diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,3895 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 2148,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0069881201956673656,
+      "grad_norm": 1.4883168935775757,
+      "learning_rate": 7.407407407407407e-07,
+      "loss": 1.0398,
+      "mean_token_accuracy": 0.6855247735977172,
+      "num_tokens": 10370767.0,
+      "step": 5
+    },
+    {
+      "epoch": 0.013976240391334731,
+      "grad_norm": 0.8389180898666382,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.0478,
+      "mean_token_accuracy": 0.6838523238897324,
+      "num_tokens": 20777189.0,
+      "step": 10
+    },
+    {
+      "epoch": 0.020964360587002098,
+      "grad_norm": 0.9392557740211487,
+      "learning_rate": 2.5925925925925925e-06,
+      "loss": 1.0355,
+      "mean_token_accuracy": 0.686600410938263,
+      "num_tokens": 31178208.0,
+      "step": 15
+    },
+    {
+      "epoch": 0.027952480782669462,
+      "grad_norm": 0.5707754492759705,
+      "learning_rate": 3.5185185185185187e-06,
+      "loss": 1.0352,
+      "mean_token_accuracy": 0.686421462893486,
+      "num_tokens": 41592308.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.03494060097833683,
+      "grad_norm": 0.6185972094535828,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.032,
+      "mean_token_accuracy": 0.6872807204723358,
+      "num_tokens": 51979715.0,
+      "step": 25
+    },
+    {
+      "epoch": 0.041928721174004195,
+      "grad_norm": 0.463251531124115,
+      "learning_rate": 5.370370370370371e-06,
+      "loss": 1.0152,
+      "mean_token_accuracy": 0.6908860236406327,
+      "num_tokens": 62394982.0,
+      "step": 30
+    },
+    {
+      "epoch": 0.04891684136967156,
+      "grad_norm": 0.3636445999145508,
+      "learning_rate": 6.296296296296297e-06,
+      "loss": 1.0179,
+      "mean_token_accuracy": 0.6905231744050979,
+      "num_tokens": 72803742.0,
+      "step": 35
+    },
+    {
+      "epoch": 0.055904961565338925,
+      "grad_norm": 0.2937327027320862,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 1.0266,
+      "mean_token_accuracy": 0.6888427168130875,
+      "num_tokens": 83202891.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.06289308176100629,
+      "grad_norm": 0.2677152156829834,
+      "learning_rate": 8.148148148148148e-06,
+      "loss": 1.0124,
+      "mean_token_accuracy": 0.6918304771184921,
+      "num_tokens": 93587913.0,
+      "step": 45
+    },
+    {
+      "epoch": 0.06988120195667366,
+      "grad_norm": 0.2599441707134247,
+      "learning_rate": 9.074074074074075e-06,
+      "loss": 1.0038,
+      "mean_token_accuracy": 0.6940747231245041,
+      "num_tokens": 103964666.0,
+      "step": 50
+    },
+    {
+      "epoch": 0.07686932215234102,
+      "grad_norm": 0.2430637627840042,
+      "learning_rate": 1e-05,
+      "loss": 0.9976,
+      "mean_token_accuracy": 0.6953895181417465,
+      "num_tokens": 114353247.0,
+      "step": 55
+    },
+    {
+      "epoch": 0.08385744234800839,
+      "grad_norm": 0.25146690011024475,
+      "learning_rate": 1.0925925925925926e-05,
+      "loss": 1.0005,
+      "mean_token_accuracy": 0.694663581252098,
+      "num_tokens": 124731779.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.09084556254367575,
+      "grad_norm": 0.22198280692100525,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 1.0101,
+      "mean_token_accuracy": 0.6926915943622589,
+      "num_tokens": 135144687.0,
+      "step": 65
+    },
+    {
+      "epoch": 0.09783368273934312,
+      "grad_norm": 0.22740793228149414,
+      "learning_rate": 1.2777777777777777e-05,
+      "loss": 1.0097,
+      "mean_token_accuracy": 0.6922657936811447,
+      "num_tokens": 145526205.0,
+      "step": 70
+    },
+    {
+      "epoch": 0.10482180293501048,
+      "grad_norm": 0.22107183933258057,
+      "learning_rate": 1.3703703703703706e-05,
+      "loss": 1.0119,
+      "mean_token_accuracy": 0.6913794606924057,
+      "num_tokens": 155919187.0,
+      "step": 75
+    },
+    {
+      "epoch": 0.11180992313067785,
+      "grad_norm": 0.27914658188819885,
+      "learning_rate": 1.4629629629629631e-05,
+      "loss": 1.0028,
+      "mean_token_accuracy": 0.6938924849033355,
+      "num_tokens": 166311147.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.1187980433263452,
+      "grad_norm": 0.2629740238189697,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 1.0155,
+      "mean_token_accuracy": 0.6906299769878388,
+      "num_tokens": 176702317.0,
+      "step": 85
+    },
+    {
+      "epoch": 0.12578616352201258,
+      "grad_norm": 0.2621951997280121,
+      "learning_rate": 1.6481481481481482e-05,
+      "loss": 1.0055,
+      "mean_token_accuracy": 0.6934900671243668,
+      "num_tokens": 187121773.0,
+      "step": 90
+    },
+    {
+      "epoch": 0.13277428371767994,
+      "grad_norm": 0.24768415093421936,
+      "learning_rate": 1.740740740740741e-05,
+      "loss": 1.0067,
+      "mean_token_accuracy": 0.6931978613138199,
+      "num_tokens": 197538039.0,
+      "step": 95
+    },
+    {
+      "epoch": 0.13976240391334732,
+      "grad_norm": 0.279287189245224,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 1.0027,
+      "mean_token_accuracy": 0.6943970769643784,
+      "num_tokens": 207934697.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.14675052410901468,
+      "grad_norm": 0.4129040241241455,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 1.0033,
+      "mean_token_accuracy": 0.6932864665985108,
+      "num_tokens": 218334496.0,
+      "step": 105
+    },
+    {
+      "epoch": 0.15373864430468204,
+      "grad_norm": 0.5416187644004822,
+      "learning_rate": 1.9999991106543355e-05,
+      "loss": 1.0084,
+      "mean_token_accuracy": 0.6928850650787354,
+      "num_tokens": 228759368.0,
+      "step": 110
+    },
+    {
+      "epoch": 0.1607267645003494,
+      "grad_norm": 0.4506029188632965,
+      "learning_rate": 1.999967983777526e-05,
+      "loss": 1.0188,
+      "mean_token_accuracy": 0.690243837237358,
+      "num_tokens": 239138531.0,
+      "step": 115
+    },
+    {
+      "epoch": 0.16771488469601678,
+      "grad_norm": 0.273882120847702,
+      "learning_rate": 1.9998923917266123e-05,
+      "loss": 1.0004,
+      "mean_token_accuracy": 0.6942124694585801,
+      "num_tokens": 249518869.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.17470300489168414,
+      "grad_norm": 0.32800787687301636,
+      "learning_rate": 1.999772338983404e-05,
+      "loss": 0.9993,
+      "mean_token_accuracy": 0.6946051716804504,
+      "num_tokens": 259936610.0,
+      "step": 125
+    },
+    {
+      "epoch": 0.1816911250873515,
+      "grad_norm": 0.35255053639411926,
+      "learning_rate": 1.9996078326657592e-05,
+      "loss": 1.0021,
+      "mean_token_accuracy": 0.6943468749523163,
+      "num_tokens": 270341806.0,
+      "step": 130
+    },
+    {
+      "epoch": 0.18867924528301888,
+      "grad_norm": 0.35514041781425476,
+      "learning_rate": 1.9993988825271643e-05,
+      "loss": 1.0001,
+      "mean_token_accuracy": 0.6942067295312881,
+      "num_tokens": 280749074.0,
+      "step": 135
+    },
+    {
+      "epoch": 0.19566736547868624,
+      "grad_norm": 0.31045007705688477,
+      "learning_rate": 1.9991455009561535e-05,
+      "loss": 1.0009,
+      "mean_token_accuracy": 0.694583398103714,
+      "num_tokens": 291068981.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.2026554856743536,
+      "grad_norm": 0.37079402804374695,
+      "learning_rate": 1.9988477029755756e-05,
+      "loss": 0.9966,
+      "mean_token_accuracy": 0.6956235706806183,
+      "num_tokens": 301442780.0,
+      "step": 145
+    },
+    {
+      "epoch": 0.20964360587002095,
+      "grad_norm": 0.3016902208328247,
+      "learning_rate": 1.9985055062417034e-05,
+      "loss": 1.0004,
+      "mean_token_accuracy": 0.6947520196437835,
+      "num_tokens": 311853604.0,
+      "step": 150
+    },
+    {
+      "epoch": 0.21663172606568834,
+      "grad_norm": 0.3417947292327881,
+      "learning_rate": 1.9981189310431857e-05,
+      "loss": 0.9993,
+      "mean_token_accuracy": 0.6947573721408844,
+      "num_tokens": 322261710.0,
+      "step": 155
+    },
+    {
+      "epoch": 0.2236198462613557,
+      "grad_norm": 0.3535808324813843,
+      "learning_rate": 1.9976880002998463e-05,
+      "loss": 0.9922,
+      "mean_token_accuracy": 0.6957294464111328,
+      "num_tokens": 332668576.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.23060796645702306,
+      "grad_norm": 0.2367454469203949,
+      "learning_rate": 1.9972127395613225e-05,
+      "loss": 0.9972,
+      "mean_token_accuracy": 0.6954801857471467,
+      "num_tokens": 343060718.0,
+      "step": 165
+    },
+    {
+      "epoch": 0.2375960866526904,
+      "grad_norm": 0.26334941387176514,
+      "learning_rate": 1.9966931770055523e-05,
+      "loss": 0.995,
+      "mean_token_accuracy": 0.6958191096782684,
+      "num_tokens": 353461725.0,
+      "step": 170
+    },
+    {
+      "epoch": 0.2445842068483578,
+      "grad_norm": 0.3441227078437805,
+      "learning_rate": 1.9961293434371044e-05,
+      "loss": 1.0044,
+      "mean_token_accuracy": 0.6938858300447464,
+      "num_tokens": 363838201.0,
+      "step": 175
+    },
+    {
+      "epoch": 0.25157232704402516,
+      "grad_norm": 0.2843206822872162,
+      "learning_rate": 1.9955212722853487e-05,
+      "loss": 0.9995,
+      "mean_token_accuracy": 0.6946384638547898,
+      "num_tokens": 374258478.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.2585604472396925,
+      "grad_norm": 0.2655826210975647,
+      "learning_rate": 1.9948689996024773e-05,
+      "loss": 0.9981,
+      "mean_token_accuracy": 0.6951499044895172,
+      "num_tokens": 384649889.0,
+      "step": 185
+    },
+    {
+      "epoch": 0.2655485674353599,
+      "grad_norm": 0.3091617822647095,
+      "learning_rate": 1.9941725640613655e-05,
+      "loss": 0.9909,
+      "mean_token_accuracy": 0.6968458265066146,
+      "num_tokens": 395029124.0,
+      "step": 190
+    },
+    {
+      "epoch": 0.27253668763102723,
+      "grad_norm": 0.3030296564102173,
+      "learning_rate": 1.9934320069532797e-05,
+      "loss": 0.9942,
+      "mean_token_accuracy": 0.6959166437387466,
+      "num_tokens": 405425365.0,
+      "step": 195
+    },
+    {
+      "epoch": 0.27952480782669464,
+      "grad_norm": 0.25540319085121155,
+      "learning_rate": 1.9926473721854284e-05,
+      "loss": 1.0046,
+      "mean_token_accuracy": 0.6931773543357849,
+      "num_tokens": 415831263.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.286512928022362,
+      "grad_norm": 0.33107173442840576,
+      "learning_rate": 1.99181870627836e-05,
+      "loss": 1.0116,
+      "mean_token_accuracy": 0.6916185826063156,
+      "num_tokens": 426180521.0,
+      "step": 205
+    },
+    {
+      "epoch": 0.29350104821802936,
+      "grad_norm": 0.39433753490448,
+      "learning_rate": 1.990946058363202e-05,
+      "loss": 0.9969,
+      "mean_token_accuracy": 0.6951790243387223,
+      "num_tokens": 436550729.0,
+      "step": 210
+    },
+    {
+      "epoch": 0.3004891684136967,
+      "grad_norm": 0.382136732339859,
+      "learning_rate": 1.9900294801787537e-05,
+      "loss": 0.9978,
+      "mean_token_accuracy": 0.6952319622039795,
+      "num_tokens": 446953885.0,
+      "step": 215
+    },
+    {
+      "epoch": 0.3074772886093641,
+      "grad_norm": 0.3471049666404724,
+      "learning_rate": 1.9890690260684122e-05,
+      "loss": 1.009,
+      "mean_token_accuracy": 0.6918054640293121,
+      "num_tokens": 457368923.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.31446540880503143,
+      "grad_norm": 0.2898043692111969,
+      "learning_rate": 1.9880647529769536e-05,
+      "loss": 0.9957,
+      "mean_token_accuracy": 0.6954202592372895,
+      "num_tokens": 467745509.0,
+      "step": 225
+    },
+    {
+      "epoch": 0.3214535290006988,
+      "grad_norm": 0.2959822118282318,
+      "learning_rate": 1.987016720447157e-05,
+      "loss": 1.0166,
+      "mean_token_accuracy": 0.6907585173845291,
+      "num_tokens": 478125444.0,
+      "step": 230
+    },
+    {
+      "epoch": 0.3284416491963662,
+      "grad_norm": 0.33595606684684753,
+      "learning_rate": 1.985924990616274e-05,
+      "loss": 0.9994,
+      "mean_token_accuracy": 0.6947229593992233,
+      "num_tokens": 488548530.0,
+      "step": 235
+    },
+    {
+      "epoch": 0.33542976939203356,
+      "grad_norm": 0.3360680937767029,
+      "learning_rate": 1.9847896282123435e-05,
+      "loss": 0.9982,
+      "mean_token_accuracy": 0.6951431065797806,
+      "num_tokens": 498959347.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.3424178895877009,
+      "grad_norm": 0.2827513813972473,
+      "learning_rate": 1.9836107005503543e-05,
+      "loss": 0.9966,
+      "mean_token_accuracy": 0.6951984494924546,
+      "num_tokens": 509357796.0,
+      "step": 245
+    },
+    {
+      "epoch": 0.3494060097833683,
+      "grad_norm": 0.426392138004303,
+      "learning_rate": 1.982388277528256e-05,
+      "loss": 0.9975,
+      "mean_token_accuracy": 0.6952957093715668,
+      "num_tokens": 519757159.0,
+      "step": 250
+    },
+    {
+      "epoch": 0.35639412997903563,
+      "grad_norm": 0.36645272374153137,
+      "learning_rate": 1.981122431622813e-05,
+      "loss": 0.9955,
+      "mean_token_accuracy": 0.6955067068338394,
+      "num_tokens": 530137537.0,
+      "step": 255
+    },
+    {
+      "epoch": 0.363382250174703,
+      "grad_norm": 0.30606377124786377,
+      "learning_rate": 1.979813237885306e-05,
+      "loss": 1.0024,
+      "mean_token_accuracy": 0.6938055425882339,
+      "num_tokens": 540516735.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.33657437562942505,
+      "learning_rate": 1.9784607739370854e-05,
+      "loss": 1.0008,
+      "mean_token_accuracy": 0.6939801007509232,
+      "num_tokens": 550924844.0,
+      "step": 265
+    },
+    {
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.3700354993343353,
+      "learning_rate": 1.9770651199649684e-05,
+      "loss": 1.008,
+      "mean_token_accuracy": 0.6923987686634063,
