diff --git "a/model/trainer_state.json" "b/model/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/model/trainer_state.json"
@@ -0,0 +1,39882 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.4999799269340399,
+  "eval_steps": 500,
+  "global_step": 49816,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00010036532980047372,
+      "learning_rate": 0.00019998193424063592,
+      "loss": 1.5103,
+      "mean_token_accuracy": 0.633021330833435,
+      "num_tokens": 33810.0,
+      "step": 10
+    },
+    {
+      "epoch": 0.00020073065960094744,
+      "learning_rate": 0.00019996186117467583,
+      "loss": 1.2727,
+      "mean_token_accuracy": 0.677767938375473,
+      "num_tokens": 68079.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.0003010959894014212,
+      "learning_rate": 0.00019994178810871574,
+      "loss": 1.1203,
+      "mean_token_accuracy": 0.7041535198688507,
+      "num_tokens": 102274.0,
+      "step": 30
+    },
+    {
+      "epoch": 0.0004014613192018949,
+      "learning_rate": 0.00019992171504275562,
+      "loss": 1.0614,
+      "mean_token_accuracy": 0.7179206430912017,
+      "num_tokens": 135784.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.0005018266490023686,
+      "learning_rate": 0.00019990164197679556,
+      "loss": 1.0812,
+      "mean_token_accuracy": 0.7120111465454102,
+      "num_tokens": 169612.0,
+      "step": 50
+    },
+    {
+      "epoch": 0.0006021919788028423,
+      "learning_rate": 0.00019988156891083544,
+      "loss": 1.0127,
+      "mean_token_accuracy": 0.7204696774482727,
+      "num_tokens": 203734.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.0007025573086033161,
+      "learning_rate": 0.00019986149584487535,
+      "loss": 0.9832,
+      "mean_token_accuracy": 0.7345041453838348,
+      "num_tokens": 237931.0,
+      "step": 70
+    },
+    {
+      "epoch": 0.0008029226384037898,
+      "learning_rate": 0.00019984142277891526,
+      "loss": 0.9677,
+      "mean_token_accuracy": 0.7350185573101043,
+      "num_tokens": 272629.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.0009032879682042635,
+      "learning_rate": 0.00019982134971295516,
+      "loss": 0.997,
+      "mean_token_accuracy": 0.7259666502475739,
+      "num_tokens": 306161.0,
+      "step": 90
+    },
+    {
+      "epoch": 0.0010036532980047372,
+      "learning_rate": 0.00019980127664699507,
+      "loss": 1.0068,
+      "mean_token_accuracy": 0.7253551065921784,
+      "num_tokens": 339884.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.001104018627805211,
+      "learning_rate": 0.00019978120358103498,
+      "loss": 0.9897,
+      "mean_token_accuracy": 0.7253098130226135,
+      "num_tokens": 373256.0,
+      "step": 110
+    },
+    {
+      "epoch": 0.0012043839576056847,
+      "learning_rate": 0.0001997611305150749,
+      "loss": 0.9673,
+      "mean_token_accuracy": 0.729346650838852,
+      "num_tokens": 407123.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.0013047492874061583,
+      "learning_rate": 0.00019974105744911477,
+      "loss": 0.9385,
+      "mean_token_accuracy": 0.7343133509159088,
+      "num_tokens": 441186.0,
+      "step": 130
+    },
+    {
+      "epoch": 0.0014051146172066322,
+      "learning_rate": 0.0001997209843831547,
+      "loss": 0.9334,
+      "mean_token_accuracy": 0.7334938704967499,
+      "num_tokens": 476076.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.0015054799470071059,
+      "learning_rate": 0.0001997009113171946,
+      "loss": 0.9647,
+      "mean_token_accuracy": 0.7314290404319763,
+      "num_tokens": 510012.0,
+      "step": 150
+    },
+    {
+      "epoch": 0.0016058452768075795,
+      "learning_rate": 0.0001996808382512345,
+      "loss": 0.9372,
+      "mean_token_accuracy": 0.7441855669021606,
+      "num_tokens": 543926.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.0017062106066080534,
+      "learning_rate": 0.0001996607651852744,
+      "loss": 0.9317,
+      "mean_token_accuracy": 0.7420060276985169,
+      "num_tokens": 577963.0,
+      "step": 170
+    },
+    {
+      "epoch": 0.001806575936408527,
+      "learning_rate": 0.00019964069211931431,
+      "loss": 0.9181,
+      "mean_token_accuracy": 0.7427789747714997,
+      "num_tokens": 611919.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.0019069412662090007,
+      "learning_rate": 0.00019962061905335422,
+      "loss": 0.9343,
+      "mean_token_accuracy": 0.7363346517086029,
+      "num_tokens": 645919.0,
+      "step": 190
+    },
+    {
+      "epoch": 0.0020073065960094743,
+      "learning_rate": 0.00019960054598739413,
+      "loss": 0.9205,
+      "mean_token_accuracy": 0.7433876693248749,
+      "num_tokens": 679456.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.002107671925809948,
+      "learning_rate": 0.000199580472921434,
+      "loss": 0.8805,
+      "mean_token_accuracy": 0.7510487496852875,
+      "num_tokens": 713397.0,
+      "step": 210
+    },
+    {
+      "epoch": 0.002208037255610422,
+      "learning_rate": 0.00019956039985547395,
+      "loss": 0.9355,
+      "mean_token_accuracy": 0.7372931838035583,
+      "num_tokens": 746693.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.0023084025854108957,
+      "learning_rate": 0.00019954032678951386,
+      "loss": 0.9282,
+      "mean_token_accuracy": 0.7390501201152802,
+      "num_tokens": 780480.0,
+      "step": 230
+    },
+    {
+      "epoch": 0.0024087679152113694,
+      "learning_rate": 0.00019952025372355374,
+      "loss": 0.8976,
+      "mean_token_accuracy": 0.7474514424800873,
+      "num_tokens": 814570.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.002509133245011843,
+      "learning_rate": 0.00019950018065759367,
+      "loss": 0.9076,
+      "mean_token_accuracy": 0.7378260135650635,
+      "num_tokens": 848776.0,
+      "step": 250
+    },
+    {
+      "epoch": 0.0026094985748123167,
+      "learning_rate": 0.00019948010759163355,
+      "loss": 0.9324,
+      "mean_token_accuracy": 0.7361941576004029,
+      "num_tokens": 881643.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.0027098639046127903,
+      "learning_rate": 0.00019946003452567346,
+      "loss": 0.9155,
+      "mean_token_accuracy": 0.7408255219459534,
+      "num_tokens": 915814.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.0028102292344132644,
+      "learning_rate": 0.00019943996145971337,
+      "loss": 0.8939,
+      "mean_token_accuracy": 0.7481561243534088,
+      "num_tokens": 950660.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.002910594564213738,
+      "learning_rate": 0.00019941988839375328,
+      "loss": 0.9425,
+      "mean_token_accuracy": 0.735199373960495,
+      "num_tokens": 984112.0,
+      "step": 290
+    },
+    {
+      "epoch": 0.0030109598940142117,
+      "learning_rate": 0.00019939981532779316,
+      "loss": 0.899,
+      "mean_token_accuracy": 0.743339866399765,
+      "num_tokens": 1018221.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.0031113252238146854,
+      "learning_rate": 0.0001993797422618331,
+      "loss": 0.9045,
+      "mean_token_accuracy": 0.7475386559963226,
+      "num_tokens": 1051883.0,
+      "step": 310
+    },
+    {
+      "epoch": 0.003211690553615159,
+      "learning_rate": 0.00019935966919587298,
+      "loss": 0.9073,
+      "mean_token_accuracy": 0.7443807065486908,
+      "num_tokens": 1086327.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.0033120558834156327,
+      "learning_rate": 0.00019933959612991289,
+      "loss": 0.9139,
+      "mean_token_accuracy": 0.7400292158126831,
+      "num_tokens": 1120924.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.0034124212132161068,
+      "learning_rate": 0.0001993195230639528,
+      "loss": 0.8737,
+      "mean_token_accuracy": 0.7494464218616486,
+      "num_tokens": 1154916.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.0035127865430165804,
+      "learning_rate": 0.0001992994499979927,
+      "loss": 0.8745,
+      "mean_token_accuracy": 0.7504120409488678,
+      "num_tokens": 1189348.0,
+      "step": 350
+    },
+    {
+      "epoch": 0.003613151872817054,
+      "learning_rate": 0.0001992793769320326,
+      "loss": 0.9187,
+      "mean_token_accuracy": 0.7410196900367737,
+      "num_tokens": 1223853.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.0037135172026175277,
+      "learning_rate": 0.00019925930386607252,
+      "loss": 0.8834,
+      "mean_token_accuracy": 0.749067735671997,
+      "num_tokens": 1257683.0,
+      "step": 370
+    },
+    {
+      "epoch": 0.0038138825324180014,
+      "learning_rate": 0.00019923923080011243,
+      "loss": 0.8981,
+      "mean_token_accuracy": 0.7470409989356994,
+      "num_tokens": 1292294.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.0039142478622184755,
+      "learning_rate": 0.0001992191577341523,
+      "loss": 0.9096,
+      "mean_token_accuracy": 0.7398540675640106,
+      "num_tokens": 1326683.0,
+      "step": 390
+    },
+    {
+      "epoch": 0.004014613192018949,
+      "learning_rate": 0.00019919908466819225,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.7491674959659577,
+      "num_tokens": 1360379.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.004114978521819423,
+      "learning_rate": 0.00019917901160223213,
+      "loss": 0.8677,
+      "mean_token_accuracy": 0.7554733157157898,
+      "num_tokens": 1394724.0,
+      "step": 410
+    },
+    {
+      "epoch": 0.004215343851619896,
+      "learning_rate": 0.00019915893853627203,
+      "loss": 0.894,
+      "mean_token_accuracy": 0.7477222561836243,
+      "num_tokens": 1429795.0,
+      "step": 420
+    },
+    {
+      "epoch": 0.00431570918142037,
+      "learning_rate": 0.00019913886547031194,
+      "loss": 0.8876,
+      "mean_token_accuracy": 0.747435849905014,
+      "num_tokens": 1463279.0,
+      "step": 430
+    },
+    {
+      "epoch": 0.004416074511220844,
+      "learning_rate": 0.00019911879240435185,
+      "loss": 0.9133,
+      "mean_token_accuracy": 0.7419423460960388,
+      "num_tokens": 1497656.0,
+      "step": 440
+    },
+    {
+      "epoch": 0.004516439841021317,
+      "learning_rate": 0.00019909871933839176,
+      "loss": 0.9105,
+      "mean_token_accuracy": 0.7445949018001556,
+      "num_tokens": 1531438.0,
+      "step": 450
+    },
+    {
+      "epoch": 0.0046168051708217915,
+      "learning_rate": 0.00019907864627243167,
+      "loss": 0.9147,
+      "mean_token_accuracy": 0.7372195899486542,
+      "num_tokens": 1564860.0,
+      "step": 460
+    },
+    {
+      "epoch": 0.004717170500622265,
+      "learning_rate": 0.00019905857320647158,
+      "loss": 0.8822,
+      "mean_token_accuracy": 0.7499520599842071,
+      "num_tokens": 1598515.0,
+      "step": 470
+    },
+    {
+      "epoch": 0.004817535830422739,
+      "learning_rate": 0.00019903850014051146,
+      "loss": 0.886,
+      "mean_token_accuracy": 0.7500568509101868,
+      "num_tokens": 1632684.0,
+      "step": 480
+    },
+    {
+      "epoch": 0.004917901160223213,
+      "learning_rate": 0.0001990184270745514,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7524400293827057,
+      "num_tokens": 1666653.0,
+      "step": 490
+    },
+    {
+      "epoch": 0.005018266490023686,
+      "learning_rate": 0.00019899835400859128,
+      "loss": 0.8954,
+      "mean_token_accuracy": 0.7409865975379943,
+      "num_tokens": 1700533.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.00511863181982416,
+      "learning_rate": 0.00019897828094263118,
+      "loss": 0.932,
+      "mean_token_accuracy": 0.7376771807670593,
+      "num_tokens": 1734093.0,
+      "step": 510
+    },
+    {
+      "epoch": 0.005218997149624633,
+      "learning_rate": 0.0001989582078766711,
+      "loss": 0.8562,
+      "mean_token_accuracy": 0.7520005285739899,
+      "num_tokens": 1767804.0,
+      "step": 520
+    },
+    {
+      "epoch": 0.0053193624794251075,
+      "learning_rate": 0.000198938134810711,
+      "loss": 0.8675,
+      "mean_token_accuracy": 0.7533258378505707,
+      "num_tokens": 1801441.0,
+      "step": 530
+    },
+    {
+      "epoch": 0.005419727809225581,
+      "learning_rate": 0.00019891806174475088,
+      "loss": 0.8802,
+      "mean_token_accuracy": 0.752521938085556,
+      "num_tokens": 1835240.0,
+      "step": 540
+    },
+    {
+      "epoch": 0.005520093139026055,
+      "learning_rate": 0.00019889798867879082,
+      "loss": 0.8772,
+      "mean_token_accuracy": 0.7515844464302063,
+      "num_tokens": 1868712.0,
+      "step": 550
+    },
+    {
+      "epoch": 0.005620458468826529,
+      "learning_rate": 0.0001988779156128307,
+      "loss": 0.8726,
+      "mean_token_accuracy": 0.7543687939643859,
+      "num_tokens": 1902913.0,
+      "step": 560
+    },
+    {
+      "epoch": 0.005720823798627002,
+      "learning_rate": 0.0001988578425468706,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7525355458259583,
+      "num_tokens": 1936217.0,
+      "step": 570
+    },
+    {
+      "epoch": 0.005821189128427476,
+      "learning_rate": 0.00019883776948091054,
+      "loss": 0.9133,
+      "mean_token_accuracy": 0.7481994688510895,
+      "num_tokens": 1969679.0,
+      "step": 580
+    },
+    {
+      "epoch": 0.005921554458227949,
+      "learning_rate": 0.00019881769641495042,
+      "loss": 0.8996,
+      "mean_token_accuracy": 0.7466781139373779,
+      "num_tokens": 2003835.0,
+      "step": 590
+    },
+    {
+      "epoch": 0.0060219197880284235,
+      "learning_rate": 0.00019879762334899033,
+      "loss": 0.9024,
+      "mean_token_accuracy": 0.7528743803501129,
+      "num_tokens": 2037954.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.0061222851178288976,
+      "learning_rate": 0.00019877755028303024,
+      "loss": 0.9052,
+      "mean_token_accuracy": 0.7427110850811005,
+      "num_tokens": 2072301.0,
+      "step": 610
+    },
+    {
+      "epoch": 0.006222650447629371,
+      "learning_rate": 0.00019875747721707015,
+      "loss": 0.886,
+      "mean_token_accuracy": 0.7461163103580475,
+      "num_tokens": 2106896.0,
+      "step": 620
+    },
+    {
+      "epoch": 0.006323015777429845,
+      "learning_rate": 0.00019873740415111003,
+      "loss": 0.9005,
+      "mean_token_accuracy": 0.7445556104183197,
+      "num_tokens": 2140725.0,
+      "step": 630
+    },
+    {
+      "epoch": 0.006423381107230318,
+      "learning_rate": 0.00019871733108514997,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7516779065132141,
+      "num_tokens": 2175119.0,
+      "step": 640
+    },
+    {
+      "epoch": 0.006523746437030792,
+      "learning_rate": 0.00019869725801918985,
+      "loss": 0.9102,
+      "mean_token_accuracy": 0.7461168050765992,
+      "num_tokens": 2209647.0,
+      "step": 650
+    },
+    {
+      "epoch": 0.006624111766831265,
+      "learning_rate": 0.00019867718495322976,
+      "loss": 0.8439,
+      "mean_token_accuracy": 0.756019902229309,
+      "num_tokens": 2243954.0,
+      "step": 660
+    },
+    {
+      "epoch": 0.0067244770966317395,
+      "learning_rate": 0.00019865711188726966,
+      "loss": 0.8892,
+      "mean_token_accuracy": 0.7439179301261902,
+      "num_tokens": 2278114.0,
+      "step": 670
+    },
+    {
+      "epoch": 0.0068248424264322136,
+      "learning_rate": 0.00019863703882130957,
+      "loss": 0.9091,
+      "mean_token_accuracy": 0.744629728794098,
+      "num_tokens": 2312064.0,
+      "step": 680
+    },
+    {
+      "epoch": 0.006925207756232687,
+      "learning_rate": 0.00019861696575534948,
+      "loss": 0.8235,
+      "mean_token_accuracy": 0.7626911222934722,
+      "num_tokens": 2345670.0,
+      "step": 690
+    },
+    {
+      "epoch": 0.007025573086033161,
+      "learning_rate": 0.0001985968926893894,
+      "loss": 0.8921,
+      "mean_token_accuracy": 0.7451446413993835,
+      "num_tokens": 2379617.0,
+      "step": 700
+    },
+    {
+      "epoch": 0.007125938415833634,
+      "learning_rate": 0.0001985768196234293,
+      "loss": 0.8448,
+      "mean_token_accuracy": 0.7558365881443023,
+      "num_tokens": 2414653.0,
+      "step": 710
+    },
+    {
+      "epoch": 0.007226303745634108,
+      "learning_rate": 0.00019855674655746918,
+      "loss": 0.8762,
+      "mean_token_accuracy": 0.7509781241416931,
+      "num_tokens": 2448981.0,
+      "step": 720
+    },
+    {
+      "epoch": 0.007326669075434582,
+      "learning_rate": 0.00019853667349150911,
+      "loss": 0.8753,
+      "mean_token_accuracy": 0.7493642330169678,
+      "num_tokens": 2483413.0,
+      "step": 730
+    },
+    {
+      "epoch": 0.0074270344052350555,
+      "learning_rate": 0.000198516600425549,
+      "loss": 0.8811,
+      "mean_token_accuracy": 0.7473379790782928,
+      "num_tokens": 2516947.0,
+      "step": 740
+    },
+    {
+      "epoch": 0.0075273997350355296,
+      "learning_rate": 0.00019849652735958893,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.7478324592113494,
+      "num_tokens": 2551303.0,
+      "step": 750
+    },
+    {
+      "epoch": 0.007627765064836003,
+      "learning_rate": 0.0001984764542936288,
+      "loss": 0.9073,
+      "mean_token_accuracy": 0.7422046661376953,
+      "num_tokens": 2585751.0,
+      "step": 760
+    },
+    {
+      "epoch": 0.007728130394636477,
+      "learning_rate": 0.00019845638122766872,
+      "loss": 0.8407,
+      "mean_token_accuracy": 0.7600900292396545,
+      "num_tokens": 2620871.0,
+      "step": 770
+    },
+    {
+      "epoch": 0.007828495724436951,
+      "learning_rate": 0.00019843630816170863,
+      "loss": 0.8666,
+      "mean_token_accuracy": 0.7502162098884583,
+      "num_tokens": 2654342.0,
+      "step": 780
+    },
+    {
+      "epoch": 0.007928861054237424,
+      "learning_rate": 0.00019841623509574854,
+      "loss": 0.8833,
+      "mean_token_accuracy": 0.7511953830718994,
+      "num_tokens": 2689101.0,
+      "step": 790
+    },
+    {
+      "epoch": 0.008029226384037897,
+      "learning_rate": 0.00019839616202978845,
+      "loss": 0.8698,
+      "mean_token_accuracy": 0.7504532694816589,
+      "num_tokens": 2723619.0,
+      "step": 800
+    },
+    {
+      "epoch": 0.008129591713838372,
+      "learning_rate": 0.00019837608896382836,
+      "loss": 0.8612,
+      "mean_token_accuracy": 0.7555593311786651,
+      "num_tokens": 2756956.0,
+      "step": 810
+    },
+    {
+      "epoch": 0.008229957043638846,
+      "learning_rate": 0.00019835601589786826,
+      "loss": 0.8864,
+      "mean_token_accuracy": 0.7532445669174195,
+      "num_tokens": 2790576.0,
+      "step": 820
+    },
+    {
+      "epoch": 0.008330322373439319,
+      "learning_rate": 0.00019833594283190815,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7430309534072876,
+      "num_tokens": 2823827.0,
+      "step": 830
+    },
+    {
+      "epoch": 0.008430687703239792,
+      "learning_rate": 0.00019831586976594808,
+      "loss": 0.8871,
+      "mean_token_accuracy": 0.7462802588939667,
+      "num_tokens": 2858359.0,
+      "step": 840
+    },
+    {
+      "epoch": 0.008531053033040267,
+      "learning_rate": 0.00019829579669998796,
+      "loss": 0.8695,
+      "mean_token_accuracy": 0.7541916847229004,
+      "num_tokens": 2892919.0,
+      "step": 850
+    },
+    {
+      "epoch": 0.00863141836284074,
+      "learning_rate": 0.00019827572363402787,
+      "loss": 0.8838,
+      "mean_token_accuracy": 0.7506275475025177,
+      "num_tokens": 2927106.0,
+      "step": 860
+    },
+    {
+      "epoch": 0.008731783692641213,
+      "learning_rate": 0.00019825565056806778,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7548351526260376,
+      "num_tokens": 2960466.0,
+      "step": 870
+    },
+    {
+      "epoch": 0.008832149022441688,
+      "learning_rate": 0.0001982355775021077,
+      "loss": 0.8557,
+      "mean_token_accuracy": 0.7527715265750885,
+      "num_tokens": 2993569.0,
+      "step": 880
+    },
+    {
+      "epoch": 0.008932514352242162,
+      "learning_rate": 0.00019821550443614757,
+      "loss": 0.8773,
+      "mean_token_accuracy": 0.7506238102912903,
+      "num_tokens": 3028181.0,
+      "step": 890
+    },
+    {
+      "epoch": 0.009032879682042635,
+      "learning_rate": 0.0001981954313701875,
+      "loss": 0.8718,
+      "mean_token_accuracy": 0.7520479619503021,
+      "num_tokens": 3061934.0,
+      "step": 900
+    },
+    {
+      "epoch": 0.00913324501184311,
+      "learning_rate": 0.00019817535830422739,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.7517975449562073,
+      "num_tokens": 3095089.0,
+      "step": 910
+    },
+    {
+      "epoch": 0.009233610341643583,
+      "learning_rate": 0.0001981552852382673,
+      "loss": 0.8492,
+      "mean_token_accuracy": 0.7479332447052002,
+      "num_tokens": 3129099.0,
+      "step": 920
+    },
+    {
+      "epoch": 0.009333975671444056,
+      "learning_rate": 0.00019813521217230723,
+      "loss": 0.8679,
+      "mean_token_accuracy": 0.749633401632309,
+      "num_tokens": 3162147.0,
+      "step": 930
+    },
+    {
+      "epoch": 0.00943434100124453,
+      "learning_rate": 0.0001981151391063471,
+      "loss": 0.8895,
+      "mean_token_accuracy": 0.7500181376934052,
+      "num_tokens": 3196315.0,
+      "step": 940
+    },
+    {
+      "epoch": 0.009534706331045004,
+      "learning_rate": 0.00019809506604038702,
+      "loss": 0.8793,
+      "mean_token_accuracy": 0.7507654368877411,
+      "num_tokens": 3230752.0,
+      "step": 950
+    },
+    {
+      "epoch": 0.009635071660845478,
+      "learning_rate": 0.00019807499297442693,
+      "loss": 0.848,
+      "mean_token_accuracy": 0.7560262858867646,
+      "num_tokens": 3264343.0,
+      "step": 960
+    },
+    {
+      "epoch": 0.00973543699064595,
+      "learning_rate": 0.00019805491990846684,
+      "loss": 0.8647,
+      "mean_token_accuracy": 0.7528027355670929,
+      "num_tokens": 3298563.0,
+      "step": 970
+    },
+    {
+      "epoch": 0.009835802320446426,
+      "learning_rate": 0.00019803484684250672,
+      "loss": 0.8385,
+      "mean_token_accuracy": 0.7556717336177826,
+      "num_tokens": 3332120.0,
+      "step": 980
+    },
+    {
+      "epoch": 0.009936167650246899,
+      "learning_rate": 0.00019801477377654665,
+      "loss": 0.8657,
+      "mean_token_accuracy": 0.7579919815063476,
+      "num_tokens": 3366253.0,
+      "step": 990
+    },
+    {
+      "epoch": 0.010036532980047372,
+      "learning_rate": 0.00019799470071058653,
+      "loss": 0.8592,
+      "mean_token_accuracy": 0.7515231788158416,
+      "num_tokens": 3399434.0,
+      "step": 1000
+    },
+    {
+      "epoch": 0.010136898309847845,
+      "learning_rate": 0.00019797462764462644,
+      "loss": 0.8361,
+      "mean_token_accuracy": 0.7587453186511993,
+      "num_tokens": 3433578.0,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01023726363964832,
+      "learning_rate": 0.00019795455457866635,
+      "loss": 0.834,
+      "mean_token_accuracy": 0.7555178880691529,
+      "num_tokens": 3467942.0,
+      "step": 1020
+    },
+    {
+      "epoch": 0.010337628969448794,
+      "learning_rate": 0.00019793448151270626,
+      "loss": 0.8569,
+      "mean_token_accuracy": 0.7506053686141968,
+      "num_tokens": 3502285.0,
+      "step": 1030
+    },
+    {
+      "epoch": 0.010437994299249267,
+      "learning_rate": 0.00019791440844674617,
+      "loss": 0.8725,
+      "mean_token_accuracy": 0.7484077334403991,
+      "num_tokens": 3536182.0,
+      "step": 1040
+    },
+    {
+      "epoch": 0.010538359629049742,
+      "learning_rate": 0.00019789433538078608,
+      "loss": 0.8061,
+      "mean_token_accuracy": 0.7640980839729309,
+      "num_tokens": 3570133.0,
+      "step": 1050
+    },
+    {
+      "epoch": 0.010638724958850215,
+      "learning_rate": 0.00019787426231482598,
+      "loss": 0.8913,
+      "mean_token_accuracy": 0.7447064459323883,
+      "num_tokens": 3604046.0,
+      "step": 1060
+    },
+    {
+      "epoch": 0.010739090288650688,
+      "learning_rate": 0.00019785418924886587,
+      "loss": 0.8695,
+      "mean_token_accuracy": 0.7490169048309326,
+      "num_tokens": 3638337.0,
+      "step": 1070
+    },
+    {
+      "epoch": 0.010839455618451161,
+      "learning_rate": 0.0001978341161829058,
+      "loss": 0.861,
+      "mean_token_accuracy": 0.7469825208187103,
+      "num_tokens": 3672695.0,
+      "step": 1080
+    },
+    {
+      "epoch": 0.010939820948251636,
+      "learning_rate": 0.00019781404311694568,
+      "loss": 0.8901,
+      "mean_token_accuracy": 0.7456781506538391,
+      "num_tokens": 3707086.0,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01104018627805211,
+      "learning_rate": 0.0001977939700509856,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.7588579893112183,
+      "num_tokens": 3740519.0,
+      "step": 1100
+    },
+    {
+      "epoch": 0.011140551607852583,
+      "learning_rate": 0.0001977738969850255,
+      "loss": 0.8728,
+      "mean_token_accuracy": 0.7522070944309235,
+      "num_tokens": 3774450.0,
+      "step": 1110
+    },
+    {
+      "epoch": 0.011240916937653058,
+      "learning_rate": 0.0001977538239190654,
+      "loss": 0.8837,
+      "mean_token_accuracy": 0.7503505706787109,
+      "num_tokens": 3808768.0,
+      "step": 1120
+    },
+    {
+      "epoch": 0.011341282267453531,
+      "learning_rate": 0.00019773375085310532,
+      "loss": 0.8359,
+      "mean_token_accuracy": 0.7603037297725678,
+      "num_tokens": 3843046.0,
+      "step": 1130
+    },
+    {
+      "epoch": 0.011441647597254004,
+      "learning_rate": 0.00019771367778714523,
+      "loss": 0.8845,
+      "mean_token_accuracy": 0.7438742220401764,
+      "num_tokens": 3876186.0,
+      "step": 1140
+    },
+    {
+      "epoch": 0.011542012927054479,
+      "learning_rate": 0.00019769360472118513,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7548528850078583,
+      "num_tokens": 3910140.0,
+      "step": 1150
+    },
+    {
+      "epoch": 0.011642378256854952,
+      "learning_rate": 0.00019767353165522502,
+      "loss": 0.8441,
+      "mean_token_accuracy": 0.7579459309577942,
+      "num_tokens": 3944252.0,
+      "step": 1160
+    },
+    {
+      "epoch": 0.011742743586655426,
+      "learning_rate": 0.00019765345858926495,
+      "loss": 0.8564,
+      "mean_token_accuracy": 0.7527571260929108,
+      "num_tokens": 3978821.0,
+      "step": 1170
+    },
+    {
+      "epoch": 0.011843108916455899,
+      "learning_rate": 0.00019763338552330483,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.7558556854724884,
+      "num_tokens": 4013127.0,
+      "step": 1180
+    },
+    {
+      "epoch": 0.011943474246256374,
+      "learning_rate": 0.00019761331245734474,
+      "loss": 0.8622,
+      "mean_token_accuracy": 0.7480232536792755,
+      "num_tokens": 4047220.0,
+      "step": 1190
+    },
+    {
+      "epoch": 0.012043839576056847,
+      "learning_rate": 0.00019759323939138465,
+      "loss": 0.8616,
+      "mean_token_accuracy": 0.7525281727313995,
+      "num_tokens": 4081046.0,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01214420490585732,
+      "learning_rate": 0.00019757316632542456,
+      "loss": 0.867,
+      "mean_token_accuracy": 0.748983871936798,
+      "num_tokens": 4115169.0,
+      "step": 1210
+    },
+    {
+      "epoch": 0.012244570235657795,
+      "learning_rate": 0.00019755309325946444,
+      "loss": 0.8851,
+      "mean_token_accuracy": 0.7456744372844696,
+      "num_tokens": 4149891.0,
+      "step": 1220
+    },
+    {
+      "epoch": 0.012344935565458268,
+      "learning_rate": 0.00019753302019350437,
+      "loss": 0.8321,
+      "mean_token_accuracy": 0.757876992225647,
+      "num_tokens": 4183258.0,
+      "step": 1230
+    },
+    {
+      "epoch": 0.012445300895258742,
+      "learning_rate": 0.00019751294712754426,
+      "loss": 0.8615,
+      "mean_token_accuracy": 0.7554275155067444,
+      "num_tokens": 4217359.0,
+      "step": 1240
+    },
+    {
+      "epoch": 0.012545666225059215,
+      "learning_rate": 0.00019749287406158416,
+      "loss": 0.856,
+      "mean_token_accuracy": 0.7524965465068817,
+      "num_tokens": 4251484.0,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01264603155485969,
+      "learning_rate": 0.00019747280099562407,
+      "loss": 0.8659,
+      "mean_token_accuracy": 0.7591691255569458,
+      "num_tokens": 4285841.0,
+      "step": 1260
+    },
+    {
+      "epoch": 0.012746396884660163,
+      "learning_rate": 0.00019745272792966398,
+      "loss": 0.8621,
+      "mean_token_accuracy": 0.7493607640266419,
+      "num_tokens": 4318785.0,
+      "step": 1270
+    },
+    {
+      "epoch": 0.012846762214460636,
+      "learning_rate": 0.0001974326548637039,
+      "loss": 0.8471,
+      "mean_token_accuracy": 0.7562038958072662,
+      "num_tokens": 4352516.0,
+      "step": 1280
+    },
+    {
+      "epoch": 0.012947127544261111,
+      "learning_rate": 0.0001974125817977438,
+      "loss": 0.8829,
+      "mean_token_accuracy": 0.7481812596321106,
+      "num_tokens": 4385434.0,
+      "step": 1290
+    },
+    {
+      "epoch": 0.013047492874061584,
+      "learning_rate": 0.0001973925087317837,
+      "loss": 0.882,
+      "mean_token_accuracy": 0.7556095838546752,
+      "num_tokens": 4419873.0,
+      "step": 1300
+    },
+    {
+      "epoch": 0.013147858203862058,
+      "learning_rate": 0.00019737243566582361,
+      "loss": 0.84,
+      "mean_token_accuracy": 0.7592958688735962,
+      "num_tokens": 4454568.0,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01324822353366253,
+      "learning_rate": 0.00019735236259986352,
+      "loss": 0.8413,
+      "mean_token_accuracy": 0.7540211915969849,
+      "num_tokens": 4488603.0,
+      "step": 1320
+    },
+    {
+      "epoch": 0.013348588863463006,
+      "learning_rate": 0.0001973322895339034,
+      "loss": 0.8775,
+      "mean_token_accuracy": 0.7487682044506073,
+      "num_tokens": 4522062.0,
+      "step": 1330
+    },
+    {
+      "epoch": 0.013448954193263479,
+      "learning_rate": 0.00019731221646794334,
+      "loss": 0.8799,
+      "mean_token_accuracy": 0.7497659981250763,
+      "num_tokens": 4556010.0,
+      "step": 1340
+    },
+    {
+      "epoch": 0.013549319523063952,
+      "learning_rate": 0.00019729214340198322,
+      "loss": 0.8476,
+      "mean_token_accuracy": 0.7548844814300537,
+      "num_tokens": 4589180.0,
+      "step": 1350
+    },
+    {
+      "epoch": 0.013649684852864427,
+      "learning_rate": 0.00019727207033602313,
+      "loss": 0.8683,
+      "mean_token_accuracy": 0.7509661138057708,
+      "num_tokens": 4621813.0,
+      "step": 1360
+    },
+    {
+      "epoch": 0.0137500501826649,
+      "learning_rate": 0.00019725199727006304,
+      "loss": 0.8475,
+      "mean_token_accuracy": 0.7586887240409851,
+      "num_tokens": 4655223.0,
+      "step": 1370
+    },
+    {
+      "epoch": 0.013850415512465374,
+      "learning_rate": 0.00019723192420410295,
+      "loss": 0.8623,
+      "mean_token_accuracy": 0.7530815541744232,
+      "num_tokens": 4688942.0,
+      "step": 1380
+    },
+    {
+      "epoch": 0.013950780842265849,
+      "learning_rate": 0.00019721185113814285,
+      "loss": 0.8928,
+      "mean_token_accuracy": 0.7447740018367768,
+      "num_tokens": 4722711.0,
+      "step": 1390
+    },
+    {
+      "epoch": 0.014051146172066322,
+      "learning_rate": 0.00019719177807218276,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7587530732154846,
+      "num_tokens": 4756986.0,
+      "step": 1400
+    },
+    {
+      "epoch": 0.014151511501866795,
+      "learning_rate": 0.00019717170500622267,
+      "loss": 0.8618,
+      "mean_token_accuracy": 0.7547706186771392,
+      "num_tokens": 4790634.0,
+      "step": 1410
+    },
+    {
+      "epoch": 0.014251876831667268,
+      "learning_rate": 0.00019715163194026255,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.7552112579345703,
+      "num_tokens": 4824610.0,
+      "step": 1420
+    },
+    {
+      "epoch": 0.014352242161467743,
+      "learning_rate": 0.0001971315588743025,
+      "loss": 0.8669,
+      "mean_token_accuracy": 0.7512627184391022,
+      "num_tokens": 4858592.0,
+      "step": 1430
+    },
+    {
+      "epoch": 0.014452607491268216,
+      "learning_rate": 0.00019711148580834237,
+      "loss": 0.8912,
+      "mean_token_accuracy": 0.743174260854721,
+      "num_tokens": 4891961.0,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01455297282106869,
+      "learning_rate": 0.00019709141274238228,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7583801984786988,
+      "num_tokens": 4925217.0,
+      "step": 1450
+    },
+    {
+      "epoch": 0.014653338150869165,
+      "learning_rate": 0.0001970713396764222,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7593223810195923,
+      "num_tokens": 4958103.0,
+      "step": 1460
+    },
+    {
+      "epoch": 0.014753703480669638,
+      "learning_rate": 0.0001970512666104621,
+      "loss": 0.8856,
+      "mean_token_accuracy": 0.7498394906520843,
+      "num_tokens": 4991826.0,
+      "step": 1470
+    },
+    {
+      "epoch": 0.014854068810470111,
+      "learning_rate": 0.000197031193544502,
+      "loss": 0.8543,
+      "mean_token_accuracy": 0.7504751682281494,
+      "num_tokens": 5024795.0,
+      "step": 1480
+    },
+    {
+      "epoch": 0.014954434140270584,
+      "learning_rate": 0.0001970111204785419,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7564951300621032,
+      "num_tokens": 5058977.0,
+      "step": 1490
+    },
+    {
+      "epoch": 0.015054799470071059,
+      "learning_rate": 0.00019699104741258182,
+      "loss": 0.8904,
+      "mean_token_accuracy": 0.7444755256175994,
+      "num_tokens": 5093915.0,
+      "step": 1500
+    },
+    {
+      "epoch": 0.015155164799871532,
+      "learning_rate": 0.0001969709743466217,
+      "loss": 0.8556,
+      "mean_token_accuracy": 0.7554027199745178,
+      "num_tokens": 5128235.0,
+      "step": 1510
+    },
+    {
+      "epoch": 0.015255530129672006,
+      "learning_rate": 0.00019695090128066164,
+      "loss": 0.8499,
+      "mean_token_accuracy": 0.7527681648731231,
+      "num_tokens": 5163167.0,
+      "step": 1520
+    },
+    {
+      "epoch": 0.01535589545947248,
+      "learning_rate": 0.00019693082821470152,
+      "loss": 0.8481,
+      "mean_token_accuracy": 0.7494793772697449,
+      "num_tokens": 5197445.0,
+      "step": 1530
+    },
+    {
+      "epoch": 0.015456260789272954,
+      "learning_rate": 0.00019691075514874143,
+      "loss": 0.8476,
+      "mean_token_accuracy": 0.7548111736774444,
+      "num_tokens": 5231705.0,
+      "step": 1540
+    },
+    {
+      "epoch": 0.015556626119073427,
+      "learning_rate": 0.00019689068208278134,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.7536470830440521,
+      "num_tokens": 5266220.0,
+      "step": 1550
+    },
+    {
+      "epoch": 0.015656991448873902,
+      "learning_rate": 0.00019687060901682124,
+      "loss": 0.8563,
+      "mean_token_accuracy": 0.753504478931427,
+      "num_tokens": 5300098.0,
+      "step": 1560
+    },
+    {
+      "epoch": 0.015757356778674375,
+      "learning_rate": 0.00019685053595086113,
+      "loss": 0.8322,
+      "mean_token_accuracy": 0.7565286755561829,
+      "num_tokens": 5334353.0,
+      "step": 1570
+    },
+    {
+      "epoch": 0.01585772210847485,
+      "learning_rate": 0.00019683046288490106,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7614298403263092,
+      "num_tokens": 5368411.0,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01595808743827532,
+      "learning_rate": 0.00019681038981894094,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.755518627166748,
+      "num_tokens": 5402236.0,
+      "step": 1590
+    },
+    {
+      "epoch": 0.016058452768075795,
+      "learning_rate": 0.00019679031675298085,
+      "loss": 0.8521,
+      "mean_token_accuracy": 0.7516325056552887,
+      "num_tokens": 5436012.0,
+      "step": 1600
+    },
+    {
+      "epoch": 0.016158818097876268,
+      "learning_rate": 0.00019677024368702076,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7498721599578857,
+      "num_tokens": 5470369.0,
+      "step": 1610
+    },
+    {
+      "epoch": 0.016259183427676745,
+      "learning_rate": 0.00019675017062106067,
+      "loss": 0.8265,
+      "mean_token_accuracy": 0.7556636273860932,
+      "num_tokens": 5504487.0,
+      "step": 1620
+    },
+    {
+      "epoch": 0.016359548757477218,
+      "learning_rate": 0.00019673009755510058,
+      "loss": 0.8231,
+      "mean_token_accuracy": 0.75765740275383,
+      "num_tokens": 5538679.0,
+      "step": 1630
+    },
+    {
+      "epoch": 0.01645991408727769,
+      "learning_rate": 0.00019671002448914048,
+      "loss": 0.8756,
+      "mean_token_accuracy": 0.7481065690517426,
+      "num_tokens": 5573500.0,
+      "step": 1640
+    },
+    {
+      "epoch": 0.016560279417078164,
+      "learning_rate": 0.0001966899514231804,
+      "loss": 0.8371,
+      "mean_token_accuracy": 0.7557444095611572,
+      "num_tokens": 5607047.0,
+      "step": 1650
+    },
+    {
+      "epoch": 0.016660644746878638,
+      "learning_rate": 0.00019666987835722027,
+      "loss": 0.8114,
+      "mean_token_accuracy": 0.7623445272445679,
+      "num_tokens": 5641450.0,
+      "step": 1660
+    },
+    {
+      "epoch": 0.01676101007667911,
+      "learning_rate": 0.0001966498052912602,
+      "loss": 0.8551,
+      "mean_token_accuracy": 0.7489903330802917,
+      "num_tokens": 5674856.0,
+      "step": 1670
+    },
+    {
+      "epoch": 0.016861375406479584,
+      "learning_rate": 0.0001966297322253001,
+      "loss": 0.8021,
+      "mean_token_accuracy": 0.7655105769634247,
+      "num_tokens": 5708295.0,
+      "step": 1680
+    },
+    {
+      "epoch": 0.01696174073628006,
+      "learning_rate": 0.00019660965915934,
+      "loss": 0.8195,
+      "mean_token_accuracy": 0.7606033682823181,
+      "num_tokens": 5743318.0,
+      "step": 1690
+    },
+    {
+      "epoch": 0.017062106066080534,
+      "learning_rate": 0.0001965895860933799,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.7644929349422455,
+      "num_tokens": 5776744.0,
+      "step": 1700
+    },
+    {
+      "epoch": 0.017162471395881007,
+      "learning_rate": 0.00019656951302741982,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7594692170619964,
+      "num_tokens": 5810694.0,
+      "step": 1710
+    },
+    {
+      "epoch": 0.01726283672568148,
+      "learning_rate": 0.00019654943996145972,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.7587471783161164,
+      "num_tokens": 5844392.0,
+      "step": 1720
+    },
+    {
+      "epoch": 0.017363202055481954,
+      "learning_rate": 0.00019652936689549963,
+      "loss": 0.8481,
+      "mean_token_accuracy": 0.7577917516231537,
+      "num_tokens": 5878431.0,
+      "step": 1730
+    },
+    {
+      "epoch": 0.017463567385282427,
+      "learning_rate": 0.00019650929382953954,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.7509515821933747,
+      "num_tokens": 5912731.0,
+      "step": 1740
+    },
+    {
+      "epoch": 0.017563932715082903,
+      "learning_rate": 0.00019648922076357942,
+      "loss": 0.8096,
+      "mean_token_accuracy": 0.7576501190662384,
+      "num_tokens": 5947053.0,
+      "step": 1750
+    },
+    {
+      "epoch": 0.017664298044883377,
+      "learning_rate": 0.00019646914769761936,
+      "loss": 0.8436,
+      "mean_token_accuracy": 0.7566677570343018,
+      "num_tokens": 5980908.0,
+      "step": 1760
+    },
+    {
+      "epoch": 0.01776466337468385,
+      "learning_rate": 0.00019644907463165924,
+      "loss": 0.8523,
+      "mean_token_accuracy": 0.7544825732707977,
+      "num_tokens": 6014535.0,
+      "step": 1770
+    },
+    {
+      "epoch": 0.017865028704484323,
+      "learning_rate": 0.00019642900156569915,
+      "loss": 0.8506,
+      "mean_token_accuracy": 0.7533735692501068,
+      "num_tokens": 6048756.0,
+      "step": 1780
+    },
+    {
+      "epoch": 0.017965394034284796,
+      "learning_rate": 0.00019640892849973906,
+      "loss": 0.8556,
+      "mean_token_accuracy": 0.752546352148056,
+      "num_tokens": 6082752.0,
+      "step": 1790
+    },
+    {
+      "epoch": 0.01806575936408527,
+      "learning_rate": 0.00019638885543377897,
+      "loss": 0.8173,
+      "mean_token_accuracy": 0.7617668688297272,
+      "num_tokens": 6116502.0,
+      "step": 1800
+    },
+    {
+      "epoch": 0.018166124693885743,
+      "learning_rate": 0.00019636878236781885,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.7598871469497681,
+      "num_tokens": 6150826.0,
+      "step": 1810
+    },
+    {
+      "epoch": 0.01826649002368622,
+      "learning_rate": 0.00019634870930185878,
+      "loss": 0.8139,
+      "mean_token_accuracy": 0.7583870530128479,
+      "num_tokens": 6185500.0,
+      "step": 1820
+    },
+    {
+      "epoch": 0.018366855353486693,
+      "learning_rate": 0.0001963286362358987,
+      "loss": 0.8235,
+      "mean_token_accuracy": 0.7577989399433136,
+      "num_tokens": 6219531.0,
+      "step": 1830
+    },
+    {
+      "epoch": 0.018467220683287166,
+      "learning_rate": 0.00019630856316993857,
+      "loss": 0.851,
+      "mean_token_accuracy": 0.7582792520523072,
+      "num_tokens": 6253286.0,
+      "step": 1840
+    },
+    {
+      "epoch": 0.01856758601308764,
+      "learning_rate": 0.0001962884901039785,
+      "loss": 0.8811,
+      "mean_token_accuracy": 0.7497155666351318,
+      "num_tokens": 6287058.0,
+      "step": 1850
+    },
+    {
+      "epoch": 0.018667951342888112,
+      "learning_rate": 0.0001962684170380184,
+      "loss": 0.8278,
+      "mean_token_accuracy": 0.7637346804141998,
+      "num_tokens": 6321137.0,
+      "step": 1860
+    },
+    {
+      "epoch": 0.018768316672688586,
+      "learning_rate": 0.0001962483439720583,
+      "loss": 0.8389,
+      "mean_token_accuracy": 0.7530131101608276,
+      "num_tokens": 6355801.0,
+      "step": 1870
+    },
+    {
+      "epoch": 0.01886868200248906,
+      "learning_rate": 0.0001962282709060982,
+      "loss": 0.8367,
+      "mean_token_accuracy": 0.7559464156627655,
+      "num_tokens": 6388657.0,
+      "step": 1880
+    },
+    {
+      "epoch": 0.018969047332289535,
+      "learning_rate": 0.00019620819784013811,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7582254111766815,
+      "num_tokens": 6422783.0,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01906941266209001,
+      "learning_rate": 0.00019618812477417802,
+      "loss": 0.8133,
+      "mean_token_accuracy": 0.7625649392604827,
+      "num_tokens": 6457489.0,
+      "step": 1900
+    },
+    {
+      "epoch": 0.019169777991890482,
+      "learning_rate": 0.00019616805170821793,
+      "loss": 0.8093,
+      "mean_token_accuracy": 0.7617171883583069,
+      "num_tokens": 6491382.0,
+      "step": 1910
+    },
+    {
+      "epoch": 0.019270143321690955,
+      "learning_rate": 0.0001961479786422578,
+      "loss": 0.8071,
+      "mean_token_accuracy": 0.7619174361228943,
+      "num_tokens": 6525576.0,
+      "step": 1920
+    },
+    {
+      "epoch": 0.01937050865149143,
+      "learning_rate": 0.00019612790557629775,
+      "loss": 0.8224,
+      "mean_token_accuracy": 0.7651695072650909,
+      "num_tokens": 6559225.0,
+      "step": 1930
+    },
+    {
+      "epoch": 0.0194708739812919,
+      "learning_rate": 0.00019610783251033763,
+      "loss": 0.855,
+      "mean_token_accuracy": 0.7547216534614563,
+      "num_tokens": 6593907.0,
+      "step": 1940
+    },
+    {
+      "epoch": 0.019571239311092375,
+      "learning_rate": 0.00019608775944437754,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7579889893531799,
+      "num_tokens": 6628286.0,
+      "step": 1950
+    },
+    {
+      "epoch": 0.01967160464089285,
+      "learning_rate": 0.00019606768637841745,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.7559162437915802,
+      "num_tokens": 6662442.0,
+      "step": 1960
+    },
+    {
+      "epoch": 0.019771969970693325,
+      "learning_rate": 0.00019604761331245735,
+      "loss": 0.8528,
+      "mean_token_accuracy": 0.7507839500904083,
+      "num_tokens": 6696329.0,
+      "step": 1970
+    },
+    {
+      "epoch": 0.019872335300493798,
+      "learning_rate": 0.00019602754024649726,
+      "loss": 0.8579,
+      "mean_token_accuracy": 0.7551623106002807,
+      "num_tokens": 6730903.0,
+      "step": 1980
+    },
+    {
+      "epoch": 0.01997270063029427,
+      "learning_rate": 0.00019600746718053717,
+      "loss": 0.8307,
+      "mean_token_accuracy": 0.7526181817054749,
+      "num_tokens": 6763865.0,
+      "step": 1990
+    },
+    {
+      "epoch": 0.020073065960094744,
+      "learning_rate": 0.00019598739411457708,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.7611814558506012,
+      "num_tokens": 6798513.0,
+      "step": 2000
+    },
+    {
+      "epoch": 0.020173431289895218,
+      "learning_rate": 0.00019596732104861696,
+      "loss": 0.8208,
+      "mean_token_accuracy": 0.7589188933372497,
+      "num_tokens": 6832618.0,
+      "step": 2010
+    },
+    {
+      "epoch": 0.02027379661969569,
+      "learning_rate": 0.0001959472479826569,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.749679434299469,
+      "num_tokens": 6866166.0,
+      "step": 2020
+    },
+    {
+      "epoch": 0.020374161949496167,
+      "learning_rate": 0.00019592717491669678,
+      "loss": 0.8788,
+      "mean_token_accuracy": 0.7494403779506683,
+      "num_tokens": 6899501.0,
+      "step": 2030
+    },
+    {
+      "epoch": 0.02047452727929664,
+      "learning_rate": 0.00019590710185073669,
+      "loss": 0.8044,
+      "mean_token_accuracy": 0.7660996496677399,
+      "num_tokens": 6933758.0,
+      "step": 2040
+    },
+    {
+      "epoch": 0.020574892609097114,
+      "learning_rate": 0.0001958870287847766,
+      "loss": 0.8321,
+      "mean_token_accuracy": 0.7600698232650757,
+      "num_tokens": 6967869.0,
+      "step": 2050
+    },
+    {
+      "epoch": 0.020675257938897587,
+      "learning_rate": 0.0001958669557188165,
+      "loss": 0.8138,
+      "mean_token_accuracy": 0.7660489320755005,
+      "num_tokens": 7002140.0,
+      "step": 2060
+    },
+    {
+      "epoch": 0.02077562326869806,
+      "learning_rate": 0.0001958468826528564,
+      "loss": 0.8206,
+      "mean_token_accuracy": 0.7583360552787781,
+      "num_tokens": 7035964.0,
+      "step": 2070
+    },
+    {
+      "epoch": 0.020875988598498534,
+      "learning_rate": 0.00019582680958689632,
+      "loss": 0.8539,
+      "mean_token_accuracy": 0.7558043360710144,
+      "num_tokens": 7069864.0,
+      "step": 2080
+    },
+    {
+      "epoch": 0.020976353928299007,
+      "learning_rate": 0.00019580673652093623,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.754914653301239,
+      "num_tokens": 7104474.0,
+      "step": 2090
+    },
+    {
+      "epoch": 0.021076719258099483,
+      "learning_rate": 0.0001957866634549761,
+      "loss": 0.8469,
+      "mean_token_accuracy": 0.75110724568367,
+      "num_tokens": 7139229.0,
+      "step": 2100
+    },
+    {
+      "epoch": 0.021177084587899957,
+      "learning_rate": 0.00019576659038901605,
+      "loss": 0.8357,
+      "mean_token_accuracy": 0.755988621711731,
+      "num_tokens": 7173015.0,
+      "step": 2110
+    },
+    {
+      "epoch": 0.02127744991770043,
+      "learning_rate": 0.00019574651732305593,
+      "loss": 0.8501,
+      "mean_token_accuracy": 0.75484619140625,
+      "num_tokens": 7206350.0,
+      "step": 2120
+    },
+    {
+      "epoch": 0.021377815247500903,
+      "learning_rate": 0.00019572644425709584,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.7622098982334137,
+      "num_tokens": 7240468.0,
+      "step": 2130
+    },
+    {
+      "epoch": 0.021478180577301376,
+      "learning_rate": 0.00019570637119113574,
+      "loss": 0.8346,
+      "mean_token_accuracy": 0.7651883065700531,
+      "num_tokens": 7274498.0,
+      "step": 2140
+    },
+    {
+      "epoch": 0.02157854590710185,
+      "learning_rate": 0.00019568629812517565,
+      "loss": 0.8617,
+      "mean_token_accuracy": 0.7555427491664887,
+      "num_tokens": 7307634.0,
+      "step": 2150
+    },
+    {
+      "epoch": 0.021678911236902323,
+      "learning_rate": 0.00019566622505921553,
+      "loss": 0.8433,
+      "mean_token_accuracy": 0.7512418270111084,
+      "num_tokens": 7341779.0,
+      "step": 2160
+    },
+    {
+      "epoch": 0.0217792765667028,
+      "learning_rate": 0.00019564615199325547,
+      "loss": 0.8554,
+      "mean_token_accuracy": 0.753146630525589,
+      "num_tokens": 7375947.0,
+      "step": 2170
+    },
+    {
+      "epoch": 0.021879641896503273,
+      "learning_rate": 0.00019562607892729538,
+      "loss": 0.8531,
+      "mean_token_accuracy": 0.7535173892974854,
+      "num_tokens": 7410782.0,
+      "step": 2180
+    },
+    {
+      "epoch": 0.021980007226303746,
+      "learning_rate": 0.00019560600586133526,
+      "loss": 0.8368,
+      "mean_token_accuracy": 0.7546063840389252,
+      "num_tokens": 7443934.0,
+      "step": 2190
+    },
+    {
+      "epoch": 0.02208037255610422,
+      "learning_rate": 0.0001955859327953752,
+      "loss": 0.8564,
+      "mean_token_accuracy": 0.756107634305954,
+      "num_tokens": 7478420.0,
+      "step": 2200
+    },
+    {
+      "epoch": 0.022180737885904692,
+      "learning_rate": 0.00019556585972941508,
+      "loss": 0.8275,
+      "mean_token_accuracy": 0.7583844363689423,
+      "num_tokens": 7512780.0,
+      "step": 2210
+    },
+    {
+      "epoch": 0.022281103215705166,
+      "learning_rate": 0.00019554578666345498,
+      "loss": 0.8519,
+      "mean_token_accuracy": 0.7547212541103363,
+      "num_tokens": 7547171.0,
+      "step": 2220
+    },
+    {
+      "epoch": 0.022381468545505642,
+      "learning_rate": 0.0001955257135974949,
+      "loss": 0.8197,
+      "mean_token_accuracy": 0.757226973772049,
+      "num_tokens": 7581359.0,
+      "step": 2230
+    },
+    {
+      "epoch": 0.022481833875306115,
+      "learning_rate": 0.0001955056405315348,
+      "loss": 0.8461,
+      "mean_token_accuracy": 0.7656774163246155,
+      "num_tokens": 7615372.0,
+      "step": 2240
+    },
+    {
+      "epoch": 0.02258219920510659,
+      "learning_rate": 0.00019548556746557468,
+      "loss": 0.8458,
+      "mean_token_accuracy": 0.7544702410697937,
+      "num_tokens": 7649479.0,
+      "step": 2250
+    },
+    {
+      "epoch": 0.022682564534907062,
+      "learning_rate": 0.00019546549439961462,
+      "loss": 0.8542,
+      "mean_token_accuracy": 0.7525253236293793,
+      "num_tokens": 7683870.0,
+      "step": 2260
+    },
+    {
+      "epoch": 0.022782929864707535,
+      "learning_rate": 0.0001954454213336545,
+      "loss": 0.8423,
+      "mean_token_accuracy": 0.7568866968154907,
+      "num_tokens": 7717550.0,
+      "step": 2270
+    },
+    {
+      "epoch": 0.02288329519450801,
+      "learning_rate": 0.0001954253482676944,
+      "loss": 0.822,
+      "mean_token_accuracy": 0.7563924074172974,
+      "num_tokens": 7751661.0,
+      "step": 2280
+    },
+    {
+      "epoch": 0.02298366052430848,
+      "learning_rate": 0.00019540527520173432,
+      "loss": 0.8308,
+      "mean_token_accuracy": 0.7572165787220001,
+      "num_tokens": 7786291.0,
+      "step": 2290
+    },
+    {
+      "epoch": 0.023084025854108958,
+      "learning_rate": 0.00019538520213577422,
+      "loss": 0.793,
+      "mean_token_accuracy": 0.7661900997161866,
+      "num_tokens": 7820722.0,
+      "step": 2300
+    },
+    {
+      "epoch": 0.02318439118390943,
+      "learning_rate": 0.00019536512906981413,
+      "loss": 0.8043,
+      "mean_token_accuracy": 0.7659907758235931,
+      "num_tokens": 7855180.0,
+      "step": 2310
+    },
+    {
+      "epoch": 0.023284756513709905,
+      "learning_rate": 0.00019534505600385404,
+      "loss": 0.7967,
+      "mean_token_accuracy": 0.7661457180976867,
+      "num_tokens": 7888552.0,
+      "step": 2320
+    },
+    {
+      "epoch": 0.023385121843510378,
+      "learning_rate": 0.00019532498293789395,
+      "loss": 0.8339,
+      "mean_token_accuracy": 0.7558695316314697,
+      "num_tokens": 7921169.0,
+      "step": 2330
+    },
+    {
+      "epoch": 0.02348548717331085,
+      "learning_rate": 0.00019530490987193383,
+      "loss": 0.8095,
+      "mean_token_accuracy": 0.76091068983078,
+      "num_tokens": 7955024.0,
+      "step": 2340
+    },
+    {
+      "epoch": 0.023585852503111324,
+      "learning_rate": 0.00019528483680597377,
+      "loss": 0.8205,
+      "mean_token_accuracy": 0.7617243111133576,
+      "num_tokens": 7988653.0,
+      "step": 2350
+    },
+    {
+      "epoch": 0.023686217832911798,
+      "learning_rate": 0.00019526476374001365,
+      "loss": 0.826,
+      "mean_token_accuracy": 0.7584989905357361,
+      "num_tokens": 8022319.0,
+      "step": 2360
+    },
+    {
+      "epoch": 0.023786583162712274,
+      "learning_rate": 0.00019524469067405356,
+      "loss": 0.827,
+      "mean_token_accuracy": 0.7621460676193237,
+      "num_tokens": 8055876.0,
+      "step": 2370
+    },
+    {
+      "epoch": 0.023886948492512747,
+      "learning_rate": 0.00019522461760809346,
+      "loss": 0.797,
+      "mean_token_accuracy": 0.7639261364936829,
+      "num_tokens": 8089693.0,
+      "step": 2380
+    },
+    {
+      "epoch": 0.02398731382231322,
+      "learning_rate": 0.00019520454454213337,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.7636609375476837,
+      "num_tokens": 8124115.0,
+      "step": 2390
+    },
+    {
+      "epoch": 0.024087679152113694,
+      "learning_rate": 0.00019518447147617328,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7648244261741638,
+      "num_tokens": 8158023.0,
+      "step": 2400
+    },
+    {
+      "epoch": 0.024188044481914167,
+      "learning_rate": 0.0001951643984102132,
+      "loss": 0.8149,
+      "mean_token_accuracy": 0.7608605980873108,
+      "num_tokens": 8192278.0,
+      "step": 2410
+    },
+    {
+      "epoch": 0.02428840981171464,
+      "learning_rate": 0.0001951443253442531,
+      "loss": 0.8373,
+      "mean_token_accuracy": 0.755948281288147,
+      "num_tokens": 8226602.0,
+      "step": 2420
+    },
+    {
+      "epoch": 0.024388775141515114,
+      "learning_rate": 0.00019512425227829298,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.7608046889305115,
+      "num_tokens": 8260675.0,
+      "step": 2430
+    },
+    {
+      "epoch": 0.02448914047131559,
+      "learning_rate": 0.00019510417921233292,
+      "loss": 0.7979,
+      "mean_token_accuracy": 0.7635708749294281,
+      "num_tokens": 8294714.0,
+      "step": 2440
+    },
+    {
+      "epoch": 0.024589505801116063,
+      "learning_rate": 0.0001950841061463728,
+      "loss": 0.853,
+      "mean_token_accuracy": 0.7571332097053528,
+      "num_tokens": 8328444.0,
+      "step": 2450
+    },
+    {
+      "epoch": 0.024689871130916537,
+      "learning_rate": 0.00019506403308041273,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.744694834947586,
+      "num_tokens": 8362891.0,
+      "step": 2460
+    },
+    {
+      "epoch": 0.02479023646071701,
+      "learning_rate": 0.0001950439600144526,
+      "loss": 0.8332,
+      "mean_token_accuracy": 0.7588581085205078,
+      "num_tokens": 8397433.0,
+      "step": 2470
+    },
+    {
+      "epoch": 0.024890601790517483,
+      "learning_rate": 0.00019502388694849252,
+      "loss": 0.8089,
+      "mean_token_accuracy": 0.7658759117126465,
+      "num_tokens": 8431278.0,
+      "step": 2480
+    },
+    {
+      "epoch": 0.024990967120317956,
+      "learning_rate": 0.00019500381388253243,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.759179824590683,
+      "num_tokens": 8465674.0,
+      "step": 2490
+    },
+    {
+      "epoch": 0.02509133245011843,
+      "learning_rate": 0.00019498374081657234,
+      "loss": 0.8507,
+      "mean_token_accuracy": 0.7583933115005493,
+      "num_tokens": 8499186.0,
+      "step": 2500
+    },
+    {
+      "epoch": 0.025191697779918906,
+      "learning_rate": 0.00019496366775061222,
+      "loss": 0.8371,
+      "mean_token_accuracy": 0.7609995067119598,
+      "num_tokens": 8533547.0,
+      "step": 2510
+    },
+    {
+      "epoch": 0.02529206310971938,
+      "learning_rate": 0.00019494359468465216,
+      "loss": 0.8217,
+      "mean_token_accuracy": 0.7622859716415405,
+      "num_tokens": 8568232.0,
+      "step": 2520
+    },
+    {
+      "epoch": 0.025392428439519853,
+      "learning_rate": 0.00019492352161869206,
+      "loss": 0.8346,
+      "mean_token_accuracy": 0.7621785938739777,
+      "num_tokens": 8602586.0,
+      "step": 2530
+    },
+    {
+      "epoch": 0.025492793769320326,
+      "learning_rate": 0.00019490344855273195,
+      "loss": 0.8109,
+      "mean_token_accuracy": 0.7586685180664062,
+      "num_tokens": 8636640.0,
+      "step": 2540
+    },
+    {
+      "epoch": 0.0255931590991208,
+      "learning_rate": 0.00019488337548677188,
+      "loss": 0.8157,
+      "mean_token_accuracy": 0.7647418260574341,
+      "num_tokens": 8670420.0,
+      "step": 2550
+    },
+    {
+      "epoch": 0.025693524428921272,
+      "learning_rate": 0.00019486330242081176,
+      "loss": 0.8071,
+      "mean_token_accuracy": 0.7660414576530457,
+      "num_tokens": 8704404.0,
+      "step": 2560
+    },
+    {
+      "epoch": 0.025793889758721746,
+      "learning_rate": 0.00019484322935485167,
+      "loss": 0.8118,
+      "mean_token_accuracy": 0.763026088476181,
+      "num_tokens": 8738337.0,
+      "step": 2570
+    },
+    {
+      "epoch": 0.025894255088522222,
+      "learning_rate": 0.00019482315628889158,
+      "loss": 0.8186,
+      "mean_token_accuracy": 0.7586533546447753,
+      "num_tokens": 8772155.0,
+      "step": 2580
+    },
+    {
+      "epoch": 0.025994620418322695,
+      "learning_rate": 0.0001948030832229315,
+      "loss": 0.8472,
+      "mean_token_accuracy": 0.7485378384590149,
+      "num_tokens": 8805291.0,
+      "step": 2590
+    },
+    {
+      "epoch": 0.02609498574812317,
+      "learning_rate": 0.00019478301015697137,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.7523237645626069,
+      "num_tokens": 8838352.0,
+      "step": 2600
+    },
+    {
+      "epoch": 0.026195351077923642,
+      "learning_rate": 0.0001947629370910113,
+      "loss": 0.834,
+      "mean_token_accuracy": 0.7580128014087677,
+      "num_tokens": 8871752.0,
+      "step": 2610
+    },
+    {
+      "epoch": 0.026295716407724115,
+      "learning_rate": 0.00019474286402505119,
+      "loss": 0.8085,
+      "mean_token_accuracy": 0.7614288806915284,
+      "num_tokens": 8906014.0,
+      "step": 2620
+    },
+    {
+      "epoch": 0.02639608173752459,
+      "learning_rate": 0.0001947227909590911,
+      "loss": 0.8158,
+      "mean_token_accuracy": 0.7616944968700409,
+      "num_tokens": 8941204.0,
+      "step": 2630
+    },
+    {
+      "epoch": 0.02649644706732506,
+      "learning_rate": 0.000194702717893131,
+      "loss": 0.8104,
+      "mean_token_accuracy": 0.7650873243808747,
+      "num_tokens": 8974391.0,
+      "step": 2640
+    },
+    {
+      "epoch": 0.026596812397125538,
+      "learning_rate": 0.0001946826448271709,
+      "loss": 0.7893,
+      "mean_token_accuracy": 0.7691265761852264,
+      "num_tokens": 9008066.0,
+      "step": 2650
+    },
+    {
+      "epoch": 0.02669717772692601,
+      "learning_rate": 0.00019466257176121082,
+      "loss": 0.7948,
+      "mean_token_accuracy": 0.7700909733772278,
+      "num_tokens": 9042380.0,
+      "step": 2660
+    },
+    {
+      "epoch": 0.026797543056726485,
+      "learning_rate": 0.00019464249869525073,
+      "loss": 0.8391,
+      "mean_token_accuracy": 0.7572174370288849,
+      "num_tokens": 9076897.0,
+      "step": 2670
+    },
+    {
+      "epoch": 0.026897908386526958,
+      "learning_rate": 0.00019462242562929064,
+      "loss": 0.8382,
+      "mean_token_accuracy": 0.7565006077289581,
+      "num_tokens": 9110755.0,
+      "step": 2680
+    },
+    {
+      "epoch": 0.02699827371632743,
+      "learning_rate": 0.00019460235256333052,
+      "loss": 0.8443,
+      "mean_token_accuracy": 0.7534610092639923,
+      "num_tokens": 9144780.0,
+      "step": 2690
+    },
+    {
+      "epoch": 0.027098639046127904,
+      "learning_rate": 0.00019458227949737045,
+      "loss": 0.8376,
+      "mean_token_accuracy": 0.751280415058136,
+      "num_tokens": 9178297.0,
+      "step": 2700
+    },
+    {
+      "epoch": 0.02719900437592838,
+      "learning_rate": 0.00019456220643141033,
+      "loss": 0.8161,
+      "mean_token_accuracy": 0.7588511765003204,
+      "num_tokens": 9212281.0,
+      "step": 2710
+    },
+    {
+      "epoch": 0.027299369705728854,
+      "learning_rate": 0.00019454213336545024,
+      "loss": 0.8257,
+      "mean_token_accuracy": 0.7590669870376587,
+      "num_tokens": 9246800.0,
+      "step": 2720
+    },
+    {
+      "epoch": 0.027399735035529327,
+      "learning_rate": 0.00019452206029949015,
+      "loss": 0.8108,
+      "mean_token_accuracy": 0.7625915348529816,
+      "num_tokens": 9280862.0,
+      "step": 2730
+    },
+    {
+      "epoch": 0.0275001003653298,
+      "learning_rate": 0.00019450198723353006,
+      "loss": 0.7889,
+      "mean_token_accuracy": 0.7649468660354615,
+      "num_tokens": 9316013.0,
+      "step": 2740
+    },
+    {
+      "epoch": 0.027600465695130274,
+      "learning_rate": 0.00019448191416756997,
+      "loss": 0.8283,
+      "mean_token_accuracy": 0.7558753788471222,
+      "num_tokens": 9349913.0,
+      "step": 2750
+    },
+    {
+      "epoch": 0.027700831024930747,
+      "learning_rate": 0.00019446184110160988,
+      "loss": 0.7758,
+      "mean_token_accuracy": 0.7654128670692444,
+      "num_tokens": 9384092.0,
+      "step": 2760
+    },
+    {
+      "epoch": 0.02780119635473122,
+      "learning_rate": 0.00019444176803564979,
+      "loss": 0.7991,
+      "mean_token_accuracy": 0.7651127219200134,
+      "num_tokens": 9418417.0,
+      "step": 2770
+    },
+    {
+      "epoch": 0.027901561684531697,
+      "learning_rate": 0.00019442169496968967,
+      "loss": 0.8285,
+      "mean_token_accuracy": 0.7555812895298004,
+      "num_tokens": 9453247.0,
+      "step": 2780
+    },
+    {
+      "epoch": 0.02800192701433217,
+      "learning_rate": 0.0001944016219037296,
+      "loss": 0.8166,
+      "mean_token_accuracy": 0.758303964138031,
+      "num_tokens": 9487349.0,
+      "step": 2790
+    },
+    {
+      "epoch": 0.028102292344132643,
+      "learning_rate": 0.00019438154883776948,
+      "loss": 0.7975,
+      "mean_token_accuracy": 0.7672683656215668,
+      "num_tokens": 9521217.0,
+      "step": 2800
+    },
+    {
+      "epoch": 0.028202657673933117,
+      "learning_rate": 0.0001943614757718094,
+      "loss": 0.8159,
+      "mean_token_accuracy": 0.7600618541240692,
+      "num_tokens": 9554836.0,
+      "step": 2810
+    },
+    {
+      "epoch": 0.02830302300373359,
+      "learning_rate": 0.0001943414027058493,
+      "loss": 0.8429,
+      "mean_token_accuracy": 0.7548216879367828,
+      "num_tokens": 9589253.0,
+      "step": 2820
+    },
+    {
+      "epoch": 0.028403388333534063,
+      "learning_rate": 0.0001943213296398892,
+      "loss": 0.8392,
+      "mean_token_accuracy": 0.7594896078109741,
+      "num_tokens": 9623090.0,
+      "step": 2830
+    },
+    {
+      "epoch": 0.028503753663334536,
+      "learning_rate": 0.0001943012565739291,
+      "loss": 0.8061,
+      "mean_token_accuracy": 0.7643302738666534,
+      "num_tokens": 9656910.0,
+      "step": 2840
+    },
+    {
+      "epoch": 0.028604118993135013,
+      "learning_rate": 0.00019428118350796903,
+      "loss": 0.8187,
+      "mean_token_accuracy": 0.7644733011722564,
+      "num_tokens": 9689992.0,
+      "step": 2850
+    },
+    {
+      "epoch": 0.028704484322935486,
+      "learning_rate": 0.0001942611104420089,
+      "loss": 0.8526,
+      "mean_token_accuracy": 0.7513968527317048,
+      "num_tokens": 9723413.0,
+      "step": 2860
+    },
+    {
+      "epoch": 0.02880484965273596,
+      "learning_rate": 0.00019424103737604882,
+      "loss": 0.8517,
+      "mean_token_accuracy": 0.7544140577316284,
+      "num_tokens": 9757325.0,
+      "step": 2870
+    },
+    {
+      "epoch": 0.028905214982536433,
+      "learning_rate": 0.00019422096431008875,
+      "loss": 0.8198,
+      "mean_token_accuracy": 0.7595404148101806,
+      "num_tokens": 9791668.0,
+      "step": 2880
+    },
+    {
+      "epoch": 0.029005580312336906,
+      "learning_rate": 0.00019420089124412863,
+      "loss": 0.8164,
+      "mean_token_accuracy": 0.7569047451019287,
+      "num_tokens": 9826222.0,
+      "step": 2890
+    },
+    {
+      "epoch": 0.02910594564213738,
+      "learning_rate": 0.00019418081817816854,
+      "loss": 0.8688,
+      "mean_token_accuracy": 0.7560630202293396,
+      "num_tokens": 9860177.0,
+      "step": 2900
+    },
+    {
+      "epoch": 0.029206310971937852,
+      "learning_rate": 0.00019416074511220845,
+      "loss": 0.8121,
+      "mean_token_accuracy": 0.7619596660137177,
+      "num_tokens": 9894213.0,
+      "step": 2910
+    },
+    {
+      "epoch": 0.02930667630173833,
+      "learning_rate": 0.00019414067204624836,
+      "loss": 0.7963,
+      "mean_token_accuracy": 0.7662225306034088,
+      "num_tokens": 9927913.0,
+      "step": 2920
+    },
+    {
+      "epoch": 0.029407041631538802,
+      "learning_rate": 0.00019412059898028824,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7606053411960602,
+      "num_tokens": 9961850.0,
+      "step": 2930
+    },
+    {
+      "epoch": 0.029507406961339275,
+      "learning_rate": 0.00019410052591432817,
+      "loss": 0.8381,
+      "mean_token_accuracy": 0.7542055785655976,
+      "num_tokens": 9996592.0,
+      "step": 2940
+    },
+    {
+      "epoch": 0.02960777229113975,
+      "learning_rate": 0.00019408045284836806,
+      "loss": 0.8097,
+      "mean_token_accuracy": 0.7627164542675018,
+      "num_tokens": 10030623.0,
+      "step": 2950
+    },
+    {
+      "epoch": 0.029708137620940222,
+      "learning_rate": 0.00019406037978240796,
+      "loss": 0.8179,
+      "mean_token_accuracy": 0.759533429145813,
+      "num_tokens": 10065371.0,
+      "step": 2960
+    },
+    {
+      "epoch": 0.029808502950740695,
+      "learning_rate": 0.00019404030671644787,
+      "loss": 0.7864,
+      "mean_token_accuracy": 0.76891188621521,
+      "num_tokens": 10098693.0,
+      "step": 2970
+    },
+    {
+      "epoch": 0.02990886828054117,
+      "learning_rate": 0.00019402023365048778,
+      "loss": 0.8138,
+      "mean_token_accuracy": 0.7631476044654846,
+      "num_tokens": 10132077.0,
+      "step": 2980
+    },
+    {
+      "epoch": 0.030009233610341645,
+      "learning_rate": 0.0001940001605845277,
+      "loss": 0.8085,
+      "mean_token_accuracy": 0.7628714203834533,
+      "num_tokens": 10167034.0,
+      "step": 2990
+    },
+    {
+      "epoch": 0.030109598940142118,
+      "learning_rate": 0.0001939800875185676,
+      "loss": 0.8067,
+      "mean_token_accuracy": 0.76093710064888,
+      "num_tokens": 10200994.0,
+      "step": 3000
+    },
+    {
+      "epoch": 0.03020996426994259,
+      "learning_rate": 0.0001939600144526075,
+      "loss": 0.8056,
+      "mean_token_accuracy": 0.7656333386898041,
+      "num_tokens": 10235556.0,
+      "step": 3010
+    },
+    {
+      "epoch": 0.030310329599743065,
+      "learning_rate": 0.00019393994138664741,
+      "loss": 0.8413,
+      "mean_token_accuracy": 0.7463635027408599,
+      "num_tokens": 10269354.0,
+      "step": 3020
+    },
+    {
+      "epoch": 0.030410694929543538,
+      "learning_rate": 0.00019391986832068732,
+      "loss": 0.8107,
+      "mean_token_accuracy": 0.7647533237934112,
+      "num_tokens": 10303415.0,
+      "step": 3030
+    },
+    {
+      "epoch": 0.03051106025934401,
+      "learning_rate": 0.0001938997952547272,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7526327192783355,
+      "num_tokens": 10336602.0,
+      "step": 3040
+    },
+    {
+      "epoch": 0.030611425589144484,
+      "learning_rate": 0.00019387972218876714,
+      "loss": 0.8229,
+      "mean_token_accuracy": 0.7578138887882233,
+      "num_tokens": 10369955.0,
+      "step": 3050
+    },
+    {
+      "epoch": 0.03071179091894496,
+      "learning_rate": 0.00019385964912280702,
+      "loss": 0.8348,
+      "mean_token_accuracy": 0.7555415332317352,
+      "num_tokens": 10404298.0,
+      "step": 3060
+    },
+    {
+      "epoch": 0.030812156248745434,
+      "learning_rate": 0.00019383957605684693,
+      "loss": 0.8314,
+      "mean_token_accuracy": 0.7591325759887695,
+      "num_tokens": 10438557.0,
+      "step": 3070
+    },
+    {
+      "epoch": 0.030912521578545907,
+      "learning_rate": 0.00019381950299088684,
+      "loss": 0.8409,
+      "mean_token_accuracy": 0.7593413949012756,
+      "num_tokens": 10472895.0,
+      "step": 3080
+    },
+    {
+      "epoch": 0.03101288690834638,
+      "learning_rate": 0.00019379942992492675,
+      "loss": 0.8414,
+      "mean_token_accuracy": 0.7592377960681915,
+      "num_tokens": 10507101.0,
+      "step": 3090
+    },
+    {
+      "epoch": 0.031113252238146854,
+      "learning_rate": 0.00019377935685896666,
+      "loss": 0.7824,
+      "mean_token_accuracy": 0.7731334388256073,
+      "num_tokens": 10541341.0,
+      "step": 3100
+    },
+    {
+      "epoch": 0.031213617567947327,
+      "learning_rate": 0.00019375928379300656,
+      "loss": 0.8333,
+      "mean_token_accuracy": 0.7570551693439483,
+      "num_tokens": 10575589.0,
+      "step": 3110
+    },
+    {
+      "epoch": 0.031313982897747804,
+      "learning_rate": 0.00019373921072704647,
+      "loss": 0.8084,
+      "mean_token_accuracy": 0.7645867586135864,
+      "num_tokens": 10609867.0,
+      "step": 3120
+    },
+    {
+      "epoch": 0.031414348227548274,
+      "learning_rate": 0.00019371913766108635,
+      "loss": 0.8437,
+      "mean_token_accuracy": 0.7533444285392761,
+      "num_tokens": 10643442.0,
+      "step": 3130
+    },
+    {
+      "epoch": 0.03151471355734875,
+      "learning_rate": 0.0001936990645951263,
+      "loss": 0.8101,
+      "mean_token_accuracy": 0.7585695147514343,
+      "num_tokens": 10677750.0,
+      "step": 3140
+    },
+    {
+      "epoch": 0.03161507888714922,
+      "learning_rate": 0.00019367899152916617,
+      "loss": 0.8215,
+      "mean_token_accuracy": 0.7583079099655151,
+      "num_tokens": 10711656.0,
+      "step": 3150
+    },
+    {
+      "epoch": 0.0317154442169497,
+      "learning_rate": 0.00019365891846320608,
+      "loss": 0.8354,
+      "mean_token_accuracy": 0.7559507787227631,
+      "num_tokens": 10745944.0,
+      "step": 3160
+    },
+    {
+      "epoch": 0.03181580954675017,
+      "learning_rate": 0.000193638845397246,
+      "loss": 0.8082,
+      "mean_token_accuracy": 0.7597402095794678,
+      "num_tokens": 10780421.0,
+      "step": 3170
+    },
+    {
+      "epoch": 0.03191617487655064,
+      "learning_rate": 0.0001936187723312859,
+      "loss": 0.8305,
+      "mean_token_accuracy": 0.7550306975841522,
+      "num_tokens": 10813665.0,
+      "step": 3180
+    },
+    {
+      "epoch": 0.03201654020635112,
+      "learning_rate": 0.00019359869926532578,
+      "loss": 0.8174,
+      "mean_token_accuracy": 0.7608563065528869,
+      "num_tokens": 10848641.0,
+      "step": 3190
+    },
+    {
+      "epoch": 0.03211690553615159,
+      "learning_rate": 0.0001935786261993657,
+      "loss": 0.8205,
+      "mean_token_accuracy": 0.7681922614574432,
+      "num_tokens": 10883127.0,
+      "step": 3200
+    },
+    {
+      "epoch": 0.032217270865952066,
+      "learning_rate": 0.0001935585531334056,
+      "loss": 0.8385,
+      "mean_token_accuracy": 0.7587725281715393,
+      "num_tokens": 10916869.0,
+      "step": 3210
+    },
+    {
+      "epoch": 0.032317636195752536,
+      "learning_rate": 0.0001935384800674455,
+      "loss": 0.8125,
+      "mean_token_accuracy": 0.7589336574077606,
+      "num_tokens": 10951318.0,
+      "step": 3220
+    },
+    {
+      "epoch": 0.03241800152555301,
+      "learning_rate": 0.00019351840700148544,
+      "loss": 0.8238,
+      "mean_token_accuracy": 0.7592109203338623,
+      "num_tokens": 10985495.0,
+      "step": 3230
+    },
+    {
+      "epoch": 0.03251836685535349,
+      "learning_rate": 0.00019349833393552532,
+      "loss": 0.8172,
+      "mean_token_accuracy": 0.7625975668430328,
+      "num_tokens": 11018921.0,
+      "step": 3240
+    },
+    {
+      "epoch": 0.03261873218515396,
+      "learning_rate": 0.00019347826086956523,
+      "loss": 0.8189,
+      "mean_token_accuracy": 0.7608864843845368,
+      "num_tokens": 11053086.0,
+      "step": 3250
+    },
+    {
+      "epoch": 0.032719097514954436,
+      "learning_rate": 0.00019345818780360514,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7709313690662384,
+      "num_tokens": 11086654.0,
+      "step": 3260
+    },
+    {
+      "epoch": 0.032819462844754906,
+      "learning_rate": 0.00019343811473764504,
+      "loss": 0.8214,
+      "mean_token_accuracy": 0.7595510005950927,
+      "num_tokens": 11120931.0,
+      "step": 3270
+    },
+    {
+      "epoch": 0.03291982817455538,
+      "learning_rate": 0.00019341804167168493,
+      "loss": 0.8331,
+      "mean_token_accuracy": 0.75942742228508,
+      "num_tokens": 11154413.0,
+      "step": 3280
+    },
+    {
+      "epoch": 0.03302019350435585,
+      "learning_rate": 0.00019339796860572486,
+      "loss": 0.7951,
+      "mean_token_accuracy": 0.7642982721328735,
+      "num_tokens": 11188382.0,
+      "step": 3290
+    },
+    {
+      "epoch": 0.03312055883415633,
