+    },
+    {
+      "epoch": 1.1073663938372653,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00014041214211247475,
+      "loss": 0.5186,
+      "step": 4600
+    },
+    {
+      "epoch": 1.1085700529610014,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00014033379041767874,
+      "loss": 0.4796,
+      "step": 4605
+    },
+    {
+      "epoch": 1.1097737120847375,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.000140255394988588,
+      "loss": 0.5297,
+      "step": 4610
+    },
+    {
+      "epoch": 1.1109773712084738,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001401769559504583,
+      "loss": 0.5243,
+      "step": 4615
+    },
+    {
+      "epoch": 1.11218103033221,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.0001400984734286151,
+      "loss": 0.4982,
+      "step": 4620
+    },
+    {
+      "epoch": 1.113384689455946,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00014001994754845337,
+      "loss": 0.5347,
+      "step": 4625
+    },
+    {
+      "epoch": 1.1145883485796821,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001399413784354373,
+      "loss": 0.549,
+      "step": 4630
+    },
+    {
+      "epoch": 1.1157920077034185,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00013986276621510025,
+      "loss": 0.5295,
+      "step": 4635
+    },
+    {
+      "epoch": 1.1169956668271546,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00013978411101304438,
+      "loss": 0.5119,
+      "step": 4640
+    },
+    {
+      "epoch": 1.1181993259508907,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00013970541295494048,
+      "loss": 0.527,
+      "step": 4645
+    },
+    {
+      "epoch": 1.1194029850746268,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00013962667216652796,
+      "loss": 0.4967,
+      "step": 4650
+    },
+    {
+      "epoch": 1.120606644198363,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00013954788877361434,
+      "loss": 0.5054,
+      "step": 4655
+    },
+    {
+      "epoch": 1.1218103033220992,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00013946906290207534,
+      "loss": 0.5073,
+      "step": 4660
+    },
+    {
+      "epoch": 1.1230139624458353,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00013939019467785452,
+      "loss": 0.5399,
+      "step": 4665
+    },
+    {
+      "epoch": 1.1242176215695716,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00013931128422696296,
+      "loss": 0.483,
+      "step": 4670
+    },
+    {
+      "epoch": 1.1254212806933077,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001392323316754794,
+      "loss": 0.5337,
+      "step": 4675
+    },
+    {
+      "epoch": 1.1266249398170438,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00013915333714954986,
+      "loss": 0.5077,
+      "step": 4680
+    },
+    {
+      "epoch": 1.12782859894078,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00013907430077538723,
+      "loss": 0.5053,
+      "step": 4685
+    },
+    {
+      "epoch": 1.129032258064516,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00013899522267927142,
+      "loss": 0.5212,
+      "step": 4690
+    },
+    {
+      "epoch": 1.1302359171882523,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00013891610298754896,
+      "loss": 0.5059,
+      "step": 4695
+    },
+    {
+      "epoch": 1.1314395763119884,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001388369418266328,
+      "loss": 0.5154,
+      "step": 4700
+    },
+    {
+      "epoch": 1.1326432354357245,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00013875773932300225,
+      "loss": 0.4965,
+      "step": 4705
+    },
+    {
+      "epoch": 1.1338468945594609,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001386784956032025,
+      "loss": 0.5163,
+      "step": 4710
+    },
+    {
+      "epoch": 1.135050553683197,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001385992107938448,
+      "loss": 0.5221,
+      "step": 4715
+    },
+    {
+      "epoch": 1.136254212806933,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00013851988502160595,
+      "loss": 0.5394,
+      "step": 4720
+    },
+    {
+      "epoch": 1.1374578719306692,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001384405184132281,
+      "loss": 0.517,
+      "step": 4725
+    },
+    {
+      "epoch": 1.1386615310544055,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00013836111109551884,
+      "loss": 0.5179,
+      "step": 4730
+    },
+    {
+      "epoch": 1.1398651901781416,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00013828166319535066,
+      "loss": 0.5361,
+      "step": 4735
+    },
+    {
+      "epoch": 1.1410688493018777,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00013820217483966095,
+      "loss": 0.498,
+      "step": 4740
+    },
+    {
+      "epoch": 1.1422725084256138,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00013812264615545175,
+      "loss": 0.5299,
+      "step": 4745
+    },
+    {
+      "epoch": 1.14347616754935,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00013804307726978946,
+      "loss": 0.5236,
+      "step": 4750
+    },
+    {
+      "epoch": 1.1446798266730862,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00013796346830980484,
+      "loss": 0.5062,
+      "step": 4755
+    },
+    {
+      "epoch": 1.1458834857968223,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001378838194026925,
+      "loss": 0.5279,
+      "step": 4760
+    },
+    {
+      "epoch": 1.1470871449205584,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00013780413067571103,
+      "loss": 0.5,
+      "step": 4765
+    },
+    {
+      "epoch": 1.1482908040442947,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.0001377244022561826,
+      "loss": 0.5132,
+      "step": 4770
+    },
+    {
+      "epoch": 1.1494944631680308,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00013764463427149278,
+      "loss": 0.4908,
+      "step": 4775
+    },
+    {
+      "epoch": 1.150698122291767,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00013756482684909032,
+      "loss": 0.5248,
+      "step": 4780
+    },
+    {
+      "epoch": 1.151901781415503,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00013748498011648705,
+      "loss": 0.5286,
+      "step": 4785
+    },
+    {
+      "epoch": 1.1531054405392394,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001374050942012576,
+      "loss": 0.5167,
+      "step": 4790
+    },
+    {
+      "epoch": 1.1543090996629755,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00013732516923103917,
+      "loss": 0.5104,
+      "step": 4795
+    },
+    {
+      "epoch": 1.1555127587867116,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00013724520533353138,
+      "loss": 0.4978,
+      "step": 4800
+    },
+    {
+      "epoch": 1.1567164179104479,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00013716520263649598,
+      "loss": 0.5154,
+      "step": 4805
+    },
+    {
+      "epoch": 1.157920077034184,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00013708516126775683,
+      "loss": 0.5168,
+      "step": 4810
+    },
+    {
+      "epoch": 1.15912373615792,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001370050813551995,
+      "loss": 0.5273,
+      "step": 4815
+    },
+    {
+      "epoch": 1.1603273952816562,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00013692496302677122,
+      "loss": 0.5251,
+      "step": 4820
+    },
+    {
+      "epoch": 1.1615310544053923,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00013684480641048044,
+      "loss": 0.5318,
+      "step": 4825
+    },
+    {
+      "epoch": 1.1627347135291286,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00013676461163439696,
+      "loss": 0.4715,
+      "step": 4830
+    },
+    {
+      "epoch": 1.1639383726528647,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00013668437882665143,
+      "loss": 0.502,
+      "step": 4835
+    },
+    {
+      "epoch": 1.1651420317766008,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00013660410811543533,
+      "loss": 0.5194,
+      "step": 4840
+    },
+    {
+      "epoch": 1.1663456909003371,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00013652379962900068,
+      "loss": 0.5127,
+      "step": 4845
+    },
+    {
+      "epoch": 1.1675493500240732,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00013644345349565984,
+      "loss": 0.535,
+      "step": 4850
+    },
+    {
+      "epoch": 1.1687530091478093,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00013636306984378537,
+      "loss": 0.4666,
+      "step": 4855
+    },
+    {
+      "epoch": 1.1699566682715454,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001362826488018097,
+      "loss": 0.5204,
+      "step": 4860
+    },
+    {
+      "epoch": 1.1711603273952818,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00013620219049822503,
+      "loss": 0.4957,
+      "step": 4865
+    },
+    {
+      "epoch": 1.1723639865190179,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00013612169506158314,
+      "loss": 0.4948,
+      "step": 4870
+    },
+    {
+      "epoch": 1.173567645642754,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00013604116262049508,
+      "loss": 0.5377,
+      "step": 4875
+    },
+    {
+      "epoch": 1.17477130476649,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00013596059330363107,
+      "loss": 0.5262,
+      "step": 4880
+    },
+    {
+      "epoch": 1.1759749638902264,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001358799872397202,
+      "loss": 0.5281,
+      "step": 4885
+    },
+    {
+      "epoch": 1.1771786230139625,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001357993445575503,
+      "loss": 0.4923,
+      "step": 4890
+    },
+    {
+      "epoch": 1.1783822821376986,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001357186653859677,
+      "loss": 0.4956,
+      "step": 4895
+    },
+    {
+      "epoch": 1.1795859412614347,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001356379498538771,
+      "loss": 0.4837,
+      "step": 4900
+    },
+    {
+      "epoch": 1.180789600385171,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00013555719809024109,
+      "loss": 0.5027,
+      "step": 4905
+    },
+    {
+      "epoch": 1.181993259508907,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0001354764102240804,
+      "loss": 0.5593,
+      "step": 4910
+    },
+    {
+      "epoch": 1.1831969186326432,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001353955863844733,
+      "loss": 0.4874,
+      "step": 4915
+    },
+    {
+      "epoch": 1.1844005777563793,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00013531472670055557,
+      "loss": 0.513,
+      "step": 4920
+    },
+    {
+      "epoch": 1.1856042368801156,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00013523383130152023,
+      "loss": 0.5019,
+      "step": 4925
+    },
+    {
+      "epoch": 1.1868078960038517,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00013515290031661744,
+      "loss": 0.5332,
+      "step": 4930
+    },
+    {
+      "epoch": 1.1880115551275878,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001350719338751541,
+      "loss": 0.5233,
+      "step": 4935
+    },
+    {
+      "epoch": 1.1892152142513241,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00013499093210649388,
+      "loss": 0.4874,
+      "step": 4940
+    },
+    {
+      "epoch": 1.1904188733750602,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00013490989514005684,
+      "loss": 0.5264,
+      "step": 4945
+    },
+    {
+      "epoch": 1.1916225324987963,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00013482882310531926,
+      "loss": 0.5037,
+      "step": 4950
+    },
+    {
+      "epoch": 1.1928261916225325,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00013474771613181347,
+      "loss": 0.4813,
+      "step": 4955
+    },
+    {
+      "epoch": 1.1940298507462686,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00013466657434912764,
+      "loss": 0.5354,
+      "step": 4960
+    },
+    {
+      "epoch": 1.1952335098700049,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001345853978869055,
+      "loss": 0.5012,
+      "step": 4965
+    },
+    {
+      "epoch": 1.196437168993741,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00013450418687484627,
+      "loss": 0.4819,
+      "step": 4970
+    },
+    {
+      "epoch": 1.197640828117477,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00013442294144270427,
+      "loss": 0.5128,
+      "step": 4975
+    },
+    {
+      "epoch": 1.1988444872412134,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00013434166172028895,
+      "loss": 0.5017,
+      "step": 4980
+    },
+    {
+      "epoch": 1.2000481463649495,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001342603478374644,
+      "loss": 0.5189,
+      "step": 4985
+    },
+    {
+      "epoch": 1.2012518054886856,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001341789999241494,
+      "loss": 0.4999,
+      "step": 4990
+    },
+    {
+      "epoch": 1.2024554646124217,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00013409761811031707,
+      "loss": 0.5187,
+      "step": 4995
+    },
+    {
+      "epoch": 1.2036591237361578,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00013401620252599466,
+      "loss": 0.5358,
+      "step": 5000
+    },
+    {
+      "epoch": 1.2036591237361578,
+      "eval_loss": 0.4536990225315094,
+      "eval_runtime": 2.3545,
+      "eval_samples_per_second": 84.944,
+      "eval_steps_per_second": 84.944,
+      "step": 5000
+    },
+    {
+      "epoch": 1.2048627828598941,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001339347533012634,
+      "loss": 0.4842,
+      "step": 5005
+    },
+    {
+      "epoch": 1.2060664419836302,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001338532705662583,
+      "loss": 0.4826,
+      "step": 5010
+    },
+    {
+      "epoch": 1.2072701011073663,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00013377175445116786,
+      "loss": 0.5059,
+      "step": 5015
+    },
+    {
+      "epoch": 1.2084737602311026,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00013369020508623394,
+      "loss": 0.504,
+      "step": 5020
+    },
+    {
+      "epoch": 1.2096774193548387,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00013360862260175153,
+      "loss": 0.5286,
+      "step": 5025
+    },
+    {
+      "epoch": 1.2108810784785748,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001335270071280685,
+      "loss": 0.518,
+      "step": 5030
+    },
+    {
+      "epoch": 1.212084737602311,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001334453587955855,
+      "loss": 0.5122,
+      "step": 5035
+    },
+    {
+      "epoch": 1.2132883967260473,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00013336367773475558,
+      "loss": 0.5292,
+      "step": 5040
+    },
+    {
+      "epoch": 1.2144920558497834,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00013328196407608414,
+      "loss": 0.5213,
+      "step": 5045
+    },
+    {
+      "epoch": 1.2156957149735195,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00013320021795012871,
+      "loss": 0.5076,
+      "step": 5050
+    },
+    {
+      "epoch": 1.2168993740972556,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0001331184394874986,
+      "loss": 0.5195,
+      "step": 5055
+    },
+    {
+      "epoch": 1.218103033220992,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00013303662881885487,
+      "loss": 0.5016,
+      "step": 5060
+    },
+    {
+      "epoch": 1.219306692344728,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00013295478607490996,
+      "loss": 0.4838,
+      "step": 5065
+    },
+    {
+      "epoch": 1.220510351468464,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00013287291138642761,
+      "loss": 0.5046,
+      "step": 5070
+    },
+    {
+      "epoch": 1.2217140105922002,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00013279100488422255,
+      "loss": 0.5018,
+      "step": 5075
+    },
+    {
+      "epoch": 1.2229176697159365,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00013270906669916042,
+      "loss": 0.5398,
+      "step": 5080
+    },
+    {
+      "epoch": 1.2241213288396726,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00013262709696215742,
+      "loss": 0.4706,
+      "step": 5085
+    },
+    {
+      "epoch": 1.2253249879634087,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00013254509580418015,
+      "loss": 0.5214,
+      "step": 5090
+    },
+    {
+      "epoch": 1.2265286470871448,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00013246306335624547,
+      "loss": 0.5027,
+      "step": 5095
+    },
+    {
+      "epoch": 1.2277323062108811,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00013238099974942022,
+      "loss": 0.5232,
+      "step": 5100
+    },
+    {
+      "epoch": 1.2289359653346172,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00013229890511482094,
+      "loss": 0.4929,
+      "step": 5105
+    },
+    {
+      "epoch": 1.2301396244583533,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001322167795836139,
+      "loss": 0.5032,
+      "step": 5110
+    },
+    {
+      "epoch": 1.2313432835820897,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00013213462328701454,
+      "loss": 0.5256,
+      "step": 5115
+    },
+    {
+      "epoch": 1.2325469427058258,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00013205243635628765,
+      "loss": 0.5033,
+      "step": 5120
+    },
+    {
+      "epoch": 1.2337506018295619,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00013197021892274685,
+      "loss": 0.4942,
+      "step": 5125
+    },
+    {
+      "epoch": 1.234954260953298,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001318879711177545,
+      "loss": 0.4887,
+      "step": 5130
+    },
+    {
+      "epoch": 1.236157920077034,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00013180569307272156,
+      "loss": 0.4551,
+      "step": 5135
+    },
+    {
+      "epoch": 1.2373615792007704,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00013172338491910718,
+      "loss": 0.5323,
+      "step": 5140
+    },
+    {
+      "epoch": 1.2385652383245065,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00013164104678841873,
+      "loss": 0.5175,
+      "step": 5145
+    },
+    {
+      "epoch": 1.2397688974482426,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00013155867881221145,
+      "loss": 0.5004,
+      "step": 5150
+    },
+    {
+      "epoch": 1.240972556571979,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00013147628112208825,
+      "loss": 0.4757,
+      "step": 5155
+    },
+    {
+      "epoch": 1.242176215695715,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001313938538496995,
+      "loss": 0.5226,
+      "step": 5160
+    },
+    {
+      "epoch": 1.2433798748194511,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00013131139712674282,
+      "loss": 0.496,
+      "step": 5165
+    },
+    {
+      "epoch": 1.2445835339431872,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00013122891108496295,
+      "loss": 0.5161,
+      "step": 5170
+    },
+    {
+      "epoch": 1.2457871930669235,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00013114639585615144,
+      "loss": 0.498,
+      "step": 5175
+    },
+    {
+      "epoch": 1.2469908521906596,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0001310638515721465,
+      "loss": 0.517,
+      "step": 5180
+    },
+    {
+      "epoch": 1.2481945113143957,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00013098127836483266,
+      "loss": 0.4963,
+      "step": 5185
+    },
+    {
+      "epoch": 1.2493981704381318,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0001308986763661408,
+      "loss": 0.4906,
+      "step": 5190
+    },
+    {
+      "epoch": 1.2506018295618682,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00013081604570804771,
+      "loss": 0.5061,
+      "step": 5195
+    },
+    {
+      "epoch": 1.2518054886856043,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00013073338652257603,
+      "loss": 0.5145,
+      "step": 5200
+    },
+    {
+      "epoch": 1.2530091478093404,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001306506989417939,
+      "loss": 0.5046,
+      "step": 5205
+    },
+    {
+      "epoch": 1.2542128069330767,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00013056798309781493,
+      "loss": 0.5017,
+      "step": 5210
+    },
+    {
+      "epoch": 1.2554164660568128,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0001304852391227978,
+      "loss": 0.4801,
+      "step": 5215
+    },
+    {
+      "epoch": 1.2566201251805489,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00013040246714894616,
+      "loss": 0.5122,
+      "step": 5220
+    },
+    {
+      "epoch": 1.257823784304285,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00013031966730850844,
+      "loss": 0.4989,
+      "step": 5225
+    },
+    {
+      "epoch": 1.259027443428021,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001302368397337776,
+      "loss": 0.4914,
+      "step": 5230
+    },
+    {
+      "epoch": 1.2602311025517574,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00013015398455709073,
+      "loss": 0.5161,
+      "step": 5235
+    },
+    {
+      "epoch": 1.2614347616754935,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001300711019108293,
+      "loss": 0.515,
+      "step": 5240
+    },
+    {
+      "epoch": 1.2626384207992296,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00012998819192741845,
+      "loss": 0.5001,
+      "step": 5245
+    },
+    {
+      "epoch": 1.263842079922966,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001299052547393271,
+      "loss": 0.5229,
+      "step": 5250
+    },
+    {
+      "epoch": 1.265045739046702,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00012982229047906764,
+      "loss": 0.5002,
+      "step": 5255
+    },
+    {
+      "epoch": 1.2662493981704381,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001297392992791957,
+      "loss": 0.5128,
+      "step": 5260
+    },
+    {
+      "epoch": 1.2674530572941742,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001296562812723099,
+      "loss": 0.5437,
+      "step": 5265
+    },
+    {
+      "epoch": 1.2686567164179103,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00012957323659105172,
+      "loss": 0.504,
+      "step": 5270
+    },
+    {
+      "epoch": 1.2698603755416467,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00012949016536810534,
+      "loss": 0.4892,
+      "step": 5275
+    },
+    {
+      "epoch": 1.2710640346653828,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00012940706773619722,
+      "loss": 0.5222,
+      "step": 5280
+    },
+    {
+      "epoch": 1.2722676937891189,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001293239438280961,
+      "loss": 0.5247,
+      "step": 5285
+    },
+    {
+      "epoch": 1.2734713529128552,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00012924079377661267,
+      "loss": 0.4893,
+      "step": 5290
+    },
+    {
+      "epoch": 1.2746750120365913,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00012915761771459938,
+      "loss": 0.5049,
+      "step": 5295
+    },
+    {
+      "epoch": 1.2758786711603274,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00012907441577495027,
+      "loss": 0.5179,
+      "step": 5300
+    },
+    {
+      "epoch": 1.2770823302840635,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00012899118809060072,
+      "loss": 0.499,
+      "step": 5305
+    },
+    {
+      "epoch": 1.2782859894077996,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00012890793479452726,
+      "loss": 0.5587,
+      "step": 5310
+    },
+    {
+      "epoch": 1.279489648531536,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00012882465601974722,
+      "loss": 0.4966,
+      "step": 5315
+    },
+    {
+      "epoch": 1.280693307655272,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00012874135189931883,
+      "loss": 0.5036,
+      "step": 5320
+    },
+    {
+      "epoch": 1.281896966779008,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00012865802256634067,
+      "loss": 0.5005,
+      "step": 5325
+    },
+    {
+      "epoch": 1.2831006259027444,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001285746681539517,
+      "loss": 0.5167,
+      "step": 5330
+    },
+    {
+      "epoch": 1.2843042850264805,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00012849128879533083,
+      "loss": 0.5216,
+      "step": 5335
+    },
+    {
+      "epoch": 1.2855079441502166,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00012840788462369695,
+      "loss": 0.5221,
+      "step": 5340
+    },
+    {
+      "epoch": 1.286711603273953,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00012832445577230854,
+      "loss": 0.4875,
+      "step": 5345
+    },
+    {
+      "epoch": 1.287915262397689,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00012824100237446352,
+      "loss": 0.4996,
+      "step": 5350
+    },
+    {
+      "epoch": 1.2891189215214252,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.000128157524563499,
+      "loss": 0.5301,
+      "step": 5355
+    },
+    {
+      "epoch": 1.2903225806451613,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00012807402247279115,
+      "loss": 0.5297,
+      "step": 5360
+    },
+    {
+      "epoch": 1.2915262397688974,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00012799049623575488,
+      "loss": 0.4997,
+      "step": 5365
+    },
+    {
+      "epoch": 1.2927298988926337,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00012790694598584373,
+      "loss": 0.4706,
+      "step": 5370
+    },
+    {
+      "epoch": 1.2939335580163698,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00012782337185654957,
+      "loss": 0.4911,
+      "step": 5375
+    },
+    {
+      "epoch": 1.2951372171401059,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001277397739814024,
+      "loss": 0.5139,
+      "step": 5380
+    },
+    {
+      "epoch": 1.2963408762638422,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001276561524939703,
+      "loss": 0.5281,
+      "step": 5385
+    },
+    {
+      "epoch": 1.2975445353875783,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00012757250752785886,
+      "loss": 0.518,
+      "step": 5390
+    },
+    {
+      "epoch": 1.2987481945113144,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00012748883921671132,
+      "loss": 0.5116,
+      "step": 5395
+    },
+    {
+      "epoch": 1.2999518536350505,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00012740514769420824,
+      "loss": 0.4575,
+      "step": 5400
+    },
+    {
+      "epoch": 1.3011555127587866,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00012732143309406713,
+      "loss": 0.5117,
+      "step": 5405
+    },
+    {
+      "epoch": 1.302359171882523,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001272376955500426,
+      "loss": 0.5249,
+      "step": 5410
+    },
+    {
+      "epoch": 1.303562831006259,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00012715393519592563,
+      "loss": 0.4805,
+      "step": 5415
+    },
+    {
+      "epoch": 1.3047664901299951,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00012707015216554388,
+      "loss": 0.4965,
+      "step": 5420
+    },
+    {
+      "epoch": 1.3059701492537314,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00012698634659276113,
+      "loss": 0.4759,
+      "step": 5425
+    },
+    {
+      "epoch": 1.3071738083774675,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0001269025186114772,
+      "loss": 0.4972,
+      "step": 5430
+    },
+    {
+      "epoch": 1.3083774675012037,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001268186683556277,
+      "loss": 0.5039,
+      "step": 5435
+    },
+    {
+      "epoch": 1.3095811266249398,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00012673479595918388,
+      "loss": 0.5127,
+      "step": 5440
+    },
+    {
+      "epoch": 1.3107847857486759,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0001266509015561523,
+      "loss": 0.5255,
+      "step": 5445
+    },
+    {
+      "epoch": 1.3119884448724122,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00012656698528057473,
+      "loss": 0.4879,
+      "step": 5450
+    },
+    {
+      "epoch": 1.3131921039961483,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00012648304726652787,
+      "loss": 0.4959,
+      "step": 5455
+    },
+    {
+      "epoch": 1.3143957631198844,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001263990876481231,
+      "loss": 0.5016,
+      "step": 5460
+    },
+    {
+      "epoch": 1.3155994222436207,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00012631510655950644,
+      "loss": 0.5054,
+      "step": 5465
+    },
+    {
+      "epoch": 1.3168030813673568,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00012623110413485808,
+      "loss": 0.4687,
+      "step": 5470
+    },
+    {
+      "epoch": 1.318006740491093,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001261470805083924,
+      "loss": 0.4957,
+      "step": 5475
+    },
+    {
+      "epoch": 1.319210399614829,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00012606303581435757,
+      "loss": 0.4938,
+      "step": 5480
+    },
+    {
+      "epoch": 1.320414058738565,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00012597897018703548,
+      "loss": 0.4824,
+      "step": 5485
+    },
+    {
+      "epoch": 1.3216177178623014,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00012589488376074152,
+      "loss": 0.5155,
+      "step": 5490
+    },
+    {
+      "epoch": 1.3228213769860375,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00012581077666982415,
+      "loss": 0.4871,
+      "step": 5495
+    },
+    {
+      "epoch": 1.3240250361097736,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.000125726649048665,
+      "loss": 0.503,
+      "step": 5500
+    },
+    {
+      "epoch": 1.3240250361097736,
+      "eval_loss": 0.4387615919113159,
+      "eval_runtime": 2.3516,
+      "eval_samples_per_second": 85.047,
+      "eval_steps_per_second": 85.047,
+      "step": 5500
+    },
+    {
+      "epoch": 1.32522869523351,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0001256425010316784,
+      "loss": 0.515,
+      "step": 5505
+    },
+    {
+      "epoch": 1.326432354357246,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00012555833275331135,
+      "loss": 0.5221,
+      "step": 5510
+    },
+    {
+      "epoch": 1.3276360134809821,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00012547414434804316,
+      "loss": 0.4718,
+      "step": 5515
+    },
+    {
+      "epoch": 1.3288396726047185,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00012538993595038537,
+      "loss": 0.4931,
+      "step": 5520
+    },
+    {
+      "epoch": 1.3300433317284546,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0001253057076948813,
+      "loss": 0.5102,
+      "step": 5525
+    },
+    {
+      "epoch": 1.3312469908521907,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001252214597161062,
+      "loss": 0.4856,
+      "step": 5530
+    },
+    {
+      "epoch": 1.3324506499759268,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001251371921486667,
+      "loss": 0.4995,
+      "step": 5535
+    },
+    {
+      "epoch": 1.3336543090996629,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00012505290512720082,
+      "loss": 0.4825,
+      "step": 5540
+    },
+    {
+      "epoch": 1.3348579682233992,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001249685987863776,
+      "loss": 0.5072,
+      "step": 5545
+    },
+    {
+      "epoch": 1.3360616273471353,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001248842732608969,
+      "loss": 0.5069,
+      "step": 5550
+    },
+    {
+      "epoch": 1.3372652864708714,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00012479992868548936,
+      "loss": 0.4969,
+      "step": 5555
+    },
+    {
+      "epoch": 1.3384689455946077,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00012471556519491592,
+      "loss": 0.5154,
+      "step": 5560
+    },
+    {
+      "epoch": 1.3396726047183438,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001246311829239679,
+      "loss": 0.5276,
+      "step": 5565
+    },
+    {
+      "epoch": 1.34087626384208,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0001245467820074665,
+      "loss": 0.4811,
+      "step": 5570
+    },
+    {
+      "epoch": 1.342079922965816,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00012446236258026276,
+      "loss": 0.5205,
+      "step": 5575
+    },
+    {
+      "epoch": 1.3432835820895521,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00012437792477723724,
+      "loss": 0.4981,
+      "step": 5580
+    },
+    {
+      "epoch": 1.3444872412132884,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00012429346873329993,
+      "loss": 0.4763,
+      "step": 5585
+    },
