diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,3797 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 1000,
+  "global_step": 1874,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.6057937383651733,
+      "epoch": 0.0026680896478121665,
+      "grad_norm": 40.55179553800617,
+      "learning_rate": 4.2553191489361704e-07,
+      "loss": 2.3272,
+      "mean_token_accuracy": 0.5414203643798828,
+      "num_tokens": 314269.0,
+      "step": 5
+    },
+    {
+      "entropy": 1.6647388219833374,
+      "epoch": 0.005336179295624333,
+      "grad_norm": 43.06403399421523,
+      "learning_rate": 9.574468085106384e-07,
+      "loss": 2.1578,
+      "mean_token_accuracy": 0.5552075743675232,
+      "num_tokens": 626887.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.6634229183197022,
+      "epoch": 0.0080042689434365,
+      "grad_norm": 23.775506080571247,
+      "learning_rate": 1.4893617021276596e-06,
+      "loss": 1.753,
+      "mean_token_accuracy": 0.6102993965148926,
+      "num_tokens": 938506.0,
+      "step": 15
+    },
+    {
+      "entropy": 1.3806034088134767,
+      "epoch": 0.010672358591248666,
+      "grad_norm": 15.144339226914607,
+      "learning_rate": 2.021276595744681e-06,
+      "loss": 1.254,
+      "mean_token_accuracy": 0.7032846212387085,
+      "num_tokens": 1253413.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.8211405515670777,
+      "epoch": 0.013340448239060833,
+      "grad_norm": 3.9107969207090885,
+      "learning_rate": 2.553191489361702e-06,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.8092005133628846,
+      "num_tokens": 1567018.0,
+      "step": 25
+    },
+    {
+      "entropy": 0.4903128743171692,
+      "epoch": 0.016008537886873,
+      "grad_norm": 2.675059421456141,
+      "learning_rate": 3.0851063829787237e-06,
+      "loss": 0.4516,
+      "mean_token_accuracy": 0.8781713724136353,
+      "num_tokens": 1879087.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.3075927495956421,
+      "epoch": 0.018676627534685165,
+      "grad_norm": 1.6301455165037353,
+      "learning_rate": 3.6170212765957453e-06,
+      "loss": 0.2837,
+      "mean_token_accuracy": 0.9133403658866882,
+      "num_tokens": 2193371.0,
+      "step": 35
+    },
+    {
+      "entropy": 0.22246569097042085,
+      "epoch": 0.021344717182497332,
+      "grad_norm": 1.1432473923317141,
+      "learning_rate": 4.148936170212766e-06,
+      "loss": 0.2158,
+      "mean_token_accuracy": 0.9255786180496216,
+      "num_tokens": 2505600.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.20077281594276428,
+      "epoch": 0.0240128068303095,
+      "grad_norm": 1.323878562688497,
+      "learning_rate": 4.680851063829788e-06,
+      "loss": 0.1979,
+      "mean_token_accuracy": 0.9290306806564331,
+      "num_tokens": 2816122.0,
+      "step": 45
+    },
+    {
+      "entropy": 0.18968549370765686,
+      "epoch": 0.026680896478121666,
+      "grad_norm": 1.2515659600838258,
+      "learning_rate": 5.212765957446809e-06,
+      "loss": 0.1896,
+      "mean_token_accuracy": 0.9300891041755677,
+      "num_tokens": 3123042.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.18321285843849183,
+      "epoch": 0.02934898612593383,
+      "grad_norm": 0.7840850203555432,
+      "learning_rate": 5.744680851063831e-06,
+      "loss": 0.1803,
+      "mean_token_accuracy": 0.9325991153717041,
+      "num_tokens": 3437738.0,
+      "step": 55
+    },
+    {
+      "entropy": 0.17623865604400635,
+      "epoch": 0.032017075773746,
+      "grad_norm": 0.5801801335679965,
+      "learning_rate": 6.276595744680851e-06,
+      "loss": 0.1764,
+      "mean_token_accuracy": 0.9335901737213135,
+      "num_tokens": 3748405.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.17049630880355834,
+      "epoch": 0.03468516542155817,
+      "grad_norm": 0.4992046369679618,
+      "learning_rate": 6.808510638297873e-06,
+      "loss": 0.1713,
+      "mean_token_accuracy": 0.9349357962608338,
+      "num_tokens": 4064833.0,
+      "step": 65
+    },
+    {
+      "entropy": 0.1735348016023636,
+      "epoch": 0.03735325506937033,
+      "grad_norm": 0.6222605436746601,
+      "learning_rate": 7.340425531914894e-06,
+      "loss": 0.1741,
+      "mean_token_accuracy": 0.9334915518760681,
+      "num_tokens": 4374963.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.17080763280391692,
+      "epoch": 0.040021344717182494,
+      "grad_norm": 0.5219735259020631,
+      "learning_rate": 7.872340425531916e-06,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9345339298248291,
+      "num_tokens": 4686411.0,
+      "step": 75
+    },
+    {
+      "entropy": 0.1672375589609146,
+      "epoch": 0.042689434364994665,
+      "grad_norm": 0.7168995037550425,
+      "learning_rate": 8.404255319148937e-06,
+      "loss": 0.1692,
+      "mean_token_accuracy": 0.9349394798278808,
+      "num_tokens": 4999106.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.16997842788696288,
+      "epoch": 0.04535752401280683,
+      "grad_norm": 0.5830639448822376,
+      "learning_rate": 8.936170212765958e-06,
+      "loss": 0.1711,
+      "mean_token_accuracy": 0.9341935753822327,
+      "num_tokens": 5314232.0,
+      "step": 85
+    },
+    {
+      "entropy": 0.17112229466438295,
+      "epoch": 0.048025613660619,
+      "grad_norm": 0.47473599128466865,
+      "learning_rate": 9.46808510638298e-06,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9356699824333191,
+      "num_tokens": 5627433.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.16550145745277406,
+      "epoch": 0.05069370330843116,
+      "grad_norm": 0.48443350956433884,
+      "learning_rate": 1e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9359866142272949,
+      "num_tokens": 5943052.0,
+      "step": 95
+    },
+    {
+      "entropy": 0.1691408634185791,
+      "epoch": 0.05336179295624333,
+      "grad_norm": 0.4462949644678706,
+      "learning_rate": 1.0531914893617022e-05,
+      "loss": 0.1661,
+      "mean_token_accuracy": 0.9358510494232177,
+      "num_tokens": 6256522.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.16405502259731292,
+      "epoch": 0.056029882604055496,
+      "grad_norm": 0.5476258963434574,
+      "learning_rate": 1.1063829787234044e-05,
+      "loss": 0.1648,
+      "mean_token_accuracy": 0.9364481329917907,
+      "num_tokens": 6564699.0,
+      "step": 105
+    },
+    {
+      "entropy": 0.16443894803524017,
+      "epoch": 0.05869797225186766,
+      "grad_norm": 0.4080625646763937,
+      "learning_rate": 1.1595744680851065e-05,
+      "loss": 0.1642,
+      "mean_token_accuracy": 0.9364275813102723,
+      "num_tokens": 6880930.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.16441120505332946,
+      "epoch": 0.06136606189967983,
+      "grad_norm": 0.40503117807901384,
+      "learning_rate": 1.2127659574468087e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.9364349961280822,
+      "num_tokens": 7195225.0,
+      "step": 115
+    },
+    {
+      "entropy": 0.16429885029792785,
+      "epoch": 0.064034151547492,
+      "grad_norm": 0.3948105540476529,
+      "learning_rate": 1.2659574468085108e-05,
+      "loss": 0.1637,
+      "mean_token_accuracy": 0.9367253661155701,
+      "num_tokens": 7505840.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.16231226325035095,
+      "epoch": 0.06670224119530416,
+      "grad_norm": 0.3707777775484136,
+      "learning_rate": 1.3191489361702127e-05,
+      "loss": 0.1634,
+      "mean_token_accuracy": 0.93640695810318,
+      "num_tokens": 7819265.0,
+      "step": 125
+    },
+    {
+      "entropy": 0.16570849120616912,
+      "epoch": 0.06937033084311633,
+      "grad_norm": 0.3440472771757231,
+      "learning_rate": 1.372340425531915e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9364205598831177,
+      "num_tokens": 8129451.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.16542819142341614,
+      "epoch": 0.0720384204909285,
+      "grad_norm": 0.4679394891059294,
+      "learning_rate": 1.425531914893617e-05,
+      "loss": 0.1663,
+      "mean_token_accuracy": 0.935624098777771,
+      "num_tokens": 8439222.0,
+      "step": 135
+    },
+    {
+      "entropy": 0.16903293132781982,
+      "epoch": 0.07470651013874066,
+      "grad_norm": 11.697052768897619,
+      "learning_rate": 1.4787234042553193e-05,
+      "loss": 0.1761,
+      "mean_token_accuracy": 0.9346019506454468,
+      "num_tokens": 8753787.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.18029699921607972,
+      "epoch": 0.07737459978655283,
+      "grad_norm": 3.2701838952682594,
+      "learning_rate": 1.5319148936170214e-05,
+      "loss": 0.1896,
+      "mean_token_accuracy": 0.928356921672821,
+      "num_tokens": 9065692.0,
+      "step": 145
+    },
+    {
+      "entropy": 0.1697184056043625,
+      "epoch": 0.08004268943436499,
+      "grad_norm": 0.5940807634293421,
+      "learning_rate": 1.5851063829787235e-05,
+      "loss": 0.1674,
+      "mean_token_accuracy": 0.9348207354545593,
+      "num_tokens": 9379119.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.16092735528945923,
+      "epoch": 0.08271077908217717,
+      "grad_norm": 0.3832395625093914,
+      "learning_rate": 1.6382978723404255e-05,
+      "loss": 0.1624,
+      "mean_token_accuracy": 0.9370127081871032,
+      "num_tokens": 9691655.0,
+      "step": 155
+    },
+    {
+      "entropy": 0.16433717608451842,
+      "epoch": 0.08537886872998933,
+      "grad_norm": 0.5605090252886916,
+      "learning_rate": 1.6914893617021276e-05,
+      "loss": 0.1642,
+      "mean_token_accuracy": 0.9362092971801758,
+      "num_tokens": 10005418.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.16236138641834258,
+      "epoch": 0.08804695837780149,
+      "grad_norm": 0.5371554214637537,
+      "learning_rate": 1.74468085106383e-05,
+      "loss": 0.1643,
+      "mean_token_accuracy": 0.9369156122207641,
+      "num_tokens": 10314605.0,
+      "step": 165
+    },
+    {
+      "entropy": 0.16272085011005402,
+      "epoch": 0.09071504802561366,
+      "grad_norm": 0.6100043143177375,
+      "learning_rate": 1.797872340425532e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9354394793510437,
+      "num_tokens": 10627864.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.16350803077220916,
+      "epoch": 0.09338313767342583,
+      "grad_norm": 0.7545661359169421,
+      "learning_rate": 1.8510638297872342e-05,
+      "loss": 0.1705,
+      "mean_token_accuracy": 0.935249638557434,
+      "num_tokens": 10942449.0,
+      "step": 175
+    },
+    {
+      "entropy": 0.16734544038772584,
+      "epoch": 0.096051227321238,
+      "grad_norm": 0.47886967480805687,
+      "learning_rate": 1.9042553191489363e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.9357113361358642,
+      "num_tokens": 11254628.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.16646199226379393,
+      "epoch": 0.09871931696905016,
+      "grad_norm": 0.8950938356931862,
+      "learning_rate": 1.9574468085106384e-05,
+      "loss": 0.1661,
+      "mean_token_accuracy": 0.9356560349464417,
+      "num_tokens": 11565553.0,
+      "step": 185
+    },
+    {
+      "entropy": 0.16376915574073792,
+      "epoch": 0.10138740661686232,
+      "grad_norm": 0.49417911396521913,
+      "learning_rate": 1.9999982639809622e-05,
+      "loss": 0.1651,
+      "mean_token_accuracy": 0.9368755459785462,
+      "num_tokens": 11876584.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.16318838894367219,
+      "epoch": 0.10405549626467449,
+      "grad_norm": 0.39442525016455715,
+      "learning_rate": 1.9999375039475278e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.9363178849220276,
+      "num_tokens": 12190388.0,
+      "step": 195
+    },
+    {
+      "entropy": 0.1615314543247223,
+      "epoch": 0.10672358591248667,
+      "grad_norm": 0.4034435704573861,
+      "learning_rate": 1.999789948989634e-05,
+      "loss": 0.1632,
+      "mean_token_accuracy": 0.9353866577148438,
+      "num_tokens": 12503581.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.1635851889848709,
+      "epoch": 0.10939167556029883,
+      "grad_norm": 0.322362019370726,
+      "learning_rate": 1.9995556119151032e-05,
+      "loss": 0.1632,
+      "mean_token_accuracy": 0.9364137291908264,
+      "num_tokens": 12815167.0,
+      "step": 205
+    },
+    {
+      "entropy": 0.1606872260570526,
+      "epoch": 0.11205976520811099,
+      "grad_norm": 0.7322225604735055,
+      "learning_rate": 1.999234513064475e-05,
+      "loss": 0.1605,
+      "mean_token_accuracy": 0.9371747970581055,
+      "num_tokens": 13133511.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.16151292622089386,
+      "epoch": 0.11472785485592316,
+      "grad_norm": 0.38224695727957775,
+      "learning_rate": 1.998826680309242e-05,
+      "loss": 0.1626,
+      "mean_token_accuracy": 0.9366207599639893,
+      "num_tokens": 13448143.0,
+      "step": 215
+    },
+    {
+      "entropy": 0.16145381927490235,
+      "epoch": 0.11739594450373532,
+      "grad_norm": 0.44622077648668956,
+      "learning_rate": 1.9983321490494292e-05,
+      "loss": 0.1656,
+      "mean_token_accuracy": 0.9352792501449585,
+      "num_tokens": 13760937.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.16521975994110108,
+      "epoch": 0.1200640341515475,
+      "grad_norm": 0.35843161110444777,
+      "learning_rate": 1.9977509622105233e-05,
+      "loss": 0.1637,
+      "mean_token_accuracy": 0.936168372631073,
+      "num_tokens": 14072855.0,
+      "step": 225
+    },
+    {
+      "entropy": 0.16004987359046935,
+      "epoch": 0.12273212379935966,
+      "grad_norm": 0.29808835226604685,
+      "learning_rate": 1.9970831702397447e-05,
+      "loss": 0.161,
+      "mean_token_accuracy": 0.9368571877479553,
+      "num_tokens": 14385910.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.16051293015480042,
+      "epoch": 0.12540021344717184,
+      "grad_norm": 0.3517417621856074,
+      "learning_rate": 1.9963288311016695e-05,
+      "loss": 0.1606,
+      "mean_token_accuracy": 0.937283992767334,
+      "num_tokens": 14696377.0,