+      "learning_rate": 0.00019337789553976474,
+      "loss": 0.8003,
+      "mean_token_accuracy": 0.7682574689388275,
+      "num_tokens": 11222513.0,
+      "step": 3300
+    },
+    {
+      "epoch": 0.033220924163956805,
+      "learning_rate": 0.00019335782247380465,
+      "loss": 0.7966,
+      "mean_token_accuracy": 0.7670561671257019,
+      "num_tokens": 11257447.0,
+      "step": 3310
+    },
+    {
+      "epoch": 0.033321289493757275,
+      "learning_rate": 0.00019333774940784456,
+      "loss": 0.8133,
+      "mean_token_accuracy": 0.7631431698799134,
+      "num_tokens": 11290812.0,
+      "step": 3320
+    },
+    {
+      "epoch": 0.03342165482355775,
+      "learning_rate": 0.00019331767634188447,
+      "loss": 0.8435,
+      "mean_token_accuracy": 0.7573554813861847,
+      "num_tokens": 11324627.0,
+      "step": 3330
+    },
+    {
+      "epoch": 0.03352202015335822,
+      "learning_rate": 0.00019329760327592438,
+      "loss": 0.8221,
+      "mean_token_accuracy": 0.7549627840518951,
+      "num_tokens": 11358198.0,
+      "step": 3340
+    },
+    {
+      "epoch": 0.0336223854831587,
+      "learning_rate": 0.00019327753020996428,
+      "loss": 0.79,
+      "mean_token_accuracy": 0.7678623855113983,
+      "num_tokens": 11392606.0,
+      "step": 3350
+    },
+    {
+      "epoch": 0.03372275081295917,
+      "learning_rate": 0.0001932574571440042,
+      "loss": 0.8377,
+      "mean_token_accuracy": 0.7548870742321014,
+      "num_tokens": 11427108.0,
+      "step": 3360
+    },
+    {
+      "epoch": 0.033823116142759645,
+      "learning_rate": 0.00019323738407804407,
+      "loss": 0.821,
+      "mean_token_accuracy": 0.7596335887908936,
+      "num_tokens": 11460882.0,
+      "step": 3370
+    },
+    {
+      "epoch": 0.03392348147256012,
+      "learning_rate": 0.000193217311012084,
+      "loss": 0.8233,
+      "mean_token_accuracy": 0.7633667171001435,
+      "num_tokens": 11495166.0,
+      "step": 3380
+    },
+    {
+      "epoch": 0.03402384680236059,
+      "learning_rate": 0.0001931972379461239,
+      "loss": 0.8403,
+      "mean_token_accuracy": 0.7576093554496766,
+      "num_tokens": 11528914.0,
+      "step": 3390
+    },
+    {
+      "epoch": 0.03412421213216107,
+      "learning_rate": 0.0001931771648801638,
+      "loss": 0.8033,
+      "mean_token_accuracy": 0.7655123353004456,
+      "num_tokens": 11562638.0,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03422457746196154,
+      "learning_rate": 0.0001931570918142037,
+      "loss": 0.8692,
+      "mean_token_accuracy": 0.7498386919498443,
+      "num_tokens": 11596992.0,
+      "step": 3410
+    },
+    {
+      "epoch": 0.034324942791762014,
+      "learning_rate": 0.00019313701874824362,
+      "loss": 0.8059,
+      "mean_token_accuracy": 0.7620149791240692,
+      "num_tokens": 11630807.0,
+      "step": 3420
+    },
+    {
+      "epoch": 0.034425308121562484,
+      "learning_rate": 0.00019311694568228353,
+      "loss": 0.8336,
+      "mean_token_accuracy": 0.7583272874355316,
+      "num_tokens": 11664527.0,
+      "step": 3430
+    },
+    {
+      "epoch": 0.03452567345136296,
+      "learning_rate": 0.00019309687261632343,
+      "loss": 0.8161,
+      "mean_token_accuracy": 0.7626073122024536,
+      "num_tokens": 11698423.0,
+      "step": 3440
+    },
+    {
+      "epoch": 0.03462603878116344,
+      "learning_rate": 0.00019307679955036334,
+      "loss": 0.7976,
+      "mean_token_accuracy": 0.7633013188838959,
+      "num_tokens": 11732303.0,
+      "step": 3450
+    },
+    {
+      "epoch": 0.03472640411096391,
+      "learning_rate": 0.00019305672648440322,
+      "loss": 0.8361,
+      "mean_token_accuracy": 0.7670786142349243,
+      "num_tokens": 11766441.0,
+      "step": 3460
+    },
+    {
+      "epoch": 0.034826769440764384,
+      "learning_rate": 0.00019303665341844316,
+      "loss": 0.8558,
+      "mean_token_accuracy": 0.7520733714103699,
+      "num_tokens": 11800699.0,
+      "step": 3470
+    },
+    {
+      "epoch": 0.034927134770564854,
+      "learning_rate": 0.00019301658035248304,
+      "loss": 0.8169,
+      "mean_token_accuracy": 0.7605992496013642,
+      "num_tokens": 11834888.0,
+      "step": 3480
+    },
+    {
+      "epoch": 0.03502750010036533,
+      "learning_rate": 0.00019299650728652295,
+      "loss": 0.8118,
+      "mean_token_accuracy": 0.7634969353675842,
+      "num_tokens": 11869188.0,
+      "step": 3490
+    },
+    {
+      "epoch": 0.03512786543016581,
+      "learning_rate": 0.00019297643422056286,
+      "loss": 0.8435,
+      "mean_token_accuracy": 0.7566763997077942,
+      "num_tokens": 11903158.0,
+      "step": 3500
+    },
+    {
+      "epoch": 0.03522823075996628,
+      "learning_rate": 0.00019295636115460277,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.7639656007289887,
+      "num_tokens": 11936631.0,
+      "step": 3510
+    },
+    {
+      "epoch": 0.03532859608976675,
+      "learning_rate": 0.00019293628808864265,
+      "loss": 0.8302,
+      "mean_token_accuracy": 0.7581159353256226,
+      "num_tokens": 11971564.0,
+      "step": 3520
+    },
+    {
+      "epoch": 0.03542896141956722,
+      "learning_rate": 0.00019291621502268258,
+      "loss": 0.7918,
+      "mean_token_accuracy": 0.771466726064682,
+      "num_tokens": 12005548.0,
+      "step": 3530
+    },
+    {
+      "epoch": 0.0355293267493677,
+      "learning_rate": 0.00019289614195672246,
+      "loss": 0.8442,
+      "mean_token_accuracy": 0.7578944623470306,
+      "num_tokens": 12039494.0,
+      "step": 3540
+    },
+    {
+      "epoch": 0.03562969207916817,
+      "learning_rate": 0.00019287606889076237,
+      "loss": 0.8083,
+      "mean_token_accuracy": 0.7562072098255157,
+      "num_tokens": 12073597.0,
+      "step": 3550
+    },
+    {
+      "epoch": 0.035730057408968646,
+      "learning_rate": 0.00019285599582480228,
+      "loss": 0.7956,
+      "mean_token_accuracy": 0.769059157371521,
+      "num_tokens": 12107546.0,
+      "step": 3560
+    },
+    {
+      "epoch": 0.03583042273876912,
+      "learning_rate": 0.0001928359227588422,
+      "loss": 0.8408,
+      "mean_token_accuracy": 0.7596532464027405,
+      "num_tokens": 12141065.0,
+      "step": 3570
+    },
+    {
+      "epoch": 0.03593078806856959,
+      "learning_rate": 0.0001928158496928821,
+      "loss": 0.8233,
+      "mean_token_accuracy": 0.7560832560062408,
+      "num_tokens": 12174410.0,
+      "step": 3580
+    },
+    {
+      "epoch": 0.03603115339837007,
+      "learning_rate": 0.000192795776626922,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.7569872558116912,
+      "num_tokens": 12207748.0,
+      "step": 3590
+    },
+    {
+      "epoch": 0.03613151872817054,
+      "learning_rate": 0.00019277570356096191,
+      "loss": 0.8257,
+      "mean_token_accuracy": 0.7610264837741851,
+      "num_tokens": 12241312.0,
+      "step": 3600
+    },
+    {
+      "epoch": 0.036231884057971016,
+      "learning_rate": 0.00019275563049500182,
+      "loss": 0.8012,
+      "mean_token_accuracy": 0.7594651341438293,
+      "num_tokens": 12274973.0,
+      "step": 3610
+    },
+    {
+      "epoch": 0.036332249387771486,
+      "learning_rate": 0.00019273555742904173,
+      "loss": 0.8264,
+      "mean_token_accuracy": 0.7608615458011627,
+      "num_tokens": 12308354.0,
+      "step": 3620
+    },
+    {
+      "epoch": 0.03643261471757196,
+      "learning_rate": 0.0001927154843630816,
+      "loss": 0.8471,
+      "mean_token_accuracy": 0.7569738268852234,
+      "num_tokens": 12341595.0,
+      "step": 3630
+    },
+    {
+      "epoch": 0.03653298004737244,
+      "learning_rate": 0.00019269541129712155,
+      "loss": 0.8124,
+      "mean_token_accuracy": 0.7671969056129455,
+      "num_tokens": 12375909.0,
+      "step": 3640
+    },
+    {
+      "epoch": 0.03663334537717291,
+      "learning_rate": 0.00019267533823116143,
+      "loss": 0.83,
+      "mean_token_accuracy": 0.7595268547534942,
+      "num_tokens": 12409430.0,
+      "step": 3650
+    },
+    {
+      "epoch": 0.036733710706973385,
+      "learning_rate": 0.00019265526516520134,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.7604560315608978,
+      "num_tokens": 12443177.0,
+      "step": 3660
+    },
+    {
+      "epoch": 0.036834076036773855,
+      "learning_rate": 0.00019263519209924125,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7625544607639313,
+      "num_tokens": 12477266.0,
+      "step": 3670
+    },
+    {
+      "epoch": 0.03693444136657433,
+      "learning_rate": 0.00019261511903328115,
+      "loss": 0.8131,
+      "mean_token_accuracy": 0.7592061042785645,
+      "num_tokens": 12510949.0,
+      "step": 3680
+    },
+    {
+      "epoch": 0.0370348066963748,
+      "learning_rate": 0.00019259504596732106,
+      "loss": 0.8476,
+      "mean_token_accuracy": 0.7568705081939697,
+      "num_tokens": 12544909.0,
+      "step": 3690
+    },
+    {
+      "epoch": 0.03713517202617528,
+      "learning_rate": 0.00019257497290136097,
+      "loss": 0.8184,
+      "mean_token_accuracy": 0.7610429286956787,
+      "num_tokens": 12578295.0,
+      "step": 3700
+    },
+    {
+      "epoch": 0.037235537355975755,
+      "learning_rate": 0.00019255489983540088,
+      "loss": 0.8185,
+      "mean_token_accuracy": 0.7600680828094483,
+      "num_tokens": 12612334.0,
+      "step": 3710
+    },
+    {
+      "epoch": 0.037335902685776225,
+      "learning_rate": 0.00019253482676944076,
+      "loss": 0.7809,
+      "mean_token_accuracy": 0.7732314109802246,
+      "num_tokens": 12646883.0,
+      "step": 3720
+    },
+    {
+      "epoch": 0.0374362680155767,
+      "learning_rate": 0.0001925147537034807,
+      "loss": 0.8161,
+      "mean_token_accuracy": 0.7603020191192627,
+      "num_tokens": 12681201.0,
+      "step": 3730
+    },
+    {
+      "epoch": 0.03753663334537717,
+      "learning_rate": 0.00019249468063752058,
+      "loss": 0.8437,
+      "mean_token_accuracy": 0.755571311712265,
+      "num_tokens": 12714592.0,
+      "step": 3740
+    },
+    {
+      "epoch": 0.03763699867517765,
+      "learning_rate": 0.0001924746075715605,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.7415451884269715,
+      "num_tokens": 12748583.0,
+      "step": 3750
+    },
+    {
+      "epoch": 0.03773736400497812,
+      "learning_rate": 0.0001924545345056004,
+      "loss": 0.8537,
+      "mean_token_accuracy": 0.7552414834499359,
+      "num_tokens": 12781649.0,
+      "step": 3760
+    },
+    {
+      "epoch": 0.037837729334778594,
+      "learning_rate": 0.0001924344614396403,
+      "loss": 0.8194,
+      "mean_token_accuracy": 0.7623369455337524,
+      "num_tokens": 12814395.0,
+      "step": 3770
+    },
+    {
+      "epoch": 0.03793809466457907,
+      "learning_rate": 0.0001924143883736802,
+      "loss": 0.8169,
+      "mean_token_accuracy": 0.7644883871078492,
+      "num_tokens": 12847968.0,
+      "step": 3780
+    },
+    {
+      "epoch": 0.03803845999437954,
+      "learning_rate": 0.00019239431530772012,
+      "loss": 0.8128,
+      "mean_token_accuracy": 0.7627908825874329,
+      "num_tokens": 12882042.0,
+      "step": 3790
+    },
+    {
+      "epoch": 0.03813882532418002,
+      "learning_rate": 0.00019237424224176003,
+      "loss": 0.8135,
+      "mean_token_accuracy": 0.7674229919910431,
+      "num_tokens": 12915962.0,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03823919065398049,
+      "learning_rate": 0.0001923541691757999,
+      "loss": 0.8152,
+      "mean_token_accuracy": 0.7644250571727753,
+      "num_tokens": 12950108.0,
+      "step": 3810
+    },
+    {
+      "epoch": 0.038339555983780964,
+      "learning_rate": 0.00019233409610983985,
+      "loss": 0.8407,
+      "mean_token_accuracy": 0.7574917793273925,
+      "num_tokens": 12983602.0,
+      "step": 3820
+    },
+    {
+      "epoch": 0.038439921313581434,
+      "learning_rate": 0.00019231402304387973,
+      "loss": 0.8331,
+      "mean_token_accuracy": 0.7552855551242829,
+      "num_tokens": 13017565.0,
+      "step": 3830
+    },
+    {
+      "epoch": 0.03854028664338191,
+      "learning_rate": 0.00019229394997791964,
+      "loss": 0.8185,
+      "mean_token_accuracy": 0.7646740972995758,
+      "num_tokens": 13051376.0,
+      "step": 3840
+    },
+    {
+      "epoch": 0.03864065197318239,
+      "learning_rate": 0.00019227387691195954,
+      "loss": 0.8064,
+      "mean_token_accuracy": 0.762885594367981,
+      "num_tokens": 13085818.0,
+      "step": 3850
+    },
+    {
+      "epoch": 0.03874101730298286,
+      "learning_rate": 0.00019225380384599945,
+      "loss": 0.79,
+      "mean_token_accuracy": 0.7621721744537353,
+      "num_tokens": 13119886.0,
+      "step": 3860
+    },
+    {
+      "epoch": 0.03884138263278333,
+      "learning_rate": 0.00019223373078003933,
+      "loss": 0.8072,
+      "mean_token_accuracy": 0.7587761878967285,
+      "num_tokens": 13154493.0,
+      "step": 3870
+    },
+    {
+      "epoch": 0.0389417479625838,
+      "learning_rate": 0.00019221365771407927,
+      "loss": 0.7764,
+      "mean_token_accuracy": 0.7698749780654908,
+      "num_tokens": 13188706.0,
+      "step": 3880
+    },
+    {
+      "epoch": 0.03904211329238428,
+      "learning_rate": 0.00019219358464811915,
+      "loss": 0.8168,
+      "mean_token_accuracy": 0.7628201723098755,
+      "num_tokens": 13222666.0,
+      "step": 3890
+    },
+    {
+      "epoch": 0.03914247862218475,
+      "learning_rate": 0.00019217351158215906,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.7653783857822418,
+      "num_tokens": 13256982.0,
+      "step": 3900
+    },
+    {
+      "epoch": 0.039242843951985226,
+      "learning_rate": 0.00019215343851619897,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7596830606460572,
+      "num_tokens": 13290715.0,
+      "step": 3910
+    },
+    {
+      "epoch": 0.0393432092817857,
+      "learning_rate": 0.00019213336545023888,
+      "loss": 0.8311,
+      "mean_token_accuracy": 0.758405339717865,
+      "num_tokens": 13325075.0,
+      "step": 3920
+    },
+    {
+      "epoch": 0.03944357461158617,
+      "learning_rate": 0.00019211329238427878,
+      "loss": 0.812,
+      "mean_token_accuracy": 0.7626287162303924,
+      "num_tokens": 13358187.0,
+      "step": 3930
+    },
+    {
+      "epoch": 0.03954393994138665,
+      "learning_rate": 0.0001920932193183187,
+      "loss": 0.8166,
+      "mean_token_accuracy": 0.7628508031368255,
+      "num_tokens": 13392473.0,
+      "step": 3940
+    },
+    {
+      "epoch": 0.03964430527118712,
+      "learning_rate": 0.0001920731462523586,
+      "loss": 0.8274,
+      "mean_token_accuracy": 0.7561886072158813,
+      "num_tokens": 13426117.0,
+      "step": 3950
+    },
+    {
+      "epoch": 0.039744670600987596,
+      "learning_rate": 0.00019205307318639848,
+      "loss": 0.8141,
+      "mean_token_accuracy": 0.7664759397506714,
+      "num_tokens": 13459599.0,
+      "step": 3960
+    },
+    {
+      "epoch": 0.039845035930788066,
+      "learning_rate": 0.00019203300012043842,
+      "loss": 0.809,
+      "mean_token_accuracy": 0.7616986215114594,
+      "num_tokens": 13493007.0,
+      "step": 3970
+    },
+    {
+      "epoch": 0.03994540126058854,
+      "learning_rate": 0.0001920129270544783,
+      "loss": 0.7988,
+      "mean_token_accuracy": 0.7667593657970428,
+      "num_tokens": 13526790.0,
+      "step": 3980
+    },
+    {
+      "epoch": 0.04004576659038902,
+      "learning_rate": 0.0001919928539885182,
+      "loss": 0.805,
+      "mean_token_accuracy": 0.7676305174827576,
+      "num_tokens": 13560353.0,
+      "step": 3990
+    },
+    {
+      "epoch": 0.04014613192018949,
+      "learning_rate": 0.00019197278092255812,
+      "loss": 0.8011,
+      "mean_token_accuracy": 0.7701820015907288,
+      "num_tokens": 13594409.0,
+      "step": 4000
+    },
+    {
+      "epoch": 0.040246497249989965,
+      "learning_rate": 0.00019195270785659802,
+      "loss": 0.8209,
+      "mean_token_accuracy": 0.7571946740150451,
+      "num_tokens": 13628492.0,
+      "step": 4010
+    },
+    {
+      "epoch": 0.040346862579790435,
+      "learning_rate": 0.00019193263479063793,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.7589803755283355,
+      "num_tokens": 13661763.0,
+      "step": 4020
+    },
+    {
+      "epoch": 0.04044722790959091,
+      "learning_rate": 0.00019191256172467784,
+      "loss": 0.814,
+      "mean_token_accuracy": 0.7596362948417663,
+      "num_tokens": 13695269.0,
+      "step": 4030
+    },
+    {
+      "epoch": 0.04054759323939138,
+      "learning_rate": 0.00019189248865871775,
+      "loss": 0.7936,
+      "mean_token_accuracy": 0.7661742925643921,
+      "num_tokens": 13728774.0,
+      "step": 4040
+    },
+    {
+      "epoch": 0.04064795856919186,
+      "learning_rate": 0.00019187241559275763,
+      "loss": 0.7932,
+      "mean_token_accuracy": 0.7682706117630005,
+      "num_tokens": 13762389.0,
+      "step": 4050
+    },
+    {
+      "epoch": 0.040748323898992335,
+      "learning_rate": 0.00019185234252679757,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.7650616884231567,
+      "num_tokens": 13796268.0,
+      "step": 4060
+    },
+    {
+      "epoch": 0.040848689228792805,
+      "learning_rate": 0.00019183226946083745,
+      "loss": 0.8043,
+      "mean_token_accuracy": 0.7669219195842742,
+      "num_tokens": 13829749.0,
+      "step": 4070
+    },
+    {
+      "epoch": 0.04094905455859328,
+      "learning_rate": 0.00019181219639487736,
+      "loss": 0.8131,
+      "mean_token_accuracy": 0.7608605742454528,
+      "num_tokens": 13863689.0,
+      "step": 4080
+    },
+    {
+      "epoch": 0.04104941988839375,
+      "learning_rate": 0.00019179212332891727,
+      "loss": 0.7973,
+      "mean_token_accuracy": 0.7637056112289429,
+      "num_tokens": 13897105.0,
+      "step": 4090
+    },
+    {
+      "epoch": 0.04114978521819423,
+      "learning_rate": 0.00019177205026295717,
+      "loss": 0.8293,
+      "mean_token_accuracy": 0.7573158740997314,
+      "num_tokens": 13930387.0,
+      "step": 4100
+    },
+    {
+      "epoch": 0.0412501505479947,
+      "learning_rate": 0.00019175197719699705,
+      "loss": 0.804,
+      "mean_token_accuracy": 0.7686714947223663,
+      "num_tokens": 13964275.0,
+      "step": 4110
+    },
+    {
+      "epoch": 0.041350515877795174,
+      "learning_rate": 0.000191731904131037,
+      "loss": 0.7523,
+      "mean_token_accuracy": 0.7722469091415405,
+      "num_tokens": 13998231.0,
+      "step": 4120
+    },
+    {
+      "epoch": 0.04145088120759565,
+      "learning_rate": 0.0001917118310650769,
+      "loss": 0.7834,
+      "mean_token_accuracy": 0.7652732789516449,
+      "num_tokens": 14032704.0,
+      "step": 4130
+    },
+    {
+      "epoch": 0.04155124653739612,
+      "learning_rate": 0.00019169175799911678,
+      "loss": 0.7995,
+      "mean_token_accuracy": 0.7662194430828094,
+      "num_tokens": 14066622.0,
+      "step": 4140
+    },
+    {
+      "epoch": 0.0416516118671966,
+      "learning_rate": 0.00019167168493315672,
+      "loss": 0.8139,
+      "mean_token_accuracy": 0.7601560235023499,
+      "num_tokens": 14101121.0,
+      "step": 4150
+    },
+    {
+      "epoch": 0.04175197719699707,
+      "learning_rate": 0.0001916516118671966,
+      "loss": 0.8401,
+      "mean_token_accuracy": 0.7569827854633331,
+      "num_tokens": 14135345.0,
+      "step": 4160
+    },
+    {
+      "epoch": 0.041852342526797544,
+      "learning_rate": 0.00019163153880123653,
+      "loss": 0.7911,
+      "mean_token_accuracy": 0.7664677023887634,
+      "num_tokens": 14169416.0,
+      "step": 4170
+    },
+    {
+      "epoch": 0.041952707856598014,
+      "learning_rate": 0.00019161146573527641,
+      "loss": 0.7994,
+      "mean_token_accuracy": 0.7639586329460144,
+      "num_tokens": 14204283.0,
+      "step": 4180
+    },
+    {
+      "epoch": 0.04205307318639849,
+      "learning_rate": 0.00019159139266931632,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7709622740745544,
+      "num_tokens": 14239003.0,
+      "step": 4190
+    },
+    {
+      "epoch": 0.04215343851619897,
+      "learning_rate": 0.00019157131960335623,
+      "loss": 0.8221,
+      "mean_token_accuracy": 0.758518660068512,
+      "num_tokens": 14272386.0,
+      "step": 4200
+    },
+    {
+      "epoch": 0.04225380384599944,
+      "learning_rate": 0.00019155124653739614,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.7598104774951935,
+      "num_tokens": 14305825.0,
+      "step": 4210
+    },
+    {
+      "epoch": 0.04235416917579991,
+      "learning_rate": 0.00019153117347143602,
+      "loss": 0.8086,
+      "mean_token_accuracy": 0.7655596077442169,
+      "num_tokens": 14340525.0,
+      "step": 4220
+    },
+    {
+      "epoch": 0.04245453450560038,
+      "learning_rate": 0.00019151110040547596,
+      "loss": 0.8273,
+      "mean_token_accuracy": 0.7571021318435669,
+      "num_tokens": 14374981.0,
+      "step": 4230
+    },
+    {
+      "epoch": 0.04255489983540086,
+      "learning_rate": 0.00019149102733951584,
+      "loss": 0.7844,
+      "mean_token_accuracy": 0.7691513001918793,
+      "num_tokens": 14408364.0,
+      "step": 4240
+    },
+    {
+      "epoch": 0.04265526516520133,
+      "learning_rate": 0.00019147095427355575,
+      "loss": 0.8186,
+      "mean_token_accuracy": 0.7621715486049652,
+      "num_tokens": 14441851.0,
+      "step": 4250
+    },
+    {
+      "epoch": 0.042755630495001806,
+      "learning_rate": 0.00019145088120759565,
+      "loss": 0.8229,
+      "mean_token_accuracy": 0.762150514125824,
+      "num_tokens": 14475783.0,
+      "step": 4260
+    },
+    {
+      "epoch": 0.04285599582480228,
+      "learning_rate": 0.00019143080814163556,
+      "loss": 0.8216,
+      "mean_token_accuracy": 0.7614712715148926,
+      "num_tokens": 14509383.0,
+      "step": 4270
+    },
+    {
+      "epoch": 0.04295636115460275,
+      "learning_rate": 0.00019141073507567547,
+      "loss": 0.8182,
+      "mean_token_accuracy": 0.7644801437854767,
+      "num_tokens": 14543595.0,
+      "step": 4280
+    },
+    {
+      "epoch": 0.04305672648440323,
+      "learning_rate": 0.00019139066200971538,
+      "loss": 0.7619,
+      "mean_token_accuracy": 0.781198114156723,
+      "num_tokens": 14577480.0,
+      "step": 4290
+    },
+    {
+      "epoch": 0.0431570918142037,
+      "learning_rate": 0.0001913705889437553,
+      "loss": 0.7718,
+      "mean_token_accuracy": 0.7747524976730347,
+      "num_tokens": 14612021.0,
+      "step": 4300
+    },
+    {
+      "epoch": 0.043257457144004176,
+      "learning_rate": 0.00019135051587779517,
+      "loss": 0.8136,
+      "mean_token_accuracy": 0.7664998054504395,
+      "num_tokens": 14646010.0,
+      "step": 4310
+    },
+    {
+      "epoch": 0.043357822473804646,
+      "learning_rate": 0.0001913304428118351,
+      "loss": 0.7982,
+      "mean_token_accuracy": 0.7697717070579528,
+      "num_tokens": 14679159.0,
+      "step": 4320
+    },
+    {
+      "epoch": 0.04345818780360512,
+      "learning_rate": 0.00019131036974587499,
+      "loss": 0.8073,
+      "mean_token_accuracy": 0.7655534505844116,
+      "num_tokens": 14712845.0,
+      "step": 4330
+    },
+    {
+      "epoch": 0.0435585531334056,
+      "learning_rate": 0.0001912902966799149,
+      "loss": 0.8319,
+      "mean_token_accuracy": 0.760830146074295,
+      "num_tokens": 14747070.0,
+      "step": 4340
+    },
+    {
+      "epoch": 0.04365891846320607,
+      "learning_rate": 0.0001912702236139548,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.7607136905193329,
+      "num_tokens": 14780892.0,
+      "step": 4350
+    },
+    {
+      "epoch": 0.043759283793006545,
+      "learning_rate": 0.0001912501505479947,
+      "loss": 0.821,
+      "mean_token_accuracy": 0.7604890406131745,
+      "num_tokens": 14814844.0,
+      "step": 4360
+    },
+    {
+      "epoch": 0.043859649122807015,
+      "learning_rate": 0.00019123007748203462,
+      "loss": 0.8344,
+      "mean_token_accuracy": 0.7620271623134613,
+      "num_tokens": 14849397.0,
+      "step": 4370
+    },
+    {
+      "epoch": 0.04396001445260749,
+      "learning_rate": 0.00019121000441607453,
+      "loss": 0.8168,
+      "mean_token_accuracy": 0.7646451532840729,
+      "num_tokens": 14884507.0,
+      "step": 4380
+    },
+    {
+      "epoch": 0.04406037978240796,
+      "learning_rate": 0.00019118993135011444,
+      "loss": 0.8326,
+      "mean_token_accuracy": 0.7582274377346039,
+      "num_tokens": 14917952.0,
+      "step": 4390
+    },
+    {
+      "epoch": 0.04416074511220844,
+      "learning_rate": 0.00019116985828415432,
+      "loss": 0.7943,
+      "mean_token_accuracy": 0.7630038142204285,
+      "num_tokens": 14951593.0,
+      "step": 4400
+    },
+    {
+      "epoch": 0.044261110442008915,
+      "learning_rate": 0.00019114978521819425,
+      "loss": 0.8004,
+      "mean_token_accuracy": 0.7646883130073547,
+      "num_tokens": 14986092.0,
+      "step": 4410
+    },
+    {
+      "epoch": 0.044361475771809385,
+      "learning_rate": 0.00019112971215223414,
+      "loss": 0.7422,
+      "mean_token_accuracy": 0.7789775013923645,
+      "num_tokens": 15020485.0,
+      "step": 4420
+    },
+    {
+      "epoch": 0.04446184110160986,
+      "learning_rate": 0.00019110963908627404,
+      "loss": 0.8134,
+      "mean_token_accuracy": 0.7608291804790497,
+      "num_tokens": 15054328.0,
+      "step": 4430
+    },
+    {
+      "epoch": 0.04456220643141033,
+      "learning_rate": 0.00019108956602031395,
+      "loss": 0.8157,
+      "mean_token_accuracy": 0.7607727348804474,
+      "num_tokens": 15088045.0,
+      "step": 4440
+    },
+    {
+      "epoch": 0.04466257176121081,
+      "learning_rate": 0.00019106949295435386,
+      "loss": 0.8194,
+      "mean_token_accuracy": 0.7610091507434845,
+      "num_tokens": 15120903.0,
+      "step": 4450
+    },
+    {
+      "epoch": 0.044762937091011284,
+      "learning_rate": 0.00019104941988839374,
+      "loss": 0.8549,
+      "mean_token_accuracy": 0.7528641760349274,
+      "num_tokens": 15154821.0,
+      "step": 4460
+    },
+    {
+      "epoch": 0.044863302420811754,
+      "learning_rate": 0.00019102934682243368,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.7704406261444092,
+      "num_tokens": 15188544.0,
+      "step": 4470
+    },
+    {
+      "epoch": 0.04496366775061223,
+      "learning_rate": 0.00019100927375647359,
+      "loss": 0.8244,
+      "mean_token_accuracy": 0.7597645819187164,
+      "num_tokens": 15222804.0,
+      "step": 4480
+    },
+    {
+      "epoch": 0.0450640330804127,
+      "learning_rate": 0.00019098920069051347,
+      "loss": 0.7967,
+      "mean_token_accuracy": 0.7645935952663422,
+      "num_tokens": 15257498.0,
+      "step": 4490
+    },
+    {
+      "epoch": 0.04516439841021318,
+      "learning_rate": 0.0001909691276245534,
+      "loss": 0.7711,
+      "mean_token_accuracy": 0.7738887727260589,
+      "num_tokens": 15291150.0,
+      "step": 4500
+    },
+    {
+      "epoch": 0.04526476374001365,
+      "learning_rate": 0.00019094905455859328,
+      "loss": 0.8196,
+      "mean_token_accuracy": 0.7628287553787232,
+      "num_tokens": 15325608.0,
+      "step": 4510
+    },
+    {
+      "epoch": 0.045365129069814124,
+      "learning_rate": 0.0001909289814926332,
+      "loss": 0.8097,
+      "mean_token_accuracy": 0.7615756452083587,
+      "num_tokens": 15359817.0,
+      "step": 4520
+    },
+    {
+      "epoch": 0.0454654943996146,
+      "learning_rate": 0.0001909089084266731,
+      "loss": 0.7865,
+      "mean_token_accuracy": 0.7654139697551727,
+      "num_tokens": 15393899.0,
+      "step": 4530
+    },
+    {
+      "epoch": 0.04556585972941507,
+      "learning_rate": 0.000190888835360713,
+      "loss": 0.8154,
+      "mean_token_accuracy": 0.7623386085033417,
+      "num_tokens": 15427234.0,
+      "step": 4540
+    },
+    {
+      "epoch": 0.04566622505921555,
+      "learning_rate": 0.0001908687622947529,
+      "loss": 0.8044,
+      "mean_token_accuracy": 0.7596213340759277,
+      "num_tokens": 15460609.0,
+      "step": 4550
+    },
+    {
+      "epoch": 0.04576659038901602,
+      "learning_rate": 0.00019084868922879283,
+      "loss": 0.7949,
+      "mean_token_accuracy": 0.7677416563034057,
+      "num_tokens": 15494983.0,
+      "step": 4560
+    },
+    {
+      "epoch": 0.04586695571881649,
+      "learning_rate": 0.0001908286161628327,
+      "loss": 0.7998,
+      "mean_token_accuracy": 0.759462857246399,
+      "num_tokens": 15527836.0,
+      "step": 4570
+    },
+    {
+      "epoch": 0.04596732104861696,
+      "learning_rate": 0.00019080854309687262,
+      "loss": 0.8461,
+      "mean_token_accuracy": 0.7568655669689178,
+      "num_tokens": 15561348.0,
+      "step": 4580
+    },
+    {
+      "epoch": 0.04606768637841744,
+      "learning_rate": 0.00019078847003091252,
+      "loss": 0.8358,
+      "mean_token_accuracy": 0.7562499463558197,
+      "num_tokens": 15595890.0,
+      "step": 4590
+    },
+    {
+      "epoch": 0.046168051708217916,
+      "learning_rate": 0.00019076839696495243,
+      "loss": 0.841,
+      "mean_token_accuracy": 0.7540477454662323,
+      "num_tokens": 15630814.0,
+      "step": 4600
+    },
+    {
+      "epoch": 0.046268417038018386,
+      "learning_rate": 0.00019074832389899234,
+      "loss": 0.8379,
+      "mean_token_accuracy": 0.7596671104431152,
+      "num_tokens": 15665342.0,
+      "step": 4610
+    },
+    {
+      "epoch": 0.04636878236781886,
+      "learning_rate": 0.00019072825083303225,
+      "loss": 0.7917,
+      "mean_token_accuracy": 0.7665965020656585,
+      "num_tokens": 15699006.0,
+      "step": 4620
+    },
+    {
+      "epoch": 0.04646914769761933,
+      "learning_rate": 0.00019070817776707216,
+      "loss": 0.806,
+      "mean_token_accuracy": 0.7658841371536255,
+      "num_tokens": 15733812.0,
+      "step": 4630
+    },
+    {
+      "epoch": 0.04656951302741981,
+      "learning_rate": 0.00019068810470111204,
+      "loss": 0.7696,
+      "mean_token_accuracy": 0.7712407052516937,
+      "num_tokens": 15768185.0,
+      "step": 4640
+    },
+    {
+      "epoch": 0.04666987835722028,
+      "learning_rate": 0.00019066803163515197,
+      "loss": 0.7868,
+      "mean_token_accuracy": 0.7670133650302887,
+      "num_tokens": 15802046.0,
+      "step": 4650
+    },
+    {
+      "epoch": 0.046770243687020756,
+      "learning_rate": 0.00019064795856919186,
+      "loss": 0.8039,
+      "mean_token_accuracy": 0.7610798418521881,
+      "num_tokens": 15836439.0,
+      "step": 4660
+    },
+    {
+      "epoch": 0.04687060901682123,
+      "learning_rate": 0.00019062788550323176,
+      "loss": 0.7943,
+      "mean_token_accuracy": 0.7643402874469757,
+      "num_tokens": 15869996.0,
+      "step": 4670
+    },
+    {
+      "epoch": 0.0469709743466217,
+      "learning_rate": 0.00019060781243727167,
+      "loss": 0.826,
+      "mean_token_accuracy": 0.7603221535682678,
+      "num_tokens": 15903732.0,
+      "step": 4680
+    },
+    {
+      "epoch": 0.04707133967642218,
+      "learning_rate": 0.00019058773937131158,
+      "loss": 0.7764,
+      "mean_token_accuracy": 0.7736131608486175,
+      "num_tokens": 15938228.0,
+      "step": 4690
+    },
+    {
+      "epoch": 0.04717170500622265,
+      "learning_rate": 0.0001905676663053515,
+      "loss": 0.8019,
+      "mean_token_accuracy": 0.7617711782455444,
+      "num_tokens": 15971639.0,
+      "step": 4700
+    },
+    {
+      "epoch": 0.047272070336023125,
+      "learning_rate": 0.0001905475932393914,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.765626859664917,
+      "num_tokens": 16005456.0,
+      "step": 4710
+    },
+    {
+      "epoch": 0.047372435665823595,
+      "learning_rate": 0.0001905275201734313,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.7578832507133484,
+      "num_tokens": 16039943.0,
+      "step": 4720
+    },
+    {
+      "epoch": 0.04747280099562407,
+      "learning_rate": 0.00019050744710747122,
+      "loss": 0.8159,
+      "mean_token_accuracy": 0.7593542397022247,
+      "num_tokens": 16073997.0,
+      "step": 4730
+    },
+    {
+      "epoch": 0.04757316632542455,
+      "learning_rate": 0.00019048737404151112,
+      "loss": 0.8401,
+      "mean_token_accuracy": 0.7576812863349914,
+      "num_tokens": 16108978.0,
+      "step": 4740
+    },
+    {
+      "epoch": 0.04767353165522502,
+      "learning_rate": 0.000190467300975551,
+      "loss": 0.802,
+      "mean_token_accuracy": 0.7665064334869385,
+      "num_tokens": 16143274.0,
+      "step": 4750
+    },
+    {
+      "epoch": 0.047773896985025495,
+      "learning_rate": 0.00019044722790959094,
+      "loss": 0.7668,
+      "mean_token_accuracy": 0.7717243552207946,
+      "num_tokens": 16177844.0,
+      "step": 4760
+    },
+    {
+      "epoch": 0.047874262314825965,
+      "learning_rate": 0.00019042715484363082,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7543803930282593,
+      "num_tokens": 16212293.0,
+      "step": 4770
+    },
+    {
+      "epoch": 0.04797462764462644,
+      "learning_rate": 0.00019040708177767073,
+      "loss": 0.7838,
+      "mean_token_accuracy": 0.766845291852951,
+      "num_tokens": 16246046.0,
+      "step": 4780
+    },
+    {
+      "epoch": 0.04807499297442691,
+      "learning_rate": 0.00019038700871171064,
+      "loss": 0.7978,
+      "mean_token_accuracy": 0.764685869216919,
+      "num_tokens": 16279922.0,
+      "step": 4790
+    },
+    {
+      "epoch": 0.04817535830422739,
+      "learning_rate": 0.00019036693564575055,
+      "loss": 0.8131,
+      "mean_token_accuracy": 0.762106591463089,
+      "num_tokens": 16313301.0,
+      "step": 4800
+    },
+    {
+      "epoch": 0.048275723634027864,
+      "learning_rate": 0.00019034686257979043,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.762081903219223,
+      "num_tokens": 16347030.0,
+      "step": 4810
+    },
+    {
+      "epoch": 0.048376088963828334,
+      "learning_rate": 0.00019032678951383036,
+      "loss": 0.8192,
+      "mean_token_accuracy": 0.7583778619766235,
+      "num_tokens": 16380294.0,
+      "step": 4820
+    },
+    {
+      "epoch": 0.04847645429362881,
+      "learning_rate": 0.00019030671644787027,
+      "loss": 0.8274,
+      "mean_token_accuracy": 0.7571999251842498,
+      "num_tokens": 16414833.0,
+      "step": 4830
+    },
+    {
+      "epoch": 0.04857681962342928,
+      "learning_rate": 0.00019028664338191015,
+      "loss": 0.808,
+      "mean_token_accuracy": 0.7681930124759674,
+      "num_tokens": 16448628.0,
+      "step": 4840
+    },
+    {
+      "epoch": 0.04867718495322976,
+      "learning_rate": 0.0001902665703159501,
+      "loss": 0.7956,
+      "mean_token_accuracy": 0.7681915581226348,
+      "num_tokens": 16482605.0,
+      "step": 4850
+    },
+    {
+      "epoch": 0.04877755028303023,
+      "learning_rate": 0.00019024649724998997,
+      "loss": 0.7698,
+      "mean_token_accuracy": 0.7775806069374085,
+      "num_tokens": 16516846.0,
+      "step": 4860
+    },
+    {
+      "epoch": 0.048877915612830704,
+      "learning_rate": 0.00019022642418402988,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.7511551916599274,
+      "num_tokens": 16550932.0,
+      "step": 4870
+    },
+    {
+      "epoch": 0.04897828094263118,
+      "learning_rate": 0.0001902063511180698,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.7689824163913727,
+      "num_tokens": 16585156.0,
+      "step": 4880
+    },
+    {
+      "epoch": 0.04907864627243165,
+      "learning_rate": 0.0001901862780521097,
+      "loss": 0.8578,
+      "mean_token_accuracy": 0.7557364404201508,
+      "num_tokens": 16619347.0,
+      "step": 4890
+    },
+    {
+      "epoch": 0.04917901160223213,
+      "learning_rate": 0.00019016620498614958,
+      "loss": 0.7943,
+      "mean_token_accuracy": 0.7654554963111877,
+      "num_tokens": 16652773.0,
+      "step": 4900
+    },
+    {
+      "epoch": 0.0492793769320326,
+      "learning_rate": 0.0001901461319201895,
+      "loss": 0.8235,
+      "mean_token_accuracy": 0.7565623998641968,
+      "num_tokens": 16686677.0,
+      "step": 4910
+    },
+    {
+      "epoch": 0.04937974226183307,
+      "learning_rate": 0.0001901260588542294,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7688578844070435,
+      "num_tokens": 16720701.0,
+      "step": 4920
+    },
+    {
+      "epoch": 0.04948010759163354,
+      "learning_rate": 0.0001901059857882693,
+      "loss": 0.8117,
+      "mean_token_accuracy": 0.761589401960373,
+      "num_tokens": 16754870.0,
+      "step": 4930
+    },
+    {
+      "epoch": 0.04958047292143402,
+      "learning_rate": 0.0001900859127223092,
+      "loss": 0.8042,
+      "mean_token_accuracy": 0.7648223102092743,
+      "num_tokens": 16789281.0,
+      "step": 4940
+    },
+    {
+      "epoch": 0.049680838251234496,
+      "learning_rate": 0.00019006583965634912,
+      "loss": 0.8184,
+      "mean_token_accuracy": 0.7596215605735779,
+      "num_tokens": 16823379.0,
+      "step": 4950
+    },
+    {
+      "epoch": 0.049781203581034966,
+      "learning_rate": 0.00019004576659038903,
+      "loss": 0.8424,
+      "mean_token_accuracy": 0.7572415292263031,
+      "num_tokens": 16856561.0,
+      "step": 4960
+    },
+    {
+      "epoch": 0.04988156891083544,
+      "learning_rate": 0.00019002569352442894,
+      "loss": 0.8194,
+      "mean_token_accuracy": 0.7631000757217408,
+      "num_tokens": 16889736.0,
+      "step": 4970
+    },
+    {
+      "epoch": 0.04998193424063591,
+      "learning_rate": 0.00019000562045846884,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7635967433452606,
+      "num_tokens": 16922931.0,
+      "step": 4980
+    },
+    {
+      "epoch": 0.05008229957043639,
+      "learning_rate": 0.00018998554739250873,
+      "loss": 0.7788,
+      "mean_token_accuracy": 0.7682659387588501,
+      "num_tokens": 16957466.0,
+      "step": 4990
+    },
+    {
+      "epoch": 0.05018266490023686,
+      "learning_rate": 0.00018996547432654866,
+      "loss": 0.7996,
+      "mean_token_accuracy": 0.7649333596229553,
+      "num_tokens": 16991761.0,
+      "step": 5000
+    },
+    {
+      "epoch": 0.050283030230037336,
+      "learning_rate": 0.00018994540126058854,
+      "loss": 0.8438,
+      "mean_token_accuracy": 0.7554546117782592,
+      "num_tokens": 17026294.0,
+      "step": 5010
+    },
+    {
+      "epoch": 0.05038339555983781,
+      "learning_rate": 0.00018992532819462845,
+      "loss": 0.8006,
+      "mean_token_accuracy": 0.7630142390727996,
+      "num_tokens": 17060231.0,
+      "step": 5020
+    },
+    {
+      "epoch": 0.05048376088963828,
+      "learning_rate": 0.00018990525512866836,
+      "loss": 0.7855,
+      "mean_token_accuracy": 0.7714180409908294,
+      "num_tokens": 17095013.0,
+      "step": 5030
+    },
+    {
+      "epoch": 0.05058412621943876,
+      "learning_rate": 0.00018988518206270827,
+      "loss": 0.821,
+      "mean_token_accuracy": 0.7537748515605927,
+      "num_tokens": 17128747.0,
+      "step": 5040
+    },
+    {
+      "epoch": 0.05068449154923923,
+      "learning_rate": 0.00018986510899674818,
+      "loss": 0.8498,
+      "mean_token_accuracy": 0.7544196486473084,
+      "num_tokens": 17162701.0,
+      "step": 5050
+    },
+    {
+      "epoch": 0.050784856879039705,
+      "learning_rate": 0.00018984503593078809,
+      "loss": 0.7982,
+      "mean_token_accuracy": 0.7683187246322631,
+      "num_tokens": 17196760.0,
+      "step": 5060
+    },
+    {
+      "epoch": 0.050885222208840175,
+      "learning_rate": 0.000189824962864828,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.7660270392894745,
+      "num_tokens": 17230284.0,
+      "step": 5070
+    },
+    {
+      "epoch": 0.05098558753864065,
+      "learning_rate": 0.00018980488979886787,
+      "loss": 0.821,
+      "mean_token_accuracy": 0.7614680528640747,
+      "num_tokens": 17264549.0,
+      "step": 5080
+    },
+    {
+      "epoch": 0.05108595286844113,
+      "learning_rate": 0.0001897848167329078,
+      "loss": 0.8164,
+      "mean_token_accuracy": 0.7596682369709015,
+      "num_tokens": 17298632.0,
+      "step": 5090
+    },
+    {
+      "epoch": 0.0511863181982416,
+      "learning_rate": 0.0001897647436669477,
+      "loss": 0.8164,
+      "mean_token_accuracy": 0.7643806874752045,
+      "num_tokens": 17332557.0,
+      "step": 5100
+    },
+    {
+      "epoch": 0.051286683528042075,
+      "learning_rate": 0.0001897446706009876,
+      "loss": 0.8234,
+      "mean_token_accuracy": 0.7651040554046631,
+      "num_tokens": 17365809.0,
+      "step": 5110
+    },
+    {
+      "epoch": 0.051387048857842545,
+      "learning_rate": 0.0001897245975350275,
+      "loss": 0.8118,
+      "mean_token_accuracy": 0.7640089511871337,
+      "num_tokens": 17400122.0,
+      "step": 5120
+    },
+    {
+      "epoch": 0.05148741418764302,
+      "learning_rate": 0.00018970452446906742,
+      "loss": 0.7918,
+      "mean_token_accuracy": 0.7671834468841553,
+      "num_tokens": 17434037.0,
+      "step": 5130
+    },
+    {
+      "epoch": 0.05158777951744349,
+      "learning_rate": 0.0001896844514031073,
+      "loss": 0.7833,
+      "mean_token_accuracy": 0.7701202273368836,
+      "num_tokens": 17467825.0,
+      "step": 5140
+    },
+    {
+      "epoch": 0.05168814484724397,
+      "learning_rate": 0.00018966437833714723,
+      "loss": 0.7972,
+      "mean_token_accuracy": 0.7589609801769257,
+      "num_tokens": 17501028.0,
+      "step": 5150
+    },
+    {
+      "epoch": 0.051788510177044444,
+      "learning_rate": 0.00018964430527118712,
+      "loss": 0.8106,
+      "mean_token_accuracy": 0.7608213603496552,
+      "num_tokens": 17535782.0,
+      "step": 5160
+    },
+    {
+      "epoch": 0.051888875506844914,
+      "learning_rate": 0.00018962423220522702,
+      "loss": 0.8292,
+      "mean_token_accuracy": 0.7518568873405457,
+      "num_tokens": 17569420.0,
+      "step": 5170
+    },
+    {
+      "epoch": 0.05198924083664539,
+      "learning_rate": 0.00018960415913926696,
+      "loss": 0.8025,
+      "mean_token_accuracy": 0.7592430770397186,
+      "num_tokens": 17603583.0,
+      "step": 5180
+    },
+    {
+      "epoch": 0.05208960616644586,
+      "learning_rate": 0.00018958408607330684,
+      "loss": 0.8245,
+      "mean_token_accuracy": 0.7620608747005463,
+      "num_tokens": 17637921.0,
+      "step": 5190
+    },
+    {
+      "epoch": 0.05218997149624634,
+      "learning_rate": 0.00018956401300734675,
+      "loss": 0.8265,
+      "mean_token_accuracy": 0.7579002261161805,
+      "num_tokens": 17671881.0,
+      "step": 5200
+    },
+    {
+      "epoch": 0.05229033682604681,
+      "learning_rate": 0.00018954393994138666,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.776001226902008,
+      "num_tokens": 17705190.0,
+      "step": 5210
+    },
+    {
+      "epoch": 0.052390702155847284,
+      "learning_rate": 0.00018952386687542657,
+      "loss": 0.8015,
+      "mean_token_accuracy": 0.7666917979717255,
+      "num_tokens": 17738155.0,
+      "step": 5220
+    },
+    {
+      "epoch": 0.05249106748564776,
+      "learning_rate": 0.00018950379380946645,
+      "loss": 0.7992,
+      "mean_token_accuracy": 0.7608095228672027,
+      "num_tokens": 17772226.0,
+      "step": 5230
+    },
+    {
+      "epoch": 0.05259143281544823,
+      "learning_rate": 0.00018948372074350638,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.777134358882904,
+      "num_tokens": 17805812.0,
+      "step": 5240
+    },
+    {
+      "epoch": 0.05269179814524871,
+      "learning_rate": 0.00018946364767754626,
+      "loss": 0.8004,
+      "mean_token_accuracy": 0.7648840487003327,
+      "num_tokens": 17841031.0,
+      "step": 5250
+    },
+    {
+      "epoch": 0.05279216347504918,
+      "learning_rate": 0.00018944357461158617,
+      "loss": 0.8193,
+      "mean_token_accuracy": 0.7593564450740814,
+      "num_tokens": 17875531.0,
+      "step": 5260
+    },
+    {
+      "epoch": 0.05289252880484965,
+      "learning_rate": 0.00018942350154562608,
+      "loss": 0.8054,
+      "mean_token_accuracy": 0.7660769641399383,
+      "num_tokens": 17909494.0,
+      "step": 5270
+    },
+    {
+      "epoch": 0.05299289413465012,
+      "learning_rate": 0.000189403428479666,
+      "loss": 0.8046,
+      "mean_token_accuracy": 0.7677799701690674,
+      "num_tokens": 17944588.0,
+      "step": 5280
+    },
+    {
+      "epoch": 0.0530932594644506,
+      "learning_rate": 0.0001893833554137059,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.7590183019638062,
+      "num_tokens": 17978905.0,
+      "step": 5290
+    },
+    {
+      "epoch": 0.053193624794251076,
+      "learning_rate": 0.0001893632823477458,
+      "loss": 0.8164,
+      "mean_token_accuracy": 0.7615972459316254,
+      "num_tokens": 18013101.0,
+      "step": 5300
+    },
+    {
+      "epoch": 0.053293990124051546,
+      "learning_rate": 0.00018934320928178571,
+      "loss": 0.8286,
+      "mean_token_accuracy": 0.7581937134265899,
+      "num_tokens": 18046969.0,
+      "step": 5310
+    },
+    {
+      "epoch": 0.05339435545385202,
+      "learning_rate": 0.00018932313621582562,
+      "loss": 0.8018,
+      "mean_token_accuracy": 0.7623614311218262,
+      "num_tokens": 18079866.0,
+      "step": 5320
+    },
+    {
+      "epoch": 0.05349472078365249,
+      "learning_rate": 0.00018930306314986553,
+      "loss": 0.8099,
+      "mean_token_accuracy": 0.7652176499366761,
+      "num_tokens": 18114087.0,
+      "step": 5330
+    },
+    {
+      "epoch": 0.05359508611345297,
+      "learning_rate": 0.0001892829900839054,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7658193945884705,
+      "num_tokens": 18147960.0,
+      "step": 5340
+    },
+    {
+      "epoch": 0.05369545144325344,
+      "learning_rate": 0.00018926291701794535,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.7609344720840454,
+      "num_tokens": 18181170.0,
+      "step": 5350
+    },
+    {
+      "epoch": 0.053795816773053916,
+      "learning_rate": 0.00018924284395198523,
+      "loss": 0.8241,
+      "mean_token_accuracy": 0.761503380537033,
+      "num_tokens": 18216036.0,
+      "step": 5360
+    },
+    {
+      "epoch": 0.05389618210285439,
+      "learning_rate": 0.00018922277088602514,
+      "loss": 0.7819,
+      "mean_token_accuracy": 0.7689382612705231,
+      "num_tokens": 18249563.0,
+      "step": 5370
+    },
+    {
+      "epoch": 0.05399654743265486,
+      "learning_rate": 0.00018920269782006505,
+      "loss": 0.7923,
+      "mean_token_accuracy": 0.7651596367359161,
+      "num_tokens": 18282234.0,
+      "step": 5380
+    },
+    {
+      "epoch": 0.05409691276245534,
+      "learning_rate": 0.00018918262475410496,
+      "loss": 0.8435,
+      "mean_token_accuracy": 0.7542744994163513,
+      "num_tokens": 18315301.0,
+      "step": 5390
+    },
+    {
+      "epoch": 0.05419727809225581,
+      "learning_rate": 0.00018916255168814486,
+      "loss": 0.805,
+      "mean_token_accuracy": 0.7645738780498504,
+      "num_tokens": 18349370.0,
+      "step": 5400
+    },
+    {
+      "epoch": 0.054297643422056285,
+      "learning_rate": 0.00018914247862218477,
+      "loss": 0.837,
+      "mean_token_accuracy": 0.7534904599189758,
+      "num_tokens": 18383501.0,
+      "step": 5410
+    },
+    {
+      "epoch": 0.05439800875185676,
+      "learning_rate": 0.00018912240555622468,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7718036055564881,
+      "num_tokens": 18417804.0,
+      "step": 5420
+    },
+    {
+      "epoch": 0.05449837408165723,
+      "learning_rate": 0.00018910233249026456,
+      "loss": 0.8023,
+      "mean_token_accuracy": 0.7666881501674652,
+      "num_tokens": 18450802.0,
+      "step": 5430
+    },
+    {
+      "epoch": 0.05459873941145771,
+      "learning_rate": 0.0001890822594243045,
+      "loss": 0.8038,
+      "mean_token_accuracy": 0.7667160868644715,
+      "num_tokens": 18485355.0,
+      "step": 5440
+    },
+    {
+      "epoch": 0.05469910474125818,
+      "learning_rate": 0.00018906218635834438,
+      "loss": 0.8123,
+      "mean_token_accuracy": 0.761625474691391,
+      "num_tokens": 18519045.0,
+      "step": 5450
+    },
+    {
+      "epoch": 0.054799470071058655,
+      "learning_rate": 0.0001890421132923843,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7566151797771454,
+      "num_tokens": 18553858.0,
+      "step": 5460
+    },
+    {
+      "epoch": 0.054899835400859125,
+      "learning_rate": 0.0001890220402264242,
+      "loss": 0.8256,
+      "mean_token_accuracy": 0.7587904393672943,
+      "num_tokens": 18587941.0,
+      "step": 5470
+    },
+    {
+      "epoch": 0.0550002007306596,
+      "learning_rate": 0.0001890019671604641,
+      "loss": 0.829,
+      "mean_token_accuracy": 0.7643349528312683,
+      "num_tokens": 18621442.0,
+      "step": 5480
+    },
+    {
+      "epoch": 0.05510056606046008,
+      "learning_rate": 0.00018898189409450399,
+      "loss": 0.7805,
+      "mean_token_accuracy": 0.7764574468135834,
+      "num_tokens": 18655346.0,
+      "step": 5490
+    },
+    {
+      "epoch": 0.05520093139026055,
+      "learning_rate": 0.00018896182102854392,
+      "loss": 0.7657,
+      "mean_token_accuracy": 0.7703535377979278,
+      "num_tokens": 18689412.0,
+      "step": 5500
+    },
+    {
+      "epoch": 0.055301296720061024,
+      "learning_rate": 0.0001889417479625838,
+      "loss": 0.7821,
+      "mean_token_accuracy": 0.763988482952118,
+      "num_tokens": 18722288.0,
+      "step": 5510
+    },
+    {
+      "epoch": 0.055401662049861494,
+      "learning_rate": 0.0001889216748966237,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7680335581302643,
+      "num_tokens": 18756723.0,
+      "step": 5520
+    },
+    {
+      "epoch": 0.05550202737966197,
+      "learning_rate": 0.00018890160183066365,
+      "loss": 0.8263,
+      "mean_token_accuracy": 0.7613557398319244,
+      "num_tokens": 18791263.0,
+      "step": 5530
+    },
+    {
+      "epoch": 0.05560239270946244,
+      "learning_rate": 0.00018888152876470353,
+      "loss": 0.8333,
+      "mean_token_accuracy": 0.7546876013278961,
+      "num_tokens": 18825148.0,
+      "step": 5540
+    },
+    {
+      "epoch": 0.05570275803926292,
+      "learning_rate": 0.00018886145569874344,
+      "loss": 0.8102,
+      "mean_token_accuracy": 0.7602154076099396,
+      "num_tokens": 18859192.0,
+      "step": 5550
+    },
+    {
+      "epoch": 0.055803123369063394,
+      "learning_rate": 0.00018884138263278334,
+      "loss": 0.8232,
+      "mean_token_accuracy": 0.7617478847503663,
+      "num_tokens": 18893079.0,
+      "step": 5560
+    },
+    {
+      "epoch": 0.055903488698863864,
+      "learning_rate": 0.00018882130956682325,
+      "loss": 0.8078,
+      "mean_token_accuracy": 0.7636166334152221,
+      "num_tokens": 18927014.0,
+      "step": 5570
+    },
+    {
+      "epoch": 0.05600385402866434,
+      "learning_rate": 0.00018880123650086313,
+      "loss": 0.8533,
+      "mean_token_accuracy": 0.7467453896999359,
+      "num_tokens": 18962129.0,
+      "step": 5580
+    },
+    {
+      "epoch": 0.05610421935846481,
+      "learning_rate": 0.00018878116343490307,
+      "loss": 0.7834,
+      "mean_token_accuracy": 0.7670481979846955,
+      "num_tokens": 18995617.0,
+      "step": 5590
+    },
+    {
+      "epoch": 0.05620458468826529,
+      "learning_rate": 0.00018876109036894295,
+      "loss": 0.7936,
+      "mean_token_accuracy": 0.7654320299625397,
+      "num_tokens": 19029412.0,
+      "step": 5600
+    },
+    {
+      "epoch": 0.05630495001806576,
+      "learning_rate": 0.00018874101730298286,
+      "loss": 0.7918,
+      "mean_token_accuracy": 0.7685347139835358,
+      "num_tokens": 19063413.0,
+      "step": 5610
+    },
+    {
+      "epoch": 0.05640531534786623,
+      "learning_rate": 0.00018872094423702277,
+      "loss": 0.7651,
+      "mean_token_accuracy": 0.7734270513057708,
+      "num_tokens": 19097359.0,
+      "step": 5620
+    },
+    {
+      "epoch": 0.05650568067766671,
+      "learning_rate": 0.00018870087117106268,
+      "loss": 0.7928,
+      "mean_token_accuracy": 0.7665821790695191,
+      "num_tokens": 19131252.0,
+      "step": 5630
+    },
+    {
+      "epoch": 0.05660604600746718,
+      "learning_rate": 0.00018868079810510258,
+      "loss": 0.7886,
+      "mean_token_accuracy": 0.7623964071273803,
+      "num_tokens": 19165014.0,
+      "step": 5640
+    },
+    {
+      "epoch": 0.056706411337267656,
+      "learning_rate": 0.0001886607250391425,
+      "loss": 0.767,
+      "mean_token_accuracy": 0.7693248212337493,
+      "num_tokens": 19198196.0,
+      "step": 5650
+    },
+    {
+      "epoch": 0.056806776667068126,
+      "learning_rate": 0.0001886406519731824,
+      "loss": 0.8134,
+      "mean_token_accuracy": 0.7667059004306793,
+      "num_tokens": 19231774.0,
+      "step": 5660
+    },
+    {
+      "epoch": 0.0569071419968686,
+      "learning_rate": 0.00018862057890722228,
+      "loss": 0.7965,
+      "mean_token_accuracy": 0.7676830589771271,
+      "num_tokens": 19266436.0,
+      "step": 5670
+    },
+    {
+      "epoch": 0.05700750732666907,
+      "learning_rate": 0.00018860050584126222,
+      "loss": 0.7978,
+      "mean_token_accuracy": 0.7694530487060547,
+      "num_tokens": 19299446.0,
+      "step": 5680
+    },
+    {
+      "epoch": 0.05710787265646955,
+      "learning_rate": 0.0001885804327753021,
+      "loss": 0.8187,
+      "mean_token_accuracy": 0.7586834967136383,
+      "num_tokens": 19333807.0,
+      "step": 5690
+    },
+    {
+      "epoch": 0.057208237986270026,
+      "learning_rate": 0.000188560359709342,
+      "loss": 0.772,
+      "mean_token_accuracy": 0.7735353410243988,
+      "num_tokens": 19368038.0,
+      "step": 5700
+    },
+    {
+      "epoch": 0.057308603316070496,
+      "learning_rate": 0.00018854028664338192,
+      "loss": 0.8039,
+      "mean_token_accuracy": 0.7609234035015107,
+      "num_tokens": 19402947.0,
+      "step": 5710
+    },
+    {
+      "epoch": 0.05740896864587097,
+      "learning_rate": 0.00018852021357742183,
+      "loss": 0.7999,
+      "mean_token_accuracy": 0.7657185137271881,
+      "num_tokens": 19437044.0,
+      "step": 5720
+    },
+    {
+      "epoch": 0.05750933397567144,
+      "learning_rate": 0.00018850014051146173,
+      "loss": 0.7695,
+      "mean_token_accuracy": 0.7703326880931854,
+      "num_tokens": 19470934.0,
+      "step": 5730
+    },
+    {
+      "epoch": 0.05760969930547192,
+      "learning_rate": 0.00018848006744550164,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.7622361481189728,
+      "num_tokens": 19503986.0,
+      "step": 5740
+    },
+    {
+      "epoch": 0.05771006463527239,
+      "learning_rate": 0.00018845999437954155,
+      "loss": 0.8077,
+      "mean_token_accuracy": 0.768898606300354,
+      "num_tokens": 19538084.0,
+      "step": 5750
+    },
+    {
+      "epoch": 0.057810429965072865,
+      "learning_rate": 0.00018843992131358143,
+      "loss": 0.7856,
+      "mean_token_accuracy": 0.7670232474803924,
+      "num_tokens": 19572080.0,
+      "step": 5760
+    },
+    {
+      "epoch": 0.05791079529487334,
+      "learning_rate": 0.00018841984824762137,
+      "loss": 0.8298,
+      "mean_token_accuracy": 0.7527221262454986,
+      "num_tokens": 19606159.0,
+      "step": 5770
+    },
+    {
+      "epoch": 0.05801116062467381,
+      "learning_rate": 0.00018839977518166125,
+      "loss": 0.8162,
+      "mean_token_accuracy": 0.7591395795345306,
+      "num_tokens": 19640673.0,
+      "step": 5780
+    },
+    {
+      "epoch": 0.05811152595447429,
+      "learning_rate": 0.00018837970211570116,
+      "loss": 0.7917,
+      "mean_token_accuracy": 0.7707108378410339,
+      "num_tokens": 19674604.0,
+      "step": 5790
+    },
+    {
+      "epoch": 0.05821189128427476,
+      "learning_rate": 0.00018835962904974107,
+      "loss": 0.8416,
+      "mean_token_accuracy": 0.7613542258739472,
+      "num_tokens": 19707873.0,
+      "step": 5800
+    },
+    {
+      "epoch": 0.058312256614075235,
+      "learning_rate": 0.00018833955598378097,
+      "loss": 0.8096,
+      "mean_token_accuracy": 0.75683753490448,
+      "num_tokens": 19742363.0,
+      "step": 5810
+    },
+    {
+      "epoch": 0.058412621943875705,
+      "learning_rate": 0.00018831948291782086,
+      "loss": 0.81,
+      "mean_token_accuracy": 0.7634640634059906,
+      "num_tokens": 19776866.0,
+      "step": 5820
+    },
+    {
+      "epoch": 0.05851298727367618,
+      "learning_rate": 0.0001882994098518608,
+      "loss": 0.7921,
+      "mean_token_accuracy": 0.7701465368270874,
+      "num_tokens": 19810644.0,
+      "step": 5830
+    },
+    {
+      "epoch": 0.05861335260347666,
+      "learning_rate": 0.00018827933678590067,
+      "loss": 0.7958,
+      "mean_token_accuracy": 0.7674748599529266,
+      "num_tokens": 19845030.0,
+      "step": 5840
+    },
+    {
+      "epoch": 0.05871371793327713,
+      "learning_rate": 0.00018825926371994058,
+      "loss": 0.7968,
+      "mean_token_accuracy": 0.7612373232841492,
+      "num_tokens": 19878641.0,
+      "step": 5850
+    },
+    {
+      "epoch": 0.058814083263077604,
+      "learning_rate": 0.0001882391906539805,
+      "loss": 0.8012,
+      "mean_token_accuracy": 0.7664192676544189,
+      "num_tokens": 19912612.0,
+      "step": 5860
+    },
+    {
+      "epoch": 0.058914448592878074,
+      "learning_rate": 0.0001882191175880204,
+      "loss": 0.8111,
+      "mean_token_accuracy": 0.7672783672809601,
+      "num_tokens": 19946931.0,
+      "step": 5870
+    },
+    {
+      "epoch": 0.05901481392267855,
+      "learning_rate": 0.0001881990445220603,
+      "loss": 0.7656,
+      "mean_token_accuracy": 0.7663740038871765,
+      "num_tokens": 19980517.0,
+      "step": 5880
+    },
+    {
+      "epoch": 0.05911517925247902,
+      "learning_rate": 0.00018817897145610021,
+      "loss": 0.814,
+      "mean_token_accuracy": 0.7602796375751495,
+      "num_tokens": 20014767.0,
+      "step": 5890
+    },
+    {
+      "epoch": 0.0592155445822795,
+      "learning_rate": 0.00018815889839014012,
+      "loss": 0.7862,
+      "mean_token_accuracy": 0.768184608221054,
+      "num_tokens": 20049856.0,
+      "step": 5900
+    },
+    {
+      "epoch": 0.059315909912079974,
+      "learning_rate": 0.00018813882532418003,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.762929767370224,
+      "num_tokens": 20084071.0,
+      "step": 5910
+    },
+    {
+      "epoch": 0.059416275241880444,
+      "learning_rate": 0.00018811875225821994,
+      "loss": 0.8411,
+      "mean_token_accuracy": 0.7571568787097931,
+      "num_tokens": 20117537.0,
+      "step": 5920
+    },
+    {
+      "epoch": 0.05951664057168092,
+      "learning_rate": 0.00018809867919225982,
+      "loss": 0.7808,
+      "mean_token_accuracy": 0.7661379158496857,
+      "num_tokens": 20151452.0,
+      "step": 5930
+    },
+    {
+      "epoch": 0.05961700590148139,
+      "learning_rate": 0.00018807860612629976,
+      "loss": 0.8045,
+      "mean_token_accuracy": 0.7613177597522736,
+      "num_tokens": 20186055.0,
+      "step": 5940
+    },
+    {
+      "epoch": 0.05971737123128187,
+      "learning_rate": 0.00018805853306033964,
+      "loss": 0.8308,
+      "mean_token_accuracy": 0.7547689199447631,
+      "num_tokens": 20219685.0,
+      "step": 5950
+    },
+    {
+      "epoch": 0.05981773656108234,
+      "learning_rate": 0.00018803845999437955,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7654559493064881,
+      "num_tokens": 20252861.0,
+      "step": 5960
+    },
+    {
+      "epoch": 0.05991810189088281,
+      "learning_rate": 0.00018801838692841945,
+      "loss": 0.8142,
+      "mean_token_accuracy": 0.7605828762054443,
+      "num_tokens": 20287153.0,
+      "step": 5970
+    },
+    {
+      "epoch": 0.06001846722068329,
+      "learning_rate": 0.00018799831386245936,
+      "loss": 0.7865,
+      "mean_token_accuracy": 0.7667762041091919,
+      "num_tokens": 20320973.0,
+      "step": 5980
+    },
+    {
+      "epoch": 0.06011883255048376,
+      "learning_rate": 0.00018797824079649927,
+      "loss": 0.8286,
+      "mean_token_accuracy": 0.760528302192688,
+      "num_tokens": 20355336.0,
+      "step": 5990
+    },
+    {
+      "epoch": 0.060219197880284236,
+      "learning_rate": 0.00018795816773053918,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7689983487129212,
+      "num_tokens": 20389318.0,
+      "step": 6000
+    },
+    {
+      "epoch": 0.060319563210084706,
+      "learning_rate": 0.0001879380946645791,
+      "loss": 0.7804,
+      "mean_token_accuracy": 0.7755901873111725,
+      "num_tokens": 20423818.0,
+      "step": 6010
+    },
+    {
+      "epoch": 0.06041992853988518,
+      "learning_rate": 0.00018791802159861897,
+      "loss": 0.842,
+      "mean_token_accuracy": 0.7587439954280853,
+      "num_tokens": 20457672.0,
+      "step": 6020
+    },
+    {
+      "epoch": 0.06052029386968565,
+      "learning_rate": 0.0001878979485326589,
+      "loss": 0.8397,
+      "mean_token_accuracy": 0.7565126895904541,
+      "num_tokens": 20491344.0,
+      "step": 6030
+    },
+    {
+      "epoch": 0.06062065919948613,
+      "learning_rate": 0.0001878778754666988,
+      "loss": 0.7981,
+      "mean_token_accuracy": 0.7695219695568085,
+      "num_tokens": 20526064.0,
+      "step": 6040
+    },
+    {
+      "epoch": 0.060721024529286606,
+      "learning_rate": 0.0001878578024007387,
+      "loss": 0.7987,
+      "mean_token_accuracy": 0.7648639619350434,
+      "num_tokens": 20559452.0,
+      "step": 6050
+    },
+    {
+      "epoch": 0.060821389859087076,
+      "learning_rate": 0.0001878377293347786,
+      "loss": 0.809,
+      "mean_token_accuracy": 0.7631817996501923,
+      "num_tokens": 20593229.0,
+      "step": 6060
+    },
+    {
+      "epoch": 0.06092175518888755,
+      "learning_rate": 0.0001878176562688185,
+      "loss": 0.7956,
+      "mean_token_accuracy": 0.7655171811580658,
+      "num_tokens": 20626658.0,
+      "step": 6070
+    },
+    {
+      "epoch": 0.06102212051868802,
+      "learning_rate": 0.00018779758320285842,
+      "loss": 0.815,
+      "mean_token_accuracy": 0.76067014336586,
+      "num_tokens": 20660950.0,
+      "step": 6080
+    },
+    {
+      "epoch": 0.0611224858484885,
+      "learning_rate": 0.00018777751013689833,
+      "loss": 0.7855,
+      "mean_token_accuracy": 0.7704148888587952,
+      "num_tokens": 20695240.0,
+      "step": 6090
+    },
+    {
+      "epoch": 0.06122285117828897,
+      "learning_rate": 0.00018775743707093824,
+      "loss": 0.777,
+      "mean_token_accuracy": 0.7698096513748169,
+      "num_tokens": 20729588.0,
+      "step": 6100
+    },
+    {
+      "epoch": 0.061323216508089445,
+      "learning_rate": 0.00018773736400497812,
+      "loss": 0.8143,
+      "mean_token_accuracy": 0.7658511698246002,
+      "num_tokens": 20763084.0,
+      "step": 6110
+    },
+    {
+      "epoch": 0.06142358183788992,
+      "learning_rate": 0.00018771729093901805,
+      "loss": 0.7913,
+      "mean_token_accuracy": 0.7673853039741516,
+      "num_tokens": 20797083.0,
+      "step": 6120
+    },
+    {
+      "epoch": 0.06152394716769039,
+      "learning_rate": 0.00018769721787305794,
+      "loss": 0.7981,
+      "mean_token_accuracy": 0.7638794362545014,
+      "num_tokens": 20831358.0,
+      "step": 6130
+    },
+    {
+      "epoch": 0.06162431249749087,
+      "learning_rate": 0.00018767714480709784,
+      "loss": 0.8172,
+      "mean_token_accuracy": 0.7574215590953827,
+      "num_tokens": 20866454.0,
+      "step": 6140
+    },
+    {
+      "epoch": 0.06172467782729134,
+      "learning_rate": 0.00018765707174113775,
+      "loss": 0.7934,
+      "mean_token_accuracy": 0.768745094537735,
+      "num_tokens": 20899962.0,
+      "step": 6150
+    },
+    {
+      "epoch": 0.061825043157091815,
+      "learning_rate": 0.00018763699867517766,
+      "loss": 0.8202,
+      "mean_token_accuracy": 0.7607693433761596,
+      "num_tokens": 20933697.0,
+      "step": 6160
+    },
+    {
+      "epoch": 0.061925408486892285,
+      "learning_rate": 0.00018761692560921754,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7652881443500519,
+      "num_tokens": 20968060.0,
+      "step": 6170
+    },
+    {
+      "epoch": 0.06202577381669276,
+      "learning_rate": 0.00018759685254325748,
+      "loss": 0.7731,
+      "mean_token_accuracy": 0.7703279852867126,
+      "num_tokens": 21001696.0,
+      "step": 6180
+    },
+    {
+      "epoch": 0.06212613914649324,
+      "learning_rate": 0.00018757677947729736,
+      "loss": 0.8066,
+      "mean_token_accuracy": 0.7638004124164581,
+      "num_tokens": 21035406.0,
+      "step": 6190
+    },
+    {
+      "epoch": 0.06222650447629371,
+      "learning_rate": 0.00018755670641133727,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.7551860272884369,
+      "num_tokens": 21069683.0,
+      "step": 6200
+    },
+    {
+      "epoch": 0.062326869806094184,
+      "learning_rate": 0.00018753663334537718,
+      "loss": 0.7973,
+      "mean_token_accuracy": 0.7640761077404022,
+      "num_tokens": 21103087.0,
+      "step": 6210
+    },
+    {
+      "epoch": 0.062427235135894654,
+      "learning_rate": 0.00018751656027941708,
+      "loss": 0.8224,
+      "mean_token_accuracy": 0.7616025328636169,
+      "num_tokens": 21137491.0,
+      "step": 6220
+    },
+    {
+      "epoch": 0.06252760046569512,
+      "learning_rate": 0.000187496487213457,
+      "loss": 0.7994,
+      "mean_token_accuracy": 0.7634743452072144,
+      "num_tokens": 21170793.0,
+      "step": 6230
+    },
+    {
+      "epoch": 0.06262796579549561,
+      "learning_rate": 0.0001874764141474969,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.7705293118953704,
+      "num_tokens": 21204817.0,
+      "step": 6240
+    },
+    {
+      "epoch": 0.06272833112529608,
+      "learning_rate": 0.0001874563410815368,
+      "loss": 0.7837,
+      "mean_token_accuracy": 0.7658221244812011,
+      "num_tokens": 21238410.0,
+      "step": 6250
+    },
+    {
+      "epoch": 0.06282869645509655,
+      "learning_rate": 0.0001874362680155767,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7699131309986115,
+      "num_tokens": 21272207.0,
+      "step": 6260
+    },
+    {
+      "epoch": 0.06292906178489703,
+      "learning_rate": 0.00018741619494961663,
+      "loss": 0.8096,
+      "mean_token_accuracy": 0.7599456548690796,
+      "num_tokens": 21306565.0,
+      "step": 6270
+    },
+    {
+      "epoch": 0.0630294271146975,
+      "learning_rate": 0.0001873961218836565,
+      "loss": 0.8294,
+      "mean_token_accuracy": 0.7624841749668121,
+      "num_tokens": 21340759.0,
+      "step": 6280
+    },
+    {
+      "epoch": 0.06312979244449797,
+      "learning_rate": 0.00018737604881769642,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.7709579408168793,
+      "num_tokens": 21374808.0,
+      "step": 6290
+    },
+    {
+      "epoch": 0.06323015777429844,
+      "learning_rate": 0.00018735597575173632,
+      "loss": 0.8122,
+      "mean_token_accuracy": 0.768221527338028,
+      "num_tokens": 21408934.0,
+      "step": 6300
+    },
+    {
+      "epoch": 0.06333052310409892,
+      "learning_rate": 0.00018733590268577623,
+      "loss": 0.7999,
+      "mean_token_accuracy": 0.7624027013778687,
+      "num_tokens": 21443766.0,
+      "step": 6310
+    },
+    {
+      "epoch": 0.0634308884338994,
+      "learning_rate": 0.00018731582961981614,
+      "loss": 0.8013,
+      "mean_token_accuracy": 0.7636422693729401,
+      "num_tokens": 21476818.0,
+      "step": 6320
+    },
+    {
+      "epoch": 0.06353125376369986,
+      "learning_rate": 0.00018729575655385605,
+      "loss": 0.8473,
+      "mean_token_accuracy": 0.7541101336479187,
+      "num_tokens": 21510554.0,
+      "step": 6330
+    },
+    {
+      "epoch": 0.06363161909350035,
+      "learning_rate": 0.00018727568348789596,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7720778703689575,
+      "num_tokens": 21544946.0,
+      "step": 6340
+    },
+    {
+      "epoch": 0.06373198442330082,
+      "learning_rate": 0.00018725561042193584,
+      "loss": 0.8102,
+      "mean_token_accuracy": 0.7652732551097869,
+      "num_tokens": 21578855.0,
+      "step": 6350
+    },
+    {
+      "epoch": 0.06383234975310129,
+      "learning_rate": 0.00018723553735597578,
+      "loss": 0.8019,
+      "mean_token_accuracy": 0.7628812253475189,
+      "num_tokens": 21612750.0,
+      "step": 6360
+    },
+    {
+      "epoch": 0.06393271508290176,
+      "learning_rate": 0.00018721546429001566,
+      "loss": 0.8047,
+      "mean_token_accuracy": 0.7649907410144806,
+      "num_tokens": 21646545.0,
+      "step": 6370
+    },
+    {
+      "epoch": 0.06403308041270224,
+      "learning_rate": 0.00018719539122405556,
+      "loss": 0.7387,
+      "mean_token_accuracy": 0.7732436716556549,
+      "num_tokens": 21679576.0,
+      "step": 6380
+    },
+    {
+      "epoch": 0.06413344574250271,
+      "learning_rate": 0.00018717531815809547,
+      "loss": 0.7835,
+      "mean_token_accuracy": 0.7719547331333161,
+      "num_tokens": 21713415.0,
+      "step": 6390
+    },
+    {
+      "epoch": 0.06423381107230318,
+      "learning_rate": 0.00018715524509213538,
+      "loss": 0.8116,
+      "mean_token_accuracy": 0.7636037707328797,
+      "num_tokens": 21748100.0,
+      "step": 6400
+    },
+    {
+      "epoch": 0.06433417640210366,
+      "learning_rate": 0.00018713517202617526,
+      "loss": 0.8141,
+      "mean_token_accuracy": 0.7599149942398071,
+      "num_tokens": 21782455.0,
+      "step": 6410
+    },
+    {
+      "epoch": 0.06443454173190413,
+      "learning_rate": 0.0001871150989602152,
+      "loss": 0.7932,
+      "mean_token_accuracy": 0.7646901726722717,
+      "num_tokens": 21815891.0,
+      "step": 6420
+    },
+    {
+      "epoch": 0.0645349070617046,
+      "learning_rate": 0.0001870950258942551,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.7630632758140564,
+      "num_tokens": 21850242.0,
+      "step": 6430
+    },
+    {
+      "epoch": 0.06463527239150507,
+      "learning_rate": 0.00018707495282829502,
+      "loss": 0.8234,
+      "mean_token_accuracy": 0.7595869898796082,
+      "num_tokens": 21884118.0,
+      "step": 6440
+    },
+    {
+      "epoch": 0.06473563772130556,
+      "learning_rate": 0.00018705487976233492,
+      "loss": 0.8086,
+      "mean_token_accuracy": 0.7603508055210113,
+      "num_tokens": 21918370.0,
+      "step": 6450
+    },
+    {
+      "epoch": 0.06483600305110603,
+      "learning_rate": 0.0001870348066963748,
+      "loss": 0.8086,
+      "mean_token_accuracy": 0.7649601578712464,
+      "num_tokens": 21952680.0,
+      "step": 6460
+    },
+    {
+      "epoch": 0.0649363683809065,
+      "learning_rate": 0.00018701473363041474,
+      "loss": 0.7808,
+      "mean_token_accuracy": 0.7692488193511963,
+      "num_tokens": 21986556.0,
+      "step": 6470
+    },
+    {
+      "epoch": 0.06503673371070698,
+      "learning_rate": 0.00018699466056445462,
+      "loss": 0.7904,
+      "mean_token_accuracy": 0.7678544044494628,
+      "num_tokens": 22020264.0,
+      "step": 6480
+    },
+    {
+      "epoch": 0.06513709904050745,
+      "learning_rate": 0.00018697458749849453,
+      "loss": 0.8102,
+      "mean_token_accuracy": 0.757061505317688,
+      "num_tokens": 22054571.0,
+      "step": 6490
+    },
+    {
+      "epoch": 0.06523746437030792,
+      "learning_rate": 0.00018695451443253444,
+      "loss": 0.8058,
+      "mean_token_accuracy": 0.7630744695663452,
+      "num_tokens": 22088121.0,
+      "step": 6500
+    },
+    {
+      "epoch": 0.06533782970010839,
+      "learning_rate": 0.00018693444136657435,
+      "loss": 0.7819,
+      "mean_token_accuracy": 0.7664906919002533,
+      "num_tokens": 22121552.0,
+      "step": 6510
+    },
+    {
+      "epoch": 0.06543819502990887,
+      "learning_rate": 0.00018691436830061423,
+      "loss": 0.8029,
+      "mean_token_accuracy": 0.7638392865657806,
+      "num_tokens": 22155401.0,
+      "step": 6520
+    },
+    {
+      "epoch": 0.06553856035970934,
+      "learning_rate": 0.00018689429523465416,
+      "loss": 0.8128,
+      "mean_token_accuracy": 0.7656901359558106,
+      "num_tokens": 22190370.0,
+      "step": 6530
+    },
+    {
+      "epoch": 0.06563892568950981,
+      "learning_rate": 0.00018687422216869405,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7664203882217407,
+      "num_tokens": 22224623.0,
+      "step": 6540
+    },
+    {
+      "epoch": 0.0657392910193103,
+      "learning_rate": 0.00018685414910273395,
+      "loss": 0.8051,
+      "mean_token_accuracy": 0.7663968801498413,
+      "num_tokens": 22257909.0,
+      "step": 6550
+    },
+    {
+      "epoch": 0.06583965634911076,
+      "learning_rate": 0.00018683407603677386,
+      "loss": 0.7753,
+      "mean_token_accuracy": 0.7702917993068695,
+      "num_tokens": 22292308.0,
+      "step": 6560
+    },
+    {
+      "epoch": 0.06594002167891123,
+      "learning_rate": 0.00018681400297081377,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.7632131099700927,
+      "num_tokens": 22326205.0,
+      "step": 6570
+    },
+    {
+      "epoch": 0.0660403870087117,
+      "learning_rate": 0.00018679392990485368,
+      "loss": 0.7981,
+      "mean_token_accuracy": 0.7664714992046356,
+      "num_tokens": 22360295.0,
+      "step": 6580
+    },
+    {
+      "epoch": 0.06614075233851219,
+      "learning_rate": 0.0001867738568388936,
+      "loss": 0.8162,
+      "mean_token_accuracy": 0.7592804372310639,