+    {
+      "epoch": 1.3456909003370245,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001242089945833899,
+      "loss": 0.4831,
+      "step": 5590
+    },
+    {
+      "epoch": 1.3468945594607606,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00012412450246247528,
+      "loss": 0.538,
+      "step": 5595
+    },
+    {
+      "epoch": 1.348098218584497,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00012403999250555273,
+      "loss": 0.5154,
+      "step": 5600
+    },
+    {
+      "epoch": 1.349301877708233,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00012395546484764757,
+      "loss": 0.4815,
+      "step": 5605
+    },
+    {
+      "epoch": 1.3505055368319692,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00012387091962381324,
+      "loss": 0.4986,
+      "step": 5610
+    },
+    {
+      "epoch": 1.3517091959557053,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001237863569691314,
+      "loss": 0.4928,
+      "step": 5615
+    },
+    {
+      "epoch": 1.3529128550794414,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00012370177701871149,
+      "loss": 0.4881,
+      "step": 5620
+    },
+    {
+      "epoch": 1.3541165142031777,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00012361717990769057,
+      "loss": 0.5174,
+      "step": 5625
+    },
+    {
+      "epoch": 1.3553201733269138,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00012353256577123315,
+      "loss": 0.4902,
+      "step": 5630
+    },
+    {
+      "epoch": 1.35652383245065,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001234479347445309,
+      "loss": 0.4853,
+      "step": 5635
+    },
+    {
+      "epoch": 1.3577274915743862,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00012336328696280254,
+      "loss": 0.5045,
+      "step": 5640
+    },
+    {
+      "epoch": 1.3589311506981223,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001232786225612935,
+      "loss": 0.5301,
+      "step": 5645
+    },
+    {
+      "epoch": 1.3601348098218584,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00012319394167527583,
+      "loss": 0.528,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3613384689455947,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0001231092444400478,
+      "loss": 0.493,
+      "step": 5655
+    },
+    {
+      "epoch": 1.3625421280693308,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.000123024530990934,
+      "loss": 0.4953,
+      "step": 5660
+    },
+    {
+      "epoch": 1.363745787193067,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00012293980146328468,
+      "loss": 0.493,
+      "step": 5665
+    },
+    {
+      "epoch": 1.364949446316803,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00012285505599247598,
+      "loss": 0.487,
+      "step": 5670
+    },
+    {
+      "epoch": 1.3661531054405391,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00012277029471390944,
+      "loss": 0.4955,
+      "step": 5675
+    },
+    {
+      "epoch": 1.3673567645642755,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001226855177630118,
+      "loss": 0.4748,
+      "step": 5680
+    },
+    {
+      "epoch": 1.3685604236880116,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00012260072527523491,
+      "loss": 0.5096,
+      "step": 5685
+    },
+    {
+      "epoch": 1.3697640828117477,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00012251591738605548,
+      "loss": 0.4603,
+      "step": 5690
+    },
+    {
+      "epoch": 1.370967741935484,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00012243109423097472,
+      "loss": 0.5298,
+      "step": 5695
+    },
+    {
+      "epoch": 1.37217140105922,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00012234625594551832,
+      "loss": 0.5053,
+      "step": 5700
+    },
+    {
+      "epoch": 1.3733750601829562,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001222614026652361,
+      "loss": 0.4938,
+      "step": 5705
+    },
+    {
+      "epoch": 1.3745787193066923,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00012217653452570185,
+      "loss": 0.4735,
+      "step": 5710
+    },
+    {
+      "epoch": 1.3757823784304284,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00012209165166251308,
+      "loss": 0.5206,
+      "step": 5715
+    },
+    {
+      "epoch": 1.3769860375541647,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00012200675421129087,
+      "loss": 0.484,
+      "step": 5720
+    },
+    {
+      "epoch": 1.3781896966779008,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00012192184230767959,
+      "loss": 0.5006,
+      "step": 5725
+    },
+    {
+      "epoch": 1.379393355801637,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001218369160873467,
+      "loss": 0.4873,
+      "step": 5730
+    },
+    {
+      "epoch": 1.3805970149253732,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001217519756859825,
+      "loss": 0.5143,
+      "step": 5735
+    },
+    {
+      "epoch": 1.3818006740491093,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00012166702123929998,
+      "loss": 0.5015,
+      "step": 5740
+    },
+    {
+      "epoch": 1.3830043331728454,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00012158205288303457,
+      "loss": 0.467,
+      "step": 5745
+    },
+    {
+      "epoch": 1.3842079922965815,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00012149707075294393,
+      "loss": 0.5055,
+      "step": 5750
+    },
+    {
+      "epoch": 1.3854116514203176,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00012141207498480772,
+      "loss": 0.5042,
+      "step": 5755
+    },
+    {
+      "epoch": 1.386615310544054,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00012132706571442743,
+      "loss": 0.5204,
+      "step": 5760
+    },
+    {
+      "epoch": 1.38781896966779,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00012124204307762598,
+      "loss": 0.4756,
+      "step": 5765
+    },
+    {
+      "epoch": 1.3890226287915262,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00012115700721024783,
+      "loss": 0.4829,
+      "step": 5770
+    },
+    {
+      "epoch": 1.3902262879152625,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00012107195824815846,
+      "loss": 0.5153,
+      "step": 5775
+    },
+    {
+      "epoch": 1.3914299470389986,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00012098689632724433,
+      "loss": 0.4649,
+      "step": 5780
+    },
+    {
+      "epoch": 1.3926336061627347,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00012090182158341258,
+      "loss": 0.5136,
+      "step": 5785
+    },
+    {
+      "epoch": 1.393837265286471,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00012081673415259083,
+      "loss": 0.4893,
+      "step": 5790
+    },
+    {
+      "epoch": 1.395040924410207,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00012073163417072698,
+      "loss": 0.4856,
+      "step": 5795
+    },
+    {
+      "epoch": 1.3962445835339432,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.000120646521773789,
+      "loss": 0.4911,
+      "step": 5800
+    },
+    {
+      "epoch": 1.3974482426576793,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00012056139709776468,
+      "loss": 0.5176,
+      "step": 5805
+    },
+    {
+      "epoch": 1.3986519017814154,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00012047626027866143,
+      "loss": 0.5161,
+      "step": 5810
+    },
+    {
+      "epoch": 1.3998555609051517,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00012039111145250606,
+      "loss": 0.5026,
+      "step": 5815
+    },
+    {
+      "epoch": 1.4010592200288878,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00012030595075534455,
+      "loss": 0.4906,
+      "step": 5820
+    },
+    {
+      "epoch": 1.402262879152624,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00012022077832324187,
+      "loss": 0.4992,
+      "step": 5825
+    },
+    {
+      "epoch": 1.4034665382763603,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00012013559429228176,
+      "loss": 0.5064,
+      "step": 5830
+    },
+    {
+      "epoch": 1.4046701974000964,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00012005039879856641,
+      "loss": 0.5087,
+      "step": 5835
+    },
+    {
+      "epoch": 1.4058738565238325,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00011996519197821648,
+      "loss": 0.4936,
+      "step": 5840
+    },
+    {
+      "epoch": 1.4070775156475686,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00011987997396737051,
+      "loss": 0.4609,
+      "step": 5845
+    },
+    {
+      "epoch": 1.4082811747713047,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00011979474490218507,
+      "loss": 0.4639,
+      "step": 5850
+    },
+    {
+      "epoch": 1.409484833895041,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00011970950491883439,
+      "loss": 0.4731,
+      "step": 5855
+    },
+    {
+      "epoch": 1.410688493018777,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00011962425415351009,
+      "loss": 0.467,
+      "step": 5860
+    },
+    {
+      "epoch": 1.4118921521425132,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00011953899274242107,
+      "loss": 0.4699,
+      "step": 5865
+    },
+    {
+      "epoch": 1.4130958112662495,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00011945372082179316,
+      "loss": 0.477,
+      "step": 5870
+    },
+    {
+      "epoch": 1.4142994703899856,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00011936843852786902,
+      "loss": 0.4717,
+      "step": 5875
+    },
+    {
+      "epoch": 1.4155031295137217,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00011928314599690795,
+      "loss": 0.4816,
+      "step": 5880
+    },
+    {
+      "epoch": 1.4167067886374578,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00011919784336518553,
+      "loss": 0.4641,
+      "step": 5885
+    },
+    {
+      "epoch": 1.417910447761194,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00011911253076899347,
+      "loss": 0.4853,
+      "step": 5890
+    },
+    {
+      "epoch": 1.4191141068849302,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001190272083446395,
+      "loss": 0.4879,
+      "step": 5895
+    },
+    {
+      "epoch": 1.4203177660086663,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00011894187622844685,
+      "loss": 0.4662,
+      "step": 5900
+    },
+    {
+      "epoch": 1.4215214251324024,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001188565345567545,
+      "loss": 0.4762,
+      "step": 5905
+    },
+    {
+      "epoch": 1.4227250842561388,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001187711834659165,
+      "loss": 0.4984,
+      "step": 5910
+    },
+    {
+      "epoch": 1.4239287433798749,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00011868582309230205,
+      "loss": 0.5133,
+      "step": 5915
+    },
+    {
+      "epoch": 1.425132402503611,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00011860045357229508,
+      "loss": 0.4922,
+      "step": 5920
+    },
+    {
+      "epoch": 1.4263360616273473,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00011851507504229425,
+      "loss": 0.4734,
+      "step": 5925
+    },
+    {
+      "epoch": 1.4275397207510834,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00011842968763871255,
+      "loss": 0.4833,
+      "step": 5930
+    },
+    {
+      "epoch": 1.4287433798748195,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00011834429149797717,
+      "loss": 0.4907,
+      "step": 5935
+    },
+    {
+      "epoch": 1.4299470389985556,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00011825888675652923,
+      "loss": 0.4982,
+      "step": 5940
+    },
+    {
+      "epoch": 1.4311506981222917,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00011817347355082364,
+      "loss": 0.493,
+      "step": 5945
+    },
+    {
+      "epoch": 1.432354357246028,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00011808805201732878,
+      "loss": 0.4807,
+      "step": 5950
+    },
+    {
+      "epoch": 1.433558016369764,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001180026222925264,
+      "loss": 0.5133,
+      "step": 5955
+    },
+    {
+      "epoch": 1.4347616754935002,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00011791718451291126,
+      "loss": 0.5,
+      "step": 5960
+    },
+    {
+      "epoch": 1.4359653346172365,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011783173881499106,
+      "loss": 0.4896,
+      "step": 5965
+    },
+    {
+      "epoch": 1.4371689937409726,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00011774628533528607,
+      "loss": 0.469,
+      "step": 5970
+    },
+    {
+      "epoch": 1.4383726528647087,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00011766082421032915,
+      "loss": 0.4911,
+      "step": 5975
+    },
+    {
+      "epoch": 1.4395763119884448,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00011757535557666514,
+      "loss": 0.511,
+      "step": 5980
+    },
+    {
+      "epoch": 1.440779971112181,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00011748987957085109,
+      "loss": 0.4991,
+      "step": 5985
+    },
+    {
+      "epoch": 1.4419836302359172,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00011740439632945572,
+      "loss": 0.4857,
+      "step": 5990
+    },
+    {
+      "epoch": 1.4431872893596533,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00011731890598905935,
+      "loss": 0.5143,
+      "step": 5995
+    },
+    {
+      "epoch": 1.4443909484833894,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00011723340868625362,
+      "loss": 0.5001,
+      "step": 6000
+    },
+    {
+      "epoch": 1.4443909484833894,
+      "eval_loss": 0.4288594126701355,
+      "eval_runtime": 2.3521,
+      "eval_samples_per_second": 85.03,
+      "eval_steps_per_second": 85.03,
+      "step": 6000
+    },
+    {
+      "epoch": 1.4455946076071258,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00011714790455764127,
+      "loss": 0.4968,
+      "step": 6005
+    },
+    {
+      "epoch": 1.4467982667308619,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00011706239373983604,
+      "loss": 0.4785,
+      "step": 6010
+    },
+    {
+      "epoch": 1.448001925854598,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00011697687636946226,
+      "loss": 0.478,
+      "step": 6015
+    },
+    {
+      "epoch": 1.449205584978334,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00011689135258315475,
+      "loss": 0.4734,
+      "step": 6020
+    },
+    {
+      "epoch": 1.4504092441020702,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00011680582251755865,
+      "loss": 0.5085,
+      "step": 6025
+    },
+    {
+      "epoch": 1.4516129032258065,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00011672028630932902,
+      "loss": 0.4644,
+      "step": 6030
+    },
+    {
+      "epoch": 1.4528165623495426,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0001166347440951308,
+      "loss": 0.4911,
+      "step": 6035
+    },
+    {
+      "epoch": 1.4540202214732787,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00011654919601163857,
+      "loss": 0.455,
+      "step": 6040
+    },
+    {
+      "epoch": 1.455223880597015,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00011646364219553618,
+      "loss": 0.4715,
+      "step": 6045
+    },
+    {
+      "epoch": 1.4564275397207511,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00011637808278351675,
+      "loss": 0.4931,
+      "step": 6050
+    },
+    {
+      "epoch": 1.4576311988444872,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00011629251791228225,
+      "loss": 0.521,
+      "step": 6055
+    },
+    {
+      "epoch": 1.4588348579682233,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001162069477185434,
+      "loss": 0.4943,
+      "step": 6060
+    },
+    {
+      "epoch": 1.4600385170919594,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00011612137233901946,
+      "loss": 0.4828,
+      "step": 6065
+    },
+    {
+      "epoch": 1.4612421762156957,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00011603579191043795,
+      "loss": 0.4734,
+      "step": 6070
+    },
+    {
+      "epoch": 1.4624458353394318,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00011595020656953443,
+      "loss": 0.488,
+      "step": 6075
+    },
+    {
+      "epoch": 1.463649494463168,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001158646164530524,
+      "loss": 0.5259,
+      "step": 6080
+    },
+    {
+      "epoch": 1.4648531535869043,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001157790216977428,
+      "loss": 0.4766,
+      "step": 6085
+    },
+    {
+      "epoch": 1.4660568127106404,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00011569342244036422,
+      "loss": 0.472,
+      "step": 6090
+    },
+    {
+      "epoch": 1.4672604718343765,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00011560781881768228,
+      "loss": 0.4969,
+      "step": 6095
+    },
+    {
+      "epoch": 1.4684641309581128,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00011552221096646963,
+      "loss": 0.4899,
+      "step": 6100
+    },
+    {
+      "epoch": 1.469667790081849,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00011543659902350574,
+      "loss": 0.4906,
+      "step": 6105
+    },
+    {
+      "epoch": 1.470871449205585,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001153509831255764,
+      "loss": 0.4671,
+      "step": 6110
+    },
+    {
+      "epoch": 1.472075108329321,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00011526536340947398,
+      "loss": 0.504,
+      "step": 6115
+    },
+    {
+      "epoch": 1.4732787674530572,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00011517974001199682,
+      "loss": 0.5031,
+      "step": 6120
+    },
+    {
+      "epoch": 1.4744824265767935,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00011509411306994916,
+      "loss": 0.4837,
+      "step": 6125
+    },
+    {
+      "epoch": 1.4756860857005296,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00011500848272014087,
+      "loss": 0.4904,
+      "step": 6130
+    },
+    {
+      "epoch": 1.4768897448242657,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00011492284909938733,
+      "loss": 0.5054,
+      "step": 6135
+    },
+    {
+      "epoch": 1.478093403948002,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00011483721234450912,
+      "loss": 0.513,
+      "step": 6140
+    },
+    {
+      "epoch": 1.4792970630717381,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00011475157259233177,
+      "loss": 0.5087,
+      "step": 6145
+    },
+    {
+      "epoch": 1.4805007221954742,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00011466592997968568,
+      "loss": 0.5239,
+      "step": 6150
+    },
+    {
+      "epoch": 1.4817043813192103,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00011458028464340583,
+      "loss": 0.452,
+      "step": 6155
+    },
+    {
+      "epoch": 1.4829080404429464,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0001144946367203315,
+      "loss": 0.4743,
+      "step": 6160
+    },
+    {
+      "epoch": 1.4841116995666828,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00011440898634730602,
+      "loss": 0.499,
+      "step": 6165
+    },
+    {
+      "epoch": 1.4853153586904189,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00011432333366117681,
+      "loss": 0.513,
+      "step": 6170
+    },
+    {
+      "epoch": 1.486519017814155,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00011423767879879492,
+      "loss": 0.489,
+      "step": 6175
+    },
+    {
+      "epoch": 1.4877226769378913,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00011415202189701486,
+      "loss": 0.4873,
+      "step": 6180
+    },
+    {
+      "epoch": 1.4889263360616274,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00011406636309269433,
+      "loss": 0.4876,
+      "step": 6185
+    },
+    {
+      "epoch": 1.4901299951853635,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00011398070252269422,
+      "loss": 0.4835,
+      "step": 6190
+    },
+    {
+      "epoch": 1.4913336543090996,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001138950403238781,
+      "loss": 0.4885,
+      "step": 6195
+    },
+    {
+      "epoch": 1.4925373134328357,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00011380937663311228,
+      "loss": 0.5127,
+      "step": 6200
+    },
+    {
+      "epoch": 1.493740972556572,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001137237115872653,
+      "loss": 0.4743,
+      "step": 6205
+    },
+    {
+      "epoch": 1.4949446316803081,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.000113638045323208,
+      "loss": 0.5125,
+      "step": 6210
+    },
+    {
+      "epoch": 1.4961482908040442,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00011355237797781306,
+      "loss": 0.4641,
+      "step": 6215
+    },
+    {
+      "epoch": 1.4973519499277805,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00011346670968795497,
+      "loss": 0.489,
+      "step": 6220
+    },
+    {
+      "epoch": 1.4985556090515166,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00011338104059050968,
+      "loss": 0.4705,
+      "step": 6225
+    },
+    {
+      "epoch": 1.4997592681752527,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00011329537082235446,
+      "loss": 0.4795,
+      "step": 6230
+    },
+    {
+      "epoch": 1.500962927298989,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00011320970052036765,
+      "loss": 0.4708,
+      "step": 6235
+    },
+    {
+      "epoch": 1.502166586422725,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00011312402982142839,
+      "loss": 0.4994,
+      "step": 6240
+    },
+    {
+      "epoch": 1.5033702455464613,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00011303835886241655,
+      "loss": 0.4612,
+      "step": 6245
+    },
+    {
+      "epoch": 1.5045739046701974,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011295268778021228,
+      "loss": 0.4355,
+      "step": 6250
+    },
+    {
+      "epoch": 1.5057775637939335,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00011286701671169608,
+      "loss": 0.4846,
+      "step": 6255
+    },
+    {
+      "epoch": 1.5069812229176698,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00011278134579374833,
+      "loss": 0.489,
+      "step": 6260
+    },
+    {
+      "epoch": 1.5081848820414059,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00011269567516324919,
+      "loss": 0.4974,
+      "step": 6265
+    },
+    {
+      "epoch": 1.509388541165142,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00011261000495707838,
+      "loss": 0.5069,
+      "step": 6270
+    },
+    {
+      "epoch": 1.5105922002888783,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00011252433531211492,
+      "loss": 0.494,
+      "step": 6275
+    },
+    {
+      "epoch": 1.5117958594126142,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00011243866636523691,
+      "loss": 0.4812,
+      "step": 6280
+    },
+    {
+      "epoch": 1.5129995185363505,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00011235299825332142,
+      "loss": 0.5093,
+      "step": 6285
+    },
+    {
+      "epoch": 1.5142031776600868,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011226733111324412,
+      "loss": 0.4921,
+      "step": 6290
+    },
+    {
+      "epoch": 1.5154068367838227,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00011218166508187913,
+      "loss": 0.4636,
+      "step": 6295
+    },
+    {
+      "epoch": 1.516610495907559,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00011209600029609879,
+      "loss": 0.4773,
+      "step": 6300
+    },
+    {
+      "epoch": 1.5178141550312951,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00011201033689277348,
+      "loss": 0.4698,
+      "step": 6305
+    },
+    {
+      "epoch": 1.5190178141550312,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00011192467500877134,
+      "loss": 0.4947,
+      "step": 6310
+    },
+    {
+      "epoch": 1.5202214732787676,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00011183901478095815,
+      "loss": 0.5108,
+      "step": 6315
+    },
+    {
+      "epoch": 1.5214251324025037,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00011175335634619695,
+      "loss": 0.5156,
+      "step": 6320
+    },
+    {
+      "epoch": 1.5226287915262398,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.000111667699841348,
+      "loss": 0.4832,
+      "step": 6325
+    },
+    {
+      "epoch": 1.523832450649976,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00011158204540326837,
+      "loss": 0.4964,
+      "step": 6330
+    },
+    {
+      "epoch": 1.525036109773712,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00011149639316881197,
+      "loss": 0.4952,
+      "step": 6335
+    },
+    {
+      "epoch": 1.5262397688974483,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00011141074327482907,
+      "loss": 0.4872,
+      "step": 6340
+    },
+    {
+      "epoch": 1.5274434280211844,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011132509585816626,
+      "loss": 0.4515,
+      "step": 6345
+    },
+    {
+      "epoch": 1.5286470871449205,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00011123945105566613,
+      "loss": 0.4838,
+      "step": 6350
+    },
+    {
+      "epoch": 1.5298507462686568,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00011115380900416715,
+      "loss": 0.4489,
+      "step": 6355
+    },
+    {
+      "epoch": 1.531054405392393,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00011106816984050336,
+      "loss": 0.4812,
+      "step": 6360
+    },
+    {
+      "epoch": 1.532258064516129,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00011098253370150419,
+      "loss": 0.4542,
+      "step": 6365
+    },
+    {
+      "epoch": 1.5334617236398653,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00011089690072399423,
+      "loss": 0.4335,
+      "step": 6370
+    },
+    {
+      "epoch": 1.5346653827636012,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00011081127104479302,
+      "loss": 0.5006,
+      "step": 6375
+    },
+    {
+      "epoch": 1.5358690418873375,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00011072564480071481,
+      "loss": 0.442,
+      "step": 6380
+    },
+    {
+      "epoch": 1.5370727010110736,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00011064002212856843,
+      "loss": 0.4852,
+      "step": 6385
+    },
+    {
+      "epoch": 1.5382763601348097,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00011055440316515699,
+      "loss": 0.4801,
+      "step": 6390
+    },
+    {
+      "epoch": 1.539480019258546,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011046878804727757,
+      "loss": 0.51,
+      "step": 6395
+    },
+    {
+      "epoch": 1.5406836783822822,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00011038317691172122,
+      "loss": 0.4616,
+      "step": 6400
+    },
+    {
+      "epoch": 1.5418873375060183,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00011029756989527258,
+      "loss": 0.4843,
+      "step": 6405
+    },
+    {
+      "epoch": 1.5430909966297546,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00011021196713470973,
+      "loss": 0.4987,
+      "step": 6410
+    },
+    {
+      "epoch": 1.5442946557534905,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00011012636876680393,
+      "loss": 0.4663,
+      "step": 6415
+    },
+    {
+      "epoch": 1.5454983148772268,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00011004077492831943,
+      "loss": 0.5064,
+      "step": 6420
+    },
+    {
+      "epoch": 1.5467019740009629,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001099551857560132,
+      "loss": 0.4703,
+      "step": 6425
+    },
+    {
+      "epoch": 1.547905633124699,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00010986960138663487,
+      "loss": 0.5145,
+      "step": 6430
+    },
+    {
+      "epoch": 1.5491092922484353,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00010978402195692629,
+      "loss": 0.4831,
+      "step": 6435
+    },
+    {
+      "epoch": 1.5503129513721714,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00010969844760362141,
+      "loss": 0.4762,
+      "step": 6440
+    },
+    {
+      "epoch": 1.5515166104959075,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00010961287846344617,
+      "loss": 0.476,
+      "step": 6445
+    },
+    {
+      "epoch": 1.5527202696196438,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010952731467311808,
+      "loss": 0.4894,
+      "step": 6450
+    },
+    {
+      "epoch": 1.55392392874338,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010944175636934618,
+      "loss": 0.483,
+      "step": 6455
+    },
+    {
+      "epoch": 1.555127587867116,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001093562036888307,
+      "loss": 0.4539,
+      "step": 6460
+    },
+    {
+      "epoch": 1.5563312469908523,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00010927065676826285,
+      "loss": 0.4881,
+      "step": 6465
+    },
+    {
+      "epoch": 1.5575349061145882,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00010918511574432468,
+      "loss": 0.4733,
+      "step": 6470
+    },
+    {
+      "epoch": 1.5587385652383245,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00010909958075368888,
+      "loss": 0.4957,
+      "step": 6475
+    },
+    {
+      "epoch": 1.5599422243620606,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00010901405193301842,
+      "loss": 0.4715,
+      "step": 6480
+    },
+    {
+      "epoch": 1.5611458834857967,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00010892852941896639,
+      "loss": 0.4872,
+      "step": 6485
+    },
+    {
+      "epoch": 1.562349542609533,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00010884301334817582,
+      "loss": 0.5172,
+      "step": 6490
+    },
+    {
+      "epoch": 1.5635532017332692,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00010875750385727954,
+      "loss": 0.4736,
+      "step": 6495
+    },
+    {
+      "epoch": 1.5647568608570053,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001086720010828998,
+      "loss": 0.5011,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5647568608570053,
+      "eval_loss": 0.4124249219894409,
+      "eval_runtime": 2.3537,
+      "eval_samples_per_second": 84.974,
+      "eval_steps_per_second": 84.974,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5659605199807416,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00010858650516164806,
+      "loss": 0.4938,
+      "step": 6505
+    },
+    {
+      "epoch": 1.5671641791044775,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00010850101623012493,
+      "loss": 0.4868,
+      "step": 6510
+    },
+    {
+      "epoch": 1.5683678382282138,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00010841553442491976,
+      "loss": 0.4827,
+      "step": 6515
+    },
+    {
+      "epoch": 1.56957149735195,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00010833005988261058,
+      "loss": 0.4808,
+      "step": 6520
+    },
+    {
+      "epoch": 1.570775156475686,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00010824459273976385,
+      "loss": 0.4448,
+      "step": 6525
+    },
+    {
+      "epoch": 1.5719788155994223,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00010815913313293407,
+      "loss": 0.5128,
+      "step": 6530
+    },
+    {
+      "epoch": 1.5731824747231584,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00010807368119866385,
+      "loss": 0.4939,
+      "step": 6535
+    },
+    {
+      "epoch": 1.5743861338468945,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001079882370734834,
+      "loss": 0.4713,
+      "step": 6540
+    },
+    {
+      "epoch": 1.5755897929706308,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010790280089391061,
+      "loss": 0.4881,
+      "step": 6545
+    },
+    {
+      "epoch": 1.5767934520943667,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010781737279645057,
+      "loss": 0.4741,
+      "step": 6550
+    },
+    {
+      "epoch": 1.577997111218103,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00010773195291759545,
+      "loss": 0.4708,
+      "step": 6555
+    },
+    {
+      "epoch": 1.5792007703418391,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010764654139382435,
+      "loss": 0.467,
+      "step": 6560
+    },
+    {
+      "epoch": 1.5804044294655752,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00010756113836160297,
+      "loss": 0.4721,
+      "step": 6565
+    },
+    {
+      "epoch": 1.5816080885893116,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010747574395738347,
+      "loss": 0.4874,
+      "step": 6570
+    },
+    {
+      "epoch": 1.5828117477130477,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010739035831760425,
+      "loss": 0.5051,
+      "step": 6575
+    },
