diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8243 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 820,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3228955268859863,
+      "epoch": 0.0024390243902439024,
+      "grad_norm": 2.4187467098236084,
+      "learning_rate": 0.0,
+      "loss": 2.5739,
+      "mean_token_accuracy": 0.5249772071838379,
+      "num_tokens": 13188.0,
+      "step": 1
+    },
+    {
+      "entropy": 1.3434489965438843,
+      "epoch": 0.004878048780487805,
+      "grad_norm": 2.1801252365112305,
+      "learning_rate": 1.2195121951219514e-07,
+      "loss": 2.5104,
+      "mean_token_accuracy": 0.5338258147239685,
+      "num_tokens": 26877.0,
+      "step": 2
+    },
+    {
+      "entropy": 1.338788390159607,
+      "epoch": 0.007317073170731708,
+      "grad_norm": 2.2766218185424805,
+      "learning_rate": 2.439024390243903e-07,
+      "loss": 2.539,
+      "mean_token_accuracy": 0.5276630520820618,
+      "num_tokens": 40449.0,
+      "step": 3
+    },
+    {
+      "entropy": 1.3349497318267822,
+      "epoch": 0.00975609756097561,
+      "grad_norm": 2.353504180908203,
+      "learning_rate": 3.6585365853658536e-07,
+      "loss": 2.5631,
+      "mean_token_accuracy": 0.5275854468345642,
+      "num_tokens": 53606.0,
+      "step": 4
+    },
+    {
+      "entropy": 1.3273138999938965,
+      "epoch": 0.012195121951219513,
+      "grad_norm": 2.307615041732788,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 2.5412,
+      "mean_token_accuracy": 0.5292984843254089,
+      "num_tokens": 67121.0,
+      "step": 5
+    },
+    {
+      "entropy": 1.3538919687271118,
+      "epoch": 0.014634146341463415,
+      "grad_norm": 2.3148252964019775,
+      "learning_rate": 6.097560975609757e-07,
+      "loss": 2.5457,
+      "mean_token_accuracy": 0.5319257378578186,
+      "num_tokens": 80872.0,
+      "step": 6
+    },
+    {
+      "entropy": 1.3218836784362793,
+      "epoch": 0.01707317073170732,
+      "grad_norm": 2.2965075969696045,
+      "learning_rate": 7.317073170731707e-07,
+      "loss": 2.5325,
+      "mean_token_accuracy": 0.5365909337997437,
+      "num_tokens": 94088.0,
+      "step": 7
+    },
+    {
+      "entropy": 1.3360720872879028,
+      "epoch": 0.01951219512195122,
+      "grad_norm": 2.219707727432251,
+      "learning_rate": 8.53658536585366e-07,
+      "loss": 2.5082,
+      "mean_token_accuracy": 0.5307488441467285,
+      "num_tokens": 107698.0,
+      "step": 8
+    },
+    {
+      "entropy": 1.3242058753967285,
+      "epoch": 0.02195121951219512,
+      "grad_norm": 2.1611530780792236,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 2.4874,
+      "mean_token_accuracy": 0.5399101376533508,
+      "num_tokens": 121733.0,
+      "step": 9
+    },
+    {
+      "entropy": 1.3375442028045654,
+      "epoch": 0.024390243902439025,
+      "grad_norm": 2.1313178539276123,
+      "learning_rate": 1.0975609756097562e-06,
+      "loss": 2.4416,
+      "mean_token_accuracy": 0.5365526676177979,
+      "num_tokens": 135934.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3702865839004517,
+      "epoch": 0.026829268292682926,
+      "grad_norm": 2.212796688079834,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 2.5007,
+      "mean_token_accuracy": 0.5316236615180969,
+      "num_tokens": 149832.0,
+      "step": 11
+    },
+    {
+      "entropy": 1.3510187864303589,
+      "epoch": 0.02926829268292683,
+      "grad_norm": 2.1613738536834717,
+      "learning_rate": 1.3414634146341465e-06,
+      "loss": 2.4974,
+      "mean_token_accuracy": 0.5329979062080383,
+      "num_tokens": 163364.0,
+      "step": 12
+    },
+    {
+      "entropy": 1.3398377895355225,
+      "epoch": 0.03170731707317073,
+      "grad_norm": 2.3386099338531494,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 2.5261,
+      "mean_token_accuracy": 0.5286726355552673,
+      "num_tokens": 176511.0,
+      "step": 13
+    },
+    {
+      "entropy": 1.3582022190093994,
+      "epoch": 0.03414634146341464,
+      "grad_norm": 2.2629709243774414,
+      "learning_rate": 1.5853658536585368e-06,
+      "loss": 2.507,
+      "mean_token_accuracy": 0.5306152701377869,
+      "num_tokens": 190115.0,
+      "step": 14
+    },
+    {
+      "entropy": 1.3705586194992065,
+      "epoch": 0.036585365853658534,
+      "grad_norm": 2.2367234230041504,
+      "learning_rate": 1.707317073170732e-06,
+      "loss": 2.4833,
+      "mean_token_accuracy": 0.5337990522384644,
+      "num_tokens": 203874.0,
+      "step": 15
+    },
+    {
+      "entropy": 1.3703287839889526,
+      "epoch": 0.03902439024390244,
+      "grad_norm": 2.342644691467285,
+      "learning_rate": 1.8292682926829268e-06,
+      "loss": 2.5473,
+      "mean_token_accuracy": 0.5246161818504333,
+      "num_tokens": 217113.0,
+      "step": 16
+    },
+    {
+      "entropy": 1.3541935682296753,
+      "epoch": 0.041463414634146344,
+      "grad_norm": 2.2277114391326904,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 2.4667,
+      "mean_token_accuracy": 0.5292628407478333,
+      "num_tokens": 230952.0,
+      "step": 17
+    },
+    {
+      "entropy": 1.3334211111068726,
+      "epoch": 0.04390243902439024,
+      "grad_norm": 2.170741081237793,
+      "learning_rate": 2.073170731707317e-06,
+      "loss": 2.4062,
+      "mean_token_accuracy": 0.5387417078018188,
+      "num_tokens": 244971.0,
+      "step": 18
+    },
+    {
+      "entropy": 1.338529109954834,
+      "epoch": 0.046341463414634146,
+      "grad_norm": 2.3073835372924805,
+      "learning_rate": 2.1951219512195125e-06,
+      "loss": 2.456,
+      "mean_token_accuracy": 0.5352984666824341,
+      "num_tokens": 258557.0,
+      "step": 19
+    },
+    {
+      "entropy": 1.347283124923706,
+      "epoch": 0.04878048780487805,
+      "grad_norm": 2.3032565116882324,
+      "learning_rate": 2.317073170731708e-06,
+      "loss": 2.4649,
+      "mean_token_accuracy": 0.5279008150100708,
+      "num_tokens": 272121.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3316829204559326,
+      "epoch": 0.05121951219512195,
+      "grad_norm": 2.3000826835632324,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 2.4087,
+      "mean_token_accuracy": 0.5390345454216003,
+      "num_tokens": 285830.0,
+      "step": 21
+    },
+    {
+      "entropy": 1.329493761062622,
+      "epoch": 0.05365853658536585,
+      "grad_norm": 2.2780513763427734,
+      "learning_rate": 2.5609756097560977e-06,
+      "loss": 2.4079,
+      "mean_token_accuracy": 0.54287189245224,
+      "num_tokens": 299398.0,
+      "step": 22
+    },
+    {
+      "entropy": 1.3588396310806274,
+      "epoch": 0.05609756097560976,
+      "grad_norm": 2.227916955947876,
+      "learning_rate": 2.682926829268293e-06,
+      "loss": 2.3605,
+      "mean_token_accuracy": 0.5377548933029175,
+      "num_tokens": 313783.0,
+      "step": 23
+    },
+    {
+      "entropy": 1.3629426956176758,
+      "epoch": 0.05853658536585366,
+      "grad_norm": 2.3609344959259033,
+      "learning_rate": 2.8048780487804884e-06,
+      "loss": 2.3981,
+      "mean_token_accuracy": 0.53337162733078,
+      "num_tokens": 327718.0,
+      "step": 24
+    },
+    {
+      "entropy": 1.3671380281448364,
+      "epoch": 0.06097560975609756,
+      "grad_norm": 2.4124526977539062,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 2.3878,
+      "mean_token_accuracy": 0.5356799364089966,
+      "num_tokens": 341257.0,
+      "step": 25
+    },
+    {
+      "entropy": 1.3848786354064941,
+      "epoch": 0.06341463414634146,
+      "grad_norm": 2.4480044841766357,
+      "learning_rate": 3.0487804878048782e-06,
+      "loss": 2.3921,
+      "mean_token_accuracy": 0.5313789248466492,
+      "num_tokens": 355088.0,
+      "step": 26
+    },
+    {
+      "entropy": 1.344458818435669,
+      "epoch": 0.06585365853658537,
+      "grad_norm": 2.3848752975463867,
+      "learning_rate": 3.1707317073170736e-06,
+      "loss": 2.3555,
+      "mean_token_accuracy": 0.5381656885147095,
+      "num_tokens": 368624.0,
+      "step": 27
+    },
+    {
+      "entropy": 1.350754976272583,
+      "epoch": 0.06829268292682927,
+      "grad_norm": 2.3436882495880127,
+      "learning_rate": 3.292682926829269e-06,
+      "loss": 2.3379,
+      "mean_token_accuracy": 0.5364958047866821,
+      "num_tokens": 382258.0,
+      "step": 28
+    },
+    {
+      "entropy": 1.370115876197815,
+      "epoch": 0.07073170731707316,
+      "grad_norm": 2.510852336883545,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 2.3477,
+      "mean_token_accuracy": 0.5365764498710632,
+      "num_tokens": 395944.0,
+      "step": 29
+    },
+    {
+      "entropy": 1.3533802032470703,
+      "epoch": 0.07317073170731707,
+      "grad_norm": 2.3963871002197266,
+      "learning_rate": 3.5365853658536588e-06,
+      "loss": 2.2746,
+      "mean_token_accuracy": 0.5382014513015747,
+      "num_tokens": 409860.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3506858348846436,
+      "epoch": 0.07560975609756097,
+      "grad_norm": 2.451052665710449,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 2.3136,
+      "mean_token_accuracy": 0.5398555994033813,
+      "num_tokens": 423174.0,
+      "step": 31
+    },
+    {
+      "entropy": 1.3611050844192505,
+      "epoch": 0.07804878048780488,
+      "grad_norm": 2.2884960174560547,
+      "learning_rate": 3.780487804878049e-06,
+      "loss": 2.2667,
+      "mean_token_accuracy": 0.544750988483429,
+      "num_tokens": 437145.0,
+      "step": 32
+    },
+    {
+      "entropy": 1.3767828941345215,
+      "epoch": 0.08048780487804878,
+      "grad_norm": 2.235347270965576,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 2.2752,
+      "mean_token_accuracy": 0.5403356552124023,
+      "num_tokens": 450747.0,
+      "step": 33
+    },
+    {
+      "entropy": 1.3706166744232178,
+      "epoch": 0.08292682926829269,
+      "grad_norm": 2.2171695232391357,
+      "learning_rate": 4.024390243902439e-06,
+      "loss": 2.2598,
+      "mean_token_accuracy": 0.5407041907310486,
+      "num_tokens": 464140.0,
+      "step": 34
+    },
+    {
+      "entropy": 1.365933895111084,
+      "epoch": 0.08536585365853659,
+      "grad_norm": 2.0118744373321533,
+      "learning_rate": 4.146341463414634e-06,
+      "loss": 2.1878,
+      "mean_token_accuracy": 0.5497426986694336,
+      "num_tokens": 478148.0,
+      "step": 35
+    },
+    {
+      "entropy": 1.3866227865219116,
+      "epoch": 0.08780487804878048,
+      "grad_norm": 2.1824421882629395,
+      "learning_rate": 4.268292682926829e-06,
+      "loss": 2.2444,
+      "mean_token_accuracy": 0.540566623210907,
+      "num_tokens": 491648.0,
+      "step": 36
+    },
+    {
+      "entropy": 1.3471767902374268,
+      "epoch": 0.09024390243902439,
+      "grad_norm": 2.1132051944732666,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 2.1395,
+      "mean_token_accuracy": 0.5577816367149353,
+      "num_tokens": 505440.0,
+      "step": 37
+    },
+    {
+      "entropy": 1.3639988899230957,
+      "epoch": 0.09268292682926829,
+      "grad_norm": 2.2273340225219727,
+      "learning_rate": 4.51219512195122e-06,
+      "loss": 2.1572,
+      "mean_token_accuracy": 0.5511302351951599,
+      "num_tokens": 519391.0,
+      "step": 38
+    },
+    {
+      "entropy": 1.3907160758972168,
+      "epoch": 0.0951219512195122,
+      "grad_norm": 2.156001567840576,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 2.1303,
+      "mean_token_accuracy": 0.550815761089325,
+      "num_tokens": 533566.0,
+      "step": 39
+    },
+    {
+      "entropy": 1.4015930891036987,
+      "epoch": 0.0975609756097561,
+      "grad_norm": 2.23702335357666,
+      "learning_rate": 4.75609756097561e-06,
+      "loss": 2.1161,
+      "mean_token_accuracy": 0.5525180697441101,
+      "num_tokens": 547263.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.4148375988006592,
+      "epoch": 0.1,
+      "grad_norm": 2.0875184535980225,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 2.1296,
+      "mean_token_accuracy": 0.5491034984588623,
+      "num_tokens": 560832.0,
+      "step": 41
+    },
+    {
+      "entropy": 1.3900272846221924,
+      "epoch": 0.1024390243902439,
+      "grad_norm": 1.8485113382339478,
+      "learning_rate": 5e-06,
+      "loss": 2.0327,
+      "mean_token_accuracy": 0.5592963099479675,
+      "num_tokens": 575343.0,
+      "step": 42
+    },
+    {
+      "entropy": 1.4016730785369873,
+      "epoch": 0.1048780487804878,
+      "grad_norm": 1.9560102224349976,
+      "learning_rate": 5.121951219512195e-06,
+      "loss": 2.0757,
+      "mean_token_accuracy": 0.5556688904762268,
+      "num_tokens": 589083.0,
+      "step": 43
+    },
+    {
+      "entropy": 1.4096416234970093,
+      "epoch": 0.1073170731707317,
+      "grad_norm": 1.9796462059020996,
+      "learning_rate": 5.243902439024391e-06,
+      "loss": 2.0705,
+      "mean_token_accuracy": 0.5589586496353149,
+      "num_tokens": 602812.0,
+      "step": 44
+    },
+    {
+      "entropy": 1.3986564874649048,
+      "epoch": 0.10975609756097561,
+      "grad_norm": 1.8606263399124146,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 2.0227,
+      "mean_token_accuracy": 0.5688395500183105,
+      "num_tokens": 616512.0,
+      "step": 45
+    },
+    {
+      "entropy": 1.3871984481811523,
+      "epoch": 0.11219512195121951,
+      "grad_norm": 1.9525058269500732,
+      "learning_rate": 5.487804878048781e-06,
+      "loss": 2.0,
+      "mean_token_accuracy": 0.5685309767723083,
+      "num_tokens": 629938.0,
+      "step": 46
+    },
+    {
+      "entropy": 1.409059762954712,
+      "epoch": 0.11463414634146342,
+      "grad_norm": 1.9902775287628174,
+      "learning_rate": 5.609756097560977e-06,
+      "loss": 1.9671,
+      "mean_token_accuracy": 0.5718567371368408,
+      "num_tokens": 643634.0,
+      "step": 47
+    },
+    {
+      "entropy": 1.4133785963058472,
+      "epoch": 0.11707317073170732,
+      "grad_norm": 1.8982359170913696,
+      "learning_rate": 5.731707317073171e-06,
+      "loss": 1.95,
+      "mean_token_accuracy": 0.5621143579483032,
+      "num_tokens": 657101.0,
+      "step": 48
+    },
+    {
+      "entropy": 1.4381654262542725,
+      "epoch": 0.11951219512195121,
+      "grad_norm": 1.773857593536377,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 1.9087,
+      "mean_token_accuracy": 0.5688751935958862,
+      "num_tokens": 671048.0,
+      "step": 49
+    },
+    {
+      "entropy": 1.4031553268432617,
+      "epoch": 0.12195121951219512,
+      "grad_norm": 1.817074179649353,
+      "learning_rate": 5.9756097560975615e-06,
+      "loss": 1.8871,
+      "mean_token_accuracy": 0.5758497714996338,
+      "num_tokens": 684538.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.4128111600875854,
+      "epoch": 0.12439024390243902,
+      "grad_norm": 1.7912412881851196,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 1.8403,
+      "mean_token_accuracy": 0.5762076377868652,
+      "num_tokens": 698155.0,
+      "step": 51
+    },
+    {
+      "entropy": 1.4453322887420654,
+      "epoch": 0.12682926829268293,
+      "grad_norm": 1.7575210332870483,
+      "learning_rate": 6.219512195121951e-06,
+      "loss": 1.8211,
+      "mean_token_accuracy": 0.5743882060050964,
+      "num_tokens": 712024.0,
+      "step": 52
+    },
+    {
+      "entropy": 1.4498295783996582,
+      "epoch": 0.12926829268292683,
+      "grad_norm": 1.9219143390655518,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 1.8002,
+      "mean_token_accuracy": 0.57662433385849,
+      "num_tokens": 725815.0,
+      "step": 53
+    },
+    {
+      "entropy": 1.4735064506530762,
+      "epoch": 0.13170731707317074,
+      "grad_norm": 1.833960771560669,
+      "learning_rate": 6.463414634146342e-06,
+      "loss": 1.7672,
+      "mean_token_accuracy": 0.5728835463523865,
+      "num_tokens": 739675.0,
+      "step": 54
+    },
+    {
+      "entropy": 1.4403457641601562,
+      "epoch": 0.13414634146341464,
+      "grad_norm": 1.7752681970596313,
+      "learning_rate": 6.585365853658538e-06,
+      "loss": 1.7276,
+      "mean_token_accuracy": 0.5748123526573181,
+      "num_tokens": 753412.0,
+      "step": 55
+    },
+    {
+      "entropy": 1.4736558198928833,
+      "epoch": 0.13658536585365855,
+      "grad_norm": 1.6332184076309204,
+      "learning_rate": 6.707317073170733e-06,
+      "loss": 1.7191,
+      "mean_token_accuracy": 0.5755113363265991,
+      "num_tokens": 767704.0,
+      "step": 56
+    },
+    {
+      "entropy": 1.4722124338150024,
+      "epoch": 0.13902439024390245,
+      "grad_norm": 1.4422837495803833,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 1.6896,
+      "mean_token_accuracy": 0.5844070315361023,
+      "num_tokens": 781380.0,
+      "step": 57
+    },
+    {
+      "entropy": 1.471705675125122,
+      "epoch": 0.14146341463414633,
+      "grad_norm": 1.4130431413650513,
+      "learning_rate": 6.951219512195122e-06,
+      "loss": 1.6775,
+      "mean_token_accuracy": 0.579021692276001,
+      "num_tokens": 794848.0,
+      "step": 58
+    },
+    {
+      "entropy": 1.4454843997955322,
+      "epoch": 0.14390243902439023,
+      "grad_norm": 1.0664691925048828,
+      "learning_rate": 7.0731707317073175e-06,
+      "loss": 1.6227,
+      "mean_token_accuracy": 0.5937726497650146,
+      "num_tokens": 808674.0,
+      "step": 59
+    },
+    {
+      "entropy": 1.4620258808135986,
+      "epoch": 0.14634146341463414,
+      "grad_norm": 0.9696555137634277,
+      "learning_rate": 7.1951219512195125e-06,
+      "loss": 1.6272,
+      "mean_token_accuracy": 0.5890450477600098,
+      "num_tokens": 822492.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.4692648649215698,
+      "epoch": 0.14878048780487804,
+      "grad_norm": 0.9255425930023193,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 1.6179,
+      "mean_token_accuracy": 0.5979862809181213,
+      "num_tokens": 836214.0,
+      "step": 61
+    },
+    {
+      "entropy": 1.447242259979248,
+      "epoch": 0.15121951219512195,
+      "grad_norm": 1.013901710510254,
+      "learning_rate": 7.439024390243903e-06,
+      "loss": 1.6052,
+      "mean_token_accuracy": 0.5923852324485779,
+      "num_tokens": 849625.0,
+      "step": 62
+    },
+    {
+      "entropy": 1.4720077514648438,
+      "epoch": 0.15365853658536585,
+      "grad_norm": 0.9797464609146118,
+      "learning_rate": 7.560975609756098e-06,
+      "loss": 1.6078,
+      "mean_token_accuracy": 0.5976994633674622,
+      "num_tokens": 863290.0,
+      "step": 63
+    },
+    {
+      "entropy": 1.4570248126983643,
+      "epoch": 0.15609756097560976,
+      "grad_norm": 0.8936851024627686,
+      "learning_rate": 7.682926829268293e-06,
+      "loss": 1.582,
+      "mean_token_accuracy": 0.5977935791015625,
+      "num_tokens": 877356.0,
+      "step": 64
+    },
+    {
+      "entropy": 1.4674687385559082,
+      "epoch": 0.15853658536585366,
+      "grad_norm": 0.8669166564941406,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 1.5665,
+      "mean_token_accuracy": 0.5966126322746277,
+      "num_tokens": 891247.0,
+      "step": 65
+    },
+    {
+      "entropy": 1.4274413585662842,
+      "epoch": 0.16097560975609757,
+      "grad_norm": 0.7330517172813416,
+      "learning_rate": 7.926829268292685e-06,
+      "loss": 1.5316,
+      "mean_token_accuracy": 0.6071735620498657,
+      "num_tokens": 905231.0,
+      "step": 66
+    },
+    {
+      "entropy": 1.4401626586914062,
+      "epoch": 0.16341463414634147,
+      "grad_norm": 0.690156102180481,
+      "learning_rate": 8.048780487804879e-06,
+      "loss": 1.531,
+      "mean_token_accuracy": 0.6064469218254089,
+      "num_tokens": 919052.0,
+      "step": 67
+    },
+    {
+      "entropy": 1.4117504358291626,
+      "epoch": 0.16585365853658537,
+      "grad_norm": 0.6785106062889099,
+      "learning_rate": 8.170731707317073e-06,
+      "loss": 1.5115,
+      "mean_token_accuracy": 0.611976146697998,
+      "num_tokens": 932812.0,
+      "step": 68
+    },
+    {
+      "entropy": 1.4311751127243042,
+      "epoch": 0.16829268292682928,
+      "grad_norm": 0.6311522722244263,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 1.5089,
+      "mean_token_accuracy": 0.6128394603729248,
+      "num_tokens": 946489.0,
+      "step": 69
+    },
+    {
+      "entropy": 1.4168494939804077,
+      "epoch": 0.17073170731707318,
+      "grad_norm": 0.6667038798332214,
+      "learning_rate": 8.414634146341464e-06,
+      "loss": 1.5077,
+      "mean_token_accuracy": 0.6107392311096191,
+      "num_tokens": 960100.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.3982901573181152,
+      "epoch": 0.17317073170731706,
+      "grad_norm": 0.6248977184295654,
+      "learning_rate": 8.536585365853658e-06,
+      "loss": 1.4698,
+      "mean_token_accuracy": 0.6139276623725891,
+      "num_tokens": 974160.0,
+      "step": 71
+    },
+    {
+      "entropy": 1.413541555404663,
+      "epoch": 0.17560975609756097,
+      "grad_norm": 0.6343597173690796,
+      "learning_rate": 8.658536585365854e-06,
+      "loss": 1.4917,
+      "mean_token_accuracy": 0.6131435036659241,
+      "num_tokens": 987734.0,
+      "step": 72
+    },
+    {
+      "entropy": 1.4195303916931152,
+      "epoch": 0.17804878048780487,
+      "grad_norm": 0.6111395359039307,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 1.4663,
+      "mean_token_accuracy": 0.6168110370635986,
+      "num_tokens": 1001717.0,
+      "step": 73
+    },
+    {
+      "entropy": 1.40428626537323,
+      "epoch": 0.18048780487804877,
+      "grad_norm": 0.6199101805686951,
+      "learning_rate": 8.902439024390244e-06,
+      "loss": 1.4651,
+      "mean_token_accuracy": 0.6148563623428345,
+      "num_tokens": 1015276.0,
+      "step": 74
+    },
+    {
+      "entropy": 1.3776400089263916,
+      "epoch": 0.18292682926829268,
+      "grad_norm": 0.5942406058311462,
+      "learning_rate": 9.02439024390244e-06,
+      "loss": 1.4377,
+      "mean_token_accuracy": 0.6237963438034058,
+      "num_tokens": 1029000.0,
+      "step": 75
+    },
+    {
+      "entropy": 1.3991613388061523,
+      "epoch": 0.18536585365853658,
+      "grad_norm": 0.5812368392944336,
+      "learning_rate": 9.146341463414635e-06,
+      "loss": 1.4434,
+      "mean_token_accuracy": 0.6211662292480469,
+      "num_tokens": 1042547.0,
+      "step": 76
+    },
+    {
+      "entropy": 1.3818302154541016,
+      "epoch": 0.1878048780487805,
+      "grad_norm": 0.5597359538078308,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 1.427,
+      "mean_token_accuracy": 0.6249622702598572,
+      "num_tokens": 1055823.0,
+      "step": 77
+    },
+    {
+      "entropy": 1.3505898714065552,
+      "epoch": 0.1902439024390244,
+      "grad_norm": 0.5411974787712097,
+      "learning_rate": 9.390243902439025e-06,
+      "loss": 1.382,
+      "mean_token_accuracy": 0.6296323537826538,
+      "num_tokens": 1069412.0,
+      "step": 78
+    },
+    {
+      "entropy": 1.3784689903259277,
+      "epoch": 0.1926829268292683,
+      "grad_norm": 0.5365982055664062,
+      "learning_rate": 9.51219512195122e-06,
+      "loss": 1.4017,
+      "mean_token_accuracy": 0.6252776384353638,
+      "num_tokens": 1082934.0,
+      "step": 79
+    },
+    {
+      "entropy": 1.3679686784744263,
+      "epoch": 0.1951219512195122,
+      "grad_norm": 0.5300019383430481,
+      "learning_rate": 9.634146341463415e-06,
+      "loss": 1.402,
+      "mean_token_accuracy": 0.6283849477767944,
+      "num_tokens": 1096281.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.3440438508987427,
+      "epoch": 0.1975609756097561,
+      "grad_norm": 0.5176823139190674,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 1.3634,
+      "mean_token_accuracy": 0.6327404975891113,
+      "num_tokens": 1110298.0,
+      "step": 81
+    },
+    {
+      "entropy": 1.3412446975708008,
+      "epoch": 0.2,
+      "grad_norm": 0.5266833305358887,
+      "learning_rate": 9.878048780487805e-06,
+      "loss": 1.3741,
+      "mean_token_accuracy": 0.6278718113899231,
+      "num_tokens": 1124199.0,
+      "step": 82
+    },
+    {
+      "entropy": 1.3485190868377686,
+      "epoch": 0.20243902439024392,
+      "grad_norm": 0.5412361025810242,
+      "learning_rate": 1e-05,
+      "loss": 1.3719,
+      "mean_token_accuracy": 0.6318795680999756,
+      "num_tokens": 1137697.0,
+      "step": 83
+    },
+    {
+      "entropy": 1.3585059642791748,
+      "epoch": 0.2048780487804878,
+      "grad_norm": 0.5267719626426697,
+      "learning_rate": 9.986449864498647e-06,
+      "loss": 1.3713,
+      "mean_token_accuracy": 0.6309584379196167,
+      "num_tokens": 1151454.0,
+      "step": 84
+    },
+    {
+      "entropy": 1.275984287261963,
+      "epoch": 0.2073170731707317,
+      "grad_norm": 0.5076718330383301,
+      "learning_rate": 9.972899728997291e-06,
+      "loss": 1.3172,
+      "mean_token_accuracy": 0.639977216720581,
+      "num_tokens": 1165533.0,
+      "step": 85
+    },
+    {
+      "entropy": 1.308915376663208,
+      "epoch": 0.2097560975609756,
+      "grad_norm": 0.5083034634590149,
+      "learning_rate": 9.959349593495936e-06,
+      "loss": 1.3337,
+      "mean_token_accuracy": 0.6340206265449524,
+      "num_tokens": 1179129.0,
+      "step": 86
+    },
+    {
+      "entropy": 1.3241677284240723,
+      "epoch": 0.2121951219512195,
+      "grad_norm": 0.4542701840400696,
+      "learning_rate": 9.94579945799458e-06,
+      "loss": 1.3551,
+      "mean_token_accuracy": 0.6313815116882324,
+      "num_tokens": 1192739.0,
+      "step": 87
+    },
+    {
+      "entropy": 1.314582347869873,
+      "epoch": 0.2146341463414634,
+      "grad_norm": 0.48066484928131104,
+      "learning_rate": 9.932249322493226e-06,
+      "loss": 1.337,
+      "mean_token_accuracy": 0.6390174627304077,
+      "num_tokens": 1206149.0,
+      "step": 88
+    },
+    {
+      "entropy": 1.3120391368865967,
+      "epoch": 0.21707317073170732,
+      "grad_norm": 0.4342902600765228,
+      "learning_rate": 9.91869918699187e-06,
+      "loss": 1.3378,
+      "mean_token_accuracy": 0.636769711971283,
+      "num_tokens": 1220492.0,
+      "step": 89
+    },
+    {
+      "entropy": 1.270530104637146,
+      "epoch": 0.21951219512195122,
+      "grad_norm": 0.43451136350631714,
+      "learning_rate": 9.905149051490516e-06,
+      "loss": 1.3036,
+      "mean_token_accuracy": 0.6418837904930115,
+      "num_tokens": 1234671.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2754387855529785,
+      "epoch": 0.22195121951219512,
+      "grad_norm": 0.4188333749771118,
+      "learning_rate": 9.89159891598916e-06,
+      "loss": 1.3121,
+      "mean_token_accuracy": 0.641899585723877,
+      "num_tokens": 1248711.0,
+      "step": 91
+    },
+    {
+      "entropy": 1.2386054992675781,
+      "epoch": 0.22439024390243903,
+      "grad_norm": 0.4193291962146759,
+      "learning_rate": 9.878048780487805e-06,
+      "loss": 1.2657,
+      "mean_token_accuracy": 0.6525153517723083,
+      "num_tokens": 1262224.0,
+      "step": 92
+    },
+    {
+      "entropy": 1.2513689994812012,
+      "epoch": 0.22682926829268293,
+      "grad_norm": 0.428231805562973,
+      "learning_rate": 9.864498644986451e-06,
+      "loss": 1.2844,
+      "mean_token_accuracy": 0.6457121968269348,
+      "num_tokens": 1275825.0,
+      "step": 93
+    },
+    {
+      "entropy": 1.266829252243042,
+      "epoch": 0.22926829268292684,
+      "grad_norm": 0.42669400572776794,
+      "learning_rate": 9.850948509485095e-06,
+      "loss": 1.2964,
+      "mean_token_accuracy": 0.6454920768737793,
+      "num_tokens": 1289395.0,
+      "step": 94
+    },
+    {
+      "entropy": 1.2397089004516602,
+      "epoch": 0.23170731707317074,
+      "grad_norm": 0.409270703792572,
+      "learning_rate": 9.837398373983741e-06,
+      "loss": 1.2499,
+      "mean_token_accuracy": 0.653494119644165,
+      "num_tokens": 1303177.0,
+      "step": 95
+    },
+    {
+      "entropy": 1.2290325164794922,
+      "epoch": 0.23414634146341465,
+      "grad_norm": 0.4218335747718811,
+      "learning_rate": 9.823848238482386e-06,
+      "loss": 1.2648,
+      "mean_token_accuracy": 0.6454984545707703,
+      "num_tokens": 1317066.0,
+      "step": 96
+    },
+    {
+      "entropy": 1.2639315128326416,
+      "epoch": 0.23658536585365852,
+      "grad_norm": 0.41416656970977783,
+      "learning_rate": 9.81029810298103e-06,
+      "loss": 1.2838,
+      "mean_token_accuracy": 0.6459396481513977,
+      "num_tokens": 1330800.0,
+      "step": 97
+    },
+    {
+      "entropy": 1.244269847869873,
+      "epoch": 0.23902439024390243,
+      "grad_norm": 0.40852072834968567,
+      "learning_rate": 9.796747967479675e-06,
+      "loss": 1.2441,
+      "mean_token_accuracy": 0.6508611440658569,
+      "num_tokens": 1344693.0,
+      "step": 98
+    },
+    {
+      "entropy": 1.2480473518371582,
+      "epoch": 0.24146341463414633,
+      "grad_norm": 0.41581472754478455,
+      "learning_rate": 9.78319783197832e-06,
+      "loss": 1.2642,
+      "mean_token_accuracy": 0.6478216648101807,
+      "num_tokens": 1358435.0,
+      "step": 99
+    },
+    {
+      "entropy": 1.2283647060394287,
+      "epoch": 0.24390243902439024,
+      "grad_norm": 0.40649130940437317,
+      "learning_rate": 9.769647696476967e-06,
+      "loss": 1.2457,
+      "mean_token_accuracy": 0.6516292691230774,
+      "num_tokens": 1371862.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2162929773330688,
+      "epoch": 0.24634146341463414,
+      "grad_norm": 0.40584856271743774,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 1.2369,
+      "mean_token_accuracy": 0.6583905816078186,
+      "num_tokens": 1385411.0,
+      "step": 101
+    },
+    {
+      "entropy": 1.208768606185913,
+      "epoch": 0.24878048780487805,
+      "grad_norm": 0.39489707350730896,
+      "learning_rate": 9.742547425474255e-06,
+      "loss": 1.2277,
+      "mean_token_accuracy": 0.6547741889953613,
+      "num_tokens": 1399021.0,
+      "step": 102
+    },
+    {
+      "entropy": 1.2198774814605713,
+      "epoch": 0.25121951219512195,
+      "grad_norm": 0.383748322725296,
+      "learning_rate": 9.7289972899729e-06,
+      "loss": 1.2335,
+      "mean_token_accuracy": 0.6584837436676025,
+      "num_tokens": 1412887.0,
+      "step": 103
+    },
+    {