+      "num_tokens": 22393878.0,
+      "step": 6590
+    },
+    {
+      "epoch": 0.06624111766831266,
+      "learning_rate": 0.0001867537837729335,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.7708893895149231,
+      "num_tokens": 22427401.0,
+      "step": 6600
+    },
+    {
+      "epoch": 0.06634148299811313,
+      "learning_rate": 0.00018673371070697338,
+      "loss": 0.7533,
+      "mean_token_accuracy": 0.7742358803749084,
+      "num_tokens": 22461573.0,
+      "step": 6610
+    },
+    {
+      "epoch": 0.06644184832791361,
+      "learning_rate": 0.0001867136376410133,
+      "loss": 0.7825,
+      "mean_token_accuracy": 0.7690136432647705,
+      "num_tokens": 22495932.0,
+      "step": 6620
+    },
+    {
+      "epoch": 0.06654221365771408,
+      "learning_rate": 0.0001866935645750532,
+      "loss": 0.7695,
+      "mean_token_accuracy": 0.772958654165268,
+      "num_tokens": 22530271.0,
+      "step": 6630
+    },
+    {
+      "epoch": 0.06664257898751455,
+      "learning_rate": 0.0001866734915090931,
+      "loss": 0.7696,
+      "mean_token_accuracy": 0.7698410391807556,
+      "num_tokens": 22564450.0,
+      "step": 6640
+    },
+    {
+      "epoch": 0.06674294431731502,
+      "learning_rate": 0.000186653418443133,
+      "loss": 0.7832,
+      "mean_token_accuracy": 0.7672110676765442,
+      "num_tokens": 22598384.0,
+      "step": 6650
+    },
+    {
+      "epoch": 0.0668433096471155,
+      "learning_rate": 0.00018663334537717292,
+      "loss": 0.8045,
+      "mean_token_accuracy": 0.7631735801696777,
+      "num_tokens": 22632404.0,
+      "step": 6660
+    },
+    {
+      "epoch": 0.06694367497691597,
+      "learning_rate": 0.00018661327231121283,
+      "loss": 0.8113,
+      "mean_token_accuracy": 0.7584132015705108,
+      "num_tokens": 22666582.0,
+      "step": 6670
+    },
+    {
+      "epoch": 0.06704404030671644,
+      "learning_rate": 0.00018659319924525274,
+      "loss": 0.8197,
+      "mean_token_accuracy": 0.7588787019252777,
+      "num_tokens": 22700042.0,
+      "step": 6680
+    },
+    {
+      "epoch": 0.06714440563651693,
+      "learning_rate": 0.00018657312617929265,
+      "loss": 0.7848,
+      "mean_token_accuracy": 0.7639056861400604,
+      "num_tokens": 22734093.0,
+      "step": 6690
+    },
+    {
+      "epoch": 0.0672447709663174,
+      "learning_rate": 0.00018655305311333253,
+      "loss": 0.7943,
+      "mean_token_accuracy": 0.768567031621933,
+      "num_tokens": 22768710.0,
+      "step": 6700
+    },
+    {
+      "epoch": 0.06734513629611787,
+      "learning_rate": 0.00018653298004737246,
+      "loss": 0.7933,
+      "mean_token_accuracy": 0.77075697183609,
+      "num_tokens": 22802873.0,
+      "step": 6710
+    },
+    {
+      "epoch": 0.06744550162591834,
+      "learning_rate": 0.00018651290698141234,
+      "loss": 0.7788,
+      "mean_token_accuracy": 0.7654812693595886,
+      "num_tokens": 22837076.0,
+      "step": 6720
+    },
+    {
+      "epoch": 0.06754586695571882,
+      "learning_rate": 0.00018649283391545225,
+      "loss": 0.7879,
+      "mean_token_accuracy": 0.767909950017929,
+      "num_tokens": 22870632.0,
+      "step": 6730
+    },
+    {
+      "epoch": 0.06764623228551929,
+      "learning_rate": 0.00018647276084949216,
+      "loss": 0.8248,
+      "mean_token_accuracy": 0.7626499176025391,
+      "num_tokens": 22904468.0,
+      "step": 6740
+    },
+    {
+      "epoch": 0.06774659761531976,
+      "learning_rate": 0.00018645268778353207,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.7642915844917297,
+      "num_tokens": 22938185.0,
+      "step": 6750
+    },
+    {
+      "epoch": 0.06784696294512024,
+      "learning_rate": 0.00018643261471757195,
+      "loss": 0.8006,
+      "mean_token_accuracy": 0.7666011452674866,
+      "num_tokens": 22971572.0,
+      "step": 6760
+    },
+    {
+      "epoch": 0.06794732827492071,
+      "learning_rate": 0.00018641254165161189,
+      "loss": 0.8288,
+      "mean_token_accuracy": 0.7597508609294892,
+      "num_tokens": 23005463.0,
+      "step": 6770
+    },
+    {
+      "epoch": 0.06804769360472118,
+      "learning_rate": 0.0001863924685856518,
+      "loss": 0.8122,
+      "mean_token_accuracy": 0.7596763968467712,
+      "num_tokens": 23040406.0,
+      "step": 6780
+    },
+    {
+      "epoch": 0.06814805893452165,
+      "learning_rate": 0.00018637239551969168,
+      "loss": 0.8016,
+      "mean_token_accuracy": 0.7657992899417877,
+      "num_tokens": 23074368.0,
+      "step": 6790
+    },
+    {
+      "epoch": 0.06824842426432214,
+      "learning_rate": 0.0001863523224537316,
+      "loss": 0.7887,
+      "mean_token_accuracy": 0.7637387990951539,
+      "num_tokens": 23108197.0,
+      "step": 6800
+    },
+    {
+      "epoch": 0.0683487895941226,
+      "learning_rate": 0.0001863322493877715,
+      "loss": 0.7786,
+      "mean_token_accuracy": 0.7647190570831299,
+      "num_tokens": 23142110.0,
+      "step": 6810
+    },
+    {
+      "epoch": 0.06844915492392308,
+      "learning_rate": 0.0001863121763218114,
+      "loss": 0.7971,
+      "mean_token_accuracy": 0.7630793511867523,
+      "num_tokens": 23176423.0,
+      "step": 6820
+    },
+    {
+      "epoch": 0.06854952025372356,
+      "learning_rate": 0.0001862921032558513,
+      "loss": 0.8011,
+      "mean_token_accuracy": 0.7649743497371674,
+      "num_tokens": 23210845.0,
+      "step": 6830
+    },
+    {
+      "epoch": 0.06864988558352403,
+      "learning_rate": 0.00018627203018989122,
+      "loss": 0.7749,
+      "mean_token_accuracy": 0.7704343676567078,
+      "num_tokens": 23244552.0,
+      "step": 6840
+    },
+    {
+      "epoch": 0.0687502509133245,
+      "learning_rate": 0.0001862519571239311,
+      "loss": 0.7721,
+      "mean_token_accuracy": 0.7698425590991974,
+      "num_tokens": 23277814.0,
+      "step": 6850
+    },
+    {
+      "epoch": 0.06885061624312497,
+      "learning_rate": 0.00018623188405797103,
+      "loss": 0.7882,
+      "mean_token_accuracy": 0.7701736092567444,
+      "num_tokens": 23312519.0,
+      "step": 6860
+    },
+    {
+      "epoch": 0.06895098157292545,
+      "learning_rate": 0.00018621181099201092,
+      "loss": 0.7998,
+      "mean_token_accuracy": 0.7650366485118866,
+      "num_tokens": 23345465.0,
+      "step": 6870
+    },
+    {
+      "epoch": 0.06905134690272592,
+      "learning_rate": 0.00018619173792605082,
+      "loss": 0.812,
+      "mean_token_accuracy": 0.7643380880355835,
+      "num_tokens": 23379659.0,
+      "step": 6880
+    },
+    {
+      "epoch": 0.06915171223252639,
+      "learning_rate": 0.00018617166486009073,
+      "loss": 0.7994,
+      "mean_token_accuracy": 0.7671465694904327,
+      "num_tokens": 23413846.0,
+      "step": 6890
+    },
+    {
+      "epoch": 0.06925207756232687,
+      "learning_rate": 0.00018615159179413064,
+      "loss": 0.7822,
+      "mean_token_accuracy": 0.7706863880157471,
+      "num_tokens": 23448483.0,
+      "step": 6900
+    },
+    {
+      "epoch": 0.06935244289212734,
+      "learning_rate": 0.00018613151872817055,
+      "loss": 0.8162,
+      "mean_token_accuracy": 0.7635308921337127,
+      "num_tokens": 23482708.0,
+      "step": 6910
+    },
+    {
+      "epoch": 0.06945280822192781,
+      "learning_rate": 0.00018611144566221046,
+      "loss": 0.795,
+      "mean_token_accuracy": 0.7642087519168854,
+      "num_tokens": 23516631.0,
+      "step": 6920
+    },
+    {
+      "epoch": 0.0695531735517283,
+      "learning_rate": 0.00018609137259625037,
+      "loss": 0.7962,
+      "mean_token_accuracy": 0.7656026661396027,
+      "num_tokens": 23549636.0,
+      "step": 6930
+    },
+    {
+      "epoch": 0.06965353888152877,
+      "learning_rate": 0.00018607129953029025,
+      "loss": 0.7884,
+      "mean_token_accuracy": 0.7681177496910095,
+      "num_tokens": 23583937.0,
+      "step": 6940
+    },
+    {
+      "epoch": 0.06975390421132924,
+      "learning_rate": 0.00018605122646433018,
+      "loss": 0.7803,
+      "mean_token_accuracy": 0.7695082247257232,
+      "num_tokens": 23617720.0,
+      "step": 6950
+    },
+    {
+      "epoch": 0.06985426954112971,
+      "learning_rate": 0.00018603115339837006,
+      "loss": 0.8068,
+      "mean_token_accuracy": 0.7647221088409424,
+      "num_tokens": 23652403.0,
+      "step": 6960
+    },
+    {
+      "epoch": 0.06995463487093019,
+      "learning_rate": 0.00018601108033240997,
+      "loss": 0.804,
+      "mean_token_accuracy": 0.7643355369567871,
+      "num_tokens": 23686213.0,
+      "step": 6970
+    },
+    {
+      "epoch": 0.07005500020073066,
+      "learning_rate": 0.00018599100726644988,
+      "loss": 0.7881,
+      "mean_token_accuracy": 0.7634240865707398,
+      "num_tokens": 23720434.0,
+      "step": 6980
+    },
+    {
+      "epoch": 0.07015536553053113,
+      "learning_rate": 0.0001859709342004898,
+      "loss": 0.7431,
+      "mean_token_accuracy": 0.7756571233272552,
+      "num_tokens": 23754142.0,
+      "step": 6990
+    },
+    {
+      "epoch": 0.07025573086033161,
+      "learning_rate": 0.0001859508611345297,
+      "loss": 0.7755,
+      "mean_token_accuracy": 0.7678376853466033,
+      "num_tokens": 23788474.0,
+      "step": 7000
+    },
+    {
+      "epoch": 0.07035609619013208,
+      "learning_rate": 0.0001859307880685696,
+      "loss": 0.7939,
+      "mean_token_accuracy": 0.7618476390838623,
+      "num_tokens": 23822495.0,
+      "step": 7010
+    },
+    {
+      "epoch": 0.07045646151993255,
+      "learning_rate": 0.00018591071500260952,
+      "loss": 0.7935,
+      "mean_token_accuracy": 0.7679507672786713,
+      "num_tokens": 23856653.0,
+      "step": 7020
+    },
+    {
+      "epoch": 0.07055682684973302,
+      "learning_rate": 0.00018589064193664942,
+      "loss": 0.8062,
+      "mean_token_accuracy": 0.7621446549892426,
+      "num_tokens": 23889845.0,
+      "step": 7030
+    },
+    {
+      "epoch": 0.0706571921795335,
+      "learning_rate": 0.00018587056887068933,
+      "loss": 0.7955,
+      "mean_token_accuracy": 0.7612752199172974,
+      "num_tokens": 23923933.0,
+      "step": 7040
+    },
+    {
+      "epoch": 0.07075755750933398,
+      "learning_rate": 0.0001858504958047292,
+      "loss": 0.8045,
+      "mean_token_accuracy": 0.7675998091697693,
+      "num_tokens": 23958128.0,
+      "step": 7050
+    },
+    {
+      "epoch": 0.07085792283913445,
+      "learning_rate": 0.00018583042273876915,
+      "loss": 0.7695,
+      "mean_token_accuracy": 0.7677767932415008,
+      "num_tokens": 23991852.0,
+      "step": 7060
+    },
+    {
+      "epoch": 0.07095828816893493,
+      "learning_rate": 0.00018581034967280903,
+      "loss": 0.7968,
+      "mean_token_accuracy": 0.7666783332824707,
+      "num_tokens": 24025509.0,
+      "step": 7070
+    },
+    {
+      "epoch": 0.0710586534987354,
+      "learning_rate": 0.00018579027660684894,
+      "loss": 0.7608,
+      "mean_token_accuracy": 0.7742595195770263,
+      "num_tokens": 24059977.0,
+      "step": 7080
+    },
+    {
+      "epoch": 0.07115901882853587,
+      "learning_rate": 0.00018577020354088885,
+      "loss": 0.8159,
+      "mean_token_accuracy": 0.7611697912216187,
+      "num_tokens": 24094490.0,
+      "step": 7090
+    },
+    {
+      "epoch": 0.07125938415833634,
+      "learning_rate": 0.00018575013047492876,
+      "loss": 0.7988,
+      "mean_token_accuracy": 0.7634016036987304,
+      "num_tokens": 24127862.0,
+      "step": 7100
+    },
+    {
+      "epoch": 0.07135974948813682,
+      "learning_rate": 0.00018573005740896864,
+      "loss": 0.8365,
+      "mean_token_accuracy": 0.75287726521492,
+      "num_tokens": 24162831.0,
+      "step": 7110
+    },
+    {
+      "epoch": 0.07146011481793729,
+      "learning_rate": 0.00018570998434300857,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7638778507709503,
+      "num_tokens": 24196205.0,
+      "step": 7120
+    },
+    {
+      "epoch": 0.07156048014773776,
+      "learning_rate": 0.00018568991127704848,
+      "loss": 0.816,
+      "mean_token_accuracy": 0.7571133255958558,
+      "num_tokens": 24230360.0,
+      "step": 7130
+    },
+    {
+      "epoch": 0.07166084547753825,
+      "learning_rate": 0.00018566983821108836,
+      "loss": 0.8029,
+      "mean_token_accuracy": 0.7698984265327453,
+      "num_tokens": 24264184.0,
+      "step": 7140
+    },
+    {
+      "epoch": 0.07176121080733872,
+      "learning_rate": 0.0001856497651451283,
+      "loss": 0.8,
+      "mean_token_accuracy": 0.7596933901309967,
+      "num_tokens": 24298488.0,
+      "step": 7150
+    },
+    {
+      "epoch": 0.07186157613713919,
+      "learning_rate": 0.00018562969207916818,
+      "loss": 0.7838,
+      "mean_token_accuracy": 0.7674479365348816,
+      "num_tokens": 24332168.0,
+      "step": 7160
+    },
+    {
+      "epoch": 0.07196194146693966,
+      "learning_rate": 0.0001856096190132081,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7701567053794861,
+      "num_tokens": 24367218.0,
+      "step": 7170
+    },
+    {
+      "epoch": 0.07206230679674014,
+      "learning_rate": 0.000185589545947248,
+      "loss": 0.8054,
+      "mean_token_accuracy": 0.7614714801311493,
+      "num_tokens": 24401414.0,
+      "step": 7180
+    },
+    {
+      "epoch": 0.07216267212654061,
+      "learning_rate": 0.0001855694728812879,
+      "loss": 0.8169,
+      "mean_token_accuracy": 0.7615014910697937,
+      "num_tokens": 24434667.0,
+      "step": 7190
+    },
+    {
+      "epoch": 0.07226303745634108,
+      "learning_rate": 0.00018554939981532779,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.7534982144832612,
+      "num_tokens": 24467459.0,
+      "step": 7200
+    },
+    {
+      "epoch": 0.07236340278614156,
+      "learning_rate": 0.00018552932674936772,
+      "loss": 0.825,
+      "mean_token_accuracy": 0.7568074822425842,
+      "num_tokens": 24501863.0,
+      "step": 7210
+    },
+    {
+      "epoch": 0.07246376811594203,
+      "learning_rate": 0.0001855092536834076,
+      "loss": 0.8232,
+      "mean_token_accuracy": 0.7650467276573181,
+      "num_tokens": 24536523.0,
+      "step": 7220
+    },
+    {
+      "epoch": 0.0725641334457425,
+      "learning_rate": 0.0001854891806174475,
+      "loss": 0.7662,
+      "mean_token_accuracy": 0.7714950680732727,
+      "num_tokens": 24569789.0,
+      "step": 7230
+    },
+    {
+      "epoch": 0.07266449877554297,
+      "learning_rate": 0.00018546910755148742,
+      "loss": 0.7812,
+      "mean_token_accuracy": 0.7700926840305329,
+      "num_tokens": 24603356.0,
+      "step": 7240
+    },
+    {
+      "epoch": 0.07276486410534345,
+      "learning_rate": 0.00018544903448552733,
+      "loss": 0.8097,
+      "mean_token_accuracy": 0.7624179661273957,
+      "num_tokens": 24636880.0,
+      "step": 7250
+    },
+    {
+      "epoch": 0.07286522943514392,
+      "learning_rate": 0.00018542896141956724,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.7721749603748321,
+      "num_tokens": 24670012.0,
+      "step": 7260
+    },
+    {
+      "epoch": 0.0729655947649444,
+      "learning_rate": 0.00018540888835360714,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7649307548999786,
+      "num_tokens": 24705021.0,
+      "step": 7270
+    },
+    {
+      "epoch": 0.07306596009474488,
+      "learning_rate": 0.00018538881528764705,
+      "loss": 0.8028,
+      "mean_token_accuracy": 0.7616039216518402,
+      "num_tokens": 24739152.0,
+      "step": 7280
+    },
+    {
+      "epoch": 0.07316632542454535,
+      "learning_rate": 0.00018536874222168693,
+      "loss": 0.7838,
+      "mean_token_accuracy": 0.7636917471885681,
+      "num_tokens": 24772941.0,
+      "step": 7290
+    },
+    {
+      "epoch": 0.07326669075434582,
+      "learning_rate": 0.00018534866915572687,
+      "loss": 0.7688,
+      "mean_token_accuracy": 0.7758508145809173,
+      "num_tokens": 24807079.0,
+      "step": 7300
+    },
+    {
+      "epoch": 0.07336705608414629,
+      "learning_rate": 0.00018532859608976675,
+      "loss": 0.7727,
+      "mean_token_accuracy": 0.7693508982658386,
+      "num_tokens": 24841006.0,
+      "step": 7310
+    },
+    {
+      "epoch": 0.07346742141394677,
+      "learning_rate": 0.00018530852302380666,
+      "loss": 0.8136,
+      "mean_token_accuracy": 0.7646665930747986,
+      "num_tokens": 24874708.0,
+      "step": 7320
+    },
+    {
+      "epoch": 0.07356778674374724,
+      "learning_rate": 0.00018528844995784657,
+      "loss": 0.7939,
+      "mean_token_accuracy": 0.7661986827850342,
+      "num_tokens": 24907841.0,
+      "step": 7330
+    },
+    {
+      "epoch": 0.07366815207354771,
+      "learning_rate": 0.00018526837689188648,
+      "loss": 0.8078,
+      "mean_token_accuracy": 0.76460902094841,
+      "num_tokens": 24941688.0,
+      "step": 7340
+    },
+    {
+      "epoch": 0.0737685174033482,
+      "learning_rate": 0.00018524830382592639,
+      "loss": 0.7928,
+      "mean_token_accuracy": 0.7639065086841583,
+      "num_tokens": 24975598.0,
+      "step": 7350
+    },
+    {
+      "epoch": 0.07386888273314866,
+      "learning_rate": 0.0001852282307599663,
+      "loss": 0.7972,
+      "mean_token_accuracy": 0.7660596668720245,
+      "num_tokens": 25009904.0,
+      "step": 7360
+    },
+    {
+      "epoch": 0.07396924806294913,
+      "learning_rate": 0.0001852081576940062,
+      "loss": 0.8195,
+      "mean_token_accuracy": 0.761925333738327,
+      "num_tokens": 25044157.0,
+      "step": 7370
+    },
+    {
+      "epoch": 0.0740696133927496,
+      "learning_rate": 0.00018518808462804608,
+      "loss": 0.8299,
+      "mean_token_accuracy": 0.7579504311084747,
+      "num_tokens": 25078167.0,
+      "step": 7380
+    },
+    {
+      "epoch": 0.07416997872255009,
+      "learning_rate": 0.00018516801156208602,
+      "loss": 0.8098,
+      "mean_token_accuracy": 0.7582357406616211,
+      "num_tokens": 25113228.0,
+      "step": 7390
+    },
+    {
+      "epoch": 0.07427034405235056,
+      "learning_rate": 0.0001851479384961259,
+      "loss": 0.8331,
+      "mean_token_accuracy": 0.7541563868522644,
+      "num_tokens": 25146882.0,
+      "step": 7400
+    },
+    {
+      "epoch": 0.07437070938215103,
+      "learning_rate": 0.0001851278654301658,
+      "loss": 0.8157,
+      "mean_token_accuracy": 0.7633441150188446,
+      "num_tokens": 25181415.0,
+      "step": 7410
+    },
+    {
+      "epoch": 0.07447107471195151,
+      "learning_rate": 0.00018510779236420572,
+      "loss": 0.7911,
+      "mean_token_accuracy": 0.7619963705539703,
+      "num_tokens": 25214350.0,
+      "step": 7420
+    },
+    {
+      "epoch": 0.07457144004175198,
+      "learning_rate": 0.00018508771929824563,
+      "loss": 0.8051,
+      "mean_token_accuracy": 0.7647240459918976,
+      "num_tokens": 25247448.0,
+      "step": 7430
+    },
+    {
+      "epoch": 0.07467180537155245,
+      "learning_rate": 0.0001850676462322855,
+      "loss": 0.7904,
+      "mean_token_accuracy": 0.76084805727005,
+      "num_tokens": 25281821.0,
+      "step": 7440
+    },
+    {
+      "epoch": 0.07477217070135292,
+      "learning_rate": 0.00018504757316632544,
+      "loss": 0.7992,
+      "mean_token_accuracy": 0.7663967788219452,
+      "num_tokens": 25315809.0,
+      "step": 7450
+    },
+    {
+      "epoch": 0.0748725360311534,
+      "learning_rate": 0.00018502750010036532,
+      "loss": 0.7854,
+      "mean_token_accuracy": 0.7667467296123505,
+      "num_tokens": 25349531.0,
+      "step": 7460
+    },
+    {
+      "epoch": 0.07497290136095387,
+      "learning_rate": 0.00018500742703440523,
+      "loss": 0.7869,
+      "mean_token_accuracy": 0.7643426060676575,
+      "num_tokens": 25384776.0,
+      "step": 7470
+    },
+    {
+      "epoch": 0.07507326669075434,
+      "learning_rate": 0.00018498735396844514,
+      "loss": 0.8214,
+      "mean_token_accuracy": 0.7597658336162567,
+      "num_tokens": 25418496.0,
+      "step": 7480
+    },
+    {
+      "epoch": 0.07517363202055483,
+      "learning_rate": 0.00018496728090248505,
+      "loss": 0.795,
+      "mean_token_accuracy": 0.7658503949642181,
+      "num_tokens": 25452606.0,
+      "step": 7490
+    },
+    {
+      "epoch": 0.0752739973503553,
+      "learning_rate": 0.00018494720783652496,
+      "loss": 0.8209,
+      "mean_token_accuracy": 0.7629762351512909,
+      "num_tokens": 25487228.0,
+      "step": 7500
+    },
+    {
+      "epoch": 0.07537436268015577,
+      "learning_rate": 0.00018492713477056487,
+      "loss": 0.8016,
+      "mean_token_accuracy": 0.7633273899555206,
+      "num_tokens": 25521640.0,
+      "step": 7510
+    },
+    {
+      "epoch": 0.07547472800995624,
+      "learning_rate": 0.00018490706170460477,
+      "loss": 0.7946,
+      "mean_token_accuracy": 0.768276309967041,
+      "num_tokens": 25555665.0,
+      "step": 7520
+    },
+    {
+      "epoch": 0.07557509333975672,
+      "learning_rate": 0.00018488698863864466,
+      "loss": 0.7937,
+      "mean_token_accuracy": 0.7652099788188934,
+      "num_tokens": 25589412.0,
+      "step": 7530
+    },
+    {
+      "epoch": 0.07567545866955719,
+      "learning_rate": 0.0001848669155726846,
+      "loss": 0.8144,
+      "mean_token_accuracy": 0.764784699678421,
+      "num_tokens": 25623812.0,
+      "step": 7540
+    },
+    {
+      "epoch": 0.07577582399935766,
+      "learning_rate": 0.00018484684250672447,
+      "loss": 0.7747,
+      "mean_token_accuracy": 0.7708411574363708,
+      "num_tokens": 25657610.0,
+      "step": 7550
+    },
+    {
+      "epoch": 0.07587618932915814,
+      "learning_rate": 0.00018482676944076438,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7676181554794311,
+      "num_tokens": 25692063.0,
+      "step": 7560
+    },
+    {
+      "epoch": 0.07597655465895861,
+      "learning_rate": 0.0001848066963748043,
+      "loss": 0.7648,
+      "mean_token_accuracy": 0.7707294762134552,
+      "num_tokens": 25726717.0,
+      "step": 7570
+    },
+    {
+      "epoch": 0.07607691998875908,
+      "learning_rate": 0.0001847866233088442,
+      "loss": 0.8065,
+      "mean_token_accuracy": 0.7685012996196747,
+      "num_tokens": 25760066.0,
+      "step": 7580
+    },
+    {
+      "epoch": 0.07617728531855955,
+      "learning_rate": 0.0001847665502428841,
+      "loss": 0.8118,
+      "mean_token_accuracy": 0.7668388724327088,
+      "num_tokens": 25794087.0,
+      "step": 7590
+    },
+    {
+      "epoch": 0.07627765064836003,
+      "learning_rate": 0.00018474647717692401,
+      "loss": 0.8276,
+      "mean_token_accuracy": 0.7590035438537598,
+      "num_tokens": 25827788.0,
+      "step": 7600
+    },
+    {
+      "epoch": 0.0763780159781605,
+      "learning_rate": 0.00018472640411096392,
+      "loss": 0.7972,
+      "mean_token_accuracy": 0.7652108430862427,
+      "num_tokens": 25861486.0,
+      "step": 7610
+    },
+    {
+      "epoch": 0.07647838130796097,
+      "learning_rate": 0.00018470633104500383,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7683241486549377,
+      "num_tokens": 25896175.0,
+      "step": 7620
+    },
+    {
+      "epoch": 0.07657874663776146,
+      "learning_rate": 0.00018468625797904374,
+      "loss": 0.7888,
+      "mean_token_accuracy": 0.7642484068870544,
+      "num_tokens": 25930565.0,
+      "step": 7630
+    },
+    {
+      "epoch": 0.07667911196756193,
+      "learning_rate": 0.00018466618491308362,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.7603274822235108,
+      "num_tokens": 25964200.0,
+      "step": 7640
+    },
+    {
+      "epoch": 0.0767794772973624,
+      "learning_rate": 0.00018464611184712356,
+      "loss": 0.7992,
+      "mean_token_accuracy": 0.7627223074436188,
+      "num_tokens": 25998157.0,
+      "step": 7650
+    },
+    {
+      "epoch": 0.07687984262716287,
+      "learning_rate": 0.00018462603878116344,
+      "loss": 0.8036,
+      "mean_token_accuracy": 0.7680601716041565,
+      "num_tokens": 26032645.0,
+      "step": 7660
+    },
+    {
+      "epoch": 0.07698020795696335,
+      "learning_rate": 0.00018460596571520335,
+      "loss": 0.7804,
+      "mean_token_accuracy": 0.7695116221904754,
+      "num_tokens": 26066749.0,
+      "step": 7670
+    },
+    {
+      "epoch": 0.07708057328676382,
+      "learning_rate": 0.00018458589264924325,
+      "loss": 0.8028,
+      "mean_token_accuracy": 0.7702985823154449,
+      "num_tokens": 26100853.0,
+      "step": 7680
+    },
+    {
+      "epoch": 0.07718093861656429,
+      "learning_rate": 0.00018456581958328316,
+      "loss": 0.7806,
+      "mean_token_accuracy": 0.7728835463523864,
+      "num_tokens": 26134719.0,
+      "step": 7690
+    },
+    {
+      "epoch": 0.07728130394636477,
+      "learning_rate": 0.00018454574651732307,
+      "loss": 0.7949,
+      "mean_token_accuracy": 0.7665352702140809,
+      "num_tokens": 26169126.0,
+      "step": 7700
+    },
+    {
+      "epoch": 0.07738166927616524,
+      "learning_rate": 0.00018452567345136298,
+      "loss": 0.7926,
+      "mean_token_accuracy": 0.7647746622562408,
+      "num_tokens": 26202249.0,
+      "step": 7710
+    },
+    {
+      "epoch": 0.07748203460596571,
+      "learning_rate": 0.0001845056003854029,
+      "loss": 0.8082,
+      "mean_token_accuracy": 0.7587997853755951,
+      "num_tokens": 26236838.0,
+      "step": 7720
+    },
+    {
+      "epoch": 0.07758239993576618,
+      "learning_rate": 0.00018448552731944277,
+      "loss": 0.7877,
+      "mean_token_accuracy": 0.7682729542255402,
+      "num_tokens": 26271047.0,
+      "step": 7730
+    },
+    {
+      "epoch": 0.07768276526556667,
+      "learning_rate": 0.0001844654542534827,
+      "loss": 0.825,
+      "mean_token_accuracy": 0.7565823674201966,
+      "num_tokens": 26305741.0,
+      "step": 7740
+    },
+    {
+      "epoch": 0.07778313059536714,
+      "learning_rate": 0.0001844453811875226,
+      "loss": 0.826,
+      "mean_token_accuracy": 0.7599569141864777,
+      "num_tokens": 26339915.0,
+      "step": 7750
+    },
+    {
+      "epoch": 0.0778834959251676,
+      "learning_rate": 0.0001844253081215625,
+      "loss": 0.7724,
+      "mean_token_accuracy": 0.7702906250953674,
+      "num_tokens": 26374166.0,
+      "step": 7760
+    },
+    {
+      "epoch": 0.07798386125496809,
+      "learning_rate": 0.0001844052350556024,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.769506961107254,
+      "num_tokens": 26407465.0,
+      "step": 7770
+    },
+    {
+      "epoch": 0.07808422658476856,
+      "learning_rate": 0.0001843851619896423,
+      "loss": 0.8037,
+      "mean_token_accuracy": 0.7590090334415436,
+      "num_tokens": 26441910.0,
+      "step": 7780
+    },
+    {
+      "epoch": 0.07818459191456903,
+      "learning_rate": 0.0001843650889236822,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.7618598759174346,
+      "num_tokens": 26475249.0,
+      "step": 7790
+    },
+    {
+      "epoch": 0.0782849572443695,
+      "learning_rate": 0.00018434501585772213,
+      "loss": 0.7804,
+      "mean_token_accuracy": 0.7672445714473725,
+      "num_tokens": 26510028.0,
+      "step": 7800
+    },
+    {
+      "epoch": 0.07838532257416998,
+      "learning_rate": 0.000184324942791762,
+      "loss": 0.821,
+      "mean_token_accuracy": 0.764594703912735,
+      "num_tokens": 26543386.0,
+      "step": 7810
+    },
+    {
+      "epoch": 0.07848568790397045,
+      "learning_rate": 0.00018430486972580192,
+      "loss": 0.8064,
+      "mean_token_accuracy": 0.7608125746250153,
+      "num_tokens": 26576560.0,
+      "step": 7820
+    },
+    {
+      "epoch": 0.07858605323377092,
+      "learning_rate": 0.00018428479665984183,
+      "loss": 0.8145,
+      "mean_token_accuracy": 0.7621243953704834,
+      "num_tokens": 26609899.0,
+      "step": 7830
+    },
+    {
+      "epoch": 0.0786864185635714,
+      "learning_rate": 0.00018426472359388174,
+      "loss": 0.7989,
+      "mean_token_accuracy": 0.7658592283725738,
+      "num_tokens": 26644506.0,
+      "step": 7840
+    },
+    {
+      "epoch": 0.07878678389337188,
+      "learning_rate": 0.00018424465052792164,
+      "loss": 0.7831,
+      "mean_token_accuracy": 0.7703222453594207,
+      "num_tokens": 26678803.0,
+      "step": 7850
+    },
+    {
+      "epoch": 0.07888714922317235,
+      "learning_rate": 0.00018422457746196155,
+      "loss": 0.7862,
+      "mean_token_accuracy": 0.7615216493606567,
+      "num_tokens": 26712506.0,
+      "step": 7860
+    },
+    {
+      "epoch": 0.07898751455297282,
+      "learning_rate": 0.00018420450439600146,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7699089527130127,
+      "num_tokens": 26746081.0,
+      "step": 7870
+    },
+    {
+      "epoch": 0.0790878798827733,
+      "learning_rate": 0.00018418443133004134,
+      "loss": 0.7888,
+      "mean_token_accuracy": 0.7717827081680297,
+      "num_tokens": 26779973.0,
+      "step": 7880
+    },
+    {
+      "epoch": 0.07918824521257377,
+      "learning_rate": 0.00018416435826408128,
+      "loss": 0.8071,
+      "mean_token_accuracy": 0.7643128275871277,
+      "num_tokens": 26814769.0,
+      "step": 7890
+    },
+    {
+      "epoch": 0.07928861054237424,
+      "learning_rate": 0.00018414428519812116,
+      "loss": 0.7867,
+      "mean_token_accuracy": 0.7633980870246887,
+      "num_tokens": 26848312.0,
+      "step": 7900
+    },
+    {
+      "epoch": 0.07938897587217472,
+      "learning_rate": 0.00018412421213216107,
+      "loss": 0.8387,
+      "mean_token_accuracy": 0.7578362107276917,
+      "num_tokens": 26881942.0,
+      "step": 7910
+    },
+    {
+      "epoch": 0.07948934120197519,
+      "learning_rate": 0.00018410413906620098,
+      "loss": 0.8043,
+      "mean_token_accuracy": 0.7660013616085053,
+      "num_tokens": 26916642.0,
+      "step": 7920
+    },
+    {
+      "epoch": 0.07958970653177566,
+      "learning_rate": 0.00018408406600024088,
+      "loss": 0.7709,
+      "mean_token_accuracy": 0.7725791335105896,
+      "num_tokens": 26950942.0,
+      "step": 7930
+    },
+    {
+      "epoch": 0.07969007186157613,
+      "learning_rate": 0.0001840639929342808,
+      "loss": 0.8346,
+      "mean_token_accuracy": 0.7565326988697052,
+      "num_tokens": 26985030.0,
+      "step": 7940
+    },
+    {
+      "epoch": 0.07979043719137661,
+      "learning_rate": 0.0001840439198683207,
+      "loss": 0.781,
+      "mean_token_accuracy": 0.7720654487609864,
+      "num_tokens": 27018067.0,
+      "step": 7950
+    },
+    {
+      "epoch": 0.07989080252117708,
+      "learning_rate": 0.0001840238468023606,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.75545853972435,
+      "num_tokens": 27051615.0,
+      "step": 7960
+    },
+    {
+      "epoch": 0.07999116785097755,
+      "learning_rate": 0.0001840037737364005,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7600622177124023,
+      "num_tokens": 27085181.0,
+      "step": 7970
+    },
+    {
+      "epoch": 0.08009153318077804,
+      "learning_rate": 0.00018398370067044043,
+      "loss": 0.814,
+      "mean_token_accuracy": 0.7612695872783661,
+      "num_tokens": 27119116.0,
+      "step": 7980
+    },
+    {
+      "epoch": 0.08019189851057851,
+      "learning_rate": 0.0001839636276044803,
+      "loss": 0.8058,
+      "mean_token_accuracy": 0.7696108520030975,
+      "num_tokens": 27154478.0,
+      "step": 7990
+    },
+    {
+      "epoch": 0.08029226384037898,
+      "learning_rate": 0.00018394355453852022,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7640083312988282,
+      "num_tokens": 27188890.0,
+      "step": 8000
+    },
+    {
+      "epoch": 0.08039262917017945,
+      "learning_rate": 0.00018392348147256012,
+      "loss": 0.8061,
+      "mean_token_accuracy": 0.7589396297931671,
+      "num_tokens": 27222650.0,
+      "step": 8010
+    },
+    {
+      "epoch": 0.08049299449997993,
+      "learning_rate": 0.00018390340840660003,
+      "loss": 0.8015,
+      "mean_token_accuracy": 0.7603662014007568,
+      "num_tokens": 27256163.0,
+      "step": 8020
+    },
+    {
+      "epoch": 0.0805933598297804,
+      "learning_rate": 0.00018388333534063994,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.766388326883316,
+      "num_tokens": 27289939.0,
+      "step": 8030
+    },
+    {
+      "epoch": 0.08069372515958087,
+      "learning_rate": 0.00018386326227467985,
+      "loss": 0.7853,
+      "mean_token_accuracy": 0.7673959374427796,
+      "num_tokens": 27324482.0,
+      "step": 8040
+    },
+    {
+      "epoch": 0.08079409048938135,
+      "learning_rate": 0.00018384318920871976,
+      "loss": 0.8211,
+      "mean_token_accuracy": 0.7607272922992706,
+      "num_tokens": 27358513.0,
+      "step": 8050
+    },
+    {
+      "epoch": 0.08089445581918182,
+      "learning_rate": 0.00018382311614275964,
+      "loss": 0.801,
+      "mean_token_accuracy": 0.7661828458309173,
+      "num_tokens": 27392044.0,
+      "step": 8060
+    },
+    {
+      "epoch": 0.0809948211489823,
+      "learning_rate": 0.00018380304307679958,
+      "loss": 0.8001,
+      "mean_token_accuracy": 0.7657620906829834,
+      "num_tokens": 27425618.0,
+      "step": 8070
+    },
+    {
+      "epoch": 0.08109518647878276,
+      "learning_rate": 0.00018378297001083946,
+      "loss": 0.7737,
+      "mean_token_accuracy": 0.773236620426178,
+      "num_tokens": 27460506.0,
+      "step": 8080
+    },
+    {
+      "epoch": 0.08119555180858325,
+      "learning_rate": 0.00018376289694487937,
+      "loss": 0.7914,
+      "mean_token_accuracy": 0.7647178828716278,
+      "num_tokens": 27494448.0,
+      "step": 8090
+    },
+    {
+      "epoch": 0.08129591713838372,
+      "learning_rate": 0.00018374282387891927,
+      "loss": 0.8235,
+      "mean_token_accuracy": 0.7590629518032074,
+      "num_tokens": 27529463.0,
+      "step": 8100
+    },
+    {
+      "epoch": 0.08139628246818419,
+      "learning_rate": 0.00018372275081295918,
+      "loss": 0.7411,
+      "mean_token_accuracy": 0.7821636378765107,
+      "num_tokens": 27564132.0,
+      "step": 8110
+    },
+    {
+      "epoch": 0.08149664779798467,
+      "learning_rate": 0.00018370267774699906,
+      "loss": 0.8074,
+      "mean_token_accuracy": 0.7685580670833587,
+      "num_tokens": 27597888.0,
+      "step": 8120
+    },
+    {
+      "epoch": 0.08159701312778514,
+      "learning_rate": 0.000183682604681039,
+      "loss": 0.8054,
+      "mean_token_accuracy": 0.765894103050232,
+      "num_tokens": 27631531.0,
+      "step": 8130
+    },
+    {
+      "epoch": 0.08169737845758561,
+      "learning_rate": 0.00018366253161507888,
+      "loss": 0.7771,
+      "mean_token_accuracy": 0.7699494779109954,
+      "num_tokens": 27665523.0,
+      "step": 8140
+    },
+    {
+      "epoch": 0.08179774378738608,
+      "learning_rate": 0.00018364245854911882,
+      "loss": 0.7586,
+      "mean_token_accuracy": 0.773178231716156,
+      "num_tokens": 27699754.0,
+      "step": 8150
+    },
+    {
+      "epoch": 0.08189810911718656,
+      "learning_rate": 0.0001836223854831587,
+      "loss": 0.7507,
+      "mean_token_accuracy": 0.7754209101200104,
+      "num_tokens": 27733210.0,
+      "step": 8160
+    },
+    {
+      "epoch": 0.08199847444698703,
+      "learning_rate": 0.0001836023124171986,
+      "loss": 0.7712,
+      "mean_token_accuracy": 0.7670442402362824,
+      "num_tokens": 27767777.0,
+      "step": 8170
+    },
+    {
+      "epoch": 0.0820988397767875,
+      "learning_rate": 0.00018358223935123851,
+      "loss": 0.8371,
+      "mean_token_accuracy": 0.7562036275863647,
+      "num_tokens": 27801779.0,
+      "step": 8180
+    },
+    {
+      "epoch": 0.08219920510658799,
+      "learning_rate": 0.00018356216628527842,
+      "loss": 0.7753,
+      "mean_token_accuracy": 0.7712762773036956,
+      "num_tokens": 27835493.0,
+      "step": 8190
+    },
+    {
+      "epoch": 0.08229957043638846,
+      "learning_rate": 0.00018354209321931833,
+      "loss": 0.7857,
+      "mean_token_accuracy": 0.7714332640171051,
+      "num_tokens": 27868848.0,
+      "step": 8200
+    },
+    {
+      "epoch": 0.08239993576618893,
+      "learning_rate": 0.00018352202015335824,
+      "loss": 0.7889,
+      "mean_token_accuracy": 0.764836972951889,
+      "num_tokens": 27902676.0,
+      "step": 8210
+    },
+    {
+      "epoch": 0.0825003010959894,
+      "learning_rate": 0.00018350194708739815,
+      "loss": 0.7917,
+      "mean_token_accuracy": 0.7639157354831696,
+      "num_tokens": 27936893.0,
+      "step": 8220
+    },
+    {
+      "epoch": 0.08260066642578988,
+      "learning_rate": 0.00018348187402143803,
+      "loss": 0.8124,
+      "mean_token_accuracy": 0.7611936211585999,
+      "num_tokens": 27971349.0,
+      "step": 8230
+    },
+    {
+      "epoch": 0.08270103175559035,
+      "learning_rate": 0.00018346180095547796,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7635602414608001,
+      "num_tokens": 28005360.0,
+      "step": 8240
+    },
+    {
+      "epoch": 0.08280139708539082,
+      "learning_rate": 0.00018344172788951785,
+      "loss": 0.8182,
+      "mean_token_accuracy": 0.7626757025718689,
+      "num_tokens": 28039496.0,
+      "step": 8250
+    },
+    {
+      "epoch": 0.0829017624151913,
+      "learning_rate": 0.00018342165482355775,
+      "loss": 0.8021,
+      "mean_token_accuracy": 0.7612757861614228,
+      "num_tokens": 28073988.0,
+      "step": 8260
+    },
+    {
+      "epoch": 0.08300212774499177,
+      "learning_rate": 0.00018340158175759766,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7759433329105377,
+      "num_tokens": 28108105.0,
+      "step": 8270
+    },
+    {
+      "epoch": 0.08310249307479224,
+      "learning_rate": 0.00018338150869163757,
+      "loss": 0.783,
+      "mean_token_accuracy": 0.7666752696037292,
+      "num_tokens": 28141570.0,
+      "step": 8280
+    },
+    {
+      "epoch": 0.08320285840459271,
+      "learning_rate": 0.00018336143562567748,
+      "loss": 0.7812,
+      "mean_token_accuracy": 0.7679860413074493,
+      "num_tokens": 28175667.0,
+      "step": 8290
+    },
+    {
+      "epoch": 0.0833032237343932,
+      "learning_rate": 0.0001833413625597174,
+      "loss": 0.7937,
+      "mean_token_accuracy": 0.771624231338501,
+      "num_tokens": 28209112.0,
+      "step": 8300
+    },
+    {
+      "epoch": 0.08340358906419366,
+      "learning_rate": 0.0001833212894937573,
+      "loss": 0.7765,
+      "mean_token_accuracy": 0.7699212431907654,
+      "num_tokens": 28242271.0,
+      "step": 8310
+    },
+    {
+      "epoch": 0.08350395439399413,
+      "learning_rate": 0.00018330121642779718,
+      "loss": 0.8077,
+      "mean_token_accuracy": 0.7627488255500794,
+      "num_tokens": 28275861.0,
+      "step": 8320
+    },
+    {
+      "epoch": 0.08360431972379462,
+      "learning_rate": 0.0001832811433618371,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.757368391752243,
+      "num_tokens": 28308668.0,
+      "step": 8330
+    },
+    {
+      "epoch": 0.08370468505359509,
+      "learning_rate": 0.000183261070295877,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.766782796382904,
+      "num_tokens": 28342565.0,
+      "step": 8340
+    },
+    {
+      "epoch": 0.08380505038339556,
+      "learning_rate": 0.0001832409972299169,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7720321476459503,
+      "num_tokens": 28376767.0,
+      "step": 8350
+    },
+    {
+      "epoch": 0.08390541571319603,
+      "learning_rate": 0.0001832209241639568,
+      "loss": 0.7766,
+      "mean_token_accuracy": 0.7695994555950165,
+      "num_tokens": 28409804.0,
+      "step": 8360
+    },
+    {
+      "epoch": 0.08400578104299651,
+      "learning_rate": 0.00018320085109799672,
+      "loss": 0.815,
+      "mean_token_accuracy": 0.7584184944629669,
+      "num_tokens": 28444956.0,
+      "step": 8370
+    },
+    {
+      "epoch": 0.08410614637279698,
+      "learning_rate": 0.00018318077803203663,
+      "loss": 0.825,
+      "mean_token_accuracy": 0.7594221532344818,
+      "num_tokens": 28478777.0,
+      "step": 8380
+    },
+    {
+      "epoch": 0.08420651170259745,
+      "learning_rate": 0.00018316070496607654,
+      "loss": 0.8124,
+      "mean_token_accuracy": 0.761949211359024,
+      "num_tokens": 28513212.0,
+      "step": 8390
+    },
+    {
+      "epoch": 0.08430687703239793,
+      "learning_rate": 0.00018314063190011645,
+      "loss": 0.7748,
+      "mean_token_accuracy": 0.7677722752094269,
+      "num_tokens": 28547306.0,
+      "step": 8400
+    },
+    {
+      "epoch": 0.0844072423621984,
+      "learning_rate": 0.00018312055883415633,
+      "loss": 0.8043,
+      "mean_token_accuracy": 0.7592228293418884,
+      "num_tokens": 28580947.0,
+      "step": 8410
+    },
+    {
+      "epoch": 0.08450760769199887,
+      "learning_rate": 0.00018310048576819626,
+      "loss": 0.7855,
+      "mean_token_accuracy": 0.7644309282302857,
+      "num_tokens": 28615511.0,
+      "step": 8420
+    },
+    {
+      "epoch": 0.08460797302179934,
+      "learning_rate": 0.00018308041270223614,
+      "loss": 0.7721,
+      "mean_token_accuracy": 0.7723537027835846,
+      "num_tokens": 28650137.0,
+      "step": 8430
+    },
+    {
+      "epoch": 0.08470833835159983,
+      "learning_rate": 0.00018306033963627605,
+      "loss": 0.85,
+      "mean_token_accuracy": 0.7568152248859406,
+      "num_tokens": 28684894.0,
+      "step": 8440
+    },
+    {
+      "epoch": 0.0848087036814003,
+      "learning_rate": 0.00018304026657031596,
+      "loss": 0.7921,
+      "mean_token_accuracy": 0.766416597366333,
+      "num_tokens": 28718933.0,
+      "step": 8450
+    },
+    {
+      "epoch": 0.08490906901120077,
+      "learning_rate": 0.00018302019350435587,
+      "loss": 0.8001,
+      "mean_token_accuracy": 0.7644974291324615,
+      "num_tokens": 28752719.0,
+      "step": 8460
+    },
+    {
+      "epoch": 0.08500943434100125,
+      "learning_rate": 0.00018300012043839575,
+      "loss": 0.8108,
+      "mean_token_accuracy": 0.7660545587539673,
+      "num_tokens": 28787113.0,
+      "step": 8470
+    },
+    {
+      "epoch": 0.08510979967080172,
+      "learning_rate": 0.00018298004737243569,
+      "loss": 0.7598,
+      "mean_token_accuracy": 0.7736972510814667,
+      "num_tokens": 28821222.0,
+      "step": 8480
+    },
+    {
+      "epoch": 0.08521016500060219,
+      "learning_rate": 0.00018295997430647557,
+      "loss": 0.7866,
+      "mean_token_accuracy": 0.7729543089866638,
+      "num_tokens": 28855377.0,
+      "step": 8490
+    },
+    {
+      "epoch": 0.08531053033040266,
+      "learning_rate": 0.00018293990124051548,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.7719655096530914,
+      "num_tokens": 28889960.0,
+      "step": 8500
+    },
+    {
+      "epoch": 0.08541089566020314,
+      "learning_rate": 0.00018291982817455538,
+      "loss": 0.8095,
+      "mean_token_accuracy": 0.7601831316947937,
+      "num_tokens": 28924470.0,
+      "step": 8510
+    },
+    {
+      "epoch": 0.08551126099000361,
+      "learning_rate": 0.0001828997551085953,
+      "loss": 0.8041,
+      "mean_token_accuracy": 0.7582235693931579,
+      "num_tokens": 28957673.0,
+      "step": 8520
+    },
+    {
+      "epoch": 0.08561162631980408,
+      "learning_rate": 0.0001828796820426352,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7658786296844482,
+      "num_tokens": 28991285.0,
+      "step": 8530
+    },
+    {
+      "epoch": 0.08571199164960457,
+      "learning_rate": 0.0001828596089766751,
+      "loss": 0.7883,
+      "mean_token_accuracy": 0.766554594039917,
+      "num_tokens": 29025102.0,
+      "step": 8540
+    },
+    {
+      "epoch": 0.08581235697940504,
+      "learning_rate": 0.00018283953591071502,
+      "loss": 0.7781,
+      "mean_token_accuracy": 0.7628990948200226,
+      "num_tokens": 29059866.0,
+      "step": 8550
+    },
+    {
+      "epoch": 0.0859127223092055,
+      "learning_rate": 0.0001828194628447549,
+      "loss": 0.792,
+      "mean_token_accuracy": 0.7650415420532226,
+      "num_tokens": 29094399.0,
+      "step": 8560
+    },
+    {
+      "epoch": 0.08601308763900598,
+      "learning_rate": 0.00018279938977879483,
+      "loss": 0.8129,
+      "mean_token_accuracy": 0.7575330376625061,
+      "num_tokens": 29128148.0,
+      "step": 8570
+    },
+    {
+      "epoch": 0.08611345296880646,
+      "learning_rate": 0.00018277931671283472,
+      "loss": 0.8154,
+      "mean_token_accuracy": 0.7631605803966522,
+      "num_tokens": 29162146.0,
+      "step": 8580
+    },
+    {
+      "epoch": 0.08621381829860693,
+      "learning_rate": 0.00018275924364687462,
+      "loss": 0.7924,
+      "mean_token_accuracy": 0.7658519327640534,
+      "num_tokens": 29197231.0,
+      "step": 8590
+    },
+    {
+      "epoch": 0.0863141836284074,
+      "learning_rate": 0.00018273917058091453,
+      "loss": 0.7935,
+      "mean_token_accuracy": 0.7624973893165589,
+      "num_tokens": 29231195.0,
+      "step": 8600
+    },
+    {
+      "epoch": 0.08641454895820788,
+      "learning_rate": 0.00018271909751495444,
+      "loss": 0.8187,
+      "mean_token_accuracy": 0.7632297515869141,
+      "num_tokens": 29265717.0,
+      "step": 8610
+    },
+    {
+      "epoch": 0.08651491428800835,
+      "learning_rate": 0.00018269902444899435,
+      "loss": 0.7962,
+      "mean_token_accuracy": 0.7655292451381683,
+      "num_tokens": 29299063.0,
+      "step": 8620
+    },
+    {
+      "epoch": 0.08661527961780882,
+      "learning_rate": 0.00018267895138303426,
+      "loss": 0.779,
+      "mean_token_accuracy": 0.7777730166912079,
+      "num_tokens": 29333094.0,
+      "step": 8630
+    },
+    {
+      "epoch": 0.08671564494760929,
+      "learning_rate": 0.00018265887831707417,
+      "loss": 0.8155,
+      "mean_token_accuracy": 0.7669682502746582,
+      "num_tokens": 29367809.0,
+      "step": 8640
+    },
+    {
+      "epoch": 0.08681601027740977,
+      "learning_rate": 0.00018263880525111405,
+      "loss": 0.813,
+      "mean_token_accuracy": 0.7619749963283539,
+      "num_tokens": 29401517.0,
+      "step": 8650
+    },
+    {
+      "epoch": 0.08691637560721024,
+      "learning_rate": 0.00018261873218515398,
+      "loss": 0.7736,
+      "mean_token_accuracy": 0.7691554009914399,
+      "num_tokens": 29435888.0,
+      "step": 8660
+    },
+    {
+      "epoch": 0.08701674093701071,
+      "learning_rate": 0.00018259865911919386,
+      "loss": 0.7964,
+      "mean_token_accuracy": 0.7643998801708222,
+      "num_tokens": 29469495.0,
+      "step": 8670
+    },
+    {
+      "epoch": 0.0871171062668112,
+      "learning_rate": 0.00018257858605323377,
+      "loss": 0.803,
+      "mean_token_accuracy": 0.7638863921165466,
+      "num_tokens": 29503322.0,
+      "step": 8680
+    },
+    {
+      "epoch": 0.08721747159661167,
+      "learning_rate": 0.00018255851298727368,
+      "loss": 0.7765,
+      "mean_token_accuracy": 0.7672784090042114,
+      "num_tokens": 29537549.0,
+      "step": 8690
+    },
+    {
+      "epoch": 0.08731783692641214,
+      "learning_rate": 0.0001825384399213136,
+      "loss": 0.8106,
+      "mean_token_accuracy": 0.7580951035022736,
+      "num_tokens": 29572158.0,
+      "step": 8700
+    },
+    {
+      "epoch": 0.08741820225621261,
+      "learning_rate": 0.0001825183668553535,
+      "loss": 0.8063,
+      "mean_token_accuracy": 0.7586085200309753,
+      "num_tokens": 29606342.0,
+      "step": 8710
+    },
+    {
+      "epoch": 0.08751856758601309,
+      "learning_rate": 0.0001824982937893934,
+      "loss": 0.8092,
+      "mean_token_accuracy": 0.76483913064003,
+      "num_tokens": 29639627.0,
+      "step": 8720
+    },
+    {
+      "epoch": 0.08761893291581356,
+      "learning_rate": 0.00018247822072343332,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7659261226654053,
+      "num_tokens": 29673673.0,
+      "step": 8730
+    },
+    {
+      "epoch": 0.08771929824561403,
+      "learning_rate": 0.00018245814765747322,
+      "loss": 0.773,
+      "mean_token_accuracy": 0.7755686104297638,
+      "num_tokens": 29707564.0,
+      "step": 8740
+    },
+    {
+      "epoch": 0.08781966357541451,
+      "learning_rate": 0.00018243807459151313,
+      "loss": 0.7927,
+      "mean_token_accuracy": 0.7612753689289093,
+      "num_tokens": 29741585.0,
+      "step": 8750
+    },
+    {
+      "epoch": 0.08792002890521498,
+      "learning_rate": 0.00018241800152555301,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7699299335479737,
+      "num_tokens": 29775306.0,
+      "step": 8760
+    },
+    {
+      "epoch": 0.08802039423501545,
+      "learning_rate": 0.00018239792845959295,
+      "loss": 0.7927,
+      "mean_token_accuracy": 0.7664937555789948,
+      "num_tokens": 29808521.0,
+      "step": 8770
+    },
+    {
+      "epoch": 0.08812075956481592,
+      "learning_rate": 0.00018237785539363283,
+      "loss": 0.781,
+      "mean_token_accuracy": 0.7734834969043731,
+      "num_tokens": 29841888.0,
+      "step": 8780
+    },
+    {
+      "epoch": 0.0882211248946164,
+      "learning_rate": 0.00018235778232767274,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.7714802920818329,
+      "num_tokens": 29875937.0,
+      "step": 8790
+    },
+    {
+      "epoch": 0.08832149022441688,
+      "learning_rate": 0.00018233770926171265,
+      "loss": 0.7641,
+      "mean_token_accuracy": 0.7700235188007355,
+      "num_tokens": 29909628.0,
+      "step": 8800
+    },
+    {
+      "epoch": 0.08842185555421735,
+      "learning_rate": 0.00018231763619575256,
+      "loss": 0.7819,
+      "mean_token_accuracy": 0.7715400278568267,
+      "num_tokens": 29943880.0,
+      "step": 8810
+    },
+    {
+      "epoch": 0.08852222088401783,
+      "learning_rate": 0.00018229756312979244,
+      "loss": 0.7568,
+      "mean_token_accuracy": 0.7760605990886689,
+      "num_tokens": 29977827.0,
+      "step": 8820
+    },
+    {
+      "epoch": 0.0886225862138183,
+      "learning_rate": 0.00018227749006383237,
+      "loss": 0.7517,
+      "mean_token_accuracy": 0.780300235748291,
+      "num_tokens": 30011993.0,
+      "step": 8830
+    },
+    {
+      "epoch": 0.08872295154361877,
+      "learning_rate": 0.00018225741699787225,
+      "loss": 0.7856,
+      "mean_token_accuracy": 0.763048404455185,
+      "num_tokens": 30046264.0,
+      "step": 8840
+    },
+    {
+      "epoch": 0.08882331687341925,
+      "learning_rate": 0.00018223734393191216,
+      "loss": 0.8117,
+      "mean_token_accuracy": 0.7608951687812805,
+      "num_tokens": 30080946.0,
+      "step": 8850
+    },
+    {
+      "epoch": 0.08892368220321972,
+      "learning_rate": 0.00018221727086595207,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7685857653617859,
+      "num_tokens": 30114139.0,
+      "step": 8860
+    },
+    {
+      "epoch": 0.08902404753302019,
+      "learning_rate": 0.00018219719779999198,
+      "loss": 0.7778,
+      "mean_token_accuracy": 0.7680936872959137,
+      "num_tokens": 30148943.0,
+      "step": 8870
+    },
+    {
+      "epoch": 0.08912441286282066,
+      "learning_rate": 0.0001821771247340319,
+      "loss": 0.7883,
+      "mean_token_accuracy": 0.7649091064929963,
+      "num_tokens": 30182765.0,
+      "step": 8880
+    },
+    {
+      "epoch": 0.08922477819262115,
+      "learning_rate": 0.0001821570516680718,
+      "loss": 0.7993,
+      "mean_token_accuracy": 0.7652966022491455,
+      "num_tokens": 30216428.0,
+      "step": 8890
+    },
+    {
+      "epoch": 0.08932514352242162,
+      "learning_rate": 0.0001821369786021117,
+      "loss": 0.805,
+      "mean_token_accuracy": 0.7664391279220581,
+      "num_tokens": 30249886.0,
+      "step": 8900
+    },
+    {
+      "epoch": 0.08942550885222209,
+      "learning_rate": 0.00018211690553615159,
+      "loss": 0.7606,
+      "mean_token_accuracy": 0.7713489532470703,
+      "num_tokens": 30284478.0,
+      "step": 8910
+    },
+    {
+      "epoch": 0.08952587418202257,
+      "learning_rate": 0.00018209683247019152,
+      "loss": 0.8324,
+      "mean_token_accuracy": 0.7595540165901185,
+      "num_tokens": 30318152.0,
+      "step": 8920
+    },
+    {
+      "epoch": 0.08962623951182304,
+      "learning_rate": 0.0001820767594042314,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.7607889533042907,
+      "num_tokens": 30352097.0,
+      "step": 8930
+    },
+    {
+      "epoch": 0.08972660484162351,
+      "learning_rate": 0.0001820566863382713,
+      "loss": 0.7659,
+      "mean_token_accuracy": 0.7716902375221253,
+      "num_tokens": 30386993.0,
+      "step": 8940
+    },
+    {
+      "epoch": 0.08982697017142398,
+      "learning_rate": 0.00018203661327231122,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.7665991187095642,
+      "num_tokens": 30421404.0,
+      "step": 8950
+    },
+    {
+      "epoch": 0.08992733550122446,
+      "learning_rate": 0.00018201654020635113,
+      "loss": 0.8137,
+      "mean_token_accuracy": 0.7588241100311279,
+      "num_tokens": 30455514.0,
+      "step": 8960
+    },
+    {
+      "epoch": 0.09002770083102493,
+      "learning_rate": 0.00018199646714039104,
+      "loss": 0.7789,
+      "mean_token_accuracy": 0.7674042224884033,
+      "num_tokens": 30488335.0,
+      "step": 8970
+    },
+    {
+      "epoch": 0.0901280661608254,
+      "learning_rate": 0.00018197639407443094,
+      "loss": 0.7598,
+      "mean_token_accuracy": 0.7746587514877319,
+      "num_tokens": 30522594.0,
+      "step": 8980
+    },
+    {
+      "epoch": 0.09022843149062588,
+      "learning_rate": 0.00018195632100847085,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7681942880153656,
+      "num_tokens": 30556175.0,
+      "step": 8990
+    },
+    {
+      "epoch": 0.09032879682042635,
+      "learning_rate": 0.00018193624794251073,
+      "loss": 0.8158,
+      "mean_token_accuracy": 0.7650823891162872,
+      "num_tokens": 30590301.0,
+      "step": 9000
+    },
+    {
+      "epoch": 0.09042916215022682,
+      "learning_rate": 0.00018191617487655067,
+      "loss": 0.7983,
+      "mean_token_accuracy": 0.7667717099189758,
+      "num_tokens": 30624330.0,
+      "step": 9010
+    },
+    {
+      "epoch": 0.0905295274800273,
+      "learning_rate": 0.00018189610181059055,
+      "loss": 0.7941,
+      "mean_token_accuracy": 0.7668941855430603,
+      "num_tokens": 30658103.0,
+      "step": 9020
+    },
+    {
+      "epoch": 0.09062989280982778,
+      "learning_rate": 0.00018187602874463046,
+      "loss": 0.7656,
+      "mean_token_accuracy": 0.7736150443553924,
+      "num_tokens": 30692457.0,
+      "step": 9030
+    },
+    {
+      "epoch": 0.09073025813962825,
+      "learning_rate": 0.00018185595567867037,
+      "loss": 0.8097,
+      "mean_token_accuracy": 0.7703277707099915,
+      "num_tokens": 30725813.0,
+      "step": 9040
+    },
+    {
+      "epoch": 0.09083062346942872,
+      "learning_rate": 0.00018183588261271028,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7693876445293426,
+      "num_tokens": 30760046.0,
+      "step": 9050
+    },
+    {
+      "epoch": 0.0909309887992292,
+      "learning_rate": 0.00018181580954675016,
+      "loss": 0.7914,
+      "mean_token_accuracy": 0.7648736119270325,
+      "num_tokens": 30794043.0,
+      "step": 9060
+    },
+    {
+      "epoch": 0.09103135412902967,
+      "learning_rate": 0.0001817957364807901,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7657828867435456,
+      "num_tokens": 30827493.0,
+      "step": 9070
+    },
+    {
+      "epoch": 0.09113171945883014,
+      "learning_rate": 0.00018177566341482998,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.7668934762477875,
+      "num_tokens": 30861987.0,
+      "step": 9080
+    },
+    {
+      "epoch": 0.09123208478863061,
+      "learning_rate": 0.00018175559034886988,
+      "loss": 0.847,
+      "mean_token_accuracy": 0.7515221536159515,
+      "num_tokens": 30895773.0,
+      "step": 9090
+    },
+    {
+      "epoch": 0.0913324501184311,
+      "learning_rate": 0.00018173551728290982,
+      "loss": 0.7897,
+      "mean_token_accuracy": 0.7714468240737915,
+      "num_tokens": 30930474.0,
+      "step": 9100
+    },
+    {
+      "epoch": 0.09143281544823156,
+      "learning_rate": 0.0001817154442169497,
+      "loss": 0.7664,
+      "mean_token_accuracy": 0.7724806427955627,
+      "num_tokens": 30964582.0,
+      "step": 9110
+    },
+    {
+      "epoch": 0.09153318077803203,
+      "learning_rate": 0.0001816953711509896,
+      "loss": 0.7767,
+      "mean_token_accuracy": 0.7632918655872345,
+      "num_tokens": 30998736.0,
+      "step": 9120
+    },
+    {
+      "epoch": 0.09163354610783252,
+      "learning_rate": 0.00018167529808502952,
+      "loss": 0.8133,
+      "mean_token_accuracy": 0.7608244955539704,
+      "num_tokens": 31033467.0,
+      "step": 9130
+    },
+    {
+      "epoch": 0.09173391143763299,
+      "learning_rate": 0.00018165522501906943,
+      "loss": 0.7775,
+      "mean_token_accuracy": 0.7722931623458862,
+      "num_tokens": 31067147.0,
+      "step": 9140
+    },
+    {
+      "epoch": 0.09183427676743346,
+      "learning_rate": 0.0001816351519531093,
+      "loss": 0.787,
+      "mean_token_accuracy": 0.7637144982814789,
+      "num_tokens": 31101075.0,
+      "step": 9150
+    },
+    {
+      "epoch": 0.09193464209723393,
+      "learning_rate": 0.00018161507888714924,
+      "loss": 0.8038,
+      "mean_token_accuracy": 0.7647253572940826,
+      "num_tokens": 31134763.0,
+      "step": 9160
+    },
+    {
+      "epoch": 0.09203500742703441,
+      "learning_rate": 0.00018159500582118912,
+      "loss": 0.7955,
+      "mean_token_accuracy": 0.7666331231594086,
+      "num_tokens": 31169302.0,
+      "step": 9170
+    },
+    {
+      "epoch": 0.09213537275683488,
+      "learning_rate": 0.00018157493275522903,
+      "loss": 0.7631,
+      "mean_token_accuracy": 0.7757180273532868,
+      "num_tokens": 31203530.0,
+      "step": 9180
+    },
+    {
+      "epoch": 0.09223573808663535,
+      "learning_rate": 0.00018155485968926894,
+      "loss": 0.7848,
+      "mean_token_accuracy": 0.7622224748134613,
+      "num_tokens": 31237590.0,
+      "step": 9190
+    },
+    {
+      "epoch": 0.09233610341643583,
+      "learning_rate": 0.00018153478662330885,
+      "loss": 0.8142,
+      "mean_token_accuracy": 0.7676658689975738,
+      "num_tokens": 31272006.0,
+      "step": 9200
+    },
+    {
+      "epoch": 0.0924364687462363,
+      "learning_rate": 0.00018151471355734876,
+      "loss": 0.7727,
+      "mean_token_accuracy": 0.768430107831955,
+      "num_tokens": 31306088.0,
+      "step": 9210
+    },
+    {
+      "epoch": 0.09253683407603677,
+      "learning_rate": 0.00018149464049138867,
+      "loss": 0.7906,
+      "mean_token_accuracy": 0.7649446725845337,
+      "num_tokens": 31339380.0,
+      "step": 9220
+    },
+    {
+      "epoch": 0.09263719940583724,
+      "learning_rate": 0.00018147456742542857,
+      "loss": 0.8076,
+      "mean_token_accuracy": 0.768731153011322,
+      "num_tokens": 31374097.0,
+      "step": 9230
+    },
+    {
+      "epoch": 0.09273756473563773,
+      "learning_rate": 0.00018145449435946846,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7694973766803741,
+      "num_tokens": 31407852.0,
+      "step": 9240
+    },
+    {
+      "epoch": 0.0928379300654382,
+      "learning_rate": 0.0001814344212935084,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.7713075160980225,
+      "num_tokens": 31441940.0,
+      "step": 9250
+    },
+    {
+      "epoch": 0.09293829539523867,
+      "learning_rate": 0.00018141434822754827,
+      "loss": 0.7901,
+      "mean_token_accuracy": 0.7695594906806946,
+      "num_tokens": 31476335.0,
+      "step": 9260
+    },
+    {
+      "epoch": 0.09303866072503915,
+      "learning_rate": 0.00018139427516158818,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7696987748146057,
+      "num_tokens": 31510293.0,
+      "step": 9270
+    },
+    {
+      "epoch": 0.09313902605483962,
+      "learning_rate": 0.0001813742020956281,
+      "loss": 0.7658,
+      "mean_token_accuracy": 0.770842558145523,
+      "num_tokens": 31543727.0,
+      "step": 9280
+    },
+    {
+      "epoch": 0.09323939138464009,
+      "learning_rate": 0.000181354129029668,
+      "loss": 0.7833,
+      "mean_token_accuracy": 0.7661703586578369,
+      "num_tokens": 31577350.0,
+      "step": 9290
+    },
+    {
+      "epoch": 0.09333975671444056,
+      "learning_rate": 0.0001813340559637079,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.7651128113269806,
+      "num_tokens": 31611382.0,
+      "step": 9300
+    },
+    {
+      "epoch": 0.09344012204424104,
+      "learning_rate": 0.00018131398289774781,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7651427328586579,
+      "num_tokens": 31645425.0,
+      "step": 9310
+    },
+    {
+      "epoch": 0.09354048737404151,
+      "learning_rate": 0.00018129390983178772,
+      "loss": 0.7876,
+      "mean_token_accuracy": 0.763335132598877,
+      "num_tokens": 31679613.0,
+      "step": 9320
+    },
+    {
+      "epoch": 0.09364085270384198,
+      "learning_rate": 0.00018127383676582763,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.7680640757083893,
+      "num_tokens": 31713418.0,
+      "step": 9330
+    },
+    {
+      "epoch": 0.09374121803364246,
+      "learning_rate": 0.00018125376369986754,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7704873025417328,
+      "num_tokens": 31747103.0,
+      "step": 9340
+    },
+    {
+      "epoch": 0.09384158336344293,
+      "learning_rate": 0.00018123369063390742,
+      "loss": 0.7988,
+      "mean_token_accuracy": 0.7634690701961517,
+      "num_tokens": 31780966.0,
+      "step": 9350
+    },
+    {
+      "epoch": 0.0939419486932434,
+      "learning_rate": 0.00018121361756794736,
+      "loss": 0.8216,
+      "mean_token_accuracy": 0.7603338420391083,
+      "num_tokens": 31814446.0,
+      "step": 9360
+    },
+    {
+      "epoch": 0.09404231402304387,
+      "learning_rate": 0.00018119354450198724,
+      "loss": 0.7842,
+      "mean_token_accuracy": 0.7677425026893616,
+      "num_tokens": 31849355.0,
+      "step": 9370
+    },
+    {
+      "epoch": 0.09414267935284436,
+      "learning_rate": 0.00018117347143602715,
+      "loss": 0.7905,
+      "mean_token_accuracy": 0.768053936958313,
+      "num_tokens": 31883524.0,
+      "step": 9380
+    },
+    {
+      "epoch": 0.09424304468264483,
+      "learning_rate": 0.00018115339837006706,
+      "loss": 0.818,
+      "mean_token_accuracy": 0.7665216565132141,
+      "num_tokens": 31916837.0,
+      "step": 9390
+    },
+    {
+      "epoch": 0.0943434100124453,
+      "learning_rate": 0.00018113332530410696,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.7733565986156463,
+      "num_tokens": 31950958.0,
+      "step": 9400
+    },
+    {
+      "epoch": 0.09444377534224578,
+      "learning_rate": 0.00018111325223814685,
+      "loss": 0.8145,
+      "mean_token_accuracy": 0.7630878150463104,
+      "num_tokens": 31985483.0,
+      "step": 9410
+    },
+    {
+      "epoch": 0.09454414067204625,
+      "learning_rate": 0.00018109317917218678,
+      "loss": 0.7871,
+      "mean_token_accuracy": 0.7698763608932495,
+      "num_tokens": 32019177.0,
+      "step": 9420
+    },
+    {
+      "epoch": 0.09464450600184672,
+      "learning_rate": 0.00018107310610622666,
+      "loss": 0.797,
+      "mean_token_accuracy": 0.7613940715789795,
+      "num_tokens": 32052859.0,
+      "step": 9430
+    },
+    {
+      "epoch": 0.09474487133164719,
+      "learning_rate": 0.00018105303304026657,
+      "loss": 0.7935,
+      "mean_token_accuracy": 0.7656313002109527,
+      "num_tokens": 32086478.0,
+      "step": 9440
+    },
+    {
+      "epoch": 0.09484523666144767,
+      "learning_rate": 0.0001810329599743065,
+      "loss": 0.8287,
+      "mean_token_accuracy": 0.7582162857055664,
+      "num_tokens": 32120084.0,
+      "step": 9450
+    },
+    {
+      "epoch": 0.09494560199124814,
+      "learning_rate": 0.0001810128869083464,
+      "loss": 0.7766,
+      "mean_token_accuracy": 0.7686365067958831,
+      "num_tokens": 32154693.0,
+      "step": 9460
+    },
+    {
+      "epoch": 0.09504596732104861,
+      "learning_rate": 0.0001809928138423863,
+      "loss": 0.8127,
+      "mean_token_accuracy": 0.7605416655540467,
+      "num_tokens": 32188248.0,
+      "step": 9470
+    },
+    {
+      "epoch": 0.0951463326508491,
+      "learning_rate": 0.0001809727407764262,
+      "loss": 0.7957,
+      "mean_token_accuracy": 0.7632535398006439,
+      "num_tokens": 32222370.0,
+      "step": 9480
+    },
+    {
+      "epoch": 0.09524669798064957,
+      "learning_rate": 0.0001809526677104661,
+      "loss": 0.7844,
+      "mean_token_accuracy": 0.7648251771926879,
+      "num_tokens": 32256138.0,
+      "step": 9490
+    },
+    {
+      "epoch": 0.09534706331045004,
+      "learning_rate": 0.000180932594644506,
+      "loss": 0.7997,
+      "mean_token_accuracy": 0.7668353676795959,
+      "num_tokens": 32290337.0,
+      "step": 9500
+    },
+    {
+      "epoch": 0.0954474286402505,
+      "learning_rate": 0.00018091252157854593,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7700516879558563,
+      "num_tokens": 32324631.0,
+      "step": 9510
+    },
+    {
+      "epoch": 0.09554779397005099,
+      "learning_rate": 0.0001808924485125858,
+      "loss": 0.7645,
+      "mean_token_accuracy": 0.7741646885871887,
+      "num_tokens": 32358596.0,
+      "step": 9520
+    },
+    {
+      "epoch": 0.09564815929985146,
+      "learning_rate": 0.00018087237544662572,
+      "loss": 0.7946,
+      "mean_token_accuracy": 0.7669793128967285,
+      "num_tokens": 32393177.0,
+      "step": 9530
+    },
+    {
+      "epoch": 0.09574852462965193,
+      "learning_rate": 0.00018085230238066563,
+      "loss": 0.7961,
+      "mean_token_accuracy": 0.7640884757041931,
+      "num_tokens": 32427640.0,
+      "step": 9540
+    },
+    {
+      "epoch": 0.09584888995945241,
+      "learning_rate": 0.00018083222931470554,
+      "loss": 0.7889,
+      "mean_token_accuracy": 0.76604523062706,
+      "num_tokens": 32462185.0,
+      "step": 9550
+    },
+    {
+      "epoch": 0.09594925528925288,
+      "learning_rate": 0.00018081215624874544,
+      "loss": 0.7562,
+      "mean_token_accuracy": 0.7752231001853943,
+      "num_tokens": 32496524.0,
+      "step": 9560
+    },
+    {
+      "epoch": 0.09604962061905335,
+      "learning_rate": 0.00018079208318278535,
+      "loss": 0.7754,
+      "mean_token_accuracy": 0.7691753149032593,
+      "num_tokens": 32530189.0,
+      "step": 9570
+    },
+    {
+      "epoch": 0.09614998594885382,
+      "learning_rate": 0.00018077201011682526,
+      "loss": 0.7747,
+      "mean_token_accuracy": 0.7719720363616943,
+      "num_tokens": 32564461.0,
+      "step": 9580
+    },
+    {
+      "epoch": 0.0962503512786543,
+      "learning_rate": 0.00018075193705086514,
+      "loss": 0.7645,
+      "mean_token_accuracy": 0.7691185414791107,
+      "num_tokens": 32598437.0,
+      "step": 9590
+    },
+    {
+      "epoch": 0.09635071660845478,
+      "learning_rate": 0.00018073186398490508,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.7618482708930969,
+      "num_tokens": 32632841.0,
+      "step": 9600
+    },
+    {
+      "epoch": 0.09645108193825525,
+      "learning_rate": 0.00018071179091894496,
+      "loss": 0.8236,
+      "mean_token_accuracy": 0.7573092579841614,
+      "num_tokens": 32667525.0,
+      "step": 9610
+    },
+    {
+      "epoch": 0.09655144726805573,
+      "learning_rate": 0.00018069171785298487,
+      "loss": 0.7857,
+      "mean_token_accuracy": 0.7663219451904297,
+      "num_tokens": 32701330.0,
+      "step": 9620
+    },
+    {
+      "epoch": 0.0966518125978562,
+      "learning_rate": 0.00018067164478702478,
+      "loss": 0.7591,
+      "mean_token_accuracy": 0.7785889625549316,
+      "num_tokens": 32736175.0,
+      "step": 9630
+    },
+    {
+      "epoch": 0.09675217792765667,
+      "learning_rate": 0.00018065157172106468,
+      "loss": 0.7752,
+      "mean_token_accuracy": 0.7701292335987091,
+      "num_tokens": 32769996.0,
+      "step": 9640
+    },
+    {
+      "epoch": 0.09685254325745714,
+      "learning_rate": 0.0001806314986551046,
+      "loss": 0.7483,
+      "mean_token_accuracy": 0.7770219802856445,
+      "num_tokens": 32803281.0,
+      "step": 9650
+    },
+    {
+      "epoch": 0.09695290858725762,
+      "learning_rate": 0.0001806114255891445,
+      "loss": 0.7839,
+      "mean_token_accuracy": 0.769777649641037,
+      "num_tokens": 32837087.0,
+      "step": 9660
+    },
+    {
+      "epoch": 0.09705327391705809,
+      "learning_rate": 0.0001805913525231844,
+      "loss": 0.7771,
+      "mean_token_accuracy": 0.7717701971530915,
+      "num_tokens": 32871869.0,
+      "step": 9670
+    },
+    {
+      "epoch": 0.09715363924685856,
+      "learning_rate": 0.0001805712794572243,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7715424120426178,
+      "num_tokens": 32905679.0,
+      "step": 9680
+    },
+    {
+      "epoch": 0.09725400457665904,
+      "learning_rate": 0.00018055120639126423,
+      "loss": 0.7865,
+      "mean_token_accuracy": 0.7647394895553589,
+      "num_tokens": 32939742.0,
+      "step": 9690
+    },
+    {
+      "epoch": 0.09735436990645951,
+      "learning_rate": 0.0001805311333253041,
+      "loss": 0.787,
+      "mean_token_accuracy": 0.7655929684638977,
+      "num_tokens": 32974249.0,
+      "step": 9700
+    },
+    {
+      "epoch": 0.09745473523625998,
+      "learning_rate": 0.00018051106025934402,
+      "loss": 0.8307,
+      "mean_token_accuracy": 0.7551734566688537,
+      "num_tokens": 33007680.0,
+      "step": 9710
+    },
+    {
+      "epoch": 0.09755510056606045,
+      "learning_rate": 0.00018049098719338393,
+      "loss": 0.7873,
+      "mean_token_accuracy": 0.7685339212417602,
+      "num_tokens": 33041980.0,
+      "step": 9720
+    },
+    {
+      "epoch": 0.09765546589586094,
+      "learning_rate": 0.00018047091412742383,
+      "loss": 0.7787,
+      "mean_token_accuracy": 0.7710682153701782,
+      "num_tokens": 33077135.0,
+      "step": 9730
+    },
+    {
+      "epoch": 0.09775583122566141,
+      "learning_rate": 0.00018045084106146372,
+      "loss": 0.7827,
+      "mean_token_accuracy": 0.7675726175308227,
+      "num_tokens": 33110794.0,
+      "step": 9740
+    },
+    {
+      "epoch": 0.09785619655546188,
+      "learning_rate": 0.00018043076799550365,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.7666629195213318,
+      "num_tokens": 33144064.0,
+      "step": 9750
+    },
+    {
+      "epoch": 0.09795656188526236,
+      "learning_rate": 0.00018041069492954353,
+      "loss": 0.7768,
+      "mean_token_accuracy": 0.7662809491157532,
+      "num_tokens": 33178610.0,
+      "step": 9760
+    },
+    {
+      "epoch": 0.09805692721506283,
+      "learning_rate": 0.00018039062186358344,
+      "loss": 0.7484,
+      "mean_token_accuracy": 0.7748051345348358,
+      "num_tokens": 33211709.0,
+      "step": 9770
+    },
+    {
+      "epoch": 0.0981572925448633,
+      "learning_rate": 0.00018037054879762335,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.7520861685276031,
+      "num_tokens": 33245552.0,
+      "step": 9780
+    },
+    {
+      "epoch": 0.09825765787466377,
+      "learning_rate": 0.00018035047573166326,
+      "loss": 0.7679,
+      "mean_token_accuracy": 0.771792608499527,
+      "num_tokens": 33280308.0,
+      "step": 9790
+    },
+    {
+      "epoch": 0.09835802320446425,
+      "learning_rate": 0.00018033040266570317,
+      "loss": 0.7574,
+      "mean_token_accuracy": 0.7754374146461487,
+      "num_tokens": 33313523.0,
+      "step": 9800
+    },
+    {
+      "epoch": 0.09845838853426472,
+      "learning_rate": 0.00018031032959974307,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7665438055992126,
+      "num_tokens": 33348278.0,
+      "step": 9810
+    },
+    {
+      "epoch": 0.0985587538640652,
+      "learning_rate": 0.00018029025653378298,
+      "loss": 0.7522,
+      "mean_token_accuracy": 0.7775404155254364,
+      "num_tokens": 33383632.0,
+      "step": 9820
+    },
+    {
+      "epoch": 0.09865911919386568,
+      "learning_rate": 0.00018027018346782286,
+      "loss": 0.7928,
+      "mean_token_accuracy": 0.7668094635009766,
+      "num_tokens": 33418353.0,
+      "step": 9830
+    },
+    {
+      "epoch": 0.09875948452366615,
+      "learning_rate": 0.0001802501104018628,
+      "loss": 0.8022,
+      "mean_token_accuracy": 0.7708871126174927,
+      "num_tokens": 33452423.0,
+      "step": 9840
+    },
+    {
+      "epoch": 0.09885984985346662,
+      "learning_rate": 0.00018023003733590268,
+      "loss": 0.7816,
+      "mean_token_accuracy": 0.7649633705615997,
+      "num_tokens": 33486446.0,
+      "step": 9850
+    },
+    {
+      "epoch": 0.09896021518326709,
+      "learning_rate": 0.00018020996426994262,
+      "loss": 0.8215,
+      "mean_token_accuracy": 0.7617051899433136,
+      "num_tokens": 33519984.0,
+      "step": 9860
+    },
+    {
+      "epoch": 0.09906058051306757,
+      "learning_rate": 0.0001801898912039825,
+      "loss": 0.8101,
+      "mean_token_accuracy": 0.7568509519100189,
+      "num_tokens": 33554988.0,
+      "step": 9870
+    },
+    {
+      "epoch": 0.09916094584286804,
+      "learning_rate": 0.0001801698181380224,
+      "loss": 0.7937,
+      "mean_token_accuracy": 0.7650300979614257,
+      "num_tokens": 33589544.0,
+      "step": 9880
+    },
+    {
+      "epoch": 0.09926131117266851,
+      "learning_rate": 0.00018014974507206231,
+      "loss": 0.7901,
+      "mean_token_accuracy": 0.7662486851215362,
+      "num_tokens": 33623148.0,
