diff --git "a/checkpoint-426/trainer_state.json" "b/checkpoint-426/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-426/trainer_state.json"
@@ -0,0 +1,4327 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 426,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.78125,
+      "epoch": 0.007042253521126761,
+      "grad_norm": 5.375,
+      "learning_rate": 5e-05,
+      "loss": 2.342320680618286,
+      "mean_token_accuracy": 0.5398398637771606,
+      "num_tokens": 2631.0,
+      "step": 1
+    },
+    {
+      "entropy": 1.7265625,
+      "epoch": 0.014084507042253521,
+      "grad_norm": 4.28125,
+      "learning_rate": 5e-05,
+      "loss": 2.184455394744873,
+      "mean_token_accuracy": 0.5707964897155762,
+      "num_tokens": 5577.0,
+      "step": 2
+    },
+    {
+      "entropy": 1.5859375,
+      "epoch": 0.02112676056338028,
+      "grad_norm": 3.796875,
+      "learning_rate": 5e-05,
+      "loss": 1.9440301656723022,
+      "mean_token_accuracy": 0.6223160624504089,
+      "num_tokens": 8426.0,
+      "step": 3
+    },
+    {
+      "entropy": 1.75,
+      "epoch": 0.028169014084507043,
+      "grad_norm": 3.234375,
+      "learning_rate": 5e-05,
+      "loss": 2.034520149230957,
+      "mean_token_accuracy": 0.5993841886520386,
+      "num_tokens": 11357.0,
+      "step": 4
+    },
+    {
+      "entropy": 1.8984375,
+      "epoch": 0.035211267605633804,
+      "grad_norm": 3.3125,
+      "learning_rate": 5e-05,
+      "loss": 2.151127338409424,
+      "mean_token_accuracy": 0.563842236995697,
+      "num_tokens": 13926.0,
+      "step": 5
+    },
+    {
+      "entropy": 1.890625,
+      "epoch": 0.04225352112676056,
+      "grad_norm": 2.546875,
+      "learning_rate": 5e-05,
+      "loss": 2.066777229309082,
+      "mean_token_accuracy": 0.5754261612892151,
+      "num_tokens": 17043.0,
+      "step": 6
+    },
+    {
+      "entropy": 2.03125,
+      "epoch": 0.04929577464788732,
+      "grad_norm": 2.9375,
+      "learning_rate": 5e-05,
+      "loss": 2.2290213108062744,
+      "mean_token_accuracy": 0.5708707571029663,
+      "num_tokens": 19612.0,
+      "step": 7
+    },
+    {
+      "entropy": 1.8046875,
+      "epoch": 0.056338028169014086,
+      "grad_norm": 2.59375,
+      "learning_rate": 5e-05,
+      "loss": 1.9348505735397339,
+      "mean_token_accuracy": 0.6066350936889648,
+      "num_tokens": 22363.0,
+      "step": 8
+    },
+    {
+      "entropy": 1.7734375,
+      "epoch": 0.06338028169014084,
+      "grad_norm": 2.4375,
+      "learning_rate": 5e-05,
+      "loss": 1.8786349296569824,
+      "mean_token_accuracy": 0.628695011138916,
+      "num_tokens": 25145.0,
+      "step": 9
+    },
+    {
+      "entropy": 1.8515625,
+      "epoch": 0.07042253521126761,
+      "grad_norm": 2.296875,
+      "learning_rate": 5e-05,
+      "loss": 1.9190694093704224,
+      "mean_token_accuracy": 0.6125850081443787,
+      "num_tokens": 28093.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.8515625,
+      "epoch": 0.07746478873239436,
+      "grad_norm": 2.171875,
+      "learning_rate": 5e-05,
+      "loss": 1.891284704208374,
+      "mean_token_accuracy": 0.6322028040885925,
+      "num_tokens": 31138.0,
+      "step": 11
+    },
+    {
+      "entropy": 1.953125,
+      "epoch": 0.08450704225352113,
+      "grad_norm": 2.421875,
+      "learning_rate": 5e-05,
+      "loss": 1.948248267173767,
+      "mean_token_accuracy": 0.6148879528045654,
+      "num_tokens": 33779.0,
+      "step": 12
+    },
+    {
+      "entropy": 1.9609375,
+      "epoch": 0.09154929577464789,
+      "grad_norm": 2.671875,
+      "learning_rate": 5e-05,
+      "loss": 1.968241810798645,
+      "mean_token_accuracy": 0.6172575950622559,
+      "num_tokens": 36128.0,
+      "step": 13
+    },
+    {
+      "entropy": 1.8828125,
+      "epoch": 0.09859154929577464,
+      "grad_norm": 2.015625,
+      "learning_rate": 5e-05,
+      "loss": 1.902205228805542,
+      "mean_token_accuracy": 0.6156116724014282,
+      "num_tokens": 39185.0,
+      "step": 14
+    },
+    {
+      "entropy": 1.703125,
+      "epoch": 0.1056338028169014,
+      "grad_norm": 2.328125,
+      "learning_rate": 5e-05,
+      "loss": 1.6213911771774292,
+      "mean_token_accuracy": 0.6750187277793884,
+      "num_tokens": 41867.0,
+      "step": 15
+    },
+    {
+      "entropy": 2.015625,
+      "epoch": 0.11267605633802817,
+      "grad_norm": 2.234375,
+      "learning_rate": 5e-05,
+      "loss": 1.9176678657531738,
+      "mean_token_accuracy": 0.6337976455688477,
+      "num_tokens": 44603.0,
+      "step": 16
+    },
+    {
+      "entropy": 1.921875,
+      "epoch": 0.11971830985915492,
+      "grad_norm": 1.921875,
+      "learning_rate": 5e-05,
+      "loss": 1.8336955308914185,
+      "mean_token_accuracy": 0.6507288813591003,
+      "num_tokens": 48041.0,
+      "step": 17
+    },
+    {
+      "entropy": 1.828125,
+      "epoch": 0.1267605633802817,
+      "grad_norm": 2.046875,
+      "learning_rate": 5e-05,
+      "loss": 1.7068480253219604,
+      "mean_token_accuracy": 0.6735284924507141,
+      "num_tokens": 51158.0,
+      "step": 18
+    },
+    {
+      "entropy": 1.7421875,
+      "epoch": 0.13380281690140844,
+      "grad_norm": 1.8984375,
+      "learning_rate": 5e-05,
+      "loss": 1.6770858764648438,
+      "mean_token_accuracy": 0.6662946343421936,
+      "num_tokens": 54750.0,
+      "step": 19
+    },
+    {
+      "entropy": 2.109375,
+      "epoch": 0.14084507042253522,
+      "grad_norm": 2.828125,
+      "learning_rate": 5e-05,
+      "loss": 1.9946776628494263,
+      "mean_token_accuracy": 0.6195476651191711,
+      "num_tokens": 57234.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.6328125,
+      "epoch": 0.14788732394366197,
+      "grad_norm": 1.90625,
+      "learning_rate": 5e-05,
+      "loss": 1.5617338418960571,
+      "mean_token_accuracy": 0.6951566934585571,
+      "num_tokens": 60752.0,
+      "step": 21
+    },
+    {
+      "entropy": 1.78125,
+      "epoch": 0.15492957746478872,
+      "grad_norm": 2.515625,
+      "learning_rate": 5e-05,
+      "loss": 1.7242701053619385,
+      "mean_token_accuracy": 0.6592110395431519,
+      "num_tokens": 63219.0,
+      "step": 22
+    },
+    {
+      "entropy": 1.6796875,
+      "epoch": 0.1619718309859155,
+      "grad_norm": 1.9375,
+      "learning_rate": 5e-05,
+      "loss": 1.6089009046554565,
+      "mean_token_accuracy": 0.6697189807891846,
+      "num_tokens": 66394.0,
+      "step": 23
+    },
+    {
+      "entropy": 1.9296875,
+      "epoch": 0.16901408450704225,
+      "grad_norm": 2.125,
+      "learning_rate": 5e-05,
+      "loss": 1.8391156196594238,
+      "mean_token_accuracy": 0.6323254704475403,
+      "num_tokens": 69081.0,
+      "step": 24
+    },
+    {
+      "entropy": 1.7265625,
+      "epoch": 0.176056338028169,
+      "grad_norm": 2.15625,
+      "learning_rate": 5e-05,
+      "loss": 1.6465730667114258,
+      "mean_token_accuracy": 0.654959499835968,
+      "num_tokens": 71680.0,
+      "step": 25
+    },
+    {
+      "entropy": 1.9609375,
+      "epoch": 0.18309859154929578,
+      "grad_norm": 2.03125,
+      "learning_rate": 5e-05,
+      "loss": 1.873896598815918,
+      "mean_token_accuracy": 0.6338028311729431,
+      "num_tokens": 74670.0,
+      "step": 26
+    },
+    {
+      "entropy": 1.6875,
+      "epoch": 0.19014084507042253,
+      "grad_norm": 1.84375,
+      "learning_rate": 5e-05,
+      "loss": 1.629098653793335,
+      "mean_token_accuracy": 0.6646864414215088,
+      "num_tokens": 77708.0,
+      "step": 27
+    },
+    {
+      "entropy": 1.609375,
+      "epoch": 0.19718309859154928,
+      "grad_norm": 1.8671875,
+      "learning_rate": 5e-05,
+      "loss": 1.5465565919876099,
+      "mean_token_accuracy": 0.6922129988670349,
+      "num_tokens": 80965.0,
+      "step": 28
+    },
+    {
+      "entropy": 1.59375,
+      "epoch": 0.20422535211267606,
+      "grad_norm": 2.140625,
+      "learning_rate": 5e-05,
+      "loss": 1.5382331609725952,
+      "mean_token_accuracy": 0.6933178901672363,
+      "num_tokens": 83562.0,
+      "step": 29
+    },
+    {
+      "entropy": 1.5390625,
+      "epoch": 0.2112676056338028,
+      "grad_norm": 1.9921875,
+      "learning_rate": 5e-05,
+      "loss": 1.4340285062789917,
+      "mean_token_accuracy": 0.6944348216056824,
+      "num_tokens": 86463.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.5,
+      "epoch": 0.21830985915492956,
+      "grad_norm": 1.953125,
+      "learning_rate": 5e-05,
+      "loss": 1.41927969455719,
+      "mean_token_accuracy": 0.7242607474327087,
+      "num_tokens": 89278.0,
+      "step": 31
+    },
+    {
+      "entropy": 1.484375,
+      "epoch": 0.22535211267605634,
+      "grad_norm": 2.328125,
+      "learning_rate": 5e-05,
+      "loss": 1.395577311515808,
+      "mean_token_accuracy": 0.7233359813690186,
+      "num_tokens": 91780.0,
+      "step": 32
+    },
+    {
+      "entropy": 1.5234375,
+      "epoch": 0.2323943661971831,
+      "grad_norm": 1.8671875,
+      "learning_rate": 5e-05,
+      "loss": 1.4273605346679688,
+      "mean_token_accuracy": 0.7213010191917419,
+      "num_tokens": 94924.0,
+      "step": 33
+    },
+    {
+      "entropy": 1.640625,
+      "epoch": 0.23943661971830985,
+      "grad_norm": 1.9375,
+      "learning_rate": 5e-05,
+      "loss": 1.549876093864441,
+      "mean_token_accuracy": 0.6991525292396545,
+      "num_tokens": 98236.0,
+      "step": 34
+    },
+    {
+      "entropy": 1.3984375,
+      "epoch": 0.24647887323943662,
+      "grad_norm": 2.5625,
+      "learning_rate": 5e-05,
+      "loss": 1.2571226358413696,
+      "mean_token_accuracy": 0.7764654159545898,
+      "num_tokens": 100530.0,
+      "step": 35
+    },
+    {
+      "entropy": 1.484375,
+      "epoch": 0.2535211267605634,
+      "grad_norm": 1.9921875,
+      "learning_rate": 5e-05,
+      "loss": 1.430403709411621,
+      "mean_token_accuracy": 0.7238532304763794,
+      "num_tokens": 103808.0,
+      "step": 36
+    },
+    {
+      "entropy": 1.4375,
+      "epoch": 0.2605633802816901,
+      "grad_norm": 1.9296875,
+      "learning_rate": 5e-05,
+      "loss": 1.370211124420166,
+      "mean_token_accuracy": 0.7603058218955994,
+      "num_tokens": 106824.0,
+      "step": 37
+    },
+    {
+      "entropy": 1.3046875,
+      "epoch": 0.2676056338028169,
+      "grad_norm": 1.9921875,
+      "learning_rate": 5e-05,
+      "loss": 1.160237431526184,
+      "mean_token_accuracy": 0.7812739610671997,
+      "num_tokens": 109438.0,
+      "step": 38
+    },
+    {
+      "entropy": 1.40625,
+      "epoch": 0.2746478873239437,
+      "grad_norm": 2.265625,
+      "learning_rate": 5e-05,
+      "loss": 1.3372321128845215,
+      "mean_token_accuracy": 0.7477554678916931,
+      "num_tokens": 111785.0,
+      "step": 39
+    },
+    {
+      "entropy": 1.3671875,
+      "epoch": 0.28169014084507044,
+      "grad_norm": 1.84375,
+      "learning_rate": 5e-05,
+      "loss": 1.3203842639923096,
+      "mean_token_accuracy": 0.7538297176361084,
+      "num_tokens": 114600.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.21875,
+      "epoch": 0.2887323943661972,
+      "grad_norm": 2.203125,
+      "learning_rate": 5e-05,
+      "loss": 1.1055949926376343,
+      "mean_token_accuracy": 0.7860545516014099,
+      "num_tokens": 116917.0,
+      "step": 41
+    },
+    {
+      "entropy": 1.484375,
+      "epoch": 0.29577464788732394,
+      "grad_norm": 1.546875,
+      "learning_rate": 5e-05,
+      "loss": 1.403846025466919,
+      "mean_token_accuracy": 0.7451103925704956,
+      "num_tokens": 120095.0,
+      "step": 42
+    },
+    {
+      "entropy": 1.328125,
+      "epoch": 0.3028169014084507,
+      "grad_norm": 1.65625,
+      "learning_rate": 5e-05,
+      "loss": 1.3071074485778809,
+      "mean_token_accuracy": 0.7589316964149475,
+      "num_tokens": 122986.0,
+      "step": 43
+    },
+    {
+      "entropy": 1.3203125,
+      "epoch": 0.30985915492957744,
+      "grad_norm": 1.8359375,
+      "learning_rate": 5e-05,
+      "loss": 1.2948389053344727,
+      "mean_token_accuracy": 0.7675830125808716,
+      "num_tokens": 125795.0,
+      "step": 44
+    },
+    {
+      "entropy": 1.3125,
+      "epoch": 0.31690140845070425,
+      "grad_norm": 1.875,
+      "learning_rate": 5e-05,
+      "loss": 1.2394049167633057,
+      "mean_token_accuracy": 0.7697368264198303,
+      "num_tokens": 128235.0,
+      "step": 45
+    },
+    {
+      "entropy": 1.3203125,
+      "epoch": 0.323943661971831,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5e-05,
+      "loss": 1.3117918968200684,
+      "mean_token_accuracy": 0.7587719559669495,
+      "num_tokens": 131663.0,
+      "step": 46
+    },
+    {
+      "entropy": 1.3515625,
+      "epoch": 0.33098591549295775,
+      "grad_norm": 1.453125,
+      "learning_rate": 5e-05,
+      "loss": 1.3100031614303589,
+      "mean_token_accuracy": 0.7717799544334412,
+      "num_tokens": 134598.0,
+      "step": 47
+    },
+    {
+      "entropy": 1.3203125,
+      "epoch": 0.3380281690140845,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5e-05,
+      "loss": 1.28191077709198,
+      "mean_token_accuracy": 0.7734912037849426,
+      "num_tokens": 137224.0,
+      "step": 48
+    },
+    {
+      "entropy": 1.3828125,
+      "epoch": 0.34507042253521125,
+      "grad_norm": 1.515625,
+      "learning_rate": 5e-05,
+      "loss": 1.3600674867630005,
+      "mean_token_accuracy": 0.7438323497772217,
+      "num_tokens": 140191.0,
+      "step": 49
+    },
+    {
+      "entropy": 1.21875,
+      "epoch": 0.352112676056338,
+      "grad_norm": 1.46875,
+      "learning_rate": 5e-05,
+      "loss": 1.1239631175994873,
+      "mean_token_accuracy": 0.788614809513092,
+      "num_tokens": 142834.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.4765625,
+      "epoch": 0.3591549295774648,
+      "grad_norm": 1.46875,
+      "learning_rate": 5e-05,
+      "loss": 1.3836950063705444,
+      "mean_token_accuracy": 0.7538405060768127,
+      "num_tokens": 145576.0,
+      "step": 51
+    },
+    {
+      "entropy": 1.1328125,
+      "epoch": 0.36619718309859156,
+      "grad_norm": 1.59375,
+      "learning_rate": 5e-05,
+      "loss": 1.1020089387893677,
+      "mean_token_accuracy": 0.8136998414993286,
+      "num_tokens": 147876.0,
+      "step": 52
+    },
+    {
+      "entropy": 1.3671875,
+      "epoch": 0.3732394366197183,
+      "grad_norm": 1.15625,
+      "learning_rate": 5e-05,
+      "loss": 1.3127013444900513,