+      "step": 235
+    },
+    {
+      "entropy": 0.15953898727893828,
+      "epoch": 0.128068303094984,
+      "grad_norm": 0.3030873385645047,
+      "learning_rate": 1.995488010273198e-05,
+      "loss": 0.161,
+      "mean_token_accuracy": 0.9370529532432557,
+      "num_tokens": 15007831.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.1591504067182541,
+      "epoch": 0.13073639274279616,
+      "grad_norm": 0.3185241274110224,
+      "learning_rate": 1.9945607807378717e-05,
+      "loss": 0.1614,
+      "mean_token_accuracy": 0.9368512272834778,
+      "num_tokens": 15319086.0,
+      "step": 245
+    },
+    {
+      "entropy": 0.16048753559589385,
+      "epoch": 0.13340448239060831,
+      "grad_norm": 0.30739582985703917,
+      "learning_rate": 1.9935472229795386e-05,
+      "loss": 0.1598,
+      "mean_token_accuracy": 0.9372594594955445,
+      "num_tokens": 15631096.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.15844869911670684,
+      "epoch": 0.1360725720384205,
+      "grad_norm": 0.28671630040460405,
+      "learning_rate": 1.9924474249753656e-05,
+      "loss": 0.1592,
+      "mean_token_accuracy": 0.9375737190246582,
+      "num_tokens": 15948578.0,
+      "step": 255
+    },
+    {
+      "entropy": 0.15891876220703124,
+      "epoch": 0.13874066168623267,
+      "grad_norm": 0.33123746638711155,
+      "learning_rate": 1.991261482188203e-05,
+      "loss": 0.1603,
+      "mean_token_accuracy": 0.9368638515472412,
+      "num_tokens": 16264642.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.16164786219596863,
+      "epoch": 0.14140875133404482,
+      "grad_norm": 0.3180618039206802,
+      "learning_rate": 1.9899894975582997e-05,
+      "loss": 0.1613,
+      "mean_token_accuracy": 0.9373221755027771,
+      "num_tokens": 16574897.0,
+      "step": 265
+    },
+    {
+      "entropy": 0.15974415242671966,
+      "epoch": 0.144076840981857,
+      "grad_norm": 0.31502539896563436,
+      "learning_rate": 1.988631581494365e-05,
+      "loss": 0.1617,
+      "mean_token_accuracy": 0.9362964034080505,
+      "num_tokens": 16885412.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.16087310314178466,
+      "epoch": 0.14674493062966915,
+      "grad_norm": 0.3042949940918383,
+      "learning_rate": 1.9871878518639875e-05,
+      "loss": 0.16,
+      "mean_token_accuracy": 0.9370861053466797,
+      "num_tokens": 17196171.0,
+      "step": 275
+    },
+    {
+      "entropy": 0.15577465891838074,
+      "epoch": 0.14941302027748132,
+      "grad_norm": 0.2837059066482121,
+      "learning_rate": 1.985658433983403e-05,
+      "loss": 0.158,
+      "mean_token_accuracy": 0.938635504245758,
+      "num_tokens": 17512272.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.1611830174922943,
+      "epoch": 0.1520811099252935,
+      "grad_norm": 2.07322747055291,
+      "learning_rate": 1.9840434606066182e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9359327912330627,
+      "num_tokens": 17824349.0,
+      "step": 285
+    },
+    {
+      "entropy": 0.16292799413204193,
+      "epoch": 0.15474919957310565,
+      "grad_norm": 0.3567691951667491,
+      "learning_rate": 1.9823430719138856e-05,
+      "loss": 0.1618,
+      "mean_token_accuracy": 0.9365374684333801,
+      "num_tokens": 18134612.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.15836259722709656,
+      "epoch": 0.15741728922091783,
+      "grad_norm": 0.3306976011964114,
+      "learning_rate": 1.980557415499537e-05,
+      "loss": 0.1613,
+      "mean_token_accuracy": 0.9377595663070679,
+      "num_tokens": 18444174.0,
+      "step": 295
+    },
+    {
+      "entropy": 0.16409131288528442,
+      "epoch": 0.16008537886872998,
+      "grad_norm": 0.8223816526857904,
+      "learning_rate": 1.978686646359173e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9360223889350892,
+      "num_tokens": 18757377.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.16573355793952943,
+      "epoch": 0.16275346851654215,
+      "grad_norm": 0.4570923592099133,
+      "learning_rate": 1.976730926876208e-05,
+      "loss": 0.1647,
+      "mean_token_accuracy": 0.9364067673683166,
+      "num_tokens": 19071020.0,
+      "step": 305
+    },
+    {
+      "entropy": 0.1585041582584381,
+      "epoch": 0.16542155816435433,
+      "grad_norm": 0.3847935585188154,
+      "learning_rate": 1.9746904268077766e-05,
+      "loss": 0.1608,
+      "mean_token_accuracy": 0.9372907161712647,
+      "num_tokens": 19384510.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.15850078761577607,
+      "epoch": 0.16808964781216648,
+      "grad_norm": 0.27644993691755454,
+      "learning_rate": 1.9725653232699962e-05,
+      "loss": 0.1586,
+      "mean_token_accuracy": 0.9377372860908508,
+      "num_tokens": 19700894.0,
+      "step": 315
+    },
+    {
+      "entropy": 0.1582118660211563,
+      "epoch": 0.17075773745997866,
+      "grad_norm": 0.3068511012896705,
+      "learning_rate": 1.9703558007225964e-05,
+      "loss": 0.1601,
+      "mean_token_accuracy": 0.9376296758651733,
+      "num_tokens": 20012882.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.1584922641515732,
+      "epoch": 0.1734258271077908,
+      "grad_norm": 0.27025656430048917,
+      "learning_rate": 1.968062050952906e-05,
+      "loss": 0.1584,
+      "mean_token_accuracy": 0.9380206942558289,
+      "num_tokens": 20324153.0,
+      "step": 325
+    },
+    {
+      "entropy": 0.15942637920379638,
+      "epoch": 0.17609391675560299,
+      "grad_norm": 0.29485986752985277,
+      "learning_rate": 1.9656842730592046e-05,
+      "loss": 0.1585,
+      "mean_token_accuracy": 0.9375697135925293,
+      "num_tokens": 20635204.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.15870639085769653,
+      "epoch": 0.17876200640341516,
+      "grad_norm": 0.332325793174125,
+      "learning_rate": 1.963222673433445e-05,
+      "loss": 0.1606,
+      "mean_token_accuracy": 0.9369964957237243,
+      "num_tokens": 20946049.0,
+      "step": 335
+    },
+    {
+      "entropy": 0.1566277861595154,
+      "epoch": 0.1814300960512273,
+      "grad_norm": 0.35414878529407523,
+      "learning_rate": 1.9606774657433335e-05,
+      "loss": 0.1584,
+      "mean_token_accuracy": 0.9371545314788818,
+      "num_tokens": 21257058.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.15877649188041687,
+      "epoch": 0.1840981856990395,
+      "grad_norm": 0.2720083540153981,
+      "learning_rate": 1.958048870913786e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9378036260604858,
+      "num_tokens": 21567622.0,
+      "step": 345
+    },
+    {
+      "entropy": 0.15873935520648957,
+      "epoch": 0.18676627534685167,
+      "grad_norm": 0.33426473267402307,
+      "learning_rate": 1.9553371171077523e-05,
+      "loss": 0.1598,
+      "mean_token_accuracy": 0.937464439868927,
+      "num_tokens": 21877214.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.1577691912651062,
+      "epoch": 0.18943436499466382,
+      "grad_norm": 0.30962801826583497,
+      "learning_rate": 1.9525424397064082e-05,
+      "loss": 0.1586,
+      "mean_token_accuracy": 0.9379173636436462,
+      "num_tokens": 22188854.0,
+      "step": 355
+    },
+    {
+      "entropy": 0.15920698940753936,
+      "epoch": 0.192102454642476,
+      "grad_norm": 0.24858466270630716,
+      "learning_rate": 1.9496650812887293e-05,
+      "loss": 0.1585,
+      "mean_token_accuracy": 0.937783944606781,
+      "num_tokens": 22503817.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.15488786101341248,
+      "epoch": 0.19477054429028814,
+      "grad_norm": 0.25132425246934453,
+      "learning_rate": 1.9467052916104297e-05,
+      "loss": 0.1568,
+      "mean_token_accuracy": 0.938006055355072,
+      "num_tokens": 22820485.0,
+      "step": 365
+    },
+    {
+      "entropy": 0.1576866924762726,
+      "epoch": 0.19743863393810032,
+      "grad_norm": 0.28007319047792106,
+      "learning_rate": 1.9436633275822876e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.937552809715271,
+      "num_tokens": 23137566.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.1562004268169403,
+      "epoch": 0.2001067235859125,
+      "grad_norm": 0.26412703181082453,
+      "learning_rate": 1.9405394532478422e-05,
+      "loss": 0.1571,
+      "mean_token_accuracy": 0.9379514217376709,
+      "num_tokens": 23454287.0,
+      "step": 375
+    },
+    {
+      "entropy": 0.15889868140220642,
+      "epoch": 0.20277481323372465,
+      "grad_norm": 0.2814723241928714,
+      "learning_rate": 1.937333939760477e-05,
+      "loss": 0.1586,
+      "mean_token_accuracy": 0.9374628782272338,
+      "num_tokens": 23765850.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.15611255168914795,
+      "epoch": 0.20544290288153683,
+      "grad_norm": 0.2591709773637703,
+      "learning_rate": 1.934047065359881e-05,
+      "loss": 0.1572,
+      "mean_token_accuracy": 0.9383575320243835,
+      "num_tokens": 24079680.0,
+      "step": 385
+    },
+    {
+      "entropy": 0.15731624364852906,
+      "epoch": 0.20811099252934898,
+      "grad_norm": 0.3274322630296524,
+      "learning_rate": 1.9306791153479007e-05,
+      "loss": 0.1575,
+      "mean_token_accuracy": 0.9381836175918579,
+      "num_tokens": 24391912.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.1555784285068512,
+      "epoch": 0.21077908217716115,
+      "grad_norm": 0.26139305778817185,
+      "learning_rate": 1.9272303820637727e-05,
+      "loss": 0.157,
+      "mean_token_accuracy": 0.9382982850074768,
+      "num_tokens": 24706470.0,
+      "step": 395
+    },
+    {
+      "entropy": 0.1561396062374115,
+      "epoch": 0.21344717182497333,
+      "grad_norm": 0.2937863844138795,
+      "learning_rate": 1.9237011648587496e-05,
+      "loss": 0.1563,
+      "mean_token_accuracy": 0.9385831475257873,
+      "num_tokens": 25022583.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.1557897746562958,
+      "epoch": 0.21611526147278548,
+      "grad_norm": 0.2606893621813393,
+      "learning_rate": 1.9200917700701176e-05,
+      "loss": 0.1575,
+      "mean_token_accuracy": 0.9378427028656006,
+      "num_tokens": 25336503.0,
+      "step": 405
+    },
+    {
+      "entropy": 0.15799925923347474,
+      "epoch": 0.21878335112059766,
+      "grad_norm": 0.26497991242135527,
+      "learning_rate": 1.9164025109946042e-05,
+      "loss": 0.1572,
+      "mean_token_accuracy": 0.9383312106132508,
+      "num_tokens": 25649817.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.15797626674175264,
+      "epoch": 0.2214514407684098,
+      "grad_norm": 0.27467565279203027,
+      "learning_rate": 1.9126337078611855e-05,
+      "loss": 0.159,
+      "mean_token_accuracy": 0.9374309062957764,
+      "num_tokens": 25962652.0,
+      "step": 415
+    },
+    {
+      "entropy": 0.15986525714397432,
+      "epoch": 0.22411953041622198,
+      "grad_norm": 0.2969689886430783,
+      "learning_rate": 1.908785687803289e-05,
+      "loss": 0.1596,
+      "mean_token_accuracy": 0.9372278928756714,
+      "num_tokens": 26271673.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.1557195782661438,
+      "epoch": 0.22678762006403416,
+      "grad_norm": 0.23040213059715953,
+      "learning_rate": 1.9048587848303995e-05,
+      "loss": 0.1571,
+      "mean_token_accuracy": 0.9377826452255249,
+      "num_tokens": 26584515.0,
+      "step": 425
+    },
+    {
+      "entropy": 0.15680642426013947,
+      "epoch": 0.2294557097118463,
+      "grad_norm": 0.22629954502268448,
+      "learning_rate": 1.900853339799066e-05,
+      "loss": 0.1581,
+      "mean_token_accuracy": 0.9378763437271118,
+      "num_tokens": 26895984.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.15606102645397185,
+      "epoch": 0.2321237993596585,
+      "grad_norm": 0.23559172310943693,
+      "learning_rate": 1.8967697003833156e-05,
+      "loss": 0.1556,
+      "mean_token_accuracy": 0.9387164950370789,
+      "num_tokens": 27208162.0,
+      "step": 435
+    },
+    {
+      "entropy": 0.15733987092971802,
+      "epoch": 0.23479188900747064,
+      "grad_norm": 0.3197383053118265,
+      "learning_rate": 1.8926082210444755e-05,
+      "loss": 0.1578,
+      "mean_token_accuracy": 0.9378737568855285,
+      "num_tokens": 27519484.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.1580660492181778,
+      "epoch": 0.23745997865528282,
+      "grad_norm": 0.2737127367874776,
+      "learning_rate": 1.8883692630004046e-05,
+      "loss": 0.1578,
+      "mean_token_accuracy": 0.9377901077270507,
+      "num_tokens": 27831566.0,
+      "step": 445
+    },
+    {
+      "entropy": 0.15822867751121522,
+      "epoch": 0.240128068303095,
+      "grad_norm": 1.0659710021994102,
+      "learning_rate": 1.884053194194142e-05,
+      "loss": 0.1633,
+      "mean_token_accuracy": 0.936628234386444,
+      "num_tokens": 28140071.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.16525934636592865,
+      "epoch": 0.24279615795090714,
+      "grad_norm": 0.6779597502538339,
+      "learning_rate": 1.8796603892619668e-05,
+      "loss": 0.164,
+      "mean_token_accuracy": 0.9368445754051209,
+      "num_tokens": 28451414.0,
+      "step": 455
+    },
+    {
+      "entropy": 0.1622500389814377,
+      "epoch": 0.24546424759871932,
+      "grad_norm": 0.38034546828877175,
+      "learning_rate": 1.8751912295008804e-05,
+      "loss": 0.1623,
+      "mean_token_accuracy": 0.9371641993522644,
+      "num_tokens": 28762308.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.15895282626152038,
+      "epoch": 0.24813233724653147,
+      "grad_norm": 0.3511065822256795,
+      "learning_rate": 1.8706461028355107e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9371441841125489,
+      "num_tokens": 29072105.0,
+      "step": 465
+    },
+    {
+      "entropy": 0.15918091535568238,
+      "epoch": 0.2508004268943437,
+      "grad_norm": 0.2924598212092513,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.1575,
+      "mean_token_accuracy": 0.9381168961524964,
+      "num_tokens": 29382690.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.15696989595890046,