+      "entropy": 1.221745252609253,
+      "epoch": 0.25365853658536586,
+      "grad_norm": 0.39412277936935425,
+      "learning_rate": 9.715447154471546e-06,
+      "loss": 1.2403,
+      "mean_token_accuracy": 0.6498458981513977,
+      "num_tokens": 1426854.0,
+      "step": 104
+    },
+    {
+      "entropy": 1.217738151550293,
+      "epoch": 0.25609756097560976,
+      "grad_norm": 0.3701169192790985,
+      "learning_rate": 9.70189701897019e-06,
+      "loss": 1.2242,
+      "mean_token_accuracy": 0.6566095948219299,
+      "num_tokens": 1440426.0,
+      "step": 105
+    },
+    {
+      "entropy": 1.1941392421722412,
+      "epoch": 0.25853658536585367,
+      "grad_norm": 0.371112585067749,
+      "learning_rate": 9.688346883468836e-06,
+      "loss": 1.2115,
+      "mean_token_accuracy": 0.6609773635864258,
+      "num_tokens": 1454152.0,
+      "step": 106
+    },
+    {
+      "entropy": 1.1990996599197388,
+      "epoch": 0.26097560975609757,
+      "grad_norm": 0.365815669298172,
+      "learning_rate": 9.67479674796748e-06,
+      "loss": 1.2128,
+      "mean_token_accuracy": 0.6610169410705566,
+      "num_tokens": 1467856.0,
+      "step": 107
+    },
+    {
+      "entropy": 1.1728260517120361,
+      "epoch": 0.2634146341463415,
+      "grad_norm": 0.3645389974117279,
+      "learning_rate": 9.661246612466125e-06,
+      "loss": 1.1996,
+      "mean_token_accuracy": 0.6662229895591736,
+      "num_tokens": 1481396.0,
+      "step": 108
+    },
+    {
+      "entropy": 1.1833685636520386,
+      "epoch": 0.2658536585365854,
+      "grad_norm": 0.36212098598480225,
+      "learning_rate": 9.64769647696477e-06,
+      "loss": 1.202,
+      "mean_token_accuracy": 0.6605650186538696,
+      "num_tokens": 1495288.0,
+      "step": 109
+    },
+    {
+      "entropy": 1.1809427738189697,
+      "epoch": 0.2682926829268293,
+      "grad_norm": 0.4562148153781891,
+      "learning_rate": 9.634146341463415e-06,
+      "loss": 1.1918,
+      "mean_token_accuracy": 0.6666184663772583,
+      "num_tokens": 1509141.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.1492688655853271,
+      "epoch": 0.2707317073170732,
+      "grad_norm": 0.3741201162338257,
+      "learning_rate": 9.620596205962061e-06,
+      "loss": 1.1765,
+      "mean_token_accuracy": 0.671988844871521,
+      "num_tokens": 1522748.0,
+      "step": 111
+    },
+    {
+      "entropy": 1.1608304977416992,
+      "epoch": 0.2731707317073171,
+      "grad_norm": 0.5636874437332153,
+      "learning_rate": 9.607046070460706e-06,
+      "loss": 1.1628,
+      "mean_token_accuracy": 0.6660817265510559,
+      "num_tokens": 1536441.0,
+      "step": 112
+    },
+    {
+      "entropy": 1.2037889957427979,
+      "epoch": 0.275609756097561,
+      "grad_norm": 0.3735000491142273,
+      "learning_rate": 9.59349593495935e-06,
+      "loss": 1.2275,
+      "mean_token_accuracy": 0.6517032384872437,
+      "num_tokens": 1550313.0,
+      "step": 113
+    },
+    {
+      "entropy": 1.1596955060958862,
+      "epoch": 0.2780487804878049,
+      "grad_norm": 0.37646710872650146,
+      "learning_rate": 9.579945799457996e-06,
+      "loss": 1.1889,
+      "mean_token_accuracy": 0.6636276841163635,
+      "num_tokens": 1563930.0,
+      "step": 114
+    },
+    {
+      "entropy": 1.1520904302597046,
+      "epoch": 0.2804878048780488,
+      "grad_norm": 0.3705466687679291,
+      "learning_rate": 9.56639566395664e-06,
+      "loss": 1.1592,
+      "mean_token_accuracy": 0.671284556388855,
+      "num_tokens": 1576939.0,
+      "step": 115
+    },
+    {
+      "entropy": 1.1546409130096436,
+      "epoch": 0.28292682926829266,
+      "grad_norm": 0.3923504948616028,
+      "learning_rate": 9.552845528455286e-06,
+      "loss": 1.1643,
+      "mean_token_accuracy": 0.6669601202011108,
+      "num_tokens": 1590587.0,
+      "step": 116
+    },
+    {
+      "entropy": 1.1506261825561523,
+      "epoch": 0.28536585365853656,
+      "grad_norm": 0.3831369876861572,
+      "learning_rate": 9.53929539295393e-06,
+      "loss": 1.1585,
+      "mean_token_accuracy": 0.670432984828949,
+      "num_tokens": 1603790.0,
+      "step": 117
+    },
+    {
+      "entropy": 1.1624176502227783,
+      "epoch": 0.28780487804878047,
+      "grad_norm": 0.3876406252384186,
+      "learning_rate": 9.525745257452575e-06,
+      "loss": 1.1547,
+      "mean_token_accuracy": 0.6688105463981628,
+      "num_tokens": 1617333.0,
+      "step": 118
+    },
+    {
+      "entropy": 1.1381282806396484,
+      "epoch": 0.29024390243902437,
+      "grad_norm": 0.3799033761024475,
+      "learning_rate": 9.51219512195122e-06,
+      "loss": 1.1325,
+      "mean_token_accuracy": 0.6745951175689697,
+      "num_tokens": 1631181.0,
+      "step": 119
+    },
+    {
+      "entropy": 1.1472123861312866,
+      "epoch": 0.2926829268292683,
+      "grad_norm": 0.40002092719078064,
+      "learning_rate": 9.498644986449865e-06,
+      "loss": 1.131,
+      "mean_token_accuracy": 0.6772284507751465,
+      "num_tokens": 1644547.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1485984325408936,
+      "epoch": 0.2951219512195122,
+      "grad_norm": 0.41155263781547546,
+      "learning_rate": 9.485094850948512e-06,
+      "loss": 1.1314,
+      "mean_token_accuracy": 0.6737343072891235,
+      "num_tokens": 1658429.0,
+      "step": 121
+    },
+    {
+      "entropy": 1.1536389589309692,
+      "epoch": 0.2975609756097561,
+      "grad_norm": 0.39085325598716736,
+      "learning_rate": 9.471544715447156e-06,
+      "loss": 1.1478,
+      "mean_token_accuracy": 0.6709743738174438,
+      "num_tokens": 1672064.0,
+      "step": 122
+    },
+    {
+      "entropy": 1.1491937637329102,
+      "epoch": 0.3,
+      "grad_norm": 0.33603760600090027,
+      "learning_rate": 9.4579945799458e-06,
+      "loss": 1.1478,
+      "mean_token_accuracy": 0.6709640622138977,
+      "num_tokens": 1686197.0,
+      "step": 123
+    },
+    {
+      "entropy": 1.1482875347137451,
+      "epoch": 0.3024390243902439,
+      "grad_norm": 0.32015225291252136,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 1.138,
+      "mean_token_accuracy": 0.6719373464584351,
+      "num_tokens": 1700253.0,
+      "step": 124
+    },
+    {
+      "entropy": 1.153489589691162,
+      "epoch": 0.3048780487804878,
+      "grad_norm": 0.32207733392715454,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 1.1585,
+      "mean_token_accuracy": 0.6705389022827148,
+      "num_tokens": 1713870.0,
+      "step": 125
+    },
+    {
+      "entropy": 1.107184648513794,
+      "epoch": 0.3073170731707317,
+      "grad_norm": 0.3159145712852478,
+      "learning_rate": 9.417344173441735e-06,
+      "loss": 1.1136,
+      "mean_token_accuracy": 0.6776368618011475,
+      "num_tokens": 1727529.0,
+      "step": 126
+    },
+    {
+      "entropy": 1.1219724416732788,
+      "epoch": 0.3097560975609756,
+      "grad_norm": 0.2986023426055908,
+      "learning_rate": 9.403794037940381e-06,
+      "loss": 1.1253,
+      "mean_token_accuracy": 0.673386812210083,
+      "num_tokens": 1742128.0,
+      "step": 127
+    },
+    {
+      "entropy": 1.1211440563201904,
+      "epoch": 0.3121951219512195,
+      "grad_norm": 0.33409249782562256,
+      "learning_rate": 9.390243902439025e-06,
+      "loss": 1.1226,
+      "mean_token_accuracy": 0.6810810565948486,
+      "num_tokens": 1755834.0,
+      "step": 128
+    },
+    {
+      "entropy": 1.1153686046600342,
+      "epoch": 0.3146341463414634,
+      "grad_norm": 0.29576805233955383,
+      "learning_rate": 9.37669376693767e-06,
+      "loss": 1.112,
+      "mean_token_accuracy": 0.6776238083839417,
+      "num_tokens": 1769418.0,
+      "step": 129
+    },
+    {
+      "entropy": 1.138454794883728,
+      "epoch": 0.3170731707317073,
+      "grad_norm": 0.28794384002685547,
+      "learning_rate": 9.363143631436316e-06,
+      "loss": 1.1295,
+      "mean_token_accuracy": 0.6768421530723572,
+      "num_tokens": 1783127.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.0799577236175537,
+      "epoch": 0.3195121951219512,
+      "grad_norm": 0.3011602759361267,
+      "learning_rate": 9.34959349593496e-06,
+      "loss": 1.0794,
+      "mean_token_accuracy": 0.6856415867805481,
+      "num_tokens": 1796898.0,
+      "step": 131
+    },
+    {
+      "entropy": 1.1080092191696167,
+      "epoch": 0.32195121951219513,
+      "grad_norm": 0.28195154666900635,
+      "learning_rate": 9.336043360433606e-06,
+      "loss": 1.1034,
+      "mean_token_accuracy": 0.6803351044654846,
+      "num_tokens": 1810522.0,
+      "step": 132
+    },
+    {
+      "entropy": 1.0953476428985596,
+      "epoch": 0.32439024390243903,
+      "grad_norm": 0.2998756468296051,
+      "learning_rate": 9.32249322493225e-06,
+      "loss": 1.0979,
+      "mean_token_accuracy": 0.6815518140792847,
+      "num_tokens": 1824019.0,
+      "step": 133
+    },
+    {
+      "entropy": 1.1013312339782715,
+      "epoch": 0.32682926829268294,
+      "grad_norm": 0.3195270001888275,
+      "learning_rate": 9.308943089430895e-06,
+      "loss": 1.0963,
+      "mean_token_accuracy": 0.6816878914833069,
+      "num_tokens": 1837638.0,
+      "step": 134
+    },
+    {
+      "entropy": 1.097846508026123,
+      "epoch": 0.32926829268292684,
+      "grad_norm": 0.31614819169044495,
+      "learning_rate": 9.29539295392954e-06,
+      "loss": 1.0846,
+      "mean_token_accuracy": 0.6832810044288635,
+      "num_tokens": 1851357.0,
+      "step": 135
+    },
+    {
+      "entropy": 1.1029002666473389,
+      "epoch": 0.33170731707317075,
+      "grad_norm": 0.3023568391799927,
+      "learning_rate": 9.281842818428185e-06,
+      "loss": 1.0926,
+      "mean_token_accuracy": 0.68416428565979,
+      "num_tokens": 1865032.0,
+      "step": 136
+    },
+    {
+      "entropy": 1.100654125213623,
+      "epoch": 0.33414634146341465,
+      "grad_norm": 0.30091455578804016,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 1.1102,
+      "mean_token_accuracy": 0.6823589205741882,
+      "num_tokens": 1878834.0,
+      "step": 137
+    },
+    {
+      "entropy": 1.0853767395019531,
+      "epoch": 0.33658536585365856,
+      "grad_norm": 0.28139373660087585,
+      "learning_rate": 9.254742547425476e-06,
+      "loss": 1.0822,
+      "mean_token_accuracy": 0.6865951418876648,
+      "num_tokens": 1892733.0,
+      "step": 138
+    },
+    {
+      "entropy": 1.1051311492919922,
+      "epoch": 0.33902439024390246,
+      "grad_norm": 0.2940380275249481,
+      "learning_rate": 9.24119241192412e-06,
+      "loss": 1.1011,
+      "mean_token_accuracy": 0.6799636483192444,
+      "num_tokens": 1907063.0,
+      "step": 139
+    },
+    {
+      "entropy": 1.1128774881362915,
+      "epoch": 0.34146341463414637,
+      "grad_norm": 0.2861708104610443,
+      "learning_rate": 9.227642276422764e-06,
+      "loss": 1.1021,
+      "mean_token_accuracy": 0.6820879578590393,
+      "num_tokens": 1921064.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1019363403320312,
+      "epoch": 0.3439024390243902,
+      "grad_norm": 0.2888548970222473,
+      "learning_rate": 9.21409214092141e-06,
+      "loss": 1.1021,
+      "mean_token_accuracy": 0.6837447881698608,
+      "num_tokens": 1934528.0,
+      "step": 141
+    },
+    {
+      "entropy": 1.0777734518051147,
+      "epoch": 0.3463414634146341,
+      "grad_norm": 0.3018644452095032,
+      "learning_rate": 9.200542005420055e-06,
+      "loss": 1.082,
+      "mean_token_accuracy": 0.6913608312606812,
+      "num_tokens": 1947867.0,
+      "step": 142
+    },
+    {
+      "entropy": 1.0810433626174927,
+      "epoch": 0.348780487804878,
+      "grad_norm": 0.29657524824142456,
+      "learning_rate": 9.1869918699187e-06,
+      "loss": 1.0869,
+      "mean_token_accuracy": 0.6829285621643066,
+      "num_tokens": 1962088.0,
+      "step": 143
+    },
+    {
+      "entropy": 1.0838487148284912,
+      "epoch": 0.35121951219512193,
+      "grad_norm": 0.30069437623023987,
+      "learning_rate": 9.173441734417345e-06,
+      "loss": 1.0751,
+      "mean_token_accuracy": 0.6872337460517883,
+      "num_tokens": 1975718.0,
+      "step": 144
+    },
+    {
+      "entropy": 1.0671093463897705,
+      "epoch": 0.35365853658536583,
+      "grad_norm": 0.32158926129341125,
+      "learning_rate": 9.15989159891599e-06,
+      "loss": 1.0671,
+      "mean_token_accuracy": 0.6920549273490906,
+      "num_tokens": 1989428.0,
+      "step": 145
+    },
+    {
+      "entropy": 1.0684272050857544,
+      "epoch": 0.35609756097560974,
+      "grad_norm": 0.3996563255786896,
+      "learning_rate": 9.146341463414635e-06,
+      "loss": 1.071,
+      "mean_token_accuracy": 0.6908498406410217,
+      "num_tokens": 2002952.0,
+      "step": 146
+    },
+    {
+      "entropy": 1.0824508666992188,
+      "epoch": 0.35853658536585364,
+      "grad_norm": 0.2806661128997803,
+      "learning_rate": 9.13279132791328e-06,
+      "loss": 1.0989,
+      "mean_token_accuracy": 0.6818480491638184,
+      "num_tokens": 2016669.0,
+      "step": 147
+    },
+    {
+      "entropy": 1.077519178390503,
+      "epoch": 0.36097560975609755,
+      "grad_norm": 0.29812970757484436,
+      "learning_rate": 9.119241192411926e-06,
+      "loss": 1.0606,
+      "mean_token_accuracy": 0.6873586773872375,
+      "num_tokens": 2030394.0,
+      "step": 148
+    },
+    {
+      "entropy": 1.0674171447753906,
+      "epoch": 0.36341463414634145,
+      "grad_norm": 0.3090595006942749,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 1.0757,
+      "mean_token_accuracy": 0.6799144744873047,
+      "num_tokens": 2043972.0,
+      "step": 149
+    },
+    {
+      "entropy": 1.088179349899292,
+      "epoch": 0.36585365853658536,
+      "grad_norm": 0.3295019268989563,
+      "learning_rate": 9.092140921409215e-06,
+      "loss": 1.0892,
+      "mean_token_accuracy": 0.6850416660308838,
+      "num_tokens": 2057793.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.073317050933838,
+      "epoch": 0.36829268292682926,
+      "grad_norm": 0.31593307852745056,
+      "learning_rate": 9.07859078590786e-06,
+      "loss": 1.0712,
+      "mean_token_accuracy": 0.6870791912078857,
+      "num_tokens": 2071771.0,
+      "step": 151
+    },
+    {
+      "entropy": 1.070550799369812,
+      "epoch": 0.37073170731707317,
+      "grad_norm": 0.28268736600875854,
+      "learning_rate": 9.065040650406505e-06,
+      "loss": 1.0527,
+      "mean_token_accuracy": 0.6912903785705566,
+      "num_tokens": 2085473.0,
+      "step": 152
+    },
+    {
+      "entropy": 1.0618209838867188,
+      "epoch": 0.37317073170731707,
+      "grad_norm": 0.2891680598258972,
+      "learning_rate": 9.051490514905151e-06,
+      "loss": 1.0464,
+      "mean_token_accuracy": 0.6930386424064636,
+      "num_tokens": 2099380.0,
+      "step": 153
+    },
+    {
+      "entropy": 1.0524916648864746,
+      "epoch": 0.375609756097561,
+      "grad_norm": 0.31244540214538574,
+      "learning_rate": 9.037940379403795e-06,
+      "loss": 1.0557,
+      "mean_token_accuracy": 0.6923752427101135,
+      "num_tokens": 2113062.0,
+      "step": 154
+    },
+    {
+      "entropy": 1.0701351165771484,
+      "epoch": 0.3780487804878049,
+      "grad_norm": 0.2857276201248169,
+      "learning_rate": 9.02439024390244e-06,
+      "loss": 1.0646,
+      "mean_token_accuracy": 0.6889170408248901,
+      "num_tokens": 2126486.0,
+      "step": 155
+    },
+    {
+      "entropy": 1.0808346271514893,
+      "epoch": 0.3804878048780488,
+      "grad_norm": 0.30712175369262695,
+      "learning_rate": 9.010840108401084e-06,
+      "loss": 1.0764,
+      "mean_token_accuracy": 0.6845914721488953,
+      "num_tokens": 2140319.0,
+      "step": 156
+    },
+    {
+      "entropy": 1.0551326274871826,
+      "epoch": 0.3829268292682927,
+      "grad_norm": 0.29289886355400085,
+      "learning_rate": 8.99728997289973e-06,
+      "loss": 1.0613,
+      "mean_token_accuracy": 0.6889886260032654,
+      "num_tokens": 2154148.0,
+      "step": 157
+    },
+    {
+      "entropy": 1.044635534286499,
+      "epoch": 0.3853658536585366,
+      "grad_norm": 0.2924291491508484,
+      "learning_rate": 8.983739837398374e-06,
+      "loss": 1.0475,
+      "mean_token_accuracy": 0.6969406604766846,
+      "num_tokens": 2167729.0,
+      "step": 158
+    },
+    {
+      "entropy": 1.0424062013626099,
+      "epoch": 0.3878048780487805,
+      "grad_norm": 0.33836257457733154,
+      "learning_rate": 8.970189701897019e-06,
+      "loss": 1.0477,
+      "mean_token_accuracy": 0.6928502917289734,
+      "num_tokens": 2181354.0,
+      "step": 159
+    },
+    {
+      "entropy": 1.0367578268051147,
+      "epoch": 0.3902439024390244,
+      "grad_norm": 0.29433733224868774,
+      "learning_rate": 8.956639566395665e-06,
+      "loss": 1.0343,
+      "mean_token_accuracy": 0.6925205588340759,
+      "num_tokens": 2195101.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.0720441341400146,
+      "epoch": 0.3926829268292683,
+      "grad_norm": 0.33182695508003235,
+      "learning_rate": 8.94308943089431e-06,
+      "loss": 1.0601,
+      "mean_token_accuracy": 0.6875677108764648,
+      "num_tokens": 2208960.0,
+      "step": 161
+    },
+    {
+      "entropy": 1.0333222150802612,
+      "epoch": 0.3951219512195122,
+      "grad_norm": 0.32103872299194336,
+      "learning_rate": 8.929539295392955e-06,
+      "loss": 1.0316,
+      "mean_token_accuracy": 0.6961764693260193,
+      "num_tokens": 2223073.0,
+      "step": 162
+    },
+    {
+      "entropy": 1.0387994050979614,
+      "epoch": 0.3975609756097561,
+      "grad_norm": 0.29320165514945984,
+      "learning_rate": 8.9159891598916e-06,
+      "loss": 1.045,
+      "mean_token_accuracy": 0.6899910569190979,
+      "num_tokens": 2236537.0,
+      "step": 163
+    },
+    {
+      "entropy": 1.04783296585083,
+      "epoch": 0.4,
+      "grad_norm": 0.305354505777359,
+      "learning_rate": 8.902439024390244e-06,
+      "loss": 1.0416,
+      "mean_token_accuracy": 0.6935980319976807,
+      "num_tokens": 2250205.0,
+      "step": 164
+    },
+    {
+      "entropy": 1.0558719635009766,
+      "epoch": 0.4024390243902439,
+      "grad_norm": 0.3049624264240265,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.0579,
+      "mean_token_accuracy": 0.6879761815071106,
+      "num_tokens": 2264002.0,
+      "step": 165
+    },
+    {
+      "entropy": 1.0384637117385864,
+      "epoch": 0.40487804878048783,
+      "grad_norm": 0.30772659182548523,
+      "learning_rate": 8.875338753387534e-06,
+      "loss": 1.0355,
+      "mean_token_accuracy": 0.6911885738372803,
+      "num_tokens": 2277353.0,
+      "step": 166
+    },
+    {
+      "entropy": 1.0217058658599854,
+      "epoch": 0.4073170731707317,
+      "grad_norm": 0.29792320728302,
+      "learning_rate": 8.86178861788618e-06,
+      "loss": 1.0206,
+      "mean_token_accuracy": 0.6998187303543091,
+      "num_tokens": 2290611.0,
+      "step": 167
+    },
+    {
+      "entropy": 1.0324389934539795,
+      "epoch": 0.4097560975609756,
+      "grad_norm": 0.3367704451084137,
+      "learning_rate": 8.848238482384825e-06,
+      "loss": 1.0451,
+      "mean_token_accuracy": 0.6971227526664734,
+      "num_tokens": 2303695.0,
+      "step": 168
+    },
+    {
+      "entropy": 1.0322160720825195,
+      "epoch": 0.4121951219512195,
+      "grad_norm": 0.34955868124961853,
+      "learning_rate": 8.834688346883469e-06,
+      "loss": 1.0317,
+      "mean_token_accuracy": 0.6930278539657593,
+      "num_tokens": 2317810.0,
+      "step": 169
+    },
+    {
+      "entropy": 1.0324090719223022,
+      "epoch": 0.4146341463414634,
+      "grad_norm": 0.33210307359695435,
+      "learning_rate": 8.821138211382113e-06,
+      "loss": 1.0061,
+      "mean_token_accuracy": 0.6991522312164307,
+      "num_tokens": 2331627.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.0623321533203125,
+      "epoch": 0.4170731707317073,
+      "grad_norm": 0.3067868947982788,
+      "learning_rate": 8.80758807588076e-06,
+      "loss": 1.0559,
+      "mean_token_accuracy": 0.6889886260032654,
+      "num_tokens": 2345456.0,
+      "step": 171
+    },
+    {
+      "entropy": 1.0518372058868408,
+      "epoch": 0.4195121951219512,
+      "grad_norm": 0.3096657395362854,
+      "learning_rate": 8.794037940379404e-06,
+      "loss": 1.0274,
+      "mean_token_accuracy": 0.6929498910903931,
+      "num_tokens": 2359486.0,
+      "step": 172
+    },
+    {
+      "entropy": 1.030791997909546,
+      "epoch": 0.4219512195121951,
+      "grad_norm": 0.3270096480846405,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 1.0192,
+      "mean_token_accuracy": 0.6991227865219116,
+      "num_tokens": 2373182.0,
+      "step": 173
+    },
+    {
+      "entropy": 1.0519341230392456,
+      "epoch": 0.424390243902439,
+      "grad_norm": 0.31375235319137573,
+      "learning_rate": 8.766937669376694e-06,
+      "loss": 1.0385,
+      "mean_token_accuracy": 0.6936818957328796,
+      "num_tokens": 2386968.0,
+      "step": 174
+    },
+    {
+      "entropy": 1.0423860549926758,
+      "epoch": 0.4268292682926829,
+      "grad_norm": 0.34139835834503174,
+      "learning_rate": 8.753387533875339e-06,
+      "loss": 1.0354,
+      "mean_token_accuracy": 0.694669783115387,
+      "num_tokens": 2400792.0,
+      "step": 175
+    },
+    {
+      "entropy": 1.0326882600784302,
+      "epoch": 0.4292682926829268,
+      "grad_norm": 0.3083813190460205,
+      "learning_rate": 8.739837398373985e-06,
+      "loss": 1.0252,
+      "mean_token_accuracy": 0.6961446404457092,
+      "num_tokens": 2414581.0,
+      "step": 176
+    },
+    {
+      "entropy": 1.0297658443450928,
+      "epoch": 0.4317073170731707,
+      "grad_norm": 0.3456875681877136,
+      "learning_rate": 8.726287262872629e-06,
+      "loss": 1.0427,
+      "mean_token_accuracy": 0.6909603476524353,
+      "num_tokens": 2428071.0,
+      "step": 177
+    },
+    {
+      "entropy": 1.0350985527038574,
+      "epoch": 0.43414634146341463,
+      "grad_norm": 0.3167937397956848,
+      "learning_rate": 8.712737127371275e-06,
+      "loss": 1.0333,
+      "mean_token_accuracy": 0.6956335306167603,
+      "num_tokens": 2442080.0,
+      "step": 178
+    },
+    {
+      "entropy": 1.0534954071044922,
+      "epoch": 0.43658536585365854,
+      "grad_norm": 0.34160587191581726,
+      "learning_rate": 8.69918699186992e-06,
+      "loss": 1.0478,
+      "mean_token_accuracy": 0.6857367753982544,
+      "num_tokens": 2456062.0,
+      "step": 179
+    },
+    {
+      "entropy": 1.019012689590454,
+      "epoch": 0.43902439024390244,
+      "grad_norm": 0.3717818856239319,
+      "learning_rate": 8.685636856368564e-06,
+      "loss": 1.0128,
+      "mean_token_accuracy": 0.6978288292884827,
+      "num_tokens": 2469481.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.0156989097595215,
+      "epoch": 0.44146341463414634,
+      "grad_norm": 0.3390960395336151,
+      "learning_rate": 8.67208672086721e-06,
+      "loss": 1.001,
+      "mean_token_accuracy": 0.6981892585754395,
+      "num_tokens": 2483138.0,
+      "step": 181
+    },
+    {
+      "entropy": 1.0245683193206787,
+      "epoch": 0.44390243902439025,
+      "grad_norm": 0.3798343539237976,
+      "learning_rate": 8.658536585365854e-06,
+      "loss": 1.0161,
+      "mean_token_accuracy": 0.7003815174102783,
+      "num_tokens": 2496521.0,
+      "step": 182
+    },
+    {
+      "entropy": 1.0270142555236816,
+      "epoch": 0.44634146341463415,
+      "grad_norm": 0.5196627974510193,
+      "learning_rate": 8.6449864498645e-06,
+      "loss": 1.0319,
+      "mean_token_accuracy": 0.6966584920883179,
+      "num_tokens": 2509974.0,
+      "step": 183
+    },
+    {
+      "entropy": 1.028610348701477,
+      "epoch": 0.44878048780487806,
+      "grad_norm": 0.34398967027664185,
+      "learning_rate": 8.631436314363144e-06,
+      "loss": 1.0107,
+      "mean_token_accuracy": 0.6969505548477173,
+      "num_tokens": 2524222.0,
+      "step": 184
+    },
+    {
+      "entropy": 1.0180282592773438,
+      "epoch": 0.45121951219512196,
+      "grad_norm": 0.3196260631084442,
+      "learning_rate": 8.617886178861789e-06,
+      "loss": 1.0066,
+      "mean_token_accuracy": 0.7003494501113892,
+      "num_tokens": 2537974.0,
+      "step": 185
+    },
+    {
+      "entropy": 1.0140835046768188,
+      "epoch": 0.45365853658536587,
+      "grad_norm": 0.3342556655406952,
+      "learning_rate": 8.604336043360433e-06,
+      "loss": 1.0143,
+      "mean_token_accuracy": 0.698606014251709,
+      "num_tokens": 2551620.0,
+      "step": 186
+    },
+    {
+      "entropy": 1.0282118320465088,
+      "epoch": 0.4560975609756098,
+      "grad_norm": 0.4843715727329254,
+      "learning_rate": 8.59078590785908e-06,
+      "loss": 1.0166,
+      "mean_token_accuracy": 0.6956273317337036,
+      "num_tokens": 2565632.0,
+      "step": 187
+    },
+    {
+      "entropy": 1.0135064125061035,
+      "epoch": 0.4585365853658537,
+      "grad_norm": 0.3793366849422455,
+      "learning_rate": 8.577235772357724e-06,
+      "loss": 1.0064,
+      "mean_token_accuracy": 0.6974882483482361,
+      "num_tokens": 2579264.0,
+      "step": 188
+    },
+    {
+      "entropy": 1.028588056564331,
+      "epoch": 0.4609756097560976,
+      "grad_norm": 0.40557751059532166,
+      "learning_rate": 8.56368563685637e-06,
+      "loss": 1.0335,
+      "mean_token_accuracy": 0.6939067840576172,
+      "num_tokens": 2592557.0,
+      "step": 189
+    },
+    {
+      "entropy": 1.0189435482025146,
+      "epoch": 0.4634146341463415,
+      "grad_norm": 0.3505721688270569,
+      "learning_rate": 8.550135501355014e-06,
+      "loss": 0.9996,
+      "mean_token_accuracy": 0.7037708759307861,
+      "num_tokens": 2605806.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.0245192050933838,
+      "epoch": 0.4658536585365854,
+      "grad_norm": 0.35359427332878113,
+      "learning_rate": 8.536585365853658e-06,
+      "loss": 1.0118,
+      "mean_token_accuracy": 0.6959211826324463,
+      "num_tokens": 2619723.0,
+      "step": 191
+    },
+    {
+      "entropy": 1.0338459014892578,
+      "epoch": 0.4682926829268293,
+      "grad_norm": 0.397707998752594,
+      "learning_rate": 8.523035230352304e-06,
+      "loss": 1.0245,
+      "mean_token_accuracy": 0.7052842378616333,
+      "num_tokens": 2633213.0,
+      "step": 192
+    },
+    {
+      "entropy": 1.0224543809890747,
+      "epoch": 0.47073170731707314,
+      "grad_norm": 0.35734203457832336,
+      "learning_rate": 8.509485094850949e-06,
+      "loss": 1.0285,
+      "mean_token_accuracy": 0.6969306468963623,
+      "num_tokens": 2647206.0,
+      "step": 193
+    },
+    {
+      "entropy": 1.0173115730285645,
+      "epoch": 0.47317073170731705,
+      "grad_norm": 0.34693512320518494,
+      "learning_rate": 8.495934959349595e-06,
+      "loss": 1.0232,
+      "mean_token_accuracy": 0.6953300833702087,
+      "num_tokens": 2660991.0,
+      "step": 194
+    },
+    {
+      "entropy": 1.029531717300415,
+      "epoch": 0.47560975609756095,
+      "grad_norm": 0.3539057970046997,
+      "learning_rate": 8.482384823848239e-06,
+      "loss": 1.025,
+      "mean_token_accuracy": 0.6989356875419617,
+      "num_tokens": 2674725.0,
+      "step": 195
+    },
+    {
+      "entropy": 1.0025036334991455,
+      "epoch": 0.47804878048780486,
+      "grad_norm": 0.37867292761802673,
+      "learning_rate": 8.468834688346883e-06,
+      "loss": 1.0027,
+      "mean_token_accuracy": 0.7012137770652771,
+      "num_tokens": 2688500.0,
+      "step": 196
+    },
+    {
+      "entropy": 1.015578269958496,
+      "epoch": 0.48048780487804876,
+      "grad_norm": 0.38133302330970764,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.9988,
+      "mean_token_accuracy": 0.7015800476074219,
+      "num_tokens": 2702503.0,
+      "step": 197
+    },
+    {
+      "entropy": 1.0281040668487549,
+      "epoch": 0.48292682926829267,
+      "grad_norm": 0.3827400505542755,
+      "learning_rate": 8.441734417344174e-06,
+      "loss": 1.0144,
+      "mean_token_accuracy": 0.6993362903594971,
+      "num_tokens": 2716079.0,
+      "step": 198
+    },
+    {
+      "entropy": 1.0232656002044678,
+      "epoch": 0.4853658536585366,
+      "grad_norm": 0.38402533531188965,
+      "learning_rate": 8.42818428184282e-06,
+      "loss": 1.0117,
+      "mean_token_accuracy": 0.6984891891479492,
+      "num_tokens": 2729995.0,
+      "step": 199
+    },
+    {
+      "entropy": 1.0109443664550781,
+      "epoch": 0.4878048780487805,
+      "grad_norm": 0.3873913884162903,
+      "learning_rate": 8.414634146341464e-06,
+      "loss": 1.0022,
+      "mean_token_accuracy": 0.7005252838134766,
+      "num_tokens": 2743528.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0173959732055664,
+      "epoch": 0.4902439024390244,
+      "grad_norm": 0.34564492106437683,
+      "learning_rate": 8.401084010840109e-06,
+      "loss": 1.0041,
+      "mean_token_accuracy": 0.7038676142692566,
+      "num_tokens": 2757325.0,
+      "step": 201
+    },
+    {
+      "entropy": 1.0393359661102295,
+      "epoch": 0.4926829268292683,
+      "grad_norm": 0.4502638578414917,
+      "learning_rate": 8.387533875338753e-06,
+      "loss": 1.0201,
+      "mean_token_accuracy": 0.6990713477134705,
+      "num_tokens": 2770909.0,
+      "step": 202
+    },
+    {
+      "entropy": 1.0125484466552734,
+      "epoch": 0.4951219512195122,
+      "grad_norm": 0.6057789921760559,
+      "learning_rate": 8.373983739837399e-06,
+      "loss": 0.9897,
+      "mean_token_accuracy": 0.7036902904510498,
+      "num_tokens": 2784745.0,
+      "step": 203
+    },
+    {
+      "entropy": 1.0163042545318604,
+      "epoch": 0.4975609756097561,
+      "grad_norm": 0.47165754437446594,
+      "learning_rate": 8.360433604336045e-06,
+      "loss": 0.9996,
+      "mean_token_accuracy": 0.702346682548523,
+      "num_tokens": 2797971.0,
+      "step": 204
+    },
+    {
+      "entropy": 0.9835382103919983,
+      "epoch": 0.5,
+      "grad_norm": 0.38579699397087097,
+      "learning_rate": 8.34688346883469e-06,
+      "loss": 0.988,
+      "mean_token_accuracy": 0.7067919969558716,
+      "num_tokens": 2811179.0,
+      "step": 205
+    },
+    {
+      "entropy": 0.9862759113311768,
+      "epoch": 0.5024390243902439,
+      "grad_norm": 0.3595806956291199,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.9918,
