diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/last-checkpoint/trainer_state.json"
@@ -0,0 +1,9034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.008265897387149836,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 8.265897387149836e-06,
+      "grad_norm": 1.3251980543136597,
+      "learning_rate": 5e-05,
+      "loss": 1.5516,
+      "mean_token_accuracy": 0.7008633613586426,
+      "num_tokens": 57022.0,
+      "step": 1
+    },
+    {
+      "epoch": 1.653179477429967e-05,
+      "grad_norm": 1.300803542137146,
+      "learning_rate": 4.9999793352565324e-05,
+      "loss": 1.4318,
+      "mean_token_accuracy": 0.727364182472229,
+      "num_tokens": 59016.0,
+      "step": 2
+    },
+    {
+      "epoch": 2.4797692161449507e-05,
+      "grad_norm": 1.299092411994934,
+      "learning_rate": 4.9999586705130646e-05,
+      "loss": 1.3651,
+      "mean_token_accuracy": 0.740029513835907,
+      "num_tokens": 61053.0,
+      "step": 3
+    },
+    {
+      "epoch": 3.306358954859934e-05,
+      "grad_norm": 1.2995446920394897,
+      "learning_rate": 4.999938005769597e-05,
+      "loss": 1.2178,
+      "mean_token_accuracy": 0.7601810097694397,
+      "num_tokens": 63048.0,
+      "step": 4
+    },
+    {
+      "epoch": 4.132948693574918e-05,
+      "grad_norm": 1.2878410816192627,
+      "learning_rate": 4.999917341026128e-05,
+      "loss": 1.1257,
+      "mean_token_accuracy": 0.7849516868591309,
+      "num_tokens": 65021.0,
+      "step": 5
+    },
+    {
+      "epoch": 4.9595384322899015e-05,
+      "grad_norm": 1.298051357269287,
+      "learning_rate": 4.999896676282661e-05,
+      "loss": 1.0238,
+      "mean_token_accuracy": 0.8043912053108215,
+      "num_tokens": 67031.0,
+      "step": 6
+    },
+    {
+      "epoch": 5.786128171004885e-05,
+      "grad_norm": 1.2358893156051636,
+      "learning_rate": 4.999876011539193e-05,
+      "loss": 1.0112,
+      "mean_token_accuracy": 0.8141504526138306,
+      "num_tokens": 69044.0,
+      "step": 7
+    },
+    {
+      "epoch": 6.612717909719869e-05,
+      "grad_norm": 1.2391976118087769,
+      "learning_rate": 4.999855346795725e-05,
+      "loss": 0.8735,
+      "mean_token_accuracy": 0.8337622284889221,
+      "num_tokens": 70993.0,
+      "step": 8
+    },
+    {
+      "epoch": 7.439307648434852e-05,
+      "grad_norm": 1.2356516122817993,
+      "learning_rate": 4.9998346820522576e-05,
+      "loss": 0.8059,
+      "mean_token_accuracy": 0.8553104400634766,
+      "num_tokens": 72948.0,
+      "step": 9
+    },
+    {
+      "epoch": 8.265897387149836e-05,
+      "grad_norm": 1.2190320491790771,
+      "learning_rate": 4.999814017308789e-05,
+      "loss": 0.8336,
+      "mean_token_accuracy": 0.8362318873405457,
+      "num_tokens": 75024.0,
+      "step": 10
+    },
+    {
+      "epoch": 9.09248712586482e-05,
+      "grad_norm": 1.156231164932251,
+      "learning_rate": 4.999793352565321e-05,
+      "loss": 0.7332,
+      "mean_token_accuracy": 0.8614157438278198,
+      "num_tokens": 77036.0,
+      "step": 11
+    },
+    {
+      "epoch": 9.919076864579803e-05,
+      "grad_norm": 1.14757239818573,
+      "learning_rate": 4.9997726878218534e-05,
+      "loss": 0.6888,
+      "mean_token_accuracy": 0.8694362044334412,
+      "num_tokens": 79064.0,
+      "step": 12
+    },
+    {
+      "epoch": 0.00010745666603294786,
+      "grad_norm": 1.2405941486358643,
+      "learning_rate": 4.9997520230783856e-05,
+      "loss": 0.5499,
+      "mean_token_accuracy": 0.9004085659980774,
+      "num_tokens": 81028.0,
+      "step": 13
+    },
+    {
+      "epoch": 0.0001157225634200977,
+      "grad_norm": 1.1322728395462036,
+      "learning_rate": 4.9997313583349184e-05,
+      "loss": 0.5386,
+      "mean_token_accuracy": 0.898645281791687,
+      "num_tokens": 83027.0,
+      "step": 14
+    },
+    {
+      "epoch": 0.00012398846080724754,
+      "grad_norm": 1.1357864141464233,
+      "learning_rate": 4.99971069359145e-05,
+      "loss": 0.4413,
+      "mean_token_accuracy": 0.9220713973045349,
+      "num_tokens": 85022.0,
+      "step": 15
+    },
+    {
+      "epoch": 0.00013225435819439737,
+      "grad_norm": 1.273759126663208,
+      "learning_rate": 4.999690028847982e-05,
+      "loss": 0.5348,
+      "mean_token_accuracy": 0.91167813539505,
+      "num_tokens": 87066.0,
+      "step": 16
+    },
+    {
+      "epoch": 0.0001405202555815472,
+      "grad_norm": 0.975348174571991,
+      "learning_rate": 4.999669364104514e-05,
+      "loss": 0.3922,
+      "mean_token_accuracy": 0.935516357421875,
+      "num_tokens": 89057.0,
+      "step": 17
+    },
+    {
+      "epoch": 0.00014878615296869704,
+      "grad_norm": 1.0392197370529175,
+      "learning_rate": 4.9996486993610464e-05,
+      "loss": 0.5085,
+      "mean_token_accuracy": 0.9096071720123291,
+      "num_tokens": 91176.0,
+      "step": 18
+    },
+    {
+      "epoch": 0.00015705205035584688,
+      "grad_norm": 0.7911114692687988,
+      "learning_rate": 4.9996280346175785e-05,
+      "loss": 0.3615,
+      "mean_token_accuracy": 0.9434539079666138,
+      "num_tokens": 93145.0,
+      "step": 19
+    },
+    {
+      "epoch": 0.00016531794774299672,
+      "grad_norm": 0.7152270078659058,
+      "learning_rate": 4.999607369874111e-05,
+      "loss": 0.3766,
+      "mean_token_accuracy": 0.939257025718689,
+      "num_tokens": 95143.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.00017358384513014655,
+      "grad_norm": 0.6786302328109741,
+      "learning_rate": 4.999586705130643e-05,
+      "loss": 0.3196,
+      "mean_token_accuracy": 0.9410868287086487,
+      "num_tokens": 97118.0,
+      "step": 21
+    },
+    {
+      "epoch": 0.0001818497425172964,
+      "grad_norm": 0.7003538012504578,
+      "learning_rate": 4.9995660403871744e-05,
+      "loss": 0.3482,
+      "mean_token_accuracy": 0.9361808896064758,
+      "num_tokens": 99114.0,
+      "step": 22
+    },
+    {
+      "epoch": 0.00019011563990444622,
+      "grad_norm": 0.6917778849601746,
+      "learning_rate": 4.999545375643707e-05,
+      "loss": 0.5061,
+      "mean_token_accuracy": 0.9058039784431458,
+      "num_tokens": 101222.0,
+      "step": 23
+    },
+    {
+      "epoch": 0.00019838153729159606,
+      "grad_norm": 0.5567969083786011,
+      "learning_rate": 4.9995247109002394e-05,
+      "loss": 0.3651,
+      "mean_token_accuracy": 0.9303579926490784,
+      "num_tokens": 103267.0,
+      "step": 24
+    },
+    {
+      "epoch": 0.0002066474346787459,
+      "grad_norm": 0.562647819519043,
+      "learning_rate": 4.999504046156771e-05,
+      "loss": 0.3227,
+      "mean_token_accuracy": 0.9463834166526794,
+      "num_tokens": 105250.0,
+      "step": 25
+    },
+    {
+      "epoch": 0.00021491333206589573,
+      "grad_norm": 0.6241666078567505,
+      "learning_rate": 4.999483381413304e-05,
+      "loss": 0.3877,
+      "mean_token_accuracy": 0.9294871687889099,
+      "num_tokens": 107284.0,
+      "step": 26
+    },
+    {
+      "epoch": 0.00022317922945304557,
+      "grad_norm": 0.5423364043235779,
+      "learning_rate": 4.999462716669835e-05,
+      "loss": 0.3916,
+      "mean_token_accuracy": 0.9313241243362427,
+      "num_tokens": 109314.0,
+      "step": 27
+    },
+    {
+      "epoch": 0.0002314451268401954,
+      "grad_norm": 0.6723122000694275,
+      "learning_rate": 4.9994420519263674e-05,
+      "loss": 0.5012,
+      "mean_token_accuracy": 0.9086124300956726,
+      "num_tokens": 111410.0,
+      "step": 28
+    },
+    {
+      "epoch": 0.00023971102422734524,
+      "grad_norm": 0.5182911157608032,
+      "learning_rate": 4.9994213871829e-05,
+      "loss": 0.4172,
+      "mean_token_accuracy": 0.9263360500335693,
+      "num_tokens": 113493.0,
+      "step": 29
+    },
+    {
+      "epoch": 0.00024797692161449507,
+      "grad_norm": 0.5118226408958435,
+      "learning_rate": 4.999400722439432e-05,
+      "loss": 0.376,
+      "mean_token_accuracy": 0.9312561750411987,
+      "num_tokens": 115521.0,
+      "step": 30
+    },
+    {
+      "epoch": 0.00025624281900164494,
+      "grad_norm": 0.45296454429626465,
+      "learning_rate": 4.999380057695964e-05,
+      "loss": 0.3249,
+      "mean_token_accuracy": 0.9384692311286926,
+      "num_tokens": 117526.0,
+      "step": 31
+    },
+    {
+      "epoch": 0.00026450871638879474,
+      "grad_norm": 0.48847129940986633,
+      "learning_rate": 4.999359392952496e-05,
+      "loss": 0.306,
+      "mean_token_accuracy": 0.9487309455871582,
+      "num_tokens": 119502.0,
+      "step": 32
+    },
+    {
+      "epoch": 0.0002727746137759446,
+      "grad_norm": 0.410657674074173,
+      "learning_rate": 4.999338728209028e-05,
+      "loss": 0.3191,
+      "mean_token_accuracy": 0.9454081654548645,
+      "num_tokens": 121468.0,
+      "step": 33
+    },
+    {
+      "epoch": 0.0002810405111630944,
+      "grad_norm": 0.45754924416542053,
+      "learning_rate": 4.9993180634655604e-05,
+      "loss": 0.2761,
+      "mean_token_accuracy": 0.9495155811309814,
+      "num_tokens": 123435.0,
+      "step": 34
+    },
+    {
+      "epoch": 0.0002893064085502443,
+      "grad_norm": 0.46412602066993713,
+      "learning_rate": 4.9992973987220925e-05,
+      "loss": 0.4123,
+      "mean_token_accuracy": 0.9186046719551086,
+      "num_tokens": 125505.0,
+      "step": 35
+    },
+    {
+      "epoch": 0.0002975723059373941,
+      "grad_norm": 0.41713273525238037,
+      "learning_rate": 4.999276733978625e-05,
+      "loss": 0.3499,
+      "mean_token_accuracy": 0.9345980882644653,
+      "num_tokens": 127514.0,
+      "step": 36
+    },
+    {
+      "epoch": 0.00030583820332454395,
+      "grad_norm": 0.3865780830383301,
+      "learning_rate": 4.999256069235157e-05,
+      "loss": 0.2502,
+      "mean_token_accuracy": 0.9512320160865784,
+      "num_tokens": 129468.0,
+      "step": 37
+    },
+    {
+      "epoch": 0.00031410410071169376,
+      "grad_norm": 0.3944219946861267,
+      "learning_rate": 4.999235404491689e-05,
+      "loss": 0.2821,
+      "mean_token_accuracy": 0.9468845725059509,
+      "num_tokens": 131432.0,
+      "step": 38
+    },
+    {
+      "epoch": 0.0003223699980988436,
+      "grad_norm": 0.38811054825782776,
+      "learning_rate": 4.999214739748221e-05,
+      "loss": 0.3101,
+      "mean_token_accuracy": 0.9409113526344299,
+      "num_tokens": 133435.0,
+      "step": 39
+    },
+    {
+      "epoch": 0.00033063589548599343,
+      "grad_norm": 0.6454439759254456,
+      "learning_rate": 4.9991940750047533e-05,
+      "loss": 0.4536,
+      "mean_token_accuracy": 0.9109621644020081,
+      "num_tokens": 135530.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.0003389017928731433,
+      "grad_norm": 0.44250041246414185,
+      "learning_rate": 4.9991734102612855e-05,
+      "loss": 0.3279,
+      "mean_token_accuracy": 0.9331695437431335,
+      "num_tokens": 137571.0,
+      "step": 41
+    },
+    {
+      "epoch": 0.0003471676902602931,
+      "grad_norm": 0.36981847882270813,
+      "learning_rate": 4.999152745517817e-05,
+      "loss": 0.294,
+      "mean_token_accuracy": 0.9497206807136536,
+      "num_tokens": 139546.0,
+      "step": 42
+    },
+    {
+      "epoch": 0.00035543358764744296,
+      "grad_norm": 0.4224010407924652,
+      "learning_rate": 4.99913208077435e-05,
+      "loss": 0.3451,
+      "mean_token_accuracy": 0.936873733997345,
+      "num_tokens": 141548.0,
+      "step": 43
+    },
+    {
+      "epoch": 0.0003636994850345928,
+      "grad_norm": 0.4885828197002411,
+      "learning_rate": 4.999111416030881e-05,
+      "loss": 0.3556,
+      "mean_token_accuracy": 0.9327442049980164,
+      "num_tokens": 143591.0,
+      "step": 44
+    },
+    {
+      "epoch": 0.00037196538242174264,
+      "grad_norm": 0.48182302713394165,
+      "learning_rate": 4.9990907512874135e-05,
+      "loss": 0.355,
+      "mean_token_accuracy": 0.9366834163665771,
+      "num_tokens": 145587.0,
+      "step": 45
+    },
+    {
+      "epoch": 0.00038023127980889244,
+      "grad_norm": 0.5360848903656006,
+      "learning_rate": 4.9990700865439463e-05,
+      "loss": 0.3397,
+      "mean_token_accuracy": 0.9344504475593567,
+      "num_tokens": 147622.0,
+      "step": 46
+    },
+    {
+      "epoch": 0.0003884971771960423,
+      "grad_norm": 0.5050182342529297,
+      "learning_rate": 4.999049421800478e-05,
+      "loss": 0.399,
+      "mean_token_accuracy": 0.9243902564048767,
+      "num_tokens": 149678.0,
+      "step": 47
+    },
+    {
+      "epoch": 0.0003967630745831921,
+      "grad_norm": 0.362700492143631,
+      "learning_rate": 4.99902875705701e-05,
+      "loss": 0.2671,
+      "mean_token_accuracy": 0.953109085559845,
+      "num_tokens": 151646.0,
+      "step": 48
+    },
+    {
+      "epoch": 0.000405028971970342,
+      "grad_norm": 0.37579265236854553,
+      "learning_rate": 4.999008092313542e-05,
+      "loss": 0.2596,
+      "mean_token_accuracy": 0.950276255607605,
+      "num_tokens": 153643.0,
+      "step": 49
+    },
+    {
+      "epoch": 0.0004132948693574918,
+      "grad_norm": 0.5673742294311523,
+      "learning_rate": 4.998987427570074e-05,
+      "loss": 0.5196,
+      "mean_token_accuracy": 0.8972570896148682,
+      "num_tokens": 155800.0,
+      "step": 50
+    },
+    {
+      "epoch": 0.00042156076674464165,
+      "grad_norm": 0.4164047837257385,
+      "learning_rate": 4.9989667628266065e-05,
+      "loss": 0.296,
+      "mean_token_accuracy": 0.9427998065948486,
+      "num_tokens": 157799.0,
+      "step": 51
+    },
+    {
+      "epoch": 0.00042982666413179146,
+      "grad_norm": 0.4306670129299164,
+      "learning_rate": 4.9989460980831387e-05,
+      "loss": 0.3072,
+      "mean_token_accuracy": 0.9397410154342651,
+      "num_tokens": 159813.0,
+      "step": 52
+    },
+    {
+      "epoch": 0.0004380925615189413,
+      "grad_norm": 0.4390345513820648,
+      "learning_rate": 4.998925433339671e-05,
+      "loss": 0.4327,
+      "mean_token_accuracy": 0.9159461855888367,
+      "num_tokens": 161901.0,
+      "step": 53
+    },
+    {
+      "epoch": 0.00044635845890609113,
+      "grad_norm": 0.41451117396354675,
+      "learning_rate": 4.998904768596203e-05,
+      "loss": 0.3278,
+      "mean_token_accuracy": 0.9379921555519104,
+      "num_tokens": 163939.0,
+      "step": 54
+    },
+    {
+      "epoch": 0.000454624356293241,
+      "grad_norm": 0.37657612562179565,
+      "learning_rate": 4.998884103852735e-05,
+      "loss": 0.2581,
+      "mean_token_accuracy": 0.9523565769195557,
+      "num_tokens": 165897.0,
+      "step": 55
+    },
+    {
+      "epoch": 0.0004628902536803908,
+      "grad_norm": 0.499775767326355,
+      "learning_rate": 4.998863439109267e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.9381546378135681,
+      "num_tokens": 167908.0,
+      "step": 56
+    },
+    {
+      "epoch": 0.00047115615106754067,
+      "grad_norm": 0.43703413009643555,
+      "learning_rate": 4.9988427743657995e-05,
+      "loss": 0.2726,
+      "mean_token_accuracy": 0.9463291168212891,
+      "num_tokens": 169889.0,
+      "step": 57
+    },
+    {
+      "epoch": 0.0004794220484546905,
+      "grad_norm": 0.4889462888240814,
+      "learning_rate": 4.9988221096223317e-05,
+      "loss": 0.2975,
+      "mean_token_accuracy": 0.9409409165382385,
+      "num_tokens": 171893.0,
+      "step": 58
+    },
+    {
+      "epoch": 0.00048768794584184034,
+      "grad_norm": 0.37268808484077454,
+      "learning_rate": 4.998801444878863e-05,
+      "loss": 0.2948,
+      "mean_token_accuracy": 0.9461499452590942,
+      "num_tokens": 173886.0,
+      "step": 59
+    },
+    {
+      "epoch": 0.0004959538432289901,
+      "grad_norm": 0.36821237206459045,
+      "learning_rate": 4.998780780135396e-05,
+      "loss": 0.2831,
+      "mean_token_accuracy": 0.9416205286979675,
+      "num_tokens": 175879.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.00050421974061614,
+      "grad_norm": 0.4337264895439148,
+      "learning_rate": 4.9987601153919275e-05,
+      "loss": 0.3373,
+      "mean_token_accuracy": 0.9344181418418884,
+      "num_tokens": 177913.0,
+      "step": 61
+    },
+    {
+      "epoch": 0.0005124856380032899,
+      "grad_norm": 0.4724893271923065,
+      "learning_rate": 4.9987394506484596e-05,
+      "loss": 0.3336,
+      "mean_token_accuracy": 0.9384765625,
+      "num_tokens": 179967.0,
+      "step": 62
+    },
+    {
+      "epoch": 0.0005207515353904396,
+      "grad_norm": 0.4221535623073578,
+      "learning_rate": 4.9987187859049925e-05,
+      "loss": 0.2924,
+      "mean_token_accuracy": 0.9482412338256836,
+      "num_tokens": 181963.0,
+      "step": 63
+    },
+    {
+      "epoch": 0.0005290174327775895,
+      "grad_norm": 0.38894620537757874,
+      "learning_rate": 4.998698121161524e-05,
+      "loss": 0.2423,
+      "mean_token_accuracy": 0.9458646774291992,
+      "num_tokens": 183964.0,
+      "step": 64
+    },
+    {
+      "epoch": 0.0005372833301647394,
+      "grad_norm": 0.38573014736175537,
+      "learning_rate": 4.998677456418056e-05,
+      "loss": 0.302,
+      "mean_token_accuracy": 0.9441930651664734,
+      "num_tokens": 185959.0,
+      "step": 65
+    },
+    {
+      "epoch": 0.0005455492275518892,
+      "grad_norm": 0.5437004566192627,
+      "learning_rate": 4.998656791674588e-05,
+      "loss": 0.4231,
+      "mean_token_accuracy": 0.9245103001594543,
+      "num_tokens": 188058.0,
+      "step": 66
+    },
+    {
+      "epoch": 0.000553815124939039,
+      "grad_norm": 0.4524412751197815,
+      "learning_rate": 4.9986361269311205e-05,
+      "loss": 0.3121,
+      "mean_token_accuracy": 0.9383084774017334,
+      "num_tokens": 190074.0,
+      "step": 67
+    },
+    {
+      "epoch": 0.0005620810223261888,
+      "grad_norm": 0.3265776038169861,
+      "learning_rate": 4.9986154621876526e-05,
+      "loss": 0.2249,
+      "mean_token_accuracy": 0.9607235193252563,
+      "num_tokens": 192015.0,
+      "step": 68
+    },
+    {
+      "epoch": 0.0005703469197133387,
+      "grad_norm": 0.42387452721595764,
+      "learning_rate": 4.998594797444185e-05,
+      "loss": 0.2704,
+      "mean_token_accuracy": 0.9471032619476318,
+      "num_tokens": 194006.0,
+      "step": 69
+    },
+    {
+      "epoch": 0.0005786128171004886,
+      "grad_norm": 0.3408187925815582,
+      "learning_rate": 4.998574132700717e-05,
+      "loss": 0.221,
+      "mean_token_accuracy": 0.9580777287483215,
+      "num_tokens": 195968.0,
+      "step": 70
+    },
+    {
+      "epoch": 0.0005868787144876383,
+      "grad_norm": 0.34136003255844116,
+      "learning_rate": 4.9985534679572485e-05,
+      "loss": 0.2104,
+      "mean_token_accuracy": 0.9610854983329773,
+      "num_tokens": 197927.0,
+      "step": 71
+    },
+    {
+      "epoch": 0.0005951446118747882,
+      "grad_norm": 0.4188866913318634,
+      "learning_rate": 4.998532803213781e-05,
+      "loss": 0.325,
+      "mean_token_accuracy": 0.9374090433120728,
+      "num_tokens": 199994.0,
+      "step": 72
+    },
+    {
+      "epoch": 0.000603410509261938,
+      "grad_norm": 0.361499547958374,
+      "learning_rate": 4.9985121384703135e-05,
+      "loss": 0.2085,
+      "mean_token_accuracy": 0.9612001776695251,
+      "num_tokens": 201933.0,
+      "step": 73
+    },
+    {
+      "epoch": 0.0006116764066490879,
+      "grad_norm": 0.37521591782569885,
+      "learning_rate": 4.998491473726845e-05,
+      "loss": 0.2376,
+      "mean_token_accuracy": 0.9527918696403503,
+      "num_tokens": 203909.0,
+      "step": 74
+    },
+    {
+      "epoch": 0.0006199423040362377,
+      "grad_norm": 0.3477654755115509,
+      "learning_rate": 4.998470808983378e-05,
+      "loss": 0.2242,
+      "mean_token_accuracy": 0.9529827833175659,
+      "num_tokens": 205893.0,
+      "step": 75
+    },
+    {
+      "epoch": 0.0006282082014233875,
+      "grad_norm": 0.3677995800971985,
+      "learning_rate": 4.998450144239909e-05,
+      "loss": 0.2633,
+      "mean_token_accuracy": 0.9475786089897156,
+      "num_tokens": 207902.0,
+      "step": 76
+    },
+    {
+      "epoch": 0.0006364740988105374,
+      "grad_norm": 0.4442133605480194,
+      "learning_rate": 4.998429479496442e-05,
+      "loss": 0.3044,
+      "mean_token_accuracy": 0.9407407641410828,
+      "num_tokens": 209933.0,
+      "step": 77
+    },
+    {
+      "epoch": 0.0006447399961976872,
+      "grad_norm": 0.37177133560180664,
+      "learning_rate": 4.998408814752974e-05,
+      "loss": 0.2272,
+      "mean_token_accuracy": 0.9543859362602234,
+      "num_tokens": 211934.0,
+      "step": 78
+    },
+    {
+      "epoch": 0.000653005893584837,
+      "grad_norm": 0.41530996561050415,
+      "learning_rate": 4.998388150009506e-05,
+      "loss": 0.2895,
+      "mean_token_accuracy": 0.9417910575866699,
+      "num_tokens": 213950.0,
+      "step": 79
+    },
+    {
+      "epoch": 0.0006612717909719869,
+      "grad_norm": 0.43269816040992737,
+      "learning_rate": 4.9983674852660386e-05,
+      "loss": 0.2512,
+      "mean_token_accuracy": 0.9486419558525085,
+      "num_tokens": 215981.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.0006695376883591367,
+      "grad_norm": 0.4208819270133972,
+      "learning_rate": 4.99834682052257e-05,
+      "loss": 0.2453,
+      "mean_token_accuracy": 0.9551411271095276,
+      "num_tokens": 217971.0,
+      "step": 81
+    },
+    {
+      "epoch": 0.0006778035857462866,
+      "grad_norm": 0.4460567533969879,
+      "learning_rate": 4.998326155779102e-05,
+      "loss": 0.2333,
+      "mean_token_accuracy": 0.9548223614692688,
+      "num_tokens": 219947.0,
+      "step": 82
+    },
+    {
+      "epoch": 0.0006860694831334363,
+      "grad_norm": 0.4036431312561035,
+      "learning_rate": 4.9983054910356344e-05,
+      "loss": 0.2905,
+      "mean_token_accuracy": 0.9427290558815002,
+      "num_tokens": 221961.0,
+      "step": 83
+    },
+    {
+      "epoch": 0.0006943353805205862,
+      "grad_norm": 0.3974411189556122,
+      "learning_rate": 4.9982848262921666e-05,
+      "loss": 0.2417,
+      "mean_token_accuracy": 0.9485628008842468,
+      "num_tokens": 223950.0,
+      "step": 84
+    },
+    {
+      "epoch": 0.0007026012779077361,
+      "grad_norm": 0.5355939269065857,
+      "learning_rate": 4.998264161548699e-05,
+      "loss": 0.3876,
+      "mean_token_accuracy": 0.9247830510139465,
+      "num_tokens": 226030.0,
+      "step": 85
+    },
+    {
+      "epoch": 0.0007108671752948859,
+      "grad_norm": 0.3597177565097809,
+      "learning_rate": 4.998243496805231e-05,
+      "loss": 0.2384,
+      "mean_token_accuracy": 0.9443888068199158,
+      "num_tokens": 228032.0,
+      "step": 86
+    },
+    {
+      "epoch": 0.0007191330726820357,
+      "grad_norm": 0.37672877311706543,
+      "learning_rate": 4.998222832061763e-05,
+      "loss": 0.2173,
+      "mean_token_accuracy": 0.9556122422218323,
+      "num_tokens": 229998.0,
+      "step": 87
+    },
+    {
+      "epoch": 0.0007273989700691855,
+      "grad_norm": 0.35750794410705566,
+      "learning_rate": 4.998202167318295e-05,
+      "loss": 0.2002,
+      "mean_token_accuracy": 0.9600409865379333,
+      "num_tokens": 231956.0,
+      "step": 88
+    },
+    {
+      "epoch": 0.0007356648674563354,
+      "grad_norm": 0.5296061038970947,
+      "learning_rate": 4.9981815025748274e-05,
+      "loss": 0.3698,
+      "mean_token_accuracy": 0.9266890287399292,
+      "num_tokens": 234049.0,
+      "step": 89
+    },
+    {
+      "epoch": 0.0007439307648434853,
+      "grad_norm": 0.36901018023490906,
+      "learning_rate": 4.9981608378313596e-05,
+      "loss": 0.2566,
+      "mean_token_accuracy": 0.9466532468795776,
+      "num_tokens": 236042.0,
+      "step": 90
+    },
+    {
+      "epoch": 0.000752196662230635,
+      "grad_norm": 0.3750539720058441,
+      "learning_rate": 4.998140173087891e-05,
+      "loss": 0.2034,
+      "mean_token_accuracy": 0.9589457511901855,
+      "num_tokens": 238021.0,
+      "step": 91
+    },
+    {
+      "epoch": 0.0007604625596177849,
+      "grad_norm": 0.41055604815483093,
+      "learning_rate": 4.998119508344424e-05,
+      "loss": 0.3369,
+      "mean_token_accuracy": 0.9331087470054626,
+      "num_tokens": 240105.0,
+      "step": 92
+    },
+    {
+      "epoch": 0.0007687284570049348,
+      "grad_norm": 0.4907441735267639,
+      "learning_rate": 4.9980988436009554e-05,
+      "loss": 0.2908,
+      "mean_token_accuracy": 0.9406234622001648,
+      "num_tokens": 242132.0,
+      "step": 93
+    },
+    {
+      "epoch": 0.0007769943543920846,
+      "grad_norm": 0.44093140959739685,
+      "learning_rate": 4.9980781788574876e-05,
+      "loss": 0.3768,
+      "mean_token_accuracy": 0.9267578125,
+      "num_tokens": 244186.0,
+      "step": 94
+    },
+    {
+      "epoch": 0.0007852602517792344,
+      "grad_norm": 0.3314705789089203,
+      "learning_rate": 4.9980575141140204e-05,
+      "loss": 0.2209,
+      "mean_token_accuracy": 0.9508363008499146,
+      "num_tokens": 246165.0,
+      "step": 95
+    },
+    {
+      "epoch": 0.0007935261491663842,
+      "grad_norm": 0.43607860803604126,
+      "learning_rate": 4.998036849370552e-05,
+      "loss": 0.2969,
+      "mean_token_accuracy": 0.9427005648612976,
+      "num_tokens": 248178.0,
+      "step": 96
+    },
+    {
+      "epoch": 0.0008017920465535341,
+      "grad_norm": 0.32071632146835327,
+      "learning_rate": 4.998016184627085e-05,
+      "loss": 0.1694,
+      "mean_token_accuracy": 0.9639732241630554,
+      "num_tokens": 250127.0,
+      "step": 97
+    },
+    {
+      "epoch": 0.000810057943940684,
+      "grad_norm": 0.39200201630592346,
+      "learning_rate": 4.997995519883616e-05,
+      "loss": 0.2925,
+      "mean_token_accuracy": 0.9392156600952148,
+      "num_tokens": 252173.0,
+      "step": 98
+    },
+    {
+      "epoch": 0.0008183238413278337,
+      "grad_norm": 0.3622921109199524,
+      "learning_rate": 4.9979748551401484e-05,
+      "loss": 0.2952,
+      "mean_token_accuracy": 0.942392885684967,
+      "num_tokens": 254210.0,
+      "step": 99
+    },
+    {
+      "epoch": 0.0008265897387149836,
+      "grad_norm": 0.3674699366092682,
+      "learning_rate": 4.9979541903966806e-05,
+      "loss": 0.2339,
+      "mean_token_accuracy": 0.9484999775886536,
+      "num_tokens": 256216.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.0008348556361021334,
+      "grad_norm": 0.3341938555240631,
+      "learning_rate": 4.997933525653213e-05,
+      "loss": 0.1883,
+      "mean_token_accuracy": 0.9595701098442078,
+      "num_tokens": 258176.0,
+      "step": 101
+    },
+    {
+      "epoch": 0.0008431215334892833,
+      "grad_norm": 0.4190879166126251,
+      "learning_rate": 4.997912860909745e-05,
+      "loss": 0.3421,
+      "mean_token_accuracy": 0.929892897605896,
+      "num_tokens": 260236.0,
+      "step": 102
+    },
+    {
+      "epoch": 0.0008513874308764331,
+      "grad_norm": 0.3123447299003601,
+      "learning_rate": 4.997892196166277e-05,
+      "loss": 0.2194,
+      "mean_token_accuracy": 0.95489102602005,
+      "num_tokens": 262215.0,
+      "step": 103
+    },
+    {
+      "epoch": 0.0008596533282635829,
+      "grad_norm": 0.30452197790145874,
+      "learning_rate": 4.997871531422809e-05,
+      "loss": 0.1823,
+      "mean_token_accuracy": 0.9631742835044861,
+      "num_tokens": 264149.0,
+      "step": 104
+    },
+    {
+      "epoch": 0.0008679192256507328,
+      "grad_norm": 0.3034026026725769,
+      "learning_rate": 4.9978508666793414e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9628483057022095,
+      "num_tokens": 266093.0,
+      "step": 105
+    },
+    {
+      "epoch": 0.0008761851230378826,
+      "grad_norm": 0.4742376506328583,
+      "learning_rate": 4.9978302019358736e-05,
+      "loss": 0.2472,
+      "mean_token_accuracy": 0.946107804775238,
+      "num_tokens": 268103.0,
+      "step": 106
+    },
+    {
+      "epoch": 0.0008844510204250324,
+      "grad_norm": 0.491089791059494,
+      "learning_rate": 4.997809537192406e-05,
+      "loss": 0.2678,
+      "mean_token_accuracy": 0.9438090324401855,
+      "num_tokens": 270120.0,
+      "step": 107
+    },
+    {
+      "epoch": 0.0008927169178121823,
+      "grad_norm": 0.4098670184612274,
+      "learning_rate": 4.997788872448937e-05,
+      "loss": 0.2075,
+      "mean_token_accuracy": 0.9583333134651184,
+      "num_tokens": 272118.0,
+      "step": 108
+    },
+    {
+      "epoch": 0.0009009828151993321,
+      "grad_norm": 0.3567214906215668,
+      "learning_rate": 4.99776820770547e-05,
+      "loss": 0.2323,
+      "mean_token_accuracy": 0.9554075002670288,
+      "num_tokens": 274075.0,
+      "step": 109
+    },
+    {
+      "epoch": 0.000909248712586482,
+      "grad_norm": 0.32733988761901855,
+      "learning_rate": 4.9977475429620016e-05,
+      "loss": 0.2044,
+      "mean_token_accuracy": 0.9544057250022888,
+      "num_tokens": 276033.0,
+      "step": 110
+    },
+    {
+      "epoch": 0.0009175146099736317,
+      "grad_norm": 0.44298064708709717,
+      "learning_rate": 4.997726878218534e-05,
+      "loss": 0.295,
+      "mean_token_accuracy": 0.942643404006958,
+      "num_tokens": 278044.0,
+      "step": 111
+    },
+    {
+      "epoch": 0.0009257805073607816,
+      "grad_norm": 0.4355528652667999,
+      "learning_rate": 4.9977062134750666e-05,
+      "loss": 0.2227,
+      "mean_token_accuracy": 0.956632673740387,
+      "num_tokens": 280010.0,
+      "step": 112
+    },
+    {
+      "epoch": 0.0009340464047479315,
+      "grad_norm": 0.4641549587249756,
+      "learning_rate": 4.997685548731598e-05,
+      "loss": 0.274,
+      "mean_token_accuracy": 0.9453085660934448,
+      "num_tokens": 282009.0,
+      "step": 113
+    },
+    {
+      "epoch": 0.0009423123021350813,
+      "grad_norm": 0.4358769655227661,
+      "learning_rate": 4.99766488398813e-05,
+      "loss": 0.2348,
+      "mean_token_accuracy": 0.9539006948471069,
+      "num_tokens": 283989.0,
+      "step": 114
+    },
+    {
+      "epoch": 0.0009505781995222311,
+      "grad_norm": 0.3793763518333435,
+      "learning_rate": 4.9976442192446624e-05,
+      "loss": 0.2267,
+      "mean_token_accuracy": 0.9523321390151978,
+      "num_tokens": 285946.0,
+      "step": 115
+    },
+    {
+      "epoch": 0.000958844096909381,
+      "grad_norm": 0.35620346665382385,
+      "learning_rate": 4.9976235545011946e-05,
+      "loss": 0.2093,
+      "mean_token_accuracy": 0.9576530456542969,
+      "num_tokens": 287912.0,
+      "step": 116
+    },
+    {
+      "epoch": 0.0009671099942965308,
+      "grad_norm": 0.4162020981311798,
+      "learning_rate": 4.997602889757727e-05,
+      "loss": 0.265,
+      "mean_token_accuracy": 0.9475786089897156,
+      "num_tokens": 289921.0,
+      "step": 117
+    },
+    {
+      "epoch": 0.0009753758916836807,
+      "grad_norm": 0.37774839997291565,
+      "learning_rate": 4.997582225014259e-05,
+      "loss": 0.2451,
+      "mean_token_accuracy": 0.9467604160308838,
+      "num_tokens": 291918.0,
+      "step": 118
+    },
+    {
+      "epoch": 0.0009836417890708305,
+      "grad_norm": 0.37172678112983704,
+      "learning_rate": 4.997561560270791e-05,
+      "loss": 0.1832,
+      "mean_token_accuracy": 0.9625641107559204,
+      "num_tokens": 293874.0,
+      "step": 119
+    },
+    {
+      "epoch": 0.0009919076864579803,
+      "grad_norm": 0.34664982557296753,
+      "learning_rate": 4.997540895527323e-05,
+      "loss": 0.2393,
+      "mean_token_accuracy": 0.9501259326934814,
+      "num_tokens": 295865.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.00100017358384513,
+      "grad_norm": 0.3998957574367523,
+      "learning_rate": 4.9975202307838554e-05,
+      "loss": 0.2463,
+      "mean_token_accuracy": 0.9456467032432556,
+      "num_tokens": 297858.0,
+      "step": 121
+    },
+    {
+      "epoch": 0.00100843948123228,
+      "grad_norm": 0.34160110354423523,
+      "learning_rate": 4.9974995660403875e-05,
+      "loss": 0.2268,
+      "mean_token_accuracy": 0.9513677954673767,
+      "num_tokens": 299838.0,
+      "step": 122
+    },
+    {
+      "epoch": 0.0010167053786194298,
+      "grad_norm": 0.4167325496673584,
+      "learning_rate": 4.99747890129692e-05,
+      "loss": 0.3688,
+      "mean_token_accuracy": 0.9249762296676636,
+      "num_tokens": 301950.0,
+      "step": 123
+    },
+    {
+      "epoch": 0.0010249712760065797,
+      "grad_norm": 0.3429563045501709,
+      "learning_rate": 4.997458236553452e-05,
+      "loss": 0.2127,
+      "mean_token_accuracy": 0.9597107172012329,
+      "num_tokens": 303892.0,
+      "step": 124
+    },
+    {
+      "epoch": 0.0010332371733937295,
+      "grad_norm": 0.46849849820137024,
+      "learning_rate": 4.9974375718099834e-05,
+      "loss": 0.3169,
+      "mean_token_accuracy": 0.939853310585022,