+      "mean_token_accuracy": 0.7502166032791138,
+      "num_tokens": 151347.0,
+      "step": 53
+    },
+    {
+      "entropy": 1.0390625,
+      "epoch": 0.38028169014084506,
+      "grad_norm": 1.25,
+      "learning_rate": 5e-05,
+      "loss": 1.0258498191833496,
+      "mean_token_accuracy": 0.8231140971183777,
+      "num_tokens": 154046.0,
+      "step": 54
+    },
+    {
+      "entropy": 1.296875,
+      "epoch": 0.3873239436619718,
+      "grad_norm": 1.515625,
+      "learning_rate": 5e-05,
+      "loss": 1.26340913772583,
+      "mean_token_accuracy": 0.7880998253822327,
+      "num_tokens": 156659.0,
+      "step": 55
+    },
+    {
+      "entropy": 1.3828125,
+      "epoch": 0.39436619718309857,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5e-05,
+      "loss": 1.3492928743362427,
+      "mean_token_accuracy": 0.7417179942131042,
+      "num_tokens": 160078.0,
+      "step": 56
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 0.4014084507042254,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5e-05,
+      "loss": 1.0268410444259644,
+      "mean_token_accuracy": 0.8108739256858826,
+      "num_tokens": 163084.0,
+      "step": 57
+    },
+    {
+      "entropy": 1.2734375,
+      "epoch": 0.4084507042253521,
+      "grad_norm": 1.296875,
+      "learning_rate": 5e-05,
+      "loss": 1.218086838722229,
+      "mean_token_accuracy": 0.7687548398971558,
+      "num_tokens": 165678.0,
+      "step": 58
+    },
+    {
+      "entropy": 1.265625,
+      "epoch": 0.4154929577464789,
+      "grad_norm": 1.078125,
+      "learning_rate": 5e-05,
+      "loss": 1.2634363174438477,
+      "mean_token_accuracy": 0.7484593987464905,
+      "num_tokens": 169256.0,
+      "step": 59
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 0.4225352112676056,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5e-05,
+      "loss": 0.9637977480888367,
+      "mean_token_accuracy": 0.8200221061706543,
+      "num_tokens": 171981.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.1015625,
+      "epoch": 0.4295774647887324,
+      "grad_norm": 1.09375,
+      "learning_rate": 5e-05,
+      "loss": 1.0381879806518555,
+      "mean_token_accuracy": 0.8146411776542664,
+      "num_tokens": 174762.0,
+      "step": 61
+    },
+    {
+      "entropy": 1.4609375,
+      "epoch": 0.43661971830985913,
+      "grad_norm": 1.109375,
+      "learning_rate": 5e-05,
+      "loss": 1.442441701889038,
+      "mean_token_accuracy": 0.7236841917037964,
+      "num_tokens": 177962.0,
+      "step": 62
+    },
+    {
+      "entropy": 0.9453125,
+      "epoch": 0.44366197183098594,
+      "grad_norm": 1.109375,
+      "learning_rate": 5e-05,
+      "loss": 0.9214452505111694,
+      "mean_token_accuracy": 0.8193620443344116,
+      "num_tokens": 180666.0,
+      "step": 63
+    },
+    {
+      "entropy": 1.0859375,
+      "epoch": 0.4507042253521127,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5e-05,
+      "loss": 1.1097257137298584,
+      "mean_token_accuracy": 0.8031823039054871,
+      "num_tokens": 183565.0,
+      "step": 64
+    },
+    {
+      "entropy": 1.1640625,
+      "epoch": 0.45774647887323944,
+      "grad_norm": 1.140625,
+      "learning_rate": 5e-05,
+      "loss": 1.155249834060669,
+      "mean_token_accuracy": 0.7887681126594543,
+      "num_tokens": 186333.0,
+      "step": 65
+    },
+    {
+      "entropy": 0.85546875,
+      "epoch": 0.4647887323943662,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5e-05,
+      "loss": 0.8293672800064087,
+      "mean_token_accuracy": 0.8543021082878113,
+      "num_tokens": 188956.0,
+      "step": 66
+    },
+    {
+      "entropy": 0.91015625,
+      "epoch": 0.47183098591549294,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5e-05,
+      "loss": 0.894307553768158,
+      "mean_token_accuracy": 0.8422175049781799,
+      "num_tokens": 191309.0,
+      "step": 67
+    },
+    {
+      "entropy": 1.203125,
+      "epoch": 0.4788732394366197,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5e-05,
+      "loss": 1.1631497144699097,
+      "mean_token_accuracy": 0.7861995100975037,
+      "num_tokens": 194114.0,
+      "step": 68
+    },
+    {
+      "entropy": 1.328125,
+      "epoch": 0.4859154929577465,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5e-05,
+      "loss": 1.2809975147247314,
+      "mean_token_accuracy": 0.7543198466300964,
+      "num_tokens": 197305.0,
+      "step": 69
+    },
+    {
+      "entropy": 1.015625,
+      "epoch": 0.49295774647887325,
+      "grad_norm": 1.3203125,
+      "learning_rate": 5e-05,
+      "loss": 1.0155205726623535,
+      "mean_token_accuracy": 0.834074079990387,
+      "num_tokens": 199338.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.83984375,
+      "epoch": 0.5,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5e-05,
+      "loss": 0.7589200735092163,
+      "mean_token_accuracy": 0.8618420958518982,
+      "num_tokens": 201474.0,
+      "step": 71
+    },
+    {
+      "entropy": 1.0859375,
+      "epoch": 0.5070422535211268,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5e-05,
+      "loss": 1.0090736150741577,
+      "mean_token_accuracy": 0.8233413100242615,
+      "num_tokens": 203984.0,
+      "step": 72
+    },
+    {
+      "entropy": 1.0703125,
+      "epoch": 0.5140845070422535,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5e-05,
+      "loss": 1.046358346939087,
+      "mean_token_accuracy": 0.802097737789154,
+      "num_tokens": 207615.0,
+      "step": 73
+    },
+    {
+      "entropy": 1.203125,
+      "epoch": 0.5211267605633803,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5e-05,
+      "loss": 1.14384126663208,
+      "mean_token_accuracy": 0.7883082628250122,
+      "num_tokens": 209881.0,
+      "step": 74
+    },
+    {
+      "entropy": 1.015625,
+      "epoch": 0.528169014084507,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5e-05,
+      "loss": 0.9863561987876892,
+      "mean_token_accuracy": 0.8165745735168457,
+      "num_tokens": 212604.0,
+      "step": 75
+    },
+    {
+      "entropy": 1.2421875,
+      "epoch": 0.5352112676056338,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5e-05,
+      "loss": 1.1881471872329712,
+      "mean_token_accuracy": 0.7924298048019409,
+      "num_tokens": 215069.0,
+      "step": 76
+    },
+    {
+      "entropy": 0.99609375,
+      "epoch": 0.5422535211267606,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5e-05,
+      "loss": 0.9455610513687134,
+      "mean_token_accuracy": 0.8238698244094849,
+      "num_tokens": 217842.0,
+      "step": 77
+    },
+    {
+      "entropy": 0.93359375,
+      "epoch": 0.5492957746478874,
+      "grad_norm": 1.015625,
+      "learning_rate": 5e-05,
+      "loss": 0.8940377235412598,
+      "mean_token_accuracy": 0.8325259685516357,
+      "num_tokens": 220740.0,
+      "step": 78
+    },
+    {
+      "entropy": 1.078125,
+      "epoch": 0.5563380281690141,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5e-05,
+      "loss": 1.072181224822998,
+      "mean_token_accuracy": 0.7974197864532471,
+      "num_tokens": 223616.0,
+      "step": 79
+    },
+    {
+      "entropy": 0.9609375,
+      "epoch": 0.5633802816901409,
+      "grad_norm": 1.0625,
+      "learning_rate": 5e-05,
+      "loss": 0.968547523021698,
+      "mean_token_accuracy": 0.8254417181015015,
+      "num_tokens": 226454.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.0625,
+      "epoch": 0.5704225352112676,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5e-05,
+      "loss": 1.029323935508728,
+      "mean_token_accuracy": 0.8091970086097717,
+      "num_tokens": 229376.0,
+      "step": 81
+    },
+    {
+      "entropy": 1.296875,
+      "epoch": 0.5774647887323944,
+      "grad_norm": 1.078125,
+      "learning_rate": 5e-05,
+      "loss": 1.30299711227417,
+      "mean_token_accuracy": 0.7740401029586792,
+      "num_tokens": 232327.0,
+      "step": 82
+    },
+    {
+      "entropy": 1.34375,
+      "epoch": 0.5845070422535211,
+      "grad_norm": 1.046875,
+      "learning_rate": 5e-05,
+      "loss": 1.3241432905197144,
+      "mean_token_accuracy": 0.7471368312835693,
+      "num_tokens": 235653.0,
+      "step": 83
+    },
+    {
+      "entropy": 1.1953125,
+      "epoch": 0.5915492957746479,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5e-05,
+      "loss": 1.1841082572937012,
+      "mean_token_accuracy": 0.7741320133209229,
+      "num_tokens": 238282.0,
+      "step": 84
+    },
+    {
+      "entropy": 0.96875,
+      "epoch": 0.5985915492957746,
+      "grad_norm": 1.0,
+      "learning_rate": 5e-05,
+      "loss": 0.9051790833473206,
+      "mean_token_accuracy": 0.8358089327812195,
+      "num_tokens": 240781.0,
+      "step": 85
+    },
+    {
+      "entropy": 0.81640625,
+      "epoch": 0.6056338028169014,
+      "grad_norm": 1.0625,
+      "learning_rate": 5e-05,
+      "loss": 0.8058979511260986,
+      "mean_token_accuracy": 0.8656787872314453,
+      "num_tokens": 242881.0,
+      "step": 86
+    },
+    {
+      "entropy": 1.0234375,
+      "epoch": 0.6126760563380281,
+      "grad_norm": 0.89453125,
+      "learning_rate": 5e-05,
+      "loss": 0.9673476219177246,
+      "mean_token_accuracy": 0.8199172616004944,
+      "num_tokens": 246032.0,
+      "step": 87
+    },
+    {
+      "entropy": 0.9609375,
+      "epoch": 0.6197183098591549,
+      "grad_norm": 1.2265625,
+      "learning_rate": 5e-05,
+      "loss": 0.9231598377227783,
+      "mean_token_accuracy": 0.833532452583313,
+      "num_tokens": 248551.0,
+      "step": 88
+    },
+    {
+      "entropy": 0.9921875,
+      "epoch": 0.6267605633802817,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5e-05,
+      "loss": 0.998257040977478,
+      "mean_token_accuracy": 0.8266341686248779,
+      "num_tokens": 251022.0,
+      "step": 89
+    },
+    {
+      "entropy": 1.0546875,
+      "epoch": 0.6338028169014085,
+      "grad_norm": 1.015625,
+      "learning_rate": 5e-05,
+      "loss": 1.0252176523208618,
+      "mean_token_accuracy": 0.8216797709465027,
+      "num_tokens": 253447.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.1484375,
+      "epoch": 0.6408450704225352,
+      "grad_norm": 1.0,
+      "learning_rate": 5e-05,
+      "loss": 1.164228081703186,
+      "mean_token_accuracy": 0.7702828645706177,
+      "num_tokens": 256672.0,
+      "step": 91
+    },
+    {
+      "entropy": 0.77734375,
+      "epoch": 0.647887323943662,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5e-05,
+      "loss": 0.7661694288253784,
+      "mean_token_accuracy": 0.8448406457901001,
+      "num_tokens": 259316.0,
+      "step": 92
+    },
+    {
+      "entropy": 1.15625,
+      "epoch": 0.6549295774647887,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5e-05,
+      "loss": 1.1309537887573242,
+      "mean_token_accuracy": 0.7973990440368652,
+      "num_tokens": 262246.0,
+      "step": 93
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 0.6619718309859155,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5e-05,
+      "loss": 1.0406370162963867,
+      "mean_token_accuracy": 0.8099964261054993,
+      "num_tokens": 265075.0,
+      "step": 94
+    },
+    {
+      "entropy": 1.0,
+      "epoch": 0.6690140845070423,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 0.9478747844696045,
+      "mean_token_accuracy": 0.8167044520378113,
+      "num_tokens": 267729.0,
+      "step": 95
+    },
+    {
+      "entropy": 1.125,
+      "epoch": 0.676056338028169,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5e-05,
+      "loss": 1.0806983709335327,
+      "mean_token_accuracy": 0.8070116639137268,
+      "num_tokens": 270732.0,
+      "step": 96
+    },
+    {
+      "entropy": 1.3359375,
+      "epoch": 0.6830985915492958,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5e-05,
+      "loss": 1.2827727794647217,
+      "mean_token_accuracy": 0.7681026458740234,
+      "num_tokens": 274013.0,
+      "step": 97
+    },
+    {
+      "entropy": 0.96875,
+      "epoch": 0.6901408450704225,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5e-05,
+      "loss": 0.9255345463752747,
+      "mean_token_accuracy": 0.820858895778656,
+      "num_tokens": 276466.0,
+      "step": 98
+    },
+    {
+      "entropy": 0.8828125,
+      "epoch": 0.6971830985915493,
+      "grad_norm": 0.828125,
+      "learning_rate": 5e-05,
+      "loss": 0.8368520736694336,
+      "mean_token_accuracy": 0.8478841185569763,
+      "num_tokens": 279097.0,
+      "step": 99
+    },
+    {
+      "entropy": 1.1875,
+      "epoch": 0.704225352112676,
+      "grad_norm": 0.91796875,
+      "learning_rate": 5e-05,
+      "loss": 1.1672570705413818,
+      "mean_token_accuracy": 0.7689912915229797,
+      "num_tokens": 282317.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.53125,
+      "epoch": 0.7112676056338029,
+      "grad_norm": 1.0625,
+      "learning_rate": 5e-05,
+      "loss": 1.4877268075942993,
+      "mean_token_accuracy": 0.7194092869758606,
+      "num_tokens": 285643.0,
+      "step": 101
+    },
+    {
+      "entropy": 1.0,
+      "epoch": 0.7183098591549296,
+      "grad_norm": 1.0625,
+      "learning_rate": 5e-05,
+      "loss": 0.956112265586853,
+      "mean_token_accuracy": 0.8256269693374634,
+      "num_tokens": 288203.0,
+      "step": 102
+    },
+    {
+      "entropy": 1.1953125,
+      "epoch": 0.7253521126760564,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5e-05,
+      "loss": 1.2514692544937134,
+      "mean_token_accuracy": 0.771497905254364,
+      "num_tokens": 291095.0,
+      "step": 103
+    },
+    {
+      "entropy": 1.2109375,
+      "epoch": 0.7323943661971831,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5e-05,
+      "loss": 1.1472859382629395,
+      "mean_token_accuracy": 0.7850604057312012,
+      "num_tokens": 293834.0,
+      "step": 104
+    },
+    {
+      "entropy": 0.7734375,
+      "epoch": 0.7394366197183099,
+      "grad_norm": 1.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.7254777550697327,
+      "mean_token_accuracy": 0.8632119297981262,
+      "num_tokens": 295984.0,
+      "step": 105
+    },
+    {
+      "entropy": 0.82421875,
+      "epoch": 0.7464788732394366,
+      "grad_norm": 1.078125,
+      "learning_rate": 5e-05,
+      "loss": 0.795278787612915,
+      "mean_token_accuracy": 0.8470132946968079,
+      "num_tokens": 298319.0,
+      "step": 106
+    },
+    {
+      "entropy": 1.09375,
+      "epoch": 0.7535211267605634,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 1.0427790880203247,