+      "num_tokens": 561322928.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.3843466107617051,
+      "grad_norm": 0.4612869918346405,
+      "learning_rate": 1.9756263587164823e-05,
+      "loss": 0.9964,
+      "mean_token_accuracy": 0.6956883102655411,
+      "num_tokens": 571719820.0,
+      "step": 275
+    },
+    {
+      "epoch": 0.3913347309573725,
+      "grad_norm": 0.28541725873947144,
+      "learning_rate": 1.974144575494961e-05,
+      "loss": 1.0052,
+      "mean_token_accuracy": 0.6935481756925583,
+      "num_tokens": 582133553.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.39832285115303984,
+      "grad_norm": 0.27733850479125977,
+      "learning_rate": 1.9726198581544875e-05,
+      "loss": 1.0038,
+      "mean_token_accuracy": 0.6932170450687408,
+      "num_tokens": 592552862.0,
+      "step": 285
+    },
+    {
+      "epoch": 0.4053109713487072,
+      "grad_norm": 0.25651562213897705,
+      "learning_rate": 1.971052297094683e-05,
+      "loss": 0.9981,
+      "mean_token_accuracy": 0.6951008081436157,
+      "num_tokens": 602969229.0,
+      "step": 290
+    },
+    {
+      "epoch": 0.41229909154437455,
+      "grad_norm": 0.2750900685787201,
+      "learning_rate": 1.969441985255349e-05,
+      "loss": 0.9995,
+      "mean_token_accuracy": 0.6946498632431031,
+      "num_tokens": 613327395.0,
+      "step": 295
+    },
+    {
+      "epoch": 0.4192872117400419,
+      "grad_norm": 0.3030959665775299,
+      "learning_rate": 1.967789018110956e-05,
+      "loss": 0.9938,
+      "mean_token_accuracy": 0.6957368463277817,
+      "num_tokens": 623734734.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.42627533193570927,
+      "grad_norm": 0.2968519330024719,
+      "learning_rate": 1.9660934936649836e-05,
+      "loss": 0.9986,
+      "mean_token_accuracy": 0.6947296649217606,
+      "num_tokens": 634141013.0,
+      "step": 305
+    },
+    {
+      "epoch": 0.4332634521313767,
+      "grad_norm": 0.32919952273368835,
+      "learning_rate": 1.9643555124441093e-05,
+      "loss": 0.994,
+      "mean_token_accuracy": 0.6957409083843231,
+      "num_tokens": 644493377.0,
+      "step": 310
+    },
+    {
+      "epoch": 0.44025157232704404,
+      "grad_norm": 0.36588236689567566,
+      "learning_rate": 1.9625751774922485e-05,
+      "loss": 0.9918,
+      "mean_token_accuracy": 0.6964637786149979,
+      "num_tokens": 654861392.0,
+      "step": 315
+    },
+    {
+      "epoch": 0.4472396925227114,
+      "grad_norm": 0.3859914541244507,
+      "learning_rate": 1.9607525943644458e-05,
+      "loss": 0.997,
+      "mean_token_accuracy": 0.6950535237789154,
+      "num_tokens": 665281740.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.45422781271837875,
+      "grad_norm": 0.35685864090919495,
+      "learning_rate": 1.9588878711206147e-05,
+      "loss": 0.9942,
+      "mean_token_accuracy": 0.6956727713346481,
+      "num_tokens": 675674938.0,
+      "step": 325
+    },
+    {
+      "epoch": 0.4612159329140461,
+      "grad_norm": 0.29461824893951416,
+      "learning_rate": 1.956981118319134e-05,
+      "loss": 1.002,
+      "mean_token_accuracy": 0.6937934666872024,
+      "num_tokens": 686085726.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.46820405310971347,
+      "grad_norm": 0.25779464840888977,
+      "learning_rate": 1.9550324490102893e-05,
+      "loss": 0.9996,
+      "mean_token_accuracy": 0.6942994087934494,
+      "num_tokens": 696519911.0,
+      "step": 335
+    },
+    {
+      "epoch": 0.4751921733053808,
+      "grad_norm": 0.32080188393592834,
+      "learning_rate": 1.9530419787295732e-05,
+      "loss": 1.0017,
+      "mean_token_accuracy": 0.6941420465707779,
+      "num_tokens": 706915324.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.48218029350104824,
+      "grad_norm": 0.3333379030227661,
+      "learning_rate": 1.951009825490833e-05,
+      "loss": 0.9978,
+      "mean_token_accuracy": 0.6944221198558808,
+      "num_tokens": 717279358.0,
+      "step": 345
+    },
+    {
+      "epoch": 0.4891684136967156,
+      "grad_norm": 0.2867525517940521,
+      "learning_rate": 1.9489361097792763e-05,
+      "loss": 0.9956,
+      "mean_token_accuracy": 0.6951427847146988,
+      "num_tokens": 727668296.0,
+      "step": 350
+    },
+    {
+      "epoch": 0.49615653389238296,
+      "grad_norm": 0.28376054763793945,
+      "learning_rate": 1.9468209545443243e-05,
+      "loss": 0.9947,
+      "mean_token_accuracy": 0.6956383109092712,
+      "num_tokens": 738059747.0,
+      "step": 355
+    },
+    {
+      "epoch": 0.5031446540880503,
+      "grad_norm": 0.3152044117450714,
+      "learning_rate": 1.944664485192325e-05,
+      "loss": 0.9943,
+      "mean_token_accuracy": 0.6953181773424149,
+      "num_tokens": 748421972.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.5101327742837177,
+      "grad_norm": 0.30702683329582214,
+      "learning_rate": 1.9424668295791172e-05,
+      "loss": 0.9916,
+      "mean_token_accuracy": 0.696011558175087,
+      "num_tokens": 758808879.0,
+      "step": 365
+    },
+    {
+      "epoch": 0.517120894479385,
+      "grad_norm": 0.32153916358947754,
+      "learning_rate": 1.9402281180024483e-05,
+      "loss": 0.9981,
+      "mean_token_accuracy": 0.6946611881256104,
+      "num_tokens": 769184194.0,
+      "step": 370
+    },
+    {
+      "epoch": 0.5241090146750524,
+      "grad_norm": 0.25884702801704407,
+      "learning_rate": 1.9379484831942516e-05,
+      "loss": 0.9936,
+      "mean_token_accuracy": 0.6954076588153839,
+      "num_tokens": 779577281.0,
+      "step": 375
+    },
+    {
+      "epoch": 0.5310971348707197,
+      "grad_norm": 0.33953726291656494,
+      "learning_rate": 1.9356280603127745e-05,
+      "loss": 0.9923,
+      "mean_token_accuracy": 0.6961508899927139,
+      "num_tokens": 789973447.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.5380852550663872,
+      "grad_norm": 0.26376229524612427,
+      "learning_rate": 1.9332669869345663e-05,
+      "loss": 0.9867,
+      "mean_token_accuracy": 0.697160255908966,
+      "num_tokens": 800391253.0,
+      "step": 385
+    },
+    {
+      "epoch": 0.5450733752620545,
+      "grad_norm": 0.26495668292045593,
+      "learning_rate": 1.930865403046322e-05,
+      "loss": 0.9986,
+      "mean_token_accuracy": 0.6942210704088211,
+      "num_tokens": 810774695.0,
+      "step": 390
+    },
+    {
+      "epoch": 0.5520614954577219,
+      "grad_norm": 0.35614827275276184,
+      "learning_rate": 1.9284234510365808e-05,
+      "loss": 0.9884,
+      "mean_token_accuracy": 0.6969966471195221,
+      "num_tokens": 821116961.0,
+      "step": 395
+    },
+    {
+      "epoch": 0.5590496156533893,
+      "grad_norm": 0.3040587604045868,
+      "learning_rate": 1.9259412756872843e-05,
+      "loss": 0.9894,
+      "mean_token_accuracy": 0.6961709499359131,
+      "num_tokens": 831489732.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.4788316786289215,
+      "learning_rate": 1.923419024165194e-05,
+      "loss": 1.0035,
+      "mean_token_accuracy": 0.6934469550848007,
+      "num_tokens": 841868264.0,
+      "step": 405
+    },
+    {
+      "epoch": 0.573025856044724,
+      "grad_norm": 0.26638856530189514,
+      "learning_rate": 1.9208568460131638e-05,
+      "loss": 0.988,
+      "mean_token_accuracy": 0.6969901770353317,
+      "num_tokens": 852276790.0,
+      "step": 410
+    },
+    {
+      "epoch": 0.5800139762403913,
+      "grad_norm": 0.29064929485321045,
+      "learning_rate": 1.918254893141276e-05,
+      "loss": 0.988,
+      "mean_token_accuracy": 0.6970887899398803,
+      "num_tokens": 862638829.0,
+      "step": 415
+    },
+    {
+      "epoch": 0.5870020964360587,
+      "grad_norm": 0.3532664179801941,
+      "learning_rate": 1.9156133198178325e-05,
+      "loss": 0.9957,
+      "mean_token_accuracy": 0.6949932664632797,
+      "num_tokens": 873044735.0,
+      "step": 420
+    },
+    {
+      "epoch": 0.593990216631726,
+      "grad_norm": 0.26006296277046204,
+      "learning_rate": 1.9129322826602093e-05,
+      "loss": 0.9834,
+      "mean_token_accuracy": 0.6978975415229798,
+      "num_tokens": 883461462.0,
+      "step": 425
+    },
+    {
+      "epoch": 0.6009783368273934,
+      "grad_norm": 0.30922138690948486,
+      "learning_rate": 1.9102119406255704e-05,
+      "loss": 0.9917,
+      "mean_token_accuracy": 0.6963249862194061,
+      "num_tokens": 893808723.0,
+      "step": 430
+    },
+    {
+      "epoch": 0.6079664570230608,
+      "grad_norm": 0.2605006694793701,
+      "learning_rate": 1.907452455001444e-05,
+      "loss": 0.9922,
+      "mean_token_accuracy": 0.6961820662021637,
+      "num_tokens": 904174818.0,
+      "step": 435
+    },
+    {
+      "epoch": 0.6149545772187281,
+      "grad_norm": 0.3060205578804016,
+      "learning_rate": 1.9046539893961596e-05,
+      "loss": 0.9858,
+      "mean_token_accuracy": 0.6980990767478943,
+      "num_tokens": 914558969.0,
+      "step": 440
+    },
+    {
+      "epoch": 0.6219426974143956,
+      "grad_norm": 0.3001151382923126,
+      "learning_rate": 1.9018167097291465e-05,
+      "loss": 1.0048,
+      "mean_token_accuracy": 0.6929802745580673,
+      "num_tokens": 924948007.0,
+      "step": 445
+    },
+    {
+      "epoch": 0.6289308176100629,
+      "grad_norm": 0.2708394527435303,
+      "learning_rate": 1.8989407842210986e-05,
+      "loss": 1.0065,
+      "mean_token_accuracy": 0.6926934719085693,
+      "num_tokens": 935357079.0,
+      "step": 450
+    },
+    {
+      "epoch": 0.6359189378057303,
+      "grad_norm": 0.29378563165664673,
+      "learning_rate": 1.8960263833839993e-05,
+      "loss": 0.991,
+      "mean_token_accuracy": 0.6963881731033326,
+      "num_tokens": 945773313.0,
+      "step": 455
+    },
+    {
+      "epoch": 0.6429070580013976,
+      "grad_norm": 0.34197095036506653,
+      "learning_rate": 1.8930736800110124e-05,
+      "loss": 0.9877,
+      "mean_token_accuracy": 0.6973118752241134,
+      "num_tokens": 956143314.0,
+      "step": 460
+    },
+    {
+      "epoch": 0.649895178197065,
+      "grad_norm": 0.30472254753112793,
+      "learning_rate": 1.8900828491662372e-05,
+      "loss": 0.9852,
+      "mean_token_accuracy": 0.6980044454336166,
+      "num_tokens": 966548893.0,
+      "step": 465
+    },
+    {
+      "epoch": 0.6568832983927324,
+      "grad_norm": 0.2430526465177536,
+      "learning_rate": 1.8870540681743288e-05,
+      "loss": 0.9842,
+      "mean_token_accuracy": 0.6978294104337692,
+      "num_tokens": 976941511.0,
+      "step": 470
+    },
+    {
+      "epoch": 0.6638714185883997,
+      "grad_norm": 0.25737932324409485,
+      "learning_rate": 1.8839875166099855e-05,
+      "loss": 0.9858,
+      "mean_token_accuracy": 0.6972843915224075,
+      "num_tokens": 987316088.0,
+      "step": 475
+    },
+    {
+      "epoch": 0.6708595387840671,
+      "grad_norm": 0.36384668946266174,
+      "learning_rate": 1.8808833762873006e-05,
+      "loss": 0.9855,
+      "mean_token_accuracy": 0.6972428739070893,
+      "num_tokens": 997698210.0,
+      "step": 480
+    },
+    {
+      "epoch": 0.6778476589797344,
+      "grad_norm": 0.253492146730423,
+      "learning_rate": 1.8777418312489834e-05,
+      "loss": 0.9875,
+      "mean_token_accuracy": 0.6970369219779968,
+      "num_tokens": 1008096298.0,
+      "step": 485
+    },
+    {
+      "epoch": 0.6848357791754018,
+      "grad_norm": 0.26920533180236816,
+      "learning_rate": 1.8745630677554486e-05,
+      "loss": 0.9993,
+      "mean_token_accuracy": 0.6944381654262543,
+      "num_tokens": 1018465735.0,
+      "step": 490
+    },
+    {
+      "epoch": 0.6918238993710691,
+      "grad_norm": 0.3240351974964142,
+      "learning_rate": 1.871347274273771e-05,
+      "loss": 0.983,
+      "mean_token_accuracy": 0.6982408493757248,
+      "num_tokens": 1028829779.0,
+      "step": 495
+    },
+    {
+      "epoch": 0.6988120195667366,
+      "grad_norm": 0.2850527763366699,
+      "learning_rate": 1.8680946414665117e-05,
+      "loss": 0.9924,
+      "mean_token_accuracy": 0.6957588851451874,
+      "num_tokens": 1039212930.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.705800139762404,
+      "grad_norm": 0.28837037086486816,
+      "learning_rate": 1.864805362180416e-05,
+      "loss": 0.9843,
+      "mean_token_accuracy": 0.6984927475452423,
+      "num_tokens": 1049622375.0,
+      "step": 505
+    },
+    {
+      "epoch": 0.7127882599580713,
+      "grad_norm": 0.3023139238357544,
+      "learning_rate": 1.8614796314349765e-05,
+      "loss": 0.9875,
+      "mean_token_accuracy": 0.6971228659152985,
+      "num_tokens": 1060021012.0,
+      "step": 510
+    },
+    {
+      "epoch": 0.7197763801537387,
+      "grad_norm": 0.2863965630531311,
+      "learning_rate": 1.8581176464108727e-05,
+      "loss": 1.0023,
+      "mean_token_accuracy": 0.6931641280651093,
+      "num_tokens": 1070415145.0,
+      "step": 515
+    },
+    {
+      "epoch": 0.726764500349406,
+      "grad_norm": 0.2860371470451355,
+      "learning_rate": 1.8547196064382798e-05,
+      "loss": 0.9867,
+      "mean_token_accuracy": 0.6977067708969116,
+      "num_tokens": 1080814626.0,
+      "step": 520
+    },
+    {
+      "epoch": 0.7337526205450734,
+      "grad_norm": 0.27839553356170654,
+      "learning_rate": 1.8512857129850502e-05,
+      "loss": 0.9974,
+      "mean_token_accuracy": 0.6940527439117432,
+      "num_tokens": 1091225187.0,
+      "step": 525
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.25351476669311523,
+      "learning_rate": 1.8478161696447686e-05,
+      "loss": 0.9935,
+      "mean_token_accuracy": 0.6956435739994049,
+      "num_tokens": 1101580066.0,
+      "step": 530
+    },
+    {
+      "epoch": 0.7477288609364081,
+      "grad_norm": 0.30183982849121094,
+      "learning_rate": 1.8443111821246808e-05,
+      "loss": 0.9759,
+      "mean_token_accuracy": 0.6996494948863983,
+      "num_tokens": 1111994907.0,
+      "step": 535
+    },
+    {
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.3375386893749237,
+      "learning_rate": 1.8407709582334986e-05,
+      "loss": 0.984,
+      "mean_token_accuracy": 0.698539337515831,