+      "num_tokens": 305943.0,
+      "step": 125
+    },
+    {
+      "epoch": 0.0010415030707808793,
+      "grad_norm": 0.4358327090740204,
+      "learning_rate": 4.997416907066516e-05,
+      "loss": 0.2923,
+      "mean_token_accuracy": 0.9409201145172119,
+      "num_tokens": 308014.0,
+      "step": 126
+    },
+    {
+      "epoch": 0.0010497689681680292,
+      "grad_norm": 0.40376272797584534,
+      "learning_rate": 4.997396242323048e-05,
+      "loss": 0.2575,
+      "mean_token_accuracy": 0.9454635381698608,
+      "num_tokens": 310037.0,
+      "step": 127
+    },
+    {
+      "epoch": 0.001058034865555179,
+      "grad_norm": 0.44060271978378296,
+      "learning_rate": 4.99737557757958e-05,
+      "loss": 0.2582,
+      "mean_token_accuracy": 0.9496728777885437,
+      "num_tokens": 312030.0,
+      "step": 128
+    },
+    {
+      "epoch": 0.0010663007629423287,
+      "grad_norm": 0.3269643783569336,
+      "learning_rate": 4.997354912836113e-05,
+      "loss": 0.2011,
+      "mean_token_accuracy": 0.9614213109016418,
+      "num_tokens": 314006.0,
+      "step": 129
+    },
+    {
+      "epoch": 0.0010745666603294787,
+      "grad_norm": 0.28902527689933777,
+      "learning_rate": 4.997334248092644e-05,
+      "loss": 0.1758,
+      "mean_token_accuracy": 0.9618514776229858,
+      "num_tokens": 315978.0,
+      "step": 130
+    },
+    {
+      "epoch": 0.0010828325577166285,
+      "grad_norm": 0.4907967746257782,
+      "learning_rate": 4.9973135833491764e-05,
+      "loss": 0.4331,
+      "mean_token_accuracy": 0.9163585901260376,
+      "num_tokens": 318148.0,
+      "step": 131
+    },
+    {
+      "epoch": 0.0010910984551037784,
+      "grad_norm": 0.44002026319503784,
+      "learning_rate": 4.9972929186057085e-05,
+      "loss": 0.3373,
+      "mean_token_accuracy": 0.9270488023757935,
+      "num_tokens": 320265.0,
+      "step": 132
+    },
+    {
+      "epoch": 0.0010993643524909282,
+      "grad_norm": 0.4253641664981842,
+      "learning_rate": 4.997272253862241e-05,
+      "loss": 0.3186,
+      "mean_token_accuracy": 0.9350268840789795,
+      "num_tokens": 322318.0,
+      "step": 133
+    },
+    {
+      "epoch": 0.001107630249878078,
+      "grad_norm": 0.47336700558662415,
+      "learning_rate": 4.997251589118773e-05,
+      "loss": 0.3062,
+      "mean_token_accuracy": 0.938339114189148,
+      "num_tokens": 324335.0,
+      "step": 134
+    },
+    {
+      "epoch": 0.001115896147265228,
+      "grad_norm": 0.4241746664047241,
+      "learning_rate": 4.997230924375305e-05,
+      "loss": 0.2258,
+      "mean_token_accuracy": 0.9517892599105835,
+      "num_tokens": 326353.0,
+      "step": 135
+    },
+    {
+      "epoch": 0.0011241620446523777,
+      "grad_norm": 0.41735950112342834,
+      "learning_rate": 4.997210259631837e-05,
+      "loss": 0.3644,
+      "mean_token_accuracy": 0.9337493777275085,
+      "num_tokens": 328442.0,
+      "step": 136
+    },
+    {
+      "epoch": 0.0011324279420395274,
+      "grad_norm": 0.3628050982952118,
+      "learning_rate": 4.9971895948883694e-05,
+      "loss": 0.2733,
+      "mean_token_accuracy": 0.9426877498626709,
+      "num_tokens": 330472.0,
+      "step": 137
+    },
+    {
+      "epoch": 0.0011406938394266774,
+      "grad_norm": 0.34173405170440674,
+      "learning_rate": 4.9971689301449015e-05,
+      "loss": 0.2499,
+      "mean_token_accuracy": 0.954017162322998,
+      "num_tokens": 332457.0,
+      "step": 138
+    },
+    {
+      "epoch": 0.0011489597368138271,
+      "grad_norm": 0.3606759309768677,
+      "learning_rate": 4.997148265401434e-05,
+      "loss": 0.2597,
+      "mean_token_accuracy": 0.9488752484321594,
+      "num_tokens": 334419.0,
+      "step": 139
+    },
+    {
+      "epoch": 0.0011572256342009771,
+      "grad_norm": 0.4132915735244751,
+      "learning_rate": 4.997127600657966e-05,
+      "loss": 0.2767,
+      "mean_token_accuracy": 0.9491778612136841,
+      "num_tokens": 336432.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.0011654915315881269,
+      "grad_norm": 0.4765280485153198,
+      "learning_rate": 4.997106935914498e-05,
+      "loss": 0.3741,
+      "mean_token_accuracy": 0.9291488528251648,
+      "num_tokens": 338541.0,
+      "step": 141
+    },
+    {
+      "epoch": 0.0011737574289752766,
+      "grad_norm": 0.4147401452064514,
+      "learning_rate": 4.9970862711710295e-05,
+      "loss": 0.2949,
+      "mean_token_accuracy": 0.9372905492782593,
+      "num_tokens": 340636.0,
+      "step": 142
+    },
+    {
+      "epoch": 0.0011820233263624266,
+      "grad_norm": 0.4276408553123474,
+      "learning_rate": 4.9970656064275624e-05,
+      "loss": 0.2794,
+      "mean_token_accuracy": 0.9426189064979553,
+      "num_tokens": 342681.0,
+      "step": 143
+    },
+    {
+      "epoch": 0.0011902892237495763,
+      "grad_norm": 0.32022973895072937,
+      "learning_rate": 4.9970449416840945e-05,
+      "loss": 0.1828,
+      "mean_token_accuracy": 0.9586101174354553,
+      "num_tokens": 344644.0,
+      "step": 144
+    },
+    {
+      "epoch": 0.001198555121136726,
+      "grad_norm": 0.5508199334144592,
+      "learning_rate": 4.997024276940626e-05,
+      "loss": 0.3904,
+      "mean_token_accuracy": 0.9209012389183044,
+      "num_tokens": 346736.0,
+      "step": 145
+    },
+    {
+      "epoch": 0.001206821018523876,
+      "grad_norm": 0.3436821699142456,
+      "learning_rate": 4.997003612197159e-05,
+      "loss": 0.2325,
+      "mean_token_accuracy": 0.9511826634407043,
+      "num_tokens": 348729.0,
+      "step": 146
+    },
+    {
+      "epoch": 0.0012150869159110258,
+      "grad_norm": 0.3437710404396057,
+      "learning_rate": 4.99698294745369e-05,
+      "loss": 0.2104,
+      "mean_token_accuracy": 0.958268940448761,
+      "num_tokens": 350676.0,
+      "step": 147
+    },
+    {
+      "epoch": 0.0012233528132981758,
+      "grad_norm": 0.3376191258430481,
+      "learning_rate": 4.9969622827102225e-05,
+      "loss": 0.1821,
+      "mean_token_accuracy": 0.9618163108825684,
+      "num_tokens": 352620.0,
+      "step": 148
+    },
+    {
+      "epoch": 0.0012316187106853256,
+      "grad_norm": 0.4335744082927704,
+      "learning_rate": 4.996941617966755e-05,
+      "loss": 0.3394,
+      "mean_token_accuracy": 0.930355429649353,
+      "num_tokens": 354708.0,
+      "step": 149
+    },
+    {
+      "epoch": 0.0012398846080724753,
+      "grad_norm": 0.3190179765224457,
+      "learning_rate": 4.996920953223287e-05,
+      "loss": 0.1968,
+      "mean_token_accuracy": 0.9601227045059204,
+      "num_tokens": 356670.0,
+      "step": 150
+    },
+    {
+      "epoch": 0.0012481505054596253,
+      "grad_norm": 0.37530186772346497,
+      "learning_rate": 4.996900288479819e-05,
+      "loss": 0.248,
+      "mean_token_accuracy": 0.9488208889961243,
+      "num_tokens": 358669.0,
+      "step": 151
+    },
+    {
+      "epoch": 0.001256416402846775,
+      "grad_norm": 0.4432967007160187,
+      "learning_rate": 4.996879623736351e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.9335918426513672,
+      "num_tokens": 360738.0,
+      "step": 152
+    },
+    {
+      "epoch": 0.0012646823002339248,
+      "grad_norm": 0.39701804518699646,
+      "learning_rate": 4.996858958992883e-05,
+      "loss": 0.3284,
+      "mean_token_accuracy": 0.9327852725982666,
+      "num_tokens": 362812.0,
+      "step": 153
+    },
+    {
+      "epoch": 0.0012729481976210748,
+      "grad_norm": 0.3877154588699341,
+      "learning_rate": 4.9968382942494155e-05,
+      "loss": 0.266,
+      "mean_token_accuracy": 0.9488089084625244,
+      "num_tokens": 364791.0,
+      "step": 154
+    },
+    {
+      "epoch": 0.0012812140950082245,
+      "grad_norm": 0.36334314942359924,
+      "learning_rate": 4.996817629505948e-05,
+      "loss": 0.2335,
+      "mean_token_accuracy": 0.9545224905014038,
+      "num_tokens": 366776.0,
+      "step": 155
+    },
+    {
+      "epoch": 0.0012894799923953745,
+      "grad_norm": 0.43625608086586,
+      "learning_rate": 4.99679696476248e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.9325950741767883,
+      "num_tokens": 368859.0,
+      "step": 156
+    },
+    {
+      "epoch": 0.0012977458897825242,
+      "grad_norm": 0.36922386288642883,
+      "learning_rate": 4.996776300019011e-05,
+      "loss": 0.2259,
+      "mean_token_accuracy": 0.9508113861083984,
+      "num_tokens": 370837.0,
+      "step": 157
+    },
+    {
+      "epoch": 0.001306011787169674,
+      "grad_norm": 0.4996206760406494,
+      "learning_rate": 4.996755635275544e-05,
+      "loss": 0.3793,
+      "mean_token_accuracy": 0.9264069199562073,
+      "num_tokens": 372922.0,
+      "step": 158
+    },
+    {
+      "epoch": 0.001314277684556824,
+      "grad_norm": 0.3880661427974701,
+      "learning_rate": 4.9967349705320756e-05,
+      "loss": 0.2457,
+      "mean_token_accuracy": 0.946446418762207,
+      "num_tokens": 374926.0,
+      "step": 159
+    },
+    {
+      "epoch": 0.0013225435819439737,
+      "grad_norm": 0.32943350076675415,
+      "learning_rate": 4.9967143057886085e-05,
+      "loss": 0.2156,
+      "mean_token_accuracy": 0.952858567237854,
+      "num_tokens": 376926.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.0013308094793311235,
+      "grad_norm": 0.3695654273033142,
+      "learning_rate": 4.9966936410451407e-05,
+      "loss": 0.2368,
+      "mean_token_accuracy": 0.9545681476593018,
+      "num_tokens": 378935.0,
+      "step": 161
+    },
+    {
+      "epoch": 0.0013390753767182734,
+      "grad_norm": 0.42351189255714417,
+      "learning_rate": 4.996672976301672e-05,
+      "loss": 0.2666,
+      "mean_token_accuracy": 0.9390485882759094,
+      "num_tokens": 380959.0,
+      "step": 162
+    },
+    {
+      "epoch": 0.0013473412741054232,
+      "grad_norm": 0.7745271921157837,
+      "learning_rate": 4.996652311558205e-05,
+      "loss": 0.219,
+      "mean_token_accuracy": 0.949999988079071,
+      "num_tokens": 382925.0,
+      "step": 163
+    },
+    {
+      "epoch": 0.0013556071714925732,
+      "grad_norm": 0.31670019030570984,
+      "learning_rate": 4.9966316468147365e-05,
+      "loss": 0.2137,
+      "mean_token_accuracy": 0.953406810760498,
+      "num_tokens": 384927.0,
+      "step": 164
+    },
+    {
+      "epoch": 0.001363873068879723,
+      "grad_norm": 0.32420817017555237,
+      "learning_rate": 4.9966109820712686e-05,
+      "loss": 0.1978,
+      "mean_token_accuracy": 0.9547534584999084,
+      "num_tokens": 386900.0,
+      "step": 165
+    },
+    {
+      "epoch": 0.0013721389662668727,
+      "grad_norm": 0.4137507379055023,
+      "learning_rate": 4.996590317327801e-05,
+      "loss": 0.2337,
+      "mean_token_accuracy": 0.9464373588562012,
+      "num_tokens": 388941.0,
+      "step": 166
+    },
+    {
+      "epoch": 0.0013804048636540227,
+      "grad_norm": 0.4905780553817749,
+      "learning_rate": 4.996569652584333e-05,
+      "loss": 0.2975,
+      "mean_token_accuracy": 0.9425173401832581,
+      "num_tokens": 390965.0,
+      "step": 167
+    },
+    {
+      "epoch": 0.0013886707610411724,
+      "grad_norm": 0.5064888000488281,
+      "learning_rate": 4.996548987840865e-05,
+      "loss": 0.372,
+      "mean_token_accuracy": 0.9281952977180481,
+      "num_tokens": 393060.0,
+      "step": 168
+    },
+    {
+      "epoch": 0.0013969366584283224,
+      "grad_norm": 0.3523234724998474,
+      "learning_rate": 4.996528323097397e-05,
+      "loss": 0.2051,
+      "mean_token_accuracy": 0.9609856009483337,
+      "num_tokens": 395014.0,
+      "step": 169
+    },
+    {
+      "epoch": 0.0014052025558154721,
+      "grad_norm": 0.34162983298301697,
+      "learning_rate": 4.9965076583539295e-05,
+      "loss": 0.2178,
+      "mean_token_accuracy": 0.9570055603981018,
+      "num_tokens": 396997.0,
+      "step": 170
+    },
+    {
+      "epoch": 0.0014134684532026219,
+      "grad_norm": 0.3736307621002197,
+      "learning_rate": 4.9964869936104616e-05,
+      "loss": 0.2185,
+      "mean_token_accuracy": 0.9555888175964355,
+      "num_tokens": 399007.0,
+      "step": 171
+    },
+    {
+      "epoch": 0.0014217343505897719,
+      "grad_norm": 0.3532184362411499,
+      "learning_rate": 4.996466328866994e-05,
+      "loss": 0.2133,
+      "mean_token_accuracy": 0.9600404500961304,
+      "num_tokens": 400990.0,
+      "step": 172
+    },
+    {
+      "epoch": 0.0014300002479769216,
+      "grad_norm": 0.4058968722820282,
+      "learning_rate": 4.996445664123526e-05,
+      "loss": 0.2618,
+      "mean_token_accuracy": 0.942643404006958,
+      "num_tokens": 403001.0,
+      "step": 173
+    },
+    {
+      "epoch": 0.0014382661453640714,
+      "grad_norm": 0.36838194727897644,
+      "learning_rate": 4.9964249993800575e-05,
+      "loss": 0.2952,
+      "mean_token_accuracy": 0.9451581239700317,
+      "num_tokens": 405031.0,
+      "step": 174
+    },
+    {
+      "epoch": 0.0014465320427512213,
+      "grad_norm": 0.4245535731315613,
+      "learning_rate": 4.99640433463659e-05,
+      "loss": 0.305,
+      "mean_token_accuracy": 0.9347721934318542,
+      "num_tokens": 407122.0,
+      "step": 175
+    },
+    {
+      "epoch": 0.001454797940138371,
+      "grad_norm": 0.35786592960357666,
+      "learning_rate": 4.996383669893122e-05,
+      "loss": 0.2508,
+      "mean_token_accuracy": 0.9473161101341248,
+      "num_tokens": 409140.0,
+      "step": 176
+    },
+    {
+      "epoch": 0.001463063837525521,
+      "grad_norm": 0.33934542536735535,
+      "learning_rate": 4.996363005149654e-05,
+      "loss": 0.2278,
+      "mean_token_accuracy": 0.952023983001709,
+      "num_tokens": 411147.0,
+      "step": 177
+    },
+    {
+      "epoch": 0.0014713297349126708,
+      "grad_norm": 0.36044228076934814,
+      "learning_rate": 4.996342340406187e-05,
+      "loss": 0.2051,
+      "mean_token_accuracy": 0.9536990523338318,
+      "num_tokens": 413140.0,
+      "step": 178
+    },
+    {
+      "epoch": 0.0014795956322998206,
+      "grad_norm": 0.31432509422302246,
+      "learning_rate": 4.996321675662718e-05,
+      "loss": 0.1774,
+      "mean_token_accuracy": 0.9571647047996521,
+      "num_tokens": 415107.0,
+      "step": 179
+    },
+    {
+      "epoch": 0.0014878615296869705,
+      "grad_norm": 0.43004900217056274,
+      "learning_rate": 4.996301010919251e-05,
+      "loss": 0.2464,
+      "mean_token_accuracy": 0.9493110179901123,
+      "num_tokens": 417145.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.0014961274270741203,
+      "grad_norm": 0.38657617568969727,
+      "learning_rate": 4.9962803461757826e-05,
+      "loss": 0.1988,
+      "mean_token_accuracy": 0.9582272171974182,
+      "num_tokens": 419114.0,
+      "step": 181
+    },
+    {
+      "epoch": 0.00150439332446127,
+      "grad_norm": 0.323732852935791,
+      "learning_rate": 4.996259681432315e-05,
+      "loss": 0.2199,
+      "mean_token_accuracy": 0.9575543403625488,
+      "num_tokens": 421099.0,
+      "step": 182
+    },
+    {
+      "epoch": 0.00151265922184842,
+      "grad_norm": 0.3862565755844116,
+      "learning_rate": 4.9962390166888476e-05,
+      "loss": 0.2615,
+      "mean_token_accuracy": 0.9473684430122375,
+      "num_tokens": 423138.0,
+      "step": 183
+    },
+    {
+      "epoch": 0.0015209251192355698,
+      "grad_norm": 0.44200780987739563,
+      "learning_rate": 4.996218351945379e-05,
+      "loss": 0.2477,
+      "mean_token_accuracy": 0.9512194991111755,
+      "num_tokens": 425153.0,
+      "step": 184
+    },
+    {
+      "epoch": 0.0015291910166227198,
+      "grad_norm": 0.527441680431366,
+      "learning_rate": 4.996197687201911e-05,
+      "loss": 0.2395,
+      "mean_token_accuracy": 0.9471624493598938,
+      "num_tokens": 427203.0,
+      "step": 185
+    },
+    {
+      "epoch": 0.0015374569140098695,
+      "grad_norm": 0.35225358605384827,
+      "learning_rate": 4.9961770224584434e-05,
+      "loss": 0.2079,
+      "mean_token_accuracy": 0.9567649960517883,
+      "num_tokens": 429175.0,
+      "step": 186
+    },
+    {
+      "epoch": 0.0015457228113970193,
+      "grad_norm": 0.4101203680038452,
+      "learning_rate": 4.9961563577149756e-05,
+      "loss": 0.2409,
+      "mean_token_accuracy": 0.9457171559333801,
+      "num_tokens": 431189.0,
+      "step": 187
+    },
+    {
+      "epoch": 0.0015539887087841692,
+      "grad_norm": 0.35431399941444397,
+      "learning_rate": 4.996135692971508e-05,
+      "loss": 0.2084,
+      "mean_token_accuracy": 0.956940233707428,
+      "num_tokens": 433169.0,
+      "step": 188
+    },
+    {
+      "epoch": 0.001562254606171319,
+      "grad_norm": 0.3679504692554474,
+      "learning_rate": 4.99611502822804e-05,
+      "loss": 0.1955,
+      "mean_token_accuracy": 0.9568489789962769,
+      "num_tokens": 435168.0,
+      "step": 189
+    },
+    {
+      "epoch": 0.0015705205035584687,
+      "grad_norm": 0.3775530755519867,
+      "learning_rate": 4.996094363484572e-05,
+      "loss": 0.2188,
+      "mean_token_accuracy": 0.9526686668395996,
+      "num_tokens": 437160.0,
+      "step": 190
+    },
+    {
+      "epoch": 0.0015787864009456187,
+      "grad_norm": 0.397073894739151,
+      "learning_rate": 4.9960736987411036e-05,
+      "loss": 0.2229,
+      "mean_token_accuracy": 0.9520719051361084,
+      "num_tokens": 439169.0,
+      "step": 191
+    },
+    {
+      "epoch": 0.0015870522983327685,
+      "grad_norm": 0.5781298279762268,
+      "learning_rate": 4.9960530339976364e-05,
+      "loss": 0.2773,
+      "mean_token_accuracy": 0.942460298538208,
+      "num_tokens": 441191.0,
+      "step": 192
+    },
+    {
+      "epoch": 0.0015953181957199184,
+      "grad_norm": 0.4482797086238861,
+      "learning_rate": 4.9960323692541686e-05,
+      "loss": 0.2893,
+      "mean_token_accuracy": 0.944767415523529,
+      "num_tokens": 443261.0,
+      "step": 193
+    },
+    {
+      "epoch": 0.0016035840931070682,
+      "grad_norm": 0.43917760252952576,
+      "learning_rate": 4.9960117045107e-05,
+      "loss": 0.296,
+      "mean_token_accuracy": 0.9437869787216187,
+      "num_tokens": 445295.0,
+      "step": 194
+    },
+    {
+      "epoch": 0.001611849990494218,
+      "grad_norm": 0.41199809312820435,
+      "learning_rate": 4.995991039767233e-05,
+      "loss": 0.257,
+      "mean_token_accuracy": 0.9435721039772034,
+      "num_tokens": 447339.0,
+      "step": 195
+    },
+    {
+      "epoch": 0.001620115887881368,
+      "grad_norm": 0.44698280096054077,
+      "learning_rate": 4.9959703750237644e-05,
+      "loss": 0.2319,
+      "mean_token_accuracy": 0.9520479440689087,
+      "num_tokens": 449347.0,
+      "step": 196
+    },
+    {
+      "epoch": 0.0016283817852685177,
+      "grad_norm": 0.3956550657749176,
+      "learning_rate": 4.9959497102802966e-05,
+      "loss": 0.2465,
+      "mean_token_accuracy": 0.9480712413787842,
+      "num_tokens": 451375.0,
+      "step": 197
+    },
+    {
+      "epoch": 0.0016366476826556674,
+      "grad_norm": 0.3713931441307068,
+      "learning_rate": 4.995929045536829e-05,
+      "loss": 0.2835,
+      "mean_token_accuracy": 0.9433040022850037,
+      "num_tokens": 453427.0,
+      "step": 198
+    },
+    {
+      "epoch": 0.0016449135800428174,
+      "grad_norm": 0.49990788102149963,
+      "learning_rate": 4.995908380793361e-05,
+      "loss": 0.353,
+      "mean_token_accuracy": 0.9331449866294861,
+      "num_tokens": 455557.0,
+      "step": 199
+    },
+    {
+      "epoch": 0.0016531794774299672,
+      "grad_norm": 0.40154170989990234,
+      "learning_rate": 4.995887716049893e-05,
+      "loss": 0.2324,
+      "mean_token_accuracy": 0.9506234526634216,
+      "num_tokens": 457568.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.0016614453748171171,
+      "grad_norm": 0.42626091837882996,
+      "learning_rate": 4.995867051306425e-05,
+      "loss": 0.2421,
+      "mean_token_accuracy": 0.9526656866073608,
+      "num_tokens": 459581.0,
+      "step": 201
+    },
+    {
+      "epoch": 0.0016697112722042669,
+      "grad_norm": 0.42444708943367004,
+      "learning_rate": 4.9958463865629574e-05,
+      "loss": 0.2724,
+      "mean_token_accuracy": 0.9428289532661438,
+      "num_tokens": 461616.0,
+      "step": 202
+    },
+    {
+      "epoch": 0.0016779771695914166,
+      "grad_norm": 0.3321168124675751,
+      "learning_rate": 4.9958257218194896e-05,
+      "loss": 0.2055,
+      "mean_token_accuracy": 0.9579533934593201,
+      "num_tokens": 463596.0,
+      "step": 203
+    },
+    {
+      "epoch": 0.0016862430669785666,
+      "grad_norm": 0.31841254234313965,
+      "learning_rate": 4.995805057076022e-05,
+      "loss": 0.1771,
+      "mean_token_accuracy": 0.9633401036262512,
+      "num_tokens": 465566.0,
+      "step": 204
+    },
+    {
+      "epoch": 0.0016945089643657164,
+      "grad_norm": 0.4629010856151581,
+      "learning_rate": 4.995784392332554e-05,
+      "loss": 0.3147,
+      "mean_token_accuracy": 0.9342294931411743,
+      "num_tokens": 467655.0,
+      "step": 205
+    },
+    {
+      "epoch": 0.0017027748617528661,
+      "grad_norm": 0.38947194814682007,
+      "learning_rate": 4.995763727589086e-05,
+      "loss": 0.2271,
+      "mean_token_accuracy": 0.9518255591392517,
+      "num_tokens": 469633.0,
+      "step": 206
+    },
+    {
+      "epoch": 0.001711040759140016,
+      "grad_norm": 0.39823102951049805,
+      "learning_rate": 4.995743062845618e-05,
+      "loss": 0.2338,
+      "mean_token_accuracy": 0.9496545195579529,
+      "num_tokens": 471665.0,
+      "step": 207
+    },
+    {
+      "epoch": 0.0017193066565271658,
+      "grad_norm": 0.4126001298427582,
+      "learning_rate": 4.99572239810215e-05,
+      "loss": 0.244,
+      "mean_token_accuracy": 0.9533705115318298,
+      "num_tokens": 473644.0,
+      "step": 208
+    },
+    {
+      "epoch": 0.0017275725539143158,
+      "grad_norm": 0.43648454546928406,
+      "learning_rate": 4.9957017333586826e-05,
+      "loss": 0.2394,
+      "mean_token_accuracy": 0.9514999985694885,
+      "num_tokens": 475650.0,
+      "step": 209
+    },
+    {
+      "epoch": 0.0017358384513014656,
+      "grad_norm": 0.4396474063396454,
+      "learning_rate": 4.995681068615215e-05,
+      "loss": 0.2683,
+      "mean_token_accuracy": 0.9454905986785889,
+      "num_tokens": 477674.0,
+      "step": 210
+    },
+    {
+      "epoch": 0.0017441043486886153,
+      "grad_norm": 0.389310747385025,
+      "learning_rate": 4.995660403871746e-05,
+      "loss": 0.2076,
+      "mean_token_accuracy": 0.957717776298523,
+      "num_tokens": 479643.0,
+      "step": 211
+    },
+    {
+      "epoch": 0.0017523702460757653,
+      "grad_norm": 0.5761370658874512,
+      "learning_rate": 4.995639739128279e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.9356555342674255,
+      "num_tokens": 481716.0,
+      "step": 212
+    },
+    {
+      "epoch": 0.001760636143462915,
+      "grad_norm": 0.4655229151248932,
+      "learning_rate": 4.9956190743848106e-05,
+      "loss": 0.2651,
+      "mean_token_accuracy": 0.947029709815979,
+      "num_tokens": 483742.0,
+      "step": 213
+    },
+    {
+      "epoch": 0.0017689020408500648,
+      "grad_norm": 0.44393202662467957,
+      "learning_rate": 4.995598409641343e-05,
+      "loss": 0.3115,
+      "mean_token_accuracy": 0.9448584318161011,
+      "num_tokens": 485761.0,
+      "step": 214
+    },
+    {
+      "epoch": 0.0017771679382372148,
+      "grad_norm": 0.3399536609649658,
+      "learning_rate": 4.995577744897875e-05,
+      "loss": 0.1942,
+      "mean_token_accuracy": 0.9593830108642578,
+      "num_tokens": 487712.0,
+      "step": 215
+    },
+    {
+      "epoch": 0.0017854338356243645,
+      "grad_norm": 0.44372278451919556,
+      "learning_rate": 4.995557080154407e-05,
+      "loss": 0.3806,
+      "mean_token_accuracy": 0.9252601861953735,
+      "num_tokens": 489832.0,
+      "step": 216
+    },
+    {
+      "epoch": 0.0017936997330115145,
+      "grad_norm": 0.3342445194721222,
+      "learning_rate": 4.995536415410939e-05,
+      "loss": 0.1948,
+      "mean_token_accuracy": 0.9573934674263,
+      "num_tokens": 491833.0,
+      "step": 217
+    },
+    {
+      "epoch": 0.0018019656303986642,
+      "grad_norm": 0.35612207651138306,
+      "learning_rate": 4.9955157506674714e-05,
+      "loss": 0.2363,
+      "mean_token_accuracy": 0.9495549201965332,
+      "num_tokens": 493861.0,
+      "step": 218
+    },
+    {
+      "epoch": 0.001810231527785814,
+      "grad_norm": 0.47954094409942627,
+      "learning_rate": 4.9954950859240036e-05,
+      "loss": 0.3813,
+      "mean_token_accuracy": 0.9208294153213501,
+      "num_tokens": 495989.0,
+      "step": 219
+    },
+    {
+      "epoch": 0.001818497425172964,
+      "grad_norm": 0.40823832154273987,
+      "learning_rate": 4.995474421180536e-05,
+      "loss": 0.2748,
+      "mean_token_accuracy": 0.9354679584503174,
+      "num_tokens": 498025.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.0018267633225601137,
+      "grad_norm": 0.40444284677505493,
+      "learning_rate": 4.995453756437068e-05,
+      "loss": 0.29,
+      "mean_token_accuracy": 0.9416996240615845,
+      "num_tokens": 500055.0,
+      "step": 221
+    },
+    {
+      "epoch": 0.0018350292199472635,
+      "grad_norm": 0.33328917622566223,
+      "learning_rate": 4.9954330916936e-05,
+      "loss": 0.2092,
+      "mean_token_accuracy": 0.9618473649024963,
+      "num_tokens": 502053.0,
+      "step": 222
+    },
+    {
+      "epoch": 0.0018432951173344135,
+      "grad_norm": 0.3845839202404022,
+      "learning_rate": 4.995412426950132e-05,
+      "loss": 0.1982,
+      "mean_token_accuracy": 0.9604862928390503,
+      "num_tokens": 504033.0,
+      "step": 223
+    },
+    {
+      "epoch": 0.0018515610147215632,
+      "grad_norm": 0.4249724745750427,
+      "learning_rate": 4.9953917622066644e-05,
+      "loss": 0.1636,
+      "mean_token_accuracy": 0.9634518027305603,
+      "num_tokens": 506009.0,
+      "step": 224
+    },
+    {
+      "epoch": 0.0018598269121087132,
+      "grad_norm": 0.46474647521972656,
+      "learning_rate": 4.995371097463196e-05,
+      "loss": 0.2951,
+      "mean_token_accuracy": 0.9388052225112915,
+      "num_tokens": 508074.0,
+      "step": 225
+    },
+    {
+      "epoch": 0.001868092809495863,
+      "grad_norm": 0.4238729774951935,
+      "learning_rate": 4.995350432719729e-05,
+      "loss": 0.2161,
+      "mean_token_accuracy": 0.9593908786773682,
+      "num_tokens": 510050.0,
+      "step": 226
+    },
+    {
+      "epoch": 0.0018763587068830127,
+      "grad_norm": 0.4334641396999359,
+      "learning_rate": 4.995329767976261e-05,
+      "loss": 0.2622,
+      "mean_token_accuracy": 0.9480968713760376,
+      "num_tokens": 512079.0,
+      "step": 227
+    },
+    {
+      "epoch": 0.0018846246042701627,
+      "grad_norm": 0.38220450282096863,
+      "learning_rate": 4.9953091032327924e-05,
+      "loss": 0.1968,
+      "mean_token_accuracy": 0.9569838047027588,
+      "num_tokens": 514061.0,
+      "step": 228
+    },
+    {
+      "epoch": 0.0018928905016573124,
+      "grad_norm": 0.3590328097343445,
+      "learning_rate": 4.995288438489325e-05,
+      "loss": 0.2105,
+      "mean_token_accuracy": 0.9559999704360962,
+      "num_tokens": 516067.0,
+      "step": 229
+    },
+    {
+      "epoch": 0.0019011563990444622,
+      "grad_norm": 0.365873247385025,
+      "learning_rate": 4.995267773745857e-05,
+      "loss": 0.2312,
+      "mean_token_accuracy": 0.9503012299537659,
+      "num_tokens": 518065.0,
+      "step": 230
+    },
+    {
+      "epoch": 0.0019094222964316121,
+      "grad_norm": 0.3869886100292206,
+      "learning_rate": 4.995247109002389e-05,
+      "loss": 0.2183,
+      "mean_token_accuracy": 0.9523326754570007,
+      "num_tokens": 520043.0,
+      "step": 231
+    },
+    {
+      "epoch": 0.001917688193818762,
+      "grad_norm": 0.4155399203300476,
+      "learning_rate": 4.995226444258922e-05,
+      "loss": 0.2543,
+      "mean_token_accuracy": 0.9486796259880066,
+      "num_tokens": 522056.0,
+      "step": 232
+    },
+    {
+      "epoch": 0.0019259540912059119,
+      "grad_norm": 0.385654479265213,
+      "learning_rate": 4.995205779515453e-05,
+      "loss": 0.2128,
+      "mean_token_accuracy": 0.9536390900611877,
+      "num_tokens": 524068.0,
+      "step": 233
+    },
+    {
+      "epoch": 0.0019342199885930616,
+      "grad_norm": 0.425402969121933,
+      "learning_rate": 4.9951851147719854e-05,
+      "loss": 0.2976,
+      "mean_token_accuracy": 0.9434323906898499,
+      "num_tokens": 526160.0,
+      "step": 234
+    },
+    {
+      "epoch": 0.0019424858859802114,
+      "grad_norm": 0.46727752685546875,
+      "learning_rate": 4.9951644500285175e-05,
+      "loss": 0.2954,
+      "mean_token_accuracy": 0.9373493790626526,
+      "num_tokens": 528241.0,
+      "step": 235
+    },
+    {
+      "epoch": 0.0019507517833673613,
+      "grad_norm": 0.3046579957008362,
+      "learning_rate": 4.99514378528505e-05,
+      "loss": 0.1787,
+      "mean_token_accuracy": 0.9619341492652893,
+      "num_tokens": 530191.0,
+      "step": 236
+    },
+    {
+      "epoch": 0.0019590176807545113,
+      "grad_norm": 0.44570425152778625,
+      "learning_rate": 4.995123120541582e-05,
+      "loss": 0.2982,
+      "mean_token_accuracy": 0.9380701184272766,
+      "num_tokens": 532280.0,
+      "step": 237
+    },
+    {
+      "epoch": 0.001967283578141661,
+      "grad_norm": 0.4583204686641693,
+      "learning_rate": 4.995102455798114e-05,
+      "loss": 0.267,
+      "mean_token_accuracy": 0.948319137096405,
+      "num_tokens": 534279.0,
+      "step": 238
+    },
+    {
+      "epoch": 0.001975549475528811,
+      "grad_norm": 0.48733460903167725,
+      "learning_rate": 4.995081791054646e-05,
+      "loss": 0.2298,
+      "mean_token_accuracy": 0.9539077877998352,
+      "num_tokens": 536281.0,
+      "step": 239
+    },
+    {
+      "epoch": 0.0019838153729159606,
+      "grad_norm": 0.3647000193595886,
+      "learning_rate": 4.995061126311178e-05,
+      "loss": 0.1994,
+      "mean_token_accuracy": 0.9569620490074158,
+      "num_tokens": 538262.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.0019920812703031103,
+      "grad_norm": 0.35905659198760986,
+      "learning_rate": 4.9950404615677105e-05,
+      "loss": 0.2187,
+      "mean_token_accuracy": 0.9546586871147156,
+      "num_tokens": 540275.0,
+      "step": 241
+    },
+    {
+      "epoch": 0.00200034716769026,
+      "grad_norm": 0.4354383945465088,
+      "learning_rate": 4.995019796824243e-05,
+      "loss": 0.3224,
+      "mean_token_accuracy": 0.9346781969070435,
+      "num_tokens": 542363.0,
+      "step": 242
+    },
+    {
+      "epoch": 0.0020086130650774103,
+      "grad_norm": 0.35473981499671936,
+      "learning_rate": 4.994999132080775e-05,
+      "loss": 0.1936,
+      "mean_token_accuracy": 0.9590495228767395,
+      "num_tokens": 544347.0,
+      "step": 243
+    },
+    {
+      "epoch": 0.00201687896246456,
+      "grad_norm": 0.3254944086074829,
+      "learning_rate": 4.994978467337307e-05,
+      "loss": 0.1717,
+      "mean_token_accuracy": 0.9607942700386047,
+      "num_tokens": 546317.0,
+      "step": 244
+    },
+    {
+      "epoch": 0.00202514485985171,
+      "grad_norm": 0.3838103413581848,
+      "learning_rate": 4.9949578025938385e-05,
+      "loss": 0.2207,
+      "mean_token_accuracy": 0.9522402882575989,
+      "num_tokens": 548354.0,
+      "step": 245
+    },
+    {
+      "epoch": 0.0020334107572388595,
+      "grad_norm": 0.38548266887664795,
+      "learning_rate": 4.9949371378503714e-05,
+      "loss": 0.2581,
+      "mean_token_accuracy": 0.9475747346878052,
+      "num_tokens": 550401.0,
+      "step": 246
+    },
+    {
+      "epoch": 0.0020416766546260093,
+      "grad_norm": 0.3755143880844116,
+      "learning_rate": 4.994916473106903e-05,
+      "loss": 0.2557,
+      "mean_token_accuracy": 0.9424390196800232,
+      "num_tokens": 552457.0,
+      "step": 247
+    },
+    {
+      "epoch": 0.0020499425520131595,
+      "grad_norm": 0.3793683648109436,
+      "learning_rate": 4.994895808363435e-05,
+      "loss": 0.2264,
+      "mean_token_accuracy": 0.9554880857467651,
+      "num_tokens": 554440.0,
+      "step": 248
+    },
+    {
+      "epoch": 0.0020582084494003092,
+      "grad_norm": 0.4046688973903656,
+      "learning_rate": 4.994875143619968e-05,
+      "loss": 0.2443,
+      "mean_token_accuracy": 0.9479115605354309,
+      "num_tokens": 556481.0,
+      "step": 249
+    },
+    {
+      "epoch": 0.002066474346787459,
+      "grad_norm": 0.39033177495002747,
+      "learning_rate": 4.9948544788764993e-05,
+      "loss": 0.2337,
+      "mean_token_accuracy": 0.9542943239212036,