+      "mean_token_accuracy": 0.7074604630470276,
+      "num_tokens": 2824800.0,
+      "step": 206
+    },
+    {
+      "entropy": 0.9990562200546265,
+      "epoch": 0.5048780487804878,
+      "grad_norm": 0.37352100014686584,
+      "learning_rate": 8.319783197831978e-06,
+      "loss": 0.986,
+      "mean_token_accuracy": 0.702506422996521,
+      "num_tokens": 2838860.0,
+      "step": 207
+    },
+    {
+      "entropy": 1.0180875062942505,
+      "epoch": 0.5073170731707317,
+      "grad_norm": 0.35853612422943115,
+      "learning_rate": 8.306233062330624e-06,
+      "loss": 1.0047,
+      "mean_token_accuracy": 0.7018040418624878,
+      "num_tokens": 2853011.0,
+      "step": 208
+    },
+    {
+      "entropy": 1.0083836317062378,
+      "epoch": 0.5097560975609756,
+      "grad_norm": 0.3947008550167084,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 0.9958,
+      "mean_token_accuracy": 0.7039703726768494,
+      "num_tokens": 2867056.0,
+      "step": 209
+    },
+    {
+      "entropy": 0.9842475056648254,
+      "epoch": 0.5121951219512195,
+      "grad_norm": 0.4123150408267975,
+      "learning_rate": 8.279132791327915e-06,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.7054763436317444,
+      "num_tokens": 2880256.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0007009506225586,
+      "epoch": 0.5146341463414634,
+      "grad_norm": 0.3687898814678192,
+      "learning_rate": 8.265582655826559e-06,
+      "loss": 0.99,
+      "mean_token_accuracy": 0.7061342000961304,
+      "num_tokens": 2894520.0,
+      "step": 211
+    },
+    {
+      "entropy": 1.0035302639007568,
+      "epoch": 0.5170731707317073,
+      "grad_norm": 0.42165127396583557,
+      "learning_rate": 8.252032520325203e-06,
+      "loss": 1.0181,
+      "mean_token_accuracy": 0.6972360014915466,
+      "num_tokens": 2907850.0,
+      "step": 212
+    },
+    {
+      "entropy": 1.0078985691070557,
+      "epoch": 0.5195121951219512,
+      "grad_norm": 0.36786288022994995,
+      "learning_rate": 8.23848238482385e-06,
+      "loss": 1.0063,
+      "mean_token_accuracy": 0.702234148979187,
+      "num_tokens": 2921652.0,
+      "step": 213
+    },
+    {
+      "entropy": 0.994750440120697,
+      "epoch": 0.5219512195121951,
+      "grad_norm": 0.3809518814086914,
+      "learning_rate": 8.224932249322494e-06,
+      "loss": 0.9888,
+      "mean_token_accuracy": 0.7052788138389587,
+      "num_tokens": 2935800.0,
+      "step": 214
+    },
+    {
+      "entropy": 0.9751050472259521,
+      "epoch": 0.524390243902439,
+      "grad_norm": 0.45397570729255676,
+      "learning_rate": 8.21138211382114e-06,
+      "loss": 0.9735,
+      "mean_token_accuracy": 0.712651252746582,
+      "num_tokens": 2949451.0,
+      "step": 215
+    },
+    {
+      "entropy": 1.0001088380813599,
+      "epoch": 0.526829268292683,
+      "grad_norm": 0.36515524983406067,
+      "learning_rate": 8.197831978319784e-06,
+      "loss": 1.0072,
+      "mean_token_accuracy": 0.7006496787071228,
+      "num_tokens": 2963474.0,
+      "step": 216
+    },
+    {
+      "entropy": 1.0082523822784424,
+      "epoch": 0.5292682926829269,
+      "grad_norm": 0.3793857991695404,
+      "learning_rate": 8.184281842818428e-06,
+      "loss": 1.0056,
+      "mean_token_accuracy": 0.6996766328811646,
+      "num_tokens": 2977405.0,
+      "step": 217
+    },
+    {
+      "entropy": 0.9938634634017944,
+      "epoch": 0.5317073170731708,
+      "grad_norm": 0.43802475929260254,
+      "learning_rate": 8.170731707317073e-06,
+      "loss": 0.997,
+      "mean_token_accuracy": 0.6989560127258301,
+      "num_tokens": 2990831.0,
+      "step": 218
+    },
+    {
+      "entropy": 0.989290177822113,
+      "epoch": 0.5341463414634147,
+      "grad_norm": 0.3721674978733063,
+      "learning_rate": 8.157181571815719e-06,
+      "loss": 0.9836,
+      "mean_token_accuracy": 0.7013733983039856,
+      "num_tokens": 3004754.0,
+      "step": 219
+    },
+    {
+      "entropy": 0.9925138354301453,
+      "epoch": 0.5365853658536586,
+      "grad_norm": 0.4069049656391144,
+      "learning_rate": 8.143631436314365e-06,
+      "loss": 0.9759,
+      "mean_token_accuracy": 0.7074543833732605,
+      "num_tokens": 3018091.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.9971339702606201,
+      "epoch": 0.5390243902439025,
+      "grad_norm": 0.3578735291957855,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.9816,
+      "mean_token_accuracy": 0.7043969035148621,
+      "num_tokens": 3031571.0,
+      "step": 221
+    },
+    {
+      "entropy": 0.9838342666625977,
+      "epoch": 0.5414634146341464,
+      "grad_norm": 0.38385337591171265,
+      "learning_rate": 8.116531165311653e-06,
+      "loss": 0.9589,
+      "mean_token_accuracy": 0.7182108163833618,
+      "num_tokens": 3045381.0,
+      "step": 222
+    },
+    {
+      "entropy": 0.9965362548828125,
+      "epoch": 0.5439024390243903,
+      "grad_norm": 0.396474689245224,
+      "learning_rate": 8.102981029810298e-06,
+      "loss": 0.9894,
+      "mean_token_accuracy": 0.703424870967865,
+      "num_tokens": 3058945.0,
+      "step": 223
+    },
+    {
+      "entropy": 0.9957606792449951,
+      "epoch": 0.5463414634146342,
+      "grad_norm": 0.35657909512519836,
+      "learning_rate": 8.089430894308944e-06,
+      "loss": 0.9834,
+      "mean_token_accuracy": 0.7055048942565918,
+      "num_tokens": 3072676.0,
+      "step": 224
+    },
+    {
+      "entropy": 0.9773147106170654,
+      "epoch": 0.5487804878048781,
+      "grad_norm": 0.35940998792648315,
+      "learning_rate": 8.075880758807588e-06,
+      "loss": 0.9623,
+      "mean_token_accuracy": 0.7097449898719788,
+      "num_tokens": 3086535.0,
+      "step": 225
+    },
+    {
+      "entropy": 0.9726331830024719,
+      "epoch": 0.551219512195122,
+      "grad_norm": 0.37580767273902893,
+      "learning_rate": 8.062330623306234e-06,
+      "loss": 0.9637,
+      "mean_token_accuracy": 0.7098819017410278,
+      "num_tokens": 3100435.0,
+      "step": 226
+    },
+    {
+      "entropy": 0.9689302444458008,
+      "epoch": 0.5536585365853659,
+      "grad_norm": 0.4062517285346985,
+      "learning_rate": 8.048780487804879e-06,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.7093740701675415,
+      "num_tokens": 3113743.0,
+      "step": 227
+    },
+    {
+      "entropy": 0.9909324049949646,
+      "epoch": 0.5560975609756098,
+      "grad_norm": 0.4221988022327423,
+      "learning_rate": 8.035230352303523e-06,
+      "loss": 0.9812,
+      "mean_token_accuracy": 0.7099443674087524,
+      "num_tokens": 3127415.0,
+      "step": 228
+    },
+    {
+      "entropy": 0.9781975746154785,
+      "epoch": 0.5585365853658537,
+      "grad_norm": 0.41590189933776855,
+      "learning_rate": 8.021680216802169e-06,
+      "loss": 0.9529,
+      "mean_token_accuracy": 0.7079962491989136,
+      "num_tokens": 3141400.0,
+      "step": 229
+    },
+    {
+      "entropy": 0.9742963314056396,
+      "epoch": 0.5609756097560976,
+      "grad_norm": 0.41367363929748535,
+      "learning_rate": 8.008130081300813e-06,
+      "loss": 0.979,
+      "mean_token_accuracy": 0.7071376442909241,
+      "num_tokens": 3155146.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.9977781772613525,
+      "epoch": 0.5634146341463414,
+      "grad_norm": 0.41266530752182007,
+      "learning_rate": 7.99457994579946e-06,
+      "loss": 0.9989,
+      "mean_token_accuracy": 0.7003742456436157,
+      "num_tokens": 3169323.0,
+      "step": 231
+    },
+    {
+      "entropy": 0.9736911058425903,
+      "epoch": 0.5658536585365853,
+      "grad_norm": 0.4872662127017975,
+      "learning_rate": 7.981029810298104e-06,
+      "loss": 0.9667,
+      "mean_token_accuracy": 0.7085639238357544,
+      "num_tokens": 3183071.0,
+      "step": 232
+    },
+    {
+      "entropy": 0.9661515355110168,
+      "epoch": 0.5682926829268292,
+      "grad_norm": 0.4976981580257416,
+      "learning_rate": 7.967479674796748e-06,
+      "loss": 0.9408,
+      "mean_token_accuracy": 0.7196811437606812,
+      "num_tokens": 3196511.0,
+      "step": 233
+    },
+    {
+      "entropy": 0.9788615107536316,
+      "epoch": 0.5707317073170731,
+      "grad_norm": 0.40486401319503784,
+      "learning_rate": 7.953929539295394e-06,
+      "loss": 0.9839,
+      "mean_token_accuracy": 0.7100006937980652,
+      "num_tokens": 3210496.0,
+      "step": 234
+    },
+    {
+      "entropy": 0.9701911211013794,
+      "epoch": 0.573170731707317,
+      "grad_norm": 0.3933921158313751,
+      "learning_rate": 7.940379403794039e-06,
+      "loss": 0.9784,
+      "mean_token_accuracy": 0.7084203958511353,
+      "num_tokens": 3224395.0,
+      "step": 235
+    },
+    {
+      "entropy": 0.9639068841934204,
+      "epoch": 0.5756097560975609,
+      "grad_norm": 0.4816792607307434,
+      "learning_rate": 7.926829268292685e-06,
+      "loss": 0.9585,
+      "mean_token_accuracy": 0.7153215408325195,
+      "num_tokens": 3238065.0,
+      "step": 236
+    },
+    {
+      "entropy": 0.9722325801849365,
+      "epoch": 0.5780487804878048,
+      "grad_norm": 0.5372746586799622,
+      "learning_rate": 7.913279132791329e-06,
+      "loss": 0.973,
+      "mean_token_accuracy": 0.7086333632469177,
+      "num_tokens": 3251552.0,
+      "step": 237
+    },
+    {
+      "entropy": 1.0175796747207642,
+      "epoch": 0.5804878048780487,
+      "grad_norm": 0.4492548406124115,
+      "learning_rate": 7.899728997289973e-06,
+      "loss": 1.0157,
+      "mean_token_accuracy": 0.7006236910820007,
+      "num_tokens": 3265517.0,
+      "step": 238
+    },
+    {
+      "entropy": 0.9735410213470459,
+      "epoch": 0.5829268292682926,
+      "grad_norm": 0.8624438643455505,
+      "learning_rate": 7.886178861788618e-06,
+      "loss": 0.9642,
+      "mean_token_accuracy": 0.709635853767395,
+      "num_tokens": 3279512.0,
+      "step": 239
+    },
+    {
+      "entropy": 0.9796257615089417,
+      "epoch": 0.5853658536585366,
+      "grad_norm": 0.4248017370700836,
+      "learning_rate": 7.872628726287264e-06,
+      "loss": 0.9693,
+      "mean_token_accuracy": 0.708400547504425,
+      "num_tokens": 3293170.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.9722896218299866,
+      "epoch": 0.5878048780487805,
+      "grad_norm": 0.3795738220214844,
+      "learning_rate": 7.859078590785908e-06,
+      "loss": 0.9604,
+      "mean_token_accuracy": 0.7118780612945557,
+      "num_tokens": 3306833.0,
+      "step": 241
+    },
+    {
+      "entropy": 0.9680063128471375,
+      "epoch": 0.5902439024390244,
+      "grad_norm": 0.3923177719116211,
+      "learning_rate": 7.845528455284554e-06,
+      "loss": 0.9593,
+      "mean_token_accuracy": 0.7090327143669128,
+      "num_tokens": 3320909.0,
+      "step": 242
+    },
+    {
+      "entropy": 0.9560147523880005,
+      "epoch": 0.5926829268292683,
+      "grad_norm": 0.41641271114349365,
+      "learning_rate": 7.831978319783198e-06,
+      "loss": 0.9324,
+      "mean_token_accuracy": 0.7116286158561707,
+      "num_tokens": 3334581.0,
+      "step": 243
+    },
+    {
+      "entropy": 0.9750783443450928,
+      "epoch": 0.5951219512195122,
+      "grad_norm": 0.3982071578502655,
+      "learning_rate": 7.818428184281843e-06,
+      "loss": 0.9632,
+      "mean_token_accuracy": 0.7131319046020508,
+      "num_tokens": 3348464.0,
+      "step": 244
+    },
+    {
+      "entropy": 0.9652288556098938,
+      "epoch": 0.5975609756097561,
+      "grad_norm": 0.4358637034893036,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.9391,
+      "mean_token_accuracy": 0.716895341873169,
+      "num_tokens": 3361892.0,
+      "step": 245
+    },
+    {
+      "entropy": 0.9535477161407471,
+      "epoch": 0.6,
+      "grad_norm": 0.4099048972129822,
+      "learning_rate": 7.791327913279133e-06,
+      "loss": 0.9522,
+      "mean_token_accuracy": 0.7090840935707092,
+      "num_tokens": 3375283.0,
+      "step": 246
+    },
+    {
+      "entropy": 0.9606568813323975,
+      "epoch": 0.6024390243902439,
+      "grad_norm": 0.3803389370441437,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 0.9627,
+      "mean_token_accuracy": 0.7069713473320007,
+      "num_tokens": 3389127.0,
+      "step": 247
+    },
+    {
+      "entropy": 0.9476461410522461,
+      "epoch": 0.6048780487804878,
+      "grad_norm": 0.4176231026649475,
+      "learning_rate": 7.764227642276424e-06,
+      "loss": 0.9437,
+      "mean_token_accuracy": 0.7153881192207336,
+      "num_tokens": 3402361.0,
+      "step": 248
+    },
+    {
+      "entropy": 0.9819697141647339,
+      "epoch": 0.6073170731707317,
+      "grad_norm": 0.3921288251876831,
+      "learning_rate": 7.750677506775068e-06,
+      "loss": 0.9895,
+      "mean_token_accuracy": 0.7068524956703186,
+      "num_tokens": 3416080.0,
+      "step": 249
+    },
+    {
+      "entropy": 0.976287841796875,
+      "epoch": 0.6097560975609756,
+      "grad_norm": 0.40087559819221497,
+      "learning_rate": 7.737127371273714e-06,
+      "loss": 0.9797,
+      "mean_token_accuracy": 0.709733247756958,
+      "num_tokens": 3429966.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.9725978374481201,
+      "epoch": 0.6121951219512195,
+      "grad_norm": 0.48437660932540894,
+      "learning_rate": 7.723577235772358e-06,
+      "loss": 0.967,
+      "mean_token_accuracy": 0.7101976871490479,
+      "num_tokens": 3443995.0,
+      "step": 251
+    },
+    {
+      "entropy": 0.9459497332572937,
+      "epoch": 0.6146341463414634,
+      "grad_norm": 0.5235543847084045,
+      "learning_rate": 7.710027100271004e-06,
+      "loss": 0.9323,
+      "mean_token_accuracy": 0.7215677499771118,
+      "num_tokens": 3457508.0,
+      "step": 252
+    },
+    {
+      "entropy": 0.9774629473686218,
+      "epoch": 0.6170731707317073,
+      "grad_norm": 0.3972068130970001,
+      "learning_rate": 7.696476964769649e-06,
+      "loss": 0.9694,
+      "mean_token_accuracy": 0.7105010151863098,
+      "num_tokens": 3471037.0,
+      "step": 253
+    },
+    {
+      "entropy": 0.9459494352340698,
+      "epoch": 0.6195121951219512,
+      "grad_norm": 0.4803354740142822,
+      "learning_rate": 7.682926829268293e-06,
+      "loss": 0.9277,
+      "mean_token_accuracy": 0.7180070281028748,
+      "num_tokens": 3484681.0,
+      "step": 254
+    },
+    {
+      "entropy": 0.9822655916213989,
+      "epoch": 0.6219512195121951,
+      "grad_norm": 0.5377360582351685,
+      "learning_rate": 7.669376693766937e-06,
+      "loss": 0.9671,
+      "mean_token_accuracy": 0.7102938890457153,
+      "num_tokens": 3498511.0,
+      "step": 255
+    },
+    {
+      "entropy": 0.9723156690597534,
+      "epoch": 0.624390243902439,
+      "grad_norm": 0.4719409644603729,
+      "learning_rate": 7.655826558265583e-06,
+      "loss": 0.9567,
+      "mean_token_accuracy": 0.7167521715164185,
+      "num_tokens": 3512370.0,
+      "step": 256
+    },
+    {
+      "entropy": 0.9557574391365051,
+      "epoch": 0.6268292682926829,
+      "grad_norm": 0.4471491575241089,
+      "learning_rate": 7.64227642276423e-06,
+      "loss": 0.948,
+      "mean_token_accuracy": 0.7144047617912292,
+      "num_tokens": 3525583.0,
+      "step": 257
+    },
+    {
+      "entropy": 0.974261999130249,
+      "epoch": 0.6292682926829268,
+      "grad_norm": 0.4518379271030426,
+      "learning_rate": 7.628726287262873e-06,
+      "loss": 0.9802,
+      "mean_token_accuracy": 0.7040947079658508,
+      "num_tokens": 3539617.0,
+      "step": 258
+    },
+    {
+      "entropy": 0.9723663330078125,
+      "epoch": 0.6317073170731707,
+      "grad_norm": 0.45028042793273926,
+      "learning_rate": 7.615176151761519e-06,
+      "loss": 0.9681,
+      "mean_token_accuracy": 0.7073206305503845,
+      "num_tokens": 3553416.0,
+      "step": 259
+    },
+    {
+      "entropy": 0.9442059993743896,
+      "epoch": 0.6341463414634146,
+      "grad_norm": 0.5071269273757935,
+      "learning_rate": 7.601626016260163e-06,
+      "loss": 0.9575,
+      "mean_token_accuracy": 0.7091909646987915,
+      "num_tokens": 3566695.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9819319248199463,
+      "epoch": 0.6365853658536585,
+      "grad_norm": 0.45541971921920776,
+      "learning_rate": 7.5880758807588085e-06,
+      "loss": 0.9808,
+      "mean_token_accuracy": 0.7078201770782471,
+      "num_tokens": 3580189.0,
+      "step": 261
+    },
+    {
+      "entropy": 0.9500574469566345,
+      "epoch": 0.6390243902439025,
+      "grad_norm": 0.5451672673225403,
+      "learning_rate": 7.574525745257453e-06,
+      "loss": 0.945,
+      "mean_token_accuracy": 0.7172244191169739,
+      "num_tokens": 3593622.0,
+      "step": 262
+    },
+    {
+      "entropy": 0.965298056602478,
+      "epoch": 0.6414634146341464,
+      "grad_norm": 0.45017296075820923,
+      "learning_rate": 7.560975609756098e-06,
+      "loss": 0.9552,
+      "mean_token_accuracy": 0.7146288156509399,
+      "num_tokens": 3606961.0,
+      "step": 263
+    },
+    {
+      "entropy": 0.9539521932601929,
+      "epoch": 0.6439024390243903,
+      "grad_norm": 0.47731754183769226,
+      "learning_rate": 7.547425474254744e-06,
+      "loss": 0.9404,
+      "mean_token_accuracy": 0.7130581736564636,
+      "num_tokens": 3620593.0,
+      "step": 264
+    },
+    {
+      "entropy": 0.9426309466362,
+      "epoch": 0.6463414634146342,
+      "grad_norm": 0.5125877857208252,
+      "learning_rate": 7.5338753387533885e-06,
+      "loss": 0.9448,
+      "mean_token_accuracy": 0.7107785940170288,
+      "num_tokens": 3634377.0,
+      "step": 265
+    },
+    {
+      "entropy": 0.968903660774231,
+      "epoch": 0.6487804878048781,
+      "grad_norm": 0.5131202340126038,
+      "learning_rate": 7.520325203252034e-06,
+      "loss": 0.9526,
+      "mean_token_accuracy": 0.7136064767837524,
+      "num_tokens": 3648695.0,
+      "step": 266
+    },
+    {
+      "entropy": 0.9822129011154175,
+      "epoch": 0.651219512195122,
+      "grad_norm": 0.45986267924308777,
+      "learning_rate": 7.506775067750678e-06,
+      "loss": 0.9754,
+      "mean_token_accuracy": 0.7104863524436951,
+      "num_tokens": 3662548.0,
+      "step": 267
+    },
+    {
+      "entropy": 0.9548493027687073,
+      "epoch": 0.6536585365853659,
+      "grad_norm": 0.4756591320037842,
+      "learning_rate": 7.493224932249323e-06,
+      "loss": 0.9522,
+      "mean_token_accuracy": 0.706981897354126,
+      "num_tokens": 3676099.0,
+      "step": 268
+    },
+    {
+      "entropy": 0.9379334449768066,
+      "epoch": 0.6560975609756098,
+      "grad_norm": 0.7421849370002747,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.9305,
+      "mean_token_accuracy": 0.714191734790802,
+      "num_tokens": 3689799.0,
+      "step": 269
+    },
+    {
+      "entropy": 0.9444692730903625,
+      "epoch": 0.6585365853658537,
+      "grad_norm": 0.7555662393569946,
+      "learning_rate": 7.466124661246613e-06,
+      "loss": 0.9489,
+      "mean_token_accuracy": 0.7124654054641724,
+      "num_tokens": 3703549.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.964323878288269,
+      "epoch": 0.6609756097560976,
+      "grad_norm": 0.4882325232028961,
+      "learning_rate": 7.452574525745257e-06,
+      "loss": 0.954,
+      "mean_token_accuracy": 0.714160144329071,
+      "num_tokens": 3717216.0,
+      "step": 271
+    },
+    {
+      "entropy": 0.9607169032096863,
+      "epoch": 0.6634146341463415,
+      "grad_norm": 0.6977524757385254,
+      "learning_rate": 7.439024390243903e-06,
+      "loss": 0.963,
+      "mean_token_accuracy": 0.7098829746246338,
+      "num_tokens": 3730730.0,
+      "step": 272
+    },
+    {
+      "entropy": 0.9752838611602783,
+      "epoch": 0.6658536585365854,
+      "grad_norm": 0.48511064052581787,
+      "learning_rate": 7.425474254742548e-06,
+      "loss": 0.9639,
+      "mean_token_accuracy": 0.7046083807945251,
+      "num_tokens": 3744829.0,
+      "step": 273
+    },
+    {
+      "entropy": 0.9718318581581116,
+      "epoch": 0.6682926829268293,
+      "grad_norm": 0.4946161210536957,
+      "learning_rate": 7.411924119241193e-06,
+      "loss": 0.9695,
+      "mean_token_accuracy": 0.7096002697944641,
+      "num_tokens": 3758230.0,
+      "step": 274
+    },
+    {
+      "entropy": 0.9614427089691162,
+      "epoch": 0.6707317073170732,
+      "grad_norm": 0.47201916575431824,
+      "learning_rate": 7.398373983739838e-06,
+      "loss": 0.9518,
+      "mean_token_accuracy": 0.7111756205558777,
+      "num_tokens": 3772026.0,
+      "step": 275
+    },
+    {
+      "entropy": 0.9677258729934692,
+      "epoch": 0.6731707317073171,
+      "grad_norm": 0.4112757742404938,
+      "learning_rate": 7.384823848238482e-06,
+      "loss": 0.9619,
+      "mean_token_accuracy": 0.7081936597824097,
+      "num_tokens": 3786065.0,
+      "step": 276
+    },
+    {
+      "entropy": 0.9812069535255432,
+      "epoch": 0.675609756097561,
+      "grad_norm": 0.5289693474769592,
+      "learning_rate": 7.371273712737128e-06,
+      "loss": 0.9772,
+      "mean_token_accuracy": 0.7038695812225342,
+      "num_tokens": 3799700.0,
+      "step": 277
+    },
+    {
+      "entropy": 0.9499857425689697,
+      "epoch": 0.6780487804878049,
+      "grad_norm": 0.4317754805088043,
+      "learning_rate": 7.357723577235773e-06,
+      "loss": 0.914,
+      "mean_token_accuracy": 0.7191169857978821,
+      "num_tokens": 3813170.0,
+      "step": 278
+    },
+    {
+      "entropy": 0.9827193021774292,
+      "epoch": 0.6804878048780488,
+      "grad_norm": 0.5651706457138062,
+      "learning_rate": 7.344173441734418e-06,
+      "loss": 0.9524,
+      "mean_token_accuracy": 0.7127862572669983,
+      "num_tokens": 3826810.0,
+      "step": 279
+    },
+    {
+      "entropy": 0.9605240821838379,
+      "epoch": 0.6829268292682927,
+      "grad_norm": 0.4189467430114746,
+      "learning_rate": 7.330623306233063e-06,
+      "loss": 0.9574,
+      "mean_token_accuracy": 0.7105111479759216,
+      "num_tokens": 3840716.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9565154314041138,
+      "epoch": 0.6853658536585366,
+      "grad_norm": 0.41188377141952515,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.9365,
+      "mean_token_accuracy": 0.7180280685424805,
+      "num_tokens": 3854627.0,
+      "step": 281
+    },
+    {
+      "entropy": 0.9569879770278931,
+      "epoch": 0.6878048780487804,
+      "grad_norm": 0.4937061369419098,
+      "learning_rate": 7.303523035230353e-06,
+      "loss": 0.9311,
+      "mean_token_accuracy": 0.7244262099266052,
+      "num_tokens": 3867801.0,
+      "step": 282
+    },
+    {
+      "entropy": 0.964173436164856,
+      "epoch": 0.6902439024390243,
+      "grad_norm": 0.4437481164932251,
+      "learning_rate": 7.289972899728998e-06,
+      "loss": 0.9502,
+      "mean_token_accuracy": 0.717008113861084,
+      "num_tokens": 3881093.0,
+      "step": 283
+    },
+    {
+      "entropy": 0.9317843914031982,
+      "epoch": 0.6926829268292682,
+      "grad_norm": 0.47311607003211975,
+      "learning_rate": 7.276422764227643e-06,
+      "loss": 0.9266,
+      "mean_token_accuracy": 0.7166216373443604,
+      "num_tokens": 3894808.0,
+      "step": 284
+    },
+    {
+      "entropy": 0.9279685616493225,
+      "epoch": 0.6951219512195121,
+      "grad_norm": 0.48283323645591736,
+      "learning_rate": 7.262872628726287e-06,
+      "loss": 0.9421,
+      "mean_token_accuracy": 0.7190238237380981,
+      "num_tokens": 3908633.0,
+      "step": 285
+    },
+    {
+      "entropy": 0.9323222637176514,
+      "epoch": 0.697560975609756,
+      "grad_norm": 0.41079843044281006,
+      "learning_rate": 7.2493224932249325e-06,
+      "loss": 0.9234,
+      "mean_token_accuracy": 0.7223949432373047,
+      "num_tokens": 3922478.0,
+      "step": 286
+    },
+    {
+      "entropy": 0.9287464618682861,
+      "epoch": 0.7,
+      "grad_norm": 0.44754576683044434,
+      "learning_rate": 7.2357723577235786e-06,
+      "loss": 0.9262,
+      "mean_token_accuracy": 0.7185174226760864,
+      "num_tokens": 3936200.0,
+      "step": 287
+    },
+    {
+      "entropy": 0.9283928871154785,
+      "epoch": 0.7024390243902439,
+      "grad_norm": 0.4761507511138916,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 0.9135,
+      "mean_token_accuracy": 0.7208345532417297,
+      "num_tokens": 3949828.0,
+      "step": 288
+    },
+    {
+      "entropy": 0.9376824498176575,
+      "epoch": 0.7048780487804878,
+      "grad_norm": 0.4816683828830719,
+      "learning_rate": 7.208672086720868e-06,
+      "loss": 0.9222,
+      "mean_token_accuracy": 0.7185068726539612,
+      "num_tokens": 3963855.0,
+      "step": 289
+    },
+    {
+      "entropy": 0.9271779656410217,
+      "epoch": 0.7073170731707317,
+      "grad_norm": 0.45386019349098206,
+      "learning_rate": 7.1951219512195125e-06,
+      "loss": 0.9326,
+      "mean_token_accuracy": 0.7165395021438599,
+      "num_tokens": 3977499.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9511876106262207,
+      "epoch": 0.7097560975609756,
+      "grad_norm": 0.43612465262413025,
+      "learning_rate": 7.181571815718158e-06,
+      "loss": 0.9382,
+      "mean_token_accuracy": 0.7135776281356812,
+      "num_tokens": 3990831.0,
+      "step": 291
+    },
+    {
+      "entropy": 0.934224009513855,
+      "epoch": 0.7121951219512195,
+      "grad_norm": 0.4678195118904114,
+      "learning_rate": 7.168021680216802e-06,
+      "loss": 0.9126,
+      "mean_token_accuracy": 0.7236147522926331,
+      "num_tokens": 4004491.0,
+      "step": 292
+    },
+    {
+      "entropy": 0.9473061561584473,
+      "epoch": 0.7146341463414634,
+      "grad_norm": 0.44049718976020813,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.9425,
+      "mean_token_accuracy": 0.7178494334220886,
+      "num_tokens": 4017936.0,
+      "step": 293
+    },
+    {
+      "entropy": 0.9487972259521484,
+      "epoch": 0.7170731707317073,
+      "grad_norm": 0.5107514262199402,
+      "learning_rate": 7.140921409214093e-06,
+      "loss": 0.9617,
+      "mean_token_accuracy": 0.7122842073440552,
+      "num_tokens": 4032084.0,
+      "step": 294
+    },
+    {
+      "entropy": 0.9367107152938843,
+      "epoch": 0.7195121951219512,
+      "grad_norm": 0.4118102490901947,
+      "learning_rate": 7.127371273712738e-06,
+      "loss": 0.937,
+      "mean_token_accuracy": 0.7204485535621643,
+      "num_tokens": 4046101.0,
+      "step": 295
+    },
+    {
+      "entropy": 0.9352602958679199,
+      "epoch": 0.7219512195121951,
+      "grad_norm": 0.45560237765312195,
+      "learning_rate": 7.113821138211383e-06,
+      "loss": 0.9545,
+      "mean_token_accuracy": 0.7115190029144287,
+      "num_tokens": 4059955.0,
+      "step": 296
+    },
+    {
+      "entropy": 0.9381440877914429,
+      "epoch": 0.724390243902439,
+      "grad_norm": 0.4506898522377014,
+      "learning_rate": 7.100271002710027e-06,
+      "loss": 0.9457,
+      "mean_token_accuracy": 0.7140668034553528,
+      "num_tokens": 4073677.0,
+      "step": 297
+    },
+    {
+      "entropy": 0.931757926940918,
+      "epoch": 0.7268292682926829,
+      "grad_norm": 0.47680553793907166,
+      "learning_rate": 7.086720867208673e-06,
+      "loss": 0.9383,
+      "mean_token_accuracy": 0.720937192440033,
+      "num_tokens": 4087052.0,
+      "step": 298
+    },
+    {
+      "entropy": 0.9337366819381714,
+      "epoch": 0.7292682926829268,
+      "grad_norm": 0.4333563446998596,
+      "learning_rate": 7.0731707317073175e-06,
+      "loss": 0.9283,
+      "mean_token_accuracy": 0.7173944711685181,
+      "num_tokens": 4100854.0,
+      "step": 299
+    },
+    {
+      "entropy": 0.9619350433349609,
+      "epoch": 0.7317073170731707,
+      "grad_norm": 0.45877236127853394,
+      "learning_rate": 7.059620596205963e-06,
+      "loss": 0.9604,
+      "mean_token_accuracy": 0.7089931964874268,
+      "num_tokens": 4114825.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9500558376312256,
+      "epoch": 0.7341463414634146,
+      "grad_norm": 0.608589231967926,
+      "learning_rate": 7.046070460704607e-06,
+      "loss": 0.9329,
+      "mean_token_accuracy": 0.7173042893409729,
+      "num_tokens": 4128329.0,
+      "step": 301
+    },
+    {
+      "entropy": 0.9519375562667847,
+      "epoch": 0.7365853658536585,
+      "grad_norm": 0.5312452912330627,
+      "learning_rate": 7.032520325203252e-06,
+      "loss": 0.9488,
+      "mean_token_accuracy": 0.7112184166908264,
+      "num_tokens": 4142224.0,
+      "step": 302
+    },
+    {
+      "entropy": 0.9525036811828613,
+      "epoch": 0.7390243902439024,
+      "grad_norm": 0.48584797978401184,
+      "learning_rate": 7.018970189701898e-06,
+      "loss": 0.9427,
+      "mean_token_accuracy": 0.713076114654541,
+      "num_tokens": 4155822.0,
+      "step": 303
+    },
+    {
+      "entropy": 0.9396979212760925,
+      "epoch": 0.7414634146341463,
+      "grad_norm": 0.61397385597229,
+      "learning_rate": 7.005420054200543e-06,
+      "loss": 0.9384,
+      "mean_token_accuracy": 0.7213865518569946,
+      "num_tokens": 4169599.0,
+      "step": 304
+    },
+    {
+      "entropy": 0.9650927782058716,
+      "epoch": 0.7439024390243902,
+      "grad_norm": 0.514038622379303,
+      "learning_rate": 6.991869918699188e-06,
+      "loss": 0.9353,
+      "mean_token_accuracy": 0.7186560034751892,
+      "num_tokens": 4183246.0,
+      "step": 305
+    },
+    {
+      "entropy": 0.9711487293243408,
+      "epoch": 0.7463414634146341,
+      "grad_norm": 0.45081955194473267,
+      "learning_rate": 6.978319783197832e-06,
+      "loss": 0.9557,
+      "mean_token_accuracy": 0.7125919461250305,
+      "num_tokens": 4196588.0,
+      "step": 306
+    },
+    {
+      "entropy": 0.9525703191757202,
+      "epoch": 0.748780487804878,
+      "grad_norm": 0.5370531678199768,
+      "learning_rate": 6.964769647696477e-06,
+      "loss": 0.9328,
+      "mean_token_accuracy": 0.7191635370254517,
+      "num_tokens": 4210281.0,
+      "step": 307
+    },
+    {
+      "entropy": 0.945855975151062,
+      "epoch": 0.751219512195122,
+      "grad_norm": 0.516020655632019,
+      "learning_rate": 6.951219512195122e-06,
+      "loss": 0.942,
+      "mean_token_accuracy": 0.7142351865768433,
+      "num_tokens": 4224431.0,
+      "step": 308
+    },