+      "epoch": 0.2534685165421558,
+      "grad_norm": 0.2711360547796898,
+      "learning_rate": 1.861329533425956e-05,
+      "loss": 0.1582,
+      "mean_token_accuracy": 0.9376309752464295,
+      "num_tokens": 29694957.0,
+      "step": 475
+    },
+    {
+      "entropy": 0.15944585502147673,
+      "epoch": 0.256136606189968,
+      "grad_norm": 0.2572014322194392,
+      "learning_rate": 1.8565588993632488e-05,
+      "loss": 0.1593,
+      "mean_token_accuracy": 0.9372217178344726,
+      "num_tokens": 30008241.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.15602820217609406,
+      "epoch": 0.25880469583778015,
+      "grad_norm": 0.26005795926801584,
+      "learning_rate": 1.8517139156890196e-05,
+      "loss": 0.158,
+      "mean_token_accuracy": 0.9383147120475769,
+      "num_tokens": 30319838.0,
+      "step": 485
+    },
+    {
+      "entropy": 0.15950543880462648,
+      "epoch": 0.26147278548559233,
+      "grad_norm": 0.2765594574846301,
+      "learning_rate": 1.846795002949543e-05,
+      "loss": 0.1575,
+      "mean_token_accuracy": 0.9381031513214111,
+      "num_tokens": 30633240.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.1567516565322876,
+      "epoch": 0.2641408751334045,
+      "grad_norm": 0.23855526184988263,
+      "learning_rate": 1.8418025881081612e-05,
+      "loss": 0.1583,
+      "mean_token_accuracy": 0.9379245042800903,
+      "num_tokens": 30944472.0,
+      "step": 495
+    },
+    {
+      "entropy": 0.15918792486190797,
+      "epoch": 0.26680896478121663,
+      "grad_norm": 0.2579721676572744,
+      "learning_rate": 1.8367371045082265e-05,
+      "loss": 0.1605,
+      "mean_token_accuracy": 0.937362265586853,
+      "num_tokens": 31253331.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.15768127143383026,
+      "epoch": 0.2694770544290288,
+      "grad_norm": 0.24642750080557177,
+      "learning_rate": 1.8315989918354835e-05,
+      "loss": 0.1576,
+      "mean_token_accuracy": 0.9381099939346313,
+      "num_tokens": 31567073.0,
+      "step": 505
+    },
+    {
+      "entropy": 0.15596598982810975,
+      "epoch": 0.272145144076841,
+      "grad_norm": 0.25681503988190824,
+      "learning_rate": 1.8263886960799062e-05,
+      "loss": 0.1566,
+      "mean_token_accuracy": 0.9381922602653503,
+      "num_tokens": 31880319.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.15669248402118682,
+      "epoch": 0.27481323372465316,
+      "grad_norm": 0.22065427811092975,
+      "learning_rate": 1.8211066694969855e-05,
+      "loss": 0.1572,
+      "mean_token_accuracy": 0.938100790977478,
+      "num_tokens": 32192951.0,
+      "step": 515
+    },
+    {
+      "entropy": 0.15724363029003144,
+      "epoch": 0.27748132337246534,
+      "grad_norm": 0.2390844464658894,
+      "learning_rate": 1.8157533705684724e-05,
+      "loss": 0.158,
+      "mean_token_accuracy": 0.938084626197815,
+      "num_tokens": 32503369.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.15720780789852143,
+      "epoch": 0.28014941302027746,
+      "grad_norm": 0.24549164685428548,
+      "learning_rate": 1.810329263962584e-05,
+      "loss": 0.157,
+      "mean_token_accuracy": 0.9381688714027405,
+      "num_tokens": 32817261.0,
+      "step": 525
+    },
+    {
+      "entropy": 0.15599971413612365,
+      "epoch": 0.28281750266808964,
+      "grad_norm": 0.2976212854060201,
+      "learning_rate": 1.804834820493666e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9384048819541931,
+      "num_tokens": 33128065.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.15751757621765136,
+      "epoch": 0.2854855923159018,
+      "grad_norm": 0.23632025616502753,
+      "learning_rate": 1.7992705170813306e-05,
+      "loss": 0.1582,
+      "mean_token_accuracy": 0.9379205465316772,
+      "num_tokens": 33438675.0,
+      "step": 535
+    },
+    {
+      "entropy": 0.1583792746067047,
+      "epoch": 0.288153681963714,
+      "grad_norm": 0.25593143055297085,
+      "learning_rate": 1.793636836709057e-05,
+      "loss": 0.1584,
+      "mean_token_accuracy": 0.937691605091095,
+      "num_tokens": 33749618.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.15800465345382692,
+      "epoch": 0.29082177161152617,
+      "grad_norm": 0.2572762932952313,
+      "learning_rate": 1.7879342683822683e-05,
+      "loss": 0.1584,
+      "mean_token_accuracy": 0.937437629699707,
+      "num_tokens": 34059695.0,
+      "step": 545
+    },
+    {
+      "entropy": 0.15476928949356078,
+      "epoch": 0.2934898612593383,
+      "grad_norm": 0.23558110079948988,
+      "learning_rate": 1.7821633070858855e-05,
+      "loss": 0.1559,
+      "mean_token_accuracy": 0.9386150956153869,
+      "num_tokens": 34371953.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.15539830029010773,
+      "epoch": 0.29615795090715047,
+      "grad_norm": 0.278660274756863,
+      "learning_rate": 1.776324453741365e-05,
+      "loss": 0.1556,
+      "mean_token_accuracy": 0.9385053515434265,
+      "num_tokens": 34685055.0,
+      "step": 555
+    },
+    {
+      "entropy": 0.15808961391448975,
+      "epoch": 0.29882604055496265,
+      "grad_norm": 0.27320357497494924,
+      "learning_rate": 1.770418215163215e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9380279421806336,
+      "num_tokens": 34997194.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.15648388266563415,
+      "epoch": 0.3014941302027748,
+      "grad_norm": 0.26423644669523827,
+      "learning_rate": 1.764445104015007e-05,
+      "loss": 0.1574,
+      "mean_token_accuracy": 0.9378781914710999,
+      "num_tokens": 35304669.0,
+      "step": 565
+    },
+    {
+      "entropy": 0.15613919198513032,
+      "epoch": 0.304162219850587,
+      "grad_norm": 0.2322264198496824,
+      "learning_rate": 1.7584056387648727e-05,
+      "loss": 0.1574,
+      "mean_token_accuracy": 0.9375425100326538,
+      "num_tokens": 35615302.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.15673792064189912,
+      "epoch": 0.3068303094983991,
+      "grad_norm": 0.27604927624321784,
+      "learning_rate": 1.7523003436405055e-05,
+      "loss": 0.1566,
+      "mean_token_accuracy": 0.9383435010910034,
+      "num_tokens": 35928610.0,
+      "step": 575
+    },
+    {
+      "entropy": 0.15415124595165253,
+      "epoch": 0.3094983991462113,
+      "grad_norm": 0.28251874667852384,
+      "learning_rate": 1.7461297485836537e-05,
+      "loss": 0.1548,
+      "mean_token_accuracy": 0.938884687423706,
+      "num_tokens": 36243859.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.15712933540344237,
+      "epoch": 0.3121664887940235,
+      "grad_norm": 0.22839995512002847,
+      "learning_rate": 1.7398943892041223e-05,
+      "loss": 0.1571,
+      "mean_token_accuracy": 0.9381480932235717,
+      "num_tokens": 36554595.0,
+      "step": 585
+    },
+    {
+      "entropy": 0.1551433563232422,
+      "epoch": 0.31483457844183566,
+      "grad_norm": 0.280231703238132,
+      "learning_rate": 1.7335948067332827e-05,
+      "loss": 0.1561,
+      "mean_token_accuracy": 0.9383224487304688,
+      "num_tokens": 36869542.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.15621574819087983,
+      "epoch": 0.31750266808964783,
+      "grad_norm": 0.23211378761666593,
+      "learning_rate": 1.7272315479770925e-05,
+      "loss": 0.1572,
+      "mean_token_accuracy": 0.9376968264579773,
+      "num_tokens": 37179670.0,
+      "step": 595
+    },
+    {
+      "entropy": 0.15389657020568848,
+      "epoch": 0.32017075773745995,
+      "grad_norm": 0.23153822744101846,
+      "learning_rate": 1.7208051652686335e-05,
+      "loss": 0.1533,
+      "mean_token_accuracy": 0.9396386504173279,
+      "num_tokens": 37496771.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.1562306672334671,
+      "epoch": 0.32283884738527213,
+      "grad_norm": 0.23042754503602308,
+      "learning_rate": 1.714316216420169e-05,
+      "loss": 0.1565,
+      "mean_token_accuracy": 0.9387167930603028,
+      "num_tokens": 37810099.0,
+      "step": 605
+    },
+    {
+      "entropy": 0.15647551119327546,
+      "epoch": 0.3255069370330843,
+      "grad_norm": 0.22042753143681787,
+      "learning_rate": 1.707765264674724e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9378628253936767,
+      "num_tokens": 38121465.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.1532812863588333,
+      "epoch": 0.3281750266808965,
+      "grad_norm": 0.26050597368266426,
+      "learning_rate": 1.701152878657197e-05,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.9390907406806945,
+      "num_tokens": 38434404.0,
+      "step": 615
+    },
+    {
+      "entropy": 0.15821815133094788,
+      "epoch": 0.33084311632870866,
+      "grad_norm": 0.26774466711519207,
+      "learning_rate": 1.6944796323250036e-05,
+      "loss": 0.1575,
+      "mean_token_accuracy": 0.9375768661499023,
+      "num_tokens": 38745997.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.15520448386669158,
+      "epoch": 0.3335112059765208,
+      "grad_norm": 0.2555654904241375,
+      "learning_rate": 1.687746104918255e-05,
+      "loss": 0.1565,
+      "mean_token_accuracy": 0.938270628452301,
+      "num_tokens": 39058084.0,
+      "step": 625
+    },
+    {
+      "entropy": 0.15563291013240815,
+      "epoch": 0.33617929562433296,
+      "grad_norm": 0.24561720528977696,
+      "learning_rate": 1.6809528809094808e-05,
+      "loss": 0.1554,
+      "mean_token_accuracy": 0.9383365511894226,
+      "num_tokens": 39374950.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.15381639301776887,
+      "epoch": 0.33884738527214514,
+      "grad_norm": 0.2752577369716408,
+      "learning_rate": 1.674100549952897e-05,
+      "loss": 0.1553,
+      "mean_token_accuracy": 0.9386968612670898,
+      "num_tokens": 39687975.0,
+      "step": 635
+    },
+    {
+      "entropy": 0.15738414824008942,
+      "epoch": 0.3415154749199573,
+      "grad_norm": 0.2557558969973054,
+      "learning_rate": 1.6671897068332232e-05,
+      "loss": 0.1558,
+      "mean_token_accuracy": 0.9381865382194519,
+      "num_tokens": 40001067.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.15551066398620605,
+      "epoch": 0.3441835645677695,
+      "grad_norm": 0.22019467451935834,
+      "learning_rate": 1.6602209514140552e-05,
+      "loss": 0.157,
+      "mean_token_accuracy": 0.9382826447486877,
+      "num_tokens": 40312428.0,
+      "step": 645
+    },
+    {
+      "entropy": 0.15661275684833526,
+      "epoch": 0.3468516542155816,
+      "grad_norm": 0.22547104678195906,
+      "learning_rate": 1.6531948885857957e-05,
+      "loss": 0.1568,
+      "mean_token_accuracy": 0.9382068991661072,
+      "num_tokens": 40621016.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.15573394000530244,
+      "epoch": 0.3495197438633938,
+      "grad_norm": 0.22623521473974365,
+      "learning_rate": 1.6461121282131518e-05,
+      "loss": 0.1566,
+      "mean_token_accuracy": 0.9383347630500793,
+      "num_tokens": 40935624.0,
+      "step": 655
+    },
+    {
+      "entropy": 0.15639272928237916,
+      "epoch": 0.35218783351120597,
+      "grad_norm": 0.21033908601468662,
+      "learning_rate": 1.6389732850821967e-05,
+      "loss": 0.156,
+      "mean_token_accuracy": 0.938245439529419,
+      "num_tokens": 41247131.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.15640448331832885,
+      "epoch": 0.35485592315901815,
+      "grad_norm": 0.23057738411070072,
+      "learning_rate": 1.6317789788470063e-05,
+      "loss": 0.1572,
+      "mean_token_accuracy": 0.9381119847297669,
+      "num_tokens": 41558078.0,
+      "step": 665
+    },
+    {
+      "entropy": 0.15540762543678283,
+      "epoch": 0.3575240128068303,
+      "grad_norm": 0.2034575643460659,
+      "learning_rate": 1.624529833975874e-05,
+      "loss": 0.1554,
+      "mean_token_accuracy": 0.9385368585586548,
+      "num_tokens": 41871920.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.15528062880039215,
+      "epoch": 0.36019210245464245,
+      "grad_norm": 0.20896524280441509,
+      "learning_rate": 1.617226479697105e-05,
+      "loss": 0.1559,
+      "mean_token_accuracy": 0.9384117603302002,
+      "num_tokens": 42181375.0,
+      "step": 675
+    },
+    {
+      "entropy": 0.15409398078918457,
+      "epoch": 0.3628601921024546,
+      "grad_norm": 0.2153970667083397,
+      "learning_rate": 1.609869549944401e-05,
+      "loss": 0.1554,
+      "mean_token_accuracy": 0.938829505443573,
+      "num_tokens": 42495107.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.1557753324508667,
+      "epoch": 0.3655282817502668,
+      "grad_norm": 0.25314972577609657,
+      "learning_rate": 1.6024596833018335e-05,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.938931691646576,
+      "num_tokens": 42807362.0,
+      "step": 685
+    },
+    {
+      "entropy": 0.15671408474445342,
+      "epoch": 0.368196371398079,
+      "grad_norm": 0.23786555264931808,
+      "learning_rate": 1.5949975229484132e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9379519820213318,
+      "num_tokens": 43118231.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.1554604947566986,
+      "epoch": 0.37086446104589116,
+      "grad_norm": 0.21787664753869665,
+      "learning_rate": 1.587483716602265e-05,
+      "loss": 0.1566,
+      "mean_token_accuracy": 0.9381156921386719,
+      "num_tokens": 43429978.0,
+      "step": 695
+    },
+    {
+      "entropy": 0.1558491349220276,
+      "epoch": 0.37353255069370334,
+      "grad_norm": 0.20339100359830212,
+      "learning_rate": 1.5799189164644024e-05,
+      "loss": 0.1573,
+      "mean_token_accuracy": 0.9377960801124573,
+      "num_tokens": 43739241.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.15804753303527833,
+      "epoch": 0.37620064034151546,
+      "grad_norm": 0.19712038739819182,
+      "learning_rate": 1.5723037791621193e-05,
+      "loss": 0.157,
+      "mean_token_accuracy": 0.9380033731460571,
+      "num_tokens": 44055301.0,
+      "step": 705
+    },
+    {
+      "entropy": 0.15700536370277404,
+      "epoch": 0.37886872998932764,
+      "grad_norm": 0.2134151511615338,