+      "num_tokens": 558478.0,
+      "step": 250
+    },
+    {
+      "epoch": 0.0020747402441746087,
+      "grad_norm": 0.43119877576828003,
+      "learning_rate": 4.9948338141330315e-05,
+      "loss": 0.1949,
+      "mean_token_accuracy": 0.9583756327629089,
+      "num_tokens": 560454.0,
+      "step": 251
+    },
+    {
+      "epoch": 0.0020830061415617585,
+      "grad_norm": 0.4812301695346832,
+      "learning_rate": 4.994813149389564e-05,
+      "loss": 0.1955,
+      "mean_token_accuracy": 0.957121729850769,
+      "num_tokens": 562489.0,
+      "step": 252
+    },
+    {
+      "epoch": 0.0020912720389489087,
+      "grad_norm": 0.401766300201416,
+      "learning_rate": 4.994792484646096e-05,
+      "loss": 0.2275,
+      "mean_token_accuracy": 0.953125,
+      "num_tokens": 564479.0,
+      "step": 253
+    },
+    {
+      "epoch": 0.0020995379363360584,
+      "grad_norm": 0.4104023575782776,
+      "learning_rate": 4.994771819902628e-05,
+      "loss": 0.2244,
+      "mean_token_accuracy": 0.9537773132324219,
+      "num_tokens": 566497.0,
+      "step": 254
+    },
+    {
+      "epoch": 0.002107803833723208,
+      "grad_norm": 0.37615010142326355,
+      "learning_rate": 4.99475115515916e-05,
+      "loss": 0.2366,
+      "mean_token_accuracy": 0.9547534584999084,
+      "num_tokens": 568470.0,
+      "step": 255
+    },
+    {
+      "epoch": 0.002116069731110358,
+      "grad_norm": 0.4524342715740204,
+      "learning_rate": 4.994730490415692e-05,
+      "loss": 0.2377,
+      "mean_token_accuracy": 0.9517412781715393,
+      "num_tokens": 570486.0,
+      "step": 256
+    },
+    {
+      "epoch": 0.0021243356284975077,
+      "grad_norm": 0.36174899339675903,
+      "learning_rate": 4.994709825672224e-05,
+      "loss": 0.2559,
+      "mean_token_accuracy": 0.9482329487800598,
+      "num_tokens": 572501.0,
+      "step": 257
+    },
+    {
+      "epoch": 0.0021326015258846575,
+      "grad_norm": 0.5051626563072205,
+      "learning_rate": 4.994689160928757e-05,
+      "loss": 0.2601,
+      "mean_token_accuracy": 0.9450767040252686,
+      "num_tokens": 574528.0,
+      "step": 258
+    },
+    {
+      "epoch": 0.0021408674232718077,
+      "grad_norm": 0.4387591779232025,
+      "learning_rate": 4.994668496185289e-05,
+      "loss": 0.242,
+      "mean_token_accuracy": 0.9555779695510864,
+      "num_tokens": 576515.0,
+      "step": 259
+    },
+    {
+      "epoch": 0.0021491333206589574,
+      "grad_norm": 0.46807482838630676,
+      "learning_rate": 4.99464783144182e-05,
+      "loss": 0.2313,
+      "mean_token_accuracy": 0.9486419558525085,
+      "num_tokens": 578546.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.002157399218046107,
+      "grad_norm": 0.4324467182159424,
+      "learning_rate": 4.994627166698353e-05,
+      "loss": 0.2304,
+      "mean_token_accuracy": 0.9508684873580933,
+      "num_tokens": 580567.0,
+      "step": 261
+    },
+    {
+      "epoch": 0.002165665115433257,
+      "grad_norm": 0.3726848363876343,
+      "learning_rate": 4.9946065019548847e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9652742743492126,
+      "num_tokens": 582560.0,
+      "step": 262
+    },
+    {
+      "epoch": 0.0021739310128204067,
+      "grad_norm": 0.4649844765663147,
+      "learning_rate": 4.9945858372114175e-05,
+      "loss": 0.2823,
+      "mean_token_accuracy": 0.9440078735351562,
+      "num_tokens": 584602.0,
+      "step": 263
+    },
+    {
+      "epoch": 0.002182196910207557,
+      "grad_norm": 0.3798585832118988,
+      "learning_rate": 4.994565172467949e-05,
+      "loss": 0.2385,
+      "mean_token_accuracy": 0.9526420831680298,
+      "num_tokens": 586614.0,
+      "step": 264
+    },
+    {
+      "epoch": 0.0021904628075947066,
+      "grad_norm": 0.5047613978385925,
+      "learning_rate": 4.994544507724481e-05,
+      "loss": 0.2007,
+      "mean_token_accuracy": 0.9601837396621704,
+      "num_tokens": 588579.0,
+      "step": 265
+    },
+    {
+      "epoch": 0.0021987287049818564,
+      "grad_norm": 0.37147533893585205,
+      "learning_rate": 4.994523842981014e-05,
+      "loss": 0.2324,
+      "mean_token_accuracy": 0.9545683860778809,
+      "num_tokens": 590566.0,
+      "step": 266
+    },
+    {
+      "epoch": 0.002206994602369006,
+      "grad_norm": 0.39514973759651184,
+      "learning_rate": 4.9945031782375455e-05,
+      "loss": 0.2525,
+      "mean_token_accuracy": 0.9510143399238586,
+      "num_tokens": 592593.0,
+      "step": 267
+    },
+    {
+      "epoch": 0.002215260499756156,
+      "grad_norm": 0.43914228677749634,
+      "learning_rate": 4.9944825134940776e-05,
+      "loss": 0.1881,
+      "mean_token_accuracy": 0.9605936408042908,
+      "num_tokens": 594553.0,
+      "step": 268
+    },
+    {
+      "epoch": 0.002223526397143306,
+      "grad_norm": 0.4348392188549042,
+      "learning_rate": 4.99446184875061e-05,
+      "loss": 0.233,
+      "mean_token_accuracy": 0.9477871656417847,
+      "num_tokens": 596570.0,
+      "step": 269
+    },
+    {
+      "epoch": 0.002231792294530456,
+      "grad_norm": 0.4209921360015869,
+      "learning_rate": 4.994441184007142e-05,
+      "loss": 0.2436,
+      "mean_token_accuracy": 0.955610990524292,
+      "num_tokens": 598581.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.0022400581919176056,
+      "grad_norm": 0.4016135632991791,
+      "learning_rate": 4.994420519263674e-05,
+      "loss": 0.2263,
+      "mean_token_accuracy": 0.9556650519371033,
+      "num_tokens": 600617.0,
+      "step": 271
+    },
+    {
+      "epoch": 0.0022483240893047553,
+      "grad_norm": 0.35575857758522034,
+      "learning_rate": 4.994399854520206e-05,
+      "loss": 0.2029,
+      "mean_token_accuracy": 0.9541832804679871,
+      "num_tokens": 602631.0,
+      "step": 272
+    },
+    {
+      "epoch": 0.002256589986691905,
+      "grad_norm": 0.3445430397987366,
+      "learning_rate": 4.9943791897767385e-05,
+      "loss": 0.1885,
+      "mean_token_accuracy": 0.9560059309005737,
+      "num_tokens": 604660.0,
+      "step": 273
+    },
+    {
+      "epoch": 0.002264855884079055,
+      "grad_norm": 0.36195898056030273,
+      "learning_rate": 4.99435852503327e-05,
+      "loss": 0.2284,
+      "mean_token_accuracy": 0.9558897018432617,
+      "num_tokens": 606661.0,
+      "step": 274
+    },
+    {
+      "epoch": 0.002273121781466205,
+      "grad_norm": 0.3910697102546692,
+      "learning_rate": 4.994337860289803e-05,
+      "loss": 0.2217,
+      "mean_token_accuracy": 0.9544329047203064,
+      "num_tokens": 608686.0,
+      "step": 275
+    },
+    {
+      "epoch": 0.0022813876788533548,
+      "grad_norm": 0.43754875659942627,
+      "learning_rate": 4.994317195546335e-05,
+      "loss": 0.2419,
+      "mean_token_accuracy": 0.9506480693817139,
+      "num_tokens": 610698.0,
+      "step": 276
+    },
+    {
+      "epoch": 0.0022896535762405045,
+      "grad_norm": 0.43161723017692566,
+      "learning_rate": 4.9942965308028665e-05,
+      "loss": 0.2328,
+      "mean_token_accuracy": 0.9513990879058838,
+      "num_tokens": 612741.0,
+      "step": 277
+    },
+    {
+      "epoch": 0.0022979194736276543,
+      "grad_norm": 0.258259117603302,
+      "learning_rate": 4.994275866059399e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9726521968841553,
+      "num_tokens": 614685.0,
+      "step": 278
+    },
+    {
+      "epoch": 0.002306185371014804,
+      "grad_norm": 0.4533786177635193,
+      "learning_rate": 4.994255201315931e-05,
+      "loss": 0.2567,
+      "mean_token_accuracy": 0.9435998201370239,
+      "num_tokens": 616730.0,
+      "step": 279
+    },
+    {
+      "epoch": 0.0023144512684019542,
+      "grad_norm": 0.40258121490478516,
+      "learning_rate": 4.994234536572463e-05,
+      "loss": 0.2557,
+      "mean_token_accuracy": 0.9493480324745178,
+      "num_tokens": 618730.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.002322717165789104,
+      "grad_norm": 0.41241350769996643,
+      "learning_rate": 4.994213871828996e-05,
+      "loss": 0.2194,
+      "mean_token_accuracy": 0.9514027833938599,
+      "num_tokens": 620732.0,
+      "step": 281
+    },
+    {
+      "epoch": 0.0023309830631762537,
+      "grad_norm": 0.43289700150489807,
+      "learning_rate": 4.994193207085527e-05,
+      "loss": 0.2918,
+      "mean_token_accuracy": 0.9389716386795044,
+      "num_tokens": 622819.0,
+      "step": 282
+    },
+    {
+      "epoch": 0.0023392489605634035,
+      "grad_norm": 0.4616968631744385,
+      "learning_rate": 4.9941725423420595e-05,
+      "loss": 0.2696,
+      "mean_token_accuracy": 0.9431051015853882,
+      "num_tokens": 624899.0,
+      "step": 283
+    },
+    {
+      "epoch": 0.0023475148579505532,
+      "grad_norm": 0.43764299154281616,
+      "learning_rate": 4.9941518775985916e-05,
+      "loss": 0.2455,
+      "mean_token_accuracy": 0.9458128213882446,
+      "num_tokens": 626935.0,
+      "step": 284
+    },
+    {
+      "epoch": 0.0023557807553377034,
+      "grad_norm": 0.410492479801178,
+      "learning_rate": 4.994131212855124e-05,
+      "loss": 0.2407,
+      "mean_token_accuracy": 0.9466209411621094,
+      "num_tokens": 628983.0,
+      "step": 285
+    },
+    {
+      "epoch": 0.002364046652724853,
+      "grad_norm": 0.44209814071655273,
+      "learning_rate": 4.994110548111656e-05,
+      "loss": 0.2259,
+      "mean_token_accuracy": 0.9534301161766052,
+      "num_tokens": 630986.0,
+      "step": 286
+    },
+    {
+      "epoch": 0.002372312550112003,
+      "grad_norm": 0.4304454028606415,
+      "learning_rate": 4.994089883368188e-05,
+      "loss": 0.2157,
+      "mean_token_accuracy": 0.9500494599342346,
+      "num_tokens": 633014.0,
+      "step": 287
+    },
+    {
+      "epoch": 0.0023805784474991527,
+      "grad_norm": 0.37028518319129944,
+      "learning_rate": 4.99406921862472e-05,
+      "loss": 0.2446,
+      "mean_token_accuracy": 0.9519038200378418,
+      "num_tokens": 635016.0,
+      "step": 288
+    },
+    {
+      "epoch": 0.0023888443448863025,
+      "grad_norm": 0.3473495841026306,
+      "learning_rate": 4.9940485538812525e-05,
+      "loss": 0.1871,
+      "mean_token_accuracy": 0.9602187871932983,
+      "num_tokens": 637033.0,
+      "step": 289
+    },
+    {
+      "epoch": 0.002397110242273452,
+      "grad_norm": 0.4115881621837616,
+      "learning_rate": 4.9940278891377846e-05,
+      "loss": 0.3009,
+      "mean_token_accuracy": 0.9386473298072815,
+      "num_tokens": 639109.0,
+      "step": 290
+    },
+    {
+      "epoch": 0.0024053761396606024,
+      "grad_norm": 0.39887967705726624,
+      "learning_rate": 4.994007224394317e-05,
+      "loss": 0.2498,
+      "mean_token_accuracy": 0.9518371224403381,
+      "num_tokens": 641129.0,
+      "step": 291
+    },
+    {
+      "epoch": 0.002413642037047752,
+      "grad_norm": 0.28864195942878723,
+      "learning_rate": 4.993986559650849e-05,
+      "loss": 0.1681,
+      "mean_token_accuracy": 0.9704533815383911,
+      "num_tokens": 643098.0,
+      "step": 292
+    },
+    {
+      "epoch": 0.002421907934434902,
+      "grad_norm": 0.42355039715766907,
+      "learning_rate": 4.993965894907381e-05,
+      "loss": 0.2712,
+      "mean_token_accuracy": 0.943109393119812,
+      "num_tokens": 645143.0,
+      "step": 293
+    },
+    {
+      "epoch": 0.0024301738318220517,
+      "grad_norm": 0.5223525166511536,
+      "learning_rate": 4.9939452301639126e-05,
+      "loss": 0.2075,
+      "mean_token_accuracy": 0.9607843160629272,
+      "num_tokens": 647138.0,
+      "step": 294
+    },
+    {
+      "epoch": 0.0024384397292092014,
+      "grad_norm": 0.3669191300868988,
+      "learning_rate": 4.9939245654204454e-05,
+      "loss": 0.1558,
+      "mean_token_accuracy": 0.9679586291313171,
+      "num_tokens": 649079.0,
+      "step": 295
+    },
+    {
+      "epoch": 0.0024467056265963516,
+      "grad_norm": 0.35745570063591003,
+      "learning_rate": 4.993903900676977e-05,
+      "loss": 0.2239,
+      "mean_token_accuracy": 0.9514999985694885,
+      "num_tokens": 651085.0,
+      "step": 296
+    },
+    {
+      "epoch": 0.0024549715239835014,
+      "grad_norm": 0.4034227728843689,
+      "learning_rate": 4.993883235933509e-05,
+      "loss": 0.3067,
+      "mean_token_accuracy": 0.9388942718505859,
+      "num_tokens": 653153.0,
+      "step": 297
+    },
+    {
+      "epoch": 0.002463237421370651,
+      "grad_norm": 0.40081143379211426,
+      "learning_rate": 4.993862571190042e-05,
+      "loss": 0.23,
+      "mean_token_accuracy": 0.9531639218330383,
+      "num_tokens": 655166.0,
+      "step": 298
+    },
+    {
+      "epoch": 0.002471503318757801,
+      "grad_norm": 0.3677947223186493,
+      "learning_rate": 4.9938419064465734e-05,
+      "loss": 0.1783,
+      "mean_token_accuracy": 0.9605063199996948,
+      "num_tokens": 657147.0,
+      "step": 299
+    },
+    {
+      "epoch": 0.0024797692161449506,
+      "grad_norm": 0.3393653333187103,
+      "learning_rate": 4.9938212417031056e-05,
+      "loss": 0.1867,
+      "mean_token_accuracy": 0.9632652997970581,
+      "num_tokens": 659113.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.002488035113532101,
+      "grad_norm": 0.41311779618263245,
+      "learning_rate": 4.993800576959638e-05,
+      "loss": 0.2167,
+      "mean_token_accuracy": 0.9567594528198242,
+      "num_tokens": 661131.0,
+      "step": 301
+    },
+    {
+      "epoch": 0.0024963010109192506,
+      "grad_norm": 0.3797242343425751,
+      "learning_rate": 4.99377991221617e-05,
+      "loss": 0.2078,
+      "mean_token_accuracy": 0.9559118151664734,
+      "num_tokens": 663133.0,
+      "step": 302
+    },
+    {
+      "epoch": 0.0025045669083064003,
+      "grad_norm": 0.44771602749824524,
+      "learning_rate": 4.993759247472702e-05,
+      "loss": 0.2621,
+      "mean_token_accuracy": 0.949127197265625,
+      "num_tokens": 665144.0,
+      "step": 303
+    },
+    {
+      "epoch": 0.00251283280569355,
+      "grad_norm": 0.33966875076293945,
+      "learning_rate": 4.993738582729234e-05,
+      "loss": 0.2111,
+      "mean_token_accuracy": 0.9543842077255249,
+      "num_tokens": 667123.0,
+      "step": 304
+    },
+    {
+      "epoch": 0.0025210987030807,
+      "grad_norm": 0.314005583524704,
+      "learning_rate": 4.9937179179857664e-05,
+      "loss": 0.1743,
+      "mean_token_accuracy": 0.9641393423080444,
+      "num_tokens": 669081.0,
+      "step": 305
+    },
+    {
+      "epoch": 0.0025293646004678496,
+      "grad_norm": 0.5659259557723999,
+      "learning_rate": 4.9936972532422986e-05,
+      "loss": 0.2362,
+      "mean_token_accuracy": 0.9511088728904724,
+      "num_tokens": 671071.0,
+      "step": 306
+    },
+    {
+      "epoch": 0.0025376304978549998,
+      "grad_norm": 0.3476862609386444,
+      "learning_rate": 4.993676588498831e-05,
+      "loss": 0.1537,
+      "mean_token_accuracy": 0.9676578044891357,
+      "num_tokens": 672994.0,
+      "step": 307
+    },
+    {
+      "epoch": 0.0025458963952421495,
+      "grad_norm": 0.45339474081993103,
+      "learning_rate": 4.993655923755363e-05,
+      "loss": 0.2449,
+      "mean_token_accuracy": 0.9519087672233582,
+      "num_tokens": 675017.0,
+      "step": 308
+    },
+    {
+      "epoch": 0.0025541622926292993,
+      "grad_norm": 0.3326128423213959,
+      "learning_rate": 4.993635259011895e-05,
+      "loss": 0.2103,
+      "mean_token_accuracy": 0.9533730149269104,
+      "num_tokens": 677039.0,
+      "step": 309
+    },
+    {
+      "epoch": 0.002562428190016449,
+      "grad_norm": 0.38087713718414307,
+      "learning_rate": 4.993614594268427e-05,
+      "loss": 0.1822,
+      "mean_token_accuracy": 0.9575328826904297,
+      "num_tokens": 679023.0,
+      "step": 310
+    },
+    {
+      "epoch": 0.002570694087403599,
+      "grad_norm": 0.41911280155181885,
+      "learning_rate": 4.993593929524959e-05,
+      "loss": 0.2534,
+      "mean_token_accuracy": 0.9478908181190491,
+      "num_tokens": 681044.0,
+      "step": 311
+    },
+    {
+      "epoch": 0.002578959984790749,
+      "grad_norm": 0.3845995366573334,
+      "learning_rate": 4.9935732647814916e-05,
+      "loss": 0.2271,
+      "mean_token_accuracy": 0.9553571343421936,
+      "num_tokens": 683066.0,
+      "step": 312
+    },
+    {
+      "epoch": 0.0025872258821778987,
+      "grad_norm": 0.4912784695625305,
+      "learning_rate": 4.993552600038023e-05,
+      "loss": 0.2081,
+      "mean_token_accuracy": 0.9535232186317444,
+      "num_tokens": 685073.0,
+      "step": 313
+    },
+    {
+      "epoch": 0.0025954917795650485,
+      "grad_norm": 0.35172325372695923,
+      "learning_rate": 4.993531935294555e-05,
+      "loss": 0.2089,
+      "mean_token_accuracy": 0.9572989344596863,
+      "num_tokens": 687093.0,
+      "step": 314
+    },
+    {
+      "epoch": 0.0026037576769521982,
+      "grad_norm": 0.35449934005737305,
+      "learning_rate": 4.993511270551088e-05,
+      "loss": 0.2002,
+      "mean_token_accuracy": 0.9577039480209351,
+      "num_tokens": 689085.0,
+      "step": 315
+    },
+    {
+      "epoch": 0.002612023574339348,
+      "grad_norm": 0.34856724739074707,
+      "learning_rate": 4.9934906058076196e-05,
+      "loss": 0.1863,
+      "mean_token_accuracy": 0.9597585797309875,
+      "num_tokens": 691079.0,
+      "step": 316
+    },
+    {
+      "epoch": 0.002620289471726498,
+      "grad_norm": 0.39429017901420593,
+      "learning_rate": 4.993469941064152e-05,
+      "loss": 0.264,
+      "mean_token_accuracy": 0.9471865892410278,
+      "num_tokens": 693111.0,
+      "step": 317
+    },
+    {
+      "epoch": 0.002628555369113648,
+      "grad_norm": 0.38967621326446533,
+      "learning_rate": 4.993449276320684e-05,
+      "loss": 0.2103,
+      "mean_token_accuracy": 0.9556227922439575,
+      "num_tokens": 695100.0,
+      "step": 318
+    },
+    {
+      "epoch": 0.0026368212665007977,
+      "grad_norm": 0.42454391717910767,
+      "learning_rate": 4.993428611577216e-05,
+      "loss": 0.2458,
+      "mean_token_accuracy": 0.9503759145736694,
+      "num_tokens": 697101.0,
+      "step": 319
+    },
+    {
+      "epoch": 0.0026450871638879474,
+      "grad_norm": 0.43203118443489075,
+      "learning_rate": 4.993407946833748e-05,
+      "loss": 0.209,
+      "mean_token_accuracy": 0.9589794874191284,
+      "num_tokens": 699106.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.002653353061275097,
+      "grad_norm": 0.2782335877418518,
+      "learning_rate": 4.9933872820902804e-05,
+      "loss": 0.1457,
+      "mean_token_accuracy": 0.9726239442825317,
+      "num_tokens": 701048.0,
+      "step": 321
+    },
+    {
+      "epoch": 0.002661618958662247,
+      "grad_norm": 0.34580934047698975,
+      "learning_rate": 4.9933666173468126e-05,
+      "loss": 0.1517,
+      "mean_token_accuracy": 0.9663212299346924,
+      "num_tokens": 702984.0,
+      "step": 322
+    },
+    {
+      "epoch": 0.002669884856049397,
+      "grad_norm": 0.3792515695095062,
+      "learning_rate": 4.993345952603344e-05,
+      "loss": 0.1996,
+      "mean_token_accuracy": 0.9585207104682922,
+      "num_tokens": 704991.0,
+      "step": 323
+    },
+    {
+      "epoch": 0.002678150753436547,
+      "grad_norm": 0.4265529215335846,
+      "learning_rate": 4.993325287859877e-05,
+      "loss": 0.1854,
+      "mean_token_accuracy": 0.9631336331367493,
+      "num_tokens": 706950.0,
+      "step": 324
+    },
+    {
+      "epoch": 0.0026864166508236966,
+      "grad_norm": 0.36660078167915344,
+      "learning_rate": 4.993304623116409e-05,
+      "loss": 0.2067,
+      "mean_token_accuracy": 0.9601211547851562,
+      "num_tokens": 708937.0,
+      "step": 325
+    },
+    {
+      "epoch": 0.0026946825482108464,
+      "grad_norm": 0.4065447747707367,
+      "learning_rate": 4.993283958372941e-05,
+      "loss": 0.1888,
+      "mean_token_accuracy": 0.9626216292381287,
+      "num_tokens": 710896.0,
+      "step": 326
+    },
+    {
+      "epoch": 0.002702948445597996,
+      "grad_norm": 0.29612088203430176,
+      "learning_rate": 4.9932632936294734e-05,
+      "loss": 0.1354,
+      "mean_token_accuracy": 0.9708439707756042,
+      "num_tokens": 712857.0,
+      "step": 327
+    },
+    {
+      "epoch": 0.0027112143429851463,
+      "grad_norm": 0.3609902560710907,
+      "learning_rate": 4.993242628886005e-05,
+      "loss": 0.1816,
+      "mean_token_accuracy": 0.960774302482605,
+      "num_tokens": 714826.0,
+      "step": 328
+    },
+    {
+      "epoch": 0.002719480240372296,
+      "grad_norm": 0.4500750005245209,
+      "learning_rate": 4.993221964142538e-05,
+      "loss": 0.2144,
+      "mean_token_accuracy": 0.9514798521995544,
+      "num_tokens": 716893.0,
+      "step": 329
+    },
+    {
+      "epoch": 0.002727746137759446,
+      "grad_norm": 0.416002094745636,
+      "learning_rate": 4.99320129939907e-05,
+      "loss": 0.2419,
+      "mean_token_accuracy": 0.9465424418449402,
+      "num_tokens": 718938.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.0027360120351465956,
+      "grad_norm": 0.396006315946579,
+      "learning_rate": 4.9931806346556014e-05,
+      "loss": 0.2139,
+      "mean_token_accuracy": 0.9553660750389099,
+      "num_tokens": 720938.0,
+      "step": 331
+    },
+    {
+      "epoch": 0.0027442779325337454,
+      "grad_norm": 0.4281254708766937,
+      "learning_rate": 4.993159969912134e-05,
+      "loss": 0.2475,
+      "mean_token_accuracy": 0.9524975419044495,
+      "num_tokens": 722986.0,
+      "step": 332
+    },
+    {
+      "epoch": 0.0027525438299208956,
+      "grad_norm": 0.4296991229057312,
+      "learning_rate": 4.993139305168666e-05,
+      "loss": 0.2036,
+      "mean_token_accuracy": 0.9571356773376465,
+      "num_tokens": 724975.0,
+      "step": 333
+    },
+    {
+      "epoch": 0.0027608097273080453,
+      "grad_norm": 0.41935306787490845,
+      "learning_rate": 4.993118640425198e-05,
+      "loss": 0.2116,
+      "mean_token_accuracy": 0.9563894271850586,
+      "num_tokens": 726953.0,
+      "step": 334
+    },
+    {
+      "epoch": 0.002769075624695195,
+      "grad_norm": 0.3584057092666626,
+      "learning_rate": 4.99309797568173e-05,
+      "loss": 0.1853,
+      "mean_token_accuracy": 0.9592252969741821,
+      "num_tokens": 728921.0,
+      "step": 335
+    },
+    {
+      "epoch": 0.002777341522082345,
+      "grad_norm": 0.4146767258644104,
+      "learning_rate": 4.993077310938262e-05,
+      "loss": 0.2196,
+      "mean_token_accuracy": 0.9568308591842651,
+      "num_tokens": 730896.0,
+      "step": 336
+    },
+    {
+      "epoch": 0.0027856074194694946,
+      "grad_norm": 0.3676367998123169,
+      "learning_rate": 4.9930566461947944e-05,
+      "loss": 0.1915,
+      "mean_token_accuracy": 0.9622736573219299,
+      "num_tokens": 732890.0,
+      "step": 337
+    },
+    {
+      "epoch": 0.0027938733168566448,
+      "grad_norm": 0.41326549649238586,
+      "learning_rate": 4.9930359814513265e-05,
+      "loss": 0.2046,
+      "mean_token_accuracy": 0.9549999833106995,
+      "num_tokens": 734896.0,
+      "step": 338
+    },
+    {
+      "epoch": 0.0028021392142437945,
+      "grad_norm": 0.4531644880771637,
+      "learning_rate": 4.993015316707859e-05,
+      "loss": 0.2445,
+      "mean_token_accuracy": 0.9481589198112488,
+      "num_tokens": 736966.0,
+      "step": 339
+    },
+    {
+      "epoch": 0.0028104051116309443,
+      "grad_norm": 0.38033097982406616,
+      "learning_rate": 4.992994651964391e-05,
+      "loss": 0.2096,
+      "mean_token_accuracy": 0.9564564824104309,
+      "num_tokens": 738970.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.002818671009018094,
+      "grad_norm": 0.5736963152885437,
+      "learning_rate": 4.992973987220923e-05,
+      "loss": 0.32,
+      "mean_token_accuracy": 0.9341723322868347,
+      "num_tokens": 741042.0,
+      "step": 341
+    },
+    {
+      "epoch": 0.0028269369064052438,
+      "grad_norm": 0.44328004121780396,
+      "learning_rate": 4.992953322477455e-05,
+      "loss": 0.2257,
+      "mean_token_accuracy": 0.950470507144928,
+      "num_tokens": 743067.0,
+      "step": 342
+    },
+    {
+      "epoch": 0.0028352028037923935,
+      "grad_norm": 0.36262065172195435,
+      "learning_rate": 4.992932657733987e-05,
+      "loss": 0.2283,
+      "mean_token_accuracy": 0.9575848579406738,
+      "num_tokens": 745077.0,
+      "step": 343
+    },
+    {
+      "epoch": 0.0028434687011795437,
+      "grad_norm": 0.3923065960407257,
+      "learning_rate": 4.9929119929905195e-05,
+      "loss": 0.2119,
+      "mean_token_accuracy": 0.9602015018463135,
+      "num_tokens": 747068.0,
+      "step": 344
+    },
+    {
+      "epoch": 0.0028517345985666935,
+      "grad_norm": 0.5067260265350342,
+      "learning_rate": 4.992891328247051e-05,
+      "loss": 0.2586,
+      "mean_token_accuracy": 0.945919394493103,
+      "num_tokens": 749108.0,
+      "step": 345
+    },
+    {
+      "epoch": 0.0028600004959538432,
+      "grad_norm": 0.40983039140701294,
+      "learning_rate": 4.992870663503583e-05,
+      "loss": 0.2041,
+      "mean_token_accuracy": 0.9590080976486206,
+      "num_tokens": 751090.0,
+      "step": 346
+    },
+    {
+      "epoch": 0.002868266393340993,
+      "grad_norm": 0.4184195399284363,
+      "learning_rate": 4.992849998760116e-05,
+      "loss": 0.264,
+      "mean_token_accuracy": 0.948483407497406,
+      "num_tokens": 753173.0,
+      "step": 347
+    },
+    {
+      "epoch": 0.0028765322907281427,
+      "grad_norm": 0.4780874252319336,
+      "learning_rate": 4.9928293340166475e-05,
+      "loss": 0.2839,
+      "mean_token_accuracy": 0.9410637021064758,
+      "num_tokens": 755266.0,
+      "step": 348
+    },
+    {
+      "epoch": 0.002884798188115293,
+      "grad_norm": 0.3527054786682129,
+      "learning_rate": 4.9928086692731804e-05,
+      "loss": 0.2064,
+      "mean_token_accuracy": 0.9580808281898499,
+      "num_tokens": 757252.0,
+      "step": 349
+    },
+    {
+      "epoch": 0.0028930640855024427,
+      "grad_norm": 0.4469713270664215,
+      "learning_rate": 4.992788004529712e-05,
+      "loss": 0.3162,
+      "mean_token_accuracy": 0.9341232180595398,
+      "num_tokens": 759368.0,
+      "step": 350
+    },
+    {
+      "epoch": 0.0029013299828895924,
+      "grad_norm": 0.39907026290893555,
+      "learning_rate": 4.992767339786244e-05,
+      "loss": 0.2265,
+      "mean_token_accuracy": 0.9537920355796814,
+      "num_tokens": 761365.0,
+      "step": 351
+    },
+    {
+      "epoch": 0.002909595880276742,
+      "grad_norm": 0.34887832403182983,
+      "learning_rate": 4.992746675042776e-05,
+      "loss": 0.2377,
+      "mean_token_accuracy": 0.9492682814598083,
+      "num_tokens": 763421.0,
+      "step": 352
+    },
+    {
+      "epoch": 0.002917861777663892,
+      "grad_norm": 0.4177617132663727,
+      "learning_rate": 4.9927260102993083e-05,
+      "loss": 0.2459,
+      "mean_token_accuracy": 0.9459459185600281,
+      "num_tokens": 765462.0,
+      "step": 353
+    },
+    {
+      "epoch": 0.002926127675051042,
+      "grad_norm": 0.38036975264549255,
+      "learning_rate": 4.9927053455558405e-05,
+      "loss": 0.2166,
+      "mean_token_accuracy": 0.957121729850769,
+      "num_tokens": 767497.0,
+      "step": 354
+    },
+    {
+      "epoch": 0.002934393572438192,
+      "grad_norm": 0.3349733352661133,
+      "learning_rate": 4.992684680812373e-05,
+      "loss": 0.1878,
+      "mean_token_accuracy": 0.9640688300132751,
+      "num_tokens": 769479.0,
+      "step": 355
+    },
+    {
+      "epoch": 0.0029426594698253416,
+      "grad_norm": 0.4739646911621094,
+      "learning_rate": 4.992664016068905e-05,
+      "loss": 0.2587,
+      "mean_token_accuracy": 0.9436893463134766,
+      "num_tokens": 771545.0,
+      "step": 356
+    },
+    {
+      "epoch": 0.0029509253672124914,
+      "grad_norm": 0.3559514582157135,
+      "learning_rate": 4.992643351325437e-05,
+      "loss": 0.1915,
+      "mean_token_accuracy": 0.9610655903816223,
+      "num_tokens": 773503.0,
+      "step": 357
+    },
+    {
+      "epoch": 0.002959191264599641,
+      "grad_norm": 0.4058264195919037,
+      "learning_rate": 4.992622686581969e-05,
+      "loss": 0.2615,
+      "mean_token_accuracy": 0.9463152885437012,
+      "num_tokens": 775558.0,
+      "step": 358
+    },
+    {
+      "epoch": 0.002967457161986791,
+      "grad_norm": 0.39049819111824036,
+      "learning_rate": 4.9926020218385013e-05,
+      "loss": 0.2534,
+      "mean_token_accuracy": 0.9458927512168884,
+      "num_tokens": 777597.0,
+      "step": 359
+    },
+    {
+      "epoch": 0.002975723059373941,
+      "grad_norm": 0.4811549782752991,
+      "learning_rate": 4.992581357095033e-05,
+      "loss": 0.2993,
+      "mean_token_accuracy": 0.9370155334472656,
+      "num_tokens": 779667.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.002983988956761091,
+      "grad_norm": 0.3548329174518585,
+      "learning_rate": 4.992560692351566e-05,
+      "loss": 0.1986,
+      "mean_token_accuracy": 0.9604802131652832,
+      "num_tokens": 781672.0,
+      "step": 361
+    },
+    {
+      "epoch": 0.0029922548541482406,
+      "grad_norm": 0.5780093669891357,
+      "learning_rate": 4.992540027608097e-05,
+      "loss": 0.2916,
+      "mean_token_accuracy": 0.9366230964660645,
+      "num_tokens": 783745.0,
+      "step": 362
+    },
+    {
+      "epoch": 0.0030005207515353904,
+      "grad_norm": 0.45139673352241516,
+      "learning_rate": 4.992519362864629e-05,
+      "loss": 0.2687,
+      "mean_token_accuracy": 0.9430102109909058,
+      "num_tokens": 785804.0,
+      "step": 363
+    },
+    {
+      "epoch": 0.00300878664892254,
+      "grad_norm": 0.4197750687599182,
+      "learning_rate": 4.992498698121162e-05,
+      "loss": 0.1919,
+      "mean_token_accuracy": 0.9562374353408813,
+      "num_tokens": 787798.0,
+      "step": 364
+    },
+    {
+      "epoch": 0.0030170525463096903,
+      "grad_norm": 0.4055244028568268,
+      "learning_rate": 4.9924780333776937e-05,
+      "loss": 0.2207,
+      "mean_token_accuracy": 0.9598997235298157,
+      "num_tokens": 789799.0,
+      "step": 365
+    },
+    {
+      "epoch": 0.00302531844369684,
+      "grad_norm": 0.4178362786769867,
+      "learning_rate": 4.992457368634226e-05,
+      "loss": 0.2215,
+      "mean_token_accuracy": 0.9520782232284546,
+      "num_tokens": 791850.0,
+      "step": 366
+    },
+    {
+      "epoch": 0.00303358434108399,
+      "grad_norm": 0.3683435320854187,
+      "learning_rate": 4.992436703890758e-05,
+      "loss": 0.1526,
+      "mean_token_accuracy": 0.9724675416946411,
+      "num_tokens": 793781.0,
+      "step": 367
+    },
+    {
+      "epoch": 0.0030418502384711396,
+      "grad_norm": 0.4005066454410553,
+      "learning_rate": 4.99241603914729e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9660843014717102,
+      "num_tokens": 795733.0,
+      "step": 368
+    },
+    {
+      "epoch": 0.0030501161358582893,
+      "grad_norm": 0.3928993344306946,
+      "learning_rate": 4.992395374403822e-05,
+      "loss": 0.1892,
+      "mean_token_accuracy": 0.9575971961021423,
+      "num_tokens": 797720.0,
+      "step": 369
+    },
+    {
+      "epoch": 0.0030583820332454395,
+      "grad_norm": 0.3924673795700073,
+      "learning_rate": 4.9923747096603545e-05,
+      "loss": 0.155,
+      "mean_token_accuracy": 0.9669762849807739,
+      "num_tokens": 799664.0,
+      "step": 370
+    },
+    {
+      "epoch": 0.0030666479306325893,
+      "grad_norm": 0.41587457060813904,
+      "learning_rate": 4.9923540449168867e-05,
+      "loss": 0.1917,
+      "mean_token_accuracy": 0.9566301107406616,
+      "num_tokens": 801676.0,
+      "step": 371
+    },
+    {
+      "epoch": 0.003074913828019739,
+      "grad_norm": 0.42865025997161865,
+      "learning_rate": 4.992333380173419e-05,
+      "loss": 0.2885,
+      "mean_token_accuracy": 0.9363548755645752,
+      "num_tokens": 803756.0,
+      "step": 372
+    },
+    {
+      "epoch": 0.0030831797254068888,
+      "grad_norm": 0.4900335669517517,
+      "learning_rate": 4.992312715429951e-05,
+      "loss": 0.2366,
+      "mean_token_accuracy": 0.9501991868019104,
+      "num_tokens": 805770.0,
+      "step": 373
+    },
+    {
+      "epoch": 0.0030914456227940385,
+      "grad_norm": 0.5078573822975159,
+      "learning_rate": 4.992292050686483e-05,
+      "loss": 0.2046,
+      "mean_token_accuracy": 0.9584569931030273,
+      "num_tokens": 807798.0,
+      "step": 374
+    },
+    {
+      "epoch": 0.0030997115201811883,
+      "grad_norm": 0.37238258123397827,
+      "learning_rate": 4.992271385943015e-05,
+      "loss": 0.1896,
+      "mean_token_accuracy": 0.9603255391120911,
+      "num_tokens": 809770.0,
+      "step": 375