+      "step": 9890
+    },
+    {
+      "epoch": 0.09936167650246899,
+      "learning_rate": 0.00018012967200610222,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.770133638381958,
+      "num_tokens": 33657576.0,
+      "step": 9900
+    },
+    {
+      "epoch": 0.09946204183226946,
+      "learning_rate": 0.00018010959894014213,
+      "loss": 0.8253,
+      "mean_token_accuracy": 0.7663521468639374,
+      "num_tokens": 33691433.0,
+      "step": 9910
+    },
+    {
+      "epoch": 0.09956240716206993,
+      "learning_rate": 0.00018008952587418204,
+      "loss": 0.8082,
+      "mean_token_accuracy": 0.7643501996994019,
+      "num_tokens": 33725430.0,
+      "step": 9920
+    },
+    {
+      "epoch": 0.0996627724918704,
+      "learning_rate": 0.00018006945280822195,
+      "loss": 0.7961,
+      "mean_token_accuracy": 0.7636694729328155,
+      "num_tokens": 33759779.0,
+      "step": 9930
+    },
+    {
+      "epoch": 0.09976313782167089,
+      "learning_rate": 0.00018004937974226183,
+      "loss": 0.7872,
+      "mean_token_accuracy": 0.7669819295406342,
+      "num_tokens": 33794158.0,
+      "step": 9940
+    },
+    {
+      "epoch": 0.09986350315147136,
+      "learning_rate": 0.00018002930667630177,
+      "loss": 0.757,
+      "mean_token_accuracy": 0.7733592569828034,
+      "num_tokens": 33828132.0,
+      "step": 9950
+    },
+    {
+      "epoch": 0.09996386848127183,
+      "learning_rate": 0.00018000923361034165,
+      "loss": 0.7771,
+      "mean_token_accuracy": 0.7673372983932495,
+      "num_tokens": 33862552.0,
+      "step": 9960
+    },
+    {
+      "epoch": 0.10006423381107231,
+      "learning_rate": 0.00017998916054438155,
+      "loss": 0.8106,
+      "mean_token_accuracy": 0.7603930771350861,
+      "num_tokens": 33895266.0,
+      "step": 9970
+    },
+    {
+      "epoch": 0.10016459914087278,
+      "learning_rate": 0.00017996908747842146,
+      "loss": 0.7752,
+      "mean_token_accuracy": 0.7717368602752686,
+      "num_tokens": 33928639.0,
+      "step": 9980
+    },
+    {
+      "epoch": 0.10026496447067325,
+      "learning_rate": 0.00017994901441246137,
+      "loss": 0.7536,
+      "mean_token_accuracy": 0.7732951700687408,
+      "num_tokens": 33962493.0,
+      "step": 9990
+    },
+    {
+      "epoch": 0.10036532980047372,
+      "learning_rate": 0.00017992894134650128,
+      "loss": 0.7912,
+      "mean_token_accuracy": 0.7701098620891571,
+      "num_tokens": 33995585.0,
+      "step": 10000
+    },
+    {
+      "epoch": 0.1004656951302742,
+      "learning_rate": 0.0001799088682805412,
+      "loss": 0.8131,
+      "mean_token_accuracy": 0.7622293293476105,
+      "num_tokens": 34029514.0,
+      "step": 10010
+    },
+    {
+      "epoch": 0.10056606046007467,
+      "learning_rate": 0.0001798887952145811,
+      "loss": 0.7735,
+      "mean_token_accuracy": 0.7684919655323028,
+      "num_tokens": 34062780.0,
+      "step": 10020
+    },
+    {
+      "epoch": 0.10066642578987514,
+      "learning_rate": 0.00017986872214862098,
+      "loss": 0.766,
+      "mean_token_accuracy": 0.7764488875865936,
+      "num_tokens": 34096411.0,
+      "step": 10030
+    },
+    {
+      "epoch": 0.10076679111967562,
+      "learning_rate": 0.00017984864908266091,
+      "loss": 0.7748,
+      "mean_token_accuracy": 0.7689181506633759,
+      "num_tokens": 34130781.0,
+      "step": 10040
+    },
+    {
+      "epoch": 0.1008671564494761,
+      "learning_rate": 0.0001798285760167008,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.7614941120147705,
+      "num_tokens": 34164604.0,
+      "step": 10050
+    },
+    {
+      "epoch": 0.10096752177927656,
+      "learning_rate": 0.0001798085029507407,
+      "loss": 0.8176,
+      "mean_token_accuracy": 0.7642528653144837,
+      "num_tokens": 34198526.0,
+      "step": 10060
+    },
+    {
+      "epoch": 0.10106788710907703,
+      "learning_rate": 0.0001797884298847806,
+      "loss": 0.8101,
+      "mean_token_accuracy": 0.7603436768054962,
+      "num_tokens": 34232408.0,
+      "step": 10070
+    },
+    {
+      "epoch": 0.10116825243887752,
+      "learning_rate": 0.00017976835681882052,
+      "loss": 0.8015,
+      "mean_token_accuracy": 0.7639372229576111,
+      "num_tokens": 34266649.0,
+      "step": 10080
+    },
+    {
+      "epoch": 0.10126861776867799,
+      "learning_rate": 0.0001797482837528604,
+      "loss": 0.7727,
+      "mean_token_accuracy": 0.7697113692760468,
+      "num_tokens": 34300849.0,
+      "step": 10090
+    },
+    {
+      "epoch": 0.10136898309847846,
+      "learning_rate": 0.00017972821068690034,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7766829371452332,
+      "num_tokens": 34335580.0,
+      "step": 10100
+    },
+    {
+      "epoch": 0.10146934842827894,
+      "learning_rate": 0.00017970813762094022,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.7638198852539062,
+      "num_tokens": 34370229.0,
+      "step": 10110
+    },
+    {
+      "epoch": 0.10156971375807941,
+      "learning_rate": 0.00017968806455498013,
+      "loss": 0.7881,
+      "mean_token_accuracy": 0.7672926723957062,
+      "num_tokens": 34404429.0,
+      "step": 10120
+    },
+    {
+      "epoch": 0.10167007908787988,
+      "learning_rate": 0.00017966799148902004,
+      "loss": 0.7835,
+      "mean_token_accuracy": 0.7664349377155304,
+      "num_tokens": 34437979.0,
+      "step": 10130
+    },
+    {
+      "epoch": 0.10177044441768035,
+      "learning_rate": 0.00017964791842305994,
+      "loss": 0.8127,
+      "mean_token_accuracy": 0.758906751871109,
+      "num_tokens": 34472067.0,
+      "step": 10140
+    },
+    {
+      "epoch": 0.10187080974748083,
+      "learning_rate": 0.00017962784535709985,
+      "loss": 0.7766,
+      "mean_token_accuracy": 0.7701891541481019,
+      "num_tokens": 34506834.0,
+      "step": 10150
+    },
+    {
+      "epoch": 0.1019711750772813,
+      "learning_rate": 0.00017960777229113976,
+      "loss": 0.7671,
+      "mean_token_accuracy": 0.7738985598087311,
+      "num_tokens": 34542076.0,
+      "step": 10160
+    },
+    {
+      "epoch": 0.10207154040708177,
+      "learning_rate": 0.00017958769922517967,
+      "loss": 0.7977,
+      "mean_token_accuracy": 0.7637203812599183,
+      "num_tokens": 34576739.0,
+      "step": 10170
+    },
+    {
+      "epoch": 0.10217190573688226,
+      "learning_rate": 0.00017956762615921955,
+      "loss": 0.7948,
+      "mean_token_accuracy": 0.7669504582881927,
+      "num_tokens": 34610972.0,
+      "step": 10180
+    },
+    {
+      "epoch": 0.10227227106668273,
+      "learning_rate": 0.00017954755309325949,
+      "loss": 0.74,
+      "mean_token_accuracy": 0.7822540640830994,
+      "num_tokens": 34645218.0,
+      "step": 10190
+    },
+    {
+      "epoch": 0.1023726363964832,
+      "learning_rate": 0.00017952748002729937,
+      "loss": 0.7851,
+      "mean_token_accuracy": 0.7663270890712738,
+      "num_tokens": 34679317.0,
+      "step": 10200
+    },
+    {
+      "epoch": 0.10247300172628367,
+      "learning_rate": 0.00017950740696133928,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.782097053527832,
+      "num_tokens": 34713589.0,
+      "step": 10210
+    },
+    {
+      "epoch": 0.10257336705608415,
+      "learning_rate": 0.00017948733389537918,
+      "loss": 0.7882,
+      "mean_token_accuracy": 0.7661328732967376,
+      "num_tokens": 34747582.0,
+      "step": 10220
+    },
+    {
+      "epoch": 0.10267373238588462,
+      "learning_rate": 0.0001794672608294191,
+      "loss": 0.7566,
+      "mean_token_accuracy": 0.7741984486579895,
+      "num_tokens": 34781547.0,
+      "step": 10230
+    },
+    {
+      "epoch": 0.10277409771568509,
+      "learning_rate": 0.000179447187763459,
+      "loss": 0.8143,
+      "mean_token_accuracy": 0.7658887326717376,
+      "num_tokens": 34815953.0,
+      "step": 10240
+    },
+    {
+      "epoch": 0.10287446304548557,
+      "learning_rate": 0.0001794271146974989,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7683747112751007,
+      "num_tokens": 34849623.0,
+      "step": 10250
+    },
+    {
+      "epoch": 0.10297482837528604,
+      "learning_rate": 0.00017940704163153882,
+      "loss": 0.7784,
+      "mean_token_accuracy": 0.7713934361934662,
+      "num_tokens": 34883376.0,
+      "step": 10260
+    },
+    {
+      "epoch": 0.10307519370508651,
+      "learning_rate": 0.0001793869685655787,
+      "loss": 0.7724,
+      "mean_token_accuracy": 0.7745378196239472,
+      "num_tokens": 34917673.0,
+      "step": 10270
+    },
+    {
+      "epoch": 0.10317555903488698,
+      "learning_rate": 0.00017936689549961864,
+      "loss": 0.7647,
+      "mean_token_accuracy": 0.7662047564983367,
+      "num_tokens": 34951154.0,
+      "step": 10280
+    },
+    {
+      "epoch": 0.10327592436468747,
+      "learning_rate": 0.00017934682243365852,
+      "loss": 0.8227,
+      "mean_token_accuracy": 0.7628067374229431,
+      "num_tokens": 34985037.0,
+      "step": 10290
+    },
+    {
+      "epoch": 0.10337628969448794,
+      "learning_rate": 0.00017932674936769842,
+      "loss": 0.7846,
+      "mean_token_accuracy": 0.7674853265285492,
+      "num_tokens": 35018599.0,
+      "step": 10300
+    },
+    {
+      "epoch": 0.1034766550242884,
+      "learning_rate": 0.00017930667630173833,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.7672294676303864,
+      "num_tokens": 35052701.0,
+      "step": 10310
+    },
+    {
+      "epoch": 0.10357702035408889,
+      "learning_rate": 0.00017928660323577824,
+      "loss": 0.7519,
+      "mean_token_accuracy": 0.7754966616630554,
+      "num_tokens": 35086618.0,
+      "step": 10320
+    },
+    {
+      "epoch": 0.10367738568388936,
+      "learning_rate": 0.00017926653016981812,
+      "loss": 0.7699,
+      "mean_token_accuracy": 0.771609491109848,
+      "num_tokens": 35120506.0,
+      "step": 10330
+    },
+    {
+      "epoch": 0.10377775101368983,
+      "learning_rate": 0.00017924645710385806,
+      "loss": 0.7826,
+      "mean_token_accuracy": 0.7697823762893676,
+      "num_tokens": 35154462.0,
+      "step": 10340
+    },
+    {
+      "epoch": 0.1038781163434903,
+      "learning_rate": 0.00017922638403789797,
+      "loss": 0.8021,
+      "mean_token_accuracy": 0.7638498783111572,
+      "num_tokens": 35188876.0,
+      "step": 10350
+    },
+    {
+      "epoch": 0.10397848167329078,
+      "learning_rate": 0.00017920631097193785,
+      "loss": 0.7639,
+      "mean_token_accuracy": 0.772318959236145,
+      "num_tokens": 35222612.0,
+      "step": 10360
+    },
+    {
+      "epoch": 0.10407884700309125,
+      "learning_rate": 0.00017918623790597778,
+      "loss": 0.7741,
+      "mean_token_accuracy": 0.7691188335418702,
+      "num_tokens": 35256951.0,
+      "step": 10370
+    },
+    {
+      "epoch": 0.10417921233289172,
+      "learning_rate": 0.00017916616484001767,
+      "loss": 0.7602,
+      "mean_token_accuracy": 0.7751318573951721,
+      "num_tokens": 35291125.0,
+      "step": 10380
+    },
+    {
+      "epoch": 0.1042795776626922,
+      "learning_rate": 0.00017914609177405757,
+      "loss": 0.8093,
+      "mean_token_accuracy": 0.7600766003131867,
+      "num_tokens": 35325526.0,
+      "step": 10390
+    },
+    {
+      "epoch": 0.10437994299249267,
+      "learning_rate": 0.00017912601870809748,
+      "loss": 0.7839,
+      "mean_token_accuracy": 0.770651638507843,
+      "num_tokens": 35359603.0,
+      "step": 10400
+    },
+    {
+      "epoch": 0.10448030832229314,
+      "learning_rate": 0.0001791059456421374,
+      "loss": 0.8134,
+      "mean_token_accuracy": 0.7629783511161804,
+      "num_tokens": 35394005.0,
+      "step": 10410
+    },
+    {
+      "epoch": 0.10458067365209361,
+      "learning_rate": 0.0001790858725761773,
+      "loss": 0.7882,
+      "mean_token_accuracy": 0.7687924087047577,
+      "num_tokens": 35427765.0,
+      "step": 10420
+    },
+    {
+      "epoch": 0.1046810389818941,
+      "learning_rate": 0.0001790657995102172,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7761599898338318,
+      "num_tokens": 35462276.0,
+      "step": 10430
+    },
+    {
+      "epoch": 0.10478140431169457,
+      "learning_rate": 0.0001790457264442571,
+      "loss": 0.7984,
+      "mean_token_accuracy": 0.7601453244686127,
+      "num_tokens": 35495851.0,
+      "step": 10440
+    },
+    {
+      "epoch": 0.10488176964149504,
+      "learning_rate": 0.00017902565337829702,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7701234340667724,
+      "num_tokens": 35529293.0,
+      "step": 10450
+    },
+    {
+      "epoch": 0.10498213497129552,
+      "learning_rate": 0.0001790055803123369,
+      "loss": 0.8224,
+      "mean_token_accuracy": 0.7610841393470764,
+      "num_tokens": 35562951.0,
+      "step": 10460
+    },
+    {
+      "epoch": 0.10508250030109599,
+      "learning_rate": 0.00017898550724637681,
+      "loss": 0.7961,
+      "mean_token_accuracy": 0.7697411417961121,
+      "num_tokens": 35596293.0,
+      "step": 10470
+    },
+    {
+      "epoch": 0.10518286563089646,
+      "learning_rate": 0.00017896543418041672,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7662987589836121,
+      "num_tokens": 35630647.0,
+      "step": 10480
+    },
+    {
+      "epoch": 0.10528323096069693,
+      "learning_rate": 0.00017894536111445663,
+      "loss": 0.7993,
+      "mean_token_accuracy": 0.7630761504173279,
+      "num_tokens": 35665123.0,
+      "step": 10490
+    },
+    {
+      "epoch": 0.10538359629049741,
+      "learning_rate": 0.00017892528804849654,
+      "loss": 0.7891,
+      "mean_token_accuracy": 0.7697969257831574,
+      "num_tokens": 35699146.0,
+      "step": 10500
+    },
+    {
+      "epoch": 0.10548396162029788,
+      "learning_rate": 0.00017890521498253645,
+      "loss": 0.7781,
+      "mean_token_accuracy": 0.7654927611351013,
+      "num_tokens": 35733136.0,
+      "step": 10510
+    },
+    {
+      "epoch": 0.10558432695009835,
+      "learning_rate": 0.00017888514191657636,
+      "loss": 0.795,
+      "mean_token_accuracy": 0.7694994390010834,
+      "num_tokens": 35767045.0,
+      "step": 10520
+    },
+    {
+      "epoch": 0.10568469227989884,
+      "learning_rate": 0.00017886506885061624,
+      "loss": 0.7664,
+      "mean_token_accuracy": 0.7748150587081909,
+      "num_tokens": 35799958.0,
+      "step": 10530
+    },
+    {
+      "epoch": 0.1057850576096993,
+      "learning_rate": 0.00017884499578465617,
+      "loss": 0.773,
+      "mean_token_accuracy": 0.7681718349456788,
+      "num_tokens": 35835426.0,
+      "step": 10540
+    },
+    {
+      "epoch": 0.10588542293949978,
+      "learning_rate": 0.00017882492271869605,
+      "loss": 0.7709,
+      "mean_token_accuracy": 0.7701553642749787,
+      "num_tokens": 35870072.0,
+      "step": 10550
+    },
+    {
+      "epoch": 0.10598578826930025,
+      "learning_rate": 0.00017880484965273596,
+      "loss": 0.7635,
+      "mean_token_accuracy": 0.7779048621654511,
+      "num_tokens": 35903911.0,
+      "step": 10560
+    },
+    {
+      "epoch": 0.10608615359910073,
+      "learning_rate": 0.00017878477658677587,
+      "loss": 0.7711,
+      "mean_token_accuracy": 0.7651203393936157,
+      "num_tokens": 35937883.0,
+      "step": 10570
+    },
+    {
+      "epoch": 0.1061865189289012,
+      "learning_rate": 0.00017876470352081578,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7720681846141815,
+      "num_tokens": 35971773.0,
+      "step": 10580
+    },
+    {
+      "epoch": 0.10628688425870167,
+      "learning_rate": 0.0001787446304548557,
+      "loss": 0.7987,
+      "mean_token_accuracy": 0.7652650296688079,
+      "num_tokens": 36005676.0,
+      "step": 10590
+    },
+    {
+      "epoch": 0.10638724958850215,
+      "learning_rate": 0.0001787245573888956,
+      "loss": 0.7918,
+      "mean_token_accuracy": 0.7614519000053406,
+      "num_tokens": 36040256.0,
+      "step": 10600
+    },
+    {
+      "epoch": 0.10648761491830262,
+      "learning_rate": 0.0001787044843229355,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.7717543840408325,
+      "num_tokens": 36074289.0,
+      "step": 10610
+    },
+    {
+      "epoch": 0.10658798024810309,
+      "learning_rate": 0.00017868441125697539,
+      "loss": 0.7797,
+      "mean_token_accuracy": 0.7678600788116455,
+      "num_tokens": 36108241.0,
+      "step": 10620
+    },
+    {
+      "epoch": 0.10668834557790356,
+      "learning_rate": 0.00017866433819101532,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7684706568717956,
+      "num_tokens": 36142590.0,
+      "step": 10630
+    },
+    {
+      "epoch": 0.10678871090770405,
+      "learning_rate": 0.0001786442651250552,
+      "loss": 0.7741,
+      "mean_token_accuracy": 0.7725037276744843,
+      "num_tokens": 36176355.0,
+      "step": 10640
+    },
+    {
+      "epoch": 0.10688907623750452,
+      "learning_rate": 0.0001786241920590951,
+      "loss": 0.7625,
+      "mean_token_accuracy": 0.7765066862106323,
+      "num_tokens": 36210760.0,
+      "step": 10650
+    },
+    {
+      "epoch": 0.10698944156730499,
+      "learning_rate": 0.00017860411899313502,
+      "loss": 0.8049,
+      "mean_token_accuracy": 0.7634962856769562,
+      "num_tokens": 36245787.0,
+      "step": 10660
+    },
+    {
+      "epoch": 0.10708980689710547,
+      "learning_rate": 0.00017858404592717493,
+      "loss": 0.7719,
+      "mean_token_accuracy": 0.7657606959342956,
+      "num_tokens": 36279324.0,
+      "step": 10670
+    },
+    {
+      "epoch": 0.10719017222690594,
+      "learning_rate": 0.0001785639728612148,
+      "loss": 0.7691,
+      "mean_token_accuracy": 0.7721742331981659,
+      "num_tokens": 36313393.0,
+      "step": 10680
+    },
+    {
+      "epoch": 0.10729053755670641,
+      "learning_rate": 0.00017854389979525475,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7692029237747192,
+      "num_tokens": 36347216.0,
+      "step": 10690
+    },
+    {
+      "epoch": 0.10739090288650688,
+      "learning_rate": 0.00017852382672929465,
+      "loss": 0.8148,
+      "mean_token_accuracy": 0.7642192721366883,
+      "num_tokens": 36381240.0,
+      "step": 10700
+    },
+    {
+      "epoch": 0.10749126821630736,
+      "learning_rate": 0.00017850375366333454,
+      "loss": 0.7572,
+      "mean_token_accuracy": 0.7711041688919067,
+      "num_tokens": 36415161.0,
+      "step": 10710
+    },
+    {
+      "epoch": 0.10759163354610783,
+      "learning_rate": 0.00017848368059737447,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.7689418256282806,
+      "num_tokens": 36450114.0,
+      "step": 10720
+    },
+    {
+      "epoch": 0.1076919988759083,
+      "learning_rate": 0.00017846360753141435,
+      "loss": 0.7732,
+      "mean_token_accuracy": 0.764790666103363,
+      "num_tokens": 36483827.0,
+      "step": 10730
+    },
+    {
+      "epoch": 0.10779236420570878,
+      "learning_rate": 0.00017844353446545426,
+      "loss": 0.8257,
+      "mean_token_accuracy": 0.7528440475463867,
+      "num_tokens": 36518401.0,
+      "step": 10740
+    },
+    {
+      "epoch": 0.10789272953550925,
+      "learning_rate": 0.00017842346139949417,
+      "loss": 0.799,
+      "mean_token_accuracy": 0.7689604461193085,
+      "num_tokens": 36552415.0,
+      "step": 10750
+    },
+    {
+      "epoch": 0.10799309486530972,
+      "learning_rate": 0.00017840338833353408,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.7767434477806091,
+      "num_tokens": 36585872.0,
+      "step": 10760
+    },
+    {
+      "epoch": 0.1080934601951102,
+      "learning_rate": 0.00017838331526757396,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.7629287362098693,
+      "num_tokens": 36620050.0,
+      "step": 10770
+    },
+    {
+      "epoch": 0.10819382552491068,
+      "learning_rate": 0.0001783632422016139,
+      "loss": 0.7585,
+      "mean_token_accuracy": 0.7698633432388305,
+      "num_tokens": 36653788.0,
+      "step": 10780
+    },
+    {
+      "epoch": 0.10829419085471115,
+      "learning_rate": 0.00017834316913565378,
+      "loss": 0.8034,
+      "mean_token_accuracy": 0.7642245471477509,
+      "num_tokens": 36687477.0,
+      "step": 10790
+    },
+    {
+      "epoch": 0.10839455618451162,
+      "learning_rate": 0.00017832309606969368,
+      "loss": 0.7852,
+      "mean_token_accuracy": 0.7684082210063934,
+      "num_tokens": 36721177.0,
+      "step": 10800
+    },
+    {
+      "epoch": 0.1084949215143121,
+      "learning_rate": 0.0001783030230037336,
+      "loss": 0.7901,
+      "mean_token_accuracy": 0.7634345769882203,
+      "num_tokens": 36754202.0,
+      "step": 10810
+    },
+    {
+      "epoch": 0.10859528684411257,
+      "learning_rate": 0.0001782829499377735,
+      "loss": 0.784,
+      "mean_token_accuracy": 0.7708740293979645,
+      "num_tokens": 36788811.0,
+      "step": 10820
+    },
+    {
+      "epoch": 0.10869565217391304,
+      "learning_rate": 0.0001782628768718134,
+      "loss": 0.782,
+      "mean_token_accuracy": 0.7696825981140136,
+      "num_tokens": 36822957.0,
+      "step": 10830
+    },
+    {
+      "epoch": 0.10879601750371352,
+      "learning_rate": 0.00017824280380585332,
+      "loss": 0.775,
+      "mean_token_accuracy": 0.7708654582500458,
+      "num_tokens": 36858058.0,
+      "step": 10840
+    },
+    {
+      "epoch": 0.108896382833514,
+      "learning_rate": 0.00017822273073989323,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.7714969336986541,
+      "num_tokens": 36891877.0,
+      "step": 10850
+    },
+    {
+      "epoch": 0.10899674816331446,
+      "learning_rate": 0.0001782026576739331,
+      "loss": 0.7899,
+      "mean_token_accuracy": 0.7672240257263183,
+      "num_tokens": 36926216.0,
+      "step": 10860
+    },
+    {
+      "epoch": 0.10909711349311493,
+      "learning_rate": 0.00017818258460797304,
+      "loss": 0.7896,
+      "mean_token_accuracy": 0.7663983047008515,
+      "num_tokens": 36960170.0,
+      "step": 10870
+    },
+    {
+      "epoch": 0.10919747882291542,
+      "learning_rate": 0.00017816251154201292,
+      "loss": 0.7492,
+      "mean_token_accuracy": 0.7793428659439087,
+      "num_tokens": 36994234.0,
+      "step": 10880
+    },
+    {
+      "epoch": 0.10929784415271589,
+      "learning_rate": 0.00017814243847605283,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7706003427505493,
+      "num_tokens": 37027842.0,
+      "step": 10890
+    },
+    {
+      "epoch": 0.10939820948251636,
+      "learning_rate": 0.00017812236541009274,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7689258813858032,
+      "num_tokens": 37062157.0,
+      "step": 10900
+    },
+    {
+      "epoch": 0.10949857481231684,
+      "learning_rate": 0.00017810229234413265,
+      "loss": 0.7415,
+      "mean_token_accuracy": 0.7800822496414185,
+      "num_tokens": 37096416.0,
+      "step": 10910
+    },
+    {
+      "epoch": 0.10959894014211731,
+      "learning_rate": 0.00017808221927817256,
+      "loss": 0.7996,
+      "mean_token_accuracy": 0.7646092355251313,
+      "num_tokens": 37129570.0,
+      "step": 10920
+    },
+    {
+      "epoch": 0.10969930547191778,
+      "learning_rate": 0.00017806214621221247,
+      "loss": 0.7887,
+      "mean_token_accuracy": 0.7637861371040344,
+      "num_tokens": 37163649.0,
+      "step": 10930
+    },
+    {
+      "epoch": 0.10979967080171825,
+      "learning_rate": 0.00017804207314625237,
+      "loss": 0.7577,
+      "mean_token_accuracy": 0.7768438696861267,
+      "num_tokens": 37197267.0,
+      "step": 10940
+    },
+    {
+      "epoch": 0.10990003613151873,
+      "learning_rate": 0.00017802200008029226,
+      "loss": 0.8056,
+      "mean_token_accuracy": 0.7628251135349273,
+      "num_tokens": 37231308.0,
+      "step": 10950
+    },
+    {
+      "epoch": 0.1100004014613192,
+      "learning_rate": 0.0001780019270143322,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.7645903825759888,
+      "num_tokens": 37264848.0,
+      "step": 10960
+    },
+    {
+      "epoch": 0.11010076679111967,
+      "learning_rate": 0.00017798185394837207,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7644063472747803,
+      "num_tokens": 37298911.0,
+      "step": 10970
+    },
+    {
+      "epoch": 0.11020113212092016,
+      "learning_rate": 0.000177961780882412,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.7657126247882843,
+      "num_tokens": 37332105.0,
+      "step": 10980
+    },
+    {
+      "epoch": 0.11030149745072063,
+      "learning_rate": 0.0001779417078164519,
+      "loss": 0.7851,
+      "mean_token_accuracy": 0.7642782807350159,
+      "num_tokens": 37366442.0,
+      "step": 10990
+    },
+    {
+      "epoch": 0.1104018627805211,
+      "learning_rate": 0.0001779216347504918,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.7642432451248169,
+      "num_tokens": 37400575.0,
+      "step": 11000
+    },
+    {
+      "epoch": 0.11050222811032157,
+      "learning_rate": 0.0001779015616845317,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7645276188850403,
+      "num_tokens": 37434658.0,
+      "step": 11010
+    },
+    {
+      "epoch": 0.11060259344012205,
+      "learning_rate": 0.00017788148861857162,
+      "loss": 0.8082,
+      "mean_token_accuracy": 0.7625736057758331,
+      "num_tokens": 37469762.0,
+      "step": 11020
+    },
+    {
+      "epoch": 0.11070295876992252,
+      "learning_rate": 0.0001778614155526115,
+      "loss": 0.7881,
+      "mean_token_accuracy": 0.7737322926521302,
+      "num_tokens": 37504309.0,
+      "step": 11030
+    },
+    {
+      "epoch": 0.11080332409972299,
+      "learning_rate": 0.00017784134248665143,
+      "loss": 0.7881,
+      "mean_token_accuracy": 0.7622285485267639,
+      "num_tokens": 37537634.0,
+      "step": 11040
+    },
+    {
+      "epoch": 0.11090368942952347,
+      "learning_rate": 0.00017782126942069134,
+      "loss": 0.7971,
+      "mean_token_accuracy": 0.7630579233169555,
+      "num_tokens": 37571291.0,
+      "step": 11050
+    },
+    {
+      "epoch": 0.11100405475932394,
+      "learning_rate": 0.00017780119635473122,
+      "loss": 0.7835,
+      "mean_token_accuracy": 0.7660659909248352,
+      "num_tokens": 37605243.0,
+      "step": 11060
+    },
+    {
+      "epoch": 0.11110442008912441,
+      "learning_rate": 0.00017778112328877116,
+      "loss": 0.7883,
+      "mean_token_accuracy": 0.7664074063301086,
+      "num_tokens": 37638583.0,
+      "step": 11070
+    },
+    {
+      "epoch": 0.11120478541892488,
+      "learning_rate": 0.00017776105022281104,
+      "loss": 0.8163,
+      "mean_token_accuracy": 0.7627486288547516,
+      "num_tokens": 37672756.0,
+      "step": 11080
+    },
+    {
+      "epoch": 0.11130515074872536,
+      "learning_rate": 0.00017774097715685095,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7649985313415527,
+      "num_tokens": 37706226.0,
+      "step": 11090
+    },
+    {
+      "epoch": 0.11140551607852583,
+      "learning_rate": 0.00017772090409089086,
+      "loss": 0.7781,
+      "mean_token_accuracy": 0.7736667335033417,
+      "num_tokens": 37739816.0,
+      "step": 11100
+    },
+    {
+      "epoch": 0.1115058814083263,
+      "learning_rate": 0.00017770083102493076,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7703223526477814,
+      "num_tokens": 37774363.0,
+      "step": 11110
+    },
+    {
+      "epoch": 0.11160624673812679,
+      "learning_rate": 0.00017768075795897065,
+      "loss": 0.7972,
+      "mean_token_accuracy": 0.7607849955558776,
+      "num_tokens": 37809032.0,
+      "step": 11120
+    },
+    {
+      "epoch": 0.11170661206792726,
+      "learning_rate": 0.00017766068489301058,
+      "loss": 0.7755,
+      "mean_token_accuracy": 0.773020726442337,
+      "num_tokens": 37843114.0,
+      "step": 11130
+    },
+    {
+      "epoch": 0.11180697739772773,
+      "learning_rate": 0.00017764061182705046,
+      "loss": 0.8009,
+      "mean_token_accuracy": 0.7607036411762238,
+      "num_tokens": 37876581.0,
+      "step": 11140
+    },
+    {
+      "epoch": 0.1119073427275282,
+      "learning_rate": 0.00017762053876109037,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.7713505566120148,
+      "num_tokens": 37911079.0,
+      "step": 11150
+    },
+    {
+      "epoch": 0.11200770805732868,
+      "learning_rate": 0.00017760046569513028,
+      "loss": 0.8049,
+      "mean_token_accuracy": 0.7616325080394745,
+      "num_tokens": 37945303.0,
+      "step": 11160
+    },
+    {
+      "epoch": 0.11210807338712915,
+      "learning_rate": 0.0001775803926291702,
+      "loss": 0.8093,
+      "mean_token_accuracy": 0.7597049415111542,
+      "num_tokens": 37978943.0,
+      "step": 11170
+    },
+    {
+      "epoch": 0.11220843871692962,
+      "learning_rate": 0.0001775603195632101,
+      "loss": 0.7977,
+      "mean_token_accuracy": 0.7681370735168457,
+      "num_tokens": 38013071.0,
+      "step": 11180
+    },
+    {
+      "epoch": 0.1123088040467301,
+      "learning_rate": 0.00017754024649725,
+      "loss": 0.8246,
+      "mean_token_accuracy": 0.7641762137413025,
+      "num_tokens": 38046353.0,
+      "step": 11190
+    },
+    {
+      "epoch": 0.11240916937653057,
+      "learning_rate": 0.0001775201734312899,
+      "loss": 0.8302,
+      "mean_token_accuracy": 0.7551983535289765,
+      "num_tokens": 38079892.0,
+      "step": 11200
+    },
+    {
+      "epoch": 0.11250953470633104,
+      "learning_rate": 0.0001775001003653298,
+      "loss": 0.7994,
+      "mean_token_accuracy": 0.7642681956291199,
+      "num_tokens": 38113690.0,
+      "step": 11210
+    },
+    {
+      "epoch": 0.11260990003613151,
+      "learning_rate": 0.00017748002729936973,
+      "loss": 0.7823,
+      "mean_token_accuracy": 0.7686408400535584,
+      "num_tokens": 38147364.0,
+      "step": 11220
+    },
+    {
+      "epoch": 0.112710265365932,
+      "learning_rate": 0.0001774599542334096,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7673284769058227,
+      "num_tokens": 38181641.0,
+      "step": 11230
+    },
+    {
+      "epoch": 0.11281063069573247,
+      "learning_rate": 0.00017743988116744952,
+      "loss": 0.7671,
+      "mean_token_accuracy": 0.7764685571193695,
+      "num_tokens": 38215665.0,
+      "step": 11240
+    },
+    {
+      "epoch": 0.11291099602553294,
+      "learning_rate": 0.00017741980810148943,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7562175869941712,
+      "num_tokens": 38250324.0,
+      "step": 11250
+    },
+    {
+      "epoch": 0.11301136135533342,
+      "learning_rate": 0.00017739973503552934,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.7702659010887146,
+      "num_tokens": 38284395.0,
+      "step": 11260
+    },
+    {
+      "epoch": 0.11311172668513389,
+      "learning_rate": 0.00017737966196956924,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.7622061729431152,
+      "num_tokens": 38318500.0,
+      "step": 11270
+    },
+    {
+      "epoch": 0.11321209201493436,
+      "learning_rate": 0.00017735958890360915,
+      "loss": 0.7934,
+      "mean_token_accuracy": 0.7680340647697449,
+      "num_tokens": 38352413.0,
+      "step": 11280
+    },
+    {
+      "epoch": 0.11331245734473483,
+      "learning_rate": 0.00017733951583764906,
+      "loss": 0.7642,
+      "mean_token_accuracy": 0.7741311669349671,
+      "num_tokens": 38385939.0,
+      "step": 11290
+    },
+    {
+      "epoch": 0.11341282267453531,
+      "learning_rate": 0.00017731944277168894,
+      "loss": 0.7932,
+      "mean_token_accuracy": 0.7647429883480072,
+      "num_tokens": 38420540.0,
+      "step": 11300
+    },
+    {
+      "epoch": 0.11351318800433578,
+      "learning_rate": 0.00017729936970572888,
+      "loss": 0.8164,
+      "mean_token_accuracy": 0.7598032474517822,
+      "num_tokens": 38454312.0,
+      "step": 11310
+    },
+    {
+      "epoch": 0.11361355333413625,
+      "learning_rate": 0.00017727929663976876,
+      "loss": 0.7763,
+      "mean_token_accuracy": 0.7746975064277649,
+      "num_tokens": 38488831.0,
+      "step": 11320
+    },
+    {
+      "epoch": 0.11371391866393674,
+      "learning_rate": 0.00017725922357380867,
+      "loss": 0.7958,
+      "mean_token_accuracy": 0.7677113354206085,
+      "num_tokens": 38521896.0,
+      "step": 11330
+    },
+    {
+      "epoch": 0.1138142839937372,
+      "learning_rate": 0.00017723915050784858,
+      "loss": 0.7477,
+      "mean_token_accuracy": 0.7796377956867218,
+      "num_tokens": 38556722.0,
+      "step": 11340
+    },
+    {
+      "epoch": 0.11391464932353768,
+      "learning_rate": 0.00017721907744188849,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7726167142391205,
+      "num_tokens": 38591405.0,
+      "step": 11350
+    },
+    {
+      "epoch": 0.11401501465333815,
+      "learning_rate": 0.00017719900437592837,
+      "loss": 0.7673,
+      "mean_token_accuracy": 0.7699409604072571,
+      "num_tokens": 38625518.0,
+      "step": 11360
+    },
+    {
+      "epoch": 0.11411537998313863,
+      "learning_rate": 0.0001771789313099683,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7713293433189392,
+      "num_tokens": 38659584.0,
+      "step": 11370
+    },
+    {
+      "epoch": 0.1142157453129391,
+      "learning_rate": 0.00017715885824400818,
+      "loss": 0.7973,
+      "mean_token_accuracy": 0.7608528554439544,
+      "num_tokens": 38693605.0,
+      "step": 11380
+    },
+    {
+      "epoch": 0.11431611064273957,
+      "learning_rate": 0.0001771387851780481,
+      "loss": 0.8145,
+      "mean_token_accuracy": 0.766344141960144,
+      "num_tokens": 38728476.0,
+      "step": 11390
+    },
+    {
+      "epoch": 0.11441647597254005,
+      "learning_rate": 0.00017711871211208803,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7720429956912994,
+      "num_tokens": 38762922.0,
+      "step": 11400
+    },
+    {
+      "epoch": 0.11451684130234052,
+      "learning_rate": 0.0001770986390461279,
+      "loss": 0.7659,
+      "mean_token_accuracy": 0.7721791803836823,
+      "num_tokens": 38796838.0,
+      "step": 11410
+    },
+    {
+      "epoch": 0.11461720663214099,
+      "learning_rate": 0.00017707856598016782,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7671987414360046,
+      "num_tokens": 38830903.0,
+      "step": 11420
+    },
+    {
+      "epoch": 0.11471757196194146,
+      "learning_rate": 0.00017705849291420773,
+      "loss": 0.795,
+      "mean_token_accuracy": 0.7672559738159179,
+      "num_tokens": 38865312.0,
+      "step": 11430
+    },
+    {
+      "epoch": 0.11481793729174194,
+      "learning_rate": 0.00017703841984824763,
+      "loss": 0.7741,
+      "mean_token_accuracy": 0.7725123465061188,
+      "num_tokens": 38900004.0,
+      "step": 11440
+    },
+    {
+      "epoch": 0.11491830262154241,
+      "learning_rate": 0.00017701834678228752,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7691251873970032,
+      "num_tokens": 38934537.0,
+      "step": 11450
+    },
+    {
+      "epoch": 0.11501866795134288,
+      "learning_rate": 0.00017699827371632745,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.7705145180225372,
+      "num_tokens": 38967988.0,
+      "step": 11460
+    },
+    {
+      "epoch": 0.11511903328114337,
+      "learning_rate": 0.00017697820065036733,
+      "loss": 0.756,
+      "mean_token_accuracy": 0.7800585031509399,
+      "num_tokens": 39001441.0,
+      "step": 11470
+    },
+    {
+      "epoch": 0.11521939861094384,
+      "learning_rate": 0.00017695812758440724,
+      "loss": 0.768,
+      "mean_token_accuracy": 0.7769594490528107,
+      "num_tokens": 39034929.0,
+      "step": 11480
+    },
+    {
+      "epoch": 0.11531976394074431,
+      "learning_rate": 0.00017693805451844715,
+      "loss": 0.7913,
+      "mean_token_accuracy": 0.7646923243999482,
+      "num_tokens": 39068092.0,
+      "step": 11490
+    },
+    {
+      "epoch": 0.11542012927054478,
+      "learning_rate": 0.00017691798145248706,
+      "loss": 0.7768,
+      "mean_token_accuracy": 0.7709584832191467,
+      "num_tokens": 39101540.0,
+      "step": 11500
+    },
+    {
+      "epoch": 0.11552049460034526,
+      "learning_rate": 0.00017689790838652697,
+      "loss": 0.8017,
+      "mean_token_accuracy": 0.7663677096366882,
+      "num_tokens": 39136047.0,
+      "step": 11510
+    },
+    {
+      "epoch": 0.11562085993014573,
+      "learning_rate": 0.00017687783532056687,
+      "loss": 0.7653,
+      "mean_token_accuracy": 0.7705176889896392,
+      "num_tokens": 39169501.0,
+      "step": 11520
+    },
+    {
+      "epoch": 0.1157212252599462,
+      "learning_rate": 0.00017685776225460678,
+      "loss": 0.8023,
+      "mean_token_accuracy": 0.7634245574474334,
+      "num_tokens": 39203477.0,
+      "step": 11530
+    },
+    {
+      "epoch": 0.11582159058974668,
+      "learning_rate": 0.00017683768918864666,
+      "loss": 0.7639,
+      "mean_token_accuracy": 0.7677269458770752,
+      "num_tokens": 39238060.0,
+      "step": 11540
+    },
+    {
+      "epoch": 0.11592195591954715,
+      "learning_rate": 0.0001768176161226866,
+      "loss": 0.7859,
+      "mean_token_accuracy": 0.7660326838493348,
+      "num_tokens": 39271841.0,
+      "step": 11550
+    },
+    {
+      "epoch": 0.11602232124934762,
+      "learning_rate": 0.00017679754305672648,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7717775106430054,
+      "num_tokens": 39306606.0,
+      "step": 11560
+    },
+    {
+      "epoch": 0.1161226865791481,
+      "learning_rate": 0.00017677746999076642,
+      "loss": 0.7661,
+      "mean_token_accuracy": 0.7701225757598877,
+      "num_tokens": 39339475.0,
+      "step": 11570
+    },
+    {
+      "epoch": 0.11622305190894858,
+      "learning_rate": 0.0001767573969248063,
+      "loss": 0.7757,
+      "mean_token_accuracy": 0.7728428721427918,
+      "num_tokens": 39374240.0,
+      "step": 11580
+    },
+    {
+      "epoch": 0.11632341723874905,
+      "learning_rate": 0.0001767373238588462,
+      "loss": 0.7842,
+      "mean_token_accuracy": 0.7713092923164367,
+      "num_tokens": 39408018.0,
+      "step": 11590
+    },
+    {
+      "epoch": 0.11642378256854952,
+      "learning_rate": 0.00017671725079288611,
+      "loss": 0.7551,
+      "mean_token_accuracy": 0.7804297387599946,
+      "num_tokens": 39442330.0,
+      "step": 11600
+    },
+    {
+      "epoch": 0.11652414789835,
+      "learning_rate": 0.00017669717772692602,
+      "loss": 0.7865,
+      "mean_token_accuracy": 0.7714122951030731,
+      "num_tokens": 39475806.0,
+      "step": 11610
+    },
+    {
+      "epoch": 0.11662451322815047,
+      "learning_rate": 0.00017667710466096593,
+      "loss": 0.793,
+      "mean_token_accuracy": 0.766773521900177,
+      "num_tokens": 39510580.0,
+      "step": 11620
+    },
+    {
+      "epoch": 0.11672487855795094,
+      "learning_rate": 0.00017665703159500584,
+      "loss": 0.747,
+      "mean_token_accuracy": 0.7717756271362305,
+      "num_tokens": 39544057.0,
+      "step": 11630
+    },
+    {
+      "epoch": 0.11682524388775141,
+      "learning_rate": 0.00017663695852904575,
+      "loss": 0.7983,
+      "mean_token_accuracy": 0.7713942110538483,
+      "num_tokens": 39578119.0,
+      "step": 11640
+    },
+    {
+      "epoch": 0.11692560921755189,
+      "learning_rate": 0.00017661688546308563,
+      "loss": 0.7699,
+      "mean_token_accuracy": 0.7693363547325134,
+      "num_tokens": 39612377.0,
+      "step": 11650
+    },
+    {
+      "epoch": 0.11702597454735236,
+      "learning_rate": 0.00017659681239712557,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7707055330276489,
+      "num_tokens": 39646664.0,
+      "step": 11660
+    },
+    {
+      "epoch": 0.11712633987715283,
+      "learning_rate": 0.00017657673933116545,
+      "loss": 0.7642,
+      "mean_token_accuracy": 0.7722592294216156,
+      "num_tokens": 39680694.0,
+      "step": 11670
+    },
+    {
+      "epoch": 0.11722670520695332,
+      "learning_rate": 0.00017655666626520536,
+      "loss": 0.7999,
+      "mean_token_accuracy": 0.7656277120113373,
+      "num_tokens": 39715597.0,
+      "step": 11680
+    },
+    {
+      "epoch": 0.11732707053675379,
+      "learning_rate": 0.00017653659319924526,
+      "loss": 0.7726,
+      "mean_token_accuracy": 0.7747666537761688,
+      "num_tokens": 39750466.0,
+      "step": 11690
+    },
+    {
+      "epoch": 0.11742743586655426,
+      "learning_rate": 0.00017651652013328517,
+      "loss": 0.8171,
+      "mean_token_accuracy": 0.76343132853508,
+      "num_tokens": 39784127.0,
+      "step": 11700
+    },
+    {
+      "epoch": 0.11752780119635473,
+      "learning_rate": 0.00017649644706732505,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.7643001914024353,
+      "num_tokens": 39818279.0,
+      "step": 11710
+    },
+    {
+      "epoch": 0.11762816652615521,
+      "learning_rate": 0.000176476374001365,
+      "loss": 0.8281,
+      "mean_token_accuracy": 0.764520788192749,
+      "num_tokens": 39852065.0,
+      "step": 11720
+    },
+    {
+      "epoch": 0.11772853185595568,
+      "learning_rate": 0.00017645630093540487,
+      "loss": 0.7999,
+      "mean_token_accuracy": 0.7661543011665344,
+      "num_tokens": 39886218.0,
+      "step": 11730
+    },
+    {
+      "epoch": 0.11782889718575615,
+      "learning_rate": 0.00017643622786944478,
+      "loss": 0.7623,
+      "mean_token_accuracy": 0.7742419481277466,
+      "num_tokens": 39919453.0,
+      "step": 11740
+    },
+    {
+      "epoch": 0.11792926251555663,
+      "learning_rate": 0.00017641615480348471,
+      "loss": 0.7836,
+      "mean_token_accuracy": 0.7672594904899597,
+      "num_tokens": 39954125.0,
+      "step": 11750
+    },
+    {
+      "epoch": 0.1180296278453571,
+      "learning_rate": 0.0001763960817375246,
+      "loss": 0.7726,
+      "mean_token_accuracy": 0.7788445174694061,
+      "num_tokens": 39988348.0,
+      "step": 11760
+    },
+    {
+      "epoch": 0.11812999317515757,
+      "learning_rate": 0.0001763760086715645,
+      "loss": 0.8272,
+      "mean_token_accuracy": 0.7619171619415284,
+      "num_tokens": 40022296.0,
+      "step": 11770
+    },
+    {
+      "epoch": 0.11823035850495804,
+      "learning_rate": 0.0001763559356056044,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.7701909244060516,
+      "num_tokens": 40055180.0,
+      "step": 11780
+    },
+    {
+      "epoch": 0.11833072383475852,
+      "learning_rate": 0.00017633586253964432,
+      "loss": 0.7398,
+      "mean_token_accuracy": 0.7765741109848022,
+      "num_tokens": 40089322.0,
+      "step": 11790
+    },
+    {
+      "epoch": 0.118431089164559,
+      "learning_rate": 0.0001763157894736842,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.776421868801117,
+      "num_tokens": 40123226.0,
+      "step": 11800
+    },
+    {
+      "epoch": 0.11853145449435946,
+      "learning_rate": 0.00017629571640772414,
+      "loss": 0.8092,
+      "mean_token_accuracy": 0.7604811728000641,
+      "num_tokens": 40156888.0,
+      "step": 11810
+    },
+    {
+      "epoch": 0.11863181982415995,
+      "learning_rate": 0.00017627564334176402,
+      "loss": 0.7697,
+      "mean_token_accuracy": 0.7683253526687622,
+      "num_tokens": 40190380.0,
+      "step": 11820
+    },
+    {
+      "epoch": 0.11873218515396042,
+      "learning_rate": 0.00017625557027580393,
+      "loss": 0.7996,
+      "mean_token_accuracy": 0.7628776550292968,
+      "num_tokens": 40224206.0,
+      "step": 11830
+    },
+    {
+      "epoch": 0.11883255048376089,
+      "learning_rate": 0.00017623549720984384,
+      "loss": 0.7741,
+      "mean_token_accuracy": 0.7673757433891296,
+      "num_tokens": 40257463.0,
+      "step": 11840
+    },
+    {
+      "epoch": 0.11893291581356136,
+      "learning_rate": 0.00017621542414388374,
+      "loss": 0.8199,
+      "mean_token_accuracy": 0.7614277958869934,
+      "num_tokens": 40290925.0,
+      "step": 11850
+    },
+    {
+      "epoch": 0.11903328114336184,
+      "learning_rate": 0.00017619535107792365,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.7650829792022705,
+      "num_tokens": 40324936.0,
+      "step": 11860
+    },
+    {
+      "epoch": 0.11913364647316231,
+      "learning_rate": 0.00017617527801196356,
+      "loss": 0.8039,
+      "mean_token_accuracy": 0.7632757902145386,
+      "num_tokens": 40358512.0,
+      "step": 11870
+    },
+    {
+      "epoch": 0.11923401180296278,
+      "learning_rate": 0.00017615520494600347,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7703190386295319,
+      "num_tokens": 40391900.0,
+      "step": 11880
+    },
+    {
+      "epoch": 0.11933437713276326,
+      "learning_rate": 0.00017613513188004335,
+      "loss": 0.7905,
+      "mean_token_accuracy": 0.762644600868225,
+      "num_tokens": 40425909.0,
+      "step": 11890
+    },
+    {
+      "epoch": 0.11943474246256373,
+      "learning_rate": 0.0001761150588140833,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.7715180158615113,
+      "num_tokens": 40460001.0,
+      "step": 11900
+    },
+    {
+      "epoch": 0.1195351077923642,
+      "learning_rate": 0.00017609498574812317,
+      "loss": 0.764,
+      "mean_token_accuracy": 0.7699169874191284,
+      "num_tokens": 40493921.0,
+      "step": 11910
+    },
+    {
+      "epoch": 0.11963547312216467,
+      "learning_rate": 0.00017607491268216308,
+      "loss": 0.7863,
+      "mean_token_accuracy": 0.7684384703636169,
+      "num_tokens": 40527849.0,
+      "step": 11920
+    },
+    {
+      "epoch": 0.11973583845196516,
+      "learning_rate": 0.00017605483961620298,
+      "loss": 0.7991,
+      "mean_token_accuracy": 0.7640502333641053,
+      "num_tokens": 40562200.0,
+      "step": 11930
+    },
+    {
+      "epoch": 0.11983620378176563,
+      "learning_rate": 0.0001760347665502429,
+      "loss": 0.802,
+      "mean_token_accuracy": 0.7624570906162262,
+      "num_tokens": 40595828.0,
+      "step": 11940
+    },
+    {
+      "epoch": 0.1199365691115661,
+      "learning_rate": 0.0001760146934842828,
+      "loss": 0.7966,
+      "mean_token_accuracy": 0.7684255540370941,
+      "num_tokens": 40629479.0,
+      "step": 11950
+    },
+    {
+      "epoch": 0.12003693444136658,
+      "learning_rate": 0.0001759946204183227,
+      "loss": 0.748,
+      "mean_token_accuracy": 0.7810657501220704,
+      "num_tokens": 40663397.0,
+      "step": 11960
+    },
+    {
+      "epoch": 0.12013729977116705,
+      "learning_rate": 0.00017597454735236262,
+      "loss": 0.7789,
+      "mean_token_accuracy": 0.7606176555156707,
+      "num_tokens": 40697411.0,
+      "step": 11970
+    },
+    {
+      "epoch": 0.12023766510096752,
+      "learning_rate": 0.0001759544742864025,
+      "loss": 0.7962,
+      "mean_token_accuracy": 0.7685517072677612,
+      "num_tokens": 40731045.0,
+      "step": 11980
+    },
+    {
+      "epoch": 0.12033803043076799,
+      "learning_rate": 0.00017593440122044244,
+      "loss": 0.7857,
+      "mean_token_accuracy": 0.7689754724502563,
+      "num_tokens": 40764613.0,
+      "step": 11990
+    },
+    {
+      "epoch": 0.12043839576056847,
+      "learning_rate": 0.00017591432815448232,
+      "loss": 0.7416,
+      "mean_token_accuracy": 0.7737857520580291,
+      "num_tokens": 40798558.0,
+      "step": 12000
+    },
+    {
+      "epoch": 0.12053876109036894,
+      "learning_rate": 0.00017589425508852223,
+      "loss": 0.8079,
+      "mean_token_accuracy": 0.7625069737434387,
+      "num_tokens": 40832288.0,
+      "step": 12010
+    },
+    {
+      "epoch": 0.12063912642016941,
+      "learning_rate": 0.00017587418202256213,
+      "loss": 0.8001,
+      "mean_token_accuracy": 0.7698694944381714,
+      "num_tokens": 40866218.0,
+      "step": 12020
+    },
+    {
+      "epoch": 0.1207394917499699,
+      "learning_rate": 0.00017585410895660204,
+      "loss": 0.7995,
+      "mean_token_accuracy": 0.7681800425052643,
+      "num_tokens": 40900843.0,
+      "step": 12030
+    },
+    {
+      "epoch": 0.12083985707977037,
+      "learning_rate": 0.00017583403589064192,
+      "loss": 0.8066,
+      "mean_token_accuracy": 0.7637708008289337,
+      "num_tokens": 40934399.0,
+      "step": 12040
+    },
+    {
+      "epoch": 0.12094022240957084,
+      "learning_rate": 0.00017581396282468186,
+      "loss": 0.7766,
+      "mean_token_accuracy": 0.7660633981227875,
+      "num_tokens": 40968768.0,
+      "step": 12050
+    },
+    {
+      "epoch": 0.1210405877393713,
+      "learning_rate": 0.00017579388975872174,
+      "loss": 0.7732,
+      "mean_token_accuracy": 0.7718873798847199,
+      "num_tokens": 41002363.0,
+      "step": 12060
+    },
+    {
+      "epoch": 0.12114095306917179,
+      "learning_rate": 0.00017577381669276165,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.775162810087204,
+      "num_tokens": 41037020.0,
+      "step": 12070
+    },
+    {
+      "epoch": 0.12124131839897226,
+      "learning_rate": 0.00017575374362680156,
+      "loss": 0.831,
+      "mean_token_accuracy": 0.7604571998119354,
+      "num_tokens": 41071366.0,
+      "step": 12080
+    },
+    {
+      "epoch": 0.12134168372877273,
+      "learning_rate": 0.00017573367056084147,
+      "loss": 0.8195,
+      "mean_token_accuracy": 0.7606946229934692,
+      "num_tokens": 41105049.0,
+      "step": 12090
+    },
+    {
+      "epoch": 0.12144204905857321,
+      "learning_rate": 0.00017571359749488137,
+      "loss": 0.7906,
+      "mean_token_accuracy": 0.7686561584472656,
+      "num_tokens": 41139040.0,
+      "step": 12100
+    },
+    {
+      "epoch": 0.12154241438837368,
+      "learning_rate": 0.00017569352442892128,
+      "loss": 0.794,
+      "mean_token_accuracy": 0.7679334044456482,
+      "num_tokens": 41172738.0,
+      "step": 12110
+    },
+    {
+      "epoch": 0.12164277971817415,
+      "learning_rate": 0.0001756734513629612,
+      "loss": 0.7825,
+      "mean_token_accuracy": 0.7688836455345154,
+      "num_tokens": 41207544.0,
+      "step": 12120
+    },
+    {
+      "epoch": 0.12174314504797462,
+      "learning_rate": 0.0001756533782970011,
+      "loss": 0.79,
+      "mean_token_accuracy": 0.7671124160289764,
+      "num_tokens": 41241678.0,
+      "step": 12130
+    },
+    {
+      "epoch": 0.1218435103777751,
+      "learning_rate": 0.000175633305231041,
+      "loss": 0.786,
+      "mean_token_accuracy": 0.7698677480220795,
+      "num_tokens": 41276109.0,
+      "step": 12140
+    },
+    {
+      "epoch": 0.12194387570757557,
+      "learning_rate": 0.0001756132321650809,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7711529910564423,
+      "num_tokens": 41309967.0,
+      "step": 12150
+    },
+    {
+      "epoch": 0.12204424103737604,
+      "learning_rate": 0.00017559315909912082,
+      "loss": 0.7946,
+      "mean_token_accuracy": 0.7652078151702881,
+      "num_tokens": 41344595.0,
+      "step": 12160
+    },
+    {
+      "epoch": 0.12214460636717653,
+      "learning_rate": 0.0001755730860331607,
+      "loss": 0.7935,
+      "mean_token_accuracy": 0.7659879446029663,
+      "num_tokens": 41378590.0,
+      "step": 12170
+    },
+    {
+      "epoch": 0.122244971696977,
+      "learning_rate": 0.00017555301296720061,
+      "loss": 0.7731,
+      "mean_token_accuracy": 0.770485919713974,
+      "num_tokens": 41412020.0,
+      "step": 12180
+    },
+    {
+      "epoch": 0.12234533702677747,
+      "learning_rate": 0.00017553293990124052,
+      "loss": 0.7958,
+      "mean_token_accuracy": 0.7665983974933624,
+      "num_tokens": 41446265.0,
+      "step": 12190
+    },
+    {
+      "epoch": 0.12244570235657794,
+      "learning_rate": 0.00017551286683528043,
+      "loss": 0.7606,
+      "mean_token_accuracy": 0.7740192413330078,
+      "num_tokens": 41479493.0,
+      "step": 12200
+    },
+    {
+      "epoch": 0.12254606768637842,
+      "learning_rate": 0.00017549279376932034,
+      "loss": 0.7853,
+      "mean_token_accuracy": 0.7652707397937775,
+      "num_tokens": 41513536.0,
+      "step": 12210
+    },
+    {
+      "epoch": 0.12264643301617889,
+      "learning_rate": 0.00017547272070336025,
+      "loss": 0.7618,
+      "mean_token_accuracy": 0.7741611421108245,
+      "num_tokens": 41547227.0,
+      "step": 12220
+    },
+    {
+      "epoch": 0.12274679834597936,
+      "learning_rate": 0.00017545264763740016,
+      "loss": 0.768,
+      "mean_token_accuracy": 0.7714941918849945,
+      "num_tokens": 41581441.0,
+      "step": 12230
+    },
+    {
+      "epoch": 0.12284716367577984,
+      "learning_rate": 0.00017543257457144004,
+      "loss": 0.7809,
+      "mean_token_accuracy": 0.7698556900024414,
+      "num_tokens": 41615059.0,
+      "step": 12240
+    },
+    {
+      "epoch": 0.12294752900558031,
+      "learning_rate": 0.00017541250150547997,
+      "loss": 0.7832,
+      "mean_token_accuracy": 0.7657558023929596,
+      "num_tokens": 41649299.0,
+      "step": 12250
+    },
+    {
+      "epoch": 0.12304789433538078,
+      "learning_rate": 0.00017539242843951985,
+      "loss": 0.7607,
+      "mean_token_accuracy": 0.7744082927703857,
+      "num_tokens": 41682483.0,
+      "step": 12260
+    },
+    {
+      "epoch": 0.12314825966518125,
+      "learning_rate": 0.00017537235537355976,
+      "loss": 0.808,
+      "mean_token_accuracy": 0.7617681801319123,
+      "num_tokens": 41716531.0,
+      "step": 12270
+    },
+    {
+      "epoch": 0.12324862499498174,
+      "learning_rate": 0.00017535228230759967,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.7723277449607849,
+      "num_tokens": 41750691.0,
+      "step": 12280
+    },
+    {
+      "epoch": 0.1233489903247822,
+      "learning_rate": 0.00017533220924163958,
+      "loss": 0.7608,
+      "mean_token_accuracy": 0.7751142323017121,
+      "num_tokens": 41785619.0,
+      "step": 12290
+    },
+    {
+      "epoch": 0.12344935565458268,
+      "learning_rate": 0.0001753121361756795,
+      "loss": 0.7678,
+      "mean_token_accuracy": 0.7734646201133728,
+      "num_tokens": 41818417.0,
+      "step": 12300
+    },
+    {
+      "epoch": 0.12354972098438316,
+      "learning_rate": 0.0001752920631097194,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.7770093083381653,
+      "num_tokens": 41852925.0,
+      "step": 12310
+    },
+    {
+      "epoch": 0.12365008631418363,
+      "learning_rate": 0.0001752719900437593,
+      "loss": 0.7708,
+      "mean_token_accuracy": 0.7711196780204773,
+      "num_tokens": 41887457.0,
+      "step": 12320
+    },
+    {
+      "epoch": 0.1237504516439841,
+      "learning_rate": 0.0001752519169777992,
+      "loss": 0.7318,
+      "mean_token_accuracy": 0.7761717677116394,
+      "num_tokens": 41921253.0,
+      "step": 12330
+    },
+    {
+      "epoch": 0.12385081697378457,
+      "learning_rate": 0.00017523184391183912,
+      "loss": 0.7849,
+      "mean_token_accuracy": 0.7733409106731415,
+      "num_tokens": 41955293.0,
+      "step": 12340
+    },
+    {
+      "epoch": 0.12395118230358505,
+      "learning_rate": 0.000175211770845879,
+      "loss": 0.8089,
+      "mean_token_accuracy": 0.761337673664093,
+      "num_tokens": 41989698.0,
+      "step": 12350
+    },
+    {
+      "epoch": 0.12405154763338552,
+      "learning_rate": 0.0001751916977799189,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7657052159309388,
+      "num_tokens": 42023229.0,
+      "step": 12360
+    },
+    {
+      "epoch": 0.12415191296318599,
+      "learning_rate": 0.00017517162471395882,
+      "loss": 0.7943,
+      "mean_token_accuracy": 0.7652049005031586,
+      "num_tokens": 42056794.0,
+      "step": 12370
+    },
+    {
+      "epoch": 0.12425227829298648,
+      "learning_rate": 0.00017515155164799873,
+      "loss": 0.7835,
+      "mean_token_accuracy": 0.7664673566818238,
+      "num_tokens": 42090348.0,
+      "step": 12380
+    },
+    {
+      "epoch": 0.12435264362278695,
+      "learning_rate": 0.0001751314785820386,
+      "loss": 0.7946,
+      "mean_token_accuracy": 0.7590865314006805,
+      "num_tokens": 42123920.0,
+      "step": 12390
+    },
+    {
+      "epoch": 0.12445300895258742,
+      "learning_rate": 0.00017511140551607855,
+      "loss": 0.7847,
+      "mean_token_accuracy": 0.7666195809841156,
+      "num_tokens": 42157663.0,
+      "step": 12400
+    },
+    {
+      "epoch": 0.12455337428238789,
+      "learning_rate": 0.00017509133245011843,
+      "loss": 0.7946,
+      "mean_token_accuracy": 0.765533983707428,
+      "num_tokens": 42191417.0,
+      "step": 12410
+    },
+    {
+      "epoch": 0.12465373961218837,
+      "learning_rate": 0.00017507125938415834,
+      "loss": 0.7981,
+      "mean_token_accuracy": 0.7656021595001221,
+      "num_tokens": 42225247.0,
+      "step": 12420
+    },
+    {
+      "epoch": 0.12475410494198884,
+      "learning_rate": 0.00017505118631819824,
+      "loss": 0.7686,
+      "mean_token_accuracy": 0.769657415151596,
+      "num_tokens": 42259252.0,
+      "step": 12430
+    },
+    {
+      "epoch": 0.12485447027178931,
+      "learning_rate": 0.00017503111325223815,
+      "loss": 0.8012,
+      "mean_token_accuracy": 0.7693645060062408,
+      "num_tokens": 42292686.0,
+      "step": 12440
+    },
+    {
+      "epoch": 0.12495483560158979,
+      "learning_rate": 0.00017501104018627806,
+      "loss": 0.8029,
+      "mean_token_accuracy": 0.7659431755542755,
+      "num_tokens": 42326756.0,
+      "step": 12450
+    },
+    {
+      "epoch": 0.12505520093139025,
+      "learning_rate": 0.00017499096712031797,
+      "loss": 0.7851,
+      "mean_token_accuracy": 0.767655348777771,
+      "num_tokens": 42360582.0,
+      "step": 12460
+    },
+    {
+      "epoch": 0.12515556626119073,
+      "learning_rate": 0.00017497089405435788,
+      "loss": 0.8064,
+      "mean_token_accuracy": 0.761356520652771,
+      "num_tokens": 42394863.0,
+      "step": 12470
+    },
+    {
+      "epoch": 0.12525593159099122,
+      "learning_rate": 0.00017495082098839776,
+      "loss": 0.7849,
+      "mean_token_accuracy": 0.7705150127410889,
+      "num_tokens": 42428642.0,
+      "step": 12480
+    },
+    {
+      "epoch": 0.12535629692079167,
+      "learning_rate": 0.0001749307479224377,
+      "loss": 0.7495,
+      "mean_token_accuracy": 0.7836252212524414,
+      "num_tokens": 42463108.0,
+      "step": 12490
+    },
+    {
+      "epoch": 0.12545666225059215,
+      "learning_rate": 0.00017491067485647758,
+      "loss": 0.7507,
+      "mean_token_accuracy": 0.7728869378566742,
+      "num_tokens": 42496448.0,
+      "step": 12500
+    },
+    {
+      "epoch": 0.12555702758039264,
+      "learning_rate": 0.00017489060179051748,
+      "loss": 0.76,
+      "mean_token_accuracy": 0.7759355187416077,
+      "num_tokens": 42529912.0,
+      "step": 12510
+    },
+    {
+      "epoch": 0.1256573929101931,
+      "learning_rate": 0.0001748705287245574,
+      "loss": 0.7994,
+      "mean_token_accuracy": 0.7663152515888214,
+      "num_tokens": 42563978.0,
+      "step": 12520
+    },
+    {
+      "epoch": 0.12575775823999358,
+      "learning_rate": 0.0001748504556585973,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7709584891796112,
+      "num_tokens": 42598037.0,
+      "step": 12530
+    },
+    {
+      "epoch": 0.12585812356979406,
+      "learning_rate": 0.0001748303825926372,
+      "loss": 0.8008,
+      "mean_token_accuracy": 0.7673484742641449,
+      "num_tokens": 42631788.0,
+      "step": 12540
+    },
+    {
+      "epoch": 0.12595848889959452,
+      "learning_rate": 0.00017481030952667712,
+      "loss": 0.7768,
+      "mean_token_accuracy": 0.7681606829166412,
+      "num_tokens": 42666401.0,
+      "step": 12550
+    },
+    {
+      "epoch": 0.126058854229395,
+      "learning_rate": 0.00017479023646071703,
+      "loss": 0.7879,
+      "mean_token_accuracy": 0.7740061402320861,
+      "num_tokens": 42699968.0,
+      "step": 12560
+    },
+    {
+      "epoch": 0.12615921955919548,
+      "learning_rate": 0.0001747701633947569,
+      "loss": 0.7701,
+      "mean_token_accuracy": 0.7727460920810699,
+      "num_tokens": 42733680.0,
+      "step": 12570
+    },
+    {
+      "epoch": 0.12625958488899594,
+      "learning_rate": 0.00017475009032879684,
+      "loss": 0.8036,
+      "mean_token_accuracy": 0.7588761568069458,
+      "num_tokens": 42767954.0,
+      "step": 12580
+    },
+    {
+      "epoch": 0.12635995021879642,
+      "learning_rate": 0.00017473001726283672,
+      "loss": 0.7755,
+      "mean_token_accuracy": 0.7718429625034332,
+      "num_tokens": 42802447.0,
+      "step": 12590
+    },
+    {
+      "epoch": 0.12646031554859688,
+      "learning_rate": 0.00017470994419687663,
+      "loss": 0.7585,
+      "mean_token_accuracy": 0.7763899385929107,
+      "num_tokens": 42836185.0,
+      "step": 12600
+    },
+    {
+      "epoch": 0.12656068087839736,
+      "learning_rate": 0.00017468987113091654,
+      "loss": 0.7922,
+      "mean_token_accuracy": 0.7698125064373016,
+      "num_tokens": 42870463.0,
+      "step": 12610
+    },
+    {
+      "epoch": 0.12666104620819785,
+      "learning_rate": 0.00017466979806495645,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7690317809581757,
+      "num_tokens": 42904594.0,
+      "step": 12620
+    },
+    {
+      "epoch": 0.1267614115379983,
+      "learning_rate": 0.00017464972499899633,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7723953187465668,
+      "num_tokens": 42938652.0,
+      "step": 12630
+    },
+    {
+      "epoch": 0.1268617768677988,
+      "learning_rate": 0.00017462965193303627,
+      "loss": 0.8115,
+      "mean_token_accuracy": 0.7605502486228943,
+      "num_tokens": 42972464.0,
+      "step": 12640
+    },
+    {
+      "epoch": 0.12696214219759927,
+      "learning_rate": 0.00017460957886707618,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.7675876438617706,
+      "num_tokens": 43007157.0,
+      "step": 12650
+    },
+    {
+      "epoch": 0.12706250752739973,
+      "learning_rate": 0.00017458950580111606,
+      "loss": 0.7688,
+      "mean_token_accuracy": 0.770320039987564,
+      "num_tokens": 43040658.0,
+      "step": 12660
+    },
+    {
+      "epoch": 0.1271628728572002,
+      "learning_rate": 0.000174569432735156,
+      "loss": 0.7849,
+      "mean_token_accuracy": 0.763702666759491,
+      "num_tokens": 43075908.0,
+      "step": 12670
+    },
+    {
+      "epoch": 0.1272632381870007,
+      "learning_rate": 0.00017454935966919587,
+      "loss": 0.8181,
+      "mean_token_accuracy": 0.7626882672309876,
+      "num_tokens": 43109535.0,
+      "step": 12680
+    },
+    {
+      "epoch": 0.12736360351680115,
+      "learning_rate": 0.0001745292866032358,
+      "loss": 0.7753,
+      "mean_token_accuracy": 0.7663200795650482,
+      "num_tokens": 43143526.0,
+      "step": 12690
+    },
+    {
+      "epoch": 0.12746396884660163,
+      "learning_rate": 0.0001745092135372757,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7678152918815613,
+      "num_tokens": 43176797.0,
+      "step": 12700
+    },
+    {
+      "epoch": 0.12756433417640212,
+      "learning_rate": 0.0001744891404713156,
+      "loss": 0.7946,
+      "mean_token_accuracy": 0.7646638870239257,
+      "num_tokens": 43210092.0,
+      "step": 12710
+    },
+    {
+      "epoch": 0.12766469950620257,
+      "learning_rate": 0.0001744690674053555,
+      "loss": 0.7792,
+      "mean_token_accuracy": 0.7689422249794007,
+      "num_tokens": 43244132.0,
+      "step": 12720
+    },
+    {
+      "epoch": 0.12776506483600306,
+      "learning_rate": 0.00017444899433939542,
+      "loss": 0.789,
+      "mean_token_accuracy": 0.7684858202934265,
+      "num_tokens": 43278611.0,
+      "step": 12730
+    },
+    {
+      "epoch": 0.1278654301658035,
+      "learning_rate": 0.0001744289212734353,
+      "loss": 0.7591,
+      "mean_token_accuracy": 0.7765051007270813,
+      "num_tokens": 43313373.0,
+      "step": 12740
+    },
+    {
+      "epoch": 0.127965795495604,
+      "learning_rate": 0.00017440884820747523,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.7724325597286225,
+      "num_tokens": 43347203.0,
+      "step": 12750
+    },
+    {
+      "epoch": 0.12806616082540448,
+      "learning_rate": 0.00017438877514151511,
+      "loss": 0.7674,
+      "mean_token_accuracy": 0.7737890899181366,
+      "num_tokens": 43381067.0,
+      "step": 12760
+    },
+    {
+      "epoch": 0.12816652615520494,
+      "learning_rate": 0.00017436870207555502,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7662975430488587,
+      "num_tokens": 43415482.0,
+      "step": 12770
+    },
+    {
+      "epoch": 0.12826689148500542,
+      "learning_rate": 0.00017434862900959493,
+      "loss": 0.7805,
+      "mean_token_accuracy": 0.7716028571128846,
+      "num_tokens": 43449886.0,
+      "step": 12780
+    },
+    {
+      "epoch": 0.1283672568148059,
+      "learning_rate": 0.00017432855594363484,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7711127996444702,
+      "num_tokens": 43484309.0,
+      "step": 12790
+    },
+    {
+      "epoch": 0.12846762214460636,
+      "learning_rate": 0.00017430848287767475,
+      "loss": 0.7954,
+      "mean_token_accuracy": 0.7629608392715455,
+      "num_tokens": 43518256.0,
+      "step": 12800
+    },
+    {
+      "epoch": 0.12856798747440684,
+      "learning_rate": 0.00017428840981171466,
+      "loss": 0.7683,
+      "mean_token_accuracy": 0.772105598449707,
+      "num_tokens": 43551721.0,
+      "step": 12810
+    },
+    {
+      "epoch": 0.12866835280420733,
+      "learning_rate": 0.00017426833674575456,
+      "loss": 0.7997,
+      "mean_token_accuracy": 0.7652229070663452,
+      "num_tokens": 43585449.0,
+      "step": 12820
+    },
+    {
+      "epoch": 0.12876871813400778,
+      "learning_rate": 0.00017424826367979445,
+      "loss": 0.7671,
+      "mean_token_accuracy": 0.7684259057044983,
+      "num_tokens": 43618648.0,
+      "step": 12830
+    },
+    {
+      "epoch": 0.12886908346380826,
+      "learning_rate": 0.00017422819061383438,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.7688802659511567,
+      "num_tokens": 43653129.0,
+      "step": 12840
+    },
+    {
+      "epoch": 0.12896944879360875,
+      "learning_rate": 0.00017420811754787426,
+      "loss": 0.7974,
+      "mean_token_accuracy": 0.7685005486011505,
+      "num_tokens": 43687601.0,
+      "step": 12850
+    },
+    {
+      "epoch": 0.1290698141234092,
+      "learning_rate": 0.00017418804448191417,
+      "loss": 0.7689,
+      "mean_token_accuracy": 0.7680033624172211,
+      "num_tokens": 43721180.0,
+      "step": 12860
+    },
+    {
+      "epoch": 0.1291701794532097,
+      "learning_rate": 0.00017416797141595408,
+      "loss": 0.8088,
+      "mean_token_accuracy": 0.7594873309135437,
+      "num_tokens": 43754834.0,
+      "step": 12870
+    },
+    {
+      "epoch": 0.12927054478301014,
+      "learning_rate": 0.000174147898349994,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7675781488418579,
+      "num_tokens": 43788573.0,
+      "step": 12880
+    },
+    {
+      "epoch": 0.12937091011281063,
+      "learning_rate": 0.0001741278252840339,
+      "loss": 0.7913,
+      "mean_token_accuracy": 0.7689393699169159,
+      "num_tokens": 43823352.0,
+      "step": 12890
+    },
+    {
+      "epoch": 0.1294712754426111,
+      "learning_rate": 0.0001741077522180738,
+      "loss": 0.7955,
+      "mean_token_accuracy": 0.7641589105129242,
+      "num_tokens": 43857292.0,
+      "step": 12900
+    },
+    {
+      "epoch": 0.12957164077241157,
+      "learning_rate": 0.0001740876791521137,
+      "loss": 0.7823,
+      "mean_token_accuracy": 0.7730109691619873,
+      "num_tokens": 43892081.0,
+      "step": 12910
+    },
+    {
+      "epoch": 0.12967200610221205,
+      "learning_rate": 0.0001740676060861536,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7731576800346375,
+      "num_tokens": 43925939.0,
+      "step": 12920
+    },
+    {
+      "epoch": 0.12977237143201253,
+      "learning_rate": 0.00017404753302019353,
+      "loss": 0.7879,
+      "mean_token_accuracy": 0.7724081039428711,
+      "num_tokens": 43959747.0,
+      "step": 12930
+    },
+    {
+      "epoch": 0.129872736761813,
+      "learning_rate": 0.0001740274599542334,
+      "loss": 0.7784,
+      "mean_token_accuracy": 0.7701088905334472,
+      "num_tokens": 43993034.0,
+      "step": 12940
+    },
+    {
+      "epoch": 0.12997310209161347,
+      "learning_rate": 0.00017400738688827332,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7703144550323486,
+      "num_tokens": 44026791.0,
+      "step": 12950
+    },
+    {
+      "epoch": 0.13007346742141396,
+      "learning_rate": 0.00017398731382231323,
+      "loss": 0.7712,
+      "mean_token_accuracy": 0.7708591520786285,
+      "num_tokens": 44060490.0,
+      "step": 12960
+    },
+    {
+      "epoch": 0.1301738327512144,
+      "learning_rate": 0.00017396724075635314,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.7685743629932403,
+      "num_tokens": 44095062.0,
+      "step": 12970
+    },
+    {
+      "epoch": 0.1302741980810149,
+      "learning_rate": 0.00017394716769039302,
+      "loss": 0.792,
+      "mean_token_accuracy": 0.7664016485214233,
+      "num_tokens": 44129691.0,
+      "step": 12980
+    },
+    {
+      "epoch": 0.13037456341081538,
+      "learning_rate": 0.00017392709462443295,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.7692232489585876,
+      "num_tokens": 44164031.0,
+      "step": 12990
+    },
+    {
+      "epoch": 0.13047492874061584,
+      "learning_rate": 0.00017390702155847286,
+      "loss": 0.7897,
+      "mean_token_accuracy": 0.7714962363243103,
+      "num_tokens": 44197894.0,
+      "step": 13000
+    },
+    {
+      "epoch": 0.13057529407041632,
+      "learning_rate": 0.00017388694849251274,
+      "loss": 0.791,
+      "mean_token_accuracy": 0.7653005301952363,
+      "num_tokens": 44232236.0,
+      "step": 13010
+    },
+    {
+      "epoch": 0.13067565940021678,
+      "learning_rate": 0.00017386687542655268,
+      "loss": 0.8147,
+      "mean_token_accuracy": 0.7650872945785523,
+      "num_tokens": 44267014.0,
+      "step": 13020
+    },
+    {
+      "epoch": 0.13077602473001726,
+      "learning_rate": 0.00017384680236059256,
+      "loss": 0.7727,
+      "mean_token_accuracy": 0.7726529896259308,
+      "num_tokens": 44300640.0,
+      "step": 13030
+    },
+    {
+      "epoch": 0.13087639005981774,
+      "learning_rate": 0.00017382672929463247,
+      "loss": 0.7807,
+      "mean_token_accuracy": 0.7766844749450683,
+      "num_tokens": 44335628.0,
+      "step": 13040
+    },
+    {
+      "epoch": 0.1309767553896182,
+      "learning_rate": 0.00017380665622867238,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7728781461715698,
+      "num_tokens": 44369668.0,
+      "step": 13050
+    },
+    {
+      "epoch": 0.13107712071941868,
+      "learning_rate": 0.00017378658316271229,
+      "loss": 0.7737,
+      "mean_token_accuracy": 0.7709382057189942,
+      "num_tokens": 44404473.0,
+      "step": 13060
+    },
+    {
+      "epoch": 0.13117748604921917,
+      "learning_rate": 0.00017376651009675217,
+      "loss": 0.8177,
+      "mean_token_accuracy": 0.7549030959606171,
+      "num_tokens": 44438206.0,
+      "step": 13070
+    },
+    {
+      "epoch": 0.13127785137901962,
+      "learning_rate": 0.0001737464370307921,
+      "loss": 0.7607,
+      "mean_token_accuracy": 0.7736218094825744,
+      "num_tokens": 44473666.0,
+      "step": 13080
+    },
+    {
+      "epoch": 0.1313782167088201,
+      "learning_rate": 0.00017372636396483198,
+      "loss": 0.8073,
+      "mean_token_accuracy": 0.766367393732071,
+      "num_tokens": 44507583.0,
+      "step": 13090
+    },
+    {
+      "epoch": 0.1314785820386206,
+      "learning_rate": 0.0001737062908988719,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7713600337505341,
+      "num_tokens": 44540798.0,
+      "step": 13100
+    },
+    {
+      "epoch": 0.13157894736842105,
+      "learning_rate": 0.0001736862178329118,
+      "loss": 0.7622,
+      "mean_token_accuracy": 0.7700433492660522,
+      "num_tokens": 44574823.0,
+      "step": 13110
+    },
+    {
+      "epoch": 0.13167931269822153,
+      "learning_rate": 0.0001736661447669517,
+      "loss": 0.797,
+      "mean_token_accuracy": 0.7680361986160278,
+      "num_tokens": 44608377.0,
+      "step": 13120
+    },
+    {
+      "epoch": 0.131779678028022,
+      "learning_rate": 0.00017364607170099162,
+      "loss": 0.7792,
+      "mean_token_accuracy": 0.7709780335426331,
+      "num_tokens": 44642990.0,
+      "step": 13130
+    },
+    {
+      "epoch": 0.13188004335782247,
+      "learning_rate": 0.00017362599863503153,
+      "loss": 0.7647,
+      "mean_token_accuracy": 0.7713236749172211,
+      "num_tokens": 44677186.0,
+      "step": 13140
+    },
+    {
+      "epoch": 0.13198040868762295,
+      "learning_rate": 0.00017360592556907143,
+      "loss": 0.8078,
+      "mean_token_accuracy": 0.7607882797718049,
+      "num_tokens": 44711522.0,
+      "step": 13150
+    },
+    {
+      "epoch": 0.1320807740174234,
+      "learning_rate": 0.00017358585250311132,
+      "loss": 0.7737,
+      "mean_token_accuracy": 0.7688131511211396,
+      "num_tokens": 44746232.0,
+      "step": 13160
+    },
+    {
+      "epoch": 0.1321811393472239,
+      "learning_rate": 0.00017356577943715125,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7671611368656158,
+      "num_tokens": 44779911.0,
+      "step": 13170
+    },
+    {
+      "epoch": 0.13228150467702438,
+      "learning_rate": 0.00017354570637119113,
+      "loss": 0.7855,
+      "mean_token_accuracy": 0.7711513102054596,
+      "num_tokens": 44813918.0,