+    {
+      "entropy": 0.9390691518783569,
+      "epoch": 0.7536585365853659,
+      "grad_norm": 0.5693097114562988,
+      "learning_rate": 6.937669376693768e-06,
+      "loss": 0.9329,
+      "mean_token_accuracy": 0.71612149477005,
+      "num_tokens": 4238143.0,
+      "step": 309
+    },
+    {
+      "entropy": 0.9312563538551331,
+      "epoch": 0.7560975609756098,
+      "grad_norm": 0.5105025172233582,
+      "learning_rate": 6.924119241192413e-06,
+      "loss": 0.9361,
+      "mean_token_accuracy": 0.715452253818512,
+      "num_tokens": 4252487.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9333822727203369,
+      "epoch": 0.7585365853658537,
+      "grad_norm": 0.4418947994709015,
+      "learning_rate": 6.910569105691057e-06,
+      "loss": 0.9238,
+      "mean_token_accuracy": 0.7165272831916809,
+      "num_tokens": 4266014.0,
+      "step": 311
+    },
+    {
+      "entropy": 0.9335235357284546,
+      "epoch": 0.7609756097560976,
+      "grad_norm": 0.4711912274360657,
+      "learning_rate": 6.8970189701897025e-06,
+      "loss": 0.9385,
+      "mean_token_accuracy": 0.7139325141906738,
+      "num_tokens": 4279782.0,
+      "step": 312
+    },
+    {
+      "entropy": 0.9249697327613831,
+      "epoch": 0.7634146341463415,
+      "grad_norm": 0.4940471351146698,
+      "learning_rate": 6.883468834688347e-06,
+      "loss": 0.9284,
+      "mean_token_accuracy": 0.7152402400970459,
+      "num_tokens": 4293866.0,
+      "step": 313
+    },
+    {
+      "entropy": 0.9464020729064941,
+      "epoch": 0.7658536585365854,
+      "grad_norm": 0.4517545700073242,
+      "learning_rate": 6.869918699186993e-06,
+      "loss": 0.9326,
+      "mean_token_accuracy": 0.7157577276229858,
+      "num_tokens": 4307469.0,
+      "step": 314
+    },
+    {
+      "entropy": 0.9276844263076782,
+      "epoch": 0.7682926829268293,
+      "grad_norm": 0.46661779284477234,
+      "learning_rate": 6.856368563685637e-06,
+      "loss": 0.9393,
+      "mean_token_accuracy": 0.713527262210846,
+      "num_tokens": 4321235.0,
+      "step": 315
+    },
+    {
+      "entropy": 0.9301877617835999,
+      "epoch": 0.7707317073170732,
+      "grad_norm": 0.4871988296508789,
+      "learning_rate": 6.8428184281842825e-06,
+      "loss": 0.9268,
+      "mean_token_accuracy": 0.718126118183136,
+      "num_tokens": 4335126.0,
+      "step": 316
+    },
+    {
+      "entropy": 0.9497619867324829,
+      "epoch": 0.7731707317073171,
+      "grad_norm": 0.4936281144618988,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.9576,
+      "mean_token_accuracy": 0.7093750238418579,
+      "num_tokens": 4348582.0,
+      "step": 317
+    },
+    {
+      "entropy": 0.9262542724609375,
+      "epoch": 0.775609756097561,
+      "grad_norm": 0.47619983553886414,
+      "learning_rate": 6.815718157181572e-06,
+      "loss": 0.9105,
+      "mean_token_accuracy": 0.7227988839149475,
+      "num_tokens": 4362375.0,
+      "step": 318
+    },
+    {
+      "entropy": 0.9480463266372681,
+      "epoch": 0.7780487804878049,
+      "grad_norm": 0.46537765860557556,
+      "learning_rate": 6.802168021680218e-06,
+      "loss": 0.9353,
+      "mean_token_accuracy": 0.7166230082511902,
+      "num_tokens": 4376143.0,
+      "step": 319
+    },
+    {
+      "entropy": 0.9295194745063782,
+      "epoch": 0.7804878048780488,
+      "grad_norm": 0.4315537214279175,
+      "learning_rate": 6.788617886178862e-06,
+      "loss": 0.8985,
+      "mean_token_accuracy": 0.7267556190490723,
+      "num_tokens": 4389872.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9453158378601074,
+      "epoch": 0.7829268292682927,
+      "grad_norm": 0.47282472252845764,
+      "learning_rate": 6.775067750677508e-06,
+      "loss": 0.9413,
+      "mean_token_accuracy": 0.7136433720588684,
+      "num_tokens": 4404122.0,
+      "step": 321
+    },
+    {
+      "entropy": 0.9408867359161377,
+      "epoch": 0.7853658536585366,
+      "grad_norm": 0.47678133845329285,
+      "learning_rate": 6.761517615176152e-06,
+      "loss": 0.9193,
+      "mean_token_accuracy": 0.7188437581062317,
+      "num_tokens": 4417803.0,
+      "step": 322
+    },
+    {
+      "entropy": 0.9601535797119141,
+      "epoch": 0.7878048780487805,
+      "grad_norm": 0.47788533568382263,
+      "learning_rate": 6.747967479674797e-06,
+      "loss": 0.9403,
+      "mean_token_accuracy": 0.7107544541358948,
+      "num_tokens": 4431776.0,
+      "step": 323
+    },
+    {
+      "entropy": 0.9345723986625671,
+      "epoch": 0.7902439024390244,
+      "grad_norm": 0.47635290026664734,
+      "learning_rate": 6.734417344173443e-06,
+      "loss": 0.9283,
+      "mean_token_accuracy": 0.7155382037162781,
+      "num_tokens": 4445365.0,
+      "step": 324
+    },
+    {
+      "entropy": 0.931740939617157,
+      "epoch": 0.7926829268292683,
+      "grad_norm": 0.4577711224555969,
+      "learning_rate": 6.7208672086720876e-06,
+      "loss": 0.9163,
+      "mean_token_accuracy": 0.7215069532394409,
+      "num_tokens": 4459051.0,
+      "step": 325
+    },
+    {
+      "entropy": 0.9348794221878052,
+      "epoch": 0.7951219512195122,
+      "grad_norm": 0.4833238124847412,
+      "learning_rate": 6.707317073170733e-06,
+      "loss": 0.9421,
+      "mean_token_accuracy": 0.7192904353141785,
+      "num_tokens": 4472540.0,
+      "step": 326
+    },
+    {
+      "entropy": 0.9263391494750977,
+      "epoch": 0.7975609756097561,
+      "grad_norm": 0.5299957394599915,
+      "learning_rate": 6.693766937669377e-06,
+      "loss": 0.9143,
+      "mean_token_accuracy": 0.7203274965286255,
+      "num_tokens": 4486601.0,
+      "step": 327
+    },
+    {
+      "entropy": 0.9487115144729614,
+      "epoch": 0.8,
+      "grad_norm": 0.531969428062439,
+      "learning_rate": 6.680216802168022e-06,
+      "loss": 0.936,
+      "mean_token_accuracy": 0.7118086218833923,
+      "num_tokens": 4500285.0,
+      "step": 328
+    },
+    {
+      "entropy": 0.9246724843978882,
+      "epoch": 0.802439024390244,
+      "grad_norm": 0.47073879837989807,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.9381,
+      "mean_token_accuracy": 0.7152372002601624,
+      "num_tokens": 4513814.0,
+      "step": 329
+    },
+    {
+      "entropy": 0.9244076013565063,
+      "epoch": 0.8048780487804879,
+      "grad_norm": 0.5922364592552185,
+      "learning_rate": 6.653116531165313e-06,
+      "loss": 0.9274,
+      "mean_token_accuracy": 0.718658447265625,
+      "num_tokens": 4527486.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9245865345001221,
+      "epoch": 0.8073170731707318,
+      "grad_norm": 0.48464804887771606,
+      "learning_rate": 6.639566395663957e-06,
+      "loss": 0.9285,
+      "mean_token_accuracy": 0.718628466129303,
+      "num_tokens": 4541384.0,
+      "step": 331
+    },
+    {
+      "entropy": 0.9379637241363525,
+      "epoch": 0.8097560975609757,
+      "grad_norm": 0.5395787954330444,
+      "learning_rate": 6.626016260162602e-06,
+      "loss": 0.9311,
+      "mean_token_accuracy": 0.7203213572502136,
+      "num_tokens": 4554844.0,
+      "step": 332
+    },
+    {
+      "entropy": 0.9466052055358887,
+      "epoch": 0.8121951219512196,
+      "grad_norm": 0.48719871044158936,
+      "learning_rate": 6.6124661246612474e-06,
+      "loss": 0.9501,
+      "mean_token_accuracy": 0.7128361463546753,
+      "num_tokens": 4568953.0,
+      "step": 333
+    },
+    {
+      "entropy": 0.9478181600570679,
+      "epoch": 0.8146341463414634,
+      "grad_norm": 0.4469890892505646,
+      "learning_rate": 6.598915989159892e-06,
+      "loss": 0.9348,
+      "mean_token_accuracy": 0.716035783290863,
+      "num_tokens": 4582601.0,
+      "step": 334
+    },
+    {
+      "entropy": 0.946060299873352,
+      "epoch": 0.8170731707317073,
+      "grad_norm": 0.4337994456291199,
+      "learning_rate": 6.585365853658538e-06,
+      "loss": 0.9385,
+      "mean_token_accuracy": 0.7160581946372986,
+      "num_tokens": 4596641.0,
+      "step": 335
+    },
+    {
+      "entropy": 0.9306113719940186,
+      "epoch": 0.8195121951219512,
+      "grad_norm": 0.4518454074859619,
+      "learning_rate": 6.571815718157182e-06,
+      "loss": 0.8974,
+      "mean_token_accuracy": 0.724989116191864,
+      "num_tokens": 4610431.0,
+      "step": 336
+    },
+    {
+      "entropy": 0.9236544966697693,
+      "epoch": 0.8219512195121951,
+      "grad_norm": 0.5101668238639832,
+      "learning_rate": 6.558265582655827e-06,
+      "loss": 0.9058,
+      "mean_token_accuracy": 0.7232162952423096,
+      "num_tokens": 4623692.0,
+      "step": 337
+    },
+    {
+      "entropy": 0.9380990266799927,
+      "epoch": 0.824390243902439,
+      "grad_norm": 0.43267154693603516,
+      "learning_rate": 6.544715447154472e-06,
+      "loss": 0.9283,
+      "mean_token_accuracy": 0.7211065888404846,
+      "num_tokens": 4637552.0,
+      "step": 338
+    },
+    {
+      "entropy": 0.9355394840240479,
+      "epoch": 0.8268292682926829,
+      "grad_norm": 0.476979523897171,
+      "learning_rate": 6.531165311653117e-06,
+      "loss": 0.9242,
+      "mean_token_accuracy": 0.7181625366210938,
+      "num_tokens": 4651718.0,
+      "step": 339
+    },
+    {
+      "entropy": 0.9340901374816895,
+      "epoch": 0.8292682926829268,
+      "grad_norm": 0.4357958734035492,
+      "learning_rate": 6.517615176151762e-06,
+      "loss": 0.9247,
+      "mean_token_accuracy": 0.7212997674942017,
+      "num_tokens": 4665706.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.948611855506897,
+      "epoch": 0.8317073170731707,
+      "grad_norm": 0.49279242753982544,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.9382,
+      "mean_token_accuracy": 0.7165062427520752,
+      "num_tokens": 4679747.0,
+      "step": 341
+    },
+    {
+      "entropy": 0.9157364964485168,
+      "epoch": 0.8341463414634146,
+      "grad_norm": 0.4604078531265259,
+      "learning_rate": 6.4905149051490525e-06,
+      "loss": 0.9156,
+      "mean_token_accuracy": 0.7201681733131409,
+      "num_tokens": 4693557.0,
+      "step": 342
+    },
+    {
+      "entropy": 0.9359651207923889,
+      "epoch": 0.8365853658536585,
+      "grad_norm": 0.5489866137504578,
+      "learning_rate": 6.476964769647697e-06,
+      "loss": 0.9421,
+      "mean_token_accuracy": 0.7123420238494873,
+      "num_tokens": 4707023.0,
+      "step": 343
+    },
+    {
+      "entropy": 0.9256577491760254,
+      "epoch": 0.8390243902439024,
+      "grad_norm": 0.47433802485466003,
+      "learning_rate": 6.463414634146342e-06,
+      "loss": 0.9169,
+      "mean_token_accuracy": 0.716153085231781,
+      "num_tokens": 4720733.0,
+      "step": 344
+    },
+    {
+      "entropy": 0.9283899068832397,
+      "epoch": 0.8414634146341463,
+      "grad_norm": 0.42945265769958496,
+      "learning_rate": 6.449864498644986e-06,
+      "loss": 0.9259,
+      "mean_token_accuracy": 0.7218199372291565,
+      "num_tokens": 4734420.0,
+      "step": 345
+    },
+    {
+      "entropy": 0.9272925853729248,
+      "epoch": 0.8439024390243902,
+      "grad_norm": 0.5199980139732361,
+      "learning_rate": 6.436314363143632e-06,
+      "loss": 0.9121,
+      "mean_token_accuracy": 0.7235289812088013,
+      "num_tokens": 4748083.0,
+      "step": 346
+    },
+    {
+      "entropy": 0.9043445587158203,
+      "epoch": 0.8463414634146341,
+      "grad_norm": 0.535645604133606,
+      "learning_rate": 6.422764227642278e-06,
+      "loss": 0.9132,
+      "mean_token_accuracy": 0.7166420221328735,
+      "num_tokens": 4761619.0,
+      "step": 347
+    },
+    {
+      "entropy": 0.9312768578529358,
+      "epoch": 0.848780487804878,
+      "grad_norm": 0.480049729347229,
+      "learning_rate": 6.409214092140922e-06,
+      "loss": 0.9328,
+      "mean_token_accuracy": 0.722160816192627,
+      "num_tokens": 4775204.0,
+      "step": 348
+    },
+    {
+      "entropy": 0.9094958305358887,
+      "epoch": 0.8512195121951219,
+      "grad_norm": 0.5510603785514832,
+      "learning_rate": 6.395663956639567e-06,
+      "loss": 0.9133,
+      "mean_token_accuracy": 0.7258016467094421,
+      "num_tokens": 4788630.0,
+      "step": 349
+    },
+    {
+      "entropy": 0.9357163906097412,
+      "epoch": 0.8536585365853658,
+      "grad_norm": 0.4671006500720978,
+      "learning_rate": 6.3821138211382115e-06,
+      "loss": 0.9101,
+      "mean_token_accuracy": 0.7199651002883911,
+      "num_tokens": 4802405.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.911231517791748,
+      "epoch": 0.8560975609756097,
+      "grad_norm": 0.531291127204895,
+      "learning_rate": 6.368563685636857e-06,
+      "loss": 0.8899,
+      "mean_token_accuracy": 0.7275990843772888,
+      "num_tokens": 4816070.0,
+      "step": 351
+    },
+    {
+      "entropy": 0.9376004934310913,
+      "epoch": 0.8585365853658536,
+      "grad_norm": 0.5408096313476562,
+      "learning_rate": 6.355013550135501e-06,
+      "loss": 0.9137,
+      "mean_token_accuracy": 0.7203587293624878,
+      "num_tokens": 4829800.0,
+      "step": 352
+    },
+    {
+      "entropy": 0.9320589303970337,
+      "epoch": 0.8609756097560975,
+      "grad_norm": 0.4515257477760315,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 0.9247,
+      "mean_token_accuracy": 0.719613790512085,
+      "num_tokens": 4843383.0,
+      "step": 353
+    },
+    {
+      "entropy": 0.9337896108627319,
+      "epoch": 0.8634146341463415,
+      "grad_norm": 0.5287736058235168,
+      "learning_rate": 6.3279132791327915e-06,
+      "loss": 0.9259,
+      "mean_token_accuracy": 0.716210126876831,
+      "num_tokens": 4856761.0,
+      "step": 354
+    },
+    {
+      "entropy": 0.9203476309776306,
+      "epoch": 0.8658536585365854,
+      "grad_norm": 0.5867226719856262,
+      "learning_rate": 6.314363143631437e-06,
+      "loss": 0.9017,
+      "mean_token_accuracy": 0.724420964717865,
+      "num_tokens": 4870421.0,
+      "step": 355
+    },
+    {
+      "entropy": 0.943367063999176,
+      "epoch": 0.8682926829268293,
+      "grad_norm": 0.4877321124076843,
+      "learning_rate": 6.300813008130082e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7234193086624146,
+      "num_tokens": 4884545.0,
+      "step": 356
+    },
+    {
+      "entropy": 0.9346873760223389,
+      "epoch": 0.8707317073170732,
+      "grad_norm": 0.465480238199234,
+      "learning_rate": 6.287262872628726e-06,
+      "loss": 0.9226,
+      "mean_token_accuracy": 0.7138851881027222,
+      "num_tokens": 4898115.0,
+      "step": 357
+    },
+    {
+      "entropy": 0.9186785221099854,
+      "epoch": 0.8731707317073171,
+      "grad_norm": 0.560099184513092,
+      "learning_rate": 6.273712737127372e-06,
+      "loss": 0.9178,
+      "mean_token_accuracy": 0.7197883129119873,
+      "num_tokens": 4911735.0,
+      "step": 358
+    },
+    {
+      "entropy": 0.9136354923248291,
+      "epoch": 0.875609756097561,
+      "grad_norm": 0.5416183471679688,
+      "learning_rate": 6.260162601626017e-06,
+      "loss": 0.9112,
+      "mean_token_accuracy": 0.7166678309440613,
+      "num_tokens": 4925964.0,
+      "step": 359
+    },
+    {
+      "entropy": 0.8963789343833923,
+      "epoch": 0.8780487804878049,
+      "grad_norm": 0.5411756038665771,
+      "learning_rate": 6.246612466124662e-06,
+      "loss": 0.9084,
+      "mean_token_accuracy": 0.726434051990509,
+      "num_tokens": 4939421.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9152790307998657,
+      "epoch": 0.8804878048780488,
+      "grad_norm": 0.45827537775039673,
+      "learning_rate": 6.233062330623306e-06,
+      "loss": 0.924,
+      "mean_token_accuracy": 0.719045877456665,
+      "num_tokens": 4953272.0,
+      "step": 361
+    },
+    {
+      "entropy": 0.9214059114456177,
+      "epoch": 0.8829268292682927,
+      "grad_norm": 0.5113202333450317,
+      "learning_rate": 6.219512195121951e-06,
+      "loss": 0.9158,
+      "mean_token_accuracy": 0.7190465331077576,
+      "num_tokens": 4966671.0,
+      "step": 362
+    },
+    {
+      "entropy": 0.9017857313156128,
+      "epoch": 0.8853658536585366,
+      "grad_norm": 0.49038177728652954,
+      "learning_rate": 6.205962059620597e-06,
+      "loss": 0.892,
+      "mean_token_accuracy": 0.7244783639907837,
+      "num_tokens": 4980058.0,
+      "step": 363
+    },
+    {
+      "entropy": 0.9247295260429382,
+      "epoch": 0.8878048780487805,
+      "grad_norm": 0.4466925859451294,
+      "learning_rate": 6.192411924119242e-06,
+      "loss": 0.9259,
+      "mean_token_accuracy": 0.7192449569702148,
+      "num_tokens": 4993901.0,
+      "step": 364
+    },
+    {
+      "entropy": 0.9221506118774414,
+      "epoch": 0.8902439024390244,
+      "grad_norm": 0.4737914502620697,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.9219,
+      "mean_token_accuracy": 0.7163258194923401,
+      "num_tokens": 5007852.0,
+      "step": 365
+    },
+    {
+      "entropy": 0.9240057468414307,
+      "epoch": 0.8926829268292683,
+      "grad_norm": 0.5152718424797058,
+      "learning_rate": 6.165311653116531e-06,
+      "loss": 0.9287,
+      "mean_token_accuracy": 0.7177215218544006,
+      "num_tokens": 5021298.0,
+      "step": 366
+    },
+    {
+      "entropy": 0.9130709171295166,
+      "epoch": 0.8951219512195122,
+      "grad_norm": 0.5038972496986389,
+      "learning_rate": 6.1517615176151765e-06,
+      "loss": 0.9033,
+      "mean_token_accuracy": 0.726975679397583,
+      "num_tokens": 5034474.0,
+      "step": 367
+    },
+    {
+      "entropy": 0.9465330839157104,
+      "epoch": 0.8975609756097561,
+      "grad_norm": 0.48436981439590454,
+      "learning_rate": 6.138211382113821e-06,
+      "loss": 0.9469,
+      "mean_token_accuracy": 0.7148197889328003,
+      "num_tokens": 5047864.0,
+      "step": 368
+    },
+    {
+      "entropy": 0.9248801469802856,
+      "epoch": 0.9,
+      "grad_norm": 0.523639440536499,
+      "learning_rate": 6.124661246612467e-06,
+      "loss": 0.9253,
+      "mean_token_accuracy": 0.7153012156486511,
+      "num_tokens": 5061807.0,
+      "step": 369
+    },
+    {
+      "entropy": 0.9198777675628662,
+      "epoch": 0.9024390243902439,
+      "grad_norm": 0.4940233528614044,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 0.8889,
+      "mean_token_accuracy": 0.73076331615448,
+      "num_tokens": 5074871.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9299763441085815,
+      "epoch": 0.9048780487804878,
+      "grad_norm": 0.5024548768997192,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 0.9323,
+      "mean_token_accuracy": 0.7201763987541199,
+      "num_tokens": 5088492.0,
+      "step": 371
+    },
+    {
+      "entropy": 0.9322130680084229,
+      "epoch": 0.9073170731707317,
+      "grad_norm": 0.4858999252319336,
+      "learning_rate": 6.084010840108402e-06,
+      "loss": 0.9204,
+      "mean_token_accuracy": 0.7197042107582092,
+      "num_tokens": 5102165.0,
+      "step": 372
+    },
+    {
+      "entropy": 0.9305065870285034,
+      "epoch": 0.9097560975609756,
+      "grad_norm": 0.5411167144775391,
+      "learning_rate": 6.070460704607046e-06,
+      "loss": 0.911,
+      "mean_token_accuracy": 0.7243618369102478,
+      "num_tokens": 5115735.0,
+      "step": 373
+    },
+    {
+      "entropy": 0.9265807867050171,
+      "epoch": 0.9121951219512195,
+      "grad_norm": 0.5147826075553894,
+      "learning_rate": 6.056910569105692e-06,
+      "loss": 0.911,
+      "mean_token_accuracy": 0.7187079191207886,
+      "num_tokens": 5129125.0,
+      "step": 374
+    },
+    {
+      "entropy": 0.9361748695373535,
+      "epoch": 0.9146341463414634,
+      "grad_norm": 0.4886268079280853,
+      "learning_rate": 6.043360433604336e-06,
+      "loss": 0.9358,
+      "mean_token_accuracy": 0.7172378897666931,
+      "num_tokens": 5142884.0,
+      "step": 375
+    },
+    {
+      "entropy": 0.9086503386497498,
+      "epoch": 0.9170731707317074,
+      "grad_norm": 0.5013159513473511,
+      "learning_rate": 6.0298102981029816e-06,
+      "loss": 0.883,
+      "mean_token_accuracy": 0.7264388203620911,
+      "num_tokens": 5156418.0,
+      "step": 376
+    },
+    {
+      "entropy": 0.8983087539672852,
+      "epoch": 0.9195121951219513,
+      "grad_norm": 0.5629090070724487,
+      "learning_rate": 6.016260162601627e-06,
+      "loss": 0.8767,
+      "mean_token_accuracy": 0.7258298397064209,
+      "num_tokens": 5170232.0,
+      "step": 377
+    },
+    {
+      "entropy": 0.9251675605773926,
+      "epoch": 0.9219512195121952,
+      "grad_norm": 0.6035507321357727,
+      "learning_rate": 6.002710027100271e-06,
+      "loss": 0.9105,
+      "mean_token_accuracy": 0.7216352224349976,
+      "num_tokens": 5183971.0,
+      "step": 378
+    },
+    {
+      "entropy": 0.9210293292999268,
+      "epoch": 0.9243902439024391,
+      "grad_norm": 0.46510669589042664,
+      "learning_rate": 5.989159891598917e-06,
+      "loss": 0.913,
+      "mean_token_accuracy": 0.7173804044723511,
+      "num_tokens": 5197928.0,
+      "step": 379
+    },
+    {
+      "entropy": 0.9165781140327454,
+      "epoch": 0.926829268292683,
+      "grad_norm": 0.49598658084869385,
+      "learning_rate": 5.9756097560975615e-06,
+      "loss": 0.9322,
+      "mean_token_accuracy": 0.7150620818138123,
+      "num_tokens": 5211561.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9051110744476318,
+      "epoch": 0.9292682926829269,
+      "grad_norm": 0.5174202919006348,
+      "learning_rate": 5.962059620596207e-06,
+      "loss": 0.8931,
+      "mean_token_accuracy": 0.7223758697509766,
+      "num_tokens": 5225315.0,
+      "step": 381
+    },
+    {
+      "entropy": 0.8991729021072388,
+      "epoch": 0.9317073170731708,
+      "grad_norm": 0.5201939344406128,
+      "learning_rate": 5.948509485094851e-06,
+      "loss": 0.906,
+      "mean_token_accuracy": 0.7234763503074646,
+      "num_tokens": 5239196.0,
+      "step": 382
+    },
+    {
+      "entropy": 0.9231674671173096,
+      "epoch": 0.9341463414634147,
+      "grad_norm": 0.6580784916877747,
+      "learning_rate": 5.934959349593496e-06,
+      "loss": 0.9432,
+      "mean_token_accuracy": 0.716602087020874,
+      "num_tokens": 5253150.0,
+      "step": 383
+    },
+    {
+      "entropy": 0.9015557169914246,
+      "epoch": 0.9365853658536586,
+      "grad_norm": 0.6364657878875732,
+      "learning_rate": 5.921409214092141e-06,
+      "loss": 0.8855,
+      "mean_token_accuracy": 0.7272793650627136,
+      "num_tokens": 5266887.0,
+      "step": 384
+    },
+    {
+      "entropy": 0.9314663410186768,
+      "epoch": 0.9390243902439024,
+      "grad_norm": 0.4809994399547577,
+      "learning_rate": 5.907859078590787e-06,
+      "loss": 0.9233,
+      "mean_token_accuracy": 0.7218727469444275,
+      "num_tokens": 5280893.0,
+      "step": 385
+    },
+    {
+      "entropy": 0.9162091612815857,
+      "epoch": 0.9414634146341463,
+      "grad_norm": 0.4761136472225189,
+      "learning_rate": 5.894308943089432e-06,
+      "loss": 0.9116,
+      "mean_token_accuracy": 0.7201241850852966,
+      "num_tokens": 5294758.0,
+      "step": 386
+    },
+    {
+      "entropy": 0.937978982925415,
+      "epoch": 0.9439024390243902,
+      "grad_norm": 0.5258710384368896,
+      "learning_rate": 5.880758807588076e-06,
+      "loss": 0.9526,
+      "mean_token_accuracy": 0.7101801633834839,
+      "num_tokens": 5308762.0,
+      "step": 387
+    },
+    {
+      "entropy": 0.9138582348823547,
+      "epoch": 0.9463414634146341,
+      "grad_norm": 0.5211506485939026,
+      "learning_rate": 5.867208672086721e-06,
+      "loss": 0.9006,
+      "mean_token_accuracy": 0.7287028431892395,
+      "num_tokens": 5323017.0,
+      "step": 388
+    },
+    {
+      "entropy": 0.9176620244979858,
+      "epoch": 0.948780487804878,
+      "grad_norm": 0.4472411572933197,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.9192,
+      "mean_token_accuracy": 0.7202617526054382,
+      "num_tokens": 5336635.0,
+      "step": 389
+    },
+    {
+      "entropy": 0.9260638952255249,
+      "epoch": 0.9512195121951219,
+      "grad_norm": 0.5161044597625732,
+      "learning_rate": 5.840108401084012e-06,
+      "loss": 0.9329,
+      "mean_token_accuracy": 0.7174168825149536,
+      "num_tokens": 5350247.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9278532862663269,
+      "epoch": 0.9536585365853658,
+      "grad_norm": 0.462639719247818,
+      "learning_rate": 5.826558265582656e-06,
+      "loss": 0.9066,
+      "mean_token_accuracy": 0.7207022905349731,
+      "num_tokens": 5364445.0,
+      "step": 391
+    },
+    {
+      "entropy": 0.9160816669464111,
+      "epoch": 0.9560975609756097,
+      "grad_norm": 0.4721212387084961,
+      "learning_rate": 5.813008130081301e-06,
+      "loss": 0.894,
+      "mean_token_accuracy": 0.7259811758995056,
+      "num_tokens": 5378398.0,
+      "step": 392
+    },
+    {
+      "entropy": 0.9266201257705688,
+      "epoch": 0.9585365853658536,
+      "grad_norm": 0.5438399314880371,
+      "learning_rate": 5.7994579945799465e-06,
+      "loss": 0.914,
+      "mean_token_accuracy": 0.7273541688919067,
+      "num_tokens": 5391805.0,
+      "step": 393
+    },
+    {
+      "entropy": 0.925537645816803,
+      "epoch": 0.9609756097560975,
+      "grad_norm": 0.47833213210105896,
+      "learning_rate": 5.785907859078591e-06,
+      "loss": 0.923,
+      "mean_token_accuracy": 0.7158151268959045,
+      "num_tokens": 5405365.0,
+      "step": 394
+    },
+    {
+      "entropy": 0.921244204044342,
+      "epoch": 0.9634146341463414,
+      "grad_norm": 0.5551829934120178,
+      "learning_rate": 5.772357723577237e-06,
+      "loss": 0.9059,
+      "mean_token_accuracy": 0.7213079333305359,
+      "num_tokens": 5419174.0,
+      "step": 395
+    },
+    {
+      "entropy": 0.924101710319519,
+      "epoch": 0.9658536585365853,
+      "grad_norm": 0.4825878441333771,
+      "learning_rate": 5.758807588075881e-06,
+      "loss": 0.9019,
+      "mean_token_accuracy": 0.7207282185554504,
+      "num_tokens": 5432428.0,
+      "step": 396
+    },
+    {
+      "entropy": 0.9335201978683472,
+      "epoch": 0.9682926829268292,
+      "grad_norm": 0.46904629468917847,
+      "learning_rate": 5.7452574525745265e-06,
+      "loss": 0.9403,
+      "mean_token_accuracy": 0.7109548449516296,
+      "num_tokens": 5445954.0,
+      "step": 397
+    },
+    {
+      "entropy": 0.9025465250015259,
+      "epoch": 0.9707317073170731,
+      "grad_norm": 0.5336489081382751,
+      "learning_rate": 5.731707317073171e-06,
+      "loss": 0.8739,
+      "mean_token_accuracy": 0.7340313196182251,
+      "num_tokens": 5459246.0,
+      "step": 398
+    },
+    {
+      "entropy": 0.921517014503479,
+      "epoch": 0.973170731707317,
+      "grad_norm": 0.4557330012321472,
+      "learning_rate": 5.718157181571816e-06,
+      "loss": 0.9044,
+      "mean_token_accuracy": 0.7241076827049255,
+      "num_tokens": 5472934.0,
+      "step": 399
+    },
+    {
+      "entropy": 0.9142166972160339,
+      "epoch": 0.975609756097561,
+      "grad_norm": 0.48192092776298523,
+      "learning_rate": 5.704607046070462e-06,
+      "loss": 0.9115,
+      "mean_token_accuracy": 0.7226113677024841,
+      "num_tokens": 5486797.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.9260218143463135,
+      "epoch": 0.9780487804878049,
+      "grad_norm": 0.44798484444618225,
+      "learning_rate": 5.691056910569106e-06,
+      "loss": 0.9105,
+      "mean_token_accuracy": 0.7207956314086914,
+      "num_tokens": 5500638.0,
+      "step": 401
+    },
+    {
+      "entropy": 0.9152418375015259,
+      "epoch": 0.9804878048780488,
+      "grad_norm": 0.43449974060058594,
+      "learning_rate": 5.677506775067752e-06,
+      "loss": 0.9016,
+      "mean_token_accuracy": 0.7272465229034424,
+      "num_tokens": 5514542.0,
+      "step": 402
+    },
+    {
+      "entropy": 0.9501792788505554,
+      "epoch": 0.9829268292682927,
+      "grad_norm": 0.47718504071235657,
+      "learning_rate": 5.663956639566396e-06,
+      "loss": 0.9465,
+      "mean_token_accuracy": 0.7121170163154602,
+      "num_tokens": 5528918.0,
+      "step": 403
+    },
+    {
+      "entropy": 0.9009894728660583,
+      "epoch": 0.9853658536585366,
+      "grad_norm": 0.5538046360015869,
+      "learning_rate": 5.650406504065041e-06,
+      "loss": 0.9062,
+      "mean_token_accuracy": 0.7224763631820679,
+      "num_tokens": 5542486.0,
+      "step": 404
+    },
+    {
+      "entropy": 0.8964865207672119,
+      "epoch": 0.9878048780487805,
+      "grad_norm": 0.47334983944892883,
+      "learning_rate": 5.6368563685636855e-06,
+      "loss": 0.9,
+      "mean_token_accuracy": 0.7234845757484436,
+      "num_tokens": 5556541.0,
+      "step": 405
+    },
+    {
+      "entropy": 0.9112534523010254,
+      "epoch": 0.9902439024390244,
+      "grad_norm": 0.5056594610214233,
+      "learning_rate": 5.6233062330623315e-06,
+      "loss": 0.9111,
+      "mean_token_accuracy": 0.7229964137077332,
+      "num_tokens": 5570120.0,
+      "step": 406
+    },
+    {
+      "entropy": 0.9020826816558838,
+      "epoch": 0.9926829268292683,
+      "grad_norm": 0.45458662509918213,
+      "learning_rate": 5.609756097560977e-06,
+      "loss": 0.9081,
+      "mean_token_accuracy": 0.7202485203742981,
+      "num_tokens": 5583816.0,
+      "step": 407
+    },
+    {
+      "entropy": 0.9063427448272705,
+      "epoch": 0.9951219512195122,
+      "grad_norm": 0.44870609045028687,
+      "learning_rate": 5.596205962059621e-06,
+      "loss": 0.9073,
+      "mean_token_accuracy": 0.7196942567825317,
+      "num_tokens": 5597699.0,
+      "step": 408
+    },
+    {
+      "entropy": 0.9229121208190918,
+      "epoch": 0.9975609756097561,
+      "grad_norm": 0.47767025232315063,
+      "learning_rate": 5.582655826558266e-06,
+      "loss": 0.9143,
+      "mean_token_accuracy": 0.7199122905731201,
+      "num_tokens": 5611857.0,
+      "step": 409
+    },
+    {
+      "entropy": 0.9124419093132019,
+      "epoch": 1.0,
+      "grad_norm": 0.5221594572067261,
+      "learning_rate": 5.569105691056911e-06,
+      "loss": 0.8995,
+      "mean_token_accuracy": 0.7252015471458435,
+      "num_tokens": 5625392.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.9091067314147949,
+      "epoch": 1.002439024390244,
+      "grad_norm": 0.46902021765708923,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 0.9075,