+      "mean_token_accuracy": 0.804815411567688,
+      "num_tokens": 301442.0,
+      "step": 107
+    },
+    {
+      "entropy": 1.1171875,
+      "epoch": 0.7605633802816901,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5e-05,
+      "loss": 1.104764699935913,
+      "mean_token_accuracy": 0.8097593188285828,
+      "num_tokens": 304483.0,
+      "step": 108
+    },
+    {
+      "entropy": 0.7421875,
+      "epoch": 0.7676056338028169,
+      "grad_norm": 1.0,
+      "learning_rate": 5e-05,
+      "loss": 0.6636262536048889,
+      "mean_token_accuracy": 0.861122190952301,
+      "num_tokens": 307004.0,
+      "step": 109
+    },
+    {
+      "entropy": 0.97265625,
+      "epoch": 0.7746478873239436,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 0.9813203811645508,
+      "mean_token_accuracy": 0.8114525079727173,
+      "num_tokens": 309876.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.09375,
+      "epoch": 0.7816901408450704,
+      "grad_norm": 1.078125,
+      "learning_rate": 5e-05,
+      "loss": 1.0708268880844116,
+      "mean_token_accuracy": 0.7995486855506897,
+      "num_tokens": 312543.0,
+      "step": 111
+    },
+    {
+      "entropy": 0.953125,
+      "epoch": 0.7887323943661971,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5e-05,
+      "loss": 0.9298059344291687,
+      "mean_token_accuracy": 0.8266475796699524,
+      "num_tokens": 315343.0,
+      "step": 112
+    },
+    {
+      "entropy": 0.93359375,
+      "epoch": 0.795774647887324,
+      "grad_norm": 1.2109375,
+      "learning_rate": 5e-05,
+      "loss": 0.9146865606307983,
+      "mean_token_accuracy": 0.8447837233543396,
+      "num_tokens": 317709.0,
+      "step": 113
+    },
+    {
+      "entropy": 1.0390625,
+      "epoch": 0.8028169014084507,
+      "grad_norm": 1.015625,
+      "learning_rate": 5e-05,
+      "loss": 1.0023632049560547,
+      "mean_token_accuracy": 0.8247303366661072,
+      "num_tokens": 320313.0,
+      "step": 114
+    },
+    {
+      "entropy": 1.0234375,
+      "epoch": 0.8098591549295775,
+      "grad_norm": 0.9375,
+      "learning_rate": 5e-05,
+      "loss": 1.0119502544403076,
+      "mean_token_accuracy": 0.8081619739532471,
+      "num_tokens": 323433.0,
+      "step": 115
+    },
+    {
+      "entropy": 1.1484375,
+      "epoch": 0.8169014084507042,
+      "grad_norm": 1.09375,
+      "learning_rate": 5e-05,
+      "loss": 1.129537582397461,
+      "mean_token_accuracy": 0.7915045619010925,
+      "num_tokens": 326525.0,
+      "step": 116
+    },
+    {
+      "entropy": 0.828125,
+      "epoch": 0.823943661971831,
+      "grad_norm": 0.96875,
+      "learning_rate": 5e-05,
+      "loss": 0.8433165550231934,
+      "mean_token_accuracy": 0.8409850001335144,
+      "num_tokens": 328929.0,
+      "step": 117
+    },
+    {
+      "entropy": 1.0625,
+      "epoch": 0.8309859154929577,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5e-05,
+      "loss": 1.0346664190292358,
+      "mean_token_accuracy": 0.7833219170570374,
+      "num_tokens": 331863.0,
+      "step": 118
+    },
+    {
+      "entropy": 0.98046875,
+      "epoch": 0.8380281690140845,
+      "grad_norm": 1.09375,
+      "learning_rate": 5e-05,
+      "loss": 0.950432300567627,
+      "mean_token_accuracy": 0.8241989612579346,
+      "num_tokens": 334243.0,
+      "step": 119
+    },
+    {
+      "entropy": 1.0546875,
+      "epoch": 0.8450704225352113,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 1.0502727031707764,
+      "mean_token_accuracy": 0.7917791604995728,
+      "num_tokens": 337584.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.89453125,
+      "epoch": 0.852112676056338,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5e-05,
+      "loss": 0.9121181964874268,
+      "mean_token_accuracy": 0.8309713006019592,
+      "num_tokens": 340485.0,
+      "step": 121
+    },
+    {
+      "entropy": 0.88671875,
+      "epoch": 0.8591549295774648,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5e-05,
+      "loss": 0.9219837188720703,
+      "mean_token_accuracy": 0.8140540719032288,
+      "num_tokens": 343268.0,
+      "step": 122
+    },
+    {
+      "entropy": 0.8515625,
+      "epoch": 0.8661971830985915,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5e-05,
+      "loss": 0.7965174317359924,
+      "mean_token_accuracy": 0.8607358932495117,
+      "num_tokens": 345423.0,
+      "step": 123
+    },
+    {
+      "entropy": 0.9765625,
+      "epoch": 0.8732394366197183,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5e-05,
+      "loss": 0.9578317999839783,
+      "mean_token_accuracy": 0.8050633072853088,
+      "num_tokens": 348591.0,
+      "step": 124
+    },
+    {
+      "entropy": 0.953125,
+      "epoch": 0.8802816901408451,
+      "grad_norm": 0.890625,
+      "learning_rate": 5e-05,
+      "loss": 0.9342446327209473,
+      "mean_token_accuracy": 0.8261499404907227,
+      "num_tokens": 351360.0,
+      "step": 125
+    },
+    {
+      "entropy": 1.0859375,
+      "epoch": 0.8873239436619719,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5e-05,
+      "loss": 1.0417184829711914,
+      "mean_token_accuracy": 0.7993496060371399,
+      "num_tokens": 354443.0,
+      "step": 126
+    },
+    {
+      "entropy": 1.1875,
+      "epoch": 0.8943661971830986,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5e-05,
+      "loss": 1.169641137123108,
+      "mean_token_accuracy": 0.789621889591217,
+      "num_tokens": 356937.0,
+      "step": 127
+    },
+    {
+      "entropy": 1.171875,
+      "epoch": 0.9014084507042254,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5e-05,
+      "loss": 1.0926804542541504,
+      "mean_token_accuracy": 0.78143709897995,
+      "num_tokens": 359951.0,
+      "step": 128
+    },
+    {
+      "entropy": 0.90625,
+      "epoch": 0.9084507042253521,
+      "grad_norm": 1.046875,
+      "learning_rate": 5e-05,
+      "loss": 0.8831167817115784,
+      "mean_token_accuracy": 0.8297297358512878,
+      "num_tokens": 362549.0,
+      "step": 129
+    },
+    {
+      "entropy": 1.34375,
+      "epoch": 0.9154929577464789,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5e-05,
+      "loss": 1.2982932329177856,
+      "mean_token_accuracy": 0.7599736452102661,
+      "num_tokens": 365590.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.8515625,
+      "epoch": 0.9225352112676056,
+      "grad_norm": 1.140625,
+      "learning_rate": 5e-05,
+      "loss": 0.7926298975944519,
+      "mean_token_accuracy": 0.8279292583465576,
+      "num_tokens": 368312.0,
+      "step": 131
+    },
+    {
+      "entropy": 0.85546875,
+      "epoch": 0.9295774647887324,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5e-05,
+      "loss": 0.876966655254364,
+      "mean_token_accuracy": 0.8338558077812195,
+      "num_tokens": 370872.0,
+      "step": 132
+    },
+    {
+      "entropy": 0.58203125,
+      "epoch": 0.9366197183098591,
+      "grad_norm": 1.09375,
+      "learning_rate": 5e-05,
+      "loss": 0.5232805609703064,
+      "mean_token_accuracy": 0.9088762998580933,
+      "num_tokens": 372998.0,
+      "step": 133
+    },
+    {
+      "entropy": 1.3125,
+      "epoch": 0.9436619718309859,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5e-05,
+      "loss": 1.2858319282531738,
+      "mean_token_accuracy": 0.764127790927887,
+      "num_tokens": 375855.0,
+      "step": 134
+    },
+    {
+      "entropy": 0.9453125,
+      "epoch": 0.9507042253521126,
+      "grad_norm": 0.8125,
+      "learning_rate": 5e-05,
+      "loss": 0.9115528464317322,
+      "mean_token_accuracy": 0.8201290965080261,
+      "num_tokens": 379271.0,
+      "step": 135
+    },
+    {
+      "entropy": 1.0703125,
+      "epoch": 0.9577464788732394,
+      "grad_norm": 1.3828125,
+      "learning_rate": 5e-05,
+      "loss": 1.03813898563385,
+      "mean_token_accuracy": 0.8119325637817383,
+      "num_tokens": 381592.0,
+      "step": 136
+    },
+    {
+      "entropy": 1.109375,
+      "epoch": 0.9647887323943662,
+      "grad_norm": 0.953125,
+      "learning_rate": 5e-05,
+      "loss": 1.1146811246871948,
+      "mean_token_accuracy": 0.790168285369873,
+      "num_tokens": 384631.0,
+      "step": 137
+    },
+    {
+      "entropy": 1.1640625,
+      "epoch": 0.971830985915493,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5e-05,
+      "loss": 1.173197627067566,
+      "mean_token_accuracy": 0.7781087160110474,
+      "num_tokens": 387325.0,
+      "step": 138
+    },
+    {
+      "entropy": 0.828125,
+      "epoch": 0.9788732394366197,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5e-05,
+      "loss": 0.8895862698554993,
+      "mean_token_accuracy": 0.8376736044883728,
+      "num_tokens": 389637.0,
+      "step": 139
+    },
+    {
+      "entropy": 1.1953125,
+      "epoch": 0.9859154929577465,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 1.2043834924697876,
+      "mean_token_accuracy": 0.7506383061408997,
+      "num_tokens": 393170.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.87109375,
+      "epoch": 0.9929577464788732,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5e-05,
+      "loss": 0.8573941588401794,
+      "mean_token_accuracy": 0.8464516401290894,
+      "num_tokens": 395503.0,
+      "step": 141
+    },
+    {
+      "entropy": 1.078125,
+      "epoch": 1.0,
+      "grad_norm": 0.90625,
+      "learning_rate": 5e-05,
+      "loss": 1.0620708465576172,
+      "mean_token_accuracy": 0.7814034819602966,
+      "num_tokens": 398361.0,
+      "step": 142
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9779188368055556,
+      "eval_loss": 0.9645262956619263,
+      "eval_mean_token_accuracy": 0.8152486599153943,
+      "eval_num_tokens": 398361.0,
+      "eval_runtime": 3.7052,
+      "eval_samples_per_second": 76.65,
+      "eval_steps_per_second": 9.716,
+      "step": 142
+    },
+    {
+      "entropy": 0.953125,
+      "epoch": 1.0070422535211268,
+      "grad_norm": 1.125,
+      "learning_rate": 5e-05,
+      "loss": 0.9150600433349609,
+      "mean_token_accuracy": 0.8252961039543152,
+      "num_tokens": 400733.0,
+      "step": 143
+    },
+    {
+      "entropy": 0.8046875,
+      "epoch": 1.0140845070422535,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5e-05,
+      "loss": 0.7507684230804443,
+      "mean_token_accuracy": 0.8355586528778076,
+      "num_tokens": 403587.0,
+      "step": 144
+    },
+    {
+      "entropy": 0.7265625,
+      "epoch": 1.0211267605633803,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5e-05,
+      "loss": 0.7104012370109558,
+      "mean_token_accuracy": 0.8799096941947937,
+      "num_tokens": 405810.0,
+      "step": 145
+    },
+    {
+      "entropy": 0.8515625,
+      "epoch": 1.028169014084507,
+      "grad_norm": 0.8671875,
+      "learning_rate": 5e-05,
+      "loss": 0.7941129207611084,
+      "mean_token_accuracy": 0.8590818643569946,
+      "num_tokens": 408323.0,
+      "step": 146
+    },
+    {
+      "entropy": 1.140625,
+      "epoch": 1.0352112676056338,
+      "grad_norm": 0.90625,
+      "learning_rate": 5e-05,
+      "loss": 1.110805630683899,
+      "mean_token_accuracy": 0.7991207242012024,
+      "num_tokens": 411288.0,
+      "step": 147
+    },
+    {
+      "entropy": 1.078125,
+      "epoch": 1.0422535211267605,
+      "grad_norm": 0.875,
+      "learning_rate": 5e-05,
+      "loss": 1.096816897392273,
+      "mean_token_accuracy": 0.793763279914856,
+      "num_tokens": 414118.0,
+      "step": 148
+    },
+    {
+      "entropy": 0.79296875,
+      "epoch": 1.0492957746478873,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5e-05,
+      "loss": 0.7629069089889526,
+      "mean_token_accuracy": 0.8661789894104004,
+      "num_tokens": 416450.0,
+      "step": 149
+    },
+    {
+      "entropy": 0.78125,
+      "epoch": 1.056338028169014,
+      "grad_norm": 1.03125,
+      "learning_rate": 5e-05,
+      "loss": 0.7574359178543091,
+      "mean_token_accuracy": 0.8526899814605713,
+      "num_tokens": 418800.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.71484375,
+      "epoch": 1.0633802816901408,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5e-05,
+      "loss": 0.6598190069198608,
+      "mean_token_accuracy": 0.8710424900054932,
+      "num_tokens": 421398.0,
+      "step": 151
+    },
+    {
+      "entropy": 0.96484375,
+      "epoch": 1.0704225352112675,
+      "grad_norm": 1.078125,
+      "learning_rate": 5e-05,
+      "loss": 0.9846295118331909,
+      "mean_token_accuracy": 0.8020523190498352,
+      "num_tokens": 424427.0,
+      "step": 152
+    },
+    {
+      "entropy": 0.9609375,
+      "epoch": 1.0774647887323943,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5e-05,
+      "loss": 0.982389509677887,
+      "mean_token_accuracy": 0.8070852160453796,
+      "num_tokens": 427286.0,
+      "step": 153
+    },
+    {
+      "entropy": 0.7734375,
+      "epoch": 1.084507042253521,
+      "grad_norm": 0.92578125,
+      "learning_rate": 5e-05,
+      "loss": 0.7500929236412048,
+      "mean_token_accuracy": 0.8648427724838257,
+      "num_tokens": 429743.0,
+      "step": 154
+    },
+    {
+      "entropy": 0.578125,
+      "epoch": 1.091549295774648,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5e-05,
+      "loss": 0.5609915256500244,
+      "mean_token_accuracy": 0.9083379507064819,
+      "num_tokens": 431562.0,
+      "step": 155
+    },
+    {
+      "entropy": 0.84375,
+      "epoch": 1.0985915492957747,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5e-05,
+      "loss": 0.8129755854606628,
+      "mean_token_accuracy": 0.83039790391922,
+      "num_tokens": 434636.0,
+      "step": 156
+    },
+    {
+      "entropy": 1.1640625,
+      "epoch": 1.1056338028169015,
+      "grad_norm": 1.0625,
+      "learning_rate": 5e-05,
+      "loss": 1.157644271850586,
+      "mean_token_accuracy": 0.7780413627624512,
+      "num_tokens": 437595.0,
+      "step": 157
+    },
+    {
+      "entropy": 0.87890625,
+      "epoch": 1.1126760563380282,
+      "grad_norm": 1.046875,
+      "learning_rate": 5e-05,
+      "loss": 0.865077793598175,
+      "mean_token_accuracy": 0.8418839573860168,
+      "num_tokens": 439981.0,
+      "step": 158
+    },
+    {
+      "entropy": 1.078125,
+      "epoch": 1.119718309859155,
+      "grad_norm": 1.03125,
+      "learning_rate": 5e-05,
+      "loss": 1.0671403408050537,
+      "mean_token_accuracy": 0.7895287871360779,
+      "num_tokens": 442854.0,