+      "num_tokens": 1122408439.0,
+      "step": 540
+    },
+    {
+      "epoch": 0.7617051013277428,
+      "grad_norm": 0.2689567804336548,
+      "learning_rate": 1.8371957078690772e-05,
+      "loss": 0.9839,
+      "mean_token_accuracy": 0.6976371347904206,
+      "num_tokens": 1132816803.0,
+      "step": 545
+    },
+    {
+      "epoch": 0.7686932215234102,
+      "grad_norm": 0.30557841062545776,
+      "learning_rate": 1.833585643005973e-05,
+      "loss": 0.9953,
+      "mean_token_accuracy": 0.6952255100011826,
+      "num_tokens": 1143223094.0,
+      "step": 550
+    },
+    {
+      "epoch": 0.7756813417190775,
+      "grad_norm": 0.24475328624248505,
+      "learning_rate": 1.8299409776828726e-05,
+      "loss": 0.9856,
+      "mean_token_accuracy": 0.6971423864364624,
+      "num_tokens": 1153614904.0,
+      "step": 555
+    },
+    {
+      "epoch": 0.782669461914745,
+      "grad_norm": 0.2633828818798065,
+      "learning_rate": 1.8262619279899045e-05,
+      "loss": 0.983,
+      "mean_token_accuracy": 0.6979476481676101,
+      "num_tokens": 1164008710.0,
+      "step": 560
+    },
+    {
+      "epoch": 0.7896575821104123,
+      "grad_norm": 0.31182652711868286,
+      "learning_rate": 1.8225487120558275e-05,
+      "loss": 0.9894,
+      "mean_token_accuracy": 0.6967566043138504,
+      "num_tokens": 1174400555.0,
+      "step": 565
+    },
+    {
+      "epoch": 0.7966457023060797,
+      "grad_norm": 0.26585254073143005,
+      "learning_rate": 1.8188015500350968e-05,
+      "loss": 0.9907,
+      "mean_token_accuracy": 0.6964362800121308,
+      "num_tokens": 1184789828.0,
+      "step": 570
+    },
+    {
+      "epoch": 0.803633822501747,
+      "grad_norm": 0.30836501717567444,
+      "learning_rate": 1.8150206640948114e-05,
+      "loss": 0.9882,
+      "mean_token_accuracy": 0.6964727938175201,
+      "num_tokens": 1195143536.0,
+      "step": 575
+    },
+    {
+      "epoch": 0.8106219426974144,
+      "grad_norm": 0.23804894089698792,
+      "learning_rate": 1.811206278401544e-05,
+      "loss": 1.0016,
+      "mean_token_accuracy": 0.6934799730777741,
+      "num_tokens": 1205515568.0,
+      "step": 580
+    },
+    {
+      "epoch": 0.8176100628930818,
+      "grad_norm": 0.24671317636966705,
+      "learning_rate": 1.807358619108046e-05,
+      "loss": 0.9917,
+      "mean_token_accuracy": 0.6958341568708419,
+      "num_tokens": 1215941400.0,
+      "step": 585
+    },
+    {
+      "epoch": 0.8245981830887491,
+      "grad_norm": 0.2816368341445923,
+      "learning_rate": 1.803477914339843e-05,
+      "loss": 0.9848,
+      "mean_token_accuracy": 0.6971950948238372,
+      "num_tokens": 1226344021.0,
+      "step": 590
+    },
+    {
+      "epoch": 0.8315863032844165,
+      "grad_norm": 0.26695674657821655,
+      "learning_rate": 1.7995643941817088e-05,
+      "loss": 0.9979,
+      "mean_token_accuracy": 0.6942486584186554,
+      "num_tokens": 1236748640.0,
+      "step": 595
+    },
+    {
+      "epoch": 0.8385744234800838,
+      "grad_norm": 0.2597620487213135,
+      "learning_rate": 1.795618290664021e-05,
+      "loss": 0.9846,
+      "mean_token_accuracy": 0.6978382110595703,
+      "num_tokens": 1247150254.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.8455625436757512,
+      "grad_norm": 0.2926052212715149,
+      "learning_rate": 1.7916398377490073e-05,
+      "loss": 0.9839,
+      "mean_token_accuracy": 0.6982168465852737,
+      "num_tokens": 1257534085.0,
+      "step": 605
+    },
+    {
+      "epoch": 0.8525506638714185,
+      "grad_norm": 0.2637452185153961,
+      "learning_rate": 1.7876292713168723e-05,
+      "loss": 0.9935,
+      "mean_token_accuracy": 0.695600014925003,
+      "num_tokens": 1267918085.0,
+      "step": 610
+    },
+    {
+      "epoch": 0.859538784067086,
+      "grad_norm": 0.29727014899253845,
+      "learning_rate": 1.7835868291518128e-05,
+      "loss": 0.9866,
+      "mean_token_accuracy": 0.6969569563865662,
+      "num_tokens": 1278294024.0,
+      "step": 615
+    },
+    {
+      "epoch": 0.8665269042627534,
+      "grad_norm": 0.2716880440711975,
+      "learning_rate": 1.77951275092792e-05,
+      "loss": 0.9826,
+      "mean_token_accuracy": 0.697507557272911,
+      "num_tokens": 1288661352.0,
+      "step": 620
+    },
+    {
+      "epoch": 0.8735150244584207,
+      "grad_norm": 0.2710786759853363,
+      "learning_rate": 1.7754072781949683e-05,
+      "loss": 0.9957,
+      "mean_token_accuracy": 0.6947832137346268,
+      "num_tokens": 1299045987.0,
+      "step": 625
+    },
+    {
+      "epoch": 0.8805031446540881,
+      "grad_norm": 0.24244153499603271,
+      "learning_rate": 1.771270654364095e-05,
+      "loss": 0.9912,
+      "mean_token_accuracy": 0.6964092016220093,
+      "num_tokens": 1309404095.0,
+      "step": 630
+    },
+    {
+      "epoch": 0.8874912648497554,
+      "grad_norm": 0.2675671875476837,
+      "learning_rate": 1.7671031246933684e-05,
+      "loss": 0.9907,
+      "mean_token_accuracy": 0.6958764642477036,
+      "num_tokens": 1319817107.0,
+      "step": 635
+    },
+    {
+      "epoch": 0.8944793850454228,
+      "grad_norm": 0.24089151620864868,
+      "learning_rate": 1.762904936273246e-05,
+      "loss": 0.9827,
+      "mean_token_accuracy": 0.6984389781951904,
+      "num_tokens": 1330221560.0,
+      "step": 640
+    },
+    {
+      "epoch": 0.9014675052410901,
+      "grad_norm": 0.27282679080963135,
+      "learning_rate": 1.7586763380119257e-05,
+      "loss": 0.9927,
+      "mean_token_accuracy": 0.6955932974815369,
+      "num_tokens": 1340634686.0,
+      "step": 645
+    },
+    {
+      "epoch": 0.9084556254367575,
+      "grad_norm": 0.27260783314704895,
+      "learning_rate": 1.7544175806205866e-05,
+      "loss": 0.9903,
+      "mean_token_accuracy": 0.6963932275772095,
+      "num_tokens": 1351003877.0,
+      "step": 650
+    },
+    {
+      "epoch": 0.9154437456324249,
+      "grad_norm": 0.2771081030368805,
+      "learning_rate": 1.7501289165985262e-05,
+      "loss": 0.9834,
+      "mean_token_accuracy": 0.6979602754116059,
+      "num_tokens": 1361402284.0,
+      "step": 655
+    },
+    {
+      "epoch": 0.9224318658280922,
+      "grad_norm": 0.2612958550453186,
+      "learning_rate": 1.745810600218189e-05,
+      "loss": 0.9935,
+      "mean_token_accuracy": 0.6954259991645813,
+      "num_tokens": 1371789396.0,
+      "step": 660
+    },
+    {
+      "epoch": 0.9294199860237596,
+      "grad_norm": 0.2653946578502655,
+      "learning_rate": 1.7414628875100903e-05,
+      "loss": 0.993,
+      "mean_token_accuracy": 0.6953835368156434,
+      "num_tokens": 1382207353.0,
+      "step": 665
+    },
+    {
+      "epoch": 0.9364081062194269,
+      "grad_norm": 0.2772393524646759,
+      "learning_rate": 1.7370860362476376e-05,
+      "loss": 0.9718,
+      "mean_token_accuracy": 0.7005853295326233,
+      "num_tokens": 1392570580.0,
+      "step": 670
+    },
+    {
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.24524082243442535,
+      "learning_rate": 1.732680305931847e-05,
+      "loss": 0.982,
+      "mean_token_accuracy": 0.6978916674852371,
+      "num_tokens": 1402973016.0,
+      "step": 675
+    },
+    {
+      "epoch": 0.9503843466107617,
+      "grad_norm": 0.23882606625556946,
+      "learning_rate": 1.7282459577759558e-05,
+      "loss": 0.994,
+      "mean_token_accuracy": 0.6950913339853286,
+      "num_tokens": 1413391023.0,
+      "step": 680
+    },
+    {
+      "epoch": 0.9573724668064291,
+      "grad_norm": 0.2698175013065338,
+      "learning_rate": 1.7237832546899383e-05,
+      "loss": 0.9764,
+      "mean_token_accuracy": 0.7002421110868454,
+      "num_tokens": 1423756209.0,
+      "step": 685
+    },
+    {
+      "epoch": 0.9643605870020965,
+      "grad_norm": 0.24084943532943726,
+      "learning_rate": 1.719292461264915e-05,
+      "loss": 0.9921,
+      "mean_token_accuracy": 0.6954464077949524,
+      "num_tokens": 1434141449.0,
+      "step": 690
+    },
+    {
+      "epoch": 0.9713487071977638,
+      "grad_norm": 0.2785717248916626,
+      "learning_rate": 1.7147738437574682e-05,
+      "loss": 0.9783,
+      "mean_token_accuracy": 0.699154207110405,
+      "num_tokens": 1444495476.0,
+      "step": 695
+    },
+    {
+      "epoch": 0.9783368273934312,
+      "grad_norm": 0.32852408289909363,
+      "learning_rate": 1.710227670073852e-05,
+      "loss": 0.9926,
+      "mean_token_accuracy": 0.6958128333091735,
+      "num_tokens": 1454908018.0,
+      "step": 700
+    },
+    {
+      "epoch": 0.9853249475890985,
+      "grad_norm": 0.2404838651418686,
+      "learning_rate": 1.705654209754113e-05,
+      "loss": 0.9741,
+      "mean_token_accuracy": 0.7001296132802963,
+      "num_tokens": 1465317940.0,
+      "step": 705
+    },
+    {
+      "epoch": 0.9923130677847659,
+      "grad_norm": 0.2792597711086273,
+      "learning_rate": 1.701053733956105e-05,
+      "loss": 0.9798,
+      "mean_token_accuracy": 0.6991855531930924,
+      "num_tokens": 1475718987.0,
+      "step": 710
+    },
+    {
+      "epoch": 0.9993011879804332,
+      "grad_norm": 0.28146040439605713,
+      "learning_rate": 1.696426515439414e-05,
+      "loss": 0.9803,
+      "mean_token_accuracy": 0.6980899155139924,
+      "num_tokens": 1486112797.0,
+      "step": 715
+    },
+    {
+      "epoch": 1.0055904961565338,
+      "grad_norm": 0.4210590422153473,
+      "learning_rate": 1.691772828549189e-05,
+      "loss": 0.9511,
+      "mean_token_accuracy": 0.7056844896740384,
+      "num_tokens": 1495486842.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.0125786163522013,
+      "grad_norm": 0.40947115421295166,
+      "learning_rate": 1.687092949199871e-05,
+      "loss": 0.9641,
+      "mean_token_accuracy": 0.701970437169075,
+      "num_tokens": 1505887963.0,
+      "step": 725
+    },
+    {
+      "epoch": 1.0195667365478687,
+      "grad_norm": 0.34101659059524536,
+      "learning_rate": 1.682387154858838e-05,
+      "loss": 0.9542,
+      "mean_token_accuracy": 0.7043519228696823,
+      "num_tokens": 1516294528.0,
+      "step": 730
+    },
+    {
+      "epoch": 1.026554856743536,
+      "grad_norm": 0.2541981041431427,
+      "learning_rate": 1.6776557245299532e-05,
+      "loss": 0.9457,
+      "mean_token_accuracy": 0.7069508880376816,
+      "num_tokens": 1526652289.0,
+      "step": 735
+    },
+    {
+      "epoch": 1.0335429769392033,
+      "grad_norm": 0.248768150806427,
+      "learning_rate": 1.672898938737023e-05,
+      "loss": 0.9473,
+      "mean_token_accuracy": 0.7058848321437836,
+      "num_tokens": 1537056955.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0405310971348707,
+      "grad_norm": 0.2375871241092682,
+      "learning_rate": 1.668117079507164e-05,
+      "loss": 0.9546,
+      "mean_token_accuracy": 0.7042842835187912,
+      "num_tokens": 1547469838.0,
+      "step": 745
+    },
+    {
+      "epoch": 1.047519217330538,
+      "grad_norm": 0.24888072907924652,
+      "learning_rate": 1.6633104303540842e-05,
+      "loss": 0.9358,
+      "mean_token_accuracy": 0.7089408129453659,
+      "num_tokens": 1557844495.0,
+      "step": 750
+    },
+    {
+      "epoch": 1.0545073375262055,
+      "grad_norm": 0.2507837116718292,
+      "learning_rate": 1.6584792762612706e-05,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7055849075317383,
+      "num_tokens": 1568244116.0,
+      "step": 755
+    },
+    {
+      "epoch": 1.0614954577218727,
+      "grad_norm": 0.2577778100967407,
+      "learning_rate": 1.6536239036650942e-05,
+      "loss": 0.9548,
+      "mean_token_accuracy": 0.7038821280002594,
+      "num_tokens": 1578606457.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.0684835779175401,
+      "grad_norm": 0.24672797322273254,
+      "learning_rate": 1.6487446004378273e-05,
+      "loss": 0.9427,
+      "mean_token_accuracy": 0.7073481321334839,
+      "num_tokens": 1588972856.0,
+      "step": 765
+    },
+    {
+      "epoch": 1.0754716981132075,
+      "grad_norm": 0.21917957067489624,
+      "learning_rate": 1.6438416558705767e-05,
+      "loss": 0.9592,
+      "mean_token_accuracy": 0.7032468438148498,
+      "num_tokens": 1599361363.0,
+      "step": 770
+    },
+    {
+      "epoch": 1.082459818308875,
+      "grad_norm": 0.2600899934768677,
+      "learning_rate": 1.6389153606561293e-05,
+      "loss": 0.9466,
+      "mean_token_accuracy": 0.7060303270816803,
+      "num_tokens": 1609753390.0,
+      "step": 775
+    },
+    {
+      "epoch": 1.0894479385045424,
+      "grad_norm": 0.24663616716861725,
+      "learning_rate": 1.6339660068717197e-05,
+      "loss": 0.9567,
+      "mean_token_accuracy": 0.7034341752529144,
+      "num_tokens": 1620132659.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.0964360587002095,
+      "grad_norm": 0.30449482798576355,
+      "learning_rate": 1.6289938879617114e-05,
+      "loss": 0.9377,
+      "mean_token_accuracy": 0.7081806808710098,
+      "num_tokens": 1630536804.0,
+      "step": 785
+    },
+    {
+      "epoch": 1.103424178895877,
+      "grad_norm": 0.3289632797241211,
+      "learning_rate": 1.623999298720199e-05,
+      "loss": 0.9518,
+      "mean_token_accuracy": 0.7046901375055313,
+      "num_tokens": 1640935243.0,
+      "step": 790
+    },
+    {
+      "epoch": 1.1104122990915444,
+      "grad_norm": 0.26432016491889954,
+      "learning_rate": 1.618982535273531e-05,
+      "loss": 0.9463,
+      "mean_token_accuracy": 0.7065684378147126,
+      "num_tokens": 1651324542.0,
+      "step": 795
+    },
+    {
+      "epoch": 1.1174004192872118,
+      "grad_norm": 0.243063285946846,
+      "learning_rate": 1.6139438950627513e-05,
+      "loss": 0.9534,
+      "mean_token_accuracy": 0.7047319650650025,
+      "num_tokens": 1661713439.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.1243885394828792,
+      "grad_norm": 0.24993912875652313,
+      "learning_rate": 1.608883676825965e-05,
+      "loss": 0.9484,
+      "mean_token_accuracy": 0.7057817548513412,
+      "num_tokens": 1672104702.0,
+      "step": 805
+    },
+    {
+      "epoch": 1.1313766596785464,
+      "grad_norm": 0.2645328640937805,
+      "learning_rate": 1.6038021805806257e-05,