+      "mean_token_accuracy": 0.7198294401168823,
+      "num_tokens": 5639478.0,
+      "step": 411
+    },
+    {
+      "entropy": 0.8968576192855835,
+      "epoch": 1.0048780487804878,
+      "grad_norm": 0.4639955759048462,
+      "learning_rate": 5.542005420054201e-06,
+      "loss": 0.8771,
+      "mean_token_accuracy": 0.7282954454421997,
+      "num_tokens": 5653005.0,
+      "step": 412
+    },
+    {
+      "entropy": 0.9134030342102051,
+      "epoch": 1.0073170731707317,
+      "grad_norm": 0.539460301399231,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 0.9019,
+      "mean_token_accuracy": 0.7240451574325562,
+      "num_tokens": 5666400.0,
+      "step": 413
+    },
+    {
+      "entropy": 0.9106577634811401,
+      "epoch": 1.0097560975609756,
+      "grad_norm": 0.5035801529884338,
+      "learning_rate": 5.5149051490514906e-06,
+      "loss": 0.8977,
+      "mean_token_accuracy": 0.7225700616836548,
+      "num_tokens": 5679832.0,
+      "step": 414
+    },
+    {
+      "entropy": 0.9107507467269897,
+      "epoch": 1.0121951219512195,
+      "grad_norm": 0.48390769958496094,
+      "learning_rate": 5.501355013550136e-06,
+      "loss": 0.9168,
+      "mean_token_accuracy": 0.7192770838737488,
+      "num_tokens": 5693958.0,
+      "step": 415
+    },
+    {
+      "entropy": 0.9108355045318604,
+      "epoch": 1.0146341463414634,
+      "grad_norm": 0.5258286595344543,
+      "learning_rate": 5.487804878048781e-06,
+      "loss": 0.9121,
+      "mean_token_accuracy": 0.7195926308631897,
+      "num_tokens": 5707622.0,
+      "step": 416
+    },
+    {
+      "entropy": 0.9191775918006897,
+      "epoch": 1.0170731707317073,
+      "grad_norm": 0.5931156277656555,
+      "learning_rate": 5.474254742547425e-06,
+      "loss": 0.9,
+      "mean_token_accuracy": 0.7233299612998962,
+      "num_tokens": 5721380.0,
+      "step": 417
+    },
+    {
+      "entropy": 0.9231445789337158,
+      "epoch": 1.0195121951219512,
+      "grad_norm": 0.5037340521812439,
+      "learning_rate": 5.460704607046071e-06,
+      "loss": 0.8959,
+      "mean_token_accuracy": 0.7243925929069519,
+      "num_tokens": 5734937.0,
+      "step": 418
+    },
+    {
+      "entropy": 0.9194551110267639,
+      "epoch": 1.0219512195121951,
+      "grad_norm": 0.4648572504520416,
+      "learning_rate": 5.447154471544716e-06,
+      "loss": 0.9066,
+      "mean_token_accuracy": 0.721906304359436,
+      "num_tokens": 5748319.0,
+      "step": 419
+    },
+    {
+      "entropy": 0.9239892959594727,
+      "epoch": 1.024390243902439,
+      "grad_norm": 0.49774280190467834,
+      "learning_rate": 5.433604336043361e-06,
+      "loss": 0.926,
+      "mean_token_accuracy": 0.7180191278457642,
+      "num_tokens": 5761804.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9286030530929565,
+      "epoch": 1.026829268292683,
+      "grad_norm": 0.526913046836853,
+      "learning_rate": 5.420054200542005e-06,
+      "loss": 0.9187,
+      "mean_token_accuracy": 0.7196871638298035,
+      "num_tokens": 5775758.0,
+      "step": 421
+    },
+    {
+      "entropy": 0.9007755517959595,
+      "epoch": 1.0292682926829269,
+      "grad_norm": 0.4655540883541107,
+      "learning_rate": 5.4065040650406504e-06,
+      "loss": 0.8932,
+      "mean_token_accuracy": 0.7245073914527893,
+      "num_tokens": 5789680.0,
+      "step": 422
+    },
+    {
+      "entropy": 0.9167541265487671,
+      "epoch": 1.0317073170731708,
+      "grad_norm": 0.49581092596054077,
+      "learning_rate": 5.3929539295392965e-06,
+      "loss": 0.9008,
+      "mean_token_accuracy": 0.7206687927246094,
+      "num_tokens": 5803751.0,
+      "step": 423
+    },
+    {
+      "entropy": 0.9178483486175537,
+      "epoch": 1.0341463414634147,
+      "grad_norm": 0.5530057549476624,
+      "learning_rate": 5.379403794037941e-06,
+      "loss": 0.9005,
+      "mean_token_accuracy": 0.7275153994560242,
+      "num_tokens": 5817254.0,
+      "step": 424
+    },
+    {
+      "entropy": 0.9028128385543823,
+      "epoch": 1.0365853658536586,
+      "grad_norm": 0.5000537633895874,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 0.9004,
+      "mean_token_accuracy": 0.7256110906600952,
+      "num_tokens": 5830893.0,
+      "step": 425
+    },
+    {
+      "entropy": 0.9217618107795715,
+      "epoch": 1.0390243902439025,
+      "grad_norm": 0.49565988779067993,
+      "learning_rate": 5.35230352303523e-06,
+      "loss": 0.9222,
+      "mean_token_accuracy": 0.7196056842803955,
+      "num_tokens": 5844604.0,
+      "step": 426
+    },
+    {
+      "entropy": 0.9046823978424072,
+      "epoch": 1.0414634146341464,
+      "grad_norm": 0.47147202491760254,
+      "learning_rate": 5.338753387533876e-06,
+      "loss": 0.9004,
+      "mean_token_accuracy": 0.724524974822998,
+      "num_tokens": 5858251.0,
+      "step": 427
+    },
+    {
+      "entropy": 0.8891482353210449,
+      "epoch": 1.0439024390243903,
+      "grad_norm": 0.5021435618400574,
+      "learning_rate": 5.32520325203252e-06,
+      "loss": 0.8821,
+      "mean_token_accuracy": 0.7333284020423889,
+      "num_tokens": 5871853.0,
+      "step": 428
+    },
+    {
+      "entropy": 0.9102505445480347,
+      "epoch": 1.0463414634146342,
+      "grad_norm": 0.7069030404090881,
+      "learning_rate": 5.311653116531166e-06,
+      "loss": 0.9061,
+      "mean_token_accuracy": 0.7191069722175598,
+      "num_tokens": 5885351.0,
+      "step": 429
+    },
+    {
+      "entropy": 0.8940747976303101,
+      "epoch": 1.048780487804878,
+      "grad_norm": 0.4877447783946991,
+      "learning_rate": 5.298102981029811e-06,
+      "loss": 0.9031,
+      "mean_token_accuracy": 0.7237207293510437,
+      "num_tokens": 5899418.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8975793123245239,
+      "epoch": 1.051219512195122,
+      "grad_norm": 0.49886640906333923,
+      "learning_rate": 5.2845528455284555e-06,
+      "loss": 0.9076,
+      "mean_token_accuracy": 0.7229393720626831,
+      "num_tokens": 5913301.0,
+      "step": 431
+    },
+    {
+      "entropy": 0.9058709144592285,
+      "epoch": 1.053658536585366,
+      "grad_norm": 0.5046048760414124,
+      "learning_rate": 5.271002710027101e-06,
+      "loss": 0.8963,
+      "mean_token_accuracy": 0.727680504322052,
+      "num_tokens": 5927139.0,
+      "step": 432
+    },
+    {
+      "entropy": 0.8988399505615234,
+      "epoch": 1.0560975609756098,
+      "grad_norm": 0.5127994418144226,
+      "learning_rate": 5.257452574525745e-06,
+      "loss": 0.8939,
+      "mean_token_accuracy": 0.7257552742958069,
+      "num_tokens": 5940395.0,
+      "step": 433
+    },
+    {
+      "entropy": 0.9131532907485962,
+      "epoch": 1.0585365853658537,
+      "grad_norm": 0.5204398036003113,
+      "learning_rate": 5.243902439024391e-06,
+      "loss": 0.9098,
+      "mean_token_accuracy": 0.7254641652107239,
+      "num_tokens": 5953983.0,
+      "step": 434
+    },
+    {
+      "entropy": 0.9116688966751099,
+      "epoch": 1.0609756097560976,
+      "grad_norm": 0.5195981860160828,
+      "learning_rate": 5.2303523035230355e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7243059873580933,
+      "num_tokens": 5967543.0,
+      "step": 435
+    },
+    {
+      "entropy": 0.8945907354354858,
+      "epoch": 1.0634146341463415,
+      "grad_norm": 0.4468877613544464,
+      "learning_rate": 5.216802168021681e-06,
+      "loss": 0.8946,
+      "mean_token_accuracy": 0.7255864143371582,
+      "num_tokens": 5981414.0,
+      "step": 436
+    },
+    {
+      "entropy": 0.914824366569519,
+      "epoch": 1.0658536585365854,
+      "grad_norm": 0.4595797657966614,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 0.9051,
+      "mean_token_accuracy": 0.7230360507965088,
+      "num_tokens": 5995356.0,
+      "step": 437
+    },
+    {
+      "entropy": 0.9173953533172607,
+      "epoch": 1.0682926829268293,
+      "grad_norm": 0.5476388931274414,
+      "learning_rate": 5.18970189701897e-06,
+      "loss": 0.9105,
+      "mean_token_accuracy": 0.7220655083656311,
+      "num_tokens": 6009199.0,
+      "step": 438
+    },
+    {
+      "entropy": 0.9245563745498657,
+      "epoch": 1.0707317073170732,
+      "grad_norm": 0.5139727592468262,
+      "learning_rate": 5.176151761517616e-06,
+      "loss": 0.9118,
+      "mean_token_accuracy": 0.7199491262435913,
+      "num_tokens": 6023366.0,
+      "step": 439
+    },
+    {
+      "entropy": 0.8988422751426697,
+      "epoch": 1.0731707317073171,
+      "grad_norm": 0.5930190682411194,
+      "learning_rate": 5.162601626016261e-06,
+      "loss": 0.8856,
+      "mean_token_accuracy": 0.7224959135055542,
+      "num_tokens": 6036780.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.9217547178268433,
+      "epoch": 1.075609756097561,
+      "grad_norm": 0.534253716468811,
+      "learning_rate": 5.149051490514906e-06,
+      "loss": 0.9108,
+      "mean_token_accuracy": 0.7249597907066345,
+      "num_tokens": 6050474.0,
+      "step": 441
+    },
+    {
+      "entropy": 0.9051541090011597,
+      "epoch": 1.078048780487805,
+      "grad_norm": 0.5235859751701355,
+      "learning_rate": 5.13550135501355e-06,
+      "loss": 0.8801,
+      "mean_token_accuracy": 0.7299248576164246,
+      "num_tokens": 6063927.0,
+      "step": 442
+    },
+    {
+      "entropy": 0.9091707468032837,
+      "epoch": 1.0804878048780489,
+      "grad_norm": 0.5509324669837952,
+      "learning_rate": 5.121951219512195e-06,
+      "loss": 0.9093,
+      "mean_token_accuracy": 0.7241198420524597,
+      "num_tokens": 6077293.0,
+      "step": 443
+    },
+    {
+      "entropy": 0.9139906167984009,
+      "epoch": 1.0829268292682928,
+      "grad_norm": 0.4984675943851471,
+      "learning_rate": 5.10840108401084e-06,
+      "loss": 0.8988,
+      "mean_token_accuracy": 0.7256140112876892,
+      "num_tokens": 6091559.0,
+      "step": 444
+    },
+    {
+      "entropy": 0.8962512016296387,
+      "epoch": 1.0853658536585367,
+      "grad_norm": 0.6049586534500122,
+      "learning_rate": 5.094850948509486e-06,
+      "loss": 0.8985,
+      "mean_token_accuracy": 0.7295030951499939,
+      "num_tokens": 6106004.0,
+      "step": 445
+    },
+    {
+      "entropy": 0.9080502986907959,
+      "epoch": 1.0878048780487806,
+      "grad_norm": 0.539002001285553,
+      "learning_rate": 5.081300813008131e-06,
+      "loss": 0.9128,
+      "mean_token_accuracy": 0.7214069366455078,
+      "num_tokens": 6119240.0,
+      "step": 446
+    },
+    {
+      "entropy": 0.9243647456169128,
+      "epoch": 1.0902439024390245,
+      "grad_norm": 0.5728839635848999,
+      "learning_rate": 5.067750677506775e-06,
+      "loss": 0.9147,
+      "mean_token_accuracy": 0.7210649251937866,
+      "num_tokens": 6133417.0,
+      "step": 447
+    },
+    {
+      "entropy": 0.9063730835914612,
+      "epoch": 1.0926829268292684,
+      "grad_norm": 0.5486031174659729,
+      "learning_rate": 5.0542005420054205e-06,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.7223870158195496,
+      "num_tokens": 6147593.0,
+      "step": 448
+    },
+    {
+      "entropy": 0.9127067923545837,
+      "epoch": 1.0951219512195123,
+      "grad_norm": 0.5353381037712097,
+      "learning_rate": 5.040650406504065e-06,
+      "loss": 0.908,
+      "mean_token_accuracy": 0.7239224314689636,
+      "num_tokens": 6161529.0,
+      "step": 449
+    },
+    {
+      "entropy": 0.900060772895813,
+      "epoch": 1.0975609756097562,
+      "grad_norm": 0.49772390723228455,
+      "learning_rate": 5.027100271002711e-06,
+      "loss": 0.8893,
+      "mean_token_accuracy": 0.7243174910545349,
+      "num_tokens": 6175757.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.9074562788009644,
+      "epoch": 1.1,
+      "grad_norm": 0.47336098551750183,
+      "learning_rate": 5.013550135501355e-06,
+      "loss": 0.8936,
+      "mean_token_accuracy": 0.7258216142654419,
+      "num_tokens": 6189618.0,
+      "step": 451
+    },
+    {
+      "entropy": 0.9215185642242432,
+      "epoch": 1.102439024390244,
+      "grad_norm": 0.49442532658576965,
+      "learning_rate": 5e-06,
+      "loss": 0.9186,
+      "mean_token_accuracy": 0.7188286781311035,
+      "num_tokens": 6203533.0,
+      "step": 452
+    },
+    {
+      "entropy": 0.9018423557281494,
+      "epoch": 1.104878048780488,
+      "grad_norm": 0.5464769601821899,
+      "learning_rate": 4.986449864498646e-06,
+      "loss": 0.904,
+      "mean_token_accuracy": 0.7272529602050781,
+      "num_tokens": 6217342.0,
+      "step": 453
+    },
+    {
+      "entropy": 0.9130271673202515,
+      "epoch": 1.1073170731707318,
+      "grad_norm": 0.4416399896144867,
+      "learning_rate": 4.97289972899729e-06,
+      "loss": 0.8913,
+      "mean_token_accuracy": 0.7259694933891296,
+      "num_tokens": 6231309.0,
+      "step": 454
+    },
+    {
+      "entropy": 0.8990261554718018,
+      "epoch": 1.1097560975609757,
+      "grad_norm": 0.4484340250492096,
+      "learning_rate": 4.959349593495935e-06,
+      "loss": 0.8827,
+      "mean_token_accuracy": 0.7287976741790771,
+      "num_tokens": 6245215.0,
+      "step": 455
+    },
+    {
+      "entropy": 0.8985174298286438,
+      "epoch": 1.1121951219512196,
+      "grad_norm": 0.47807013988494873,
+      "learning_rate": 4.94579945799458e-06,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7291353344917297,
+      "num_tokens": 6259190.0,
+      "step": 456
+    },
+    {
+      "entropy": 0.8863782286643982,
+      "epoch": 1.1146341463414635,
+      "grad_norm": 0.44421708583831787,
+      "learning_rate": 4.9322493224932255e-06,
+      "loss": 0.8759,
+      "mean_token_accuracy": 0.7294890880584717,
+      "num_tokens": 6273357.0,
+      "step": 457
+    },
+    {
+      "entropy": 0.8936686515808105,
+      "epoch": 1.1170731707317074,
+      "grad_norm": 0.5020765066146851,
+      "learning_rate": 4.918699186991871e-06,
+      "loss": 0.9018,
+      "mean_token_accuracy": 0.7257601618766785,
+      "num_tokens": 6286956.0,
+      "step": 458
+    },
+    {
+      "entropy": 0.8936822414398193,
+      "epoch": 1.1195121951219513,
+      "grad_norm": 0.4957248568534851,
+      "learning_rate": 4.905149051490515e-06,
+      "loss": 0.8913,
+      "mean_token_accuracy": 0.7259597778320312,
+      "num_tokens": 6300751.0,
+      "step": 459
+    },
+    {
+      "entropy": 0.8922320604324341,
+      "epoch": 1.1219512195121952,
+      "grad_norm": 0.5009384155273438,
+      "learning_rate": 4.89159891598916e-06,
+      "loss": 0.8858,
+      "mean_token_accuracy": 0.7239802479743958,
+      "num_tokens": 6314545.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.9002817869186401,
+      "epoch": 1.1243902439024391,
+      "grad_norm": 0.5000170469284058,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.8839,
+      "mean_token_accuracy": 0.7265538573265076,
+      "num_tokens": 6328092.0,
+      "step": 461
+    },
+    {
+      "entropy": 0.8917896747589111,
+      "epoch": 1.126829268292683,
+      "grad_norm": 0.4810425639152527,
+      "learning_rate": 4.86449864498645e-06,
+      "loss": 0.886,
+      "mean_token_accuracy": 0.7309262752532959,
+      "num_tokens": 6342067.0,
+      "step": 462
+    },
+    {
+      "entropy": 0.9265782833099365,
+      "epoch": 1.129268292682927,
+      "grad_norm": 0.474269300699234,
+      "learning_rate": 4.850948509485095e-06,
+      "loss": 0.9401,
+      "mean_token_accuracy": 0.7133171558380127,
+      "num_tokens": 6356095.0,
+      "step": 463
+    },
+    {
+      "entropy": 0.9106369018554688,
+      "epoch": 1.1317073170731708,
+      "grad_norm": 0.49770355224609375,
+      "learning_rate": 4.83739837398374e-06,
+      "loss": 0.8978,
+      "mean_token_accuracy": 0.7211055159568787,
+      "num_tokens": 6369643.0,
+      "step": 464
+    },
+    {
+      "entropy": 0.8969599008560181,
+      "epoch": 1.1341463414634148,
+      "grad_norm": 0.47881314158439636,
+      "learning_rate": 4.823848238482385e-06,
+      "loss": 0.9047,
+      "mean_token_accuracy": 0.7260048985481262,
+      "num_tokens": 6383068.0,
+      "step": 465
+    },
+    {
+      "entropy": 0.9027596712112427,
+      "epoch": 1.1365853658536587,
+      "grad_norm": 0.5635483264923096,
+      "learning_rate": 4.810298102981031e-06,
+      "loss": 0.9149,
+      "mean_token_accuracy": 0.7189622521400452,
+      "num_tokens": 6396922.0,
+      "step": 466
+    },
+    {
+      "entropy": 0.8914520740509033,
+      "epoch": 1.1390243902439026,
+      "grad_norm": 0.5290432572364807,
+      "learning_rate": 4.796747967479675e-06,
+      "loss": 0.8913,
+      "mean_token_accuracy": 0.7268030047416687,
+      "num_tokens": 6410873.0,
+      "step": 467
+    },
+    {
+      "entropy": 0.8867425918579102,
+      "epoch": 1.1414634146341462,
+      "grad_norm": 0.48660776019096375,
+      "learning_rate": 4.78319783197832e-06,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.729336678981781,
+      "num_tokens": 6424367.0,
+      "step": 468
+    },
+    {
+      "entropy": 0.9146620035171509,
+      "epoch": 1.1439024390243901,
+      "grad_norm": 0.5063614249229431,
+      "learning_rate": 4.769647696476965e-06,
+      "loss": 0.9136,
+      "mean_token_accuracy": 0.724082350730896,
+      "num_tokens": 6438032.0,
+      "step": 469
+    },
+    {
+      "entropy": 0.9144189357757568,
+      "epoch": 1.146341463414634,
+      "grad_norm": 0.5035169124603271,
+      "learning_rate": 4.75609756097561e-06,
+      "loss": 0.9137,
+      "mean_token_accuracy": 0.7198021411895752,
+      "num_tokens": 6451592.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.9157655835151672,
+      "epoch": 1.148780487804878,
+      "grad_norm": 0.5481422543525696,
+      "learning_rate": 4.742547425474256e-06,
+      "loss": 0.8973,
+      "mean_token_accuracy": 0.7282554507255554,
+      "num_tokens": 6465577.0,
+      "step": 471
+    },
+    {
+      "entropy": 0.8913492560386658,
+      "epoch": 1.1512195121951219,
+      "grad_norm": 0.5407872796058655,
+      "learning_rate": 4.7289972899729e-06,
+      "loss": 0.8807,
+      "mean_token_accuracy": 0.7290219068527222,
+      "num_tokens": 6478690.0,
+      "step": 472
+    },
+    {
+      "entropy": 0.9125957489013672,
+      "epoch": 1.1536585365853658,
+      "grad_norm": 0.4850672483444214,
+      "learning_rate": 4.715447154471545e-06,
+      "loss": 0.8974,
+      "mean_token_accuracy": 0.7230581641197205,
+      "num_tokens": 6492662.0,
+      "step": 473
+    },
+    {
+      "entropy": 0.9092464447021484,
+      "epoch": 1.1560975609756097,
+      "grad_norm": 0.4803924560546875,
+      "learning_rate": 4.7018970189701905e-06,
+      "loss": 0.8974,
+      "mean_token_accuracy": 0.7263503670692444,
+      "num_tokens": 6506082.0,
+      "step": 474
+    },
+    {
+      "entropy": 0.9019302129745483,
+      "epoch": 1.1585365853658536,
+      "grad_norm": 0.4694143533706665,
+      "learning_rate": 4.688346883468835e-06,
+      "loss": 0.8888,
+      "mean_token_accuracy": 0.7246914505958557,
+      "num_tokens": 6520115.0,
+      "step": 475
+    },
+    {
+      "entropy": 0.9142214059829712,
+      "epoch": 1.1609756097560975,
+      "grad_norm": 0.5448134541511536,
+      "learning_rate": 4.67479674796748e-06,
+      "loss": 0.8928,
+      "mean_token_accuracy": 0.7227379679679871,
+      "num_tokens": 6533703.0,
+      "step": 476
+    },
+    {
+      "entropy": 0.8932230472564697,
+      "epoch": 1.1634146341463414,
+      "grad_norm": 0.7330875992774963,
+      "learning_rate": 4.661246612466125e-06,
+      "loss": 0.8704,
+      "mean_token_accuracy": 0.7294765114784241,
+      "num_tokens": 6547167.0,
+      "step": 477
+    },
+    {
+      "entropy": 0.8862720727920532,
+      "epoch": 1.1658536585365853,
+      "grad_norm": 0.5509917736053467,
+      "learning_rate": 4.64769647696477e-06,
+      "loss": 0.882,
+      "mean_token_accuracy": 0.7343223094940186,
+      "num_tokens": 6560530.0,
+      "step": 478
+    },
+    {
+      "entropy": 0.8873111009597778,
+      "epoch": 1.1682926829268292,
+      "grad_norm": 0.4889533817768097,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 0.8727,
+      "mean_token_accuracy": 0.731568455696106,
+      "num_tokens": 6574069.0,
+      "step": 479
+    },
+    {
+      "entropy": 0.8986578583717346,
+      "epoch": 1.170731707317073,
+      "grad_norm": 0.4992435872554779,
+      "learning_rate": 4.62059620596206e-06,
+      "loss": 0.8813,
+      "mean_token_accuracy": 0.726729154586792,
+      "num_tokens": 6587632.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8890163898468018,
+      "epoch": 1.173170731707317,
+      "grad_norm": 0.5475744009017944,
+      "learning_rate": 4.607046070460705e-06,
+      "loss": 0.8792,
+      "mean_token_accuracy": 0.7279131412506104,
+      "num_tokens": 6601276.0,
+      "step": 481
+    },
+    {
+      "entropy": 0.8914072513580322,
+      "epoch": 1.175609756097561,
+      "grad_norm": 0.49075618386268616,
+      "learning_rate": 4.59349593495935e-06,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7293311357498169,
+      "num_tokens": 6614718.0,
+      "step": 482
+    },
+    {
+      "entropy": 0.9068721532821655,
+      "epoch": 1.1780487804878048,
+      "grad_norm": 0.5013757348060608,
+      "learning_rate": 4.579945799457995e-06,
+      "loss": 0.8965,
+      "mean_token_accuracy": 0.7248654365539551,
+      "num_tokens": 6628480.0,
+      "step": 483
+    },
+    {
+      "entropy": 0.8967253565788269,
+      "epoch": 1.1804878048780487,
+      "grad_norm": 0.450345516204834,
+      "learning_rate": 4.56639566395664e-06,
+      "loss": 0.891,
+      "mean_token_accuracy": 0.7259194850921631,
+      "num_tokens": 6642335.0,
+      "step": 484
+    },
+    {
+      "entropy": 0.9039151668548584,
+      "epoch": 1.1829268292682926,
+      "grad_norm": 0.5018711090087891,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 0.8985,
+      "mean_token_accuracy": 0.7184143662452698,
+      "num_tokens": 6656503.0,
+      "step": 485
+    },
+    {
+      "entropy": 0.8783119916915894,
+      "epoch": 1.1853658536585365,
+      "grad_norm": 0.4550807476043701,
+      "learning_rate": 4.53929539295393e-06,
+      "loss": 0.8816,
+      "mean_token_accuracy": 0.7290191650390625,
+      "num_tokens": 6670782.0,
+      "step": 486
+    },
+    {
+      "entropy": 0.8858023881912231,
+      "epoch": 1.1878048780487804,
+      "grad_norm": 0.5283997058868408,
+      "learning_rate": 4.5257452574525755e-06,
+      "loss": 0.9076,
+      "mean_token_accuracy": 0.7234495282173157,
+      "num_tokens": 6684423.0,
+      "step": 487
+    },
+    {
+      "entropy": 0.9109866619110107,
+      "epoch": 1.1902439024390243,
+      "grad_norm": 0.4794978201389313,
+      "learning_rate": 4.51219512195122e-06,
+      "loss": 0.9218,
+      "mean_token_accuracy": 0.7165172100067139,
+      "num_tokens": 6698267.0,
+      "step": 488
+    },
+    {
+      "entropy": 0.9090756177902222,
+      "epoch": 1.1926829268292682,
+      "grad_norm": 0.5581902861595154,
+      "learning_rate": 4.498644986449865e-06,
+      "loss": 0.889,
+      "mean_token_accuracy": 0.728282630443573,
+      "num_tokens": 6711786.0,
+      "step": 489
+    },
+    {
+      "entropy": 0.8913673758506775,
+      "epoch": 1.1951219512195121,
+      "grad_norm": 0.5424763560295105,
+      "learning_rate": 4.485094850948509e-06,
+      "loss": 0.886,
+      "mean_token_accuracy": 0.7309855818748474,
+      "num_tokens": 6725489.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.9181253910064697,
+      "epoch": 1.197560975609756,
+      "grad_norm": 0.4987359941005707,
+      "learning_rate": 4.471544715447155e-06,
+      "loss": 0.9156,
+      "mean_token_accuracy": 0.7176190614700317,
+      "num_tokens": 6738962.0,
+      "step": 491
+    },
+    {
+      "entropy": 0.8765039443969727,
+      "epoch": 1.2,
+      "grad_norm": 0.4425133764743805,
+      "learning_rate": 4.4579945799458e-06,
+      "loss": 0.8714,
+      "mean_token_accuracy": 0.732286810874939,
+      "num_tokens": 6752993.0,
+      "step": 492
+    },
+    {
+      "entropy": 0.9064300060272217,
+      "epoch": 1.2024390243902439,
+      "grad_norm": 0.5201969146728516,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.9121,
+      "mean_token_accuracy": 0.7229897379875183,
+      "num_tokens": 6767124.0,
+      "step": 493
+    },
+    {
+      "entropy": 0.9056813716888428,
+      "epoch": 1.2048780487804878,
+      "grad_norm": 0.6040856242179871,
+      "learning_rate": 4.43089430894309e-06,
+      "loss": 0.9076,
+      "mean_token_accuracy": 0.7237012386322021,
+      "num_tokens": 6781038.0,
+      "step": 494
+    },
+    {
+      "entropy": 0.8948585987091064,
+      "epoch": 1.2073170731707317,
+      "grad_norm": 0.4676818251609802,
+      "learning_rate": 4.4173441734417345e-06,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7293363213539124,
+      "num_tokens": 6794931.0,
+      "step": 495
+    },
+    {
+      "entropy": 0.9097365736961365,
+      "epoch": 1.2097560975609756,
+      "grad_norm": 0.506108820438385,
+      "learning_rate": 4.40379403794038e-06,
+      "loss": 0.9102,
+      "mean_token_accuracy": 0.7242258191108704,
+      "num_tokens": 6808284.0,
+      "step": 496
+    },
+    {
+      "entropy": 0.8956904411315918,
+      "epoch": 1.2121951219512195,
+      "grad_norm": 0.4974040389060974,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 0.8766,
+      "mean_token_accuracy": 0.7297356724739075,
+      "num_tokens": 6821994.0,
+      "step": 497
+    },
+    {
+      "entropy": 0.899440348148346,
+      "epoch": 1.2146341463414634,
+      "grad_norm": 0.5107272267341614,
+      "learning_rate": 4.376693766937669e-06,
+      "loss": 0.8803,
+      "mean_token_accuracy": 0.7293535470962524,
+      "num_tokens": 6835947.0,
+      "step": 498
+    },
+    {
+      "entropy": 0.9155231714248657,
+      "epoch": 1.2170731707317073,
+      "grad_norm": 0.5254876017570496,
+      "learning_rate": 4.3631436314363145e-06,
+      "loss": 0.9069,
+      "mean_token_accuracy": 0.7245771884918213,
+      "num_tokens": 6849386.0,
+      "step": 499
+    },
+    {
+      "entropy": 0.9062044620513916,
+      "epoch": 1.2195121951219512,
+      "grad_norm": 0.4622846841812134,
+      "learning_rate": 4.34959349593496e-06,
+      "loss": 0.8909,
+      "mean_token_accuracy": 0.7278857827186584,
+      "num_tokens": 6863341.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.9044604301452637,
+      "epoch": 1.221951219512195,
+      "grad_norm": 0.5532169342041016,
+      "learning_rate": 4.336043360433605e-06,
+      "loss": 0.8998,
+      "mean_token_accuracy": 0.7176086902618408,
+      "num_tokens": 6877157.0,
+      "step": 501
+    },
+    {
+      "entropy": 0.8994400501251221,
+      "epoch": 1.224390243902439,
+      "grad_norm": 0.5434163212776184,
+      "learning_rate": 4.32249322493225e-06,
+      "loss": 0.8922,
+      "mean_token_accuracy": 0.7263439893722534,
+      "num_tokens": 6890975.0,
+      "step": 502
+    },
+    {
+      "entropy": 0.9025291204452515,
+      "epoch": 1.226829268292683,
+      "grad_norm": 0.4925203323364258,
+      "learning_rate": 4.308943089430894e-06,
+      "loss": 0.8849,
+      "mean_token_accuracy": 0.7318164706230164,
+      "num_tokens": 6904176.0,
+      "step": 503
+    },
+    {
+      "entropy": 0.9300906658172607,
+      "epoch": 1.2292682926829268,
+      "grad_norm": 0.5197744369506836,
+      "learning_rate": 4.29539295392954e-06,
+      "loss": 0.9405,
+      "mean_token_accuracy": 0.7121908664703369,
+      "num_tokens": 6917899.0,
+      "step": 504
+    },
+    {
+      "entropy": 0.9055182933807373,
+      "epoch": 1.2317073170731707,
+      "grad_norm": 0.4847884476184845,
+      "learning_rate": 4.281842818428185e-06,
+      "loss": 0.9092,
+      "mean_token_accuracy": 0.722909688949585,
+      "num_tokens": 6931896.0,
+      "step": 505
+    },
+    {
+      "entropy": 0.9060989618301392,
+      "epoch": 1.2341463414634146,
+      "grad_norm": 0.4853264093399048,
+      "learning_rate": 4.268292682926829e-06,
+      "loss": 0.9044,
+      "mean_token_accuracy": 0.7221659421920776,
+      "num_tokens": 6945726.0,
+      "step": 506
+    },
+    {
+      "entropy": 0.8736509084701538,
+      "epoch": 1.2365853658536585,
+      "grad_norm": 0.48868390917778015,
+      "learning_rate": 4.254742547425474e-06,
+      "loss": 0.8564,
+      "mean_token_accuracy": 0.7347749471664429,
+      "num_tokens": 6959338.0,
+      "step": 507
+    },
+    {
+      "entropy": 0.925621747970581,
+      "epoch": 1.2390243902439024,
+      "grad_norm": 0.5540333390235901,
+      "learning_rate": 4.2411924119241196e-06,
+      "loss": 0.913,
+      "mean_token_accuracy": 0.7208184003829956,
+      "num_tokens": 6973284.0,
+      "step": 508
+    },
+    {
+      "entropy": 0.911636233329773,
+      "epoch": 1.2414634146341463,
+      "grad_norm": 0.5570753812789917,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.9143,
+      "mean_token_accuracy": 0.7139846086502075,
+      "num_tokens": 6987058.0,
+      "step": 509
+    },
+    {
+      "entropy": 0.9004988670349121,
+      "epoch": 1.2439024390243902,
+      "grad_norm": 0.5094306468963623,
+      "learning_rate": 4.21409214092141e-06,
+      "loss": 0.8758,
+      "mean_token_accuracy": 0.724961519241333,
+      "num_tokens": 7000723.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8960776329040527,
+      "epoch": 1.2463414634146341,
+      "grad_norm": 0.5077755451202393,
+      "learning_rate": 4.200542005420054e-06,
+      "loss": 0.8984,
+      "mean_token_accuracy": 0.7270680665969849,
+      "num_tokens": 7014508.0,
+      "step": 511
+    },
+    {
+      "entropy": 0.9104132056236267,
+      "epoch": 1.248780487804878,
+      "grad_norm": 0.48007309436798096,
+      "learning_rate": 4.1869918699186995e-06,
+      "loss": 0.9044,
+      "mean_token_accuracy": 0.7218325138092041,
+      "num_tokens": 7028494.0,
+      "step": 512
+    },
+    {
+      "entropy": 0.8949484825134277,
+      "epoch": 1.251219512195122,
+      "grad_norm": 0.4952548146247864,
+      "learning_rate": 4.173441734417345e-06,
+      "loss": 0.8752,
+      "mean_token_accuracy": 0.729067862033844,
+      "num_tokens": 7042639.0,
+      "step": 513
+    },
+    {
+      "entropy": 0.9143623113632202,
+      "epoch": 1.2536585365853659,
+      "grad_norm": 0.4703666567802429,