+      "step": 159
+    },
+    {
+      "entropy": 0.6875,
+      "epoch": 1.1267605633802817,
+      "grad_norm": 0.91796875,
+      "learning_rate": 5e-05,
+      "loss": 0.6512315273284912,
+      "mean_token_accuracy": 0.8766743540763855,
+      "num_tokens": 445027.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.125,
+      "epoch": 1.1338028169014085,
+      "grad_norm": 1.03125,
+      "learning_rate": 5e-05,
+      "loss": 1.1221463680267334,
+      "mean_token_accuracy": 0.7976190447807312,
+      "num_tokens": 447891.0,
+      "step": 161
+    },
+    {
+      "entropy": 0.90625,
+      "epoch": 1.1408450704225352,
+      "grad_norm": 0.90625,
+      "learning_rate": 5e-05,
+      "loss": 0.8583451509475708,
+      "mean_token_accuracy": 0.8390721082687378,
+      "num_tokens": 450658.0,
+      "step": 162
+    },
+    {
+      "entropy": 1.0078125,
+      "epoch": 1.147887323943662,
+      "grad_norm": 0.890625,
+      "learning_rate": 5e-05,
+      "loss": 1.0121499300003052,
+      "mean_token_accuracy": 0.8126649260520935,
+      "num_tokens": 454077.0,
+      "step": 163
+    },
+    {
+      "entropy": 1.109375,
+      "epoch": 1.1549295774647887,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 1.0865602493286133,
+      "mean_token_accuracy": 0.8106972575187683,
+      "num_tokens": 457656.0,
+      "step": 164
+    },
+    {
+      "entropy": 0.83984375,
+      "epoch": 1.1619718309859155,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5e-05,
+      "loss": 0.784701406955719,
+      "mean_token_accuracy": 0.8401400446891785,
+      "num_tokens": 460235.0,
+      "step": 165
+    },
+    {
+      "entropy": 0.87890625,
+      "epoch": 1.1690140845070423,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 0.8821406364440918,
+      "mean_token_accuracy": 0.8279282450675964,
+      "num_tokens": 462864.0,
+      "step": 166
+    },
+    {
+      "entropy": 1.09375,
+      "epoch": 1.176056338028169,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5e-05,
+      "loss": 1.071103572845459,
+      "mean_token_accuracy": 0.7772451043128967,
+      "num_tokens": 466001.0,
+      "step": 167
+    },
+    {
+      "entropy": 1.265625,
+      "epoch": 1.1830985915492958,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5e-05,
+      "loss": 1.2070709466934204,
+      "mean_token_accuracy": 0.7771183848381042,
+      "num_tokens": 469042.0,
+      "step": 168
+    },
+    {
+      "entropy": 1.03125,
+      "epoch": 1.1901408450704225,
+      "grad_norm": 0.84375,
+      "learning_rate": 5e-05,
+      "loss": 0.9974339604377747,
+      "mean_token_accuracy": 0.8080986142158508,
+      "num_tokens": 471890.0,
+      "step": 169
+    },
+    {
+      "entropy": 0.90625,
+      "epoch": 1.1971830985915493,
+      "grad_norm": 0.859375,
+      "learning_rate": 5e-05,
+      "loss": 0.8776722550392151,
+      "mean_token_accuracy": 0.836818516254425,
+      "num_tokens": 474576.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.96484375,
+      "epoch": 1.204225352112676,
+      "grad_norm": 0.90625,
+      "learning_rate": 5e-05,
+      "loss": 0.9560971856117249,
+      "mean_token_accuracy": 0.8277040123939514,
+      "num_tokens": 477219.0,
+      "step": 171
+    },
+    {
+      "entropy": 1.1328125,
+      "epoch": 1.2112676056338028,
+      "grad_norm": 0.9375,
+      "learning_rate": 5e-05,
+      "loss": 1.1175223588943481,
+      "mean_token_accuracy": 0.7786499261856079,
+      "num_tokens": 480412.0,
+      "step": 172
+    },
+    {
+      "entropy": 1.3984375,
+      "epoch": 1.2183098591549295,
+      "grad_norm": 1.109375,
+      "learning_rate": 5e-05,
+      "loss": 1.3395131826400757,
+      "mean_token_accuracy": 0.751716673374176,
+      "num_tokens": 483187.0,
+      "step": 173
+    },
+    {
+      "entropy": 0.9765625,
+      "epoch": 1.2253521126760563,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 1.004224419593811,
+      "mean_token_accuracy": 0.8214024901390076,
+      "num_tokens": 485933.0,
+      "step": 174
+    },
+    {
+      "entropy": 0.890625,
+      "epoch": 1.232394366197183,
+      "grad_norm": 0.7734375,
+      "learning_rate": 5e-05,
+      "loss": 0.8696950674057007,
+      "mean_token_accuracy": 0.8322895169258118,
+      "num_tokens": 488815.0,
+      "step": 175
+    },
+    {
+      "entropy": 1.109375,
+      "epoch": 1.2394366197183098,
+      "grad_norm": 0.91015625,
+      "learning_rate": 5e-05,
+      "loss": 1.0984435081481934,
+      "mean_token_accuracy": 0.7843074202537537,
+      "num_tokens": 491920.0,
+      "step": 176
+    },
+    {
+      "entropy": 0.95703125,
+      "epoch": 1.2464788732394365,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5e-05,
+      "loss": 0.9243696331977844,
+      "mean_token_accuracy": 0.8328244090080261,
+      "num_tokens": 494548.0,
+      "step": 177
+    },
+    {
+      "entropy": 1.1171875,
+      "epoch": 1.2535211267605635,
+      "grad_norm": 1.015625,
+      "learning_rate": 5e-05,
+      "loss": 1.0856032371520996,
+      "mean_token_accuracy": 0.7994632124900818,
+      "num_tokens": 497164.0,
+      "step": 178
+    },
+    {
+      "entropy": 0.8046875,
+      "epoch": 1.26056338028169,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5e-05,
+      "loss": 0.7936550378799438,
+      "mean_token_accuracy": 0.8561053276062012,
+      "num_tokens": 499375.0,
+      "step": 179
+    },
+    {
+      "entropy": 1.3125,
+      "epoch": 1.267605633802817,
+      "grad_norm": 0.94921875,
+      "learning_rate": 5e-05,
+      "loss": 1.3021771907806396,
+      "mean_token_accuracy": 0.7595491409301758,
+      "num_tokens": 502577.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.83203125,
+      "epoch": 1.2746478873239437,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5e-05,
+      "loss": 0.8171048760414124,
+      "mean_token_accuracy": 0.8462969660758972,
+      "num_tokens": 505272.0,
+      "step": 181
+    },
+    {
+      "entropy": 0.55078125,
+      "epoch": 1.2816901408450705,
+      "grad_norm": 1.15625,
+      "learning_rate": 5e-05,
+      "loss": 0.536847710609436,
+      "mean_token_accuracy": 0.9226457476615906,
+      "num_tokens": 507064.0,
+      "step": 182
+    },
+    {
+      "entropy": 1.0078125,
+      "epoch": 1.2887323943661972,
+      "grad_norm": 0.875,
+      "learning_rate": 5e-05,
+      "loss": 1.0221246480941772,
+      "mean_token_accuracy": 0.8086583018302917,
+      "num_tokens": 510098.0,
+      "step": 183
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 1.295774647887324,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5e-05,
+      "loss": 1.0499778985977173,
+      "mean_token_accuracy": 0.7971059083938599,
+      "num_tokens": 513354.0,
+      "step": 184
+    },
+    {
+      "entropy": 0.91796875,
+      "epoch": 1.3028169014084507,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5e-05,
+      "loss": 0.908235490322113,
+      "mean_token_accuracy": 0.8296475410461426,
+      "num_tokens": 516256.0,
+      "step": 185
+    },
+    {
+      "entropy": 1.0078125,
+      "epoch": 1.3098591549295775,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 0.9736582040786743,
+      "mean_token_accuracy": 0.8088183403015137,
+      "num_tokens": 519099.0,
+      "step": 186
+    },
+    {
+      "entropy": 1.03125,
+      "epoch": 1.3169014084507042,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5e-05,
+      "loss": 0.9836978316307068,
+      "mean_token_accuracy": 0.8049733638763428,
+      "num_tokens": 521922.0,
+      "step": 187
+    },
+    {
+      "entropy": 1.2265625,
+      "epoch": 1.323943661971831,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5e-05,
+      "loss": 1.236617088317871,
+      "mean_token_accuracy": 0.7495366930961609,
+      "num_tokens": 525707.0,
+      "step": 188
+    },
+    {
+      "entropy": 0.7265625,
+      "epoch": 1.3309859154929577,
+      "grad_norm": 0.78515625,
+      "learning_rate": 5e-05,
+      "loss": 0.6896946430206299,
+      "mean_token_accuracy": 0.8708636164665222,
+      "num_tokens": 528193.0,
+      "step": 189
+    },
+    {
+      "entropy": 0.7890625,
+      "epoch": 1.3380281690140845,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 0.7507163286209106,
+      "mean_token_accuracy": 0.852664589881897,
+      "num_tokens": 530753.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1328125,
+      "epoch": 1.3450704225352113,
+      "grad_norm": 0.859375,
+      "learning_rate": 5e-05,
+      "loss": 1.1465418338775635,
+      "mean_token_accuracy": 0.7830012440681458,
+      "num_tokens": 533973.0,
+      "step": 191
+    },
+    {
+      "entropy": 0.94140625,
+      "epoch": 1.352112676056338,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5e-05,
+      "loss": 0.9713115692138672,
+      "mean_token_accuracy": 0.8168444037437439,
+      "num_tokens": 536700.0,
+      "step": 192
+    },
+    {
+      "entropy": 0.98046875,
+      "epoch": 1.3591549295774648,
+      "grad_norm": 0.91796875,
+      "learning_rate": 5e-05,
+      "loss": 0.9194943904876709,
+      "mean_token_accuracy": 0.812965989112854,
+      "num_tokens": 539793.0,
+      "step": 193
+    },
+    {
+      "entropy": 1.0234375,
+      "epoch": 1.3661971830985915,
+      "grad_norm": 0.83203125,
+      "learning_rate": 5e-05,
+      "loss": 0.9923701286315918,
+      "mean_token_accuracy": 0.8061100840568542,
+      "num_tokens": 543107.0,
+      "step": 194
+    },
+    {
+      "entropy": 0.77734375,
+      "epoch": 1.3732394366197183,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 0.7561295032501221,
+      "mean_token_accuracy": 0.8627303242683411,
+      "num_tokens": 545774.0,
+      "step": 195
+    },
+    {
+      "entropy": 0.9765625,
+      "epoch": 1.380281690140845,
+      "grad_norm": 0.94921875,
+      "learning_rate": 5e-05,
+      "loss": 0.9733573198318481,
+      "mean_token_accuracy": 0.812156617641449,
+      "num_tokens": 548694.0,
+      "step": 196
+    },
+    {
+      "entropy": 0.8203125,
+      "epoch": 1.3873239436619718,
+      "grad_norm": 0.921875,
+      "learning_rate": 5e-05,
+      "loss": 0.7888185977935791,
+      "mean_token_accuracy": 0.8413514494895935,
+      "num_tokens": 551425.0,
+      "step": 197
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 1.3943661971830985,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5e-05,
+      "loss": 1.1070986986160278,
+      "mean_token_accuracy": 0.8009437918663025,
+      "num_tokens": 553764.0,
+      "step": 198
+    },
+    {
+      "entropy": 1.0,
+      "epoch": 1.4014084507042255,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5e-05,
+      "loss": 0.9817038178443909,
+      "mean_token_accuracy": 0.8074533939361572,
+      "num_tokens": 556348.0,
+      "step": 199
+    },
+    {
+      "entropy": 0.90234375,
+      "epoch": 1.408450704225352,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 0.9156808853149414,
+      "mean_token_accuracy": 0.8265197277069092,
+      "num_tokens": 559169.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.95703125,
+      "epoch": 1.415492957746479,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 0.9911283254623413,
+      "mean_token_accuracy": 0.8143333196640015,
+      "num_tokens": 562177.0,
+      "step": 201
+    },
+    {
+      "entropy": 0.93359375,
+      "epoch": 1.4225352112676055,
+      "grad_norm": 0.85546875,
+      "learning_rate": 5e-05,
+      "loss": 0.9385692477226257,
+      "mean_token_accuracy": 0.8237070441246033,
+      "num_tokens": 564834.0,
+      "step": 202
+    },
+    {
+      "entropy": 0.98828125,
+      "epoch": 1.4295774647887325,
+      "grad_norm": 0.890625,
+      "learning_rate": 5e-05,
+      "loss": 0.9537026882171631,
+      "mean_token_accuracy": 0.8196093440055847,
+      "num_tokens": 567453.0,
+      "step": 203
+    },
+    {
+      "entropy": 1.1171875,
+      "epoch": 1.436619718309859,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5e-05,
+      "loss": 1.1185535192489624,
+      "mean_token_accuracy": 0.7943561673164368,
+      "num_tokens": 570544.0,
+      "step": 204
+    },
+    {
+      "entropy": 1.2265625,
+      "epoch": 1.443661971830986,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 1.2277859449386597,
+      "mean_token_accuracy": 0.7642069458961487,
+      "num_tokens": 574089.0,
+      "step": 205
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 1.4507042253521127,
+      "grad_norm": 0.92578125,
+      "learning_rate": 5e-05,
+      "loss": 1.0433013439178467,
+      "mean_token_accuracy": 0.7854348421096802,
+      "num_tokens": 577420.0,
+      "step": 206
+    },
+    {
+      "entropy": 0.8515625,
+      "epoch": 1.4577464788732395,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5e-05,
+      "loss": 0.8301782011985779,
+      "mean_token_accuracy": 0.8421682119369507,
+      "num_tokens": 579937.0,
+      "step": 207
+    },
+    {
+      "entropy": 0.78125,
+      "epoch": 1.4647887323943662,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 0.7898612022399902,
+      "mean_token_accuracy": 0.8362168669700623,
+      "num_tokens": 582546.0,
+      "step": 208
+    },
+    {
+      "entropy": 0.984375,
+      "epoch": 1.471830985915493,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5e-05,
+      "loss": 0.9981078505516052,
+      "mean_token_accuracy": 0.8020643591880798,
+      "num_tokens": 585848.0,
+      "step": 209
+    },
+    {
+      "entropy": 0.94140625,
+      "epoch": 1.4788732394366197,
+      "grad_norm": 0.91796875,
+      "learning_rate": 5e-05,
+      "loss": 0.938367486000061,
+      "mean_token_accuracy": 0.8073078393936157,
+      "num_tokens": 588757.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.96484375,
+      "epoch": 1.4859154929577465,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5e-05,
+      "loss": 0.9669017791748047,
+      "mean_token_accuracy": 0.8119603395462036,
+      "num_tokens": 591892.0,
+      "step": 211
+    },
+    {
+      "entropy": 1.390625,
+      "epoch": 1.4929577464788732,
+      "grad_norm": 0.875,
+      "learning_rate": 5e-05,
+      "loss": 1.3462798595428467,
+      "mean_token_accuracy": 0.7334238886833191,
+      "num_tokens": 595580.0,
+      "step": 212
+    },
+    {
+      "entropy": 1.109375,
+      "epoch": 1.5,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5e-05,
+      "loss": 1.068884015083313,
+      "mean_token_accuracy": 0.7929913401603699,
+      "num_tokens": 598356.0,
+      "step": 213
+    },
+    {