+      "learning_rate": 1.5646389656919912e-05,
+      "loss": 0.1574,
+      "mean_token_accuracy": 0.937994635105133,
+      "num_tokens": 44367651.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.15613543689250947,
+      "epoch": 0.3815368196371398,
+      "grad_norm": 0.21307800352503886,
+      "learning_rate": 1.5569251413625043e-05,
+      "loss": 0.1568,
+      "mean_token_accuracy": 0.9383546113967896,
+      "num_tokens": 44679783.0,
+      "step": 715
+    },
+    {
+      "entropy": 0.1584879845380783,
+      "epoch": 0.384204909284952,
+      "grad_norm": 0.2044418124578672,
+      "learning_rate": 1.5491629757363033e-05,
+      "loss": 0.1575,
+      "mean_token_accuracy": 0.9375457048416138,
+      "num_tokens": 44990142.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.15449098348617554,
+      "epoch": 0.38687299893276417,
+      "grad_norm": 0.1916484977488963,
+      "learning_rate": 1.5413531425720747e-05,
+      "loss": 0.1549,
+      "mean_token_accuracy": 0.938965666294098,
+      "num_tokens": 45300580.0,
+      "step": 725
+    },
+    {
+      "entropy": 0.1531301200389862,
+      "epoch": 0.3895410885805763,
+      "grad_norm": 0.19224448782508818,
+      "learning_rate": 1.533496319766064e-05,
+      "loss": 0.1551,
+      "mean_token_accuracy": 0.9387568950653076,
+      "num_tokens": 45613328.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.1593233525753021,
+      "epoch": 0.39220917822838847,
+      "grad_norm": 0.22677980229806463,
+      "learning_rate": 1.5255931892932333e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.9373407602310181,
+      "num_tokens": 45919600.0,
+      "step": 735
+    },
+    {
+      "entropy": 0.15600906312465668,
+      "epoch": 0.39487726787620064,
+      "grad_norm": 0.21468441059311985,
+      "learning_rate": 1.5176444371480673e-05,
+      "loss": 0.1563,
+      "mean_token_accuracy": 0.9382573127746582,
+      "num_tokens": 46231651.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.1550221174955368,
+      "epoch": 0.3975453575240128,
+      "grad_norm": 0.2628525776434047,
+      "learning_rate": 1.5096507532850275e-05,
+      "loss": 0.1549,
+      "mean_token_accuracy": 0.9392532229423523,
+      "num_tokens": 46545633.0,
+      "step": 745
+    },
+    {
+      "entropy": 0.15294830203056337,
+      "epoch": 0.400213447171825,
+      "grad_norm": 0.23137778372846649,
+      "learning_rate": 1.501612831558664e-05,
+      "loss": 0.1541,
+      "mean_token_accuracy": 0.9392737984657288,
+      "num_tokens": 46861243.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.15558450818061828,
+      "epoch": 0.4028815368196371,
+      "grad_norm": 0.19570194195754864,
+      "learning_rate": 1.493531369663389e-05,
+      "loss": 0.1547,
+      "mean_token_accuracy": 0.9386296153068543,
+      "num_tokens": 47176657.0,
+      "step": 755
+    },
+    {
+      "entropy": 0.15616436898708344,
+      "epoch": 0.4055496264674493,
+      "grad_norm": 0.20951796443416532,
+      "learning_rate": 1.4854070690729165e-05,
+      "loss": 0.1558,
+      "mean_token_accuracy": 0.9383723020553589,
+      "num_tokens": 47490967.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.15231897830963134,
+      "epoch": 0.4082177161152615,
+      "grad_norm": 0.2117819997652048,
+      "learning_rate": 1.4772406349793744e-05,
+      "loss": 0.1551,
+      "mean_token_accuracy": 0.9383639574050904,
+      "num_tokens": 47801604.0,
+      "step": 765
+    },
+    {
+      "entropy": 0.15625073611736298,
+      "epoch": 0.41088580576307365,
+      "grad_norm": 0.21291868754948795,
+      "learning_rate": 1.4690327762320931e-05,
+      "loss": 0.1552,
+      "mean_token_accuracy": 0.9382084488868714,
+      "num_tokens": 48114796.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.15438767075538634,
+      "epoch": 0.41355389541088583,
+      "grad_norm": 0.2018230333685515,
+      "learning_rate": 1.4607842052760777e-05,
+      "loss": 0.154,
+      "mean_token_accuracy": 0.9390445947647095,
+      "num_tokens": 48429003.0,
+      "step": 775
+    },
+    {
+      "entropy": 0.1527568519115448,
+      "epoch": 0.41622198505869795,
+      "grad_norm": 0.21050953897057828,
+      "learning_rate": 1.4524956380901669e-05,
+      "loss": 0.154,
+      "mean_token_accuracy": 0.9386907339096069,
+      "num_tokens": 48742907.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.15494268834590913,
+      "epoch": 0.41889007470651013,
+      "grad_norm": 0.24431428774607528,
+      "learning_rate": 1.4441677941248862e-05,
+      "loss": 0.155,
+      "mean_token_accuracy": 0.938653540611267,
+      "num_tokens": 49054937.0,
+      "step": 785
+    },
+    {
+      "entropy": 0.15730642080307006,
+      "epoch": 0.4215581643543223,
+      "grad_norm": 0.21834695991194622,
+      "learning_rate": 1.4358013962399998e-05,
+      "loss": 0.1592,
+      "mean_token_accuracy": 0.9372829079627991,
+      "num_tokens": 49363498.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.15722571909427643,
+      "epoch": 0.4242262540021345,
+      "grad_norm": 0.20605119981337897,
+      "learning_rate": 1.4273971706417648e-05,
+      "loss": 0.1552,
+      "mean_token_accuracy": 0.9383088946342468,
+      "num_tokens": 49675352.0,
+      "step": 795
+    },
+    {
+      "entropy": 0.15412231981754304,
+      "epoch": 0.42689434364994666,
+      "grad_norm": 0.2264335394675343,
+      "learning_rate": 1.4189558468198973e-05,
+      "loss": 0.1556,
+      "mean_token_accuracy": 0.93870609998703,
+      "num_tokens": 49987958.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.1554412692785263,
+      "epoch": 0.4295624332977588,
+      "grad_norm": 0.26579388260223064,
+      "learning_rate": 1.4104781574842524e-05,
+      "loss": 0.1571,
+      "mean_token_accuracy": 0.9380474090576172,
+      "num_tokens": 50301027.0,
+      "step": 805
+    },
+    {
+      "entropy": 0.1543067067861557,
+      "epoch": 0.43223052294557096,
+      "grad_norm": 0.2559081768640021,
+      "learning_rate": 1.4019648385012243e-05,
+      "loss": 0.1537,
+      "mean_token_accuracy": 0.9395334362983704,
+      "num_tokens": 50614978.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.15630431473255157,
+      "epoch": 0.43489861259338314,
+      "grad_norm": 0.25616045587986713,
+      "learning_rate": 1.3934166288298729e-05,
+      "loss": 0.1558,
+      "mean_token_accuracy": 0.9386638164520263,
+      "num_tokens": 50929542.0,
+      "step": 815
+    },
+    {
+      "entropy": 0.15396025478839875,
+      "epoch": 0.4375667022411953,
+      "grad_norm": 0.23184443832119792,
+      "learning_rate": 1.3848342704577824e-05,
+      "loss": 0.1555,
+      "mean_token_accuracy": 0.9384626746177673,
+      "num_tokens": 51241435.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.15652641355991365,
+      "epoch": 0.4402347918890075,
+      "grad_norm": 0.22777748094798117,
+      "learning_rate": 1.3762185083366557e-05,
+      "loss": 0.156,
+      "mean_token_accuracy": 0.938478434085846,
+      "num_tokens": 51553060.0,
+      "step": 825
+    },
+    {
+      "entropy": 0.15561273396015168,
+      "epoch": 0.4429028815368196,
+      "grad_norm": 0.22271913073510122,
+      "learning_rate": 1.3675700903176534e-05,
+      "loss": 0.1558,
+      "mean_token_accuracy": 0.9381041049957275,
+      "num_tokens": 51865184.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.15462508499622346,
+      "epoch": 0.4455709711846318,
+      "grad_norm": 0.22905739195488695,
+      "learning_rate": 1.3588897670864787e-05,
+      "loss": 0.156,
+      "mean_token_accuracy": 0.9383329391479492,
+      "num_tokens": 52179580.0,
+      "step": 835
+    },
+    {
+      "entropy": 0.15535084903240204,
+      "epoch": 0.44823906083244397,
+      "grad_norm": 0.20758795386301682,
+      "learning_rate": 1.3501782920982185e-05,
+      "loss": 0.1552,
+      "mean_token_accuracy": 0.9387192964553833,
+      "num_tokens": 52493650.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.15626255571842193,
+      "epoch": 0.45090715048025615,
+      "grad_norm": 0.26357886940476394,
+      "learning_rate": 1.3414364215119436e-05,
+      "loss": 0.157,
+      "mean_token_accuracy": 0.9378098011016845,
+      "num_tokens": 52805986.0,
+      "step": 845
+    },
+    {
+      "entropy": 0.15512668788433076,
+      "epoch": 0.4535752401280683,
+      "grad_norm": 0.2687039281213037,
+      "learning_rate": 1.3326649141250731e-05,
+      "loss": 0.1547,
+      "mean_token_accuracy": 0.9390361428260803,
+      "num_tokens": 53119855.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.15508779585361482,
+      "epoch": 0.45624332977588045,
+      "grad_norm": 0.23811379871161598,
+      "learning_rate": 1.3238645313075104e-05,
+      "loss": 0.1555,
+      "mean_token_accuracy": 0.9387191772460938,
+      "num_tokens": 53430994.0,
+      "step": 855
+    },
+    {
+      "entropy": 0.15608465075492858,
+      "epoch": 0.4589114194236926,
+      "grad_norm": 0.19296201717409298,
+      "learning_rate": 1.315036036935557e-05,
+      "loss": 0.1567,
+      "mean_token_accuracy": 0.9383533000946045,
+      "num_tokens": 53743754.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.15427376329898834,
+      "epoch": 0.4615795090715048,
+      "grad_norm": 0.19057043944008742,
+      "learning_rate": 1.3061801973256066e-05,
+      "loss": 0.1544,
+      "mean_token_accuracy": 0.9388363599777222,
+      "num_tokens": 54060087.0,
+      "step": 865
+    },
+    {
+      "entropy": 0.15521352589130402,
+      "epoch": 0.464247598719317,
+      "grad_norm": 0.26017936933633373,
+      "learning_rate": 1.2972977811676286e-05,
+      "loss": 0.1557,
+      "mean_token_accuracy": 0.9386460781097412,
+      "num_tokens": 54371977.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.15631742775440216,
+      "epoch": 0.46691568836712916,
+      "grad_norm": 0.21790366736450598,
+      "learning_rate": 1.2883895594584474e-05,
+      "loss": 0.1568,
+      "mean_token_accuracy": 0.9381857514381409,
+      "num_tokens": 54682077.0,
+      "step": 875
+    },
+    {
+      "entropy": 0.15416281521320344,
+      "epoch": 0.4695837780149413,
+      "grad_norm": 0.1978743033527956,
+      "learning_rate": 1.2794563054348173e-05,
+      "loss": 0.1534,
+      "mean_token_accuracy": 0.9395832657814026,
+      "num_tokens": 54996694.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.15204716920852662,
+      "epoch": 0.47225186766275346,
+      "grad_norm": 0.20919665202064539,
+      "learning_rate": 1.270498794506307e-05,
+      "loss": 0.1535,
+      "mean_token_accuracy": 0.9389811158180237,
+      "num_tokens": 55311736.0,
+      "step": 885
+    },
+    {
+      "entropy": 0.15359111726284028,
+      "epoch": 0.47491995731056563,
+      "grad_norm": 0.1959526143243109,
+      "learning_rate": 1.2615178041879918e-05,
+      "loss": 0.1536,
+      "mean_token_accuracy": 0.9392480969429016,
+      "num_tokens": 55627584.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.1572214722633362,
+      "epoch": 0.4775880469583778,
+      "grad_norm": 0.21946848825899526,
+      "learning_rate": 1.2525141140329673e-05,
+      "loss": 0.1573,
+      "mean_token_accuracy": 0.938277006149292,
+      "num_tokens": 55937026.0,
+      "step": 895
+    },
+    {
+      "entropy": 0.15423589944839478,
+      "epoch": 0.48025613660619,
+      "grad_norm": 0.22980710459603573,
+      "learning_rate": 1.2434885055646823e-05,
+      "loss": 0.1545,
+      "mean_token_accuracy": 0.9389993071556091,
+      "num_tokens": 56250825.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.15658611953258514,
+      "epoch": 0.4829242262540021,
+      "grad_norm": 0.24487921513464042,
+      "learning_rate": 1.2344417622091032e-05,
+      "loss": 0.156,
+      "mean_token_accuracy": 0.9381351351737977,
+      "num_tokens": 56560445.0,
+      "step": 905
+    },
+    {
+      "entropy": 0.15108625292778016,
+      "epoch": 0.4855923159018143,
+      "grad_norm": 0.2301434106979841,
+      "learning_rate": 1.2253746692267119e-05,
+      "loss": 0.1528,
+      "mean_token_accuracy": 0.9396188139915467,
+      "num_tokens": 56877321.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.1589755594730377,
+      "epoch": 0.48826040554962646,
+      "grad_norm": 0.25294725895747705,
+      "learning_rate": 1.2162880136443447e-05,
+      "loss": 0.1589,
+      "mean_token_accuracy": 0.9371992588043213,
+      "num_tokens": 57185139.0,
+      "step": 915
+    },
+    {
+      "entropy": 0.15456578135490417,
+      "epoch": 0.49092849519743864,
+      "grad_norm": 0.20948010370122902,
+      "learning_rate": 1.2071825841868788e-05,
+      "loss": 0.1543,
+      "mean_token_accuracy": 0.9389463782310485,
+      "num_tokens": 57499589.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.15248561203479766,
+      "epoch": 0.4935965848452508,
+      "grad_norm": 0.19733552881311944,
+      "learning_rate": 1.19805917120877e-05,
+      "loss": 0.1534,
+      "mean_token_accuracy": 0.9392978549003601,
+      "num_tokens": 57817486.0,
+      "step": 925
+    },
+    {
+      "entropy": 0.15405131876468658,
+      "epoch": 0.49626467449306294,
+      "grad_norm": 0.18714177298536025,
+      "learning_rate": 1.1889185666254505e-05,
+      "loss": 0.1539,
+      "mean_token_accuracy": 0.9388469815254211,
+      "num_tokens": 58133143.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.15573233962059022,
+      "epoch": 0.4989327641408751,
+      "grad_norm": 0.21350853503849604,
+      "learning_rate": 1.179761563844589e-05,
+      "loss": 0.1547,
+      "mean_token_accuracy": 0.9387565493583679,
+      "num_tokens": 58448138.0,
+      "step": 935
+    },
+    {
+      "entropy": 0.15494557321071625,
+      "epoch": 0.5016008537886874,
+      "grad_norm": 0.17864580074050054,
+      "learning_rate": 1.1705889576972243e-05,
+      "loss": 0.1556,
+      "mean_token_accuracy": 0.9387572526931762,
+      "num_tokens": 58759567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.15362054109573364,
+      "epoch": 0.5042689434364994,
+      "grad_norm": 0.17875371335143,