+      "learning_rate": 4.159891598915989e-06,
+      "loss": 0.897,
+      "mean_token_accuracy": 0.7206144332885742,
+      "num_tokens": 7056131.0,
+      "step": 514
+    },
+    {
+      "entropy": 0.9043334126472473,
+      "epoch": 1.2560975609756098,
+      "grad_norm": 0.5099503397941589,
+      "learning_rate": 4.146341463414634e-06,
+      "loss": 0.8831,
+      "mean_token_accuracy": 0.7296978831291199,
+      "num_tokens": 7069717.0,
+      "step": 515
+    },
+    {
+      "entropy": 0.9112102389335632,
+      "epoch": 1.2585365853658537,
+      "grad_norm": 0.5303446054458618,
+      "learning_rate": 4.1327913279132794e-06,
+      "loss": 0.9021,
+      "mean_token_accuracy": 0.7211565971374512,
+      "num_tokens": 7083601.0,
+      "step": 516
+    },
+    {
+      "entropy": 0.8933310508728027,
+      "epoch": 1.2609756097560976,
+      "grad_norm": 0.49321338534355164,
+      "learning_rate": 4.119241192411925e-06,
+      "loss": 0.8779,
+      "mean_token_accuracy": 0.7297635078430176,
+      "num_tokens": 7097231.0,
+      "step": 517
+    },
+    {
+      "entropy": 0.8844630122184753,
+      "epoch": 1.2634146341463415,
+      "grad_norm": 0.46473798155784607,
+      "learning_rate": 4.10569105691057e-06,
+      "loss": 0.8904,
+      "mean_token_accuracy": 0.726249098777771,
+      "num_tokens": 7111457.0,
+      "step": 518
+    },
+    {
+      "entropy": 0.8918311595916748,
+      "epoch": 1.2658536585365854,
+      "grad_norm": 0.474924772977829,
+      "learning_rate": 4.092140921409214e-06,
+      "loss": 0.8927,
+      "mean_token_accuracy": 0.7301933169364929,
+      "num_tokens": 7125231.0,
+      "step": 519
+    },
+    {
+      "entropy": 0.8817993402481079,
+      "epoch": 1.2682926829268293,
+      "grad_norm": 0.601287841796875,
+      "learning_rate": 4.078590785907859e-06,
+      "loss": 0.8788,
+      "mean_token_accuracy": 0.7307068109512329,
+      "num_tokens": 7138801.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8969420194625854,
+      "epoch": 1.2707317073170732,
+      "grad_norm": 0.5099979043006897,
+      "learning_rate": 4.0650406504065046e-06,
+      "loss": 0.8869,
+      "mean_token_accuracy": 0.7282543778419495,
+      "num_tokens": 7152245.0,
+      "step": 521
+    },
+    {
+      "entropy": 0.8839466571807861,
+      "epoch": 1.273170731707317,
+      "grad_norm": 0.45876964926719666,
+      "learning_rate": 4.051490514905149e-06,
+      "loss": 0.8703,
+      "mean_token_accuracy": 0.7337388396263123,
+      "num_tokens": 7166236.0,
+      "step": 522
+    },
+    {
+      "entropy": 0.901630163192749,
+      "epoch": 1.275609756097561,
+      "grad_norm": 0.539262592792511,
+      "learning_rate": 4.037940379403794e-06,
+      "loss": 0.8866,
+      "mean_token_accuracy": 0.7242372035980225,
+      "num_tokens": 7180148.0,
+      "step": 523
+    },
+    {
+      "entropy": 0.883599042892456,
+      "epoch": 1.278048780487805,
+      "grad_norm": 0.504664421081543,
+      "learning_rate": 4.024390243902439e-06,
+      "loss": 0.8716,
+      "mean_token_accuracy": 0.7333284616470337,
+      "num_tokens": 7193900.0,
+      "step": 524
+    },
+    {
+      "entropy": 0.8952665328979492,
+      "epoch": 1.2804878048780488,
+      "grad_norm": 0.4933059811592102,
+      "learning_rate": 4.0108401084010845e-06,
+      "loss": 0.8975,
+      "mean_token_accuracy": 0.7243708372116089,
+      "num_tokens": 7207982.0,
+      "step": 525
+    },
+    {
+      "entropy": 0.8998792171478271,
+      "epoch": 1.2829268292682927,
+      "grad_norm": 0.4748615026473999,
+      "learning_rate": 3.99728997289973e-06,
+      "loss": 0.9043,
+      "mean_token_accuracy": 0.7223044037818909,
+      "num_tokens": 7222197.0,
+      "step": 526
+    },
+    {
+      "entropy": 0.8993245363235474,
+      "epoch": 1.2853658536585366,
+      "grad_norm": 0.5399959683418274,
+      "learning_rate": 3.983739837398374e-06,
+      "loss": 0.9002,
+      "mean_token_accuracy": 0.7197199463844299,
+      "num_tokens": 7236067.0,
+      "step": 527
+    },
+    {
+      "entropy": 0.8935285806655884,
+      "epoch": 1.2878048780487805,
+      "grad_norm": 0.5220499038696289,
+      "learning_rate": 3.970189701897019e-06,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.7281142473220825,
+      "num_tokens": 7250019.0,
+      "step": 528
+    },
+    {
+      "entropy": 0.8838009834289551,
+      "epoch": 1.2902439024390244,
+      "grad_norm": 0.49501916766166687,
+      "learning_rate": 3.9566395663956644e-06,
+      "loss": 0.8915,
+      "mean_token_accuracy": 0.7248112559318542,
+      "num_tokens": 7263811.0,
+      "step": 529
+    },
+    {
+      "entropy": 0.895301103591919,
+      "epoch": 1.2926829268292683,
+      "grad_norm": 0.493888795375824,
+      "learning_rate": 3.943089430894309e-06,
+      "loss": 0.8935,
+      "mean_token_accuracy": 0.7277877330780029,
+      "num_tokens": 7277243.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.9079329967498779,
+      "epoch": 1.2951219512195122,
+      "grad_norm": 0.47917982935905457,
+      "learning_rate": 3.929539295392954e-06,
+      "loss": 0.9023,
+      "mean_token_accuracy": 0.724371075630188,
+      "num_tokens": 7291013.0,
+      "step": 531
+    },
+    {
+      "entropy": 0.9105004668235779,
+      "epoch": 1.2975609756097561,
+      "grad_norm": 0.5026032328605652,
+      "learning_rate": 3.915989159891599e-06,
+      "loss": 0.9191,
+      "mean_token_accuracy": 0.7229947447776794,
+      "num_tokens": 7304693.0,
+      "step": 532
+    },
+    {
+      "entropy": 0.8809376955032349,
+      "epoch": 1.3,
+      "grad_norm": 0.47273725271224976,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.7340049743652344,
+      "num_tokens": 7318010.0,
+      "step": 533
+    },
+    {
+      "entropy": 0.8872978687286377,
+      "epoch": 1.302439024390244,
+      "grad_norm": 0.46795758605003357,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 0.8789,
+      "mean_token_accuracy": 0.726021409034729,
+      "num_tokens": 7331757.0,
+      "step": 534
+    },
+    {
+      "entropy": 0.8998485207557678,
+      "epoch": 1.3048780487804879,
+      "grad_norm": 0.5320788025856018,
+      "learning_rate": 3.875338753387534e-06,
+      "loss": 0.8863,
+      "mean_token_accuracy": 0.7274210453033447,
+      "num_tokens": 7345868.0,
+      "step": 535
+    },
+    {
+      "entropy": 0.8918326497077942,
+      "epoch": 1.3073170731707318,
+      "grad_norm": 0.5132240056991577,
+      "learning_rate": 3.861788617886179e-06,
+      "loss": 0.8678,
+      "mean_token_accuracy": 0.7323923110961914,
+      "num_tokens": 7359628.0,
+      "step": 536
+    },
+    {
+      "entropy": 0.8886317014694214,
+      "epoch": 1.3097560975609757,
+      "grad_norm": 0.5985807776451111,
+      "learning_rate": 3.848238482384824e-06,
+      "loss": 0.8806,
+      "mean_token_accuracy": 0.7288233637809753,
+      "num_tokens": 7373480.0,
+      "step": 537
+    },
+    {
+      "entropy": 0.8903034329414368,
+      "epoch": 1.3121951219512196,
+      "grad_norm": 0.549172580242157,
+      "learning_rate": 3.834688346883469e-06,
+      "loss": 0.8965,
+      "mean_token_accuracy": 0.7226969599723816,
+      "num_tokens": 7387304.0,
+      "step": 538
+    },
+    {
+      "entropy": 0.8851415514945984,
+      "epoch": 1.3146341463414635,
+      "grad_norm": 0.5681750774383545,
+      "learning_rate": 3.821138211382115e-06,
+      "loss": 0.879,
+      "mean_token_accuracy": 0.7248347401618958,
+      "num_tokens": 7400632.0,
+      "step": 539
+    },
+    {
+      "entropy": 0.882635235786438,
+      "epoch": 1.3170731707317074,
+      "grad_norm": 0.5316335558891296,
+      "learning_rate": 3.8075880758807595e-06,
+      "loss": 0.8583,
+      "mean_token_accuracy": 0.7311773300170898,
+      "num_tokens": 7414408.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8970792293548584,
+      "epoch": 1.3195121951219513,
+      "grad_norm": 0.5417896509170532,
+      "learning_rate": 3.7940379403794043e-06,
+      "loss": 0.8972,
+      "mean_token_accuracy": 0.7224768996238708,
+      "num_tokens": 7428603.0,
+      "step": 541
+    },
+    {
+      "entropy": 0.9087656736373901,
+      "epoch": 1.3219512195121952,
+      "grad_norm": 0.5408928990364075,
+      "learning_rate": 3.780487804878049e-06,
+      "loss": 0.8969,
+      "mean_token_accuracy": 0.7309739589691162,
+      "num_tokens": 7442521.0,
+      "step": 542
+    },
+    {
+      "entropy": 0.9143552780151367,
+      "epoch": 1.324390243902439,
+      "grad_norm": 0.5758176445960999,
+      "learning_rate": 3.7669376693766942e-06,
+      "loss": 0.9149,
+      "mean_token_accuracy": 0.7155868411064148,
+      "num_tokens": 7456042.0,
+      "step": 543
+    },
+    {
+      "entropy": 0.8879578113555908,
+      "epoch": 1.326829268292683,
+      "grad_norm": 0.5441621541976929,
+      "learning_rate": 3.753387533875339e-06,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.7230884432792664,
+      "num_tokens": 7469398.0,
+      "step": 544
+    },
+    {
+      "entropy": 0.8851050734519958,
+      "epoch": 1.329268292682927,
+      "grad_norm": 0.5097270607948303,
+      "learning_rate": 3.7398373983739838e-06,
+      "loss": 0.8702,
+      "mean_token_accuracy": 0.7331685423851013,
+      "num_tokens": 7483168.0,
+      "step": 545
+    },
+    {
+      "entropy": 0.8914124965667725,
+      "epoch": 1.3317073170731708,
+      "grad_norm": 0.4741462171077728,
+      "learning_rate": 3.7262872628726286e-06,
+      "loss": 0.8776,
+      "mean_token_accuracy": 0.7253215909004211,
+      "num_tokens": 7497022.0,
+      "step": 546
+    },
+    {
+      "entropy": 0.8920881748199463,
+      "epoch": 1.3341463414634147,
+      "grad_norm": 0.544121503829956,
+      "learning_rate": 3.712737127371274e-06,
+      "loss": 0.8766,
+      "mean_token_accuracy": 0.7277756929397583,
+      "num_tokens": 7510233.0,
+      "step": 547
+    },
+    {
+      "entropy": 0.8978779315948486,
+      "epoch": 1.3365853658536586,
+      "grad_norm": 0.6094754934310913,
+      "learning_rate": 3.699186991869919e-06,
+      "loss": 0.8765,
+      "mean_token_accuracy": 0.725739598274231,
+      "num_tokens": 7524006.0,
+      "step": 548
+    },
+    {
+      "entropy": 0.8959704637527466,
+      "epoch": 1.3390243902439025,
+      "grad_norm": 0.484652042388916,
+      "learning_rate": 3.685636856368564e-06,
+      "loss": 0.8906,
+      "mean_token_accuracy": 0.7273457050323486,
+      "num_tokens": 7537728.0,
+      "step": 549
+    },
+    {
+      "entropy": 0.8667632341384888,
+      "epoch": 1.3414634146341464,
+      "grad_norm": 0.5026461482048035,
+      "learning_rate": 3.672086720867209e-06,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7339208722114563,
+      "num_tokens": 7551473.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.8890734910964966,
+      "epoch": 1.34390243902439,
+      "grad_norm": 0.5242050886154175,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 0.8675,
+      "mean_token_accuracy": 0.7381078600883484,
+      "num_tokens": 7564712.0,
+      "step": 551
+    },
+    {
+      "entropy": 0.8986338376998901,
+      "epoch": 1.346341463414634,
+      "grad_norm": 0.5245202779769897,
+      "learning_rate": 3.644986449864499e-06,
+      "loss": 0.8886,
+      "mean_token_accuracy": 0.7265607118606567,
+      "num_tokens": 7578007.0,
+      "step": 552
+    },
+    {
+      "entropy": 0.9158647656440735,
+      "epoch": 1.348780487804878,
+      "grad_norm": 0.5433186888694763,
+      "learning_rate": 3.6314363143631437e-06,
+      "loss": 0.9039,
+      "mean_token_accuracy": 0.7216664552688599,
+      "num_tokens": 7592017.0,
+      "step": 553
+    },
+    {
+      "entropy": 0.876274585723877,
+      "epoch": 1.3512195121951218,
+      "grad_norm": 0.5984988808631897,
+      "learning_rate": 3.6178861788617893e-06,
+      "loss": 0.8645,
+      "mean_token_accuracy": 0.732636570930481,
+      "num_tokens": 7605524.0,
+      "step": 554
+    },
+    {
+      "entropy": 0.8815956711769104,
+      "epoch": 1.3536585365853657,
+      "grad_norm": 0.5312358140945435,
+      "learning_rate": 3.604336043360434e-06,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.7308400273323059,
+      "num_tokens": 7619123.0,
+      "step": 555
+    },
+    {
+      "entropy": 0.8996585607528687,
+      "epoch": 1.3560975609756096,
+      "grad_norm": 0.5443177819252014,
+      "learning_rate": 3.590785907859079e-06,
+      "loss": 0.8877,
+      "mean_token_accuracy": 0.727543830871582,
+      "num_tokens": 7632888.0,
+      "step": 556
+    },
+    {
+      "entropy": 0.9065992832183838,
+      "epoch": 1.3585365853658535,
+      "grad_norm": 0.5264752507209778,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.8993,
+      "mean_token_accuracy": 0.7188636064529419,
+      "num_tokens": 7646385.0,
+      "step": 557
+    },
+    {
+      "entropy": 0.8936566114425659,
+      "epoch": 1.3609756097560974,
+      "grad_norm": 0.5454937815666199,
+      "learning_rate": 3.563685636856369e-06,
+      "loss": 0.8808,
+      "mean_token_accuracy": 0.7289999127388,
+      "num_tokens": 7659770.0,
+      "step": 558
+    },
+    {
+      "entropy": 0.8959879875183105,
+      "epoch": 1.3634146341463413,
+      "grad_norm": 0.4646812379360199,
+      "learning_rate": 3.5501355013550136e-06,
+      "loss": 0.8847,
+      "mean_token_accuracy": 0.7275198698043823,
+      "num_tokens": 7673765.0,
+      "step": 559
+    },
+    {
+      "entropy": 0.8788614273071289,
+      "epoch": 1.3658536585365852,
+      "grad_norm": 0.5636532306671143,
+      "learning_rate": 3.5365853658536588e-06,
+      "loss": 0.8731,
+      "mean_token_accuracy": 0.7269564270973206,
+      "num_tokens": 7687863.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8784700632095337,
+      "epoch": 1.3682926829268292,
+      "grad_norm": 0.4874018728733063,
+      "learning_rate": 3.5230352303523035e-06,
+      "loss": 0.8777,
+      "mean_token_accuracy": 0.7319697737693787,
+      "num_tokens": 7701911.0,
+      "step": 561
+    },
+    {
+      "entropy": 0.8921349048614502,
+      "epoch": 1.370731707317073,
+      "grad_norm": 0.5778629779815674,
+      "learning_rate": 3.509485094850949e-06,
+      "loss": 0.88,
+      "mean_token_accuracy": 0.7272459864616394,
+      "num_tokens": 7715518.0,
+      "step": 562
+    },
+    {
+      "entropy": 0.8773680329322815,
+      "epoch": 1.373170731707317,
+      "grad_norm": 0.5318922996520996,
+      "learning_rate": 3.495934959349594e-06,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.7313464879989624,
+      "num_tokens": 7729392.0,
+      "step": 563
+    },
+    {
+      "entropy": 0.8853249549865723,
+      "epoch": 1.3756097560975609,
+      "grad_norm": 0.5359835624694824,
+      "learning_rate": 3.4823848238482387e-06,
+      "loss": 0.8815,
+      "mean_token_accuracy": 0.7314780354499817,
+      "num_tokens": 7742919.0,
+      "step": 564
+    },
+    {
+      "entropy": 0.886237621307373,
+      "epoch": 1.3780487804878048,
+      "grad_norm": 0.5415591597557068,
+      "learning_rate": 3.468834688346884e-06,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7315119504928589,
+      "num_tokens": 7756295.0,
+      "step": 565
+    },
+    {
+      "entropy": 0.8944040536880493,
+      "epoch": 1.3804878048780487,
+      "grad_norm": 0.5151652097702026,
+      "learning_rate": 3.4552845528455287e-06,
+      "loss": 0.9041,
+      "mean_token_accuracy": 0.7207305431365967,
+      "num_tokens": 7770054.0,
+      "step": 566
+    },
+    {
+      "entropy": 0.8833385705947876,
+      "epoch": 1.3829268292682926,
+      "grad_norm": 0.4992096722126007,
+      "learning_rate": 3.4417344173441734e-06,
+      "loss": 0.8721,
+      "mean_token_accuracy": 0.7270156741142273,
+      "num_tokens": 7783862.0,
+      "step": 567
+    },
+    {
+      "entropy": 0.9013800621032715,
+      "epoch": 1.3853658536585365,
+      "grad_norm": 0.51569664478302,
+      "learning_rate": 3.4281842818428186e-06,
+      "loss": 0.9019,
+      "mean_token_accuracy": 0.7236117124557495,
+      "num_tokens": 7797312.0,
+      "step": 568
+    },
+    {
+      "entropy": 0.8924955725669861,
+      "epoch": 1.3878048780487804,
+      "grad_norm": 0.5286757349967957,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 0.9005,
+      "mean_token_accuracy": 0.7262544631958008,
+      "num_tokens": 7810720.0,
+      "step": 569
+    },
+    {
+      "entropy": 0.8890297412872314,
+      "epoch": 1.3902439024390243,
+      "grad_norm": 0.5955765247344971,
+      "learning_rate": 3.401084010840109e-06,
+      "loss": 0.8912,
+      "mean_token_accuracy": 0.7231224775314331,
+      "num_tokens": 7824251.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8626318573951721,
+      "epoch": 1.3926829268292682,
+      "grad_norm": 0.4987097680568695,
+      "learning_rate": 3.387533875338754e-06,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7393428087234497,
+      "num_tokens": 7837779.0,
+      "step": 571
+    },
+    {
+      "entropy": 0.8848908543586731,
+      "epoch": 1.395121951219512,
+      "grad_norm": 0.4759664535522461,
+      "learning_rate": 3.3739837398373986e-06,
+      "loss": 0.8813,
+      "mean_token_accuracy": 0.7298733592033386,
+      "num_tokens": 7851533.0,
+      "step": 572
+    },
+    {
+      "entropy": 0.8894503116607666,
+      "epoch": 1.397560975609756,
+      "grad_norm": 0.4857938587665558,
+      "learning_rate": 3.3604336043360438e-06,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7303609251976013,
+      "num_tokens": 7865208.0,
+      "step": 573
+    },
+    {
+      "entropy": 0.9084067344665527,
+      "epoch": 1.4,
+      "grad_norm": 0.5191188454627991,
+      "learning_rate": 3.3468834688346886e-06,
+      "loss": 0.891,
+      "mean_token_accuracy": 0.7281098961830139,
+      "num_tokens": 7878689.0,
+      "step": 574
+    },
+    {
+      "entropy": 0.8691054582595825,
+      "epoch": 1.4024390243902438,
+      "grad_norm": 0.49921202659606934,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.7362488508224487,
+      "num_tokens": 7892013.0,
+      "step": 575
+    },
+    {
+      "entropy": 0.8847055435180664,
+      "epoch": 1.4048780487804877,
+      "grad_norm": 0.4999859035015106,
+      "learning_rate": 3.3197831978319785e-06,
+      "loss": 0.8755,
+      "mean_token_accuracy": 0.7334121465682983,
+      "num_tokens": 7905563.0,
+      "step": 576
+    },
+    {
+      "entropy": 0.9092885255813599,
+      "epoch": 1.4073170731707316,
+      "grad_norm": 0.5342906713485718,
+      "learning_rate": 3.3062330623306237e-06,
+      "loss": 0.9235,
+      "mean_token_accuracy": 0.7145213484764099,
+      "num_tokens": 7919524.0,
+      "step": 577
+    },
+    {
+      "entropy": 0.8825151920318604,
+      "epoch": 1.4097560975609755,
+      "grad_norm": 0.5005236268043518,
+      "learning_rate": 3.292682926829269e-06,
+      "loss": 0.8714,
+      "mean_token_accuracy": 0.7287511825561523,
+      "num_tokens": 7933129.0,
+      "step": 578
+    },
+    {
+      "entropy": 0.8973416686058044,
+      "epoch": 1.4121951219512194,
+      "grad_norm": 0.5246015191078186,
+      "learning_rate": 3.2791327913279137e-06,
+      "loss": 0.8874,
+      "mean_token_accuracy": 0.7312119603157043,
+      "num_tokens": 7946784.0,
+      "step": 579
+    },
+    {
+      "entropy": 0.8993310928344727,
+      "epoch": 1.4146341463414633,
+      "grad_norm": 0.5140748023986816,
+      "learning_rate": 3.2655826558265585e-06,
+      "loss": 0.9069,
+      "mean_token_accuracy": 0.7212364673614502,
+      "num_tokens": 7961002.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8946079015731812,
+      "epoch": 1.4170731707317072,
+      "grad_norm": 0.6002617478370667,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.88,
+      "mean_token_accuracy": 0.7245283126831055,
+      "num_tokens": 7974798.0,
+      "step": 581
+    },
+    {
+      "entropy": 0.8892184495925903,
+      "epoch": 1.4195121951219511,
+      "grad_norm": 0.5291239619255066,
+      "learning_rate": 3.2384823848238484e-06,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7324248552322388,
+      "num_tokens": 7988683.0,
+      "step": 582
+    },
+    {
+      "entropy": 0.8774310946464539,
+      "epoch": 1.421951219512195,
+      "grad_norm": 0.5266504287719727,
+      "learning_rate": 3.224932249322493e-06,
+      "loss": 0.8667,
+      "mean_token_accuracy": 0.7288246750831604,
+      "num_tokens": 8002406.0,
+      "step": 583
+    },
+    {
+      "entropy": 0.8831679821014404,
+      "epoch": 1.424390243902439,
+      "grad_norm": 0.5562174320220947,
+      "learning_rate": 3.211382113821139e-06,
+      "loss": 0.8765,
+      "mean_token_accuracy": 0.7314850091934204,
+      "num_tokens": 8015479.0,
+      "step": 584
+    },
+    {
+      "entropy": 0.9056977033615112,
+      "epoch": 1.4268292682926829,
+      "grad_norm": 0.5713416934013367,
+      "learning_rate": 3.1978319783197836e-06,
+      "loss": 0.8898,
+      "mean_token_accuracy": 0.7224880456924438,
+      "num_tokens": 8029080.0,
+      "step": 585
+    },
+    {
+      "entropy": 0.8931180238723755,
+      "epoch": 1.4292682926829268,
+      "grad_norm": 0.503696858882904,
+      "learning_rate": 3.1842818428184284e-06,
+      "loss": 0.8832,
+      "mean_token_accuracy": 0.7255293726921082,
+      "num_tokens": 8042602.0,
+      "step": 586
+    },
+    {
+      "entropy": 0.8890825510025024,
+      "epoch": 1.4317073170731707,
+      "grad_norm": 0.5064833760261536,
+      "learning_rate": 3.1707317073170736e-06,
+      "loss": 0.8811,
+      "mean_token_accuracy": 0.7287392020225525,
+      "num_tokens": 8056070.0,
+      "step": 587
+    },
+    {
+      "entropy": 0.8768372535705566,
+      "epoch": 1.4341463414634146,
+      "grad_norm": 0.5217829346656799,
+      "learning_rate": 3.1571815718157183e-06,
+      "loss": 0.856,
+      "mean_token_accuracy": 0.7340074777603149,
+      "num_tokens": 8069733.0,
+      "step": 588
+    },
+    {
+      "entropy": 0.8988307118415833,
+      "epoch": 1.4365853658536585,
+      "grad_norm": 0.5267525315284729,
+      "learning_rate": 3.143631436314363e-06,
+      "loss": 0.8844,
+      "mean_token_accuracy": 0.7272128462791443,
+      "num_tokens": 8083408.0,
+      "step": 589
+    },
+    {
+      "entropy": 0.8802148103713989,
+      "epoch": 1.4390243902439024,
+      "grad_norm": 0.49351492524147034,
+      "learning_rate": 3.1300813008130083e-06,
+      "loss": 0.8681,
+      "mean_token_accuracy": 0.7272524833679199,
+      "num_tokens": 8096887.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8733420372009277,
+      "epoch": 1.4414634146341463,
+      "grad_norm": 0.6216485500335693,
+      "learning_rate": 3.116531165311653e-06,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.7351359128952026,
+      "num_tokens": 8110476.0,
+      "step": 591
+    },
+    {
+      "entropy": 0.8928675651550293,
+      "epoch": 1.4439024390243902,
+      "grad_norm": 0.5383737683296204,
+      "learning_rate": 3.1029810298102987e-06,
+      "loss": 0.9043,
+      "mean_token_accuracy": 0.7218061089515686,
+      "num_tokens": 8124511.0,
+      "step": 592
+    },
+    {
+      "entropy": 0.886874794960022,
+      "epoch": 1.446341463414634,
+      "grad_norm": 0.5209896564483643,
+      "learning_rate": 3.0894308943089435e-06,
+      "loss": 0.8773,
+      "mean_token_accuracy": 0.7260102033615112,
+      "num_tokens": 8138064.0,
+      "step": 593
+    },
+    {
+      "entropy": 0.8790051937103271,
+      "epoch": 1.448780487804878,
+      "grad_norm": 0.5296643376350403,
+      "learning_rate": 3.0758807588075882e-06,
+      "loss": 0.8836,
+      "mean_token_accuracy": 0.729468584060669,
+      "num_tokens": 8151949.0,
+      "step": 594
+    },
+    {
+      "entropy": 0.88617342710495,
+      "epoch": 1.451219512195122,
+      "grad_norm": 0.5086992383003235,
+      "learning_rate": 3.0623306233062334e-06,
+      "loss": 0.8673,
+      "mean_token_accuracy": 0.7311142683029175,
+      "num_tokens": 8165454.0,
+      "step": 595
+    },
+    {
+      "entropy": 0.8895916938781738,
+      "epoch": 1.4536585365853658,
+      "grad_norm": 0.6400481462478638,
+      "learning_rate": 3.0487804878048782e-06,
+      "loss": 0.8791,
+      "mean_token_accuracy": 0.7254001498222351,
+      "num_tokens": 8179403.0,
+      "step": 596
+    },
+    {
+      "entropy": 0.8727394938468933,
+      "epoch": 1.4560975609756097,
+      "grad_norm": 0.5148288011550903,
+      "learning_rate": 3.035230352303523e-06,
+      "loss": 0.8641,
+      "mean_token_accuracy": 0.7360201478004456,
+      "num_tokens": 8193314.0,
+      "step": 597
+    },
+    {
+      "entropy": 0.8800617456436157,
+      "epoch": 1.4585365853658536,
+      "grad_norm": 0.555195152759552,
+      "learning_rate": 3.021680216802168e-06,
+      "loss": 0.8833,
+      "mean_token_accuracy": 0.7324080467224121,
+      "num_tokens": 8206731.0,
+      "step": 598
+    },
+    {
+      "entropy": 0.8606910705566406,
+      "epoch": 1.4609756097560975,
+      "grad_norm": 0.5864395499229431,
+      "learning_rate": 3.0081300813008134e-06,
+      "loss": 0.8496,
+      "mean_token_accuracy": 0.7344096899032593,
+      "num_tokens": 8220249.0,
+      "step": 599
+    },
+    {
+      "entropy": 0.8760560154914856,
+      "epoch": 1.4634146341463414,
+      "grad_norm": 0.5235090255737305,
+      "learning_rate": 2.9945799457994586e-06,
+      "loss": 0.8878,
+      "mean_token_accuracy": 0.7239957451820374,
+      "num_tokens": 8234330.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8802227973937988,
+      "epoch": 1.4658536585365853,
+      "grad_norm": 0.522666871547699,
+      "learning_rate": 2.9810298102981034e-06,
+      "loss": 0.8898,
+      "mean_token_accuracy": 0.7270638942718506,
+      "num_tokens": 8248276.0,
+      "step": 601
+    },
+    {
+      "entropy": 0.8891713619232178,
+      "epoch": 1.4682926829268292,
+      "grad_norm": 0.4861930012702942,
+      "learning_rate": 2.967479674796748e-06,
+      "loss": 0.9017,
+      "mean_token_accuracy": 0.721794843673706,
+      "num_tokens": 8262332.0,
+      "step": 602
+    },
+    {
+      "entropy": 0.8843443393707275,
+      "epoch": 1.4707317073170731,
+      "grad_norm": 0.5338602662086487,
+      "learning_rate": 2.9539295392953933e-06,
+      "loss": 0.8758,
+      "mean_token_accuracy": 0.7304834127426147,
+      "num_tokens": 8275939.0,
+      "step": 603
+    },
+    {
+      "entropy": 0.8781300783157349,
+      "epoch": 1.473170731707317,
+      "grad_norm": 0.5555716156959534,
+      "learning_rate": 2.940379403794038e-06,
+      "loss": 0.8616,
+      "mean_token_accuracy": 0.729243814945221,
+      "num_tokens": 8289192.0,
+      "step": 604
+    },
+    {
+      "entropy": 0.8766529560089111,
+      "epoch": 1.475609756097561,
+      "grad_norm": 0.6152108907699585,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.8778,
+      "mean_token_accuracy": 0.726624071598053,
+      "num_tokens": 8302662.0,
+      "step": 605
+    },
+    {
+      "entropy": 0.9065438508987427,
+      "epoch": 1.4780487804878049,
+      "grad_norm": 0.4771234095096588,
+      "learning_rate": 2.913279132791328e-06,
+      "loss": 0.8874,
+      "mean_token_accuracy": 0.7299425601959229,
+      "num_tokens": 8316775.0,
+      "step": 606
+    },
+    {
+      "entropy": 0.8783807754516602,
+      "epoch": 1.4804878048780488,
+      "grad_norm": 0.5496251583099365,
+      "learning_rate": 2.8997289972899733e-06,
+      "loss": 0.8738,
+      "mean_token_accuracy": 0.730849027633667,
+      "num_tokens": 8330289.0,
+      "step": 607
+    },
+    {
+      "entropy": 0.905014157295227,
+      "epoch": 1.4829268292682927,
+      "grad_norm": 0.5509771704673767,
+      "learning_rate": 2.8861788617886185e-06,
+      "loss": 0.8763,
+      "mean_token_accuracy": 0.7334033250808716,
+      "num_tokens": 8343636.0,
+      "step": 608
+    },
+    {
+      "entropy": 0.875667929649353,
+      "epoch": 1.4853658536585366,
+      "grad_norm": 0.5156928896903992,
+      "learning_rate": 2.8726287262872632e-06,
+      "loss": 0.8588,
+      "mean_token_accuracy": 0.7322340607643127,
+      "num_tokens": 8357358.0,
+      "step": 609
+    },
+    {
+      "entropy": 0.892330527305603,
+      "epoch": 1.4878048780487805,
+      "grad_norm": 0.5027339458465576,
+      "learning_rate": 2.859078590785908e-06,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.7313015460968018,
+      "num_tokens": 8370958.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8820470571517944,
+      "epoch": 1.4902439024390244,
+      "grad_norm": 0.48004716634750366,
+      "learning_rate": 2.845528455284553e-06,
+      "loss": 0.8773,
+      "mean_token_accuracy": 0.7273856401443481,
+      "num_tokens": 8384660.0,
+      "step": 611
+    },
+    {
+      "entropy": 0.8868808746337891,
+      "epoch": 1.4926829268292683,
+      "grad_norm": 0.4970702528953552,
+      "learning_rate": 2.831978319783198e-06,
+      "loss": 0.8651,
+      "mean_token_accuracy": 0.7320648431777954,
+      "num_tokens": 8398183.0,
+      "step": 612
+    },
+    {
+      "entropy": 0.8882201910018921,
+      "epoch": 1.4951219512195122,
+      "grad_norm": 0.5164826512336731,
+      "learning_rate": 2.8184281842818427e-06,
+      "loss": 0.885,
+      "mean_token_accuracy": 0.7296938300132751,
+      "num_tokens": 8411754.0,
+      "step": 613
+    },
+    {
+      "entropy": 0.8878587484359741,
+      "epoch": 1.497560975609756,
+      "grad_norm": 0.5206930637359619,
+      "learning_rate": 2.8048780487804884e-06,
+      "loss": 0.868,
+      "mean_token_accuracy": 0.7336882948875427,
+      "num_tokens": 8425104.0,
+      "step": 614
+    },
+    {
+      "entropy": 0.8786503076553345,
+      "epoch": 1.5,
+      "grad_norm": 0.4896388053894043,
+      "learning_rate": 2.791327913279133e-06,
+      "loss": 0.8487,
+      "mean_token_accuracy": 0.7345880270004272,
+      "num_tokens": 8438616.0,
+      "step": 615
+    },
+    {
+      "entropy": 0.8783841729164124,
+      "epoch": 1.502439024390244,
+      "grad_norm": 0.5044620037078857,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 0.8711,
+      "mean_token_accuracy": 0.7318198680877686,
+      "num_tokens": 8452067.0,
+      "step": 616
+    },
+    {
+      "entropy": 0.8775740265846252,
+      "epoch": 1.5048780487804878,