+    },
+    {
+      "epoch": 0.0031079774175683385,
+      "grad_norm": 0.45555543899536133,
+      "learning_rate": 4.9922507211995475e-05,
+      "loss": 0.2647,
+      "mean_token_accuracy": 0.949575662612915,
+      "num_tokens": 811779.0,
+      "step": 376
+    },
+    {
+      "epoch": 0.003116243314955488,
+      "grad_norm": 0.4149267077445984,
+      "learning_rate": 4.992230056456079e-05,
+      "loss": 0.2135,
+      "mean_token_accuracy": 0.9515655636787415,
+      "num_tokens": 813829.0,
+      "step": 377
+    },
+    {
+      "epoch": 0.003124509212342638,
+      "grad_norm": 0.4098355174064636,
+      "learning_rate": 4.992209391712612e-05,
+      "loss": 0.2225,
+      "mean_token_accuracy": 0.9564999938011169,
+      "num_tokens": 815835.0,
+      "step": 378
+    },
+    {
+      "epoch": 0.0031327751097297877,
+      "grad_norm": 0.44163843989372253,
+      "learning_rate": 4.992188726969143e-05,
+      "loss": 0.2849,
+      "mean_token_accuracy": 0.9403491616249084,
+      "num_tokens": 817903.0,
+      "step": 379
+    },
+    {
+      "epoch": 0.0031410410071169375,
+      "grad_norm": 0.3630768954753876,
+      "learning_rate": 4.9921680622256755e-05,
+      "loss": 0.2012,
+      "mean_token_accuracy": 0.9573721289634705,
+      "num_tokens": 819903.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.0031493069045040877,
+      "grad_norm": 0.3289450407028198,
+      "learning_rate": 4.992147397482208e-05,
+      "loss": 0.1551,
+      "mean_token_accuracy": 0.9648760557174683,
+      "num_tokens": 821845.0,
+      "step": 381
+    },
+    {
+      "epoch": 0.0031575728018912374,
+      "grad_norm": 0.3904035687446594,
+      "learning_rate": 4.99212673273874e-05,
+      "loss": 0.1955,
+      "mean_token_accuracy": 0.9593082666397095,
+      "num_tokens": 823817.0,
+      "step": 382
+    },
+    {
+      "epoch": 0.003165838699278387,
+      "grad_norm": 0.36014196276664734,
+      "learning_rate": 4.992106067995272e-05,
+      "loss": 0.264,
+      "mean_token_accuracy": 0.9508116245269775,
+      "num_tokens": 825856.0,
+      "step": 383
+    },
+    {
+      "epoch": 0.003174104596665537,
+      "grad_norm": 0.3443008363246918,
+      "learning_rate": 4.992085403251804e-05,
+      "loss": 0.1814,
+      "mean_token_accuracy": 0.9601413011550903,
+      "num_tokens": 827844.0,
+      "step": 384
+    },
+    {
+      "epoch": 0.0031823704940526867,
+      "grad_norm": 0.4116482734680176,
+      "learning_rate": 4.992064738508336e-05,
+      "loss": 0.217,
+      "mean_token_accuracy": 0.9594663977622986,
+      "num_tokens": 829799.0,
+      "step": 385
+    },
+    {
+      "epoch": 0.003190636391439837,
+      "grad_norm": 0.5228381156921387,
+      "learning_rate": 4.9920440737648685e-05,
+      "loss": 0.2936,
+      "mean_token_accuracy": 0.9364467859268188,
+      "num_tokens": 831882.0,
+      "step": 386
+    },
+    {
+      "epoch": 0.0031989022888269866,
+      "grad_norm": 0.40949776768684387,
+      "learning_rate": 4.9920234090214006e-05,
+      "loss": 0.1998,
+      "mean_token_accuracy": 0.9591419696807861,
+      "num_tokens": 833846.0,
+      "step": 387
+    },
+    {
+      "epoch": 0.0032071681862141364,
+      "grad_norm": 0.4674682915210724,
+      "learning_rate": 4.992002744277933e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9629057049751282,
+      "num_tokens": 835793.0,
+      "step": 388
+    },
+    {
+      "epoch": 0.003215434083601286,
+      "grad_norm": 0.3216884434223175,
+      "learning_rate": 4.991982079534465e-05,
+      "loss": 0.1985,
+      "mean_token_accuracy": 0.960606038570404,
+      "num_tokens": 837779.0,
+      "step": 389
+    },
+    {
+      "epoch": 0.003223699980988436,
+      "grad_norm": 0.37692657113075256,
+      "learning_rate": 4.991961414790997e-05,
+      "loss": 0.2112,
+      "mean_token_accuracy": 0.9524752497673035,
+      "num_tokens": 839805.0,
+      "step": 390
+    },
+    {
+      "epoch": 0.0032319658783755856,
+      "grad_norm": 0.38113531470298767,
+      "learning_rate": 4.991940750047529e-05,
+      "loss": 0.2331,
+      "mean_token_accuracy": 0.9517478942871094,
+      "num_tokens": 841842.0,
+      "step": 391
+    },
+    {
+      "epoch": 0.003240231775762736,
+      "grad_norm": 0.4439680874347687,
+      "learning_rate": 4.9919200853040615e-05,
+      "loss": 0.3131,
+      "mean_token_accuracy": 0.9396508932113647,
+      "num_tokens": 843853.0,
+      "step": 392
+    },
+    {
+      "epoch": 0.0032484976731498856,
+      "grad_norm": 0.33021825551986694,
+      "learning_rate": 4.9918994205605936e-05,
+      "loss": 0.21,
+      "mean_token_accuracy": 0.9591429829597473,
+      "num_tokens": 845866.0,
+      "step": 393
+    },
+    {
+      "epoch": 0.0032567635705370353,
+      "grad_norm": 0.4424145221710205,
+      "learning_rate": 4.991878755817125e-05,
+      "loss": 0.3097,
+      "mean_token_accuracy": 0.9398350119590759,
+      "num_tokens": 847933.0,
+      "step": 394
+    },
+    {
+      "epoch": 0.003265029467924185,
+      "grad_norm": 0.3730950355529785,
+      "learning_rate": 4.991858091073658e-05,
+      "loss": 0.2081,
+      "mean_token_accuracy": 0.9553884863853455,
+      "num_tokens": 849934.0,
+      "step": 395
+    },
+    {
+      "epoch": 0.003273295365311335,
+      "grad_norm": 0.4788861870765686,
+      "learning_rate": 4.99183742633019e-05,
+      "loss": 0.2004,
+      "mean_token_accuracy": 0.9563275575637817,
+      "num_tokens": 851955.0,
+      "step": 396
+    },
+    {
+      "epoch": 0.003281561262698485,
+      "grad_norm": 0.4064439833164215,
+      "learning_rate": 4.9918167615867216e-05,
+      "loss": 0.2142,
+      "mean_token_accuracy": 0.9543424248695374,
+      "num_tokens": 853976.0,
+      "step": 397
+    },
+    {
+      "epoch": 0.003289827160085635,
+      "grad_norm": 0.41253840923309326,
+      "learning_rate": 4.9917960968432544e-05,
+      "loss": 0.2113,
+      "mean_token_accuracy": 0.9553752541542053,
+      "num_tokens": 855954.0,
+      "step": 398
+    },
+    {
+      "epoch": 0.0032980930574727845,
+      "grad_norm": 0.36155927181243896,
+      "learning_rate": 4.991775432099786e-05,
+      "loss": 0.1681,
+      "mean_token_accuracy": 0.9655703902244568,
+      "num_tokens": 857906.0,
+      "step": 399
+    },
+    {
+      "epoch": 0.0033063589548599343,
+      "grad_norm": 0.36993175745010376,
+      "learning_rate": 4.991754767356318e-05,
+      "loss": 0.2004,
+      "mean_token_accuracy": 0.9560161828994751,
+      "num_tokens": 859890.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.003314624852247084,
+      "grad_norm": 0.7902066111564636,
+      "learning_rate": 4.99173410261285e-05,
+      "loss": 0.3206,
+      "mean_token_accuracy": 0.9354990720748901,
+      "num_tokens": 862020.0,
+      "step": 401
+    },
+    {
+      "epoch": 0.0033228907496342342,
+      "grad_norm": 0.4118296205997467,
+      "learning_rate": 4.9917134378693824e-05,
+      "loss": 0.2645,
+      "mean_token_accuracy": 0.9444990158081055,
+      "num_tokens": 864062.0,
+      "step": 402
+    },
+    {
+      "epoch": 0.003331156647021384,
+      "grad_norm": 0.41594621539115906,
+      "learning_rate": 4.9916927731259146e-05,
+      "loss": 0.1843,
+      "mean_token_accuracy": 0.9603612422943115,
+      "num_tokens": 866061.0,
+      "step": 403
+    },
+    {
+      "epoch": 0.0033394225444085338,
+      "grad_norm": 0.35747259855270386,
+      "learning_rate": 4.991672108382447e-05,
+      "loss": 0.1876,
+      "mean_token_accuracy": 0.962051272392273,
+      "num_tokens": 868017.0,
+      "step": 404
+    },
+    {
+      "epoch": 0.0033476884417956835,
+      "grad_norm": 0.38307011127471924,
+      "learning_rate": 4.991651443638979e-05,
+      "loss": 0.1546,
+      "mean_token_accuracy": 0.9671052694320679,
+      "num_tokens": 869999.0,
+      "step": 405
+    },
+    {
+      "epoch": 0.0033559543391828333,
+      "grad_norm": 0.37060555815696716,
+      "learning_rate": 4.991630778895511e-05,
+      "loss": 0.2177,
+      "mean_token_accuracy": 0.9546574950218201,
+      "num_tokens": 872034.0,
+      "step": 406
+    },
+    {
+      "epoch": 0.003364220236569983,
+      "grad_norm": 0.3875426650047302,
+      "learning_rate": 4.991610114152043e-05,
+      "loss": 0.2074,
+      "mean_token_accuracy": 0.9561971426010132,
+      "num_tokens": 874049.0,
+      "step": 407
+    },
+    {
+      "epoch": 0.003372486133957133,
+      "grad_norm": 0.3926350176334381,
+      "learning_rate": 4.9915894494085754e-05,
+      "loss": 0.1887,
+      "mean_token_accuracy": 0.959857702255249,
+      "num_tokens": 876023.0,
+      "step": 408
+    },
+    {
+      "epoch": 0.003380752031344283,
+      "grad_norm": 0.5841166973114014,
+      "learning_rate": 4.991568784665107e-05,
+      "loss": 0.2199,
+      "mean_token_accuracy": 0.9575757384300232,
+      "num_tokens": 878009.0,
+      "step": 409
+    },
+    {
+      "epoch": 0.0033890179287314327,
+      "grad_norm": 0.29495862126350403,
+      "learning_rate": 4.99154811992164e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9695248007774353,
+      "num_tokens": 879951.0,
+      "step": 410
+    },
+    {
+      "epoch": 0.0033972838261185825,
+      "grad_norm": 0.3995298743247986,
+      "learning_rate": 4.991527455178171e-05,
+      "loss": 0.2155,
+      "mean_token_accuracy": 0.9552238583564758,
+      "num_tokens": 881967.0,
+      "step": 411
+    },
+    {
+      "epoch": 0.0034055497235057322,
+      "grad_norm": 0.39600613713264465,
+      "learning_rate": 4.991506790434704e-05,
+      "loss": 0.2027,
+      "mean_token_accuracy": 0.9567404389381409,
+      "num_tokens": 883961.0,
+      "step": 412
+    },
+    {
+      "epoch": 0.0034138156208928824,
+      "grad_norm": 0.4439043700695038,
+      "learning_rate": 4.991486125691236e-05,
+      "loss": 0.1865,
+      "mean_token_accuracy": 0.962063729763031,
+      "num_tokens": 885944.0,
+      "step": 413
+    },
+    {
+      "epoch": 0.003422081518280032,
+      "grad_norm": 0.45049723982810974,
+      "learning_rate": 4.991465460947768e-05,
+      "loss": 0.2465,
+      "mean_token_accuracy": 0.9426934123039246,
+      "num_tokens": 888044.0,
+      "step": 414
+    },
+    {
+      "epoch": 0.003430347415667182,
+      "grad_norm": 0.33216142654418945,
+      "learning_rate": 4.9914447962043006e-05,
+      "loss": 0.1911,
+      "mean_token_accuracy": 0.9613648056983948,
+      "num_tokens": 890043.0,
+      "step": 415
+    },
+    {
+      "epoch": 0.0034386133130543317,
+      "grad_norm": 0.3986356258392334,
+      "learning_rate": 4.991424131460832e-05,
+      "loss": 0.1857,
+      "mean_token_accuracy": 0.9615573883056641,
+      "num_tokens": 892078.0,
+      "step": 416
+    },
+    {
+      "epoch": 0.0034468792104414814,
+      "grad_norm": 0.4196581542491913,
+      "learning_rate": 4.991403466717364e-05,
+      "loss": 0.2547,
+      "mean_token_accuracy": 0.9481813907623291,
+      "num_tokens": 894091.0,
+      "step": 417
+    },
+    {
+      "epoch": 0.0034551451078286316,
+      "grad_norm": 0.43260493874549866,
+      "learning_rate": 4.9913828019738964e-05,
+      "loss": 0.2203,
+      "mean_token_accuracy": 0.9589322209358215,
+      "num_tokens": 896045.0,
+      "step": 418
+    },
+    {
+      "epoch": 0.0034634110052157814,
+      "grad_norm": 0.352995365858078,
+      "learning_rate": 4.9913621372304286e-05,
+      "loss": 0.1628,
+      "mean_token_accuracy": 0.9689440727233887,
+      "num_tokens": 897983.0,
+      "step": 419
+    },
+    {
+      "epoch": 0.003471676902602931,
+      "grad_norm": 0.4919193387031555,
+      "learning_rate": 4.991341472486961e-05,
+      "loss": 0.2599,
+      "mean_token_accuracy": 0.9484337568283081,
+      "num_tokens": 900064.0,
+      "step": 420
+    },
+    {
+      "epoch": 0.003479942799990081,
+      "grad_norm": 0.37560829520225525,
+      "learning_rate": 4.991320807743493e-05,
+      "loss": 0.1983,
+      "mean_token_accuracy": 0.9549281597137451,
+      "num_tokens": 902089.0,
+      "step": 421
+    },
+    {
+      "epoch": 0.0034882086973772306,
+      "grad_norm": 0.37490546703338623,
+      "learning_rate": 4.991300143000025e-05,
+      "loss": 0.191,
+      "mean_token_accuracy": 0.9587525129318237,
+      "num_tokens": 904083.0,
+      "step": 422
+    },
+    {
+      "epoch": 0.0034964745947643804,
+      "grad_norm": 0.3598591685295105,
+      "learning_rate": 4.991279478256557e-05,
+      "loss": 0.2153,
+      "mean_token_accuracy": 0.9576482176780701,
+      "num_tokens": 906096.0,
+      "step": 423
+    },
+    {
+      "epoch": 0.0035047404921515306,
+      "grad_norm": 0.4452194273471832,
+      "learning_rate": 4.9912588135130894e-05,
+      "loss": 0.2417,
+      "mean_token_accuracy": 0.9560276865959167,
+      "num_tokens": 908126.0,
+      "step": 424
+    },
+    {
+      "epoch": 0.0035130063895386803,
+      "grad_norm": 0.3407069444656372,
+      "learning_rate": 4.9912381487696216e-05,
+      "loss": 0.1722,
+      "mean_token_accuracy": 0.9696816802024841,
+      "num_tokens": 910111.0,
+      "step": 425
+    },
+    {
+      "epoch": 0.00352127228692583,
+      "grad_norm": 0.2964221239089966,
+      "learning_rate": 4.991217484026153e-05,
+      "loss": 0.1468,
+      "mean_token_accuracy": 0.9682619571685791,
+      "num_tokens": 912102.0,
+      "step": 426
+    },
+    {
+      "epoch": 0.00352953818431298,
+      "grad_norm": 0.44236114621162415,
+      "learning_rate": 4.991196819282686e-05,
+      "loss": 0.236,
+      "mean_token_accuracy": 0.951884925365448,
+      "num_tokens": 914124.0,
+      "step": 427
+    },
+    {
+      "epoch": 0.0035378040817001296,
+      "grad_norm": 0.44241392612457275,
+      "learning_rate": 4.9911761545392174e-05,
+      "loss": 0.2342,
+      "mean_token_accuracy": 0.9520479440689087,
+      "num_tokens": 916132.0,
+      "step": 428
+    },
+    {
+      "epoch": 0.00354606997908728,
+      "grad_norm": 0.44964122772216797,
+      "learning_rate": 4.9911554897957496e-05,
+      "loss": 0.2244,
+      "mean_token_accuracy": 0.9514129757881165,
+      "num_tokens": 918155.0,
+      "step": 429
+    },
+    {
+      "epoch": 0.0035543358764744295,
+      "grad_norm": 0.4190461039543152,
+      "learning_rate": 4.9911348250522824e-05,
+      "loss": 0.1753,
+      "mean_token_accuracy": 0.9613478779792786,
+      "num_tokens": 920179.0,
+      "step": 430
+    },
+    {
+      "epoch": 0.0035626017738615793,
+      "grad_norm": 0.33894556760787964,
+      "learning_rate": 4.991114160308814e-05,
+      "loss": 0.2183,
+      "mean_token_accuracy": 0.9519563913345337,
+      "num_tokens": 922204.0,
+      "step": 431
+    },
+    {
+      "epoch": 0.003570867671248729,
+      "grad_norm": 0.40855875611305237,
+      "learning_rate": 4.991093495565347e-05,
+      "loss": 0.2215,
+      "mean_token_accuracy": 0.9560710787773132,
+      "num_tokens": 924236.0,
+      "step": 432
+    },
+    {
+      "epoch": 0.003579133568635879,
+      "grad_norm": 0.33665502071380615,
+      "learning_rate": 4.991072830821878e-05,
+      "loss": 0.159,
+      "mean_token_accuracy": 0.9615960121154785,
+      "num_tokens": 926247.0,
+      "step": 433
+    },
+    {
+      "epoch": 0.003587399466023029,
+      "grad_norm": 0.4639236330986023,
+      "learning_rate": 4.9910521660784104e-05,
+      "loss": 0.2001,
+      "mean_token_accuracy": 0.9574148058891296,
+      "num_tokens": 928249.0,
+      "step": 434
+    },
+    {
+      "epoch": 0.0035956653634101787,
+      "grad_norm": 0.33748435974121094,
+      "learning_rate": 4.991031501334943e-05,
+      "loss": 0.1833,
+      "mean_token_accuracy": 0.9631336331367493,
+      "num_tokens": 930208.0,
+      "step": 435
+    },
+    {
+      "epoch": 0.0036039312607973285,
+      "grad_norm": 0.3085154891014099,
+      "learning_rate": 4.991010836591475e-05,
+      "loss": 0.1397,
+      "mean_token_accuracy": 0.9710744023323059,
+      "num_tokens": 932150.0,
+      "step": 436
+    },
+    {
+      "epoch": 0.0036121971581844783,
+      "grad_norm": 0.31141841411590576,
+      "learning_rate": 4.990990171848007e-05,
+      "loss": 0.1502,
+      "mean_token_accuracy": 0.9713847637176514,
+      "num_tokens": 934113.0,
+      "step": 437
+    },
+    {
+      "epoch": 0.003620463055571628,
+      "grad_norm": 0.3119066655635834,
+      "learning_rate": 4.990969507104539e-05,
+      "loss": 0.1712,
+      "mean_token_accuracy": 0.9675257802009583,
+      "num_tokens": 936059.0,
+      "step": 438
+    },
+    {
+      "epoch": 0.0036287289529587778,
+      "grad_norm": 0.3771989643573761,
+      "learning_rate": 4.990948842361071e-05,
+      "loss": 0.2657,
+      "mean_token_accuracy": 0.9486803412437439,
+      "num_tokens": 938111.0,
+      "step": 439
+    },
+    {
+      "epoch": 0.003636994850345928,
+      "grad_norm": 0.39734941720962524,
+      "learning_rate": 4.9909281776176034e-05,
+      "loss": 0.1861,
+      "mean_token_accuracy": 0.9546574950218201,
+      "num_tokens": 940146.0,
+      "step": 440
+    },
+    {
+      "epoch": 0.0036452607477330777,
+      "grad_norm": 0.4957985281944275,
+      "learning_rate": 4.9909075128741355e-05,
+      "loss": 0.3424,
+      "mean_token_accuracy": 0.9304511547088623,
+      "num_tokens": 942280.0,
+      "step": 441
+    },
+    {
+      "epoch": 0.0036535266451202275,
+      "grad_norm": 0.4146612584590912,
+      "learning_rate": 4.990886848130668e-05,
+      "loss": 0.2766,
+      "mean_token_accuracy": 0.9470821022987366,
+      "num_tokens": 944308.0,
+      "step": 442
+    },
+    {
+      "epoch": 0.003661792542507377,
+      "grad_norm": 0.43134474754333496,
+      "learning_rate": 4.990866183387199e-05,
+      "loss": 0.2234,
+      "mean_token_accuracy": 0.9567404389381409,
+      "num_tokens": 946302.0,
+      "step": 443
+    },
+    {
+      "epoch": 0.003670058439894527,
+      "grad_norm": 0.3586181402206421,
+      "learning_rate": 4.990845518643732e-05,
+      "loss": 0.1804,
+      "mean_token_accuracy": 0.9633960127830505,
+      "num_tokens": 948275.0,
+      "step": 444
+    },
+    {
+      "epoch": 0.003678324337281677,
+      "grad_norm": 0.35363054275512695,
+      "learning_rate": 4.990824853900264e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.965323805809021,
+      "num_tokens": 950242.0,
+      "step": 445
+    },
+    {
+      "epoch": 0.003686590234668827,
+      "grad_norm": 0.4113800823688507,
+      "learning_rate": 4.990804189156796e-05,
+      "loss": 0.2247,
+      "mean_token_accuracy": 0.9484536051750183,
+      "num_tokens": 952285.0,
+      "step": 446
+    },
+    {
+      "epoch": 0.0036948561320559767,
+      "grad_norm": 0.4207867383956909,
+      "learning_rate": 4.9907835244133285e-05,
+      "loss": 0.2419,
+      "mean_token_accuracy": 0.9496123790740967,
+      "num_tokens": 954355.0,
+      "step": 447
+    },
+    {
+      "epoch": 0.0037031220294431264,
+      "grad_norm": 0.4202803671360016,
+      "learning_rate": 4.99076285966986e-05,
+      "loss": 0.244,
+      "mean_token_accuracy": 0.9511221647262573,
+      "num_tokens": 956366.0,
+      "step": 448
+    },
+    {
+      "epoch": 0.003711387926830276,
+      "grad_norm": 0.4157666862010956,
+      "learning_rate": 4.990742194926392e-05,
+      "loss": 0.2257,
+      "mean_token_accuracy": 0.9516527056694031,
+      "num_tokens": 958399.0,
+      "step": 449
+    },
+    {
+      "epoch": 0.0037196538242174264,
+      "grad_norm": 0.41051551699638367,
+      "learning_rate": 4.9907215301829244e-05,
+      "loss": 0.1938,
+      "mean_token_accuracy": 0.9570707082748413,
+      "num_tokens": 960385.0,
+      "step": 450
+    },
+    {
+      "epoch": 0.003727919721604576,
+      "grad_norm": 0.395478755235672,
+      "learning_rate": 4.9907008654394565e-05,
+      "loss": 0.1661,
+      "mean_token_accuracy": 0.9637279510498047,
+      "num_tokens": 962376.0,
+      "step": 451
+    },
+    {
+      "epoch": 0.003736185618991726,
+      "grad_norm": 0.385276734828949,
+      "learning_rate": 4.9906802006959894e-05,
+      "loss": 0.2134,
+      "mean_token_accuracy": 0.9554675817489624,
+      "num_tokens": 964403.0,
+      "step": 452
+    },
+    {
+      "epoch": 0.0037444515163788756,
+      "grad_norm": 0.44546204805374146,
+      "learning_rate": 4.990659535952521e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9638988971710205,
+      "num_tokens": 966348.0,
+      "step": 453
+    },
+    {
+      "epoch": 0.0037527174137660254,
+      "grad_norm": 0.4484986662864685,
+      "learning_rate": 4.990638871209053e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.968008279800415,
+      "num_tokens": 968292.0,
+      "step": 454
+    },
+    {
+      "epoch": 0.003760983311153175,
+      "grad_norm": 0.37638169527053833,
+      "learning_rate": 4.990618206465585e-05,
+      "loss": 0.1986,
+      "mean_token_accuracy": 0.9592137336730957,
+      "num_tokens": 970333.0,
+      "step": 455
+    },
+    {
+      "epoch": 0.0037692492085403253,
+      "grad_norm": 0.4274209439754486,
+      "learning_rate": 4.9905975417221174e-05,
+      "loss": 0.2232,
+      "mean_token_accuracy": 0.9558312892913818,
+      "num_tokens": 972354.0,
+      "step": 456
+    },
+    {
+      "epoch": 0.003777515105927475,
+      "grad_norm": 0.32814547419548035,
+      "learning_rate": 4.9905768769786495e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9623304605484009,
+      "num_tokens": 974351.0,
+      "step": 457
+    },
+    {
+      "epoch": 0.003785781003314625,
+      "grad_norm": 0.3761850893497467,
+      "learning_rate": 4.990556212235182e-05,
+      "loss": 0.2242,
+      "mean_token_accuracy": 0.9508599638938904,
+      "num_tokens": 976392.0,
+      "step": 458
+    },
+    {
+      "epoch": 0.0037940469007017746,
+      "grad_norm": 0.32914650440216064,
+      "learning_rate": 4.990535547491714e-05,
+      "loss": 0.1801,
+      "mean_token_accuracy": 0.9625641107559204,
+      "num_tokens": 978348.0,
+      "step": 459
+    },
+    {
+      "epoch": 0.0038023127980889243,
+      "grad_norm": 0.43531864881515503,
+      "learning_rate": 4.990514882748245e-05,
+      "loss": 0.2171,
+      "mean_token_accuracy": 0.9565646648406982,
+      "num_tokens": 980380.0,
+      "step": 460
+    },
+    {
+      "epoch": 0.0038105786954760745,
+      "grad_norm": 0.4166911244392395,
+      "learning_rate": 4.990494218004778e-05,
+      "loss": 0.267,
+      "mean_token_accuracy": 0.9429124593734741,
+      "num_tokens": 982453.0,
+      "step": 461
+    },
+    {
+      "epoch": 0.0038188445928632243,
+      "grad_norm": 0.3695017397403717,
+      "learning_rate": 4.9904735532613103e-05,
+      "loss": 0.161,
+      "mean_token_accuracy": 0.9642309546470642,
+      "num_tokens": 984416.0,
+      "step": 462
+    },
+    {
+      "epoch": 0.003827110490250374,
+      "grad_norm": 0.34344106912612915,
+      "learning_rate": 4.990452888517842e-05,
+      "loss": 0.1761,
+      "mean_token_accuracy": 0.9605394601821899,
+      "num_tokens": 986424.0,
+      "step": 463
+    },
+    {
+      "epoch": 0.003835376387637524,
+      "grad_norm": 0.3933105766773224,
+      "learning_rate": 4.990432223774375e-05,
+      "loss": 0.218,
+      "mean_token_accuracy": 0.9563472270965576,
+      "num_tokens": 988423.0,
+      "step": 464
+    },
+    {
+      "epoch": 0.0038436422850246735,
+      "grad_norm": 0.3542276918888092,
+      "learning_rate": 4.990411559030906e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9633774161338806,
+      "num_tokens": 990395.0,
+      "step": 465
+    },
+    {
+      "epoch": 0.0038519081824118237,
+      "grad_norm": 0.450093150138855,
+      "learning_rate": 4.990390894287438e-05,
+      "loss": 0.1761,
+      "mean_token_accuracy": 0.9630753397941589,
+      "num_tokens": 992378.0,
+      "step": 466
+    },
+    {
+      "epoch": 0.0038601740797989735,
+      "grad_norm": 0.3510618507862091,
+      "learning_rate": 4.9903702295439705e-05,
+      "loss": 0.194,
+      "mean_token_accuracy": 0.9634332060813904,
+      "num_tokens": 994353.0,
+      "step": 467
+    },
+    {
+      "epoch": 0.0038684399771861232,
+      "grad_norm": 0.41123032569885254,
+      "learning_rate": 4.990349564800503e-05,
+      "loss": 0.2503,
+      "mean_token_accuracy": 0.9509803652763367,
+      "num_tokens": 996399.0,
+      "step": 468
+    },
+    {
+      "epoch": 0.003876705874573273,
+      "grad_norm": 0.43580517172813416,
+      "learning_rate": 4.990328900057035e-05,
+      "loss": 0.2306,
+      "mean_token_accuracy": 0.9522872567176819,
+      "num_tokens": 998438.0,
+      "step": 469
+    },
+    {
+      "epoch": 0.0038849717719604228,
+      "grad_norm": 0.39722317457199097,
+      "learning_rate": 4.990308235313567e-05,
+      "loss": 0.1887,
+      "mean_token_accuracy": 0.9601010084152222,
+      "num_tokens": 1000424.0,
+      "step": 470
+    },
+    {
+      "epoch": 0.0038932376693475725,
+      "grad_norm": 0.4448322355747223,
+      "learning_rate": 4.990287570570099e-05,
+      "loss": 0.2317,
+      "mean_token_accuracy": 0.9544088244438171,
+      "num_tokens": 1002426.0,
+      "step": 471
+    },
+    {
+      "epoch": 0.0039015035667347227,
+      "grad_norm": 0.34188705682754517,
+      "learning_rate": 4.990266905826631e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9652568101882935,
+      "num_tokens": 1004418.0,
+      "step": 472
+    },
+    {
+      "epoch": 0.0039097694641218724,
+      "grad_norm": 0.36124542355537415,
+      "learning_rate": 4.9902462410831635e-05,
+      "loss": 0.2059,
+      "mean_token_accuracy": 0.9597585797309875,
+      "num_tokens": 1006412.0,
+      "step": 473
+    },
+    {
+      "epoch": 0.003918035361509023,
+      "grad_norm": 0.4938713014125824,
+      "learning_rate": 4.9902255763396957e-05,
+      "loss": 0.2303,
+      "mean_token_accuracy": 0.9551724195480347,
+      "num_tokens": 1008448.0,
+      "step": 474
+    },
+    {
+      "epoch": 0.003926301258896172,
+      "grad_norm": 0.39540615677833557,
+      "learning_rate": 4.990204911596228e-05,
+      "loss": 0.2264,
+      "mean_token_accuracy": 0.9539309144020081,
+      "num_tokens": 1010451.0,
+      "step": 475
+    },
+    {
+      "epoch": 0.003934567156283322,
+      "grad_norm": 0.4768139123916626,
+      "learning_rate": 4.99018424685276e-05,
+      "loss": 0.1835,
+      "mean_token_accuracy": 0.9609683752059937,
+      "num_tokens": 1012481.0,
+      "step": 476
+    },
+    {
+      "epoch": 0.0039428330536704715,
+      "grad_norm": 0.329805850982666,
+      "learning_rate": 4.9901635821092915e-05,
+      "loss": 0.1326,
+      "mean_token_accuracy": 0.9701799750328064,
+      "num_tokens": 1014432.0,
+      "step": 477
+    },
+    {
+      "epoch": 0.003951098951057622,
+      "grad_norm": 0.37058165669441223,
+      "learning_rate": 4.990142917365824e-05,
+      "loss": 0.1867,
+      "mean_token_accuracy": 0.9611111283302307,
+      "num_tokens": 1016418.0,
+      "step": 478
+    },
+    {
+      "epoch": 0.003959364848444771,
+      "grad_norm": 0.5383991599082947,
+      "learning_rate": 4.9901222526223565e-05,
+      "loss": 0.2183,
+      "mean_token_accuracy": 0.9517892599105835,
+      "num_tokens": 1018436.0,
+      "step": 479
+    },
+    {
+      "epoch": 0.003967630745831921,
+      "grad_norm": 0.670604407787323,
+      "learning_rate": 4.990101587878888e-05,
+      "loss": 0.2127,
+      "mean_token_accuracy": 0.9586276412010193,
+      "num_tokens": 1020424.0,
+      "step": 480
+    },
+    {
+      "epoch": 0.003975896643219071,
+      "grad_norm": 0.4291754961013794,
+      "learning_rate": 4.990080923135421e-05,
+      "loss": 0.2622,
+      "mean_token_accuracy": 0.9468292593955994,
+      "num_tokens": 1022480.0,
+      "step": 481
+    },
+    {
+      "epoch": 0.003984162540606221,
+      "grad_norm": 0.5822583436965942,
+      "learning_rate": 4.990060258391952e-05,
+      "loss": 0.1756,
+      "mean_token_accuracy": 0.963764488697052,
+      "num_tokens": 1024473.0,
+      "step": 482
+    },
+    {
+      "epoch": 0.003992428437993371,
+      "grad_norm": 0.4418441951274872,
+      "learning_rate": 4.9900395936484845e-05,
+      "loss": 0.2186,
+      "mean_token_accuracy": 0.9539666771888733,
+      "num_tokens": 1026521.0,
+      "step": 483
+    },
+    {
+      "epoch": 0.00400069433538052,
+      "grad_norm": 0.39106106758117676,
+      "learning_rate": 4.990018928905017e-05,
+      "loss": 0.2278,
+      "mean_token_accuracy": 0.9536159634590149,
+      "num_tokens": 1028532.0,
+      "step": 484
+    },
+    {
+      "epoch": 0.00400896023276767,
+      "grad_norm": 0.40405771136283875,
+      "learning_rate": 4.989998264161549e-05,
+      "loss": 0.1978,
+      "mean_token_accuracy": 0.9608134627342224,
+      "num_tokens": 1030554.0,
+      "step": 485
+    },
+    {
+      "epoch": 0.0040172261301548206,
+      "grad_norm": 0.503392219543457,
+      "learning_rate": 4.989977599418081e-05,
+      "loss": 0.2308,
+      "mean_token_accuracy": 0.9507537484169006,
+      "num_tokens": 1032550.0,
+      "step": 486
+    },
+    {
+      "epoch": 0.00402549202754197,
+      "grad_norm": 0.3696815073490143,
+      "learning_rate": 4.989956934674613e-05,
+      "loss": 0.1455,
+      "mean_token_accuracy": 0.9691991806030273,
+      "num_tokens": 1034504.0,
+      "step": 487
+    },
+    {
+      "epoch": 0.00403375792492912,
+      "grad_norm": 0.4207642674446106,
+      "learning_rate": 4.989936269931145e-05,
+      "loss": 0.2391,
+      "mean_token_accuracy": 0.9518190622329712,
+      "num_tokens": 1036544.0,
+      "step": 488
+    },
+    {
+      "epoch": 0.004042023822316269,
+      "grad_norm": 0.41360029578208923,
+      "learning_rate": 4.9899156051876775e-05,
+      "loss": 0.2544,
+      "mean_token_accuracy": 0.9499514102935791,
+      "num_tokens": 1038608.0,
+      "step": 489
+    },
+    {
+      "epoch": 0.00405028971970342,
+      "grad_norm": 0.4410945475101471,
+      "learning_rate": 4.9898949404442096e-05,
+      "loss": 0.2673,
+      "mean_token_accuracy": 0.9487665891647339,
+      "num_tokens": 1040722.0,
+      "step": 490
+    },
+    {
+      "epoch": 0.00405855561709057,
+      "grad_norm": 0.41227084398269653,
+      "learning_rate": 4.989874275700742e-05,
+      "loss": 0.2378,
+      "mean_token_accuracy": 0.9528255462646484,
+      "num_tokens": 1042763.0,
+      "step": 491
+    },
+    {
+      "epoch": 0.004066821514477719,
+      "grad_norm": 0.3907535970211029,
+      "learning_rate": 4.989853610957273e-05,
+      "loss": 0.154,
+      "mean_token_accuracy": 0.9658163189888,
+      "num_tokens": 1044729.0,
+      "step": 492
+    },
+    {
+      "epoch": 0.004075087411864869,
+      "grad_norm": 0.41625314950942993,
+      "learning_rate": 4.989832946213806e-05,
+      "loss": 0.2214,
+      "mean_token_accuracy": 0.949009895324707,
+      "num_tokens": 1046755.0,
+      "step": 493
+    },
+    {
+      "epoch": 0.004083353309252019,
+      "grad_norm": 0.3951849937438965,
+      "learning_rate": 4.989812281470338e-05,
+      "loss": 0.2194,
+      "mean_token_accuracy": 0.9571072459220886,
+      "num_tokens": 1048766.0,
+      "step": 494
+    },
+    {
+      "epoch": 0.004091619206639169,
+      "grad_norm": 0.38774311542510986,
+      "learning_rate": 4.9897916167268705e-05,
+      "loss": 0.1921,
+      "mean_token_accuracy": 0.9617321491241455,
+      "num_tokens": 1050758.0,
+      "step": 495
+    },
+    {
+      "epoch": 0.004099885104026319,
+      "grad_norm": 0.360399067401886,
+      "learning_rate": 4.9897709519834026e-05,
+      "loss": 0.1536,
+      "mean_token_accuracy": 0.9669007062911987,
+      "num_tokens": 1052758.0,
+      "step": 496
+    },
+    {
+      "epoch": 0.004108151001413468,
+      "grad_norm": 0.4067020118236542,
+      "learning_rate": 4.989750287239934e-05,
+      "loss": 0.2394,
+      "mean_token_accuracy": 0.9497784376144409,
+      "num_tokens": 1054795.0,
+      "step": 497
+    },
+    {
+      "epoch": 0.0041164168988006185,
+      "grad_norm": 0.38551297783851624,
+      "learning_rate": 4.989729622496467e-05,
+      "loss": 0.2336,
+      "mean_token_accuracy": 0.9515331387519836,
+      "num_tokens": 1056823.0,
+      "step": 498
+    },
+    {
+      "epoch": 0.004124682796187768,
+      "grad_norm": 0.4567355811595917,
+      "learning_rate": 4.9897089577529984e-05,
+      "loss": 0.1985,
+      "mean_token_accuracy": 0.9577252268791199,
+      "num_tokens": 1058816.0,
+      "step": 499
+    },
+    {
+      "epoch": 0.004132948693574918,
+      "grad_norm": 0.3516772389411926,
+      "learning_rate": 4.9896882930095306e-05,
+      "loss": 0.192,
+      "mean_token_accuracy": 0.9586485028266907,
+      "num_tokens": 1060805.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.004141214590962068,