+      "loss": 0.951,
+      "mean_token_accuracy": 0.7048044502735138,
+      "num_tokens": 1682485345.0,
+      "step": 810
+    },
+    {
+      "epoch": 1.1383647798742138,
+      "grad_norm": 0.22507497668266296,
+      "learning_rate": 1.5986997076057477e-05,
+      "loss": 0.9533,
+      "mean_token_accuracy": 0.7048071801662446,
+      "num_tokens": 1692860107.0,
+      "step": 815
+    },
+    {
+      "epoch": 1.1453529000698812,
+      "grad_norm": 0.3045750856399536,
+      "learning_rate": 1.593576560424045e-05,
+      "loss": 0.9445,
+      "mean_token_accuracy": 0.7067113369703293,
+      "num_tokens": 1703270041.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.1523410202655486,
+      "grad_norm": 0.2575872540473938,
+      "learning_rate": 1.588433042783992e-05,
+      "loss": 0.958,
+      "mean_token_accuracy": 0.7032164961099625,
+      "num_tokens": 1713664308.0,
+      "step": 825
+    },
+    {
+      "epoch": 1.159329140461216,
+      "grad_norm": 0.2729121148586273,
+      "learning_rate": 1.5832694596418174e-05,
+      "loss": 0.9398,
+      "mean_token_accuracy": 0.7079699605703353,
+      "num_tokens": 1724045480.0,
+      "step": 830
+    },
+    {
+      "epoch": 1.1663172606568832,
+      "grad_norm": 0.23945994675159454,
+      "learning_rate": 1.57808611714342e-05,
+      "loss": 0.9472,
+      "mean_token_accuracy": 0.7062007933855057,
+      "num_tokens": 1734427138.0,
+      "step": 835
+    },
+    {
+      "epoch": 1.1733053808525507,
+      "grad_norm": 0.2659391164779663,
+      "learning_rate": 1.5728833226062216e-05,
+      "loss": 0.952,
+      "mean_token_accuracy": 0.7049013942480087,
+      "num_tokens": 1744839569.0,
+      "step": 840
+    },
+    {
+      "epoch": 1.180293501048218,
+      "grad_norm": 0.28336286544799805,
+      "learning_rate": 1.5676613845009434e-05,
+      "loss": 0.9489,
+      "mean_token_accuracy": 0.7052598625421524,
+      "num_tokens": 1755242100.0,
+      "step": 845
+    },
+    {
+      "epoch": 1.1872816212438855,
+      "grad_norm": 0.2740361988544464,
+      "learning_rate": 1.562420612433318e-05,
+      "loss": 0.9585,
+      "mean_token_accuracy": 0.7030707478523255,
+      "num_tokens": 1765621699.0,
+      "step": 850
+    },
+    {
+      "epoch": 1.1942697414395527,
+      "grad_norm": 0.2763083577156067,
+      "learning_rate": 1.557161317125733e-05,
+      "loss": 0.9527,
+      "mean_token_accuracy": 0.7051246583461761,
+      "num_tokens": 1776038202.0,
+      "step": 855
+    },
+    {
+      "epoch": 1.20125786163522,
+      "grad_norm": 0.28946614265441895,
+      "learning_rate": 1.5518838103988075e-05,
+      "loss": 0.9455,
+      "mean_token_accuracy": 0.7068380653858185,
+      "num_tokens": 1786417813.0,
+      "step": 860
+    },
+    {
+      "epoch": 1.2082459818308875,
+      "grad_norm": 0.26109829545021057,
+      "learning_rate": 1.546588405152907e-05,
+      "loss": 0.9527,
+      "mean_token_accuracy": 0.704670849442482,
+      "num_tokens": 1796827043.0,
+      "step": 865
+    },
+    {
+      "epoch": 1.215234102026555,
+      "grad_norm": 0.24084337055683136,
+      "learning_rate": 1.5412754153495882e-05,
+      "loss": 0.959,
+      "mean_token_accuracy": 0.7029285609722138,
+      "num_tokens": 1807231537.0,
+      "step": 870
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 0.25638347864151,
+      "learning_rate": 1.5359451559929873e-05,
+      "loss": 0.9648,
+      "mean_token_accuracy": 0.7019408881664276,
+      "num_tokens": 1817615778.0,
+      "step": 875
+    },
+    {
+      "epoch": 1.2292103424178895,
+      "grad_norm": 0.2681230306625366,
+      "learning_rate": 1.5305979431111416e-05,
+      "loss": 0.9603,
+      "mean_token_accuracy": 0.7021995514631272,
+      "num_tokens": 1828001448.0,
+      "step": 880
+    },
+    {
+      "epoch": 1.236198462613557,
+      "grad_norm": 0.24295032024383545,
+      "learning_rate": 1.5252340937372542e-05,
+      "loss": 0.9682,
+      "mean_token_accuracy": 0.7013109654188157,
+      "num_tokens": 1838409867.0,
+      "step": 885
+    },
+    {
+      "epoch": 1.2431865828092243,
+      "grad_norm": 0.23449333012104034,
+      "learning_rate": 1.5198539258908955e-05,
+      "loss": 0.9408,
+      "mean_token_accuracy": 0.7074013233184815,
+      "num_tokens": 1848787748.0,
+      "step": 890
+    },
+    {
+      "epoch": 1.2501747030048918,
+      "grad_norm": 0.24246180057525635,
+      "learning_rate": 1.5144577585591496e-05,
+      "loss": 0.9672,
+      "mean_token_accuracy": 0.7012275934219361,
+      "num_tokens": 1859159875.0,
+      "step": 895
+    },
+    {
+      "epoch": 1.257162823200559,
+      "grad_norm": 0.23688940703868866,
+      "learning_rate": 1.5090459116777e-05,
+      "loss": 0.9555,
+      "mean_token_accuracy": 0.7037541270256042,
+      "num_tokens": 1869540000.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.2641509433962264,
+      "grad_norm": 0.229180246591568,
+      "learning_rate": 1.503618706111863e-05,
+      "loss": 0.9591,
+      "mean_token_accuracy": 0.7028054356575012,
+      "num_tokens": 1879933756.0,
+      "step": 905
+    },
+    {
+      "epoch": 1.2711390635918938,
+      "grad_norm": 0.22831909358501434,
+      "learning_rate": 1.498176463637561e-05,
+      "loss": 0.9504,
+      "mean_token_accuracy": 0.7053441524505615,
+      "num_tokens": 1890307582.0,
+      "step": 910
+    },
+    {
+      "epoch": 1.2781271837875612,
+      "grad_norm": 0.2654498219490051,
+      "learning_rate": 1.492719506922248e-05,
+      "loss": 0.9595,
+      "mean_token_accuracy": 0.7033340811729432,
+      "num_tokens": 1900666240.0,
+      "step": 915
+    },
+    {
+      "epoch": 1.2851153039832286,
+      "grad_norm": 0.23844845592975616,
+      "learning_rate": 1.487248159505775e-05,
+      "loss": 0.9571,
+      "mean_token_accuracy": 0.7038801491260529,
+      "num_tokens": 1911048901.0,
+      "step": 920
+    },
+    {
+      "epoch": 1.292103424178896,
+      "grad_norm": 0.21731877326965332,
+      "learning_rate": 1.4817627457812107e-05,
+      "loss": 0.9571,
+      "mean_token_accuracy": 0.703849372267723,
+      "num_tokens": 1921390824.0,
+      "step": 925
+    },
+    {
+      "epoch": 1.2990915443745632,
+      "grad_norm": 0.23090894520282745,
+      "learning_rate": 1.4762635909756071e-05,
+      "loss": 0.9514,
+      "mean_token_accuracy": 0.7047541320323945,
+      "num_tokens": 1931805342.0,
+      "step": 930
+    },
+    {
+      "epoch": 1.3060796645702306,
+      "grad_norm": 0.2666736841201782,
+      "learning_rate": 1.4707510211307165e-05,
+      "loss": 0.9587,
+      "mean_token_accuracy": 0.7034244388341904,
+      "num_tokens": 1942199423.0,
+      "step": 935
+    },
+    {
+      "epoch": 1.313067784765898,
+      "grad_norm": 0.2366342693567276,
+      "learning_rate": 1.4652253630836622e-05,
+      "loss": 0.959,
+      "mean_token_accuracy": 0.7027652233839035,
+      "num_tokens": 1952598729.0,
+      "step": 940
+    },
+    {
+      "epoch": 1.3200559049615652,
+      "grad_norm": 0.24025586247444153,
+      "learning_rate": 1.4596869444475582e-05,
+      "loss": 0.9474,
+      "mean_token_accuracy": 0.7063993692398072,
+      "num_tokens": 1962973528.0,
+      "step": 945
+    },
+    {
+      "epoch": 1.3270440251572326,
+      "grad_norm": 0.24174918234348297,
+      "learning_rate": 1.4541360935920874e-05,
+      "loss": 0.9602,
+      "mean_token_accuracy": 0.7028678983449936,
+      "num_tokens": 1973395017.0,
+      "step": 950
+    },
+    {
+      "epoch": 1.3340321453529,
+      "grad_norm": 0.2554982304573059,
+      "learning_rate": 1.448573139624032e-05,
+      "loss": 0.9548,
+      "mean_token_accuracy": 0.7040337771177292,
+      "num_tokens": 1983785157.0,
+      "step": 955
+    },
+    {
+      "epoch": 1.3410202655485675,
+      "grad_norm": 0.2739647924900055,
+      "learning_rate": 1.4429984123677605e-05,
+      "loss": 0.9589,
+      "mean_token_accuracy": 0.7027077376842499,
+      "num_tokens": 1994207711.0,
+      "step": 960
+    },
+    {
+      "epoch": 1.3480083857442349,
+      "grad_norm": 0.3043140769004822,
+      "learning_rate": 1.4374122423456731e-05,
+      "loss": 0.9661,
+      "mean_token_accuracy": 0.7012409687042236,
+      "num_tokens": 2004589901.0,
+      "step": 965
+    },
+    {
+      "epoch": 1.3549965059399023,
+      "grad_norm": 0.28174853324890137,
+      "learning_rate": 1.4318149607586052e-05,
+      "loss": 0.9654,
+      "mean_token_accuracy": 0.7013572067022323,
+      "num_tokens": 2014949898.0,
+      "step": 970
+    },
+    {
+      "epoch": 1.3619846261355695,
+      "grad_norm": 0.29067397117614746,
+      "learning_rate": 1.4262068994661902e-05,
+      "loss": 0.9541,
+      "mean_token_accuracy": 0.7043993294239044,
+      "num_tokens": 2025331279.0,
+      "step": 975
+    },
+    {
+      "epoch": 1.368972746331237,
+      "grad_norm": 0.2759871184825897,
+      "learning_rate": 1.4205883909671838e-05,
+      "loss": 0.947,
+      "mean_token_accuracy": 0.705754178762436,
+      "num_tokens": 2035743276.0,
+      "step": 980
+    },
+    {
+      "epoch": 1.3759608665269043,
+      "grad_norm": 0.25539323687553406,
+      "learning_rate": 1.4149597683797516e-05,
+      "loss": 0.9456,
+      "mean_token_accuracy": 0.706447896361351,
+      "num_tokens": 2046100659.0,
+      "step": 985
+    },
+    {
+      "epoch": 1.3829489867225715,
+      "grad_norm": 0.23867954313755035,
+      "learning_rate": 1.409321365421717e-05,
+      "loss": 0.9492,
+      "mean_token_accuracy": 0.7054832637310028,
+      "num_tokens": 2056458977.0,
+      "step": 990
+    },
+    {
+      "epoch": 1.389937106918239,
+      "grad_norm": 0.2444266527891159,
+      "learning_rate": 1.4036735163907765e-05,
+      "loss": 0.9584,
+      "mean_token_accuracy": 0.7032276690006256,
+      "num_tokens": 2066827483.0,
+      "step": 995
+    },
+    {
+      "epoch": 1.3969252271139063,
+      "grad_norm": 0.29059308767318726,
+      "learning_rate": 1.3980165561446781e-05,
+      "loss": 0.9503,
+      "mean_token_accuracy": 0.705039718747139,
+      "num_tokens": 2077249640.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4039133473095737,
+      "grad_norm": 0.2778762876987457,
+      "learning_rate": 1.3923508200813695e-05,
+      "loss": 0.9518,
+      "mean_token_accuracy": 0.7050242900848389,
+      "num_tokens": 2087646218.0,
+      "step": 1005
+    },
+    {
+      "epoch": 1.4109014675052411,
+      "grad_norm": 0.2580850124359131,
+      "learning_rate": 1.3866766441191108e-05,
+      "loss": 0.9534,
+      "mean_token_accuracy": 0.7048745006322861,
+      "num_tokens": 2098032107.0,
+      "step": 1010
+    },
+    {
+      "epoch": 1.4178895877009086,
+      "grad_norm": 0.24883116781711578,
+      "learning_rate": 1.3809943646765594e-05,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.7045016646385193,
+      "num_tokens": 2108394030.0,
+      "step": 1015
+    },
+    {
+      "epoch": 1.4248777078965758,
+      "grad_norm": 0.24177159368991852,
+      "learning_rate": 1.3753043186528228e-05,
+      "loss": 0.9628,
+      "mean_token_accuracy": 0.702374005317688,
+      "num_tokens": 2118815629.0,
+      "step": 1020
+    },
+    {
+      "epoch": 1.4318658280922432,
+      "grad_norm": 0.29814979434013367,
+      "learning_rate": 1.3696068434074858e-05,
+      "loss": 0.9427,
+      "mean_token_accuracy": 0.7067977696657181,
+      "num_tokens": 2129188452.0,
+      "step": 1025
+    },
+    {
+      "epoch": 1.4388539482879106,
+      "grad_norm": 0.29647916555404663,
+      "learning_rate": 1.3639022767406065e-05,
+      "loss": 0.9514,
+      "mean_token_accuracy": 0.7053138375282287,
+      "num_tokens": 2139616248.0,
+      "step": 1030
+    },
+    {
+      "epoch": 1.445842068483578,
+      "grad_norm": 0.26481950283050537,
+      "learning_rate": 1.3581909568726897e-05,
+      "loss": 0.9507,
+      "mean_token_accuracy": 0.7051736801862717,
+      "num_tokens": 2150019652.0,
+      "step": 1035
+    },
+    {
+      "epoch": 1.4528301886792452,
+      "grad_norm": 0.2721223533153534,
+      "learning_rate": 1.3524732224246334e-05,
+      "loss": 0.9591,
+      "mean_token_accuracy": 0.703093808889389,
+      "num_tokens": 2160401014.0,
+      "step": 1040
+    },
+    {
+      "epoch": 1.4598183088749126,
+      "grad_norm": 0.2551819384098053,
+      "learning_rate": 1.3467494123976526e-05,
+      "loss": 0.952,
+      "mean_token_accuracy": 0.7046844661235809,
+      "num_tokens": 2170810669.0,
+      "step": 1045
+    },
+    {
+      "epoch": 1.46680642907058,
+      "grad_norm": 0.23688183724880219,
+      "learning_rate": 1.3410198661531802e-05,
+      "loss": 0.963,
+      "mean_token_accuracy": 0.7021675914525985,
+      "num_tokens": 2181208872.0,
+      "step": 1050
+    },
+    {
+      "epoch": 1.4737945492662474,
+      "grad_norm": 0.2515881359577179,
+      "learning_rate": 1.3352849233927458e-05,
+      "loss": 0.9495,
+      "mean_token_accuracy": 0.7059559196233749,
+      "num_tokens": 2191619229.0,
+      "step": 1055
+    },
+    {
+      "epoch": 1.4807826694619148,
+      "grad_norm": 0.24117465317249298,
+      "learning_rate": 1.3295449241378358e-05,
+      "loss": 0.9421,
+      "mean_token_accuracy": 0.707347297668457,
+      "num_tokens": 2202012017.0,
+      "step": 1060
+    },
+    {
+      "epoch": 1.4877707896575822,
+      "grad_norm": 0.21908260881900787,
+      "learning_rate": 1.3238002087097321e-05,
+      "loss": 0.9489,
+      "mean_token_accuracy": 0.7059781879186631,
+      "num_tokens": 2212374750.0,
+      "step": 1065
+    },
+    {
+      "epoch": 1.4947589098532494,
+      "grad_norm": 0.23432235419750214,
+      "learning_rate": 1.3180511177093371e-05,
+      "loss": 0.9455,
+      "mean_token_accuracy": 0.7061308234930038,
+      "num_tokens": 2222759914.0,
+      "step": 1070
+    },
+    {
+      "epoch": 1.5017470300489169,
+      "grad_norm": 0.24842332303524017,
+      "learning_rate": 1.312297991996978e-05,
+      "loss": 0.9539,
+      "mean_token_accuracy": 0.7039518356323242,
+      "num_tokens": 2233170092.0,
+      "step": 1075
+    },
+    {
+      "epoch": 1.508735150244584,
+      "grad_norm": 0.23463116586208344,