+      "step": 13180
+    },
+    {
+      "epoch": 0.13238187000682483,
+      "learning_rate": 0.00017352563330523104,
+      "loss": 0.7996,
+      "mean_token_accuracy": 0.7630888521671295,
+      "num_tokens": 44847945.0,
+      "step": 13190
+    },
+    {
+      "epoch": 0.13248223533662531,
+      "learning_rate": 0.00017350556023927095,
+      "loss": 0.7831,
+      "mean_token_accuracy": 0.7665793359279632,
+      "num_tokens": 44882085.0,
+      "step": 13200
+    },
+    {
+      "epoch": 0.1325826006664258,
+      "learning_rate": 0.00017348548717331086,
+      "loss": 0.8153,
+      "mean_token_accuracy": 0.7603378415107727,
+      "num_tokens": 44916298.0,
+      "step": 13210
+    },
+    {
+      "epoch": 0.13268296599622625,
+      "learning_rate": 0.00017346541410735077,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7715738892555237,
+      "num_tokens": 44949774.0,
+      "step": 13220
+    },
+    {
+      "epoch": 0.13278333132602674,
+      "learning_rate": 0.00017344534104139067,
+      "loss": 0.7429,
+      "mean_token_accuracy": 0.77811558842659,
+      "num_tokens": 44983903.0,
+      "step": 13230
+    },
+    {
+      "epoch": 0.13288369665582722,
+      "learning_rate": 0.00017342526797543058,
+      "loss": 0.7714,
+      "mean_token_accuracy": 0.7692193686962128,
+      "num_tokens": 45018521.0,
+      "step": 13240
+    },
+    {
+      "epoch": 0.13298406198562768,
+      "learning_rate": 0.00017340519490947046,
+      "loss": 0.7701,
+      "mean_token_accuracy": 0.770143574476242,
+      "num_tokens": 45052925.0,
+      "step": 13250
+    },
+    {
+      "epoch": 0.13308442731542816,
+      "learning_rate": 0.0001733851218435104,
+      "loss": 0.7908,
+      "mean_token_accuracy": 0.7723272562026977,
+      "num_tokens": 45086762.0,
+      "step": 13260
+    },
+    {
+      "epoch": 0.13318479264522864,
+      "learning_rate": 0.00017336504877755028,
+      "loss": 0.7799,
+      "mean_token_accuracy": 0.770423299074173,
+      "num_tokens": 45120480.0,
+      "step": 13270
+    },
+    {
+      "epoch": 0.1332851579750291,
+      "learning_rate": 0.00017334497571159022,
+      "loss": 0.792,
+      "mean_token_accuracy": 0.7626452445983887,
+      "num_tokens": 45155019.0,
+      "step": 13280
+    },
+    {
+      "epoch": 0.13338552330482958,
+      "learning_rate": 0.0001733249026456301,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7737775027751923,
+      "num_tokens": 45190033.0,
+      "step": 13290
+    },
+    {
+      "epoch": 0.13348588863463004,
+      "learning_rate": 0.00017330482957967,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.7683071434497833,
+      "num_tokens": 45223426.0,
+      "step": 13300
+    },
+    {
+      "epoch": 0.13358625396443052,
+      "learning_rate": 0.00017328475651370992,
+      "loss": 0.7866,
+      "mean_token_accuracy": 0.7668528437614441,
+      "num_tokens": 45256840.0,
+      "step": 13310
+    },
+    {
+      "epoch": 0.133686619294231,
+      "learning_rate": 0.00017326468344774982,
+      "loss": 0.8124,
+      "mean_token_accuracy": 0.7612246215343476,
+      "num_tokens": 45290629.0,
+      "step": 13320
+    },
+    {
+      "epoch": 0.13378698462403146,
+      "learning_rate": 0.0001732446103817897,
+      "loss": 0.8165,
+      "mean_token_accuracy": 0.7597314715385437,
+      "num_tokens": 45324283.0,
+      "step": 13330
+    },
+    {
+      "epoch": 0.13388734995383195,
+      "learning_rate": 0.00017322453731582964,
+      "loss": 0.7603,
+      "mean_token_accuracy": 0.7741209208965302,
+      "num_tokens": 45358674.0,
+      "step": 13340
+    },
+    {
+      "epoch": 0.13398771528363243,
+      "learning_rate": 0.00017320446424986955,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7722501635551453,
+      "num_tokens": 45393358.0,
+      "step": 13350
+    },
+    {
+      "epoch": 0.1340880806134329,
+      "learning_rate": 0.00017318439118390943,
+      "loss": 0.8138,
+      "mean_token_accuracy": 0.759548419713974,
+      "num_tokens": 45428044.0,
+      "step": 13360
+    },
+    {
+      "epoch": 0.13418844594323337,
+      "learning_rate": 0.00017316431811794937,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7725479602813721,
+      "num_tokens": 45463074.0,
+      "step": 13370
+    },
+    {
+      "epoch": 0.13428881127303385,
+      "learning_rate": 0.00017314424505198925,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.7658045768737793,
+      "num_tokens": 45497004.0,
+      "step": 13380
+    },
+    {
+      "epoch": 0.1343891766028343,
+      "learning_rate": 0.00017312417198602916,
+      "loss": 0.7651,
+      "mean_token_accuracy": 0.7743133664131164,
+      "num_tokens": 45531498.0,
+      "step": 13390
+    },
+    {
+      "epoch": 0.1344895419326348,
+      "learning_rate": 0.00017310409892006906,
+      "loss": 0.8109,
+      "mean_token_accuracy": 0.7595912873744964,
+      "num_tokens": 45565567.0,
+      "step": 13400
+    },
+    {
+      "epoch": 0.13458990726243528,
+      "learning_rate": 0.00017308402585410897,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.772896945476532,
+      "num_tokens": 45599978.0,
+      "step": 13410
+    },
+    {
+      "epoch": 0.13469027259223573,
+      "learning_rate": 0.00017306395278814885,
+      "loss": 0.771,
+      "mean_token_accuracy": 0.7661577820777893,
+      "num_tokens": 45633574.0,
+      "step": 13420
+    },
+    {
+      "epoch": 0.13479063792203622,
+      "learning_rate": 0.0001730438797221888,
+      "loss": 0.8021,
+      "mean_token_accuracy": 0.7706047177314759,
+      "num_tokens": 45667545.0,
+      "step": 13430
+    },
+    {
+      "epoch": 0.13489100325183667,
+      "learning_rate": 0.00017302380665622867,
+      "loss": 0.7708,
+      "mean_token_accuracy": 0.7734810650348664,
+      "num_tokens": 45701413.0,
+      "step": 13440
+    },
+    {
+      "epoch": 0.13499136858163716,
+      "learning_rate": 0.00017300373359026858,
+      "loss": 0.8189,
+      "mean_token_accuracy": 0.7598798215389252,
+      "num_tokens": 45735144.0,
+      "step": 13450
+    },
+    {
+      "epoch": 0.13509173391143764,
+      "learning_rate": 0.0001729836605243085,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.7524699211120606,
+      "num_tokens": 45768067.0,
+      "step": 13460
+    },
+    {
+      "epoch": 0.1351920992412381,
+      "learning_rate": 0.0001729635874583484,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7695520401000977,
+      "num_tokens": 45802032.0,
+      "step": 13470
+    },
+    {
+      "epoch": 0.13529246457103858,
+      "learning_rate": 0.0001729435143923883,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.773700475692749,
+      "num_tokens": 45835979.0,
+      "step": 13480
+    },
+    {
+      "epoch": 0.13539282990083906,
+      "learning_rate": 0.0001729234413264282,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.772723913192749,
+      "num_tokens": 45870257.0,
+      "step": 13490
+    },
+    {
+      "epoch": 0.13549319523063952,
+      "learning_rate": 0.00017290336826046812,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.771820068359375,
+      "num_tokens": 45904177.0,
+      "step": 13500
+    },
+    {
+      "epoch": 0.13559356056044,
+      "learning_rate": 0.000172883295194508,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7701919913291931,
+      "num_tokens": 45937497.0,
+      "step": 13510
+    },
+    {
+      "epoch": 0.13569392589024049,
+      "learning_rate": 0.00017286322212854794,
+      "loss": 0.7851,
+      "mean_token_accuracy": 0.7701319694519043,
+      "num_tokens": 45971675.0,
+      "step": 13520
+    },
+    {
+      "epoch": 0.13579429122004094,
+      "learning_rate": 0.00017284314906258782,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.7749830961227417,
+      "num_tokens": 46005934.0,
+      "step": 13530
+    },
+    {
+      "epoch": 0.13589465654984142,
+      "learning_rate": 0.00017282307599662773,
+      "loss": 0.7927,
+      "mean_token_accuracy": 0.7665489256381989,
+      "num_tokens": 46040017.0,
+      "step": 13540
+    },
+    {
+      "epoch": 0.1359950218796419,
+      "learning_rate": 0.00017280300293066764,
+      "loss": 0.7818,
+      "mean_token_accuracy": 0.7697871625423431,
+      "num_tokens": 46074303.0,
+      "step": 13550
+    },
+    {
+      "epoch": 0.13609538720944236,
+      "learning_rate": 0.00017278292986470754,
+      "loss": 0.7797,
+      "mean_token_accuracy": 0.7691508531570435,
+      "num_tokens": 46107970.0,
+      "step": 13560
+    },
+    {
+      "epoch": 0.13619575253924285,
+      "learning_rate": 0.00017276285679874745,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.7740316569805146,
+      "num_tokens": 46142060.0,
+      "step": 13570
+    },
+    {
+      "epoch": 0.1362961178690433,
+      "learning_rate": 0.00017274278373278736,
+      "loss": 0.7739,
+      "mean_token_accuracy": 0.7710063636302948,
+      "num_tokens": 46176680.0,
+      "step": 13580
+    },
+    {
+      "epoch": 0.1363964831988438,
+      "learning_rate": 0.00017272271066682727,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.7800506114959717,
+      "num_tokens": 46210560.0,
+      "step": 13590
+    },
+    {
+      "epoch": 0.13649684852864427,
+      "learning_rate": 0.00017270263760086715,
+      "loss": 0.7753,
+      "mean_token_accuracy": 0.7694782733917236,
+      "num_tokens": 46244876.0,
+      "step": 13600
+    },
+    {
+      "epoch": 0.13659721385844473,
+      "learning_rate": 0.0001726825645349071,
+      "loss": 0.8214,
+      "mean_token_accuracy": 0.7506749331951141,
+      "num_tokens": 46278744.0,
+      "step": 13610
+    },
+    {
+      "epoch": 0.1366975791882452,
+      "learning_rate": 0.00017266249146894697,
+      "loss": 0.7883,
+      "mean_token_accuracy": 0.772595876455307,
+      "num_tokens": 46312450.0,
+      "step": 13620
+    },
+    {
+      "epoch": 0.1367979445180457,
+      "learning_rate": 0.00017264241840298688,
+      "loss": 0.7715,
+      "mean_token_accuracy": 0.7711716651916504,
+      "num_tokens": 46346374.0,
+      "step": 13630
+    },
+    {
+      "epoch": 0.13689830984784615,
+      "learning_rate": 0.00017262234533702679,
+      "loss": 0.7385,
+      "mean_token_accuracy": 0.7801128923892975,
+      "num_tokens": 46380150.0,
+      "step": 13640
+    },
+    {
+      "epoch": 0.13699867517764663,
+      "learning_rate": 0.0001726022722710667,
+      "loss": 0.7758,
+      "mean_token_accuracy": 0.7712238490581512,
+      "num_tokens": 46414625.0,
+      "step": 13650
+    },
+    {
+      "epoch": 0.13709904050744712,
+      "learning_rate": 0.00017258219920510657,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.7679083704948425,
+      "num_tokens": 46448971.0,
+      "step": 13660
+    },
+    {
+      "epoch": 0.13719940583724757,
+      "learning_rate": 0.0001725621261391465,
+      "loss": 0.7807,
+      "mean_token_accuracy": 0.7680625975131988,
+      "num_tokens": 46483608.0,
+      "step": 13670
+    },
+    {
+      "epoch": 0.13729977116704806,
+      "learning_rate": 0.0001725420530731864,
+      "loss": 0.7679,
+      "mean_token_accuracy": 0.7741383075714111,
+      "num_tokens": 46517840.0,
+      "step": 13680
+    },
+    {
+      "epoch": 0.13740013649684854,
+      "learning_rate": 0.0001725219800072263,
+      "loss": 0.7813,
+      "mean_token_accuracy": 0.7687681138515472,
+      "num_tokens": 46552218.0,
+      "step": 13690
+    },
+    {
+      "epoch": 0.137500501826649,
+      "learning_rate": 0.00017250190694126624,
+      "loss": 0.7856,
+      "mean_token_accuracy": 0.7702401280403137,
+      "num_tokens": 46586259.0,
+      "step": 13700
+    },
+    {
+      "epoch": 0.13760086715644948,
+      "learning_rate": 0.00017248183387530612,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7646296799182892,
+      "num_tokens": 46620640.0,
+      "step": 13710
+    },
+    {
+      "epoch": 0.13770123248624994,
+      "learning_rate": 0.00017246176080934603,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7727301299571991,
+      "num_tokens": 46654435.0,
+      "step": 13720
+    },
+    {
+      "epoch": 0.13780159781605042,
+      "learning_rate": 0.00017244168774338593,
+      "loss": 0.7903,
+      "mean_token_accuracy": 0.7676368415355682,
+      "num_tokens": 46688023.0,
+      "step": 13730
+    },
+    {
+      "epoch": 0.1379019631458509,
+      "learning_rate": 0.00017242161467742584,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7680228412151336,
+      "num_tokens": 46722401.0,
+      "step": 13740
+    },
+    {
+      "epoch": 0.13800232847565136,
+      "learning_rate": 0.00017240154161146572,
+      "loss": 0.8088,
+      "mean_token_accuracy": 0.7624124407768249,
+      "num_tokens": 46756080.0,
+      "step": 13750
+    },
+    {
+      "epoch": 0.13810269380545184,
+      "learning_rate": 0.00017238146854550566,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.7686403095722198,
+      "num_tokens": 46790453.0,
+      "step": 13760
+    },
+    {
+      "epoch": 0.13820305913525233,
+      "learning_rate": 0.00017236139547954554,
+      "loss": 0.7623,
+      "mean_token_accuracy": 0.7764238238334655,
+      "num_tokens": 46824206.0,
+      "step": 13770
+    },
+    {
+      "epoch": 0.13830342446505278,
+      "learning_rate": 0.00017234132241358545,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7711726546287536,
+      "num_tokens": 46857842.0,
+      "step": 13780
+    },
+    {
+      "epoch": 0.13840378979485327,
+      "learning_rate": 0.00017232124934762536,
+      "loss": 0.7812,
+      "mean_token_accuracy": 0.769701623916626,
+      "num_tokens": 46891852.0,
+      "step": 13790
+    },
+    {
+      "epoch": 0.13850415512465375,
+      "learning_rate": 0.00017230117628166527,
+      "loss": 0.812,
+      "mean_token_accuracy": 0.7646488845348358,
+      "num_tokens": 46925378.0,
+      "step": 13800
+    },
+    {
+      "epoch": 0.1386045204544542,
+      "learning_rate": 0.00017228110321570517,
+      "loss": 0.7333,
+      "mean_token_accuracy": 0.7735948204994202,
+      "num_tokens": 46959647.0,
+      "step": 13810
+    },
+    {
+      "epoch": 0.1387048857842547,
+      "learning_rate": 0.00017226103014974508,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.7705456852912903,
+      "num_tokens": 46993300.0,
+      "step": 13820
+    },
+    {
+      "epoch": 0.13880525111405517,
+      "learning_rate": 0.000172240957083785,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.7709096789360046,
+      "num_tokens": 47027822.0,
+      "step": 13830
+    },
+    {
+      "epoch": 0.13890561644385563,
+      "learning_rate": 0.0001722208840178249,
+      "loss": 0.7351,
+      "mean_token_accuracy": 0.7875620603561402,
+      "num_tokens": 47062331.0,
+      "step": 13840
+    },
+    {
+      "epoch": 0.1390059817736561,
+      "learning_rate": 0.0001722008109518648,
+      "loss": 0.7799,
+      "mean_token_accuracy": 0.7704236924648284,
+      "num_tokens": 47095947.0,
+      "step": 13850
+    },
+    {
+      "epoch": 0.1391063471034566,
+      "learning_rate": 0.0001721807378859047,
+      "loss": 0.7928,
+      "mean_token_accuracy": 0.7634164929389954,
+      "num_tokens": 47129887.0,
+      "step": 13860
+    },
+    {
+      "epoch": 0.13920671243325705,
+      "learning_rate": 0.00017216066481994462,
+      "loss": 0.7675,
+      "mean_token_accuracy": 0.7728378593921661,
+      "num_tokens": 47164407.0,
+      "step": 13870
+    },
+    {
+      "epoch": 0.13930707776305754,
+      "learning_rate": 0.0001721405917539845,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7675823390483856,
+      "num_tokens": 47197902.0,
+      "step": 13880
+    },
+    {
+      "epoch": 0.139407443092858,
+      "learning_rate": 0.00017212051868802441,
+      "loss": 0.7996,
+      "mean_token_accuracy": 0.7673685848712921,
+      "num_tokens": 47231665.0,
+      "step": 13890
+    },
+    {
+      "epoch": 0.13950780842265847,
+      "learning_rate": 0.00017210044562206432,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.765827351808548,
+      "num_tokens": 47265497.0,
+      "step": 13900
+    },
+    {
+      "epoch": 0.13960817375245896,
+      "learning_rate": 0.00017208037255610423,
+      "loss": 0.794,
+      "mean_token_accuracy": 0.7675317227840424,
+      "num_tokens": 47299717.0,
+      "step": 13910
+    },
+    {
+      "epoch": 0.13970853908225941,
+      "learning_rate": 0.00017206029949014414,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.775213748216629,
+      "num_tokens": 47333205.0,
+      "step": 13920
+    },
+    {
+      "epoch": 0.1398089044120599,
+      "learning_rate": 0.00017204022642418405,
+      "loss": 0.7833,
+      "mean_token_accuracy": 0.7670728325843811,
+      "num_tokens": 47367000.0,
+      "step": 13930
+    },
+    {
+      "epoch": 0.13990926974186038,
+      "learning_rate": 0.00017202015335822396,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.7673534214496612,
+      "num_tokens": 47400247.0,
+      "step": 13940
+    },
+    {
+      "epoch": 0.14000963507166084,
+      "learning_rate": 0.00017200008029226384,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7697259187698364,
+      "num_tokens": 47433823.0,
+      "step": 13950
+    },
+    {
+      "epoch": 0.14011000040146132,
+      "learning_rate": 0.00017198000722630377,
+      "loss": 0.7617,
+      "mean_token_accuracy": 0.7738004982471466,
+      "num_tokens": 47468682.0,
+      "step": 13960
+    },
+    {
+      "epoch": 0.1402103657312618,
+      "learning_rate": 0.00017195993416034366,
+      "loss": 0.7536,
+      "mean_token_accuracy": 0.7769059181213379,
+      "num_tokens": 47502876.0,
+      "step": 13970
+    },
+    {
+      "epoch": 0.14031073106106226,
+      "learning_rate": 0.00017193986109438356,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7735214591026306,
+      "num_tokens": 47537313.0,
+      "step": 13980
+    },
+    {
+      "epoch": 0.14041109639086274,
+      "learning_rate": 0.00017191978802842347,
+      "loss": 0.773,
+      "mean_token_accuracy": 0.7725825011730194,
+      "num_tokens": 47571993.0,
+      "step": 13990
+    },
+    {
+      "epoch": 0.14051146172066323,
+      "learning_rate": 0.00017189971496246338,
+      "loss": 0.8027,
+      "mean_token_accuracy": 0.7586640000343323,
+      "num_tokens": 47605283.0,
+      "step": 14000
+    },
+    {
+      "epoch": 0.14061182705046368,
+      "learning_rate": 0.00017187964189650326,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.7648543298244477,
+      "num_tokens": 47639356.0,
+      "step": 14010
+    },
+    {
+      "epoch": 0.14071219238026417,
+      "learning_rate": 0.0001718595688305432,
+      "loss": 0.7727,
+      "mean_token_accuracy": 0.7667241334915161,
+      "num_tokens": 47673439.0,
+      "step": 14020
+    },
+    {
+      "epoch": 0.14081255771006462,
+      "learning_rate": 0.00017183949576458308,
+      "loss": 0.7986,
+      "mean_token_accuracy": 0.7616683006286621,
+      "num_tokens": 47708473.0,
+      "step": 14030
+    },
+    {
+      "epoch": 0.1409129230398651,
+      "learning_rate": 0.000171819422698623,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7730585336685181,
+      "num_tokens": 47742040.0,
+      "step": 14040
+    },
+    {
+      "epoch": 0.1410132883696656,
+      "learning_rate": 0.00017179934963266292,
+      "loss": 0.7786,
+      "mean_token_accuracy": 0.7654690504074096,
+      "num_tokens": 47776275.0,
+      "step": 14050
+    },
+    {
+      "epoch": 0.14111365369946605,
+      "learning_rate": 0.0001717792765667028,
+      "loss": 0.7641,
+      "mean_token_accuracy": 0.7734162449836731,
+      "num_tokens": 47810210.0,
+      "step": 14060
+    },
+    {
+      "epoch": 0.14121401902926653,
+      "learning_rate": 0.0001717592035007427,
+      "loss": 0.8132,
+      "mean_token_accuracy": 0.7598636031150818,
+      "num_tokens": 47844089.0,
+      "step": 14070
+    },
+    {
+      "epoch": 0.141314384359067,
+      "learning_rate": 0.00017173913043478262,
+      "loss": 0.8046,
+      "mean_token_accuracy": 0.7569852828979492,
+      "num_tokens": 47878100.0,
+      "step": 14080
+    },
+    {
+      "epoch": 0.14141474968886747,
+      "learning_rate": 0.00017171905736882253,
+      "loss": 0.7758,
+      "mean_token_accuracy": 0.7700919866561889,
+      "num_tokens": 47911663.0,
+      "step": 14090
+    },
+    {
+      "epoch": 0.14151511501866795,
+      "learning_rate": 0.0001716989843028624,
+      "loss": 0.7978,
+      "mean_token_accuracy": 0.7650790452957154,
+      "num_tokens": 47945287.0,
+      "step": 14100
+    },
+    {
+      "epoch": 0.14161548034846844,
+      "learning_rate": 0.00017167891123690235,
+      "loss": 0.7581,
+      "mean_token_accuracy": 0.769956111907959,
+      "num_tokens": 47979839.0,
+      "step": 14110
+    },
+    {
+      "epoch": 0.1417158456782689,
+      "learning_rate": 0.00017165883817094223,
+      "loss": 0.8178,
+      "mean_token_accuracy": 0.7598431885242463,
+      "num_tokens": 48013703.0,
+      "step": 14120
+    },
+    {
+      "epoch": 0.14181621100806938,
+      "learning_rate": 0.00017163876510498214,
+      "loss": 0.7827,
+      "mean_token_accuracy": 0.7717898309230804,
+      "num_tokens": 48048922.0,
+      "step": 14130
+    },
+    {
+      "epoch": 0.14191657633786986,
+      "learning_rate": 0.00017161869203902204,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7692624986171722,
+      "num_tokens": 48083289.0,
+      "step": 14140
+    },
+    {
+      "epoch": 0.14201694166767032,
+      "learning_rate": 0.00017159861897306195,
+      "loss": 0.768,
+      "mean_token_accuracy": 0.7684755384922027,
+      "num_tokens": 48117632.0,
+      "step": 14150
+    },
+    {
+      "epoch": 0.1421173069974708,
+      "learning_rate": 0.00017157854590710186,
+      "loss": 0.775,
+      "mean_token_accuracy": 0.7717197239398956,
+      "num_tokens": 48151932.0,
+      "step": 14160
+    },
+    {
+      "epoch": 0.14221767232727126,
+      "learning_rate": 0.00017155847284114177,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.770142924785614,
+      "num_tokens": 48186237.0,
+      "step": 14170
+    },
+    {
+      "epoch": 0.14231803765707174,
+      "learning_rate": 0.00017153839977518168,
+      "loss": 0.7625,
+      "mean_token_accuracy": 0.774982887506485,
+      "num_tokens": 48220618.0,
+      "step": 14180
+    },
+    {
+      "epoch": 0.14241840298687222,
+      "learning_rate": 0.00017151832670922156,
+      "loss": 0.7561,
+      "mean_token_accuracy": 0.7678909063339233,
+      "num_tokens": 48254715.0,
+      "step": 14190
+    },
+    {
+      "epoch": 0.14251876831667268,
+      "learning_rate": 0.0001714982536432615,
+      "loss": 0.8216,
+      "mean_token_accuracy": 0.7582562386989593,
+      "num_tokens": 48289057.0,
+      "step": 14200
+    },
+    {
+      "epoch": 0.14261913364647316,
+      "learning_rate": 0.00017147818057730138,
+      "loss": 0.7678,
+      "mean_token_accuracy": 0.7684305965900421,
+      "num_tokens": 48322743.0,
+      "step": 14210
+    },
+    {
+      "epoch": 0.14271949897627365,
+      "learning_rate": 0.00017145810751134128,
+      "loss": 0.775,
+      "mean_token_accuracy": 0.7674304842948914,
+      "num_tokens": 48356844.0,
+      "step": 14220
+    },
+    {
+      "epoch": 0.1428198643060741,
+      "learning_rate": 0.0001714380344453812,
+      "loss": 0.7682,
+      "mean_token_accuracy": 0.7700712382793427,
+      "num_tokens": 48391103.0,
+      "step": 14230
+    },
+    {
+      "epoch": 0.14292022963587458,
+      "learning_rate": 0.0001714179613794211,
+      "loss": 0.7638,
+      "mean_token_accuracy": 0.7687857925891877,
+      "num_tokens": 48426188.0,
+      "step": 14240
+    },
+    {
+      "epoch": 0.14302059496567507,
+      "learning_rate": 0.000171397888313461,
+      "loss": 0.8046,
+      "mean_token_accuracy": 0.765725177526474,
+      "num_tokens": 48459984.0,
+      "step": 14250
+    },
+    {
+      "epoch": 0.14312096029547552,
+      "learning_rate": 0.00017137781524750092,
+      "loss": 0.7996,
+      "mean_token_accuracy": 0.7675525188446045,
+      "num_tokens": 48493797.0,
+      "step": 14260
+    },
+    {
+      "epoch": 0.143221325625276,
+      "learning_rate": 0.00017135774218154083,
+      "loss": 0.7695,
+      "mean_token_accuracy": 0.7674437403678894,
+      "num_tokens": 48527438.0,
+      "step": 14270
+    },
+    {
+      "epoch": 0.1433216909550765,
+      "learning_rate": 0.0001713376691155807,
+      "loss": 0.7818,
+      "mean_token_accuracy": 0.7690040349960328,
+      "num_tokens": 48561319.0,
+      "step": 14280
+    },
+    {
+      "epoch": 0.14342205628487695,
+      "learning_rate": 0.00017131759604962064,
+      "loss": 0.7419,
+      "mean_token_accuracy": 0.7776353180408477,
+      "num_tokens": 48596375.0,
+      "step": 14290
+    },
+    {
+      "epoch": 0.14352242161467743,
+      "learning_rate": 0.00017129752298366053,
+      "loss": 0.8173,
+      "mean_token_accuracy": 0.7611851871013642,
+      "num_tokens": 48629504.0,
+      "step": 14300
+    },
+    {
+      "epoch": 0.1436227869444779,
+      "learning_rate": 0.00017127744991770043,
+      "loss": 0.7744,
+      "mean_token_accuracy": 0.77262002825737,
+      "num_tokens": 48663457.0,
+      "step": 14310
+    },
+    {
+      "epoch": 0.14372315227427837,
+      "learning_rate": 0.00017125737685174034,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7743302881717682,
+      "num_tokens": 48696690.0,
+      "step": 14320
+    },
+    {
+      "epoch": 0.14382351760407885,
+      "learning_rate": 0.00017123730378578025,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.7791252076625824,
+      "num_tokens": 48730493.0,
+      "step": 14330
+    },
+    {
+      "epoch": 0.1439238829338793,
+      "learning_rate": 0.00017121723071982013,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7683629155158996,
+      "num_tokens": 48764417.0,
+      "step": 14340
+    },
+    {
+      "epoch": 0.1440242482636798,
+      "learning_rate": 0.00017119715765386007,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7765732944011688,
+      "num_tokens": 48798453.0,
+      "step": 14350
+    },
+    {
+      "epoch": 0.14412461359348028,
+      "learning_rate": 0.00017117708458789995,
+      "loss": 0.7683,
+      "mean_token_accuracy": 0.7712306499481201,
+      "num_tokens": 48832693.0,
+      "step": 14360
+    },
+    {
+      "epoch": 0.14422497892328073,
+      "learning_rate": 0.00017115701152193986,
+      "loss": 0.7457,
+      "mean_token_accuracy": 0.7717716634273529,
+      "num_tokens": 48866575.0,
+      "step": 14370
+    },
+    {
+      "epoch": 0.14432534425308122,
+      "learning_rate": 0.00017113693845597977,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7678154647350312,
+      "num_tokens": 48901108.0,
+      "step": 14380
+    },
+    {
+      "epoch": 0.1444257095828817,
+      "learning_rate": 0.00017111686539001967,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.7703318357467651,
+      "num_tokens": 48934997.0,
+      "step": 14390
+    },
+    {
+      "epoch": 0.14452607491268216,
+      "learning_rate": 0.0001710967923240596,
+      "loss": 0.8135,
+      "mean_token_accuracy": 0.7601113379001617,
+      "num_tokens": 48968946.0,
+      "step": 14400
+    },
+    {
+      "epoch": 0.14462644024248264,
+      "learning_rate": 0.0001710767192580995,
+      "loss": 0.7806,
+      "mean_token_accuracy": 0.7674680829048157,
+      "num_tokens": 49002596.0,
+      "step": 14410
+    },
+    {
+      "epoch": 0.14472680557228312,
+      "learning_rate": 0.0001710566461921394,
+      "loss": 0.7753,
+      "mean_token_accuracy": 0.7743801653385163,
+      "num_tokens": 49037227.0,
+      "step": 14420
+    },
+    {
+      "epoch": 0.14482717090208358,
+      "learning_rate": 0.0001710365731261793,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7689658284187317,
+      "num_tokens": 49071575.0,
+      "step": 14430
+    },
+    {
+      "epoch": 0.14492753623188406,
+      "learning_rate": 0.00017101650006021922,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7673587083816529,
+      "num_tokens": 49105269.0,
+      "step": 14440
+    },
+    {
+      "epoch": 0.14502790156168452,
+      "learning_rate": 0.0001709964269942591,
+      "loss": 0.7291,
+      "mean_token_accuracy": 0.7780017077922821,
+      "num_tokens": 49139078.0,
+      "step": 14450
+    },
+    {
+      "epoch": 0.145128266891485,
+      "learning_rate": 0.00017097635392829903,
+      "loss": 0.7903,
+      "mean_token_accuracy": 0.763459712266922,
+      "num_tokens": 49172279.0,
+      "step": 14460
+    },
+    {
+      "epoch": 0.1452286322212855,
+      "learning_rate": 0.00017095628086233891,
+      "loss": 0.7744,
+      "mean_token_accuracy": 0.7665413856506348,
+      "num_tokens": 49206375.0,
+      "step": 14470
+    },
+    {
+      "epoch": 0.14532899755108594,
+      "learning_rate": 0.00017093620779637882,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7723636865615845,
+      "num_tokens": 49239652.0,
+      "step": 14480
+    },
+    {
+      "epoch": 0.14542936288088643,
+      "learning_rate": 0.00017091613473041873,
+      "loss": 0.7851,
+      "mean_token_accuracy": 0.7731632351875305,
+      "num_tokens": 49274454.0,
+      "step": 14490
+    },
+    {
+      "epoch": 0.1455297282106869,
+      "learning_rate": 0.00017089606166445864,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.7742032051086426,
+      "num_tokens": 49308392.0,
+      "step": 14500
+    },
+    {
+      "epoch": 0.14563009354048737,
+      "learning_rate": 0.00017087598859849855,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7644475340843201,
+      "num_tokens": 49342432.0,
+      "step": 14510
+    },
+    {
+      "epoch": 0.14573045887028785,
+      "learning_rate": 0.00017085591553253846,
+      "loss": 0.762,
+      "mean_token_accuracy": 0.7713985919952393,
+      "num_tokens": 49376858.0,
+      "step": 14520
+    },
+    {
+      "epoch": 0.14583082420008833,
+      "learning_rate": 0.00017083584246657836,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7681733310222626,
+      "num_tokens": 49410766.0,
+      "step": 14530
+    },
+    {
+      "epoch": 0.1459311895298888,
+      "learning_rate": 0.00017081576940061825,
+      "loss": 0.7734,
+      "mean_token_accuracy": 0.7684456408023834,
+      "num_tokens": 49444756.0,
+      "step": 14540
+    },
+    {
+      "epoch": 0.14603155485968927,
+      "learning_rate": 0.00017079569633465818,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.7629444479942322,
+      "num_tokens": 49478718.0,
+      "step": 14550
+    },
+    {
+      "epoch": 0.14613192018948976,
+      "learning_rate": 0.00017077562326869806,
+      "loss": 0.7801,
+      "mean_token_accuracy": 0.7721074879169464,
+      "num_tokens": 49512966.0,
+      "step": 14560
+    },
+    {
+      "epoch": 0.1462322855192902,
+      "learning_rate": 0.00017075555020273797,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.7574770987033844,
+      "num_tokens": 49546625.0,
+      "step": 14570
+    },
+    {
+      "epoch": 0.1463326508490907,
+      "learning_rate": 0.00017073547713677788,
+      "loss": 0.7797,
+      "mean_token_accuracy": 0.7676370680332184,
+      "num_tokens": 49580349.0,
+      "step": 14580
+    },
+    {
+      "epoch": 0.14643301617889115,
+      "learning_rate": 0.0001707154040708178,
+      "loss": 0.7727,
+      "mean_token_accuracy": 0.7711577177047729,
+      "num_tokens": 49615199.0,
+      "step": 14590
+    },
+    {
+      "epoch": 0.14653338150869163,
+      "learning_rate": 0.0001706953310048577,
+      "loss": 0.7873,
+      "mean_token_accuracy": 0.7690609931945801,
+      "num_tokens": 49649446.0,
+      "step": 14600
+    },
+    {
+      "epoch": 0.14663374683849212,
+      "learning_rate": 0.0001706752579388976,
+      "loss": 0.8099,
+      "mean_token_accuracy": 0.7618093013763427,
+      "num_tokens": 49684222.0,
+      "step": 14610
+    },
+    {
+      "epoch": 0.14673411216829257,
+      "learning_rate": 0.00017065518487293751,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7617361068725585,
+      "num_tokens": 49718135.0,
+      "step": 14620
+    },
+    {
+      "epoch": 0.14683447749809306,
+      "learning_rate": 0.0001706351118069774,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7758596897125244,
+      "num_tokens": 49751759.0,
+      "step": 14630
+    },
+    {
+      "epoch": 0.14693484282789354,
+      "learning_rate": 0.00017061503874101733,
+      "loss": 0.7403,
+      "mean_token_accuracy": 0.779796940088272,
+      "num_tokens": 49785188.0,
+      "step": 14640
+    },
+    {
+      "epoch": 0.147035208157694,
+      "learning_rate": 0.0001705949656750572,
+      "loss": 0.7291,
+      "mean_token_accuracy": 0.7814127624034881,
+      "num_tokens": 49819304.0,
+      "step": 14650
+    },
+    {
+      "epoch": 0.14713557348749448,
+      "learning_rate": 0.00017057489260909712,
+      "loss": 0.738,
+      "mean_token_accuracy": 0.781837260723114,
+      "num_tokens": 49853347.0,
+      "step": 14660
+    },
+    {
+      "epoch": 0.14723593881729496,
+      "learning_rate": 0.00017055481954313703,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.7659143388271332,
+      "num_tokens": 49887091.0,
+      "step": 14670
+    },
+    {
+      "epoch": 0.14733630414709542,
+      "learning_rate": 0.00017053474647717694,
+      "loss": 0.7821,
+      "mean_token_accuracy": 0.762554156780243,
+      "num_tokens": 49920807.0,
+      "step": 14680
+    },
+    {
+      "epoch": 0.1474366694768959,
+      "learning_rate": 0.00017051467341121682,
+      "loss": 0.7683,
+      "mean_token_accuracy": 0.7699177026748657,
+      "num_tokens": 49954757.0,
+      "step": 14690
+    },
+    {
+      "epoch": 0.1475370348066964,
+      "learning_rate": 0.00017049460034525675,
+      "loss": 0.8059,
+      "mean_token_accuracy": 0.7631576359272003,
+      "num_tokens": 49988778.0,
+      "step": 14700
+    },
+    {
+      "epoch": 0.14763740013649684,
+      "learning_rate": 0.00017047452727929664,
+      "loss": 0.7575,
+      "mean_token_accuracy": 0.7731440722942352,
+      "num_tokens": 50022660.0,
+      "step": 14710
+    },
+    {
+      "epoch": 0.14773776546629733,
+      "learning_rate": 0.00017045445421333654,
+      "loss": 0.7977,
+      "mean_token_accuracy": 0.7649730324745179,
+      "num_tokens": 50057313.0,
+      "step": 14720
+    },
+    {
+      "epoch": 0.14783813079609778,
+      "learning_rate": 0.00017043438114737645,
+      "loss": 0.7949,
+      "mean_token_accuracy": 0.7624725639820099,
+      "num_tokens": 50091866.0,
+      "step": 14730
+    },
+    {
+      "epoch": 0.14793849612589827,
+      "learning_rate": 0.00017041430808141636,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7648936033248901,
+      "num_tokens": 50125834.0,
+      "step": 14740
+    },
+    {
+      "epoch": 0.14803886145569875,
+      "learning_rate": 0.00017039423501545627,
+      "loss": 0.7624,
+      "mean_token_accuracy": 0.7722070574760437,
+      "num_tokens": 50159601.0,
+      "step": 14750
+    },
+    {
+      "epoch": 0.1481392267854992,
+      "learning_rate": 0.00017037416194949618,
+      "loss": 0.7902,
+      "mean_token_accuracy": 0.7632073402404785,
+      "num_tokens": 50193854.0,
+      "step": 14760
+    },
+    {
+      "epoch": 0.1482395921152997,
+      "learning_rate": 0.00017035408888353609,
+      "loss": 0.7849,
+      "mean_token_accuracy": 0.7690035998821259,
+      "num_tokens": 50226519.0,
+      "step": 14770
+    },
+    {
+      "epoch": 0.14833995744510017,
+      "learning_rate": 0.00017033401581757597,
+      "loss": 0.786,
+      "mean_token_accuracy": 0.76683349609375,
+      "num_tokens": 50260778.0,
+      "step": 14780
+    },
+    {
+      "epoch": 0.14844032277490063,
+      "learning_rate": 0.0001703139427516159,
+      "loss": 0.7987,
+      "mean_token_accuracy": 0.7645675659179687,
+      "num_tokens": 50295055.0,
+      "step": 14790
+    },
+    {
+      "epoch": 0.1485406881047011,
+      "learning_rate": 0.00017029386968565578,
+      "loss": 0.7995,
+      "mean_token_accuracy": 0.7587841510772705,
+      "num_tokens": 50329292.0,
+      "step": 14800
+    },
+    {
+      "epoch": 0.1486410534345016,
+      "learning_rate": 0.0001702737966196957,
+      "loss": 0.7646,
+      "mean_token_accuracy": 0.7711651980876922,
+      "num_tokens": 50363063.0,
+      "step": 14810
+    },
+    {
+      "epoch": 0.14874141876430205,
+      "learning_rate": 0.0001702537235537356,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7693265795707702,
+      "num_tokens": 50396926.0,
+      "step": 14820
+    },
+    {
+      "epoch": 0.14884178409410254,
+      "learning_rate": 0.0001702336504877755,
+      "loss": 0.783,
+      "mean_token_accuracy": 0.7658595263957977,
+      "num_tokens": 50431613.0,
+      "step": 14830
+    },
+    {
+      "epoch": 0.14894214942390302,
+      "learning_rate": 0.00017021357742181542,
+      "loss": 0.7675,
+      "mean_token_accuracy": 0.7688614845275878,
+      "num_tokens": 50465709.0,
+      "step": 14840
+    },
+    {
+      "epoch": 0.14904251475370348,
+      "learning_rate": 0.00017019350435585533,
+      "loss": 0.7679,
+      "mean_token_accuracy": 0.7678883016109467,
+      "num_tokens": 50499759.0,
+      "step": 14850
+    },
+    {
+      "epoch": 0.14914288008350396,
+      "learning_rate": 0.00017017343128989523,
+      "loss": 0.7617,
+      "mean_token_accuracy": 0.770219212770462,
+      "num_tokens": 50532901.0,
+      "step": 14860
+    },
+    {
+      "epoch": 0.14924324541330442,
+      "learning_rate": 0.00017015335822393512,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.77974174618721,
+      "num_tokens": 50567043.0,
+      "step": 14870
+    },
+    {
+      "epoch": 0.1493436107431049,
+      "learning_rate": 0.00017013328515797505,
+      "loss": 0.7909,
+      "mean_token_accuracy": 0.7677434682846069,
+      "num_tokens": 50602267.0,
+      "step": 14880
+    },
+    {
+      "epoch": 0.14944397607290538,
+      "learning_rate": 0.00017011321209201493,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7755030393600464,
+      "num_tokens": 50636462.0,
+      "step": 14890
+    },
+    {
+      "epoch": 0.14954434140270584,
+      "learning_rate": 0.00017009313902605484,
+      "loss": 0.7802,
+      "mean_token_accuracy": 0.7705316066741943,
+      "num_tokens": 50670148.0,
+      "step": 14900
+    },
+    {
+      "epoch": 0.14964470673250632,
+      "learning_rate": 0.00017007306596009475,
+      "loss": 0.7737,
+      "mean_token_accuracy": 0.7712871432304382,
+      "num_tokens": 50703722.0,
+      "step": 14910
+    },
+    {
+      "epoch": 0.1497450720623068,
+      "learning_rate": 0.00017005299289413466,
+      "loss": 0.7586,
+      "mean_token_accuracy": 0.7706503093242645,
+      "num_tokens": 50737562.0,
+      "step": 14920
+    },
+    {
+      "epoch": 0.14984543739210726,
+      "learning_rate": 0.00017003291982817454,
+      "loss": 0.7456,
+      "mean_token_accuracy": 0.777435052394867,
+      "num_tokens": 50772332.0,
+      "step": 14930
+    },
+    {
+      "epoch": 0.14994580272190774,
+      "learning_rate": 0.00017001284676221448,
+      "loss": 0.7653,
+      "mean_token_accuracy": 0.7687119841575623,
+      "num_tokens": 50806611.0,
+      "step": 14940
+    },
+    {
+      "epoch": 0.15004616805170823,
+      "learning_rate": 0.00016999277369625438,
+      "loss": 0.7887,
+      "mean_token_accuracy": 0.7700403153896331,
+      "num_tokens": 50840571.0,
+      "step": 14950
+    },
+    {
+      "epoch": 0.15014653338150868,
+      "learning_rate": 0.0001699727006302943,
+      "loss": 0.7526,
+      "mean_token_accuracy": 0.7817915320396424,
+      "num_tokens": 50874811.0,
+      "step": 14960
+    },
+    {
+      "epoch": 0.15024689871130917,
+      "learning_rate": 0.0001699526275643342,
+      "loss": 0.7738,
+      "mean_token_accuracy": 0.768299001455307,
+      "num_tokens": 50908397.0,
+      "step": 14970
+    },
+    {
+      "epoch": 0.15034726404110965,
+      "learning_rate": 0.00016993255449837408,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.7687034428119659,
+      "num_tokens": 50942598.0,
+      "step": 14980
+    },
+    {
+      "epoch": 0.1504476293709101,
+      "learning_rate": 0.00016991248143241402,
+      "loss": 0.7641,
+      "mean_token_accuracy": 0.777588176727295,
+      "num_tokens": 50976321.0,
+      "step": 14990
+    },
+    {
+      "epoch": 0.1505479947007106,
+      "learning_rate": 0.0001698924083664539,
+      "loss": 0.7864,
+      "mean_token_accuracy": 0.7661957383155823,
+      "num_tokens": 51010290.0,
+      "step": 15000
+    },
+    {
+      "epoch": 0.15064836003051105,
+      "learning_rate": 0.0001698723353004938,
+      "loss": 0.8026,
+      "mean_token_accuracy": 0.7674146234989166,
+      "num_tokens": 51044490.0,
+      "step": 15010
+    },
+    {
+      "epoch": 0.15074872536031153,
+      "learning_rate": 0.00016985226223453372,
+      "loss": 0.7927,
+      "mean_token_accuracy": 0.7660162925720215,
+      "num_tokens": 51078398.0,
+      "step": 15020
+    },
+    {
+      "epoch": 0.15084909069011201,
+      "learning_rate": 0.00016983218916857362,
+      "loss": 0.7849,
+      "mean_token_accuracy": 0.7597988963127136,
+      "num_tokens": 51112212.0,
+      "step": 15030
+    },
+    {
+      "epoch": 0.15094945601991247,
+      "learning_rate": 0.0001698121161026135,
+      "loss": 0.7827,
+      "mean_token_accuracy": 0.769049596786499,
+      "num_tokens": 51146235.0,
+      "step": 15040
+    },
+    {
+      "epoch": 0.15104982134971295,
+      "learning_rate": 0.00016979204303665344,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7703625738620759,
+      "num_tokens": 51181217.0,
+      "step": 15050
+    },
+    {
+      "epoch": 0.15115018667951344,
+      "learning_rate": 0.00016977196997069332,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.7812409579753876,
+      "num_tokens": 51215132.0,
+      "step": 15060
+    },
+    {
+      "epoch": 0.1512505520093139,
+      "learning_rate": 0.00016975189690473323,
+      "loss": 0.7898,
+      "mean_token_accuracy": 0.7634902775287629,
+      "num_tokens": 51248088.0,
+      "step": 15070
+    },
+    {
+      "epoch": 0.15135091733911438,
+      "learning_rate": 0.00016973182383877314,
+      "loss": 0.7961,
+      "mean_token_accuracy": 0.765084159374237,
+      "num_tokens": 51282063.0,
+      "step": 15080
+    },
+    {
+      "epoch": 0.15145128266891486,
+      "learning_rate": 0.00016971175077281305,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.7728128671646118,
+      "num_tokens": 51315751.0,
+      "step": 15090
+    },
+    {
+      "epoch": 0.15155164799871532,
+      "learning_rate": 0.00016969167770685296,
+      "loss": 0.7762,
+      "mean_token_accuracy": 0.7731463789939881,
+      "num_tokens": 51350886.0,
+      "step": 15100
+    },
+    {
+      "epoch": 0.1516520133285158,
+      "learning_rate": 0.00016967160464089286,
+      "loss": 0.8285,
+      "mean_token_accuracy": 0.7553871512413025,
+      "num_tokens": 51385104.0,
+      "step": 15110
+    },
+    {
+      "epoch": 0.15175237865831628,
+      "learning_rate": 0.00016965153157493277,
+      "loss": 0.7773,
+      "mean_token_accuracy": 0.7678468465805054,
+      "num_tokens": 51418921.0,
+      "step": 15120
+    },
+    {
+      "epoch": 0.15185274398811674,
+      "learning_rate": 0.00016963145850897265,
+      "loss": 0.7839,
+      "mean_token_accuracy": 0.766031664609909,
+      "num_tokens": 51453508.0,
+      "step": 15130
+    },
+    {
+      "epoch": 0.15195310931791722,
+      "learning_rate": 0.0001696113854430126,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.7636368274688721,
+      "num_tokens": 51487504.0,
+      "step": 15140
+    },
+    {
+      "epoch": 0.15205347464771768,
+      "learning_rate": 0.00016959131237705247,
+      "loss": 0.7573,
+      "mean_token_accuracy": 0.7721414029598236,
+      "num_tokens": 51521862.0,
+      "step": 15150
+    },
+    {
+      "epoch": 0.15215383997751816,
+      "learning_rate": 0.00016957123931109238,
+      "loss": 0.7863,
+      "mean_token_accuracy": 0.7631633579730988,
+      "num_tokens": 51555799.0,
+      "step": 15160
+    },
+    {
+      "epoch": 0.15225420530731865,
+      "learning_rate": 0.0001695511662451323,
+      "loss": 0.7799,
+      "mean_token_accuracy": 0.7730798959732056,
+      "num_tokens": 51589540.0,
+      "step": 15170
+    },
+    {
+      "epoch": 0.1523545706371191,
+      "learning_rate": 0.0001695310931791722,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.770631343126297,
+      "num_tokens": 51622986.0,
+      "step": 15180
+    },
+    {
+      "epoch": 0.15245493596691959,
+      "learning_rate": 0.0001695110201132121,
+      "loss": 0.7931,
+      "mean_token_accuracy": 0.7658929049968719,
+      "num_tokens": 51657488.0,
+      "step": 15190
+    },
+    {
+      "epoch": 0.15255530129672007,
+      "learning_rate": 0.000169490947047252,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.7691733777523041,
+      "num_tokens": 51691318.0,
+      "step": 15200
+    },
+    {
+      "epoch": 0.15265566662652053,
+      "learning_rate": 0.00016947087398129192,
+      "loss": 0.7829,
+      "mean_token_accuracy": 0.7676796436309814,
+      "num_tokens": 51725571.0,
+      "step": 15210
+    },
+    {
+      "epoch": 0.152756031956321,
+      "learning_rate": 0.0001694508009153318,
+      "loss": 0.7736,
+      "mean_token_accuracy": 0.7698560178279876,
+      "num_tokens": 51760144.0,
+      "step": 15220
+    },
+    {
+      "epoch": 0.1528563972861215,
+      "learning_rate": 0.00016943072784937174,
+      "loss": 0.776,
+      "mean_token_accuracy": 0.7702752113342285,
+      "num_tokens": 51794828.0,
+      "step": 15230
+    },
+    {
+      "epoch": 0.15295676261592195,
+      "learning_rate": 0.00016941065478341162,
+      "loss": 0.7994,
+      "mean_token_accuracy": 0.7647412180900574,
+      "num_tokens": 51829991.0,
+      "step": 15240
+    },
+    {
+      "epoch": 0.15305712794572243,
+      "learning_rate": 0.00016939058171745153,
+      "loss": 0.8329,
+      "mean_token_accuracy": 0.7600164353847504,
+      "num_tokens": 51863756.0,
+      "step": 15250
+    },
+    {
+      "epoch": 0.15315749327552292,
+      "learning_rate": 0.00016937050865149144,
+      "loss": 0.7648,
+      "mean_token_accuracy": 0.7694015741348267,
+      "num_tokens": 51897795.0,
+      "step": 15260
+    },
+    {
+      "epoch": 0.15325785860532337,
+      "learning_rate": 0.00016935043558553135,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.7628164887428284,
+      "num_tokens": 51932470.0,
+      "step": 15270
+    },
+    {
+      "epoch": 0.15335822393512386,
+      "learning_rate": 0.00016933036251957123,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.7684712946414948,
+      "num_tokens": 51966083.0,
+      "step": 15280
+    },
+    {
+      "epoch": 0.1534585892649243,
+      "learning_rate": 0.00016931028945361116,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.7693732619285584,
+      "num_tokens": 52000227.0,
+      "step": 15290
+    },
+    {
+      "epoch": 0.1535589545947248,
+      "learning_rate": 0.00016929021638765107,
+      "loss": 0.7447,
+      "mean_token_accuracy": 0.7782359421253204,
+      "num_tokens": 52034381.0,
+      "step": 15300
+    },
+    {
+      "epoch": 0.15365931992452528,
+      "learning_rate": 0.00016927014332169095,
+      "loss": 0.7443,
+      "mean_token_accuracy": 0.7774894177913666,
+      "num_tokens": 52068730.0,
+      "step": 15310
+    },
+    {
+      "epoch": 0.15375968525432573,
+      "learning_rate": 0.0001692500702557309,
+      "loss": 0.7818,
+      "mean_token_accuracy": 0.7613789796829223,
+      "num_tokens": 52102583.0,
+      "step": 15320
+    },
+    {
+      "epoch": 0.15386005058412622,
+      "learning_rate": 0.00016922999718977077,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.7733263611793518,
+      "num_tokens": 52136611.0,
+      "step": 15330
+    },
+    {
+      "epoch": 0.1539604159139267,
+      "learning_rate": 0.00016920992412381068,
+      "loss": 0.8141,
+      "mean_token_accuracy": 0.7589938104152679,
+      "num_tokens": 52170087.0,
+      "step": 15340
+    },
+    {
+      "epoch": 0.15406078124372716,
+      "learning_rate": 0.00016918985105785059,
+      "loss": 0.7711,
+      "mean_token_accuracy": 0.7660266518592834,
+      "num_tokens": 52203630.0,
+      "step": 15350
+    },
+    {
+      "epoch": 0.15416114657352764,
+      "learning_rate": 0.0001691697779918905,
+      "loss": 0.7854,
+      "mean_token_accuracy": 0.767578762769699,
+      "num_tokens": 52237519.0,
+      "step": 15360
+    },
+    {
+      "epoch": 0.15426151190332812,
+      "learning_rate": 0.00016914970492593038,
+      "loss": 0.8112,
+      "mean_token_accuracy": 0.763132518529892,
+      "num_tokens": 52272011.0,
+      "step": 15370
+    },
+    {
+      "epoch": 0.15436187723312858,
+      "learning_rate": 0.0001691296318599703,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.7657889425754547,
+      "num_tokens": 52305881.0,
+      "step": 15380
+    },
+    {
+      "epoch": 0.15446224256292906,
+      "learning_rate": 0.0001691095587940102,
+      "loss": 0.7674,
+      "mean_token_accuracy": 0.7740002870559692,
+      "num_tokens": 52340078.0,
+      "step": 15390
+    },
+    {
+      "epoch": 0.15456260789272955,
+      "learning_rate": 0.0001690894857280501,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.7709534764289856,
+      "num_tokens": 52373189.0,
+      "step": 15400
+    },
+    {
+      "epoch": 0.15466297322253,
+      "learning_rate": 0.00016906941266209,
+      "loss": 0.7778,
+      "mean_token_accuracy": 0.770790982246399,
+      "num_tokens": 52407317.0,
+      "step": 15410
+    },
+    {
+      "epoch": 0.1547633385523305,
+      "learning_rate": 0.00016904933959612992,
+      "loss": 0.7848,
+      "mean_token_accuracy": 0.7632052719593048,
+      "num_tokens": 52441594.0,
+      "step": 15420
+    },
+    {
+      "epoch": 0.15486370388213094,
+      "learning_rate": 0.00016902926653016983,
+      "loss": 0.7897,
+      "mean_token_accuracy": 0.7655413448810577,
+      "num_tokens": 52476312.0,
+      "step": 15430
+    },
+    {
+      "epoch": 0.15496406921193143,
+      "learning_rate": 0.00016900919346420973,
+      "loss": 0.799,
+      "mean_token_accuracy": 0.7574868977069855,
+      "num_tokens": 52509946.0,
+      "step": 15440
+    },
+    {
+      "epoch": 0.1550644345417319,
+      "learning_rate": 0.00016898912039824964,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7688789784908294,
+      "num_tokens": 52543677.0,
+      "step": 15450
+    },
+    {
+      "epoch": 0.15516479987153237,
+      "learning_rate": 0.00016896904733228952,
+      "loss": 0.783,
+      "mean_token_accuracy": 0.7685745835304261,
+      "num_tokens": 52578088.0,
+      "step": 15460
+    },
+    {
+      "epoch": 0.15526516520133285,
+      "learning_rate": 0.00016894897426632946,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7655752122402191,
+      "num_tokens": 52611959.0,
+      "step": 15470
+    },
+    {
+      "epoch": 0.15536553053113333,
+      "learning_rate": 0.00016892890120036934,
+      "loss": 0.7806,
+      "mean_token_accuracy": 0.770868468284607,
+      "num_tokens": 52647202.0,
+      "step": 15480
+    },
+    {
+      "epoch": 0.1554658958609338,
+      "learning_rate": 0.00016890882813440925,
+      "loss": 0.7804,
+      "mean_token_accuracy": 0.7744951903820038,
+      "num_tokens": 52680968.0,
+      "step": 15490
+    },
+    {
+      "epoch": 0.15556626119073427,
+      "learning_rate": 0.00016888875506844916,
+      "loss": 0.754,
+      "mean_token_accuracy": 0.7730810880661011,
+      "num_tokens": 52715308.0,
+      "step": 15500
+    },
+    {
+      "epoch": 0.15566662652053476,
+      "learning_rate": 0.00016886868200248907,
+      "loss": 0.7843,
+      "mean_token_accuracy": 0.7653223693370819,
+      "num_tokens": 52749798.0,
+      "step": 15510
+    },
+    {
+      "epoch": 0.1557669918503352,
+      "learning_rate": 0.00016884860893652897,
+      "loss": 0.7951,
+      "mean_token_accuracy": 0.7614391505718231,
+      "num_tokens": 52782918.0,
+      "step": 15520
+    },
+    {
+      "epoch": 0.1558673571801357,
+      "learning_rate": 0.00016882853587056888,
+      "loss": 0.7665,
+      "mean_token_accuracy": 0.7676415383815766,
+      "num_tokens": 52816974.0,
+      "step": 15530
+    },
+    {
+      "epoch": 0.15596772250993618,
+      "learning_rate": 0.0001688084628046088,
+      "loss": 0.7866,
+      "mean_token_accuracy": 0.7668696820735932,
+      "num_tokens": 52851162.0,
+      "step": 15540
+    },
+    {
+      "epoch": 0.15606808783973664,
+      "learning_rate": 0.0001687883897386487,
+      "loss": 0.7694,
+      "mean_token_accuracy": 0.7691082119941711,
+      "num_tokens": 52884720.0,
+      "step": 15550
+    },
+    {
+      "epoch": 0.15616845316953712,
+      "learning_rate": 0.0001687683166726886,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.7732429683208466,
+      "num_tokens": 52917938.0,
+      "step": 15560
+    },
+    {
+      "epoch": 0.15626881849933758,
+      "learning_rate": 0.0001687482436067285,
+      "loss": 0.792,
+      "mean_token_accuracy": 0.7621237814426423,
+      "num_tokens": 52951666.0,
+      "step": 15570
+    },
+    {
+      "epoch": 0.15636918382913806,
+      "learning_rate": 0.00016872817054076843,
+      "loss": 0.7439,
+      "mean_token_accuracy": 0.7802545189857483,
+      "num_tokens": 52987559.0,
+      "step": 15580
+    },
+    {
+      "epoch": 0.15646954915893854,
+      "learning_rate": 0.0001687080974748083,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.7657202303409576,
+      "num_tokens": 53021331.0,
+      "step": 15590
+    },
+    {
+      "epoch": 0.156569914488739,
+      "learning_rate": 0.00016868802440884822,
+      "loss": 0.7732,
+      "mean_token_accuracy": 0.7699995994567871,
+      "num_tokens": 53055735.0,
+      "step": 15600
+    },
+    {
+      "epoch": 0.15667027981853948,
+      "learning_rate": 0.00016866795134288812,
+      "loss": 0.7566,
+      "mean_token_accuracy": 0.7705514073371887,
+      "num_tokens": 53089781.0,
+      "step": 15610
+    },
+    {
+      "epoch": 0.15677064514833997,
+      "learning_rate": 0.00016864787827692803,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.7708892226219177,
+      "num_tokens": 53123529.0,
+      "step": 15620
+    },
+    {
+      "epoch": 0.15687101047814042,
+      "learning_rate": 0.0001686278052109679,
+      "loss": 0.7513,
+      "mean_token_accuracy": 0.7754028618335724,
+      "num_tokens": 53157103.0,
+      "step": 15630
+    },
+    {
+      "epoch": 0.1569713758079409,
+      "learning_rate": 0.00016860773214500785,
+      "loss": 0.7515,
+      "mean_token_accuracy": 0.7748264670372009,
+      "num_tokens": 53191522.0,
+      "step": 15640
+    },
+    {
+      "epoch": 0.1570717411377414,
+      "learning_rate": 0.00016858765907904776,
+      "loss": 0.7506,
+      "mean_token_accuracy": 0.7732868671417237,
+      "num_tokens": 53224757.0,
+      "step": 15650
+    },
+    {
+      "epoch": 0.15717210646754184,
+      "learning_rate": 0.00016856758601308764,
+      "loss": 0.7498,
+      "mean_token_accuracy": 0.7753817021846772,
+      "num_tokens": 53258888.0,
+      "step": 15660
+    },
+    {
+      "epoch": 0.15727247179734233,
+      "learning_rate": 0.00016854751294712757,
+      "loss": 0.772,
+      "mean_token_accuracy": 0.7627705037593842,
+      "num_tokens": 53292080.0,
+      "step": 15670
+    },
+    {
+      "epoch": 0.1573728371271428,
+      "learning_rate": 0.00016852743988116746,
+      "loss": 0.7709,
+      "mean_token_accuracy": 0.7693831861019135,
+      "num_tokens": 53326293.0,
+      "step": 15680
+    },
+    {
+      "epoch": 0.15747320245694327,
+      "learning_rate": 0.00016850736681520736,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7701425313949585,
+      "num_tokens": 53360079.0,
+      "step": 15690
+    },
+    {
+      "epoch": 0.15757356778674375,
+      "learning_rate": 0.00016848729374924727,
+      "loss": 0.766,
+      "mean_token_accuracy": 0.7696471452713013,
+      "num_tokens": 53394078.0,
+      "step": 15700
+    },
+    {
+      "epoch": 0.15767393311654423,
+      "learning_rate": 0.00016846722068328718,
+      "loss": 0.7701,
+      "mean_token_accuracy": 0.7699676394462586,
+      "num_tokens": 53428344.0,
+      "step": 15710
+    },
+    {
+      "epoch": 0.1577742984463447,
+      "learning_rate": 0.00016844714761732706,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7695291221141816,
+      "num_tokens": 53461912.0,
+      "step": 15720
+    },
+    {
+      "epoch": 0.15787466377614517,
+      "learning_rate": 0.000168427074551367,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.7636789500713348,
+      "num_tokens": 53496403.0,
+      "step": 15730
+    },
+    {
+      "epoch": 0.15797502910594563,
+      "learning_rate": 0.00016840700148540688,
+      "loss": 0.7939,
+      "mean_token_accuracy": 0.7635593891143799,
+      "num_tokens": 53529907.0,
+      "step": 15740
+    },
+    {
+      "epoch": 0.1580753944357461,
+      "learning_rate": 0.0001683869284194468,
+      "loss": 0.7699,
+      "mean_token_accuracy": 0.7722089886665344,
+      "num_tokens": 53563525.0,
+      "step": 15750
+    },
+    {
+      "epoch": 0.1581757597655466,
+      "learning_rate": 0.0001683668553534867,
+      "loss": 0.7598,
+      "mean_token_accuracy": 0.7795711100101471,
+      "num_tokens": 53597740.0,
+      "step": 15760
+    },
+    {
+      "epoch": 0.15827612509534705,
+      "learning_rate": 0.0001683467822875266,
+      "loss": 0.7893,
+      "mean_token_accuracy": 0.7689566075801849,
+      "num_tokens": 53632244.0,
+      "step": 15770
+    },
+    {
+      "epoch": 0.15837649042514754,
+      "learning_rate": 0.0001683267092215665,
+      "loss": 0.7622,
+      "mean_token_accuracy": 0.7705060064792633,
+      "num_tokens": 53666099.0,
+      "step": 15780
+    },
+    {
+      "epoch": 0.15847685575494802,
+      "learning_rate": 0.00016830663615560642,
+      "loss": 0.7891,
+      "mean_token_accuracy": 0.7639210462570191,
+      "num_tokens": 53700311.0,
+      "step": 15790
+    },
+    {
+      "epoch": 0.15857722108474848,
+      "learning_rate": 0.00016828656308964633,
+      "loss": 0.7711,
+      "mean_token_accuracy": 0.7662554621696472,
+      "num_tokens": 53734915.0,
+      "step": 15800
+    },
+    {
+      "epoch": 0.15867758641454896,
+      "learning_rate": 0.0001682664900236862,
+      "loss": 0.7607,
+      "mean_token_accuracy": 0.7745907962322235,
+      "num_tokens": 53769792.0,
+      "step": 15810
+    },
+    {
+      "epoch": 0.15877795174434944,
+      "learning_rate": 0.00016824641695772615,
+      "loss": 0.7489,
+      "mean_token_accuracy": 0.7763136684894562,
+      "num_tokens": 53803658.0,
+      "step": 15820
+    },
+    {
+      "epoch": 0.1588783170741499,
+      "learning_rate": 0.00016822634389176603,
+      "loss": 0.7964,
+      "mean_token_accuracy": 0.7618490397930145,
+      "num_tokens": 53837440.0,
+      "step": 15830
+    },
+    {
+      "epoch": 0.15897868240395038,
+      "learning_rate": 0.00016820627082580594,
+      "loss": 0.7768,
+      "mean_token_accuracy": 0.7668090462684631,
+      "num_tokens": 53871063.0,
+      "step": 15840
+    },
+    {
+      "epoch": 0.15907904773375087,
+      "learning_rate": 0.00016818619775984584,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7736083090305328,
+      "num_tokens": 53905460.0,
+      "step": 15850
+    },
+    {
+      "epoch": 0.15917941306355132,
+      "learning_rate": 0.00016816612469388575,
+      "loss": 0.7549,
+      "mean_token_accuracy": 0.7736191630363465,
+      "num_tokens": 53939217.0,
+      "step": 15860
+    },
+    {
+      "epoch": 0.1592797783933518,
+      "learning_rate": 0.00016814605162792566,
+      "loss": 0.7334,
+      "mean_token_accuracy": 0.7773832976818085,
+      "num_tokens": 53972404.0,
+      "step": 15870
+    },
+    {
+      "epoch": 0.15938014372315226,
+      "learning_rate": 0.00016812597856196557,
+      "loss": 0.8131,
+      "mean_token_accuracy": 0.7607673048973084,
+      "num_tokens": 54007085.0,
+      "step": 15880
+    },
+    {
+      "epoch": 0.15948050905295275,
+      "learning_rate": 0.00016810590549600548,
+      "loss": 0.7581,
+      "mean_token_accuracy": 0.7750826418399811,
+      "num_tokens": 54040901.0,
+      "step": 15890
+    },
+    {
+      "epoch": 0.15958087438275323,
+      "learning_rate": 0.00016808583243004536,
+      "loss": 0.7775,
+      "mean_token_accuracy": 0.7645307302474975,
+      "num_tokens": 54075068.0,
+      "step": 15900
+    },
+    {
+      "epoch": 0.15968123971255369,
+      "learning_rate": 0.0001680657593640853,
+      "loss": 0.7823,
+      "mean_token_accuracy": 0.7664214968681335,
+      "num_tokens": 54108768.0,
+      "step": 15910
+    },
+    {
+      "epoch": 0.15978160504235417,
+      "learning_rate": 0.00016804568629812518,
+      "loss": 0.7731,
+      "mean_token_accuracy": 0.7718466520309448,
+      "num_tokens": 54142779.0,
+      "step": 15920
+    },
+    {
+      "epoch": 0.15988197037215465,
+      "learning_rate": 0.00016802561323216508,
+      "loss": 0.7995,
+      "mean_token_accuracy": 0.7627302944660187,
+      "num_tokens": 54176709.0,
+      "step": 15930
+    },
+    {
+      "epoch": 0.1599823357019551,
+      "learning_rate": 0.000168005540166205,
+      "loss": 0.7952,
+      "mean_token_accuracy": 0.7675760447978973,
+      "num_tokens": 54211077.0,
+      "step": 15940
+    },
+    {
+      "epoch": 0.1600827010317556,
+      "learning_rate": 0.0001679854671002449,
+      "loss": 0.7905,
+      "mean_token_accuracy": 0.7670733034610748,
+      "num_tokens": 54245642.0,
+      "step": 15950
+    },
+    {
+      "epoch": 0.16018306636155608,
+      "learning_rate": 0.00016796539403428478,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.773389321565628,
+      "num_tokens": 54280193.0,
+      "step": 15960
+    },
+    {
+      "epoch": 0.16028343169135653,
+      "learning_rate": 0.00016794532096832472,
+      "loss": 0.7797,
+      "mean_token_accuracy": 0.7719764292240143,
+      "num_tokens": 54314135.0,
+      "step": 15970
+    },
+    {
+      "epoch": 0.16038379702115702,
+      "learning_rate": 0.0001679252479023646,
+      "loss": 0.7562,
+      "mean_token_accuracy": 0.780254477262497,
+      "num_tokens": 54347415.0,
+      "step": 15980
+    },
+    {
+      "epoch": 0.1604841623509575,
+      "learning_rate": 0.0001679051748364045,
+      "loss": 0.7524,
+      "mean_token_accuracy": 0.7712141573429108,
+      "num_tokens": 54382058.0,
+      "step": 15990
+    },
+    {
+      "epoch": 0.16058452768075795,
+      "learning_rate": 0.00016788510177044444,
+      "loss": 0.7944,
+      "mean_token_accuracy": 0.7686404109001159,
+      "num_tokens": 54416790.0,
+      "step": 16000
+    },
+    {
+      "epoch": 0.16068489301055844,
+      "learning_rate": 0.00016786502870448433,
+      "loss": 0.7819,
+      "mean_token_accuracy": 0.7646767497062683,
+      "num_tokens": 54451156.0,
+      "step": 16010
+    },
+    {
+      "epoch": 0.1607852583403589,
+      "learning_rate": 0.00016784495563852423,
+      "loss": 0.7854,
+      "mean_token_accuracy": 0.7703452050685883,
+      "num_tokens": 54485109.0,
+      "step": 16020
+    },
+    {
+      "epoch": 0.16088562367015938,
+      "learning_rate": 0.00016782488257256414,
+      "loss": 0.7607,
+      "mean_token_accuracy": 0.7728536605834961,
+      "num_tokens": 54519121.0,
+      "step": 16030
+    },
+    {
+      "epoch": 0.16098598899995986,
+      "learning_rate": 0.00016780480950660405,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.7680536806583405,
+      "num_tokens": 54552312.0,
+      "step": 16040
+    },
+    {
+      "epoch": 0.16108635432976032,
+      "learning_rate": 0.00016778473644064393,
+      "loss": 0.7939,
+      "mean_token_accuracy": 0.763797128200531,
+      "num_tokens": 54585602.0,
+      "step": 16050
+    },
+    {
+      "epoch": 0.1611867196595608,
+      "learning_rate": 0.00016776466337468387,
+      "loss": 0.7831,
+      "mean_token_accuracy": 0.7706405282020569,
+      "num_tokens": 54620257.0,
+      "step": 16060
+    },
+    {
+      "epoch": 0.16128708498936128,
+      "learning_rate": 0.00016774459030872375,
+      "loss": 0.8012,
+      "mean_token_accuracy": 0.7677317500114441,
+      "num_tokens": 54654400.0,
+      "step": 16070
+    },
+    {
+      "epoch": 0.16138745031916174,
+      "learning_rate": 0.00016772451724276366,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7740169882774353,
+      "num_tokens": 54689429.0,
+      "step": 16080
+    },
+    {
+      "epoch": 0.16148781564896222,
+      "learning_rate": 0.00016770444417680357,
+      "loss": 0.762,
+      "mean_token_accuracy": 0.7705418944358826,
+      "num_tokens": 54723355.0,
+      "step": 16090
+    },
+    {
+      "epoch": 0.1615881809787627,
+      "learning_rate": 0.00016768437111084347,
+      "loss": 0.7956,
+      "mean_token_accuracy": 0.7653626322746276,
+      "num_tokens": 54757516.0,
+      "step": 16100
+    },
+    {
+      "epoch": 0.16168854630856316,
+      "learning_rate": 0.00016766429804488338,
+      "loss": 0.7442,
+      "mean_token_accuracy": 0.7787913143634796,
+      "num_tokens": 54791568.0,
+      "step": 16110
+    },
+    {
+      "epoch": 0.16178891163836365,
+      "learning_rate": 0.0001676442249789233,
+      "loss": 0.7666,
+      "mean_token_accuracy": 0.7725750982761384,
+      "num_tokens": 54825658.0,
+      "step": 16120
+    },
+    {
+      "epoch": 0.16188927696816413,
+      "learning_rate": 0.0001676241519129632,
+      "loss": 0.7641,
+      "mean_token_accuracy": 0.7700630605220795,
+      "num_tokens": 54859192.0,
+      "step": 16130
+    },
+    {
+      "epoch": 0.1619896422979646,
+      "learning_rate": 0.0001676040788470031,
+      "loss": 0.7685,
+      "mean_token_accuracy": 0.7750592708587647,
+      "num_tokens": 54893175.0,
+      "step": 16140
+    },
+    {
+      "epoch": 0.16209000762776507,
+      "learning_rate": 0.00016758400578104302,
+      "loss": 0.7562,
+      "mean_token_accuracy": 0.7745824337005616,
+      "num_tokens": 54927452.0,
+      "step": 16150
+    },
+    {
+      "epoch": 0.16219037295756553,
+      "learning_rate": 0.0001675639327150829,
+      "loss": 0.7662,
+      "mean_token_accuracy": 0.7729141354560852,
+      "num_tokens": 54961756.0,
+      "step": 16160
+    },
+    {
+      "epoch": 0.162290738287366,
+      "learning_rate": 0.00016754385964912283,
+      "loss": 0.7332,
+      "mean_token_accuracy": 0.7856162011623382,
+      "num_tokens": 54995486.0,
+      "step": 16170
+    },
+    {
+      "epoch": 0.1623911036171665,
+      "learning_rate": 0.00016752378658316271,
+      "loss": 0.7781,
+      "mean_token_accuracy": 0.7697393536567688,
+      "num_tokens": 55029693.0,
+      "step": 16180
+    },
+    {
+      "epoch": 0.16249146894696695,
+      "learning_rate": 0.00016750371351720262,
+      "loss": 0.7915,
+      "mean_token_accuracy": 0.7663042962551116,
+      "num_tokens": 55063475.0,
+      "step": 16190
+    },
+    {
+      "epoch": 0.16259183427676743,
+      "learning_rate": 0.00016748364045124253,
+      "loss": 0.7502,
+      "mean_token_accuracy": 0.7753411650657653,
+      "num_tokens": 55097478.0,
+      "step": 16200
+    },
+    {
+      "epoch": 0.16269219960656792,
+      "learning_rate": 0.00016746356738528244,
+      "loss": 0.7645,
+      "mean_token_accuracy": 0.7713445484638214,
+      "num_tokens": 55131686.0,
+      "step": 16210
+    },
+    {
+      "epoch": 0.16279256493636837,
+      "learning_rate": 0.00016744349431932235,
+      "loss": 0.8009,
+      "mean_token_accuracy": 0.7626453280448914,
+      "num_tokens": 55165462.0,
+      "step": 16220
+    },
+    {
+      "epoch": 0.16289293026616886,
+      "learning_rate": 0.00016742342125336226,
+      "loss": 0.7992,
+      "mean_token_accuracy": 0.7661191344261169,
+      "num_tokens": 55199348.0,
+      "step": 16230
+    },
+    {
+      "epoch": 0.16299329559596934,
+      "learning_rate": 0.00016740334818740217,
+      "loss": 0.8257,
+      "mean_token_accuracy": 0.7612239360809326,
+      "num_tokens": 55233696.0,
+      "step": 16240
+    },
+    {
+      "epoch": 0.1630936609257698,
+      "learning_rate": 0.00016738327512144205,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7717586278915405,
+      "num_tokens": 55267361.0,
+      "step": 16250
+    },
+    {
+      "epoch": 0.16319402625557028,
+      "learning_rate": 0.00016736320205548198,
+      "loss": 0.7901,
+      "mean_token_accuracy": 0.7668171525001526,
+      "num_tokens": 55301500.0,
+      "step": 16260
+    },
+    {
+      "epoch": 0.16329439158537076,
+      "learning_rate": 0.00016734312898952186,
+      "loss": 0.7327,
+      "mean_token_accuracy": 0.7806723415851593,
+      "num_tokens": 55336458.0,
+      "step": 16270
+    },
+    {
+      "epoch": 0.16339475691517122,
+      "learning_rate": 0.00016732305592356177,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.773836076259613,
+      "num_tokens": 55371171.0,
+      "step": 16280
+    },
+    {
+      "epoch": 0.1634951222449717,
+      "learning_rate": 0.00016730298285760168,
+      "loss": 0.7977,
+      "mean_token_accuracy": 0.763254064321518,
+      "num_tokens": 55405118.0,
+      "step": 16290
+    },
+    {
+      "epoch": 0.16359548757477216,
+      "learning_rate": 0.0001672829097916416,
+      "loss": 0.7896,
+      "mean_token_accuracy": 0.7666790723800659,
+      "num_tokens": 55439087.0,
+      "step": 16300
+    },
+    {
+      "epoch": 0.16369585290457264,
+      "learning_rate": 0.00016726283672568147,
+      "loss": 0.8011,
+      "mean_token_accuracy": 0.765297120809555,
+      "num_tokens": 55472738.0,
+      "step": 16310
+    },
+    {
+      "epoch": 0.16379621823437313,
+      "learning_rate": 0.0001672427636597214,
+      "loss": 0.768,
+      "mean_token_accuracy": 0.7692670464515686,
+      "num_tokens": 55507007.0,
+      "step": 16320
+    },
+    {
+      "epoch": 0.16389658356417358,
+      "learning_rate": 0.0001672226905937613,
+      "loss": 0.7608,
+      "mean_token_accuracy": 0.772177517414093,
+      "num_tokens": 55541368.0,
+      "step": 16330
+    },
+    {
+      "epoch": 0.16399694889397406,
+      "learning_rate": 0.0001672026175278012,
+      "loss": 0.7814,
+      "mean_token_accuracy": 0.7707577764987945,
+      "num_tokens": 55576173.0,
+      "step": 16340
+    },
+    {
+      "epoch": 0.16409731422377455,
+      "learning_rate": 0.00016718254446184113,
+      "loss": 0.7934,
+      "mean_token_accuracy": 0.762543672323227,
+      "num_tokens": 55609079.0,
+      "step": 16350
+    },
+    {
+      "epoch": 0.164197679553575,
+      "learning_rate": 0.000167162471395881,
+      "loss": 0.7954,
+      "mean_token_accuracy": 0.7651306629180908,
+      "num_tokens": 55642519.0,
+      "step": 16360
+    },
+    {
+      "epoch": 0.1642980448833755,
+      "learning_rate": 0.00016714239832992092,
+      "loss": 0.7601,
+      "mean_token_accuracy": 0.7750393986701966,
+      "num_tokens": 55676882.0,
+      "step": 16370
+    },
+    {
+      "epoch": 0.16439841021317597,
+      "learning_rate": 0.00016712232526396083,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.7707339823246002,
+      "num_tokens": 55711314.0,
+      "step": 16380
+    },
+    {
+      "epoch": 0.16449877554297643,
+      "learning_rate": 0.00016710225219800074,
+      "loss": 0.7589,
+      "mean_token_accuracy": 0.7756678760051727,
+      "num_tokens": 55745279.0,
+      "step": 16390
+    },
+    {
+      "epoch": 0.1645991408727769,
+      "learning_rate": 0.00016708217913204062,
+      "loss": 0.7816,
+      "mean_token_accuracy": 0.766136747598648,
+      "num_tokens": 55779482.0,
+      "step": 16400
+    },
+    {
+      "epoch": 0.1646995062025774,
+      "learning_rate": 0.00016706210606608055,
+      "loss": 0.7499,
+      "mean_token_accuracy": 0.7740479826927185,
+      "num_tokens": 55814137.0,
+      "step": 16410
+    },
+    {
+      "epoch": 0.16479987153237785,
+      "learning_rate": 0.00016704203300012044,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.7630363523960113,
+      "num_tokens": 55848446.0,
+      "step": 16420
+    },
+    {
+      "epoch": 0.16490023686217833,
+      "learning_rate": 0.00016702195993416034,
+      "loss": 0.7561,
+      "mean_token_accuracy": 0.7752853691577911,
+      "num_tokens": 55882891.0,
+      "step": 16430
+    },
+    {
+      "epoch": 0.1650006021919788,
+      "learning_rate": 0.00016700188686820025,
+      "loss": 0.7532,
+      "mean_token_accuracy": 0.770255833864212,
+      "num_tokens": 55916716.0,
+      "step": 16440
+    },
+    {
+      "epoch": 0.16510096752177927,
+      "learning_rate": 0.00016698181380224016,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7719760477542877,
+      "num_tokens": 55950412.0,
+      "step": 16450
+    },
+    {
+      "epoch": 0.16520133285157976,
+      "learning_rate": 0.00016696174073628007,
+      "loss": 0.7513,
+      "mean_token_accuracy": 0.7779366493225097,
+      "num_tokens": 55983776.0,
+      "step": 16460
+    },
+    {
+      "epoch": 0.1653016981813802,
+      "learning_rate": 0.00016694166767031998,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.7739466309547425,
+      "num_tokens": 56018200.0,
+      "step": 16470
+    },
+    {
+      "epoch": 0.1654020635111807,