+      "grad_norm": 0.3619779348373413,
+      "learning_rate": 4.9896676282660635e-05,
+      "loss": 0.2291,
+      "mean_token_accuracy": 0.9540967345237732,
+      "num_tokens": 1062837.0,
+      "step": 501
+    },
+    {
+      "epoch": 0.0041494804883492175,
+      "grad_norm": 0.3610847592353821,
+      "learning_rate": 4.989646963522595e-05,
+      "loss": 0.1744,
+      "mean_token_accuracy": 0.967725396156311,
+      "num_tokens": 1064795.0,
+      "step": 502
+    },
+    {
+      "epoch": 0.004157746385736368,
+      "grad_norm": 0.42383328080177307,
+      "learning_rate": 4.989626298779127e-05,
+      "loss": 0.2415,
+      "mean_token_accuracy": 0.9517003297805786,
+      "num_tokens": 1066830.0,
+      "step": 503
+    },
+    {
+      "epoch": 0.004166012283123517,
+      "grad_norm": 0.3487216532230377,
+      "learning_rate": 4.989605634035659e-05,
+      "loss": 0.1809,
+      "mean_token_accuracy": 0.9627016186714172,
+      "num_tokens": 1068820.0,
+      "step": 504
+    },
+    {
+      "epoch": 0.004174278180510667,
+      "grad_norm": 0.3901805281639099,
+      "learning_rate": 4.9895849692921914e-05,
+      "loss": 0.2177,
+      "mean_token_accuracy": 0.9561144113540649,
+      "num_tokens": 1070854.0,
+      "step": 505
+    },
+    {
+      "epoch": 0.004182544077897817,
+      "grad_norm": 0.4214717447757721,
+      "learning_rate": 4.9895643045487236e-05,
+      "loss": 0.1871,
+      "mean_token_accuracy": 0.9616151452064514,
+      "num_tokens": 1072866.0,
+      "step": 506
+    },
+    {
+      "epoch": 0.004190809975284967,
+      "grad_norm": 0.40214648842811584,
+      "learning_rate": 4.989543639805256e-05,
+      "loss": 0.2096,
+      "mean_token_accuracy": 0.9554656147956848,
+      "num_tokens": 1074848.0,
+      "step": 507
+    },
+    {
+      "epoch": 0.004199075872672117,
+      "grad_norm": 0.34685319662094116,
+      "learning_rate": 4.989522975061788e-05,
+      "loss": 0.1612,
+      "mean_token_accuracy": 0.9637755155563354,
+      "num_tokens": 1076814.0,
+      "step": 508
+    },
+    {
+      "epoch": 0.004207341770059266,
+      "grad_norm": 0.37900763750076294,
+      "learning_rate": 4.9895023103183194e-05,
+      "loss": 0.1585,
+      "mean_token_accuracy": 0.9673202633857727,
+      "num_tokens": 1078809.0,
+      "step": 509
+    },
+    {
+      "epoch": 0.004215607667446416,
+      "grad_norm": 0.42314958572387695,
+      "learning_rate": 4.989481645574852e-05,
+      "loss": 0.2021,
+      "mean_token_accuracy": 0.9601792097091675,
+      "num_tokens": 1080824.0,
+      "step": 510
+    },
+    {
+      "epoch": 0.004223873564833567,
+      "grad_norm": 0.6958469152450562,
+      "learning_rate": 4.9894609808313844e-05,
+      "loss": 0.244,
+      "mean_token_accuracy": 0.9466732144355774,
+      "num_tokens": 1082874.0,
+      "step": 511
+    },
+    {
+      "epoch": 0.004232139462220716,
+      "grad_norm": 0.41557443141937256,
+      "learning_rate": 4.989440316087916e-05,
+      "loss": 0.169,
+      "mean_token_accuracy": 0.9666160941123962,
+      "num_tokens": 1084857.0,
+      "step": 512
+    },
+    {
+      "epoch": 0.004240405359607866,
+      "grad_norm": 0.41902002692222595,
+      "learning_rate": 4.989419651344449e-05,
+      "loss": 0.2198,
+      "mean_token_accuracy": 0.9581465125083923,
+      "num_tokens": 1086870.0,
+      "step": 513
+    },
+    {
+      "epoch": 0.004248671256995015,
+      "grad_norm": 0.4280377924442291,
+      "learning_rate": 4.98939898660098e-05,
+      "loss": 0.1634,
+      "mean_token_accuracy": 0.9616161584854126,
+      "num_tokens": 1088856.0,
+      "step": 514
+    },
+    {
+      "epoch": 0.004256937154382166,
+      "grad_norm": 0.4948086738586426,
+      "learning_rate": 4.989378321857513e-05,
+      "loss": 0.3042,
+      "mean_token_accuracy": 0.9396718144416809,
+      "num_tokens": 1090934.0,
+      "step": 515
+    },
+    {
+      "epoch": 0.004265203051769315,
+      "grad_norm": 0.3329893946647644,
+      "learning_rate": 4.9893576571140446e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.9655172228813171,
+      "num_tokens": 1092883.0,
+      "step": 516
+    },
+    {
+      "epoch": 0.004273468949156465,
+      "grad_norm": 0.5646424293518066,
+      "learning_rate": 4.989336992370577e-05,
+      "loss": 0.1971,
+      "mean_token_accuracy": 0.9614017009735107,
+      "num_tokens": 1094858.0,
+      "step": 517
+    },
+    {
+      "epoch": 0.004281734846543615,
+      "grad_norm": 0.4989800751209259,
+      "learning_rate": 4.9893163276271096e-05,
+      "loss": 0.2192,
+      "mean_token_accuracy": 0.9535906314849854,
+      "num_tokens": 1096911.0,
+      "step": 518
+    },
+    {
+      "epoch": 0.004290000743930765,
+      "grad_norm": 0.3853658139705658,
+      "learning_rate": 4.989295662883641e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.96695476770401,
+      "num_tokens": 1098884.0,
+      "step": 519
+    },
+    {
+      "epoch": 0.004298266641317915,
+      "grad_norm": 0.4520922005176544,
+      "learning_rate": 4.989274998140173e-05,
+      "loss": 0.228,
+      "mean_token_accuracy": 0.948051929473877,
+      "num_tokens": 1100969.0,
+      "step": 520
+    },
+    {
+      "epoch": 0.004306532538705064,
+      "grad_norm": 0.35607728362083435,
+      "learning_rate": 4.9892543333967054e-05,
+      "loss": 0.1509,
+      "mean_token_accuracy": 0.9689567685127258,
+      "num_tokens": 1102940.0,
+      "step": 521
+    },
+    {
+      "epoch": 0.004314798436092214,
+      "grad_norm": 0.4449218511581421,
+      "learning_rate": 4.9892336686532376e-05,
+      "loss": 0.2816,
+      "mean_token_accuracy": 0.943668782711029,
+      "num_tokens": 1105023.0,
+      "step": 522
+    },
+    {
+      "epoch": 0.0043230643334793645,
+      "grad_norm": 0.5703555941581726,
+      "learning_rate": 4.98921300390977e-05,
+      "loss": 0.1871,
+      "mean_token_accuracy": 0.9587786197662354,
+      "num_tokens": 1106994.0,
+      "step": 523
+    },
+    {
+      "epoch": 0.004331330230866514,
+      "grad_norm": 0.33718162775039673,
+      "learning_rate": 4.989192339166302e-05,
+      "loss": 0.1851,
+      "mean_token_accuracy": 0.962063729763031,
+      "num_tokens": 1108977.0,
+      "step": 524
+    },
+    {
+      "epoch": 0.004339596128253664,
+      "grad_norm": 0.4550211429595947,
+      "learning_rate": 4.989171674422834e-05,
+      "loss": 0.2616,
+      "mean_token_accuracy": 0.9500244855880737,
+      "num_tokens": 1111024.0,
+      "step": 525
+    },
+    {
+      "epoch": 0.004347862025640813,
+      "grad_norm": 0.3400307297706604,
+      "learning_rate": 4.9891510096793656e-05,
+      "loss": 0.1954,
+      "mean_token_accuracy": 0.9562153816223145,
+      "num_tokens": 1113017.0,
+      "step": 526
+    },
+    {
+      "epoch": 0.0043561279230279635,
+      "grad_norm": 0.33124852180480957,
+      "learning_rate": 4.9891303449358984e-05,
+      "loss": 0.1957,
+      "mean_token_accuracy": 0.9598191976547241,
+      "num_tokens": 1115014.0,
+      "step": 527
+    },
+    {
+      "epoch": 0.004364393820415114,
+      "grad_norm": 0.48293545842170715,
+      "learning_rate": 4.9891096801924306e-05,
+      "loss": 0.2722,
+      "mean_token_accuracy": 0.9463053941726685,
+      "num_tokens": 1117050.0,
+      "step": 528
+    },
+    {
+      "epoch": 0.004372659717802263,
+      "grad_norm": 0.30610159039497375,
+      "learning_rate": 4.989089015448962e-05,
+      "loss": 0.1334,
+      "mean_token_accuracy": 0.9716932773590088,
+      "num_tokens": 1118999.0,
+      "step": 529
+    },
+    {
+      "epoch": 0.004380925615189413,
+      "grad_norm": 0.4187106788158417,
+      "learning_rate": 4.989068350705495e-05,
+      "loss": 0.2082,
+      "mean_token_accuracy": 0.9589178562164307,
+      "num_tokens": 1121001.0,
+      "step": 530
+    },
+    {
+      "epoch": 0.0043891915125765625,
+      "grad_norm": 0.5653623342514038,
+      "learning_rate": 4.9890476859620264e-05,
+      "loss": 0.2966,
+      "mean_token_accuracy": 0.9459459185600281,
+      "num_tokens": 1123005.0,
+      "step": 531
+    },
+    {
+      "epoch": 0.004397457409963713,
+      "grad_norm": 0.41617530584335327,
+      "learning_rate": 4.9890270212185586e-05,
+      "loss": 0.2344,
+      "mean_token_accuracy": 0.9540791511535645,
+      "num_tokens": 1125058.0,
+      "step": 532
+    },
+    {
+      "epoch": 0.004405723307350863,
+      "grad_norm": 0.40406036376953125,
+      "learning_rate": 4.9890063564750914e-05,
+      "loss": 0.1903,
+      "mean_token_accuracy": 0.9607250690460205,
+      "num_tokens": 1127050.0,
+      "step": 533
+    },
+    {
+      "epoch": 0.004413989204738012,
+      "grad_norm": 0.4404779076576233,
+      "learning_rate": 4.988985691731623e-05,
+      "loss": 0.2569,
+      "mean_token_accuracy": 0.9501726627349854,
+      "num_tokens": 1129083.0,
+      "step": 534
+    },
+    {
+      "epoch": 0.004422255102125162,
+      "grad_norm": 0.3881947696208954,
+      "learning_rate": 4.988965026988155e-05,
+      "loss": 0.2466,
+      "mean_token_accuracy": 0.950326144695282,
+      "num_tokens": 1131082.0,
+      "step": 535
+    },
+    {
+      "epoch": 0.004430520999512312,
+      "grad_norm": 0.2894626557826996,
+      "learning_rate": 4.988944362244687e-05,
+      "loss": 0.1235,
+      "mean_token_accuracy": 0.973738431930542,
+      "num_tokens": 1133030.0,
+      "step": 536
+    },
+    {
+      "epoch": 0.004438786896899462,
+      "grad_norm": 0.3985994756221771,
+      "learning_rate": 4.9889236975012194e-05,
+      "loss": 0.2072,
+      "mean_token_accuracy": 0.9604662656784058,
+      "num_tokens": 1135009.0,
+      "step": 537
+    },
+    {
+      "epoch": 0.004447052794286612,
+      "grad_norm": 0.28767383098602295,
+      "learning_rate": 4.9889030327577516e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9692937731742859,
+      "num_tokens": 1136969.0,
+      "step": 538
+    },
+    {
+      "epoch": 0.0044553186916737614,
+      "grad_norm": 0.36411911249160767,
+      "learning_rate": 4.988882368014284e-05,
+      "loss": 0.198,
+      "mean_token_accuracy": 0.9573934674263,
+      "num_tokens": 1138970.0,
+      "step": 539
+    },
+    {
+      "epoch": 0.004463584589060912,
+      "grad_norm": 0.38707268238067627,
+      "learning_rate": 4.988861703270816e-05,
+      "loss": 0.2443,
+      "mean_token_accuracy": 0.947576642036438,
+      "num_tokens": 1140998.0,
+      "step": 540
+    },
+    {
+      "epoch": 0.004471850486448061,
+      "grad_norm": 0.373520165681839,
+      "learning_rate": 4.988841038527348e-05,
+      "loss": 0.2164,
+      "mean_token_accuracy": 0.9554794430732727,
+      "num_tokens": 1143048.0,
+      "step": 541
+    },
+    {
+      "epoch": 0.004480116383835211,
+      "grad_norm": 0.479813814163208,
+      "learning_rate": 4.98882037378388e-05,
+      "loss": 0.2545,
+      "mean_token_accuracy": 0.9450549483299255,
+      "num_tokens": 1145147.0,
+      "step": 542
+    },
+    {
+      "epoch": 0.004488382281222361,
+      "grad_norm": 0.39338168501853943,
+      "learning_rate": 4.9887997090404124e-05,
+      "loss": 0.1829,
+      "mean_token_accuracy": 0.9600204825401306,
+      "num_tokens": 1147104.0,
+      "step": 543
+    },
+    {
+      "epoch": 0.004496648178609511,
+      "grad_norm": 0.42352020740509033,
+      "learning_rate": 4.9887790442969445e-05,
+      "loss": 0.1925,
+      "mean_token_accuracy": 0.9606686234474182,
+      "num_tokens": 1149144.0,
+      "step": 544
+    },
+    {
+      "epoch": 0.004504914075996661,
+      "grad_norm": 0.4865024983882904,
+      "learning_rate": 4.988758379553477e-05,
+      "loss": 0.2578,
+      "mean_token_accuracy": 0.9479882121086121,
+      "num_tokens": 1151188.0,
+      "step": 545
+    },
+    {
+      "epoch": 0.00451317997338381,
+      "grad_norm": 0.4011925756931305,
+      "learning_rate": 4.988737714810008e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9657988548278809,
+      "num_tokens": 1153153.0,
+      "step": 546
+    },
+    {
+      "epoch": 0.00452144587077096,
+      "grad_norm": 0.40737876296043396,
+      "learning_rate": 4.988717050066541e-05,
+      "loss": 0.2055,
+      "mean_token_accuracy": 0.9587365984916687,
+      "num_tokens": 1155122.0,
+      "step": 547
+    },
+    {
+      "epoch": 0.00452971176815811,
+      "grad_norm": 0.476521372795105,
+      "learning_rate": 4.9886963853230725e-05,
+      "loss": 0.2476,
+      "mean_token_accuracy": 0.9504659175872803,
+      "num_tokens": 1157167.0,
+      "step": 548
+    },
+    {
+      "epoch": 0.00453797766554526,
+      "grad_norm": 0.38565853238105774,
+      "learning_rate": 4.988675720579605e-05,
+      "loss": 0.1415,
+      "mean_token_accuracy": 0.9707070589065552,
+      "num_tokens": 1159153.0,
+      "step": 549
+    },
+    {
+      "epoch": 0.00454624356293241,
+      "grad_norm": 0.4481552243232727,
+      "learning_rate": 4.9886550558361375e-05,
+      "loss": 0.2018,
+      "mean_token_accuracy": 0.963761031627655,
+      "num_tokens": 1161201.0,
+      "step": 550
+    },
+    {
+      "epoch": 0.004554509460319559,
+      "grad_norm": 0.3850753605365753,
+      "learning_rate": 4.988634391092669e-05,
+      "loss": 0.1761,
+      "mean_token_accuracy": 0.9665465950965881,
+      "num_tokens": 1163150.0,
+      "step": 551
+    },
+    {
+      "epoch": 0.0045627753577067096,
+      "grad_norm": 0.49590855836868286,
+      "learning_rate": 4.988613726349201e-05,
+      "loss": 0.1915,
+      "mean_token_accuracy": 0.9587525129318237,
+      "num_tokens": 1165144.0,
+      "step": 552
+    },
+    {
+      "epoch": 0.004571041255093859,
+      "grad_norm": 0.4352778494358063,
+      "learning_rate": 4.9885930616057334e-05,
+      "loss": 0.2394,
+      "mean_token_accuracy": 0.9542450308799744,
+      "num_tokens": 1167117.0,
+      "step": 553
+    },
+    {
+      "epoch": 0.004579307152481009,
+      "grad_norm": 0.3296060860157013,
+      "learning_rate": 4.9885723968622655e-05,
+      "loss": 0.1448,
+      "mean_token_accuracy": 0.966292142868042,
+      "num_tokens": 1169081.0,
+      "step": 554
+    },
+    {
+      "epoch": 0.004587573049868159,
+      "grad_norm": 0.33051133155822754,
+      "learning_rate": 4.988551732118798e-05,
+      "loss": 0.1931,
+      "mean_token_accuracy": 0.9642674922943115,
+      "num_tokens": 1171046.0,
+      "step": 555
+    },
+    {
+      "epoch": 0.004595838947255309,
+      "grad_norm": 0.5116681456565857,
+      "learning_rate": 4.98853106737533e-05,
+      "loss": 0.1976,
+      "mean_token_accuracy": 0.9608433842658997,
+      "num_tokens": 1173044.0,
+      "step": 556
+    },
+    {
+      "epoch": 0.004604104844642459,
+      "grad_norm": 0.3828645348548889,
+      "learning_rate": 4.988510402631862e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.9621594548225403,
+      "num_tokens": 1175032.0,
+      "step": 557
+    },
+    {
+      "epoch": 0.004612370742029608,
+      "grad_norm": 0.6423853635787964,
+      "learning_rate": 4.988489737888394e-05,
+      "loss": 0.3258,
+      "mean_token_accuracy": 0.930343508720398,
+      "num_tokens": 1177134.0,
+      "step": 558
+    },
+    {
+      "epoch": 0.004620636639416758,
+      "grad_norm": 0.456575870513916,
+      "learning_rate": 4.9884690731449264e-05,
+      "loss": 0.2104,
+      "mean_token_accuracy": 0.9585414528846741,
+      "num_tokens": 1179142.0,
+      "step": 559
+    },
+    {
+      "epoch": 0.0046289025368039085,
+      "grad_norm": 0.40300890803337097,
+      "learning_rate": 4.9884484084014585e-05,
+      "loss": 0.1759,
+      "mean_token_accuracy": 0.9616534113883972,
+      "num_tokens": 1181156.0,
+      "step": 560
+    },
+    {
+      "epoch": 0.004637168434191058,
+      "grad_norm": 0.4320686459541321,
+      "learning_rate": 4.988427743657991e-05,
+      "loss": 0.2638,
+      "mean_token_accuracy": 0.9496793150901794,
+      "num_tokens": 1183189.0,
+      "step": 561
+    },
+    {
+      "epoch": 0.004645434331578208,
+      "grad_norm": 0.31509634852409363,
+      "learning_rate": 4.988407078914523e-05,
+      "loss": 0.1451,
+      "mean_token_accuracy": 0.9720101952552795,
+      "num_tokens": 1185160.0,
+      "step": 562
+    },
+    {
+      "epoch": 0.004653700228965357,
+      "grad_norm": 0.47020307183265686,
+      "learning_rate": 4.9883864141710543e-05,
+      "loss": 0.1906,
+      "mean_token_accuracy": 0.9631525278091431,
+      "num_tokens": 1187120.0,
+      "step": 563
+    },
+    {
+      "epoch": 0.0046619661263525075,
+      "grad_norm": 0.3651725649833679,
+      "learning_rate": 4.988365749427587e-05,
+      "loss": 0.1594,
+      "mean_token_accuracy": 0.9637462496757507,
+      "num_tokens": 1189112.0,
+      "step": 564
+    },
+    {
+      "epoch": 0.004670232023739658,
+      "grad_norm": 0.31554025411605835,
+      "learning_rate": 4.988345084684119e-05,
+      "loss": 0.1853,
+      "mean_token_accuracy": 0.9606573581695557,
+      "num_tokens": 1191126.0,
+      "step": 565
+    },
+    {
+      "epoch": 0.004678497921126807,
+      "grad_norm": 0.350764662027359,
+      "learning_rate": 4.988324419940651e-05,
+      "loss": 0.16,
+      "mean_token_accuracy": 0.9633533954620361,
+      "num_tokens": 1193124.0,
+      "step": 566
+    },
+    {
+      "epoch": 0.004686763818513957,
+      "grad_norm": 0.3196449279785156,
+      "learning_rate": 4.988303755197184e-05,
+      "loss": 0.1513,
+      "mean_token_accuracy": 0.9681724905967712,
+      "num_tokens": 1195078.0,
+      "step": 567
+    },
+    {
+      "epoch": 0.0046950297159011065,
+      "grad_norm": 0.46126583218574524,
+      "learning_rate": 4.988283090453715e-05,
+      "loss": 0.2712,
+      "mean_token_accuracy": 0.9475473761558533,
+      "num_tokens": 1197143.0,
+      "step": 568
+    },
+    {
+      "epoch": 0.004703295613288257,
+      "grad_norm": 0.4376954734325409,
+      "learning_rate": 4.988262425710247e-05,
+      "loss": 0.1901,
+      "mean_token_accuracy": 0.9618902206420898,
+      "num_tokens": 1199117.0,
+      "step": 569
+    },
+    {
+      "epoch": 0.004711561510675407,
+      "grad_norm": 0.45122748613357544,
+      "learning_rate": 4.9882417609667795e-05,
+      "loss": 0.2919,
+      "mean_token_accuracy": 0.9361900091171265,
+      "num_tokens": 1201270.0,
+      "step": 570
+    },
+    {
+      "epoch": 0.004719827408062556,
+      "grad_norm": 0.5808910131454468,
+      "learning_rate": 4.988221096223312e-05,
+      "loss": 0.3158,
+      "mean_token_accuracy": 0.939496636390686,
+      "num_tokens": 1203342.0,
+      "step": 571
+    },
+    {
+      "epoch": 0.004728093305449706,
+      "grad_norm": 0.4194670021533966,
+      "learning_rate": 4.988200431479844e-05,
+      "loss": 0.2258,
+      "mean_token_accuracy": 0.9534653425216675,
+      "num_tokens": 1205368.0,
+      "step": 572
+    },
+    {
+      "epoch": 0.004736359202836856,
+      "grad_norm": 0.4512733519077301,
+      "learning_rate": 4.988179766736376e-05,
+      "loss": 0.1957,
+      "mean_token_accuracy": 0.9579207897186279,
+      "num_tokens": 1207394.0,
+      "step": 573
+    },
+    {
+      "epoch": 0.004744625100224006,
+      "grad_norm": 0.4763815701007843,
+      "learning_rate": 4.988159101992908e-05,
+      "loss": 0.2784,
+      "mean_token_accuracy": 0.9424564838409424,
+      "num_tokens": 1209468.0,
+      "step": 574
+    },
+    {
+      "epoch": 0.004752890997611156,
+      "grad_norm": 0.3886113464832306,
+      "learning_rate": 4.9881384372494397e-05,
+      "loss": 0.1563,
+      "mean_token_accuracy": 0.9676757454872131,
+      "num_tokens": 1211423.0,
+      "step": 575
+    },
+    {
+      "epoch": 0.004761156894998305,
+      "grad_norm": 0.449587881565094,
+      "learning_rate": 4.9881177725059725e-05,
+      "loss": 0.2726,
+      "mean_token_accuracy": 0.9422610402107239,
+      "num_tokens": 1213490.0,
+      "step": 576
+    },
+    {
+      "epoch": 0.004769422792385456,
+      "grad_norm": 0.36666297912597656,
+      "learning_rate": 4.9880971077625047e-05,
+      "loss": 0.1722,
+      "mean_token_accuracy": 0.9632241725921631,
+      "num_tokens": 1215481.0,
+      "step": 577
+    },
+    {
+      "epoch": 0.004777688689772605,
+      "grad_norm": 0.4038402736186981,
+      "learning_rate": 4.988076443019037e-05,
+      "loss": 0.2087,
+      "mean_token_accuracy": 0.9565651416778564,
+      "num_tokens": 1217490.0,
+      "step": 578
+    },
+    {
+      "epoch": 0.004785954587159755,
+      "grad_norm": 0.47383275628089905,
+      "learning_rate": 4.988055778275569e-05,
+      "loss": 0.2893,
+      "mean_token_accuracy": 0.9465240836143494,
+      "num_tokens": 1219553.0,
+      "step": 579
+    },
+    {
+      "epoch": 0.004794220484546904,
+      "grad_norm": 0.5003758668899536,
+      "learning_rate": 4.9880351135321005e-05,
+      "loss": 0.2143,
+      "mean_token_accuracy": 0.9577187895774841,
+      "num_tokens": 1221593.0,
+      "step": 580
+    },
+    {
+      "epoch": 0.004802486381934055,
+      "grad_norm": 0.4269118309020996,
+      "learning_rate": 4.988014448788633e-05,
+      "loss": 0.1874,
+      "mean_token_accuracy": 0.9630738496780396,
+      "num_tokens": 1223603.0,
+      "step": 581
+    },
+    {
+      "epoch": 0.004810752279321205,
+      "grad_norm": 0.5176770091056824,
+      "learning_rate": 4.9879937840451655e-05,
+      "loss": 0.205,
+      "mean_token_accuracy": 0.9568272829055786,
+      "num_tokens": 1225601.0,
+      "step": 582
+    },
+    {
+      "epoch": 0.004819018176708354,
+      "grad_norm": 0.36790063977241516,
+      "learning_rate": 4.987973119301697e-05,
+      "loss": 0.1652,
+      "mean_token_accuracy": 0.9640287756919861,
+      "num_tokens": 1227553.0,
+      "step": 583
+    },
+    {
+      "epoch": 0.004827284074095504,
+      "grad_norm": 0.3703189492225647,
+      "learning_rate": 4.98795245455823e-05,
+      "loss": 0.1537,
+      "mean_token_accuracy": 0.9708290696144104,
+      "num_tokens": 1229513.0,
+      "step": 584
+    },
+    {
+      "epoch": 0.004835549971482654,
+      "grad_norm": 0.8259407877922058,
+      "learning_rate": 4.987931789814761e-05,
+      "loss": 0.2563,
+      "mean_token_accuracy": 0.9576746821403503,
+      "num_tokens": 1231480.0,
+      "step": 585
+    },
+    {
+      "epoch": 0.004843815868869804,
+      "grad_norm": 0.32427746057510376,
+      "learning_rate": 4.9879111250712935e-05,
+      "loss": 0.151,
+      "mean_token_accuracy": 0.9679552316665649,
+      "num_tokens": 1233452.0,
+      "step": 586
+    },
+    {
+      "epoch": 0.004852081766256954,
+      "grad_norm": 0.42049461603164673,
+      "learning_rate": 4.9878904603278256e-05,
+      "loss": 0.188,
+      "mean_token_accuracy": 0.9590931534767151,
+      "num_tokens": 1235487.0,
+      "step": 587
+    },
+    {
+      "epoch": 0.004860347663644103,
+      "grad_norm": 0.35315588116645813,
+      "learning_rate": 4.987869795584358e-05,
+      "loss": 0.1954,
+      "mean_token_accuracy": 0.9597180485725403,
+      "num_tokens": 1237479.0,
+      "step": 588
+    },
+    {
+      "epoch": 0.0048686135610312535,
+      "grad_norm": 0.35957205295562744,
+      "learning_rate": 4.98784913084089e-05,
+      "loss": 0.2247,
+      "mean_token_accuracy": 0.955610990524292,
+      "num_tokens": 1239490.0,
+      "step": 589
+    },
+    {
+      "epoch": 0.004876879458418403,
+      "grad_norm": 0.37462323904037476,
+      "learning_rate": 4.987828466097422e-05,
+      "loss": 0.2092,
+      "mean_token_accuracy": 0.9589873552322388,
+      "num_tokens": 1241471.0,
+      "step": 590
+    },
+    {
+      "epoch": 0.004885145355805553,
+      "grad_norm": 0.3435719311237335,
+      "learning_rate": 4.987807801353954e-05,
+      "loss": 0.1895,
+      "mean_token_accuracy": 0.9608040452003479,
+      "num_tokens": 1243467.0,
+      "step": 591
+    },
+    {
+      "epoch": 0.004893411253192703,
+      "grad_norm": 0.47866758704185486,
+      "learning_rate": 4.9877871366104865e-05,
+      "loss": 0.1725,
+      "mean_token_accuracy": 0.9660144448280334,
+      "num_tokens": 1245415.0,
+      "step": 592
+    },
+    {
+      "epoch": 0.0049016771505798525,
+      "grad_norm": 0.33533069491386414,
+      "learning_rate": 4.9877664718670186e-05,
+      "loss": 0.1966,
+      "mean_token_accuracy": 0.9575212597846985,
+      "num_tokens": 1247422.0,
+      "step": 593
+    },
+    {
+      "epoch": 0.004909943047967003,
+      "grad_norm": 0.34776511788368225,
+      "learning_rate": 4.987745807123551e-05,
+      "loss": 0.1694,
+      "mean_token_accuracy": 0.9621020555496216,
+      "num_tokens": 1249407.0,
+      "step": 594
+    },
+    {
+      "epoch": 0.004918208945354152,
+      "grad_norm": 0.37613072991371155,
+      "learning_rate": 4.987725142380082e-05,
+      "loss": 0.2525,
+      "mean_token_accuracy": 0.9534085392951965,
+      "num_tokens": 1251452.0,
+      "step": 595
+    },
+    {
+      "epoch": 0.004926474842741302,
+      "grad_norm": 0.3642001450061798,
+      "learning_rate": 4.987704477636615e-05,
+      "loss": 0.183,
+      "mean_token_accuracy": 0.95839524269104,
+      "num_tokens": 1253477.0,
+      "step": 596
+    },
+    {
+      "epoch": 0.004934740740128452,
+      "grad_norm": 0.34267255663871765,
+      "learning_rate": 4.9876838128931466e-05,
+      "loss": 0.1945,
+      "mean_token_accuracy": 0.9652742743492126,
+      "num_tokens": 1255470.0,
+      "step": 597
+    },
+    {
+      "epoch": 0.004943006637515602,
+      "grad_norm": 0.3232819437980652,
+      "learning_rate": 4.987663148149679e-05,
+      "loss": 0.1791,
+      "mean_token_accuracy": 0.9632981419563293,
+      "num_tokens": 1257465.0,
+      "step": 598
+    },
+    {
+      "epoch": 0.004951272534902752,
+      "grad_norm": 0.42522719502449036,
+      "learning_rate": 4.9876424834062116e-05,
+      "loss": 0.2506,
+      "mean_token_accuracy": 0.9504613876342773,
+      "num_tokens": 1259530.0,
+      "step": 599
+    },
+    {
+      "epoch": 0.004959538432289901,
+      "grad_norm": 0.4197443127632141,
+      "learning_rate": 4.987621818662743e-05,
+      "loss": 0.2668,
+      "mean_token_accuracy": 0.9411201477050781,
+      "num_tokens": 1261625.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.004967804329677051,
+      "grad_norm": 0.33340418338775635,
+      "learning_rate": 4.987601153919276e-05,
+      "loss": 0.2257,
+      "mean_token_accuracy": 0.9577114582061768,
+      "num_tokens": 1263641.0,
+      "step": 601
+    },
+    {
+      "epoch": 0.004976070227064202,
+      "grad_norm": 0.4351325035095215,
+      "learning_rate": 4.9875804891758074e-05,
+      "loss": 0.217,
+      "mean_token_accuracy": 0.9556227922439575,
+      "num_tokens": 1265630.0,
+      "step": 602
+    },
+    {
+      "epoch": 0.004984336124451351,
+      "grad_norm": 0.3091565668582916,
+      "learning_rate": 4.9875598244323396e-05,
+      "loss": 0.1436,
+      "mean_token_accuracy": 0.9680689573287964,
+      "num_tokens": 1267609.0,
+      "step": 603
+    },
+    {
+      "epoch": 0.004992602021838501,
+      "grad_norm": 0.3690471351146698,
+      "learning_rate": 4.987539159688872e-05,
+      "loss": 0.1543,
+      "mean_token_accuracy": 0.968988299369812,
+      "num_tokens": 1269582.0,
+      "step": 604
+    },
+    {
+      "epoch": 0.0050008679192256504,
+      "grad_norm": 0.5284733176231384,
+      "learning_rate": 4.987518494945404e-05,
+      "loss": 0.2306,
+      "mean_token_accuracy": 0.9521435499191284,
+      "num_tokens": 1271594.0,
+      "step": 605
+    },
+    {
+      "epoch": 0.005009133816612801,
+      "grad_norm": 0.3853582441806793,
+      "learning_rate": 4.987497830201936e-05,
+      "loss": 0.2079,
+      "mean_token_accuracy": 0.9585798978805542,
+      "num_tokens": 1273628.0,
+      "step": 606
+    },
+    {
+      "epoch": 0.005017399713999951,
+      "grad_norm": 0.44998297095298767,
+      "learning_rate": 4.987477165458468e-05,
+      "loss": 0.1937,
+      "mean_token_accuracy": 0.9611220359802246,
+      "num_tokens": 1275666.0,
+      "step": 607
+    },
+    {
+      "epoch": 0.0050256656113871,
+      "grad_norm": 0.4008032977581024,
+      "learning_rate": 4.9874565007150004e-05,
+      "loss": 0.1793,
+      "mean_token_accuracy": 0.9632652997970581,
+      "num_tokens": 1277632.0,
+      "step": 608
+    },
+    {
+      "epoch": 0.00503393150877425,
+      "grad_norm": 0.3322504758834839,
+      "learning_rate": 4.9874358359715326e-05,
+      "loss": 0.1297,
+      "mean_token_accuracy": 0.9696813821792603,
+      "num_tokens": 1279584.0,
+      "step": 609
+    },
+    {
+      "epoch": 0.0050421974061614,
+      "grad_norm": 0.8555606603622437,
+      "learning_rate": 4.987415171228065e-05,
+      "loss": 0.2322,
+      "mean_token_accuracy": 0.9585677981376648,
+      "num_tokens": 1281545.0,
+      "step": 610
+    },
+    {
+      "epoch": 0.00505046330354855,
+      "grad_norm": 0.3547472357749939,
+      "learning_rate": 4.987394506484597e-05,
+      "loss": 0.174,
+      "mean_token_accuracy": 0.96257483959198,
+      "num_tokens": 1283555.0,
+      "step": 611
+    },
+    {
+      "epoch": 0.005058729200935699,
+      "grad_norm": 0.4231840670108795,
+      "learning_rate": 4.9873738417411284e-05,
+      "loss": 0.1811,
+      "mean_token_accuracy": 0.9599800109863281,
+      "num_tokens": 1285560.0,
+      "step": 612
+    },
+    {
+      "epoch": 0.005066995098322849,
+      "grad_norm": 0.4377189576625824,
+      "learning_rate": 4.987353176997661e-05,
+      "loss": 0.1868,
+      "mean_token_accuracy": 0.9591939449310303,
+      "num_tokens": 1287551.0,
+      "step": 613
+    },
+    {
+      "epoch": 0.0050752609957099995,
+      "grad_norm": 0.4747653007507324,
+      "learning_rate": 4.987332512254193e-05,
+      "loss": 0.2138,
+      "mean_token_accuracy": 0.9557957053184509,
+      "num_tokens": 1289593.0,
+      "step": 614
+    },
+    {
+      "epoch": 0.005083526893097149,
+      "grad_norm": 0.48416292667388916,
+      "learning_rate": 4.987311847510725e-05,
+      "loss": 0.2593,
+      "mean_token_accuracy": 0.9521912336349487,
+      "num_tokens": 1291607.0,
+      "step": 615
+    },
+    {
+      "epoch": 0.005091792790484299,
+      "grad_norm": 0.4701284170150757,
+      "learning_rate": 4.987291182767258e-05,
+      "loss": 0.2189,
+      "mean_token_accuracy": 0.9538461565971375,
+      "num_tokens": 1293628.0,
+      "step": 616
+    },
+    {
+      "epoch": 0.005100058687871448,
+      "grad_norm": 0.5288640856742859,
+      "learning_rate": 4.987270518023789e-05,
+      "loss": 0.2348,
+      "mean_token_accuracy": 0.9571072459220886,
+      "num_tokens": 1295639.0,
+      "step": 617
+    },
+    {
+      "epoch": 0.0051083245852585986,
+      "grad_norm": 0.3681744337081909,
+      "learning_rate": 4.9872498532803214e-05,
+      "loss": 0.2096,
+      "mean_token_accuracy": 0.9571356773376465,
+      "num_tokens": 1297628.0,
+      "step": 618
+    },
+    {
+      "epoch": 0.005116590482645749,
+      "grad_norm": 0.3801528513431549,
+      "learning_rate": 4.9872291885368536e-05,
+      "loss": 0.1622,
+      "mean_token_accuracy": 0.9673469662666321,
+      "num_tokens": 1299594.0,
+      "step": 619
+    },
+    {
+      "epoch": 0.005124856380032898,
+      "grad_norm": 0.33789482712745667,
+      "learning_rate": 4.987208523793386e-05,
+      "loss": 0.1518,
+      "mean_token_accuracy": 0.9684842228889465,
+      "num_tokens": 1301599.0,
+      "step": 620
+    },
+    {
+      "epoch": 0.005133122277420048,
+      "grad_norm": 0.5162757635116577,
+      "learning_rate": 4.987187859049918e-05,
+      "loss": 0.278,
+      "mean_token_accuracy": 0.9440902471542358,
+      "num_tokens": 1303644.0,
+      "step": 621
+    },
+    {
+      "epoch": 0.005141388174807198,
+      "grad_norm": 0.42620551586151123,
+      "learning_rate": 4.98716719430645e-05,
+      "loss": 0.2407,
+      "mean_token_accuracy": 0.9534772038459778,
+      "num_tokens": 1305735.0,
+      "step": 622
+    },
+    {
+      "epoch": 0.005149654072194348,
+      "grad_norm": 0.35942429304122925,
+      "learning_rate": 4.987146529562982e-05,
+      "loss": 0.2066,
+      "mean_token_accuracy": 0.9572219252586365,
+      "num_tokens": 1307728.0,
+      "step": 623
+    },
+    {
+      "epoch": 0.005157919969581498,
+      "grad_norm": 0.3561047613620758,
+      "learning_rate": 4.9871258648195144e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9616742134094238,
+      "num_tokens": 1309717.0,
+      "step": 624
+    },
+    {
+      "epoch": 0.005166185866968647,
+      "grad_norm": 0.4044671058654785,
+      "learning_rate": 4.9871052000760466e-05,
+      "loss": 0.2267,
+      "mean_token_accuracy": 0.9577534794807434,
+      "num_tokens": 1311735.0,
+      "step": 625
+    },
+    {
+      "epoch": 0.0051744517643557975,