+      "learning_rate": 1.3065411726721975e-05,
+      "loss": 0.9455,
+      "mean_token_accuracy": 0.7066747874021531,
+      "num_tokens": 2243582510.0,
+      "step": 1080
+    },
+    {
+      "epoch": 1.5157232704402515,
+      "grad_norm": 0.22929295897483826,
+      "learning_rate": 1.300781001053531e-05,
+      "loss": 0.9607,
+      "mean_token_accuracy": 0.7026191800832748,
+      "num_tokens": 2253976158.0,
+      "step": 1085
+    },
+    {
+      "epoch": 1.5227113906359189,
+      "grad_norm": 0.22391878068447113,
+      "learning_rate": 1.2950178186582685e-05,
+      "loss": 0.9491,
+      "mean_token_accuracy": 0.7051207542419433,
+      "num_tokens": 2264388639.0,
+      "step": 1090
+    },
+    {
+      "epoch": 1.5296995108315863,
+      "grad_norm": 0.2603966295719147,
+      "learning_rate": 1.289251967182208e-05,
+      "loss": 0.9541,
+      "mean_token_accuracy": 0.7041330695152282,
+      "num_tokens": 2274779483.0,
+      "step": 1095
+    },
+    {
+      "epoch": 1.5366876310272537,
+      "grad_norm": 0.230450838804245,
+      "learning_rate": 1.2834837884793963e-05,
+      "loss": 0.9528,
+      "mean_token_accuracy": 0.7049372524023056,
+      "num_tokens": 2285190622.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.543675751222921,
+      "grad_norm": 0.21797870099544525,
+      "learning_rate": 1.277713624541859e-05,
+      "loss": 0.9592,
+      "mean_token_accuracy": 0.7032217621803284,
+      "num_tokens": 2295617562.0,
+      "step": 1105
+    },
+    {
+      "epoch": 1.5506638714185885,
+      "grad_norm": 0.2613143026828766,
+      "learning_rate": 1.2719418174793256e-05,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7029330909252167,
+      "num_tokens": 2306005585.0,
+      "step": 1110
+    },
+    {
+      "epoch": 1.5576519916142557,
+      "grad_norm": 0.23213548958301544,
+      "learning_rate": 1.2661687094989457e-05,
+      "loss": 0.9587,
+      "mean_token_accuracy": 0.7030121952295303,
+      "num_tokens": 2316399640.0,
+      "step": 1115
+    },
+    {
+      "epoch": 1.5646401118099231,
+      "grad_norm": 0.2354755848646164,
+      "learning_rate": 1.2603946428849995e-05,
+      "loss": 0.9503,
+      "mean_token_accuracy": 0.7047839015722275,
+      "num_tokens": 2326822403.0,
+      "step": 1120
+    },
+    {
+      "epoch": 1.5716282320055905,
+      "grad_norm": 0.2406359165906906,
+      "learning_rate": 1.2546199599786042e-05,
+      "loss": 0.9551,
+      "mean_token_accuracy": 0.704132291674614,
+      "num_tokens": 2337192629.0,
+      "step": 1125
+    },
+    {
+      "epoch": 1.5786163522012577,
+      "grad_norm": 0.2363503873348236,
+      "learning_rate": 1.248845003157416e-05,
+      "loss": 0.9579,
+      "mean_token_accuracy": 0.7029750674962998,
+      "num_tokens": 2347588209.0,
+      "step": 1130
+    },
+    {
+      "epoch": 1.5856044723969251,
+      "grad_norm": 0.23799030482769012,
+      "learning_rate": 1.2430701148153322e-05,
+      "loss": 0.9519,
+      "mean_token_accuracy": 0.7047583341598511,
+      "num_tokens": 2357969391.0,
+      "step": 1135
+    },
+    {
+      "epoch": 1.5925925925925926,
+      "grad_norm": 0.253959983587265,
+      "learning_rate": 1.237295637342189e-05,
+      "loss": 0.9422,
+      "mean_token_accuracy": 0.7074365675449371,
+      "num_tokens": 2368344869.0,
+      "step": 1140
+    },
+    {
+      "epoch": 1.59958071278826,
+      "grad_norm": 0.27008169889450073,
+      "learning_rate": 1.2315219131034637e-05,
+      "loss": 0.955,
+      "mean_token_accuracy": 0.7039465218782425,
+      "num_tokens": 2378746901.0,
+      "step": 1145
+    },
+    {
+      "epoch": 1.6065688329839274,
+      "grad_norm": 0.2541366219520569,
+      "learning_rate": 1.2257492844199736e-05,
+      "loss": 0.9567,
+      "mean_token_accuracy": 0.703971728682518,
+      "num_tokens": 2389112135.0,
+      "step": 1150
+    },
+    {
+      "epoch": 1.6135569531795948,
+      "grad_norm": 0.23742936551570892,
+      "learning_rate": 1.219978093547582e-05,
+      "loss": 0.9456,
+      "mean_token_accuracy": 0.7065177857875824,
+      "num_tokens": 2399497212.0,
+      "step": 1155
+    },
+    {
+      "epoch": 1.6205450733752622,
+      "grad_norm": 0.24232052266597748,
+      "learning_rate": 1.2142086826569043e-05,
+      "loss": 0.9587,
+      "mean_token_accuracy": 0.7030541986227036,
+      "num_tokens": 2409911323.0,
+      "step": 1160
+    },
+    {
+      "epoch": 1.6275331935709294,
+      "grad_norm": 0.23238737881183624,
+      "learning_rate": 1.2084413938130226e-05,
+      "loss": 0.9498,
+      "mean_token_accuracy": 0.7052562206983566,
+      "num_tokens": 2420296768.0,
+      "step": 1165
+    },
+    {
+      "epoch": 1.6345213137665968,
+      "grad_norm": 0.22225673496723175,
+      "learning_rate": 1.202676568955204e-05,
+      "loss": 0.9553,
+      "mean_token_accuracy": 0.7043942958116531,
+      "num_tokens": 2430700961.0,
+      "step": 1170
+    },
+    {
+      "epoch": 1.641509433962264,
+      "grad_norm": 0.2327306568622589,
+      "learning_rate": 1.1969145498766266e-05,
+      "loss": 0.9578,
+      "mean_token_accuracy": 0.7032434940338135,
+      "num_tokens": 2441087132.0,
+      "step": 1175
+    },
+    {
+      "epoch": 1.6484975541579314,
+      "grad_norm": 0.24940182268619537,
+      "learning_rate": 1.1911556782041166e-05,
+      "loss": 0.955,
+      "mean_token_accuracy": 0.7039715796709061,
+      "num_tokens": 2451459456.0,
+      "step": 1180
+    },
+    {
+      "epoch": 1.6554856743535988,
+      "grad_norm": 0.2305036038160324,
+      "learning_rate": 1.1854002953778906e-05,
+      "loss": 0.9593,
+      "mean_token_accuracy": 0.7024497330188751,
+      "num_tokens": 2461836410.0,
+      "step": 1185
+    },
+    {
+      "epoch": 1.6624737945492662,
+      "grad_norm": 0.21421563625335693,
+      "learning_rate": 1.1796487426313152e-05,
+      "loss": 0.9534,
+      "mean_token_accuracy": 0.7043396472930908,
+      "num_tokens": 2472221312.0,
+      "step": 1190
+    },
+    {
+      "epoch": 1.6694619147449337,
+      "grad_norm": 0.2404475063085556,
+      "learning_rate": 1.1739013609706729e-05,
+      "loss": 0.9612,
+      "mean_token_accuracy": 0.7030987590551376,
+      "num_tokens": 2482587402.0,
+      "step": 1195
+    },
+    {
+      "epoch": 1.676450034940601,
+      "grad_norm": 0.2337433099746704,
+      "learning_rate": 1.168158491154945e-05,
+      "loss": 0.9508,
+      "mean_token_accuracy": 0.7051290899515152,
+      "num_tokens": 2492964095.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.6834381551362685,
+      "grad_norm": 0.26063719391822815,
+      "learning_rate": 1.162420473675608e-05,
+      "loss": 0.9429,
+      "mean_token_accuracy": 0.7071120649576187,
+      "num_tokens": 2503327221.0,
+      "step": 1205
+    },
+    {
+      "epoch": 1.6904262753319357,
+      "grad_norm": 0.22970515489578247,
+      "learning_rate": 1.1566876487364462e-05,
+      "loss": 0.9551,
+      "mean_token_accuracy": 0.7043853759765625,
+      "num_tokens": 2513705668.0,
+      "step": 1210
+    },
+    {
+      "epoch": 1.697414395527603,
+      "grad_norm": 0.22828368842601776,
+      "learning_rate": 1.1509603562333817e-05,
+      "loss": 0.9463,
+      "mean_token_accuracy": 0.7062768042087555,
+      "num_tokens": 2524125309.0,
+      "step": 1215
+    },
+    {
+      "epoch": 1.7044025157232703,
+      "grad_norm": 0.2501721680164337,
+      "learning_rate": 1.1452389357343208e-05,
+      "loss": 0.95,
+      "mean_token_accuracy": 0.7051142156124115,
+      "num_tokens": 2534528216.0,
+      "step": 1220
+    },
+    {
+      "epoch": 1.7113906359189377,
+      "grad_norm": 0.21163718402385712,
+      "learning_rate": 1.139523726459022e-05,
+      "loss": 0.9517,
+      "mean_token_accuracy": 0.7048409789800644,
+      "num_tokens": 2544943369.0,
+      "step": 1225
+    },
+    {
+      "epoch": 1.718378756114605,
+      "grad_norm": 0.2533326745033264,
+      "learning_rate": 1.1338150672589847e-05,
+      "loss": 0.9519,
+      "mean_token_accuracy": 0.7046971768140793,
+      "num_tokens": 2555293671.0,
+      "step": 1230
+    },
+    {
+      "epoch": 1.7253668763102725,
+      "grad_norm": 0.23022139072418213,
+      "learning_rate": 1.1281132965973578e-05,
+      "loss": 0.9525,
+      "mean_token_accuracy": 0.7047327756881714,
+      "num_tokens": 2565688022.0,
+      "step": 1235
+    },
+    {
+      "epoch": 1.73235499650594,
+      "grad_norm": 0.24306832253932953,
+      "learning_rate": 1.1224187525288722e-05,
+      "loss": 0.9406,
+      "mean_token_accuracy": 0.7070292413234711,
+      "num_tokens": 2576050620.0,
+      "step": 1240
+    },
+    {
+      "epoch": 1.7393431167016074,
+      "grad_norm": 0.22193977236747742,
+      "learning_rate": 1.1167317726797986e-05,
+      "loss": 0.9461,
+      "mean_token_accuracy": 0.7061527252197266,
+      "num_tokens": 2586432548.0,
+      "step": 1245
+    },
+    {
+      "epoch": 1.7463312368972748,
+      "grad_norm": 0.25281763076782227,
+      "learning_rate": 1.1110526942279294e-05,
+      "loss": 0.9525,
+      "mean_token_accuracy": 0.704227501153946,
+      "num_tokens": 2596837800.0,
+      "step": 1250
+    },
+    {
+      "epoch": 1.753319357092942,
+      "grad_norm": 0.2383853644132614,
+      "learning_rate": 1.1053818538825876e-05,
+      "loss": 0.9458,
+      "mean_token_accuracy": 0.7060607433319092,
+      "num_tokens": 2607259536.0,
+      "step": 1255
+    },
+    {
+      "epoch": 1.7603074772886094,
+      "grad_norm": 0.2552628815174103,
+      "learning_rate": 1.0997195878646636e-05,
+      "loss": 0.9449,
+      "mean_token_accuracy": 0.7070667445659637,
+      "num_tokens": 2617663573.0,
+      "step": 1260
+    },
+    {
+      "epoch": 1.7672955974842768,
+      "grad_norm": 0.20744191110134125,
+      "learning_rate": 1.0940662318866808e-05,
+      "loss": 0.9474,
+      "mean_token_accuracy": 0.7057566583156586,
+      "num_tokens": 2628064038.0,
+      "step": 1265
+    },
+    {
+      "epoch": 1.774283717679944,
+      "grad_norm": 0.2790570855140686,
+      "learning_rate": 1.0884221211328914e-05,
+      "loss": 0.9442,
+      "mean_token_accuracy": 0.7067927271127701,
+      "num_tokens": 2638465498.0,
+      "step": 1270
+    },
+    {
+      "epoch": 1.7812718378756114,
+      "grad_norm": 0.22672200202941895,
+      "learning_rate": 1.0827875902394033e-05,
+      "loss": 0.9478,
+      "mean_token_accuracy": 0.7057815074920655,
+      "num_tokens": 2648882994.0,
+      "step": 1275
+    },
+    {
+      "epoch": 1.7882599580712788,
+      "grad_norm": 0.25658106803894043,
+      "learning_rate": 1.077162973274341e-05,
+      "loss": 0.9484,
+      "mean_token_accuracy": 0.7052160561084747,
+      "num_tokens": 2659259873.0,
+      "step": 1280
+    },
+    {
+      "epoch": 1.7952480782669462,
+      "grad_norm": 0.2257668524980545,
+      "learning_rate": 1.0715486037180363e-05,
+      "loss": 0.9546,
+      "mean_token_accuracy": 0.7042631804943085,
+      "num_tokens": 2669645562.0,
+      "step": 1285
+    },
+    {
+      "epoch": 1.8022361984626136,
+      "grad_norm": 0.217051699757576,
+      "learning_rate": 1.0659448144432585e-05,
+      "loss": 0.9602,
+      "mean_token_accuracy": 0.702583622932434,
+      "num_tokens": 2680036133.0,
+      "step": 1290
+    },
+    {
+      "epoch": 1.809224318658281,
+      "grad_norm": 0.22850480675697327,
+      "learning_rate": 1.060351937695479e-05,
+      "loss": 0.9421,
+      "mean_token_accuracy": 0.707316416501999,
+      "num_tokens": 2690446906.0,
+      "step": 1295
+    },
+    {
+      "epoch": 1.8162124388539485,
+      "grad_norm": 0.20470769703388214,
+      "learning_rate": 1.0547703050731707e-05,
+      "loss": 0.9415,
+      "mean_token_accuracy": 0.7072466671466827,
+      "num_tokens": 2700857094.0,
+      "step": 1300
+    },
+    {
+      "epoch": 1.8232005590496156,
+      "grad_norm": 0.2180124670267105,
+      "learning_rate": 1.049200247508149e-05,
+      "loss": 0.9444,
+      "mean_token_accuracy": 0.7065752387046814,
+      "num_tokens": 2711231823.0,
+      "step": 1305
+    },
+    {
+      "epoch": 1.830188679245283,
+      "grad_norm": 0.22791129350662231,
+      "learning_rate": 1.0436420952459508e-05,
+      "loss": 0.9598,
+      "mean_token_accuracy": 0.7031501412391663,
+      "num_tokens": 2721643248.0,
+      "step": 1310
+    },
+    {
+      "epoch": 1.8371767994409502,
+      "grad_norm": 0.20525507628917694,
+      "learning_rate": 1.0380961778262536e-05,
+      "loss": 0.9493,
+      "mean_token_accuracy": 0.7051385581493378,
+      "num_tokens": 2732049186.0,
+      "step": 1315
+    },
+    {
+      "epoch": 1.8441649196366177,
+      "grad_norm": 0.21797043085098267,
+      "learning_rate": 1.0325628240633398e-05,
+      "loss": 0.9573,
+      "mean_token_accuracy": 0.703617399930954,
+      "num_tokens": 2742450874.0,
+      "step": 1320
+    },
+    {
+      "epoch": 1.851153039832285,
+      "grad_norm": 0.20817072689533234,
+      "learning_rate": 1.0270423620265982e-05,
+      "loss": 0.9543,
+      "mean_token_accuracy": 0.7040122479200364,
+      "num_tokens": 2752825826.0,
+      "step": 1325
+    },
+    {
+      "epoch": 1.8581411600279525,
+      "grad_norm": 0.21172043681144714,
+      "learning_rate": 1.021535119021075e-05,
+      "loss": 0.945,
+      "mean_token_accuracy": 0.7063402503728866,
+      "num_tokens": 2763203546.0,
+      "step": 1330
+    },
+    {
+      "epoch": 1.86512928022362,
+      "grad_norm": 0.22424334287643433,
+      "learning_rate": 1.0160414215680674e-05,
+      "loss": 0.9507,
+      "mean_token_accuracy": 0.7050115913152695,
+      "num_tokens": 2773607371.0,
+      "step": 1335
+    },
+    {
+      "epoch": 1.8721174004192873,
+      "grad_norm": 0.24080626666545868,
+      "learning_rate": 1.0105615953857652e-05,
+      "loss": 0.9502,
+      "mean_token_accuracy": 0.7050412774085999,
+      "num_tokens": 2783983723.0,
+      "step": 1340
+    },
+    {
+      "epoch": 1.8791055206149547,
+      "grad_norm": 0.22549575567245483,
+      "learning_rate": 1.0050959653699377e-05,
+      "loss": 0.9507,
+      "mean_token_accuracy": 0.7051549941301346,