+      "learning_rate": 1.1614015443687723e-05,
+      "loss": 0.1541,
+      "mean_token_accuracy": 0.9391363382339477,
+      "num_tokens": 59072420.0,
+      "step": 945
+    },
+    {
+      "entropy": 0.15368084311485292,
+      "epoch": 0.5069370330843116,
+      "grad_norm": 0.20278372952945256,
+      "learning_rate": 1.1522001213299176e-05,
+      "loss": 0.1541,
+      "mean_token_accuracy": 0.9390037655830383,
+      "num_tokens": 59385414.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.15587687492370605,
+      "epoch": 0.5096051227321238,
+      "grad_norm": 0.20036906160941148,
+      "learning_rate": 1.1429854872673927e-05,
+      "loss": 0.1557,
+      "mean_token_accuracy": 0.9383829355239868,
+      "num_tokens": 59698209.0,
+      "step": 955
+    },
+    {
+      "entropy": 0.15657925903797149,
+      "epoch": 0.512273212379936,
+      "grad_norm": 0.2518678485978799,
+      "learning_rate": 1.133758442014651e-05,
+      "loss": 0.1575,
+      "mean_token_accuracy": 0.9377983450889588,
+      "num_tokens": 60007386.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.1564165860414505,
+      "epoch": 0.5149413020277481,
+      "grad_norm": 0.21361188248461715,
+      "learning_rate": 1.1245197864824426e-05,
+      "loss": 0.1567,
+      "mean_token_accuracy": 0.9377204418182373,
+      "num_tokens": 60317545.0,
+      "step": 965
+    },
+    {
+      "entropy": 0.1573189228773117,
+      "epoch": 0.5176093916755603,
+      "grad_norm": 0.21766212031152615,
+      "learning_rate": 1.1152703225892929e-05,
+      "loss": 0.1563,
+      "mean_token_accuracy": 0.938551950454712,
+      "num_tokens": 60629190.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.15436407625675203,
+      "epoch": 0.5202774813233725,
+      "grad_norm": 0.260368916857354,
+      "learning_rate": 1.1060108531918972e-05,
+      "loss": 0.1553,
+      "mean_token_accuracy": 0.9390477299690246,
+      "num_tokens": 60941397.0,
+      "step": 975
+    },
+    {
+      "entropy": 0.15507691204547883,
+      "epoch": 0.5229455709711847,
+      "grad_norm": 0.1962543554869463,
+      "learning_rate": 1.0967421820154319e-05,
+      "loss": 0.1564,
+      "mean_token_accuracy": 0.9386341333389282,
+      "num_tokens": 61250900.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.15573692321777344,
+      "epoch": 0.5256136606189968,
+      "grad_norm": 0.22480820207139385,
+      "learning_rate": 1.0874651135837912e-05,
+      "loss": 0.1557,
+      "mean_token_accuracy": 0.9386470794677735,
+      "num_tokens": 61562372.0,
+      "step": 985
+    },
+    {
+      "entropy": 0.15311691761016846,
+      "epoch": 0.528281750266809,
+      "grad_norm": 0.20634597184754916,
+      "learning_rate": 1.078180453149754e-05,
+      "loss": 0.1527,
+      "mean_token_accuracy": 0.9395962953567505,
+      "num_tokens": 61880598.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.15343199968338012,
+      "epoch": 0.5309498399146211,
+      "grad_norm": 0.1883659554769896,
+      "learning_rate": 1.0688890066250876e-05,
+      "loss": 0.1528,
+      "mean_token_accuracy": 0.9397060394287109,
+      "num_tokens": 62193244.0,
+      "step": 995
+    },
+    {
+      "entropy": 0.15367394387722016,
+      "epoch": 0.5336179295624333,
+      "grad_norm": 0.199812419140156,
+      "learning_rate": 1.0595915805105945e-05,
+      "loss": 0.1543,
+      "mean_token_accuracy": 0.9393554091453552,
+      "num_tokens": 62507530.0,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5336179295624333,
+      "eval_entropy": 0.1538348892067052,
+      "eval_loss": 0.15534012019634247,
+      "eval_mean_token_accuracy": 0.9388755446747888,
+      "eval_num_tokens": 62507530.0,
+      "eval_runtime": 17.6249,
+      "eval_samples_per_second": 283.69,
+      "eval_steps_per_second": 4.482,
+      "step": 1000
+    },
+    {
+      "entropy": 0.15436525344848634,
+      "epoch": 0.5362860192102454,
+      "grad_norm": 0.19576898569996617,
+      "learning_rate": 1.0502889818261075e-05,
+      "loss": 0.1554,
+      "mean_token_accuracy": 0.9384722709655762,
+      "num_tokens": 62818270.0,
+      "step": 1005
+    },
+    {
+      "entropy": 0.15489275455474855,
+      "epoch": 0.5389541088580576,
+      "grad_norm": 0.20047233121126237,
+      "learning_rate": 1.04098201804044e-05,
+      "loss": 0.1542,
+      "mean_token_accuracy": 0.9395401954650879,
+      "num_tokens": 63129173.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.15460718274116517,
+      "epoch": 0.5416221985058698,
+      "grad_norm": 0.19314743208982307,
+      "learning_rate": 1.031671497001298e-05,
+      "loss": 0.1548,
+      "mean_token_accuracy": 0.9389819502830505,
+      "num_tokens": 63440789.0,
+      "step": 1015
+    },
+    {
+      "entropy": 0.15471943616867065,
+      "epoch": 0.544290288153682,
+      "grad_norm": 0.23882474744234125,
+      "learning_rate": 1.0223582268651585e-05,
+      "loss": 0.1552,
+      "mean_token_accuracy": 0.9385024428367614,
+      "num_tokens": 63752159.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.15407437086105347,
+      "epoch": 0.5469583778014941,
+      "grad_norm": 0.20795985933204394,
+      "learning_rate": 1.0130430160271215e-05,
+      "loss": 0.1551,
+      "mean_token_accuracy": 0.9385804533958435,
+      "num_tokens": 64063845.0,
+      "step": 1025
+    },
+    {
+      "entropy": 0.1559080123901367,
+      "epoch": 0.5496264674493063,
+      "grad_norm": 0.19061876365796376,
+      "learning_rate": 1.0037266730507394e-05,
+      "loss": 0.1551,
+      "mean_token_accuracy": 0.9385924577713013,
+      "num_tokens": 64376046.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.15590772330760955,
+      "epoch": 0.5522945570971185,
+      "grad_norm": 0.19744593251536702,
+      "learning_rate": 9.944100065978351e-06,
+      "loss": 0.1555,
+      "mean_token_accuracy": 0.938746166229248,
+      "num_tokens": 64687857.0,
+      "step": 1035
+    },
+    {
+      "entropy": 0.15375302731990814,
+      "epoch": 0.5549626467449307,
+      "grad_norm": 0.19950393668783845,
+      "learning_rate": 9.850938253583102e-06,
+      "loss": 0.1549,
+      "mean_token_accuracy": 0.9387599468231201,
+      "num_tokens": 64998447.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.15291616320610046,
+      "epoch": 0.5576307363927427,
+      "grad_norm": 0.19509972625285316,
+      "learning_rate": 9.757789379799482e-06,
+      "loss": 0.1532,
+      "mean_token_accuracy": 0.9396419167518616,
+      "num_tokens": 65313380.0,
+      "step": 1045
+    },
+    {
+      "entropy": 0.15656664967536926,
+      "epoch": 0.5602988260405549,
+      "grad_norm": 0.20136336987602882,
+      "learning_rate": 9.664661529982261e-06,
+      "loss": 0.1559,
+      "mean_token_accuracy": 0.9382720947265625,
+      "num_tokens": 65624880.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.15257844924926758,
+      "epoch": 0.5629669156883671,
+      "grad_norm": 0.1796732976273371,
+      "learning_rate": 9.571562787661316e-06,
+      "loss": 0.1528,
+      "mean_token_accuracy": 0.9397455334663392,
+      "num_tokens": 65938756.0,
+      "step": 1055
+    },
+    {
+      "entropy": 0.15621959567070007,
+      "epoch": 0.5656350053361793,
+      "grad_norm": 0.190132157006254,
+      "learning_rate": 9.47850123383999e-06,
+      "loss": 0.1558,
+      "mean_token_accuracy": 0.9388103485107422,
+      "num_tokens": 66247340.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.15296345055103303,
+      "epoch": 0.5683030949839915,
+      "grad_norm": 0.17706775442899994,
+      "learning_rate": 9.385484946293636e-06,
+      "loss": 0.154,
+      "mean_token_accuracy": 0.9390620470046998,
+      "num_tokens": 66560725.0,
+      "step": 1065
+    },
+    {
+      "entropy": 0.1538401484489441,
+      "epoch": 0.5709711846318036,
+      "grad_norm": 0.23749640187460702,
+      "learning_rate": 9.292521998868498e-06,
+      "loss": 0.1542,
+      "mean_token_accuracy": 0.9392606258392334,
+      "num_tokens": 66871322.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.1535812646150589,
+      "epoch": 0.5736392742796158,
+      "grad_norm": 0.224556333435221,
+      "learning_rate": 9.199620460780867e-06,
+      "loss": 0.1542,
+      "mean_token_accuracy": 0.93891681432724,
+      "num_tokens": 67181595.0,
+      "step": 1075
+    },
+    {
+      "entropy": 0.1573965609073639,
+      "epoch": 0.576307363927428,
+      "grad_norm": 0.19815175316382136,
+      "learning_rate": 9.106788395916679e-06,
+      "loss": 0.1565,
+      "mean_token_accuracy": 0.9381605625152588,
+      "num_tokens": 67492926.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.15403612852096557,
+      "epoch": 0.5789754535752402,
+      "grad_norm": 0.1840244632498643,
+      "learning_rate": 9.014033862131583e-06,
+      "loss": 0.1536,
+      "mean_token_accuracy": 0.9394134044647217,
+      "num_tokens": 67808031.0,
+      "step": 1085
+    },
+    {
+      "entropy": 0.15346834659576417,
+      "epoch": 0.5816435432230523,
+      "grad_norm": 0.21717458002978238,
+      "learning_rate": 8.92136491055149e-06,
+      "loss": 0.1548,
+      "mean_token_accuracy": 0.9388310194015503,
+      "num_tokens": 68119639.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.15523187518119813,
+      "epoch": 0.5843116328708644,
+      "grad_norm": 0.1840200166212005,
+      "learning_rate": 8.828789584873754e-06,
+      "loss": 0.1558,
+      "mean_token_accuracy": 0.9386077761650086,
+      "num_tokens": 68429603.0,
+      "step": 1095
+    },
+    {
+      "entropy": 0.15351728796958924,
+      "epoch": 0.5869797225186766,
+      "grad_norm": 0.21035998185165955,
+      "learning_rate": 8.736315920668968e-06,
+      "loss": 0.153,
+      "mean_token_accuracy": 0.939343523979187,
+      "num_tokens": 68741574.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.15493020713329314,
+      "epoch": 0.5896478121664888,
+      "grad_norm": 0.2262869104676659,
+      "learning_rate": 8.643951944683466e-06,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.9387609839439393,
+      "num_tokens": 69052450.0,
+      "step": 1105
+    },
+    {
+      "entropy": 0.1536796510219574,
+      "epoch": 0.5923159018143009,
+      "grad_norm": 0.2588445878760737,
+      "learning_rate": 8.551705674142618e-06,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.9383743643760681,
+      "num_tokens": 69363673.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.15553466677665712,
+      "epoch": 0.5949839914621131,
+      "grad_norm": 0.24222053348739628,
+      "learning_rate": 8.459585116054903e-06,
+      "loss": 0.1559,
+      "mean_token_accuracy": 0.9381522536277771,
+      "num_tokens": 69676750.0,
+      "step": 1115
+    },
+    {
+      "entropy": 0.15311868786811828,
+      "epoch": 0.5976520811099253,
+      "grad_norm": 0.17614863141030848,
+      "learning_rate": 8.367598266516934e-06,
+      "loss": 0.1523,
+      "mean_token_accuracy": 0.9392088413238525,
+      "num_tokens": 69991603.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.15428132712841033,
+      "epoch": 0.6003201707577375,
+      "grad_norm": 0.1998512753996338,
+      "learning_rate": 8.275753110019367e-06,
+      "loss": 0.1547,
+      "mean_token_accuracy": 0.938707685470581,
+      "num_tokens": 70306692.0,
+      "step": 1125
+    },
+    {
+      "entropy": 0.15532322824001313,
+      "epoch": 0.6029882604055496,
+      "grad_norm": 0.24332810412676226,
+      "learning_rate": 8.184057618753849e-06,
+      "loss": 0.1556,
+      "mean_token_accuracy": 0.9380664229393005,
+      "num_tokens": 70618028.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.15342488884925842,
+      "epoch": 0.6056563500533618,
+      "grad_norm": 0.17933966138875423,
+      "learning_rate": 8.09251975192104e-06,
+      "loss": 0.1542,
+      "mean_token_accuracy": 0.9395575881004333,
+      "num_tokens": 70931744.0,
+      "step": 1135
+    },
+    {
+      "entropy": 0.1545114040374756,
+      "epoch": 0.608324439701174,
+      "grad_norm": 0.21466920390671304,
+      "learning_rate": 8.001147455039735e-06,
+      "loss": 0.1553,
+      "mean_token_accuracy": 0.9387154817581177,
+      "num_tokens": 71244224.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.1548311859369278,
+      "epoch": 0.6109925293489862,
+      "grad_norm": 0.24490082669063587,
+      "learning_rate": 7.909948659257213e-06,
+      "loss": 0.1536,
+      "mean_token_accuracy": 0.9394757270812988,
+      "num_tokens": 71558782.0,
+      "step": 1145
+    },
+    {
+      "entropy": 0.15591164231300353,
+      "epoch": 0.6136606189967982,
+      "grad_norm": 0.19745696634708404,
+      "learning_rate": 7.818931280660781e-06,
+      "loss": 0.1558,
+      "mean_token_accuracy": 0.9384572505950928,
+      "num_tokens": 71868069.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.15378665030002595,
+      "epoch": 0.6163287086446104,
+      "grad_norm": 0.18610092854623703,
+      "learning_rate": 7.72810321959068e-06,
+      "loss": 0.1538,
+      "mean_token_accuracy": 0.9386947751045227,
+      "num_tokens": 72179019.0,
+      "step": 1155
+    },
+    {
+      "entropy": 0.15749258995056153,
+      "epoch": 0.6189967982924226,
+      "grad_norm": 0.19344944196983696,
+      "learning_rate": 7.63747235995433e-06,
+      "loss": 0.1574,
+      "mean_token_accuracy": 0.9379115223884582,
+      "num_tokens": 72487777.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.151074355840683,
+      "epoch": 0.6216648879402348,
+      "grad_norm": 0.21666750191470913,
+      "learning_rate": 7.54704656854198e-06,
+      "loss": 0.1518,
+      "mean_token_accuracy": 0.9397258996963501,
+      "num_tokens": 72805762.0,
+      "step": 1165
+    },
+    {
+      "entropy": 0.1522180676460266,
+      "epoch": 0.624332977588047,
+      "grad_norm": 0.1819485518927904,
+      "learning_rate": 7.4568336943439055e-06,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.939635157585144,
+      "num_tokens": 73118436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.15505608022212983,
+      "epoch": 0.6270010672358591,
+      "grad_norm": 0.17663094640924515,