+      "grad_norm": 0.35080039501190186,
+      "learning_rate": 4.987084535332579e-05,
+      "loss": 0.1969,
+      "mean_token_accuracy": 0.9600197672843933,
+      "num_tokens": 1313767.0,
+      "step": 626
+    },
+    {
+      "epoch": 0.005182717661742947,
+      "grad_norm": 0.420389324426651,
+      "learning_rate": 4.987063870589111e-05,
+      "loss": 0.2062,
+      "mean_token_accuracy": 0.9584980010986328,
+      "num_tokens": 1315797.0,
+      "step": 627
+    },
+    {
+      "epoch": 0.005190983559130097,
+      "grad_norm": 0.32986149191856384,
+      "learning_rate": 4.987043205845643e-05,
+      "loss": 0.1553,
+      "mean_token_accuracy": 0.9685438871383667,
+      "num_tokens": 1317774.0,
+      "step": 628
+    },
+    {
+      "epoch": 0.005199249456517247,
+      "grad_norm": 0.36671149730682373,
+      "learning_rate": 4.9870225411021746e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.965309202671051,
+      "num_tokens": 1319769.0,
+      "step": 629
+    },
+    {
+      "epoch": 0.0052075153539043965,
+      "grad_norm": 0.42939433455467224,
+      "learning_rate": 4.9870018763587074e-05,
+      "loss": 0.227,
+      "mean_token_accuracy": 0.9533600807189941,
+      "num_tokens": 1321769.0,
+      "step": 630
+    },
+    {
+      "epoch": 0.005215781251291547,
+      "grad_norm": 0.4047471582889557,
+      "learning_rate": 4.986981211615239e-05,
+      "loss": 0.2006,
+      "mean_token_accuracy": 0.9592555165290833,
+      "num_tokens": 1323763.0,
+      "step": 631
+    },
+    {
+      "epoch": 0.005224047148678696,
+      "grad_norm": 0.31303027272224426,
+      "learning_rate": 4.986960546871771e-05,
+      "loss": 0.144,
+      "mean_token_accuracy": 0.9680851101875305,
+      "num_tokens": 1325743.0,
+      "step": 632
+    },
+    {
+      "epoch": 0.005232313046065846,
+      "grad_norm": 0.39468520879745483,
+      "learning_rate": 4.986939882128304e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9617898464202881,
+      "num_tokens": 1327738.0,
+      "step": 633
+    },
+    {
+      "epoch": 0.005240578943452996,
+      "grad_norm": 0.3944607675075531,
+      "learning_rate": 4.9869192173848354e-05,
+      "loss": 0.2016,
+      "mean_token_accuracy": 0.9624624848365784,
+      "num_tokens": 1329742.0,
+      "step": 634
+    },
+    {
+      "epoch": 0.005248844840840146,
+      "grad_norm": 0.4482612907886505,
+      "learning_rate": 4.9868985526413676e-05,
+      "loss": 0.2218,
+      "mean_token_accuracy": 0.9538461565971375,
+      "num_tokens": 1331763.0,
+      "step": 635
+    },
+    {
+      "epoch": 0.005257110738227296,
+      "grad_norm": 0.5000016689300537,
+      "learning_rate": 4.9868778878979e-05,
+      "loss": 0.2209,
+      "mean_token_accuracy": 0.9566949009895325,
+      "num_tokens": 1333778.0,
+      "step": 636
+    },
+    {
+      "epoch": 0.005265376635614445,
+      "grad_norm": 0.47637665271759033,
+      "learning_rate": 4.986857223154432e-05,
+      "loss": 0.2229,
+      "mean_token_accuracy": 0.9557086825370789,
+      "num_tokens": 1335816.0,
+      "step": 637
+    },
+    {
+      "epoch": 0.005273642533001595,
+      "grad_norm": 0.42545145750045776,
+      "learning_rate": 4.986836558410964e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9671458005905151,
+      "num_tokens": 1337770.0,
+      "step": 638
+    },
+    {
+      "epoch": 0.0052819084303887456,
+      "grad_norm": 0.6827418804168701,
+      "learning_rate": 4.986815893667496e-05,
+      "loss": 0.2117,
+      "mean_token_accuracy": 0.9604462385177612,
+      "num_tokens": 1339748.0,
+      "step": 639
+    },
+    {
+      "epoch": 0.005290174327775895,
+      "grad_norm": 0.368962824344635,
+      "learning_rate": 4.9867952289240284e-05,
+      "loss": 0.1642,
+      "mean_token_accuracy": 0.9683834910392761,
+      "num_tokens": 1341715.0,
+      "step": 640
+    },
+    {
+      "epoch": 0.005298440225163045,
+      "grad_norm": 0.4205617606639862,
+      "learning_rate": 4.9867745641805606e-05,
+      "loss": 0.2486,
+      "mean_token_accuracy": 0.9519608020782471,
+      "num_tokens": 1343761.0,
+      "step": 641
+    },
+    {
+      "epoch": 0.005306706122550194,
+      "grad_norm": 0.33469706773757935,
+      "learning_rate": 4.986753899437093e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9635000228881836,
+      "num_tokens": 1345767.0,
+      "step": 642
+    },
+    {
+      "epoch": 0.005314972019937345,
+      "grad_norm": 0.44214770197868347,
+      "learning_rate": 4.986733234693625e-05,
+      "loss": 0.1848,
+      "mean_token_accuracy": 0.9607250690460205,
+      "num_tokens": 1347759.0,
+      "step": 643
+    },
+    {
+      "epoch": 0.005323237917324494,
+      "grad_norm": 0.34376171231269836,
+      "learning_rate": 4.986712569950157e-05,
+      "loss": 0.1851,
+      "mean_token_accuracy": 0.9625374674797058,
+      "num_tokens": 1349767.0,
+      "step": 644
+    },
+    {
+      "epoch": 0.005331503814711644,
+      "grad_norm": 0.41137999296188354,
+      "learning_rate": 4.986691905206689e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9691451787948608,
+      "num_tokens": 1351750.0,
+      "step": 645
+    },
+    {
+      "epoch": 0.005339769712098794,
+      "grad_norm": 0.368313193321228,
+      "learning_rate": 4.986671240463221e-05,
+      "loss": 0.1538,
+      "mean_token_accuracy": 0.9658805727958679,
+      "num_tokens": 1353749.0,
+      "step": 646
+    },
+    {
+      "epoch": 0.005348035609485944,
+      "grad_norm": 0.3550889194011688,
+      "learning_rate": 4.9866505757197536e-05,
+      "loss": 0.2014,
+      "mean_token_accuracy": 0.9619675278663635,
+      "num_tokens": 1355727.0,
+      "step": 647
+    },
+    {
+      "epoch": 0.005356301506873094,
+      "grad_norm": 0.3864426016807556,
+      "learning_rate": 4.986629910976286e-05,
+      "loss": 0.1633,
+      "mean_token_accuracy": 0.9668874144554138,
+      "num_tokens": 1357696.0,
+      "step": 648
+    },
+    {
+      "epoch": 0.005364567404260243,
+      "grad_norm": 0.3569417893886566,
+      "learning_rate": 4.986609246232817e-05,
+      "loss": 0.1506,
+      "mean_token_accuracy": 0.9674465656280518,
+      "num_tokens": 1359668.0,
+      "step": 649
+    },
+    {
+      "epoch": 0.005372833301647393,
+      "grad_norm": 0.4566428065299988,
+      "learning_rate": 4.98658858148935e-05,
+      "loss": 0.2085,
+      "mean_token_accuracy": 0.955875039100647,
+      "num_tokens": 1361691.0,
+      "step": 650
+    },
+    {
+      "epoch": 0.0053810991990345435,
+      "grad_norm": 0.35461804270744324,
+      "learning_rate": 4.9865679167458815e-05,
+      "loss": 0.2036,
+      "mean_token_accuracy": 0.9614213109016418,
+      "num_tokens": 1363667.0,
+      "step": 651
+    },
+    {
+      "epoch": 0.005389365096421693,
+      "grad_norm": 0.3737211525440216,
+      "learning_rate": 4.986547252002414e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9645177125930786,
+      "num_tokens": 1365674.0,
+      "step": 652
+    },
+    {
+      "epoch": 0.005397630993808843,
+      "grad_norm": 0.495787113904953,
+      "learning_rate": 4.986526587258946e-05,
+      "loss": 0.24,
+      "mean_token_accuracy": 0.9544103145599365,
+      "num_tokens": 1367698.0,
+      "step": 653
+    },
+    {
+      "epoch": 0.005405896891195992,
+      "grad_norm": 0.3912745714187622,
+      "learning_rate": 4.986505922515478e-05,
+      "loss": 0.2178,
+      "mean_token_accuracy": 0.955276370048523,
+      "num_tokens": 1369694.0,
+      "step": 654
+    },
+    {
+      "epoch": 0.0054141627885831425,
+      "grad_norm": 0.4028461277484894,
+      "learning_rate": 4.98648525777201e-05,
+      "loss": 0.2061,
+      "mean_token_accuracy": 0.9557957053184509,
+      "num_tokens": 1371736.0,
+      "step": 655
+    },
+    {
+      "epoch": 0.005422428685970293,
+      "grad_norm": 0.42663925886154175,
+      "learning_rate": 4.9864645930285424e-05,
+      "loss": 0.2033,
+      "mean_token_accuracy": 0.9591429829597473,
+      "num_tokens": 1373749.0,
+      "step": 656
+    },
+    {
+      "epoch": 0.005430694583357442,
+      "grad_norm": 0.4920315742492676,
+      "learning_rate": 4.9864439282850745e-05,
+      "loss": 0.274,
+      "mean_token_accuracy": 0.940487802028656,
+      "num_tokens": 1375805.0,
+      "step": 657
+    },
+    {
+      "epoch": 0.005438960480744592,
+      "grad_norm": 0.44999560713768005,
+      "learning_rate": 4.986423263541607e-05,
+      "loss": 0.1413,
+      "mean_token_accuracy": 0.9719673991203308,
+      "num_tokens": 1377773.0,
+      "step": 658
+    },
+    {
+      "epoch": 0.0054472263781317415,
+      "grad_norm": 0.5461704134941101,
+      "learning_rate": 4.986402598798139e-05,
+      "loss": 0.2902,
+      "mean_token_accuracy": 0.9418439865112305,
+      "num_tokens": 1379894.0,
+      "step": 659
+    },
+    {
+      "epoch": 0.005455492275518892,
+      "grad_norm": 0.40486854314804077,
+      "learning_rate": 4.986381934054671e-05,
+      "loss": 0.2201,
+      "mean_token_accuracy": 0.9576693177223206,
+      "num_tokens": 1381908.0,
+      "step": 660
+    },
+    {
+      "epoch": 0.005463758172906042,
+      "grad_norm": 0.39104777574539185,
+      "learning_rate": 4.986361269311203e-05,
+      "loss": 0.199,
+      "mean_token_accuracy": 0.9594391584396362,
+      "num_tokens": 1383911.0,
+      "step": 661
+    },
+    {
+      "epoch": 0.005472024070293191,
+      "grad_norm": 0.4896831512451172,
+      "learning_rate": 4.9863406045677354e-05,
+      "loss": 0.2927,
+      "mean_token_accuracy": 0.9400094747543335,
+      "num_tokens": 1386034.0,
+      "step": 662
+    },
+    {
+      "epoch": 0.005480289967680341,
+      "grad_norm": 0.39970508217811584,
+      "learning_rate": 4.986319939824267e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.96546471118927,
+      "num_tokens": 1388009.0,
+      "step": 663
+    },
+    {
+      "epoch": 0.005488555865067491,
+      "grad_norm": 0.4880525469779968,
+      "learning_rate": 4.9862992750808e-05,
+      "loss": 0.2554,
+      "mean_token_accuracy": 0.9509470462799072,
+      "num_tokens": 1390074.0,
+      "step": 664
+    },
+    {
+      "epoch": 0.005496821762454641,
+      "grad_norm": 0.38858509063720703,
+      "learning_rate": 4.986278610337332e-05,
+      "loss": 0.2024,
+      "mean_token_accuracy": 0.9606181383132935,
+      "num_tokens": 1392086.0,
+      "step": 665
+    },
+    {
+      "epoch": 0.005505087659841791,
+      "grad_norm": 0.4237862527370453,
+      "learning_rate": 4.9862579455938633e-05,
+      "loss": 0.2149,
+      "mean_token_accuracy": 0.9590163826942444,
+      "num_tokens": 1394044.0,
+      "step": 666
+    },
+    {
+      "epoch": 0.00551335355722894,
+      "grad_norm": 0.3626960813999176,
+      "learning_rate": 4.986237280850396e-05,
+      "loss": 0.1563,
+      "mean_token_accuracy": 0.9675620794296265,
+      "num_tokens": 1396023.0,
+      "step": 667
+    },
+    {
+      "epoch": 0.005521619454616091,
+      "grad_norm": 0.43807655572891235,
+      "learning_rate": 4.986216616106928e-05,
+      "loss": 0.1969,
+      "mean_token_accuracy": 0.9539892077445984,
+      "num_tokens": 1398072.0,
+      "step": 668
+    },
+    {
+      "epoch": 0.00552988535200324,
+      "grad_norm": 0.4833122789859772,
+      "learning_rate": 4.98619595136346e-05,
+      "loss": 0.2412,
+      "mean_token_accuracy": 0.9476260542869568,
+      "num_tokens": 1400121.0,
+      "step": 669
+    },
+    {
+      "epoch": 0.00553815124939039,
+      "grad_norm": 0.28978195786476135,
+      "learning_rate": 4.986175286619992e-05,
+      "loss": 0.1628,
+      "mean_token_accuracy": 0.9713701605796814,
+      "num_tokens": 1402083.0,
+      "step": 670
+    },
+    {
+      "epoch": 0.00554641714677754,
+      "grad_norm": 0.34871920943260193,
+      "learning_rate": 4.986154621876524e-05,
+      "loss": 0.1755,
+      "mean_token_accuracy": 0.9663485884666443,
+      "num_tokens": 1404080.0,
+      "step": 671
+    },
+    {
+      "epoch": 0.00555468304416469,
+      "grad_norm": 0.32516831159591675,
+      "learning_rate": 4.9861339571330563e-05,
+      "loss": 0.1559,
+      "mean_token_accuracy": 0.9718237519264221,
+      "num_tokens": 1406038.0,
+      "step": 672
+    },
+    {
+      "epoch": 0.00556294894155184,
+      "grad_norm": 0.34836143255233765,
+      "learning_rate": 4.9861132923895885e-05,
+      "loss": 0.1902,
+      "mean_token_accuracy": 0.9575308561325073,
+      "num_tokens": 1408069.0,
+      "step": 673
+    },
+    {
+      "epoch": 0.005571214838938989,
+      "grad_norm": 0.41781359910964966,
+      "learning_rate": 4.986092627646121e-05,
+      "loss": 0.2541,
+      "mean_token_accuracy": 0.9523809552192688,
+      "num_tokens": 1410112.0,
+      "step": 674
+    },
+    {
+      "epoch": 0.005579480736326139,
+      "grad_norm": 0.43378257751464844,
+      "learning_rate": 4.986071962902653e-05,
+      "loss": 0.2344,
+      "mean_token_accuracy": 0.9537628889083862,
+      "num_tokens": 1412151.0,
+      "step": 675
+    },
+    {
+      "epoch": 0.0055877466337132895,
+      "grad_norm": 0.3090350031852722,
+      "learning_rate": 4.986051298159185e-05,
+      "loss": 0.1337,
+      "mean_token_accuracy": 0.9713408350944519,
+      "num_tokens": 1414111.0,
+      "step": 676
+    },
+    {
+      "epoch": 0.005596012531100439,
+      "grad_norm": 0.3500804901123047,
+      "learning_rate": 4.986030633415717e-05,
+      "loss": 0.188,
+      "mean_token_accuracy": 0.959857702255249,
+      "num_tokens": 1416085.0,
+      "step": 677
+    },
+    {
+      "epoch": 0.005604278428487589,
+      "grad_norm": 0.38729560375213623,
+      "learning_rate": 4.9860099686722487e-05,
+      "loss": 0.252,
+      "mean_token_accuracy": 0.9494900703430176,
+      "num_tokens": 1418150.0,
+      "step": 678
+    },
+    {
+      "epoch": 0.005612544325874738,
+      "grad_norm": 0.42061978578567505,
+      "learning_rate": 4.9859893039287815e-05,
+      "loss": 0.2006,
+      "mean_token_accuracy": 0.9550173282623291,
+      "num_tokens": 1420179.0,
+      "step": 679
+    },
+    {
+      "epoch": 0.0056208102232618885,
+      "grad_norm": 0.4347386956214905,
+      "learning_rate": 4.985968639185313e-05,
+      "loss": 0.1974,
+      "mean_token_accuracy": 0.9583333134651184,
+      "num_tokens": 1422201.0,
+      "step": 680
+    },
+    {
+      "epoch": 0.005629076120649038,
+      "grad_norm": 0.3133489489555359,
+      "learning_rate": 4.985947974441845e-05,
+      "loss": 0.1303,
+      "mean_token_accuracy": 0.9709035158157349,
+      "num_tokens": 1424166.0,
+      "step": 681
+    },
+    {
+      "epoch": 0.005637342018036188,
+      "grad_norm": 0.3719736337661743,
+      "learning_rate": 4.985927309698378e-05,
+      "loss": 0.1545,
+      "mean_token_accuracy": 0.96546471118927,
+      "num_tokens": 1426141.0,
+      "step": 682
+    },
+    {
+      "epoch": 0.005645607915423338,
+      "grad_norm": 0.34936827421188354,
+      "learning_rate": 4.9859066449549095e-05,
+      "loss": 0.1429,
+      "mean_token_accuracy": 0.9701952934265137,
+      "num_tokens": 1428093.0,
+      "step": 683
+    },
+    {
+      "epoch": 0.0056538738128104876,
+      "grad_norm": 0.46590298414230347,
+      "learning_rate": 4.985885980211442e-05,
+      "loss": 0.1757,
+      "mean_token_accuracy": 0.9638971090316772,
+      "num_tokens": 1430121.0,
+      "step": 684
+    },
+    {
+      "epoch": 0.005662139710197638,
+      "grad_norm": 0.3431544303894043,
+      "learning_rate": 4.985865315467974e-05,
+      "loss": 0.1753,
+      "mean_token_accuracy": 0.9636363387107849,
+      "num_tokens": 1432107.0,
+      "step": 685
+    },
+    {
+      "epoch": 0.005670405607584787,
+      "grad_norm": 0.3336385488510132,
+      "learning_rate": 4.985844650724506e-05,
+      "loss": 0.1895,
+      "mean_token_accuracy": 0.962705135345459,
+      "num_tokens": 1434124.0,
+      "step": 686
+    },
+    {
+      "epoch": 0.005678671504971937,
+      "grad_norm": 0.3304632008075714,
+      "learning_rate": 4.985823985981039e-05,
+      "loss": 0.1595,
+      "mean_token_accuracy": 0.9657947421073914,
+      "num_tokens": 1436118.0,
+      "step": 687
+    },
+    {
+      "epoch": 0.0056869374023590874,
+      "grad_norm": 0.5806984901428223,
+      "learning_rate": 4.98580332123757e-05,
+      "loss": 0.348,
+      "mean_token_accuracy": 0.9307909607887268,
+      "num_tokens": 1438248.0,
+      "step": 688
+    },
+    {
+      "epoch": 0.005695203299746237,
+      "grad_norm": 0.3722713887691498,
+      "learning_rate": 4.9857826564941025e-05,
+      "loss": 0.2137,
+      "mean_token_accuracy": 0.9550617337226868,
+      "num_tokens": 1440279.0,
+      "step": 689
+    },
+    {
+      "epoch": 0.005703469197133387,
+      "grad_norm": 0.4312034845352173,
+      "learning_rate": 4.9857619917506346e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9646191596984863,
+      "num_tokens": 1442320.0,
+      "step": 690
+    },
+    {
+      "epoch": 0.005711735094520536,
+      "grad_norm": 0.32912158966064453,
+      "learning_rate": 4.985741327007167e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9644444584846497,
+      "num_tokens": 1444351.0,
+      "step": 691
+    },
+    {
+      "epoch": 0.0057200009919076865,
+      "grad_norm": 0.3393937051296234,
+      "learning_rate": 4.985720662263699e-05,
+      "loss": 0.1401,
+      "mean_token_accuracy": 0.9711340069770813,
+      "num_tokens": 1446297.0,
+      "step": 692
+    },
+    {
+      "epoch": 0.005728266889294837,
+      "grad_norm": 0.3419071137905121,
+      "learning_rate": 4.985699997520231e-05,
+      "loss": 0.161,
+      "mean_token_accuracy": 0.965482234954834,
+      "num_tokens": 1448273.0,
+      "step": 693
+    },
+    {
+      "epoch": 0.005736532786681986,
+      "grad_norm": 0.400582879781723,
+      "learning_rate": 4.985679332776763e-05,
+      "loss": 0.2005,
+      "mean_token_accuracy": 0.959673285484314,
+      "num_tokens": 1450238.0,
+      "step": 694
+    },
+    {
+      "epoch": 0.005744798684069136,
+      "grad_norm": 0.31403979659080505,
+      "learning_rate": 4.985658668033295e-05,
+      "loss": 0.16,
+      "mean_token_accuracy": 0.9657288789749146,
+      "num_tokens": 1452199.0,
+      "step": 695
+    },
+    {
+      "epoch": 0.0057530645814562855,
+      "grad_norm": 0.4190968871116638,
+      "learning_rate": 4.9856380032898276e-05,
+      "loss": 0.1924,
+      "mean_token_accuracy": 0.9628687500953674,
+      "num_tokens": 1454171.0,
+      "step": 696
+    },
+    {
+      "epoch": 0.005761330478843436,
+      "grad_norm": 0.3504311144351959,
+      "learning_rate": 4.98561733854636e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9641594886779785,
+      "num_tokens": 1456158.0,
+      "step": 697
+    },
+    {
+      "epoch": 0.005769596376230586,
+      "grad_norm": 0.400316447019577,
+      "learning_rate": 4.985596673802891e-05,
+      "loss": 0.1664,
+      "mean_token_accuracy": 0.9645748734474182,
+      "num_tokens": 1458140.0,
+      "step": 698
+    },
+    {
+      "epoch": 0.005777862273617735,
+      "grad_norm": 0.36822885274887085,
+      "learning_rate": 4.985576009059424e-05,
+      "loss": 0.1905,
+      "mean_token_accuracy": 0.9655516743659973,
+      "num_tokens": 1460149.0,
+      "step": 699
+    },
+    {
+      "epoch": 0.005786128171004885,
+      "grad_norm": 0.35863491892814636,
+      "learning_rate": 4.9855553443159556e-05,
+      "loss": 0.1705,
+      "mean_token_accuracy": 0.9621976017951965,
+      "num_tokens": 1462139.0,
+      "step": 700
+    },
+    {
+      "epoch": 0.005794394068392035,
+      "grad_norm": 0.3071357011795044,
+      "learning_rate": 4.985534679572488e-05,
+      "loss": 0.1452,
+      "mean_token_accuracy": 0.9687179327011108,
+      "num_tokens": 1464095.0,
+      "step": 701
+    },
+    {
+      "epoch": 0.005802659965779185,
+      "grad_norm": 0.35435113310813904,
+      "learning_rate": 4.98551401482902e-05,
+      "loss": 0.1858,
+      "mean_token_accuracy": 0.9593373537063599,
+      "num_tokens": 1466093.0,
+      "step": 702
+    },
+    {
+      "epoch": 0.005810925863166335,
+      "grad_norm": 0.4352758526802063,
+      "learning_rate": 4.985493350085552e-05,
+      "loss": 0.241,
+      "mean_token_accuracy": 0.9529354572296143,
+      "num_tokens": 1468160.0,
+      "step": 703
+    },
+    {
+      "epoch": 0.005819191760553484,
+      "grad_norm": 0.3950839042663574,
+      "learning_rate": 4.985472685342085e-05,
+      "loss": 0.1652,
+      "mean_token_accuracy": 0.9668808579444885,
+      "num_tokens": 1470189.0,
+      "step": 704
+    },
+    {
+      "epoch": 0.0058274576579406346,
+      "grad_norm": 0.37273624539375305,
+      "learning_rate": 4.9854520205986165e-05,
+      "loss": 0.2047,
+      "mean_token_accuracy": 0.9579207897186279,
+      "num_tokens": 1472215.0,
+      "step": 705
+    },
+    {
+      "epoch": 0.005835723555327784,
+      "grad_norm": 0.5543181896209717,
+      "learning_rate": 4.9854313558551486e-05,
+      "loss": 0.2127,
+      "mean_token_accuracy": 0.9566075205802917,
+      "num_tokens": 1474249.0,
+      "step": 706
+    },
+    {
+      "epoch": 0.005843989452714934,
+      "grad_norm": 0.34716540575027466,
+      "learning_rate": 4.985410691111681e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9681233763694763,
+      "num_tokens": 1476200.0,
+      "step": 707
+    },
+    {
+      "epoch": 0.005852255350102084,
+      "grad_norm": 0.38925662636756897,
+      "learning_rate": 4.985390026368213e-05,
+      "loss": 0.2115,
+      "mean_token_accuracy": 0.9577187895774841,
+      "num_tokens": 1478240.0,
+      "step": 708
+    },
+    {
+      "epoch": 0.005860521247489234,
+      "grad_norm": 0.44542396068573,
+      "learning_rate": 4.985369361624745e-05,
+      "loss": 0.2563,
+      "mean_token_accuracy": 0.9501234292984009,
+      "num_tokens": 1480271.0,
+      "step": 709
+    },
+    {
+      "epoch": 0.005868787144876384,
+      "grad_norm": 0.3741936683654785,
+      "learning_rate": 4.985348696881277e-05,
+      "loss": 0.2109,
+      "mean_token_accuracy": 0.9556331038475037,
+      "num_tokens": 1482283.0,
+      "step": 710
+    },
+    {
+      "epoch": 0.005877053042263533,
+      "grad_norm": 0.3294547498226166,
+      "learning_rate": 4.9853280321378094e-05,
+      "loss": 0.1609,
+      "mean_token_accuracy": 0.9708737730979919,
+      "num_tokens": 1484246.0,
+      "step": 711
+    },
+    {
+      "epoch": 0.005885318939650683,
+      "grad_norm": 0.3919965922832489,
+      "learning_rate": 4.985307367394341e-05,
+      "loss": 0.1829,
+      "mean_token_accuracy": 0.9602415561676025,
+      "num_tokens": 1486239.0,
+      "step": 712
+    },
+    {
+      "epoch": 0.005893584837037833,
+      "grad_norm": 0.36469322443008423,
+      "learning_rate": 4.985286702650874e-05,
+      "loss": 0.2058,
+      "mean_token_accuracy": 0.9588972330093384,
+      "num_tokens": 1488240.0,
+      "step": 713
+    },
+    {
+      "epoch": 0.005901850734424983,
+      "grad_norm": 0.4072090685367584,
+      "learning_rate": 4.985266037907406e-05,
+      "loss": 0.2107,
+      "mean_token_accuracy": 0.9603456854820251,
+      "num_tokens": 1490213.0,
+      "step": 714
+    },
+    {
+      "epoch": 0.005910116631812133,
+      "grad_norm": 0.31470343470573425,
+      "learning_rate": 4.9852453731639374e-05,
+      "loss": 0.1809,
+      "mean_token_accuracy": 0.9616342782974243,
+      "num_tokens": 1492226.0,
+      "step": 715
+    },
+    {
+      "epoch": 0.005918382529199282,
+      "grad_norm": 0.3966371715068817,
+      "learning_rate": 4.98522470842047e-05,
+      "loss": 0.1953,
+      "mean_token_accuracy": 0.964092493057251,
+      "num_tokens": 1494265.0,
+      "step": 716
+    },
+    {
+      "epoch": 0.0059266484265864325,
+      "grad_norm": 0.3984452486038208,
+      "learning_rate": 4.985204043677002e-05,
+      "loss": 0.2752,
+      "mean_token_accuracy": 0.938804566860199,
+      "num_tokens": 1496379.0,
+      "step": 717
+    },
+    {
+      "epoch": 0.005934914323973582,
+      "grad_norm": 0.36626997590065,
+      "learning_rate": 4.985183378933534e-05,
+      "loss": 0.2399,
+      "mean_token_accuracy": 0.951604962348938,
+      "num_tokens": 1498410.0,
+      "step": 718
+    },
+    {
+      "epoch": 0.005943180221360732,
+      "grad_norm": 0.5053192377090454,
+      "learning_rate": 4.985162714190066e-05,
+      "loss": 0.2012,
+      "mean_token_accuracy": 0.9546350836753845,
+      "num_tokens": 1500444.0,
+      "step": 719
+    },
+    {
+      "epoch": 0.005951446118747882,
+      "grad_norm": 0.40764230489730835,
+      "learning_rate": 4.985142049446598e-05,
+      "loss": 0.242,
+      "mean_token_accuracy": 0.9532338380813599,
+      "num_tokens": 1502460.0,
+      "step": 720
+    },
+    {
+      "epoch": 0.0059597120161350315,
+      "grad_norm": 0.3426690101623535,
+      "learning_rate": 4.9851213847031304e-05,
+      "loss": 0.1655,
+      "mean_token_accuracy": 0.9657084941864014,
+      "num_tokens": 1504449.0,
+      "step": 721
+    },
+    {
+      "epoch": 0.005967977913522182,
+      "grad_norm": 0.37115100026130676,
+      "learning_rate": 4.9851007199596626e-05,
+      "loss": 0.1552,
+      "mean_token_accuracy": 0.9694430828094482,
+      "num_tokens": 1506484.0,
+      "step": 722
+    },
+    {
+      "epoch": 0.005976243810909331,
+      "grad_norm": 0.30470573902130127,
+      "learning_rate": 4.985080055216195e-05,
+      "loss": 0.1357,
+      "mean_token_accuracy": 0.972708523273468,
+      "num_tokens": 1508432.0,
+      "step": 723
+    },
+    {
+      "epoch": 0.005984509708296481,
+      "grad_norm": 0.42090368270874023,
+      "learning_rate": 4.985059390472727e-05,
+      "loss": 0.1951,
+      "mean_token_accuracy": 0.9596211314201355,
+      "num_tokens": 1510444.0,
+      "step": 724
+    },
+    {
+      "epoch": 0.005992775605683631,
+      "grad_norm": 0.44461822509765625,
+      "learning_rate": 4.985038725729259e-05,
+      "loss": 0.2102,
+      "mean_token_accuracy": 0.9578524827957153,
+      "num_tokens": 1512443.0,
+      "step": 725
+    },
+    {
+      "epoch": 0.006001041503070781,
+      "grad_norm": 0.30949079990386963,
+      "learning_rate": 4.985018060985791e-05,
+      "loss": 0.1453,
+      "mean_token_accuracy": 0.9671120047569275,
+      "num_tokens": 1514395.0,
+      "step": 726
+    },
+    {
+      "epoch": 0.006009307400457931,
+      "grad_norm": 0.3812202513217926,
+      "learning_rate": 4.9849973962423234e-05,
+      "loss": 0.1889,
+      "mean_token_accuracy": 0.9578736424446106,
+      "num_tokens": 1516395.0,
+      "step": 727
+    },
+    {
+      "epoch": 0.00601757329784508,
+      "grad_norm": 0.3606492877006531,
+      "learning_rate": 4.9849767314988556e-05,
+      "loss": 0.1626,
+      "mean_token_accuracy": 0.9687339663505554,
+      "num_tokens": 1518352.0,
+      "step": 728
+    },
+    {
+      "epoch": 0.00602583919523223,
+      "grad_norm": 0.4418604373931885,
+      "learning_rate": 4.984956066755387e-05,
+      "loss": 0.2059,
+      "mean_token_accuracy": 0.955610990524292,
+      "num_tokens": 1520363.0,
+      "step": 729
+    },
+    {
+      "epoch": 0.006034105092619381,
+      "grad_norm": 0.5098370909690857,
+      "learning_rate": 4.98493540201192e-05,
+      "loss": 0.2185,
+      "mean_token_accuracy": 0.9520514011383057,
+      "num_tokens": 1522392.0,
+      "step": 730
+    },
+    {
+      "epoch": 0.00604237099000653,
+      "grad_norm": 0.43088775873184204,
+      "learning_rate": 4.984914737268452e-05,
+      "loss": 0.2158,
+      "mean_token_accuracy": 0.9557788968086243,
+      "num_tokens": 1524388.0,
+      "step": 731
+    },
+    {
+      "epoch": 0.00605063688739368,
+      "grad_norm": 0.44536110758781433,
+      "learning_rate": 4.9848940725249836e-05,
+      "loss": 0.2113,
+      "mean_token_accuracy": 0.9576771855354309,
+      "num_tokens": 1526426.0,
+      "step": 732
+    },
+    {
+      "epoch": 0.006058902784780829,
+      "grad_norm": 0.3841190040111542,
+      "learning_rate": 4.9848734077815164e-05,
+      "loss": 0.1936,
+      "mean_token_accuracy": 0.9607160687446594,
+      "num_tokens": 1528443.0,
+      "step": 733
+    },
+    {
+      "epoch": 0.00606716868216798,
+      "grad_norm": 0.3974607586860657,
+      "learning_rate": 4.984852743038048e-05,
+      "loss": 0.225,
+      "mean_token_accuracy": 0.9522872567176819,
+      "num_tokens": 1530482.0,
+      "step": 734
+    },
+    {
+      "epoch": 0.00607543457955513,
+      "grad_norm": 0.37308064103126526,
+      "learning_rate": 4.98483207829458e-05,
+      "loss": 0.1955,
+      "mean_token_accuracy": 0.9607160687446594,
+      "num_tokens": 1532499.0,
+      "step": 735
+    },
+    {
+      "epoch": 0.006083700476942279,
+      "grad_norm": 0.41781359910964966,
+      "learning_rate": 4.984811413551113e-05,
+      "loss": 0.1806,
+      "mean_token_accuracy": 0.9641594886779785,
+      "num_tokens": 1534486.0,
+      "step": 736
+    },
+    {
+      "epoch": 0.006091966374329429,
+      "grad_norm": 0.4266640841960907,
+      "learning_rate": 4.9847907488076444e-05,
+      "loss": 0.2105,
+      "mean_token_accuracy": 0.9561315774917603,
+      "num_tokens": 1536498.0,
+      "step": 737
+    },
+    {
+      "epoch": 0.006100232271716579,
+      "grad_norm": 0.48743149638175964,
+      "learning_rate": 4.9847700840641766e-05,
+      "loss": 0.2462,
+      "mean_token_accuracy": 0.9492971897125244,
+      "num_tokens": 1538496.0,
+      "step": 738
+    },
+    {
+      "epoch": 0.006108498169103729,
+      "grad_norm": 0.3538137674331665,
+      "learning_rate": 4.984749419320709e-05,
+      "loss": 0.164,
+      "mean_token_accuracy": 0.9655521512031555,
+      "num_tokens": 1540476.0,
+      "step": 739
+    },
+    {
+      "epoch": 0.006116764066490879,
+      "grad_norm": 0.40453362464904785,
+      "learning_rate": 4.984728754577241e-05,
+      "loss": 0.2461,
+      "mean_token_accuracy": 0.9468982815742493,
+      "num_tokens": 1542497.0,
+      "step": 740
+    },
+    {
+      "epoch": 0.006125029963878028,
+      "grad_norm": 0.38154086470603943,
+      "learning_rate": 4.984708089833773e-05,
+      "loss": 0.183,
+      "mean_token_accuracy": 0.9572815299034119,
+      "num_tokens": 1544563.0,
+      "step": 741
+    },
+    {
+      "epoch": 0.0061332958612651785,
+      "grad_norm": 0.3262519836425781,
+      "learning_rate": 4.984687425090305e-05,
+      "loss": 0.1605,
+      "mean_token_accuracy": 0.9671385288238525,
+      "num_tokens": 1546547.0,
+      "step": 742
+    },
+    {
+      "epoch": 0.006141561758652328,
+      "grad_norm": 0.4108532965183258,
+      "learning_rate": 4.9846667603468374e-05,
+      "loss": 0.2542,
+      "mean_token_accuracy": 0.9488054513931274,
+      "num_tokens": 1548604.0,
+      "step": 743
+    },
+    {
+      "epoch": 0.006149827656039478,
+      "grad_norm": 0.35494884848594666,
+      "learning_rate": 4.984646095603369e-05,
+      "loss": 0.1819,
+      "mean_token_accuracy": 0.9633482098579407,
+      "num_tokens": 1550629.0,
+      "step": 744
+    },
+    {
+      "epoch": 0.006158093553426627,
+      "grad_norm": 0.40606263279914856,
+      "learning_rate": 4.984625430859902e-05,
+      "loss": 0.2148,
+      "mean_token_accuracy": 0.9566968679428101,
+      "num_tokens": 1552621.0,
+      "step": 745
+    },
+    {
+      "epoch": 0.0061663594508137775,
+      "grad_norm": 0.44487881660461426,
+      "learning_rate": 4.984604766116434e-05,
+      "loss": 0.3033,
+      "mean_token_accuracy": 0.9479572176933289,
+      "num_tokens": 1554683.0,
+      "step": 746
+    },
+    {
+      "epoch": 0.006174625348200928,
+      "grad_norm": 0.40845802426338196,
+      "learning_rate": 4.984584101372966e-05,
+      "loss": 0.222,
+      "mean_token_accuracy": 0.9511957168579102,
+      "num_tokens": 1556738.0,
+      "step": 747
+    },
+    {
+      "epoch": 0.006182891245588077,
+      "grad_norm": 0.3766586184501648,
+      "learning_rate": 4.984563436629498e-05,
+      "loss": 0.2169,
+      "mean_token_accuracy": 0.9517953991889954,
+      "num_tokens": 1558777.0,
+      "step": 748
+    },
+    {
+      "epoch": 0.006191157142975227,
+      "grad_norm": 0.33251988887786865,
+      "learning_rate": 4.98454277188603e-05,
+      "loss": 0.2064,
+      "mean_token_accuracy": 0.9600818753242493,
+      "num_tokens": 1560737.0,
+      "step": 749
+    },
+    {
+      "epoch": 0.0061994230403623765,
+      "grad_norm": 0.47217926383018494,
+      "learning_rate": 4.9845221071425626e-05,
+      "loss": 0.265,
+      "mean_token_accuracy": 0.944735586643219,
+      "num_tokens": 1562842.0,
+      "step": 750
+    },
+    {
+      "epoch": 0.006207688937749527,
+      "grad_norm": 0.4406071603298187,