+      "num_tokens": 2794356215.0,
+      "step": 1345
+    },
+    {
+      "epoch": 1.886093640810622,
+      "grad_norm": 0.2175074964761734,
+      "learning_rate": 9.99644855574672e-06,
+      "loss": 0.9495,
+      "mean_token_accuracy": 0.705304104089737,
+      "num_tokens": 2804730279.0,
+      "step": 1350
+    },
+    {
+      "epoch": 1.8930817610062893,
+      "grad_norm": 0.23900523781776428,
+      "learning_rate": 9.942085891931592e-06,
+      "loss": 0.9587,
+      "mean_token_accuracy": 0.7033442050218582,
+      "num_tokens": 2815149853.0,
+      "step": 1355
+    },
+    {
+      "epoch": 1.9000698812019565,
+      "grad_norm": 0.2067466527223587,
+      "learning_rate": 9.887874885385337e-06,
+      "loss": 0.9544,
+      "mean_token_accuracy": 0.7042512089014054,
+      "num_tokens": 2825562033.0,
+      "step": 1360
+    },
+    {
+      "epoch": 1.907058001397624,
+      "grad_norm": 0.20439454913139343,
+      "learning_rate": 9.833818750247615e-06,
+      "loss": 0.9609,
+      "mean_token_accuracy": 0.702758863568306,
+      "num_tokens": 2835964405.0,
+      "step": 1365
+    },
+    {
+      "epoch": 1.9140461215932913,
+      "grad_norm": 0.20674073696136475,
+      "learning_rate": 9.779920691475857e-06,
+      "loss": 0.9508,
+      "mean_token_accuracy": 0.7046973228454589,
+      "num_tokens": 2846369993.0,
+      "step": 1370
+    },
+    {
+      "epoch": 1.9210342417889588,
+      "grad_norm": 0.21767380833625793,
+      "learning_rate": 9.726183904655234e-06,
+      "loss": 0.9575,
+      "mean_token_accuracy": 0.7034873872995376,
+      "num_tokens": 2856789273.0,
+      "step": 1375
+    },
+    {
+      "epoch": 1.9280223619846262,
+      "grad_norm": 0.22383835911750793,
+      "learning_rate": 9.67261157580919e-06,
+      "loss": 0.9513,
+      "mean_token_accuracy": 0.7051774680614471,
+      "num_tokens": 2867172272.0,
+      "step": 1380
+    },
+    {
+      "epoch": 1.9350104821802936,
+      "grad_norm": 0.21862174570560455,
+      "learning_rate": 9.619206881210558e-06,
+      "loss": 0.9487,
+      "mean_token_accuracy": 0.7056363463401795,
+      "num_tokens": 2877584360.0,
+      "step": 1385
+    },
+    {
+      "epoch": 1.941998602375961,
+      "grad_norm": 0.2177773416042328,
+      "learning_rate": 9.565972987193225e-06,
+      "loss": 0.9467,
+      "mean_token_accuracy": 0.7059289753437042,
+      "num_tokens": 2888015561.0,
+      "step": 1390
+    },
+    {
+      "epoch": 1.9489867225716282,
+      "grad_norm": 0.2325785905122757,
+      "learning_rate": 9.512913049964414e-06,
+      "loss": 0.9458,
+      "mean_token_accuracy": 0.7066907495260238,
+      "num_tokens": 2898383704.0,
+      "step": 1395
+    },
+    {
+      "epoch": 1.9559748427672956,
+      "grad_norm": 0.2303362786769867,
+      "learning_rate": 9.460030215417546e-06,
+      "loss": 0.9506,
+      "mean_token_accuracy": 0.7054461807012558,
+      "num_tokens": 2908789197.0,
+      "step": 1400
+    },
+    {
+      "epoch": 1.9629629629629628,
+      "grad_norm": 0.24411572515964508,
+      "learning_rate": 9.407327618945726e-06,
+      "loss": 0.9464,
+      "mean_token_accuracy": 0.7061127424240112,
+      "num_tokens": 2919211817.0,
+      "step": 1405
+    },
+    {
+      "epoch": 1.9699510831586302,
+      "grad_norm": 0.20445261895656586,
+      "learning_rate": 9.354808385255845e-06,
+      "loss": 0.9606,
+      "mean_token_accuracy": 0.7030856341123581,
+      "num_tokens": 2929612659.0,
+      "step": 1410
+    },
+    {
+      "epoch": 1.9769392033542976,
+      "grad_norm": 0.24357956647872925,
+      "learning_rate": 9.302475628183325e-06,
+      "loss": 0.9447,
+      "mean_token_accuracy": 0.7061789721250534,
+      "num_tokens": 2940019034.0,
+      "step": 1415
+    },
+    {
+      "epoch": 1.983927323549965,
+      "grad_norm": 0.21846109628677368,
+      "learning_rate": 9.25033245050748e-06,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.7045047521591187,
+      "num_tokens": 2950402980.0,
+      "step": 1420
+    },
+    {
+      "epoch": 1.9909154437456325,
+      "grad_norm": 0.20229177176952362,
+      "learning_rate": 9.198381943767596e-06,
+      "loss": 0.9336,
+      "mean_token_accuracy": 0.7093939870595932,
+      "num_tokens": 2960814004.0,
+      "step": 1425
+    },
+    {
+      "epoch": 1.9979035639412999,
+      "grad_norm": 0.22578562796115875,
+      "learning_rate": 9.146627188079584e-06,
+      "loss": 0.9399,
+      "mean_token_accuracy": 0.7082905799150467,
+      "num_tokens": 2971180910.0,
+      "step": 1430
+    },
+    {
+      "epoch": 2.0041928721174003,
+      "grad_norm": 0.333396315574646,
+      "learning_rate": 9.0950712519534e-06,
+      "loss": 0.9168,
+      "mean_token_accuracy": 0.7133585843775008,
+      "num_tokens": 2980537865.0,
+      "step": 1435
+    },
+    {
+      "epoch": 2.0111809923130677,
+      "grad_norm": 0.3023437559604645,
+      "learning_rate": 9.043717192111097e-06,
+      "loss": 0.9179,
+      "mean_token_accuracy": 0.712645736336708,
+      "num_tokens": 2990956848.0,
+      "step": 1440
+    },
+    {
+      "epoch": 2.018169112508735,
+      "grad_norm": 0.2666738033294678,
+      "learning_rate": 8.992568053305601e-06,
+      "loss": 0.9192,
+      "mean_token_accuracy": 0.7127588331699372,
+      "num_tokens": 3001334490.0,
+      "step": 1445
+    },
+    {
+      "epoch": 2.0251572327044025,
+      "grad_norm": 0.3103075921535492,
+      "learning_rate": 8.94162686814018e-06,
+      "loss": 0.921,
+      "mean_token_accuracy": 0.7121388405561447,
+      "num_tokens": 3011735960.0,
+      "step": 1450
+    },
+    {
+      "epoch": 2.03214535290007,
+      "grad_norm": 0.3161068558692932,
+      "learning_rate": 8.890896656888664e-06,
+      "loss": 0.9164,
+      "mean_token_accuracy": 0.7127919644117355,
+      "num_tokens": 3022067629.0,
+      "step": 1455
+    },
+    {
+      "epoch": 2.0391334730957373,
+      "grad_norm": 0.2562922239303589,
+      "learning_rate": 8.84038042731634e-06,
+      "loss": 0.9189,
+      "mean_token_accuracy": 0.7123399645090103,
+      "num_tokens": 3032463160.0,
+      "step": 1460
+    },
+    {
+      "epoch": 2.0461215932914047,
+      "grad_norm": 0.24388477206230164,
+      "learning_rate": 8.79008117450166e-06,
+      "loss": 0.915,
+      "mean_token_accuracy": 0.7135502338409424,
+      "num_tokens": 3042894163.0,
+      "step": 1465
+    },
+    {
+      "epoch": 2.053109713487072,
+      "grad_norm": 0.2350393682718277,
+      "learning_rate": 8.740001880658638e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.712294626235962,
+      "num_tokens": 3053289714.0,
+      "step": 1470
+    },
+    {
+      "epoch": 2.060097833682739,
+      "grad_norm": 0.2192184180021286,
+      "learning_rate": 8.690145514960053e-06,
+      "loss": 0.916,
+      "mean_token_accuracy": 0.7133681744337081,
+      "num_tokens": 3063670311.0,
+      "step": 1475
+    },
+    {
+      "epoch": 2.0670859538784065,
+      "grad_norm": 0.21661223471164703,
+      "learning_rate": 8.640515033361391e-06,
+      "loss": 0.9152,
+      "mean_token_accuracy": 0.7137379050254822,
+      "num_tokens": 3074069500.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.074074074074074,
+      "grad_norm": 0.2259409874677658,
+      "learning_rate": 8.591113378425612e-06,
+      "loss": 0.9202,
+      "mean_token_accuracy": 0.7123195230960846,
+      "num_tokens": 3084436917.0,
+      "step": 1485
+    },
+    {
+      "epoch": 2.0810621942697414,
+      "grad_norm": 0.22233138978481293,
+      "learning_rate": 8.541943479148665e-06,
+      "loss": 0.9097,
+      "mean_token_accuracy": 0.7149380385875702,
+      "num_tokens": 3094851526.0,
+      "step": 1490
+    },
+    {
+      "epoch": 2.088050314465409,
+      "grad_norm": 0.2333570122718811,
+      "learning_rate": 8.493008250785837e-06,
+      "loss": 0.9104,
+      "mean_token_accuracy": 0.7141807585954666,
+      "num_tokens": 3105236001.0,
+      "step": 1495
+    },
+    {
+      "epoch": 2.095038434661076,
+      "grad_norm": 0.21300968527793884,
+      "learning_rate": 8.444310594678919e-06,
+      "loss": 0.9171,
+      "mean_token_accuracy": 0.7128302127122879,
+      "num_tokens": 3115627860.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.1020265548567436,
+      "grad_norm": 0.2188606709241867,
+      "learning_rate": 8.395853398084167e-06,
+      "loss": 0.9219,
+      "mean_token_accuracy": 0.7115776121616364,
+      "num_tokens": 3126018436.0,
+      "step": 1505
+    },
+    {
+      "epoch": 2.109014675052411,
+      "grad_norm": 0.2129020094871521,
+      "learning_rate": 8.34763953400114e-06,
+      "loss": 0.9211,
+      "mean_token_accuracy": 0.7127009093761444,
+      "num_tokens": 3136413374.0,
+      "step": 1510
+    },
+    {
+      "epoch": 2.1160027952480784,
+      "grad_norm": 0.2339140921831131,
+      "learning_rate": 8.29967186100234e-06,
+      "loss": 0.9054,
+      "mean_token_accuracy": 0.7164467304944993,
+      "num_tokens": 3146798645.0,
+      "step": 1515
+    },
+    {
+      "epoch": 2.1229909154437454,
+      "grad_norm": 0.210405170917511,
+      "learning_rate": 8.251953223063756e-06,
+      "loss": 0.9253,
+      "mean_token_accuracy": 0.7112928241491318,
+      "num_tokens": 3157172766.0,
+      "step": 1520
+    },
+    {
+      "epoch": 2.129979035639413,
+      "grad_norm": 0.21990114450454712,
+      "learning_rate": 8.204486449396211e-06,
+      "loss": 0.9066,
+      "mean_token_accuracy": 0.7155609846115112,
+      "num_tokens": 3167537336.0,
+      "step": 1525
+    },
+    {
+      "epoch": 2.1369671558350802,
+      "grad_norm": 0.21438851952552795,
+      "learning_rate": 8.157274354277657e-06,
+      "loss": 0.9118,
+      "mean_token_accuracy": 0.7143852919340133,
+      "num_tokens": 3177930338.0,
+      "step": 1530
+    },
+    {
+      "epoch": 2.1439552760307476,
+      "grad_norm": 0.22087393701076508,
+      "learning_rate": 8.110319736886287e-06,
+      "loss": 0.9206,
+      "mean_token_accuracy": 0.71253402531147,
+      "num_tokens": 3188312246.0,
+      "step": 1535
+    },
+    {
+      "epoch": 2.150943396226415,
+      "grad_norm": 0.22319301962852478,
+      "learning_rate": 8.063625381134593e-06,
+      "loss": 0.9205,
+      "mean_token_accuracy": 0.7122439503669739,
+      "num_tokens": 3198688206.0,
+      "step": 1540
+    },
+    {
+      "epoch": 2.1579315164220825,
+      "grad_norm": 0.210270494222641,
+      "learning_rate": 8.017194055504297e-06,
+      "loss": 0.9229,
+      "mean_token_accuracy": 0.7112518161535263,
+      "num_tokens": 3209093337.0,
+      "step": 1545
+    },
+    {
+      "epoch": 2.16491963661775,
+      "grad_norm": 0.22403500974178314,
+      "learning_rate": 7.971028512882226e-06,
+      "loss": 0.9173,
+      "mean_token_accuracy": 0.7129957526922226,
+      "num_tokens": 3219447479.0,
+      "step": 1550
+    },
+    {
+      "epoch": 2.1719077568134173,
+      "grad_norm": 0.22942931950092316,
+      "learning_rate": 7.925131490397075e-06,
+      "loss": 0.9222,
+      "mean_token_accuracy": 0.7115698724985122,
+      "num_tokens": 3229834720.0,
+      "step": 1555
+    },
+    {
+      "epoch": 2.1788958770090847,
+      "grad_norm": 0.21285541355609894,
+      "learning_rate": 7.879505709257131e-06,
+      "loss": 0.9192,
+      "mean_token_accuracy": 0.7122551679611206,
+      "num_tokens": 3240228413.0,
+      "step": 1560
+    },
+    {
+      "epoch": 2.185883997204752,
+      "grad_norm": 0.22184862196445465,
+      "learning_rate": 7.834153874588948e-06,
+      "loss": 0.9184,
+      "mean_token_accuracy": 0.7121668696403504,
+      "num_tokens": 3250638493.0,
+      "step": 1565
+    },
+    {
+      "epoch": 2.192872117400419,
+      "grad_norm": 0.2081901580095291,
+      "learning_rate": 7.789078675276934e-06,
+      "loss": 0.9215,
+      "mean_token_accuracy": 0.7117640644311904,
+      "num_tokens": 3261034976.0,
+      "step": 1570
+    },
+    {
+      "epoch": 2.1998602375960865,
+      "grad_norm": 0.22591695189476013,
+      "learning_rate": 7.744282783803961e-06,
+      "loss": 0.9163,
+      "mean_token_accuracy": 0.7132966309785843,
+      "num_tokens": 3271461658.0,
+      "step": 1575
+    },
+    {
+      "epoch": 2.206848357791754,
+      "grad_norm": 0.21045826375484467,
+      "learning_rate": 7.699768856092882e-06,
+      "loss": 0.9144,
+      "mean_token_accuracy": 0.713357812166214,
+      "num_tokens": 3281821360.0,
+      "step": 1580
+    },
+    {
+      "epoch": 2.2138364779874213,
+      "grad_norm": 0.21535682678222656,
+      "learning_rate": 7.655539531349097e-06,
+      "loss": 0.9202,
+      "mean_token_accuracy": 0.7120324313640595,
+      "num_tokens": 3292235573.0,
+      "step": 1585
+    },
+    {
+      "epoch": 2.2208245981830887,
+      "grad_norm": 0.2165815681219101,
+      "learning_rate": 7.611597431904043e-06,
+      "loss": 0.9183,
+      "mean_token_accuracy": 0.7125867277383804,
+      "num_tokens": 3302639624.0,
+      "step": 1590
+    },
+    {
+      "epoch": 2.227812718378756,
+      "grad_norm": 0.21561956405639648,
+      "learning_rate": 7.567945163059741e-06,
+      "loss": 0.9087,
+      "mean_token_accuracy": 0.7150823384523392,
+      "num_tokens": 3313021924.0,
+      "step": 1595
+    },
+    {
+      "epoch": 2.2348008385744236,
+      "grad_norm": 0.21405579149723053,
+      "learning_rate": 7.524585312934318e-06,
+      "loss": 0.9116,
+      "mean_token_accuracy": 0.7144843906164169,
+      "num_tokens": 3323394648.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.241788958770091,
+      "grad_norm": 0.22909052670001984,
+      "learning_rate": 7.481520452308566e-06,
+      "loss": 0.9124,
+      "mean_token_accuracy": 0.7143676280975342,
+      "num_tokens": 3333800522.0,
+      "step": 1605
+    },
+    {
+      "epoch": 2.2487770789657584,
+      "grad_norm": 0.21624407172203064,
+      "learning_rate": 7.438753134473508e-06,
+      "loss": 0.9197,
+      "mean_token_accuracy": 0.7123973101377488,
+      "num_tokens": 3344220363.0,
+      "step": 1610
+    },
+    {
+      "epoch": 2.2557651991614254,
+      "grad_norm": 0.2294032871723175,
+      "learning_rate": 7.396285895079038e-06,