+      "learning_rate": 0.00016692159460435989,
+      "loss": 0.7446,
+      "mean_token_accuracy": 0.7743953824043274,
+      "num_tokens": 56051751.0,
+      "step": 16480
+    },
+    {
+      "epoch": 0.16550242884098118,
+      "learning_rate": 0.00016690152153839977,
+      "loss": 0.7719,
+      "mean_token_accuracy": 0.7692020237445831,
+      "num_tokens": 56086204.0,
+      "step": 16490
+    },
+    {
+      "epoch": 0.16560279417078164,
+      "learning_rate": 0.0001668814484724397,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.76861452460289,
+      "num_tokens": 56120293.0,
+      "step": 16500
+    },
+    {
+      "epoch": 0.16570315950058212,
+      "learning_rate": 0.00016686137540647958,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.771511435508728,
+      "num_tokens": 56153683.0,
+      "step": 16510
+    },
+    {
+      "epoch": 0.1658035248303826,
+      "learning_rate": 0.0001668413023405195,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.7763586521148682,
+      "num_tokens": 56187960.0,
+      "step": 16520
+    },
+    {
+      "epoch": 0.16590389016018306,
+      "learning_rate": 0.0001668212292745594,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.762101536989212,
+      "num_tokens": 56222385.0,
+      "step": 16530
+    },
+    {
+      "epoch": 0.16600425548998354,
+      "learning_rate": 0.0001668011562085993,
+      "loss": 0.7642,
+      "mean_token_accuracy": 0.7768764853477478,
+      "num_tokens": 56256575.0,
+      "step": 16540
+    },
+    {
+      "epoch": 0.16610462081978403,
+      "learning_rate": 0.00016678108314263922,
+      "loss": 0.8119,
+      "mean_token_accuracy": 0.7595563709735871,
+      "num_tokens": 56291016.0,
+      "step": 16550
+    },
+    {
+      "epoch": 0.16620498614958448,
+      "learning_rate": 0.00016676101007667913,
+      "loss": 0.7523,
+      "mean_token_accuracy": 0.775392371416092,
+      "num_tokens": 56324925.0,
+      "step": 16560
+    },
+    {
+      "epoch": 0.16630535147938497,
+      "learning_rate": 0.00016674093701071904,
+      "loss": 0.7889,
+      "mean_token_accuracy": 0.7664380073547363,
+      "num_tokens": 56358698.0,
+      "step": 16570
+    },
+    {
+      "epoch": 0.16640571680918542,
+      "learning_rate": 0.00016672086394475892,
+      "loss": 0.7372,
+      "mean_token_accuracy": 0.7752080440521241,
+      "num_tokens": 56392057.0,
+      "step": 16580
+    },
+    {
+      "epoch": 0.1665060821389859,
+      "learning_rate": 0.00016670079087879885,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7740992903709412,
+      "num_tokens": 56425622.0,
+      "step": 16590
+    },
+    {
+      "epoch": 0.1666064474687864,
+      "learning_rate": 0.00016668071781283873,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7681378543376922,
+      "num_tokens": 56459099.0,
+      "step": 16600
+    },
+    {
+      "epoch": 0.16670681279858685,
+      "learning_rate": 0.00016666064474687864,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.7682817459106446,
+      "num_tokens": 56493058.0,
+      "step": 16610
+    },
+    {
+      "epoch": 0.16680717812838733,
+      "learning_rate": 0.00016664057168091855,
+      "loss": 0.7631,
+      "mean_token_accuracy": 0.7736068487167358,
+      "num_tokens": 56527665.0,
+      "step": 16620
+    },
+    {
+      "epoch": 0.1669075434581878,
+      "learning_rate": 0.00016662049861495846,
+      "loss": 0.7964,
+      "mean_token_accuracy": 0.7661099135875702,
+      "num_tokens": 56561558.0,
+      "step": 16630
+    },
+    {
+      "epoch": 0.16700790878798827,
+      "learning_rate": 0.00016660042554899834,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.7665910840034484,
+      "num_tokens": 56595946.0,
+      "step": 16640
+    },
+    {
+      "epoch": 0.16710827411778875,
+      "learning_rate": 0.00016658035248303828,
+      "loss": 0.7871,
+      "mean_token_accuracy": 0.7688241362571716,
+      "num_tokens": 56630198.0,
+      "step": 16650
+    },
+    {
+      "epoch": 0.16720863944758924,
+      "learning_rate": 0.00016656027941707816,
+      "loss": 0.7748,
+      "mean_token_accuracy": 0.773946350812912,
+      "num_tokens": 56664493.0,
+      "step": 16660
+    },
+    {
+      "epoch": 0.1673090047773897,
+      "learning_rate": 0.0001665402063511181,
+      "loss": 0.7656,
+      "mean_token_accuracy": 0.7693340480327606,
+      "num_tokens": 56698753.0,
+      "step": 16670
+    },
+    {
+      "epoch": 0.16740937010719018,
+      "learning_rate": 0.00016652013328515797,
+      "loss": 0.7957,
+      "mean_token_accuracy": 0.7672327339649201,
+      "num_tokens": 56732436.0,
+      "step": 16680
+    },
+    {
+      "epoch": 0.16750973543699066,
+      "learning_rate": 0.00016650006021919788,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7741209328174591,
+      "num_tokens": 56767592.0,
+      "step": 16690
+    },
+    {
+      "epoch": 0.16761010076679111,
+      "learning_rate": 0.00016647998715323782,
+      "loss": 0.7749,
+      "mean_token_accuracy": 0.7696184456348419,
+      "num_tokens": 56801909.0,
+      "step": 16700
+    },
+    {
+      "epoch": 0.1677104660965916,
+      "learning_rate": 0.0001664599140872777,
+      "loss": 0.8106,
+      "mean_token_accuracy": 0.763665896654129,
+      "num_tokens": 56836591.0,
+      "step": 16710
+    },
+    {
+      "epoch": 0.16781083142639205,
+      "learning_rate": 0.0001664398410213176,
+      "loss": 0.7606,
+      "mean_token_accuracy": 0.7729604065418243,
+      "num_tokens": 56870841.0,
+      "step": 16720
+    },
+    {
+      "epoch": 0.16791119675619254,
+      "learning_rate": 0.00016641976795535752,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7698002696037293,
+      "num_tokens": 56904225.0,
+      "step": 16730
+    },
+    {
+      "epoch": 0.16801156208599302,
+      "learning_rate": 0.00016639969488939742,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.7710990846157074,
+      "num_tokens": 56938018.0,
+      "step": 16740
+    },
+    {
+      "epoch": 0.16811192741579348,
+      "learning_rate": 0.0001663796218234373,
+      "loss": 0.7582,
+      "mean_token_accuracy": 0.775606769323349,
+      "num_tokens": 56972455.0,
+      "step": 16750
+    },
+    {
+      "epoch": 0.16821229274559396,
+      "learning_rate": 0.00016635954875747724,
+      "loss": 0.7755,
+      "mean_token_accuracy": 0.7679441034793854,
+      "num_tokens": 57006665.0,
+      "step": 16760
+    },
+    {
+      "epoch": 0.16831265807539444,
+      "learning_rate": 0.00016633947569151712,
+      "loss": 0.7999,
+      "mean_token_accuracy": 0.7675316214561463,
+      "num_tokens": 57041029.0,
+      "step": 16770
+    },
+    {
+      "epoch": 0.1684130234051949,
+      "learning_rate": 0.00016631940262555703,
+      "loss": 0.7629,
+      "mean_token_accuracy": 0.7740450620651245,
+      "num_tokens": 57074771.0,
+      "step": 16780
+    },
+    {
+      "epoch": 0.16851338873499538,
+      "learning_rate": 0.00016629932955959694,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7727279365062714,
+      "num_tokens": 57108544.0,
+      "step": 16790
+    },
+    {
+      "epoch": 0.16861375406479587,
+      "learning_rate": 0.00016627925649363685,
+      "loss": 0.7631,
+      "mean_token_accuracy": 0.7695836067199707,
+      "num_tokens": 57143451.0,
+      "step": 16800
+    },
+    {
+      "epoch": 0.16871411939459632,
+      "learning_rate": 0.00016625918342767676,
+      "loss": 0.7622,
+      "mean_token_accuracy": 0.7722588241100311,
+      "num_tokens": 57177527.0,
+      "step": 16810
+    },
+    {
+      "epoch": 0.1688144847243968,
+      "learning_rate": 0.00016623911036171666,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7699831604957581,
+      "num_tokens": 57211529.0,
+      "step": 16820
+    },
+    {
+      "epoch": 0.1689148500541973,
+      "learning_rate": 0.00016621903729575657,
+      "loss": 0.7404,
+      "mean_token_accuracy": 0.7806845128536224,
+      "num_tokens": 57245974.0,
+      "step": 16830
+    },
+    {
+      "epoch": 0.16901521538399775,
+      "learning_rate": 0.00016619896422979645,
+      "loss": 0.7864,
+      "mean_token_accuracy": 0.7693890869617462,
+      "num_tokens": 57280503.0,
+      "step": 16840
+    },
+    {
+      "epoch": 0.16911558071379823,
+      "learning_rate": 0.0001661788911638364,
+      "loss": 0.7749,
+      "mean_token_accuracy": 0.7683271586894989,
+      "num_tokens": 57314050.0,
+      "step": 16850
+    },
+    {
+      "epoch": 0.1692159460435987,
+      "learning_rate": 0.00016615881809787627,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.7672585129737854,
+      "num_tokens": 57348241.0,
+      "step": 16860
+    },
+    {
+      "epoch": 0.16931631137339917,
+      "learning_rate": 0.00016613874503191618,
+      "loss": 0.7563,
+      "mean_token_accuracy": 0.7712204933166504,
+      "num_tokens": 57382864.0,
+      "step": 16870
+    },
+    {
+      "epoch": 0.16941667670319965,
+      "learning_rate": 0.0001661186719659561,
+      "loss": 0.7912,
+      "mean_token_accuracy": 0.7650717854499817,
+      "num_tokens": 57416511.0,
+      "step": 16880
+    },
+    {
+      "epoch": 0.1695170420330001,
+      "learning_rate": 0.000166098598899996,
+      "loss": 0.7828,
+      "mean_token_accuracy": 0.7699388206005097,
+      "num_tokens": 57451135.0,
+      "step": 16890
+    },
+    {
+      "epoch": 0.1696174073628006,
+      "learning_rate": 0.0001660785258340359,
+      "loss": 0.7793,
+      "mean_token_accuracy": 0.7671032309532165,
+      "num_tokens": 57485805.0,
+      "step": 16900
+    },
+    {
+      "epoch": 0.16971777269260108,
+      "learning_rate": 0.0001660584527680758,
+      "loss": 0.7937,
+      "mean_token_accuracy": 0.7623134732246399,
+      "num_tokens": 57519798.0,
+      "step": 16910
+    },
+    {
+      "epoch": 0.16981813802240153,
+      "learning_rate": 0.00016603837970211572,
+      "loss": 0.7888,
+      "mean_token_accuracy": 0.7673300087451935,
+      "num_tokens": 57553778.0,
+      "step": 16920
+    },
+    {
+      "epoch": 0.16991850335220202,
+      "learning_rate": 0.0001660183066361556,
+      "loss": 0.7725,
+      "mean_token_accuracy": 0.7663509368896484,
+      "num_tokens": 57588052.0,
+      "step": 16930
+    },
+    {
+      "epoch": 0.1700188686820025,
+      "learning_rate": 0.00016599823357019554,
+      "loss": 0.7636,
+      "mean_token_accuracy": 0.7714050889015198,
+      "num_tokens": 57621687.0,
+      "step": 16940
+    },
+    {
+      "epoch": 0.17011923401180296,
+      "learning_rate": 0.00016597816050423542,
+      "loss": 0.7873,
+      "mean_token_accuracy": 0.7676332652568817,
+      "num_tokens": 57656223.0,
+      "step": 16950
+    },
+    {
+      "epoch": 0.17021959934160344,
+      "learning_rate": 0.00016595808743827533,
+      "loss": 0.7648,
+      "mean_token_accuracy": 0.7720965504646301,
+      "num_tokens": 57690924.0,
+      "step": 16960
+    },
+    {
+      "epoch": 0.17031996467140392,
+      "learning_rate": 0.00016593801437231524,
+      "loss": 0.7295,
+      "mean_token_accuracy": 0.7817053020000457,
+      "num_tokens": 57724855.0,
+      "step": 16970
+    },
+    {
+      "epoch": 0.17042033000120438,
+      "learning_rate": 0.00016591794130635515,
+      "loss": 0.8057,
+      "mean_token_accuracy": 0.7620230674743652,
+      "num_tokens": 57759081.0,
+      "step": 16980
+    },
+    {
+      "epoch": 0.17052069533100486,
+      "learning_rate": 0.00016589786824039503,
+      "loss": 0.7748,
+      "mean_token_accuracy": 0.7702114403247833,
+      "num_tokens": 57794046.0,
+      "step": 16990
+    },
+    {
+      "epoch": 0.17062106066080532,
+      "learning_rate": 0.00016587779517443496,
+      "loss": 0.7984,
+      "mean_token_accuracy": 0.7697321712970734,
+      "num_tokens": 57827455.0,
+      "step": 17000
+    },
+    {
+      "epoch": 0.1707214259906058,
+      "learning_rate": 0.00016585772210847484,
+      "loss": 0.7747,
+      "mean_token_accuracy": 0.7693339705467224,
+      "num_tokens": 57862864.0,
+      "step": 17010
+    },
+    {
+      "epoch": 0.17082179132040629,
+      "learning_rate": 0.00016583764904251475,
+      "loss": 0.7644,
+      "mean_token_accuracy": 0.772398829460144,
+      "num_tokens": 57896640.0,
+      "step": 17020
+    },
+    {
+      "epoch": 0.17092215665020674,
+      "learning_rate": 0.00016581757597655466,
+      "loss": 0.7361,
+      "mean_token_accuracy": 0.7789756894111634,
+      "num_tokens": 57930726.0,
+      "step": 17030
+    },
+    {
+      "epoch": 0.17102252198000722,
+      "learning_rate": 0.00016579750291059457,
+      "loss": 0.7665,
+      "mean_token_accuracy": 0.7718335926532746,
+      "num_tokens": 57963979.0,
+      "step": 17040
+    },
+    {
+      "epoch": 0.1711228873098077,
+      "learning_rate": 0.00016577742984463448,
+      "loss": 0.7406,
+      "mean_token_accuracy": 0.7737122237682342,
+      "num_tokens": 57997739.0,
+      "step": 17050
+    },
+    {
+      "epoch": 0.17122325263960816,
+      "learning_rate": 0.00016575735677867439,
+      "loss": 0.8044,
+      "mean_token_accuracy": 0.7606332123279571,
+      "num_tokens": 58030893.0,
+      "step": 17060
+    },
+    {
+      "epoch": 0.17132361796940865,
+      "learning_rate": 0.0001657372837127143,
+      "loss": 0.7719,
+      "mean_token_accuracy": 0.7710410356521606,
+      "num_tokens": 58065200.0,
+      "step": 17070
+    },
+    {
+      "epoch": 0.17142398329920913,
+      "learning_rate": 0.00016571721064675418,
+      "loss": 0.8052,
+      "mean_token_accuracy": 0.7625146746635437,
+      "num_tokens": 58098788.0,
+      "step": 17080
+    },
+    {
+      "epoch": 0.1715243486290096,
+      "learning_rate": 0.0001656971375807941,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7621822893619538,
+      "num_tokens": 58132803.0,
+      "step": 17090
+    },
+    {
+      "epoch": 0.17162471395881007,
+      "learning_rate": 0.000165677064514834,
+      "loss": 0.7644,
+      "mean_token_accuracy": 0.7708954095840455,
+      "num_tokens": 58166826.0,
+      "step": 17100
+    },
+    {
+      "epoch": 0.17172507928861055,
+      "learning_rate": 0.0001656569914488739,
+      "loss": 0.7519,
+      "mean_token_accuracy": 0.7677281260490417,
+      "num_tokens": 58201048.0,
+      "step": 17110
+    },
+    {
+      "epoch": 0.171825444618411,
+      "learning_rate": 0.0001656369183829138,
+      "loss": 0.7655,
+      "mean_token_accuracy": 0.7702815532684326,
+      "num_tokens": 58234789.0,
+      "step": 17120
+    },
+    {
+      "epoch": 0.1719258099482115,
+      "learning_rate": 0.00016561684531695372,
+      "loss": 0.7789,
+      "mean_token_accuracy": 0.7679742932319641,
+      "num_tokens": 58268627.0,
+      "step": 17130
+    },
+    {
+      "epoch": 0.17202617527801195,
+      "learning_rate": 0.00016559677225099363,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.7711734771728516,
+      "num_tokens": 58301797.0,
+      "step": 17140
+    },
+    {
+      "epoch": 0.17212654060781243,
+      "learning_rate": 0.00016557669918503353,
+      "loss": 0.808,
+      "mean_token_accuracy": 0.7599303185939789,
+      "num_tokens": 58336030.0,
+      "step": 17150
+    },
+    {
+      "epoch": 0.17222690593761292,
+      "learning_rate": 0.00016555662611907344,
+      "loss": 0.793,
+      "mean_token_accuracy": 0.759373950958252,
+      "num_tokens": 58369393.0,
+      "step": 17160
+    },
+    {
+      "epoch": 0.17232727126741337,
+      "learning_rate": 0.00016553655305311332,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7763037025928498,
+      "num_tokens": 58403831.0,
+      "step": 17170
+    },
+    {
+      "epoch": 0.17242763659721386,
+      "learning_rate": 0.00016551647998715326,
+      "loss": 0.7921,
+      "mean_token_accuracy": 0.7688264667987823,
+      "num_tokens": 58437571.0,
+      "step": 17180
+    },
+    {
+      "epoch": 0.17252800192701434,
+      "learning_rate": 0.00016549640692119314,
+      "loss": 0.7961,
+      "mean_token_accuracy": 0.7655426859855652,
+      "num_tokens": 58470734.0,
+      "step": 17190
+    },
+    {
+      "epoch": 0.1726283672568148,
+      "learning_rate": 0.00016547633385523305,
+      "loss": 0.7725,
+      "mean_token_accuracy": 0.7713332056999207,
+      "num_tokens": 58504723.0,
+      "step": 17200
+    },
+    {
+      "epoch": 0.17272873258661528,
+      "learning_rate": 0.00016545626078927296,
+      "loss": 0.7972,
+      "mean_token_accuracy": 0.7661644756793976,
+      "num_tokens": 58539048.0,
+      "step": 17210
+    },
+    {
+      "epoch": 0.17282909791641576,
+      "learning_rate": 0.00016543618772331287,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7684025287628173,
+      "num_tokens": 58574095.0,
+      "step": 17220
+    },
+    {
+      "epoch": 0.17292946324621622,
+      "learning_rate": 0.00016541611465735275,
+      "loss": 0.7969,
+      "mean_token_accuracy": 0.7652607381343841,
+      "num_tokens": 58607984.0,
+      "step": 17230
+    },
+    {
+      "epoch": 0.1730298285760167,
+      "learning_rate": 0.00016539604159139268,
+      "loss": 0.7865,
+      "mean_token_accuracy": 0.7680651903152466,
+      "num_tokens": 58641860.0,
+      "step": 17240
+    },
+    {
+      "epoch": 0.1731301939058172,
+      "learning_rate": 0.0001653759685254326,
+      "loss": 0.799,
+      "mean_token_accuracy": 0.761265105009079,
+      "num_tokens": 58675411.0,
+      "step": 17250
+    },
+    {
+      "epoch": 0.17323055923561764,
+      "learning_rate": 0.0001653558954594725,
+      "loss": 0.7661,
+      "mean_token_accuracy": 0.7691290080547333,
+      "num_tokens": 58709040.0,
+      "step": 17260
+    },
+    {
+      "epoch": 0.17333092456541813,
+      "learning_rate": 0.0001653358223935124,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.7651784241199493,
+      "num_tokens": 58743377.0,
+      "step": 17270
+    },
+    {
+      "epoch": 0.17343128989521858,
+      "learning_rate": 0.0001653157493275523,
+      "loss": 0.7788,
+      "mean_token_accuracy": 0.7682713389396667,
+      "num_tokens": 58776804.0,
+      "step": 17280
+    },
+    {
+      "epoch": 0.17353165522501907,
+      "learning_rate": 0.00016529567626159223,
+      "loss": 0.7371,
+      "mean_token_accuracy": 0.7746301293373108,
+      "num_tokens": 58810322.0,
+      "step": 17290
+    },
+    {
+      "epoch": 0.17363202055481955,
+      "learning_rate": 0.0001652756031956321,
+      "loss": 0.7754,
+      "mean_token_accuracy": 0.769923746585846,
+      "num_tokens": 58844019.0,
+      "step": 17300
+    },
+    {
+      "epoch": 0.17373238588462,
+      "learning_rate": 0.00016525553012967202,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7680152416229248,
+      "num_tokens": 58878254.0,
+      "step": 17310
+    },
+    {
+      "epoch": 0.1738327512144205,
+      "learning_rate": 0.00016523545706371192,
+      "loss": 0.7867,
+      "mean_token_accuracy": 0.7672698497772217,
+      "num_tokens": 58913152.0,
+      "step": 17320
+    },
+    {
+      "epoch": 0.17393311654422097,
+      "learning_rate": 0.00016521538399775183,
+      "loss": 0.7675,
+      "mean_token_accuracy": 0.769359028339386,
+      "num_tokens": 58947238.0,
+      "step": 17330
+    },
+    {
+      "epoch": 0.17403348187402143,
+      "learning_rate": 0.00016519531093179171,
+      "loss": 0.7503,
+      "mean_token_accuracy": 0.7771645188331604,
+      "num_tokens": 58980919.0,
+      "step": 17340
+    },
+    {
+      "epoch": 0.1741338472038219,
+      "learning_rate": 0.00016517523786583165,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.7696813583374024,
+      "num_tokens": 59014821.0,
+      "step": 17350
+    },
+    {
+      "epoch": 0.1742342125336224,
+      "learning_rate": 0.00016515516479987153,
+      "loss": 0.7638,
+      "mean_token_accuracy": 0.7662936210632324,
+      "num_tokens": 59048470.0,
+      "step": 17360
+    },
+    {
+      "epoch": 0.17433457786342285,
+      "learning_rate": 0.00016513509173391144,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7683330059051514,
+      "num_tokens": 59081997.0,
+      "step": 17370
+    },
+    {
+      "epoch": 0.17443494319322334,
+      "learning_rate": 0.00016511501866795135,
+      "loss": 0.784,
+      "mean_token_accuracy": 0.7709557950496674,
+      "num_tokens": 59116091.0,
+      "step": 17380
+    },
+    {
+      "epoch": 0.17453530852302382,
+      "learning_rate": 0.00016509494560199126,
+      "loss": 0.78,
+      "mean_token_accuracy": 0.7627904713153839,
+      "num_tokens": 59150435.0,
+      "step": 17390
+    },
+    {
+      "epoch": 0.17463567385282427,
+      "learning_rate": 0.00016507487253603116,
+      "loss": 0.7695,
+      "mean_token_accuracy": 0.7699004292488099,
+      "num_tokens": 59184392.0,
+      "step": 17400
+    },
+    {
+      "epoch": 0.17473603918262476,
+      "learning_rate": 0.00016505479947007107,
+      "loss": 0.7657,
+      "mean_token_accuracy": 0.7727609097957611,
+      "num_tokens": 59218019.0,
+      "step": 17410
+    },
+    {
+      "epoch": 0.17483640451242521,
+      "learning_rate": 0.00016503472640411098,
+      "loss": 0.779,
+      "mean_token_accuracy": 0.7669853568077087,
+      "num_tokens": 59252988.0,
+      "step": 17420
+    },
+    {
+      "epoch": 0.1749367698422257,
+      "learning_rate": 0.00016501465333815086,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7775791406631469,
+      "num_tokens": 59287957.0,
+      "step": 17430
+    },
+    {
+      "epoch": 0.17503713517202618,
+      "learning_rate": 0.0001649945802721908,
+      "loss": 0.773,
+      "mean_token_accuracy": 0.7717095196247101,
+      "num_tokens": 59321479.0,
+      "step": 17440
+    },
+    {
+      "epoch": 0.17513750050182664,
+      "learning_rate": 0.00016497450720623068,
+      "loss": 0.7786,
+      "mean_token_accuracy": 0.7710772454738617,
+      "num_tokens": 59355315.0,
+      "step": 17450
+    },
+    {
+      "epoch": 0.17523786583162712,
+      "learning_rate": 0.0001649544341402706,
+      "loss": 0.7544,
+      "mean_token_accuracy": 0.779824960231781,
+      "num_tokens": 59388451.0,
+      "step": 17460
+    },
+    {
+      "epoch": 0.1753382311614276,
+      "learning_rate": 0.0001649343610743105,
+      "loss": 0.7758,
+      "mean_token_accuracy": 0.7695206284523011,
+      "num_tokens": 59421993.0,
+      "step": 17470
+    },
+    {
+      "epoch": 0.17543859649122806,
+      "learning_rate": 0.0001649142880083504,
+      "loss": 0.7765,
+      "mean_token_accuracy": 0.7642025172710418,
+      "num_tokens": 59455941.0,
+      "step": 17480
+    },
+    {
+      "epoch": 0.17553896182102854,
+      "learning_rate": 0.0001648942149423903,
+      "loss": 0.7675,
+      "mean_token_accuracy": 0.7715909838676452,
+      "num_tokens": 59489901.0,
+      "step": 17490
+    },
+    {
+      "epoch": 0.17563932715082903,
+      "learning_rate": 0.00016487414187643022,
+      "loss": 0.7584,
+      "mean_token_accuracy": 0.7685763359069824,
+      "num_tokens": 59523829.0,
+      "step": 17500
+    },
+    {
+      "epoch": 0.17573969248062948,
+      "learning_rate": 0.00016485406881047013,
+      "loss": 0.796,
+      "mean_token_accuracy": 0.7640321314334869,
+      "num_tokens": 59557675.0,
+      "step": 17510
+    },
+    {
+      "epoch": 0.17584005781042997,
+      "learning_rate": 0.00016483399574451,
+      "loss": 0.79,
+      "mean_token_accuracy": 0.768464207649231,
+      "num_tokens": 59591396.0,
+      "step": 17520
+    },
+    {
+      "epoch": 0.17594042314023045,
+      "learning_rate": 0.00016481392267854995,
+      "loss": 0.7907,
+      "mean_token_accuracy": 0.7635212242603302,
+      "num_tokens": 59624772.0,
+      "step": 17530
+    },
+    {
+      "epoch": 0.1760407884700309,
+      "learning_rate": 0.00016479384961258983,
+      "loss": 0.7781,
+      "mean_token_accuracy": 0.7647325575351716,
+      "num_tokens": 59658505.0,
+      "step": 17540
+    },
+    {
+      "epoch": 0.1761411537998314,
+      "learning_rate": 0.00016477377654662974,
+      "loss": 0.8023,
+      "mean_token_accuracy": 0.7604661285877228,
+      "num_tokens": 59693441.0,
+      "step": 17550
+    },
+    {
+      "epoch": 0.17624151912963185,
+      "learning_rate": 0.00016475370348066964,
+      "loss": 0.7838,
+      "mean_token_accuracy": 0.7707070887088776,
+      "num_tokens": 59726652.0,
+      "step": 17560
+    },
+    {
+      "epoch": 0.17634188445943233,
+      "learning_rate": 0.00016473363041470955,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7668064892292022,
+      "num_tokens": 59760757.0,
+      "step": 17570
+    },
+    {
+      "epoch": 0.1764422497892328,
+      "learning_rate": 0.00016471355734874943,
+      "loss": 0.7494,
+      "mean_token_accuracy": 0.7720450818538666,
+      "num_tokens": 59795084.0,
+      "step": 17580
+    },
+    {
+      "epoch": 0.17654261511903327,
+      "learning_rate": 0.00016469348428278937,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7722791850566864,
+      "num_tokens": 59828670.0,
+      "step": 17590
+    },
+    {
+      "epoch": 0.17664298044883375,
+      "learning_rate": 0.00016467341121682928,
+      "loss": 0.786,
+      "mean_token_accuracy": 0.7623280882835388,
+      "num_tokens": 59862929.0,
+      "step": 17600
+    },
+    {
+      "epoch": 0.17674334577863424,
+      "learning_rate": 0.00016465333815086916,
+      "loss": 0.7719,
+      "mean_token_accuracy": 0.7731631696224213,
+      "num_tokens": 59896994.0,
+      "step": 17610
+    },
+    {
+      "epoch": 0.1768437111084347,
+      "learning_rate": 0.0001646332650849091,
+      "loss": 0.7781,
+      "mean_token_accuracy": 0.7665385782718659,
+      "num_tokens": 59931868.0,
+      "step": 17620
+    },
+    {
+      "epoch": 0.17694407643823518,
+      "learning_rate": 0.00016461319201894898,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7663361430168152,
+      "num_tokens": 59965609.0,
+      "step": 17630
+    },
+    {
+      "epoch": 0.17704444176803566,
+      "learning_rate": 0.00016459311895298889,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.772375512123108,
+      "num_tokens": 60000021.0,
+      "step": 17640
+    },
+    {
+      "epoch": 0.17714480709783612,
+      "learning_rate": 0.0001645730458870288,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.775952297449112,
+      "num_tokens": 60033294.0,
+      "step": 17650
+    },
+    {
+      "epoch": 0.1772451724276366,
+      "learning_rate": 0.0001645529728210687,
+      "loss": 0.734,
+      "mean_token_accuracy": 0.7766773700714111,
+      "num_tokens": 60066604.0,
+      "step": 17660
+    },
+    {
+      "epoch": 0.17734553775743708,
+      "learning_rate": 0.00016453289975510858,
+      "loss": 0.7978,
+      "mean_token_accuracy": 0.7641145765781403,
+      "num_tokens": 60101585.0,
+      "step": 17670
+    },
+    {
+      "epoch": 0.17744590308723754,
+      "learning_rate": 0.00016451282668914852,
+      "loss": 0.7697,
+      "mean_token_accuracy": 0.7713691294193268,
+      "num_tokens": 60135502.0,
+      "step": 17680
+    },
+    {
+      "epoch": 0.17754626841703802,
+      "learning_rate": 0.0001644927536231884,
+      "loss": 0.7834,
+      "mean_token_accuracy": 0.7656452357769012,
+      "num_tokens": 60168785.0,
+      "step": 17690
+    },
+    {
+      "epoch": 0.1776466337468385,
+      "learning_rate": 0.0001644726805572283,
+      "loss": 0.7761,
+      "mean_token_accuracy": 0.7686882853507996,
+      "num_tokens": 60203573.0,
+      "step": 17700
+    },
+    {
+      "epoch": 0.17774699907663896,
+      "learning_rate": 0.00016445260749126822,
+      "loss": 0.7113,
+      "mean_token_accuracy": 0.7895864546298981,
+      "num_tokens": 60236447.0,
+      "step": 17710
+    },
+    {
+      "epoch": 0.17784736440643945,
+      "learning_rate": 0.00016443253442530813,
+      "loss": 0.8131,
+      "mean_token_accuracy": 0.7609871804714203,
+      "num_tokens": 60270141.0,
+      "step": 17720
+    },
+    {
+      "epoch": 0.1779477297362399,
+      "learning_rate": 0.00016441246135934803,
+      "loss": 0.7605,
+      "mean_token_accuracy": 0.7759998798370361,
+      "num_tokens": 60304104.0,
+      "step": 17730
+    },
+    {
+      "epoch": 0.17804809506604038,
+      "learning_rate": 0.00016439238829338794,
+      "loss": 0.7951,
+      "mean_token_accuracy": 0.7703262507915497,
+      "num_tokens": 60339135.0,
+      "step": 17740
+    },
+    {
+      "epoch": 0.17814846039584087,
+      "learning_rate": 0.00016437231522742785,
+      "loss": 0.7537,
+      "mean_token_accuracy": 0.7827511548995971,
+      "num_tokens": 60372199.0,
+      "step": 17750
+    },
+    {
+      "epoch": 0.17824882572564132,
+      "learning_rate": 0.00016435224216146773,
+      "loss": 0.7809,
+      "mean_token_accuracy": 0.7661384582519531,
+      "num_tokens": 60406203.0,
+      "step": 17760
+    },
+    {
+      "epoch": 0.1783491910554418,
+      "learning_rate": 0.00016433216909550767,
+      "loss": 0.7537,
+      "mean_token_accuracy": 0.7733368217945099,
+      "num_tokens": 60440957.0,
+      "step": 17770
+    },
+    {
+      "epoch": 0.1784495563852423,
+      "learning_rate": 0.00016431209602954755,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7707456469535827,
+      "num_tokens": 60474679.0,
+      "step": 17780
+    },
+    {
+      "epoch": 0.17854992171504275,
+      "learning_rate": 0.00016429202296358746,
+      "loss": 0.7746,
+      "mean_token_accuracy": 0.7717380285263061,
+      "num_tokens": 60508245.0,
+      "step": 17790
+    },
+    {
+      "epoch": 0.17865028704484323,
+      "learning_rate": 0.00016427194989762737,
+      "loss": 0.7562,
+      "mean_token_accuracy": 0.7732972860336303,
+      "num_tokens": 60542678.0,
+      "step": 17800
+    },
+    {
+      "epoch": 0.17875065237464371,
+      "learning_rate": 0.00016425187683166727,
+      "loss": 0.7824,
+      "mean_token_accuracy": 0.7712754130363464,
+      "num_tokens": 60576155.0,
+      "step": 17810
+    },
+    {
+      "epoch": 0.17885101770444417,
+      "learning_rate": 0.00016423180376570718,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.7724056899547577,
+      "num_tokens": 60610482.0,
+      "step": 17820
+    },
+    {
+      "epoch": 0.17895138303424465,
+      "learning_rate": 0.0001642117306997471,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7675942838191986,
+      "num_tokens": 60644434.0,
+      "step": 17830
+    },
+    {
+      "epoch": 0.17905174836404514,
+      "learning_rate": 0.000164191657633787,
+      "loss": 0.7697,
+      "mean_token_accuracy": 0.7697039902210235,
+      "num_tokens": 60678529.0,
+      "step": 17840
+    },
+    {
+      "epoch": 0.1791521136938456,
+      "learning_rate": 0.0001641715845678269,
+      "loss": 0.7635,
+      "mean_token_accuracy": 0.7731961727142334,
+      "num_tokens": 60712066.0,
+      "step": 17850
+    },
+    {
+      "epoch": 0.17925247902364608,
+      "learning_rate": 0.00016415151150186682,
+      "loss": 0.7939,
+      "mean_token_accuracy": 0.7673891663551331,
+      "num_tokens": 60746415.0,
+      "step": 17860
+    },
+    {
+      "epoch": 0.17935284435344653,
+      "learning_rate": 0.0001641314384359067,
+      "loss": 0.7814,
+      "mean_token_accuracy": 0.7660547375679017,
+      "num_tokens": 60780453.0,
+      "step": 17870
+    },
+    {
+      "epoch": 0.17945320968324702,
+      "learning_rate": 0.00016411136536994663,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.7775905013084412,
+      "num_tokens": 60815219.0,
+      "step": 17880
+    },
+    {
+      "epoch": 0.1795535750130475,
+      "learning_rate": 0.00016409129230398651,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7698129117488861,
+      "num_tokens": 60849708.0,
+      "step": 17890
+    },
+    {
+      "epoch": 0.17965394034284796,
+      "learning_rate": 0.00016407121923802642,
+      "loss": 0.8327,
+      "mean_token_accuracy": 0.7578640341758728,
+      "num_tokens": 60882991.0,
+      "step": 17900
+    },
+    {
+      "epoch": 0.17975430567264844,
+      "learning_rate": 0.00016405114617206633,
+      "loss": 0.7803,
+      "mean_token_accuracy": 0.7644511163234711,
+      "num_tokens": 60916793.0,
+      "step": 17910
+    },
+    {
+      "epoch": 0.17985467100244892,
+      "learning_rate": 0.00016403107310610624,
+      "loss": 0.7698,
+      "mean_token_accuracy": 0.7728886902332306,
+      "num_tokens": 60951248.0,
+      "step": 17920
+    },
+    {
+      "epoch": 0.17995503633224938,
+      "learning_rate": 0.00016401100004014612,
+      "loss": 0.78,
+      "mean_token_accuracy": 0.7688422322273254,
+      "num_tokens": 60984506.0,
+      "step": 17930
+    },
+    {
+      "epoch": 0.18005540166204986,
+      "learning_rate": 0.00016399092697418606,
+      "loss": 0.7535,
+      "mean_token_accuracy": 0.7761100590229034,
+      "num_tokens": 61017911.0,
+      "step": 17940
+    },
+    {
+      "epoch": 0.18015576699185035,
+      "learning_rate": 0.00016397085390822597,
+      "loss": 0.8006,
+      "mean_token_accuracy": 0.7644067168235779,
+      "num_tokens": 61053525.0,
+      "step": 17950
+    },
+    {
+      "epoch": 0.1802561323216508,
+      "learning_rate": 0.00016395078084226585,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7717255651950836,
+      "num_tokens": 61087497.0,
+      "step": 17960
+    },
+    {
+      "epoch": 0.1803564976514513,
+      "learning_rate": 0.00016393070777630578,
+      "loss": 0.795,
+      "mean_token_accuracy": 0.7677553355693817,
+      "num_tokens": 61122792.0,
+      "step": 17970
+    },
+    {
+      "epoch": 0.18045686298125177,
+      "learning_rate": 0.00016391063471034566,
+      "loss": 0.7721,
+      "mean_token_accuracy": 0.7706233382225036,
+      "num_tokens": 61157415.0,
+      "step": 17980
+    },
+    {
+      "epoch": 0.18055722831105223,
+      "learning_rate": 0.00016389056164438557,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7725520610809327,
+      "num_tokens": 61190629.0,
+      "step": 17990
+    },
+    {
+      "epoch": 0.1806575936408527,
+      "learning_rate": 0.00016387048857842548,
+      "loss": 0.7434,
+      "mean_token_accuracy": 0.7789816677570343,
+      "num_tokens": 61224902.0,
+      "step": 18000
+    },
+    {
+      "epoch": 0.18075795897065317,
+      "learning_rate": 0.0001638504155124654,
+      "loss": 0.7332,
+      "mean_token_accuracy": 0.7772551894187927,
+      "num_tokens": 61259118.0,
+      "step": 18010
+    },
+    {
+      "epoch": 0.18085832430045365,
+      "learning_rate": 0.00016383034244650527,
+      "loss": 0.8069,
+      "mean_token_accuracy": 0.7600550413131714,
+      "num_tokens": 61293836.0,
+      "step": 18020
+    },
+    {
+      "epoch": 0.18095868963025413,
+      "learning_rate": 0.0001638102693805452,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.7729291975498199,
+      "num_tokens": 61327575.0,
+      "step": 18030
+    },
+    {
+      "epoch": 0.1810590549600546,
+      "learning_rate": 0.0001637901963145851,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7731200933456421,
+      "num_tokens": 61361526.0,
+      "step": 18040
+    },
+    {
+      "epoch": 0.18115942028985507,
+      "learning_rate": 0.000163770123248625,
+      "loss": 0.7899,
+      "mean_token_accuracy": 0.7629842936992646,
+      "num_tokens": 61396000.0,
+      "step": 18050
+    },
+    {
+      "epoch": 0.18125978561965556,
+      "learning_rate": 0.0001637500501826649,
+      "loss": 0.7853,
+      "mean_token_accuracy": 0.7604344427585602,
+      "num_tokens": 61430117.0,
+      "step": 18060
+    },
+    {
+      "epoch": 0.181360150949456,
+      "learning_rate": 0.0001637299771167048,
+      "loss": 0.7893,
+      "mean_token_accuracy": 0.766669875383377,
+      "num_tokens": 61463966.0,
+      "step": 18070
+    },
+    {
+      "epoch": 0.1814605162792565,
+      "learning_rate": 0.00016370990405074472,
+      "loss": 0.7824,
+      "mean_token_accuracy": 0.7696913361549378,
+      "num_tokens": 61497639.0,
+      "step": 18080
+    },
+    {
+      "epoch": 0.18156088160905698,
+      "learning_rate": 0.00016368983098478463,
+      "loss": 0.7629,
+      "mean_token_accuracy": 0.77036554813385,
+      "num_tokens": 61532293.0,
+      "step": 18090
+    },
+    {
+      "epoch": 0.18166124693885743,
+      "learning_rate": 0.00016366975791882454,
+      "loss": 0.7699,
+      "mean_token_accuracy": 0.7703576445579529,
+      "num_tokens": 61567287.0,
+      "step": 18100
+    },
+    {
+      "epoch": 0.18176161226865792,
+      "learning_rate": 0.00016364968485286442,
+      "loss": 0.7715,
+      "mean_token_accuracy": 0.7734453499317169,
+      "num_tokens": 61601581.0,
+      "step": 18110
+    },
+    {
+      "epoch": 0.1818619775984584,
+      "learning_rate": 0.00016362961178690435,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7735493004322052,
+      "num_tokens": 61636230.0,
+      "step": 18120
+    },
+    {
+      "epoch": 0.18196234292825886,
+      "learning_rate": 0.00016360953872094424,
+      "loss": 0.7957,
+      "mean_token_accuracy": 0.7671381592750549,
+      "num_tokens": 61670977.0,
+      "step": 18130
+    },
+    {
+      "epoch": 0.18206270825805934,
+      "learning_rate": 0.00016358946565498414,
+      "loss": 0.7564,
+      "mean_token_accuracy": 0.7743266701698304,
+      "num_tokens": 61704761.0,
+      "step": 18140
+    },
+    {
+      "epoch": 0.1821630735878598,
+      "learning_rate": 0.00016356939258902405,
+      "loss": 0.793,
+      "mean_token_accuracy": 0.7678810358047485,
+      "num_tokens": 61738979.0,
+      "step": 18150
+    },
+    {
+      "epoch": 0.18226343891766028,
+      "learning_rate": 0.00016354931952306396,
+      "loss": 0.7911,
+      "mean_token_accuracy": 0.7643441200256348,
+      "num_tokens": 61772803.0,
+      "step": 18160
+    },
+    {
+      "epoch": 0.18236380424746076,
+      "learning_rate": 0.00016352924645710387,
+      "loss": 0.7843,
+      "mean_token_accuracy": 0.7699290454387665,
+      "num_tokens": 61806363.0,
+      "step": 18170
+    },
+    {
+      "epoch": 0.18246416957726122,
+      "learning_rate": 0.00016350917339114378,
+      "loss": 0.7534,
+      "mean_token_accuracy": 0.774201774597168,
+      "num_tokens": 61839436.0,
+      "step": 18180
+    },
+    {
+      "epoch": 0.1825645349070617,
+      "learning_rate": 0.0001634891003251837,
+      "loss": 0.738,
+      "mean_token_accuracy": 0.7790593385696412,
+      "num_tokens": 61873826.0,
+      "step": 18190
+    },
+    {
+      "epoch": 0.1826649002368622,
+      "learning_rate": 0.00016346902725922357,
+      "loss": 0.7757,
+      "mean_token_accuracy": 0.769185584783554,
+      "num_tokens": 61907947.0,
+      "step": 18200
+    },
+    {
+      "epoch": 0.18276526556666264,
+      "learning_rate": 0.0001634489541932635,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7664050936698914,
+      "num_tokens": 61942529.0,
+      "step": 18210
+    },
+    {
+      "epoch": 0.18286563089646313,
+      "learning_rate": 0.00016342888112730338,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.7730357885360718,
+      "num_tokens": 61976876.0,
+      "step": 18220
+    },
+    {
+      "epoch": 0.1829659962262636,
+      "learning_rate": 0.0001634088080613433,
+      "loss": 0.8061,
+      "mean_token_accuracy": 0.7635744273662567,
+      "num_tokens": 62011317.0,
+      "step": 18230
+    },
+    {
+      "epoch": 0.18306636155606407,
+      "learning_rate": 0.0001633887349953832,
+      "loss": 0.7827,
+      "mean_token_accuracy": 0.7664428949356079,
+      "num_tokens": 62045627.0,
+      "step": 18240
+    },
+    {
+      "epoch": 0.18316672688586455,
+      "learning_rate": 0.0001633686619294231,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.7628428339958191,
+      "num_tokens": 62079421.0,
+      "step": 18250
+    },
+    {
+      "epoch": 0.18326709221566503,
+      "learning_rate": 0.000163348588863463,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7762756168842315,
+      "num_tokens": 62114007.0,
+      "step": 18260
+    },
+    {
+      "epoch": 0.1833674575454655,
+      "learning_rate": 0.00016332851579750293,
+      "loss": 0.7622,
+      "mean_token_accuracy": 0.7699788212776184,
+      "num_tokens": 62148214.0,
+      "step": 18270
+    },
+    {
+      "epoch": 0.18346782287526597,
+      "learning_rate": 0.0001633084427315428,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7681151330471039,
+      "num_tokens": 62182380.0,
+      "step": 18280
+    },
+    {
+      "epoch": 0.18356818820506643,
+      "learning_rate": 0.00016328836966558272,
+      "loss": 0.7658,
+      "mean_token_accuracy": 0.7687394440174102,
+      "num_tokens": 62216320.0,
+      "step": 18290
+    },
+    {
+      "epoch": 0.1836685535348669,
+      "learning_rate": 0.00016326829659962263,
+      "loss": 0.786,
+      "mean_token_accuracy": 0.7645105063915253,
+      "num_tokens": 62249609.0,
+      "step": 18300
+    },
+    {
+      "epoch": 0.1837689188646674,
+      "learning_rate": 0.00016324822353366253,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.7732066512107849,
+      "num_tokens": 62283605.0,
+      "step": 18310
+    },
+    {
+      "epoch": 0.18386928419446785,
+      "learning_rate": 0.00016322815046770244,
+      "loss": 0.8099,
+      "mean_token_accuracy": 0.7657205998897553,
+      "num_tokens": 62318429.0,
+      "step": 18320
+    },
+    {
+      "epoch": 0.18396964952426834,
+      "learning_rate": 0.00016320807740174235,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.7653166830539704,
+      "num_tokens": 62352475.0,
+      "step": 18330
+    },
+    {
+      "epoch": 0.18407001485406882,
+      "learning_rate": 0.00016318800433578226,
+      "loss": 0.7986,
+      "mean_token_accuracy": 0.7597263157367706,
+      "num_tokens": 62386501.0,
+      "step": 18340
+    },
+    {
+      "epoch": 0.18417038018386928,
+      "learning_rate": 0.00016316793126982214,
+      "loss": 0.7243,
+      "mean_token_accuracy": 0.7759540557861329,
+      "num_tokens": 62420011.0,
+      "step": 18350
+    },
+    {
+      "epoch": 0.18427074551366976,
+      "learning_rate": 0.00016314785820386208,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7748587667942047,
+      "num_tokens": 62454044.0,
+      "step": 18360
+    },
+    {
+      "epoch": 0.18437111084347024,
+      "learning_rate": 0.00016312778513790196,
+      "loss": 0.7768,
+      "mean_token_accuracy": 0.7700540661811829,
+      "num_tokens": 62488487.0,
+      "step": 18370
+    },
+    {
+      "epoch": 0.1844714761732707,
+      "learning_rate": 0.0001631077120719419,
+      "loss": 0.7639,
+      "mean_token_accuracy": 0.7720458507537842,
+      "num_tokens": 62522346.0,
+      "step": 18380
+    },
+    {
+      "epoch": 0.18457184150307118,
+      "learning_rate": 0.00016308763900598177,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.768638151884079,
+      "num_tokens": 62556694.0,
+      "step": 18390
+    },
+    {
+      "epoch": 0.18467220683287167,
+      "learning_rate": 0.00016306756594002168,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7727836549282074,
+      "num_tokens": 62590509.0,
+      "step": 18400
+    },
+    {
+      "epoch": 0.18477257216267212,
+      "learning_rate": 0.0001630474928740616,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7713655471801758,
+      "num_tokens": 62624925.0,
+      "step": 18410
+    },
+    {
+      "epoch": 0.1848729374924726,
+      "learning_rate": 0.0001630274198081015,
+      "loss": 0.8011,
+      "mean_token_accuracy": 0.7590881168842316,
+      "num_tokens": 62659270.0,
+      "step": 18420
+    },
+    {
+      "epoch": 0.18497330282227306,
+      "learning_rate": 0.0001630073467421414,
+      "loss": 0.7709,
+      "mean_token_accuracy": 0.7708830296993255,
+      "num_tokens": 62693910.0,
+      "step": 18430
+    },
+    {
+      "epoch": 0.18507366815207354,
+      "learning_rate": 0.00016298727367618132,
+      "loss": 0.7835,
+      "mean_token_accuracy": 0.7661713480949401,
+      "num_tokens": 62727733.0,
+      "step": 18440
+    },
+    {
+      "epoch": 0.18517403348187403,
+      "learning_rate": 0.00016296720061022122,
+      "loss": 0.7698,
+      "mean_token_accuracy": 0.7711425065994263,
+      "num_tokens": 62762110.0,
+      "step": 18450
+    },
+    {
+      "epoch": 0.18527439881167448,
+      "learning_rate": 0.0001629471275442611,
+      "loss": 0.7355,
+      "mean_token_accuracy": 0.7776850581169128,
+      "num_tokens": 62796340.0,
+      "step": 18460
+    },
+    {
+      "epoch": 0.18537476414147497,
+      "learning_rate": 0.00016292705447830104,
+      "loss": 0.8002,
+      "mean_token_accuracy": 0.76320121884346,
+      "num_tokens": 62830746.0,
+      "step": 18470
+    },
+    {
+      "epoch": 0.18547512947127545,
+      "learning_rate": 0.00016290698141234092,
+      "loss": 0.8161,
+      "mean_token_accuracy": 0.7639877259731293,
+      "num_tokens": 62864457.0,
+      "step": 18480
+    },
+    {
+      "epoch": 0.1855754948010759,
+      "learning_rate": 0.00016288690834638083,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7661233425140381,
+      "num_tokens": 62897935.0,
+      "step": 18490
+    },
+    {
+      "epoch": 0.1856758601308764,
+      "learning_rate": 0.00016286683528042074,
+      "loss": 0.7715,
+      "mean_token_accuracy": 0.7745231449604034,
+      "num_tokens": 62931317.0,
+      "step": 18500
+    },
+    {
+      "epoch": 0.18577622546067687,
+      "learning_rate": 0.00016284676221446065,
+      "loss": 0.764,
+      "mean_token_accuracy": 0.7681089520454407,
+      "num_tokens": 62966225.0,
+      "step": 18510
+    },
+    {
+      "epoch": 0.18587659079047733,
+      "learning_rate": 0.00016282668914850056,
+      "loss": 0.7418,
+      "mean_token_accuracy": 0.7758630812168121,
+      "num_tokens": 62999601.0,
+      "step": 18520
+    },
+    {
+      "epoch": 0.18597695612027781,
+      "learning_rate": 0.00016280661608254047,
+      "loss": 0.7502,
+      "mean_token_accuracy": 0.7772676169872283,
+      "num_tokens": 63034028.0,
+      "step": 18530
+    },
+    {
+      "epoch": 0.1860773214500783,
+      "learning_rate": 0.00016278654301658037,
+      "loss": 0.7602,
+      "mean_token_accuracy": 0.77999187707901,
+      "num_tokens": 63068275.0,
+      "step": 18540
+    },
+    {
+      "epoch": 0.18617768677987875,
+      "learning_rate": 0.00016276646995062025,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7712823390960694,
+      "num_tokens": 63102226.0,
+      "step": 18550
+    },
+    {
+      "epoch": 0.18627805210967924,
+      "learning_rate": 0.0001627463968846602,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.7757633328437805,
+      "num_tokens": 63136322.0,
+      "step": 18560
+    },
+    {
+      "epoch": 0.1863784174394797,
+      "learning_rate": 0.00016272632381870007,
+      "loss": 0.8076,
+      "mean_token_accuracy": 0.7684038341045379,
+      "num_tokens": 63169950.0,
+      "step": 18570
+    },
+    {
+      "epoch": 0.18647878276928018,
+      "learning_rate": 0.00016270625075273998,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.7748810112476349,
+      "num_tokens": 63204442.0,
+      "step": 18580
+    },
+    {
+      "epoch": 0.18657914809908066,
+      "learning_rate": 0.0001626861776867799,
+      "loss": 0.7866,
+      "mean_token_accuracy": 0.7683030843734742,
+      "num_tokens": 63237776.0,
+      "step": 18590
+    },
+    {
+      "epoch": 0.18667951342888112,
+      "learning_rate": 0.0001626661046208198,
+      "loss": 0.7459,
+      "mean_token_accuracy": 0.7781949758529663,
+      "num_tokens": 63271701.0,
+      "step": 18600
+    },
+    {
+      "epoch": 0.1867798787586816,
+      "learning_rate": 0.00016264603155485968,
+      "loss": 0.772,
+      "mean_token_accuracy": 0.7715380012989044,
+      "num_tokens": 63305993.0,
+      "step": 18610
+    },
+    {
+      "epoch": 0.18688024408848208,
+      "learning_rate": 0.00016262595848889961,
+      "loss": 0.7893,
+      "mean_token_accuracy": 0.7674874544143677,
+      "num_tokens": 63340482.0,
+      "step": 18620
+    },
+    {
+      "epoch": 0.18698060941828254,
+      "learning_rate": 0.0001626058854229395,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7720118045806885,
+      "num_tokens": 63374747.0,
+      "step": 18630
+    },
+    {
+      "epoch": 0.18708097474808302,
+      "learning_rate": 0.0001625858123569794,
+      "loss": 0.7554,
+      "mean_token_accuracy": 0.7776407539844513,
+      "num_tokens": 63409758.0,
+      "step": 18640
+    },
+    {
+      "epoch": 0.1871813400778835,
+      "learning_rate": 0.0001625657392910193,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.7663829922676086,
+      "num_tokens": 63443948.0,
+      "step": 18650
+    },
+    {
+      "epoch": 0.18728170540768396,
+      "learning_rate": 0.00016254566622505922,
+      "loss": 0.7333,
+      "mean_token_accuracy": 0.7760569155216217,
+      "num_tokens": 63477292.0,
+      "step": 18660
+    },
+    {
+      "epoch": 0.18738207073748445,
+      "learning_rate": 0.00016252559315909913,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7722366213798523,
+      "num_tokens": 63512263.0,
+      "step": 18670
+    },
+    {
+      "epoch": 0.18748243606728493,
+      "learning_rate": 0.00016250552009313904,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.7722975015640259,
+      "num_tokens": 63546456.0,
+      "step": 18680
+    },
+    {
+      "epoch": 0.18758280139708539,
+      "learning_rate": 0.00016248544702717895,
+      "loss": 0.7483,
+      "mean_token_accuracy": 0.7726555287837982,
+      "num_tokens": 63580150.0,
+      "step": 18690
+    },
+    {
+      "epoch": 0.18768316672688587,
+      "learning_rate": 0.00016246537396121883,
+      "loss": 0.7964,
+      "mean_token_accuracy": 0.7653996467590332,
+      "num_tokens": 63613575.0,
+      "step": 18700
+    },
+    {
+      "epoch": 0.18778353205668633,
+      "learning_rate": 0.00016244530089525876,
+      "loss": 0.7967,
+      "mean_token_accuracy": 0.7680726945400238,
+      "num_tokens": 63647897.0,
+      "step": 18710
+    },
+    {
+      "epoch": 0.1878838973864868,
+      "learning_rate": 0.00016242522782929864,
+      "loss": 0.761,
+      "mean_token_accuracy": 0.7686357796192169,
+      "num_tokens": 63682485.0,
+      "step": 18720
+    },
+    {
+      "epoch": 0.1879842627162873,
+      "learning_rate": 0.00016240515476333855,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7725031673908234,
+      "num_tokens": 63716708.0,
+      "step": 18730
+    },
+    {
+      "epoch": 0.18808462804608775,
+      "learning_rate": 0.00016238508169737846,
+      "loss": 0.7876,
+      "mean_token_accuracy": 0.7653063297271728,
+      "num_tokens": 63750589.0,
+      "step": 18740
+    },
+    {
+      "epoch": 0.18818499337588823,
+      "learning_rate": 0.00016236500863141837,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.7706561863422394,
+      "num_tokens": 63784726.0,
+      "step": 18750
+    },
+    {
+      "epoch": 0.18828535870568872,
+      "learning_rate": 0.00016234493556545828,
+      "loss": 0.7524,
+      "mean_token_accuracy": 0.7732706964015961,
+      "num_tokens": 63818732.0,
+      "step": 18760
+    },
+    {
+      "epoch": 0.18838572403548917,
+      "learning_rate": 0.00016232486249949819,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.771875262260437,
+      "num_tokens": 63852581.0,
+      "step": 18770
+    },
+    {
+      "epoch": 0.18848608936528966,
+      "learning_rate": 0.0001623047894335381,
+      "loss": 0.7364,
+      "mean_token_accuracy": 0.7823958218097686,
+      "num_tokens": 63887370.0,
+      "step": 18780
+    },
+    {
+      "epoch": 0.18858645469509014,
+      "learning_rate": 0.00016228471636757798,
+      "loss": 0.7522,
+      "mean_token_accuracy": 0.7735142648220062,
+      "num_tokens": 63921195.0,
+      "step": 18790
+    },
+    {
+      "epoch": 0.1886868200248906,
+      "learning_rate": 0.0001622646433016179,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.7688505411148071,
+      "num_tokens": 63955282.0,
+      "step": 18800
+    },
+    {
+      "epoch": 0.18878718535469108,
+      "learning_rate": 0.0001622445702356578,
+      "loss": 0.7245,
+      "mean_token_accuracy": 0.7823804557323456,
+      "num_tokens": 63989952.0,
+      "step": 18810
+    },
+    {
+      "epoch": 0.18888755068449156,
+      "learning_rate": 0.0001622244971696977,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.7732816517353058,
+      "num_tokens": 64024831.0,
+      "step": 18820
+    },
+    {
+      "epoch": 0.18898791601429202,
+      "learning_rate": 0.0001622044241037376,
+      "loss": 0.7837,
+      "mean_token_accuracy": 0.7630086898803711,
+      "num_tokens": 64057698.0,
+      "step": 18830
+    },
+    {
+      "epoch": 0.1890882813440925,
+      "learning_rate": 0.00016218435103777752,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7682810366153717,
+      "num_tokens": 64091622.0,
+      "step": 18840
+    },
+    {
+      "epoch": 0.18918864667389296,
+      "learning_rate": 0.00016216427797181743,
+      "loss": 0.774,
+      "mean_token_accuracy": 0.763655811548233,
+      "num_tokens": 64125909.0,
+      "step": 18850
+    },
+    {
+      "epoch": 0.18928901200369344,
+      "learning_rate": 0.00016214420490585733,
+      "loss": 0.7757,
+      "mean_token_accuracy": 0.7707290172576904,
+      "num_tokens": 64159547.0,
+      "step": 18860
+    },
+    {
+      "epoch": 0.18938937733349392,
+      "learning_rate": 0.00016212413183989724,
+      "loss": 0.7333,
+      "mean_token_accuracy": 0.7797725558280945,
+      "num_tokens": 64194312.0,
+      "step": 18870
+    },
+    {
+      "epoch": 0.18948974266329438,
+      "learning_rate": 0.00016210405877393712,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.7734246671199798,
+      "num_tokens": 64228034.0,
+      "step": 18880
+    },
+    {
+      "epoch": 0.18959010799309486,
+      "learning_rate": 0.00016208398570797706,
+      "loss": 0.7897,
+      "mean_token_accuracy": 0.7694403171539307,
+      "num_tokens": 64261670.0,
+      "step": 18890
+    },
+    {
+      "epoch": 0.18969047332289535,
+      "learning_rate": 0.00016206391264201694,
+      "loss": 0.7826,
+      "mean_token_accuracy": 0.7667472302913666,
+      "num_tokens": 64295493.0,
+      "step": 18900
+    },
+    {
+      "epoch": 0.1897908386526958,
+      "learning_rate": 0.00016204383957605685,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7765673577785492,
+      "num_tokens": 64330705.0,
+      "step": 18910
+    },
+    {
+      "epoch": 0.1898912039824963,
+      "learning_rate": 0.00016202376651009676,
+      "loss": 0.7754,
+      "mean_token_accuracy": 0.7735772848129272,
+      "num_tokens": 64364839.0,
+      "step": 18920
+    },
+    {
+      "epoch": 0.18999156931229677,
+      "learning_rate": 0.00016200369344413667,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7750263512134552,
+      "num_tokens": 64399309.0,
+      "step": 18930
+    },
+    {
+      "epoch": 0.19009193464209723,
+      "learning_rate": 0.00016198362037817658,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7775311589241027,
+      "num_tokens": 64432659.0,
+      "step": 18940
+    },
+    {
+      "epoch": 0.1901922999718977,
+      "learning_rate": 0.00016196354731221648,
+      "loss": 0.7947,
+      "mean_token_accuracy": 0.7694822609424591,
+      "num_tokens": 64467264.0,
+      "step": 18950
+    },
+    {
+      "epoch": 0.1902926653016982,
+      "learning_rate": 0.00016194347424625637,
+      "loss": 0.7435,
+      "mean_token_accuracy": 0.7792628645896912,
+      "num_tokens": 64500619.0,
+      "step": 18960
+    },
+    {
+      "epoch": 0.19039303063149865,
+      "learning_rate": 0.0001619234011802963,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7780649542808533,
+      "num_tokens": 64535044.0,
+      "step": 18970
+    },
+    {
+      "epoch": 0.19049339596129913,
+      "learning_rate": 0.00016190332811433618,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.7738030016422272,
+      "num_tokens": 64568192.0,
+      "step": 18980
+    },
+    {
+      "epoch": 0.1905937612910996,
+      "learning_rate": 0.0001618832550483761,
+      "loss": 0.7508,
+      "mean_token_accuracy": 0.7778998613357544,
+      "num_tokens": 64602388.0,
+      "step": 18990
+    },
+    {
+      "epoch": 0.19069412662090007,
+      "learning_rate": 0.000161863181982416,
+      "loss": 0.7433,
+      "mean_token_accuracy": 0.7757618129253387,
+      "num_tokens": 64636406.0,
+      "step": 19000
+    },
+    {
+      "epoch": 0.19079449195070056,
+      "learning_rate": 0.0001618431089164559,
+      "loss": 0.7362,
+      "mean_token_accuracy": 0.7791922152042389,
+      "num_tokens": 64670618.0,
+      "step": 19010
+    },
+    {
+      "epoch": 0.190894857280501,
+      "learning_rate": 0.00016182303585049582,
+      "loss": 0.757,
+      "mean_token_accuracy": 0.7783075511455536,
+      "num_tokens": 64704532.0,
+      "step": 19020
+    },
+    {
+      "epoch": 0.1909952226103015,
+      "learning_rate": 0.00016180296278453572,
+      "loss": 0.7496,
+      "mean_token_accuracy": 0.7750849246978759,
+      "num_tokens": 64738653.0,
+      "step": 19030
+    },
+    {
+      "epoch": 0.19109558794010198,
+      "learning_rate": 0.00016178288971857563,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7745764195919037,
+      "num_tokens": 64772906.0,
+      "step": 19040
+    },
+    {
+      "epoch": 0.19119595326990244,
+      "learning_rate": 0.00016176281665261551,
+      "loss": 0.7985,
+      "mean_token_accuracy": 0.765564888715744,
+      "num_tokens": 64806675.0,
+      "step": 19050
+    },
+    {
+      "epoch": 0.19129631859970292,
+      "learning_rate": 0.00016174274358665545,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7681283593177796,
+      "num_tokens": 64840702.0,
+      "step": 19060
+    },
+    {
+      "epoch": 0.1913966839295034,
+      "learning_rate": 0.00016172267052069533,
+      "loss": 0.7764,
+      "mean_token_accuracy": 0.7724027931690216,
+      "num_tokens": 64874883.0,
+      "step": 19070
+    },
+    {
+      "epoch": 0.19149704925930386,
+      "learning_rate": 0.00016170259745473524,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7691495954990387,
+      "num_tokens": 64909325.0,
+      "step": 19080
+    },
+    {
+      "epoch": 0.19159741458910434,
+      "learning_rate": 0.00016168252438877515,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7718911945819855,
+      "num_tokens": 64943396.0,
+      "step": 19090
+    },
+    {
+      "epoch": 0.19169777991890483,
+      "learning_rate": 0.00016166245132281506,
+      "loss": 0.7724,
+      "mean_token_accuracy": 0.773424506187439,
+      "num_tokens": 64976834.0,
+      "step": 19100
+    },
+    {
+      "epoch": 0.19179814524870528,
+      "learning_rate": 0.00016164237825685496,
+      "loss": 0.7492,
+      "mean_token_accuracy": 0.7783820271492005,
+      "num_tokens": 65011251.0,
+      "step": 19110
+    },
+    {
+      "epoch": 0.19189851057850577,
+      "learning_rate": 0.00016162230519089487,
+      "loss": 0.7803,
+      "mean_token_accuracy": 0.7681094467639923,
+      "num_tokens": 65045076.0,
+      "step": 19120
+    },
+    {
+      "epoch": 0.19199887590830622,
+      "learning_rate": 0.00016160223212493478,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.7738582372665406,
+      "num_tokens": 65079194.0,
+      "step": 19130
+    },
+    {
+      "epoch": 0.1920992412381067,
+      "learning_rate": 0.00016158215905897466,
+      "loss": 0.7479,
+      "mean_token_accuracy": 0.7765959143638611,
+      "num_tokens": 65113027.0,
+      "step": 19140
+    },
+    {
+      "epoch": 0.1921996065679072,
+      "learning_rate": 0.0001615620859930146,
+      "loss": 0.7552,
+      "mean_token_accuracy": 0.7722527205944061,
+      "num_tokens": 65147200.0,
+      "step": 19150
+    },
+    {
+      "epoch": 0.19229997189770764,
+      "learning_rate": 0.00016154201292705448,
+      "loss": 0.7623,
+      "mean_token_accuracy": 0.7735209167003632,
+      "num_tokens": 65181510.0,
+      "step": 19160
+    },
+    {
+      "epoch": 0.19240033722750813,
+      "learning_rate": 0.0001615219398610944,
+      "loss": 0.7746,
+      "mean_token_accuracy": 0.7687154591083527,
+      "num_tokens": 65216910.0,
+      "step": 19170
+    },
+    {
+      "epoch": 0.1925007025573086,
+      "learning_rate": 0.0001615018667951343,
+      "loss": 0.7587,
+      "mean_token_accuracy": 0.7720073342323304,
+      "num_tokens": 65250578.0,
+      "step": 19180
+    },
+    {
+      "epoch": 0.19260106788710907,
+      "learning_rate": 0.0001614817937291742,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7646001219749451,
+      "num_tokens": 65285306.0,
+      "step": 19190
+    },
+    {
+      "epoch": 0.19270143321690955,
+      "learning_rate": 0.0001614617206632141,
+      "loss": 0.7715,
+      "mean_token_accuracy": 0.7696290969848633,
+      "num_tokens": 65318962.0,
+      "step": 19200
+    },
+    {
+      "epoch": 0.19280179854671003,
+      "learning_rate": 0.00016144164759725402,
+      "loss": 0.7646,
+      "mean_token_accuracy": 0.7771562337875366,
+      "num_tokens": 65353318.0,
+      "step": 19210
+    },
+    {
+      "epoch": 0.1929021638765105,
+      "learning_rate": 0.00016142157453129393,
+      "loss": 0.7875,
+      "mean_token_accuracy": 0.7628707885742188,
+      "num_tokens": 65387345.0,
+      "step": 19220
+    },
+    {
+      "epoch": 0.19300252920631097,
+      "learning_rate": 0.0001614015014653338,
+      "loss": 0.7945,
+      "mean_token_accuracy": 0.7688746809959411,
+      "num_tokens": 65421580.0,
+      "step": 19230
+    },
+    {
+      "epoch": 0.19310289453611146,
+      "learning_rate": 0.00016138142839937375,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7717042803764343,
+      "num_tokens": 65455984.0,
+      "step": 19240
+    },
+    {
+      "epoch": 0.1932032598659119,
+      "learning_rate": 0.00016136135533341363,
+      "loss": 0.7943,
+      "mean_token_accuracy": 0.759830641746521,
+      "num_tokens": 65490077.0,
+      "step": 19250
+    },
+    {
+      "epoch": 0.1933036251957124,
+      "learning_rate": 0.00016134128226745354,
+      "loss": 0.7432,
+      "mean_token_accuracy": 0.7727313756942749,
+      "num_tokens": 65524484.0,
+      "step": 19260
+    },
+    {
+      "epoch": 0.19340399052551285,
+      "learning_rate": 0.00016132120920149345,
+      "loss": 0.7773,
+      "mean_token_accuracy": 0.7688717007637024,
+      "num_tokens": 65558779.0,
+      "step": 19270
+    },
+    {
+      "epoch": 0.19350435585531334,
+      "learning_rate": 0.00016130113613553335,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7738625407218933,
+      "num_tokens": 65593308.0,
+      "step": 19280
+    },
+    {
+      "epoch": 0.19360472118511382,
+      "learning_rate": 0.00016128106306957324,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7738423943519592,
+      "num_tokens": 65627835.0,
+      "step": 19290
+    },
+    {
+      "epoch": 0.19370508651491428,
+      "learning_rate": 0.00016126099000361317,
+      "loss": 0.8207,
+      "mean_token_accuracy": 0.7625119268894196,
+      "num_tokens": 65661353.0,
+      "step": 19300
+    },
+    {
+      "epoch": 0.19380545184471476,
+      "learning_rate": 0.00016124091693765305,
+      "loss": 0.7023,
+      "mean_token_accuracy": 0.7863344430923462,
+      "num_tokens": 65696158.0,
+      "step": 19310
+    },
+    {
+      "epoch": 0.19390581717451524,
+      "learning_rate": 0.00016122084387169296,
+      "loss": 0.7513,
+      "mean_token_accuracy": 0.776838755607605,
+      "num_tokens": 65730366.0,
+      "step": 19320
+    },
+    {
+      "epoch": 0.1940061825043157,
+      "learning_rate": 0.00016120077080573287,
+      "loss": 0.7921,
+      "mean_token_accuracy": 0.7652410626411438,
+      "num_tokens": 65764302.0,
+      "step": 19330
+    },
+    {
+      "epoch": 0.19410654783411618,
+      "learning_rate": 0.00016118069773977278,
+      "loss": 0.7633,
+      "mean_token_accuracy": 0.7726734757423401,
+      "num_tokens": 65798873.0,
+      "step": 19340
+    },
+    {
+      "epoch": 0.19420691316391667,
+      "learning_rate": 0.00016116062467381269,
+      "loss": 0.8107,
+      "mean_token_accuracy": 0.7613295316696167,
+      "num_tokens": 65833515.0,
+      "step": 19350
+    },
+    {
+      "epoch": 0.19430727849371712,
+      "learning_rate": 0.0001611405516078526,
+      "loss": 0.7282,
+      "mean_token_accuracy": 0.78021360039711,
+      "num_tokens": 65868430.0,
+      "step": 19360
+    },
+    {
+      "epoch": 0.1944076438235176,
+      "learning_rate": 0.0001611204785418925,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.773417180776596,
+      "num_tokens": 65902908.0,
+      "step": 19370
+    },
+    {
+      "epoch": 0.1945080091533181,
+      "learning_rate": 0.00016110040547593238,
+      "loss": 0.764,
+      "mean_token_accuracy": 0.7677555322647095,
+      "num_tokens": 65937131.0,
+      "step": 19380
+    },
+    {
+      "epoch": 0.19460837448311855,
+      "learning_rate": 0.00016108033240997232,
+      "loss": 0.7787,
+      "mean_token_accuracy": 0.7665543496608734,
+      "num_tokens": 65971478.0,
+      "step": 19390
+    },
+    {
+      "epoch": 0.19470873981291903,
+      "learning_rate": 0.0001610602593440122,
+      "loss": 0.7957,
+      "mean_token_accuracy": 0.7618605971336365,
+      "num_tokens": 66006556.0,
+      "step": 19400
+    },
+    {
+      "epoch": 0.19480910514271949,
+      "learning_rate": 0.0001610401862780521,
+      "loss": 0.7613,
+      "mean_token_accuracy": 0.7710905432701111,
+      "num_tokens": 66041114.0,
+      "step": 19410
+    },
+    {
+      "epoch": 0.19490947047251997,
+      "learning_rate": 0.00016102011321209202,
+      "loss": 0.802,
+      "mean_token_accuracy": 0.7617272436618805,
+      "num_tokens": 66075592.0,
+      "step": 19420
+    },
+    {
+      "epoch": 0.19500983580232045,
+      "learning_rate": 0.00016100004014613193,
+      "loss": 0.7734,
+      "mean_token_accuracy": 0.7753029108047486,
+      "num_tokens": 66110273.0,
+      "step": 19430
+    },
+    {
+      "epoch": 0.1951102011321209,
+      "learning_rate": 0.00016097996708017183,
+      "loss": 0.7799,
+      "mean_token_accuracy": 0.7730140388011932,
+      "num_tokens": 66144736.0,
+      "step": 19440
+    },
+    {
+      "epoch": 0.1952105664619214,
+      "learning_rate": 0.00016095989401421174,
+      "loss": 0.8066,
+      "mean_token_accuracy": 0.7628728866577148,
+      "num_tokens": 66178161.0,
+      "step": 19450
+    },
+    {
+      "epoch": 0.19531093179172188,
+      "learning_rate": 0.00016093982094825165,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7740750789642334,
+      "num_tokens": 66212869.0,
+      "step": 19460
+    },
+    {
+      "epoch": 0.19541129712152233,
+      "learning_rate": 0.00016091974788229153,
+      "loss": 0.7671,
+      "mean_token_accuracy": 0.7667285203933716,
+      "num_tokens": 66246749.0,
+      "step": 19470
+    },
+    {
+      "epoch": 0.19551166245132282,
+      "learning_rate": 0.00016089967481633147,
+      "loss": 0.7931,
+      "mean_token_accuracy": 0.7644418716430664,
+      "num_tokens": 66281090.0,
+      "step": 19480
+    },
+    {
+      "epoch": 0.1956120277811233,
+      "learning_rate": 0.00016087960175037135,
+      "loss": 0.7689,
+      "mean_token_accuracy": 0.7702011048793793,
+      "num_tokens": 66314725.0,
+      "step": 19490
+    },
+    {
+      "epoch": 0.19571239311092375,
+      "learning_rate": 0.00016085952868441126,
+      "loss": 0.7489,
+      "mean_token_accuracy": 0.7723415434360504,
+      "num_tokens": 66348900.0,
+      "step": 19500
+    },
+    {
+      "epoch": 0.19581275844072424,
+      "learning_rate": 0.00016083945561845117,
+      "loss": 0.7949,
+      "mean_token_accuracy": 0.764240151643753,
+      "num_tokens": 66384488.0,
+      "step": 19510
+    },
+    {
+      "epoch": 0.19591312377052472,
+      "learning_rate": 0.00016081938255249107,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7781309485435486,
+      "num_tokens": 66417923.0,
+      "step": 19520
+    },
+    {
+      "epoch": 0.19601348910032518,
+      "learning_rate": 0.00016079930948653098,
+      "loss": 0.7758,
+      "mean_token_accuracy": 0.7700519561767578,
+      "num_tokens": 66451905.0,
+      "step": 19530
+    },
+    {
+      "epoch": 0.19611385443012566,
+      "learning_rate": 0.0001607792364205709,
+      "loss": 0.7327,
+      "mean_token_accuracy": 0.779544734954834,
+      "num_tokens": 66486623.0,
+      "step": 19540
+    },
+    {
+      "epoch": 0.19621421975992612,
+      "learning_rate": 0.0001607591633546108,
+      "loss": 0.8051,
+      "mean_token_accuracy": 0.7675462424755096,
+      "num_tokens": 66520599.0,
+      "step": 19550
+    },
+    {
+      "epoch": 0.1963145850897266,
+      "learning_rate": 0.0001607390902886507,
+      "loss": 0.7556,
+      "mean_token_accuracy": 0.7777894735336304,
+      "num_tokens": 66554324.0,
+      "step": 19560
+    },
+    {
+      "epoch": 0.19641495041952708,
+      "learning_rate": 0.00016071901722269062,
+      "loss": 0.8006,
+      "mean_token_accuracy": 0.7677490472793579,
+      "num_tokens": 66588213.0,
+      "step": 19570
+    },
+    {
+      "epoch": 0.19651531574932754,
+      "learning_rate": 0.0001606989441567305,
+      "loss": 0.7556,
+      "mean_token_accuracy": 0.7706791758537292,
+      "num_tokens": 66622343.0,
+      "step": 19580
+    },
+    {
+      "epoch": 0.19661568107912802,
+      "learning_rate": 0.00016067887109077043,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7715229392051697,
+      "num_tokens": 66656268.0,
+      "step": 19590
+    },
+    {
+      "epoch": 0.1967160464089285,
+      "learning_rate": 0.00016065879802481032,
+      "loss": 0.7393,
+      "mean_token_accuracy": 0.775808721780777,
+      "num_tokens": 66690563.0,
+      "step": 19600
+    },
+    {
+      "epoch": 0.19681641173872896,
+      "learning_rate": 0.00016063872495885022,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.7682813704013824,
+      "num_tokens": 66724181.0,
+      "step": 19610
+    },
+    {
+      "epoch": 0.19691677706852945,
+      "learning_rate": 0.00016061865189289013,
+      "loss": 0.7803,
+      "mean_token_accuracy": 0.7706619620323181,
+      "num_tokens": 66758485.0,
+      "step": 19620
+    },
+    {
+      "epoch": 0.19701714239832993,
+      "learning_rate": 0.00016059857882693004,
+      "loss": 0.7465,
+      "mean_token_accuracy": 0.7719769418239594,
+      "num_tokens": 66792463.0,
+      "step": 19630
+    },
+    {
+      "epoch": 0.1971175077281304,
+      "learning_rate": 0.00016057850576096992,
+      "loss": 0.7843,
+      "mean_token_accuracy": 0.7723612189292908,
+      "num_tokens": 66827091.0,
+      "step": 19640
+    },
+    {
+      "epoch": 0.19721787305793087,
+      "learning_rate": 0.00016055843269500986,
+      "loss": 0.7754,
+      "mean_token_accuracy": 0.768305379152298,
+      "num_tokens": 66861157.0,
+      "step": 19650
+    },
+    {
+      "epoch": 0.19731823838773135,
+      "learning_rate": 0.00016053835962904974,
+      "loss": 0.7689,
+      "mean_token_accuracy": 0.773998099565506,
+      "num_tokens": 66895205.0,
+      "step": 19660
+    },
+    {
+      "epoch": 0.1974186037175318,
+      "learning_rate": 0.00016051828656308965,
+      "loss": 0.8079,
+      "mean_token_accuracy": 0.7620140492916108,
+      "num_tokens": 66928321.0,
+      "step": 19670
+    },
+    {
+      "epoch": 0.1975189690473323,
+      "learning_rate": 0.00016049821349712956,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.7748634397983551,
+      "num_tokens": 66962433.0,
+      "step": 19680
+    },
+    {
+      "epoch": 0.19761933437713278,
+      "learning_rate": 0.00016047814043116946,
+      "loss": 0.7832,
+      "mean_token_accuracy": 0.7628565311431885,
+      "num_tokens": 66996228.0,
+      "step": 19690
+    },
+    {
+      "epoch": 0.19771969970693323,
+      "learning_rate": 0.00016045806736520937,
+      "loss": 0.7924,
+      "mean_token_accuracy": 0.7635167181491852,
+      "num_tokens": 67030281.0,
+      "step": 19700
+    },
+    {
+      "epoch": 0.19782006503673372,
+      "learning_rate": 0.00016043799429924928,
+      "loss": 0.7802,
+      "mean_token_accuracy": 0.7688194334506988,
+      "num_tokens": 67063354.0,
+      "step": 19710
+    },
+    {
+      "epoch": 0.19792043036653417,
+      "learning_rate": 0.0001604179212332892,
+      "loss": 0.7444,
+      "mean_token_accuracy": 0.7749076604843139,
+      "num_tokens": 67098215.0,
+      "step": 19720
+    },
+    {
+      "epoch": 0.19802079569633466,
+      "learning_rate": 0.00016039784816732907,
+      "loss": 0.7757,
+      "mean_token_accuracy": 0.7693403422832489,
+      "num_tokens": 67132673.0,
+      "step": 19730
+    },
+    {
+      "epoch": 0.19812116102613514,
+      "learning_rate": 0.000160377775101369,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7703332483768464,
+      "num_tokens": 67166083.0,
+      "step": 19740
+    },
+    {
+      "epoch": 0.1982215263559356,
+      "learning_rate": 0.0001603577020354089,
+      "loss": 0.7883,
+      "mean_token_accuracy": 0.7647324919700622,
+      "num_tokens": 67199412.0,
+      "step": 19750
+    },
+    {
+      "epoch": 0.19832189168573608,
+      "learning_rate": 0.0001603376289694488,
+      "loss": 0.8099,
+      "mean_token_accuracy": 0.7631685435771942,
+      "num_tokens": 67234387.0,
+      "step": 19760
+    },
+    {
+      "epoch": 0.19842225701553656,
+      "learning_rate": 0.0001603175559034887,
+      "loss": 0.8004,