+      "entropy": 1.1015625,
+      "epoch": 1.5070422535211268,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5e-05,
+      "loss": 1.0299583673477173,
+      "mean_token_accuracy": 0.7973333597183228,
+      "num_tokens": 601364.0,
+      "step": 214
+    },
+    {
+      "entropy": 1.0703125,
+      "epoch": 1.5140845070422535,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5e-05,
+      "loss": 1.041951298713684,
+      "mean_token_accuracy": 0.8040791153907776,
+      "num_tokens": 604608.0,
+      "step": 215
+    },
+    {
+      "entropy": 0.8671875,
+      "epoch": 1.5211267605633803,
+      "grad_norm": 0.84375,
+      "learning_rate": 5e-05,
+      "loss": 0.8255084156990051,
+      "mean_token_accuracy": 0.8424015045166016,
+      "num_tokens": 607281.0,
+      "step": 216
+    },
+    {
+      "entropy": 0.76953125,
+      "epoch": 1.528169014084507,
+      "grad_norm": 0.89453125,
+      "learning_rate": 5e-05,
+      "loss": 0.7329130172729492,
+      "mean_token_accuracy": 0.8575052618980408,
+      "num_tokens": 609654.0,
+      "step": 217
+    },
+    {
+      "entropy": 1.0859375,
+      "epoch": 1.5352112676056338,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 1.0870012044906616,
+      "mean_token_accuracy": 0.7985919713973999,
+      "num_tokens": 612929.0,
+      "step": 218
+    },
+    {
+      "entropy": 1.2421875,
+      "epoch": 1.5422535211267605,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 1.22186279296875,
+      "mean_token_accuracy": 0.7604916095733643,
+      "num_tokens": 616273.0,
+      "step": 219
+    },
+    {
+      "entropy": 1.1640625,
+      "epoch": 1.5492957746478875,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 1.1332319974899292,
+      "mean_token_accuracy": 0.7764706015586853,
+      "num_tokens": 619426.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0546875,
+      "epoch": 1.556338028169014,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 1.0643455982208252,
+      "mean_token_accuracy": 0.7792642116546631,
+      "num_tokens": 622424.0,
+      "step": 221
+    },
+    {
+      "entropy": 0.6875,
+      "epoch": 1.563380281690141,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5e-05,
+      "loss": 0.6475476622581482,
+      "mean_token_accuracy": 0.8877917528152466,
+      "num_tokens": 624660.0,
+      "step": 222
+    },
+    {
+      "entropy": 0.7890625,
+      "epoch": 1.5704225352112675,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5e-05,
+      "loss": 0.7602906823158264,
+      "mean_token_accuracy": 0.8367270231246948,
+      "num_tokens": 627320.0,
+      "step": 223
+    },
+    {
+      "entropy": 0.9140625,
+      "epoch": 1.5774647887323945,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 0.89766526222229,
+      "mean_token_accuracy": 0.8361436724662781,
+      "num_tokens": 630056.0,
+      "step": 224
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 1.584507042253521,
+      "grad_norm": 0.85546875,
+      "learning_rate": 5e-05,
+      "loss": 1.0017104148864746,
+      "mean_token_accuracy": 0.800404965877533,
+      "num_tokens": 633521.0,
+      "step": 225
+    },
+    {
+      "entropy": 0.3515625,
+      "epoch": 1.591549295774648,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5e-05,
+      "loss": 0.3341764211654663,
+      "mean_token_accuracy": 0.9567307829856873,
+      "num_tokens": 635193.0,
+      "step": 226
+    },
+    {
+      "entropy": 0.95703125,
+      "epoch": 1.5985915492957745,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 0.9153825640678406,
+      "mean_token_accuracy": 0.8223153948783875,
+      "num_tokens": 638302.0,
+      "step": 227
+    },
+    {
+      "entropy": 0.98046875,
+      "epoch": 1.6056338028169015,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5e-05,
+      "loss": 0.9880096316337585,
+      "mean_token_accuracy": 0.8147650957107544,
+      "num_tokens": 641290.0,
+      "step": 228
+    },
+    {
+      "entropy": 1.140625,
+      "epoch": 1.612676056338028,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5e-05,
+      "loss": 1.1274977922439575,
+      "mean_token_accuracy": 0.7871226072311401,
+      "num_tokens": 644544.0,
+      "step": 229
+    },
+    {
+      "entropy": 1.3828125,
+      "epoch": 1.619718309859155,
+      "grad_norm": 0.96875,
+      "learning_rate": 5e-05,
+      "loss": 1.3756309747695923,
+      "mean_token_accuracy": 0.7243735790252686,
+      "num_tokens": 648503.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.140625,
+      "epoch": 1.6267605633802817,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5e-05,
+      "loss": 1.129349946975708,
+      "mean_token_accuracy": 0.7763996124267578,
+      "num_tokens": 651494.0,
+      "step": 231
+    },
+    {
+      "entropy": 0.90625,
+      "epoch": 1.6338028169014085,
+      "grad_norm": 1.046875,
+      "learning_rate": 5e-05,
+      "loss": 0.8873183131217957,
+      "mean_token_accuracy": 0.8282105326652527,
+      "num_tokens": 653877.0,
+      "step": 232
+    },
+    {
+      "entropy": 0.80078125,
+      "epoch": 1.6408450704225352,
+      "grad_norm": 0.94921875,
+      "learning_rate": 5e-05,
+      "loss": 0.7888442277908325,
+      "mean_token_accuracy": 0.8641618490219116,
+      "num_tokens": 655961.0,
+      "step": 233
+    },
+    {
+      "entropy": 1.0078125,
+      "epoch": 1.647887323943662,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5e-05,
+      "loss": 0.9641861319541931,
+      "mean_token_accuracy": 0.8236775994300842,
+      "num_tokens": 658351.0,
+      "step": 234
+    },
+    {
+      "entropy": 1.140625,
+      "epoch": 1.6549295774647887,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5e-05,
+      "loss": 1.1152194738388062,
+      "mean_token_accuracy": 0.7858839631080627,
+      "num_tokens": 661306.0,
+      "step": 235
+    },
+    {
+      "entropy": 1.0,
+      "epoch": 1.6619718309859155,
+      "grad_norm": 0.9375,
+      "learning_rate": 5e-05,
+      "loss": 0.9575698971748352,
+      "mean_token_accuracy": 0.8209785223007202,
+      "num_tokens": 664012.0,
+      "step": 236
+    },
+    {
+      "entropy": 0.6171875,
+      "epoch": 1.6690140845070423,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 0.5924875140190125,
+      "mean_token_accuracy": 0.8942976593971252,
+      "num_tokens": 666177.0,
+      "step": 237
+    },
+    {
+      "entropy": 1.1484375,
+      "epoch": 1.676056338028169,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5e-05,
+      "loss": 1.1048884391784668,
+      "mean_token_accuracy": 0.7807835936546326,
+      "num_tokens": 669401.0,
+      "step": 238
+    },
+    {
+      "entropy": 0.8515625,
+      "epoch": 1.6830985915492958,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5e-05,
+      "loss": 0.7597204446792603,
+      "mean_token_accuracy": 0.8503086566925049,
+      "num_tokens": 672001.0,
+      "step": 239
+    },
+    {
+      "entropy": 1.3203125,
+      "epoch": 1.6901408450704225,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5e-05,
+      "loss": 1.3505604267120361,
+      "mean_token_accuracy": 0.7473288774490356,
+      "num_tokens": 675472.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1328125,
+      "epoch": 1.6971830985915493,
+      "grad_norm": 0.91015625,
+      "learning_rate": 5e-05,
+      "loss": 1.0587811470031738,
+      "mean_token_accuracy": 0.801033616065979,
+      "num_tokens": 678189.0,
+      "step": 241
+    },
+    {
+      "entropy": 0.890625,
+      "epoch": 1.704225352112676,
+      "grad_norm": 1.2421875,
+      "learning_rate": 5e-05,
+      "loss": 0.8245605230331421,
+      "mean_token_accuracy": 0.8421264886856079,
+      "num_tokens": 680680.0,
+      "step": 242
+    },
+    {
+      "entropy": 1.15625,
+      "epoch": 1.711267605633803,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5e-05,
+      "loss": 1.1660860776901245,
+      "mean_token_accuracy": 0.7721893787384033,
+      "num_tokens": 683730.0,
+      "step": 243
+    },
+    {
+      "entropy": 0.87109375,
+      "epoch": 1.7183098591549295,
+      "grad_norm": 0.91796875,
+      "learning_rate": 5e-05,
+      "loss": 0.8355469703674316,
+      "mean_token_accuracy": 0.8397717475891113,
+      "num_tokens": 686016.0,
+      "step": 244
+    },
+    {
+      "entropy": 0.9921875,
+      "epoch": 1.7253521126760565,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5e-05,
+      "loss": 0.9942226409912109,
+      "mean_token_accuracy": 0.8151007890701294,
+      "num_tokens": 688355.0,
+      "step": 245
+    },
+    {
+      "entropy": 0.80078125,
+      "epoch": 1.732394366197183,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 0.748877763748169,
+      "mean_token_accuracy": 0.8575118184089661,
+      "num_tokens": 690686.0,
+      "step": 246
+    },
+    {
+      "entropy": 0.67578125,
+      "epoch": 1.73943661971831,
+      "grad_norm": 0.859375,
+      "learning_rate": 5e-05,
+      "loss": 0.6232599020004272,
+      "mean_token_accuracy": 0.8659953474998474,
+      "num_tokens": 693276.0,
+      "step": 247
+    },
+    {
+      "entropy": 0.89453125,
+      "epoch": 1.7464788732394365,
+      "grad_norm": 0.84375,
+      "learning_rate": 5e-05,
+      "loss": 0.8559207916259766,
+      "mean_token_accuracy": 0.8167107105255127,
+      "num_tokens": 696312.0,
+      "step": 248
+    },
+    {
+      "entropy": 0.98828125,
+      "epoch": 1.7535211267605635,
+      "grad_norm": 0.84375,
+      "learning_rate": 5e-05,
+      "loss": 0.9271271824836731,
+      "mean_token_accuracy": 0.8255208134651184,
+      "num_tokens": 699392.0,
+      "step": 249
+    },
+    {
+      "entropy": 0.67578125,
+      "epoch": 1.76056338028169,
+      "grad_norm": 0.859375,
+      "learning_rate": 5e-05,
+      "loss": 0.6688521504402161,
+      "mean_token_accuracy": 0.8663651347160339,
+      "num_tokens": 701832.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1015625,
+      "epoch": 1.767605633802817,
+      "grad_norm": 0.92578125,
+      "learning_rate": 5e-05,
+      "loss": 1.0740944147109985,
+      "mean_token_accuracy": 0.7920154333114624,
+      "num_tokens": 704946.0,
+      "step": 251
+    },
+    {
+      "entropy": 0.92578125,
+      "epoch": 1.7746478873239435,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5e-05,
+      "loss": 0.9155241847038269,
+      "mean_token_accuracy": 0.8214963674545288,
+      "num_tokens": 707587.0,
+      "step": 252
+    },
+    {
+      "entropy": 1.0234375,
+      "epoch": 1.7816901408450705,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 1.0119163990020752,
+      "mean_token_accuracy": 0.8100035786628723,
+      "num_tokens": 710374.0,
+      "step": 253
+    },
+    {
+      "entropy": 1.1484375,
+      "epoch": 1.788732394366197,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 1.098153829574585,
+      "mean_token_accuracy": 0.7915265560150146,
+      "num_tokens": 713356.0,
+      "step": 254
+    },
+    {
+      "entropy": 1.21875,
+      "epoch": 1.795774647887324,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 1.1987107992172241,
+      "mean_token_accuracy": 0.77190762758255,
+      "num_tokens": 716525.0,
+      "step": 255
+    },
+    {
+      "entropy": 1.1171875,
+      "epoch": 1.8028169014084507,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 1.1219631433486938,
+      "mean_token_accuracy": 0.7778891324996948,
+      "num_tokens": 719527.0,
+      "step": 256
+    },
+    {
+      "entropy": 1.0390625,
+      "epoch": 1.8098591549295775,
+      "grad_norm": 0.953125,
+      "learning_rate": 5e-05,
+      "loss": 1.0219084024429321,
+      "mean_token_accuracy": 0.8117510676383972,
+      "num_tokens": 722122.0,
+      "step": 257
+    },
+    {
+      "entropy": 0.92578125,
+      "epoch": 1.8169014084507042,
+      "grad_norm": 1.09375,
+      "learning_rate": 5e-05,
+      "loss": 0.8954271078109741,
+      "mean_token_accuracy": 0.8421261310577393,
+      "num_tokens": 724651.0,
+      "step": 258
+    },
+    {
+      "entropy": 0.79296875,
+      "epoch": 1.823943661971831,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5e-05,
+      "loss": 0.7704645395278931,
+      "mean_token_accuracy": 0.8589590787887573,
+      "num_tokens": 727176.0,
+      "step": 259
+    },
+    {
+      "entropy": 0.63671875,
+      "epoch": 1.8309859154929577,
+      "grad_norm": 1.03125,
+      "learning_rate": 5e-05,
+      "loss": 0.5992922782897949,
+      "mean_token_accuracy": 0.8863636255264282,
+      "num_tokens": 729516.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1484375,
+      "epoch": 1.8380281690140845,
+      "grad_norm": 1.171875,
+      "learning_rate": 5e-05,
+      "loss": 1.0654540061950684,
+      "mean_token_accuracy": 0.8068057298660278,
+      "num_tokens": 732257.0,
+      "step": 261
+    },
+    {
+      "entropy": 0.64453125,
+      "epoch": 1.8450704225352113,
+      "grad_norm": 0.828125,
+      "learning_rate": 5e-05,
+      "loss": 0.6201009154319763,
+      "mean_token_accuracy": 0.8818508386611938,
+      "num_tokens": 734491.0,
+      "step": 262
+    },
+    {
+      "entropy": 1.0,
+      "epoch": 1.852112676056338,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 0.9687674045562744,
+      "mean_token_accuracy": 0.8077198266983032,
+      "num_tokens": 737297.0,
+      "step": 263
+    },
+    {
+      "entropy": 0.5546875,
+      "epoch": 1.8591549295774648,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5e-05,
+      "loss": 0.5199809074401855,
+      "mean_token_accuracy": 0.900547444820404,
+      "num_tokens": 739497.0,
+      "step": 264
+    },
+    {
+      "entropy": 0.81640625,
+      "epoch": 1.8661971830985915,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5e-05,
+      "loss": 0.7673903107643127,
+      "mean_token_accuracy": 0.8505041599273682,
+      "num_tokens": 741786.0,
+      "step": 265
+    },
+    {
+      "entropy": 1.203125,
+      "epoch": 1.8732394366197183,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5e-05,
+      "loss": 1.2104980945587158,
+      "mean_token_accuracy": 0.7729952931404114,
+      "num_tokens": 745186.0,
+      "step": 266
+    },
+    {
+      "entropy": 0.921875,
+      "epoch": 1.880281690140845,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 0.9426409006118774,
+      "mean_token_accuracy": 0.8229434490203857,
+      "num_tokens": 748306.0,
+      "step": 267
+    },
+    {
+      "entropy": 0.92578125,
+      "epoch": 1.887323943661972,
+      "grad_norm": 0.91015625,
+      "learning_rate": 5e-05,