+    {
+      "epoch": 1.5840154068367838,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00010730498157868957,
+      "loss": 0.4875,
+      "step": 6580
+    },
+    {
+      "epoch": 1.58521906596052,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00010721961387704972,
+      "loss": 0.4675,
+      "step": 6585
+    },
+    {
+      "epoch": 1.5864227250842562,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00010713425534908028,
+      "loss": 0.4837,
+      "step": 6590
+    },
+    {
+      "epoch": 1.5876263842079923,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00010704890613116234,
+      "loss": 0.4638,
+      "step": 6595
+    },
+    {
+      "epoch": 1.5888300433317286,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00010696356635966209,
+      "loss": 0.4632,
+      "step": 6600
+    },
+    {
+      "epoch": 1.5900337024554645,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00010687823617093058,
+      "loss": 0.4934,
+      "step": 6605
+    },
+    {
+      "epoch": 1.5912373615792008,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00010679291570130362,
+      "loss": 0.476,
+      "step": 6610
+    },
+    {
+      "epoch": 1.592441020702937,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00010670760508710139,
+      "loss": 0.5106,
+      "step": 6615
+    },
+    {
+      "epoch": 1.593644679826673,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010662230446462838,
+      "loss": 0.4749,
+      "step": 6620
+    },
+    {
+      "epoch": 1.5948483389504093,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001065370139701732,
+      "loss": 0.4272,
+      "step": 6625
+    },
+    {
+      "epoch": 1.5960519980741454,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001064517337400081,
+      "loss": 0.4753,
+      "step": 6630
+    },
+    {
+      "epoch": 1.5972556571978815,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001063664639103891,
+      "loss": 0.4324,
+      "step": 6635
+    },
+    {
+      "epoch": 1.5984593163216179,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00010628120461755546,
+      "loss": 0.4633,
+      "step": 6640
+    },
+    {
+      "epoch": 1.5996629754453537,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001061959559977297,
+      "loss": 0.4943,
+      "step": 6645
+    },
+    {
+      "epoch": 1.60086663456909,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010611071818711735,
+      "loss": 0.4756,
+      "step": 6650
+    },
+    {
+      "epoch": 1.6020702936928262,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001060254913219064,
+      "loss": 0.4873,
+      "step": 6655
+    },
+    {
+      "epoch": 1.6032739528165623,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010594027553826772,
+      "loss": 0.5023,
+      "step": 6660
+    },
+    {
+      "epoch": 1.6044776119402986,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00010585507097235414,
+      "loss": 0.5088,
+      "step": 6665
+    },
+    {
+      "epoch": 1.6056812710640347,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00010576987776030081,
+      "loss": 0.5043,
+      "step": 6670
+    },
+    {
+      "epoch": 1.6068849301877708,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001056846960382246,
+      "loss": 0.4875,
+      "step": 6675
+    },
+    {
+      "epoch": 1.608088589311507,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00010559952594222405,
+      "loss": 0.522,
+      "step": 6680
+    },
+    {
+      "epoch": 1.609292248435243,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010551436760837918,
+      "loss": 0.4958,
+      "step": 6685
+    },
+    {
+      "epoch": 1.6104959075589793,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00010542922117275117,
+      "loss": 0.4595,
+      "step": 6690
+    },
+    {
+      "epoch": 1.6116995666827154,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00010534408677138218,
+      "loss": 0.4668,
+      "step": 6695
+    },
+    {
+      "epoch": 1.6129032258064515,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00010525896454029514,
+      "loss": 0.4807,
+      "step": 6700
+    },
+    {
+      "epoch": 1.6141068849301878,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00010517385461549358,
+      "loss": 0.4873,
+      "step": 6705
+    },
+    {
+      "epoch": 1.615310544053924,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010508875713296138,
+      "loss": 0.4647,
+      "step": 6710
+    },
+    {
+      "epoch": 1.61651420317766,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00010500367222866245,
+      "loss": 0.427,
+      "step": 6715
+    },
+    {
+      "epoch": 1.6177178623013964,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010491860003854065,
+      "loss": 0.5016,
+      "step": 6720
+    },
+    {
+      "epoch": 1.6189215214251322,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00010483354069851956,
+      "loss": 0.4748,
+      "step": 6725
+    },
+    {
+      "epoch": 1.6201251805488686,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00010474849434450218,
+      "loss": 0.4419,
+      "step": 6730
+    },
+    {
+      "epoch": 1.6213288396726049,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00010466346111237082,
+      "loss": 0.4921,
+      "step": 6735
+    },
+    {
+      "epoch": 1.6225324987963408,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00010457844113798672,
+      "loss": 0.4725,
+      "step": 6740
+    },
+    {
+      "epoch": 1.623736157920077,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00010449343455719007,
+      "loss": 0.4811,
+      "step": 6745
+    },
+    {
+      "epoch": 1.6249398170438132,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00010440844150579957,
+      "loss": 0.4848,
+      "step": 6750
+    },
+    {
+      "epoch": 1.6261434761675493,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001043234621196123,
+      "loss": 0.4992,
+      "step": 6755
+    },
+    {
+      "epoch": 1.6273471352912856,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00010423849653440359,
+      "loss": 0.484,
+      "step": 6760
+    },
+    {
+      "epoch": 1.6285507944150217,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00010415354488592661,
+      "loss": 0.4921,
+      "step": 6765
+    },
+    {
+      "epoch": 1.6297544535387578,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00010406860730991234,
+      "loss": 0.4796,
+      "step": 6770
+    },
+    {
+      "epoch": 1.6309581126624941,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00010398368394206926,
+      "loss": 0.4552,
+      "step": 6775
+    },
+    {
+      "epoch": 1.63216177178623,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00010389877491808314,
+      "loss": 0.4549,
+      "step": 6780
+    },
+    {
+      "epoch": 1.6333654309099663,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010381388037361683,
+      "loss": 0.4465,
+      "step": 6785
+    },
+    {
+      "epoch": 1.6345690900337024,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00010372900044431003,
+      "loss": 0.4927,
+      "step": 6790
+    },
+    {
+      "epoch": 1.6357727491574385,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010364413526577915,
+      "loss": 0.4789,
+      "step": 6795
+    },
+    {
+      "epoch": 1.6369764082811749,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00010355928497361696,
+      "loss": 0.4934,
+      "step": 6800
+    },
+    {
+      "epoch": 1.638180067404911,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00010347444970339244,
+      "loss": 0.4821,
+      "step": 6805
+    },
+    {
+      "epoch": 1.639383726528647,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00010338962959065069,
+      "loss": 0.461,
+      "step": 6810
+    },
+    {
+      "epoch": 1.6405873856523834,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00010330482477091243,
+      "loss": 0.4625,
+      "step": 6815
+    },
+    {
+      "epoch": 1.6417910447761193,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00010322003537967406,
+      "loss": 0.4604,
+      "step": 6820
+    },
+    {
+      "epoch": 1.6429947038998556,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001031352615524073,
+      "loss": 0.4654,
+      "step": 6825
+    },
+    {
+      "epoch": 1.6441983630235917,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00010305050342455897,
+      "loss": 0.486,
+      "step": 6830
+    },
+    {
+      "epoch": 1.6454020221473278,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00010296576113155084,
+      "loss": 0.4581,
+      "step": 6835
+    },
+    {
+      "epoch": 1.646605681271064,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00010288103480877935,
+      "loss": 0.4845,
+      "step": 6840
+    },
+    {
+      "epoch": 1.6478093403948002,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00010279632459161548,
+      "loss": 0.4575,
+      "step": 6845
+    },
+    {
+      "epoch": 1.6490129995185363,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00010271163061540445,
+      "loss": 0.5021,
+      "step": 6850
+    },
+    {
+      "epoch": 1.6502166586422726,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00010262695301546543,
+      "loss": 0.4466,
+      "step": 6855
+    },
+    {
+      "epoch": 1.6514203177660085,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00010254229192709164,
+      "loss": 0.4566,
+      "step": 6860
+    },
+    {
+      "epoch": 1.6526239768897448,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001024576474855497,
+      "loss": 0.4657,
+      "step": 6865
+    },
+    {
+      "epoch": 1.653827636013481,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00010237301982607981,
+      "loss": 0.5034,
+      "step": 6870
+    },
+    {
+      "epoch": 1.655031295137217,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001022884090838952,
+      "loss": 0.4257,
+      "step": 6875
+    },
+    {
+      "epoch": 1.6562349542609534,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00010220381539418218,
+      "loss": 0.4822,
+      "step": 6880
+    },
+    {
+      "epoch": 1.6574386133846895,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00010211923889209983,
+      "loss": 0.5075,
+      "step": 6885
+    },
+    {
+      "epoch": 1.6586422725084256,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001020346797127796,
+      "loss": 0.4946,
+      "step": 6890
+    },
+    {
+      "epoch": 1.6598459316321619,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001019501379913255,
+      "loss": 0.4642,
+      "step": 6895
+    },
+    {
+      "epoch": 1.661049590755898,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00010186561386281345,
+      "loss": 0.4639,
+      "step": 6900
+    },
+    {
+      "epoch": 1.662253249879634,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001017811074622914,
+      "loss": 0.4764,
+      "step": 6905
+    },
+    {
+      "epoch": 1.6634569090033704,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00010169661892477887,
+      "loss": 0.4607,
+      "step": 6910
+    },
+    {
+      "epoch": 1.6646605681271063,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00010161214838526686,
+      "loss": 0.4905,
+      "step": 6915
+    },
+    {
+      "epoch": 1.6658642272508426,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00010152769597871774,
+      "loss": 0.4737,
+      "step": 6920
+    },
+    {
+      "epoch": 1.6670678863745787,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010144326184006468,
+      "loss": 0.4716,
+      "step": 6925
+    },
+    {
+      "epoch": 1.6682715454983148,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001013588461042119,
+      "loss": 0.5117,
+      "step": 6930
+    },
+    {
+      "epoch": 1.6694752046220511,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010127444890603404,
+      "loss": 0.4528,
+      "step": 6935
+    },
+    {
+      "epoch": 1.6706788637457872,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00010119007038037618,
+      "loss": 0.4587,
+      "step": 6940
+    },
+    {
+      "epoch": 1.6718825228695233,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00010110571066205365,
+      "loss": 0.4969,
+      "step": 6945
+    },
+    {
+      "epoch": 1.6730861819932596,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00010102136988585153,
+      "loss": 0.4338,
+      "step": 6950
+    },
+    {
+      "epoch": 1.6742898411169955,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00010093704818652488,
+      "loss": 0.4835,
+      "step": 6955
+    },
+    {
+      "epoch": 1.6754935002407318,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001008527456987981,
+      "loss": 0.448,
+      "step": 6960
+    },
+    {
+      "epoch": 1.676697159364468,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00010076846255736495,
+      "loss": 0.494,
+      "step": 6965
+    },
+    {
+      "epoch": 1.677900818488204,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00010068419889688831,
+      "loss": 0.4518,
+      "step": 6970
+    },
+    {
+      "epoch": 1.6791044776119404,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00010059995485199988,
+      "loss": 0.4678,
+      "step": 6975
+    },
+    {
+      "epoch": 1.6803081367356765,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00010051573055730008,
+      "loss": 0.4377,
+      "step": 6980
+    },
+    {
+      "epoch": 1.6815117958594126,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00010043152614735774,
+      "loss": 0.4676,
+      "step": 6985
+    },
+    {
+      "epoch": 1.682715454983149,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00010034734175670987,
+      "loss": 0.4992,
+      "step": 6990
+    },
+    {
+      "epoch": 1.6839191141068848,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00010026317751986162,
+      "loss": 0.4562,
+      "step": 6995
+    },
+    {
+      "epoch": 1.685122773230621,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010017903357128577,
+      "loss": 0.4707,
+      "step": 7000
+    },
+    {
+      "epoch": 1.685122773230621,
+      "eval_loss": 0.40290209650993347,
+      "eval_runtime": 2.3488,
+      "eval_samples_per_second": 85.151,
+      "eval_steps_per_second": 85.151,
+      "step": 7000
+    },
+    {
+      "epoch": 1.6863264323543572,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001000949100454229,
+      "loss": 0.4751,
+      "step": 7005
+    },
+    {
+      "epoch": 1.6875300914780933,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00010001080707668073,
+      "loss": 0.4645,
+      "step": 7010
+    },
+    {
+      "epoch": 1.6887337506018296,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.99267247994343e-05,
+      "loss": 0.4886,
+      "step": 7015
+    },
+    {
+      "epoch": 1.6899374097255657,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.984266334802554e-05,
+      "loss": 0.4251,
+      "step": 7020
+    },
+    {
+      "epoch": 1.6911410688493018,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.975862285676306e-05,
+      "loss": 0.4545,
+      "step": 7025
+    },
+    {
+      "epoch": 1.6923447279730381,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.967460345992207e-05,
+      "loss": 0.4555,
+      "step": 7030
+    },
+    {
+      "epoch": 1.6935483870967742,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.959060529174396e-05,
+      "loss": 0.4874,
+      "step": 7035
+    },
+    {
+      "epoch": 1.6947520462205103,
+      "grad_norm": 1.125,
+      "learning_rate": 9.950662848643635e-05,
+      "loss": 0.4654,
+      "step": 7040
+    },
+    {
+      "epoch": 1.6959557053442467,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.942267317817261e-05,
+      "loss": 0.4806,
+      "step": 7045
+    },
+    {
+      "epoch": 1.6971593644679825,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.933873950109178e-05,
+      "loss": 0.4521,
+      "step": 7050
+    },
+    {
+      "epoch": 1.6983630235917189,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.925482758929841e-05,
+      "loss": 0.445,
+      "step": 7055
+    },
+    {
+      "epoch": 1.699566682715455,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.917093757686215e-05,
+      "loss": 0.4613,
+      "step": 7060
+    },
+    {
+      "epoch": 1.700770341839191,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.908706959781779e-05,
+      "loss": 0.4411,
+      "step": 7065
+    },
+    {
+      "epoch": 1.7019740009629274,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.900322378616488e-05,
+      "loss": 0.4746,
+      "step": 7070
+    },
+    {
+      "epoch": 1.7031776600866635,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.891940027586747e-05,
+      "loss": 0.4972,
+      "step": 7075
+    },
+    {
+      "epoch": 1.7043813192103996,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.883559920085408e-05,
+      "loss": 0.4312,
+      "step": 7080
+    },
+    {
+      "epoch": 1.705584978334136,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.875182069501733e-05,
+      "loss": 0.4539,
+      "step": 7085
+    },
+    {
+      "epoch": 1.7067886374578718,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.866806489221378e-05,
+      "loss": 0.4543,
+      "step": 7090
+    },
+    {
+      "epoch": 1.7079922965816081,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.858433192626381e-05,
+      "loss": 0.4403,
+      "step": 7095
+    },
+    {
+      "epoch": 1.7091959557053442,
+      "grad_norm": 1.125,
+      "learning_rate": 9.850062193095115e-05,
+      "loss": 0.4739,
+      "step": 7100
+    },
+    {
+      "epoch": 1.7103996148290803,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.841693504002295e-05,
+      "loss": 0.4529,
+      "step": 7105
+    },
+    {
+      "epoch": 1.7116032739528166,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.833327138718938e-05,
+      "loss": 0.485,
+      "step": 7110
+    },
+    {
+      "epoch": 1.7128069330765527,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.824963110612354e-05,
+      "loss": 0.4641,
+      "step": 7115
+    },
+    {
+      "epoch": 1.7140105922002888,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.816601433046117e-05,
+      "loss": 0.4743,
+      "step": 7120
+    },
+    {
+      "epoch": 1.7152142513240252,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.808242119380038e-05,
+      "loss": 0.4361,
+      "step": 7125
+    },
+    {
+      "epoch": 1.716417910447761,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.799885182970167e-05,
+      "loss": 0.5017,
+      "step": 7130
+    },
+    {
+      "epoch": 1.7176215695714974,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.791530637168735e-05,
+      "loss": 0.4815,
+      "step": 7135
+    },
+    {
+      "epoch": 1.7188252286952335,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.783178495324173e-05,
+      "loss": 0.4753,
+      "step": 7140
+    },
+    {
+      "epoch": 1.7200288878189696,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.774828770781058e-05,
+      "loss": 0.4376,
+      "step": 7145
+    },
+    {
+      "epoch": 1.7212325469427059,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.766481476880111e-05,
+      "loss": 0.4664,
+      "step": 7150
+    },
+    {
+      "epoch": 1.722436206066442,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.75813662695817e-05,
+      "loss": 0.4686,
+      "step": 7155
+    },
+    {
+      "epoch": 1.723639865190178,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.749794234348159e-05,
+      "loss": 0.4841,
+      "step": 7160
+    },
+    {
+      "epoch": 1.7248435243139144,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.741454312379094e-05,
+      "loss": 0.4352,
+      "step": 7165
+    },
+    {
+      "epoch": 1.7260471834376505,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.733116874376018e-05,
+      "loss": 0.4566,
+      "step": 7170
+    },
+    {
+      "epoch": 1.7272508425613866,
+      "grad_norm": 1.25,
+      "learning_rate": 9.724781933660033e-05,
+      "loss": 0.5107,
+      "step": 7175
+    },
+    {
+      "epoch": 1.728454501685123,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.716449503548231e-05,
+      "loss": 0.445,
+      "step": 7180
+    },
+    {
+      "epoch": 1.7296581608088588,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.708119597353697e-05,
+      "loss": 0.4901,
+      "step": 7185
+    },
+    {
+      "epoch": 1.7308618199325951,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.699792228385495e-05,
+      "loss": 0.5007,
+      "step": 7190
+    },
+    {
+      "epoch": 1.7320654790563312,
+      "grad_norm": 1.125,
+      "learning_rate": 9.691467409948614e-05,
+      "loss": 0.444,
+      "step": 7195
+    },
+    {
+      "epoch": 1.7332691381800673,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.683145155343987e-05,
+      "loss": 0.4686,
+      "step": 7200
+    },
+    {
+      "epoch": 1.7344727973038037,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.674825477868445e-05,
+      "loss": 0.4509,
+      "step": 7205
+    },
+    {
+      "epoch": 1.7356764564275398,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.666508390814692e-05,
+      "loss": 0.4759,
+      "step": 7210
+    },
+    {
+      "epoch": 1.7368801155512759,
+      "grad_norm": 1.2890625,
+      "learning_rate": 9.658193907471311e-05,
+      "loss": 0.4304,
+      "step": 7215
+    },
+    {
+      "epoch": 1.7380837746750122,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.649882041122704e-05,
+      "loss": 0.4717,
+      "step": 7220
+    },
+    {
+      "epoch": 1.739287433798748,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.641572805049113e-05,
+      "loss": 0.4895,
+      "step": 7225
+    },
+    {
+      "epoch": 1.7404910929224844,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.633266212526563e-05,
+      "loss": 0.4505,
+      "step": 7230
+    },
+    {
+      "epoch": 1.7416947520462205,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.624962276826856e-05,
+      "loss": 0.4761,
+      "step": 7235
+    },
+    {
+      "epoch": 1.7428984111699566,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.616661011217558e-05,
+      "loss": 0.4738,
+      "step": 7240
+    },
+    {
+      "epoch": 1.744102070293693,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.608362428961959e-05,
+      "loss": 0.4677,
+      "step": 7245
+    },
+    {
+      "epoch": 1.745305729417429,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.600066543319069e-05,
+      "loss": 0.4675,
+      "step": 7250
+    },
+    {
+      "epoch": 1.746509388541165,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.591773367543587e-05,
+      "loss": 0.4501,
+      "step": 7255
+    },
+    {
+      "epoch": 1.7477130476649014,
+      "grad_norm": 1.125,
+      "learning_rate": 9.583482914885877e-05,
+      "loss": 0.4613,
+      "step": 7260
+    },
+    {
+      "epoch": 1.7489167067886373,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.575195198591965e-05,
+      "loss": 0.478,
+      "step": 7265
+    },
+    {
+      "epoch": 1.7501203659123736,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.566910231903487e-05,
+      "loss": 0.4599,
+      "step": 7270
+    },
+    {
+      "epoch": 1.7513240250361097,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.558628028057707e-05,
+      "loss": 0.4457,
+      "step": 7275
+    },
+    {
+      "epoch": 1.7525276841598458,
+      "grad_norm": 1.125,
+      "learning_rate": 9.550348600287457e-05,
+      "loss": 0.4465,
+      "step": 7280
+    },
+    {
+      "epoch": 1.7537313432835822,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.542071961821139e-05,
+      "loss": 0.4525,
+      "step": 7285
+    },
+    {
+      "epoch": 1.7549350024073183,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.533798125882704e-05,
+      "loss": 0.451,
+      "step": 7290
+    },
+    {
+      "epoch": 1.7561386615310544,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.525527105691617e-05,
+      "loss": 0.446,
+      "step": 7295
+    },
+    {
+      "epoch": 1.7573423206547907,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.51725891446285e-05,
+      "loss": 0.4422,
+      "step": 7300
+    },
+    {
+      "epoch": 1.7585459797785266,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.508993565406854e-05,
+      "loss": 0.4566,
+      "step": 7305
+    },
+    {
+      "epoch": 1.7597496389022629,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.500731071729535e-05,
+      "loss": 0.493,
+      "step": 7310
+    },
+    {
+      "epoch": 1.7609532980259992,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.492471446632245e-05,
+      "loss": 0.4364,
+      "step": 7315
+    },
+    {
+      "epoch": 1.762156957149735,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.484214703311737e-05,
+      "loss": 0.4506,
+      "step": 7320
+    },
+    {
+      "epoch": 1.7633606162734714,
+      "grad_norm": 1.125,
+      "learning_rate": 9.475960854960182e-05,
+      "loss": 0.4845,
+      "step": 7325
+    },
+    {
+      "epoch": 1.7645642753972075,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.46770991476511e-05,
+      "loss": 0.448,
+      "step": 7330
+    },
+    {
+      "epoch": 1.7657679345209436,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.459461895909403e-05,
+      "loss": 0.4888,
+      "step": 7335
+    },
+    {
+      "epoch": 1.76697159364468,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.451216811571288e-05,
+      "loss": 0.4663,
+      "step": 7340
+    },
+    {
+      "epoch": 1.768175252768416,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.44297467492429e-05,
+      "loss": 0.4857,
+      "step": 7345
+    },
+    {
+      "epoch": 1.7693789118921521,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.434735499137234e-05,
+      "loss": 0.446,
+      "step": 7350
+    },
+    {
+      "epoch": 1.7705825710158885,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.426499297374213e-05,
+      "loss": 0.4694,
+      "step": 7355
+    },
+    {
+      "epoch": 1.7717862301396243,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.418266082794563e-05,
+      "loss": 0.4568,
+      "step": 7360
+    },
+    {
+      "epoch": 1.7729898892633607,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.41003586855285e-05,
+      "loss": 0.4902,
+      "step": 7365
+    },
+    {
+      "epoch": 1.7741935483870968,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.401808667798844e-05,
+      "loss": 0.4444,
+      "step": 7370
+    },
+    {
+      "epoch": 1.7753972075108329,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.393584493677512e-05,
+      "loss": 0.4728,
+      "step": 7375
+    },
+    {
+      "epoch": 1.7766008666345692,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.385363359328963e-05,
+      "loss": 0.4763,
+      "step": 7380
+    },
+    {
+      "epoch": 1.7778045257583053,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.377145277888473e-05,
+      "loss": 0.4566,
+      "step": 7385
+    },
+    {
+      "epoch": 1.7790081848820414,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.368930262486427e-05,
+      "loss": 0.4589,
+      "step": 7390
+    },
+    {
+      "epoch": 1.7802118440057777,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.360718326248307e-05,
+      "loss": 0.4537,
+      "step": 7395
+    },
+    {
+      "epoch": 1.7814155031295136,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.352509482294692e-05,
+      "loss": 0.4597,
+      "step": 7400
+    },
+    {
+      "epoch": 1.78261916225325,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.344303743741201e-05,
+      "loss": 0.4747,
+      "step": 7405
+    },
+    {
+      "epoch": 1.783822821376986,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.336101123698506e-05,
+      "loss": 0.4555,
+      "step": 7410
+    },
+    {
+      "epoch": 1.785026480500722,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.327901635272294e-05,
+      "loss": 0.4594,
+      "step": 7415
+    },
+    {
+      "epoch": 1.7862301396244584,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.319705291563237e-05,
+      "loss": 0.474,
+      "step": 7420
+    },
+    {
+      "epoch": 1.7874337987481945,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.311512105667e-05,
+      "loss": 0.4418,
+      "step": 7425
+    },
+    {
+      "epoch": 1.7886374578719306,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.303322090674185e-05,
+      "loss": 0.4606,
+      "step": 7430
+    },
+    {
+      "epoch": 1.789841116995667,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.295135259670342e-05,
+      "loss": 0.4831,
+      "step": 7435
+    },
+    {
+      "epoch": 1.7910447761194028,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.286951625735929e-05,
+      "loss": 0.4549,
+      "step": 7440
+    },
+    {
+      "epoch": 1.7922484352431391,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.278771201946291e-05,
+      "loss": 0.4608,
+      "step": 7445
+    },
+    {
+      "epoch": 1.7934520943668752,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.270594001371654e-05,
+      "loss": 0.4771,
+      "step": 7450
+    },
+    {
+      "epoch": 1.7946557534906113,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.262420037077078e-05,
+      "loss": 0.4548,
+      "step": 7455
+    },
+    {
+      "epoch": 1.7958594126143477,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.254249322122474e-05,
+      "loss": 0.4413,
+      "step": 7460
+    },
+    {
+      "epoch": 1.7970630717380838,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.246081869562543e-05,
+      "loss": 0.4634,
+      "step": 7465
+    },
+    {
+      "epoch": 1.7982667308618199,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.237917692446784e-05,
+      "loss": 0.4422,
+      "step": 7470
+    },
+    {
+      "epoch": 1.7994703899855562,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.229756803819458e-05,
+      "loss": 0.478,
+      "step": 7475
+    },
+    {
+      "epoch": 1.8006740491092923,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.221599216719573e-05,
+      "loss": 0.4449,
+      "step": 7480
+    },
+    {
+      "epoch": 1.8018777082330284,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.213444944180858e-05,
+      "loss": 0.4792,
+      "step": 7485
+    },
+    {
+      "epoch": 1.8030813673567647,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.205293999231758e-05,
+      "loss": 0.4494,
+      "step": 7490
+    },
+    {
+      "epoch": 1.8042850264805006,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.197146394895386e-05,
+      "loss": 0.4398,
+      "step": 7495
+    },
+    {
+      "epoch": 1.805488685604237,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.189002144189533e-05,
+      "loss": 0.4357,
+      "step": 7500
+    },
+    {
+      "epoch": 1.805488685604237,
+      "eval_loss": 0.39568030834198,
+      "eval_runtime": 2.3503,
+      "eval_samples_per_second": 85.097,
+      "eval_steps_per_second": 85.097,
+      "step": 7500
+    },
+    {
+      "epoch": 1.806692344727973,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.180861260126613e-05,
+      "loss": 0.49,
+      "step": 7505
+    },
+    {
+      "epoch": 1.8078960038517091,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.172723755713681e-05,
+      "loss": 0.4472,
+      "step": 7510
+    },
+    {
+      "epoch": 1.8090996629754454,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.164589643952379e-05,
+      "loss": 0.4608,
+      "step": 7515
+    },
+    {
+      "epoch": 1.8103033220991815,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.156458937838935e-05,
+      "loss": 0.4465,
+      "step": 7520
+    },
+    {
+      "epoch": 1.8115069812229176,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.148331650364132e-05,
+      "loss": 0.4368,
+      "step": 7525
+    },
+    {
+      "epoch": 1.812710640346654,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.14020779451329e-05,
+      "loss": 0.4707,
+      "step": 7530
+    },
+    {
+      "epoch": 1.8139142994703898,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.132087383266247e-05,
+      "loss": 0.496,
+      "step": 7535
+    },
+    {
+      "epoch": 1.8151179585941262,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.123970429597343e-05,
+      "loss": 0.4682,
+      "step": 7540
+    },
+    {
+      "epoch": 1.8163216177178623,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.115856946475389e-05,
+      "loss": 0.4773,
+      "step": 7545
+    },
+    {
+      "epoch": 1.8175252768415984,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.107746946863647e-05,
+      "loss": 0.4476,
+      "step": 7550
+    },
+    {
+      "epoch": 1.8187289359653347,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.099640443719819e-05,
+      "loss": 0.4957,
+      "step": 7555
+    },
+    {
+      "epoch": 1.8199325950890708,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.091537449996017e-05,
+      "loss": 0.4474,
+      "step": 7560
+    },
+    {
+      "epoch": 1.821136254212807,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.083437978638755e-05,
+      "loss": 0.4621,