+      "loss": 0.9177,
+      "mean_token_accuracy": 0.7133157104253769,
+      "num_tokens": 3354615793.0,
+      "step": 1615
+    },
+    {
+      "epoch": 2.262753319357093,
+      "grad_norm": 0.2232784926891327,
+      "learning_rate": 7.354121251983562e-06,
+      "loss": 0.9196,
+      "mean_token_accuracy": 0.712562283873558,
+      "num_tokens": 3365013962.0,
+      "step": 1620
+    },
+    {
+      "epoch": 2.26974143955276,
+      "grad_norm": 0.21506592631340027,
+      "learning_rate": 7.312261705104725e-06,
+      "loss": 0.92,
+      "mean_token_accuracy": 0.7120138168334961,
+      "num_tokens": 3375415677.0,
+      "step": 1625
+    },
+    {
+      "epoch": 2.2767295597484276,
+      "grad_norm": 0.2243955284357071,
+      "learning_rate": 7.270709736271191e-06,
+      "loss": 0.9242,
+      "mean_token_accuracy": 0.7115818649530411,
+      "num_tokens": 3385793817.0,
+      "step": 1630
+    },
+    {
+      "epoch": 2.283717679944095,
+      "grad_norm": 0.23249688744544983,
+      "learning_rate": 7.229467809075501e-06,
+      "loss": 0.9209,
+      "mean_token_accuracy": 0.7121629118919373,
+      "num_tokens": 3396147298.0,
+      "step": 1635
+    },
+    {
+      "epoch": 2.2907058001397624,
+      "grad_norm": 0.25106868147850037,
+      "learning_rate": 7.188538368727998e-06,
+      "loss": 0.9259,
+      "mean_token_accuracy": 0.7110367149114609,
+      "num_tokens": 3406552609.0,
+      "step": 1640
+    },
+    {
+      "epoch": 2.29769392033543,
+      "grad_norm": 0.22599752247333527,
+      "learning_rate": 7.147923841911859e-06,
+      "loss": 0.9132,
+      "mean_token_accuracy": 0.7141028851270675,
+      "num_tokens": 3416969976.0,
+      "step": 1645
+    },
+    {
+      "epoch": 2.3046820405310973,
+      "grad_norm": 0.2180515080690384,
+      "learning_rate": 7.107626636639224e-06,
+      "loss": 0.918,
+      "mean_token_accuracy": 0.7125603795051575,
+      "num_tokens": 3427380676.0,
+      "step": 1650
+    },
+    {
+      "epoch": 2.3116701607267647,
+      "grad_norm": 0.21943214535713196,
+      "learning_rate": 7.067649142108414e-06,
+      "loss": 0.9286,
+      "mean_token_accuracy": 0.7103828489780426,
+      "num_tokens": 3437787781.0,
+      "step": 1655
+    },
+    {
+      "epoch": 2.318658280922432,
+      "grad_norm": 0.224712535738945,
+      "learning_rate": 7.027993728562282e-06,
+      "loss": 0.9227,
+      "mean_token_accuracy": 0.7115794867277145,
+      "num_tokens": 3448190522.0,
+      "step": 1660
+    },
+    {
+      "epoch": 2.325646401118099,
+      "grad_norm": 0.21609260141849518,
+      "learning_rate": 6.988662747147685e-06,
+      "loss": 0.9124,
+      "mean_token_accuracy": 0.7146041959524154,
+      "num_tokens": 3458566741.0,
+      "step": 1665
+    },
+    {
+      "epoch": 2.3326345213137665,
+      "grad_norm": 0.21408966183662415,
+      "learning_rate": 6.949658529776083e-06,
+      "loss": 0.9206,
+      "mean_token_accuracy": 0.7124300539493561,
+      "num_tokens": 3468965946.0,
+      "step": 1670
+    },
+    {
+      "epoch": 2.339622641509434,
+      "grad_norm": 0.21197549998760223,
+      "learning_rate": 6.9109833889852804e-06,
+      "loss": 0.9214,
+      "mean_token_accuracy": 0.7117568552494049,
+      "num_tokens": 3479375801.0,
+      "step": 1675
+    },
+    {
+      "epoch": 2.3466107617051013,
+      "grad_norm": 0.21107982099056244,
+      "learning_rate": 6.872639617802327e-06,
+      "loss": 0.9164,
+      "mean_token_accuracy": 0.7129042834043503,
+      "num_tokens": 3489780868.0,
+      "step": 1680
+    },
+    {
+      "epoch": 2.3535988819007687,
+      "grad_norm": 0.2257632613182068,
+      "learning_rate": 6.834629489607547e-06,
+      "loss": 0.9254,
+      "mean_token_accuracy": 0.7109665483236313,
+      "num_tokens": 3500196102.0,
+      "step": 1685
+    },
+    {
+      "epoch": 2.360587002096436,
+      "grad_norm": 0.22178520262241364,
+      "learning_rate": 6.796955257999769e-06,
+      "loss": 0.9138,
+      "mean_token_accuracy": 0.7136050283908844,
+      "num_tokens": 3510590489.0,
+      "step": 1690
+    },
+    {
+      "epoch": 2.3675751222921035,
+      "grad_norm": 0.21775364875793457,
+      "learning_rate": 6.759619156662699e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.712714570760727,
+      "num_tokens": 3520949287.0,
+      "step": 1695
+    },
+    {
+      "epoch": 2.374563242487771,
+      "grad_norm": 0.2097868025302887,
+      "learning_rate": 6.722623399232498e-06,
+      "loss": 0.9183,
+      "mean_token_accuracy": 0.7127634763717652,
+      "num_tokens": 3531350527.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.381551362683438,
+      "grad_norm": 0.2364913821220398,
+      "learning_rate": 6.685970179166526e-06,
+      "loss": 0.9291,
+      "mean_token_accuracy": 0.7098951876163483,
+      "num_tokens": 3541732110.0,
+      "step": 1705
+    },
+    {
+      "epoch": 2.3885394828791053,
+      "grad_norm": 0.2230016142129898,
+      "learning_rate": 6.649661669613303e-06,
+      "loss": 0.9243,
+      "mean_token_accuracy": 0.711082261800766,
+      "num_tokens": 3552132624.0,
+      "step": 1710
+    },
+    {
+      "epoch": 2.3955276030747727,
+      "grad_norm": 0.20846879482269287,
+      "learning_rate": 6.6137000232836515e-06,
+      "loss": 0.9258,
+      "mean_token_accuracy": 0.7104093581438065,
+      "num_tokens": 3562541950.0,
+      "step": 1715
+    },
+    {
+      "epoch": 2.40251572327044,
+      "grad_norm": 0.2259218990802765,
+      "learning_rate": 6.578087372323081e-06,
+      "loss": 0.9202,
+      "mean_token_accuracy": 0.7123449355363846,
+      "num_tokens": 3572913217.0,
+      "step": 1720
+    },
+    {
+      "epoch": 2.4095038434661076,
+      "grad_norm": 0.2155347317457199,
+      "learning_rate": 6.542825828185355e-06,
+      "loss": 0.9116,
+      "mean_token_accuracy": 0.7139072418212891,
+      "num_tokens": 3583298270.0,
+      "step": 1725
+    },
+    {
+      "epoch": 2.416491963661775,
+      "grad_norm": 0.21978759765625,
+      "learning_rate": 6.507917481507322e-06,
+      "loss": 0.9156,
+      "mean_token_accuracy": 0.7135312378406524,
+      "num_tokens": 3593703423.0,
+      "step": 1730
+    },
+    {
+      "epoch": 2.4234800838574424,
+      "grad_norm": 0.23815348744392395,
+      "learning_rate": 6.473364401984947e-06,
+      "loss": 0.9254,
+      "mean_token_accuracy": 0.710757565498352,
+      "num_tokens": 3604044031.0,
+      "step": 1735
+    },
+    {
+      "epoch": 2.43046820405311,
+      "grad_norm": 0.22524483501911163,
+      "learning_rate": 6.439168638250615e-06,
+      "loss": 0.9238,
+      "mean_token_accuracy": 0.7114168167114258,
+      "num_tokens": 3614426228.0,
+      "step": 1740
+    },
+    {
+      "epoch": 2.4374563242487772,
+      "grad_norm": 0.2189108431339264,
+      "learning_rate": 6.4053322177516585e-06,
+      "loss": 0.9141,
+      "mean_token_accuracy": 0.7138565927743912,
+      "num_tokens": 3624785064.0,
+      "step": 1745
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 0.2147430032491684,
+      "learning_rate": 6.371857146630151e-06,
+      "loss": 0.9257,
+      "mean_token_accuracy": 0.7105766355991363,
+      "num_tokens": 3635155170.0,
+      "step": 1750
+    },
+    {
+      "epoch": 2.451432564640112,
+      "grad_norm": 0.21530896425247192,
+      "learning_rate": 6.338745409603976e-06,
+      "loss": 0.9173,
+      "mean_token_accuracy": 0.7129728078842164,
+      "num_tokens": 3645471686.0,
+      "step": 1755
+    },
+    {
+      "epoch": 2.458420684835779,
+      "grad_norm": 0.2143326848745346,
+      "learning_rate": 6.305998969849134e-06,
+      "loss": 0.9143,
+      "mean_token_accuracy": 0.7136251211166382,
+      "num_tokens": 3655843687.0,
+      "step": 1760
+    },
+    {
+      "epoch": 2.4654088050314464,
+      "grad_norm": 0.21387705206871033,
+      "learning_rate": 6.2736197688833724e-06,
+      "loss": 0.9332,
+      "mean_token_accuracy": 0.7091861546039582,
+      "num_tokens": 3666240232.0,
+      "step": 1765
+    },
+    {
+      "epoch": 2.472396925227114,
+      "grad_norm": 0.21236532926559448,
+      "learning_rate": 6.241609726451045e-06,
+      "loss": 0.912,
+      "mean_token_accuracy": 0.7143011152744293,
+      "num_tokens": 3676651293.0,
+      "step": 1770
+    },
+    {
+      "epoch": 2.4793850454227813,
+      "grad_norm": 0.21449092030525208,
+      "learning_rate": 6.209970740409321e-06,
+      "loss": 0.9209,
+      "mean_token_accuracy": 0.7120813369750977,
+      "num_tokens": 3687062327.0,
+      "step": 1775
+    },
+    {
+      "epoch": 2.4863731656184487,
+      "grad_norm": 0.21442821621894836,
+      "learning_rate": 6.178704686615637e-06,
+      "loss": 0.9174,
+      "mean_token_accuracy": 0.713122546672821,
+      "num_tokens": 3697470376.0,
+      "step": 1780
+    },
+    {
+      "epoch": 2.493361285814116,
+      "grad_norm": 0.21565364301204681,
+      "learning_rate": 6.147813418816498e-06,
+      "loss": 0.9108,
+      "mean_token_accuracy": 0.7147074729204178,
+      "num_tokens": 3707874673.0,
+      "step": 1785
+    },
+    {
+      "epoch": 2.5003494060097835,
+      "grad_norm": 0.21220961213111877,
+      "learning_rate": 6.117298768537552e-06,
+      "loss": 0.9205,
+      "mean_token_accuracy": 0.7123848766088485,
+      "num_tokens": 3718240853.0,
+      "step": 1790
+    },
+    {
+      "epoch": 2.5073375262054505,
+      "grad_norm": 0.2127240151166916,
+      "learning_rate": 6.087162544975018e-06,
+      "loss": 0.9249,
+      "mean_token_accuracy": 0.7108253628015518,
+      "num_tokens": 3728655146.0,
+      "step": 1795
+    },
+    {
+      "epoch": 2.514325646401118,
+      "grad_norm": 0.2235720455646515,
+      "learning_rate": 6.057406534888403e-06,
+      "loss": 0.9257,
+      "mean_token_accuracy": 0.7108543246984482,
+      "num_tokens": 3739022782.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.5213137665967853,
+      "grad_norm": 0.2347782701253891,
+      "learning_rate": 6.028032502494578e-06,
+      "loss": 0.9197,
+      "mean_token_accuracy": 0.7129616022109986,
+      "num_tokens": 3749431904.0,
+      "step": 1805
+    },
+    {
+      "epoch": 2.5283018867924527,
+      "grad_norm": 0.23844227194786072,
+      "learning_rate": 5.999042189363176e-06,
+      "loss": 0.9158,
+      "mean_token_accuracy": 0.7129719346761704,
+      "num_tokens": 3759845315.0,
+      "step": 1810
+    },
+    {
+      "epoch": 2.53529000698812,
+      "grad_norm": 0.22466135025024414,
+      "learning_rate": 5.970437314313328e-06,
+      "loss": 0.919,
+      "mean_token_accuracy": 0.712509247660637,
+      "num_tokens": 3770259622.0,
+      "step": 1815
+    },
+    {
+      "epoch": 2.5422781271837875,
+      "grad_norm": 0.21150830388069153,
+      "learning_rate": 5.9422195733117685e-06,
+      "loss": 0.9221,
+      "mean_token_accuracy": 0.7120893627405167,
+      "num_tokens": 3780638703.0,
+      "step": 1820
+    },
+    {
+      "epoch": 2.549266247379455,
+      "grad_norm": 0.20506353676319122,
+      "learning_rate": 5.914390639372266e-06,
+      "loss": 0.9209,
+      "mean_token_accuracy": 0.7126324087381363,
+      "num_tokens": 3791047016.0,
+      "step": 1825
+    },
+    {
+      "epoch": 2.5562543675751224,
+      "grad_norm": 0.2072431892156601,
+      "learning_rate": 5.8869521624564496e-06,
+      "loss": 0.9129,
+      "mean_token_accuracy": 0.7142052918672561,
+      "num_tokens": 3801447729.0,
+      "step": 1830
+    },
+    {
+      "epoch": 2.56324248777079,
+      "grad_norm": 0.20922759175300598,
+      "learning_rate": 5.859905769375965e-06,
+      "loss": 0.9236,
+      "mean_token_accuracy": 0.7109846025705338,
+      "num_tokens": 3811848994.0,
+      "step": 1835
+    },
+    {
+      "epoch": 2.570230607966457,
+      "grad_norm": 0.214607372879982,
+      "learning_rate": 5.83325306369604e-06,
+      "loss": 0.9087,
+      "mean_token_accuracy": 0.7150654703378677,
+      "num_tokens": 3822222315.0,
+      "step": 1840
+    },
+    {
+      "epoch": 2.5772187281621246,
+      "grad_norm": 0.2122754007577896,
+      "learning_rate": 5.806995625640393e-06,
+      "loss": 0.9193,
+      "mean_token_accuracy": 0.7120439529418945,
+      "num_tokens": 3832658214.0,
+      "step": 1845
+    },
+    {
+      "epoch": 2.584206848357792,
+      "grad_norm": 0.2132931798696518,
+      "learning_rate": 5.781135011997554e-06,
+      "loss": 0.9229,
+      "mean_token_accuracy": 0.711396324634552,
+      "num_tokens": 3843058336.0,
+      "step": 1850
+    },
+    {
+      "epoch": 2.591194968553459,
+      "grad_norm": 0.22513744235038757,
+      "learning_rate": 5.755672756028558e-06,
+      "loss": 0.9153,
+      "mean_token_accuracy": 0.7129842311143875,
+      "num_tokens": 3853440632.0,
+      "step": 1855
+    },
+    {
+      "epoch": 2.5981830887491264,
+      "grad_norm": 0.21373115479946136,
+      "learning_rate": 5.730610367376047e-06,
+      "loss": 0.9211,
+      "mean_token_accuracy": 0.7117935091257095,
+      "num_tokens": 3863875661.0,
+      "step": 1860
+    },
+    {
+      "epoch": 2.605171208944794,
+      "grad_norm": 0.20069323480129242,
+      "learning_rate": 5.705949331974746e-06,
+      "loss": 0.9165,
+      "mean_token_accuracy": 0.7130931288003921,
+      "num_tokens": 3874289974.0,
+      "step": 1865
+    },
+    {
+      "epoch": 2.6121593291404612,
+      "grad_norm": 0.2091623991727829,
+      "learning_rate": 5.681691111963388e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7131164252758027,
+      "num_tokens": 3884708109.0,
+      "step": 1870
+    },
+    {
+      "epoch": 2.6191474493361286,
+      "grad_norm": 0.22306294739246368,
+      "learning_rate": 5.657837145598001e-06,
+      "loss": 0.9324,
+      "mean_token_accuracy": 0.7094870746135712,
+      "num_tokens": 3895062073.0,
+      "step": 1875
+    },
+    {
+      "epoch": 2.626135569531796,
+      "grad_norm": 0.20904283225536346,
+      "learning_rate": 5.634388847166651e-06,
+      "loss": 0.9235,
+      "mean_token_accuracy": 0.7111514359712601,
+      "num_tokens": 3905470817.0,
+      "step": 1880
+    },
+    {
+      "epoch": 2.6331236897274635,