+      "loss": 0.9220845103263855,
+      "mean_token_accuracy": 0.8194718360900879,
+      "num_tokens": 750851.0,
+      "step": 268
+    },
+    {
+      "entropy": 1.0234375,
+      "epoch": 1.8943661971830985,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5e-05,
+      "loss": 1.0415750741958618,
+      "mean_token_accuracy": 0.7877483367919922,
+      "num_tokens": 753879.0,
+      "step": 269
+    },
+    {
+      "entropy": 0.87109375,
+      "epoch": 1.9014084507042255,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5e-05,
+      "loss": 0.858406126499176,
+      "mean_token_accuracy": 0.8365448713302612,
+      "num_tokens": 756897.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.0625,
+      "epoch": 1.908450704225352,
+      "grad_norm": 0.875,
+      "learning_rate": 5e-05,
+      "loss": 1.1015472412109375,
+      "mean_token_accuracy": 0.7784239649772644,
+      "num_tokens": 760344.0,
+      "step": 271
+    },
+    {
+      "entropy": 1.03125,
+      "epoch": 1.915492957746479,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5e-05,
+      "loss": 1.0051850080490112,
+      "mean_token_accuracy": 0.8070366382598877,
+      "num_tokens": 763109.0,
+      "step": 272
+    },
+    {
+      "entropy": 1.140625,
+      "epoch": 1.9225352112676055,
+      "grad_norm": 0.89453125,
+      "learning_rate": 5e-05,
+      "loss": 1.1266851425170898,
+      "mean_token_accuracy": 0.7730742692947388,
+      "num_tokens": 765999.0,
+      "step": 273
+    },
+    {
+      "entropy": 0.8046875,
+      "epoch": 1.9295774647887325,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5e-05,
+      "loss": 0.7725271582603455,
+      "mean_token_accuracy": 0.855555534362793,
+      "num_tokens": 768527.0,
+      "step": 274
+    },
+    {
+      "entropy": 0.74609375,
+      "epoch": 1.936619718309859,
+      "grad_norm": 0.78125,
+      "learning_rate": 5e-05,
+      "loss": 0.7248237133026123,
+      "mean_token_accuracy": 0.8687350749969482,
+      "num_tokens": 771049.0,
+      "step": 275
+    },
+    {
+      "entropy": 0.82421875,
+      "epoch": 1.943661971830986,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5e-05,
+      "loss": 0.8026857972145081,
+      "mean_token_accuracy": 0.8561615943908691,
+      "num_tokens": 773532.0,
+      "step": 276
+    },
+    {
+      "entropy": 1.0,
+      "epoch": 1.9507042253521125,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5e-05,
+      "loss": 0.9862289428710938,
+      "mean_token_accuracy": 0.8059446811676025,
+      "num_tokens": 776467.0,
+      "step": 277
+    },
+    {
+      "entropy": 1.1328125,
+      "epoch": 1.9577464788732395,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5e-05,
+      "loss": 1.1140096187591553,
+      "mean_token_accuracy": 0.7820383310317993,
+      "num_tokens": 779448.0,
+      "step": 278
+    },
+    {
+      "entropy": 1.0859375,
+      "epoch": 1.9647887323943662,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 1.0614746809005737,
+      "mean_token_accuracy": 0.7838625907897949,
+      "num_tokens": 782542.0,
+      "step": 279
+    },
+    {
+      "entropy": 1.0625,
+      "epoch": 1.971830985915493,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5e-05,
+      "loss": 1.053848385810852,
+      "mean_token_accuracy": 0.7868378758430481,
+      "num_tokens": 785665.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.0625,
+      "epoch": 1.9788732394366197,
+      "grad_norm": 1.015625,
+      "learning_rate": 5e-05,
+      "loss": 1.0552152395248413,
+      "mean_token_accuracy": 0.7965670824050903,
+      "num_tokens": 788819.0,
+      "step": 281
+    },
+    {
+      "entropy": 1.1171875,
+      "epoch": 1.9859154929577465,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5e-05,
+      "loss": 1.146568775177002,
+      "mean_token_accuracy": 0.7714470624923706,
+      "num_tokens": 791916.0,
+      "step": 282
+    },
+    {
+      "entropy": 0.875,
+      "epoch": 1.9929577464788732,
+      "grad_norm": 1.015625,
+      "learning_rate": 5e-05,
+      "loss": 0.8557621240615845,
+      "mean_token_accuracy": 0.8393825888633728,
+      "num_tokens": 794321.0,
+      "step": 283
+    },
+    {
+      "entropy": 0.69140625,
+      "epoch": 2.0,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5e-05,
+      "loss": 0.694361686706543,
+      "mean_token_accuracy": 0.8591725826263428,
+      "num_tokens": 796722.0,
+      "step": 284
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9236111111111112,
+      "eval_loss": 0.9322065114974976,
+      "eval_mean_token_accuracy": 0.8198160545693504,
+      "eval_num_tokens": 796722.0,
+      "eval_runtime": 3.7129,
+      "eval_samples_per_second": 76.489,
+      "eval_steps_per_second": 9.696,
+      "step": 284
+    },
+    {
+      "entropy": 0.73828125,
+      "epoch": 2.007042253521127,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5e-05,
+      "loss": 0.7374175786972046,
+      "mean_token_accuracy": 0.8610056638717651,
+      "num_tokens": 798838.0,
+      "step": 285
+    },
+    {
+      "entropy": 0.8828125,
+      "epoch": 2.0140845070422535,
+      "grad_norm": 0.91015625,
+      "learning_rate": 5e-05,
+      "loss": 0.881466269493103,
+      "mean_token_accuracy": 0.8392930626869202,
+      "num_tokens": 801279.0,
+      "step": 286
+    },
+    {
+      "entropy": 1.0859375,
+      "epoch": 2.0211267605633805,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5e-05,
+      "loss": 1.0629348754882812,
+      "mean_token_accuracy": 0.7871087789535522,
+      "num_tokens": 804514.0,
+      "step": 287
+    },
+    {
+      "entropy": 0.75,
+      "epoch": 2.028169014084507,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5e-05,
+      "loss": 0.7385261654853821,
+      "mean_token_accuracy": 0.8663028478622437,
+      "num_tokens": 806721.0,
+      "step": 288
+    },
+    {
+      "entropy": 0.87890625,
+      "epoch": 2.035211267605634,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5e-05,
+      "loss": 0.8331909775733948,
+      "mean_token_accuracy": 0.8295155763626099,
+      "num_tokens": 809392.0,
+      "step": 289
+    },
+    {
+      "entropy": 0.92578125,
+      "epoch": 2.0422535211267605,
+      "grad_norm": 0.921875,
+      "learning_rate": 5e-05,
+      "loss": 0.9309926629066467,
+      "mean_token_accuracy": 0.817585825920105,
+      "num_tokens": 811993.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.015625,
+      "epoch": 2.0492957746478875,
+      "grad_norm": 1.015625,
+      "learning_rate": 5e-05,
+      "loss": 1.0413026809692383,
+      "mean_token_accuracy": 0.80402010679245,
+      "num_tokens": 814588.0,
+      "step": 291
+    },
+    {
+      "entropy": 0.69140625,
+      "epoch": 2.056338028169014,
+      "grad_norm": 1.15625,
+      "learning_rate": 5e-05,
+      "loss": 0.6655716896057129,
+      "mean_token_accuracy": 0.8746465444564819,
+      "num_tokens": 816718.0,
+      "step": 292
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 2.063380281690141,
+      "grad_norm": 0.921875,
+      "learning_rate": 5e-05,
+      "loss": 1.0103598833084106,
+      "mean_token_accuracy": 0.7893930077552795,
+      "num_tokens": 819988.0,
+      "step": 293
+    },
+    {
+      "entropy": 0.65625,
+      "epoch": 2.0704225352112675,
+      "grad_norm": 0.8671875,
+      "learning_rate": 5e-05,
+      "loss": 0.6369961500167847,
+      "mean_token_accuracy": 0.8664092421531677,
+      "num_tokens": 822586.0,
+      "step": 294
+    },
+    {
+      "entropy": 0.90234375,
+      "epoch": 2.0774647887323945,
+      "grad_norm": 0.92578125,
+      "learning_rate": 5e-05,
+      "loss": 0.851521372795105,
+      "mean_token_accuracy": 0.8140385746955872,
+      "num_tokens": 825600.0,
+      "step": 295
+    },
+    {
+      "entropy": 0.65234375,
+      "epoch": 2.084507042253521,
+      "grad_norm": 0.8671875,
+      "learning_rate": 5e-05,
+      "loss": 0.5919080972671509,
+      "mean_token_accuracy": 0.8811880946159363,
+      "num_tokens": 828133.0,
+      "step": 296
+    },
+    {
+      "entropy": 0.640625,
+      "epoch": 2.091549295774648,
+      "grad_norm": 0.84375,
+      "learning_rate": 5e-05,
+      "loss": 0.5999734997749329,
+      "mean_token_accuracy": 0.8843873739242554,
+      "num_tokens": 830165.0,
+      "step": 297
+    },
+    {
+      "entropy": 0.79296875,
+      "epoch": 2.0985915492957745,
+      "grad_norm": 0.765625,
+      "learning_rate": 5e-05,
+      "loss": 0.7580751180648804,
+      "mean_token_accuracy": 0.8627010583877563,
+      "num_tokens": 832846.0,
+      "step": 298
+    },
+    {
+      "entropy": 0.78125,
+      "epoch": 2.1056338028169015,
+      "grad_norm": 0.953125,
+      "learning_rate": 5e-05,
+      "loss": 0.7869684100151062,
+      "mean_token_accuracy": 0.8528945446014404,
+      "num_tokens": 835376.0,
+      "step": 299
+    },
+    {
+      "entropy": 0.74609375,
+      "epoch": 2.112676056338028,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5e-05,
+      "loss": 0.7340086102485657,
+      "mean_token_accuracy": 0.8652892708778381,
+      "num_tokens": 837804.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9765625,
+      "epoch": 2.119718309859155,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 0.9712856411933899,
+      "mean_token_accuracy": 0.8149620294570923,
+      "num_tokens": 840579.0,
+      "step": 301
+    },
+    {
+      "entropy": 0.8046875,
+      "epoch": 2.1267605633802815,
+      "grad_norm": 0.96875,
+      "learning_rate": 5e-05,
+      "loss": 0.7940016388893127,
+      "mean_token_accuracy": 0.8557435274124146,
+      "num_tokens": 842833.0,
+      "step": 302
+    },
+    {
+      "entropy": 1.1640625,
+      "epoch": 2.1338028169014085,
+      "grad_norm": 0.9375,
+      "learning_rate": 5e-05,
+      "loss": 1.1491037607192993,
+      "mean_token_accuracy": 0.7900171875953674,
+      "num_tokens": 845746.0,
+      "step": 303
+    },
+    {
+      "entropy": 1.2109375,
+      "epoch": 2.140845070422535,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5e-05,
+      "loss": 1.1838632822036743,
+      "mean_token_accuracy": 0.7814313173294067,
+      "num_tokens": 848856.0,
+      "step": 304
+    },
+    {
+      "entropy": 0.71484375,
+      "epoch": 2.147887323943662,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5e-05,
+      "loss": 0.6575556993484497,
+      "mean_token_accuracy": 0.865422785282135,
+      "num_tokens": 851383.0,
+      "step": 305
+    },
+    {
+      "entropy": 0.8984375,
+      "epoch": 2.1549295774647885,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 0.8470981121063232,
+      "mean_token_accuracy": 0.8228431344032288,
+      "num_tokens": 854648.0,
+      "step": 306
+    },
+    {
+      "entropy": 1.265625,
+      "epoch": 2.1619718309859155,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5e-05,
+      "loss": 1.2397805452346802,
+      "mean_token_accuracy": 0.7695154547691345,
+      "num_tokens": 857628.0,
+      "step": 307
+    },
+    {
+      "entropy": 0.9609375,
+      "epoch": 2.169014084507042,
+      "grad_norm": 0.921875,
+      "learning_rate": 5e-05,
+      "loss": 1.025403380393982,
+      "mean_token_accuracy": 0.8165938854217529,
+      "num_tokens": 860384.0,
+      "step": 308
+    },
+    {
+      "entropy": 1.1484375,
+      "epoch": 2.176056338028169,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 1.1041961908340454,
+      "mean_token_accuracy": 0.7777073979377747,
+      "num_tokens": 863550.0,
+      "step": 309
+    },
+    {
+      "entropy": 0.98828125,
+      "epoch": 2.183098591549296,
+      "grad_norm": 0.77734375,
+      "learning_rate": 5e-05,
+      "loss": 0.9599985480308533,
+      "mean_token_accuracy": 0.8179652690887451,
+      "num_tokens": 866497.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.92578125,
+      "epoch": 2.1901408450704225,
+      "grad_norm": 0.890625,
+      "learning_rate": 5e-05,
+      "loss": 0.8720007538795471,
+      "mean_token_accuracy": 0.8374785780906677,
+      "num_tokens": 868837.0,
+      "step": 311
+    },
+    {
+      "entropy": 1.1953125,
+      "epoch": 2.1971830985915495,
+      "grad_norm": 0.921875,
+      "learning_rate": 5e-05,
+      "loss": 1.1740869283676147,
+      "mean_token_accuracy": 0.7726989984512329,
+      "num_tokens": 872061.0,
+      "step": 312
+    },
+    {
+      "entropy": 1.2265625,
+      "epoch": 2.204225352112676,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5e-05,
+      "loss": 1.2219297885894775,
+      "mean_token_accuracy": 0.755230724811554,
+      "num_tokens": 875558.0,
+      "step": 313
+    },
+    {
+      "entropy": 0.9453125,
+      "epoch": 2.211267605633803,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 0.8945049047470093,
+      "mean_token_accuracy": 0.8340791463851929,
+      "num_tokens": 878471.0,
+      "step": 314
+    },
+    {
+      "entropy": 0.6796875,
+      "epoch": 2.2183098591549295,
+      "grad_norm": 1.15625,
+      "learning_rate": 5e-05,
+      "loss": 0.6467694640159607,
+      "mean_token_accuracy": 0.8709545135498047,
+      "num_tokens": 880920.0,
+      "step": 315
+    },
+    {
+      "entropy": 0.95703125,
+      "epoch": 2.2253521126760565,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5e-05,
+      "loss": 0.9356101155281067,
+      "mean_token_accuracy": 0.8094237446784973,
+      "num_tokens": 884260.0,
+      "step": 316
+    },
+    {
+      "entropy": 0.87109375,
+      "epoch": 2.232394366197183,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5e-05,
+      "loss": 0.8017354011535645,
+      "mean_token_accuracy": 0.8435049653053284,
+      "num_tokens": 887201.0,
+      "step": 317
+    },
+    {
+      "entropy": 0.875,
+      "epoch": 2.23943661971831,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5e-05,
+      "loss": 0.9189158082008362,
+      "mean_token_accuracy": 0.8133284449577332,
+      "num_tokens": 889925.0,
+      "step": 318
+    },
+    {
+      "entropy": 1.1796875,
+      "epoch": 2.2464788732394365,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5e-05,
+      "loss": 1.1888455152511597,
+      "mean_token_accuracy": 0.7918660044670105,
+      "num_tokens": 892859.0,
+      "step": 319
+    },
+    {
+      "entropy": 0.828125,
+      "epoch": 2.2535211267605635,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5e-05,
+      "loss": 0.8145902156829834,
+      "mean_token_accuracy": 0.8462405800819397,
+      "num_tokens": 895527.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.87109375,