+      "learning_rate": 7.3668415678690745e-06,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.9385998487472534,
+      "num_tokens": 73431334.0,
+      "step": 1175
+    },
+    {
+      "entropy": 0.15457534790039062,
+      "epoch": 0.6296691568836713,
+      "grad_norm": 0.19274527895852187,
+      "learning_rate": 7.277078000465487e-06,
+      "loss": 0.1549,
+      "mean_token_accuracy": 0.9389039039611816,
+      "num_tokens": 73742853.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.15383411943912506,
+      "epoch": 0.6323372465314835,
+      "grad_norm": 0.2045826633338559,
+      "learning_rate": 7.187550783642141e-06,
+      "loss": 0.1539,
+      "mean_token_accuracy": 0.9392240285873413,
+      "num_tokens": 74056292.0,
+      "step": 1185
+    },
+    {
+      "entropy": 0.15572645366191865,
+      "epoch": 0.6350053361792957,
+      "grad_norm": 0.1740911980701786,
+      "learning_rate": 7.098267688392702e-06,
+      "loss": 0.1554,
+      "mean_token_accuracy": 0.9382194876670837,
+      "num_tokens": 74369090.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.15480996966361998,
+      "epoch": 0.6376734258271078,
+      "grad_norm": 0.1958267352216853,
+      "learning_rate": 7.009236464521021e-06,
+      "loss": 0.1552,
+      "mean_token_accuracy": 0.9386586546897888,
+      "num_tokens": 74680567.0,
+      "step": 1195
+    },
+    {
+      "entropy": 0.1536591023206711,
+      "epoch": 0.6403415154749199,
+      "grad_norm": 0.20969055435632733,
+      "learning_rate": 6.920464839968405e-06,
+      "loss": 0.1534,
+      "mean_token_accuracy": 0.9394211530685425,
+      "num_tokens": 74990800.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.1548810452222824,
+      "epoch": 0.6430096051227321,
+      "grad_norm": 0.17777373225867352,
+      "learning_rate": 6.831960520142868e-06,
+      "loss": 0.1542,
+      "mean_token_accuracy": 0.9388859510421753,
+      "num_tokens": 75303707.0,
+      "step": 1205
+    },
+    {
+      "entropy": 0.15442072451114655,
+      "epoch": 0.6456776947705443,
+      "grad_norm": 0.2054630769538203,
+      "learning_rate": 6.743731187250262e-06,
+      "loss": 0.1548,
+      "mean_token_accuracy": 0.9383596777915955,
+      "num_tokens": 75615973.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.15551903247833251,
+      "epoch": 0.6483457844183564,
+      "grad_norm": 0.2388470280787671,
+      "learning_rate": 6.655784499627491e-06,
+      "loss": 0.1555,
+      "mean_token_accuracy": 0.9386810302734375,
+      "num_tokens": 75926254.0,
+      "step": 1215
+    },
+    {
+      "entropy": 0.15150526762008668,
+      "epoch": 0.6510138740661686,
+      "grad_norm": 0.1896593043296809,
+      "learning_rate": 6.56812809107775e-06,
+      "loss": 0.1522,
+      "mean_token_accuracy": 0.9392547249794007,
+      "num_tokens": 76241390.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.15170768201351165,
+      "epoch": 0.6536819637139808,
+      "grad_norm": 0.17975435195878658,
+      "learning_rate": 6.480769570207897e-06,
+      "loss": 0.1523,
+      "mean_token_accuracy": 0.939859139919281,
+      "num_tokens": 76556998.0,
+      "step": 1225
+    },
+    {
+      "entropy": 0.15482379198074342,
+      "epoch": 0.656350053361793,
+      "grad_norm": 0.2067120427304762,
+      "learning_rate": 6.393716519768047e-06,
+      "loss": 0.156,
+      "mean_token_accuracy": 0.9383298635482789,
+      "num_tokens": 76868707.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.1551853358745575,
+      "epoch": 0.6590181430096052,
+      "grad_norm": 0.17199612564283812,
+      "learning_rate": 6.306976495993373e-06,
+      "loss": 0.1531,
+      "mean_token_accuracy": 0.9393068075180053,
+      "num_tokens": 77181461.0,
+      "step": 1235
+    },
+    {
+      "entropy": 0.1531525045633316,
+      "epoch": 0.6616862326574173,
+      "grad_norm": 0.20032722239251838,
+      "learning_rate": 6.220557027948222e-06,
+      "loss": 0.1527,
+      "mean_token_accuracy": 0.940034294128418,
+      "num_tokens": 77494717.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.1529313951730728,
+      "epoch": 0.6643543223052295,
+      "grad_norm": 0.18132538819196686,
+      "learning_rate": 6.134465616872598e-06,
+      "loss": 0.1539,
+      "mean_token_accuracy": 0.9392167925834656,
+      "num_tokens": 77807893.0,
+      "step": 1245
+    },
+    {
+      "entropy": 0.154258593916893,
+      "epoch": 0.6670224119530416,
+      "grad_norm": 0.18572593550286834,
+      "learning_rate": 6.04870973553103e-06,
+      "loss": 0.155,
+      "mean_token_accuracy": 0.9384752988815308,
+      "num_tokens": 78115560.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.157048237323761,
+      "epoch": 0.6696905016008537,
+      "grad_norm": 0.19007712881646377,
+      "learning_rate": 5.963296827563969e-06,
+      "loss": 0.1565,
+      "mean_token_accuracy": 0.938205361366272,
+      "num_tokens": 78425131.0,
+      "step": 1255
+    },
+    {
+      "entropy": 0.15400311648845671,
+      "epoch": 0.6723585912486659,
+      "grad_norm": 0.17992885462267433,
+      "learning_rate": 5.878234306841637e-06,
+      "loss": 0.1539,
+      "mean_token_accuracy": 0.9387377262115478,
+      "num_tokens": 78740041.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.15463869869709015,
+      "epoch": 0.6750266808964781,
+      "grad_norm": 0.16967767258074606,
+      "learning_rate": 5.793529556820538e-06,
+      "loss": 0.1542,
+      "mean_token_accuracy": 0.9388965487480163,
+      "num_tokens": 79050088.0,
+      "step": 1265
+    },
+    {
+      "entropy": 0.1539893090724945,
+      "epoch": 0.6776947705442903,
+      "grad_norm": 0.2124727900050765,
+      "learning_rate": 5.709189929902551e-06,
+      "loss": 0.1541,
+      "mean_token_accuracy": 0.9387233018875122,
+      "num_tokens": 79363843.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.15476358234882354,
+      "epoch": 0.6803628601921025,
+      "grad_norm": 0.20572465697212475,
+      "learning_rate": 5.62522274679673e-06,
+      "loss": 0.1545,
+      "mean_token_accuracy": 0.9390263795852661,
+      "num_tokens": 79675480.0,
+      "step": 1275
+    },
+    {
+      "entropy": 0.15502644181251526,
+      "epoch": 0.6830309498399146,
+      "grad_norm": 0.16772458380589314,
+      "learning_rate": 5.541635295883889e-06,
+      "loss": 0.1547,
+      "mean_token_accuracy": 0.9388966798782349,
+      "num_tokens": 79988876.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.15343300998210907,
+      "epoch": 0.6856990394877268,
+      "grad_norm": 0.17283095303374088,
+      "learning_rate": 5.45843483258395e-06,
+      "loss": 0.1541,
+      "mean_token_accuracy": 0.9392352104187012,
+      "num_tokens": 80301888.0,
+      "step": 1285
+    },
+    {
+      "entropy": 0.15344253182411194,
+      "epoch": 0.688367129135539,
+      "grad_norm": 0.1947170351124985,
+      "learning_rate": 5.375628578726181e-06,
+      "loss": 0.1537,
+      "mean_token_accuracy": 0.9390166163444519,
+      "num_tokens": 80613143.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.1538625419139862,
+      "epoch": 0.6910352187833512,
+      "grad_norm": 0.18817183057447215,
+      "learning_rate": 5.293223721922319e-06,
+      "loss": 0.1536,
+      "mean_token_accuracy": 0.9387935757637024,
+      "num_tokens": 80926795.0,
+      "step": 1295
+    },
+    {
+      "entropy": 0.15280040502548217,
+      "epoch": 0.6937033084311632,
+      "grad_norm": 0.1717728161068434,
+      "learning_rate": 5.21122741494271e-06,
+      "loss": 0.1532,
+      "mean_token_accuracy": 0.9387828230857849,
+      "num_tokens": 81239514.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.15432974696159363,
+      "epoch": 0.6963713980789754,
+      "grad_norm": 0.21208211624998147,
+      "learning_rate": 5.129646775095432e-06,
+      "loss": 0.1536,
+      "mean_token_accuracy": 0.9388492107391357,
+      "num_tokens": 81551899.0,
+      "step": 1305
+    },
+    {
+      "entropy": 0.15256355106830596,
+      "epoch": 0.6990394877267876,
+      "grad_norm": 0.18456842506003865,
+      "learning_rate": 5.048488883608496e-06,
+      "loss": 0.153,
+      "mean_token_accuracy": 0.9395634889602661,
+      "num_tokens": 81868789.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.15440889596939086,
+      "epoch": 0.7017075773745998,
+      "grad_norm": 0.1800146180845423,
+      "learning_rate": 4.96776078501523e-06,
+      "loss": 0.1548,
+      "mean_token_accuracy": 0.939136517047882,
+      "num_tokens": 82179997.0,
+      "step": 1315
+    },
+    {
+      "entropy": 0.15329372584819795,
+      "epoch": 0.7043756670224119,
+      "grad_norm": 0.19968494109740412,
+      "learning_rate": 4.8874694865427676e-06,
+      "loss": 0.1527,
+      "mean_token_accuracy": 0.939085054397583,
+      "num_tokens": 82496847.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.15478273034095763,
+      "epoch": 0.7070437566702241,
+      "grad_norm": 0.16598727253128187,
+      "learning_rate": 4.807621957503853e-06,
+      "loss": 0.1544,
+      "mean_token_accuracy": 0.938567292690277,
+      "num_tokens": 82809223.0,
+      "step": 1325
+    },
+    {
+      "entropy": 0.15093289017677308,
+      "epoch": 0.7097118463180363,
+      "grad_norm": 0.17117692227910006,
+      "learning_rate": 4.728225128691888e-06,
+      "loss": 0.152,
+      "mean_token_accuracy": 0.9395733594894409,
+      "num_tokens": 83123357.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.1547090470790863,
+      "epoch": 0.7123799359658485,
+      "grad_norm": 0.19535886780678244,
+      "learning_rate": 4.649285891779327e-06,
+      "loss": 0.1534,
+      "mean_token_accuracy": 0.9391082644462585,
+      "num_tokens": 83436086.0,
+      "step": 1335
+    },
+    {
+      "entropy": 0.15347299575805665,
+      "epoch": 0.7150480256136607,
+      "grad_norm": 0.17698403460692924,
+      "learning_rate": 4.570811098719502e-06,
+      "loss": 0.1533,
+      "mean_token_accuracy": 0.9396489024162292,
+      "num_tokens": 83749042.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.15381872951984404,
+      "epoch": 0.7177161152614728,
+      "grad_norm": 0.17372376390391425,
+      "learning_rate": 4.492807561151841e-06,
+      "loss": 0.1547,
+      "mean_token_accuracy": 0.938352620601654,
+      "num_tokens": 84063674.0,
+      "step": 1345
+    },
+    {
+      "entropy": 0.15317289531230927,
+      "epoch": 0.7203842049092849,
+      "grad_norm": 0.18379765550528124,
+      "learning_rate": 4.415282049810644e-06,
+      "loss": 0.1534,
+      "mean_token_accuracy": 0.9391525387763977,
+      "num_tokens": 84378034.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.15499058365821838,
+      "epoch": 0.7230522945570971,
+      "grad_norm": 0.16958667153352305,
+      "learning_rate": 4.338241293937362e-06,
+      "loss": 0.1543,
+      "mean_token_accuracy": 0.9386964082717896,
+      "num_tokens": 84690531.0,
+      "step": 1355
+    },
+    {
+      "entropy": 0.15222392082214356,
+      "epoch": 0.7257203842049093,
+      "grad_norm": 0.19889195844150365,
+      "learning_rate": 4.261691980696502e-06,
+      "loss": 0.1522,
+      "mean_token_accuracy": 0.939692759513855,
+      "num_tokens": 85003751.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.1543875366449356,
+      "epoch": 0.7283884738527214,
+      "grad_norm": 0.17327562385573972,
+      "learning_rate": 4.185640754595183e-06,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.9386188387870789,
+      "num_tokens": 85315782.0,
+      "step": 1365
+    },
+    {
+      "entropy": 0.15334829092025756,
+      "epoch": 0.7310565635005336,
+      "grad_norm": 0.18169650993402348,
+      "learning_rate": 4.110094216906377e-06,
+      "loss": 0.1537,
+      "mean_token_accuracy": 0.9394148349761963,
+      "num_tokens": 85629325.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.15514211356639862,
+      "epoch": 0.7337246531483458,
+      "grad_norm": 0.1769358845517928,
+      "learning_rate": 4.03505892509594e-06,
+      "loss": 0.1556,
+      "mean_token_accuracy": 0.9386692762374877,
+      "num_tokens": 85936223.0,
+      "step": 1375
+    },
+    {
+      "entropy": 0.15412222743034362,
+      "epoch": 0.736392742796158,
+      "grad_norm": 0.18204823506406237,
+      "learning_rate": 3.960541392253387e-06,
+      "loss": 0.1533,
+      "mean_token_accuracy": 0.9394712686538697,
+      "num_tokens": 86249477.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.15360663831233978,
+      "epoch": 0.7390608324439701,
+      "grad_norm": 0.14263985064175763,
+      "learning_rate": 3.886548086526594e-06,
+      "loss": 0.153,
+      "mean_token_accuracy": 0.9389709830284119,
+      "num_tokens": 86559839.0,
+      "step": 1385
+    },
+    {
+      "entropy": 0.1553943246603012,
+      "epoch": 0.7417289220917823,
+      "grad_norm": 0.17680317343436647,
+      "learning_rate": 3.8130854305603325e-06,
+      "loss": 0.1551,
+      "mean_token_accuracy": 0.938764214515686,
+      "num_tokens": 86870592.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.15353876650333403,
+      "epoch": 0.7443970117395945,
+      "grad_norm": 0.18044428125254167,
+      "learning_rate": 3.740159800938784e-06,
+      "loss": 0.155,
+      "mean_token_accuracy": 0.9386942744255066,
+      "num_tokens": 87178900.0,
+      "step": 1395
+    },
+    {
+      "entropy": 0.15736522674560546,
+      "epoch": 0.7470651013874067,
+      "grad_norm": 0.19735522854846968,
+      "learning_rate": 3.667777527632066e-06,
+      "loss": 0.1564,
+      "mean_token_accuracy": 0.9383513927459717,
+      "num_tokens": 87489168.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.1552681565284729,
+      "epoch": 0.7497331910352187,
+      "grad_norm": 0.16086661998527077,
+      "learning_rate": 3.595944893446768e-06,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.9392289519309998,
+      "num_tokens": 87800261.0,
+      "step": 1405
+    },
+    {
+      "entropy": 0.15531840324401855,
+      "epoch": 0.7524012806830309,
+      "grad_norm": 0.1626277417036322,
+      "learning_rate": 3.5246681334806177e-06,
+      "loss": 0.1558,