+      "step": 7565
+    },
+    {
+      "epoch": 1.8223399133365432,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.075342042588907e-05,
+      "loss": 0.4471,
+      "step": 7570
+    },
+    {
+      "epoch": 1.823543572460279,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.067249654781703e-05,
+      "loss": 0.4626,
+      "step": 7575
+    },
+    {
+      "epoch": 1.8247472315840154,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.059160828146701e-05,
+      "loss": 0.458,
+      "step": 7580
+    },
+    {
+      "epoch": 1.8259508907077515,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.051075575607781e-05,
+      "loss": 0.4532,
+      "step": 7585
+    },
+    {
+      "epoch": 1.8271545498314876,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.042993910083096e-05,
+      "loss": 0.4586,
+      "step": 7590
+    },
+    {
+      "epoch": 1.828358208955224,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.03491584448508e-05,
+      "loss": 0.4783,
+      "step": 7595
+    },
+    {
+      "epoch": 1.82956186807896,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.02684139172041e-05,
+      "loss": 0.4357,
+      "step": 7600
+    },
+    {
+      "epoch": 1.8307655272026961,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.018770564689988e-05,
+      "loss": 0.4525,
+      "step": 7605
+    },
+    {
+      "epoch": 1.8319691863264325,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.010703376288933e-05,
+      "loss": 0.47,
+      "step": 7610
+    },
+    {
+      "epoch": 1.8331728454501686,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.00263983940654e-05,
+      "loss": 0.4505,
+      "step": 7615
+    },
+    {
+      "epoch": 1.8343765045739047,
+      "grad_norm": 1.2578125,
+      "learning_rate": 8.994579966926274e-05,
+      "loss": 0.4681,
+      "step": 7620
+    },
+    {
+      "epoch": 1.835580163697641,
+      "grad_norm": 1.21875,
+      "learning_rate": 8.986523771725747e-05,
+      "loss": 0.4418,
+      "step": 7625
+    },
+    {
+      "epoch": 1.8367838228213769,
+      "grad_norm": 1.2421875,
+      "learning_rate": 8.978471266676691e-05,
+      "loss": 0.4744,
+      "step": 7630
+    },
+    {
+      "epoch": 1.8379874819451132,
+      "grad_norm": 1.28125,
+      "learning_rate": 8.970422464644951e-05,
+      "loss": 0.4439,
+      "step": 7635
+    },
+    {
+      "epoch": 1.8391911410688493,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.962377378490439e-05,
+      "loss": 0.4812,
+      "step": 7640
+    },
+    {
+      "epoch": 1.8403948001925854,
+      "grad_norm": 1.1171875,
+      "learning_rate": 8.954336021067146e-05,
+      "loss": 0.4641,
+      "step": 7645
+    },
+    {
+      "epoch": 1.8415984593163217,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.946298405223105e-05,
+      "loss": 0.4797,
+      "step": 7650
+    },
+    {
+      "epoch": 1.8428021184400578,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.938264543800356e-05,
+      "loss": 0.4551,
+      "step": 7655
+    },
+    {
+      "epoch": 1.844005777563794,
+      "grad_norm": 1.203125,
+      "learning_rate": 8.930234449634958e-05,
+      "loss": 0.4606,
+      "step": 7660
+    },
+    {
+      "epoch": 1.8452094366875302,
+      "grad_norm": 1.1953125,
+      "learning_rate": 8.922208135556936e-05,
+      "loss": 0.4816,
+      "step": 7665
+    },
+    {
+      "epoch": 1.8464130958112661,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.914185614390285e-05,
+      "loss": 0.4715,
+      "step": 7670
+    },
+    {
+      "epoch": 1.8476167549350024,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.90616689895294e-05,
+      "loss": 0.4704,
+      "step": 7675
+    },
+    {
+      "epoch": 1.8488204140587385,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.898152002056751e-05,
+      "loss": 0.4278,
+      "step": 7680
+    },
+    {
+      "epoch": 1.8500240731824746,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.89014093650747e-05,
+      "loss": 0.4681,
+      "step": 7685
+    },
+    {
+      "epoch": 1.851227732306211,
+      "grad_norm": 1.25,
+      "learning_rate": 8.88213371510472e-05,
+      "loss": 0.4809,
+      "step": 7690
+    },
+    {
+      "epoch": 1.852431391429947,
+      "grad_norm": 1.1015625,
+      "learning_rate": 8.874130350641992e-05,
+      "loss": 0.4337,
+      "step": 7695
+    },
+    {
+      "epoch": 1.8536350505536832,
+      "grad_norm": 1.234375,
+      "learning_rate": 8.866130855906615e-05,
+      "loss": 0.467,
+      "step": 7700
+    },
+    {
+      "epoch": 1.8548387096774195,
+      "grad_norm": 1.203125,
+      "learning_rate": 8.858135243679725e-05,
+      "loss": 0.4486,
+      "step": 7705
+    },
+    {
+      "epoch": 1.8560423688011554,
+      "grad_norm": 1.0859375,
+      "learning_rate": 8.850143526736265e-05,
+      "loss": 0.4463,
+      "step": 7710
+    },
+    {
+      "epoch": 1.8572460279248917,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.842155717844943e-05,
+      "loss": 0.4714,
+      "step": 7715
+    },
+    {
+      "epoch": 1.8584496870486278,
+      "grad_norm": 1.2109375,
+      "learning_rate": 8.834171829768236e-05,
+      "loss": 0.4678,
+      "step": 7720
+    },
+    {
+      "epoch": 1.8596533461723639,
+      "grad_norm": 1.15625,
+      "learning_rate": 8.826191875262353e-05,
+      "loss": 0.4437,
+      "step": 7725
+    },
+    {
+      "epoch": 1.8608570052961002,
+      "grad_norm": 1.2734375,
+      "learning_rate": 8.818215867077208e-05,
+      "loss": 0.4542,
+      "step": 7730
+    },
+    {
+      "epoch": 1.8620606644198363,
+      "grad_norm": 1.203125,
+      "learning_rate": 8.810243817956424e-05,
+      "loss": 0.4759,
+      "step": 7735
+    },
+    {
+      "epoch": 1.8632643235435724,
+      "grad_norm": 1.1171875,
+      "learning_rate": 8.802275740637285e-05,
+      "loss": 0.4581,
+      "step": 7740
+    },
+    {
+      "epoch": 1.8644679826673087,
+      "grad_norm": 1.0859375,
+      "learning_rate": 8.794311647850739e-05,
+      "loss": 0.4238,
+      "step": 7745
+    },
+    {
+      "epoch": 1.8656716417910446,
+      "grad_norm": 1.21875,
+      "learning_rate": 8.78635155232137e-05,
+      "loss": 0.4728,
+      "step": 7750
+    },
+    {
+      "epoch": 1.866875300914781,
+      "grad_norm": 1.21875,
+      "learning_rate": 8.778395466767365e-05,
+      "loss": 0.4451,
+      "step": 7755
+    },
+    {
+      "epoch": 1.8680789600385173,
+      "grad_norm": 1.359375,
+      "learning_rate": 8.770443403900507e-05,
+      "loss": 0.457,
+      "step": 7760
+    },
+    {
+      "epoch": 1.8692826191622531,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.762495376426158e-05,
+      "loss": 0.4512,
+      "step": 7765
+    },
+    {
+      "epoch": 1.8704862782859895,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.754551397043226e-05,
+      "loss": 0.4497,
+      "step": 7770
+    },
+    {
+      "epoch": 1.8716899374097256,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.746611478444157e-05,
+      "loss": 0.4503,
+      "step": 7775
+    },
+    {
+      "epoch": 1.8728935965334617,
+      "grad_norm": 1.2890625,
+      "learning_rate": 8.738675633314902e-05,
+      "loss": 0.442,
+      "step": 7780
+    },
+    {
+      "epoch": 1.874097255657198,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.730743874334909e-05,
+      "loss": 0.4566,
+      "step": 7785
+    },
+    {
+      "epoch": 1.875300914780934,
+      "grad_norm": 1.21875,
+      "learning_rate": 8.722816214177093e-05,
+      "loss": 0.4417,
+      "step": 7790
+    },
+    {
+      "epoch": 1.8765045739046702,
+      "grad_norm": 1.125,
+      "learning_rate": 8.714892665507825e-05,
+      "loss": 0.4951,
+      "step": 7795
+    },
+    {
+      "epoch": 1.8777082330284065,
+      "grad_norm": 1.265625,
+      "learning_rate": 8.706973240986906e-05,
+      "loss": 0.4616,
+      "step": 7800
+    },
+    {
+      "epoch": 1.8789118921521424,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.699057953267541e-05,
+      "loss": 0.4855,
+      "step": 7805
+    },
+    {
+      "epoch": 1.8801155512758787,
+      "grad_norm": 1.1171875,
+      "learning_rate": 8.691146814996337e-05,
+      "loss": 0.4607,
+      "step": 7810
+    },
+    {
+      "epoch": 1.8813192103996148,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.683239838813261e-05,
+      "loss": 0.4378,
+      "step": 7815
+    },
+    {
+      "epoch": 1.882522869523351,
+      "grad_norm": 1.2265625,
+      "learning_rate": 8.675337037351635e-05,
+      "loss": 0.4872,
+      "step": 7820
+    },
+    {
+      "epoch": 1.8837265286470872,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.667438423238108e-05,
+      "loss": 0.4662,
+      "step": 7825
+    },
+    {
+      "epoch": 1.8849301877708233,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.659544009092642e-05,
+      "loss": 0.4819,
+      "step": 7830
+    },
+    {
+      "epoch": 1.8861338468945594,
+      "grad_norm": 1.265625,
+      "learning_rate": 8.651653807528489e-05,
+      "loss": 0.4656,
+      "step": 7835
+    },
+    {
+      "epoch": 1.8873375060182958,
+      "grad_norm": 1.15625,
+      "learning_rate": 8.643767831152166e-05,
+      "loss": 0.4723,
+      "step": 7840
+    },
+    {
+      "epoch": 1.8885411651420316,
+      "grad_norm": 1.1953125,
+      "learning_rate": 8.635886092563444e-05,
+      "loss": 0.4626,
+      "step": 7845
+    },
+    {
+      "epoch": 1.889744824265768,
+      "grad_norm": 1.25,
+      "learning_rate": 8.628008604355316e-05,
+      "loss": 0.4311,
+      "step": 7850
+    },
+    {
+      "epoch": 1.890948483389504,
+      "grad_norm": 1.1875,
+      "learning_rate": 8.620135379113992e-05,
+      "loss": 0.4525,
+      "step": 7855
+    },
+    {
+      "epoch": 1.8921521425132402,
+      "grad_norm": 1.203125,
+      "learning_rate": 8.61226642941887e-05,
+      "loss": 0.4317,
+      "step": 7860
+    },
+    {
+      "epoch": 1.8933558016369765,
+      "grad_norm": 1.125,
+      "learning_rate": 8.60440176784251e-05,
+      "loss": 0.4601,
+      "step": 7865
+    },
+    {
+      "epoch": 1.8945594607607126,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.59654140695063e-05,
+      "loss": 0.4347,
+      "step": 7870
+    },
+    {
+      "epoch": 1.8957631198844487,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.588685359302063e-05,
+      "loss": 0.4727,
+      "step": 7875
+    },
+    {
+      "epoch": 1.896966779008185,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.580833637448769e-05,
+      "loss": 0.4099,
+      "step": 7880
+    },
+    {
+      "epoch": 1.8981704381319209,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.572986253935781e-05,
+      "loss": 0.4509,
+      "step": 7885
+    },
+    {
+      "epoch": 1.8993740972556572,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.565143221301208e-05,
+      "loss": 0.4716,
+      "step": 7890
+    },
+    {
+      "epoch": 1.9005777563793933,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.557304552076206e-05,
+      "loss": 0.4672,
+      "step": 7895
+    },
+    {
+      "epoch": 1.9017814155031294,
+      "grad_norm": 1.0546875,
+      "learning_rate": 8.549470258784957e-05,
+      "loss": 0.4225,
+      "step": 7900
+    },
+    {
+      "epoch": 1.9029850746268657,
+      "grad_norm": 1.2421875,
+      "learning_rate": 8.541640353944654e-05,
+      "loss": 0.4699,
+      "step": 7905
+    },
+    {
+      "epoch": 1.9041887337506018,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.53381485006548e-05,
+      "loss": 0.4432,
+      "step": 7910
+    },
+    {
+      "epoch": 1.905392392874338,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.525993759650581e-05,
+      "loss": 0.4421,
+      "step": 7915
+    },
+    {
+      "epoch": 1.9065960519980742,
+      "grad_norm": 1.1875,
+      "learning_rate": 8.518177095196058e-05,
+      "loss": 0.466,
+      "step": 7920
+    },
+    {
+      "epoch": 1.9077997111218103,
+      "grad_norm": 1.2578125,
+      "learning_rate": 8.510364869190934e-05,
+      "loss": 0.4965,
+      "step": 7925
+    },
+    {
+      "epoch": 1.9090033702455464,
+      "grad_norm": 1.1953125,
+      "learning_rate": 8.502557094117142e-05,
+      "loss": 0.4582,
+      "step": 7930
+    },
+    {
+      "epoch": 1.9102070293692828,
+      "grad_norm": 1.2578125,
+      "learning_rate": 8.494753782449512e-05,
+      "loss": 0.4507,
+      "step": 7935
+    },
+    {
+      "epoch": 1.9114106884930187,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.486954946655726e-05,
+      "loss": 0.4378,
+      "step": 7940
+    },
+    {
+      "epoch": 1.912614347616755,
+      "grad_norm": 1.2109375,
+      "learning_rate": 8.479160599196334e-05,
+      "loss": 0.4681,
+      "step": 7945
+    },
+    {
+      "epoch": 1.913818006740491,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.471370752524695e-05,
+      "loss": 0.4373,
+      "step": 7950
+    },
+    {
+      "epoch": 1.9150216658642272,
+      "grad_norm": 1.25,
+      "learning_rate": 8.463585419086999e-05,
+      "loss": 0.4622,
+      "step": 7955
+    },
+    {
+      "epoch": 1.9162253249879635,
+      "grad_norm": 1.1015625,
+      "learning_rate": 8.455804611322205e-05,
+      "loss": 0.4669,
+      "step": 7960
+    },
+    {
+      "epoch": 1.9174289841116996,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.448028341662049e-05,
+      "loss": 0.4499,
+      "step": 7965
+    },
+    {
+      "epoch": 1.9186326432354357,
+      "grad_norm": 0.98046875,
+      "learning_rate": 8.440256622531019e-05,
+      "loss": 0.4136,
+      "step": 7970
+    },
+    {
+      "epoch": 1.919836302359172,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.432489466346327e-05,
+      "loss": 0.4402,
+      "step": 7975
+    },
+    {
+      "epoch": 1.921039961482908,
+      "grad_norm": 1.296875,
+      "learning_rate": 8.4247268855179e-05,
+      "loss": 0.4496,
+      "step": 7980
+    },
+    {
+      "epoch": 1.9222436206066442,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.416968892448354e-05,
+      "loss": 0.4861,
+      "step": 7985
+    },
+    {
+      "epoch": 1.9234472797303803,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.409215499532969e-05,
+      "loss": 0.4525,
+      "step": 7990
+    },
+    {
+      "epoch": 1.9246509388541164,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.401466719159685e-05,
+      "loss": 0.4661,
+      "step": 7995
+    },
+    {
+      "epoch": 1.9258545979778527,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.393722563709057e-05,
+      "loss": 0.436,
+      "step": 8000
+    },
+    {
+      "epoch": 1.9258545979778527,
+      "eval_loss": 0.38646742701530457,
+      "eval_runtime": 2.3414,
+      "eval_samples_per_second": 85.418,
+      "eval_steps_per_second": 85.418,
+      "step": 8000
+    },
+    {
+      "epoch": 1.9270582571015888,
+      "grad_norm": 1.328125,
+      "learning_rate": 8.385983045554271e-05,
+      "loss": 0.4722,
+      "step": 8005
+    },
+    {
+      "epoch": 1.928261916225325,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.378248177061085e-05,
+      "loss": 0.4317,
+      "step": 8010
+    },
+    {
+      "epoch": 1.9294655753490613,
+      "grad_norm": 1.2109375,
+      "learning_rate": 8.370517970587842e-05,
+      "loss": 0.4586,
+      "step": 8015
+    },
+    {
+      "epoch": 1.9306692344727971,
+      "grad_norm": 1.265625,
+      "learning_rate": 8.362792438485427e-05,
+      "loss": 0.4688,
+      "step": 8020
+    },
+    {
+      "epoch": 1.9318728935965335,
+      "grad_norm": 1.125,
+      "learning_rate": 8.355071593097257e-05,
+      "loss": 0.4873,
+      "step": 8025
+    },
+    {
+      "epoch": 1.9330765527202696,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.347355446759267e-05,
+      "loss": 0.489,
+      "step": 8030
+    },
+    {
+      "epoch": 1.9342802118440057,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.339644011799883e-05,
+      "loss": 0.4165,
+      "step": 8035
+    },
+    {
+      "epoch": 1.935483870967742,
+      "grad_norm": 1.1875,
+      "learning_rate": 8.331937300539996e-05,
+      "loss": 0.4854,
+      "step": 8040
+    },
+    {
+      "epoch": 1.936687530091478,
+      "grad_norm": 1.25,
+      "learning_rate": 8.324235325292962e-05,
+      "loss": 0.4409,
+      "step": 8045
+    },
+    {
+      "epoch": 1.9378911892152142,
+      "grad_norm": 1.2421875,
+      "learning_rate": 8.316538098364553e-05,
+      "loss": 0.4517,
+      "step": 8050
+    },
+    {
+      "epoch": 1.9390948483389505,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.308845632052976e-05,
+      "loss": 0.4462,
+      "step": 8055
+    },
+    {
+      "epoch": 1.9402985074626866,
+      "grad_norm": 1.1015625,
+      "learning_rate": 8.301157938648811e-05,
+      "loss": 0.4558,
+      "step": 8060
+    },
+    {
+      "epoch": 1.9415021665864227,
+      "grad_norm": 1.125,
+      "learning_rate": 8.293475030435027e-05,
+      "loss": 0.4457,
+      "step": 8065
+    },
+    {
+      "epoch": 1.942705825710159,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.28579691968694e-05,
+      "loss": 0.429,
+      "step": 8070
+    },
+    {
+      "epoch": 1.943909484833895,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.278123618672199e-05,
+      "loss": 0.4432,
+      "step": 8075
+    },
+    {
+      "epoch": 1.9451131439576312,
+      "grad_norm": 1.2578125,
+      "learning_rate": 8.270455139650777e-05,
+      "loss": 0.4277,
+      "step": 8080
+    },
+    {
+      "epoch": 1.9463168030813673,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.262791494874934e-05,
+      "loss": 0.4068,
+      "step": 8085
+    },
+    {
+      "epoch": 1.9475204622051034,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.255132696589212e-05,
+      "loss": 0.4259,
+      "step": 8090
+    },
+    {
+      "epoch": 1.9487241213288398,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.247478757030408e-05,
+      "loss": 0.4313,
+      "step": 8095
+    },
+    {
+      "epoch": 1.9499277804525759,
+      "grad_norm": 1.375,
+      "learning_rate": 8.239829688427552e-05,
+      "loss": 0.4646,
+      "step": 8100
+    },
+    {
+      "epoch": 1.951131439576312,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.232185503001901e-05,
+      "loss": 0.4925,
+      "step": 8105
+    },
+    {
+      "epoch": 1.9523350987000483,
+      "grad_norm": 1.15625,
+      "learning_rate": 8.224546212966896e-05,
+      "loss": 0.4387,
+      "step": 8110
+    },
+    {
+      "epoch": 1.9535387578237842,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.216911830528171e-05,
+      "loss": 0.4515,
+      "step": 8115
+    },
+    {
+      "epoch": 1.9547424169475205,
+      "grad_norm": 1.0859375,
+      "learning_rate": 8.209282367883507e-05,
+      "loss": 0.431,
+      "step": 8120
+    },
+    {
+      "epoch": 1.9559460760712566,
+      "grad_norm": 1.234375,
+      "learning_rate": 8.20165783722283e-05,
+      "loss": 0.4543,
+      "step": 8125
+    },
+    {
+      "epoch": 1.9571497351949927,
+      "grad_norm": 1.34375,
+      "learning_rate": 8.19403825072819e-05,
+      "loss": 0.4505,
+      "step": 8130
+    },
+    {
+      "epoch": 1.958353394318729,
+      "grad_norm": 1.1953125,
+      "learning_rate": 8.186423620573726e-05,
+      "loss": 0.4771,
+      "step": 8135
+    },
+    {
+      "epoch": 1.9595570534424651,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.17881395892567e-05,
+      "loss": 0.4488,
+      "step": 8140
+    },
+    {
+      "epoch": 1.9607607125662012,
+      "grad_norm": 1.203125,
+      "learning_rate": 8.17120927794231e-05,
+      "loss": 0.4494,
+      "step": 8145
+    },
+    {
+      "epoch": 1.9619643716899375,
+      "grad_norm": 1.2265625,
+      "learning_rate": 8.163609589773973e-05,
+      "loss": 0.4186,
+      "step": 8150
+    },
+    {
+      "epoch": 1.9631680308136734,
+      "grad_norm": 1.1953125,
+      "learning_rate": 8.15601490656302e-05,
+      "loss": 0.4626,
+      "step": 8155
+    },
+    {
+      "epoch": 1.9643716899374097,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.148425240443799e-05,
+      "loss": 0.4411,
+      "step": 8160
+    },
+    {
+      "epoch": 1.9655753490611458,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.140840603542657e-05,
+      "loss": 0.4392,
+      "step": 8165
+    },
+    {
+      "epoch": 1.966779008184882,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.133261007977897e-05,
+      "loss": 0.443,
+      "step": 8170
+    },
+    {
+      "epoch": 1.9679826673086183,
+      "grad_norm": 1.1171875,
+      "learning_rate": 8.125686465859771e-05,
+      "loss": 0.4552,
+      "step": 8175
+    },
+    {
+      "epoch": 1.9691863264323544,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.118116989290456e-05,
+      "loss": 0.4753,
+      "step": 8180
+    },
+    {
+      "epoch": 1.9703899855560905,
+      "grad_norm": 1.1875,
+      "learning_rate": 8.11055259036403e-05,
+      "loss": 0.4581,
+      "step": 8185
+    },
+    {
+      "epoch": 1.9715936446798268,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.102993281166469e-05,
+      "loss": 0.4529,
+      "step": 8190
+    },
+    {
+      "epoch": 1.9727973038035629,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.095439073775611e-05,
+      "loss": 0.4403,
+      "step": 8195
+    },
+    {
+      "epoch": 1.974000962927299,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.087889980261139e-05,
+      "loss": 0.4428,
+      "step": 8200
+    },
+    {
+      "epoch": 1.9752046220510353,
+      "grad_norm": 1.1875,
+      "learning_rate": 8.080346012684573e-05,
+      "loss": 0.4792,
+      "step": 8205
+    },
+    {
+      "epoch": 1.9764082811747712,
+      "grad_norm": 1.21875,
+      "learning_rate": 8.072807183099237e-05,
+      "loss": 0.4788,
+      "step": 8210
+    },
+    {
+      "epoch": 1.9776119402985075,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.065273503550247e-05,
+      "loss": 0.4231,
+      "step": 8215
+    },
+    {
+      "epoch": 1.9788155994222436,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.057744986074497e-05,
+      "loss": 0.4536,
+      "step": 8220
+    },
+    {
+      "epoch": 1.9800192585459797,
+      "grad_norm": 1.0546875,
+      "learning_rate": 8.050221642700624e-05,
+      "loss": 0.4245,
+      "step": 8225
+    },
+    {
+      "epoch": 1.981222917669716,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.042703485449003e-05,
+      "loss": 0.4479,
+      "step": 8230
+    },
+    {
+      "epoch": 1.9824265767934521,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.035190526331718e-05,
+      "loss": 0.4422,
+      "step": 8235
+    },
+    {
+      "epoch": 1.9836302359171882,
+      "grad_norm": 1.0546875,
+      "learning_rate": 8.027682777352556e-05,
+      "loss": 0.4483,
+      "step": 8240
+    },
+    {
+      "epoch": 1.9848338950409246,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.020180250506978e-05,
+      "loss": 0.4408,
+      "step": 8245
+    },
+    {
+      "epoch": 1.9860375541646604,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.01268295778209e-05,
+      "loss": 0.4745,
+      "step": 8250
+    },
+    {
+      "epoch": 1.9872412132883968,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.005190911156654e-05,
+      "loss": 0.4414,
+      "step": 8255
+    },
+    {
+      "epoch": 1.9884448724121329,
+      "grad_norm": 1.1328125,
+      "learning_rate": 7.997704122601031e-05,
+      "loss": 0.4462,
+      "step": 8260
+    },
+    {
+      "epoch": 1.989648531535869,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.990222604077194e-05,
+      "loss": 0.4435,
+      "step": 8265
+    },
+    {
+      "epoch": 1.9908521906596053,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.982746367538694e-05,
+      "loss": 0.465,
+      "step": 8270
+    },
+    {
+      "epoch": 1.9920558497833414,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.975275424930638e-05,
+      "loss": 0.4433,
+      "step": 8275
+    },
+    {
+      "epoch": 1.9932595089070775,
+      "grad_norm": 1.1953125,
+      "learning_rate": 7.967809788189675e-05,
+      "loss": 0.4534,
+      "step": 8280
+    },
+    {
+      "epoch": 1.9944631680308138,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.96034946924398e-05,
+      "loss": 0.4665,
+      "step": 8285
+    },
+    {
+      "epoch": 1.9956668271545497,
+      "grad_norm": 1.1953125,
+      "learning_rate": 7.952894480013234e-05,
+      "loss": 0.456,
+      "step": 8290
+    },
+    {
+      "epoch": 1.996870486278286,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.945444832408592e-05,
+      "loss": 0.4459,
+      "step": 8295
+    },
+    {
+      "epoch": 1.998074145402022,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.938000538332684e-05,
+      "loss": 0.4198,
+      "step": 8300
+    },
+    {
+      "epoch": 1.9992778045257582,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.930561609679588e-05,
+      "loss": 0.4287,
+      "step": 8305
+    },
+    {
+      "epoch": 1.9995185363505055,
+      "eval_loss": 0.3824025094509125,
+      "eval_runtime": 2.3407,
+      "eval_samples_per_second": 85.446,
+      "eval_steps_per_second": 85.446,
+      "step": 8306
+    },
+    {
+      "epoch": 2.0004814636494945,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.923128058334797e-05,
+      "loss": 0.4334,
+      "step": 8310
+    },
+    {
+      "epoch": 2.0016851227732304,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.915699896175228e-05,
+      "loss": 0.3898,
+      "step": 8315
+    },
+    {
+      "epoch": 2.0028887818969667,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.908277135069175e-05,
+      "loss": 0.4385,
+      "step": 8320
+    },
+    {
+      "epoch": 2.004092441020703,
+      "grad_norm": 1.125,
+      "learning_rate": 7.900859786876308e-05,
+      "loss": 0.4072,
+      "step": 8325
+    },
+    {
+      "epoch": 2.005296100144439,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.893447863447651e-05,
+      "loss": 0.3937,
+      "step": 8330
+    },
+    {
+      "epoch": 2.0064997592681753,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.886041376625557e-05,
+      "loss": 0.4192,
+      "step": 8335
+    },
+    {
+      "epoch": 2.0077034183919116,
+      "grad_norm": 1.1796875,
+      "learning_rate": 7.87864033824369e-05,
+      "loss": 0.4446,
+      "step": 8340
+    },
+    {
+      "epoch": 2.0089070775156475,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.871244760127011e-05,
+      "loss": 0.4018,
+      "step": 8345
+    },
+    {
+      "epoch": 2.0101107366393838,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.863854654091763e-05,
+      "loss": 0.4013,
+      "step": 8350
+    },
+    {
+      "epoch": 2.01131439576312,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.85647003194544e-05,
+      "loss": 0.4188,
+      "step": 8355
+    },
+    {
+      "epoch": 2.012518054886856,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.84909090548677e-05,
+      "loss": 0.3993,
+      "step": 8360
+    },
+    {
+      "epoch": 2.0137217140105923,
+      "grad_norm": 1.1015625,
+      "learning_rate": 7.841717286505714e-05,
+      "loss": 0.4066,
+      "step": 8365
+    },
+    {
+      "epoch": 2.014925373134328,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.834349186783416e-05,
+      "loss": 0.3951,
+      "step": 8370
+    },
+    {
+      "epoch": 2.0161290322580645,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.826986618092216e-05,
+      "loss": 0.4215,
+      "step": 8375
+    },
+    {
+      "epoch": 2.017332691381801,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.819629592195612e-05,
+      "loss": 0.3705,
+      "step": 8380
+    },
+    {
+      "epoch": 2.0185363505055367,
+      "grad_norm": 1.2421875,
+      "learning_rate": 7.812278120848243e-05,
+      "loss": 0.424,
+      "step": 8385
+    },
+    {
+      "epoch": 2.019740009629273,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.804932215795877e-05,
+      "loss": 0.3994,
+      "step": 8390
+    },
+    {
+      "epoch": 2.0209436687530093,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.797591888775384e-05,
+      "loss": 0.3877,
+      "step": 8395
+    },
+    {
+      "epoch": 2.0221473278767452,
+      "grad_norm": 1.1796875,
+      "learning_rate": 7.790257151514726e-05,
+      "loss": 0.4092,
+      "step": 8400
+    },
+    {
+      "epoch": 2.0233509870004815,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.782928015732937e-05,
+      "loss": 0.4255,
+      "step": 8405
+    },
+    {
+      "epoch": 2.0245546461242174,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.775604493140092e-05,
+      "loss": 0.4021,
+      "step": 8410
+    },
+    {
+      "epoch": 2.0257583052479537,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.768286595437306e-05,
+      "loss": 0.3863,
+      "step": 8415
+    },
+    {
+      "epoch": 2.02696196437169,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.760974334316702e-05,
+      "loss": 0.3939,
+      "step": 8420
+    },
+    {
+      "epoch": 2.028165623495426,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.753667721461396e-05,
+      "loss": 0.4396,
+      "step": 8425
+    },
+    {
+      "epoch": 2.0293692826191623,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.74636676854549e-05,
+      "loss": 0.3985,
+      "step": 8430
+    },
+    {
+      "epoch": 2.0305729417428986,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.739071487234031e-05,
+      "loss": 0.4124,
+      "step": 8435
+    },
+    {
+      "epoch": 2.0317766008666345,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.731781889183004e-05,
+      "loss": 0.4312,
+      "step": 8440
+    },
+    {
+      "epoch": 2.032980259990371,
+      "grad_norm": 1.1015625,
+      "learning_rate": 7.724497986039326e-05,
+      "loss": 0.4224,
+      "step": 8445
+    },
+    {
+      "epoch": 2.0341839191141067,
+      "grad_norm": 1.3203125,
+      "learning_rate": 7.7172197894408e-05,
+      "loss": 0.4149,
+      "step": 8450
+    },
+    {
+      "epoch": 2.035387578237843,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.709947311016122e-05,
+      "loss": 0.4007,
+      "step": 8455
+    },
+    {
+      "epoch": 2.0365912373615793,
+      "grad_norm": 1.1953125,
+      "learning_rate": 7.702680562384848e-05,
+      "loss": 0.4025,
+      "step": 8460
+    },
+    {
+      "epoch": 2.037794896485315,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.695419555157379e-05,
+      "loss": 0.4085,
+      "step": 8465
+    },
+    {
+      "epoch": 2.0389985556090515,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.68816430093494e-05,
+      "loss": 0.413,
+      "step": 8470
+    },
+    {
+      "epoch": 2.040202214732788,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.680914811309574e-05,
+      "loss": 0.3988,
+      "step": 8475
+    },
+    {
+      "epoch": 2.0414058738565237,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.673671097864102e-05,
+      "loss": 0.3933,
+      "step": 8480
+    },
+    {
+      "epoch": 2.04260953298026,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.666433172172121e-05,
+      "loss": 0.4104,
+      "step": 8485
+    },
+    {
+      "epoch": 2.043813192103996,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.659201045797986e-05,
+      "loss": 0.4226,
+      "step": 8490
+    },
+    {
+      "epoch": 2.0450168512277322,
+      "grad_norm": 1.171875,
+      "learning_rate": 7.651974730296779e-05,
+      "loss": 0.399,
+      "step": 8495
+    },
+    {
+      "epoch": 2.0462205103514686,
+      "grad_norm": 1.203125,
+      "learning_rate": 7.644754237214296e-05,
+      "loss": 0.417,
+      "step": 8500
+    },
+    {
+      "epoch": 2.0462205103514686,
+      "eval_loss": 0.38362109661102295,
+      "eval_runtime": 2.3419,
+      "eval_samples_per_second": 85.401,
+      "eval_steps_per_second": 85.401,
+      "step": 8500
+    },
+    {
+      "epoch": 2.0474241694752044,
+      "grad_norm": 1.1328125,
+      "learning_rate": 7.637539578087033e-05,
+      "loss": 0.4,
+      "step": 8505
+    },
+    {
+      "epoch": 2.0486278285989408,
+      "grad_norm": 1.2109375,
+      "learning_rate": 7.63033076444217e-05,
+      "loss": 0.4213,
+      "step": 8510
+    },
+    {
+      "epoch": 2.049831487722677,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.623127807797544e-05,
+      "loss": 0.4076,
+      "step": 8515
+    },
+    {
+      "epoch": 2.051035146846413,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.615930719661628e-05,
+      "loss": 0.4243,
+      "step": 8520
+    },
+    {
+      "epoch": 2.0522388059701493,
+      "grad_norm": 1.1875,
+      "learning_rate": 7.608739511533531e-05,
+      "loss": 0.3875,
+      "step": 8525
+    },
+    {
+      "epoch": 2.0534424650938856,
+      "grad_norm": 0.98046875,
+      "learning_rate": 7.601554194902953e-05,
+      "loss": 0.3694,
+      "step": 8530
+    },
+    {
+      "epoch": 2.0546461242176215,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.594374781250192e-05,
+      "loss": 0.4145,
+      "step": 8535
+    },
+    {
+      "epoch": 2.055849783341358,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.58720128204611e-05,