+      "epoch": 2.26056338028169,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 0.8571617603302002,
+      "mean_token_accuracy": 0.8290506601333618,
+      "num_tokens": 898337.0,
+      "step": 321
+    },
+    {
+      "entropy": 1.171875,
+      "epoch": 2.267605633802817,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5e-05,
+      "loss": 1.1506730318069458,
+      "mean_token_accuracy": 0.7653835415840149,
+      "num_tokens": 901904.0,
+      "step": 322
+    },
+    {
+      "entropy": 1.3984375,
+      "epoch": 2.2746478873239435,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 1.3748078346252441,
+      "mean_token_accuracy": 0.7246376872062683,
+      "num_tokens": 905431.0,
+      "step": 323
+    },
+    {
+      "entropy": 0.9375,
+      "epoch": 2.2816901408450705,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 0.9095409512519836,
+      "mean_token_accuracy": 0.8207698464393616,
+      "num_tokens": 907933.0,
+      "step": 324
+    },
+    {
+      "entropy": 0.921875,
+      "epoch": 2.288732394366197,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5e-05,
+      "loss": 0.9583578705787659,
+      "mean_token_accuracy": 0.8119794726371765,
+      "num_tokens": 911063.0,
+      "step": 325
+    },
+    {
+      "entropy": 0.953125,
+      "epoch": 2.295774647887324,
+      "grad_norm": 0.828125,
+      "learning_rate": 5e-05,
+      "loss": 0.9665497541427612,
+      "mean_token_accuracy": 0.8108496069908142,
+      "num_tokens": 914555.0,
+      "step": 326
+    },
+    {
+      "entropy": 1.125,
+      "epoch": 2.3028169014084505,
+      "grad_norm": 0.921875,
+      "learning_rate": 5e-05,
+      "loss": 1.0927993059158325,
+      "mean_token_accuracy": 0.7958276271820068,
+      "num_tokens": 917487.0,
+      "step": 327
+    },
+    {
+      "entropy": 0.87109375,
+      "epoch": 2.3098591549295775,
+      "grad_norm": 0.75,
+      "learning_rate": 5e-05,
+      "loss": 0.8782572150230408,
+      "mean_token_accuracy": 0.8208358883857727,
+      "num_tokens": 920749.0,
+      "step": 328
+    },
+    {
+      "entropy": 0.9765625,
+      "epoch": 2.316901408450704,
+      "grad_norm": 0.8671875,
+      "learning_rate": 5e-05,
+      "loss": 0.9704670906066895,
+      "mean_token_accuracy": 0.8154719471931458,
+      "num_tokens": 923575.0,
+      "step": 329
+    },
+    {
+      "entropy": 1.078125,
+      "epoch": 2.323943661971831,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 1.0923304557800293,
+      "mean_token_accuracy": 0.8025941252708435,
+      "num_tokens": 926744.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.171875,
+      "epoch": 2.330985915492958,
+      "grad_norm": 0.890625,
+      "learning_rate": 5e-05,
+      "loss": 1.156880259513855,
+      "mean_token_accuracy": 0.766771137714386,
+      "num_tokens": 929942.0,
+      "step": 331
+    },
+    {
+      "entropy": 0.484375,
+      "epoch": 2.3380281690140845,
+      "grad_norm": 1.203125,
+      "learning_rate": 5e-05,
+      "loss": 0.5406565070152283,
+      "mean_token_accuracy": 0.9252772927284241,
+      "num_tokens": 931663.0,
+      "step": 332
+    },
+    {
+      "entropy": 0.8984375,
+      "epoch": 2.345070422535211,
+      "grad_norm": 0.76953125,
+      "learning_rate": 5e-05,
+      "loss": 0.8872555494308472,
+      "mean_token_accuracy": 0.8348463773727417,
+      "num_tokens": 934535.0,
+      "step": 333
+    },
+    {
+      "entropy": 0.60546875,
+      "epoch": 2.352112676056338,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5e-05,
+      "loss": 0.5910133123397827,
+      "mean_token_accuracy": 0.9001062512397766,
+      "num_tokens": 936425.0,
+      "step": 334
+    },
+    {
+      "entropy": 0.796875,
+      "epoch": 2.359154929577465,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 0.7814599275588989,
+      "mean_token_accuracy": 0.8306627869606018,
+      "num_tokens": 939179.0,
+      "step": 335
+    },
+    {
+      "entropy": 0.984375,
+      "epoch": 2.3661971830985915,
+      "grad_norm": 0.828125,
+      "learning_rate": 5e-05,
+      "loss": 0.9610760807991028,
+      "mean_token_accuracy": 0.8171220421791077,
+      "num_tokens": 941932.0,
+      "step": 336
+    },
+    {
+      "entropy": 0.58203125,
+      "epoch": 2.3732394366197185,
+      "grad_norm": 0.984375,
+      "learning_rate": 5e-05,
+      "loss": 0.6330059170722961,
+      "mean_token_accuracy": 0.894197940826416,
+      "num_tokens": 943991.0,
+      "step": 337
+    },
+    {
+      "entropy": 0.95703125,
+      "epoch": 2.380281690140845,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5e-05,
+      "loss": 0.9380885362625122,
+      "mean_token_accuracy": 0.8218390941619873,
+      "num_tokens": 946609.0,
+      "step": 338
+    },
+    {
+      "entropy": 0.91796875,
+      "epoch": 2.387323943661972,
+      "grad_norm": 0.9375,
+      "learning_rate": 5e-05,
+      "loss": 0.8984470963478088,
+      "mean_token_accuracy": 0.8211155533790588,
+      "num_tokens": 949127.0,
+      "step": 339
+    },
+    {
+      "entropy": 1.0234375,
+      "epoch": 2.3943661971830985,
+      "grad_norm": 0.859375,
+      "learning_rate": 5e-05,
+      "loss": 1.0115832090377808,
+      "mean_token_accuracy": 0.814687967300415,
+      "num_tokens": 951763.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.8828125,
+      "epoch": 2.4014084507042255,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 0.8474752306938171,
+      "mean_token_accuracy": 0.8142760396003723,
+      "num_tokens": 954727.0,
+      "step": 341
+    },
+    {
+      "entropy": 1.1015625,
+      "epoch": 2.408450704225352,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5e-05,
+      "loss": 1.0535743236541748,
+      "mean_token_accuracy": 0.8000690340995789,
+      "num_tokens": 957631.0,
+      "step": 342
+    },
+    {
+      "entropy": 0.6328125,
+      "epoch": 2.415492957746479,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5e-05,
+      "loss": 0.6323789954185486,
+      "mean_token_accuracy": 0.8735529780387878,
+      "num_tokens": 959885.0,
+      "step": 343
+    },
+    {
+      "entropy": 0.796875,
+      "epoch": 2.4225352112676055,
+      "grad_norm": 0.85546875,
+      "learning_rate": 5e-05,
+      "loss": 0.7820839881896973,
+      "mean_token_accuracy": 0.8398364782333374,
+      "num_tokens": 962584.0,
+      "step": 344
+    },
+    {
+      "entropy": 0.98046875,
+      "epoch": 2.4295774647887325,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 0.9701536297798157,
+      "mean_token_accuracy": 0.8102932572364807,
+      "num_tokens": 965934.0,
+      "step": 345
+    },
+    {
+      "entropy": 1.1484375,
+      "epoch": 2.436619718309859,
+      "grad_norm": 0.83203125,
+      "learning_rate": 5e-05,
+      "loss": 1.173242449760437,
+      "mean_token_accuracy": 0.7778127193450928,
+      "num_tokens": 969124.0,
+      "step": 346
+    },
+    {
+      "entropy": 1.078125,
+      "epoch": 2.443661971830986,
+      "grad_norm": 0.8125,
+      "learning_rate": 5e-05,
+      "loss": 1.0826599597930908,
+      "mean_token_accuracy": 0.7962716221809387,
+      "num_tokens": 972136.0,
+      "step": 347
+    },
+    {
+      "entropy": 0.87109375,
+      "epoch": 2.4507042253521125,
+      "grad_norm": 0.89453125,
+      "learning_rate": 5e-05,
+      "loss": 0.8673123717308044,
+      "mean_token_accuracy": 0.8210601210594177,
+      "num_tokens": 974955.0,
+      "step": 348
+    },
+    {
+      "entropy": 0.69921875,
+      "epoch": 2.4577464788732395,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5e-05,
+      "loss": 0.6696112751960754,
+      "mean_token_accuracy": 0.8591090440750122,
+      "num_tokens": 977724.0,
+      "step": 349
+    },
+    {
+      "entropy": 0.9375,
+      "epoch": 2.464788732394366,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5e-05,
+      "loss": 0.8982508778572083,
+      "mean_token_accuracy": 0.8321678042411804,
+      "num_tokens": 980306.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.71875,
+      "epoch": 2.471830985915493,
+      "grad_norm": 0.89453125,
+      "learning_rate": 5e-05,
+      "loss": 0.7076046466827393,
+      "mean_token_accuracy": 0.8737216591835022,
+      "num_tokens": 982563.0,
+      "step": 351
+    },
+    {
+      "entropy": 0.4375,
+      "epoch": 2.4788732394366195,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5e-05,
+      "loss": 0.40601345896720886,
+      "mean_token_accuracy": 0.930973470211029,
+      "num_tokens": 984266.0,
+      "step": 352
+    },
+    {
+      "entropy": 0.96875,
+      "epoch": 2.4859154929577465,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5e-05,
+      "loss": 0.9373111128807068,
+      "mean_token_accuracy": 0.8173550367355347,
+      "num_tokens": 986913.0,
+      "step": 353
+    },
+    {
+      "entropy": 0.81640625,
+      "epoch": 2.492957746478873,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5e-05,
+      "loss": 0.7937692403793335,
+      "mean_token_accuracy": 0.8562434315681458,
+      "num_tokens": 989780.0,
+      "step": 354
+    },
+    {
+      "entropy": 0.87109375,
+      "epoch": 2.5,
+      "grad_norm": 0.83203125,
+      "learning_rate": 5e-05,
+      "loss": 0.8554614186286926,
+      "mean_token_accuracy": 0.8344972133636475,
+      "num_tokens": 992652.0,
+      "step": 355
+    },
+    {
+      "entropy": 0.93359375,
+      "epoch": 2.507042253521127,
+      "grad_norm": 0.96875,
+      "learning_rate": 5e-05,
+      "loss": 0.9242957234382629,
+      "mean_token_accuracy": 0.8118355870246887,
+      "num_tokens": 995482.0,
+      "step": 356
+    },
+    {
+      "entropy": 1.0703125,
+      "epoch": 2.5140845070422535,
+      "grad_norm": 0.92578125,
+      "learning_rate": 5e-05,
+      "loss": 1.0817320346832275,
+      "mean_token_accuracy": 0.7886710166931152,
+      "num_tokens": 998244.0,
+      "step": 357
+    },
+    {
+      "entropy": 0.65625,
+      "epoch": 2.52112676056338,
+      "grad_norm": 0.921875,
+      "learning_rate": 5e-05,
+      "loss": 0.6246756911277771,
+      "mean_token_accuracy": 0.8944069147109985,
+      "num_tokens": 1000326.0,
+      "step": 358
+    },
+    {
+      "entropy": 1.2265625,
+      "epoch": 2.528169014084507,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5e-05,
+      "loss": 1.2332262992858887,
+      "mean_token_accuracy": 0.7636415958404541,
+      "num_tokens": 1003816.0,
+      "step": 359
+    },
+    {
+      "entropy": 1.1015625,
+      "epoch": 2.535211267605634,
+      "grad_norm": 0.875,
+      "learning_rate": 5e-05,
+      "loss": 1.0355029106140137,
+      "mean_token_accuracy": 0.8010130524635315,
+      "num_tokens": 1006588.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.92578125,
+      "epoch": 2.5422535211267605,
+      "grad_norm": 0.7890625,
+      "learning_rate": 5e-05,
+      "loss": 0.8966702818870544,
+      "mean_token_accuracy": 0.8360433578491211,
+      "num_tokens": 1009548.0,
+      "step": 361
+    },
+    {
+      "entropy": 0.87109375,
+      "epoch": 2.5492957746478875,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5e-05,
+      "loss": 0.8771792054176331,
+      "mean_token_accuracy": 0.8336964249610901,
+      "num_tokens": 1012310.0,
+      "step": 362
+    },
+    {
+      "entropy": 0.84375,
+      "epoch": 2.556338028169014,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5e-05,
+      "loss": 0.8473308086395264,
+      "mean_token_accuracy": 0.8333333134651184,
+      "num_tokens": 1014880.0,
+      "step": 363
+    },
+    {
+      "entropy": 1.0234375,
+      "epoch": 2.563380281690141,
+      "grad_norm": 0.8125,
+      "learning_rate": 5e-05,
+      "loss": 1.00178062915802,
+      "mean_token_accuracy": 0.8125210404396057,
+      "num_tokens": 1017859.0,
+      "step": 364
+    },
+    {
+      "entropy": 1.0390625,
+      "epoch": 2.5704225352112675,
+      "grad_norm": 0.85546875,
+      "learning_rate": 5e-05,
+      "loss": 1.007246971130371,
+      "mean_token_accuracy": 0.801886796951294,
+      "num_tokens": 1020835.0,
+      "step": 365
+    },
+    {
+      "entropy": 1.03125,
+      "epoch": 2.5774647887323945,
+      "grad_norm": 0.90625,
+      "learning_rate": 5e-05,
+      "loss": 0.9952259659767151,
+      "mean_token_accuracy": 0.7855361700057983,
+      "num_tokens": 1024051.0,
+      "step": 366
+    },
+    {
+      "entropy": 0.98046875,
+      "epoch": 2.584507042253521,
+      "grad_norm": 0.83203125,
+      "learning_rate": 5e-05,
+      "loss": 0.9216870665550232,
+      "mean_token_accuracy": 0.8177199363708496,
+      "num_tokens": 1027208.0,
+      "step": 367
+    },
+    {
+      "entropy": 0.83984375,
+      "epoch": 2.591549295774648,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 0.8254913091659546,
+      "mean_token_accuracy": 0.8484081029891968,
+      "num_tokens": 1029980.0,
+      "step": 368
+    },
+    {
+      "entropy": 1.03125,
+      "epoch": 2.5985915492957745,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 1.01625657081604,
+      "mean_token_accuracy": 0.8179768323898315,
+      "num_tokens": 1033092.0,
+      "step": 369
+    },
+    {
+      "entropy": 1.015625,
+      "epoch": 2.6056338028169015,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5e-05,
+      "loss": 0.9791706800460815,
+      "mean_token_accuracy": 0.8002439737319946,
+      "num_tokens": 1036379.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.95703125,
+      "epoch": 2.612676056338028,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5e-05,
+      "loss": 0.9625215530395508,
+      "mean_token_accuracy": 0.8120957016944885,
+      "num_tokens": 1039479.0,
+      "step": 371
+    },
+    {
+      "entropy": 0.90625,
+      "epoch": 2.619718309859155,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5e-05,
+      "loss": 0.8910521268844604,
+      "mean_token_accuracy": 0.8163344860076904,
+      "num_tokens": 1042046.0,
+      "step": 372
+    },
+    {
+      "entropy": 0.64453125,
+      "epoch": 2.626760563380282,
+      "grad_norm": 1.0625,
+      "learning_rate": 5e-05,
+      "loss": 0.6820624470710754,
+      "mean_token_accuracy": 0.8745020031929016,
+      "num_tokens": 1044062.0,
+      "step": 373
+    },
+    {
+      "entropy": 1.0234375,
+      "epoch": 2.6338028169014085,
+      "grad_norm": 0.953125,
+      "learning_rate": 5e-05,
+      "loss": 1.0213501453399658,
+      "mean_token_accuracy": 0.8043197989463806,
+      "num_tokens": 1047172.0,
+      "step": 374
+    },
+    {