+      "grad_norm": 0.5294600129127502,
+      "learning_rate": 2.764227642276423e-06,
+      "loss": 0.8636,
+      "mean_token_accuracy": 0.7268987894058228,
+      "num_tokens": 8466184.0,
+      "step": 617
+    },
+    {
+      "entropy": 0.8924665451049805,
+      "epoch": 1.5073170731707317,
+      "grad_norm": 0.6135087609291077,
+      "learning_rate": 2.750677506775068e-06,
+      "loss": 0.9149,
+      "mean_token_accuracy": 0.7222589254379272,
+      "num_tokens": 8479817.0,
+      "step": 618
+    },
+    {
+      "entropy": 0.8739150762557983,
+      "epoch": 1.5097560975609756,
+      "grad_norm": 0.4934721887111664,
+      "learning_rate": 2.7371273712737127e-06,
+      "loss": 0.8695,
+      "mean_token_accuracy": 0.730917751789093,
+      "num_tokens": 8493301.0,
+      "step": 619
+    },
+    {
+      "entropy": 0.8825638890266418,
+      "epoch": 1.5121951219512195,
+      "grad_norm": 0.4939766526222229,
+      "learning_rate": 2.723577235772358e-06,
+      "loss": 0.8731,
+      "mean_token_accuracy": 0.730536699295044,
+      "num_tokens": 8506881.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8853427767753601,
+      "epoch": 1.5146341463414634,
+      "grad_norm": 0.5113255977630615,
+      "learning_rate": 2.7100271002710026e-06,
+      "loss": 0.8848,
+      "mean_token_accuracy": 0.7268093824386597,
+      "num_tokens": 8520631.0,
+      "step": 621
+    },
+    {
+      "entropy": 0.882167398929596,
+      "epoch": 1.5170731707317073,
+      "grad_norm": 0.5379652976989746,
+      "learning_rate": 2.6964769647696482e-06,
+      "loss": 0.8778,
+      "mean_token_accuracy": 0.7336709499359131,
+      "num_tokens": 8534273.0,
+      "step": 622
+    },
+    {
+      "entropy": 0.8745768070220947,
+      "epoch": 1.5195121951219512,
+      "grad_norm": 0.47744041681289673,
+      "learning_rate": 2.682926829268293e-06,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.7373600602149963,
+      "num_tokens": 8547600.0,
+      "step": 623
+    },
+    {
+      "entropy": 0.8744478225708008,
+      "epoch": 1.5219512195121951,
+      "grad_norm": 0.48806169629096985,
+      "learning_rate": 2.669376693766938e-06,
+      "loss": 0.8473,
+      "mean_token_accuracy": 0.7403261661529541,
+      "num_tokens": 8561106.0,
+      "step": 624
+    },
+    {
+      "entropy": 0.8737915754318237,
+      "epoch": 1.524390243902439,
+      "grad_norm": 0.5594715476036072,
+      "learning_rate": 2.655826558265583e-06,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7284777760505676,
+      "num_tokens": 8575003.0,
+      "step": 625
+    },
+    {
+      "entropy": 0.8842061161994934,
+      "epoch": 1.526829268292683,
+      "grad_norm": 0.5108963251113892,
+      "learning_rate": 2.6422764227642278e-06,
+      "loss": 0.8712,
+      "mean_token_accuracy": 0.7290242910385132,
+      "num_tokens": 8588773.0,
+      "step": 626
+    },
+    {
+      "entropy": 0.8844512701034546,
+      "epoch": 1.5292682926829269,
+      "grad_norm": 0.5226480960845947,
+      "learning_rate": 2.6287262872628725e-06,
+      "loss": 0.8935,
+      "mean_token_accuracy": 0.7259301543235779,
+      "num_tokens": 8602738.0,
+      "step": 627
+    },
+    {
+      "entropy": 0.8773095011711121,
+      "epoch": 1.5317073170731708,
+      "grad_norm": 0.530270516872406,
+      "learning_rate": 2.6151761517615177e-06,
+      "loss": 0.8718,
+      "mean_token_accuracy": 0.7299329042434692,
+      "num_tokens": 8616321.0,
+      "step": 628
+    },
+    {
+      "entropy": 0.883411169052124,
+      "epoch": 1.5341463414634147,
+      "grad_norm": 0.5226531624794006,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 0.8696,
+      "mean_token_accuracy": 0.7328861951828003,
+      "num_tokens": 8630054.0,
+      "step": 629
+    },
+    {
+      "entropy": 0.8826172947883606,
+      "epoch": 1.5365853658536586,
+      "grad_norm": 0.5279987454414368,
+      "learning_rate": 2.588075880758808e-06,
+      "loss": 0.8648,
+      "mean_token_accuracy": 0.73455411195755,
+      "num_tokens": 8643941.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8816993236541748,
+      "epoch": 1.5390243902439025,
+      "grad_norm": 0.4635319113731384,
+      "learning_rate": 2.574525745257453e-06,
+      "loss": 0.8719,
+      "mean_token_accuracy": 0.7324573993682861,
+      "num_tokens": 8658037.0,
+      "step": 631
+    },
+    {
+      "entropy": 0.8873631954193115,
+      "epoch": 1.5414634146341464,
+      "grad_norm": 0.5214371681213379,
+      "learning_rate": 2.5609756097560977e-06,
+      "loss": 0.8807,
+      "mean_token_accuracy": 0.7238868474960327,
+      "num_tokens": 8672337.0,
+      "step": 632
+    },
+    {
+      "entropy": 0.8695685863494873,
+      "epoch": 1.5439024390243903,
+      "grad_norm": 0.522845447063446,
+      "learning_rate": 2.547425474254743e-06,
+      "loss": 0.8614,
+      "mean_token_accuracy": 0.7336251139640808,
+      "num_tokens": 8686063.0,
+      "step": 633
+    },
+    {
+      "entropy": 0.8817265033721924,
+      "epoch": 1.5463414634146342,
+      "grad_norm": 0.5790318250656128,
+      "learning_rate": 2.5338753387533876e-06,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.735949695110321,
+      "num_tokens": 8699762.0,
+      "step": 634
+    },
+    {
+      "entropy": 0.8943449258804321,
+      "epoch": 1.548780487804878,
+      "grad_norm": 0.5461540818214417,
+      "learning_rate": 2.5203252032520324e-06,
+      "loss": 0.8953,
+      "mean_token_accuracy": 0.7255430817604065,
+      "num_tokens": 8713128.0,
+      "step": 635
+    },
+    {
+      "entropy": 0.87168288230896,
+      "epoch": 1.551219512195122,
+      "grad_norm": 0.50957190990448,
+      "learning_rate": 2.5067750677506776e-06,
+      "loss": 0.8859,
+      "mean_token_accuracy": 0.7287718653678894,
+      "num_tokens": 8726970.0,
+      "step": 636
+    },
+    {
+      "entropy": 0.8847887516021729,
+      "epoch": 1.553658536585366,
+      "grad_norm": 0.4929066002368927,
+      "learning_rate": 2.493224932249323e-06,
+      "loss": 0.8804,
+      "mean_token_accuracy": 0.7283905148506165,
+      "num_tokens": 8740649.0,
+      "step": 637
+    },
+    {
+      "entropy": 0.8921957015991211,
+      "epoch": 1.5560975609756098,
+      "grad_norm": 0.48889023065567017,
+      "learning_rate": 2.4796747967479676e-06,
+      "loss": 0.8764,
+      "mean_token_accuracy": 0.726151168346405,
+      "num_tokens": 8754607.0,
+      "step": 638
+    },
+    {
+      "entropy": 0.8531458973884583,
+      "epoch": 1.5585365853658537,
+      "grad_norm": 0.5207129716873169,
+      "learning_rate": 2.4661246612466128e-06,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7364062666893005,
+      "num_tokens": 8768269.0,
+      "step": 639
+    },
+    {
+      "entropy": 0.8851905465126038,
+      "epoch": 1.5609756097560976,
+      "grad_norm": 0.5498895049095154,
+      "learning_rate": 2.4525745257452575e-06,
+      "loss": 0.8783,
+      "mean_token_accuracy": 0.7295770645141602,
+      "num_tokens": 8782093.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8695498704910278,
+      "epoch": 1.5634146341463415,
+      "grad_norm": 0.4955178499221802,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 0.8765,
+      "mean_token_accuracy": 0.7328490614891052,
+      "num_tokens": 8795738.0,
+      "step": 641
+    },
+    {
+      "entropy": 0.8819414973258972,
+      "epoch": 1.5658536585365854,
+      "grad_norm": 0.5156698822975159,
+      "learning_rate": 2.4254742547425475e-06,
+      "loss": 0.8585,
+      "mean_token_accuracy": 0.7310495376586914,
+      "num_tokens": 8809474.0,
+      "step": 642
+    },
+    {
+      "entropy": 0.8857159614562988,
+      "epoch": 1.5682926829268293,
+      "grad_norm": 0.5351740121841431,
+      "learning_rate": 2.4119241192411927e-06,
+      "loss": 0.8919,
+      "mean_token_accuracy": 0.7224662899971008,
+      "num_tokens": 8823146.0,
+      "step": 643
+    },
+    {
+      "entropy": 0.8779886364936829,
+      "epoch": 1.5707317073170732,
+      "grad_norm": 0.5071956515312195,
+      "learning_rate": 2.3983739837398375e-06,
+      "loss": 0.8734,
+      "mean_token_accuracy": 0.7339435815811157,
+      "num_tokens": 8836708.0,
+      "step": 644
+    },
+    {
+      "entropy": 0.9025756120681763,
+      "epoch": 1.5731707317073171,
+      "grad_norm": 0.5226132273674011,
+      "learning_rate": 2.3848238482384827e-06,
+      "loss": 0.8932,
+      "mean_token_accuracy": 0.7239575386047363,
+      "num_tokens": 8850298.0,
+      "step": 645
+    },
+    {
+      "entropy": 0.8937682509422302,
+      "epoch": 1.575609756097561,
+      "grad_norm": 0.5704896450042725,
+      "learning_rate": 2.371273712737128e-06,
+      "loss": 0.8891,
+      "mean_token_accuracy": 0.721930980682373,
+      "num_tokens": 8864048.0,
+      "step": 646
+    },
+    {
+      "entropy": 0.8963336944580078,
+      "epoch": 1.578048780487805,
+      "grad_norm": 0.5066708326339722,
+      "learning_rate": 2.3577235772357727e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7227367162704468,
+      "num_tokens": 8877993.0,
+      "step": 647
+    },
+    {
+      "entropy": 0.8906478881835938,
+      "epoch": 1.5804878048780489,
+      "grad_norm": 0.47832056879997253,
+      "learning_rate": 2.3441734417344174e-06,
+      "loss": 0.8827,
+      "mean_token_accuracy": 0.728906512260437,
+      "num_tokens": 8891141.0,
+      "step": 648
+    },
+    {
+      "entropy": 0.8920251131057739,
+      "epoch": 1.5829268292682928,
+      "grad_norm": 0.5094121098518372,
+      "learning_rate": 2.3306233062330626e-06,
+      "loss": 0.8819,
+      "mean_token_accuracy": 0.7286123037338257,
+      "num_tokens": 8905137.0,
+      "step": 649
+    },
+    {
+      "entropy": 0.9012845158576965,
+      "epoch": 1.5853658536585367,
+      "grad_norm": 0.5223912000656128,
+      "learning_rate": 2.317073170731708e-06,
+      "loss": 0.8765,
+      "mean_token_accuracy": 0.7305706739425659,
+      "num_tokens": 8919101.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.8928289413452148,
+      "epoch": 1.5878048780487806,
+      "grad_norm": 0.5515636801719666,
+      "learning_rate": 2.3035230352303526e-06,
+      "loss": 0.8951,
+      "mean_token_accuracy": 0.724931538105011,
+      "num_tokens": 8932630.0,
+      "step": 651
+    },
+    {
+      "entropy": 0.8886030912399292,
+      "epoch": 1.5902439024390245,
+      "grad_norm": 0.49004659056663513,
+      "learning_rate": 2.2899728997289974e-06,
+      "loss": 0.8717,
+      "mean_token_accuracy": 0.7325581312179565,
+      "num_tokens": 8946406.0,
+      "step": 652
+    },
+    {
+      "entropy": 0.8970048427581787,
+      "epoch": 1.5926829268292684,
+      "grad_norm": 0.5707719326019287,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 0.8839,
+      "mean_token_accuracy": 0.725605845451355,
+      "num_tokens": 8960493.0,
+      "step": 653
+    },
+    {
+      "entropy": 0.8691211938858032,
+      "epoch": 1.5951219512195123,
+      "grad_norm": 0.5564479827880859,
+      "learning_rate": 2.2628726287262878e-06,
+      "loss": 0.8485,
+      "mean_token_accuracy": 0.7343336343765259,
+      "num_tokens": 8974105.0,
+      "step": 654
+    },
+    {
+      "entropy": 0.8998392224311829,
+      "epoch": 1.5975609756097562,
+      "grad_norm": 0.5155988931655884,
+      "learning_rate": 2.2493224932249325e-06,
+      "loss": 0.9102,
+      "mean_token_accuracy": 0.7190064787864685,
+      "num_tokens": 8988011.0,
+      "step": 655
+    },
+    {
+      "entropy": 0.8792390823364258,
+      "epoch": 1.6,
+      "grad_norm": 0.548163115978241,
+      "learning_rate": 2.2357723577235773e-06,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7319982051849365,
+      "num_tokens": 9001359.0,
+      "step": 656
+    },
+    {
+      "entropy": 0.9006599187850952,
+      "epoch": 1.602439024390244,
+      "grad_norm": 0.5012010931968689,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 0.8916,
+      "mean_token_accuracy": 0.7241604924201965,
+      "num_tokens": 9015133.0,
+      "step": 657
+    },
+    {
+      "entropy": 0.8827446699142456,
+      "epoch": 1.604878048780488,
+      "grad_norm": 0.562204122543335,
+      "learning_rate": 2.2086720867208673e-06,
+      "loss": 0.879,
+      "mean_token_accuracy": 0.7299302220344543,
+      "num_tokens": 9028764.0,
+      "step": 658
+    },
+    {
+      "entropy": 0.8815902471542358,
+      "epoch": 1.6073170731707318,
+      "grad_norm": 0.5246950387954712,
+      "learning_rate": 2.1951219512195125e-06,
+      "loss": 0.8775,
+      "mean_token_accuracy": 0.7348490357398987,
+      "num_tokens": 9042459.0,
+      "step": 659
+    },
+    {
+      "entropy": 0.9139642715454102,
+      "epoch": 1.6097560975609757,
+      "grad_norm": 0.5334653258323669,
+      "learning_rate": 2.1815718157181572e-06,
+      "loss": 0.9022,
+      "mean_token_accuracy": 0.724953293800354,
+      "num_tokens": 9056389.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8966798186302185,
+      "epoch": 1.6121951219512196,
+      "grad_norm": 0.5264739394187927,
+      "learning_rate": 2.1680216802168024e-06,
+      "loss": 0.8956,
+      "mean_token_accuracy": 0.726248562335968,
+      "num_tokens": 9070341.0,
+      "step": 661
+    },
+    {
+      "entropy": 0.8982292413711548,
+      "epoch": 1.6146341463414635,
+      "grad_norm": 0.5185431838035583,
+      "learning_rate": 2.154471544715447e-06,
+      "loss": 0.8945,
+      "mean_token_accuracy": 0.7259421944618225,
+      "num_tokens": 9084022.0,
+      "step": 662
+    },
+    {
+      "entropy": 0.9089407920837402,
+      "epoch": 1.6170731707317074,
+      "grad_norm": 0.500324547290802,
+      "learning_rate": 2.1409214092140924e-06,
+      "loss": 0.8895,
+      "mean_token_accuracy": 0.7267783284187317,
+      "num_tokens": 9097829.0,
+      "step": 663
+    },
+    {
+      "entropy": 0.8799450397491455,
+      "epoch": 1.6195121951219513,
+      "grad_norm": 0.5376121401786804,
+      "learning_rate": 2.127371273712737e-06,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.73411625623703,
+      "num_tokens": 9111129.0,
+      "step": 664
+    },
+    {
+      "entropy": 0.8976962566375732,
+      "epoch": 1.6219512195121952,
+      "grad_norm": 0.48157837986946106,
+      "learning_rate": 2.1138211382113824e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7249734997749329,
+      "num_tokens": 9125300.0,
+      "step": 665
+    },
+    {
+      "entropy": 0.8926389217376709,
+      "epoch": 1.6243902439024391,
+      "grad_norm": 0.5121070742607117,
+      "learning_rate": 2.100271002710027e-06,
+      "loss": 0.8901,
+      "mean_token_accuracy": 0.7253190279006958,
+      "num_tokens": 9138717.0,
+      "step": 666
+    },
+    {
+      "entropy": 0.8846919536590576,
+      "epoch": 1.626829268292683,
+      "grad_norm": 0.4743853807449341,
+      "learning_rate": 2.0867208672086723e-06,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7336145639419556,
+      "num_tokens": 9152007.0,
+      "step": 667
+    },
+    {
+      "entropy": 0.8994081020355225,
+      "epoch": 1.629268292682927,
+      "grad_norm": 0.46882563829421997,
+      "learning_rate": 2.073170731707317e-06,
+      "loss": 0.8859,
+      "mean_token_accuracy": 0.7267160415649414,
+      "num_tokens": 9165251.0,
+      "step": 668
+    },
+    {
+      "entropy": 0.8839161396026611,
+      "epoch": 1.6317073170731708,
+      "grad_norm": 0.5257059335708618,
+      "learning_rate": 2.0596205962059623e-06,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7352175712585449,
+      "num_tokens": 9178712.0,
+      "step": 669
+    },
+    {
+      "entropy": 0.8784297704696655,
+      "epoch": 1.6341463414634148,
+      "grad_norm": 0.519122838973999,
+      "learning_rate": 2.046070460704607e-06,
+      "loss": 0.8662,
+      "mean_token_accuracy": 0.7308939099311829,
+      "num_tokens": 9192611.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8900331258773804,
+      "epoch": 1.6365853658536587,
+      "grad_norm": 0.5005249977111816,
+      "learning_rate": 2.0325203252032523e-06,
+      "loss": 0.89,
+      "mean_token_accuracy": 0.7258846163749695,
+      "num_tokens": 9206249.0,
+      "step": 671
+    },
+    {
+      "entropy": 0.8765691518783569,
+      "epoch": 1.6390243902439026,
+      "grad_norm": 0.5093433856964111,
+      "learning_rate": 2.018970189701897e-06,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.7366489171981812,
+      "num_tokens": 9219616.0,
+      "step": 672
+    },
+    {
+      "entropy": 0.873883843421936,
+      "epoch": 1.6414634146341465,
+      "grad_norm": 0.5480417609214783,
+      "learning_rate": 2.0054200542005423e-06,
+      "loss": 0.8643,
+      "mean_token_accuracy": 0.7316689491271973,
+      "num_tokens": 9233611.0,
+      "step": 673
+    },
+    {
+      "entropy": 0.8792591094970703,
+      "epoch": 1.6439024390243904,
+      "grad_norm": 0.5551111698150635,
+      "learning_rate": 1.991869918699187e-06,
+      "loss": 0.869,
+      "mean_token_accuracy": 0.7298508286476135,
+      "num_tokens": 9247238.0,
+      "step": 674
+    },
+    {
+      "entropy": 0.8775755167007446,
+      "epoch": 1.6463414634146343,
+      "grad_norm": 0.5572032928466797,
+      "learning_rate": 1.9783197831978322e-06,
+      "loss": 0.8712,
+      "mean_token_accuracy": 0.7310796976089478,
+      "num_tokens": 9261128.0,
+      "step": 675
+    },
+    {
+      "entropy": 0.8921653032302856,
+      "epoch": 1.6487804878048782,
+      "grad_norm": 0.5530563592910767,
+      "learning_rate": 1.964769647696477e-06,
+      "loss": 0.8956,
+      "mean_token_accuracy": 0.7238720655441284,
+      "num_tokens": 9275152.0,
+      "step": 676
+    },
+    {
+      "entropy": 0.8693351745605469,
+      "epoch": 1.651219512195122,
+      "grad_norm": 0.5165068507194519,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 0.8687,
+      "mean_token_accuracy": 0.7313920855522156,
+      "num_tokens": 9288630.0,
+      "step": 677
+    },
+    {
+      "entropy": 0.8753268122673035,
+      "epoch": 1.653658536585366,
+      "grad_norm": 0.4899144470691681,
+      "learning_rate": 1.937669376693767e-06,
+      "loss": 0.8646,
+      "mean_token_accuracy": 0.7322003841400146,
+      "num_tokens": 9302298.0,
+      "step": 678
+    },
+    {
+      "entropy": 0.8960176706314087,
+      "epoch": 1.65609756097561,
+      "grad_norm": 0.5682179927825928,
+      "learning_rate": 1.924119241192412e-06,
+      "loss": 0.9055,
+      "mean_token_accuracy": 0.7215538620948792,
+      "num_tokens": 9316112.0,
+      "step": 679
+    },
+    {
+      "entropy": 0.875866174697876,
+      "epoch": 1.6585365853658538,
+      "grad_norm": 0.48258838057518005,
+      "learning_rate": 1.9105691056910574e-06,
+      "loss": 0.8567,
+      "mean_token_accuracy": 0.7342554330825806,
+      "num_tokens": 9329720.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8869270086288452,
+      "epoch": 1.6609756097560977,
+      "grad_norm": 0.5566992163658142,
+      "learning_rate": 1.8970189701897021e-06,
+      "loss": 0.8982,
+      "mean_token_accuracy": 0.7254682779312134,
+      "num_tokens": 9343137.0,
+      "step": 681
+    },
+    {
+      "entropy": 0.8802525997161865,
+      "epoch": 1.6634146341463416,
+      "grad_norm": 0.5332781076431274,
+      "learning_rate": 1.8834688346883471e-06,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.733722984790802,
+      "num_tokens": 9356669.0,
+      "step": 682
+    },
+    {
+      "entropy": 0.8921730518341064,
+      "epoch": 1.6658536585365855,
+      "grad_norm": 0.5774820446968079,
+      "learning_rate": 1.8699186991869919e-06,
+      "loss": 0.8853,
+      "mean_token_accuracy": 0.7285693883895874,
+      "num_tokens": 9370637.0,
+      "step": 683
+    },
+    {
+      "entropy": 0.8870189785957336,
+      "epoch": 1.6682926829268294,
+      "grad_norm": 0.5511258840560913,
+      "learning_rate": 1.856368563685637e-06,
+      "loss": 0.8722,
+      "mean_token_accuracy": 0.729683518409729,
+      "num_tokens": 9384681.0,
+      "step": 684
+    },
+    {
+      "entropy": 0.8922406435012817,
+      "epoch": 1.6707317073170733,
+      "grad_norm": 0.6266551613807678,
+      "learning_rate": 1.842818428184282e-06,
+      "loss": 0.8926,
+      "mean_token_accuracy": 0.7260046005249023,
+      "num_tokens": 9398533.0,
+      "step": 685
+    },
+    {
+      "entropy": 0.877333402633667,
+      "epoch": 1.6731707317073172,
+      "grad_norm": 0.5099342465400696,
+      "learning_rate": 1.8292682926829268e-06,
+      "loss": 0.8647,
+      "mean_token_accuracy": 0.7367627024650574,
+      "num_tokens": 9411807.0,
+      "step": 686
+    },
+    {
+      "entropy": 0.8942214250564575,
+      "epoch": 1.6756097560975611,
+      "grad_norm": 0.6306934356689453,
+      "learning_rate": 1.8157181571815718e-06,
+      "loss": 0.8553,
+      "mean_token_accuracy": 0.7317629456520081,
+      "num_tokens": 9425408.0,
+      "step": 687
+    },
+    {
+      "entropy": 0.892951250076294,
+      "epoch": 1.678048780487805,
+      "grad_norm": 0.5098548531532288,
+      "learning_rate": 1.802168021680217e-06,
+      "loss": 0.8804,
+      "mean_token_accuracy": 0.7273315191268921,
+      "num_tokens": 9439342.0,
+      "step": 688
+    },
+    {
+      "entropy": 0.8908568620681763,
+      "epoch": 1.680487804878049,
+      "grad_norm": 0.5078380703926086,
+      "learning_rate": 1.788617886178862e-06,
+      "loss": 0.8821,
+      "mean_token_accuracy": 0.7286223769187927,
+      "num_tokens": 9453099.0,
+      "step": 689
+    },
+    {
+      "entropy": 0.8855090737342834,
+      "epoch": 1.6829268292682928,
+      "grad_norm": 0.547004222869873,
+      "learning_rate": 1.7750677506775068e-06,
+      "loss": 0.8704,
+      "mean_token_accuracy": 0.7313100695610046,
+      "num_tokens": 9467053.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.906796932220459,
+      "epoch": 1.6853658536585368,
+      "grad_norm": 0.5479020476341248,
+      "learning_rate": 1.7615176151761518e-06,
+      "loss": 0.878,
+      "mean_token_accuracy": 0.7282412052154541,
+      "num_tokens": 9480868.0,
+      "step": 691
+    },
+    {
+      "entropy": 0.893460750579834,
+      "epoch": 1.6878048780487804,
+      "grad_norm": 0.5498223304748535,
+      "learning_rate": 1.747967479674797e-06,
+      "loss": 0.8838,
+      "mean_token_accuracy": 0.7282806038856506,
+      "num_tokens": 9495226.0,
+      "step": 692
+    },
+    {
+      "entropy": 0.88077712059021,
+      "epoch": 1.6902439024390243,
+      "grad_norm": 0.646657407283783,
+      "learning_rate": 1.734417344173442e-06,
+      "loss": 0.8828,
+      "mean_token_accuracy": 0.7274883389472961,
+      "num_tokens": 9508735.0,
+      "step": 693
+    },
+    {
+      "entropy": 0.8923465013504028,
+      "epoch": 1.6926829268292682,
+      "grad_norm": 0.6305309534072876,
+      "learning_rate": 1.7208672086720867e-06,
+      "loss": 0.8933,
+      "mean_token_accuracy": 0.7270920872688293,
+      "num_tokens": 9522338.0,
+      "step": 694
+    },
+    {
+      "entropy": 0.8856834173202515,
+      "epoch": 1.6951219512195121,
+      "grad_norm": 0.5028641819953918,
+      "learning_rate": 1.707317073170732e-06,
+      "loss": 0.874,
+      "mean_token_accuracy": 0.7312750220298767,
+      "num_tokens": 9536346.0,
+      "step": 695
+    },
+    {
+      "entropy": 0.8918694257736206,
+      "epoch": 1.697560975609756,
+      "grad_norm": 0.5215321183204651,
+      "learning_rate": 1.693766937669377e-06,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7307663559913635,
+      "num_tokens": 9549867.0,
+      "step": 696
+    },
+    {
+      "entropy": 0.891266942024231,
+      "epoch": 1.7,
+      "grad_norm": 0.5111386775970459,
+      "learning_rate": 1.6802168021680219e-06,
+      "loss": 0.8796,
+      "mean_token_accuracy": 0.7272476553916931,
+      "num_tokens": 9564398.0,
+      "step": 697
+    },
+    {
+      "entropy": 0.861628532409668,
+      "epoch": 1.7024390243902439,
+      "grad_norm": 0.5218578577041626,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.8463,
+      "mean_token_accuracy": 0.7348657846450806,
+      "num_tokens": 9578422.0,
+      "step": 698
+    },
+    {
+      "entropy": 0.8941474556922913,
+      "epoch": 1.7048780487804878,
+      "grad_norm": 0.5603755712509155,
+      "learning_rate": 1.6531165311653119e-06,
+      "loss": 0.8736,
+      "mean_token_accuracy": 0.7347380518913269,
+      "num_tokens": 9592296.0,
+      "step": 699
+    },
+    {
+      "entropy": 0.8734927177429199,
+      "epoch": 1.7073170731707317,
+      "grad_norm": 0.49485158920288086,
+      "learning_rate": 1.6395663956639568e-06,
+      "loss": 0.8656,
+      "mean_token_accuracy": 0.734604001045227,
+      "num_tokens": 9606163.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.8836231827735901,
+      "epoch": 1.7097560975609756,
+      "grad_norm": 0.5523728728294373,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 0.8877,
+      "mean_token_accuracy": 0.7232486009597778,
+      "num_tokens": 9619711.0,
+      "step": 701
+    },
+    {
+      "entropy": 0.8645951747894287,
+      "epoch": 1.7121951219512195,
+      "grad_norm": 0.5394639372825623,
+      "learning_rate": 1.6124661246612466e-06,
+      "loss": 0.8535,
+      "mean_token_accuracy": 0.7320195436477661,
+      "num_tokens": 9633631.0,
+      "step": 702
+    },
+    {
+      "entropy": 0.8609342575073242,
+      "epoch": 1.7146341463414634,
+      "grad_norm": 0.5509612560272217,
+      "learning_rate": 1.5989159891598918e-06,
+      "loss": 0.8642,
+      "mean_token_accuracy": 0.7359808087348938,
+      "num_tokens": 9647396.0,
+      "step": 703
+    },
+    {
+      "entropy": 0.8949425220489502,
+      "epoch": 1.7170731707317073,
+      "grad_norm": 0.5197579860687256,
+      "learning_rate": 1.5853658536585368e-06,
+      "loss": 0.8879,
+      "mean_token_accuracy": 0.7241796851158142,
+      "num_tokens": 9661461.0,
+      "step": 704
+    },
+    {
+      "entropy": 0.8748670816421509,
+      "epoch": 1.7195121951219512,
+      "grad_norm": 0.5404813289642334,
+      "learning_rate": 1.5718157181571816e-06,
+      "loss": 0.889,
+      "mean_token_accuracy": 0.7258481979370117,
+      "num_tokens": 9675389.0,
+      "step": 705
+    },
+    {
+      "entropy": 0.8763769865036011,
+      "epoch": 1.721951219512195,
+      "grad_norm": 0.5087519288063049,
+      "learning_rate": 1.5582655826558265e-06,
+      "loss": 0.8703,
+      "mean_token_accuracy": 0.7309332489967346,
+      "num_tokens": 9689238.0,
+      "step": 706
+    },
+    {
+      "entropy": 0.873427152633667,
+      "epoch": 1.724390243902439,
+      "grad_norm": 0.5565934181213379,
+      "learning_rate": 1.5447154471544717e-06,
+      "loss": 0.8629,
+      "mean_token_accuracy": 0.7338494062423706,
+      "num_tokens": 9703077.0,
+      "step": 707
+    },
+    {
+      "entropy": 0.874050498008728,
+      "epoch": 1.726829268292683,
+      "grad_norm": 0.5677209496498108,
+      "learning_rate": 1.5311653116531167e-06,
+      "loss": 0.8541,
+      "mean_token_accuracy": 0.7351242303848267,
+      "num_tokens": 9716941.0,
+      "step": 708
+    },
+    {
+      "entropy": 0.8907315731048584,
+      "epoch": 1.7292682926829268,
+      "grad_norm": 0.560185432434082,
+      "learning_rate": 1.5176151761517615e-06,
+      "loss": 0.8859,
+      "mean_token_accuracy": 0.7275949120521545,
+      "num_tokens": 9730782.0,
+      "step": 709
+    },
+    {
+      "entropy": 0.8864357471466064,
+      "epoch": 1.7317073170731707,
+      "grad_norm": 0.5551504492759705,
+      "learning_rate": 1.5040650406504067e-06,
+      "loss": 0.8893,
+      "mean_token_accuracy": 0.7227503061294556,
+      "num_tokens": 9745211.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8754174709320068,
+      "epoch": 1.7341463414634146,
+      "grad_norm": 0.4636332094669342,
+      "learning_rate": 1.4905149051490517e-06,
+      "loss": 0.852,
+      "mean_token_accuracy": 0.735327422618866,
+      "num_tokens": 9759369.0,
+      "step": 711
+    },
+    {
+      "entropy": 0.8752076625823975,
+      "epoch": 1.7365853658536585,
+      "grad_norm": 0.5738482475280762,
+      "learning_rate": 1.4769647696476967e-06,
+      "loss": 0.8555,
+      "mean_token_accuracy": 0.7349178791046143,
+      "num_tokens": 9773143.0,
+      "step": 712
+    },
+    {
+      "entropy": 0.8808815479278564,
+      "epoch": 1.7390243902439024,
+      "grad_norm": 0.4846593141555786,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 0.8672,
+      "mean_token_accuracy": 0.7282733917236328,
+      "num_tokens": 9787059.0,
+      "step": 713
+    },
+    {
+      "entropy": 0.8802102208137512,
+      "epoch": 1.7414634146341463,
+      "grad_norm": 0.5256415605545044,
+      "learning_rate": 1.4498644986449866e-06,
+      "loss": 0.8564,
+      "mean_token_accuracy": 0.7356405258178711,
+      "num_tokens": 9800829.0,
+      "step": 714
+    },
+    {
+      "entropy": 0.8865002393722534,
+      "epoch": 1.7439024390243902,
+      "grad_norm": 0.6085943579673767,
+      "learning_rate": 1.4363143631436316e-06,
+      "loss": 0.8862,
+      "mean_token_accuracy": 0.7306849360466003,
+      "num_tokens": 9814992.0,
+      "step": 715
+    },
+    {
+      "entropy": 0.8770885467529297,
+      "epoch": 1.7463414634146341,
+      "grad_norm": 0.5122337341308594,
+      "learning_rate": 1.4227642276422766e-06,
+      "loss": 0.8617,
+      "mean_token_accuracy": 0.7321338057518005,
+      "num_tokens": 9828847.0,
+      "step": 716
+    },
+    {
+      "entropy": 0.8982193470001221,
+      "epoch": 1.748780487804878,
+      "grad_norm": 0.5098617672920227,
+      "learning_rate": 1.4092140921409214e-06,
+      "loss": 0.8877,
+      "mean_token_accuracy": 0.7284758687019348,
+      "num_tokens": 9842766.0,
+      "step": 717
+    },
+    {
+      "entropy": 0.8702014088630676,
+      "epoch": 1.751219512195122,
+      "grad_norm": 0.5830314755439758,
+      "learning_rate": 1.3956639566395666e-06,
+      "loss": 0.8695,
+      "mean_token_accuracy": 0.7296539545059204,
+      "num_tokens": 9856335.0,
+      "step": 718
+    },
+    {
+      "entropy": 0.8794569969177246,
+      "epoch": 1.7536585365853659,
+      "grad_norm": 0.5575631260871887,
+      "learning_rate": 1.3821138211382116e-06,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.734868049621582,