+      "mean_token_accuracy": 0.7648675858974456,
+      "num_tokens": 67268247.0,
+      "step": 19770
+    },
+    {
+      "epoch": 0.19852262234533702,
+      "learning_rate": 0.0001602974828375286,
+      "loss": 0.7787,
+      "mean_token_accuracy": 0.7660826325416565,
+      "num_tokens": 67301830.0,
+      "step": 19780
+    },
+    {
+      "epoch": 0.1986229876751375,
+      "learning_rate": 0.00016027740977156852,
+      "loss": 0.7433,
+      "mean_token_accuracy": 0.7775061130523682,
+      "num_tokens": 67336474.0,
+      "step": 19790
+    },
+    {
+      "epoch": 0.19872335300493799,
+      "learning_rate": 0.00016025733670560843,
+      "loss": 0.7207,
+      "mean_token_accuracy": 0.7795002400875092,
+      "num_tokens": 67369765.0,
+      "step": 19800
+    },
+    {
+      "epoch": 0.19882371833473844,
+      "learning_rate": 0.00016023726363964834,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7733273029327392,
+      "num_tokens": 67404040.0,
+      "step": 19810
+    },
+    {
+      "epoch": 0.19892408366453893,
+      "learning_rate": 0.00016021719057368822,
+      "loss": 0.7694,
+      "mean_token_accuracy": 0.7676566243171692,
+      "num_tokens": 67438489.0,
+      "step": 19820
+    },
+    {
+      "epoch": 0.1990244489943394,
+      "learning_rate": 0.00016019711750772816,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7730227828025817,
+      "num_tokens": 67472735.0,
+      "step": 19830
+    },
+    {
+      "epoch": 0.19912481432413986,
+      "learning_rate": 0.00016017704444176804,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.7656783998012543,
+      "num_tokens": 67506887.0,
+      "step": 19840
+    },
+    {
+      "epoch": 0.19922517965394035,
+      "learning_rate": 0.00016015697137580794,
+      "loss": 0.7802,
+      "mean_token_accuracy": 0.7670118987560273,
+      "num_tokens": 67540833.0,
+      "step": 19850
+    },
+    {
+      "epoch": 0.1993255449837408,
+      "learning_rate": 0.00016013689830984785,
+      "loss": 0.7698,
+      "mean_token_accuracy": 0.7682573080062867,
+      "num_tokens": 67575295.0,
+      "step": 19860
+    },
+    {
+      "epoch": 0.1994259103135413,
+      "learning_rate": 0.00016011682524388776,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.7683188498020173,
+      "num_tokens": 67610099.0,
+      "step": 19870
+    },
+    {
+      "epoch": 0.19952627564334177,
+      "learning_rate": 0.00016009675217792764,
+      "loss": 0.761,
+      "mean_token_accuracy": 0.7709443926811218,
+      "num_tokens": 67644359.0,
+      "step": 19880
+    },
+    {
+      "epoch": 0.19962664097314223,
+      "learning_rate": 0.00016007667911196758,
+      "loss": 0.7494,
+      "mean_token_accuracy": 0.7775312781333923,
+      "num_tokens": 67678947.0,
+      "step": 19890
+    },
+    {
+      "epoch": 0.1997270063029427,
+      "learning_rate": 0.00016005660604600746,
+      "loss": 0.7914,
+      "mean_token_accuracy": 0.7651115536689759,
+      "num_tokens": 67714030.0,
+      "step": 19900
+    },
+    {
+      "epoch": 0.1998273716327432,
+      "learning_rate": 0.00016003653298004737,
+      "loss": 0.7577,
+      "mean_token_accuracy": 0.7703204929828644,
+      "num_tokens": 67749045.0,
+      "step": 19910
+    },
+    {
+      "epoch": 0.19992773696254365,
+      "learning_rate": 0.0001600164599140873,
+      "loss": 0.739,
+      "mean_token_accuracy": 0.7767483949661255,
+      "num_tokens": 67782181.0,
+      "step": 19920
+    },
+    {
+      "epoch": 0.20002810229234413,
+      "learning_rate": 0.00015999638684812719,
+      "loss": 0.767,
+      "mean_token_accuracy": 0.7694602489471436,
+      "num_tokens": 67816160.0,
+      "step": 19930
+    },
+    {
+      "epoch": 0.20012846762214462,
+      "learning_rate": 0.0001599763137821671,
+      "loss": 0.7796,
+      "mean_token_accuracy": 0.7599666953086853,
+      "num_tokens": 67850447.0,
+      "step": 19940
+    },
+    {
+      "epoch": 0.20022883295194507,
+      "learning_rate": 0.000159956240716207,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.7668049991130829,
+      "num_tokens": 67883705.0,
+      "step": 19950
+    },
+    {
+      "epoch": 0.20032919828174556,
+      "learning_rate": 0.0001599361676502469,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7709909677505493,
+      "num_tokens": 67917374.0,
+      "step": 19960
+    },
+    {
+      "epoch": 0.20042956361154604,
+      "learning_rate": 0.0001599160945842868,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7740794658660889,
+      "num_tokens": 67951524.0,
+      "step": 19970
+    },
+    {
+      "epoch": 0.2005299289413465,
+      "learning_rate": 0.00015989602151832673,
+      "loss": 0.7665,
+      "mean_token_accuracy": 0.7670549333095551,
+      "num_tokens": 67985802.0,
+      "step": 19980
+    },
+    {
+      "epoch": 0.20063029427114698,
+      "learning_rate": 0.0001598759484523666,
+      "loss": 0.7429,
+      "mean_token_accuracy": 0.7795494019985199,
+      "num_tokens": 68019235.0,
+      "step": 19990
+    },
+    {
+      "epoch": 0.20073065960094744,
+      "learning_rate": 0.00015985587538640652,
+      "loss": 0.757,
+      "mean_token_accuracy": 0.7741480469703674,
+      "num_tokens": 68052749.0,
+      "step": 20000
+    },
+    {
+      "epoch": 0.20083102493074792,
+      "learning_rate": 0.00015983580232044643,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7676588833332062,
+      "num_tokens": 68087359.0,
+      "step": 20010
+    },
+    {
+      "epoch": 0.2009313902605484,
+      "learning_rate": 0.00015981572925448633,
+      "loss": 0.7994,
+      "mean_token_accuracy": 0.7637075424194336,
+      "num_tokens": 68121983.0,
+      "step": 20020
+    },
+    {
+      "epoch": 0.20103175559034886,
+      "learning_rate": 0.00015979565618852624,
+      "loss": 0.7415,
+      "mean_token_accuracy": 0.7765156745910644,
+      "num_tokens": 68156097.0,
+      "step": 20030
+    },
+    {
+      "epoch": 0.20113212092014934,
+      "learning_rate": 0.00015977558312256615,
+      "loss": 0.7706,
+      "mean_token_accuracy": 0.7695318102836609,
+      "num_tokens": 68190128.0,
+      "step": 20040
+    },
+    {
+      "epoch": 0.20123248624994983,
+      "learning_rate": 0.00015975551005660606,
+      "loss": 0.7692,
+      "mean_token_accuracy": 0.7701895475387573,
+      "num_tokens": 68222825.0,
+      "step": 20050
+    },
+    {
+      "epoch": 0.20133285157975028,
+      "learning_rate": 0.00015973543699064594,
+      "loss": 0.7545,
+      "mean_token_accuracy": 0.7765472412109375,
+      "num_tokens": 68256662.0,
+      "step": 20060
+    },
+    {
+      "epoch": 0.20143321690955077,
+      "learning_rate": 0.00015971536392468588,
+      "loss": 0.7519,
+      "mean_token_accuracy": 0.7768332183361053,
+      "num_tokens": 68290858.0,
+      "step": 20070
+    },
+    {
+      "epoch": 0.20153358223935125,
+      "learning_rate": 0.00015969529085872576,
+      "loss": 0.79,
+      "mean_token_accuracy": 0.7622937262058258,
+      "num_tokens": 68324739.0,
+      "step": 20080
+    },
+    {
+      "epoch": 0.2016339475691517,
+      "learning_rate": 0.0001596752177927657,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.775265508890152,
+      "num_tokens": 68359074.0,
+      "step": 20090
+    },
+    {
+      "epoch": 0.2017343128989522,
+      "learning_rate": 0.00015965514472680557,
+      "loss": 0.7679,
+      "mean_token_accuracy": 0.769006222486496,
+      "num_tokens": 68393197.0,
+      "step": 20100
+    },
+    {
+      "epoch": 0.20183467822875267,
+      "learning_rate": 0.00015963507166084548,
+      "loss": 0.7573,
+      "mean_token_accuracy": 0.7712887763977051,
+      "num_tokens": 68427457.0,
+      "step": 20110
+    },
+    {
+      "epoch": 0.20193504355855313,
+      "learning_rate": 0.0001596149985948854,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7706013262271881,
+      "num_tokens": 68462288.0,
+      "step": 20120
+    },
+    {
+      "epoch": 0.2020354088883536,
+      "learning_rate": 0.0001595949255289253,
+      "loss": 0.7678,
+      "mean_token_accuracy": 0.7678113400936126,
+      "num_tokens": 68496126.0,
+      "step": 20130
+    },
+    {
+      "epoch": 0.20213577421815407,
+      "learning_rate": 0.0001595748524629652,
+      "loss": 0.784,
+      "mean_token_accuracy": 0.7700542449951172,
+      "num_tokens": 68530714.0,
+      "step": 20140
+    },
+    {
+      "epoch": 0.20223613954795455,
+      "learning_rate": 0.00015955477939700512,
+      "loss": 0.7465,
+      "mean_token_accuracy": 0.7755346357822418,
+      "num_tokens": 68565073.0,
+      "step": 20150
+    },
+    {
+      "epoch": 0.20233650487775504,
+      "learning_rate": 0.00015953470633104503,
+      "loss": 0.7371,
+      "mean_token_accuracy": 0.779209417104721,
+      "num_tokens": 68598624.0,
+      "step": 20160
+    },
+    {
+      "epoch": 0.2024368702075555,
+      "learning_rate": 0.0001595146332650849,
+      "loss": 0.7631,
+      "mean_token_accuracy": 0.7787334799766541,
+      "num_tokens": 68632613.0,
+      "step": 20170
+    },
+    {
+      "epoch": 0.20253723553735598,
+      "learning_rate": 0.00015949456019912484,
+      "loss": 0.7452,
+      "mean_token_accuracy": 0.7790575742721557,
+      "num_tokens": 68666550.0,
+      "step": 20180
+    },
+    {
+      "epoch": 0.20263760086715646,
+      "learning_rate": 0.00015947448713316472,
+      "loss": 0.7786,
+      "mean_token_accuracy": 0.7660065054893493,
+      "num_tokens": 68700364.0,
+      "step": 20190
+    },
+    {
+      "epoch": 0.20273796619695691,
+      "learning_rate": 0.00015945441406720463,
+      "loss": 0.7645,
+      "mean_token_accuracy": 0.7723218679428101,
+      "num_tokens": 68734109.0,
+      "step": 20200
+    },
+    {
+      "epoch": 0.2028383315267574,
+      "learning_rate": 0.00015943434100124454,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7738436937332154,
+      "num_tokens": 68768553.0,
+      "step": 20210
+    },
+    {
+      "epoch": 0.20293869685655788,
+      "learning_rate": 0.00015941426793528445,
+      "loss": 0.7487,
+      "mean_token_accuracy": 0.7739782512187958,
+      "num_tokens": 68802618.0,
+      "step": 20220
+    },
+    {
+      "epoch": 0.20303906218635834,
+      "learning_rate": 0.00015939419486932433,
+      "loss": 0.7602,
+      "mean_token_accuracy": 0.7701372802257538,
+      "num_tokens": 68837329.0,
+      "step": 20230
+    },
+    {
+      "epoch": 0.20313942751615882,
+      "learning_rate": 0.00015937412180336427,
+      "loss": 0.7383,
+      "mean_token_accuracy": 0.7742105484008789,
+      "num_tokens": 68870823.0,
+      "step": 20240
+    },
+    {
+      "epoch": 0.2032397928459593,
+      "learning_rate": 0.00015935404873740415,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7661495864391327,
+      "num_tokens": 68905264.0,
+      "step": 20250
+    },
+    {
+      "epoch": 0.20334015817575976,
+      "learning_rate": 0.00015933397567144406,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7653391361236572,
+      "num_tokens": 68939106.0,
+      "step": 20260
+    },
+    {
+      "epoch": 0.20344052350556024,
+      "learning_rate": 0.000159313902605484,
+      "loss": 0.8013,
+      "mean_token_accuracy": 0.7635286152362823,
+      "num_tokens": 68973161.0,
+      "step": 20270
+    },
+    {
+      "epoch": 0.2035408888353607,
+      "learning_rate": 0.00015929382953952387,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7765633702278137,
+      "num_tokens": 69008120.0,
+      "step": 20280
+    },
+    {
+      "epoch": 0.20364125416516118,
+      "learning_rate": 0.00015927375647356378,
+      "loss": 0.7635,
+      "mean_token_accuracy": 0.7679502248764039,
+      "num_tokens": 69041517.0,
+      "step": 20290
+    },
+    {
+      "epoch": 0.20374161949496167,
+      "learning_rate": 0.0001592536834076037,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7693223118782043,
+      "num_tokens": 69075388.0,
+      "step": 20300
+    },
+    {
+      "epoch": 0.20384198482476212,
+      "learning_rate": 0.0001592336103416436,
+      "loss": 0.7792,
+      "mean_token_accuracy": 0.7693504750728607,
+      "num_tokens": 69109863.0,
+      "step": 20310
+    },
+    {
+      "epoch": 0.2039423501545626,
+      "learning_rate": 0.00015921353727568348,
+      "loss": 0.7868,
+      "mean_token_accuracy": 0.7634381890296936,
+      "num_tokens": 69144241.0,
+      "step": 20320
+    },
+    {
+      "epoch": 0.2040427154843631,
+      "learning_rate": 0.00015919346420972341,
+      "loss": 0.7538,
+      "mean_token_accuracy": 0.7786667466163635,
+      "num_tokens": 69178931.0,
+      "step": 20330
+    },
+    {
+      "epoch": 0.20414308081416355,
+      "learning_rate": 0.0001591733911437633,
+      "loss": 0.7668,
+      "mean_token_accuracy": 0.7745855212211609,
+      "num_tokens": 69212797.0,
+      "step": 20340
+    },
+    {
+      "epoch": 0.20424344614396403,
+      "learning_rate": 0.0001591533180778032,
+      "loss": 0.7896,
+      "mean_token_accuracy": 0.7649648904800415,
+      "num_tokens": 69245945.0,
+      "step": 20350
+    },
+    {
+      "epoch": 0.20434381147376451,
+      "learning_rate": 0.0001591332450118431,
+      "loss": 0.7594,
+      "mean_token_accuracy": 0.7718109905719757,
+      "num_tokens": 69279613.0,
+      "step": 20360
+    },
+    {
+      "epoch": 0.20444417680356497,
+      "learning_rate": 0.00015911317194588302,
+      "loss": 0.7308,
+      "mean_token_accuracy": 0.7783885061740875,
+      "num_tokens": 69313556.0,
+      "step": 20370
+    },
+    {
+      "epoch": 0.20454454213336545,
+      "learning_rate": 0.00015909309887992293,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7700179278850555,
+      "num_tokens": 69347036.0,
+      "step": 20380
+    },
+    {
+      "epoch": 0.20464490746316594,
+      "learning_rate": 0.00015907302581396284,
+      "loss": 0.7425,
+      "mean_token_accuracy": 0.7761937737464905,
+      "num_tokens": 69381310.0,
+      "step": 20390
+    },
+    {
+      "epoch": 0.2047452727929664,
+      "learning_rate": 0.00015905295274800275,
+      "loss": 0.7714,
+      "mean_token_accuracy": 0.7757660686969757,
+      "num_tokens": 69415250.0,
+      "step": 20400
+    },
+    {
+      "epoch": 0.20484563812276688,
+      "learning_rate": 0.00015903287968204263,
+      "loss": 0.7726,
+      "mean_token_accuracy": 0.7738589107990265,
+      "num_tokens": 69449151.0,
+      "step": 20410
+    },
+    {
+      "epoch": 0.20494600345256733,
+      "learning_rate": 0.00015901280661608256,
+      "loss": 0.7692,
+      "mean_token_accuracy": 0.7708847939968109,
+      "num_tokens": 69484146.0,
+      "step": 20420
+    },
+    {
+      "epoch": 0.20504636878236782,
+      "learning_rate": 0.00015899273355012244,
+      "loss": 0.7575,
+      "mean_token_accuracy": 0.7732176184654236,
+      "num_tokens": 69518455.0,
+      "step": 20430
+    },
+    {
+      "epoch": 0.2051467341121683,
+      "learning_rate": 0.00015897266048416235,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.7802337765693664,
+      "num_tokens": 69552681.0,
+      "step": 20440
+    },
+    {
+      "epoch": 0.20524709944196876,
+      "learning_rate": 0.00015895258741820226,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.7741704881191254,
+      "num_tokens": 69585877.0,
+      "step": 20450
+    },
+    {
+      "epoch": 0.20534746477176924,
+      "learning_rate": 0.00015893251435224217,
+      "loss": 0.7792,
+      "mean_token_accuracy": 0.7710673213005066,
+      "num_tokens": 69620188.0,
+      "step": 20460
+    },
+    {
+      "epoch": 0.20544783010156972,
+      "learning_rate": 0.00015891244128628208,
+      "loss": 0.7827,
+      "mean_token_accuracy": 0.7680623292922973,
+      "num_tokens": 69655389.0,
+      "step": 20470
+    },
+    {
+      "epoch": 0.20554819543137018,
+      "learning_rate": 0.000158892368220322,
+      "loss": 0.7646,
+      "mean_token_accuracy": 0.7719215154647827,
+      "num_tokens": 69688903.0,
+      "step": 20480
+    },
+    {
+      "epoch": 0.20564856076117066,
+      "learning_rate": 0.0001588722951543619,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.7672102153301239,
+      "num_tokens": 69723016.0,
+      "step": 20490
+    },
+    {
+      "epoch": 0.20574892609097115,
+      "learning_rate": 0.00015885222208840178,
+      "loss": 0.7832,
+      "mean_token_accuracy": 0.7706070244312286,
+      "num_tokens": 69757382.0,
+      "step": 20500
+    },
+    {
+      "epoch": 0.2058492914207716,
+      "learning_rate": 0.0001588321490224417,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7733456909656524,
+      "num_tokens": 69791715.0,
+      "step": 20510
+    },
+    {
+      "epoch": 0.20594965675057209,
+      "learning_rate": 0.0001588120759564816,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7724246740341186,
+      "num_tokens": 69826059.0,
+      "step": 20520
+    },
+    {
+      "epoch": 0.20605002208037257,
+      "learning_rate": 0.0001587920028905215,
+      "loss": 0.7458,
+      "mean_token_accuracy": 0.7753987908363342,
+      "num_tokens": 69860872.0,
+      "step": 20530
+    },
+    {
+      "epoch": 0.20615038741017302,
+      "learning_rate": 0.0001587719298245614,
+      "loss": 0.7804,
+      "mean_token_accuracy": 0.7636359870433808,
+      "num_tokens": 69895328.0,
+      "step": 20540
+    },
+    {
+      "epoch": 0.2062507527399735,
+      "learning_rate": 0.00015875185675860132,
+      "loss": 0.8072,
+      "mean_token_accuracy": 0.767308634519577,
+      "num_tokens": 69929541.0,
+      "step": 20550
+    },
+    {
+      "epoch": 0.20635111806977396,
+      "learning_rate": 0.0001587317836926412,
+      "loss": 0.7573,
+      "mean_token_accuracy": 0.7705384433269501,
+      "num_tokens": 69962917.0,
+      "step": 20560
+    },
+    {
+      "epoch": 0.20645148339957445,
+      "learning_rate": 0.00015871171062668114,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.7707615435123444,
+      "num_tokens": 69996444.0,
+      "step": 20570
+    },
+    {
+      "epoch": 0.20655184872937493,
+      "learning_rate": 0.00015869163756072102,
+      "loss": 0.7804,
+      "mean_token_accuracy": 0.7686228096485138,
+      "num_tokens": 70030414.0,
+      "step": 20580
+    },
+    {
+      "epoch": 0.2066522140591754,
+      "learning_rate": 0.00015867156449476093,
+      "loss": 0.7433,
+      "mean_token_accuracy": 0.7765547573566437,
+      "num_tokens": 70065297.0,
+      "step": 20590
+    },
+    {
+      "epoch": 0.20675257938897587,
+      "learning_rate": 0.00015865149142880083,
+      "loss": 0.7711,
+      "mean_token_accuracy": 0.7739952743053437,
+      "num_tokens": 70099550.0,
+      "step": 20600
+    },
+    {
+      "epoch": 0.20685294471877635,
+      "learning_rate": 0.00015863141836284074,
+      "loss": 0.7765,
+      "mean_token_accuracy": 0.7641831517219544,
+      "num_tokens": 70133719.0,
+      "step": 20610
+    },
+    {
+      "epoch": 0.2069533100485768,
+      "learning_rate": 0.00015861134529688065,
+      "loss": 0.7923,
+      "mean_token_accuracy": 0.7672351777553559,
+      "num_tokens": 70167631.0,
+      "step": 20620
+    },
+    {
+      "epoch": 0.2070536753783773,
+      "learning_rate": 0.00015859127223092056,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7753488838672637,
+      "num_tokens": 70202415.0,
+      "step": 20630
+    },
+    {
+      "epoch": 0.20715404070817778,
+      "learning_rate": 0.00015857119916496047,
+      "loss": 0.767,
+      "mean_token_accuracy": 0.7673940360546112,
+      "num_tokens": 70236336.0,
+      "step": 20640
+    },
+    {
+      "epoch": 0.20725440603797823,
+      "learning_rate": 0.00015855112609900038,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7706686735153199,
+      "num_tokens": 70269739.0,
+      "step": 20650
+    },
+    {
+      "epoch": 0.20735477136777872,
+      "learning_rate": 0.00015853105303304028,
+      "loss": 0.7129,
+      "mean_token_accuracy": 0.7834236145019531,
+      "num_tokens": 70303515.0,
+      "step": 20660
+    },
+    {
+      "epoch": 0.2074551366975792,
+      "learning_rate": 0.00015851097996708017,
+      "loss": 0.7369,
+      "mean_token_accuracy": 0.7737632632255554,
+      "num_tokens": 70336878.0,
+      "step": 20670
+    },
+    {
+      "epoch": 0.20755550202737966,
+      "learning_rate": 0.0001584909069011201,
+      "loss": 0.7726,
+      "mean_token_accuracy": 0.7719765722751617,
+      "num_tokens": 70370896.0,
+      "step": 20680
+    },
+    {
+      "epoch": 0.20765586735718014,
+      "learning_rate": 0.00015847083383515998,
+      "loss": 0.7577,
+      "mean_token_accuracy": 0.7750647962093353,
+      "num_tokens": 70404729.0,
+      "step": 20690
+    },
+    {
+      "epoch": 0.2077562326869806,
+      "learning_rate": 0.0001584507607691999,
+      "loss": 0.7725,
+      "mean_token_accuracy": 0.7715784668922424,
+      "num_tokens": 70439114.0,
+      "step": 20700
+    },
+    {
+      "epoch": 0.20785659801678108,
+      "learning_rate": 0.0001584306877032398,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.7785567104816437,
+      "num_tokens": 70472634.0,
+      "step": 20710
+    },
+    {
+      "epoch": 0.20795696334658156,
+      "learning_rate": 0.0001584106146372797,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.7646165311336517,
+      "num_tokens": 70506601.0,
+      "step": 20720
+    },
+    {
+      "epoch": 0.20805732867638202,
+      "learning_rate": 0.00015839054157131962,
+      "loss": 0.8001,
+      "mean_token_accuracy": 0.7662071764469147,
+      "num_tokens": 70540508.0,
+      "step": 20730
+    },
+    {
+      "epoch": 0.2081576940061825,
+      "learning_rate": 0.00015837046850535952,
+      "loss": 0.789,
+      "mean_token_accuracy": 0.7709681332111359,
+      "num_tokens": 70575172.0,
+      "step": 20740
+    },
+    {
+      "epoch": 0.208258059335983,
+      "learning_rate": 0.00015835039543939943,
+      "loss": 0.7394,
+      "mean_token_accuracy": 0.7774100303649902,
+      "num_tokens": 70609049.0,
+      "step": 20750
+    },
+    {
+      "epoch": 0.20835842466578344,
+      "learning_rate": 0.00015833032237343931,
+      "loss": 0.7373,
+      "mean_token_accuracy": 0.7811862468719483,
+      "num_tokens": 70643102.0,
+      "step": 20760
+    },
+    {
+      "epoch": 0.20845878999558393,
+      "learning_rate": 0.00015831024930747925,
+      "loss": 0.7404,
+      "mean_token_accuracy": 0.7779871642589569,
+      "num_tokens": 70676816.0,
+      "step": 20770
+    },
+    {
+      "epoch": 0.2085591553253844,
+      "learning_rate": 0.00015829017624151913,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7701539635658264,
+      "num_tokens": 70711046.0,
+      "step": 20780
+    },
+    {
+      "epoch": 0.20865952065518487,
+      "learning_rate": 0.00015827010317555904,
+      "loss": 0.7852,
+      "mean_token_accuracy": 0.764276772737503,
+      "num_tokens": 70744074.0,
+      "step": 20790
+    },
+    {
+      "epoch": 0.20875988598498535,
+      "learning_rate": 0.00015825003010959895,
+      "loss": 0.8007,
+      "mean_token_accuracy": 0.7678519129753113,
+      "num_tokens": 70777768.0,
+      "step": 20800
+    },
+    {
+      "epoch": 0.20886025131478583,
+      "learning_rate": 0.00015822995704363886,
+      "loss": 0.7754,
+      "mean_token_accuracy": 0.7766441702842712,
+      "num_tokens": 70810973.0,
+      "step": 20810
+    },
+    {
+      "epoch": 0.2089606166445863,
+      "learning_rate": 0.00015820988397767876,
+      "loss": 0.7906,
+      "mean_token_accuracy": 0.7657183766365051,
+      "num_tokens": 70844540.0,
+      "step": 20820
+    },
+    {
+      "epoch": 0.20906098197438677,
+      "learning_rate": 0.00015818981091171867,
+      "loss": 0.7371,
+      "mean_token_accuracy": 0.777416217327118,
+      "num_tokens": 70878634.0,
+      "step": 20830
+    },
+    {
+      "epoch": 0.20916134730418723,
+      "learning_rate": 0.00015816973784575858,
+      "loss": 0.7398,
+      "mean_token_accuracy": 0.7815571427345276,
+      "num_tokens": 70912370.0,
+      "step": 20840
+    },
+    {
+      "epoch": 0.2092617126339877,
+      "learning_rate": 0.00015814966477979846,
+      "loss": 0.7539,
+      "mean_token_accuracy": 0.7748958528041839,
+      "num_tokens": 70945512.0,
+      "step": 20850
+    },
+    {
+      "epoch": 0.2093620779637882,
+      "learning_rate": 0.0001581295917138384,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.7678313195705414,
+      "num_tokens": 70979404.0,
+      "step": 20860
+    },
+    {
+      "epoch": 0.20946244329358865,
+      "learning_rate": 0.00015810951864787828,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.7691116333007812,
+      "num_tokens": 71013184.0,
+      "step": 20870
+    },
+    {
+      "epoch": 0.20956280862338914,
+      "learning_rate": 0.0001580894455819182,
+      "loss": 0.7721,
+      "mean_token_accuracy": 0.7686834394931793,
+      "num_tokens": 71046758.0,
+      "step": 20880
+    },
+    {
+      "epoch": 0.20966317395318962,
+      "learning_rate": 0.0001580693725159581,
+      "loss": 0.7106,
+      "mean_token_accuracy": 0.7875912904739379,
+      "num_tokens": 71081019.0,
+      "step": 20890
+    },
+    {
+      "epoch": 0.20976353928299007,
+      "learning_rate": 0.000158049299449998,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7696977019309997,
+      "num_tokens": 71115195.0,
+      "step": 20900
+    },
+    {
+      "epoch": 0.20986390461279056,
+      "learning_rate": 0.0001580292263840379,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7728437840938568,
+      "num_tokens": 71148512.0,
+      "step": 20910
+    },
+    {
+      "epoch": 0.20996426994259104,
+      "learning_rate": 0.00015800915331807782,
+      "loss": 0.7693,
+      "mean_token_accuracy": 0.7716422200202941,
+      "num_tokens": 71182644.0,
+      "step": 20920
+    },
+    {
+      "epoch": 0.2100646352723915,
+      "learning_rate": 0.0001579890802521177,
+      "loss": 0.7809,
+      "mean_token_accuracy": 0.7700718164443969,
+      "num_tokens": 71216736.0,
+      "step": 20930
+    },
+    {
+      "epoch": 0.21016500060219198,
+      "learning_rate": 0.0001579690071861576,
+      "loss": 0.7831,
+      "mean_token_accuracy": 0.7644835889339447,
+      "num_tokens": 71251048.0,
+      "step": 20940
+    },
+    {
+      "epoch": 0.21026536593199247,
+      "learning_rate": 0.00015794893412019752,
+      "loss": 0.7408,
+      "mean_token_accuracy": 0.7768671214580536,
+      "num_tokens": 71285103.0,
+      "step": 20950
+    },
+    {
+      "epoch": 0.21036573126179292,
+      "learning_rate": 0.00015792886105423743,
+      "loss": 0.7712,
+      "mean_token_accuracy": 0.7657074928283691,
+      "num_tokens": 71319516.0,
+      "step": 20960
+    },
+    {
+      "epoch": 0.2104660965915934,
+      "learning_rate": 0.00015790878798827734,
+      "loss": 0.7708,
+      "mean_token_accuracy": 0.7705487966537475,
+      "num_tokens": 71352939.0,
+      "step": 20970
+    },
+    {
+      "epoch": 0.21056646192139386,
+      "learning_rate": 0.00015788871492231725,
+      "loss": 0.7799,
+      "mean_token_accuracy": 0.7726272165775299,
+      "num_tokens": 71388027.0,
+      "step": 20980
+    },
+    {
+      "epoch": 0.21066682725119434,
+      "learning_rate": 0.00015786864185635715,
+      "loss": 0.769,
+      "mean_token_accuracy": 0.7712588965892792,
+      "num_tokens": 71421342.0,
+      "step": 20990
+    },
+    {
+      "epoch": 0.21076719258099483,
+      "learning_rate": 0.00015784856879039704,
+      "loss": 0.7768,
+      "mean_token_accuracy": 0.7639996528625488,
+      "num_tokens": 71455351.0,
+      "step": 21000
+    },
+    {
+      "epoch": 0.21086755791079528,
+      "learning_rate": 0.00015782849572443697,
+      "loss": 0.7589,
+      "mean_token_accuracy": 0.7738116085529327,
+      "num_tokens": 71489546.0,
+      "step": 21010
+    },
+    {
+      "epoch": 0.21096792324059577,
+      "learning_rate": 0.00015780842265847685,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7762113988399506,
+      "num_tokens": 71522857.0,
+      "step": 21020
+    },
+    {
+      "epoch": 0.21106828857039625,
+      "learning_rate": 0.00015778834959251676,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7659790396690369,
+      "num_tokens": 71556864.0,
+      "step": 21030
+    },
+    {
+      "epoch": 0.2111686539001967,
+      "learning_rate": 0.00015776827652655667,
+      "loss": 0.749,
+      "mean_token_accuracy": 0.7743443191051483,
+      "num_tokens": 71590124.0,
+      "step": 21040
+    },
+    {
+      "epoch": 0.2112690192299972,
+      "learning_rate": 0.00015774820346059658,
+      "loss": 0.7793,
+      "mean_token_accuracy": 0.7672454655170441,
+      "num_tokens": 71625632.0,
+      "step": 21050
+    },
+    {
+      "epoch": 0.21136938455979767,
+      "learning_rate": 0.00015772813039463649,
+      "loss": 0.736,
+      "mean_token_accuracy": 0.7821826696395874,
+      "num_tokens": 71659529.0,
+      "step": 21060
+    },
+    {
+      "epoch": 0.21146974988959813,
+      "learning_rate": 0.0001577080573286764,
+      "loss": 0.7416,
+      "mean_token_accuracy": 0.780550765991211,
+      "num_tokens": 71693522.0,
+      "step": 21070
+    },
+    {
+      "epoch": 0.2115701152193986,
+      "learning_rate": 0.0001576879842627163,
+      "loss": 0.7656,
+      "mean_token_accuracy": 0.7747459173202514,
+      "num_tokens": 71727633.0,
+      "step": 21080
+    },
+    {
+      "epoch": 0.2116704805491991,
+      "learning_rate": 0.00015766791119675618,
+      "loss": 0.7822,
+      "mean_token_accuracy": 0.7633500158786773,
+      "num_tokens": 71761519.0,
+      "step": 21090
+    },
+    {
+      "epoch": 0.21177084587899955,
+      "learning_rate": 0.00015764783813079612,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.7736807346343995,
+      "num_tokens": 71795381.0,
+      "step": 21100
+    },
+    {
+      "epoch": 0.21187121120880004,
+      "learning_rate": 0.000157627765064836,
+      "loss": 0.7485,
+      "mean_token_accuracy": 0.7741257667541503,
+      "num_tokens": 71829951.0,
+      "step": 21110
+    },
+    {
+      "epoch": 0.2119715765386005,
+      "learning_rate": 0.0001576076919988759,
+      "loss": 0.7816,
+      "mean_token_accuracy": 0.7627883315086365,
+      "num_tokens": 71864229.0,
+      "step": 21120
+    },
+    {
+      "epoch": 0.21207194186840098,
+      "learning_rate": 0.00015758761893291582,
+      "loss": 0.769,
+      "mean_token_accuracy": 0.770053106546402,
+      "num_tokens": 71898017.0,
+      "step": 21130
+    },
+    {
+      "epoch": 0.21217230719820146,
+      "learning_rate": 0.00015756754586695573,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7741436541080475,
+      "num_tokens": 71932994.0,
+      "step": 21140
+    },
+    {
+      "epoch": 0.21227267252800192,
+      "learning_rate": 0.00015754747280099563,
+      "loss": 0.766,
+      "mean_token_accuracy": 0.7751873314380646,
+      "num_tokens": 71967639.0,
+      "step": 21150
+    },
+    {
+      "epoch": 0.2123730378578024,
+      "learning_rate": 0.00015752739973503554,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7707334399223328,
+      "num_tokens": 72001602.0,
+      "step": 21160
+    },
+    {
+      "epoch": 0.21247340318760288,
+      "learning_rate": 0.00015750732666907545,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.7694995701313019,
+      "num_tokens": 72035995.0,
+      "step": 21170
+    },
+    {
+      "epoch": 0.21257376851740334,
+      "learning_rate": 0.00015748725360311533,
+      "loss": 0.7566,
+      "mean_token_accuracy": 0.7757935225963593,
+      "num_tokens": 72070180.0,
+      "step": 21180
+    },
+    {
+      "epoch": 0.21267413384720382,
+      "learning_rate": 0.00015746718053715527,
+      "loss": 0.7736,
+      "mean_token_accuracy": 0.7682357907295227,
+      "num_tokens": 72104000.0,
+      "step": 21190
+    },
+    {
+      "epoch": 0.2127744991770043,
+      "learning_rate": 0.00015744710747119515,
+      "loss": 0.7167,
+      "mean_token_accuracy": 0.7844587206840515,
+      "num_tokens": 72138835.0,
+      "step": 21200
+    },
+    {
+      "epoch": 0.21287486450680476,
+      "learning_rate": 0.00015742703440523506,
+      "loss": 0.7839,
+      "mean_token_accuracy": 0.7647915065288544,
+      "num_tokens": 72172464.0,
+      "step": 21210
+    },
+    {
+      "epoch": 0.21297522983660525,
+      "learning_rate": 0.00015740696133927497,
+      "loss": 0.778,
+      "mean_token_accuracy": 0.7714803338050842,
+      "num_tokens": 72207750.0,
+      "step": 21220
+    },
+    {
+      "epoch": 0.21307559516640573,
+      "learning_rate": 0.00015738688827331488,
+      "loss": 0.7477,
+      "mean_token_accuracy": 0.7713371753692627,
+      "num_tokens": 72241130.0,
+      "step": 21230
+    },
+    {
+      "epoch": 0.21317596049620618,
+      "learning_rate": 0.00015736681520735478,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.7683083772659302,
+      "num_tokens": 72275096.0,
+      "step": 21240
+    },
+    {
+      "epoch": 0.21327632582600667,
+      "learning_rate": 0.0001573467421413947,
+      "loss": 0.7589,
+      "mean_token_accuracy": 0.7768821716308594,
+      "num_tokens": 72309015.0,
+      "step": 21250
+    },
+    {
+      "epoch": 0.21337669115580712,
+      "learning_rate": 0.00015732666907543457,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7699033319950104,
+      "num_tokens": 72343249.0,
+      "step": 21260
+    },
+    {
+      "epoch": 0.2134770564856076,
+      "learning_rate": 0.0001573065960094745,
+      "loss": 0.77,
+      "mean_token_accuracy": 0.771565067768097,
+      "num_tokens": 72377915.0,
+      "step": 21270
+    },
+    {
+      "epoch": 0.2135774218154081,
+      "learning_rate": 0.0001572865229435144,
+      "loss": 0.7458,
+      "mean_token_accuracy": 0.7759322106838227,
+      "num_tokens": 72410908.0,
+      "step": 21280
+    },
+    {
+      "epoch": 0.21367778714520855,
+      "learning_rate": 0.0001572664498775543,
+      "loss": 0.7901,
+      "mean_token_accuracy": 0.7675193965435028,
+      "num_tokens": 72445208.0,
+      "step": 21290
+    },
+    {
+      "epoch": 0.21377815247500903,
+      "learning_rate": 0.0001572463768115942,
+      "loss": 0.7449,
+      "mean_token_accuracy": 0.7791765749454498,
+      "num_tokens": 72479582.0,
+      "step": 21300
+    },
+    {
+      "epoch": 0.21387851780480951,
+      "learning_rate": 0.00015722630374563412,
+      "loss": 0.7656,
+      "mean_token_accuracy": 0.7676794767379761,
+      "num_tokens": 72513685.0,
+      "step": 21310
+    },
+    {
+      "epoch": 0.21397888313460997,
+      "learning_rate": 0.00015720623067967402,
+      "loss": 0.7499,
+      "mean_token_accuracy": 0.7740518152713776,
+      "num_tokens": 72547705.0,
+      "step": 21320
+    },
+    {
+      "epoch": 0.21407924846441045,
+      "learning_rate": 0.00015718615761371393,
+      "loss": 0.7408,
+      "mean_token_accuracy": 0.7749380826950073,
+      "num_tokens": 72581639.0,
+      "step": 21330
+    },
+    {
+      "epoch": 0.21417961379421094,
+      "learning_rate": 0.00015716608454775384,
+      "loss": 0.759,
+      "mean_token_accuracy": 0.7758741557598114,
+      "num_tokens": 72615597.0,
+      "step": 21340
+    },
+    {
+      "epoch": 0.2142799791240114,
+      "learning_rate": 0.00015714601148179372,
+      "loss": 0.7965,
+      "mean_token_accuracy": 0.7671333730220795,
+      "num_tokens": 72649300.0,
+      "step": 21350
+    },
+    {
+      "epoch": 0.21438034445381188,
+      "learning_rate": 0.00015712593841583366,
+      "loss": 0.7664,
+      "mean_token_accuracy": 0.7697389960289002,
+      "num_tokens": 72683135.0,
+      "step": 21360
+    },
+    {
+      "epoch": 0.21448070978361236,
+      "learning_rate": 0.00015710586534987354,
+      "loss": 0.742,
+      "mean_token_accuracy": 0.7795991182327271,
+      "num_tokens": 72717613.0,
+      "step": 21370
+    },
+    {
+      "epoch": 0.21458107511341282,
+      "learning_rate": 0.00015708579228391345,
+      "loss": 0.7424,
+      "mean_token_accuracy": 0.7717825472354889,
+      "num_tokens": 72751495.0,
+      "step": 21380
+    },
+    {
+      "epoch": 0.2146814404432133,
+      "learning_rate": 0.00015706571921795336,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.7756230950355529,
+      "num_tokens": 72785472.0,
+      "step": 21390
+    },
+    {
+      "epoch": 0.21478180577301376,
+      "learning_rate": 0.00015704564615199326,
+      "loss": 0.7617,
+      "mean_token_accuracy": 0.7696384608745575,
+      "num_tokens": 72819465.0,
+      "step": 21400
+    },
+    {
+      "epoch": 0.21488217110281424,
+      "learning_rate": 0.00015702557308603317,
+      "loss": 0.7665,
+      "mean_token_accuracy": 0.7731923341751099,
+      "num_tokens": 72852889.0,
+      "step": 21410
+    },
+    {
+      "epoch": 0.21498253643261472,
+      "learning_rate": 0.00015700550002007308,
+      "loss": 0.7428,
+      "mean_token_accuracy": 0.7707126796245575,
+      "num_tokens": 72886839.0,
+      "step": 21420
+    },
+    {
+      "epoch": 0.21508290176241518,
+      "learning_rate": 0.000156985426954113,
+      "loss": 0.7258,
+      "mean_token_accuracy": 0.7753783881664276,
+      "num_tokens": 72921136.0,
+      "step": 21430
+    },
+    {
+      "epoch": 0.21518326709221566,
+      "learning_rate": 0.00015696535388815287,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7778366923332214,
+      "num_tokens": 72955468.0,
+      "step": 21440
+    },
+    {
+      "epoch": 0.21528363242201615,
+      "learning_rate": 0.0001569452808221928,
+      "loss": 0.7584,
+      "mean_token_accuracy": 0.770528769493103,
+      "num_tokens": 72989801.0,
+      "step": 21450
+    },
+    {
+      "epoch": 0.2153839977518166,
+      "learning_rate": 0.0001569252077562327,
+      "loss": 0.7724,
+      "mean_token_accuracy": 0.7657631397247314,
+      "num_tokens": 73023755.0,
+      "step": 21460
+    },
+    {
+      "epoch": 0.2154843630816171,
+      "learning_rate": 0.0001569051346902726,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7705927014350891,
+      "num_tokens": 73057480.0,
+      "step": 21470
+    },
+    {
+      "epoch": 0.21558472841141757,
+      "learning_rate": 0.0001568850616243125,
+      "loss": 0.7788,
+      "mean_token_accuracy": 0.7675239384174347,
+      "num_tokens": 73091640.0,
+      "step": 21480
+    },
+    {
+      "epoch": 0.21568509374121803,
+      "learning_rate": 0.0001568649885583524,
+      "loss": 0.7724,
+      "mean_token_accuracy": 0.7599137783050537,
+      "num_tokens": 73125111.0,
+      "step": 21490
+    },
+    {
+      "epoch": 0.2157854590710185,
+      "learning_rate": 0.0001568449154923923,
+      "loss": 0.7513,
+      "mean_token_accuracy": 0.7712138295173645,
+      "num_tokens": 73159105.0,
+      "step": 21500
+    },
+    {
+      "epoch": 0.215885824400819,
+      "learning_rate": 0.00015682484242643223,
+      "loss": 0.7356,
+      "mean_token_accuracy": 0.7795074224472046,
+      "num_tokens": 73192588.0,
+      "step": 21510
+    },
+    {
+      "epoch": 0.21598618973061945,
+      "learning_rate": 0.00015680476936047214,
+      "loss": 0.7927,
+      "mean_token_accuracy": 0.7678205966949463,
+      "num_tokens": 73225956.0,
+      "step": 21520
+    },
+    {
+      "epoch": 0.21608655506041993,
+      "learning_rate": 0.00015678469629451202,
+      "loss": 0.7737,
+      "mean_token_accuracy": 0.7745775699615478,
+      "num_tokens": 73259268.0,
+      "step": 21530
+    },
+    {
+      "epoch": 0.2161869203902204,
+      "learning_rate": 0.00015676462322855196,
+      "loss": 0.7183,
+      "mean_token_accuracy": 0.7816416978836059,
+      "num_tokens": 73293677.0,
+      "step": 21540
+    },
+    {
+      "epoch": 0.21628728572002087,
+      "learning_rate": 0.00015674455016259184,
+      "loss": 0.7924,
+      "mean_token_accuracy": 0.7706844091415406,
+      "num_tokens": 73327223.0,
+      "step": 21550
+    },
+    {
+      "epoch": 0.21638765104982136,
+      "learning_rate": 0.00015672447709663175,
+      "loss": 0.7988,
+      "mean_token_accuracy": 0.7607482850551606,
+      "num_tokens": 73361275.0,
+      "step": 21560
+    },
+    {
+      "epoch": 0.2164880163796218,
+      "learning_rate": 0.00015670440403067165,
+      "loss": 0.7277,
+      "mean_token_accuracy": 0.7849364280700684,
+      "num_tokens": 73395290.0,
+      "step": 21570
+    },
+    {
+      "epoch": 0.2165883817094223,
+      "learning_rate": 0.00015668433096471156,
+      "loss": 0.7582,
+      "mean_token_accuracy": 0.77339026927948,
+      "num_tokens": 73429209.0,
+      "step": 21580
+    },
+    {
+      "epoch": 0.21668874703922278,
+      "learning_rate": 0.00015666425789875144,
+      "loss": 0.766,
+      "mean_token_accuracy": 0.7696726858615875,
+      "num_tokens": 73463512.0,
+      "step": 21590
+    },
+    {
+      "epoch": 0.21678911236902323,
+      "learning_rate": 0.00015664418483279138,
+      "loss": 0.752,
+      "mean_token_accuracy": 0.7735897541046143,
+      "num_tokens": 73497731.0,
+      "step": 21600
+    },
+    {
+      "epoch": 0.21688947769882372,
+      "learning_rate": 0.00015662411176683126,
+      "loss": 0.7642,
+      "mean_token_accuracy": 0.7691661059856415,
+      "num_tokens": 73532012.0,
+      "step": 21610
+    },
+    {
+      "epoch": 0.2169898430286242,
+      "learning_rate": 0.00015660403870087117,
+      "loss": 0.7731,
+      "mean_token_accuracy": 0.7705914676189423,
+      "num_tokens": 73566607.0,
+      "step": 21620
+    },
+    {
+      "epoch": 0.21709020835842466,
+      "learning_rate": 0.00015658396563491108,
+      "loss": 0.7882,
+      "mean_token_accuracy": 0.7648298919200898,
+      "num_tokens": 73601959.0,
+      "step": 21630
+    },
+    {
+      "epoch": 0.21719057368822514,
+      "learning_rate": 0.00015656389256895099,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7717965483665467,
+      "num_tokens": 73635868.0,
+      "step": 21640
+    },
+    {
+      "epoch": 0.21729093901802563,
+      "learning_rate": 0.0001565438195029909,
+      "loss": 0.7722,
+      "mean_token_accuracy": 0.770391458272934,
+      "num_tokens": 73669533.0,
+      "step": 21650
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "learning_rate": 0.0001565237464370308,
+      "loss": 0.7659,
+      "mean_token_accuracy": 0.7714470684528351,
+      "num_tokens": 73704028.0,
+      "step": 21660
+    },
+    {
+      "epoch": 0.21749166967762656,
+      "learning_rate": 0.0001565036733710707,
+      "loss": 0.7889,
+      "mean_token_accuracy": 0.7684951424598694,
+      "num_tokens": 73738628.0,
+      "step": 21670
+    },
+    {
+      "epoch": 0.21759203500742705,
+      "learning_rate": 0.0001564836003051106,
+      "loss": 0.7517,
+      "mean_token_accuracy": 0.7769899666309357,
+      "num_tokens": 73772208.0,
+      "step": 21680
+    },
+    {
+      "epoch": 0.2176924003372275,
+      "learning_rate": 0.00015646352723915053,
+      "loss": 0.7497,
+      "mean_token_accuracy": 0.7768558323383331,
+      "num_tokens": 73806372.0,
+      "step": 21690
+    },
+    {
+      "epoch": 0.217792765667028,
+      "learning_rate": 0.0001564434541731904,
+      "loss": 0.7573,
+      "mean_token_accuracy": 0.7735520601272583,
+      "num_tokens": 73840098.0,
+      "step": 21700
+    },
+    {
+      "epoch": 0.21789313099682844,
+      "learning_rate": 0.00015642338110723032,
+      "loss": 0.7249,
+      "mean_token_accuracy": 0.781202632188797,
+      "num_tokens": 73873739.0,
+      "step": 21710
+    },
+    {
+      "epoch": 0.21799349632662893,
+      "learning_rate": 0.00015640330804127023,
+      "loss": 0.7925,
+      "mean_token_accuracy": 0.7633719384670258,
+      "num_tokens": 73907501.0,
+      "step": 21720
+    },
+    {
+      "epoch": 0.2180938616564294,
+      "learning_rate": 0.00015638323497531013,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7692427694797516,
+      "num_tokens": 73941303.0,
+      "step": 21730
+    },
+    {
+      "epoch": 0.21819422698622987,
+      "learning_rate": 0.00015636316190935004,
+      "loss": 0.8132,
+      "mean_token_accuracy": 0.7652038037776947,
+      "num_tokens": 73975219.0,
+      "step": 21740
+    },
+    {
+      "epoch": 0.21829459231603035,
+      "learning_rate": 0.00015634308884338995,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7756443738937377,
+      "num_tokens": 74009478.0,
+      "step": 21750
+    },
+    {
+      "epoch": 0.21839495764583083,
+      "learning_rate": 0.00015632301577742986,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7766941070556641,
+      "num_tokens": 74043682.0,
+      "step": 21760
+    },
+    {
+      "epoch": 0.2184953229756313,
+      "learning_rate": 0.00015630294271146974,
+      "loss": 0.7673,
+      "mean_token_accuracy": 0.7681203067302704,
+      "num_tokens": 74077408.0,
+      "step": 21770
+    },
+    {
+      "epoch": 0.21859568830543177,
+      "learning_rate": 0.00015628286964550968,
+      "loss": 0.7186,
+      "mean_token_accuracy": 0.7839077115058899,
+      "num_tokens": 74111317.0,
+      "step": 21780
+    },
+    {
+      "epoch": 0.21869605363523226,
+      "learning_rate": 0.00015626279657954956,
+      "loss": 0.771,
+      "mean_token_accuracy": 0.7729612588882446,
+      "num_tokens": 74145562.0,
+      "step": 21790
+    },
+    {
+      "epoch": 0.2187964189650327,
+      "learning_rate": 0.0001562427235135895,
+      "loss": 0.7508,
+      "mean_token_accuracy": 0.7716224253177643,
+      "num_tokens": 74179412.0,
+      "step": 21800
+    },
+    {
+      "epoch": 0.2188967842948332,
+      "learning_rate": 0.00015622265044762937,
+      "loss": 0.7946,
+      "mean_token_accuracy": 0.7600009620189667,
+      "num_tokens": 74212780.0,
+      "step": 21810
+    },
+    {
+      "epoch": 0.21899714962463368,
+      "learning_rate": 0.00015620257738166928,
+      "loss": 0.7515,
+      "mean_token_accuracy": 0.7766862094402314,
+      "num_tokens": 74246670.0,
+      "step": 21820
+    },
+    {
+      "epoch": 0.21909751495443414,
+      "learning_rate": 0.0001561825043157092,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7634067952632904,
+      "num_tokens": 74280425.0,
+      "step": 21830
+    },
+    {
+      "epoch": 0.21919788028423462,
+      "learning_rate": 0.0001561624312497491,
+      "loss": 0.7571,
+      "mean_token_accuracy": 0.772942304611206,
+      "num_tokens": 74314451.0,
+      "step": 21840
+    },
+    {
+      "epoch": 0.21929824561403508,
+      "learning_rate": 0.00015614235818378898,
+      "loss": 0.8118,
+      "mean_token_accuracy": 0.7635380804538727,
+      "num_tokens": 74348625.0,
+      "step": 21850
+    },
+    {
+      "epoch": 0.21939861094383556,
+      "learning_rate": 0.00015612228511782892,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.7696048498153687,
+      "num_tokens": 74382687.0,
+      "step": 21860
+    },
+    {
+      "epoch": 0.21949897627363604,
+      "learning_rate": 0.00015610221205186883,
+      "loss": 0.7441,
+      "mean_token_accuracy": 0.7728609502315521,
+      "num_tokens": 74416683.0,
+      "step": 21870
+    },
+    {
+      "epoch": 0.2195993416034365,
+      "learning_rate": 0.0001560821389859087,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.7727586686611175,
+      "num_tokens": 74450751.0,
+      "step": 21880
+    },
+    {
+      "epoch": 0.21969970693323698,
+      "learning_rate": 0.00015606206591994864,
+      "loss": 0.749,
+      "mean_token_accuracy": 0.7788787484169006,
+      "num_tokens": 74484170.0,
+      "step": 21890
+    },
+    {
+      "epoch": 0.21980007226303747,
+      "learning_rate": 0.00015604199285398852,
+      "loss": 0.7495,
+      "mean_token_accuracy": 0.7776457905769348,
+      "num_tokens": 74518553.0,
+      "step": 21900
+    },
+    {
+      "epoch": 0.21990043759283792,
+      "learning_rate": 0.00015602191978802843,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.7708371758460999,
+      "num_tokens": 74553087.0,
+      "step": 21910
+    },
+    {
+      "epoch": 0.2200008029226384,
+      "learning_rate": 0.00015600184672206834,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.7738921821117402,
+      "num_tokens": 74587675.0,
+      "step": 21920
+    },
+    {
+      "epoch": 0.2201011682524389,
+      "learning_rate": 0.00015598177365610825,
+      "loss": 0.7449,
+      "mean_token_accuracy": 0.7725287139415741,
+      "num_tokens": 74621433.0,
+      "step": 21930
+    },
+    {
+      "epoch": 0.22020153358223934,
+      "learning_rate": 0.00015596170059014813,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.7794537603855133,
+      "num_tokens": 74655139.0,
+      "step": 21940
+    },
+    {
+      "epoch": 0.22030189891203983,
+      "learning_rate": 0.00015594162752418807,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7764513313770294,
+      "num_tokens": 74689015.0,
+      "step": 21950
+    },
+    {
+      "epoch": 0.2204022642418403,
+      "learning_rate": 0.00015592155445822795,
+      "loss": 0.7209,
+      "mean_token_accuracy": 0.7809753298759461,
+      "num_tokens": 74722946.0,
+      "step": 21960
+    },
+    {
+      "epoch": 0.22050262957164077,
+      "learning_rate": 0.00015590148139226786,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.7701947748661041,
+      "num_tokens": 74756312.0,
+      "step": 21970
+    },
+    {
+      "epoch": 0.22060299490144125,
+      "learning_rate": 0.00015588140832630776,
+      "loss": 0.7618,
+      "mean_token_accuracy": 0.7662958800792694,
+      "num_tokens": 74790810.0,
+      "step": 21980
+    },
+    {
+      "epoch": 0.2207033602312417,
+      "learning_rate": 0.00015586133526034767,
+      "loss": 0.7343,
+      "mean_token_accuracy": 0.7780974447727204,
+      "num_tokens": 74825050.0,
+      "step": 21990
+    },
+    {
+      "epoch": 0.2208037255610422,
+      "learning_rate": 0.00015584126219438758,
+      "loss": 0.7648,
+      "mean_token_accuracy": 0.7687854588031768,
+      "num_tokens": 74858386.0,
+      "step": 22000
+    },
+    {
+      "epoch": 0.22090409089084267,
+      "learning_rate": 0.0001558211891284275,
+      "loss": 0.7859,
+      "mean_token_accuracy": 0.7641824066638947,
+      "num_tokens": 74892002.0,
+      "step": 22010
+    },
+    {
+      "epoch": 0.22100445622064313,
+      "learning_rate": 0.0001558011160624674,
+      "loss": 0.7778,
+      "mean_token_accuracy": 0.7700320959091187,
+      "num_tokens": 74926198.0,
+      "step": 22020
+    },
+    {
+      "epoch": 0.22110482155044361,
+      "learning_rate": 0.00015578104299650728,
+      "loss": 0.7513,
+      "mean_token_accuracy": 0.77521892786026,
+      "num_tokens": 74960848.0,
+      "step": 22030
+    },
+    {
+      "epoch": 0.2212051868802441,
+      "learning_rate": 0.00015576096993054721,
+      "loss": 0.7329,
+      "mean_token_accuracy": 0.7798405706882476,
+      "num_tokens": 74994390.0,
+      "step": 22040
+    },
+    {
+      "epoch": 0.22130555221004455,
+      "learning_rate": 0.0001557408968645871,
+      "loss": 0.7424,
+      "mean_token_accuracy": 0.7813081204891205,
+      "num_tokens": 75028365.0,
+      "step": 22050
+    },
+    {
+      "epoch": 0.22140591753984504,
+      "learning_rate": 0.000155720823798627,
+      "loss": 0.743,
+      "mean_token_accuracy": 0.7755156219005584,
+      "num_tokens": 75062163.0,
+      "step": 22060
+    },
+    {
+      "epoch": 0.22150628286964552,
+      "learning_rate": 0.0001557007507326669,
+      "loss": 0.7459,
+      "mean_token_accuracy": 0.7786991477012635,
+      "num_tokens": 75095932.0,
+      "step": 22070
+    },
+    {
+      "epoch": 0.22160664819944598,
+      "learning_rate": 0.00015568067766670682,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7725139379501342,
+      "num_tokens": 75130192.0,
+      "step": 22080
+    },
+    {
+      "epoch": 0.22170701352924646,
+      "learning_rate": 0.00015566060460074673,
+      "loss": 0.7378,
+      "mean_token_accuracy": 0.7728904247283935,
+      "num_tokens": 75163781.0,
+      "step": 22090
+    },
+    {
+      "epoch": 0.22180737885904694,
+      "learning_rate": 0.00015564053153478664,
+      "loss": 0.7264,
+      "mean_token_accuracy": 0.7793911337852478,
+      "num_tokens": 75197120.0,
+      "step": 22100
+    },
+    {
+      "epoch": 0.2219077441888474,
+      "learning_rate": 0.00015562045846882655,
+      "loss": 0.7322,
+      "mean_token_accuracy": 0.7795130431652069,
+      "num_tokens": 75230906.0,
+      "step": 22110
+    },
+    {
+      "epoch": 0.22200810951864788,
+      "learning_rate": 0.00015560038540286643,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.7698107719421386,
+      "num_tokens": 75264321.0,
+      "step": 22120
+    },
+    {
+      "epoch": 0.22210847484844834,
+      "learning_rate": 0.00015558031233690636,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.7709541559219361,
+      "num_tokens": 75298107.0,
+      "step": 22130
+    },
+    {
+      "epoch": 0.22220884017824882,
+      "learning_rate": 0.00015556023927094624,
+      "loss": 0.7832,
+      "mean_token_accuracy": 0.7653647840023041,
+      "num_tokens": 75332461.0,
+      "step": 22140
+    },
+    {
+      "epoch": 0.2223092055080493,
+      "learning_rate": 0.00015554016620498615,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7696703612804413,
+      "num_tokens": 75366882.0,
+      "step": 22150
+    },
+    {
+      "epoch": 0.22240957083784976,
+      "learning_rate": 0.00015552009313902606,
+      "loss": 0.7857,
+      "mean_token_accuracy": 0.7655081987380982,
+      "num_tokens": 75401726.0,
+      "step": 22160
+    },
+    {
+      "epoch": 0.22250993616765025,
+      "learning_rate": 0.00015550002007306597,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7751639306545257,
+      "num_tokens": 75435612.0,
+      "step": 22170
+    },
+    {
+      "epoch": 0.22261030149745073,
+      "learning_rate": 0.00015547994700710585,
+      "loss": 0.7773,
+      "mean_token_accuracy": 0.771637785434723,
+      "num_tokens": 75469010.0,
+      "step": 22180
+    },
+    {
+      "epoch": 0.22271066682725119,
+      "learning_rate": 0.0001554598739411458,
+      "loss": 0.7459,
+      "mean_token_accuracy": 0.7756258308887481,
+      "num_tokens": 75502834.0,
+      "step": 22190
+    },
+    {
+      "epoch": 0.22281103215705167,
+      "learning_rate": 0.00015543980087518567,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7758987307548523,
+      "num_tokens": 75537291.0,
+      "step": 22200
+    },
+    {
+      "epoch": 0.22291139748685215,
+      "learning_rate": 0.00015541972780922558,
+      "loss": 0.7639,
+      "mean_token_accuracy": 0.7698519408702851,
+      "num_tokens": 75570830.0,
+      "step": 22210
+    },
+    {
+      "epoch": 0.2230117628166526,
+      "learning_rate": 0.0001553996547432655,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.7741851449012757,
+      "num_tokens": 75604702.0,
+      "step": 22220
+    },
+    {
+      "epoch": 0.2231121281464531,
+      "learning_rate": 0.0001553795816773054,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.7693216919898986,
+      "num_tokens": 75638956.0,
+      "step": 22230
+    },
+    {
+      "epoch": 0.22321249347625358,
+      "learning_rate": 0.0001553595086113453,
+      "loss": 0.7907,
+      "mean_token_accuracy": 0.7661739885807037,
+      "num_tokens": 75673235.0,
+      "step": 22240
+    },
+    {
+      "epoch": 0.22331285880605403,
+      "learning_rate": 0.0001553394355453852,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.776247090101242,
+      "num_tokens": 75707292.0,
+      "step": 22250
+    },
+    {
+      "epoch": 0.22341322413585452,
+      "learning_rate": 0.00015531936247942512,
+      "loss": 0.7721,
+      "mean_token_accuracy": 0.7690892219543457,
+      "num_tokens": 75741060.0,
+      "step": 22260
+    },
+    {
+      "epoch": 0.22351358946565497,
+      "learning_rate": 0.000155299289413465,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.7747559785842896,
+      "num_tokens": 75775464.0,
+      "step": 22270
+    },
+    {
+      "epoch": 0.22361395479545546,
+      "learning_rate": 0.00015527921634750494,
+      "loss": 0.7736,
+      "mean_token_accuracy": 0.769670695066452,
+      "num_tokens": 75810173.0,
+      "step": 22280
+    },
+    {
+      "epoch": 0.22371432012525594,
+      "learning_rate": 0.00015525914328154482,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.7710635662078857,
+      "num_tokens": 75844231.0,
+      "step": 22290
+    },
+    {
+      "epoch": 0.2238146854550564,
+      "learning_rate": 0.00015523907021558473,
+      "loss": 0.7926,
+      "mean_token_accuracy": 0.7665982961654663,
+      "num_tokens": 75878387.0,
+      "step": 22300
+    },
+    {
+      "epoch": 0.22391505078485688,
+      "learning_rate": 0.00015521899714962463,
+      "loss": 0.7763,
+      "mean_token_accuracy": 0.7694206535816193,
+      "num_tokens": 75912285.0,
+      "step": 22310
+    },
+    {
+      "epoch": 0.22401541611465736,
+      "learning_rate": 0.00015519892408366454,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7772053599357605,
+      "num_tokens": 75946814.0,
+      "step": 22320
+    },
+    {
+      "epoch": 0.22411578144445782,
+      "learning_rate": 0.00015517885101770445,
+      "loss": 0.7324,
+      "mean_token_accuracy": 0.7818355023860931,
+      "num_tokens": 75981088.0,
+      "step": 22330
+    },
+    {
+      "epoch": 0.2242161467742583,
+      "learning_rate": 0.00015515877795174436,
+      "loss": 0.76,
+      "mean_token_accuracy": 0.7736691057682037,
+      "num_tokens": 76015104.0,
+      "step": 22340
+    },
+    {
+      "epoch": 0.22431651210405879,
+      "learning_rate": 0.00015513870488578427,
+      "loss": 0.7855,
+      "mean_token_accuracy": 0.7721585512161255,
+      "num_tokens": 76049101.0,
+      "step": 22350
+    },
+    {
+      "epoch": 0.22441687743385924,
+      "learning_rate": 0.00015511863181982418,
+      "loss": 0.8257,
+      "mean_token_accuracy": 0.7583268821239472,
+      "num_tokens": 76082977.0,
+      "step": 22360
+    },
+    {
+      "epoch": 0.22451724276365972,
+      "learning_rate": 0.00015509855875386408,
+      "loss": 0.736,
+      "mean_token_accuracy": 0.7773957967758178,
+      "num_tokens": 76116739.0,
+      "step": 22370
+    },
+    {
+      "epoch": 0.2246176080934602,
+      "learning_rate": 0.00015507848568790397,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.7722367703914642,
+      "num_tokens": 76151238.0,
+      "step": 22380
+    },
+    {
+      "epoch": 0.22471797342326066,
+      "learning_rate": 0.0001550584126219439,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.7752468824386597,
+      "num_tokens": 76185106.0,
+      "step": 22390
+    },
+    {
+      "epoch": 0.22481833875306115,
+      "learning_rate": 0.00015503833955598378,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.7812665343284607,
+      "num_tokens": 76219283.0,
+      "step": 22400
+    },
+    {
+      "epoch": 0.2249187040828616,
+      "learning_rate": 0.0001550182664900237,
+      "loss": 0.7457,
+      "mean_token_accuracy": 0.7715763866901397,
+      "num_tokens": 76254503.0,
+      "step": 22410
+    },
+    {
+      "epoch": 0.2250190694126621,
+      "learning_rate": 0.0001549981934240636,
+      "loss": 0.7923,
+      "mean_token_accuracy": 0.7659672141075134,
+      "num_tokens": 76288084.0,
+      "step": 22420
+    },
+    {
+      "epoch": 0.22511943474246257,
+      "learning_rate": 0.0001549781203581035,
+      "loss": 0.7366,
+      "mean_token_accuracy": 0.7778996825218201,
+      "num_tokens": 76322732.0,
+      "step": 22430
+    },
+    {
+      "epoch": 0.22521980007226303,
+      "learning_rate": 0.00015495804729214342,
+      "loss": 0.7949,
+      "mean_token_accuracy": 0.7588162422180176,
+      "num_tokens": 76357108.0,
+      "step": 22440
+    },
+    {
+      "epoch": 0.2253201654020635,
+      "learning_rate": 0.00015493797422618332,
+      "loss": 0.7349,
+      "mean_token_accuracy": 0.7827002227306366,
+      "num_tokens": 76391121.0,
+      "step": 22450
+    },
+    {
+      "epoch": 0.225420530731864,
+      "learning_rate": 0.00015491790116022323,
+      "loss": 0.7445,
+      "mean_token_accuracy": 0.7849261403083801,
+      "num_tokens": 76425458.0,
+      "step": 22460
+    },
+    {
+      "epoch": 0.22552089606166445,
+      "learning_rate": 0.00015489782809426311,
+      "loss": 0.7508,
+      "mean_token_accuracy": 0.7736548244953155,
+      "num_tokens": 76458851.0,
+      "step": 22470
+    },
+    {
+      "epoch": 0.22562126139146493,
+      "learning_rate": 0.00015487775502830305,
+      "loss": 0.764,
+      "mean_token_accuracy": 0.7733363151550293,
+      "num_tokens": 76492919.0,
+      "step": 22480
+    },
+    {
+      "epoch": 0.22572162672126542,
+      "learning_rate": 0.00015485768196234293,
+      "loss": 0.7402,
+      "mean_token_accuracy": 0.7776598811149598,
+      "num_tokens": 76526610.0,
+      "step": 22490
+    },
+    {
+      "epoch": 0.22582199205106587,
+      "learning_rate": 0.00015483760889638284,
+      "loss": 0.769,
+      "mean_token_accuracy": 0.7713018953800201,
+      "num_tokens": 76560058.0,
+      "step": 22500
+    },
+    {
+      "epoch": 0.22592235738086636,
+      "learning_rate": 0.00015481753583042275,
+      "loss": 0.756,
+      "mean_token_accuracy": 0.7778894782066346,
+      "num_tokens": 76593949.0,
+      "step": 22510
+    },
+    {
+      "epoch": 0.22602272271066684,
+      "learning_rate": 0.00015479746276446266,
+      "loss": 0.7392,
+      "mean_token_accuracy": 0.7810545206069947,
+      "num_tokens": 76627038.0,
+      "step": 22520
+    },
+    {
+      "epoch": 0.2261230880404673,
+      "learning_rate": 0.00015477738969850254,
+      "loss": 0.7421,
+      "mean_token_accuracy": 0.7785213947296142,
+      "num_tokens": 76660999.0,
+      "step": 22530
+    },
+    {
+      "epoch": 0.22622345337026778,
+      "learning_rate": 0.00015475731663254247,
+      "loss": 0.7655,
+      "mean_token_accuracy": 0.7720098912715911,
+      "num_tokens": 76695646.0,
+      "step": 22540
+    },
+    {
+      "epoch": 0.22632381870006824,
+      "learning_rate": 0.00015473724356658236,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.7670884728431702,
+      "num_tokens": 76729311.0,
+      "step": 22550
+    },
+    {
+      "epoch": 0.22642418402986872,
+      "learning_rate": 0.00015471717050062226,
+      "loss": 0.7619,
+      "mean_token_accuracy": 0.7754532396793365,
+      "num_tokens": 76762880.0,
+      "step": 22560
+    },
+    {
+      "epoch": 0.2265245493596692,
+      "learning_rate": 0.0001546970974346622,
+      "loss": 0.7963,
+      "mean_token_accuracy": 0.7628134310245513,
+      "num_tokens": 76796916.0,
+      "step": 22570
+    },
+    {
+      "epoch": 0.22662491468946966,
+      "learning_rate": 0.00015467702436870208,
+      "loss": 0.7719,
+      "mean_token_accuracy": 0.765484744310379,
+      "num_tokens": 76831375.0,
+      "step": 22580
+    },
+    {
+      "epoch": 0.22672528001927014,
+      "learning_rate": 0.000154656951302742,
+      "loss": 0.7535,
+      "mean_token_accuracy": 0.7687456011772156,
+      "num_tokens": 76864517.0,
+      "step": 22590
+    },
+    {
+      "epoch": 0.22682564534907063,
+      "learning_rate": 0.0001546368782367819,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.7701222836971283,
+      "num_tokens": 76900131.0,
+      "step": 22600
+    },
+    {
+      "epoch": 0.22692601067887108,
+      "learning_rate": 0.0001546168051708218,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.765713906288147,
+      "num_tokens": 76933999.0,
+      "step": 22610
+    },
+    {
+      "epoch": 0.22702637600867157,
+      "learning_rate": 0.0001545967321048617,
+      "loss": 0.7807,
+      "mean_token_accuracy": 0.7662304699420929,
+      "num_tokens": 76968304.0,
+      "step": 22620
+    },
+    {
+      "epoch": 0.22712674133847205,
+      "learning_rate": 0.00015457665903890162,
+      "loss": 0.7814,
+      "mean_token_accuracy": 0.7684104800224304,
+      "num_tokens": 77002282.0,
+      "step": 22630
+    },
+    {
+      "epoch": 0.2272271066682725,
+      "learning_rate": 0.0001545565859729415,
+      "loss": 0.8023,
+      "mean_token_accuracy": 0.7641406416893005,
+      "num_tokens": 77036795.0,
+      "step": 22640
+    },
+    {
+      "epoch": 0.227327471998073,
+      "learning_rate": 0.0001545365129069814,
+      "loss": 0.7435,
+      "mean_token_accuracy": 0.7800127029418945,
+      "num_tokens": 77071678.0,
+      "step": 22650
+    },
+    {
+      "epoch": 0.22742783732787347,
+      "learning_rate": 0.00015451643984102132,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7709397971630096,
+      "num_tokens": 77105827.0,
+      "step": 22660
+    },
+    {
+      "epoch": 0.22752820265767393,
+      "learning_rate": 0.00015449636677506123,
+      "loss": 0.7517,
+      "mean_token_accuracy": 0.7780755877494812,
+      "num_tokens": 77139690.0,
+      "step": 22670
+    },
+    {
+      "epoch": 0.2276285679874744,
+      "learning_rate": 0.00015447629370910114,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7782335638999939,
+      "num_tokens": 77173944.0,
+      "step": 22680
+    },
+    {
+      "epoch": 0.22772893331727487,
+      "learning_rate": 0.00015445622064314105,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.7662321627140045,
+      "num_tokens": 77207324.0,
+      "step": 22690
+    },
+    {
+      "epoch": 0.22782929864707535,
+      "learning_rate": 0.00015443614757718095,
+      "loss": 0.7143,
+      "mean_token_accuracy": 0.7842252373695373,
+      "num_tokens": 77241537.0,
+      "step": 22700
+    },
+    {
+      "epoch": 0.22792966397687583,
+      "learning_rate": 0.00015441607451122084,
+      "loss": 0.7524,
+      "mean_token_accuracy": 0.7744047999382019,
+      "num_tokens": 77275735.0,
+      "step": 22710
+    },
+    {
+      "epoch": 0.2280300293066763,
+      "learning_rate": 0.00015439600144526077,
+      "loss": 0.7689,
+      "mean_token_accuracy": 0.7665367484092712,
+      "num_tokens": 77309976.0,
+      "step": 22720
+    },
+    {
+      "epoch": 0.22813039463647677,
+      "learning_rate": 0.00015437592837930065,
+      "loss": 0.7763,
+      "mean_token_accuracy": 0.7648578226566315,
+      "num_tokens": 77343643.0,
+      "step": 22730
+    },
+    {
+      "epoch": 0.22823075996627726,
+      "learning_rate": 0.00015435585531334056,
+      "loss": 0.7909,
+      "mean_token_accuracy": 0.7639325499534607,
+      "num_tokens": 77377381.0,
+      "step": 22740
+    },
+    {
+      "epoch": 0.2283311252960777,
+      "learning_rate": 0.00015433578224738047,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.7704145252704621,
+      "num_tokens": 77411698.0,
+      "step": 22750
+    },
+    {
+      "epoch": 0.2284314906258782,
+      "learning_rate": 0.00015431570918142038,
+      "loss": 0.7714,
+      "mean_token_accuracy": 0.7725685954093933,
+      "num_tokens": 77445414.0,
+      "step": 22760
+    },
+    {
+      "epoch": 0.22853185595567868,
+      "learning_rate": 0.00015429563611546029,
+      "loss": 0.7976,
+      "mean_token_accuracy": 0.7643962144851685,
+      "num_tokens": 77479790.0,
+      "step": 22770
+    },
+    {
+      "epoch": 0.22863222128547914,
+      "learning_rate": 0.0001542755630495002,
+      "loss": 0.742,
+      "mean_token_accuracy": 0.7726198673248291,
+      "num_tokens": 77513380.0,
+      "step": 22780
+    },
+    {
+      "epoch": 0.22873258661527962,
+      "learning_rate": 0.0001542554899835401,
+      "loss": 0.7357,
+      "mean_token_accuracy": 0.7799278557300567,
+      "num_tokens": 77547445.0,
+      "step": 22790
+    },
+    {
+      "epoch": 0.2288329519450801,
+      "learning_rate": 0.00015423541691757998,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7717826008796692,
+      "num_tokens": 77580805.0,
+      "step": 22800
+    },
+    {
+      "epoch": 0.22893331727488056,
+      "learning_rate": 0.00015421534385161992,
+      "loss": 0.7819,
+      "mean_token_accuracy": 0.7631809949874878,
+      "num_tokens": 77614939.0,
+      "step": 22810
+    },
+    {
+      "epoch": 0.22903368260468104,
+      "learning_rate": 0.0001541952707856598,
+      "loss": 0.765,
+      "mean_token_accuracy": 0.7719576239585877,
+      "num_tokens": 77648513.0,
+      "step": 22820
+    },
+    {
+      "epoch": 0.2291340479344815,
+      "learning_rate": 0.0001541751977196997,
+      "loss": 0.7378,
+      "mean_token_accuracy": 0.7783514261245728,
+      "num_tokens": 77682577.0,
+      "step": 22830
+    },
+    {
+      "epoch": 0.22923441326428198,
+      "learning_rate": 0.00015415512465373962,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7713116824626922,
+      "num_tokens": 77716512.0,
+      "step": 22840
+    },
+    {
+      "epoch": 0.22933477859408247,
+      "learning_rate": 0.00015413505158777953,
+      "loss": 0.7539,
+      "mean_token_accuracy": 0.7750553786754608,
+      "num_tokens": 77750751.0,
+      "step": 22850
+    },
+    {
+      "epoch": 0.22943514392388292,
+      "learning_rate": 0.0001541149785218194,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7759535074234009,
+      "num_tokens": 77784657.0,
+      "step": 22860
+    },
+    {
+      "epoch": 0.2295355092536834,
+      "learning_rate": 0.00015409490545585934,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7716870188713074,
+      "num_tokens": 77818195.0,
+      "step": 22870
+    },
+    {
+      "epoch": 0.2296358745834839,
+      "learning_rate": 0.00015407483238989922,
+      "loss": 0.744,
+      "mean_token_accuracy": 0.7780516982078552,
+      "num_tokens": 77852683.0,
+      "step": 22880
+    },
+    {
+      "epoch": 0.22973623991328435,
+      "learning_rate": 0.00015405475932393913,
+      "loss": 0.799,
+      "mean_token_accuracy": 0.763060474395752,
+      "num_tokens": 77887523.0,
+      "step": 22890
+    },
+    {
+      "epoch": 0.22983660524308483,
+      "learning_rate": 0.00015403468625797904,
+      "loss": 0.7758,
+      "mean_token_accuracy": 0.7726968407630921,
+      "num_tokens": 77921201.0,
+      "step": 22900
+    },
+    {
+      "epoch": 0.2299369705728853,
+      "learning_rate": 0.00015401461319201895,
+      "loss": 0.7513,
+      "mean_token_accuracy": 0.782043993473053,
+      "num_tokens": 77955086.0,
+      "step": 22910
+    },
+    {
+      "epoch": 0.23003733590268577,
+      "learning_rate": 0.00015399454012605889,
+      "loss": 0.7712,
+      "mean_token_accuracy": 0.7686130344867707,
+      "num_tokens": 77989111.0,
+      "step": 22920
+    },
+    {
+      "epoch": 0.23013770123248625,
+      "learning_rate": 0.00015397446706009877,
+      "loss": 0.7319,
+      "mean_token_accuracy": 0.7798277974128723,
+      "num_tokens": 78022530.0,
+      "step": 22930
+    },
+    {
+      "epoch": 0.23023806656228674,
+      "learning_rate": 0.00015395439399413868,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7693477272987366,
+      "num_tokens": 78057045.0,
+      "step": 22940
+    },
+    {
+      "epoch": 0.2303384318920872,
+      "learning_rate": 0.00015393432092817858,
+      "loss": 0.7808,
+      "mean_token_accuracy": 0.7679920077323914,
+      "num_tokens": 78091534.0,
+      "step": 22950
+    },
+    {
+      "epoch": 0.23043879722188768,
+      "learning_rate": 0.0001539142478622185,
+      "loss": 0.8153,
+      "mean_token_accuracy": 0.7571540176868439,
+      "num_tokens": 78124619.0,
+      "step": 22960
+    },
+    {
+      "epoch": 0.23053916255168813,
+      "learning_rate": 0.00015389417479625837,
+      "loss": 0.8063,
+      "mean_token_accuracy": 0.7597503304481507,
+      "num_tokens": 78158312.0,
+      "step": 22970
+    },
+    {
+      "epoch": 0.23063952788148862,
+      "learning_rate": 0.0001538741017302983,
+      "loss": 0.7322,
+      "mean_token_accuracy": 0.781629455089569,
+      "num_tokens": 78193093.0,
+      "step": 22980
+    },
+    {
+      "epoch": 0.2307398932112891,
+      "learning_rate": 0.0001538540286643382,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7688167214393615,
+      "num_tokens": 78227432.0,
+      "step": 22990
+    },
+    {
+      "epoch": 0.23084025854108955,
+      "learning_rate": 0.0001538339555983781,
+      "loss": 0.7679,
+      "mean_token_accuracy": 0.772001850605011,
+      "num_tokens": 78261054.0,
+      "step": 23000
+    },
+    {
+      "epoch": 0.23094062387089004,
+      "learning_rate": 0.000153813882532418,
+      "loss": 0.8045,
+      "mean_token_accuracy": 0.7627824485301972,
+      "num_tokens": 78294990.0,
+      "step": 23010
+    },
+    {
+      "epoch": 0.23104098920069052,
+      "learning_rate": 0.00015379380946645792,
+      "loss": 0.7661,
+      "mean_token_accuracy": 0.773594182729721,
+      "num_tokens": 78329086.0,
+      "step": 23020
+    },
+    {
+      "epoch": 0.23114135453049098,
+      "learning_rate": 0.00015377373640049782,
+      "loss": 0.7651,
+      "mean_token_accuracy": 0.7671305358409881,
+      "num_tokens": 78362541.0,
+      "step": 23030
+    },
+    {
+      "epoch": 0.23124171986029146,
+      "learning_rate": 0.00015375366333453773,
+      "loss": 0.7923,
+      "mean_token_accuracy": 0.7729197859764099,
+      "num_tokens": 78397177.0,
+      "step": 23040
+    },
+    {
+      "epoch": 0.23134208519009195,
+      "learning_rate": 0.00015373359026857764,
+      "loss": 0.7604,
+      "mean_token_accuracy": 0.7704223334789276,
+      "num_tokens": 78431194.0,
+      "step": 23050
+    },
+    {
+      "epoch": 0.2314424505198924,
+      "learning_rate": 0.00015371351720261752,
+      "loss": 0.7749,
+      "mean_token_accuracy": 0.7729601800441742,
+      "num_tokens": 78465215.0,