+      "learning_rate": 4.984501442399094e-05,
+      "loss": 0.2143,
+      "mean_token_accuracy": 0.9543269276618958,
+      "num_tokens": 1564928.0,
+      "step": 751
+    },
+    {
+      "epoch": 0.006215954835136677,
+      "grad_norm": 0.41328927874565125,
+      "learning_rate": 4.984480777655626e-05,
+      "loss": 0.2182,
+      "mean_token_accuracy": 0.9553526043891907,
+      "num_tokens": 1566905.0,
+      "step": 752
+    },
+    {
+      "epoch": 0.006224220732523826,
+      "grad_norm": 0.3252612054347992,
+      "learning_rate": 4.984460112912159e-05,
+      "loss": 0.1835,
+      "mean_token_accuracy": 0.9661957621574402,
+      "num_tokens": 1568893.0,
+      "step": 753
+    },
+    {
+      "epoch": 0.006232486629910976,
+      "grad_norm": 0.46232733130455017,
+      "learning_rate": 4.9844394481686905e-05,
+      "loss": 0.1971,
+      "mean_token_accuracy": 0.9598413705825806,
+      "num_tokens": 1570916.0,
+      "step": 754
+    },
+    {
+      "epoch": 0.006240752527298126,
+      "grad_norm": 0.5029892325401306,
+      "learning_rate": 4.984418783425223e-05,
+      "loss": 0.267,
+      "mean_token_accuracy": 0.9429529905319214,
+      "num_tokens": 1573008.0,
+      "step": 755
+    },
+    {
+      "epoch": 0.006249018424685276,
+      "grad_norm": 0.4175070822238922,
+      "learning_rate": 4.984398118681755e-05,
+      "loss": 0.1851,
+      "mean_token_accuracy": 0.9622736573219299,
+      "num_tokens": 1575002.0,
+      "step": 756
+    },
+    {
+      "epoch": 0.006257284322072426,
+      "grad_norm": 0.4172671139240265,
+      "learning_rate": 4.984377453938287e-05,
+      "loss": 0.1826,
+      "mean_token_accuracy": 0.9615960121154785,
+      "num_tokens": 1577013.0,
+      "step": 757
+    },
+    {
+      "epoch": 0.0062655502194595755,
+      "grad_norm": 0.46920254826545715,
+      "learning_rate": 4.984356789194819e-05,
+      "loss": 0.1991,
+      "mean_token_accuracy": 0.9572271108627319,
+      "num_tokens": 1579053.0,
+      "step": 758
+    },
+    {
+      "epoch": 0.006273816116846726,
+      "grad_norm": 0.5021358132362366,
+      "learning_rate": 4.9843361244513514e-05,
+      "loss": 0.2116,
+      "mean_token_accuracy": 0.957446813583374,
+      "num_tokens": 1581080.0,
+      "step": 759
+    },
+    {
+      "epoch": 0.006282082014233875,
+      "grad_norm": 0.5324921607971191,
+      "learning_rate": 4.9843154597078835e-05,
+      "loss": 0.2597,
+      "mean_token_accuracy": 0.9497762322425842,
+      "num_tokens": 1583097.0,
+      "step": 760
+    },
+    {
+      "epoch": 0.006290347911621025,
+      "grad_norm": 0.46022337675094604,
+      "learning_rate": 4.984294794964415e-05,
+      "loss": 0.211,
+      "mean_token_accuracy": 0.95556640625,
+      "num_tokens": 1585151.0,
+      "step": 761
+    },
+    {
+      "epoch": 0.006298613809008175,
+      "grad_norm": 0.5305765867233276,
+      "learning_rate": 4.984274130220948e-05,
+      "loss": 0.2442,
+      "mean_token_accuracy": 0.9534769654273987,
+      "num_tokens": 1587199.0,
+      "step": 762
+    },
+    {
+      "epoch": 0.006306879706395325,
+      "grad_norm": 0.45197245478630066,
+      "learning_rate": 4.98425346547748e-05,
+      "loss": 0.1875,
+      "mean_token_accuracy": 0.9638554453849792,
+      "num_tokens": 1589197.0,
+      "step": 763
+    },
+    {
+      "epoch": 0.006315145603782475,
+      "grad_norm": 0.6175044178962708,
+      "learning_rate": 4.9842328007340115e-05,
+      "loss": 0.2469,
+      "mean_token_accuracy": 0.9531639218330383,
+      "num_tokens": 1591210.0,
+      "step": 764
+    },
+    {
+      "epoch": 0.006323411501169624,
+      "grad_norm": 0.4388083517551422,
+      "learning_rate": 4.9842121359905444e-05,
+      "loss": 0.1925,
+      "mean_token_accuracy": 0.9600202441215515,
+      "num_tokens": 1593192.0,
+      "step": 765
+    },
+    {
+      "epoch": 0.006331677398556774,
+      "grad_norm": 0.31428417563438416,
+      "learning_rate": 4.984191471247076e-05,
+      "loss": 0.15,
+      "mean_token_accuracy": 0.9669883251190186,
+      "num_tokens": 1595167.0,
+      "step": 766
+    },
+    {
+      "epoch": 0.0063399432959439245,
+      "grad_norm": 0.3937251567840576,
+      "learning_rate": 4.984170806503609e-05,
+      "loss": 0.2185,
+      "mean_token_accuracy": 0.9576399326324463,
+      "num_tokens": 1597156.0,
+      "step": 767
+    },
+    {
+      "epoch": 0.006348209193331074,
+      "grad_norm": 0.604350745677948,
+      "learning_rate": 4.98415014176014e-05,
+      "loss": 0.2357,
+      "mean_token_accuracy": 0.9501222372055054,
+      "num_tokens": 1599207.0,
+      "step": 768
+    },
+    {
+      "epoch": 0.006356475090718224,
+      "grad_norm": 0.4153241813182831,
+      "learning_rate": 4.9841294770166723e-05,
+      "loss": 0.223,
+      "mean_token_accuracy": 0.9524281620979309,
+      "num_tokens": 1601231.0,
+      "step": 769
+    },
+    {
+      "epoch": 0.006364740988105373,
+      "grad_norm": 0.348738431930542,
+      "learning_rate": 4.984108812273205e-05,
+      "loss": 0.1687,
+      "mean_token_accuracy": 0.9661102890968323,
+      "num_tokens": 1603214.0,
+      "step": 770
+    },
+    {
+      "epoch": 0.0063730068854925236,
+      "grad_norm": 0.3460107147693634,
+      "learning_rate": 4.984088147529737e-05,
+      "loss": 0.2187,
+      "mean_token_accuracy": 0.9582914710044861,
+      "num_tokens": 1605210.0,
+      "step": 771
+    },
+    {
+      "epoch": 0.006381272782879674,
+      "grad_norm": 0.2992319166660309,
+      "learning_rate": 4.984067482786269e-05,
+      "loss": 0.176,
+      "mean_token_accuracy": 0.9650025963783264,
+      "num_tokens": 1607159.0,
+      "step": 772
+    },
+    {
+      "epoch": 0.006389538680266823,
+      "grad_norm": 0.39421042799949646,
+      "learning_rate": 4.984046818042801e-05,
+      "loss": 0.1806,
+      "mean_token_accuracy": 0.9619145393371582,
+      "num_tokens": 1609108.0,
+      "step": 773
+    },
+    {
+      "epoch": 0.006397804577653973,
+      "grad_norm": 0.48183563351631165,
+      "learning_rate": 4.984026153299333e-05,
+      "loss": 0.2403,
+      "mean_token_accuracy": 0.9539215564727783,
+      "num_tokens": 1611154.0,
+      "step": 774
+    },
+    {
+      "epoch": 0.006406070475041123,
+      "grad_norm": 0.48275473713874817,
+      "learning_rate": 4.9840054885558653e-05,
+      "loss": 0.2541,
+      "mean_token_accuracy": 0.95068359375,
+      "num_tokens": 1613208.0,
+      "step": 775
+    },
+    {
+      "epoch": 0.006414336372428273,
+      "grad_norm": 0.3698482811450958,
+      "learning_rate": 4.9839848238123975e-05,
+      "loss": 0.1622,
+      "mean_token_accuracy": 0.9680905938148499,
+      "num_tokens": 1615157.0,
+      "step": 776
+    },
+    {
+      "epoch": 0.006422602269815422,
+      "grad_norm": 0.47950881719589233,
+      "learning_rate": 4.98396415906893e-05,
+      "loss": 0.2692,
+      "mean_token_accuracy": 0.9491771459579468,
+      "num_tokens": 1617229.0,
+      "step": 777
+    },
+    {
+      "epoch": 0.006430868167202572,
+      "grad_norm": 0.49546265602111816,
+      "learning_rate": 4.983943494325461e-05,
+      "loss": 0.2288,
+      "mean_token_accuracy": 0.9608826637268066,
+      "num_tokens": 1619229.0,
+      "step": 778
+    },
+    {
+      "epoch": 0.0064391340645897225,
+      "grad_norm": 0.524799644947052,
+      "learning_rate": 4.983922829581994e-05,
+      "loss": 0.2265,
+      "mean_token_accuracy": 0.9513034820556641,
+      "num_tokens": 1621268.0,
+      "step": 779
+    },
+    {
+      "epoch": 0.006447399961976872,
+      "grad_norm": 0.3272992968559265,
+      "learning_rate": 4.983902164838526e-05,
+      "loss": 0.1507,
+      "mean_token_accuracy": 0.9721074104309082,
+      "num_tokens": 1623210.0,
+      "step": 780
+    },
+    {
+      "epoch": 0.006455665859364022,
+      "grad_norm": 0.4449158310890198,
+      "learning_rate": 4.9838815000950577e-05,
+      "loss": 0.1857,
+      "mean_token_accuracy": 0.95839524269104,
+      "num_tokens": 1625235.0,
+      "step": 781
+    },
+    {
+      "epoch": 0.006463931756751171,
+      "grad_norm": 0.4281235337257385,
+      "learning_rate": 4.9838608353515905e-05,
+      "loss": 0.2276,
+      "mean_token_accuracy": 0.9552311301231384,
+      "num_tokens": 1627296.0,
+      "step": 782
+    },
+    {
+      "epoch": 0.0064721976541383215,
+      "grad_norm": 0.3788045644760132,
+      "learning_rate": 4.983840170608122e-05,
+      "loss": 0.2067,
+      "mean_token_accuracy": 0.9588701725006104,
+      "num_tokens": 1629320.0,
+      "step": 783
+    },
+    {
+      "epoch": 0.006480463551525472,
+      "grad_norm": 0.393130898475647,
+      "learning_rate": 4.983819505864654e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9674631357192993,
+      "num_tokens": 1631293.0,
+      "step": 784
+    },
+    {
+      "epoch": 0.006488729448912621,
+      "grad_norm": 0.399342805147171,
+      "learning_rate": 4.983798841121187e-05,
+      "loss": 0.1712,
+      "mean_token_accuracy": 0.9649649858474731,
+      "num_tokens": 1633297.0,
+      "step": 785
+    },
+    {
+      "epoch": 0.006496995346299771,
+      "grad_norm": 0.40883710980415344,
+      "learning_rate": 4.9837781763777185e-05,
+      "loss": 0.1974,
+      "mean_token_accuracy": 0.9562841653823853,
+      "num_tokens": 1635316.0,
+      "step": 786
+    },
+    {
+      "epoch": 0.0065052612436869205,
+      "grad_norm": 0.3678262531757355,
+      "learning_rate": 4.983757511634251e-05,
+      "loss": 0.1625,
+      "mean_token_accuracy": 0.9673635959625244,
+      "num_tokens": 1637283.0,
+      "step": 787
+    },
+    {
+      "epoch": 0.006513527141074071,
+      "grad_norm": 0.3532010614871979,
+      "learning_rate": 4.983736846890783e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.9671551585197449,
+      "num_tokens": 1639268.0,
+      "step": 788
+    },
+    {
+      "epoch": 0.006521793038461221,
+      "grad_norm": 0.4240894913673401,
+      "learning_rate": 4.983716182147315e-05,
+      "loss": 0.2367,
+      "mean_token_accuracy": 0.9500489830970764,
+      "num_tokens": 1641316.0,
+      "step": 789
+    },
+    {
+      "epoch": 0.00653005893584837,
+      "grad_norm": 0.5130672454833984,
+      "learning_rate": 4.983695517403847e-05,
+      "loss": 0.2024,
+      "mean_token_accuracy": 0.9548192620277405,
+      "num_tokens": 1643314.0,
+      "step": 790
+    },
+    {
+      "epoch": 0.00653832483323552,
+      "grad_norm": 0.3678853213787079,
+      "learning_rate": 4.983674852660379e-05,
+      "loss": 0.1492,
+      "mean_token_accuracy": 0.9689092636108398,
+      "num_tokens": 1645282.0,
+      "step": 791
+    },
+    {
+      "epoch": 0.00654659073062267,
+      "grad_norm": 0.3945569396018982,
+      "learning_rate": 4.9836541879169115e-05,
+      "loss": 0.1749,
+      "mean_token_accuracy": 0.9648689031600952,
+      "num_tokens": 1647309.0,
+      "step": 792
+    },
+    {
+      "epoch": 0.00655485662800982,
+      "grad_norm": 0.48077425360679626,
+      "learning_rate": 4.9836335231734436e-05,
+      "loss": 0.1986,
+      "mean_token_accuracy": 0.9593781232833862,
+      "num_tokens": 1649309.0,
+      "step": 793
+    },
+    {
+      "epoch": 0.00656312252539697,
+      "grad_norm": 0.4411054849624634,
+      "learning_rate": 4.983612858429976e-05,
+      "loss": 0.1864,
+      "mean_token_accuracy": 0.9590842723846436,
+      "num_tokens": 1651368.0,
+      "step": 794
+    },
+    {
+      "epoch": 0.006571388422784119,
+      "grad_norm": 0.3010132312774658,
+      "learning_rate": 4.983592193686508e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9670050740242004,
+      "num_tokens": 1653344.0,
+      "step": 795
+    },
+    {
+      "epoch": 0.00657965432017127,
+      "grad_norm": 0.4571327567100525,
+      "learning_rate": 4.98357152894304e-05,
+      "loss": 0.1738,
+      "mean_token_accuracy": 0.9667689204216003,
+      "num_tokens": 1655306.0,
+      "step": 796
+    },
+    {
+      "epoch": 0.006587920217558419,
+      "grad_norm": 0.4823872148990631,
+      "learning_rate": 4.983550864199572e-05,
+      "loss": 0.269,
+      "mean_token_accuracy": 0.9479768872261047,
+      "num_tokens": 1657388.0,
+      "step": 797
+    },
+    {
+      "epoch": 0.006596186114945569,
+      "grad_norm": 0.405563622713089,
+      "learning_rate": 4.983530199456104e-05,
+      "loss": 0.191,
+      "mean_token_accuracy": 0.9608142375946045,
+      "num_tokens": 1659359.0,
+      "step": 798
+    },
+    {
+      "epoch": 0.006604452012332719,
+      "grad_norm": 0.4063994288444519,
+      "learning_rate": 4.9835095347126366e-05,
+      "loss": 0.1593,
+      "mean_token_accuracy": 0.9664121866226196,
+      "num_tokens": 1661330.0,
+      "step": 799
+    },
+    {
+      "epoch": 0.006612717909719869,
+      "grad_norm": 0.4480558931827545,
+      "learning_rate": 4.983488869969168e-05,
+      "loss": 0.1881,
+      "mean_token_accuracy": 0.9653121829032898,
+      "num_tokens": 1663354.0,
+      "step": 800
+    },
+    {
+      "epoch": 0.006620983807107019,
+      "grad_norm": 0.4080691337585449,
+      "learning_rate": 4.9834682052257e-05,
+      "loss": 0.1976,
+      "mean_token_accuracy": 0.9613478779792786,
+      "num_tokens": 1665378.0,
+      "step": 801
+    },
+    {
+      "epoch": 0.006629249704494168,
+      "grad_norm": 0.5544767379760742,
+      "learning_rate": 4.983447540482233e-05,
+      "loss": 0.1827,
+      "mean_token_accuracy": 0.9623202681541443,
+      "num_tokens": 1667401.0,
+      "step": 802
+    },
+    {
+      "epoch": 0.006637515601881318,
+      "grad_norm": 0.4864117205142975,
+      "learning_rate": 4.9834268757387646e-05,
+      "loss": 0.1838,
+      "mean_token_accuracy": 0.9603658318519592,
+      "num_tokens": 1669375.0,
+      "step": 803
+    },
+    {
+      "epoch": 0.0066457814992684685,
+      "grad_norm": 0.34529808163642883,
+      "learning_rate": 4.983406210995297e-05,
+      "loss": 0.1526,
+      "mean_token_accuracy": 0.9655880928039551,
+      "num_tokens": 1671328.0,
+      "step": 804
+    },
+    {
+      "epoch": 0.006654047396655618,
+      "grad_norm": 0.39255163073539734,
+      "learning_rate": 4.983385546251829e-05,
+      "loss": 0.1954,
+      "mean_token_accuracy": 0.959919810295105,
+      "num_tokens": 1673330.0,
+      "step": 805
+    },
+    {
+      "epoch": 0.006662313294042768,
+      "grad_norm": 0.350628137588501,
+      "learning_rate": 4.983364881508361e-05,
+      "loss": 0.1415,
+      "mean_token_accuracy": 0.9679062366485596,
+      "num_tokens": 1675299.0,
+      "step": 806
+    },
+    {
+      "epoch": 0.006670579191429917,
+      "grad_norm": 0.35066065192222595,
+      "learning_rate": 4.983344216764893e-05,
+      "loss": 0.1912,
+      "mean_token_accuracy": 0.9628687500953674,
+      "num_tokens": 1677271.0,
+      "step": 807
+    },
+    {
+      "epoch": 0.0066788450888170675,
+      "grad_norm": 0.40352028608322144,
+      "learning_rate": 4.9833235520214255e-05,
+      "loss": 0.2173,
+      "mean_token_accuracy": 0.9588559865951538,
+      "num_tokens": 1679270.0,
+      "step": 808
+    },
+    {
+      "epoch": 0.006687110986204217,
+      "grad_norm": 0.34992119669914246,
+      "learning_rate": 4.9833028872779576e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9632089734077454,
+      "num_tokens": 1681233.0,
+      "step": 809
+    },
+    {
+      "epoch": 0.006695376883591367,
+      "grad_norm": 0.3540070652961731,
+      "learning_rate": 4.98328222253449e-05,
+      "loss": 0.1474,
+      "mean_token_accuracy": 0.9678256511688232,
+      "num_tokens": 1683166.0,
+      "step": 810
+    },
+    {
+      "epoch": 0.006703642780978517,
+      "grad_norm": 0.4656151235103607,
+      "learning_rate": 4.983261557791022e-05,
+      "loss": 0.1912,
+      "mean_token_accuracy": 0.9620128273963928,
+      "num_tokens": 1685199.0,
+      "step": 811
+    },
+    {
+      "epoch": 0.0067119086783656665,
+      "grad_norm": 0.5217596292495728,
+      "learning_rate": 4.983240893047554e-05,
+      "loss": 0.2457,
+      "mean_token_accuracy": 0.9494799375534058,
+      "num_tokens": 1687224.0,
+      "step": 812
+    },
+    {
+      "epoch": 0.006720174575752817,
+      "grad_norm": 0.39535465836524963,
+      "learning_rate": 4.983220228304086e-05,
+      "loss": 0.2054,
+      "mean_token_accuracy": 0.9591132998466492,
+      "num_tokens": 1689260.0,
+      "step": 813
+    },
+    {
+      "epoch": 0.006728440473139966,
+      "grad_norm": 0.38129371404647827,
+      "learning_rate": 4.9831995635606185e-05,
+      "loss": 0.1646,
+      "mean_token_accuracy": 0.9711099863052368,
+      "num_tokens": 1691239.0,
+      "step": 814
+    },
+    {
+      "epoch": 0.006736706370527116,
+      "grad_norm": 0.291294127702713,
+      "learning_rate": 4.98317889881715e-05,
+      "loss": 0.1511,
+      "mean_token_accuracy": 0.9708788394927979,
+      "num_tokens": 1693168.0,
+      "step": 815
+    },
+    {
+      "epoch": 0.006744972267914266,
+      "grad_norm": 0.466017484664917,
+      "learning_rate": 4.983158234073683e-05,
+      "loss": 0.2231,
+      "mean_token_accuracy": 0.9538003206253052,
+      "num_tokens": 1695187.0,
+      "step": 816
+    },
+    {
+      "epoch": 0.006753238165301416,
+      "grad_norm": 0.4086066484451294,
+      "learning_rate": 4.983137569330214e-05,
+      "loss": 0.1947,
+      "mean_token_accuracy": 0.9597585797309875,
+      "num_tokens": 1697181.0,
+      "step": 817
+    },
+    {
+      "epoch": 0.006761504062688566,
+      "grad_norm": 0.45003077387809753,
+      "learning_rate": 4.9831169045867464e-05,
+      "loss": 0.185,
+      "mean_token_accuracy": 0.9616742134094238,
+      "num_tokens": 1699170.0,
+      "step": 818
+    },
+    {
+      "epoch": 0.006769769960075715,
+      "grad_norm": 0.35875362157821655,
+      "learning_rate": 4.983096239843279e-05,
+      "loss": 0.1443,
+      "mean_token_accuracy": 0.9687026739120483,
+      "num_tokens": 1701157.0,
+      "step": 819
+    },
+    {
+      "epoch": 0.006778035857462865,
+      "grad_norm": 0.37253376841545105,
+      "learning_rate": 4.983075575099811e-05,
+      "loss": 0.2001,
+      "mean_token_accuracy": 0.9568272829055786,
+      "num_tokens": 1703155.0,
+      "step": 820
+    },
+    {
+      "epoch": 0.006786301754850016,
+      "grad_norm": 0.4945686459541321,
+      "learning_rate": 4.983054910356343e-05,
+      "loss": 0.3514,
+      "mean_token_accuracy": 0.9342105388641357,
+      "num_tokens": 1705289.0,
+      "step": 821
+    },
+    {
+      "epoch": 0.006794567652237165,
+      "grad_norm": 0.37332653999328613,
+      "learning_rate": 4.983034245612875e-05,
+      "loss": 0.2146,
+      "mean_token_accuracy": 0.9567354917526245,
+      "num_tokens": 1707329.0,
+      "step": 822
+    },
+    {
+      "epoch": 0.006802833549624315,
+      "grad_norm": 0.33233872056007385,
+      "learning_rate": 4.983013580869407e-05,
+      "loss": 0.1426,
+      "mean_token_accuracy": 0.9694656729698181,
+      "num_tokens": 1709300.0,
+      "step": 823
+    },
+    {
+      "epoch": 0.0068110994470114644,
+      "grad_norm": 0.48606806993484497,
+      "learning_rate": 4.9829929161259394e-05,
+      "loss": 0.2377,
+      "mean_token_accuracy": 0.9518013596534729,
+      "num_tokens": 1711360.0,
+      "step": 824
+    },
+    {
+      "epoch": 0.006819365344398615,
+      "grad_norm": 0.4521882236003876,
+      "learning_rate": 4.9829722513824716e-05,
+      "loss": 0.2607,
+      "mean_token_accuracy": 0.9503166079521179,
+      "num_tokens": 1713419.0,
+      "step": 825
+    },
+    {
+      "epoch": 0.006827631241785765,
+      "grad_norm": 0.3311852216720581,
+      "learning_rate": 4.982951586639004e-05,
+      "loss": 0.153,
+      "mean_token_accuracy": 0.9684317708015442,
+      "num_tokens": 1715389.0,
+      "step": 826
+    },
+    {
+      "epoch": 0.006835897139172914,
+      "grad_norm": 0.5356886982917786,
+      "learning_rate": 4.982930921895535e-05,
+      "loss": 0.2402,
+      "mean_token_accuracy": 0.9520348906517029,
+      "num_tokens": 1717459.0,
+      "step": 827
+    },
+    {
+      "epoch": 0.006844163036560064,
+      "grad_norm": 0.534487783908844,
+      "learning_rate": 4.982910257152068e-05,
+      "loss": 0.2511,
+      "mean_token_accuracy": 0.9495880007743835,
+      "num_tokens": 1719528.0,
+      "step": 828
+    },
+    {
+      "epoch": 0.006852428933947214,
+      "grad_norm": 0.5094696879386902,
+      "learning_rate": 4.9828895924086e-05,
+      "loss": 0.2083,
+      "mean_token_accuracy": 0.9571926593780518,
+      "num_tokens": 1721543.0,
+      "step": 829
+    },
+    {
+      "epoch": 0.006860694831334364,
+      "grad_norm": 0.45827627182006836,
+      "learning_rate": 4.9828689276651324e-05,
+      "loss": 0.2699,
+      "mean_token_accuracy": 0.94482421875,
+      "num_tokens": 1723597.0,
+      "step": 830
+    },
+    {
+      "epoch": 0.006868960728721514,
+      "grad_norm": 0.3388739824295044,
+      "learning_rate": 4.9828482629216646e-05,
+      "loss": 0.1718,
+      "mean_token_accuracy": 0.9656044244766235,
+      "num_tokens": 1725580.0,
+      "step": 831
+    },
+    {
+      "epoch": 0.006877226626108663,
+      "grad_norm": 0.32214492559432983,
+      "learning_rate": 4.982827598178196e-05,
+      "loss": 0.1905,
+      "mean_token_accuracy": 0.9608237147331238,
+      "num_tokens": 1727577.0,
+      "step": 832
+    },
+    {
+      "epoch": 0.0068854925234958135,
+      "grad_norm": 0.32804158329963684,
+      "learning_rate": 4.982806933434729e-05,
+      "loss": 0.1822,
+      "mean_token_accuracy": 0.9662808179855347,
+      "num_tokens": 1729570.0,
+      "step": 833
+    },
+    {
+      "epoch": 0.006893758420882963,
+      "grad_norm": 0.3797645568847656,
+      "learning_rate": 4.982786268691261e-05,
+      "loss": 0.2355,
+      "mean_token_accuracy": 0.9515484571456909,
+      "num_tokens": 1731578.0,
+      "step": 834
+    },
+    {
+      "epoch": 0.006902024318270113,
+      "grad_norm": 0.375230073928833,
+      "learning_rate": 4.9827656039477926e-05,
+      "loss": 0.1893,
+      "mean_token_accuracy": 0.9611503481864929,
+      "num_tokens": 1733566.0,
+      "step": 835
+    },
+    {
+      "epoch": 0.006910290215657263,
+      "grad_norm": 0.3135709762573242,
+      "learning_rate": 4.9827449392043254e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9642309546470642,
+      "num_tokens": 1735529.0,
+      "step": 836
+    },
+    {
+      "epoch": 0.0069185561130444126,
+      "grad_norm": 0.3929509222507477,
+      "learning_rate": 4.982724274460857e-05,
+      "loss": 0.2435,
+      "mean_token_accuracy": 0.9504901766777039,
+      "num_tokens": 1737575.0,
+      "step": 837
+    },
+    {
+      "epoch": 0.006926822010431563,
+      "grad_norm": 0.31875646114349365,
+      "learning_rate": 4.982703609717389e-05,
+      "loss": 0.1706,
+      "mean_token_accuracy": 0.9661017060279846,
+      "num_tokens": 1739587.0,
+      "step": 838
+    },
+    {
+      "epoch": 0.006935087907818712,
+      "grad_norm": 0.4902570843696594,
+      "learning_rate": 4.982682944973921e-05,
+      "loss": 0.2164,
+      "mean_token_accuracy": 0.9524975419044495,
+      "num_tokens": 1741635.0,
+      "step": 839
+    },
+    {
+      "epoch": 0.006943353805205862,
+      "grad_norm": 0.396259069442749,
+      "learning_rate": 4.9826622802304534e-05,
+      "loss": 0.1781,
+      "mean_token_accuracy": 0.9652917385101318,
+      "num_tokens": 1743629.0,
+      "step": 840
+    },
+    {
+      "epoch": 0.0069516197025930124,
+      "grad_norm": 0.3869590163230896,
+      "learning_rate": 4.9826416154869856e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9647355079650879,
+      "num_tokens": 1745620.0,
+      "step": 841
+    },
+    {
+      "epoch": 0.006959885599980162,
+      "grad_norm": 0.2901887595653534,
+      "learning_rate": 4.982620950743518e-05,
+      "loss": 0.1595,
+      "mean_token_accuracy": 0.9716055989265442,
+      "num_tokens": 1747563.0,
+      "step": 842
+    },
+    {
+      "epoch": 0.006968151497367312,
+      "grad_norm": 0.34304308891296387,
+      "learning_rate": 4.98260028600005e-05,
+      "loss": 0.1602,
+      "mean_token_accuracy": 0.9674465656280518,
+      "num_tokens": 1749535.0,
+      "step": 843
+    },
+    {
+      "epoch": 0.006976417394754461,
+      "grad_norm": 0.508950412273407,
+      "learning_rate": 4.982579621256582e-05,
+      "loss": 0.2147,
+      "mean_token_accuracy": 0.9550341963768005,
+      "num_tokens": 1751587.0,
+      "step": 844
+    },
+    {
+      "epoch": 0.0069846832921416115,
+      "grad_norm": 0.315696656703949,
+      "learning_rate": 4.982558956513114e-05,
+      "loss": 0.1596,
+      "mean_token_accuracy": 0.9683893918991089,
+      "num_tokens": 1753586.0,
+      "step": 845
+    },
+    {
+      "epoch": 0.006992949189528761,
+      "grad_norm": 0.37910759449005127,
+      "learning_rate": 4.9825382917696464e-05,
+      "loss": 0.1651,
+      "mean_token_accuracy": 0.9651817679405212,
+      "num_tokens": 1755545.0,
+      "step": 846
+    },
+    {
+      "epoch": 0.007001215086915911,
+      "grad_norm": 0.474913090467453,
+      "learning_rate": 4.982517627026178e-05,
+      "loss": 0.2151,
+      "mean_token_accuracy": 0.9575424790382385,
+      "num_tokens": 1757553.0,
+      "step": 847
+    },
+    {
+      "epoch": 0.007009480984303061,
+      "grad_norm": 0.4384135603904724,
+      "learning_rate": 4.982496962282711e-05,
+      "loss": 0.2018,
+      "mean_token_accuracy": 0.9589589834213257,
+      "num_tokens": 1759557.0,
+      "step": 848
+    },
+    {
+      "epoch": 0.0070177468816902105,
+      "grad_norm": 0.4935932457447052,
+      "learning_rate": 4.982476297539242e-05,
+      "loss": 0.2468,
+      "mean_token_accuracy": 0.9507557153701782,
+      "num_tokens": 1761614.0,
+      "step": 849
+    },
+    {
+      "epoch": 0.007026012779077361,
+      "grad_norm": 0.42140674591064453,
+      "learning_rate": 4.982455632795775e-05,
+      "loss": 0.1652,
+      "mean_token_accuracy": 0.9658859372138977,
+      "num_tokens": 1763584.0,
+      "step": 850
+    },
+    {
+      "epoch": 0.00703427867646451,
+      "grad_norm": 0.589418888092041,
+      "learning_rate": 4.982434968052307e-05,
+      "loss": 0.1986,
+      "mean_token_accuracy": 0.9621136784553528,
+      "num_tokens": 1765596.0,
+      "step": 851
+    },
+    {
+      "epoch": 0.00704254457385166,
+      "grad_norm": 0.5515609979629517,
+      "learning_rate": 4.982414303308839e-05,
+      "loss": 0.2175,
+      "mean_token_accuracy": 0.9505574703216553,
+      "num_tokens": 1767665.0,
+      "step": 852
+    },
+    {
+      "epoch": 0.00705081047123881,
+      "grad_norm": 0.42219194769859314,
+      "learning_rate": 4.9823936385653716e-05,
+      "loss": 0.17,
+      "mean_token_accuracy": 0.9626262784004211,
+      "num_tokens": 1769651.0,
+      "step": 853
+    },
+    {
+      "epoch": 0.00705907636862596,
+      "grad_norm": 0.3146474063396454,
+      "learning_rate": 4.982372973821903e-05,
+      "loss": 0.1443,
+      "mean_token_accuracy": 0.9693158864974976,
+      "num_tokens": 1771645.0,
+      "step": 854
+    },
+    {
+      "epoch": 0.00706734226601311,
+      "grad_norm": 0.4332464933395386,
+      "learning_rate": 4.982352309078435e-05,
+      "loss": 0.1856,
+      "mean_token_accuracy": 0.9673802256584167,
+      "num_tokens": 1773613.0,
+      "step": 855
+    },
+    {
+      "epoch": 0.007075608163400259,
+      "grad_norm": 0.38029155135154724,
+      "learning_rate": 4.9823316443349674e-05,
+      "loss": 0.1604,
+      "mean_token_accuracy": 0.9666160941123962,
+      "num_tokens": 1775596.0,
+      "step": 856
+    },
+    {
+      "epoch": 0.007083874060787409,
+      "grad_norm": 0.33078211545944214,
+      "learning_rate": 4.9823109795914995e-05,
+      "loss": 0.1634,
+      "mean_token_accuracy": 0.9691451787948608,
+      "num_tokens": 1777579.0,
+      "step": 857
+    },
+    {
+      "epoch": 0.00709213995817456,
+      "grad_norm": 0.47529563307762146,
+      "learning_rate": 4.982290314848032e-05,
+      "loss": 0.1634,
+      "mean_token_accuracy": 0.9662978053092957,
+      "num_tokens": 1779573.0,
+      "step": 858
+    },
+    {
+      "epoch": 0.007100405855561709,
+      "grad_norm": 0.3246769607067108,
+      "learning_rate": 4.982269650104564e-05,
+      "loss": 0.1794,
+      "mean_token_accuracy": 0.9668028354644775,
+      "num_tokens": 1781537.0,
+      "step": 859
+    },
+    {
+      "epoch": 0.007108671752948859,
+      "grad_norm": 0.4095306396484375,
+      "learning_rate": 4.982248985361096e-05,
+      "loss": 0.1861,
+      "mean_token_accuracy": 0.9620829224586487,
+      "num_tokens": 1783521.0,
+      "step": 860
+    },
+    {
+      "epoch": 0.007116937650336008,
+      "grad_norm": 0.45311596989631653,
+      "learning_rate": 4.982228320617628e-05,
+      "loss": 0.22,
+      "mean_token_accuracy": 0.9541015625,
+      "num_tokens": 1785575.0,
+      "step": 861
+    },
+    {
+      "epoch": 0.007125203547723159,
+      "grad_norm": 0.3538917601108551,
+      "learning_rate": 4.9822076558741604e-05,
+      "loss": 0.1652,
+      "mean_token_accuracy": 0.9666666388511658,
+      "num_tokens": 1787531.0,
+      "step": 862
+    },
+    {
+      "epoch": 0.007133469445110309,
+      "grad_norm": 0.3990963101387024,
+      "learning_rate": 4.9821869911306925e-05,
+      "loss": 0.2461,
+      "mean_token_accuracy": 0.9486552476882935,
+      "num_tokens": 1789582.0,
+      "step": 863
+    },
+    {
+      "epoch": 0.007141735342497458,
+      "grad_norm": 0.44176051020622253,
+      "learning_rate": 4.982166326387224e-05,
+      "loss": 0.2116,
+      "mean_token_accuracy": 0.9554047584533691,
+      "num_tokens": 1791651.0,
+      "step": 864
+    },
+    {
+      "epoch": 0.007150001239884608,
+      "grad_norm": 0.44992202520370483,
+      "learning_rate": 4.982145661643757e-05,
+      "loss": 0.2221,
+      "mean_token_accuracy": 0.9524742960929871,
+      "num_tokens": 1793698.0,
+      "step": 865
+    },
+    {
+      "epoch": 0.007158267137271758,
+      "grad_norm": 0.3285636305809021,
+      "learning_rate": 4.9821249969002884e-05,
+      "loss": 0.1481,
+      "mean_token_accuracy": 0.9697122573852539,
+      "num_tokens": 1795685.0,
+      "step": 866
+    },
+    {
+      "epoch": 0.007166533034658908,
+      "grad_norm": 0.4576401710510254,
+      "learning_rate": 4.9821043321568205e-05,
+      "loss": 0.1854,
+      "mean_token_accuracy": 0.9635728597640991,
+      "num_tokens": 1797695.0,
+      "step": 867
+    },
+    {
+      "epoch": 0.007174798932046058,
+      "grad_norm": 0.4985065460205078,
+      "learning_rate": 4.9820836674133534e-05,
+      "loss": 0.1948,
+      "mean_token_accuracy": 0.9574571847915649,
+      "num_tokens": 1799746.0,
+      "step": 868
+    },
+    {
+      "epoch": 0.007183064829433207,
+      "grad_norm": 0.38755112886428833,
+      "learning_rate": 4.982063002669885e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9661538600921631,
+      "num_tokens": 1801702.0,
+      "step": 869
+    },
+    {
+      "epoch": 0.0071913307268203575,
+      "grad_norm": 0.4477129876613617,
+      "learning_rate": 4.982042337926417e-05,
+      "loss": 0.1848,
+      "mean_token_accuracy": 0.9620189666748047,
+      "num_tokens": 1803709.0,
+      "step": 870
+    },
+    {
+      "epoch": 0.007199596624207507,
+      "grad_norm": 0.38782238960266113,
+      "learning_rate": 4.982021673182949e-05,
+      "loss": 0.1865,
+      "mean_token_accuracy": 0.9643584489822388,
+      "num_tokens": 1805679.0,
+      "step": 871
+    },
+    {
+      "epoch": 0.007207862521594657,
+      "grad_norm": 0.45049068331718445,
+      "learning_rate": 4.9820010084394814e-05,
+      "loss": 0.2122,
+      "mean_token_accuracy": 0.9574787616729736,
+      "num_tokens": 1807684.0,
+      "step": 872
+    },
+    {
+      "epoch": 0.007216128418981807,
+      "grad_norm": 0.44215312600135803,
+      "learning_rate": 4.9819803436960135e-05,
+      "loss": 0.1988,
+      "mean_token_accuracy": 0.9601413011550903,
+      "num_tokens": 1809672.0,
+      "step": 873
+    },
+    {
+      "epoch": 0.0072243943163689565,
+      "grad_norm": 0.4946404993534088,
+      "learning_rate": 4.981959678952546e-05,
+      "loss": 0.2441,
+      "mean_token_accuracy": 0.9473175406455994,
+      "num_tokens": 1811747.0,
+      "step": 874
+    },
+    {
+      "epoch": 0.007232660213756107,
+      "grad_norm": 0.3955952823162079,
+      "learning_rate": 4.981939014209078e-05,
+      "loss": 0.161,
+      "mean_token_accuracy": 0.9659489393234253,
+      "num_tokens": 1813750.0,
+      "step": 875
+    },
+    {
+      "epoch": 0.007240926111143256,
+      "grad_norm": 0.9727440476417542,