+      "loss": 0.397,
+      "step": 8540
+    },
+    {
+      "epoch": 2.0570534424650937,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.580033708752123e-05,
+      "loss": 0.3991,
+      "step": 8545
+    },
+    {
+      "epoch": 2.05825710158883,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.572872072820169e-05,
+      "loss": 0.3759,
+      "step": 8550
+    },
+    {
+      "epoch": 2.0594607607125663,
+      "grad_norm": 1.234375,
+      "learning_rate": 7.56571638569271e-05,
+      "loss": 0.4309,
+      "step": 8555
+    },
+    {
+      "epoch": 2.060664419836302,
+      "grad_norm": 1.1328125,
+      "learning_rate": 7.558566658802695e-05,
+      "loss": 0.4206,
+      "step": 8560
+    },
+    {
+      "epoch": 2.0618680789600385,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.551422903573566e-05,
+      "loss": 0.4061,
+      "step": 8565
+    },
+    {
+      "epoch": 2.063071738083775,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.5442851314192e-05,
+      "loss": 0.4041,
+      "step": 8570
+    },
+    {
+      "epoch": 2.0642753972075107,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.537153353743936e-05,
+      "loss": 0.3775,
+      "step": 8575
+    },
+    {
+      "epoch": 2.065479056331247,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.530027581942522e-05,
+      "loss": 0.4169,
+      "step": 8580
+    },
+    {
+      "epoch": 2.066682715454983,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.522907827400114e-05,
+      "loss": 0.4011,
+      "step": 8585
+    },
+    {
+      "epoch": 2.0678863745787193,
+      "grad_norm": 1.125,
+      "learning_rate": 7.515794101492259e-05,
+      "loss": 0.4128,
+      "step": 8590
+    },
+    {
+      "epoch": 2.0690900337024556,
+      "grad_norm": 1.1328125,
+      "learning_rate": 7.508686415584865e-05,
+      "loss": 0.3937,
+      "step": 8595
+    },
+    {
+      "epoch": 2.0702936928261915,
+      "grad_norm": 1.125,
+      "learning_rate": 7.501584781034188e-05,
+      "loss": 0.4162,
+      "step": 8600
+    },
+    {
+      "epoch": 2.071497351949928,
+      "grad_norm": 1.125,
+      "learning_rate": 7.494489209186828e-05,
+      "loss": 0.4208,
+      "step": 8605
+    },
+    {
+      "epoch": 2.072701011073664,
+      "grad_norm": 1.1953125,
+      "learning_rate": 7.48739971137968e-05,
+      "loss": 0.4115,
+      "step": 8610
+    },
+    {
+      "epoch": 2.0739046701974,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.480316298939953e-05,
+      "loss": 0.4011,
+      "step": 8615
+    },
+    {
+      "epoch": 2.0751083293211363,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.473238983185118e-05,
+      "loss": 0.3847,
+      "step": 8620
+    },
+    {
+      "epoch": 2.076311988444872,
+      "grad_norm": 1.2890625,
+      "learning_rate": 7.466167775422916e-05,
+      "loss": 0.4411,
+      "step": 8625
+    },
+    {
+      "epoch": 2.0775156475686085,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.45910268695132e-05,
+      "loss": 0.4003,
+      "step": 8630
+    },
+    {
+      "epoch": 2.078719306692345,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.45204372905853e-05,
+      "loss": 0.4185,
+      "step": 8635
+    },
+    {
+      "epoch": 2.0799229658160807,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.444990913022957e-05,
+      "loss": 0.4144,
+      "step": 8640
+    },
+    {
+      "epoch": 2.081126624939817,
+      "grad_norm": 1.1015625,
+      "learning_rate": 7.437944250113187e-05,
+      "loss": 0.3826,
+      "step": 8645
+    },
+    {
+      "epoch": 2.0823302840635534,
+      "grad_norm": 1.1015625,
+      "learning_rate": 7.430903751587986e-05,
+      "loss": 0.3917,
+      "step": 8650
+    },
+    {
+      "epoch": 2.0835339431872892,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.423869428696263e-05,
+      "loss": 0.3866,
+      "step": 8655
+    },
+    {
+      "epoch": 2.0847376023110256,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.41684129267706e-05,
+      "loss": 0.41,
+      "step": 8660
+    },
+    {
+      "epoch": 2.085941261434762,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.409819354759541e-05,
+      "loss": 0.3893,
+      "step": 8665
+    },
+    {
+      "epoch": 2.0871449205584978,
+      "grad_norm": 1.1875,
+      "learning_rate": 7.40280362616296e-05,
+      "loss": 0.4038,
+      "step": 8670
+    },
+    {
+      "epoch": 2.088348579682234,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.395794118096655e-05,
+      "loss": 0.4041,
+      "step": 8675
+    },
+    {
+      "epoch": 2.08955223880597,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.38879084176002e-05,
+      "loss": 0.3963,
+      "step": 8680
+    },
+    {
+      "epoch": 2.0907558979297063,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.381793808342497e-05,
+      "loss": 0.3936,
+      "step": 8685
+    },
+    {
+      "epoch": 2.0919595570534426,
+      "grad_norm": 1.125,
+      "learning_rate": 7.374803029023554e-05,
+      "loss": 0.4216,
+      "step": 8690
+    },
+    {
+      "epoch": 2.0931632161771785,
+      "grad_norm": 1.1953125,
+      "learning_rate": 7.367818514972661e-05,
+      "loss": 0.3984,
+      "step": 8695
+    },
+    {
+      "epoch": 2.094366875300915,
+      "grad_norm": 1.171875,
+      "learning_rate": 7.360840277349284e-05,
+      "loss": 0.3999,
+      "step": 8700
+    },
+    {
+      "epoch": 2.095570534424651,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.353868327302854e-05,
+      "loss": 0.3724,
+      "step": 8705
+    },
+    {
+      "epoch": 2.096774193548387,
+      "grad_norm": 1.125,
+      "learning_rate": 7.346902675972762e-05,
+      "loss": 0.3835,
+      "step": 8710
+    },
+    {
+      "epoch": 2.0979778526721233,
+      "grad_norm": 1.2265625,
+      "learning_rate": 7.339943334488335e-05,
+      "loss": 0.4211,
+      "step": 8715
+    },
+    {
+      "epoch": 2.099181511795859,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.332990313968812e-05,
+      "loss": 0.4278,
+      "step": 8720
+    },
+    {
+      "epoch": 2.1003851709195955,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.326043625523341e-05,
+      "loss": 0.4328,
+      "step": 8725
+    },
+    {
+      "epoch": 2.101588830043332,
+      "grad_norm": 1.171875,
+      "learning_rate": 7.319103280250947e-05,
+      "loss": 0.4016,
+      "step": 8730
+    },
+    {
+      "epoch": 2.1027924891670677,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.312169289240526e-05,
+      "loss": 0.4046,
+      "step": 8735
+    },
+    {
+      "epoch": 2.103996148290804,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.305241663570813e-05,
+      "loss": 0.4007,
+      "step": 8740
+    },
+    {
+      "epoch": 2.1051998074145404,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.298320414310382e-05,
+      "loss": 0.3916,
+      "step": 8745
+    },
+    {
+      "epoch": 2.1064034665382763,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.291405552517615e-05,
+      "loss": 0.3941,
+      "step": 8750
+    },
+    {
+      "epoch": 2.1076071256620126,
+      "grad_norm": 1.1875,
+      "learning_rate": 7.284497089240689e-05,
+      "loss": 0.3787,
+      "step": 8755
+    },
+    {
+      "epoch": 2.1088107847857485,
+      "grad_norm": 1.1953125,
+      "learning_rate": 7.277595035517552e-05,
+      "loss": 0.3666,
+      "step": 8760
+    },
+    {
+      "epoch": 2.110014443909485,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.270699402375924e-05,
+      "loss": 0.3993,
+      "step": 8765
+    },
+    {
+      "epoch": 2.111218103033221,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.263810200833252e-05,
+      "loss": 0.4071,
+      "step": 8770
+    },
+    {
+      "epoch": 2.112421762156957,
+      "grad_norm": 1.203125,
+      "learning_rate": 7.25692744189672e-05,
+      "loss": 0.4174,
+      "step": 8775
+    },
+    {
+      "epoch": 2.1136254212806933,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.250051136563206e-05,
+      "loss": 0.4113,
+      "step": 8780
+    },
+    {
+      "epoch": 2.1148290804044296,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.243181295819291e-05,
+      "loss": 0.4229,
+      "step": 8785
+    },
+    {
+      "epoch": 2.1160327395281655,
+      "grad_norm": 1.125,
+      "learning_rate": 7.236317930641211e-05,
+      "loss": 0.4348,
+      "step": 8790
+    },
+    {
+      "epoch": 2.117236398651902,
+      "grad_norm": 1.1875,
+      "learning_rate": 7.229461051994869e-05,
+      "loss": 0.4058,
+      "step": 8795
+    },
+    {
+      "epoch": 2.118440057775638,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.222610670835802e-05,
+      "loss": 0.3793,
+      "step": 8800
+    },
+    {
+      "epoch": 2.119643716899374,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.215766798109156e-05,
+      "loss": 0.3985,
+      "step": 8805
+    },
+    {
+      "epoch": 2.1208473760231104,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.208929444749692e-05,
+      "loss": 0.425,
+      "step": 8810
+    },
+    {
+      "epoch": 2.1220510351468462,
+      "grad_norm": 1.2734375,
+      "learning_rate": 7.202098621681746e-05,
+      "loss": 0.4327,
+      "step": 8815
+    },
+    {
+      "epoch": 2.1232546942705826,
+      "grad_norm": 1.2265625,
+      "learning_rate": 7.195274339819216e-05,
+      "loss": 0.4094,
+      "step": 8820
+    },
+    {
+      "epoch": 2.124458353394319,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.188456610065564e-05,
+      "loss": 0.4066,
+      "step": 8825
+    },
+    {
+      "epoch": 2.1256620125180548,
+      "grad_norm": 1.1796875,
+      "learning_rate": 7.181645443313766e-05,
+      "loss": 0.4166,
+      "step": 8830
+    },
+    {
+      "epoch": 2.126865671641791,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.17484085044633e-05,
+      "loss": 0.3624,
+      "step": 8835
+    },
+    {
+      "epoch": 2.1280693307655274,
+      "grad_norm": 1.2421875,
+      "learning_rate": 7.168042842335241e-05,
+      "loss": 0.414,
+      "step": 8840
+    },
+    {
+      "epoch": 2.1292729898892633,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.161251429841979e-05,
+      "loss": 0.405,
+      "step": 8845
+    },
+    {
+      "epoch": 2.1304766490129996,
+      "grad_norm": 1.2578125,
+      "learning_rate": 7.15446662381748e-05,
+      "loss": 0.3779,
+      "step": 8850
+    },
+    {
+      "epoch": 2.1316803081367355,
+      "grad_norm": 1.1953125,
+      "learning_rate": 7.147688435102122e-05,
+      "loss": 0.3961,
+      "step": 8855
+    },
+    {
+      "epoch": 2.132883967260472,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.140916874525718e-05,
+      "loss": 0.3901,
+      "step": 8860
+    },
+    {
+      "epoch": 2.134087626384208,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.134151952907482e-05,
+      "loss": 0.4149,
+      "step": 8865
+    },
+    {
+      "epoch": 2.135291285507944,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.127393681056024e-05,
+      "loss": 0.3962,
+      "step": 8870
+    },
+    {
+      "epoch": 2.1364949446316803,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.120642069769334e-05,
+      "loss": 0.3885,
+      "step": 8875
+    },
+    {
+      "epoch": 2.1376986037554166,
+      "grad_norm": 1.1015625,
+      "learning_rate": 7.113897129834749e-05,
+      "loss": 0.3891,
+      "step": 8880
+    },
+    {
+      "epoch": 2.1389022628791525,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.107158872028961e-05,
+      "loss": 0.4024,
+      "step": 8885
+    },
+    {
+      "epoch": 2.140105922002889,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.100427307117975e-05,
+      "loss": 0.3853,
+      "step": 8890
+    },
+    {
+      "epoch": 2.1413095811266247,
+      "grad_norm": 1.2890625,
+      "learning_rate": 7.093702445857105e-05,
+      "loss": 0.3992,
+      "step": 8895
+    },
+    {
+      "epoch": 2.142513240250361,
+      "grad_norm": 1.1875,
+      "learning_rate": 7.08698429899096e-05,
+      "loss": 0.4195,
+      "step": 8900
+    },
+    {
+      "epoch": 2.1437168993740974,
+      "grad_norm": 1.171875,
+      "learning_rate": 7.080272877253412e-05,
+      "loss": 0.3901,
+      "step": 8905
+    },
+    {
+      "epoch": 2.1449205584978333,
+      "grad_norm": 1.234375,
+      "learning_rate": 7.073568191367596e-05,
+      "loss": 0.4154,
+      "step": 8910
+    },
+    {
+      "epoch": 2.1461242176215696,
+      "grad_norm": 1.171875,
+      "learning_rate": 7.066870252045878e-05,
+      "loss": 0.4217,
+      "step": 8915
+    },
+    {
+      "epoch": 2.147327876745306,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.06017906998985e-05,
+      "loss": 0.3903,
+      "step": 8920
+    },
+    {
+      "epoch": 2.1485315358690418,
+      "grad_norm": 1.1796875,
+      "learning_rate": 7.053494655890306e-05,
+      "loss": 0.3995,
+      "step": 8925
+    },
+    {
+      "epoch": 2.149735194992778,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.046817020427223e-05,
+      "loss": 0.3909,
+      "step": 8930
+    },
+    {
+      "epoch": 2.1509388541165144,
+      "grad_norm": 1.1796875,
+      "learning_rate": 7.040146174269755e-05,
+      "loss": 0.3965,
+      "step": 8935
+    },
+    {
+      "epoch": 2.1521425132402503,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.033482128076199e-05,
+      "loss": 0.406,
+      "step": 8940
+    },
+    {
+      "epoch": 2.1533461723639866,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.026824892493997e-05,
+      "loss": 0.4156,
+      "step": 8945
+    },
+    {
+      "epoch": 2.1545498314877225,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.020174478159706e-05,
+      "loss": 0.3951,
+      "step": 8950
+    },
+    {
+      "epoch": 2.155753490611459,
+      "grad_norm": 1.2109375,
+      "learning_rate": 7.013530895698976e-05,
+      "loss": 0.3775,
+      "step": 8955
+    },
+    {
+      "epoch": 2.156957149735195,
+      "grad_norm": 1.1796875,
+      "learning_rate": 7.006894155726557e-05,
+      "loss": 0.4095,
+      "step": 8960
+    },
+    {
+      "epoch": 2.158160808858931,
+      "grad_norm": 1.1796875,
+      "learning_rate": 7.00026426884625e-05,
+      "loss": 0.3725,
+      "step": 8965
+    },
+    {
+      "epoch": 2.1593644679826673,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.993641245650924e-05,
+      "loss": 0.4198,
+      "step": 8970
+    },
+    {
+      "epoch": 2.1605681271064037,
+      "grad_norm": 1.21875,
+      "learning_rate": 6.987025096722466e-05,
+      "loss": 0.3956,
+      "step": 8975
+    },
+    {
+      "epoch": 2.1617717862301395,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.980415832631785e-05,
+      "loss": 0.3812,
+      "step": 8980
+    },
+    {
+      "epoch": 2.162975445353876,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.973813463938794e-05,
+      "loss": 0.3799,
+      "step": 8985
+    },
+    {
+      "epoch": 2.1641791044776117,
+      "grad_norm": 1.265625,
+      "learning_rate": 6.967218001192384e-05,
+      "loss": 0.4223,
+      "step": 8990
+    },
+    {
+      "epoch": 2.165382763601348,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.960629454930416e-05,
+      "loss": 0.3907,
+      "step": 8995
+    },
+    {
+      "epoch": 2.1665864227250844,
+      "grad_norm": 1.2578125,
+      "learning_rate": 6.954047835679694e-05,
+      "loss": 0.4085,
+      "step": 9000
+    },
+    {
+      "epoch": 2.1665864227250844,
+      "eval_loss": 0.37983912229537964,
+      "eval_runtime": 2.3776,
+      "eval_samples_per_second": 84.12,
+      "eval_steps_per_second": 84.12,
+      "step": 9000
+    },
+    {
+      "epoch": 2.1677900818488203,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.947473153955961e-05,
+      "loss": 0.3915,
+      "step": 9005
+    },
+    {
+      "epoch": 2.1689937409725566,
+      "grad_norm": 1.171875,
+      "learning_rate": 6.940905420263873e-05,
+      "loss": 0.4191,
+      "step": 9010
+    },
+    {
+      "epoch": 2.170197400096293,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.934344645096984e-05,
+      "loss": 0.394,
+      "step": 9015
+    },
+    {
+      "epoch": 2.171401059220029,
+      "grad_norm": 1.15625,
+      "learning_rate": 6.927790838937733e-05,
+      "loss": 0.3988,
+      "step": 9020
+    },
+    {
+      "epoch": 2.172604718343765,
+      "grad_norm": 1.2109375,
+      "learning_rate": 6.921244012257418e-05,
+      "loss": 0.3957,
+      "step": 9025
+    },
+    {
+      "epoch": 2.173808377467501,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.914704175516192e-05,
+      "loss": 0.421,
+      "step": 9030
+    },
+    {
+      "epoch": 2.1750120365912373,
+      "grad_norm": 1.1953125,
+      "learning_rate": 6.90817133916304e-05,
+      "loss": 0.4044,
+      "step": 9035
+    },
+    {
+      "epoch": 2.1762156957149736,
+      "grad_norm": 1.1796875,
+      "learning_rate": 6.901645513635753e-05,
+      "loss": 0.4083,
+      "step": 9040
+    },
+    {
+      "epoch": 2.1774193548387095,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.895126709360935e-05,
+      "loss": 0.4041,
+      "step": 9045
+    },
+    {
+      "epoch": 2.178623013962446,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.88861493675396e-05,
+      "loss": 0.4073,
+      "step": 9050
+    },
+    {
+      "epoch": 2.179826673086182,
+      "grad_norm": 1.203125,
+      "learning_rate": 6.882110206218971e-05,
+      "loss": 0.3887,
+      "step": 9055
+    },
+    {
+      "epoch": 2.181030332209918,
+      "grad_norm": 1.125,
+      "learning_rate": 6.875612528148862e-05,
+      "loss": 0.3788,
+      "step": 9060
+    },
+    {
+      "epoch": 2.1822339913336544,
+      "grad_norm": 1.2109375,
+      "learning_rate": 6.869121912925255e-05,
+      "loss": 0.4137,
+      "step": 9065
+    },
+    {
+      "epoch": 2.1834376504573907,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.862638370918493e-05,
+      "loss": 0.3982,
+      "step": 9070
+    },
+    {
+      "epoch": 2.1846413095811266,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.856161912487608e-05,
+      "loss": 0.4012,
+      "step": 9075
+    },
+    {
+      "epoch": 2.185844968704863,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.849692547980326e-05,
+      "loss": 0.4167,
+      "step": 9080
+    },
+    {
+      "epoch": 2.1870486278285988,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.843230287733031e-05,
+      "loss": 0.3923,
+      "step": 9085
+    },
+    {
+      "epoch": 2.188252286952335,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.836775142070754e-05,
+      "loss": 0.4036,
+      "step": 9090
+    },
+    {
+      "epoch": 2.1894559460760714,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.83032712130717e-05,
+      "loss": 0.4049,
+      "step": 9095
+    },
+    {
+      "epoch": 2.1906596051998073,
+      "grad_norm": 1.21875,
+      "learning_rate": 6.823886235744558e-05,
+      "loss": 0.4188,
+      "step": 9100
+    },
+    {
+      "epoch": 2.1918632643235436,
+      "grad_norm": 1.3046875,
+      "learning_rate": 6.817452495673803e-05,
+      "loss": 0.4083,
+      "step": 9105
+    },
+    {
+      "epoch": 2.19306692344728,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.811025911374376e-05,
+      "loss": 0.3937,
+      "step": 9110
+    },
+    {
+      "epoch": 2.194270582571016,
+      "grad_norm": 1.0546875,
+      "learning_rate": 6.804606493114308e-05,
+      "loss": 0.4093,
+      "step": 9115
+    },
+    {
+      "epoch": 2.195474241694752,
+      "grad_norm": 1.234375,
+      "learning_rate": 6.798194251150184e-05,
+      "loss": 0.4323,
+      "step": 9120
+    },
+    {
+      "epoch": 2.196677900818488,
+      "grad_norm": 1.125,
+      "learning_rate": 6.791789195727122e-05,
+      "loss": 0.4055,
+      "step": 9125
+    },
+    {
+      "epoch": 2.1978815599422243,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.785391337078763e-05,
+      "loss": 0.4068,
+      "step": 9130
+    },
+    {
+      "epoch": 2.1990852190659607,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.77900068542724e-05,
+      "loss": 0.4158,
+      "step": 9135
+    },
+    {
+      "epoch": 2.2002888781896965,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.772617250983177e-05,
+      "loss": 0.412,
+      "step": 9140
+    },
+    {
+      "epoch": 2.201492537313433,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.766241043945668e-05,
+      "loss": 0.4121,
+      "step": 9145
+    },
+    {
+      "epoch": 2.202696196437169,
+      "grad_norm": 1.2265625,
+      "learning_rate": 6.759872074502254e-05,
+      "loss": 0.426,
+      "step": 9150
+    },
+    {
+      "epoch": 2.203899855560905,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.753510352828918e-05,
+      "loss": 0.4078,
+      "step": 9155
+    },
+    {
+      "epoch": 2.2051035146846414,
+      "grad_norm": 1.171875,
+      "learning_rate": 6.747155889090062e-05,
+      "loss": 0.4036,
+      "step": 9160
+    },
+    {
+      "epoch": 2.2063071738083773,
+      "grad_norm": 1.1953125,
+      "learning_rate": 6.740808693438485e-05,
+      "loss": 0.3795,
+      "step": 9165
+    },
+    {
+      "epoch": 2.2075108329321136,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.734468776015385e-05,
+      "loss": 0.3897,
+      "step": 9170
+    },
+    {
+      "epoch": 2.20871449205585,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.728136146950318e-05,
+      "loss": 0.3972,
+      "step": 9175
+    },
+    {
+      "epoch": 2.209918151179586,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.721810816361207e-05,
+      "loss": 0.3788,
+      "step": 9180
+    },
+    {
+      "epoch": 2.211121810303322,
+      "grad_norm": 1.2578125,
+      "learning_rate": 6.715492794354308e-05,
+      "loss": 0.4073,
+      "step": 9185
+    },
+    {
+      "epoch": 2.2123254694270584,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.709182091024196e-05,
+      "loss": 0.3923,
+      "step": 9190
+    },
+    {
+      "epoch": 2.2135291285507943,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.702878716453764e-05,
+      "loss": 0.3862,
+      "step": 9195
+    },
+    {
+      "epoch": 2.2147327876745306,
+      "grad_norm": 1.3125,
+      "learning_rate": 6.696582680714181e-05,
+      "loss": 0.4135,
+      "step": 9200
+    },
+    {
+      "epoch": 2.215936446798267,
+      "grad_norm": 1.21875,
+      "learning_rate": 6.690293993864907e-05,
+      "loss": 0.3883,
+      "step": 9205
+    },
+    {
+      "epoch": 2.217140105922003,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.684012665953645e-05,
+      "loss": 0.3833,
+      "step": 9210
+    },
+    {
+      "epoch": 2.218343765045739,
+      "grad_norm": 1.3359375,
+      "learning_rate": 6.677738707016347e-05,
+      "loss": 0.4578,
+      "step": 9215
+    },
+    {
+      "epoch": 2.219547424169475,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.671472127077195e-05,
+      "loss": 0.3936,
+      "step": 9220
+    },
+    {
+      "epoch": 2.2207510832932114,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.665212936148573e-05,
+      "loss": 0.4076,
+      "step": 9225
+    },
+    {
+      "epoch": 2.2219547424169477,
+      "grad_norm": 1.234375,
+      "learning_rate": 6.658961144231066e-05,
+      "loss": 0.4269,
+      "step": 9230
+    },
+    {
+      "epoch": 2.2231584015406836,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.652716761313433e-05,
+      "loss": 0.407,
+      "step": 9235
+    },
+    {
+      "epoch": 2.22436206066442,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.646479797372603e-05,
+      "loss": 0.3887,
+      "step": 9240
+    },
+    {
+      "epoch": 2.2255657197881558,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.640250262373638e-05,
+      "loss": 0.3935,
+      "step": 9245
+    },
+    {
+      "epoch": 2.226769378911892,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.63402816626974e-05,
+      "loss": 0.39,
+      "step": 9250
+    },
+    {
+      "epoch": 2.2279730380356284,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.627813519002227e-05,
+      "loss": 0.388,
+      "step": 9255
+    },
+    {
+      "epoch": 2.2291766971593643,
+      "grad_norm": 1.1953125,
+      "learning_rate": 6.621606330500507e-05,
+      "loss": 0.4123,
+      "step": 9260
+    },
+    {
+      "epoch": 2.2303803562831006,
+      "grad_norm": 1.2421875,
+      "learning_rate": 6.615406610682079e-05,
+      "loss": 0.3847,
+      "step": 9265
+    },
+    {
+      "epoch": 2.231584015406837,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.609214369452507e-05,
+      "loss": 0.3888,
+      "step": 9270
+    },
+    {
+      "epoch": 2.232787674530573,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.603029616705402e-05,
+      "loss": 0.3791,
+      "step": 9275
+    },
+    {
+      "epoch": 2.233991333654309,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.596852362322418e-05,
+      "loss": 0.4157,
+      "step": 9280
+    },
+    {
+      "epoch": 2.2351949927780455,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.590682616173218e-05,
+      "loss": 0.3867,
+      "step": 9285
+    },
+    {
+      "epoch": 2.2363986519017813,
+      "grad_norm": 1.2734375,
+      "learning_rate": 6.58452038811548e-05,
+      "loss": 0.4108,
+      "step": 9290
+    },
+    {
+      "epoch": 2.2376023110255177,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.578365687994862e-05,
+      "loss": 0.4389,
+      "step": 9295
+    },
+    {
+      "epoch": 2.2388059701492535,
+      "grad_norm": 1.109375,
+      "learning_rate": 6.572218525644996e-05,
+      "loss": 0.4346,
+      "step": 9300
+    },
+    {
+      "epoch": 2.24000962927299,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.566078910887475e-05,
+      "loss": 0.388,
+      "step": 9305
+    },
+    {
+      "epoch": 2.241213288396726,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.559946853531826e-05,
+      "loss": 0.3964,
+      "step": 9310
+    },
+    {
+      "epoch": 2.242416947520462,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.553822363375506e-05,
+      "loss": 0.3974,
+      "step": 9315
+    },
+    {
+      "epoch": 2.2436206066441984,
+      "grad_norm": 1.171875,
+      "learning_rate": 6.547705450203882e-05,
+      "loss": 0.4184,
+      "step": 9320
+    },
+    {
+      "epoch": 2.2448242657679347,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.541596123790211e-05,
+      "loss": 0.4027,
+      "step": 9325
+    },
+    {
+      "epoch": 2.2460279248916706,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.535494393895632e-05,
+      "loss": 0.4171,
+      "step": 9330
+    },
+    {
+      "epoch": 2.247231584015407,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.529400270269144e-05,
+      "loss": 0.3795,
+      "step": 9335
+    },
+    {
+      "epoch": 2.2484352431391432,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.523313762647595e-05,
+      "loss": 0.426,
+      "step": 9340
+    },
+    {
+      "epoch": 2.249638902262879,
+      "grad_norm": 1.125,
+      "learning_rate": 6.517234880755663e-05,
+      "loss": 0.4125,
+      "step": 9345
+    },
+    {
+      "epoch": 2.2508425613866154,
+      "grad_norm": 1.21875,
+      "learning_rate": 6.511163634305844e-05,
+      "loss": 0.422,
+      "step": 9350
+    },
+    {
+      "epoch": 2.2520462205103513,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.505100032998434e-05,
+      "loss": 0.3871,
+      "step": 9355
+    },
+    {
+      "epoch": 2.2532498796340876,
+      "grad_norm": 1.15625,
+      "learning_rate": 6.499044086521509e-05,
+      "loss": 0.4342,
+      "step": 9360
+    },
+    {
+      "epoch": 2.254453538757824,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.492995804550923e-05,
+      "loss": 0.3903,
+      "step": 9365
+    },
+    {
+      "epoch": 2.25565719788156,
+      "grad_norm": 1.125,
+      "learning_rate": 6.486955196750282e-05,
+      "loss": 0.3923,
+      "step": 9370
+    },
+    {
+      "epoch": 2.256860857005296,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.480922272770924e-05,
+      "loss": 0.3975,
+      "step": 9375
+    },
+    {
+      "epoch": 2.258064516129032,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.474897042251918e-05,
+      "loss": 0.4123,
+      "step": 9380
+    },
+    {
+      "epoch": 2.2592681752527684,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.468879514820038e-05,
+      "loss": 0.4151,
+      "step": 9385
+    },
+    {
+      "epoch": 2.2604718343765047,
+      "grad_norm": 1.203125,
+      "learning_rate": 6.462869700089747e-05,
+      "loss": 0.3918,
+      "step": 9390
+    },
+    {
+      "epoch": 2.2616754935002406,
+      "grad_norm": 1.3046875,
+      "learning_rate": 6.456867607663193e-05,
+      "loss": 0.3907,
+      "step": 9395
+    },
+    {
+      "epoch": 2.262879152623977,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.450873247130179e-05,
+      "loss": 0.3894,
+      "step": 9400
+    },
+    {
+      "epoch": 2.264082811747713,
+      "grad_norm": 1.21875,
+      "learning_rate": 6.444886628068155e-05,
+      "loss": 0.3874,
+      "step": 9405
+    },
+    {
+      "epoch": 2.265286470871449,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.438907760042204e-05,
+      "loss": 0.3837,
+      "step": 9410
+    },
+    {
+      "epoch": 2.2664901299951854,
+      "grad_norm": 1.21875,
+      "learning_rate": 6.432936652605025e-05,
+      "loss": 0.4001,
+      "step": 9415
+    },
+    {
+      "epoch": 2.2676937891189217,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.426973315296917e-05,
+      "loss": 0.4088,
+      "step": 9420
+    },
+    {
+      "epoch": 2.2688974482426576,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.421017757645764e-05,
+      "loss": 0.4058,
+      "step": 9425
+    },
+    {
+      "epoch": 2.270101107366394,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.415069989167023e-05,
+      "loss": 0.3805,
+      "step": 9430
+    },
+    {
+      "epoch": 2.27130476649013,
+      "grad_norm": 1.25,
+      "learning_rate": 6.409130019363698e-05,
+      "loss": 0.4057,
+      "step": 9435
+    },
+    {
+      "epoch": 2.272508425613866,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.403197857726344e-05,
+      "loss": 0.401,
+      "step": 9440
+    },
+    {
+      "epoch": 2.2737120847376024,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.397273513733029e-05,
+      "loss": 0.3913,
+      "step": 9445
+    },
+    {
+      "epoch": 2.2749157438613383,
+      "grad_norm": 1.2578125,
+      "learning_rate": 6.39135699684934e-05,
+      "loss": 0.423,
+      "step": 9450
+    },
+    {
+      "epoch": 2.2761194029850746,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.385448316528354e-05,
+      "loss": 0.3915,
+      "step": 9455
+    },
+    {
+      "epoch": 2.277323062108811,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.379547482210624e-05,
+      "loss": 0.3817,
+      "step": 9460
+    },
+    {
+      "epoch": 2.278526721232547,
+      "grad_norm": 1.234375,
+      "learning_rate": 6.373654503324175e-05,
+      "loss": 0.3925,
+      "step": 9465
+    },
+    {
+      "epoch": 2.279730380356283,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.367769389284472e-05,
+      "loss": 0.4074,
+      "step": 9470
+    },
+    {
+      "epoch": 2.2809340394800195,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.361892149494422e-05,
+      "loss": 0.4199,
+      "step": 9475
+    },
+    {
+      "epoch": 2.2821376986037554,
+      "grad_norm": 1.3671875,
+      "learning_rate": 6.356022793344349e-05,
+      "loss": 0.4211,
+      "step": 9480
+    },
+    {
+      "epoch": 2.2833413577274917,
+      "grad_norm": 1.2734375,
+      "learning_rate": 6.350161330211973e-05,
+      "loss": 0.3933,
+      "step": 9485
+    },
+    {
+      "epoch": 2.2845450168512276,
+      "grad_norm": 1.15625,
+      "learning_rate": 6.344307769462417e-05,
+      "loss": 0.4069,
+      "step": 9490
+    },
+    {
+      "epoch": 2.285748675974964,
+      "grad_norm": 1.171875,
+      "learning_rate": 6.338462120448164e-05,
+      "loss": 0.4346,
+      "step": 9495
+    },
+    {
+      "epoch": 2.2869523350987,
+      "grad_norm": 1.2265625,
+      "learning_rate": 6.332624392509068e-05,
+      "loss": 0.4202,
+      "step": 9500
+    },
+    {
+      "epoch": 2.2869523350987,
+      "eval_loss": 0.3756787180900574,
+      "eval_runtime": 2.344,
+      "eval_samples_per_second": 85.324,
+      "eval_steps_per_second": 85.324,
+      "step": 9500
+    },
+    {
+      "epoch": 2.288155994222436,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.326794594972318e-05,
+      "loss": 0.3907,
+      "step": 9505
+    },
+    {
+      "epoch": 2.2893596533461724,
+      "grad_norm": 1.0546875,
+      "learning_rate": 6.320972737152439e-05,
+      "loss": 0.3857,
+      "step": 9510
+    },
+    {
+      "epoch": 2.2905633124699083,
+      "grad_norm": 1.2265625,
+      "learning_rate": 6.315158828351266e-05,
+      "loss": 0.3914,
+      "step": 9515
+    },
+    {
+      "epoch": 2.2917669715936446,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.309352877857933e-05,
+      "loss": 0.4172,
+      "step": 9520