+      "mean_token_accuracy": 0.9387146353721618,
+      "num_tokens": 88110224.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.153993684053421,
+      "epoch": 0.7550693703308431,
+      "grad_norm": 0.17744646490172236,
+      "learning_rate": 3.453953434581271e-06,
+      "loss": 0.1538,
+      "mean_token_accuracy": 0.9392658829689026,
+      "num_tokens": 88421782.0,
+      "step": 1415
+    },
+    {
+      "entropy": 0.15304548144340516,
+      "epoch": 0.7577374599786553,
+      "grad_norm": 0.16562692978370455,
+      "learning_rate": 3.383806934809274e-06,
+      "loss": 0.1527,
+      "mean_token_accuracy": 0.939368200302124,
+      "num_tokens": 88735130.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.15535095632076262,
+      "epoch": 0.7604055496264674,
+      "grad_norm": 0.19753091851214719,
+      "learning_rate": 3.314234722905302e-06,
+      "loss": 0.1552,
+      "mean_token_accuracy": 0.9386848092079163,
+      "num_tokens": 89046661.0,
+      "step": 1425
+    },
+    {
+      "entropy": 0.15178467333316803,
+      "epoch": 0.7630736392742796,
+      "grad_norm": 0.2163597483413995,
+      "learning_rate": 3.2452428377616373e-06,
+      "loss": 0.153,
+      "mean_token_accuracy": 0.939287257194519,
+      "num_tokens": 89360012.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.153414848446846,
+      "epoch": 0.7657417289220918,
+      "grad_norm": 0.1642251137191811,
+      "learning_rate": 3.176837267898002e-06,
+      "loss": 0.1526,
+      "mean_token_accuracy": 0.9395386815071106,
+      "num_tokens": 89673339.0,
+      "step": 1435
+    },
+    {
+      "entropy": 0.1548197388648987,
+      "epoch": 0.768409818569904,
+      "grad_norm": 0.18337676685613638,
+      "learning_rate": 3.1090239509417364e-06,
+      "loss": 0.1552,
+      "mean_token_accuracy": 0.9383951425552368,
+      "num_tokens": 89982235.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.15227739810943602,
+      "epoch": 0.7710779082177162,
+      "grad_norm": 0.15525725696787246,
+      "learning_rate": 3.0418087731124324e-06,
+      "loss": 0.1523,
+      "mean_token_accuracy": 0.9399517059326172,
+      "num_tokens": 90294967.0,
+      "step": 1445
+    },
+    {
+      "entropy": 0.15378317534923552,
+      "epoch": 0.7737459978655283,
+      "grad_norm": 0.16334353326521453,
+      "learning_rate": 2.9751975687109956e-06,
+      "loss": 0.1523,
+      "mean_token_accuracy": 0.9399662017822266,
+      "num_tokens": 90610717.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.15547354519367218,
+      "epoch": 0.7764140875133404,
+      "grad_norm": 0.18839341400355433,
+      "learning_rate": 2.909196119613218e-06,
+      "loss": 0.1554,
+      "mean_token_accuracy": 0.9385842323303223,
+      "num_tokens": 90923000.0,
+      "step": 1455
+    },
+    {
+      "entropy": 0.15278392732143403,
+      "epoch": 0.7790821771611526,
+      "grad_norm": 0.17757852890585457,
+      "learning_rate": 2.843810154767932e-06,
+      "loss": 0.1541,
+      "mean_token_accuracy": 0.9389323234558106,
+      "num_tokens": 91235879.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.15397576689720155,
+      "epoch": 0.7817502668089648,
+      "grad_norm": 0.15492266609314415,
+      "learning_rate": 2.779045349699708e-06,
+      "loss": 0.1544,
+      "mean_token_accuracy": 0.9384764671325684,
+      "num_tokens": 91547713.0,
+      "step": 1465
+    },
+    {
+      "entropy": 0.1529139518737793,
+      "epoch": 0.7844183564567769,
+      "grad_norm": 0.17336896184779682,
+      "learning_rate": 2.7149073260162416e-06,
+      "loss": 0.1528,
+      "mean_token_accuracy": 0.9394277930259705,
+      "num_tokens": 91859166.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.15577988028526307,
+      "epoch": 0.7870864461045891,
+      "grad_norm": 0.18392391814457118,
+      "learning_rate": 2.6514016509203843e-06,
+      "loss": 0.1537,
+      "mean_token_accuracy": 0.939464557170868,
+      "num_tokens": 92173028.0,
+      "step": 1475
+    },
+    {
+      "entropy": 0.15315627455711364,
+      "epoch": 0.7897545357524013,
+      "grad_norm": 0.16770812808389177,
+      "learning_rate": 2.588533836726901e-06,
+      "loss": 0.153,
+      "mean_token_accuracy": 0.9396589040756226,
+      "num_tokens": 92487360.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.15086018443107604,
+      "epoch": 0.7924226254002135,
+      "grad_norm": 0.1710231697211091,
+      "learning_rate": 2.5263093403840145e-06,
+      "loss": 0.1515,
+      "mean_token_accuracy": 0.9400036692619324,
+      "num_tokens": 92799729.0,
+      "step": 1485
+    },
+    {
+      "entropy": 0.15368268489837647,
+      "epoch": 0.7950907150480256,
+      "grad_norm": 0.16328574540278648,
+      "learning_rate": 2.464733562999737e-06,
+      "loss": 0.1538,
+      "mean_token_accuracy": 0.9391217470169068,
+      "num_tokens": 93109737.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.15320818424224852,
+      "epoch": 0.7977588046958378,
+      "grad_norm": 0.16400834989615978,
+      "learning_rate": 2.4038118493730366e-06,
+      "loss": 0.1528,
+      "mean_token_accuracy": 0.9395870327949524,
+      "num_tokens": 93418942.0,
+      "step": 1495
+    },
+    {
+      "entropy": 0.1534988135099411,
+      "epoch": 0.80042689434365,
+      "grad_norm": 0.17327724145407505,
+      "learning_rate": 2.3435494875299315e-06,
+      "loss": 0.1529,
+      "mean_token_accuracy": 0.9397768259048462,
+      "num_tokens": 93734428.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.15331260859966278,
+      "epoch": 0.8030949839914621,
+      "grad_norm": 0.1772078817717979,
+      "learning_rate": 2.283951708264468e-06,
+      "loss": 0.1537,
+      "mean_token_accuracy": 0.9394746541976928,
+      "num_tokens": 94046957.0,
+      "step": 1505
+    },
+    {
+      "entropy": 0.15226254165172576,
+      "epoch": 0.8057630736392742,
+      "grad_norm": 0.148092043865225,
+      "learning_rate": 2.2250236846846964e-06,
+      "loss": 0.1515,
+      "mean_token_accuracy": 0.9395742535591125,
+      "num_tokens": 94360546.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.1530079126358032,
+      "epoch": 0.8084311632870864,
+      "grad_norm": 0.16893142328693095,
+      "learning_rate": 2.1667705317636333e-06,
+      "loss": 0.153,
+      "mean_token_accuracy": 0.9392720460891724,
+      "num_tokens": 94675654.0,
+      "step": 1515
+    },
+    {
+      "entropy": 0.15376291871070863,
+      "epoch": 0.8110992529348986,
+      "grad_norm": 0.16634852143118453,
+      "learning_rate": 2.109197305895295e-06,
+      "loss": 0.1544,
+      "mean_token_accuracy": 0.9388089418411255,
+      "num_tokens": 94985985.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.15282466411590576,
+      "epoch": 0.8137673425827108,
+      "grad_norm": 0.16251016165730156,
+      "learning_rate": 2.0523090044557824e-06,
+      "loss": 0.1529,
+      "mean_token_accuracy": 0.9392230749130249,
+      "num_tokens": 95299780.0,
+      "step": 1525
+    },
+    {
+      "entropy": 0.15135302543640136,
+      "epoch": 0.816435432230523,
+      "grad_norm": 0.16157546003997317,
+      "learning_rate": 1.996110565369527e-06,
+      "loss": 0.1512,
+      "mean_token_accuracy": 0.9394175171852112,
+      "num_tokens": 95614445.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.15374623239040375,
+      "epoch": 0.8191035218783351,
+      "grad_norm": 0.18259722867426076,
+      "learning_rate": 1.940606866680663e-06,
+      "loss": 0.152,
+      "mean_token_accuracy": 0.9394948601722717,
+      "num_tokens": 95931805.0,
+      "step": 1535
+    },
+    {
+      "entropy": 0.1537939041852951,
+      "epoch": 0.8217716115261473,
+      "grad_norm": 0.16334117902599396,
+      "learning_rate": 1.8858027261296108e-06,
+      "loss": 0.1528,
+      "mean_token_accuracy": 0.939476215839386,
+      "num_tokens": 96245824.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.15337570607662201,
+      "epoch": 0.8244397011739595,
+      "grad_norm": 0.19316772677973434,
+      "learning_rate": 1.8317029007349086e-06,
+      "loss": 0.1541,
+      "mean_token_accuracy": 0.9392568707466126,
+      "num_tokens": 96556337.0,
+      "step": 1545
+    },
+    {
+      "entropy": 0.15260836780071257,
+      "epoch": 0.8271077908217717,
+      "grad_norm": 0.17986671972446405,
+      "learning_rate": 1.778312086380285e-06,
+      "loss": 0.1529,
+      "mean_token_accuracy": 0.9390943288803101,
+      "num_tokens": 96868925.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.15477918088436127,
+      "epoch": 0.8297758804695837,
+      "grad_norm": 0.16799503209068373,
+      "learning_rate": 1.7256349174070685e-06,
+      "loss": 0.1543,
+      "mean_token_accuracy": 0.9389909029006958,
+      "num_tokens": 97179267.0,
+      "step": 1555
+    },
+    {
+      "entropy": 0.15401456654071807,
+      "epoch": 0.8324439701173959,
+      "grad_norm": 0.1720200370324775,
+      "learning_rate": 1.6736759662119183e-06,
+      "loss": 0.1544,
+      "mean_token_accuracy": 0.9388466477394104,
+      "num_tokens": 97489742.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.15319326817989348,
+      "epoch": 0.8351120597652081,
+      "grad_norm": 0.15450019714705887,
+      "learning_rate": 1.6224397428499394e-06,
+      "loss": 0.1533,
+      "mean_token_accuracy": 0.939484441280365,
+      "num_tokens": 97801684.0,
+      "step": 1565
+    },
+    {
+      "entropy": 0.15360201597213746,
+      "epoch": 0.8377801494130203,
+      "grad_norm": 0.177662173258833,
+      "learning_rate": 1.571930694643209e-06,
+      "loss": 0.1537,
+      "mean_token_accuracy": 0.9389990210533142,
+      "num_tokens": 98114314.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.15317412316799164,
+      "epoch": 0.8404482390608324,
+      "grad_norm": 0.15123693889172668,
+      "learning_rate": 1.522153205794742e-06,
+      "loss": 0.1526,
+      "mean_token_accuracy": 0.9393292784690856,
+      "num_tokens": 98431194.0,
+      "step": 1575
+    },
+    {
+      "entropy": 0.1568188637495041,
+      "epoch": 0.8431163287086446,
+      "grad_norm": 0.17518595443649973,
+      "learning_rate": 1.4731115970079557e-06,
+      "loss": 0.1566,
+      "mean_token_accuracy": 0.9382672905921936,
+      "num_tokens": 98738019.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.15534501969814302,
+      "epoch": 0.8457844183564568,
+      "grad_norm": 0.1789791259956949,
+      "learning_rate": 1.4248101251116075e-06,
+      "loss": 0.1547,
+      "mean_token_accuracy": 0.9392199039459228,
+      "num_tokens": 99046421.0,
+      "step": 1585
+    },
+    {
+      "entropy": 0.15555815100669862,
+      "epoch": 0.848452508004269,
+      "grad_norm": 0.16712869095236838,
+      "learning_rate": 1.377252982690327e-06,
+      "loss": 0.1548,
+      "mean_token_accuracy": 0.9391135573387146,
+      "num_tokens": 99356306.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.15490398705005645,
+      "epoch": 0.8511205976520811,
+      "grad_norm": 0.1854224178958908,
+      "learning_rate": 1.3304442977206822e-06,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.9389813899993896,
+      "num_tokens": 99664488.0,
+      "step": 1595
+    },
+    {
+      "entropy": 0.15554279088974,
+      "epoch": 0.8537886872998933,
+      "grad_norm": 0.16236663288469788,
+      "learning_rate": 1.2843881332128649e-06,
+      "loss": 0.155,
+      "mean_token_accuracy": 0.9386430382728577,
+      "num_tokens": 99976581.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.151815003156662,
+      "epoch": 0.8564567769477054,
+      "grad_norm": 0.1704052427325928,
+      "learning_rate": 1.23908848685804e-06,
+      "loss": 0.1523,
+      "mean_token_accuracy": 0.9397621631622315,
+      "num_tokens": 100292769.0,
+      "step": 1605
+    },
+    {
+      "entropy": 0.15176697075366974,
+      "epoch": 0.8591248665955176,
+      "grad_norm": 0.1786854779796353,
+      "learning_rate": 1.1945492906813228e-06,
+      "loss": 0.1513,
+      "mean_token_accuracy": 0.940117597579956,
+      "num_tokens": 100607024.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.15364796221256255,
+      "epoch": 0.8617929562433297,
+      "grad_norm": 0.15429994926849608,
+      "learning_rate": 1.1507744107004937e-06,
+      "loss": 0.1533,
+      "mean_token_accuracy": 0.9390743613243103,
+      "num_tokens": 100919943.0,
+      "step": 1615
+    },
+    {
+      "entropy": 0.1543932318687439,
+      "epoch": 0.8644610458911419,
+      "grad_norm": 0.15625280035216202,
+      "learning_rate": 1.1077676465904209e-06,
+      "loss": 0.1538,
+      "mean_token_accuracy": 0.9391561985015869,
+      "num_tokens": 101227280.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.15176737308502197,
+      "epoch": 0.8671291355389541,
+      "grad_norm": 0.1456026421545509,
+      "learning_rate": 1.0655327313532393e-06,
+      "loss": 0.1515,
+      "mean_token_accuracy": 0.9398553967475891,
+      "num_tokens": 101541751.0,
+      "step": 1625
+    },
+    {
+      "entropy": 0.1538877785205841,
+      "epoch": 0.8697972251867663,
+      "grad_norm": 0.18290798792723664,
+      "learning_rate": 1.0240733309943408e-06,
+      "loss": 0.1536,
+      "mean_token_accuracy": 0.9390178799629212,
+      "num_tokens": 101852766.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.15422819554805756,
+      "epoch": 0.8724653148345785,
+      "grad_norm": 0.17873737576182977,
+      "learning_rate": 9.833930442041506e-07,
+      "loss": 0.1533,
+      "mean_token_accuracy": 0.9390836715698242,
+      "num_tokens": 102163302.0,
+      "step": 1635
+    },
+    {
+      "entropy": 0.15171377062797547,
+      "epoch": 0.8751334044823906,
+      "grad_norm": 0.15855802539068972,
+      "learning_rate": 9.434954020457676e-07,
+      "loss": 0.1518,
+      "mean_token_accuracy": 0.9397657155990601,
+      "num_tokens": 102475125.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.15260193645954132,
+      "epoch": 0.8778014941302028,
+      "grad_norm": 0.1726076552166529,
+      "learning_rate": 9.043838676484562e-07,
+      "loss": 0.1521,
+      "mean_token_accuracy": 0.9395378589630127,