+      "num_tokens": 9870295.0,
+      "step": 719
+    },
+    {
+      "entropy": 0.8718886375427246,
+      "epoch": 1.7560975609756098,
+      "grad_norm": 0.7530406713485718,
+      "learning_rate": 1.3685636856368563e-06,
+      "loss": 0.8599,
+      "mean_token_accuracy": 0.7317982912063599,
+      "num_tokens": 9884252.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8773154020309448,
+      "epoch": 1.7585365853658537,
+      "grad_norm": 0.5159140825271606,
+      "learning_rate": 1.3550135501355013e-06,
+      "loss": 0.8548,
+      "mean_token_accuracy": 0.7374392151832581,
+      "num_tokens": 9897842.0,
+      "step": 721
+    },
+    {
+      "entropy": 0.8738135099411011,
+      "epoch": 1.7609756097560976,
+      "grad_norm": 0.5728605389595032,
+      "learning_rate": 1.3414634146341465e-06,
+      "loss": 0.8595,
+      "mean_token_accuracy": 0.7306110262870789,
+      "num_tokens": 9910984.0,
+      "step": 722
+    },
+    {
+      "entropy": 0.8791817426681519,
+      "epoch": 1.7634146341463415,
+      "grad_norm": 0.5503786206245422,
+      "learning_rate": 1.3279132791327915e-06,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.72954261302948,
+      "num_tokens": 9924577.0,
+      "step": 723
+    },
+    {
+      "entropy": 0.8843265771865845,
+      "epoch": 1.7658536585365854,
+      "grad_norm": 0.5542715787887573,
+      "learning_rate": 1.3143631436314363e-06,
+      "loss": 0.8702,
+      "mean_token_accuracy": 0.7308476567268372,
+      "num_tokens": 9938325.0,
+      "step": 724
+    },
+    {
+      "entropy": 0.865561306476593,
+      "epoch": 1.7682926829268293,
+      "grad_norm": 0.5139968991279602,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 0.8481,
+      "mean_token_accuracy": 0.7372664213180542,
+      "num_tokens": 9951613.0,
+      "step": 725
+    },
+    {
+      "entropy": 0.8823000192642212,
+      "epoch": 1.7707317073170732,
+      "grad_norm": 0.5414228439331055,
+      "learning_rate": 1.2872628726287264e-06,
+      "loss": 0.8788,
+      "mean_token_accuracy": 0.7271595001220703,
+      "num_tokens": 9965278.0,
+      "step": 726
+    },
+    {
+      "entropy": 0.8546561598777771,
+      "epoch": 1.773170731707317,
+      "grad_norm": 0.6240255832672119,
+      "learning_rate": 1.2737127371273714e-06,
+      "loss": 0.8431,
+      "mean_token_accuracy": 0.7430485486984253,
+      "num_tokens": 9979176.0,
+      "step": 727
+    },
+    {
+      "entropy": 0.8734099268913269,
+      "epoch": 1.775609756097561,
+      "grad_norm": 0.5208917856216431,
+      "learning_rate": 1.2601626016260162e-06,
+      "loss": 0.8677,
+      "mean_token_accuracy": 0.7244963645935059,
+      "num_tokens": 9993141.0,
+      "step": 728
+    },
+    {
+      "entropy": 0.8741912841796875,
+      "epoch": 1.778048780487805,
+      "grad_norm": 0.5388087630271912,
+      "learning_rate": 1.2466124661246614e-06,
+      "loss": 0.8604,
+      "mean_token_accuracy": 0.7358125448226929,
+      "num_tokens": 10006602.0,
+      "step": 729
+    },
+    {
+      "entropy": 0.8812603950500488,
+      "epoch": 1.7804878048780488,
+      "grad_norm": 0.5764947533607483,
+      "learning_rate": 1.2330623306233064e-06,
+      "loss": 0.8866,
+      "mean_token_accuracy": 0.727005124092102,
+      "num_tokens": 10020208.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.883576512336731,
+      "epoch": 1.7829268292682927,
+      "grad_norm": 0.562300443649292,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 0.8772,
+      "mean_token_accuracy": 0.7265163064002991,
+      "num_tokens": 10033925.0,
+      "step": 731
+    },
+    {
+      "entropy": 0.864028811454773,
+      "epoch": 1.7853658536585366,
+      "grad_norm": 0.5029560327529907,
+      "learning_rate": 1.2059620596205964e-06,
+      "loss": 0.8683,
+      "mean_token_accuracy": 0.7317820191383362,
+      "num_tokens": 10047650.0,
+      "step": 732
+    },
+    {
+      "entropy": 0.8827657699584961,
+      "epoch": 1.7878048780487805,
+      "grad_norm": 0.5882710218429565,
+      "learning_rate": 1.1924119241192413e-06,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7273064255714417,
+      "num_tokens": 10061150.0,
+      "step": 733
+    },
+    {
+      "entropy": 0.8744525909423828,
+      "epoch": 1.7902439024390244,
+      "grad_norm": 0.5701075792312622,
+      "learning_rate": 1.1788617886178863e-06,
+      "loss": 0.865,
+      "mean_token_accuracy": 0.7322956919670105,
+      "num_tokens": 10074722.0,
+      "step": 734
+    },
+    {
+      "entropy": 0.8803761005401611,
+      "epoch": 1.7926829268292683,
+      "grad_norm": 0.5309422612190247,
+      "learning_rate": 1.1653116531165313e-06,
+      "loss": 0.8675,
+      "mean_token_accuracy": 0.7303227782249451,
+      "num_tokens": 10088091.0,
+      "step": 735
+    },
+    {
+      "entropy": 0.8677628040313721,
+      "epoch": 1.7951219512195122,
+      "grad_norm": 0.5347625613212585,
+      "learning_rate": 1.1517615176151763e-06,
+      "loss": 0.8531,
+      "mean_token_accuracy": 0.7351138591766357,
+      "num_tokens": 10101811.0,
+      "step": 736
+    },
+    {
+      "entropy": 0.8758652806282043,
+      "epoch": 1.7975609756097561,
+      "grad_norm": 0.5840176343917847,
+      "learning_rate": 1.1382113821138213e-06,
+      "loss": 0.8708,
+      "mean_token_accuracy": 0.7307803630828857,
+      "num_tokens": 10115667.0,
+      "step": 737
+    },
+    {
+      "entropy": 0.8817602396011353,
+      "epoch": 1.8,
+      "grad_norm": 0.574613630771637,
+      "learning_rate": 1.1246612466124663e-06,
+      "loss": 0.8888,
+      "mean_token_accuracy": 0.7227194309234619,
+      "num_tokens": 10128980.0,
+      "step": 738
+    },
+    {
+      "entropy": 0.8836100697517395,
+      "epoch": 1.802439024390244,
+      "grad_norm": 0.5252755880355835,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.7268431186676025,
+      "num_tokens": 10142750.0,
+      "step": 739
+    },
+    {
+      "entropy": 0.8818948268890381,
+      "epoch": 1.8048780487804879,
+      "grad_norm": 0.6356107592582703,
+      "learning_rate": 1.0975609756097562e-06,
+      "loss": 0.8776,
+      "mean_token_accuracy": 0.7285158038139343,
+      "num_tokens": 10155857.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.883324384689331,
+      "epoch": 1.8073170731707318,
+      "grad_norm": 0.6597152352333069,
+      "learning_rate": 1.0840108401084012e-06,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.7296566963195801,
+      "num_tokens": 10169563.0,
+      "step": 741
+    },
+    {
+      "entropy": 0.8747758269309998,
+      "epoch": 1.8097560975609757,
+      "grad_norm": 0.5430500507354736,
+      "learning_rate": 1.0704607046070462e-06,
+      "loss": 0.862,
+      "mean_token_accuracy": 0.7310535311698914,
+      "num_tokens": 10182972.0,
+      "step": 742
+    },
+    {
+      "entropy": 0.8945927023887634,
+      "epoch": 1.8121951219512196,
+      "grad_norm": 0.5791915655136108,
+      "learning_rate": 1.0569105691056912e-06,
+      "loss": 0.8915,
+      "mean_token_accuracy": 0.7228506207466125,
+      "num_tokens": 10196515.0,
+      "step": 743
+    },
+    {
+      "entropy": 0.889089047908783,
+      "epoch": 1.8146341463414632,
+      "grad_norm": 0.5456008315086365,
+      "learning_rate": 1.0433604336043362e-06,
+      "loss": 0.877,
+      "mean_token_accuracy": 0.728965163230896,
+      "num_tokens": 10210282.0,
+      "step": 744
+    },
+    {
+      "entropy": 0.8740544319152832,
+      "epoch": 1.8170731707317072,
+      "grad_norm": 0.660473644733429,
+      "learning_rate": 1.0298102981029812e-06,
+      "loss": 0.8764,
+      "mean_token_accuracy": 0.7304914593696594,
+      "num_tokens": 10224420.0,
+      "step": 745
+    },
+    {
+      "entropy": 0.8994358777999878,
+      "epoch": 1.819512195121951,
+      "grad_norm": 0.5640230774879456,
+      "learning_rate": 1.0162601626016261e-06,
+      "loss": 0.8883,
+      "mean_token_accuracy": 0.7228577733039856,
+      "num_tokens": 10238335.0,
+      "step": 746
+    },
+    {
+      "entropy": 0.8852430582046509,
+      "epoch": 1.821951219512195,
+      "grad_norm": 0.5081997513771057,
+      "learning_rate": 1.0027100271002711e-06,
+      "loss": 0.8734,
+      "mean_token_accuracy": 0.7313060164451599,
+      "num_tokens": 10251965.0,
+      "step": 747
+    },
+    {
+      "entropy": 0.8647916316986084,
+      "epoch": 1.8243902439024389,
+      "grad_norm": 0.5371124744415283,
+      "learning_rate": 9.891598915989161e-07,
+      "loss": 0.8464,
+      "mean_token_accuracy": 0.7346433401107788,
+      "num_tokens": 10265314.0,
+      "step": 748
+    },
+    {
+      "entropy": 0.8955444097518921,
+      "epoch": 1.8268292682926828,
+      "grad_norm": 0.5369889736175537,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 0.8706,
+      "mean_token_accuracy": 0.731004536151886,
+      "num_tokens": 10279386.0,
+      "step": 749
+    },
+    {
+      "entropy": 0.8667882680892944,
+      "epoch": 1.8292682926829267,
+      "grad_norm": 0.5496275424957275,
+      "learning_rate": 9.62059620596206e-07,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.7305599451065063,
+      "num_tokens": 10293368.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8811326026916504,
+      "epoch": 1.8317073170731706,
+      "grad_norm": 0.5627800822257996,
+      "learning_rate": 9.485094850948511e-07,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7355487942695618,
+      "num_tokens": 10306895.0,
+      "step": 751
+    },
+    {
+      "entropy": 0.8759801387786865,
+      "epoch": 1.8341463414634145,
+      "grad_norm": 0.5209669470787048,
+      "learning_rate": 9.349593495934959e-07,
+      "loss": 0.8723,
+      "mean_token_accuracy": 0.7327739596366882,
+      "num_tokens": 10320379.0,
+      "step": 752
+    },
+    {
+      "entropy": 0.9030628204345703,
+      "epoch": 1.8365853658536584,
+      "grad_norm": 0.5620582699775696,
+      "learning_rate": 9.21409214092141e-07,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7353246212005615,
+      "num_tokens": 10333887.0,
+      "step": 753
+    },
+    {
+      "entropy": 0.8929505348205566,
+      "epoch": 1.8390243902439023,
+      "grad_norm": 0.5281407833099365,
+      "learning_rate": 9.078590785907859e-07,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.7236402630805969,
+      "num_tokens": 10347968.0,
+      "step": 754
+    },
+    {
+      "entropy": 0.8875263929367065,
+      "epoch": 1.8414634146341462,
+      "grad_norm": 0.5603488087654114,
+      "learning_rate": 8.94308943089431e-07,
+      "loss": 0.8783,
+      "mean_token_accuracy": 0.730212390422821,
+      "num_tokens": 10361591.0,
+      "step": 755
+    },
+    {
+      "entropy": 0.8743438720703125,
+      "epoch": 1.84390243902439,
+      "grad_norm": 0.515800416469574,
+      "learning_rate": 8.807588075880759e-07,
+      "loss": 0.8518,
+      "mean_token_accuracy": 0.7341781258583069,
+      "num_tokens": 10375180.0,
+      "step": 756
+    },
+    {
+      "entropy": 0.8860392570495605,
+      "epoch": 1.846341463414634,
+      "grad_norm": 0.5203832387924194,
+      "learning_rate": 8.67208672086721e-07,
+      "loss": 0.8708,
+      "mean_token_accuracy": 0.729315459728241,
+      "num_tokens": 10389615.0,
+      "step": 757
+    },
+    {
+      "entropy": 0.9075171947479248,
+      "epoch": 1.848780487804878,
+      "grad_norm": 0.5129781365394592,
+      "learning_rate": 8.53658536585366e-07,
+      "loss": 0.8972,
+      "mean_token_accuracy": 0.723143458366394,
+      "num_tokens": 10403501.0,
+      "step": 758
+    },
+    {
+      "entropy": 0.8773412108421326,
+      "epoch": 1.8512195121951218,
+      "grad_norm": 0.5214099884033203,
+      "learning_rate": 8.401084010840109e-07,
+      "loss": 0.8701,
+      "mean_token_accuracy": 0.7358790040016174,
+      "num_tokens": 10416742.0,
+      "step": 759
+    },
+    {
+      "entropy": 0.9007905721664429,
+      "epoch": 1.8536585365853657,
+      "grad_norm": 0.5532057285308838,
+      "learning_rate": 8.265582655826559e-07,
+      "loss": 0.885,
+      "mean_token_accuracy": 0.727091908454895,
+      "num_tokens": 10430334.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.875899076461792,
+      "epoch": 1.8560975609756096,
+      "grad_norm": 0.4984937608242035,
+      "learning_rate": 8.130081300813009e-07,
+      "loss": 0.8598,
+      "mean_token_accuracy": 0.7356676459312439,
+      "num_tokens": 10444287.0,
+      "step": 761
+    },
+    {
+      "entropy": 0.879464328289032,
+      "epoch": 1.8585365853658535,
+      "grad_norm": 0.5563964247703552,
+      "learning_rate": 7.994579945799459e-07,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7309806942939758,
+      "num_tokens": 10457763.0,
+      "step": 762
+    },
+    {
+      "entropy": 0.8621103167533875,
+      "epoch": 1.8609756097560974,
+      "grad_norm": 0.614388644695282,
+      "learning_rate": 7.859078590785908e-07,
+      "loss": 0.8607,
+      "mean_token_accuracy": 0.7339669466018677,
+      "num_tokens": 10471563.0,
+      "step": 763
+    },
+    {
+      "entropy": 0.8521906137466431,
+      "epoch": 1.8634146341463413,
+      "grad_norm": 0.6084858775138855,
+      "learning_rate": 7.723577235772359e-07,
+      "loss": 0.8401,
+      "mean_token_accuracy": 0.7348809242248535,
+      "num_tokens": 10484890.0,
+      "step": 764
+    },
+    {
+      "entropy": 0.8763301372528076,
+      "epoch": 1.8658536585365852,
+      "grad_norm": 0.5668750405311584,
+      "learning_rate": 7.588075880758807e-07,
+      "loss": 0.8895,
+      "mean_token_accuracy": 0.7239844799041748,
+      "num_tokens": 10499061.0,
+      "step": 765
+    },
+    {
+      "entropy": 0.8991358280181885,
+      "epoch": 1.8682926829268292,
+      "grad_norm": 0.48762112855911255,
+      "learning_rate": 7.452574525745258e-07,
+      "loss": 0.8837,
+      "mean_token_accuracy": 0.7247573733329773,
+      "num_tokens": 10512883.0,
+      "step": 766
+    },
+    {
+      "entropy": 0.8535343408584595,
+      "epoch": 1.870731707317073,
+      "grad_norm": 0.5635075569152832,
+      "learning_rate": 7.317073170731707e-07,
+      "loss": 0.8385,
+      "mean_token_accuracy": 0.7378661632537842,
+      "num_tokens": 10526930.0,
+      "step": 767
+    },
+    {
+      "entropy": 0.8730718493461609,
+      "epoch": 1.873170731707317,
+      "grad_norm": 0.5588259696960449,
+      "learning_rate": 7.181571815718158e-07,
+      "loss": 0.859,
+      "mean_token_accuracy": 0.734580397605896,
+      "num_tokens": 10540792.0,
+      "step": 768
+    },
+    {
+      "entropy": 0.863853931427002,
+      "epoch": 1.8756097560975609,
+      "grad_norm": 0.5849318504333496,
+      "learning_rate": 7.046070460704607e-07,
+      "loss": 0.8562,
+      "mean_token_accuracy": 0.7336525321006775,
+      "num_tokens": 10554174.0,
+      "step": 769
+    },
+    {
+      "entropy": 0.8770521283149719,
+      "epoch": 1.8780487804878048,
+      "grad_norm": 0.5464377999305725,
+      "learning_rate": 6.910569105691058e-07,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7362645864486694,
+      "num_tokens": 10567495.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.887363076210022,
+      "epoch": 1.8804878048780487,
+      "grad_norm": 0.5217662453651428,
+      "learning_rate": 6.775067750677507e-07,
+      "loss": 0.8955,
+      "mean_token_accuracy": 0.7222620844841003,
+      "num_tokens": 10581445.0,
+      "step": 771
+    },
+    {
+      "entropy": 0.8654752969741821,
+      "epoch": 1.8829268292682926,
+      "grad_norm": 0.5389561653137207,
+      "learning_rate": 6.639566395663957e-07,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.7368183732032776,
+      "num_tokens": 10594756.0,
+      "step": 772
+    },
+    {
+      "entropy": 0.9031565189361572,
+      "epoch": 1.8853658536585365,
+      "grad_norm": 0.5442070960998535,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 0.9152,
+      "mean_token_accuracy": 0.7197484970092773,
+      "num_tokens": 10608449.0,
+      "step": 773
+    },
+    {
+      "entropy": 0.8798193335533142,
+      "epoch": 1.8878048780487804,
+      "grad_norm": 0.5139967203140259,
+      "learning_rate": 6.368563685636857e-07,
+      "loss": 0.8465,
+      "mean_token_accuracy": 0.7399256825447083,
+      "num_tokens": 10622188.0,
+      "step": 774
+    },
+    {
+      "entropy": 0.8660321235656738,
+      "epoch": 1.8902439024390243,
+      "grad_norm": 0.5604297518730164,
+      "learning_rate": 6.233062330623307e-07,
+      "loss": 0.8621,
+      "mean_token_accuracy": 0.7294635772705078,
+      "num_tokens": 10635607.0,
+      "step": 775
+    },
+    {
+      "entropy": 0.8818001747131348,
+      "epoch": 1.8926829268292682,
+      "grad_norm": 0.5093931555747986,
+      "learning_rate": 6.097560975609757e-07,
+      "loss": 0.8791,
+      "mean_token_accuracy": 0.7306883335113525,
+      "num_tokens": 10649410.0,
+      "step": 776
+    },
+    {
+      "entropy": 0.879349946975708,
+      "epoch": 1.895121951219512,
+      "grad_norm": 0.5525664687156677,
+      "learning_rate": 5.962059620596207e-07,
+      "loss": 0.859,
+      "mean_token_accuracy": 0.7347868084907532,
+      "num_tokens": 10662819.0,
+      "step": 777
+    },
+    {
+      "entropy": 0.8722898960113525,
+      "epoch": 1.897560975609756,
+      "grad_norm": 0.5286668539047241,
+      "learning_rate": 5.826558265582657e-07,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.7305213809013367,
+      "num_tokens": 10676491.0,
+      "step": 778
+    },
+    {
+      "entropy": 0.8877685070037842,
+      "epoch": 1.9,
+      "grad_norm": 0.531743586063385,
+      "learning_rate": 5.691056910569106e-07,
+      "loss": 0.8699,
+      "mean_token_accuracy": 0.7307401299476624,
+      "num_tokens": 10689721.0,
+      "step": 779
+    },
+    {
+      "entropy": 0.8780038356781006,
+      "epoch": 1.9024390243902438,
+      "grad_norm": 0.5183002948760986,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 0.8738,
+      "mean_token_accuracy": 0.72763991355896,
+      "num_tokens": 10703601.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8894438743591309,
+      "epoch": 1.9048780487804877,
+      "grad_norm": 0.4965713918209076,
+      "learning_rate": 5.420054200542006e-07,
+      "loss": 0.8886,
+      "mean_token_accuracy": 0.7262105941772461,
+      "num_tokens": 10717226.0,
+      "step": 781
+    },
+    {
+      "entropy": 0.8759099245071411,
+      "epoch": 1.9073170731707316,
+      "grad_norm": 0.504848301410675,
+      "learning_rate": 5.284552845528456e-07,
+      "loss": 0.871,
+      "mean_token_accuracy": 0.7337351441383362,
+      "num_tokens": 10731014.0,
+      "step": 782
+    },
+    {
+      "entropy": 0.8838939666748047,
+      "epoch": 1.9097560975609755,
+      "grad_norm": 0.5675578713417053,
+      "learning_rate": 5.149051490514906e-07,
+      "loss": 0.873,
+      "mean_token_accuracy": 0.731378436088562,
+      "num_tokens": 10745046.0,
+      "step": 783
+    },
+    {
+      "entropy": 0.8941045999526978,
+      "epoch": 1.9121951219512194,
+      "grad_norm": 0.609192967414856,
+      "learning_rate": 5.013550135501356e-07,
+      "loss": 0.895,
+      "mean_token_accuracy": 0.7223051190376282,
+      "num_tokens": 10759135.0,
+      "step": 784
+    },
+    {
+      "entropy": 0.8828786611557007,
+      "epoch": 1.9146341463414633,
+      "grad_norm": 0.5097105503082275,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 0.8647,
+      "mean_token_accuracy": 0.7296620607376099,
+      "num_tokens": 10772734.0,
+      "step": 785
+    },
+    {
+      "entropy": 0.8686960339546204,
+      "epoch": 1.9170731707317072,
+      "grad_norm": 0.5239219665527344,
+      "learning_rate": 4.7425474254742553e-07,
+      "loss": 0.8578,
+      "mean_token_accuracy": 0.7318073511123657,
+      "num_tokens": 10786162.0,
+      "step": 786
+    },
+    {
+      "entropy": 0.8628572225570679,
+      "epoch": 1.9195121951219511,
+      "grad_norm": 0.5133153200149536,
+      "learning_rate": 4.607046070460705e-07,
+      "loss": 0.835,
+      "mean_token_accuracy": 0.7363176941871643,
+      "num_tokens": 10799827.0,
+      "step": 787
+    },
+    {
+      "entropy": 0.8920273780822754,
+      "epoch": 1.921951219512195,
+      "grad_norm": 0.5285261869430542,
+      "learning_rate": 4.471544715447155e-07,
+      "loss": 0.8851,
+      "mean_token_accuracy": 0.7281093001365662,
+      "num_tokens": 10813753.0,
+      "step": 788
+    },
+    {
+      "entropy": 0.8801295757293701,
+      "epoch": 1.924390243902439,
+      "grad_norm": 0.5520910620689392,
+      "learning_rate": 4.336043360433605e-07,
+      "loss": 0.8872,
+      "mean_token_accuracy": 0.7253822088241577,
+      "num_tokens": 10827570.0,
+      "step": 789
+    },
+    {
+      "entropy": 0.8774034380912781,
+      "epoch": 1.9268292682926829,
+      "grad_norm": 0.5181596279144287,
+      "learning_rate": 4.2005420054200547e-07,
+      "loss": 0.8544,
+      "mean_token_accuracy": 0.7363961935043335,
+      "num_tokens": 10841277.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8727174997329712,
+      "epoch": 1.9292682926829268,
+      "grad_norm": 0.5676617622375488,
+      "learning_rate": 4.0650406504065046e-07,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7332050800323486,
+      "num_tokens": 10854809.0,
+      "step": 791
+    },
+    {
+      "entropy": 0.8603472113609314,
+      "epoch": 1.9317073170731707,
+      "grad_norm": 0.5061153173446655,
+      "learning_rate": 3.929539295392954e-07,
+      "loss": 0.8337,
+      "mean_token_accuracy": 0.7372038960456848,
+      "num_tokens": 10868208.0,
+      "step": 792
+    },
+    {
+      "entropy": 0.8648924827575684,
+      "epoch": 1.9341463414634146,
+      "grad_norm": 0.516345202922821,
+      "learning_rate": 3.794037940379404e-07,
+      "loss": 0.8779,
+      "mean_token_accuracy": 0.7320489287376404,
+      "num_tokens": 10882290.0,
+      "step": 793
+    },
+    {
+      "entropy": 0.887532651424408,
+      "epoch": 1.9365853658536585,
+      "grad_norm": 0.5096654295921326,
+      "learning_rate": 3.6585365853658536e-07,
+      "loss": 0.8717,
+      "mean_token_accuracy": 0.7271854877471924,
+      "num_tokens": 10895850.0,
+      "step": 794
+    },
+    {
+      "entropy": 0.8896082639694214,
+      "epoch": 1.9390243902439024,
+      "grad_norm": 0.5067712664604187,
+      "learning_rate": 3.5230352303523034e-07,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7332432866096497,
+      "num_tokens": 10909935.0,
+      "step": 795
+    },
+    {
+      "entropy": 0.8808606863021851,
+      "epoch": 1.9414634146341463,
+      "grad_norm": 0.5321662425994873,
+      "learning_rate": 3.3875338753387533e-07,
+      "loss": 0.8754,
+      "mean_token_accuracy": 0.7316710352897644,
+      "num_tokens": 10923386.0,
+      "step": 796
+    },
+    {
+      "entropy": 0.882371187210083,
+      "epoch": 1.9439024390243902,
+      "grad_norm": 0.6009424924850464,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 0.8694,
+      "mean_token_accuracy": 0.7282980680465698,
+      "num_tokens": 10936258.0,
+      "step": 797
+    },
+    {
+      "entropy": 0.8814548254013062,
+      "epoch": 1.946341463414634,
+      "grad_norm": 0.533541202545166,
+      "learning_rate": 3.1165311653116535e-07,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.732546865940094,
+      "num_tokens": 10949667.0,
+      "step": 798
+    },
+    {
+      "entropy": 0.8886417150497437,
+      "epoch": 1.948780487804878,
+      "grad_norm": 0.540417492389679,
+      "learning_rate": 2.9810298102981034e-07,
+      "loss": 0.8919,
+      "mean_token_accuracy": 0.7254874110221863,
+      "num_tokens": 10963890.0,
+      "step": 799
+    },
+    {
+      "entropy": 0.8837313652038574,
+      "epoch": 1.951219512195122,
+      "grad_norm": 0.5418463349342346,
+      "learning_rate": 2.845528455284553e-07,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7310764193534851,
+      "num_tokens": 10977553.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8748241662979126,
+      "epoch": 1.9536585365853658,
+      "grad_norm": 0.49068158864974976,
+      "learning_rate": 2.710027100271003e-07,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7294101119041443,
+      "num_tokens": 10991605.0,
+      "step": 801
+    },
+    {
+      "entropy": 0.8987839818000793,
+      "epoch": 1.9560975609756097,
+      "grad_norm": 0.5240376591682434,
+      "learning_rate": 2.574525745257453e-07,
+      "loss": 0.897,
+      "mean_token_accuracy": 0.7241867184638977,
+      "num_tokens": 11005054.0,
+      "step": 802
+    },
+    {
+      "entropy": 0.8850489258766174,
+      "epoch": 1.9585365853658536,
+      "grad_norm": 0.5221152305603027,
+      "learning_rate": 2.439024390243903e-07,
+      "loss": 0.8724,
+      "mean_token_accuracy": 0.7294400930404663,
+      "num_tokens": 11018786.0,
+      "step": 803
+    },
+    {
+      "entropy": 0.86225825548172,
+      "epoch": 1.9609756097560975,
+      "grad_norm": 0.5298704504966736,
+      "learning_rate": 2.3035230352303526e-07,
+      "loss": 0.8594,
+      "mean_token_accuracy": 0.7317895889282227,
+      "num_tokens": 11032146.0,
+      "step": 804
+    },
+    {
+      "entropy": 0.8769630193710327,
+      "epoch": 1.9634146341463414,
+      "grad_norm": 0.5109516382217407,
+      "learning_rate": 2.1680216802168024e-07,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7320917844772339,
+      "num_tokens": 11045801.0,
+      "step": 805
+    },
+    {
+      "entropy": 0.8887900114059448,
+      "epoch": 1.9658536585365853,
+      "grad_norm": 0.4967120289802551,
+      "learning_rate": 2.0325203252032523e-07,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.733670175075531,
+      "num_tokens": 11059672.0,
+      "step": 806
+    },
+    {
+      "entropy": 0.8606371879577637,
+      "epoch": 1.9682926829268292,
+      "grad_norm": 0.532614529132843,
+      "learning_rate": 1.897018970189702e-07,
+      "loss": 0.8465,
+      "mean_token_accuracy": 0.736371636390686,
+      "num_tokens": 11073336.0,
+      "step": 807
+    },
+    {
+      "entropy": 0.8847682476043701,
+      "epoch": 1.9707317073170731,
+      "grad_norm": 0.5335534811019897,
+      "learning_rate": 1.7615176151761517e-07,
+      "loss": 0.8846,
+      "mean_token_accuracy": 0.7342714667320251,
+      "num_tokens": 11086783.0,
+      "step": 808
+    },
+    {
+      "entropy": 0.8636947870254517,
+      "epoch": 1.973170731707317,
+      "grad_norm": 0.5089491009712219,
+      "learning_rate": 1.6260162601626018e-07,
+      "loss": 0.8487,
+      "mean_token_accuracy": 0.738890528678894,
+      "num_tokens": 11100571.0,
+      "step": 809
+    },
+    {
+      "entropy": 0.8890416622161865,
+      "epoch": 1.975609756097561,
+      "grad_norm": 0.5388813018798828,
+      "learning_rate": 1.4905149051490517e-07,
+      "loss": 0.8864,
+      "mean_token_accuracy": 0.7321293354034424,
+      "num_tokens": 11113765.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9044409394264221,
+      "epoch": 1.9780487804878049,
+      "grad_norm": 0.5027598738670349,
+      "learning_rate": 1.3550135501355015e-07,
+      "loss": 0.8914,
+      "mean_token_accuracy": 0.7259988784790039,
+      "num_tokens": 11127697.0,
+      "step": 811
+    },
+    {
+      "entropy": 0.8642138242721558,
+      "epoch": 1.9804878048780488,
+      "grad_norm": 0.5696573853492737,
+      "learning_rate": 1.2195121951219514e-07,
+      "loss": 0.8505,
+      "mean_token_accuracy": 0.7409389019012451,
+      "num_tokens": 11140984.0,
+      "step": 812
+    },
+    {
+      "entropy": 0.8789657354354858,
+      "epoch": 1.9829268292682927,
+      "grad_norm": 0.5053057670593262,
+      "learning_rate": 1.0840108401084012e-07,
+      "loss": 0.8638,
+      "mean_token_accuracy": 0.735129714012146,
+      "num_tokens": 11154954.0,
+      "step": 813
+    },
+    {
+      "entropy": 0.8782299160957336,
+      "epoch": 1.9853658536585366,
+      "grad_norm": 0.5316551327705383,
+      "learning_rate": 9.48509485094851e-08,
+      "loss": 0.8749,
+      "mean_token_accuracy": 0.7280279994010925,
+      "num_tokens": 11168692.0,
+      "step": 814
+    },
+    {
+      "entropy": 0.8809655904769897,
+      "epoch": 1.9878048780487805,
+      "grad_norm": 0.5319744944572449,
+      "learning_rate": 8.130081300813009e-08,
+      "loss": 0.8773,
+      "mean_token_accuracy": 0.7296779751777649,
+      "num_tokens": 11182806.0,
+      "step": 815
+    },
+    {
+      "entropy": 0.874808132648468,
+      "epoch": 1.9902439024390244,
+      "grad_norm": 0.5373929738998413,
+      "learning_rate": 6.775067750677508e-08,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7337278127670288,
+      "num_tokens": 11196680.0,
+      "step": 816
+    },
+    {
+      "entropy": 0.8780685663223267,
+      "epoch": 1.9926829268292683,
+      "grad_norm": 0.5213566422462463,
+      "learning_rate": 5.420054200542006e-08,
+      "loss": 0.86,
+      "mean_token_accuracy": 0.734074056148529,
+      "num_tokens": 11210196.0,
+      "step": 817
+    },
+    {
+      "entropy": 0.8767932057380676,
+      "epoch": 1.9951219512195122,
+      "grad_norm": 0.5122414827346802,
+      "learning_rate": 4.0650406504065046e-08,
+      "loss": 0.8454,
+      "mean_token_accuracy": 0.7340317368507385,
+      "num_tokens": 11223958.0,
+      "step": 818
+    },
+    {
+      "entropy": 0.8675880432128906,
+      "epoch": 1.997560975609756,
+      "grad_norm": 0.5748156309127808,
+      "learning_rate": 2.710027100271003e-08,
+      "loss": 0.8494,
+      "mean_token_accuracy": 0.7352919578552246,
+      "num_tokens": 11237453.0,
+      "step": 819
+    },
+    {
+      "entropy": 0.8685159683227539,
+      "epoch": 2.0,
+      "grad_norm": 0.535193145275116,
+      "learning_rate": 1.3550135501355015e-08,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7290802597999573,
+      "num_tokens": 11251451.0,
+      "step": 820
+    },
+    {
+      "epoch": 2.0,
+      "step": 820,
+      "total_flos": 5.656792731347845e+17,
+      "train_loss": 1.044341852534108,
+      "train_runtime": 2841.2294,
+      "train_samples_per_second": 4.611,
+      "train_steps_per_second": 0.289
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 820,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.656792731347845e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}