+    },
+    {
+      "epoch": 2.292970630717381,
+      "grad_norm": 1.125,
+      "learning_rate": 6.303554894948859e-05,
+      "loss": 0.3929,
+      "step": 9525
+    },
+    {
+      "epoch": 2.294174289841117,
+      "grad_norm": 1.125,
+      "learning_rate": 6.297764888887743e-05,
+      "loss": 0.4198,
+      "step": 9530
+    },
+    {
+      "epoch": 2.295377948964853,
+      "grad_norm": 1.1796875,
+      "learning_rate": 6.291982868925522e-05,
+      "loss": 0.4168,
+      "step": 9535
+    },
+    {
+      "epoch": 2.2965816080885895,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.286208844300387e-05,
+      "loss": 0.4076,
+      "step": 9540
+    },
+    {
+      "epoch": 2.2977852672123253,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.280442824237744e-05,
+      "loss": 0.4014,
+      "step": 9545
+    },
+    {
+      "epoch": 2.2989889263360617,
+      "grad_norm": 1.2421875,
+      "learning_rate": 6.274684817950216e-05,
+      "loss": 0.4014,
+      "step": 9550
+    },
+    {
+      "epoch": 2.300192585459798,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.268934834637626e-05,
+      "loss": 0.392,
+      "step": 9555
+    },
+    {
+      "epoch": 2.301396244583534,
+      "grad_norm": 1.1953125,
+      "learning_rate": 6.26319288348697e-05,
+      "loss": 0.4122,
+      "step": 9560
+    },
+    {
+      "epoch": 2.30259990370727,
+      "grad_norm": 1.1953125,
+      "learning_rate": 6.257458973672414e-05,
+      "loss": 0.3761,
+      "step": 9565
+    },
+    {
+      "epoch": 2.303803562831006,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.251733114355274e-05,
+      "loss": 0.4226,
+      "step": 9570
+    },
+    {
+      "epoch": 2.3050072219547424,
+      "grad_norm": 1.109375,
+      "learning_rate": 6.246015314684007e-05,
+      "loss": 0.376,
+      "step": 9575
+    },
+    {
+      "epoch": 2.3062108810784787,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.240305583794193e-05,
+      "loss": 0.4237,
+      "step": 9580
+    },
+    {
+      "epoch": 2.3074145402022146,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.234603930808515e-05,
+      "loss": 0.3921,
+      "step": 9585
+    },
+    {
+      "epoch": 2.308618199325951,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.228910364836756e-05,
+      "loss": 0.4114,
+      "step": 9590
+    },
+    {
+      "epoch": 2.3098218584496872,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.223224894975772e-05,
+      "loss": 0.4328,
+      "step": 9595
+    },
+    {
+      "epoch": 2.311025517573423,
+      "grad_norm": 1.125,
+      "learning_rate": 6.217547530309486e-05,
+      "loss": 0.4088,
+      "step": 9600
+    },
+    {
+      "epoch": 2.3122291766971594,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.211878279908876e-05,
+      "loss": 0.3963,
+      "step": 9605
+    },
+    {
+      "epoch": 2.3134328358208958,
+      "grad_norm": 1.125,
+      "learning_rate": 6.206217152831948e-05,
+      "loss": 0.3749,
+      "step": 9610
+    },
+    {
+      "epoch": 2.3146364949446316,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.200564158123732e-05,
+      "loss": 0.3965,
+      "step": 9615
+    },
+    {
+      "epoch": 2.315840154068368,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.19491930481626e-05,
+      "loss": 0.4348,
+      "step": 9620
+    },
+    {
+      "epoch": 2.317043813192104,
+      "grad_norm": 1.203125,
+      "learning_rate": 6.189282601928567e-05,
+      "loss": 0.3908,
+      "step": 9625
+    },
+    {
+      "epoch": 2.31824747231584,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.183654058466657e-05,
+      "loss": 0.4179,
+      "step": 9630
+    },
+    {
+      "epoch": 2.3194511314395765,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.178033683423499e-05,
+      "loss": 0.3856,
+      "step": 9635
+    },
+    {
+      "epoch": 2.3206547905633124,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.172421485779012e-05,
+      "loss": 0.3889,
+      "step": 9640
+    },
+    {
+      "epoch": 2.3218584496870487,
+      "grad_norm": 1.0,
+      "learning_rate": 6.166817474500047e-05,
+      "loss": 0.3951,
+      "step": 9645
+    },
+    {
+      "epoch": 2.3230621088107846,
+      "grad_norm": 1.234375,
+      "learning_rate": 6.161221658540383e-05,
+      "loss": 0.4012,
+      "step": 9650
+    },
+    {
+      "epoch": 2.324265767934521,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.155634046840693e-05,
+      "loss": 0.3987,
+      "step": 9655
+    },
+    {
+      "epoch": 2.325469427058257,
+      "grad_norm": 1.1796875,
+      "learning_rate": 6.150054648328552e-05,
+      "loss": 0.3923,
+      "step": 9660
+    },
+    {
+      "epoch": 2.326673086181993,
+      "grad_norm": 1.15625,
+      "learning_rate": 6.144483471918406e-05,
+      "loss": 0.3805,
+      "step": 9665
+    },
+    {
+      "epoch": 2.3278767453057294,
+      "grad_norm": 1.2421875,
+      "learning_rate": 6.138920526511571e-05,
+      "loss": 0.4214,
+      "step": 9670
+    },
+    {
+      "epoch": 2.3290804044294657,
+      "grad_norm": 1.109375,
+      "learning_rate": 6.133365820996204e-05,
+      "loss": 0.4041,
+      "step": 9675
+    },
+    {
+      "epoch": 2.3302840635532016,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.127819364247298e-05,
+      "loss": 0.4002,
+      "step": 9680
+    },
+    {
+      "epoch": 2.331487722676938,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.122281165126675e-05,
+      "loss": 0.4156,
+      "step": 9685
+    },
+    {
+      "epoch": 2.3326913818006743,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.116751232482955e-05,
+      "loss": 0.3841,
+      "step": 9690
+    },
+    {
+      "epoch": 2.33389504092441,
+      "grad_norm": 1.203125,
+      "learning_rate": 6.111229575151552e-05,
+      "loss": 0.3953,
+      "step": 9695
+    },
+    {
+      "epoch": 2.3350987000481465,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.105716201954659e-05,
+      "loss": 0.4067,
+      "step": 9700
+    },
+    {
+      "epoch": 2.3363023591718823,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.1002111217012335e-05,
+      "loss": 0.3868,
+      "step": 9705
+    },
+    {
+      "epoch": 2.3375060182956187,
+      "grad_norm": 1.1953125,
+      "learning_rate": 6.094714343186982e-05,
+      "loss": 0.3945,
+      "step": 9710
+    },
+    {
+      "epoch": 2.338709677419355,
+      "grad_norm": 1.15625,
+      "learning_rate": 6.089225875194349e-05,
+      "loss": 0.3912,
+      "step": 9715
+    },
+    {
+      "epoch": 2.339913336543091,
+      "grad_norm": 1.125,
+      "learning_rate": 6.083745726492499e-05,
+      "loss": 0.3853,
+      "step": 9720
+    },
+    {
+      "epoch": 2.341116995666827,
+      "grad_norm": 1.265625,
+      "learning_rate": 6.078273905837305e-05,
+      "loss": 0.4007,
+      "step": 9725
+    },
+    {
+      "epoch": 2.3423206547905635,
+      "grad_norm": 1.15625,
+      "learning_rate": 6.072810421971328e-05,
+      "loss": 0.403,
+      "step": 9730
+    },
+    {
+      "epoch": 2.3435243139142994,
+      "grad_norm": 1.125,
+      "learning_rate": 6.067355283623823e-05,
+      "loss": 0.4096,
+      "step": 9735
+    },
+    {
+      "epoch": 2.3447279730380357,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.0619084995106985e-05,
+      "loss": 0.3982,
+      "step": 9740
+    },
+    {
+      "epoch": 2.345931632161772,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.0564700783345177e-05,
+      "loss": 0.3865,
+      "step": 9745
+    },
+    {
+      "epoch": 2.347135291285508,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.051040028784489e-05,
+      "loss": 0.4359,
+      "step": 9750
+    },
+    {
+      "epoch": 2.3483389504092442,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.045618359536434e-05,
+      "loss": 0.3731,
+      "step": 9755
+    },
+    {
+      "epoch": 2.34954260953298,
+      "grad_norm": 1.125,
+      "learning_rate": 6.0402050792527914e-05,
+      "loss": 0.3772,
+      "step": 9760
+    },
+    {
+      "epoch": 2.3507462686567164,
+      "grad_norm": 1.28125,
+      "learning_rate": 6.0348001965825986e-05,
+      "loss": 0.398,
+      "step": 9765
+    },
+    {
+      "epoch": 2.3519499277804528,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.02940372016147e-05,
+      "loss": 0.3803,
+      "step": 9770
+    },
+    {
+      "epoch": 2.3531535869041886,
+      "grad_norm": 1.2890625,
+      "learning_rate": 6.0240156586115926e-05,
+      "loss": 0.3942,
+      "step": 9775
+    },
+    {
+      "epoch": 2.354357246027925,
+      "grad_norm": 1.2109375,
+      "learning_rate": 6.0186360205417044e-05,
+      "loss": 0.3934,
+      "step": 9780
+    },
+    {
+      "epoch": 2.355560905151661,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.0132648145470914e-05,
+      "loss": 0.4019,
+      "step": 9785
+    },
+    {
+      "epoch": 2.356764564275397,
+      "grad_norm": 1.21875,
+      "learning_rate": 6.0079020492095636e-05,
+      "loss": 0.4345,
+      "step": 9790
+    },
+    {
+      "epoch": 2.3579682233991335,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.002547733097443e-05,
+      "loss": 0.3959,
+      "step": 9795
+    },
+    {
+      "epoch": 2.3591718825228694,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.9972018747655576e-05,
+      "loss": 0.3901,
+      "step": 9800
+    },
+    {
+      "epoch": 2.3603755416466057,
+      "grad_norm": 1.125,
+      "learning_rate": 5.9918644827552145e-05,
+      "loss": 0.3859,
+      "step": 9805
+    },
+    {
+      "epoch": 2.361579200770342,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.9865355655942e-05,
+      "loss": 0.3907,
+      "step": 9810
+    },
+    {
+      "epoch": 2.362782859894078,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.98121513179676e-05,
+      "loss": 0.4194,
+      "step": 9815
+    },
+    {
+      "epoch": 2.363986519017814,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.9759031898635787e-05,
+      "loss": 0.4023,
+      "step": 9820
+    },
+    {
+      "epoch": 2.3651901781415505,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.9705997482817825e-05,
+      "loss": 0.415,
+      "step": 9825
+    },
+    {
+      "epoch": 2.3663938372652864,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.96530481552491e-05,
+      "loss": 0.3732,
+      "step": 9830
+    },
+    {
+      "epoch": 2.3675974963890227,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.9600184000529025e-05,
+      "loss": 0.4401,
+      "step": 9835
+    },
+    {
+      "epoch": 2.3688011555127586,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.954740510312103e-05,
+      "loss": 0.402,
+      "step": 9840
+    },
+    {
+      "epoch": 2.370004814636495,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.9494711547352204e-05,
+      "loss": 0.3932,
+      "step": 9845
+    },
+    {
+      "epoch": 2.3712084737602313,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.94421034174134e-05,
+      "loss": 0.4304,
+      "step": 9850
+    },
+    {
+      "epoch": 2.372412132883967,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.9389580797358866e-05,
+      "loss": 0.3816,
+      "step": 9855
+    },
+    {
+      "epoch": 2.3736157920077035,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.933714377110631e-05,
+      "loss": 0.382,
+      "step": 9860
+    },
+    {
+      "epoch": 2.3748194511314393,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.928479242243668e-05,
+      "loss": 0.3949,
+      "step": 9865
+    },
+    {
+      "epoch": 2.3760231102551757,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.923252683499397e-05,
+      "loss": 0.3954,
+      "step": 9870
+    },
+    {
+      "epoch": 2.377226769378912,
+      "grad_norm": 1.296875,
+      "learning_rate": 5.9180347092285224e-05,
+      "loss": 0.3966,
+      "step": 9875
+    },
+    {
+      "epoch": 2.3784304285026483,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.912825327768029e-05,
+      "loss": 0.4403,
+      "step": 9880
+    },
+    {
+      "epoch": 2.379634087626384,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.907624547441171e-05,
+      "loss": 0.3975,
+      "step": 9885
+    },
+    {
+      "epoch": 2.3808377467501205,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.9024323765574615e-05,
+      "loss": 0.3871,
+      "step": 9890
+    },
+    {
+      "epoch": 2.3820414058738564,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.8972488234126614e-05,
+      "loss": 0.4278,
+      "step": 9895
+    },
+    {
+      "epoch": 2.3832450649975927,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.89207389628876e-05,
+      "loss": 0.3843,
+      "step": 9900
+    },
+    {
+      "epoch": 2.384448724121329,
+      "grad_norm": 1.125,
+      "learning_rate": 5.88690760345396e-05,
+      "loss": 0.3924,
+      "step": 9905
+    },
+    {
+      "epoch": 2.385652383245065,
+      "grad_norm": 1.28125,
+      "learning_rate": 5.881749953162678e-05,
+      "loss": 0.3939,
+      "step": 9910
+    },
+    {
+      "epoch": 2.3868560423688012,
+      "grad_norm": 1.125,
+      "learning_rate": 5.8766009536555125e-05,
+      "loss": 0.4135,
+      "step": 9915
+    },
+    {
+      "epoch": 2.388059701492537,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.871460613159248e-05,
+      "loss": 0.3938,
+      "step": 9920
+    },
+    {
+      "epoch": 2.3892633606162734,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.866328939886831e-05,
+      "loss": 0.3952,
+      "step": 9925
+    },
+    {
+      "epoch": 2.3904670197400097,
+      "grad_norm": 1.2265625,
+      "learning_rate": 5.861205942037359e-05,
+      "loss": 0.3796,
+      "step": 9930
+    },
+    {
+      "epoch": 2.3916706788637456,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.856091627796069e-05,
+      "loss": 0.3852,
+      "step": 9935
+    },
+    {
+      "epoch": 2.392874337987482,
+      "grad_norm": 1.2890625,
+      "learning_rate": 5.850986005334323e-05,
+      "loss": 0.3841,
+      "step": 9940
+    },
+    {
+      "epoch": 2.3940779971112183,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.845889082809597e-05,
+      "loss": 0.3893,
+      "step": 9945
+    },
+    {
+      "epoch": 2.395281656234954,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.84080086836547e-05,
+      "loss": 0.4146,
+      "step": 9950
+    },
+    {
+      "epoch": 2.3964853153586905,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.835721370131599e-05,
+      "loss": 0.3858,
+      "step": 9955
+    },
+    {
+      "epoch": 2.397688974482427,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.8306505962237257e-05,
+      "loss": 0.4173,
+      "step": 9960
+    },
+    {
+      "epoch": 2.3988926336061627,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.825588554743643e-05,
+      "loss": 0.4106,
+      "step": 9965
+    },
+    {
+      "epoch": 2.400096292729899,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.820535253779196e-05,
+      "loss": 0.3961,
+      "step": 9970
+    },
+    {
+      "epoch": 2.401299951853635,
+      "grad_norm": 1.2265625,
+      "learning_rate": 5.815490701404266e-05,
+      "loss": 0.4355,
+      "step": 9975
+    },
+    {
+      "epoch": 2.402503610977371,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.8104549056787514e-05,
+      "loss": 0.386,
+      "step": 9980
+    },
+    {
+      "epoch": 2.4037072701011075,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.805427874648568e-05,
+      "loss": 0.4063,
+      "step": 9985
+    },
+    {
+      "epoch": 2.4049109292248434,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.800409616345619e-05,
+      "loss": 0.3664,
+      "step": 9990
+    },
+    {
+      "epoch": 2.4061145883485797,
+      "grad_norm": 1.1875,
+      "learning_rate": 5.7954001387877925e-05,
+      "loss": 0.3639,
+      "step": 9995
+    },
+    {
+      "epoch": 2.4073182474723156,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.790399449978955e-05,
+      "loss": 0.4059,
+      "step": 10000
+    },
+    {
+      "epoch": 2.4073182474723156,
+      "eval_loss": 0.36942535638809204,
+      "eval_runtime": 2.3535,
+      "eval_samples_per_second": 84.981,
+      "eval_steps_per_second": 84.981,
+      "step": 10000
+    },
+    {
+      "epoch": 2.408521906596052,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.78540755790892e-05,
+      "loss": 0.3825,
+      "step": 10005
+    },
+    {
+      "epoch": 2.4097255657197882,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.7804244705534553e-05,
+      "loss": 0.3795,
+      "step": 10010
+    },
+    {
+      "epoch": 2.410929224843524,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.775450195874255e-05,
+      "loss": 0.3916,
+      "step": 10015
+    },
+    {
+      "epoch": 2.4121328839672604,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.770484741818935e-05,
+      "loss": 0.394,
+      "step": 10020
+    },
+    {
+      "epoch": 2.4133365430909968,
+      "grad_norm": 1.2421875,
+      "learning_rate": 5.765528116321019e-05,
+      "loss": 0.3868,
+      "step": 10025
+    },
+    {
+      "epoch": 2.4145402022147326,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.760580327299922e-05,
+      "loss": 0.416,
+      "step": 10030
+    },
+    {
+      "epoch": 2.415743861338469,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.7556413826609464e-05,
+      "loss": 0.3958,
+      "step": 10035
+    },
+    {
+      "epoch": 2.4169475204622053,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.750711290295257e-05,
+      "loss": 0.3945,
+      "step": 10040
+    },
+    {
+      "epoch": 2.418151179585941,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.7457900580798746e-05,
+      "loss": 0.423,
+      "step": 10045
+    },
+    {
+      "epoch": 2.4193548387096775,
+      "grad_norm": 1.2734375,
+      "learning_rate": 5.7408776938776725e-05,
+      "loss": 0.4141,
+      "step": 10050
+    },
+    {
+      "epoch": 2.4205584978334134,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.7359742055373445e-05,
+      "loss": 0.3918,
+      "step": 10055
+    },
+    {
+      "epoch": 2.4217621569571497,
+      "grad_norm": 1.1875,
+      "learning_rate": 5.731079600893412e-05,
+      "loss": 0.4105,
+      "step": 10060
+    },
+    {
+      "epoch": 2.422965816080886,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.726193887766194e-05,
+      "loss": 0.3852,
+      "step": 10065
+    },
+    {
+      "epoch": 2.424169475204622,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.721317073961811e-05,
+      "loss": 0.384,
+      "step": 10070
+    },
+    {
+      "epoch": 2.425373134328358,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.716449167272162e-05,
+      "loss": 0.3943,
+      "step": 10075
+    },
+    {
+      "epoch": 2.4265767934520945,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.71159017547491e-05,
+      "loss": 0.3999,
+      "step": 10080
+    },
+    {
+      "epoch": 2.4277804525758304,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.706740106333482e-05,
+      "loss": 0.4005,
+      "step": 10085
+    },
+    {
+      "epoch": 2.4289841116995667,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.701898967597045e-05,
+      "loss": 0.3771,
+      "step": 10090
+    },
+    {
+      "epoch": 2.430187770823303,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.697066767000494e-05,
+      "loss": 0.3732,
+      "step": 10095
+    },
+    {
+      "epoch": 2.431391429947039,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.6922435122644514e-05,
+      "loss": 0.4007,
+      "step": 10100
+    },
+    {
+      "epoch": 2.4325950890707753,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.687429211095238e-05,
+      "loss": 0.4237,
+      "step": 10105
+    },
+    {
+      "epoch": 2.433798748194511,
+      "grad_norm": 1.0,
+      "learning_rate": 5.6826238711848764e-05,
+      "loss": 0.3836,
+      "step": 10110
+    },
+    {
+      "epoch": 2.4350024073182475,
+      "grad_norm": 1.28125,
+      "learning_rate": 5.677827500211065e-05,
+      "loss": 0.4104,
+      "step": 10115
+    },
+    {
+      "epoch": 2.436206066441984,
+      "grad_norm": 1.25,
+      "learning_rate": 5.673040105837176e-05,
+      "loss": 0.3988,
+      "step": 10120
+    },
+    {
+      "epoch": 2.4374097255657197,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.668261695712239e-05,
+      "loss": 0.394,
+      "step": 10125
+    },
+    {
+      "epoch": 2.438613384689456,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.663492277470928e-05,
+      "loss": 0.3955,
+      "step": 10130
+    },
+    {
+      "epoch": 2.439817043813192,
+      "grad_norm": 1.2265625,
+      "learning_rate": 5.658731858733551e-05,
+      "loss": 0.3973,
+      "step": 10135
+    },
+    {
+      "epoch": 2.441020702936928,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.653980447106035e-05,
+      "loss": 0.3938,
+      "step": 10140
+    },
+    {
+      "epoch": 2.4422243620606645,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.649238050179922e-05,
+      "loss": 0.4094,
+      "step": 10145
+    },
+    {
+      "epoch": 2.4434280211844004,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.6445046755323414e-05,
+      "loss": 0.4017,
+      "step": 10150
+    },
+    {
+      "epoch": 2.4446316803081367,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.639780330726015e-05,
+      "loss": 0.3798,
+      "step": 10155
+    },
+    {
+      "epoch": 2.445835339431873,
+      "grad_norm": 1.265625,
+      "learning_rate": 5.635065023309236e-05,
+      "loss": 0.4175,
+      "step": 10160
+    },
+    {
+      "epoch": 2.447038998555609,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.6303587608158535e-05,
+      "loss": 0.3966,
+      "step": 10165
+    },
+    {
+      "epoch": 2.4482426576793452,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.625661550765272e-05,
+      "loss": 0.4021,
+      "step": 10170
+    },
+    {
+      "epoch": 2.4494463168030816,
+      "grad_norm": 1.2421875,
+      "learning_rate": 5.620973400662427e-05,
+      "loss": 0.4069,
+      "step": 10175
+    },
+    {
+      "epoch": 2.4506499759268174,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.616294317997779e-05,
+      "loss": 0.4031,
+      "step": 10180
+    },
+    {
+      "epoch": 2.4518536350505538,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.611624310247307e-05,
+      "loss": 0.3878,
+      "step": 10185
+    },
+    {
+      "epoch": 2.4530572941742896,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.606963384872483e-05,
+      "loss": 0.3885,
+      "step": 10190
+    },
+    {
+      "epoch": 2.454260953298026,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.6023115493202725e-05,
+      "loss": 0.4029,
+      "step": 10195
+    },
+    {
+      "epoch": 2.4554646124217623,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.5976688110231156e-05,
+      "loss": 0.3871,
+      "step": 10200
+    },
+    {
+      "epoch": 2.456668271545498,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.5930351773989164e-05,
+      "loss": 0.3943,
+      "step": 10205
+    },
+    {
+      "epoch": 2.4578719306692345,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.588410655851036e-05,
+      "loss": 0.4,
+      "step": 10210
+    },
+    {
+      "epoch": 2.459075589792971,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.5837952537682724e-05,
+      "loss": 0.4252,
+      "step": 10215
+    },
+    {
+      "epoch": 2.4602792489167067,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.579188978524856e-05,
+      "loss": 0.4152,
+      "step": 10220
+    },
+    {
+      "epoch": 2.461482908040443,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.574591837480434e-05,
+      "loss": 0.3914,
+      "step": 10225
+    },
+    {
+      "epoch": 2.4626865671641793,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.570003837980058e-05,
+      "loss": 0.3967,
+      "step": 10230
+    },
+    {
+      "epoch": 2.463890226287915,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.565424987354178e-05,
+      "loss": 0.3968,
+      "step": 10235
+    },
+    {
+      "epoch": 2.4650938854116515,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.56085529291862e-05,
+      "loss": 0.3802,
+      "step": 10240
+    },
+    {
+      "epoch": 2.4662975445353874,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.5562947619745864e-05,
+      "loss": 0.4232,
+      "step": 10245
+    },
+    {
+      "epoch": 2.4675012036591237,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.551743401808637e-05,
+      "loss": 0.3887,
+      "step": 10250
+    },
+    {
+      "epoch": 2.46870486278286,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.5472012196926744e-05,
+      "loss": 0.4019,
+      "step": 10255
+    },
+    {
+      "epoch": 2.469908521906596,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.5426682228839476e-05,
+      "loss": 0.4006,
+      "step": 10260
+    },
+    {
+      "epoch": 2.4711121810303323,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.538144418625017e-05,
+      "loss": 0.4005,
+      "step": 10265
+    },
+    {
+      "epoch": 2.472315840154068,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.533629814143768e-05,
+      "loss": 0.403,
+      "step": 10270
+    },
+    {
+      "epoch": 2.4735194992778045,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.529124416653376e-05,
+      "loss": 0.413,
+      "step": 10275
+    },
+    {
+      "epoch": 2.474723158401541,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.524628233352315e-05,
+      "loss": 0.3953,
+      "step": 10280
+    },
+    {
+      "epoch": 2.4759268175252767,
+      "grad_norm": 1.2421875,
+      "learning_rate": 5.5201412714243324e-05,
+      "loss": 0.3951,
+      "step": 10285
+    },
+    {
+      "epoch": 2.477130476649013,
+      "grad_norm": 1.265625,
+      "learning_rate": 5.5156635380384427e-05,
+      "loss": 0.3856,
+      "step": 10290
+    },
+    {
+      "epoch": 2.4783341357727493,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.511195040348916e-05,
+      "loss": 0.3876,
+      "step": 10295
+    },
+    {
+      "epoch": 2.479537794896485,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.506735785495266e-05,
+      "loss": 0.3904,
+      "step": 10300
+    },
+    {
+      "epoch": 2.4807414540202215,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.50228578060224e-05,
+      "loss": 0.4175,
+      "step": 10305
+    },
+    {
+      "epoch": 2.481945113143958,
+      "grad_norm": 1.125,
+      "learning_rate": 5.4978450327798036e-05,
+      "loss": 0.3739,
+      "step": 10310
+    },
+    {
+      "epoch": 2.4831487722676937,
+      "grad_norm": 1.2734375,
+      "learning_rate": 5.4934135491231325e-05,
+      "loss": 0.4026,
+      "step": 10315
+    },
+    {
+      "epoch": 2.48435243139143,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.488991336712604e-05,
+      "loss": 0.3891,
+      "step": 10320
+    },
+    {
+      "epoch": 2.485556090515166,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.4845784026137775e-05,
+      "loss": 0.4144,
+      "step": 10325
+    },
+    {
+      "epoch": 2.4867597496389022,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.4801747538773916e-05,
+      "loss": 0.3816,
+      "step": 10330
+    },
+    {
+      "epoch": 2.4879634087626386,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.47578039753935e-05,
+      "loss": 0.3974,
+      "step": 10335
+    },
+    {
+      "epoch": 2.4891670678863744,
+      "grad_norm": 1.3125,
+      "learning_rate": 5.471395340620705e-05,
+      "loss": 0.3987,
+      "step": 10340
+    },
+    {
+      "epoch": 2.4903707270101108,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.467019590127654e-05,
+      "loss": 0.435,
+      "step": 10345
+    },
+    {
+      "epoch": 2.491574386133847,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.462653153051525e-05,
+      "loss": 0.375,
+      "step": 10350
+    },
+    {
+      "epoch": 2.492778045257583,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.4582960363687656e-05,
+      "loss": 0.3907,
+      "step": 10355
+    },
+    {
+      "epoch": 2.4939817043813193,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.453948247040931e-05,
+      "loss": 0.4016,
+      "step": 10360
+    },
+    {
+      "epoch": 2.4951853635050556,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.4496097920146724e-05,
+      "loss": 0.4186,
+      "step": 10365
+    },
+    {
+      "epoch": 2.4963890226287915,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.445280678221731e-05,
+      "loss": 0.4148,
+      "step": 10370
+    },
+    {
+      "epoch": 2.497592681752528,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.4409609125789176e-05,
+      "loss": 0.4069,
+      "step": 10375
+    },
+    {
+      "epoch": 2.4987963408762637,
+      "grad_norm": 1.28125,
+      "learning_rate": 5.436650501988115e-05,
+      "loss": 0.39,
+      "step": 10380
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.4323494533362504e-05,
+      "loss": 0.4253,
+      "step": 10385
+    },
+    {
+      "epoch": 2.5012036591237363,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.428057773495297e-05,
+      "loss": 0.3715,
+      "step": 10390
+    },
+    {
+      "epoch": 2.502407318247472,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.4237754693222645e-05,
+      "loss": 0.44,
+      "step": 10395
+    },
+    {
+      "epoch": 2.5036109773712085,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.41950254765917e-05,
+      "loss": 0.3835,
+      "step": 10400
+    },
+    {
+      "epoch": 2.5048146364949444,
+      "grad_norm": 1.125,
+      "learning_rate": 5.415239015333052e-05,
+      "loss": 0.384,
+      "step": 10405
+    },
+    {
+      "epoch": 2.5060182956186807,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.41098487915594e-05,
+      "loss": 0.3911,
+      "step": 10410
+    },
+    {
+      "epoch": 2.507221954742417,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.406740145924852e-05,
+      "loss": 0.3934,
+      "step": 10415
+    },
+    {
+      "epoch": 2.5084256138661534,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.4025048224217864e-05,
+      "loss": 0.4006,
+      "step": 10420
+    },
+    {
+      "epoch": 2.5096292729898892,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.3982789154137016e-05,
+      "loss": 0.4099,
+      "step": 10425
+    },
+    {
+      "epoch": 2.5108329321136256,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.394062431652516e-05,
+      "loss": 0.4089,
+      "step": 10430
+    },
+    {
+      "epoch": 2.5120365912373614,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.389855377875087e-05,
+      "loss": 0.3837,
+      "step": 10435
+    },
+    {
+      "epoch": 2.5132402503610978,
+      "grad_norm": 1.1875,
+      "learning_rate": 5.3856577608032104e-05,
+      "loss": 0.38,
+      "step": 10440
+    },
+    {
+      "epoch": 2.514443909484834,
+      "grad_norm": 1.2578125,
+      "learning_rate": 5.381469587143603e-05,
+      "loss": 0.414,
+      "step": 10445
+    },
+    {
+      "epoch": 2.51564756860857,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.37729086358789e-05,
+      "loss": 0.3798,
+      "step": 10450
+    },
+    {
+      "epoch": 2.5168512277323063,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.373121596812603e-05,
+      "loss": 0.3993,
+      "step": 10455
+    },
+    {
+      "epoch": 2.518054886856042,
+      "grad_norm": 1.1875,
+      "learning_rate": 5.3689617934791586e-05,
+      "loss": 0.3905,
+      "step": 10460
+    },
+    {
+      "epoch": 2.5192585459797785,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.364811460233859e-05,
+      "loss": 0.396,
+      "step": 10465
+    },
+    {
+      "epoch": 2.520462205103515,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.3606706037078704e-05,
+      "loss": 0.3868,
+      "step": 10470
+    },
+    {
+      "epoch": 2.5216658642272507,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.356539230517222e-05,
+      "loss": 0.4062,
+      "step": 10475
+    },
+    {
+      "epoch": 2.522869523350987,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.352417347262789e-05,
+      "loss": 0.3789,
+      "step": 10480
+    },
+    {
+      "epoch": 2.524073182474723,
+      "grad_norm": 1.125,
+      "learning_rate": 5.3483049605302805e-05,
+      "loss": 0.386,
+      "step": 10485
+    },
+    {
+      "epoch": 2.525276841598459,
+      "grad_norm": 1.2578125,
+      "learning_rate": 5.34420207689024e-05,
+      "loss": 0.381,
+      "step": 10490
+    },
+    {
+      "epoch": 2.5264805007221955,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.340108702898024e-05,
+      "loss": 0.3819,
+      "step": 10495
+    },
+    {
+      "epoch": 2.527684159845932,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.336024845093793e-05,
+      "loss": 0.3858,
+      "step": 10500
+    },
+    {
+      "epoch": 2.527684159845932,
+      "eval_loss": 0.36653754115104675,
+      "eval_runtime": 2.3578,
+      "eval_samples_per_second": 84.826,
+      "eval_steps_per_second": 84.826,
+      "step": 10500
+    },
+    {
+      "epoch": 2.5288878189696677,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.331950510002507e-05,