+      "entropy": 1.1328125,
+      "epoch": 2.640845070422535,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5e-05,
+      "loss": 1.163074016571045,
+      "mean_token_accuracy": 0.7688495516777039,
+      "num_tokens": 1050005.0,
+      "step": 375
+    },
+    {
+      "entropy": 0.81640625,
+      "epoch": 2.647887323943662,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5e-05,
+      "loss": 0.8177797198295593,
+      "mean_token_accuracy": 0.8428515195846558,
+      "num_tokens": 1052552.0,
+      "step": 376
+    },
+    {
+      "entropy": 1.015625,
+      "epoch": 2.654929577464789,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5e-05,
+      "loss": 0.9992722868919373,
+      "mean_token_accuracy": 0.80116206407547,
+      "num_tokens": 1055658.0,
+      "step": 377
+    },
+    {
+      "entropy": 0.89453125,
+      "epoch": 2.6619718309859155,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5e-05,
+      "loss": 0.925032913684845,
+      "mean_token_accuracy": 0.8376479148864746,
+      "num_tokens": 1058370.0,
+      "step": 378
+    },
+    {
+      "entropy": 1.375,
+      "epoch": 2.669014084507042,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5e-05,
+      "loss": 1.3769747018814087,
+      "mean_token_accuracy": 0.7368420958518982,
+      "num_tokens": 1061665.0,
+      "step": 379
+    },
+    {
+      "entropy": 0.7109375,
+      "epoch": 2.676056338028169,
+      "grad_norm": 0.80859375,
+      "learning_rate": 5e-05,
+      "loss": 0.6777468919754028,
+      "mean_token_accuracy": 0.8668290972709656,
+      "num_tokens": 1064136.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0078125,
+      "epoch": 2.683098591549296,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5e-05,
+      "loss": 0.9769162535667419,
+      "mean_token_accuracy": 0.8099352121353149,
+      "num_tokens": 1067385.0,
+      "step": 381
+    },
+    {
+      "entropy": 1.21875,
+      "epoch": 2.6901408450704225,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5e-05,
+      "loss": 1.1385142803192139,
+      "mean_token_accuracy": 0.7784450054168701,
+      "num_tokens": 1070557.0,
+      "step": 382
+    },
+    {
+      "entropy": 0.3125,
+      "epoch": 2.697183098591549,
+      "grad_norm": 1.2421875,
+      "learning_rate": 5e-05,
+      "loss": 0.3028402626514435,
+      "mean_token_accuracy": 0.9576837420463562,
+      "num_tokens": 1072361.0,
+      "step": 383
+    },
+    {
+      "entropy": 0.82421875,
+      "epoch": 2.704225352112676,
+      "grad_norm": 0.796875,
+      "learning_rate": 5e-05,
+      "loss": 0.8440111875534058,
+      "mean_token_accuracy": 0.8486763834953308,
+      "num_tokens": 1074900.0,
+      "step": 384
+    },
+    {
+      "entropy": 0.859375,
+      "epoch": 2.711267605633803,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5e-05,
+      "loss": 0.8036064505577087,
+      "mean_token_accuracy": 0.8372649550437927,
+      "num_tokens": 1077833.0,
+      "step": 385
+    },
+    {
+      "entropy": 0.5703125,
+      "epoch": 2.7183098591549295,
+      "grad_norm": 0.91015625,
+      "learning_rate": 5e-05,
+      "loss": 0.5942102670669556,
+      "mean_token_accuracy": 0.9045275449752808,
+      "num_tokens": 1079873.0,
+      "step": 386
+    },
+    {
+      "entropy": 0.703125,
+      "epoch": 2.7253521126760565,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 0.7002112865447998,
+      "mean_token_accuracy": 0.8748914003372192,
+      "num_tokens": 1082183.0,
+      "step": 387
+    },
+    {
+      "entropy": 0.69140625,
+      "epoch": 2.732394366197183,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5e-05,
+      "loss": 0.6743981838226318,
+      "mean_token_accuracy": 0.8800690174102783,
+      "num_tokens": 1084509.0,
+      "step": 388
+    },
+    {
+      "entropy": 1.171875,
+      "epoch": 2.73943661971831,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5e-05,
+      "loss": 1.1858654022216797,
+      "mean_token_accuracy": 0.771606981754303,
+      "num_tokens": 1087560.0,
+      "step": 389
+    },
+    {
+      "entropy": 0.76953125,
+      "epoch": 2.7464788732394365,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5e-05,
+      "loss": 0.7856656908988953,
+      "mean_token_accuracy": 0.8549124002456665,
+      "num_tokens": 1090194.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.125,
+      "epoch": 2.7535211267605635,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5e-05,
+      "loss": 1.0827059745788574,
+      "mean_token_accuracy": 0.7939068078994751,
+      "num_tokens": 1092992.0,
+      "step": 391
+    },
+    {
+      "entropy": 1.359375,
+      "epoch": 2.76056338028169,
+      "grad_norm": 0.94921875,
+      "learning_rate": 5e-05,
+      "loss": 1.3576359748840332,
+      "mean_token_accuracy": 0.7380163073539734,
+      "num_tokens": 1096317.0,
+      "step": 392
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 2.767605633802817,
+      "grad_norm": 0.953125,
+      "learning_rate": 5e-05,
+      "loss": 1.0619326829910278,
+      "mean_token_accuracy": 0.7976710200309753,
+      "num_tokens": 1099073.0,
+      "step": 393
+    },
+    {
+      "entropy": 0.54296875,
+      "epoch": 2.7746478873239435,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5e-05,
+      "loss": 0.4828389286994934,
+      "mean_token_accuracy": 0.8992869853973389,
+      "num_tokens": 1101325.0,
+      "step": 394
+    },
+    {
+      "entropy": 0.83203125,
+      "epoch": 2.7816901408450705,
+      "grad_norm": 0.7578125,
+      "learning_rate": 5e-05,
+      "loss": 0.7946297526359558,
+      "mean_token_accuracy": 0.8518151640892029,
+      "num_tokens": 1104363.0,
+      "step": 395
+    },
+    {
+      "entropy": 1.0703125,
+      "epoch": 2.788732394366197,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5e-05,
+      "loss": 1.0588973760604858,
+      "mean_token_accuracy": 0.7770973443984985,
+      "num_tokens": 1107637.0,
+      "step": 396
+    },
+    {
+      "entropy": 0.78125,
+      "epoch": 2.795774647887324,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5e-05,
+      "loss": 0.7428227066993713,
+      "mean_token_accuracy": 0.8583959937095642,
+      "num_tokens": 1110039.0,
+      "step": 397
+    },
+    {
+      "entropy": 1.1328125,
+      "epoch": 2.802816901408451,
+      "grad_norm": 0.96875,
+      "learning_rate": 5e-05,
+      "loss": 1.1262364387512207,
+      "mean_token_accuracy": 0.7762812376022339,
+      "num_tokens": 1113091.0,
+      "step": 398
+    },
+    {
+      "entropy": 1.1015625,
+      "epoch": 2.8098591549295775,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5e-05,
+      "loss": 1.0824552774429321,
+      "mean_token_accuracy": 0.7742782235145569,
+      "num_tokens": 1116528.0,
+      "step": 399
+    },
+    {
+      "entropy": 1.1796875,
+      "epoch": 2.816901408450704,
+      "grad_norm": 0.875,
+      "learning_rate": 5e-05,
+      "loss": 1.1277594566345215,
+      "mean_token_accuracy": 0.7675762176513672,
+      "num_tokens": 1119651.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.1796875,
+      "epoch": 2.823943661971831,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 1.170253872871399,
+      "mean_token_accuracy": 0.7603855133056641,
+      "num_tokens": 1122668.0,
+      "step": 401
+    },
+    {
+      "entropy": 1.0234375,
+      "epoch": 2.830985915492958,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5e-05,
+      "loss": 1.0204730033874512,
+      "mean_token_accuracy": 0.7934080958366394,
+      "num_tokens": 1125619.0,
+      "step": 402
+    },
+    {
+      "entropy": 0.76953125,
+      "epoch": 2.8380281690140845,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5e-05,
+      "loss": 0.7402415871620178,
+      "mean_token_accuracy": 0.8549243807792664,
+      "num_tokens": 1128074.0,
+      "step": 403
+    },
+    {
+      "entropy": 1.03125,
+      "epoch": 2.845070422535211,
+      "grad_norm": 0.91015625,
+      "learning_rate": 5e-05,
+      "loss": 1.0297044515609741,
+      "mean_token_accuracy": 0.8018761873245239,
+      "num_tokens": 1130747.0,
+      "step": 404
+    },
+    {
+      "entropy": 0.765625,
+      "epoch": 2.852112676056338,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 0.7556635141372681,
+      "mean_token_accuracy": 0.8628131151199341,
+      "num_tokens": 1133350.0,
+      "step": 405
+    },
+    {
+      "entropy": 0.74609375,
+      "epoch": 2.859154929577465,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 0.6611390709877014,
+      "mean_token_accuracy": 0.8590927124023438,
+      "num_tokens": 1135849.0,
+      "step": 406
+    },
+    {
+      "entropy": 0.859375,
+      "epoch": 2.8661971830985915,
+      "grad_norm": 0.85546875,
+      "learning_rate": 5e-05,
+      "loss": 0.8384908437728882,
+      "mean_token_accuracy": 0.8369603753089905,
+      "num_tokens": 1138660.0,
+      "step": 407
+    },
+    {
+      "entropy": 0.984375,
+      "epoch": 2.873239436619718,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 0.9605262875556946,
+      "mean_token_accuracy": 0.809036374092102,
+      "num_tokens": 1141501.0,
+      "step": 408
+    },
+    {
+      "entropy": 1.046875,
+      "epoch": 2.880281690140845,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5e-05,
+      "loss": 1.0197346210479736,
+      "mean_token_accuracy": 0.8027757406234741,
+      "num_tokens": 1144247.0,
+      "step": 409
+    },
+    {
+      "entropy": 0.734375,
+      "epoch": 2.887323943661972,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5e-05,
+      "loss": 0.7191402912139893,
+      "mean_token_accuracy": 0.8674846887588501,
+      "num_tokens": 1146700.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.6875,
+      "epoch": 2.8943661971830985,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5e-05,
+      "loss": 0.6714113354682922,
+      "mean_token_accuracy": 0.8849679231643677,
+      "num_tokens": 1148890.0,
+      "step": 411
+    },
+    {
+      "entropy": 0.95703125,
+      "epoch": 2.9014084507042255,
+      "grad_norm": 0.84375,
+      "learning_rate": 5e-05,
+      "loss": 0.930925190448761,
+      "mean_token_accuracy": 0.809787929058075,
+      "num_tokens": 1151963.0,
+      "step": 412
+    },
+    {
+      "entropy": 0.90234375,
+      "epoch": 2.908450704225352,
+      "grad_norm": 0.828125,
+      "learning_rate": 5e-05,
+      "loss": 0.9057172536849976,
+      "mean_token_accuracy": 0.8207200765609741,
+      "num_tokens": 1154693.0,
+      "step": 413
+    },
+    {
+      "entropy": 0.95703125,
+      "epoch": 2.915492957746479,
+      "grad_norm": 0.828125,
+      "learning_rate": 5e-05,
+      "loss": 0.8770691156387329,
+      "mean_token_accuracy": 0.8179959058761597,
+      "num_tokens": 1158124.0,
+      "step": 414
+    },
+    {
+      "entropy": 1.1015625,
+      "epoch": 2.9225352112676055,
+      "grad_norm": 0.875,
+      "learning_rate": 5e-05,
+      "loss": 1.0502822399139404,
+      "mean_token_accuracy": 0.7834715843200684,
+      "num_tokens": 1161508.0,
+      "step": 415
+    },
+    {
+      "entropy": 1.1640625,
+      "epoch": 2.9295774647887325,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5e-05,
+      "loss": 1.1527771949768066,
+      "mean_token_accuracy": 0.785365879535675,
+      "num_tokens": 1164591.0,
+      "step": 416
+    },
+    {
+      "entropy": 1.125,
+      "epoch": 2.936619718309859,
+      "grad_norm": 0.83203125,
+      "learning_rate": 5e-05,
+      "loss": 1.095367193222046,
+      "mean_token_accuracy": 0.7831994891166687,
+      "num_tokens": 1167837.0,
+      "step": 417
+    },
+    {
+      "entropy": 0.9453125,
+      "epoch": 2.943661971830986,
+      "grad_norm": 0.94921875,
+      "learning_rate": 5e-05,
+      "loss": 0.9723469018936157,
+      "mean_token_accuracy": 0.8173178434371948,
+      "num_tokens": 1170686.0,
+      "step": 418
+    },
+    {
+      "entropy": 1.125,
+      "epoch": 2.9507042253521125,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5e-05,
+      "loss": 1.0903615951538086,
+      "mean_token_accuracy": 0.7964138984680176,
+      "num_tokens": 1173371.0,
+      "step": 419
+    },
+    {
+      "entropy": 1.0,
+      "epoch": 2.9577464788732395,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5e-05,
+      "loss": 1.0091233253479004,
+      "mean_token_accuracy": 0.8044652342796326,
+      "num_tokens": 1176156.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.1640625,
+      "epoch": 2.964788732394366,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 1.1239677667617798,
+      "mean_token_accuracy": 0.7919235229492188,
+      "num_tokens": 1179408.0,
+      "step": 421
+    },
+    {
+      "entropy": 1.0,
+      "epoch": 2.971830985915493,
+      "grad_norm": 0.8359375,
+      "learning_rate": 5e-05,
+      "loss": 0.9868678450584412,
+      "mean_token_accuracy": 0.7931495904922485,
+      "num_tokens": 1183153.0,
+      "step": 422
+    },
+    {
+      "entropy": 0.8046875,
+      "epoch": 2.97887323943662,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5e-05,
+      "loss": 0.7784821391105652,
+      "mean_token_accuracy": 0.8386287689208984,
+      "num_tokens": 1185553.0,
+      "step": 423
+    },
+    {
+      "entropy": 1.078125,
+      "epoch": 2.9859154929577465,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5e-05,
+      "loss": 1.0243606567382812,
+      "mean_token_accuracy": 0.8010540008544922,
+      "num_tokens": 1188597.0,
+      "step": 424
+    },
+    {
+      "entropy": 0.8671875,
+      "epoch": 2.992957746478873,
+      "grad_norm": 0.84375,
+      "learning_rate": 5e-05,
+      "loss": 0.8591823577880859,
+      "mean_token_accuracy": 0.838732123374939,
+      "num_tokens": 1191476.0,
+      "step": 425
+    },
+    {
+      "entropy": 1.1484375,
+      "epoch": 3.0,
+      "grad_norm": 0.80859375,
+      "learning_rate": 5e-05,
+      "loss": 1.1457493305206299,
+      "mean_token_accuracy": 0.7729924917221069,
+      "num_tokens": 1195083.0,
+      "step": 426
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.9166124131944444,
+      "eval_loss": 0.9163780808448792,
+      "eval_mean_token_accuracy": 0.8217091543806924,
+      "eval_num_tokens": 1195083.0,
+      "eval_runtime": 3.7209,
+      "eval_samples_per_second": 76.326,
+      "eval_steps_per_second": 9.675,
+      "step": 426
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 710,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1102096202858496.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}