+      "num_tokens": 102788987.0,
+      "step": 1645
+    },
+    {
+      "entropy": 0.15317799746990204,
+      "epoch": 0.880469583778015,
+      "grad_norm": 0.16488636114365046,
+      "learning_rate": 8.660618359070605e-07,
+      "loss": 0.1527,
+      "mean_token_accuracy": 0.9392109036445617,
+      "num_tokens": 103101810.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.15190463960170747,
+      "epoch": 0.8831376734258272,
+      "grad_norm": 0.18861785294109573,
+      "learning_rate": 8.285326331873156e-07,
+      "loss": 0.1522,
+      "mean_token_accuracy": 0.9398165941238403,
+      "num_tokens": 103415395.0,
+      "step": 1655
+    },
+    {
+      "entropy": 0.15620527565479278,
+      "epoch": 0.8858057630736392,
+      "grad_norm": 0.1661916421753367,
+      "learning_rate": 7.917995170371196e-07,
+      "loss": 0.1557,
+      "mean_token_accuracy": 0.9382588505744934,
+      "num_tokens": 103726299.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.15179810523986817,
+      "epoch": 0.8884738527214514,
+      "grad_norm": 0.156852390106292,
+      "learning_rate": 7.558656759037796e-07,
+      "loss": 0.152,
+      "mean_token_accuracy": 0.9394199013710022,
+      "num_tokens": 104041096.0,
+      "step": 1665
+    },
+    {
+      "entropy": 0.15225241482257842,
+      "epoch": 0.8911419423692636,
+      "grad_norm": 0.19547586460594443,
+      "learning_rate": 7.207342288572505e-07,
+      "loss": 0.1528,
+      "mean_token_accuracy": 0.9393445611000061,
+      "num_tokens": 104354220.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.15344537496566774,
+      "epoch": 0.8938100320170758,
+      "grad_norm": 0.16355590508798246,
+      "learning_rate": 6.864082253194027e-07,
+      "loss": 0.1531,
+      "mean_token_accuracy": 0.939066469669342,
+      "num_tokens": 104665581.0,
+      "step": 1675
+    },
+    {
+      "entropy": 0.1540704697370529,
+      "epoch": 0.8964781216648879,
+      "grad_norm": 0.14321416528661213,
+      "learning_rate": 6.528906447993289e-07,
+      "loss": 0.1542,
+      "mean_token_accuracy": 0.9388607501983642,
+      "num_tokens": 104975597.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.15550175607204436,
+      "epoch": 0.8991462113127001,
+      "grad_norm": 0.15605416020222349,
+      "learning_rate": 6.201843966347176e-07,
+      "loss": 0.1539,
+      "mean_token_accuracy": 0.9388849854469299,
+      "num_tokens": 105288459.0,
+      "step": 1685
+    },
+    {
+      "entropy": 0.15473953485488892,
+      "epoch": 0.9018143009605123,
+      "grad_norm": 0.17414650381201083,
+      "learning_rate": 5.882923197393331e-07,
+      "loss": 0.1539,
+      "mean_token_accuracy": 0.9390599966049195,
+      "num_tokens": 105602719.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.15401538014411925,
+      "epoch": 0.9044823906083245,
+      "grad_norm": 0.15830882724974832,
+      "learning_rate": 5.572171823565797e-07,
+      "loss": 0.1527,
+      "mean_token_accuracy": 0.9393938302993774,
+      "num_tokens": 105914271.0,
+      "step": 1695
+    },
+    {
+      "entropy": 0.15360169410705565,
+      "epoch": 0.9071504802561366,
+      "grad_norm": 0.15926528205721613,
+      "learning_rate": 5.269616818192335e-07,
+      "loss": 0.1535,
+      "mean_token_accuracy": 0.9388334393501282,
+      "num_tokens": 106227117.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.15220762491226197,
+      "epoch": 0.9098185699039488,
+      "grad_norm": 0.1791864739200382,
+      "learning_rate": 4.975284443153062e-07,
+      "loss": 0.1517,
+      "mean_token_accuracy": 0.9400452375411987,
+      "num_tokens": 106540833.0,
+      "step": 1705
+    },
+    {
+      "entropy": 0.15139750838279725,
+      "epoch": 0.9124866595517609,
+      "grad_norm": 0.17296662165772123,
+      "learning_rate": 4.6892002466008666e-07,
+      "loss": 0.1519,
+      "mean_token_accuracy": 0.9396615028381348,
+      "num_tokens": 106854815.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.1526773154735565,
+      "epoch": 0.9151547491995731,
+      "grad_norm": 0.16734739463180334,
+      "learning_rate": 4.4113890607439315e-07,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9392900347709656,
+      "num_tokens": 107167462.0,
+      "step": 1715
+    },
+    {
+      "entropy": 0.15315515995025636,
+      "epoch": 0.9178228388473852,
+      "grad_norm": 0.1537273254608466,
+      "learning_rate": 4.141874999690143e-07,
+      "loss": 0.1533,
+      "mean_token_accuracy": 0.9392496705055237,
+      "num_tokens": 107479911.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.15481925010681152,
+      "epoch": 0.9204909284951974,
+      "grad_norm": 0.17050365912615373,
+      "learning_rate": 3.8806814573541185e-07,
+      "loss": 0.1548,
+      "mean_token_accuracy": 0.9386554837226868,
+      "num_tokens": 107788344.0,
+      "step": 1725
+    },
+    {
+      "entropy": 0.15229442715644836,
+      "epoch": 0.9231590181430096,
+      "grad_norm": 0.17679327443076584,
+      "learning_rate": 3.627831105426527e-07,
+      "loss": 0.1521,
+      "mean_token_accuracy": 0.9396791219711303,
+      "num_tokens": 108105431.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.15279924273490905,
+      "epoch": 0.9258271077908218,
+      "grad_norm": 0.16359512895424472,
+      "learning_rate": 3.383345891406176e-07,
+      "loss": 0.1518,
+      "mean_token_accuracy": 0.9397311925888061,
+      "num_tokens": 108419352.0,
+      "step": 1735
+    },
+    {
+      "entropy": 0.15270444452762605,
+      "epoch": 0.928495197438634,
+      "grad_norm": 0.16633402809686326,
+      "learning_rate": 3.147247036695034e-07,
+      "loss": 0.1518,
+      "mean_token_accuracy": 0.9400157809257508,
+      "num_tokens": 108732363.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.15398914813995362,
+      "epoch": 0.9311632870864461,
+      "grad_norm": 0.18184961530665641,
+      "learning_rate": 2.919555034756083e-07,
+      "loss": 0.1534,
+      "mean_token_accuracy": 0.9391607284545899,
+      "num_tokens": 109045942.0,
+      "step": 1745
+    },
+    {
+      "entropy": 0.15341159105300903,
+      "epoch": 0.9338313767342583,
+      "grad_norm": 0.15610763899581404,
+      "learning_rate": 2.7002896493346196e-07,
+      "loss": 0.1531,
+      "mean_token_accuracy": 0.9388864755630493,
+      "num_tokens": 109357605.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.15386421084403992,
+      "epoch": 0.9364994663820705,
+      "grad_norm": 0.17064992241605773,
+      "learning_rate": 2.489469912742637e-07,
+      "loss": 0.1539,
+      "mean_token_accuracy": 0.9389752864837646,
+      "num_tokens": 109669499.0,
+      "step": 1755
+    },
+    {
+      "entropy": 0.1526539146900177,
+      "epoch": 0.9391675560298826,
+      "grad_norm": 0.15419786967921834,
+      "learning_rate": 2.2871141242068794e-07,
+      "loss": 0.1526,
+      "mean_token_accuracy": 0.9394209742546081,
+      "num_tokens": 109982520.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.15407957434654235,
+      "epoch": 0.9418356456776947,
+      "grad_norm": 0.1607104970352119,
+      "learning_rate": 2.0932398482804374e-07,
+      "loss": 0.154,
+      "mean_token_accuracy": 0.9395723819732666,
+      "num_tokens": 110291773.0,
+      "step": 1765
+    },
+    {
+      "entropy": 0.1525971472263336,
+      "epoch": 0.9445037353255069,
+      "grad_norm": 0.16079202187464134,
+      "learning_rate": 1.9078639133181532e-07,
+      "loss": 0.1525,
+      "mean_token_accuracy": 0.9390587091445923,
+      "num_tokens": 110606679.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.15261533558368684,
+      "epoch": 0.9471718249733191,
+      "grad_norm": 0.14880163094849888,
+      "learning_rate": 1.731002410015914e-07,
+      "loss": 0.1529,
+      "mean_token_accuracy": 0.9391334652900696,
+      "num_tokens": 110916597.0,
+      "step": 1775
+    },
+    {
+      "entropy": 0.15149607956409455,
+      "epoch": 0.9498399146211313,
+      "grad_norm": 0.1592225130522692,
+      "learning_rate": 1.5626706900139344e-07,
+      "loss": 0.1506,
+      "mean_token_accuracy": 0.9401805758476257,
+      "num_tokens": 111232923.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.15343788266181946,
+      "epoch": 0.9525080042689434,
+      "grad_norm": 0.17284457111616702,
+      "learning_rate": 1.4028833645643113e-07,
+      "loss": 0.1529,
+      "mean_token_accuracy": 0.9393085956573486,
+      "num_tokens": 111542260.0,
+      "step": 1785
+    },
+    {
+      "entropy": 0.15213734805583953,
+      "epoch": 0.9551760939167556,
+      "grad_norm": 0.18244908291766967,
+      "learning_rate": 1.2516543032626393e-07,
+      "loss": 0.1516,
+      "mean_token_accuracy": 0.9401678204536438,
+      "num_tokens": 111855933.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.1516744911670685,
+      "epoch": 0.9578441835645678,
+      "grad_norm": 0.15801809576170697,
+      "learning_rate": 1.1089966328442836e-07,
+      "loss": 0.1518,
+      "mean_token_accuracy": 0.9402112603187561,
+      "num_tokens": 112167971.0,
+      "step": 1795
+    },
+    {
+      "entropy": 0.15328632593154906,
+      "epoch": 0.96051227321238,
+      "grad_norm": 0.1582365808562706,
+      "learning_rate": 9.749227360448143e-08,
+      "loss": 0.153,
+      "mean_token_accuracy": 0.9395403981208801,
+      "num_tokens": 112479502.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.1515852242708206,
+      "epoch": 0.9631803628601922,
+      "grad_norm": 0.157383395307447,
+      "learning_rate": 8.494442505252998e-08,
+      "loss": 0.1517,
+      "mean_token_accuracy": 0.939778745174408,
+      "num_tokens": 112794798.0,
+      "step": 1805
+    },
+    {
+      "entropy": 0.15518470108509064,
+      "epoch": 0.9658484525080042,
+      "grad_norm": 0.1608204572688974,
+      "learning_rate": 7.325720678620807e-08,
+      "loss": 0.1541,
+      "mean_token_accuracy": 0.938652515411377,
+      "num_tokens": 113106794.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.15322840213775635,
+      "epoch": 0.9685165421558164,
+      "grad_norm": 0.1729188603078381,
+      "learning_rate": 6.243163326014268e-08,
+      "loss": 0.153,
+      "mean_token_accuracy": 0.9392426013946533,
+      "num_tokens": 113417776.0,
+      "step": 1815
+    },
+    {
+      "entropy": 0.1549760401248932,
+      "epoch": 0.9711846318036286,
+      "grad_norm": 0.1553030515697785,
+      "learning_rate": 5.2468644137894096e-08,
+      "loss": 0.1544,
+      "mean_token_accuracy": 0.9385666251182556,
+      "num_tokens": 113727813.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.15276177525520324,
+      "epoch": 0.9738527214514408,
+      "grad_norm": 0.16417299638392785,
+      "learning_rate": 4.336910421039897e-08,
+      "loss": 0.1523,
+      "mean_token_accuracy": 0.9392683863639831,
+      "num_tokens": 114040983.0,
+      "step": 1825
+    },
+    {
+      "entropy": 0.15231212675571443,
+      "epoch": 0.9765208110992529,
+      "grad_norm": 0.16508718962458294,
+      "learning_rate": 3.5133803320897e-08,
+      "loss": 0.1533,
+      "mean_token_accuracy": 0.9389925718307495,
+      "num_tokens": 114355331.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.1533448129892349,
+      "epoch": 0.9791889007470651,
+      "grad_norm": 0.15896056804887643,
+      "learning_rate": 2.776345629638355e-08,
+      "loss": 0.1523,
+      "mean_token_accuracy": 0.9398018479347229,
+      "num_tokens": 114668137.0,
+      "step": 1835
+    },
+    {
+      "entropy": 0.15175502002239227,
+      "epoch": 0.9818569903948773,
+      "grad_norm": 0.15113393376615555,
+      "learning_rate": 2.1258702885551542e-08,
+      "loss": 0.152,
+      "mean_token_accuracy": 0.9397669315338135,
+      "num_tokens": 114980396.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.15319037735462188,
+      "epoch": 0.9845250800426895,
+      "grad_norm": 0.15994427039320552,
+      "learning_rate": 1.562010770326916e-08,
+      "loss": 0.1521,
+      "mean_token_accuracy": 0.9401495337486268,
+      "num_tokens": 115291239.0,
+      "step": 1845
+    },
+    {
+      "entropy": 0.153715580701828,
+      "epoch": 0.9871931696905016,
+      "grad_norm": 0.16228745702016076,
+      "learning_rate": 1.084816018156576e-08,
+      "loss": 0.1529,
+      "mean_token_accuracy": 0.9394945859909057,
+      "num_tokens": 115604910.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.1548271507024765,
+      "epoch": 0.9898612593383138,
+      "grad_norm": 0.16123378641612207,
+      "learning_rate": 6.9432745271535986e-09,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.9384868979454041,
+      "num_tokens": 115915463.0,
+      "step": 1855
+    },
+    {
+      "entropy": 0.15217451751232147,
+      "epoch": 0.9925293489861259,
+      "grad_norm": 0.16271968941599155,
+      "learning_rate": 3.905789685471062e-09,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9392788529396057,
+      "num_tokens": 116228046.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.15162586271762848,
+      "epoch": 0.9951974386339381,
+      "grad_norm": 0.1602092482993034,
+      "learning_rate": 1.7359693112606323e-09,
+      "loss": 0.1517,
+      "mean_token_accuracy": 0.9397766590118408,
+      "num_tokens": 116542893.0,
+      "step": 1865
+    },
+    {
+      "entropy": 0.15208709836006165,
+      "epoch": 0.9978655282817502,
+      "grad_norm": 0.16954975469697164,
+      "learning_rate": 4.3400174569052657e-10,
+      "loss": 0.1517,
+      "mean_token_accuracy": 0.9398512244224548,
+      "num_tokens": 116858029.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.15305225551128387,
+      "epoch": 1.0,
+      "mean_token_accuracy": 0.9391315132379532,
+      "num_tokens": 117107840.0,
+      "step": 1874,
+      "total_flos": 246067485081600.0,
+      "train_loss": 0.17793139662661192,
+      "train_runtime": 3132.2908,
+      "train_samples_per_second": 76.578,
+      "train_steps_per_second": 0.598
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1874,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 246067485081600.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}