+      "loss": 0.3956,
+      "step": 10505
+    },
+    {
+      "epoch": 2.530091478093404,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.327885704133906e-05,
+      "loss": 0.4002,
+      "step": 10510
+    },
+    {
+      "epoch": 2.53129513721714,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.3238304339825105e-05,
+      "loss": 0.3865,
+      "step": 10515
+    },
+    {
+      "epoch": 2.5324987963408763,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.319784706027602e-05,
+      "loss": 0.3823,
+      "step": 10520
+    },
+    {
+      "epoch": 2.5337024554646126,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.3157485267332136e-05,
+      "loss": 0.3943,
+      "step": 10525
+    },
+    {
+      "epoch": 2.5349061145883485,
+      "grad_norm": 1.125,
+      "learning_rate": 5.31172190254813e-05,
+      "loss": 0.3661,
+      "step": 10530
+    },
+    {
+      "epoch": 2.536109773712085,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.30770483990586e-05,
+      "loss": 0.3944,
+      "step": 10535
+    },
+    {
+      "epoch": 2.5373134328358207,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.3036973452246435e-05,
+      "loss": 0.3947,
+      "step": 10540
+    },
+    {
+      "epoch": 2.538517091959557,
+      "grad_norm": 1.125,
+      "learning_rate": 5.299699424907428e-05,
+      "loss": 0.3935,
+      "step": 10545
+    },
+    {
+      "epoch": 2.5397207510832933,
+      "grad_norm": 1.1875,
+      "learning_rate": 5.295711085341864e-05,
+      "loss": 0.3934,
+      "step": 10550
+    },
+    {
+      "epoch": 2.5409244102070296,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.2917323329003e-05,
+      "loss": 0.4008,
+      "step": 10555
+    },
+    {
+      "epoch": 2.5421280693307655,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.2877631739397565e-05,
+      "loss": 0.3874,
+      "step": 10560
+    },
+    {
+      "epoch": 2.543331728454502,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.2838036148019377e-05,
+      "loss": 0.3848,
+      "step": 10565
+    },
+    {
+      "epoch": 2.5445353875782377,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.279853661813198e-05,
+      "loss": 0.3887,
+      "step": 10570
+    },
+    {
+      "epoch": 2.545739046701974,
+      "grad_norm": 1.125,
+      "learning_rate": 5.2759133212845574e-05,
+      "loss": 0.3643,
+      "step": 10575
+    },
+    {
+      "epoch": 2.5469427058257104,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.271982599511664e-05,
+      "loss": 0.4055,
+      "step": 10580
+    },
+    {
+      "epoch": 2.5481463649494462,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.268061502774804e-05,
+      "loss": 0.4327,
+      "step": 10585
+    },
+    {
+      "epoch": 2.5493500240731826,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.2641500373388875e-05,
+      "loss": 0.3987,
+      "step": 10590
+    },
+    {
+      "epoch": 2.5505536831969184,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.2602482094534305e-05,
+      "loss": 0.4275,
+      "step": 10595
+    },
+    {
+      "epoch": 2.5517573423206548,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.2563560253525564e-05,
+      "loss": 0.3902,
+      "step": 10600
+    },
+    {
+      "epoch": 2.552961001444391,
+      "grad_norm": 1.328125,
+      "learning_rate": 5.2524734912549775e-05,
+      "loss": 0.4187,
+      "step": 10605
+    },
+    {
+      "epoch": 2.554164660568127,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.248600613363985e-05,
+      "loss": 0.3815,
+      "step": 10610
+    },
+    {
+      "epoch": 2.5553683196918633,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.244737397867449e-05,
+      "loss": 0.3901,
+      "step": 10615
+    },
+    {
+      "epoch": 2.556571978815599,
+      "grad_norm": 1.125,
+      "learning_rate": 5.2408838509377936e-05,
+      "loss": 0.3754,
+      "step": 10620
+    },
+    {
+      "epoch": 2.5577756379393355,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.237039978732003e-05,
+      "loss": 0.3847,
+      "step": 10625
+    },
+    {
+      "epoch": 2.558979297063072,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.233205787391596e-05,
+      "loss": 0.4114,
+      "step": 10630
+    },
+    {
+      "epoch": 2.560182956186808,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.229381283042629e-05,
+      "loss": 0.4031,
+      "step": 10635
+    },
+    {
+      "epoch": 2.561386615310544,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.2255664717956805e-05,
+      "loss": 0.3975,
+      "step": 10640
+    },
+    {
+      "epoch": 2.5625902744342803,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.2217613597458384e-05,
+      "loss": 0.4185,
+      "step": 10645
+    },
+    {
+      "epoch": 2.563793933558016,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.217965952972697e-05,
+      "loss": 0.3871,
+      "step": 10650
+    },
+    {
+      "epoch": 2.5649975926817525,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.214180257540346e-05,
+      "loss": 0.3932,
+      "step": 10655
+    },
+    {
+      "epoch": 2.566201251805489,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.210404279497353e-05,
+      "loss": 0.3969,
+      "step": 10660
+    },
+    {
+      "epoch": 2.5674049109292247,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.206638024876766e-05,
+      "loss": 0.4153,
+      "step": 10665
+    },
+    {
+      "epoch": 2.568608570052961,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.202881499696091e-05,
+      "loss": 0.3996,
+      "step": 10670
+    },
+    {
+      "epoch": 2.569812229176697,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.199134709957295e-05,
+      "loss": 0.3858,
+      "step": 10675
+    },
+    {
+      "epoch": 2.5710158883004333,
+      "grad_norm": 1.125,
+      "learning_rate": 5.195397661646787e-05,
+      "loss": 0.3619,
+      "step": 10680
+    },
+    {
+      "epoch": 2.5722195474241696,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.191670360735409e-05,
+      "loss": 0.3843,
+      "step": 10685
+    },
+    {
+      "epoch": 2.573423206547906,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.187952813178438e-05,
+      "loss": 0.4048,
+      "step": 10690
+    },
+    {
+      "epoch": 2.574626865671642,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.184245024915557e-05,
+      "loss": 0.4132,
+      "step": 10695
+    },
+    {
+      "epoch": 2.575830524795378,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.1805470018708635e-05,
+      "loss": 0.3826,
+      "step": 10700
+    },
+    {
+      "epoch": 2.577034183919114,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.176858749952851e-05,
+      "loss": 0.3977,
+      "step": 10705
+    },
+    {
+      "epoch": 2.5782378430428503,
+      "grad_norm": 1.2578125,
+      "learning_rate": 5.1731802750543986e-05,
+      "loss": 0.4028,
+      "step": 10710
+    },
+    {
+      "epoch": 2.5794415021665866,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.169511583052767e-05,
+      "loss": 0.4002,
+      "step": 10715
+    },
+    {
+      "epoch": 2.5806451612903225,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.165852679809585e-05,
+      "loss": 0.4226,
+      "step": 10720
+    },
+    {
+      "epoch": 2.581848820414059,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.1622035711708453e-05,
+      "loss": 0.3834,
+      "step": 10725
+    },
+    {
+      "epoch": 2.5830524795377947,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.158564262966883e-05,
+      "loss": 0.4073,
+      "step": 10730
+    },
+    {
+      "epoch": 2.584256138661531,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.154934761012382e-05,
+      "loss": 0.4149,
+      "step": 10735
+    },
+    {
+      "epoch": 2.5854597977852674,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.151315071106359e-05,
+      "loss": 0.3749,
+      "step": 10740
+    },
+    {
+      "epoch": 2.5866634569090032,
+      "grad_norm": 1.1875,
+      "learning_rate": 5.147705199032145e-05,
+      "loss": 0.4127,
+      "step": 10745
+    },
+    {
+      "epoch": 2.5878671160327396,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.144105150557392e-05,
+      "loss": 0.3747,
+      "step": 10750
+    },
+    {
+      "epoch": 2.5890707751564754,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.1405149314340575e-05,
+      "loss": 0.399,
+      "step": 10755
+    },
+    {
+      "epoch": 2.5902744342802118,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5.136934547398388e-05,
+      "loss": 0.3917,
+      "step": 10760
+    },
+    {
+      "epoch": 2.591478093403948,
+      "grad_norm": 1.2265625,
+      "learning_rate": 5.133364004170922e-05,
+      "loss": 0.3636,
+      "step": 10765
+    },
+    {
+      "epoch": 2.5926817525276844,
+      "grad_norm": 1.25,
+      "learning_rate": 5.129803307456468e-05,
+      "loss": 0.3905,
+      "step": 10770
+    },
+    {
+      "epoch": 2.5938854116514203,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.126252462944111e-05,
+      "loss": 0.3639,
+      "step": 10775
+    },
+    {
+      "epoch": 2.5950890707751566,
+      "grad_norm": 1.2890625,
+      "learning_rate": 5.12271147630719e-05,
+      "loss": 0.425,
+      "step": 10780
+    },
+    {
+      "epoch": 2.5962927298988925,
+      "grad_norm": 1.125,
+      "learning_rate": 5.1191803532032915e-05,
+      "loss": 0.3718,
+      "step": 10785
+    },
+    {
+      "epoch": 2.597496389022629,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.1156590992742465e-05,
+      "loss": 0.4002,
+      "step": 10790
+    },
+    {
+      "epoch": 2.598700048146365,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.112147720146115e-05,
+      "loss": 0.3949,
+      "step": 10795
+    },
+    {
+      "epoch": 2.599903707270101,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.1086462214291816e-05,
+      "loss": 0.3942,
+      "step": 10800
+    },
+    {
+      "epoch": 2.6011073663938373,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.1051546087179426e-05,
+      "loss": 0.3761,
+      "step": 10805
+    },
+    {
+      "epoch": 2.602311025517573,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.101672887591101e-05,
+      "loss": 0.409,
+      "step": 10810
+    },
+    {
+      "epoch": 2.6035146846413095,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.0982010636115545e-05,
+      "loss": 0.3918,
+      "step": 10815
+    },
+    {
+      "epoch": 2.604718343765046,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.094739142326389e-05,
+      "loss": 0.3767,
+      "step": 10820
+    },
+    {
+      "epoch": 2.605922002888782,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.0912871292668646e-05,
+      "loss": 0.4152,
+      "step": 10825
+    },
+    {
+      "epoch": 2.607125662012518,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.087845029948413e-05,
+      "loss": 0.4117,
+      "step": 10830
+    },
+    {
+      "epoch": 2.6083293211362544,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.0844128498706314e-05,
+      "loss": 0.4213,
+      "step": 10835
+    },
+    {
+      "epoch": 2.6095329802599903,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.08099059451726e-05,
+      "loss": 0.3752,
+      "step": 10840
+    },
+    {
+      "epoch": 2.6107366393837266,
+      "grad_norm": 1.1875,
+      "learning_rate": 5.077578269356184e-05,
+      "loss": 0.408,
+      "step": 10845
+    },
+    {
+      "epoch": 2.611940298507463,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.0741758798394284e-05,
+      "loss": 0.4146,
+      "step": 10850
+    },
+    {
+      "epoch": 2.6131439576311988,
+      "grad_norm": 1.125,
+      "learning_rate": 5.070783431403136e-05,
+      "loss": 0.399,
+      "step": 10855
+    },
+    {
+      "epoch": 2.614347616754935,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.06740092946757e-05,
+      "loss": 0.402,
+      "step": 10860
+    },
+    {
+      "epoch": 2.615551275878671,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.064028379437105e-05,
+      "loss": 0.4006,
+      "step": 10865
+    },
+    {
+      "epoch": 2.6167549350024073,
+      "grad_norm": 1.125,
+      "learning_rate": 5.060665786700206e-05,
+      "loss": 0.3777,
+      "step": 10870
+    },
+    {
+      "epoch": 2.6179585941261436,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.057313156629439e-05,
+      "loss": 0.4108,
+      "step": 10875
+    },
+    {
+      "epoch": 2.6191622532498795,
+      "grad_norm": 1.2421875,
+      "learning_rate": 5.053970494581444e-05,
+      "loss": 0.4123,
+      "step": 10880
+    },
+    {
+      "epoch": 2.620365912373616,
+      "grad_norm": 1.2421875,
+      "learning_rate": 5.05063780589694e-05,
+      "loss": 0.4031,
+      "step": 10885
+    },
+    {
+      "epoch": 2.6215695714973517,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.04731509590071e-05,
+      "loss": 0.383,
+      "step": 10890
+    },
+    {
+      "epoch": 2.622773230621088,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.0440023699015906e-05,
+      "loss": 0.3954,
+      "step": 10895
+    },
+    {
+      "epoch": 2.6239768897448243,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.040699633192469e-05,
+      "loss": 0.3613,
+      "step": 10900
+    },
+    {
+      "epoch": 2.6251805488685607,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.037406891050272e-05,
+      "loss": 0.3987,
+      "step": 10905
+    },
+    {
+      "epoch": 2.6263842079922965,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.0341241487359573e-05,
+      "loss": 0.4063,
+      "step": 10910
+    },
+    {
+      "epoch": 2.627587867116033,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.0308514114945074e-05,
+      "loss": 0.383,
+      "step": 10915
+    },
+    {
+      "epoch": 2.6287915262397687,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.0275886845549155e-05,
+      "loss": 0.3873,
+      "step": 10920
+    },
+    {
+      "epoch": 2.629995185363505,
+      "grad_norm": 1.1875,
+      "learning_rate": 5.0243359731301835e-05,
+      "loss": 0.3911,
+      "step": 10925
+    },
+    {
+      "epoch": 2.6311988444872414,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.0210932824173105e-05,
+      "loss": 0.3908,
+      "step": 10930
+    },
+    {
+      "epoch": 2.6324025036109773,
+      "grad_norm": 1.1875,
+      "learning_rate": 5.0178606175972834e-05,
+      "loss": 0.4026,
+      "step": 10935
+    },
+    {
+      "epoch": 2.6336061627347136,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.0146379838350745e-05,
+      "loss": 0.4183,
+      "step": 10940
+    },
+    {
+      "epoch": 2.6348098218584495,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.011425386279626e-05,
+      "loss": 0.398,
+      "step": 10945
+    },
+    {
+      "epoch": 2.636013480982186,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.0082228300638444e-05,
+      "loss": 0.4177,
+      "step": 10950
+    },
+    {
+      "epoch": 2.637217140105922,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.005030320304591e-05,
+      "loss": 0.3913,
+      "step": 10955
+    },
+    {
+      "epoch": 2.638420799229658,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.001847862102683e-05,
+      "loss": 0.3883,
+      "step": 10960
+    },
+    {
+      "epoch": 2.6396244583533943,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.9986754605428706e-05,
+      "loss": 0.4112,
+      "step": 10965
+    },
+    {
+      "epoch": 2.64082811747713,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.9955131206938366e-05,
+      "loss": 0.4387,
+      "step": 10970
+    },
+    {
+      "epoch": 2.6420317766008665,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.9923608476081915e-05,
+      "loss": 0.3736,
+      "step": 10975
+    },
+    {
+      "epoch": 2.643235435724603,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.9892186463224584e-05,
+      "loss": 0.3938,
+      "step": 10980
+    },
+    {
+      "epoch": 2.644439094848339,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.98608652185707e-05,
+      "loss": 0.3771,
+      "step": 10985
+    },
+    {
+      "epoch": 2.645642753972075,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.982964479216359e-05,
+      "loss": 0.4309,
+      "step": 10990
+    },
+    {
+      "epoch": 2.6468464130958114,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.9798525233885485e-05,
+      "loss": 0.4135,
+      "step": 10995
+    },
+    {
+      "epoch": 2.6480500722195472,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.9767506593457463e-05,
+      "loss": 0.4049,
+      "step": 11000
+    },
+    {
+      "epoch": 2.6480500722195472,
+      "eval_loss": 0.3635382056236267,
+      "eval_runtime": 2.3559,
+      "eval_samples_per_second": 84.894,
+      "eval_steps_per_second": 84.894,
+      "step": 11000
+    },
+    {
+      "epoch": 2.6492537313432836,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.973658892043935e-05,
+      "loss": 0.3901,
+      "step": 11005
+    },
+    {
+      "epoch": 2.65045739046702,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.9705772264229656e-05,
+      "loss": 0.3768,
+      "step": 11010
+    },
+    {
+      "epoch": 2.6516610495907558,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.9675056674065534e-05,
+      "loss": 0.4015,
+      "step": 11015
+    },
+    {
+      "epoch": 2.652864708714492,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.964444219902259e-05,
+      "loss": 0.4087,
+      "step": 11020
+    },
+    {
+      "epoch": 2.654068367838228,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.9613928888014915e-05,
+      "loss": 0.4151,
+      "step": 11025
+    },
+    {
+      "epoch": 2.6552720269619643,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.958351678979496e-05,
+      "loss": 0.3746,
+      "step": 11030
+    },
+    {
+      "epoch": 2.6564756860857006,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.955320595295347e-05,
+      "loss": 0.3983,
+      "step": 11035
+    },
+    {
+      "epoch": 2.657679345209437,
+      "grad_norm": 1.2578125,
+      "learning_rate": 4.952299642591937e-05,
+      "loss": 0.3974,
+      "step": 11040
+    },
+    {
+      "epoch": 2.658883004333173,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.949288825695975e-05,
+      "loss": 0.3775,
+      "step": 11045
+    },
+    {
+      "epoch": 2.660086663456909,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.946288149417977e-05,
+      "loss": 0.3934,
+      "step": 11050
+    },
+    {
+      "epoch": 2.661290322580645,
+      "grad_norm": 1.125,
+      "learning_rate": 4.943297618552253e-05,
+      "loss": 0.3777,
+      "step": 11055
+    },
+    {
+      "epoch": 2.6624939817043813,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.9403172378769016e-05,
+      "loss": 0.4239,
+      "step": 11060
+    },
+    {
+      "epoch": 2.6636976408281177,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.9373470121538103e-05,
+      "loss": 0.3757,
+      "step": 11065
+    },
+    {
+      "epoch": 2.6649012999518535,
+      "grad_norm": 1.2734375,
+      "learning_rate": 4.934386946128637e-05,
+      "loss": 0.4289,
+      "step": 11070
+    },
+    {
+      "epoch": 2.66610495907559,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.9314370445308105e-05,
+      "loss": 0.3909,
+      "step": 11075
+    },
+    {
+      "epoch": 2.6673086181993257,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.928497312073515e-05,
+      "loss": 0.4065,
+      "step": 11080
+    },
+    {
+      "epoch": 2.668512277323062,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.9255677534536906e-05,
+      "loss": 0.4246,
+      "step": 11085
+    },
+    {
+      "epoch": 2.6697159364467984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.92264837335202e-05,
+      "loss": 0.4447,
+      "step": 11090
+    },
+    {
+      "epoch": 2.6709195955705343,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.9197391764329246e-05,
+      "loss": 0.3842,
+      "step": 11095
+    },
+    {
+      "epoch": 2.6721232546942706,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.916840167344557e-05,
+      "loss": 0.3733,
+      "step": 11100
+    },
+    {
+      "epoch": 2.6733269138180065,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.9139513507187894e-05,
+      "loss": 0.384,
+      "step": 11105
+    },
+    {
+      "epoch": 2.674530572941743,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.9110727311712086e-05,
+      "loss": 0.3897,
+      "step": 11110
+    },
+    {
+      "epoch": 2.675734232065479,
+      "grad_norm": 1.2421875,
+      "learning_rate": 4.9082043133011115e-05,
+      "loss": 0.4143,
+      "step": 11115
+    },
+    {
+      "epoch": 2.6769378911892154,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.9053461016914944e-05,
+      "loss": 0.4066,
+      "step": 11120
+    },
+    {
+      "epoch": 2.6781415503129513,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.9024981009090494e-05,
+      "loss": 0.3885,
+      "step": 11125
+    },
+    {
+      "epoch": 2.6793452094366876,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.899660315504149e-05,
+      "loss": 0.3929,
+      "step": 11130
+    },
+    {
+      "epoch": 2.6805488685604235,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.8968327500108476e-05,
+      "loss": 0.364,
+      "step": 11135
+    },
+    {
+      "epoch": 2.68175252768416,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.894015408946869e-05,
+      "loss": 0.3784,
+      "step": 11140
+    },
+    {
+      "epoch": 2.682956186807896,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.891208296813603e-05,
+      "loss": 0.3999,
+      "step": 11145
+    },
+    {
+      "epoch": 2.684159845931632,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.8884114180960946e-05,
+      "loss": 0.3888,
+      "step": 11150
+    },
+    {
+      "epoch": 2.6853635050553684,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.885624777263042e-05,
+      "loss": 0.3868,
+      "step": 11155
+    },
+    {
+      "epoch": 2.6865671641791042,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.882848378766777e-05,
+      "loss": 0.3863,
+      "step": 11160
+    },
+    {
+      "epoch": 2.6877708233028406,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.880082227043278e-05,
+      "loss": 0.3974,
+      "step": 11165
+    },
+    {
+      "epoch": 2.688974482426577,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.8773263265121436e-05,
+      "loss": 0.41,
+      "step": 11170
+    },
+    {
+      "epoch": 2.690178141550313,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.874580681576597e-05,
+      "loss": 0.3997,
+      "step": 11175
+    },
+    {
+      "epoch": 2.691381800674049,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.871845296623475e-05,
+      "loss": 0.3755,
+      "step": 11180
+    },
+    {
+      "epoch": 2.6925854597977854,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.869120176023221e-05,
+      "loss": 0.3829,
+      "step": 11185
+    },
+    {
+      "epoch": 2.6937891189215213,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.866405324129881e-05,
+      "loss": 0.3873,
+      "step": 11190
+    },
+    {
+      "epoch": 2.6949927780452576,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.863700745281089e-05,
+      "loss": 0.4194,
+      "step": 11195
+    },
+    {
+      "epoch": 2.696196437168994,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.861006443798074e-05,
+      "loss": 0.3864,
+      "step": 11200
+    },
+    {
+      "epoch": 2.69740009629273,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.858322423985633e-05,
+      "loss": 0.3747,
+      "step": 11205
+    },
+    {
+      "epoch": 2.698603755416466,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.8556486901321454e-05,
+      "loss": 0.3988,
+      "step": 11210
+    },
+    {
+      "epoch": 2.699807414540202,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.852985246509552e-05,
+      "loss": 0.3403,
+      "step": 11215
+    },
+    {
+      "epoch": 2.7010110736639383,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.850332097373351e-05,
+      "loss": 0.3799,
+      "step": 11220
+    },
+    {
+      "epoch": 2.7022147327876747,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.8476892469625974e-05,
+      "loss": 0.358,
+      "step": 11225
+    },
+    {
+      "epoch": 2.7034183919114105,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.845056699499886e-05,
+      "loss": 0.3873,
+      "step": 11230
+    },
+    {
+      "epoch": 2.704622051035147,
+      "grad_norm": 1.125,
+      "learning_rate": 4.842434459191356e-05,
+      "loss": 0.3851,
+      "step": 11235
+    },
+    {
+      "epoch": 2.7058257101588827,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.839822530226671e-05,
+      "loss": 0.411,
+      "step": 11240
+    },
+    {
+      "epoch": 2.707029369282619,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.8372209167790256e-05,
+      "loss": 0.388,
+      "step": 11245
+    },
+    {
+      "epoch": 2.7082330284063554,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.834629623005135e-05,
+      "loss": 0.4201,
+      "step": 11250
+    },
+    {
+      "epoch": 2.7094366875300917,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.8320486530452155e-05,
+      "loss": 0.3806,
+      "step": 11255
+    },
+    {
+      "epoch": 2.7106403466538276,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.8294780110230024e-05,
+      "loss": 0.3789,
+      "step": 11260
+    },
+    {
+      "epoch": 2.711844005777564,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.826917701045717e-05,
+      "loss": 0.38,
+      "step": 11265
+    },
+    {
+      "epoch": 2.7130476649013,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.8243677272040824e-05,
+      "loss": 0.3812,
+      "step": 11270
+    },
+    {
+      "epoch": 2.714251324025036,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.8218280935723e-05,
+      "loss": 0.4132,
+      "step": 11275
+    },
+    {
+      "epoch": 2.7154549831487724,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.8192988042080534e-05,
+      "loss": 0.3824,
+      "step": 11280
+    },
+    {
+      "epoch": 2.7166586422725083,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.8167798631525015e-05,
+      "loss": 0.383,
+      "step": 11285
+    },
+    {
+      "epoch": 2.7178623013962446,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.814271274430264e-05,
+      "loss": 0.3768,
+      "step": 11290
+    },
+    {
+      "epoch": 2.7190659605199805,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.811773042049425e-05,
+      "loss": 0.4089,
+      "step": 11295
+    },
+    {
+      "epoch": 2.720269619643717,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.809285170001515e-05,
+      "loss": 0.3842,
+      "step": 11300
+    },
+    {
+      "epoch": 2.721473278767453,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.80680766226152e-05,
+      "loss": 0.4126,
+      "step": 11305
+    },
+    {
+      "epoch": 2.7226769378911895,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.8043405227878604e-05,
+      "loss": 0.3973,
+      "step": 11310
+    },
+    {
+      "epoch": 2.7238805970149254,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.8018837555223934e-05,
+      "loss": 0.3732,
+      "step": 11315
+    },
+    {
+      "epoch": 2.7250842561386617,
+      "grad_norm": 1.3125,
+      "learning_rate": 4.7994373643904027e-05,
+      "loss": 0.3947,
+      "step": 11320
+    },
+    {
+      "epoch": 2.7262879152623976,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.797001353300594e-05,
+      "loss": 0.3856,
+      "step": 11325
+    },
+    {
+      "epoch": 2.727491574386134,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.794575726145088e-05,
+      "loss": 0.3788,
+      "step": 11330
+    },
+    {
+      "epoch": 2.72869523350987,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.7921604867994177e-05,
+      "loss": 0.4013,
+      "step": 11335
+    },
+    {
+      "epoch": 2.729898892633606,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.789755639122513e-05,
+      "loss": 0.3841,
+      "step": 11340
+    },
+    {
+      "epoch": 2.7311025517573424,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.787361186956707e-05,
+      "loss": 0.3777,
+      "step": 11345
+    },
+    {
+      "epoch": 2.7323062108810783,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.784977134127719e-05,
+      "loss": 0.3689,
+      "step": 11350
+    },
+    {
+      "epoch": 2.7335098700048146,
+      "grad_norm": 1.125,
+      "learning_rate": 4.782603484444653e-05,
+      "loss": 0.3712,
+      "step": 11355
+    },
+    {
+      "epoch": 2.734713529128551,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.780240241699996e-05,
+      "loss": 0.3911,
+      "step": 11360
+    },
+    {
+      "epoch": 2.735917188252287,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.777887409669601e-05,
+      "loss": 0.4111,
+      "step": 11365
+    },
+    {
+      "epoch": 2.737120847376023,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.775544992112692e-05,
+      "loss": 0.3969,
+      "step": 11370
+    },
+    {
+      "epoch": 2.738324506499759,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.773212992771851e-05,
+      "loss": 0.4193,
+      "step": 11375
+    },
+    {
+      "epoch": 2.7395281656234953,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.7708914153730156e-05,
+      "loss": 0.3762,
+      "step": 11380
+    },
+    {
+      "epoch": 2.7407318247472316,
+      "grad_norm": 1.125,
+      "learning_rate": 4.768580263625472e-05,
+      "loss": 0.3743,
+      "step": 11385
+    },
+    {
+      "epoch": 2.741935483870968,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.7662795412218464e-05,
+      "loss": 0.4056,
+      "step": 11390
+    },
+    {
+      "epoch": 2.743139142994704,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.7639892518381056e-05,
+      "loss": 0.3928,
+      "step": 11395
+    },
+    {
+      "epoch": 2.74434280211844,
+      "grad_norm": 1.125,
+      "learning_rate": 4.761709399133542e-05,
+      "loss": 0.3929,
+      "step": 11400
+    },
+    {
+      "epoch": 2.745546461242176,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.759439986750778e-05,
+      "loss": 0.4081,
+      "step": 11405
+    },
+    {
+      "epoch": 2.7467501203659124,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.757181018315753e-05,
+      "loss": 0.4217,
+      "step": 11410
+    },
+    {
+      "epoch": 2.7479537794896487,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.754932497437718e-05,
+      "loss": 0.3866,
+      "step": 11415
+    },
+    {
+      "epoch": 2.7491574386133846,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.7526944277092356e-05,
+      "loss": 0.4032,
+      "step": 11420
+    },
+    {
+      "epoch": 2.750361097737121,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.750466812706165e-05,
+      "loss": 0.3796,
+      "step": 11425
+    },
+    {
+      "epoch": 2.7515647568608568,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7482496559876636e-05,
+      "loss": 0.4032,
+      "step": 11430
+    },
+    {
+      "epoch": 2.752768415984593,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.746042961096181e-05,
+      "loss": 0.4044,
+      "step": 11435
+    },
+    {
+      "epoch": 2.7539720751083294,
+      "grad_norm": 1.125,
+      "learning_rate": 4.743846731557449e-05,
+      "loss": 0.3798,
+      "step": 11440
+    },
+    {
+      "epoch": 2.7551757342320657,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.7416609708804815e-05,
+      "loss": 0.393,
+      "step": 11445
+    },
+    {
+      "epoch": 2.7563793933558016,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.7394856825575606e-05,
+      "loss": 0.4015,
+      "step": 11450
+    },
+    {
+      "epoch": 2.757583052479538,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.7373208700642395e-05,
+      "loss": 0.3922,
+      "step": 11455
+    },
+    {
+      "epoch": 2.758786711603274,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.7351665368593374e-05,
+      "loss": 0.3676,
+      "step": 11460
+    },
+    {
+      "epoch": 2.75999037072701,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.733022686384922e-05,
+      "loss": 0.3733,
+      "step": 11465
+    },
+    {
+      "epoch": 2.7611940298507465,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.730889322066321e-05,
+      "loss": 0.3956,
+      "step": 11470
+    },
+    {
+      "epoch": 2.7623976889744823,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.7287664473121e-05,
+      "loss": 0.3956,
+      "step": 11475
+    },
+    {
+      "epoch": 2.7636013480982187,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.726654065514071e-05,
+      "loss": 0.3852,
+      "step": 11480
+    },
+    {
+      "epoch": 2.7648050072219545,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.7245521800472784e-05,
+      "loss": 0.3916,
+      "step": 11485
+    },
+    {
+      "epoch": 2.766008666345691,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.722460794269993e-05,