+      "learning_rate": 4.98191834946561e-05,
+      "loss": 0.1469,
+      "mean_token_accuracy": 0.9674022197723389,
+      "num_tokens": 1815750.0,
+      "step": 876
+    },
+    {
+      "epoch": 0.007249192008530406,
+      "grad_norm": 0.46547815203666687,
+      "learning_rate": 4.981897684722142e-05,
+      "loss": 0.1907,
+      "mean_token_accuracy": 0.9638190865516663,
+      "num_tokens": 1817746.0,
+      "step": 877
+    },
+    {
+      "epoch": 0.0072574579059175555,
+      "grad_norm": 0.43833407759666443,
+      "learning_rate": 4.9818770199786743e-05,
+      "loss": 0.1968,
+      "mean_token_accuracy": 0.9588559865951538,
+      "num_tokens": 1819745.0,
+      "step": 878
+    },
+    {
+      "epoch": 0.007265723803304706,
+      "grad_norm": 0.36661940813064575,
+      "learning_rate": 4.9818563552352065e-05,
+      "loss": 0.1883,
+      "mean_token_accuracy": 0.9613478779792786,
+      "num_tokens": 1821769.0,
+      "step": 879
+    },
+    {
+      "epoch": 0.007273989700691856,
+      "grad_norm": 0.4462507665157318,
+      "learning_rate": 4.981835690491739e-05,
+      "loss": 0.299,
+      "mean_token_accuracy": 0.941662609577179,
+      "num_tokens": 1823832.0,
+      "step": 880
+    },
+    {
+      "epoch": 0.007282255598079005,
+      "grad_norm": 0.45812344551086426,
+      "learning_rate": 4.98181502574827e-05,
+      "loss": 0.214,
+      "mean_token_accuracy": 0.9591633677482605,
+      "num_tokens": 1825846.0,
+      "step": 881
+    },
+    {
+      "epoch": 0.007290521495466155,
+      "grad_norm": 0.40240493416786194,
+      "learning_rate": 4.981794361004803e-05,
+      "loss": 0.1456,
+      "mean_token_accuracy": 0.970271646976471,
+      "num_tokens": 1827803.0,
+      "step": 882
+    },
+    {
+      "epoch": 0.007298787392853305,
+      "grad_norm": 0.3636603355407715,
+      "learning_rate": 4.9817736962613345e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.9663437008857727,
+      "num_tokens": 1829770.0,
+      "step": 883
+    },
+    {
+      "epoch": 0.007307053290240455,
+      "grad_norm": 0.3355780839920044,
+      "learning_rate": 4.981753031517867e-05,
+      "loss": 0.1535,
+      "mean_token_accuracy": 0.971222996711731,
+      "num_tokens": 1831722.0,
+      "step": 884
+    },
+    {
+      "epoch": 0.007315319187627605,
+      "grad_norm": 0.5163689851760864,
+      "learning_rate": 4.9817323667743995e-05,
+      "loss": 0.2789,
+      "mean_token_accuracy": 0.9381783604621887,
+      "num_tokens": 1833847.0,
+      "step": 885
+    },
+    {
+      "epoch": 0.007323585085014754,
+      "grad_norm": 0.4665575623512268,
+      "learning_rate": 4.981711702030931e-05,
+      "loss": 0.1892,
+      "mean_token_accuracy": 0.9579746723175049,
+      "num_tokens": 1835828.0,
+      "step": 886
+    },
+    {
+      "epoch": 0.007331850982401905,
+      "grad_norm": 0.35391056537628174,
+      "learning_rate": 4.981691037287463e-05,
+      "loss": 0.1571,
+      "mean_token_accuracy": 0.9679389595985413,
+      "num_tokens": 1837799.0,
+      "step": 887
+    },
+    {
+      "epoch": 0.007340116879789054,
+      "grad_norm": 0.37589630484580994,
+      "learning_rate": 4.981670372543995e-05,
+      "loss": 0.1775,
+      "mean_token_accuracy": 0.9660415649414062,
+      "num_tokens": 1839778.0,
+      "step": 888
+    },
+    {
+      "epoch": 0.007348382777176204,
+      "grad_norm": 0.2645249664783478,
+      "learning_rate": 4.9816497078005275e-05,
+      "loss": 0.1357,
+      "mean_token_accuracy": 0.9729039072990417,
+      "num_tokens": 1841740.0,
+      "step": 889
+    },
+    {
+      "epoch": 0.007356648674563354,
+      "grad_norm": 0.37919551134109497,
+      "learning_rate": 4.9816290430570597e-05,
+      "loss": 0.1744,
+      "mean_token_accuracy": 0.9636091589927673,
+      "num_tokens": 1843752.0,
+      "step": 890
+    },
+    {
+      "epoch": 0.007364914571950504,
+      "grad_norm": 0.36111271381378174,
+      "learning_rate": 4.981608378313592e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9647467732429504,
+      "num_tokens": 1845772.0,
+      "step": 891
+    },
+    {
+      "epoch": 0.007373180469337654,
+      "grad_norm": 0.4354729652404785,
+      "learning_rate": 4.981587713570124e-05,
+      "loss": 0.2511,
+      "mean_token_accuracy": 0.9476801753044128,
+      "num_tokens": 1847804.0,
+      "step": 892
+    },
+    {
+      "epoch": 0.007381446366724803,
+      "grad_norm": 0.30511441826820374,
+      "learning_rate": 4.981567048826656e-05,
+      "loss": 0.1633,
+      "mean_token_accuracy": 0.968988299369812,
+      "num_tokens": 1849777.0,
+      "step": 893
+    },
+    {
+      "epoch": 0.007389712264111953,
+      "grad_norm": 0.46899569034576416,
+      "learning_rate": 4.981546384083188e-05,
+      "loss": 0.1945,
+      "mean_token_accuracy": 0.9620758295059204,
+      "num_tokens": 1851787.0,
+      "step": 894
+    },
+    {
+      "epoch": 0.0073979781614991035,
+      "grad_norm": 0.4520535469055176,
+      "learning_rate": 4.9815257193397205e-05,
+      "loss": 0.1975,
+      "mean_token_accuracy": 0.9593052268028259,
+      "num_tokens": 1853808.0,
+      "step": 895
+    },
+    {
+      "epoch": 0.007406244058886253,
+      "grad_norm": 0.3573121428489685,
+      "learning_rate": 4.9815050545962527e-05,
+      "loss": 0.155,
+      "mean_token_accuracy": 0.9676113128662109,
+      "num_tokens": 1855790.0,
+      "step": 896
+    },
+    {
+      "epoch": 0.007414509956273403,
+      "grad_norm": 0.4316529631614685,
+      "learning_rate": 4.981484389852785e-05,
+      "loss": 0.1926,
+      "mean_token_accuracy": 0.9633718132972717,
+      "num_tokens": 1857789.0,
+      "step": 897
+    },
+    {
+      "epoch": 0.007422775853660552,
+      "grad_norm": 0.33973968029022217,
+      "learning_rate": 4.981463725109316e-05,
+      "loss": 0.1448,
+      "mean_token_accuracy": 0.9667003154754639,
+      "num_tokens": 1859777.0,
+      "step": 898
+    },
+    {
+      "epoch": 0.0074310417510477025,
+      "grad_norm": 0.3881532549858093,
+      "learning_rate": 4.981443060365849e-05,
+      "loss": 0.1802,
+      "mean_token_accuracy": 0.9585414528846741,
+      "num_tokens": 1861785.0,
+      "step": 899
+    },
+    {
+      "epoch": 0.007439307648434853,
+      "grad_norm": 0.39037537574768066,
+      "learning_rate": 4.981422395622381e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9659035801887512,
+      "num_tokens": 1863844.0,
+      "step": 900
+    },
+    {
+      "epoch": 0.007447573545822002,
+      "grad_norm": 0.3870031237602234,
+      "learning_rate": 4.981401730878913e-05,
+      "loss": 0.2114,
+      "mean_token_accuracy": 0.9571285843849182,
+      "num_tokens": 1865856.0,
+      "step": 901
+    },
+    {
+      "epoch": 0.007455839443209152,
+      "grad_norm": 0.39852142333984375,
+      "learning_rate": 4.9813810661354456e-05,
+      "loss": 0.1754,
+      "mean_token_accuracy": 0.9642126560211182,
+      "num_tokens": 1867818.0,
+      "step": 902
+    },
+    {
+      "epoch": 0.0074641053405963016,
+      "grad_norm": 0.3147163689136505,
+      "learning_rate": 4.981360401391977e-05,
+      "loss": 0.1273,
+      "mean_token_accuracy": 0.9753340482711792,
+      "num_tokens": 1869770.0,
+      "step": 903
+    },
+    {
+      "epoch": 0.007472371237983452,
+      "grad_norm": 0.4568729102611542,
+      "learning_rate": 4.981339736648509e-05,
+      "loss": 0.19,
+      "mean_token_accuracy": 0.9620000123977661,
+      "num_tokens": 1871776.0,
+      "step": 904
+    },
+    {
+      "epoch": 0.007480637135370602,
+      "grad_norm": 0.501317024230957,
+      "learning_rate": 4.9813190719050415e-05,
+      "loss": 0.2305,
+      "mean_token_accuracy": 0.9508993625640869,
+      "num_tokens": 1873839.0,
+      "step": 905
+    },
+    {
+      "epoch": 0.007488903032757751,
+      "grad_norm": 0.5853086113929749,
+      "learning_rate": 4.9812984071615736e-05,
+      "loss": 0.3194,
+      "mean_token_accuracy": 0.9332392811775208,
+      "num_tokens": 1875972.0,
+      "step": 906
+    },
+    {
+      "epoch": 0.0074971689301449014,
+      "grad_norm": 0.4364166259765625,
+      "learning_rate": 4.981277742418106e-05,
+      "loss": 0.2387,
+      "mean_token_accuracy": 0.9537037014961243,
+      "num_tokens": 1878030.0,
+      "step": 907
+    },
+    {
+      "epoch": 0.007505434827532051,
+      "grad_norm": 0.44608074426651,
+      "learning_rate": 4.981257077674638e-05,
+      "loss": 0.2287,
+      "mean_token_accuracy": 0.9534301161766052,
+      "num_tokens": 1880033.0,
+      "step": 908
+    },
+    {
+      "epoch": 0.007513700724919201,
+      "grad_norm": 0.3870539963245392,
+      "learning_rate": 4.98123641293117e-05,
+      "loss": 0.2242,
+      "mean_token_accuracy": 0.9556331038475037,
+      "num_tokens": 1882045.0,
+      "step": 909
+    },
+    {
+      "epoch": 0.00752196662230635,
+      "grad_norm": 0.37544041872024536,
+      "learning_rate": 4.981215748187702e-05,
+      "loss": 0.1775,
+      "mean_token_accuracy": 0.9661957621574402,
+      "num_tokens": 1884033.0,
+      "step": 910
+    },
+    {
+      "epoch": 0.0075302325196935005,
+      "grad_norm": 0.35216644406318665,
+      "learning_rate": 4.9811950834442345e-05,
+      "loss": 0.1689,
+      "mean_token_accuracy": 0.9621020555496216,
+      "num_tokens": 1886018.0,
+      "step": 911
+    },
+    {
+      "epoch": 0.007538498417080651,
+      "grad_norm": 0.4044584333896637,
+      "learning_rate": 4.9811744187007666e-05,
+      "loss": 0.193,
+      "mean_token_accuracy": 0.9575757384300232,
+      "num_tokens": 1888004.0,
+      "step": 912
+    },
+    {
+      "epoch": 0.0075467643144678,
+      "grad_norm": 0.35822993516921997,
+      "learning_rate": 4.981153753957299e-05,
+      "loss": 0.1556,
+      "mean_token_accuracy": 0.965777575969696,
+      "num_tokens": 1889997.0,
+      "step": 913
+    },
+    {
+      "epoch": 0.00755503021185495,
+      "grad_norm": 0.32807430624961853,
+      "learning_rate": 4.981133089213831e-05,
+      "loss": 0.1529,
+      "mean_token_accuracy": 0.9670885801315308,
+      "num_tokens": 1891978.0,
+      "step": 914
+    },
+    {
+      "epoch": 0.0075632961092420995,
+      "grad_norm": 0.34989964962005615,
+      "learning_rate": 4.9811124244703624e-05,
+      "loss": 0.1857,
+      "mean_token_accuracy": 0.9636363387107849,
+      "num_tokens": 1893964.0,
+      "step": 915
+    },
+    {
+      "epoch": 0.00757156200662925,
+      "grad_norm": 0.3590420186519623,
+      "learning_rate": 4.981091759726895e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9637255072593689,
+      "num_tokens": 1896010.0,
+      "step": 916
+    },
+    {
+      "epoch": 0.0075798279040164,
+      "grad_norm": 0.41440725326538086,
+      "learning_rate": 4.9810710949834275e-05,
+      "loss": 0.286,
+      "mean_token_accuracy": 0.9443916082382202,
+      "num_tokens": 1898120.0,
+      "step": 917
+    },
+    {
+      "epoch": 0.007588093801403549,
+      "grad_norm": 0.3265133202075958,
+      "learning_rate": 4.981050430239959e-05,
+      "loss": 0.1578,
+      "mean_token_accuracy": 0.9685279130935669,
+      "num_tokens": 1900096.0,
+      "step": 918
+    },
+    {
+      "epoch": 0.007596359698790699,
+      "grad_norm": 0.39172786474227905,
+      "learning_rate": 4.981029765496492e-05,
+      "loss": 0.2176,
+      "mean_token_accuracy": 0.9558680653572083,
+      "num_tokens": 1902164.0,
+      "step": 919
+    },
+    {
+      "epoch": 0.007604625596177849,
+      "grad_norm": 0.3920506536960602,
+      "learning_rate": 4.981009100753023e-05,
+      "loss": 0.1834,
+      "mean_token_accuracy": 0.963873565196991,
+      "num_tokens": 1904163.0,
+      "step": 920
+    },
+    {
+      "epoch": 0.007612891493564999,
+      "grad_norm": 0.40847542881965637,
+      "learning_rate": 4.9809884360095554e-05,
+      "loss": 0.2257,
+      "mean_token_accuracy": 0.9582712650299072,
+      "num_tokens": 1906182.0,
+      "step": 921
+    },
+    {
+      "epoch": 0.007621157390952149,
+      "grad_norm": 0.31902965903282166,
+      "learning_rate": 4.9809677712660876e-05,
+      "loss": 0.1502,
+      "mean_token_accuracy": 0.9720101952552795,
+      "num_tokens": 1908153.0,
+      "step": 922
+    },
+    {
+      "epoch": 0.007629423288339298,
+      "grad_norm": 0.45197224617004395,
+      "learning_rate": 4.98094710652262e-05,
+      "loss": 0.2017,
+      "mean_token_accuracy": 0.9601196646690369,
+      "num_tokens": 1910165.0,
+      "step": 923
+    },
+    {
+      "epoch": 0.007637689185726449,
+      "grad_norm": 0.40337854623794556,
+      "learning_rate": 4.980926441779152e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9667171239852905,
+      "num_tokens": 1912154.0,
+      "step": 924
+    },
+    {
+      "epoch": 0.007645955083113598,
+      "grad_norm": 0.39368268847465515,
+      "learning_rate": 4.980905777035684e-05,
+      "loss": 0.2121,
+      "mean_token_accuracy": 0.9588559865951538,
+      "num_tokens": 1914153.0,
+      "step": 925
+    },
+    {
+      "epoch": 0.007654220980500748,
+      "grad_norm": 0.4261470139026642,
+      "learning_rate": 4.980885112292216e-05,
+      "loss": 0.2412,
+      "mean_token_accuracy": 0.9512194991111755,
+      "num_tokens": 1916250.0,
+      "step": 926
+    },
+    {
+      "epoch": 0.007662486877887898,
+      "grad_norm": 0.5017535090446472,
+      "learning_rate": 4.9808644475487484e-05,
+      "loss": 0.2617,
+      "mean_token_accuracy": 0.9467980265617371,
+      "num_tokens": 1918286.0,
+      "step": 927
+    },
+    {
+      "epoch": 0.007670752775275048,
+      "grad_norm": 0.3660185933113098,
+      "learning_rate": 4.9808437828052806e-05,
+      "loss": 0.201,
+      "mean_token_accuracy": 0.9614814519882202,
+      "num_tokens": 1920317.0,
+      "step": 928
+    },
+    {
+      "epoch": 0.007679018672662198,
+      "grad_norm": 0.3857627809047699,
+      "learning_rate": 4.980823118061813e-05,
+      "loss": 0.1726,
+      "mean_token_accuracy": 0.9633960127830505,
+      "num_tokens": 1922290.0,
+      "step": 929
+    },
+    {
+      "epoch": 0.007687284570049347,
+      "grad_norm": 0.4505801796913147,
+      "learning_rate": 4.980802453318344e-05,
+      "loss": 0.2391,
+      "mean_token_accuracy": 0.9511241316795349,
+      "num_tokens": 1924342.0,
+      "step": 930
+    },
+    {
+      "epoch": 0.007695550467436497,
+      "grad_norm": 0.5029123425483704,
+      "learning_rate": 4.980781788574877e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9662298560142517,
+      "num_tokens": 1926332.0,
+      "step": 931
+    },
+    {
+      "epoch": 0.0077038163648236475,
+      "grad_norm": 0.41090860962867737,
+      "learning_rate": 4.9807611238314086e-05,
+      "loss": 0.1592,
+      "mean_token_accuracy": 0.9629072546958923,
+      "num_tokens": 1928333.0,
+      "step": 932
+    },
+    {
+      "epoch": 0.007712082262210797,
+      "grad_norm": 0.45338863134384155,
+      "learning_rate": 4.980740459087941e-05,
+      "loss": 0.2259,
+      "mean_token_accuracy": 0.9506595134735107,
+      "num_tokens": 1930386.0,
+      "step": 933
+    },
+    {
+      "epoch": 0.007720348159597947,
+      "grad_norm": 0.3854855000972748,
+      "learning_rate": 4.9807197943444736e-05,
+      "loss": 0.2261,
+      "mean_token_accuracy": 0.9510791301727295,
+      "num_tokens": 1932477.0,
+      "step": 934
+    },
+    {
+      "epoch": 0.007728614056985096,
+      "grad_norm": 0.38385406136512756,
+      "learning_rate": 4.980699129601005e-05,
+      "loss": 0.1519,
+      "mean_token_accuracy": 0.9644288420677185,
+      "num_tokens": 1934479.0,
+      "step": 935
+    },
+    {
+      "epoch": 0.0077368799543722465,
+      "grad_norm": 0.39230501651763916,
+      "learning_rate": 4.980678464857538e-05,
+      "loss": 0.1455,
+      "mean_token_accuracy": 0.9668674468994141,
+      "num_tokens": 1936477.0,
+      "step": 936
+    },
+    {
+      "epoch": 0.007745145851759397,
+      "grad_norm": 0.41803377866744995,
+      "learning_rate": 4.9806578001140694e-05,
+      "loss": 0.1982,
+      "mean_token_accuracy": 0.9587525129318237,
+      "num_tokens": 1938471.0,
+      "step": 937
+    },
+    {
+      "epoch": 0.007753411749146546,
+      "grad_norm": 0.4002348780632019,
+      "learning_rate": 4.9806371353706016e-05,
+      "loss": 0.1612,
+      "mean_token_accuracy": 0.970933198928833,
+      "num_tokens": 1940438.0,
+      "step": 938
+    },
+    {
+      "epoch": 0.007761677646533696,
+      "grad_norm": 0.7597206830978394,
+      "learning_rate": 4.9806164706271344e-05,
+      "loss": 0.2265,
+      "mean_token_accuracy": 0.957198441028595,
+      "num_tokens": 1942500.0,
+      "step": 939
+    },
+    {
+      "epoch": 0.0077699435439208455,
+      "grad_norm": 0.3641800284385681,
+      "learning_rate": 4.980595805883666e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.9631685018539429,
+      "num_tokens": 1944488.0,
+      "step": 940
+    },
+    {
+      "epoch": 0.007778209441307996,
+      "grad_norm": 0.43571937084198,
+      "learning_rate": 4.980575141140198e-05,
+      "loss": 0.2031,
+      "mean_token_accuracy": 0.9577187895774841,
+      "num_tokens": 1946528.0,
+      "step": 941
+    },
+    {
+      "epoch": 0.007786475338695145,
+      "grad_norm": 0.3917180299758911,
+      "learning_rate": 4.98055447639673e-05,
+      "loss": 0.223,
+      "mean_token_accuracy": 0.9562406539916992,
+      "num_tokens": 1948545.0,
+      "step": 942
+    },
+    {
+      "epoch": 0.007794741236082295,
+      "grad_norm": 0.3620387017726898,
+      "learning_rate": 4.9805338116532624e-05,
+      "loss": 0.1382,
+      "mean_token_accuracy": 0.9690198302268982,
+      "num_tokens": 1950520.0,
+      "step": 943
+    },
+    {
+      "epoch": 0.007803007133469445,
+      "grad_norm": 0.4767442047595978,
+      "learning_rate": 4.9805131469097946e-05,
+      "loss": 0.2268,
+      "mean_token_accuracy": 0.9569733142852783,
+      "num_tokens": 1952548.0,
+      "step": 944
+    },
+    {
+      "epoch": 0.007811273030856595,
+      "grad_norm": 0.36579081416130066,
+      "learning_rate": 4.980492482166327e-05,
+      "loss": 0.1507,
+      "mean_token_accuracy": 0.9702621102333069,
+      "num_tokens": 1954538.0,
+      "step": 945
+    },
+    {
+      "epoch": 0.007819538928243745,
+      "grad_norm": 0.43252885341644287,
+      "learning_rate": 4.980471817422859e-05,
+      "loss": 0.2563,
+      "mean_token_accuracy": 0.9483430981636047,
+      "num_tokens": 1956596.0,
+      "step": 946
+    },
+    {
+      "epoch": 0.007827804825630894,
+      "grad_norm": 0.6051336526870728,
+      "learning_rate": 4.9804511526793904e-05,
+      "loss": 0.2769,
+      "mean_token_accuracy": 0.9462729692459106,
+      "num_tokens": 1958668.0,
+      "step": 947
+    },
+    {
+      "epoch": 0.007836070723018045,
+      "grad_norm": 0.3463861346244812,
+      "learning_rate": 4.980430487935923e-05,
+      "loss": 0.1485,
+      "mean_token_accuracy": 0.9695122241973877,
+      "num_tokens": 1960642.0,
+      "step": 948
+    },
+    {
+      "epoch": 0.007844336620405195,
+      "grad_norm": 0.3766689598560333,
+      "learning_rate": 4.9804098231924554e-05,
+      "loss": 0.1924,
+      "mean_token_accuracy": 0.9621890783309937,
+      "num_tokens": 1962658.0,
+      "step": 949
+    },
+    {
+      "epoch": 0.007852602517792344,
+      "grad_norm": 0.4464820623397827,
+      "learning_rate": 4.980389158448987e-05,
+      "loss": 0.1612,
+      "mean_token_accuracy": 0.9684368968009949,
+      "num_tokens": 1964660.0,
+      "step": 950
+    },
+    {
+      "epoch": 0.007860868415179493,
+      "grad_norm": 0.2868194282054901,
+      "learning_rate": 4.98036849370552e-05,
+      "loss": 0.1509,
+      "mean_token_accuracy": 0.9711246490478516,
+      "num_tokens": 1966640.0,
+      "step": 951
+    },
+    {
+      "epoch": 0.007869134312566644,
+      "grad_norm": 0.2970883846282959,
+      "learning_rate": 4.980347828962051e-05,
+      "loss": 0.1549,
+      "mean_token_accuracy": 0.969774603843689,
+      "num_tokens": 1968598.0,
+      "step": 952
+    },
+    {
+      "epoch": 0.007877400209953794,
+      "grad_norm": 0.39536306262016296,
+      "learning_rate": 4.9803271642185834e-05,
+      "loss": 0.1863,
+      "mean_token_accuracy": 0.9632840156555176,
+      "num_tokens": 1970565.0,
+      "step": 953
+    },
+    {
+      "epoch": 0.007885666107340943,
+      "grad_norm": 0.4233575165271759,
+      "learning_rate": 4.9803064994751156e-05,
+      "loss": 0.1838,
+      "mean_token_accuracy": 0.9616171717643738,
+      "num_tokens": 1972525.0,
+      "step": 954
+    },
+    {
+      "epoch": 0.007893932004728094,
+      "grad_norm": 0.505504846572876,
+      "learning_rate": 4.980285834731648e-05,
+      "loss": 0.2767,
+      "mean_token_accuracy": 0.9472422003746033,
+      "num_tokens": 1974616.0,
+      "step": 955
+    },
+    {
+      "epoch": 0.007902197902115243,
+      "grad_norm": 0.4857965409755707,
+      "learning_rate": 4.9802651699881806e-05,
+      "loss": 0.1712,
+      "mean_token_accuracy": 0.9655172228813171,
+      "num_tokens": 1976565.0,
+      "step": 956
+    },
+    {
+      "epoch": 0.007910463799502393,
+      "grad_norm": 0.42851853370666504,
+      "learning_rate": 4.980244505244712e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.961422860622406,
+      "num_tokens": 1978567.0,
+      "step": 957
+    },
+    {
+      "epoch": 0.007918729696889542,
+      "grad_norm": 0.3855806291103363,
+      "learning_rate": 4.980223840501244e-05,
+      "loss": 0.185,
+      "mean_token_accuracy": 0.9627268314361572,
+      "num_tokens": 1980612.0,
+      "step": 958
+    },
+    {
+      "epoch": 0.007926995594276693,
+      "grad_norm": 0.44890686869621277,
+      "learning_rate": 4.9802031757577764e-05,
+      "loss": 0.2755,
+      "mean_token_accuracy": 0.942307710647583,
+      "num_tokens": 1982750.0,
+      "step": 959
+    },
+    {
+      "epoch": 0.007935261491663842,
+      "grad_norm": 0.3707360327243805,
+      "learning_rate": 4.9801825110143085e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9668028354644775,
+      "num_tokens": 1984714.0,
+      "step": 960
+    },
+    {
+      "epoch": 0.007943527389050992,
+      "grad_norm": 0.457932710647583,
+      "learning_rate": 4.980161846270841e-05,
+      "loss": 0.1874,
+      "mean_token_accuracy": 0.9626865386962891,
+      "num_tokens": 1986730.0,
+      "step": 961
+    },
+    {
+      "epoch": 0.007951793286438143,
+      "grad_norm": 0.3275972604751587,
+      "learning_rate": 4.980141181527373e-05,
+      "loss": 0.1438,
+      "mean_token_accuracy": 0.9697285890579224,
+      "num_tokens": 1988652.0,
+      "step": 962
+    },
+    {
+      "epoch": 0.007960059183825292,
+      "grad_norm": 0.39132949709892273,
+      "learning_rate": 4.980120516783905e-05,
+      "loss": 0.1927,
+      "mean_token_accuracy": 0.9604009985923767,
+      "num_tokens": 1990653.0,
+      "step": 963
+    },
+    {
+      "epoch": 0.007968325081212441,
+      "grad_norm": 0.4453590214252472,
+      "learning_rate": 4.9800998520404365e-05,
+      "loss": 0.2627,
+      "mean_token_accuracy": 0.9438679218292236,
+      "num_tokens": 1992779.0,
+      "step": 964
+    },
+    {
+      "epoch": 0.007976590978599592,
+      "grad_norm": 0.4846074879169464,
+      "learning_rate": 4.9800791872969694e-05,
+      "loss": 0.1997,
+      "mean_token_accuracy": 0.9597585797309875,
+      "num_tokens": 1994773.0,
+      "step": 965
+    },
+    {
+      "epoch": 0.007984856875986742,
+      "grad_norm": 0.3702278137207031,
+      "learning_rate": 4.9800585225535015e-05,
+      "loss": 0.1423,
+      "mean_token_accuracy": 0.9713135361671448,
+      "num_tokens": 1996766.0,
+      "step": 966
+    },
+    {
+      "epoch": 0.007993122773373891,
+      "grad_norm": 0.3540761172771454,
+      "learning_rate": 4.980037857810033e-05,
+      "loss": 0.1977,
+      "mean_token_accuracy": 0.9621514081954956,
+      "num_tokens": 1998780.0,
+      "step": 967
+    },
+    {
+      "epoch": 0.00800138867076104,
+      "grad_norm": 0.3455694019794464,
+      "learning_rate": 4.980017193066566e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9707808494567871,
+      "num_tokens": 2000771.0,
+      "step": 968
+    },
+    {
+      "epoch": 0.008009654568148191,
+      "grad_norm": 0.37254098057746887,
+      "learning_rate": 4.9799965283230974e-05,
+      "loss": 0.1699,
+      "mean_token_accuracy": 0.9640393853187561,
+      "num_tokens": 2002807.0,
+      "step": 969
+    },
+    {
+      "epoch": 0.00801792046553534,
+      "grad_norm": 0.3742963969707489,
+      "learning_rate": 4.9799758635796295e-05,
+      "loss": 0.1905,
+      "mean_token_accuracy": 0.9588235020637512,
+      "num_tokens": 2004853.0,
+      "step": 970
+    },
+    {
+      "epoch": 0.00802618636292249,
+      "grad_norm": 0.3958035111427307,
+      "learning_rate": 4.979955198836162e-05,
+      "loss": 0.1985,
+      "mean_token_accuracy": 0.9617296457290649,
+      "num_tokens": 2006871.0,
+      "step": 971
+    },
+    {
+      "epoch": 0.008034452260309641,
+      "grad_norm": 0.3543177843093872,
+      "learning_rate": 4.979934534092694e-05,
+      "loss": 0.1898,
+      "mean_token_accuracy": 0.9626680016517639,
+      "num_tokens": 2008886.0,
+      "step": 972
+    },
+    {
+      "epoch": 0.00804271815769679,
+      "grad_norm": 0.3872028589248657,
+      "learning_rate": 4.979913869349226e-05,
+      "loss": 0.1995,
+      "mean_token_accuracy": 0.9607458114624023,
+      "num_tokens": 2010930.0,
+      "step": 973
+    },
+    {
+      "epoch": 0.00805098405508394,
+      "grad_norm": 0.5003104209899902,
+      "learning_rate": 4.979893204605758e-05,
+      "loss": 0.2237,
+      "mean_token_accuracy": 0.9536489248275757,
+      "num_tokens": 2012964.0,
+      "step": 974
+    },
+    {
+      "epoch": 0.00805924995247109,
+      "grad_norm": 0.41863206028938293,
+      "learning_rate": 4.9798725398622904e-05,
+      "loss": 0.1941,
+      "mean_token_accuracy": 0.9597180485725403,
+      "num_tokens": 2014956.0,
+      "step": 975
+    },
+    {
+      "epoch": 0.00806751584985824,
+      "grad_norm": 0.4648338556289673,
+      "learning_rate": 4.9798518751188225e-05,
+      "loss": 0.2449,
+      "mean_token_accuracy": 0.9497805833816528,
+      "num_tokens": 2017013.0,
+      "step": 976
+    },
+    {
+      "epoch": 0.00807578174724539,
+      "grad_norm": 0.3554210960865021,
+      "learning_rate": 4.979831210375355e-05,
+      "loss": 0.1655,
+      "mean_token_accuracy": 0.9632571935653687,
+      "num_tokens": 2019033.0,
+      "step": 977
+    },
+    {
+      "epoch": 0.008084047644632539,
+      "grad_norm": 0.3880045413970947,
+      "learning_rate": 4.979810545631887e-05,
+      "loss": 0.1476,
+      "mean_token_accuracy": 0.9674837589263916,
+      "num_tokens": 2021038.0,
+      "step": 978
+    },
+    {
+      "epoch": 0.00809231354201969,
+      "grad_norm": 0.3297460377216339,
+      "learning_rate": 4.979789880888419e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9670885801315308,
+      "num_tokens": 2023019.0,
+      "step": 979
+    },
+    {
+      "epoch": 0.00810057943940684,
+      "grad_norm": 0.3558855652809143,
+      "learning_rate": 4.979769216144951e-05,
+      "loss": 0.1882,
+      "mean_token_accuracy": 0.9620758295059204,
+      "num_tokens": 2025029.0,
+      "step": 980
+    },
+    {
+      "epoch": 0.008108845336793988,
+      "grad_norm": 0.3962882459163666,
+      "learning_rate": 4.979748551401483e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9603960514068604,
+      "num_tokens": 2027055.0,
+      "step": 981
+    },
+    {
+      "epoch": 0.00811711123418114,
+      "grad_norm": 0.5223516225814819,
+      "learning_rate": 4.9797278866580155e-05,
+      "loss": 0.1931,
+      "mean_token_accuracy": 0.9574257135391235,
+      "num_tokens": 2029081.0,
+      "step": 982
+    },
+    {
+      "epoch": 0.008125377131568289,
+      "grad_norm": 0.4435831606388092,
+      "learning_rate": 4.979707221914548e-05,
+      "loss": 0.2031,
+      "mean_token_accuracy": 0.9564152956008911,
+      "num_tokens": 2031129.0,
+      "step": 983
+    },
+    {
+      "epoch": 0.008133643028955438,
+      "grad_norm": 0.3897659480571747,
+      "learning_rate": 4.979686557171079e-05,
+      "loss": 0.1775,
+      "mean_token_accuracy": 0.9674962162971497,
+      "num_tokens": 2033104.0,
+      "step": 984
+    },
+    {
+      "epoch": 0.00814190892634259,
+      "grad_norm": 0.5357649326324463,
+      "learning_rate": 4.979665892427612e-05,
+      "loss": 0.1784,
+      "mean_token_accuracy": 0.959132194519043,
+      "num_tokens": 2035092.0,
+      "step": 985
+    },
+    {
+      "epoch": 0.008150174823729739,
+      "grad_norm": 0.40512484312057495,
+      "learning_rate": 4.9796452276841435e-05,
+      "loss": 0.1939,
+      "mean_token_accuracy": 0.9609929323196411,
+      "num_tokens": 2037072.0,
+      "step": 986
+    },
+    {
+      "epoch": 0.008158440721116888,
+      "grad_norm": 0.49325162172317505,
+      "learning_rate": 4.979624562940676e-05,
+      "loss": 0.1884,
+      "mean_token_accuracy": 0.9609929323196411,
+      "num_tokens": 2039052.0,
+      "step": 987
+    },
+    {
+      "epoch": 0.008166706618504037,
+      "grad_norm": 0.49558088183403015,
+      "learning_rate": 4.9796038981972085e-05,
+      "loss": 0.2092,
+      "mean_token_accuracy": 0.9586241245269775,
+      "num_tokens": 2041064.0,
+      "step": 988
+    },
+    {
+      "epoch": 0.008174972515891188,
+      "grad_norm": 0.34452107548713684,
+      "learning_rate": 4.97958323345374e-05,
+      "loss": 0.1755,
+      "mean_token_accuracy": 0.9629629850387573,
+      "num_tokens": 2043041.0,
+      "step": 989
+    },
+    {
+      "epoch": 0.008183238413278338,
+      "grad_norm": 0.4018665850162506,
+      "learning_rate": 4.979562568710272e-05,
+      "loss": 0.2419,
+      "mean_token_accuracy": 0.951361894607544,
+      "num_tokens": 2045103.0,
+      "step": 990
+    },
+    {
+      "epoch": 0.008191504310665487,
+      "grad_norm": 0.42745086550712585,
+      "learning_rate": 4.979541903966804e-05,
+      "loss": 0.2779,
+      "mean_token_accuracy": 0.9447406530380249,
+      "num_tokens": 2047172.0,
+      "step": 991
+    },
+    {
+      "epoch": 0.008199770208052638,
+      "grad_norm": 0.37178975343704224,
+      "learning_rate": 4.9795212392233365e-05,
+      "loss": 0.1647,
+      "mean_token_accuracy": 0.9659318923950195,
+      "num_tokens": 2049174.0,
+      "step": 992
+    },
+    {
+      "epoch": 0.008208036105439787,
+      "grad_norm": 0.3532976508140564,
+      "learning_rate": 4.979500574479869e-05,
+      "loss": 0.1984,
+      "mean_token_accuracy": 0.9630005359649658,
+      "num_tokens": 2051153.0,
+      "step": 993
+    },
+    {
+      "epoch": 0.008216302002826937,
+      "grad_norm": 0.3711738586425781,
+      "learning_rate": 4.979479909736401e-05,
+      "loss": 0.219,
+      "mean_token_accuracy": 0.9540566802024841,
+      "num_tokens": 2053205.0,
+      "step": 994
+    },
+    {
+      "epoch": 0.008224567900214086,
+      "grad_norm": 0.256096750497818,
+      "learning_rate": 4.979459244992933e-05,
+      "loss": 0.1299,
+      "mean_token_accuracy": 0.974452555179596,
+      "num_tokens": 2055129.0,
+      "step": 995
+    },
+    {
+      "epoch": 0.008232833797601237,
+      "grad_norm": 0.40168118476867676,
+      "learning_rate": 4.9794385802494645e-05,
+      "loss": 0.1703,
+      "mean_token_accuracy": 0.9630566835403442,
+      "num_tokens": 2057111.0,
+      "step": 996
+    },
+    {
+      "epoch": 0.008241099694988386,
+      "grad_norm": 0.436128169298172,
+      "learning_rate": 4.979417915505997e-05,
+      "loss": 0.24,
+      "mean_token_accuracy": 0.9478811621665955,
+      "num_tokens": 2059170.0,
+      "step": 997
+    },
+    {
+      "epoch": 0.008249365592375536,
+      "grad_norm": 0.3948153555393219,
+      "learning_rate": 4.9793972507625295e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.9680203199386597,
+      "num_tokens": 2061146.0,
+      "step": 998
+    },
+    {
+      "epoch": 0.008257631489762687,
+      "grad_norm": 0.3767959475517273,
+      "learning_rate": 4.9793765860190617e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9674962162971497,
+      "num_tokens": 2063121.0,
+      "step": 999
+    },
+    {
+      "epoch": 0.008265897387149836,
+      "grad_norm": 0.4880599081516266,
+      "learning_rate": 4.979355921275594e-05,
+      "loss": 0.278,
+      "mean_token_accuracy": 0.9402697682380676,
+      "num_tokens": 2065203.0,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 241958,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.729254097643712e+16,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}