+      "grad_norm": 0.21988101303577423,
+      "learning_rate": 5.611347606905577e-06,
+      "loss": 0.9284,
+      "mean_token_accuracy": 0.7102498173713684,
+      "num_tokens": 3915890767.0,
+      "step": 1885
+    },
+    {
+      "epoch": 2.6401118099231304,
+      "grad_norm": 0.21687500178813934,
+      "learning_rate": 5.58871479091678e-06,
+      "loss": 0.9221,
+      "mean_token_accuracy": 0.7118266075849533,
+      "num_tokens": 3926285033.0,
+      "step": 1890
+    },
+    {
+      "epoch": 2.647099930118798,
+      "grad_norm": 0.21978770196437836,
+      "learning_rate": 5.566491741087012e-06,
+      "loss": 0.9242,
+      "mean_token_accuracy": 0.7114381521940232,
+      "num_tokens": 3936684569.0,
+      "step": 1895
+    },
+    {
+      "epoch": 2.6540880503144653,
+      "grad_norm": 0.21627257764339447,
+      "learning_rate": 5.5446797750082235e-06,
+      "loss": 0.9151,
+      "mean_token_accuracy": 0.7139195591211319,
+      "num_tokens": 3947042426.0,
+      "step": 1900
+    },
+    {
+      "epoch": 2.6610761705101327,
+      "grad_norm": 0.2161075621843338,
+      "learning_rate": 5.523280185899449e-06,
+      "loss": 0.9239,
+      "mean_token_accuracy": 0.711063238978386,
+      "num_tokens": 3957432187.0,
+      "step": 1905
+    },
+    {
+      "epoch": 2.6680642907058,
+      "grad_norm": 0.22316473722457886,
+      "learning_rate": 5.5022942425301205e-06,
+      "loss": 0.9233,
+      "mean_token_accuracy": 0.7108366757631301,
+      "num_tokens": 3967825089.0,
+      "step": 1910
+    },
+    {
+      "epoch": 2.6750524109014675,
+      "grad_norm": 0.2171812802553177,
+      "learning_rate": 5.481723189144858e-06,
+      "loss": 0.9177,
+      "mean_token_accuracy": 0.7128182649612427,
+      "num_tokens": 3978228911.0,
+      "step": 1915
+    },
+    {
+      "epoch": 2.682040531097135,
+      "grad_norm": 0.224291130900383,
+      "learning_rate": 5.46156824538968e-06,
+      "loss": 0.918,
+      "mean_token_accuracy": 0.7131029039621353,
+      "num_tokens": 3988602589.0,
+      "step": 1920
+    },
+    {
+      "epoch": 2.6890286512928023,
+      "grad_norm": 0.23003533482551575,
+      "learning_rate": 5.441830606239715e-06,
+      "loss": 0.9207,
+      "mean_token_accuracy": 0.7122260361909867,
+      "num_tokens": 3998998048.0,
+      "step": 1925
+    },
+    {
+      "epoch": 2.6960167714884697,
+      "grad_norm": 0.22974514961242676,
+      "learning_rate": 5.4225114419283274e-06,
+      "loss": 0.9121,
+      "mean_token_accuracy": 0.7137720167636872,
+      "num_tokens": 4009386456.0,
+      "step": 1930
+    },
+    {
+      "epoch": 2.703004891684137,
+      "grad_norm": 0.2437218725681305,
+      "learning_rate": 5.4036118978777604e-06,
+      "loss": 0.9094,
+      "mean_token_accuracy": 0.7151933252811432,
+      "num_tokens": 4019758598.0,
+      "step": 1935
+    },
+    {
+      "epoch": 2.7099930118798046,
+      "grad_norm": 0.21390853822231293,
+      "learning_rate": 5.385133094631197e-06,
+      "loss": 0.9168,
+      "mean_token_accuracy": 0.7131306231021881,
+      "num_tokens": 4030152895.0,
+      "step": 1940
+    },
+    {
+      "epoch": 2.7169811320754715,
+      "grad_norm": 0.2056519240140915,
+      "learning_rate": 5.367076127786349e-06,
+      "loss": 0.9174,
+      "mean_token_accuracy": 0.7127089440822602,
+      "num_tokens": 4040504482.0,
+      "step": 1945
+    },
+    {
+      "epoch": 2.723969252271139,
+      "grad_norm": 0.21333739161491394,
+      "learning_rate": 5.34944206793048e-06,
+      "loss": 0.9259,
+      "mean_token_accuracy": 0.7104780018329621,
+      "num_tokens": 4050875953.0,
+      "step": 1950
+    },
+    {
+      "epoch": 2.7309573724668064,
+      "grad_norm": 0.2510630786418915,
+      "learning_rate": 5.332231960576948e-06,
+      "loss": 0.911,
+      "mean_token_accuracy": 0.7144548803567886,
+      "num_tokens": 4061259135.0,
+      "step": 1955
+    },
+    {
+      "epoch": 2.737945492662474,
+      "grad_norm": 0.22863414883613586,
+      "learning_rate": 5.3154468261032e-06,
+      "loss": 0.92,
+      "mean_token_accuracy": 0.7120320320129394,
+      "num_tokens": 4071660276.0,
+      "step": 1960
+    },
+    {
+      "epoch": 2.744933612858141,
+      "grad_norm": 0.221303790807724,
+      "learning_rate": 5.2990876596902895e-06,
+      "loss": 0.9177,
+      "mean_token_accuracy": 0.7134216696023941,
+      "num_tokens": 4082088769.0,
+      "step": 1965
+    },
+    {
+      "epoch": 2.7519217330538086,
+      "grad_norm": 0.21982347965240479,
+      "learning_rate": 5.283155431263865e-06,
+      "loss": 0.9267,
+      "mean_token_accuracy": 0.7102663546800614,
+      "num_tokens": 4092426308.0,
+      "step": 1970
+    },
+    {
+      "epoch": 2.758909853249476,
+      "grad_norm": 0.235822394490242,
+      "learning_rate": 5.267651085436665e-06,
+      "loss": 0.9206,
+      "mean_token_accuracy": 0.7118010431528091,
+      "num_tokens": 4102831184.0,
+      "step": 1975
+    },
+    {
+      "epoch": 2.765897973445143,
+      "grad_norm": 0.2101694792509079,
+      "learning_rate": 5.25257554145251e-06,
+      "loss": 0.9142,
+      "mean_token_accuracy": 0.7137985199689865,
+      "num_tokens": 4113206852.0,
+      "step": 1980
+    },
+    {
+      "epoch": 2.7728860936408104,
+      "grad_norm": 0.20930089056491852,
+      "learning_rate": 5.237929693131809e-06,
+      "loss": 0.9237,
+      "mean_token_accuracy": 0.7110803365707398,
+      "num_tokens": 4123598405.0,
+      "step": 1985
+    },
+    {
+      "epoch": 2.779874213836478,
+      "grad_norm": 0.21099254488945007,
+      "learning_rate": 5.223714408818557e-06,
+      "loss": 0.9144,
+      "mean_token_accuracy": 0.7142311304807663,
+      "num_tokens": 4133993093.0,
+      "step": 1990
+    },
+    {
+      "epoch": 2.7868623340321452,
+      "grad_norm": 0.21179494261741638,
+      "learning_rate": 5.2099305313288535e-06,
+      "loss": 0.9219,
+      "mean_token_accuracy": 0.711805310845375,
+      "num_tokens": 4144414884.0,
+      "step": 1995
+    },
+    {
+      "epoch": 2.7938504542278126,
+      "grad_norm": 0.21408036351203918,
+      "learning_rate": 5.196578877900937e-06,
+      "loss": 0.9244,
+      "mean_token_accuracy": 0.7111784070730209,
+      "num_tokens": 4154806748.0,
+      "step": 2000
+    },
+    {
+      "epoch": 2.80083857442348,
+      "grad_norm": 0.20586340129375458,
+      "learning_rate": 5.183660240146726e-06,
+      "loss": 0.922,
+      "mean_token_accuracy": 0.7115540742874146,
+      "num_tokens": 4165215882.0,
+      "step": 2005
+    },
+    {
+      "epoch": 2.8078266946191475,
+      "grad_norm": 0.22240835428237915,
+      "learning_rate": 5.171175384004884e-06,
+      "loss": 0.9123,
+      "mean_token_accuracy": 0.7141046315431595,
+      "num_tokens": 4175639017.0,
+      "step": 2010
+    },
+    {
+      "epoch": 2.814814814814815,
+      "grad_norm": 0.21101967990398407,
+      "learning_rate": 5.159125049695412e-06,
+      "loss": 0.9223,
+      "mean_token_accuracy": 0.7121775299310684,
+      "num_tokens": 4186054800.0,
+      "step": 2015
+    },
+    {
+      "epoch": 2.8218029350104823,
+      "grad_norm": 0.2063722461462021,
+      "learning_rate": 5.147509951675764e-06,
+      "loss": 0.928,
+      "mean_token_accuracy": 0.7096574634313584,
+      "num_tokens": 4196418295.0,
+      "step": 2020
+    },
+    {
+      "epoch": 2.8287910552061497,
+      "grad_norm": 0.20113684237003326,
+      "learning_rate": 5.1363307785984736e-06,
+      "loss": 0.9245,
+      "mean_token_accuracy": 0.7111555904150009,
+      "num_tokens": 4206840950.0,
+      "step": 2025
+    },
+    {
+      "epoch": 2.835779175401817,
+      "grad_norm": 0.2063385397195816,
+      "learning_rate": 5.125588193270341e-06,
+      "loss": 0.9288,
+      "mean_token_accuracy": 0.7106180161237716,
+      "num_tokens": 4217228644.0,
+      "step": 2030
+    },
+    {
+      "epoch": 2.8427672955974845,
+      "grad_norm": 0.20373384654521942,
+      "learning_rate": 5.1152828326131235e-06,
+      "loss": 0.9213,
+      "mean_token_accuracy": 0.7126687705516815,
+      "num_tokens": 4227613729.0,
+      "step": 2035
+    },
+    {
+      "epoch": 2.8497554157931515,
+      "grad_norm": 0.21870306134223938,
+      "learning_rate": 5.105415307625777e-06,
+      "loss": 0.9162,
+      "mean_token_accuracy": 0.7128043532371521,
+      "num_tokens": 4238020240.0,
+      "step": 2040
+    },
+    {
+      "epoch": 2.856743535988819,
+      "grad_norm": 0.20790112018585205,
+      "learning_rate": 5.095986203348233e-06,
+      "loss": 0.9153,
+      "mean_token_accuracy": 0.7130325943231582,
+      "num_tokens": 4248451466.0,
+      "step": 2045
+    },
+    {
+      "epoch": 2.8637316561844863,
+      "grad_norm": 0.20638597011566162,
+      "learning_rate": 5.086996078826707e-06,
+      "loss": 0.9204,
+      "mean_token_accuracy": 0.7117985427379608,
+      "num_tokens": 4258835788.0,
+      "step": 2050
+    },
+    {
+      "epoch": 2.8707197763801537,
+      "grad_norm": 0.21053378283977509,
+      "learning_rate": 5.078445467080554e-06,
+      "loss": 0.9235,
+      "mean_token_accuracy": 0.7111718267202377,
+      "num_tokens": 4269198156.0,
+      "step": 2055
+    },
+    {
+      "epoch": 2.877707896575821,
+      "grad_norm": 0.21624252200126648,
+      "learning_rate": 5.07033487507067e-06,
+      "loss": 0.9258,
+      "mean_token_accuracy": 0.7103535771369934,
+      "num_tokens": 4279623393.0,
+      "step": 2060
+    },
+    {
+      "epoch": 2.8846960167714886,
+      "grad_norm": 0.22071868181228638,
+      "learning_rate": 5.06266478366943e-06,
+      "loss": 0.9297,
+      "mean_token_accuracy": 0.7100573688745498,
+      "num_tokens": 4290014832.0,
+      "step": 2065
+    },
+    {
+      "epoch": 2.891684136967156,
+      "grad_norm": 0.21099479496479034,
+      "learning_rate": 5.055435647632179e-06,
+      "loss": 0.9269,
+      "mean_token_accuracy": 0.7104677796363831,
+      "num_tokens": 4300433907.0,
+      "step": 2070
+    },
+    {
+      "epoch": 2.898672257162823,
+      "grad_norm": 0.2020501047372818,
+      "learning_rate": 5.048647895570273e-06,
+      "loss": 0.9242,
+      "mean_token_accuracy": 0.7116045236587525,
+      "num_tokens": 4310825237.0,
+      "step": 2075
+    },
+    {
+      "epoch": 2.9056603773584904,
+      "grad_norm": 0.20341072976589203,
+      "learning_rate": 5.042301929925656e-06,
+      "loss": 0.9191,
+      "mean_token_accuracy": 0.712513867020607,
+      "num_tokens": 4321192226.0,
+      "step": 2080
+    },
+    {
+      "epoch": 2.912648497554158,
+      "grad_norm": 0.2014591544866562,
+      "learning_rate": 5.0363981269470175e-06,
+      "loss": 0.9214,
+      "mean_token_accuracy": 0.7120217561721802,
+      "num_tokens": 4331623259.0,
+      "step": 2085
+    },
+    {
+      "epoch": 2.919636617749825,
+      "grad_norm": 0.2104942500591278,
+      "learning_rate": 5.030936836667467e-06,
+      "loss": 0.9222,
+      "mean_token_accuracy": 0.7120204120874405,
+      "num_tokens": 4342027121.0,
+      "step": 2090
+    },
+    {
+      "epoch": 2.9266247379454926,
+      "grad_norm": 0.21690668165683746,
+      "learning_rate": 5.025918382883791e-06,
+      "loss": 0.9269,
+      "mean_token_accuracy": 0.7105346918106079,
+      "num_tokens": 4352427294.0,
+      "step": 2095
+    },
+    {
+      "epoch": 2.93361285814116,
+      "grad_norm": 0.2217000424861908,
+      "learning_rate": 5.021343063137254e-06,
+      "loss": 0.9181,
+      "mean_token_accuracy": 0.712472003698349,
+      "num_tokens": 4362825286.0,
+      "step": 2100
+    },
+    {
+      "epoch": 2.9406009783368274,
+      "grad_norm": 0.2240072786808014,
+      "learning_rate": 5.017211148695954e-06,
+      "loss": 0.9277,
+      "mean_token_accuracy": 0.7105086803436279,
+      "num_tokens": 4373218481.0,
+      "step": 2105
+    },
+    {
+      "epoch": 2.947589098532495,
+      "grad_norm": 0.23679901659488678,
+      "learning_rate": 5.013522884538742e-06,
+      "loss": 0.9145,
+      "mean_token_accuracy": 0.7134937226772309,
+      "num_tokens": 4383588535.0,
+      "step": 2110
+    },
+    {
+      "epoch": 2.9545772187281623,
+      "grad_norm": 0.2157946079969406,
+      "learning_rate": 5.0102784893406965e-06,
+      "loss": 0.9037,
+      "mean_token_accuracy": 0.7164124727249146,
+      "num_tokens": 4393978891.0,
+      "step": 2115
+    },
+    {
+      "epoch": 2.9615653389238297,
+      "grad_norm": 0.20823746919631958,
+      "learning_rate": 5.0074781554601595e-06,
+      "loss": 0.927,
+      "mean_token_accuracy": 0.7102025151252747,
+      "num_tokens": 4404364352.0,
+      "step": 2120
+    },
+    {
+      "epoch": 2.968553459119497,
+      "grad_norm": 0.2096376121044159,
+      "learning_rate": 5.005122048927329e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7095394015312195,
+      "num_tokens": 4414766826.0,
+      "step": 2125
+    },
+    {
+      "epoch": 2.9755415793151645,
+      "grad_norm": 0.20438523590564728,
+      "learning_rate": 5.0032103094344185e-06,
+      "loss": 0.9246,
+      "mean_token_accuracy": 0.7110236048698425,
+      "num_tokens": 4425117863.0,
+      "step": 2130
+    },
+    {
+      "epoch": 2.9825296995108315,
+      "grad_norm": 0.20655548572540283,
+      "learning_rate": 5.001743050327372e-06,
+      "loss": 0.924,
+      "mean_token_accuracy": 0.7113671660423279,
+      "num_tokens": 4435506710.0,
+      "step": 2135
+    },
+    {
+      "epoch": 2.989517819706499,
+      "grad_norm": 0.21492992341518402,
+      "learning_rate": 5.000720358599146e-06,
+      "loss": 0.9213,
+      "mean_token_accuracy": 0.7119130939245224,
+      "num_tokens": 4445898940.0,
+      "step": 2140
+    },
+    {
+      "epoch": 2.9965059399021663,
+      "grad_norm": 0.20215627551078796,
+      "learning_rate": 5.000142294884549e-06,
+      "loss": 0.9134,
+      "mean_token_accuracy": 0.7140958666801452,
+      "num_tokens": 4456285249.0,
+      "step": 2145
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2148,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 450,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.410669021545628e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}