+      "loss": 0.4224,
+      "step": 11490
+    },
+    {
+      "epoch": 2.767212325469427,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.720379911523719e-05,
+      "loss": 0.376,
+      "step": 11495
+    },
+    {
+      "epoch": 2.768415984593163,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.718309535133169e-05,
+      "loss": 0.3794,
+      "step": 11500
+    },
+    {
+      "epoch": 2.768415984593163,
+      "eval_loss": 0.3630373477935791,
+      "eval_runtime": 2.3435,
+      "eval_samples_per_second": 85.344,
+      "eval_steps_per_second": 85.344,
+      "step": 11500
+    },
+    {
+      "epoch": 2.7696196437168994,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.716249668406274e-05,
+      "loss": 0.3653,
+      "step": 11505
+    },
+    {
+      "epoch": 2.7708233028406353,
+      "grad_norm": 1.125,
+      "learning_rate": 4.714200314634176e-05,
+      "loss": 0.3893,
+      "step": 11510
+    },
+    {
+      "epoch": 2.7720269619643716,
+      "grad_norm": 1.28125,
+      "learning_rate": 4.7121614770912134e-05,
+      "loss": 0.3769,
+      "step": 11515
+    },
+    {
+      "epoch": 2.773230621088108,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.7101331590349296e-05,
+      "loss": 0.3879,
+      "step": 11520
+    },
+    {
+      "epoch": 2.7744342802118442,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.708115363706054e-05,
+      "loss": 0.3831,
+      "step": 11525
+    },
+    {
+      "epoch": 2.77563793933558,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.706108094328506e-05,
+      "loss": 0.3808,
+      "step": 11530
+    },
+    {
+      "epoch": 2.7768415984593164,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.704111354109391e-05,
+      "loss": 0.3977,
+      "step": 11535
+    },
+    {
+      "epoch": 2.7780452575830523,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.702125146238985e-05,
+      "loss": 0.4004,
+      "step": 11540
+    },
+    {
+      "epoch": 2.7792489167067886,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.700149473890739e-05,
+      "loss": 0.3839,
+      "step": 11545
+    },
+    {
+      "epoch": 2.780452575830525,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.6981843402212716e-05,
+      "loss": 0.407,
+      "step": 11550
+    },
+    {
+      "epoch": 2.781656234954261,
+      "grad_norm": 1.28125,
+      "learning_rate": 4.6962297483703634e-05,
+      "loss": 0.3937,
+      "step": 11555
+    },
+    {
+      "epoch": 2.782859894077997,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.69428570146095e-05,
+      "loss": 0.4046,
+      "step": 11560
+    },
+    {
+      "epoch": 2.784063553201733,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.69235220259912e-05,
+      "loss": 0.3875,
+      "step": 11565
+    },
+    {
+      "epoch": 2.7852672123254694,
+      "grad_norm": 1.125,
+      "learning_rate": 4.690429254874111e-05,
+      "loss": 0.3818,
+      "step": 11570
+    },
+    {
+      "epoch": 2.7864708714492057,
+      "grad_norm": 1.2734375,
+      "learning_rate": 4.6885168613582965e-05,
+      "loss": 0.3988,
+      "step": 11575
+    },
+    {
+      "epoch": 2.787674530572942,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.6866150251071955e-05,
+      "loss": 0.3745,
+      "step": 11580
+    },
+    {
+      "epoch": 2.788878189696678,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.684723749159452e-05,
+      "loss": 0.4021,
+      "step": 11585
+    },
+    {
+      "epoch": 2.790081848820414,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.68284303653684e-05,
+      "loss": 0.3919,
+      "step": 11590
+    },
+    {
+      "epoch": 2.79128550794415,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.680972890244258e-05,
+      "loss": 0.4028,
+      "step": 11595
+    },
+    {
+      "epoch": 2.7924891670678864,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.679113313269719e-05,
+      "loss": 0.3992,
+      "step": 11600
+    },
+    {
+      "epoch": 2.7936928261916227,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.677264308584351e-05,
+      "loss": 0.3931,
+      "step": 11605
+    },
+    {
+      "epoch": 2.7948964853153586,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.675425879142389e-05,
+      "loss": 0.3863,
+      "step": 11610
+    },
+    {
+      "epoch": 2.796100144439095,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.6735980278811716e-05,
+      "loss": 0.3739,
+      "step": 11615
+    },
+    {
+      "epoch": 2.797303803562831,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.671780757721139e-05,
+      "loss": 0.4002,
+      "step": 11620
+    },
+    {
+      "epoch": 2.798507462686567,
+      "grad_norm": 1.25,
+      "learning_rate": 4.6699740715658183e-05,
+      "loss": 0.4051,
+      "step": 11625
+    },
+    {
+      "epoch": 2.7997111218103035,
+      "grad_norm": 1.234375,
+      "learning_rate": 4.668177972301835e-05,
+      "loss": 0.3831,
+      "step": 11630
+    },
+    {
+      "epoch": 2.8009147809340393,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.6663924627988944e-05,
+      "loss": 0.379,
+      "step": 11635
+    },
+    {
+      "epoch": 2.8021184400577757,
+      "grad_norm": 1.2421875,
+      "learning_rate": 4.664617545909782e-05,
+      "loss": 0.3973,
+      "step": 11640
+    },
+    {
+      "epoch": 2.8033220991815115,
+      "grad_norm": 1.125,
+      "learning_rate": 4.6628532244703614e-05,
+      "loss": 0.3647,
+      "step": 11645
+    },
+    {
+      "epoch": 2.804525758305248,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.661099501299563e-05,
+      "loss": 0.3971,
+      "step": 11650
+    },
+    {
+      "epoch": 2.805729417428984,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.6593563791993904e-05,
+      "loss": 0.395,
+      "step": 11655
+    },
+    {
+      "epoch": 2.8069330765527205,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.657623860954904e-05,
+      "loss": 0.3814,
+      "step": 11660
+    },
+    {
+      "epoch": 2.8081367356764564,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.655901949334225e-05,
+      "loss": 0.3839,
+      "step": 11665
+    },
+    {
+      "epoch": 2.8093403948001927,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.6541906470885245e-05,
+      "loss": 0.4111,
+      "step": 11670
+    },
+    {
+      "epoch": 2.8105440539239286,
+      "grad_norm": 1.125,
+      "learning_rate": 4.652489956952027e-05,
+      "loss": 0.4018,
+      "step": 11675
+    },
+    {
+      "epoch": 2.811747713047665,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.650799881641999e-05,
+      "loss": 0.3954,
+      "step": 11680
+    },
+    {
+      "epoch": 2.8129513721714012,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.6491204238587454e-05,
+      "loss": 0.4026,
+      "step": 11685
+    },
+    {
+      "epoch": 2.814155031295137,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.6474515862856124e-05,
+      "loss": 0.4223,
+      "step": 11690
+    },
+    {
+      "epoch": 2.8153586904188734,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.645793371588975e-05,
+      "loss": 0.3726,
+      "step": 11695
+    },
+    {
+      "epoch": 2.8165623495426093,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.6441457824182304e-05,
+      "loss": 0.4033,
+      "step": 11700
+    },
+    {
+      "epoch": 2.8177660086663456,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.6425088214058085e-05,
+      "loss": 0.3833,
+      "step": 11705
+    },
+    {
+      "epoch": 2.818969667790082,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.64088249116715e-05,
+      "loss": 0.4062,
+      "step": 11710
+    },
+    {
+      "epoch": 2.8201733269138183,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.639266794300718e-05,
+      "loss": 0.3728,
+      "step": 11715
+    },
+    {
+      "epoch": 2.821376986037554,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.6376617333879767e-05,
+      "loss": 0.3784,
+      "step": 11720
+    },
+    {
+      "epoch": 2.8225806451612905,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.6360673109934046e-05,
+      "loss": 0.3902,
+      "step": 11725
+    },
+    {
+      "epoch": 2.8237843042850264,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.6344835296644816e-05,
+      "loss": 0.3874,
+      "step": 11730
+    },
+    {
+      "epoch": 2.8249879634087627,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.632910391931681e-05,
+      "loss": 0.411,
+      "step": 11735
+    },
+    {
+      "epoch": 2.826191622532499,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.6313479003084764e-05,
+      "loss": 0.3941,
+      "step": 11740
+    },
+    {
+      "epoch": 2.827395281656235,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.629796057291328e-05,
+      "loss": 0.3811,
+      "step": 11745
+    },
+    {
+      "epoch": 2.828598940779971,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.62825486535968e-05,
+      "loss": 0.3855,
+      "step": 11750
+    },
+    {
+      "epoch": 2.829802599903707,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.626724326975967e-05,
+      "loss": 0.3715,
+      "step": 11755
+    },
+    {
+      "epoch": 2.8310062590274434,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.625204444585594e-05,
+      "loss": 0.3914,
+      "step": 11760
+    },
+    {
+      "epoch": 2.8322099181511797,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.6236952206169446e-05,
+      "loss": 0.404,
+      "step": 11765
+    },
+    {
+      "epoch": 2.8334135772749156,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.622196657481371e-05,
+      "loss": 0.4047,
+      "step": 11770
+    },
+    {
+      "epoch": 2.834617236398652,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.620708757573193e-05,
+      "loss": 0.388,
+      "step": 11775
+    },
+    {
+      "epoch": 2.835820895522388,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.619231523269695e-05,
+      "loss": 0.3735,
+      "step": 11780
+    },
+    {
+      "epoch": 2.837024554646124,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.617764956931115e-05,
+      "loss": 0.3948,
+      "step": 11785
+    },
+    {
+      "epoch": 2.8382282137698605,
+      "grad_norm": 1.234375,
+      "learning_rate": 4.6163090609006536e-05,
+      "loss": 0.3964,
+      "step": 11790
+    },
+    {
+      "epoch": 2.8394318728935968,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.6148638375044584e-05,
+      "loss": 0.408,
+      "step": 11795
+    },
+    {
+      "epoch": 2.8406355320173327,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.613429289051625e-05,
+      "loss": 0.3863,
+      "step": 11800
+    },
+    {
+      "epoch": 2.841839191141069,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.6120054178341954e-05,
+      "loss": 0.3856,
+      "step": 11805
+    },
+    {
+      "epoch": 2.843042850264805,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.610592226127149e-05,
+      "loss": 0.3929,
+      "step": 11810
+    },
+    {
+      "epoch": 2.844246509388541,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.609189716188406e-05,
+      "loss": 0.3822,
+      "step": 11815
+    },
+    {
+      "epoch": 2.8454501685122775,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.607797890258817e-05,
+      "loss": 0.408,
+      "step": 11820
+    },
+    {
+      "epoch": 2.8466538276360134,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.6064167505621615e-05,
+      "loss": 0.449,
+      "step": 11825
+    },
+    {
+      "epoch": 2.8478574867597497,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.605046299305149e-05,
+      "loss": 0.3883,
+      "step": 11830
+    },
+    {
+      "epoch": 2.8490611458834856,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.603686538677408e-05,
+      "loss": 0.3782,
+      "step": 11835
+    },
+    {
+      "epoch": 2.850264805007222,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.602337470851487e-05,
+      "loss": 0.3858,
+      "step": 11840
+    },
+    {
+      "epoch": 2.8514684641309582,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.600999097982851e-05,
+      "loss": 0.3777,
+      "step": 11845
+    },
+    {
+      "epoch": 2.8526721232546945,
+      "grad_norm": 1.2578125,
+      "learning_rate": 4.5996714222098786e-05,
+      "loss": 0.3919,
+      "step": 11850
+    },
+    {
+      "epoch": 2.8538757823784304,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.598354445653853e-05,
+      "loss": 0.4115,
+      "step": 11855
+    },
+    {
+      "epoch": 2.8550794415021667,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.5970481704189654e-05,
+      "loss": 0.3761,
+      "step": 11860
+    },
+    {
+      "epoch": 2.8562831006259026,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.595752598592309e-05,
+      "loss": 0.3877,
+      "step": 11865
+    },
+    {
+      "epoch": 2.857486759749639,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.594467732243876e-05,
+      "loss": 0.3993,
+      "step": 11870
+    },
+    {
+      "epoch": 2.8586904188733753,
+      "grad_norm": 1.125,
+      "learning_rate": 4.593193573426552e-05,
+      "loss": 0.3994,
+      "step": 11875
+    },
+    {
+      "epoch": 2.859894077997111,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.591930124176118e-05,
+      "loss": 0.3997,
+      "step": 11880
+    },
+    {
+      "epoch": 2.8610977371208475,
+      "grad_norm": 1.2421875,
+      "learning_rate": 4.590677386511242e-05,
+      "loss": 0.4357,
+      "step": 11885
+    },
+    {
+      "epoch": 2.8623013962445834,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.589435362433476e-05,
+      "loss": 0.4057,
+      "step": 11890
+    },
+    {
+      "epoch": 2.8635050553683197,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.5882040539272574e-05,
+      "loss": 0.3927,
+      "step": 11895
+    },
+    {
+      "epoch": 2.864708714492056,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.5869834629599006e-05,
+      "loss": 0.4123,
+      "step": 11900
+    },
+    {
+      "epoch": 2.865912373615792,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.585773591481599e-05,
+      "loss": 0.3836,
+      "step": 11905
+    },
+    {
+      "epoch": 2.867116032739528,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.5845744414254135e-05,
+      "loss": 0.3762,
+      "step": 11910
+    },
+    {
+      "epoch": 2.868319691863264,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.583386014707282e-05,
+      "loss": 0.3659,
+      "step": 11915
+    },
+    {
+      "epoch": 2.8695233509870004,
+      "grad_norm": 1.2578125,
+      "learning_rate": 4.582208313226003e-05,
+      "loss": 0.3944,
+      "step": 11920
+    },
+    {
+      "epoch": 2.8707270101107367,
+      "grad_norm": 1.125,
+      "learning_rate": 4.581041338863245e-05,
+      "loss": 0.3668,
+      "step": 11925
+    },
+    {
+      "epoch": 2.871930669234473,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.579885093483531e-05,
+      "loss": 0.3869,
+      "step": 11930
+    },
+    {
+      "epoch": 2.873134328358209,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.5787395789342436e-05,
+      "loss": 0.3973,
+      "step": 11935
+    },
+    {
+      "epoch": 2.8743379874819452,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.5776047970456265e-05,
+      "loss": 0.3618,
+      "step": 11940
+    },
+    {
+      "epoch": 2.875541646605681,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.576480749630765e-05,
+      "loss": 0.3859,
+      "step": 11945
+    },
+    {
+      "epoch": 2.8767453057294174,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.575367438485602e-05,
+      "loss": 0.3745,
+      "step": 11950
+    },
+    {
+      "epoch": 2.8779489648531538,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.5742648653889204e-05,
+      "loss": 0.356,
+      "step": 11955
+    },
+    {
+      "epoch": 2.8791526239768896,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.573173032102351e-05,
+      "loss": 0.4155,
+      "step": 11960
+    },
+    {
+      "epoch": 2.880356283100626,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.572091940370366e-05,
+      "loss": 0.3779,
+      "step": 11965
+    },
+    {
+      "epoch": 2.881559942224362,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.571021591920267e-05,
+      "loss": 0.4197,
+      "step": 11970
+    },
+    {
+      "epoch": 2.882763601348098,
+      "grad_norm": 1.2421875,
+      "learning_rate": 4.569961988462203e-05,
+      "loss": 0.3803,
+      "step": 11975
+    },
+    {
+      "epoch": 2.8839672604718345,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.568913131689144e-05,
+      "loss": 0.4214,
+      "step": 11980
+    },
+    {
+      "epoch": 2.8851709195955704,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.5678750232768954e-05,
+      "loss": 0.3551,
+      "step": 11985
+    },
+    {
+      "epoch": 2.8863745787193067,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.56684766488409e-05,
+      "loss": 0.3932,
+      "step": 11990
+    },
+    {
+      "epoch": 2.887578237843043,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.565831058152179e-05,
+      "loss": 0.3653,
+      "step": 11995
+    },
+    {
+      "epoch": 2.888781896966779,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.564825204705445e-05,
+      "loss": 0.3705,
+      "step": 12000
+    },
+    {
+      "epoch": 2.888781896966779,
+      "eval_loss": 0.36297592520713806,
+      "eval_runtime": 2.3567,
+      "eval_samples_per_second": 84.863,
+      "eval_steps_per_second": 84.863,
+      "step": 12000
+    },
+    {
+      "epoch": 2.889985556090515,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.563830106150981e-05,
+      "loss": 0.3619,
+      "step": 12005
+    },
+    {
+      "epoch": 2.8911892152142515,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.562845764078699e-05,
+      "loss": 0.4069,
+      "step": 12010
+    },
+    {
+      "epoch": 2.8923928743379874,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.561872180061326e-05,
+      "loss": 0.3982,
+      "step": 12015
+    },
+    {
+      "epoch": 2.8935965334617237,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.5609093556543985e-05,
+      "loss": 0.376,
+      "step": 12020
+    },
+    {
+      "epoch": 2.8948001925854596,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.559957292396265e-05,
+      "loss": 0.386,
+      "step": 12025
+    },
+    {
+      "epoch": 2.896003851709196,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.559015991808078e-05,
+      "loss": 0.3997,
+      "step": 12030
+    },
+    {
+      "epoch": 2.8972075108329323,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.5580854553937935e-05,
+      "loss": 0.3895,
+      "step": 12035
+    },
+    {
+      "epoch": 2.898411169956668,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.557165684640171e-05,
+      "loss": 0.3929,
+      "step": 12040
+    },
+    {
+      "epoch": 2.8996148290804045,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.556256681016768e-05,
+      "loss": 0.3843,
+      "step": 12045
+    },
+    {
+      "epoch": 2.9008184882041403,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.555358445975939e-05,
+      "loss": 0.4044,
+      "step": 12050
+    },
+    {
+      "epoch": 2.9020221473278767,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.5544709809528343e-05,
+      "loss": 0.4108,
+      "step": 12055
+    },
+    {
+      "epoch": 2.903225806451613,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.553594287365394e-05,
+      "loss": 0.3778,
+      "step": 12060
+    },
+    {
+      "epoch": 2.9044294655753493,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.552728366614352e-05,
+      "loss": 0.3852,
+      "step": 12065
+    },
+    {
+      "epoch": 2.905633124699085,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.551873220083226e-05,
+      "loss": 0.3932,
+      "step": 12070
+    },
+    {
+      "epoch": 2.9068367838228215,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.551028849138321e-05,
+      "loss": 0.3745,
+      "step": 12075
+    },
+    {
+      "epoch": 2.9080404429465574,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.5501952551287255e-05,
+      "loss": 0.3862,
+      "step": 12080
+    },
+    {
+      "epoch": 2.9092441020702937,
+      "grad_norm": 1.2734375,
+      "learning_rate": 4.5493724393863104e-05,
+      "loss": 0.3694,
+      "step": 12085
+    },
+    {
+      "epoch": 2.91044776119403,
+      "grad_norm": 1.234375,
+      "learning_rate": 4.548560403225722e-05,
+      "loss": 0.4124,
+      "step": 12090
+    },
+    {
+      "epoch": 2.911651420317766,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.547759147944386e-05,
+      "loss": 0.3763,
+      "step": 12095
+    },
+    {
+      "epoch": 2.9128550794415022,
+      "grad_norm": 1.234375,
+      "learning_rate": 4.5469686748225054e-05,
+      "loss": 0.3835,
+      "step": 12100
+    },
+    {
+      "epoch": 2.914058738565238,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.546188985123051e-05,
+      "loss": 0.4051,
+      "step": 12105
+    },
+    {
+      "epoch": 2.9152623976889744,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.545420080091769e-05,
+      "loss": 0.3837,
+      "step": 12110
+    },
+    {
+      "epoch": 2.9164660568127108,
+      "grad_norm": 1.125,
+      "learning_rate": 4.5446619609571724e-05,
+      "loss": 0.3746,
+      "step": 12115
+    },
+    {
+      "epoch": 2.9176697159364466,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.543914628930539e-05,
+      "loss": 0.4243,
+      "step": 12120
+    },
+    {
+      "epoch": 2.918873375060183,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.5431780852059166e-05,
+      "loss": 0.3876,
+      "step": 12125
+    },
+    {
+      "epoch": 2.920077034183919,
+      "grad_norm": 1.296875,
+      "learning_rate": 4.5424523309601115e-05,
+      "loss": 0.4298,
+      "step": 12130
+    },
+    {
+      "epoch": 2.921280693307655,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.541737367352694e-05,
+      "loss": 0.3514,
+      "step": 12135
+    },
+    {
+      "epoch": 2.9224843524313915,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.541033195525991e-05,
+      "loss": 0.4289,
+      "step": 12140
+    },
+    {
+      "epoch": 2.923688011555128,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.540339816605091e-05,
+      "loss": 0.3888,
+      "step": 12145
+    },
+    {
+      "epoch": 2.9248916706788637,
+      "grad_norm": 1.3046875,
+      "learning_rate": 4.5396572316978325e-05,
+      "loss": 0.4116,
+      "step": 12150
+    },
+    {
+      "epoch": 2.9260953298026,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.538985441894814e-05,
+      "loss": 0.3848,
+      "step": 12155
+    },
+    {
+      "epoch": 2.927298988926336,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.538324448269383e-05,
+      "loss": 0.3879,
+      "step": 12160
+    },
+    {
+      "epoch": 2.928502648050072,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.5376742518776384e-05,
+      "loss": 0.3769,
+      "step": 12165
+    },
+    {
+      "epoch": 2.9297063071738085,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.5370348537584245e-05,
+      "loss": 0.3912,
+      "step": 12170
+    },
+    {
+      "epoch": 2.9309099662975444,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.5364062549333376e-05,
+      "loss": 0.4036,
+      "step": 12175
+    },
+    {
+      "epoch": 2.9321136254212807,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.535788456406718e-05,
+      "loss": 0.4089,
+      "step": 12180
+    },
+    {
+      "epoch": 2.9333172845450166,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.535181459165649e-05,
+      "loss": 0.395,
+      "step": 12185
+    },
+    {
+      "epoch": 2.934520943668753,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.534585264179955e-05,
+      "loss": 0.4056,
+      "step": 12190
+    },
+    {
+      "epoch": 2.9357246027924893,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.533999872402204e-05,
+      "loss": 0.3726,
+      "step": 12195
+    },
+    {
+      "epoch": 2.9369282619162256,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.533425284767704e-05,
+      "loss": 0.3937,
+      "step": 12200
+    },
+    {
+      "epoch": 2.9381319210399615,
+      "grad_norm": 1.3046875,
+      "learning_rate": 4.532861502194495e-05,
+      "loss": 0.3941,
+      "step": 12205
+    },
+    {
+      "epoch": 2.939335580163698,
+      "grad_norm": 1.25,
+      "learning_rate": 4.5323085255833595e-05,
+      "loss": 0.4077,
+      "step": 12210
+    },
+    {
+      "epoch": 2.9405392392874337,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.531766355817812e-05,
+      "loss": 0.3771,
+      "step": 12215
+    },
+    {
+      "epoch": 2.94174289841117,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.5312349937640994e-05,
+      "loss": 0.3888,
+      "step": 12220
+    },
+    {
+      "epoch": 2.9429465575349063,
+      "grad_norm": 1.125,
+      "learning_rate": 4.530714440271203e-05,
+      "loss": 0.4036,
+      "step": 12225
+    },
+    {
+      "epoch": 2.944150216658642,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.530204696170835e-05,
+      "loss": 0.3822,
+      "step": 12230
+    },
+    {
+      "epoch": 2.9453538757823785,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.529705762277434e-05,
+      "loss": 0.4212,
+      "step": 12235
+    },
+    {
+      "epoch": 2.9465575349061144,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.529217639388169e-05,
+      "loss": 0.3818,
+      "step": 12240
+    },
+    {
+      "epoch": 2.9477611940298507,
+      "grad_norm": 1.125,
+      "learning_rate": 4.528740328282936e-05,
+      "loss": 0.362,
+      "step": 12245
+    },
+    {
+      "epoch": 2.948964853153587,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.5282738297243556e-05,
+      "loss": 0.3889,
+      "step": 12250
+    },
+    {
+      "epoch": 2.950168512277323,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.527818144457772e-05,
+      "loss": 0.3724,
+      "step": 12255
+    },
+    {
+      "epoch": 2.9513721714010592,
+      "grad_norm": 1.2734375,
+      "learning_rate": 4.527373273211254e-05,
+      "loss": 0.4102,
+      "step": 12260
+    },
+    {
+      "epoch": 2.952575830524795,
+      "grad_norm": 1.125,
+      "learning_rate": 4.5269392166955915e-05,
+      "loss": 0.3812,
+      "step": 12265
+    },
+    {
+      "epoch": 2.9537794896485314,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.5265159756042956e-05,
+      "loss": 0.3913,
+      "step": 12270
+    },
+    {
+      "epoch": 2.9549831487722678,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.526103550613598e-05,
+      "loss": 0.3818,
+      "step": 12275
+    },
+    {
+      "epoch": 2.956186807896004,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.525701942382447e-05,
+      "loss": 0.4098,
+      "step": 12280
+    },
+    {
+      "epoch": 2.95739046701974,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.5253111515525094e-05,
+      "loss": 0.394,
+      "step": 12285
+    },
+    {
+      "epoch": 2.9585941261434763,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.5249311787481704e-05,
+      "loss": 0.364,
+      "step": 12290
+    },
+    {
+      "epoch": 2.959797785267212,
+      "grad_norm": 1.296875,
+      "learning_rate": 4.5245620245765275e-05,
+      "loss": 0.385,
+      "step": 12295
+    },
+    {
+      "epoch": 2.9610014443909485,
+      "grad_norm": 1.125,
+      "learning_rate": 4.5242036896273964e-05,
+      "loss": 0.3821,
+      "step": 12300
+    },
+    {
+      "epoch": 2.962205103514685,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.5238561744733015e-05,
+      "loss": 0.3661,
+      "step": 12305
+    },
+    {
+      "epoch": 2.9634087626384207,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.523519479669485e-05,
+      "loss": 0.3844,
+      "step": 12310
+    },
+    {
+      "epoch": 2.964612421762157,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.5231936057538984e-05,
+      "loss": 0.3873,
+      "step": 12315
+    },
+    {
+      "epoch": 2.965816080885893,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.5228785532472045e-05,
+      "loss": 0.4045,
+      "step": 12320
+    },
+    {
+      "epoch": 2.967019740009629,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.522574322652777e-05,
+      "loss": 0.3682,
+      "step": 12325
+    },
+    {
+      "epoch": 2.9682233991333655,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.5222809144566984e-05,
+      "loss": 0.3824,
+      "step": 12330
+    },
+    {
+      "epoch": 2.969427058257102,
+      "grad_norm": 0.953125,
+      "learning_rate": 4.521998329127758e-05,
+      "loss": 0.4006,
+      "step": 12335
+    },
+    {
+      "epoch": 2.9706307173808377,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.521726567117457e-05,
+      "loss": 0.4114,
+      "step": 12340
+    },
+    {
+      "epoch": 2.971834376504574,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.5214656288600014e-05,
+      "loss": 0.3867,
+      "step": 12345
+    },
+    {
+      "epoch": 2.97303803562831,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.521215514772301e-05,
+      "loss": 0.3931,
+      "step": 12350
+    },
+    {
+      "epoch": 2.9742416947520463,
+      "grad_norm": 1.125,
+      "learning_rate": 4.5209762252539775e-05,
+      "loss": 0.3825,
+      "step": 12355
+    },
+    {
+      "epoch": 2.9754453538757826,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.5207477606873514e-05,
+      "loss": 0.405,
+      "step": 12360
+    },
+    {
+      "epoch": 2.9766490129995185,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.520530121437452e-05,
+      "loss": 0.387,
+      "step": 12365
+    },
+    {
+      "epoch": 2.9778526721232548,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.5203233078520115e-05,
+      "loss": 0.3653,
+      "step": 12370
+    },
+    {
+      "epoch": 2.9790563312469907,
+      "grad_norm": 1.296875,
+      "learning_rate": 4.520127320261463e-05,
+      "loss": 0.3806,
+      "step": 12375
+    },
+    {
+      "epoch": 2.980259990370727,
+      "grad_norm": 1.2734375,
+      "learning_rate": 4.519942158978947e-05,
+      "loss": 0.3833,
+      "step": 12380
+    },
+    {
+      "epoch": 2.9814636494944633,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.519767824300301e-05,
+      "loss": 0.4026,
+      "step": 12385
+    },
+    {
+      "epoch": 2.982667308618199,
+      "grad_norm": 1.0,
+      "learning_rate": 4.519604316504069e-05,
+      "loss": 0.3843,
+      "step": 12390
+    },
+    {
+      "epoch": 2.9838709677419355,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.519451635851493e-05,
+      "loss": 0.4132,
+      "step": 12395
+    },
+    {
+      "epoch": 2.9850746268656714,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.519309782586519e-05,
+      "loss": 0.3989,
+      "step": 12400
+    },
+    {
+      "epoch": 2.9862782859894077,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.519178756935792e-05,
+      "loss": 0.3793,
+      "step": 12405
+    },
+    {
+      "epoch": 2.987481945113144,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.519058559108658e-05,
+      "loss": 0.405,
+      "step": 12410
+    },
+    {
+      "epoch": 2.9886856042368803,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.5189491892971596e-05,
+      "loss": 0.3882,
+      "step": 12415
+    },
+    {
+      "epoch": 2.9898892633606162,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.518850647676045e-05,
+      "loss": 0.398,
+      "step": 12420
+    },
+    {
+      "epoch": 2.9910929224843525,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.518762934402757e-05,
+      "loss": 0.4008,
+      "step": 12425
+    },
+    {
+      "epoch": 2.9922965816080884,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.5186860496174374e-05,
+      "loss": 0.4006,
+      "step": 12430
+    },
+    {
+      "epoch": 2.9935002407318247,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.5186199934429314e-05,
+      "loss": 0.3831,
+      "step": 12435
+    },
+    {
+      "epoch": 2.994703899855561,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.518564765984778e-05,
+      "loss": 0.4147,
+      "step": 12440
+    },
+    {
+      "epoch": 2.995907558979297,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.518520367331216e-05,
+      "loss": 0.3909,
+      "step": 12445
+    },
+    {
+      "epoch": 2.9971112181030333,
+      "grad_norm": 1.28125,
+      "learning_rate": 4.518486797553185e-05,
+      "loss": 0.4093,
+      "step": 12450
+    },
+    {
+      "epoch": 2.998314877226769,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.5184640567043195e-05,
+      "loss": 0.3735,
+      "step": 12455
+    },
+    {
+      "epoch": 2.999277804525758,
+      "eval_loss": 0.3596603274345398,
+      "eval_runtime": 2.3319,
+      "eval_samples_per_second": 85.768,
+      "eval_steps_per_second": 85.768,
+      "step": 12459
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 12462,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.12509450944512e+17,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}