+      "step": 23060
+    },
+    {
+      "epoch": 0.23154281584969288,
+      "learning_rate": 0.00015369344413665746,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.7727331995964051,
+      "num_tokens": 78499820.0,
+      "step": 23070
+    },
+    {
+      "epoch": 0.23164318117949337,
+      "learning_rate": 0.00015367337107069734,
+      "loss": 0.7887,
+      "mean_token_accuracy": 0.7687535941600799,
+      "num_tokens": 78534030.0,
+      "step": 23080
+    },
+    {
+      "epoch": 0.23174354650929382,
+      "learning_rate": 0.00015365329800473725,
+      "loss": 0.76,
+      "mean_token_accuracy": 0.7724514603614807,
+      "num_tokens": 78568289.0,
+      "step": 23090
+    },
+    {
+      "epoch": 0.2318439118390943,
+      "learning_rate": 0.00015363322493877716,
+      "loss": 0.7763,
+      "mean_token_accuracy": 0.7684251010417938,
+      "num_tokens": 78602670.0,
+      "step": 23100
+    },
+    {
+      "epoch": 0.23194427716889476,
+      "learning_rate": 0.00015361315187281706,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.7708824157714844,
+      "num_tokens": 78637550.0,
+      "step": 23110
+    },
+    {
+      "epoch": 0.23204464249869525,
+      "learning_rate": 0.00015359307880685697,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7683057010173797,
+      "num_tokens": 78671462.0,
+      "step": 23120
+    },
+    {
+      "epoch": 0.23214500782849573,
+      "learning_rate": 0.00015357300574089688,
+      "loss": 0.7418,
+      "mean_token_accuracy": 0.7737292468547821,
+      "num_tokens": 78705589.0,
+      "step": 23130
+    },
+    {
+      "epoch": 0.2322453731582962,
+      "learning_rate": 0.0001535529326749368,
+      "loss": 0.765,
+      "mean_token_accuracy": 0.7699859380722046,
+      "num_tokens": 78739648.0,
+      "step": 23140
+    },
+    {
+      "epoch": 0.23234573848809667,
+      "learning_rate": 0.00015353285960897667,
+      "loss": 0.7744,
+      "mean_token_accuracy": 0.7685719311237336,
+      "num_tokens": 78773935.0,
+      "step": 23150
+    },
+    {
+      "epoch": 0.23244610381789715,
+      "learning_rate": 0.0001535127865430166,
+      "loss": 0.7922,
+      "mean_token_accuracy": 0.7677593052387237,
+      "num_tokens": 78808407.0,
+      "step": 23160
+    },
+    {
+      "epoch": 0.2325464691476976,
+      "learning_rate": 0.0001534927134770565,
+      "loss": 0.7943,
+      "mean_token_accuracy": 0.765311723947525,
+      "num_tokens": 78842722.0,
+      "step": 23170
+    },
+    {
+      "epoch": 0.2326468344774981,
+      "learning_rate": 0.0001534726404110964,
+      "loss": 0.7831,
+      "mean_token_accuracy": 0.7717867314815521,
+      "num_tokens": 78876265.0,
+      "step": 23180
+    },
+    {
+      "epoch": 0.23274719980729858,
+      "learning_rate": 0.0001534525673451363,
+      "loss": 0.7645,
+      "mean_token_accuracy": 0.7726698100566864,
+      "num_tokens": 78910424.0,
+      "step": 23190
+    },
+    {
+      "epoch": 0.23284756513709903,
+      "learning_rate": 0.00015343249427917621,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7792783617973328,
+      "num_tokens": 78944040.0,
+      "step": 23200
+    },
+    {
+      "epoch": 0.23294793046689952,
+      "learning_rate": 0.0001534124212132161,
+      "loss": 0.7767,
+      "mean_token_accuracy": 0.7686317384243011,
+      "num_tokens": 78977131.0,
+      "step": 23210
+    },
+    {
+      "epoch": 0.2330482957967,
+      "learning_rate": 0.00015339234814725603,
+      "loss": 0.7369,
+      "mean_token_accuracy": 0.776541656255722,
+      "num_tokens": 79011287.0,
+      "step": 23220
+    },
+    {
+      "epoch": 0.23314866112650046,
+      "learning_rate": 0.0001533722750812959,
+      "loss": 0.7657,
+      "mean_token_accuracy": 0.7776280224323273,
+      "num_tokens": 79044620.0,
+      "step": 23230
+    },
+    {
+      "epoch": 0.23324902645630094,
+      "learning_rate": 0.00015335220201533582,
+      "loss": 0.7912,
+      "mean_token_accuracy": 0.7631687164306641,
+      "num_tokens": 79078598.0,
+      "step": 23240
+    },
+    {
+      "epoch": 0.2333493917861014,
+      "learning_rate": 0.00015333212894937573,
+      "loss": 0.7749,
+      "mean_token_accuracy": 0.7703653693199157,
+      "num_tokens": 79112267.0,
+      "step": 23250
+    },
+    {
+      "epoch": 0.23344975711590188,
+      "learning_rate": 0.00015331205588341564,
+      "loss": 0.723,
+      "mean_token_accuracy": 0.7808611869812012,
+      "num_tokens": 79146140.0,
+      "step": 23260
+    },
+    {
+      "epoch": 0.23355012244570236,
+      "learning_rate": 0.00015329198281745555,
+      "loss": 0.7877,
+      "mean_token_accuracy": 0.7686753094196319,
+      "num_tokens": 79179376.0,
+      "step": 23270
+    },
+    {
+      "epoch": 0.23365048777550282,
+      "learning_rate": 0.00015327190975149545,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.7692699015140534,
+      "num_tokens": 79212928.0,
+      "step": 23280
+    },
+    {
+      "epoch": 0.2337508531053033,
+      "learning_rate": 0.00015325183668553536,
+      "loss": 0.7714,
+      "mean_token_accuracy": 0.7678778171539307,
+      "num_tokens": 79247276.0,
+      "step": 23290
+    },
+    {
+      "epoch": 0.23385121843510379,
+      "learning_rate": 0.00015323176361957524,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7694761216640472,
+      "num_tokens": 79280809.0,
+      "step": 23300
+    },
+    {
+      "epoch": 0.23395158376490424,
+      "learning_rate": 0.00015321169055361518,
+      "loss": 0.7777,
+      "mean_token_accuracy": 0.7689460635185241,
+      "num_tokens": 79315215.0,
+      "step": 23310
+    },
+    {
+      "epoch": 0.23405194909470473,
+      "learning_rate": 0.00015319161748765506,
+      "loss": 0.7812,
+      "mean_token_accuracy": 0.7662648558616638,
+      "num_tokens": 79348428.0,
+      "step": 23320
+    },
+    {
+      "epoch": 0.2341523144245052,
+      "learning_rate": 0.00015317154442169497,
+      "loss": 0.7765,
+      "mean_token_accuracy": 0.7612001657485962,
+      "num_tokens": 79382310.0,
+      "step": 23330
+    },
+    {
+      "epoch": 0.23425267975430566,
+      "learning_rate": 0.00015315147135573488,
+      "loss": 0.7873,
+      "mean_token_accuracy": 0.7655379593372345,
+      "num_tokens": 79416640.0,
+      "step": 23340
+    },
+    {
+      "epoch": 0.23435304508410615,
+      "learning_rate": 0.00015313139828977479,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7735551476478577,
+      "num_tokens": 79450449.0,
+      "step": 23350
+    },
+    {
+      "epoch": 0.23445341041390663,
+      "learning_rate": 0.0001531113252238147,
+      "loss": 0.7624,
+      "mean_token_accuracy": 0.7742606103420258,
+      "num_tokens": 79484109.0,
+      "step": 23360
+    },
+    {
+      "epoch": 0.2345537757437071,
+      "learning_rate": 0.0001530912521578546,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7740523397922516,
+      "num_tokens": 79517766.0,
+      "step": 23370
+    },
+    {
+      "epoch": 0.23465414107350757,
+      "learning_rate": 0.0001530711790918945,
+      "loss": 0.7388,
+      "mean_token_accuracy": 0.7769231557846069,
+      "num_tokens": 79550985.0,
+      "step": 23380
+    },
+    {
+      "epoch": 0.23475450640330803,
+      "learning_rate": 0.0001530511060259344,
+      "loss": 0.7587,
+      "mean_token_accuracy": 0.7782913088798523,
+      "num_tokens": 79584552.0,
+      "step": 23390
+    },
+    {
+      "epoch": 0.2348548717331085,
+      "learning_rate": 0.00015303103295997433,
+      "loss": 0.8073,
+      "mean_token_accuracy": 0.7672119975090027,
+      "num_tokens": 79618051.0,
+      "step": 23400
+    },
+    {
+      "epoch": 0.234955237062909,
+      "learning_rate": 0.0001530109598940142,
+      "loss": 0.7406,
+      "mean_token_accuracy": 0.7801278173923493,
+      "num_tokens": 79652321.0,
+      "step": 23410
+    },
+    {
+      "epoch": 0.23505560239270945,
+      "learning_rate": 0.00015299088682805412,
+      "loss": 0.7625,
+      "mean_token_accuracy": 0.7745743751525879,
+      "num_tokens": 79685474.0,
+      "step": 23420
+    },
+    {
+      "epoch": 0.23515596772250993,
+      "learning_rate": 0.00015297081376209403,
+      "loss": 0.7483,
+      "mean_token_accuracy": 0.7806742548942566,
+      "num_tokens": 79718907.0,
+      "step": 23430
+    },
+    {
+      "epoch": 0.23525633305231042,
+      "learning_rate": 0.00015295074069613393,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.7705367624759674,
+      "num_tokens": 79752851.0,
+      "step": 23440
+    },
+    {
+      "epoch": 0.23535669838211087,
+      "learning_rate": 0.00015293066763017382,
+      "loss": 0.7445,
+      "mean_token_accuracy": 0.778471952676773,
+      "num_tokens": 79786545.0,
+      "step": 23450
+    },
+    {
+      "epoch": 0.23545706371191136,
+      "learning_rate": 0.00015291059456421375,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7745211839675903,
+      "num_tokens": 79820372.0,
+      "step": 23460
+    },
+    {
+      "epoch": 0.23555742904171184,
+      "learning_rate": 0.00015289052149825366,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7718680500984192,
+      "num_tokens": 79854552.0,
+      "step": 23470
+    },
+    {
+      "epoch": 0.2356577943715123,
+      "learning_rate": 0.00015287044843229354,
+      "loss": 0.7902,
+      "mean_token_accuracy": 0.7693752467632293,
+      "num_tokens": 79888384.0,
+      "step": 23480
+    },
+    {
+      "epoch": 0.23575815970131278,
+      "learning_rate": 0.00015285037536633348,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7773387908935547,
+      "num_tokens": 79921343.0,
+      "step": 23490
+    },
+    {
+      "epoch": 0.23585852503111326,
+      "learning_rate": 0.00015283030230037336,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.7736146032810212,
+      "num_tokens": 79954869.0,
+      "step": 23500
+    },
+    {
+      "epoch": 0.23595889036091372,
+      "learning_rate": 0.0001528102292344133,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.7789220571517944,
+      "num_tokens": 79988787.0,
+      "step": 23510
+    },
+    {
+      "epoch": 0.2360592556907142,
+      "learning_rate": 0.00015279015616845318,
+      "loss": 0.748,
+      "mean_token_accuracy": 0.7819363057613373,
+      "num_tokens": 80022579.0,
+      "step": 23520
+    },
+    {
+      "epoch": 0.23615962102051466,
+      "learning_rate": 0.00015277008310249308,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7654751539230347,
+      "num_tokens": 80056791.0,
+      "step": 23530
+    },
+    {
+      "epoch": 0.23625998635031514,
+      "learning_rate": 0.000152750010036533,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.7784739851951599,
+      "num_tokens": 80091293.0,
+      "step": 23540
+    },
+    {
+      "epoch": 0.23636035168011563,
+      "learning_rate": 0.0001527299369705729,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7739460527896881,
+      "num_tokens": 80125243.0,
+      "step": 23550
+    },
+    {
+      "epoch": 0.23646071700991608,
+      "learning_rate": 0.00015270986390461278,
+      "loss": 0.7448,
+      "mean_token_accuracy": 0.7724997520446777,
+      "num_tokens": 80159651.0,
+      "step": 23560
+    },
+    {
+      "epoch": 0.23656108233971657,
+      "learning_rate": 0.00015268979083865272,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.773425680398941,
+      "num_tokens": 80194166.0,
+      "step": 23570
+    },
+    {
+      "epoch": 0.23666144766951705,
+      "learning_rate": 0.0001526697177726926,
+      "loss": 0.7402,
+      "mean_token_accuracy": 0.7749640464782714,
+      "num_tokens": 80227835.0,
+      "step": 23580
+    },
+    {
+      "epoch": 0.2367618129993175,
+      "learning_rate": 0.0001526496447067325,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.7723572850227356,
+      "num_tokens": 80262461.0,
+      "step": 23590
+    },
+    {
+      "epoch": 0.236862178329118,
+      "learning_rate": 0.00015262957164077242,
+      "loss": 0.7917,
+      "mean_token_accuracy": 0.7664845049381256,
+      "num_tokens": 80296052.0,
+      "step": 23600
+    },
+    {
+      "epoch": 0.23696254365891847,
+      "learning_rate": 0.00015260949857481232,
+      "loss": 0.8065,
+      "mean_token_accuracy": 0.76147341132164,
+      "num_tokens": 80329560.0,
+      "step": 23610
+    },
+    {
+      "epoch": 0.23706290898871893,
+      "learning_rate": 0.00015258942550885223,
+      "loss": 0.7748,
+      "mean_token_accuracy": 0.7736762523651123,
+      "num_tokens": 80364268.0,
+      "step": 23620
+    },
+    {
+      "epoch": 0.2371632743185194,
+      "learning_rate": 0.00015256935244289214,
+      "loss": 0.7829,
+      "mean_token_accuracy": 0.7689502716064454,
+      "num_tokens": 80398126.0,
+      "step": 23630
+    },
+    {
+      "epoch": 0.2372636396483199,
+      "learning_rate": 0.00015254927937693205,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.7803487658500672,
+      "num_tokens": 80431691.0,
+      "step": 23640
+    },
+    {
+      "epoch": 0.23736400497812035,
+      "learning_rate": 0.00015252920631097193,
+      "loss": 0.7832,
+      "mean_token_accuracy": 0.7693206608295441,
+      "num_tokens": 80464821.0,
+      "step": 23650
+    },
+    {
+      "epoch": 0.23746437030792084,
+      "learning_rate": 0.00015250913324501187,
+      "loss": 0.7436,
+      "mean_token_accuracy": 0.7709475219249725,
+      "num_tokens": 80499063.0,
+      "step": 23660
+    },
+    {
+      "epoch": 0.23756473563772132,
+      "learning_rate": 0.00015248906017905175,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7674477159976959,
+      "num_tokens": 80533670.0,
+      "step": 23670
+    },
+    {
+      "epoch": 0.23766510096752178,
+      "learning_rate": 0.00015246898711309166,
+      "loss": 0.7706,
+      "mean_token_accuracy": 0.7716367185115814,
+      "num_tokens": 80567369.0,
+      "step": 23680
+    },
+    {
+      "epoch": 0.23776546629732226,
+      "learning_rate": 0.00015244891404713156,
+      "loss": 0.7813,
+      "mean_token_accuracy": 0.7622873067855835,
+      "num_tokens": 80601216.0,
+      "step": 23690
+    },
+    {
+      "epoch": 0.23786583162712271,
+      "learning_rate": 0.00015242884098117147,
+      "loss": 0.7957,
+      "mean_token_accuracy": 0.7627099812030792,
+      "num_tokens": 80634983.0,
+      "step": 23700
+    },
+    {
+      "epoch": 0.2379661969569232,
+      "learning_rate": 0.00015240876791521138,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7720392763614654,
+      "num_tokens": 80668706.0,
+      "step": 23710
+    },
+    {
+      "epoch": 0.23806656228672368,
+      "learning_rate": 0.0001523886948492513,
+      "loss": 0.7461,
+      "mean_token_accuracy": 0.7805128991603851,
+      "num_tokens": 80702813.0,
+      "step": 23720
+    },
+    {
+      "epoch": 0.23816692761652414,
+      "learning_rate": 0.0001523686217832912,
+      "loss": 0.7692,
+      "mean_token_accuracy": 0.7717490255832672,
+      "num_tokens": 80736386.0,
+      "step": 23730
+    },
+    {
+      "epoch": 0.23826729294632462,
+      "learning_rate": 0.00015234854871733108,
+      "loss": 0.7556,
+      "mean_token_accuracy": 0.7743633389472961,
+      "num_tokens": 80769866.0,
+      "step": 23740
+    },
+    {
+      "epoch": 0.2383676582761251,
+      "learning_rate": 0.00015232847565137101,
+      "loss": 0.7392,
+      "mean_token_accuracy": 0.7824662148952484,
+      "num_tokens": 80804598.0,
+      "step": 23750
+    },
+    {
+      "epoch": 0.23846802360592556,
+      "learning_rate": 0.0001523084025854109,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.7698829948902131,
+      "num_tokens": 80839414.0,
+      "step": 23760
+    },
+    {
+      "epoch": 0.23856838893572604,
+      "learning_rate": 0.0001522883295194508,
+      "loss": 0.758,
+      "mean_token_accuracy": 0.7751199185848237,
+      "num_tokens": 80873310.0,
+      "step": 23770
+    },
+    {
+      "epoch": 0.23866875426552653,
+      "learning_rate": 0.0001522682564534907,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7692757785320282,
+      "num_tokens": 80907353.0,
+      "step": 23780
+    },
+    {
+      "epoch": 0.23876911959532698,
+      "learning_rate": 0.00015224818338753062,
+      "loss": 0.7719,
+      "mean_token_accuracy": 0.7673554956912995,
+      "num_tokens": 80941677.0,
+      "step": 23790
+    },
+    {
+      "epoch": 0.23886948492512747,
+      "learning_rate": 0.0001522281103215705,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7721669137477875,
+      "num_tokens": 80975300.0,
+      "step": 23800
+    },
+    {
+      "epoch": 0.23896985025492795,
+      "learning_rate": 0.00015220803725561044,
+      "loss": 0.7427,
+      "mean_token_accuracy": 0.7750360429286957,
+      "num_tokens": 81009743.0,
+      "step": 23810
+    },
+    {
+      "epoch": 0.2390702155847284,
+      "learning_rate": 0.00015218796418965035,
+      "loss": 0.7502,
+      "mean_token_accuracy": 0.7714247167110443,
+      "num_tokens": 81043156.0,
+      "step": 23820
+    },
+    {
+      "epoch": 0.2391705809145289,
+      "learning_rate": 0.00015216789112369023,
+      "loss": 0.7379,
+      "mean_token_accuracy": 0.7796186149120331,
+      "num_tokens": 81077157.0,
+      "step": 23830
+    },
+    {
+      "epoch": 0.23927094624432935,
+      "learning_rate": 0.00015214781805773016,
+      "loss": 0.7561,
+      "mean_token_accuracy": 0.7781562328338623,
+      "num_tokens": 81111061.0,
+      "step": 23840
+    },
+    {
+      "epoch": 0.23937131157412983,
+      "learning_rate": 0.00015212774499177005,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.7698342204093933,
+      "num_tokens": 81144849.0,
+      "step": 23850
+    },
+    {
+      "epoch": 0.23947167690393031,
+      "learning_rate": 0.00015210767192580995,
+      "loss": 0.7429,
+      "mean_token_accuracy": 0.7787699818611145,
+      "num_tokens": 81178995.0,
+      "step": 23860
+    },
+    {
+      "epoch": 0.23957204223373077,
+      "learning_rate": 0.00015208759885984986,
+      "loss": 0.7224,
+      "mean_token_accuracy": 0.7845096230506897,
+      "num_tokens": 81212540.0,
+      "step": 23870
+    },
+    {
+      "epoch": 0.23967240756353125,
+      "learning_rate": 0.00015206752579388977,
+      "loss": 0.7611,
+      "mean_token_accuracy": 0.7693464875221252,
+      "num_tokens": 81245572.0,
+      "step": 23880
+    },
+    {
+      "epoch": 0.23977277289333174,
+      "learning_rate": 0.00015204745272792965,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7700681567192078,
+      "num_tokens": 81279958.0,
+      "step": 23890
+    },
+    {
+      "epoch": 0.2398731382231322,
+      "learning_rate": 0.0001520273796619696,
+      "loss": 0.7998,
+      "mean_token_accuracy": 0.7623650908470154,
+      "num_tokens": 81314120.0,
+      "step": 23900
+    },
+    {
+      "epoch": 0.23997350355293268,
+      "learning_rate": 0.00015200730659600947,
+      "loss": 0.7459,
+      "mean_token_accuracy": 0.7785841584205627,
+      "num_tokens": 81348417.0,
+      "step": 23910
+    },
+    {
+      "epoch": 0.24007386888273316,
+      "learning_rate": 0.00015198723353004938,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7697364985942841,
+      "num_tokens": 81382691.0,
+      "step": 23920
+    },
+    {
+      "epoch": 0.24017423421253362,
+      "learning_rate": 0.00015196716046408929,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.7758682548999787,
+      "num_tokens": 81416407.0,
+      "step": 23930
+    },
+    {
+      "epoch": 0.2402745995423341,
+      "learning_rate": 0.0001519470873981292,
+      "loss": 0.758,
+      "mean_token_accuracy": 0.7686584532260895,
+      "num_tokens": 81450577.0,
+      "step": 23940
+    },
+    {
+      "epoch": 0.24037496487213458,
+      "learning_rate": 0.0001519270143321691,
+      "loss": 0.7846,
+      "mean_token_accuracy": 0.7708744466304779,
+      "num_tokens": 81483488.0,
+      "step": 23950
+    },
+    {
+      "epoch": 0.24047533020193504,
+      "learning_rate": 0.000151906941266209,
+      "loss": 0.7611,
+      "mean_token_accuracy": 0.7697698056697846,
+      "num_tokens": 81517080.0,
+      "step": 23960
+    },
+    {
+      "epoch": 0.24057569553173552,
+      "learning_rate": 0.00015188686820024892,
+      "loss": 0.7582,
+      "mean_token_accuracy": 0.7723502278327942,
+      "num_tokens": 81550744.0,
+      "step": 23970
+    },
+    {
+      "epoch": 0.24067606086153598,
+      "learning_rate": 0.0001518667951342888,
+      "loss": 0.7491,
+      "mean_token_accuracy": 0.7779267430305481,
+      "num_tokens": 81583661.0,
+      "step": 23980
+    },
+    {
+      "epoch": 0.24077642619133646,
+      "learning_rate": 0.00015184672206832874,
+      "loss": 0.7448,
+      "mean_token_accuracy": 0.7716735482215882,
+      "num_tokens": 81617350.0,
+      "step": 23990
+    },
+    {
+      "epoch": 0.24087679152113695,
+      "learning_rate": 0.00015182664900236862,
+      "loss": 0.7594,
+      "mean_token_accuracy": 0.7715551853179932,
+      "num_tokens": 81652052.0,
+      "step": 24000
+    },
+    {
+      "epoch": 0.2409771568509374,
+      "learning_rate": 0.00015180657593640853,
+      "loss": 0.752,
+      "mean_token_accuracy": 0.7722446501255036,
+      "num_tokens": 81686582.0,
+      "step": 24010
+    },
+    {
+      "epoch": 0.24107752218073789,
+      "learning_rate": 0.00015178650287044843,
+      "loss": 0.7922,
+      "mean_token_accuracy": 0.7614984273910522,
+      "num_tokens": 81720532.0,
+      "step": 24020
+    },
+    {
+      "epoch": 0.24117788751053837,
+      "learning_rate": 0.00015176642980448834,
+      "loss": 0.7255,
+      "mean_token_accuracy": 0.7767630457878113,
+      "num_tokens": 81753436.0,
+      "step": 24030
+    },
+    {
+      "epoch": 0.24127825284033882,
+      "learning_rate": 0.00015174635673852825,
+      "loss": 0.7811,
+      "mean_token_accuracy": 0.7708664476871491,
+      "num_tokens": 81786624.0,
+      "step": 24040
+    },
+    {
+      "epoch": 0.2413786181701393,
+      "learning_rate": 0.00015172628367256816,
+      "loss": 0.7344,
+      "mean_token_accuracy": 0.7771263420581818,
+      "num_tokens": 81820742.0,
+      "step": 24050
+    },
+    {
+      "epoch": 0.2414789834999398,
+      "learning_rate": 0.00015170621060660807,
+      "loss": 0.7361,
+      "mean_token_accuracy": 0.7774999976158142,
+      "num_tokens": 81854862.0,
+      "step": 24060
+    },
+    {
+      "epoch": 0.24157934882974025,
+      "learning_rate": 0.00015168613754064798,
+      "loss": 0.7712,
+      "mean_token_accuracy": 0.7666206061840057,
+      "num_tokens": 81889274.0,
+      "step": 24070
+    },
+    {
+      "epoch": 0.24167971415954073,
+      "learning_rate": 0.00015166606447468788,
+      "loss": 0.7633,
+      "mean_token_accuracy": 0.7750555813312531,
+      "num_tokens": 81923286.0,
+      "step": 24080
+    },
+    {
+      "epoch": 0.24178007948934122,
+      "learning_rate": 0.00015164599140872777,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.7750057756900788,
+      "num_tokens": 81957647.0,
+      "step": 24090
+    },
+    {
+      "epoch": 0.24188044481914167,
+      "learning_rate": 0.0001516259183427677,
+      "loss": 0.7657,
+      "mean_token_accuracy": 0.7737397193908692,
+      "num_tokens": 81992262.0,
+      "step": 24100
+    },
+    {
+      "epoch": 0.24198081014894215,
+      "learning_rate": 0.00015160584527680758,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7746361017227172,
+      "num_tokens": 82026134.0,
+      "step": 24110
+    },
+    {
+      "epoch": 0.2420811754787426,
+      "learning_rate": 0.0001515857722108475,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.7760856509208679,
+      "num_tokens": 82059412.0,
+      "step": 24120
+    },
+    {
+      "epoch": 0.2421815408085431,
+      "learning_rate": 0.0001515656991448874,
+      "loss": 0.7867,
+      "mean_token_accuracy": 0.7642060816287994,
+      "num_tokens": 82093614.0,
+      "step": 24130
+    },
+    {
+      "epoch": 0.24228190613834358,
+      "learning_rate": 0.0001515456260789273,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7733182191848755,
+      "num_tokens": 82127152.0,
+      "step": 24140
+    },
+    {
+      "epoch": 0.24238227146814403,
+      "learning_rate": 0.0001515255530129672,
+      "loss": 0.7538,
+      "mean_token_accuracy": 0.7726780831813812,
+      "num_tokens": 82161045.0,
+      "step": 24150
+    },
+    {
+      "epoch": 0.24248263679794452,
+      "learning_rate": 0.00015150547994700713,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7732154488563537,
+      "num_tokens": 82195039.0,
+      "step": 24160
+    },
+    {
+      "epoch": 0.242583002127745,
+      "learning_rate": 0.00015148540688104703,
+      "loss": 0.7853,
+      "mean_token_accuracy": 0.7675738036632538,
+      "num_tokens": 82229105.0,
+      "step": 24170
+    },
+    {
+      "epoch": 0.24268336745754546,
+      "learning_rate": 0.00015146533381508692,
+      "loss": 0.7686,
+      "mean_token_accuracy": 0.7697836279869079,
+      "num_tokens": 82262613.0,
+      "step": 24180
+    },
+    {
+      "epoch": 0.24278373278734594,
+      "learning_rate": 0.00015144526074912685,
+      "loss": 0.7689,
+      "mean_token_accuracy": 0.7739278435707092,
+      "num_tokens": 82296703.0,
+      "step": 24190
+    },
+    {
+      "epoch": 0.24288409811714642,
+      "learning_rate": 0.00015142518768316673,
+      "loss": 0.7673,
+      "mean_token_accuracy": 0.7727110087871552,
+      "num_tokens": 82331068.0,
+      "step": 24200
+    },
+    {
+      "epoch": 0.24298446344694688,
+      "learning_rate": 0.00015140511461720664,
+      "loss": 0.7503,
+      "mean_token_accuracy": 0.7769576132297515,
+      "num_tokens": 82364505.0,
+      "step": 24210
+    },
+    {
+      "epoch": 0.24308482877674736,
+      "learning_rate": 0.00015138504155124655,
+      "loss": 0.745,
+      "mean_token_accuracy": 0.7776174426078797,
+      "num_tokens": 82398500.0,
+      "step": 24220
+    },
+    {
+      "epoch": 0.24318519410654785,
+      "learning_rate": 0.00015136496848528646,
+      "loss": 0.7814,
+      "mean_token_accuracy": 0.7694048345088959,
+      "num_tokens": 82431808.0,
+      "step": 24230
+    },
+    {
+      "epoch": 0.2432855594363483,
+      "learning_rate": 0.00015134489541932634,
+      "loss": 0.7956,
+      "mean_token_accuracy": 0.7689190626144409,
+      "num_tokens": 82466117.0,
+      "step": 24240
+    },
+    {
+      "epoch": 0.2433859247661488,
+      "learning_rate": 0.00015132482235336627,
+      "loss": 0.7453,
+      "mean_token_accuracy": 0.780043751001358,
+      "num_tokens": 82500190.0,
+      "step": 24250
+    },
+    {
+      "epoch": 0.24348629009594924,
+      "learning_rate": 0.00015130474928740616,
+      "loss": 0.7847,
+      "mean_token_accuracy": 0.7688788890838623,
+      "num_tokens": 82533370.0,
+      "step": 24260
+    },
+    {
+      "epoch": 0.24358665542574973,
+      "learning_rate": 0.00015128467622144606,
+      "loss": 0.7662,
+      "mean_token_accuracy": 0.7725997388362884,
+      "num_tokens": 82567679.0,
+      "step": 24270
+    },
+    {
+      "epoch": 0.2436870207555502,
+      "learning_rate": 0.00015126460315548597,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.77141774892807,
+      "num_tokens": 82601856.0,
+      "step": 24280
+    },
+    {
+      "epoch": 0.24378738608535067,
+      "learning_rate": 0.00015124453008952588,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.7707475662231446,
+      "num_tokens": 82635356.0,
+      "step": 24290
+    },
+    {
+      "epoch": 0.24388775141515115,
+      "learning_rate": 0.0001512244570235658,
+      "loss": 0.7545,
+      "mean_token_accuracy": 0.7701977491378784,
+      "num_tokens": 82669879.0,
+      "step": 24300
+    },
+    {
+      "epoch": 0.24398811674495163,
+      "learning_rate": 0.0001512043839576057,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.7706035733222961,
+      "num_tokens": 82704522.0,
+      "step": 24310
+    },
+    {
+      "epoch": 0.2440884820747521,
+      "learning_rate": 0.0001511843108916456,
+      "loss": 0.7971,
+      "mean_token_accuracy": 0.7634208917617797,
+      "num_tokens": 82737999.0,
+      "step": 24320
+    },
+    {
+      "epoch": 0.24418884740455257,
+      "learning_rate": 0.0001511642378256855,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.778376430273056,
+      "num_tokens": 82771533.0,
+      "step": 24330
+    },
+    {
+      "epoch": 0.24428921273435306,
+      "learning_rate": 0.00015114416475972542,
+      "loss": 0.7487,
+      "mean_token_accuracy": 0.7745313465595245,
+      "num_tokens": 82805949.0,
+      "step": 24340
+    },
+    {
+      "epoch": 0.2443895780641535,
+      "learning_rate": 0.0001511240916937653,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7709367036819458,
+      "num_tokens": 82840834.0,
+      "step": 24350
+    },
+    {
+      "epoch": 0.244489943393954,
+      "learning_rate": 0.0001511040186278052,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7755620777606964,
+      "num_tokens": 82874182.0,
+      "step": 24360
+    },
+    {
+      "epoch": 0.24459030872375448,
+      "learning_rate": 0.00015108394556184512,
+      "loss": 0.7832,
+      "mean_token_accuracy": 0.7668637633323669,
+      "num_tokens": 82907934.0,
+      "step": 24370
+    },
+    {
+      "epoch": 0.24469067405355494,
+      "learning_rate": 0.00015106387249588503,
+      "loss": 0.7701,
+      "mean_token_accuracy": 0.7740091979503632,
+      "num_tokens": 82942363.0,
+      "step": 24380
+    },
+    {
+      "epoch": 0.24479103938335542,
+      "learning_rate": 0.00015104379942992494,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.775358909368515,
+      "num_tokens": 82976781.0,
+      "step": 24390
+    },
+    {
+      "epoch": 0.24489140471315587,
+      "learning_rate": 0.00015102372636396485,
+      "loss": 0.7905,
+      "mean_token_accuracy": 0.7667368113994598,
+      "num_tokens": 83010721.0,
+      "step": 24400
+    },
+    {
+      "epoch": 0.24499177004295636,
+      "learning_rate": 0.00015100365329800475,
+      "loss": 0.7485,
+      "mean_token_accuracy": 0.7746308982372284,
+      "num_tokens": 83044831.0,
+      "step": 24410
+    },
+    {
+      "epoch": 0.24509213537275684,
+      "learning_rate": 0.00015098358023204464,
+      "loss": 0.7526,
+      "mean_token_accuracy": 0.7761373102664948,
+      "num_tokens": 83078730.0,
+      "step": 24420
+    },
+    {
+      "epoch": 0.2451925007025573,
+      "learning_rate": 0.00015096350716608457,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.7750726759433746,
+      "num_tokens": 83112963.0,
+      "step": 24430
+    },
+    {
+      "epoch": 0.24529286603235778,
+      "learning_rate": 0.00015094343410012445,
+      "loss": 0.7528,
+      "mean_token_accuracy": 0.7696741640567779,
+      "num_tokens": 83146238.0,
+      "step": 24440
+    },
+    {
+      "epoch": 0.24539323136215827,
+      "learning_rate": 0.00015092336103416436,
+      "loss": 0.7571,
+      "mean_token_accuracy": 0.7693495810031891,
+      "num_tokens": 83179916.0,
+      "step": 24450
+    },
+    {
+      "epoch": 0.24549359669195872,
+      "learning_rate": 0.00015090328796820427,
+      "loss": 0.7639,
+      "mean_token_accuracy": 0.7746648609638214,
+      "num_tokens": 83214052.0,
+      "step": 24460
+    },
+    {
+      "epoch": 0.2455939620217592,
+      "learning_rate": 0.00015088321490224418,
+      "loss": 0.7327,
+      "mean_token_accuracy": 0.7817089438438416,
+      "num_tokens": 83248328.0,
+      "step": 24470
+    },
+    {
+      "epoch": 0.2456943273515597,
+      "learning_rate": 0.00015086314183628406,
+      "loss": 0.7763,
+      "mean_token_accuracy": 0.7671709716320038,
+      "num_tokens": 83281506.0,
+      "step": 24480
+    },
+    {
+      "epoch": 0.24579469268136014,
+      "learning_rate": 0.000150843068770324,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7677108585834503,
+      "num_tokens": 83315624.0,
+      "step": 24490
+    },
+    {
+      "epoch": 0.24589505801116063,
+      "learning_rate": 0.00015082299570436388,
+      "loss": 0.7814,
+      "mean_token_accuracy": 0.7697063982486725,
+      "num_tokens": 83349795.0,
+      "step": 24500
+    },
+    {
+      "epoch": 0.2459954233409611,
+      "learning_rate": 0.00015080292263840378,
+      "loss": 0.7506,
+      "mean_token_accuracy": 0.7754936277866363,
+      "num_tokens": 83383555.0,
+      "step": 24510
+    },
+    {
+      "epoch": 0.24609578867076157,
+      "learning_rate": 0.00015078284957244372,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7674134135246277,
+      "num_tokens": 83417458.0,
+      "step": 24520
+    },
+    {
+      "epoch": 0.24619615400056205,
+      "learning_rate": 0.0001507627765064836,
+      "loss": 0.7682,
+      "mean_token_accuracy": 0.7745970547199249,
+      "num_tokens": 83452068.0,
+      "step": 24530
+    },
+    {
+      "epoch": 0.2462965193303625,
+      "learning_rate": 0.0001507427034405235,
+      "loss": 0.7397,
+      "mean_token_accuracy": 0.7787245750427246,
+      "num_tokens": 83486232.0,
+      "step": 24540
+    },
+    {
+      "epoch": 0.246396884660163,
+      "learning_rate": 0.00015072263037456342,
+      "loss": 0.7854,
+      "mean_token_accuracy": 0.7659784495830536,
+      "num_tokens": 83520412.0,
+      "step": 24550
+    },
+    {
+      "epoch": 0.24649724998996347,
+      "learning_rate": 0.00015070255730860333,
+      "loss": 0.7673,
+      "mean_token_accuracy": 0.7728880882263184,
+      "num_tokens": 83554383.0,
+      "step": 24560
+    },
+    {
+      "epoch": 0.24659761531976393,
+      "learning_rate": 0.0001506824842426432,
+      "loss": 0.7763,
+      "mean_token_accuracy": 0.7732985317707062,
+      "num_tokens": 83588112.0,
+      "step": 24570
+    },
+    {
+      "epoch": 0.2466979806495644,
+      "learning_rate": 0.00015066241117668314,
+      "loss": 0.7909,
+      "mean_token_accuracy": 0.7652292132377625,
+      "num_tokens": 83622427.0,
+      "step": 24580
+    },
+    {
+      "epoch": 0.2467983459793649,
+      "learning_rate": 0.00015064233811072303,
+      "loss": 0.7231,
+      "mean_token_accuracy": 0.7823591709136963,
+      "num_tokens": 83656834.0,
+      "step": 24590
+    },
+    {
+      "epoch": 0.24689871130916535,
+      "learning_rate": 0.00015062226504476293,
+      "loss": 0.7636,
+      "mean_token_accuracy": 0.7728028655052185,
+      "num_tokens": 83690537.0,
+      "step": 24600
+    },
+    {
+      "epoch": 0.24699907663896584,
+      "learning_rate": 0.00015060219197880284,
+      "loss": 0.7269,
+      "mean_token_accuracy": 0.7785695910453796,
+      "num_tokens": 83724980.0,
+      "step": 24610
+    },
+    {
+      "epoch": 0.24709944196876632,
+      "learning_rate": 0.00015058211891284275,
+      "loss": 0.7324,
+      "mean_token_accuracy": 0.7791578233242035,
+      "num_tokens": 83759574.0,
+      "step": 24620
+    },
+    {
+      "epoch": 0.24719980729856678,
+      "learning_rate": 0.00015056204584688266,
+      "loss": 0.7934,
+      "mean_token_accuracy": 0.7672521889209747,
+      "num_tokens": 83793398.0,
+      "step": 24630
+    },
+    {
+      "epoch": 0.24730017262836726,
+      "learning_rate": 0.00015054197278092257,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7744177103042602,
+      "num_tokens": 83828140.0,
+      "step": 24640
+    },
+    {
+      "epoch": 0.24740053795816774,
+      "learning_rate": 0.00015052189971496248,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.768083530664444,
+      "num_tokens": 83862329.0,
+      "step": 24650
+    },
+    {
+      "epoch": 0.2475009032879682,
+      "learning_rate": 0.00015050182664900238,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.7712616503238678,
+      "num_tokens": 83896804.0,
+      "step": 24660
+    },
+    {
+      "epoch": 0.24760126861776868,
+      "learning_rate": 0.0001504817535830423,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7727303802967072,
+      "num_tokens": 83931689.0,
+      "step": 24670
+    },
+    {
+      "epoch": 0.24770163394756914,
+      "learning_rate": 0.00015046168051708217,
+      "loss": 0.7554,
+      "mean_token_accuracy": 0.7718530833721161,
+      "num_tokens": 83965817.0,
+      "step": 24680
+    },
+    {
+      "epoch": 0.24780199927736962,
+      "learning_rate": 0.0001504416074511221,
+      "loss": 0.766,
+      "mean_token_accuracy": 0.7704915702342987,
+      "num_tokens": 83999083.0,
+      "step": 24690
+    },
+    {
+      "epoch": 0.2479023646071701,
+      "learning_rate": 0.000150421534385162,
+      "loss": 0.7429,
+      "mean_token_accuracy": 0.7724758744239807,
+      "num_tokens": 84033306.0,
+      "step": 24700
+    },
+    {
+      "epoch": 0.24800272993697056,
+      "learning_rate": 0.0001504014613192019,
+      "loss": 0.7658,
+      "mean_token_accuracy": 0.7693391263484954,
+      "num_tokens": 84066880.0,
+      "step": 24710
+    },
+    {
+      "epoch": 0.24810309526677105,
+      "learning_rate": 0.0001503813882532418,
+      "loss": 0.7494,
+      "mean_token_accuracy": 0.770747721195221,
+      "num_tokens": 84100580.0,
+      "step": 24720
+    },
+    {
+      "epoch": 0.24820346059657153,
+      "learning_rate": 0.00015036131518728172,
+      "loss": 0.7875,
+      "mean_token_accuracy": 0.7645402133464814,
+      "num_tokens": 84134540.0,
+      "step": 24730
+    },
+    {
+      "epoch": 0.24830382592637198,
+      "learning_rate": 0.00015034124212132162,
+      "loss": 0.7692,
+      "mean_token_accuracy": 0.7743058323860168,
+      "num_tokens": 84168575.0,
+      "step": 24740
+    },
+    {
+      "epoch": 0.24840419125617247,
+      "learning_rate": 0.00015032116905536153,
+      "loss": 0.7812,
+      "mean_token_accuracy": 0.766360980272293,
+      "num_tokens": 84202924.0,
+      "step": 24750
+    },
+    {
+      "epoch": 0.24850455658597295,
+      "learning_rate": 0.00015030109598940144,
+      "loss": 0.7697,
+      "mean_token_accuracy": 0.7704414248466491,
+      "num_tokens": 84237238.0,
+      "step": 24760
+    },
+    {
+      "epoch": 0.2486049219157734,
+      "learning_rate": 0.00015028102292344132,
+      "loss": 0.7391,
+      "mean_token_accuracy": 0.7736336052417755,
+      "num_tokens": 84271592.0,
+      "step": 24770
+    },
+    {
+      "epoch": 0.2487052872455739,
+      "learning_rate": 0.00015026094985748126,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7742735326290131,
+      "num_tokens": 84305917.0,
+      "step": 24780
+    },
+    {
+      "epoch": 0.24880565257537438,
+      "learning_rate": 0.00015024087679152114,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7742458462715149,
+      "num_tokens": 84339607.0,
+      "step": 24790
+    },
+    {
+      "epoch": 0.24890601790517483,
+      "learning_rate": 0.00015022080372556105,
+      "loss": 0.7454,
+      "mean_token_accuracy": 0.7744569897651672,
+      "num_tokens": 84374014.0,
+      "step": 24800
+    },
+    {
+      "epoch": 0.24900638323497531,
+      "learning_rate": 0.00015020073065960096,
+      "loss": 0.78,
+      "mean_token_accuracy": 0.7679212808609008,
+      "num_tokens": 84407966.0,
+      "step": 24810
+    },
+    {
+      "epoch": 0.24910674856477577,
+      "learning_rate": 0.00015018065759364087,
+      "loss": 0.7903,
+      "mean_token_accuracy": 0.7653574228286744,
+      "num_tokens": 84442126.0,
+      "step": 24820
+    },
+    {
+      "epoch": 0.24920711389457625,
+      "learning_rate": 0.00015016058452768075,
+      "loss": 0.7613,
+      "mean_token_accuracy": 0.7751851081848145,
+      "num_tokens": 84476405.0,
+      "step": 24830
+    },
+    {
+      "epoch": 0.24930747922437674,
+      "learning_rate": 0.00015014051146172068,
+      "loss": 0.7459,
+      "mean_token_accuracy": 0.7735485136508942,
+      "num_tokens": 84510655.0,
+      "step": 24840
+    },
+    {
+      "epoch": 0.2494078445541772,
+      "learning_rate": 0.00015012043839576056,
+      "loss": 0.7649,
+      "mean_token_accuracy": 0.7721766114234925,
+      "num_tokens": 84544543.0,
+      "step": 24850
+    },
+    {
+      "epoch": 0.24950820988397768,
+      "learning_rate": 0.00015010036532980047,
+      "loss": 0.7673,
+      "mean_token_accuracy": 0.7733584702014923,
+      "num_tokens": 84578016.0,
+      "step": 24860
+    },
+    {
+      "epoch": 0.24960857521377816,
+      "learning_rate": 0.0001500802922638404,
+      "loss": 0.7727,
+      "mean_token_accuracy": 0.7719229996204376,
+      "num_tokens": 84612899.0,
+      "step": 24870
+    },
+    {
+      "epoch": 0.24970894054357862,
+      "learning_rate": 0.0001500602191978803,
+      "loss": 0.7685,
+      "mean_token_accuracy": 0.769759064912796,
+      "num_tokens": 84647798.0,
+      "step": 24880
+    },
+    {
+      "epoch": 0.2498093058733791,
+      "learning_rate": 0.0001500401461319202,
+      "loss": 0.754,
+      "mean_token_accuracy": 0.7807211995124816,
+      "num_tokens": 84682489.0,
+      "step": 24890
+    },
+    {
+      "epoch": 0.24990967120317958,
+      "learning_rate": 0.0001500200730659601,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7719671905040741,
+      "num_tokens": 84716528.0,
+      "step": 24900
+    },
+    {
+      "epoch": 0.25001003653298004,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.7432,
+      "mean_token_accuracy": 0.7741679668426513,
+      "num_tokens": 84750516.0,
+      "step": 24910
+    },
+    {
+      "epoch": 0.2501104018627805,
+      "learning_rate": 0.0001499799269340399,
+      "loss": 0.7741,
+      "mean_token_accuracy": 0.769940984249115,
+      "num_tokens": 84784272.0,
+      "step": 24920
+    },
+    {
+      "epoch": 0.250210767192581,
+      "learning_rate": 0.00014995985386807983,
+      "loss": 0.7687,
+      "mean_token_accuracy": 0.7691887438297271,
+      "num_tokens": 84819003.0,
+      "step": 24930
+    },
+    {
+      "epoch": 0.25031113252238146,
+      "learning_rate": 0.0001499397808021197,
+      "loss": 0.7266,
+      "mean_token_accuracy": 0.7773842871189117,
+      "num_tokens": 84852763.0,
+      "step": 24940
+    },
+    {
+      "epoch": 0.2504114978521819,
+      "learning_rate": 0.00014991970773615962,
+      "loss": 0.7593,
+      "mean_token_accuracy": 0.7744570195674896,
+      "num_tokens": 84886921.0,
+      "step": 24950
+    },
+    {
+      "epoch": 0.25051186318198243,
+      "learning_rate": 0.00014989963467019953,
+      "loss": 0.7699,
+      "mean_token_accuracy": 0.7752615630626678,
+      "num_tokens": 84921113.0,
+      "step": 24960
+    },
+    {
+      "epoch": 0.2506122285117829,
+      "learning_rate": 0.00014987956160423944,
+      "loss": 0.7212,
+      "mean_token_accuracy": 0.7837283551692963,
+      "num_tokens": 84955449.0,
+      "step": 24970
+    },
+    {
+      "epoch": 0.25071259384158334,
+      "learning_rate": 0.00014985948853827935,
+      "loss": 0.7636,
+      "mean_token_accuracy": 0.7748461961746216,
+      "num_tokens": 84989985.0,
+      "step": 24980
+    },
+    {
+      "epoch": 0.25081295917138385,
+      "learning_rate": 0.00014983941547231925,
+      "loss": 0.7679,
+      "mean_token_accuracy": 0.7788638472557068,
+      "num_tokens": 85023754.0,
+      "step": 24990
+    },
+    {
+      "epoch": 0.2509133245011843,
+      "learning_rate": 0.00014981934240635916,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7738966584205628,
+      "num_tokens": 85057485.0,
+      "step": 25000
+    },
+    {
+      "epoch": 0.25101368983098477,
+      "learning_rate": 0.00014979926934039904,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.7705737590789795,
+      "num_tokens": 85092345.0,
+      "step": 25010
+    },
+    {
+      "epoch": 0.2511140551607853,
+      "learning_rate": 0.00014977919627443898,
+      "loss": 0.8036,
+      "mean_token_accuracy": 0.7606657981872559,
+      "num_tokens": 85126415.0,
+      "step": 25020
+    },
+    {
+      "epoch": 0.25121442049058573,
+      "learning_rate": 0.00014975912320847886,
+      "loss": 0.7839,
+      "mean_token_accuracy": 0.7678379952907562,
+      "num_tokens": 85160265.0,
+      "step": 25030
+    },
+    {
+      "epoch": 0.2513147858203862,
+      "learning_rate": 0.00014973905014251877,
+      "loss": 0.7654,
+      "mean_token_accuracy": 0.7709371447563171,
+      "num_tokens": 85194420.0,
+      "step": 25040
+    },
+    {
+      "epoch": 0.2514151511501867,
+      "learning_rate": 0.00014971897707655868,
+      "loss": 0.7382,
+      "mean_token_accuracy": 0.7809635996818542,
+      "num_tokens": 85227880.0,
+      "step": 25050
+    },
+    {
+      "epoch": 0.25151551647998716,
+      "learning_rate": 0.00014969890401059859,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.7718024134635926,
+      "num_tokens": 85261717.0,
+      "step": 25060
+    },
+    {
+      "epoch": 0.2516158818097876,
+      "learning_rate": 0.0001496788309446385,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.7697980642318726,
+      "num_tokens": 85295733.0,
+      "step": 25070
+    },
+    {
+      "epoch": 0.2517162471395881,
+      "learning_rate": 0.0001496587578786784,
+      "loss": 0.775,
+      "mean_token_accuracy": 0.771453458070755,
+      "num_tokens": 85329861.0,
+      "step": 25080
+    },
+    {
+      "epoch": 0.2518166124693886,
+      "learning_rate": 0.0001496386848127183,
+      "loss": 0.7659,
+      "mean_token_accuracy": 0.7687193214893341,
+      "num_tokens": 85364228.0,
+      "step": 25090
+    },
+    {
+      "epoch": 0.25191697779918903,
+      "learning_rate": 0.0001496186117467582,
+      "loss": 0.745,
+      "mean_token_accuracy": 0.7802204549312591,
+      "num_tokens": 85397878.0,
+      "step": 25100
+    },
+    {
+      "epoch": 0.25201734312898955,
+      "learning_rate": 0.00014959853868079813,
+      "loss": 0.7364,
+      "mean_token_accuracy": 0.7754930555820465,
+      "num_tokens": 85431906.0,
+      "step": 25110
+    },
+    {
+      "epoch": 0.25211770845879,
+      "learning_rate": 0.000149578465614838,
+      "loss": 0.8023,
+      "mean_token_accuracy": 0.7620309591293335,
+      "num_tokens": 85466672.0,
+      "step": 25120
+    },
+    {
+      "epoch": 0.25221807378859046,
+      "learning_rate": 0.00014955839254887792,
+      "loss": 0.7639,
+      "mean_token_accuracy": 0.7732936799526214,
+      "num_tokens": 85501948.0,
+      "step": 25130
+    },
+    {
+      "epoch": 0.25231843911839097,
+      "learning_rate": 0.00014953831948291783,
+      "loss": 0.7285,
+      "mean_token_accuracy": 0.7803778707981109,
+      "num_tokens": 85535594.0,
+      "step": 25140
+    },
+    {
+      "epoch": 0.2524188044481914,
+      "learning_rate": 0.00014951824641695774,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7724705159664154,
+      "num_tokens": 85570745.0,
+      "step": 25150
+    },
+    {
+      "epoch": 0.2525191697779919,
+      "learning_rate": 0.00014949817335099762,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.7735279023647308,
+      "num_tokens": 85604587.0,
+      "step": 25160
+    },
+    {
+      "epoch": 0.25261953510779234,
+      "learning_rate": 0.00014947810028503755,
+      "loss": 0.7619,
+      "mean_token_accuracy": 0.7704692482948303,
+      "num_tokens": 85638073.0,
+      "step": 25170
+    },
+    {
+      "epoch": 0.25271990043759285,
+      "learning_rate": 0.00014945802721907743,
+      "loss": 0.7564,
+      "mean_token_accuracy": 0.7743849337100983,
+      "num_tokens": 85671963.0,
+      "step": 25180
+    },
+    {
+      "epoch": 0.2528202657673933,
+      "learning_rate": 0.00014943795415311734,
+      "loss": 0.7575,
+      "mean_token_accuracy": 0.774717116355896,
+      "num_tokens": 85706538.0,
+      "step": 25190
+    },
+    {
+      "epoch": 0.25292063109719376,
+      "learning_rate": 0.00014941788108715725,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7788683176040649,
+      "num_tokens": 85740414.0,
+      "step": 25200
+    },
+    {
+      "epoch": 0.25302099642699427,
+      "learning_rate": 0.00014939780802119716,
+      "loss": 0.7752,
+      "mean_token_accuracy": 0.7676084637641907,
+      "num_tokens": 85774633.0,
+      "step": 25210
+    },
+    {
+      "epoch": 0.2531213617567947,
+      "learning_rate": 0.0001493777349552371,
+      "loss": 0.749,
+      "mean_token_accuracy": 0.7740694999694824,
+      "num_tokens": 85808692.0,
+      "step": 25220
+    },
+    {
+      "epoch": 0.2532217270865952,
+      "learning_rate": 0.00014935766188927698,
+      "loss": 0.7251,
+      "mean_token_accuracy": 0.7860980331897736,
+      "num_tokens": 85842126.0,
+      "step": 25230
+    },
+    {
+      "epoch": 0.2533220924163957,
+      "learning_rate": 0.00014933758882331688,
+      "loss": 0.7751,
+      "mean_token_accuracy": 0.7714491665363312,
+      "num_tokens": 85875593.0,
+      "step": 25240
+    },
+    {
+      "epoch": 0.25342245774619615,
+      "learning_rate": 0.0001493175157573568,
+      "loss": 0.7671,
+      "mean_token_accuracy": 0.7751310586929321,
+      "num_tokens": 85909507.0,
+      "step": 25250
+    },
+    {
+      "epoch": 0.2535228230759966,
+      "learning_rate": 0.0001492974426913967,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.7769211292266845,
+      "num_tokens": 85943937.0,
+      "step": 25260
+    },
+    {
+      "epoch": 0.2536231884057971,
+      "learning_rate": 0.00014927736962543658,
+      "loss": 0.7508,
+      "mean_token_accuracy": 0.7736414492130279,
+      "num_tokens": 85977417.0,
+      "step": 25270
+    },
+    {
+      "epoch": 0.2537235537355976,
+      "learning_rate": 0.00014925729655947652,
+      "loss": 0.8044,
+      "mean_token_accuracy": 0.7656659841537475,
+      "num_tokens": 86010732.0,
+      "step": 25280
+    },
+    {
+      "epoch": 0.25382391906539803,
+      "learning_rate": 0.0001492372234935164,
+      "loss": 0.7445,
+      "mean_token_accuracy": 0.7763683319091796,
+      "num_tokens": 86044764.0,
+      "step": 25290
+    },
+    {
+      "epoch": 0.25392428439519854,
+      "learning_rate": 0.0001492171504275563,
+      "loss": 0.7432,
+      "mean_token_accuracy": 0.7775234937667846,
+      "num_tokens": 86078781.0,
+      "step": 25300
+    },
+    {
+      "epoch": 0.254024649724999,
+      "learning_rate": 0.00014919707736159622,
+      "loss": 0.7654,
+      "mean_token_accuracy": 0.7700466334819793,
+      "num_tokens": 86113326.0,
+      "step": 25310
+    },
+    {
+      "epoch": 0.25412501505479945,
+      "learning_rate": 0.00014917700429563612,
+      "loss": 0.7856,
+      "mean_token_accuracy": 0.7683400988578797,
+      "num_tokens": 86147025.0,
+      "step": 25320
+    },
+    {
+      "epoch": 0.25422538038459996,
+      "learning_rate": 0.00014915693122967603,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.7748342633247376,
+      "num_tokens": 86181539.0,
+      "step": 25330
+    },
+    {
+      "epoch": 0.2543257457144004,
+      "learning_rate": 0.00014913685816371594,
+      "loss": 0.7805,
+      "mean_token_accuracy": 0.768756777048111,
+      "num_tokens": 86215963.0,
+      "step": 25340
+    },
+    {
+      "epoch": 0.2544261110442009,
+      "learning_rate": 0.00014911678509775585,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7703627109527588,
+      "num_tokens": 86249684.0,
+      "step": 25350
+    },
+    {
+      "epoch": 0.2545264763740014,
+      "learning_rate": 0.00014909671203179573,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7684246659278869,
+      "num_tokens": 86282686.0,
+      "step": 25360
+    },
+    {
+      "epoch": 0.25462684170380184,
+      "learning_rate": 0.00014907663896583567,
+      "loss": 0.7586,
+      "mean_token_accuracy": 0.7752669513225555,
+      "num_tokens": 86317014.0,
+      "step": 25370
+    },
+    {
+      "epoch": 0.2547272070336023,
+      "learning_rate": 0.00014905656589987555,
+      "loss": 0.7911,
+      "mean_token_accuracy": 0.7662883758544922,
+      "num_tokens": 86351353.0,
+      "step": 25380
+    },
+    {
+      "epoch": 0.2548275723634028,
+      "learning_rate": 0.00014903649283391546,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7772423982620239,
+      "num_tokens": 86385009.0,
+      "step": 25390
+    },
+    {
+      "epoch": 0.25492793769320327,
+      "learning_rate": 0.00014901641976795536,
+      "loss": 0.757,
+      "mean_token_accuracy": 0.776004284620285,
+      "num_tokens": 86419271.0,
+      "step": 25400
+    },
+    {
+      "epoch": 0.2550283030230037,
+      "learning_rate": 0.00014899634670199527,
+      "loss": 0.756,
+      "mean_token_accuracy": 0.7729967117309571,
+      "num_tokens": 86453343.0,
+      "step": 25410
+    },
+    {
+      "epoch": 0.25512866835280423,
+      "learning_rate": 0.00014897627363603518,
+      "loss": 0.7826,
+      "mean_token_accuracy": 0.7646670699119568,
+      "num_tokens": 86487554.0,
+      "step": 25420
+    },
+    {
+      "epoch": 0.2552290336826047,
+      "learning_rate": 0.0001489562005700751,
+      "loss": 0.7624,
+      "mean_token_accuracy": 0.7803102493286133,
+      "num_tokens": 86521925.0,
+      "step": 25430
+    },
+    {
+      "epoch": 0.25532939901240514,
+      "learning_rate": 0.000148936127504115,
+      "loss": 0.7857,
+      "mean_token_accuracy": 0.7645636975765229,
+      "num_tokens": 86555158.0,
+      "step": 25440
+    },
+    {
+      "epoch": 0.2554297643422056,
+      "learning_rate": 0.00014891605443815488,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.7731352686882019,
+      "num_tokens": 86589507.0,
+      "step": 25450
+    },
+    {
+      "epoch": 0.2555301296720061,
+      "learning_rate": 0.00014889598137219482,
+      "loss": 0.7557,
+      "mean_token_accuracy": 0.7728032350540162,
+      "num_tokens": 86622919.0,
+      "step": 25460
+    },
+    {
+      "epoch": 0.25563049500180657,
+      "learning_rate": 0.0001488759083062347,
+      "loss": 0.8059,
+      "mean_token_accuracy": 0.7628725171089172,
+      "num_tokens": 86656473.0,
+      "step": 25470
+    },
+    {
+      "epoch": 0.255730860331607,
+      "learning_rate": 0.0001488558352402746,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7711473762989044,
+      "num_tokens": 86690523.0,
+      "step": 25480
+    },
+    {
+      "epoch": 0.25583122566140754,
+      "learning_rate": 0.0001488357621743145,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.7721888720989227,
+      "num_tokens": 86724478.0,
+      "step": 25490
+    },
+    {
+      "epoch": 0.255931590991208,
+      "learning_rate": 0.00014881568910835442,
+      "loss": 0.8016,
+      "mean_token_accuracy": 0.7639282763004303,
+      "num_tokens": 86757913.0,
+      "step": 25500
+    },
+    {
+      "epoch": 0.25603195632100845,
+      "learning_rate": 0.0001487956160423943,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.7748159408569336,
+      "num_tokens": 86791797.0,
+      "step": 25510
+    },
+    {
+      "epoch": 0.25613232165080896,
+      "learning_rate": 0.00014877554297643424,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7691547334194183,
+      "num_tokens": 86825417.0,
+      "step": 25520
+    },
+    {
+      "epoch": 0.2562326869806094,
+      "learning_rate": 0.00014875546991047412,
+      "loss": 0.7442,
+      "mean_token_accuracy": 0.7793104350566864,
+      "num_tokens": 86858769.0,
+      "step": 25530
+    },
+    {
+      "epoch": 0.25633305231040987,
+      "learning_rate": 0.00014873539684451403,
+      "loss": 0.7636,
+      "mean_token_accuracy": 0.7719816446304322,
+      "num_tokens": 86892597.0,
+      "step": 25540
+    },
+    {
+      "epoch": 0.2564334176402104,
+      "learning_rate": 0.00014871532377855394,
+      "loss": 0.7282,
+      "mean_token_accuracy": 0.7823039889335632,
+      "num_tokens": 86926765.0,
+      "step": 25550
+    },
+    {
+      "epoch": 0.25653378297001084,
+      "learning_rate": 0.00014869525071259385,
+      "loss": 0.7331,
+      "mean_token_accuracy": 0.778553056716919,
+      "num_tokens": 86962121.0,
+      "step": 25560
+    },
+    {
+      "epoch": 0.2566341482998113,
+      "learning_rate": 0.00014867517764663375,
+      "loss": 0.7989,
+      "mean_token_accuracy": 0.7632017970085144,
+      "num_tokens": 86996587.0,
+      "step": 25570
+    },
+    {
+      "epoch": 0.2567345136296118,
+      "learning_rate": 0.00014865510458067366,
+      "loss": 0.8135,
+      "mean_token_accuracy": 0.7593931913375854,
+      "num_tokens": 87029765.0,
+      "step": 25580
+    },
+    {
+      "epoch": 0.25683487895941226,
+      "learning_rate": 0.00014863503151471357,
+      "loss": 0.7807,
+      "mean_token_accuracy": 0.7670421421527862,
+      "num_tokens": 87063804.0,
+      "step": 25590
+    },
+    {
+      "epoch": 0.2569352442892127,
+      "learning_rate": 0.00014861495844875345,
+      "loss": 0.7511,
+      "mean_token_accuracy": 0.7708527326583863,
+      "num_tokens": 87098792.0,
+      "step": 25600
+    },
+    {
+      "epoch": 0.25703560961901323,
+      "learning_rate": 0.0001485948853827934,
+      "loss": 0.7642,
+      "mean_token_accuracy": 0.7667521178722382,
+      "num_tokens": 87133042.0,
+      "step": 25610
+    },
+    {
+      "epoch": 0.2571359749488137,
+      "learning_rate": 0.00014857481231683327,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.7766914963722229,
+      "num_tokens": 87166899.0,
+      "step": 25620
+    },
+    {
+      "epoch": 0.25723634027861414,
+      "learning_rate": 0.00014855473925087318,
+      "loss": 0.7379,
+      "mean_token_accuracy": 0.7837460577487946,
+      "num_tokens": 87201686.0,
+      "step": 25630
+    },
+    {
+      "epoch": 0.25733670560841465,
+      "learning_rate": 0.00014853466618491309,
+      "loss": 0.7499,
+      "mean_token_accuracy": 0.7803411364555359,
+      "num_tokens": 87235747.0,
+      "step": 25640
+    },
+    {
+      "epoch": 0.2574370709382151,
+      "learning_rate": 0.000148514593118953,
+      "loss": 0.7564,
+      "mean_token_accuracy": 0.7728347659111023,
+      "num_tokens": 87270207.0,
+      "step": 25650
+    },
+    {
+      "epoch": 0.25753743626801556,
+      "learning_rate": 0.0001484945200529929,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7803968906402587,
+      "num_tokens": 87303823.0,
+      "step": 25660
+    },
+    {
+      "epoch": 0.2576378015978161,
+      "learning_rate": 0.0001484744469870328,
+      "loss": 0.7389,
+      "mean_token_accuracy": 0.7769669115543365,
+      "num_tokens": 87338022.0,
+      "step": 25670
+    },
+    {
+      "epoch": 0.25773816692761653,
+      "learning_rate": 0.00014845437392107272,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7749365508556366,
+      "num_tokens": 87371875.0,
+      "step": 25680
+    },
+    {
+      "epoch": 0.257838532257417,
+      "learning_rate": 0.0001484343008551126,
+      "loss": 0.7358,
+      "mean_token_accuracy": 0.7795381069183349,
+      "num_tokens": 87406221.0,
+      "step": 25690
+    },
+    {
+      "epoch": 0.2579388975872175,
+      "learning_rate": 0.00014841422778915254,
+      "loss": 0.7875,
+      "mean_token_accuracy": 0.7627764105796814,
+      "num_tokens": 87440019.0,
+      "step": 25700
+    },
+    {
+      "epoch": 0.25803926291701795,
+      "learning_rate": 0.00014839415472319242,
+      "loss": 0.7575,
+      "mean_token_accuracy": 0.7729061365127563,
+      "num_tokens": 87473672.0,
+      "step": 25710
+    },
+    {
+      "epoch": 0.2581396282468184,
+      "learning_rate": 0.00014837408165723233,
+      "loss": 0.7385,
+      "mean_token_accuracy": 0.7753385066986084,
+      "num_tokens": 87507791.0,
+      "step": 25720
+    },
+    {
+      "epoch": 0.2582399935766189,
+      "learning_rate": 0.00014835400859127223,
+      "loss": 0.7725,
+      "mean_token_accuracy": 0.769771808385849,
+      "num_tokens": 87541696.0,
+      "step": 25730
+    },
+    {
+      "epoch": 0.2583403589064194,
+      "learning_rate": 0.00014833393552531214,
+      "loss": 0.7594,
+      "mean_token_accuracy": 0.7752532303333283,
+      "num_tokens": 87575872.0,
+      "step": 25740
+    },
+    {
+      "epoch": 0.25844072423621983,
+      "learning_rate": 0.00014831386245935202,
+      "loss": 0.7869,
+      "mean_token_accuracy": 0.7630620658397674,
+      "num_tokens": 87609585.0,
+      "step": 25750
+    },
+    {
+      "epoch": 0.2585410895660203,
+      "learning_rate": 0.00014829378939339196,
+      "loss": 0.7837,
+      "mean_token_accuracy": 0.7641562402248383,
+      "num_tokens": 87642769.0,
+      "step": 25760
+    },
+    {
+      "epoch": 0.2586414548958208,
+      "learning_rate": 0.00014827371632743187,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7799721121788025,
+      "num_tokens": 87677019.0,
+      "step": 25770
+    },
+    {
+      "epoch": 0.25874182022562126,
+      "learning_rate": 0.00014825364326147178,
+      "loss": 0.7809,
+      "mean_token_accuracy": 0.7631854057312012,
+      "num_tokens": 87710569.0,
+      "step": 25780
+    },
+    {
+      "epoch": 0.2588421855554217,
+      "learning_rate": 0.00014823357019551169,
+      "loss": 0.7645,
+      "mean_token_accuracy": 0.768880182504654,
+      "num_tokens": 87744559.0,
+      "step": 25790
+    },
+    {
+      "epoch": 0.2589425508852222,
+      "learning_rate": 0.00014821349712955157,
+      "loss": 0.7331,
+      "mean_token_accuracy": 0.7779667556285859,
+      "num_tokens": 87779111.0,
+      "step": 25800
+    },
+    {
+      "epoch": 0.2590429162150227,
+      "learning_rate": 0.0001481934240635915,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7720450103282929,
+      "num_tokens": 87812979.0,
+      "step": 25810
+    },
+    {
+      "epoch": 0.25914328154482313,
+      "learning_rate": 0.00014817335099763138,
+      "loss": 0.7678,
+      "mean_token_accuracy": 0.7701455950737,
+      "num_tokens": 87846804.0,
+      "step": 25820
+    },
+    {
+      "epoch": 0.25924364687462365,
+      "learning_rate": 0.0001481532779316713,
+      "loss": 0.7376,
+      "mean_token_accuracy": 0.7828764975070953,
+      "num_tokens": 87880594.0,
+      "step": 25830
+    },
+    {
+      "epoch": 0.2593440122044241,
+      "learning_rate": 0.0001481332048657112,
+      "loss": 0.7419,
+      "mean_token_accuracy": 0.7778504431247711,
+      "num_tokens": 87915015.0,
+      "step": 25840
+    },
+    {
+      "epoch": 0.25944437753422456,
+      "learning_rate": 0.0001481131317997511,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7739579916000366,
+      "num_tokens": 87949304.0,
+      "step": 25850
+    },
+    {
+      "epoch": 0.25954474286402507,
+      "learning_rate": 0.000148093058733791,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7708935976028443,
+      "num_tokens": 87984007.0,
+      "step": 25860
+    },
+    {
+      "epoch": 0.2596451081938255,
+      "learning_rate": 0.00014807298566783093,
+      "loss": 0.7483,
+      "mean_token_accuracy": 0.7781088590621948,
+      "num_tokens": 88017751.0,
+      "step": 25870
+    },
+    {
+      "epoch": 0.259745473523626,
+      "learning_rate": 0.0001480529126018708,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7717514097690582,
+      "num_tokens": 88051964.0,
+      "step": 25880
+    },
+    {
+      "epoch": 0.2598458388534265,
+      "learning_rate": 0.00014803283953591072,
+      "loss": 0.7754,
+      "mean_token_accuracy": 0.7723719537258148,
+      "num_tokens": 88086205.0,
+      "step": 25890
+    },
+    {
+      "epoch": 0.25994620418322695,
+      "learning_rate": 0.00014801276646995062,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.7652353048324585,
+      "num_tokens": 88120072.0,
+      "step": 25900
+    },
+    {
+      "epoch": 0.2600465695130274,
+      "learning_rate": 0.00014799269340399053,
+      "loss": 0.7664,
+      "mean_token_accuracy": 0.7733864307403564,
+      "num_tokens": 88154084.0,
+      "step": 25910
+    },
+    {
+      "epoch": 0.2601469348428279,
+      "learning_rate": 0.00014797262033803044,
+      "loss": 0.7458,
+      "mean_token_accuracy": 0.7812437653541565,
+      "num_tokens": 88187507.0,
+      "step": 25920
+    },
+    {
+      "epoch": 0.26024730017262837,
+      "learning_rate": 0.00014795254727207035,
+      "loss": 0.7518,
+      "mean_token_accuracy": 0.781405258178711,
+      "num_tokens": 88222091.0,
+      "step": 25930
+    },
+    {
+      "epoch": 0.2603476655024288,
+      "learning_rate": 0.00014793247420611026,
+      "loss": 0.758,
+      "mean_token_accuracy": 0.7747311294078827,
+      "num_tokens": 88256756.0,
+      "step": 25940
+    },
+    {
+      "epoch": 0.26044803083222934,
+      "learning_rate": 0.00014791240114015014,
+      "loss": 0.7998,
+      "mean_token_accuracy": 0.76486234664917,
+      "num_tokens": 88291051.0,
+      "step": 25950
+    },
+    {
+      "epoch": 0.2605483961620298,
+      "learning_rate": 0.00014789232807419007,
+      "loss": 0.7442,
+      "mean_token_accuracy": 0.774928230047226,
+      "num_tokens": 88325971.0,
+      "step": 25960
+    },
+    {
+      "epoch": 0.26064876149183025,
+      "learning_rate": 0.00014787225500822996,
+      "loss": 0.7897,
+      "mean_token_accuracy": 0.7625066220760346,
+      "num_tokens": 88359212.0,
+      "step": 25970
+    },
+    {
+      "epoch": 0.26074912682163076,
+      "learning_rate": 0.00014785218194226986,
+      "loss": 0.7635,
+      "mean_token_accuracy": 0.7737499475479126,
+      "num_tokens": 88393800.0,
+      "step": 25980
+    },
+    {
+      "epoch": 0.2608494921514312,
+      "learning_rate": 0.00014783210887630977,
+      "loss": 0.7666,
+      "mean_token_accuracy": 0.7730528473854065,
+      "num_tokens": 88427377.0,
+      "step": 25990
+    },
+    {
+      "epoch": 0.2609498574812317,
+      "learning_rate": 0.00014781203581034968,
+      "loss": 0.754,
+      "mean_token_accuracy": 0.7699731945991516,
+      "num_tokens": 88461009.0,
+      "step": 26000
+    },
+    {
+      "epoch": 0.2610502228110322,
+      "learning_rate": 0.0001477919627443896,
+      "loss": 0.7432,
+      "mean_token_accuracy": 0.7773293673992157,
+      "num_tokens": 88495592.0,
+      "step": 26010
+    },
+    {
+      "epoch": 0.26115058814083264,
+      "learning_rate": 0.0001477718896784295,
+      "loss": 0.7381,
+      "mean_token_accuracy": 0.7789890348911286,
+      "num_tokens": 88529869.0,
+      "step": 26020
+    },
+    {
+      "epoch": 0.2612509534706331,
+      "learning_rate": 0.0001477518166124694,
+      "loss": 0.7281,
+      "mean_token_accuracy": 0.7793543338775635,
+      "num_tokens": 88564353.0,
+      "step": 26030
+    },
+    {
+      "epoch": 0.26135131880043355,
+      "learning_rate": 0.0001477317435465093,
+      "loss": 0.7974,
+      "mean_token_accuracy": 0.7633599936962128,
+      "num_tokens": 88597465.0,
+      "step": 26040
+    },
+    {
+      "epoch": 0.26145168413023406,
+      "learning_rate": 0.00014771167048054922,
+      "loss": 0.7659,
+      "mean_token_accuracy": 0.7736822187900543,
+      "num_tokens": 88632088.0,
+      "step": 26050
+    },
+    {
+      "epoch": 0.2615520494600345,
+      "learning_rate": 0.0001476915974145891,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.775991004705429,
+      "num_tokens": 88666349.0,
+      "step": 26060
+    },
+    {
+      "epoch": 0.261652414789835,
+      "learning_rate": 0.000147671524348629,
+      "loss": 0.7854,
+      "mean_token_accuracy": 0.767848151922226,
+      "num_tokens": 88700496.0,
+      "step": 26070
+    },
+    {
+      "epoch": 0.2617527801196355,
+      "learning_rate": 0.00014765145128266892,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7757873892784118,
+      "num_tokens": 88734242.0,
+      "step": 26080
+    },
+    {
+      "epoch": 0.26185314544943594,
+      "learning_rate": 0.00014763137821670883,
+      "loss": 0.7732,
+      "mean_token_accuracy": 0.7637600004673004,
+      "num_tokens": 88768767.0,
+      "step": 26090
+    },
+    {
+      "epoch": 0.2619535107792364,
+      "learning_rate": 0.0001476113051507487,
+      "loss": 0.7515,
+      "mean_token_accuracy": 0.7808138072490692,
+      "num_tokens": 88802694.0,
+      "step": 26100
+    },
+    {
+      "epoch": 0.2620538761090369,
+      "learning_rate": 0.00014759123208478865,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7718406736850738,
+      "num_tokens": 88837667.0,
+      "step": 26110
+    },
+    {
+      "epoch": 0.26215424143883737,
+      "learning_rate": 0.00014757115901882856,
+      "loss": 0.7647,
+      "mean_token_accuracy": 0.7676180183887482,
+      "num_tokens": 88871970.0,
+      "step": 26120
+    },
+    {
+      "epoch": 0.2622546067686378,
+      "learning_rate": 0.00014755108595286844,
+      "loss": 0.763,
+      "mean_token_accuracy": 0.7729875087738037,
+      "num_tokens": 88906196.0,
+      "step": 26130
+    },
+    {
+      "epoch": 0.26235497209843833,
+      "learning_rate": 0.00014753101288690837,
+      "loss": 0.7668,
+      "mean_token_accuracy": 0.765146690607071,
+      "num_tokens": 88940835.0,
+      "step": 26140
+    },
+    {
+      "epoch": 0.2624553374282388,
+      "learning_rate": 0.00014751093982094825,
+      "loss": 0.7629,
+      "mean_token_accuracy": 0.7752687871456146,
+      "num_tokens": 88974742.0,
+      "step": 26150
+    },
+    {
+      "epoch": 0.26255570275803924,
+      "learning_rate": 0.00014749086675498816,
+      "loss": 0.7757,
+      "mean_token_accuracy": 0.766808158159256,
+      "num_tokens": 89008704.0,
+      "step": 26160
+    },
+    {
+      "epoch": 0.26265606808783976,
+      "learning_rate": 0.00014747079368902807,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7748014807701111,
+      "num_tokens": 89042137.0,
+      "step": 26170
+    },
+    {
+      "epoch": 0.2627564334176402,
+      "learning_rate": 0.00014745072062306798,
+      "loss": 0.777,
+      "mean_token_accuracy": 0.7709549844264985,
+      "num_tokens": 89076907.0,
+      "step": 26180
+    },
+    {
+      "epoch": 0.26285679874744067,
+      "learning_rate": 0.00014743064755710786,
+      "loss": 0.7771,
+      "mean_token_accuracy": 0.769685173034668,
+      "num_tokens": 89111152.0,
+      "step": 26190
+    },
+    {
+      "epoch": 0.2629571640772412,
+      "learning_rate": 0.0001474105744911478,
+      "loss": 0.7893,
+      "mean_token_accuracy": 0.7698112249374389,
+      "num_tokens": 89144779.0,
+      "step": 26200
+    },
+    {
+      "epoch": 0.26305752940704163,
+      "learning_rate": 0.00014739050142518768,
+      "loss": 0.7556,
+      "mean_token_accuracy": 0.7724555194377899,
+      "num_tokens": 89179208.0,
+      "step": 26210
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "learning_rate": 0.00014737042835922759,
+      "loss": 0.764,
+      "mean_token_accuracy": 0.7716877400875092,
+      "num_tokens": 89214342.0,
+      "step": 26220
+    },
+    {
+      "epoch": 0.2632582600666426,
+      "learning_rate": 0.0001473503552932675,
+      "loss": 0.7617,
+      "mean_token_accuracy": 0.7650292932987213,
+      "num_tokens": 89248605.0,
+      "step": 26230
+    },
+    {
+      "epoch": 0.26335862539644306,
+      "learning_rate": 0.0001473302822273074,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.7738565146923065,
+      "num_tokens": 89283124.0,
+      "step": 26240
+    },
+    {
+      "epoch": 0.2634589907262435,
+      "learning_rate": 0.0001473102091613473,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.7719832837581635,
+      "num_tokens": 89317935.0,
+      "step": 26250
+    },
+    {
+      "epoch": 0.263559356056044,
+      "learning_rate": 0.00014729013609538722,
+      "loss": 0.7356,
+      "mean_token_accuracy": 0.7735422611236572,
+      "num_tokens": 89351818.0,
+      "step": 26260
+    },
+    {
+      "epoch": 0.2636597213858445,
+      "learning_rate": 0.00014727006302942713,
+      "loss": 0.7358,
+      "mean_token_accuracy": 0.7762411177158356,
+      "num_tokens": 89386435.0,
+      "step": 26270
+    },
+    {
+      "epoch": 0.26376008671564494,
+      "learning_rate": 0.000147249989963467,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7716344475746155,
+      "num_tokens": 89419913.0,
+      "step": 26280
+    },
+    {
+      "epoch": 0.26386045204544545,
+      "learning_rate": 0.00014722991689750694,
+      "loss": 0.756,
+      "mean_token_accuracy": 0.7759276568889618,
+      "num_tokens": 89453983.0,
+      "step": 26290
+    },
+    {
+      "epoch": 0.2639608173752459,
+      "learning_rate": 0.00014720984383154683,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7660773396492004,
+      "num_tokens": 89488193.0,
+      "step": 26300
+    },
+    {
+      "epoch": 0.26406118270504636,
+      "learning_rate": 0.00014718977076558673,
+      "loss": 0.7792,
+      "mean_token_accuracy": 0.7708575546741485,
+      "num_tokens": 89522827.0,
+      "step": 26310
+    },
+    {
+      "epoch": 0.2641615480348468,
+      "learning_rate": 0.00014716969769962664,
+      "loss": 0.7586,
+      "mean_token_accuracy": 0.7754275083541871,
+      "num_tokens": 89557222.0,
+      "step": 26320
+    },
+    {
+      "epoch": 0.2642619133646473,
+      "learning_rate": 0.00014714962463366655,
+      "loss": 0.7269,
+      "mean_token_accuracy": 0.7784863173961639,
+      "num_tokens": 89590824.0,
+      "step": 26330
+    },
+    {
+      "epoch": 0.2643622786944478,
+      "learning_rate": 0.00014712955156770646,
+      "loss": 0.7862,
+      "mean_token_accuracy": 0.7694959044456482,
+      "num_tokens": 89624242.0,
+      "step": 26340
+    },
+    {
+      "epoch": 0.26446264402424824,
+      "learning_rate": 0.00014710947850174637,
+      "loss": 0.774,
+      "mean_token_accuracy": 0.7698050498962402,
+      "num_tokens": 89658183.0,
+      "step": 26350
+    },
+    {
+      "epoch": 0.26456300935404875,
+      "learning_rate": 0.00014708940543578628,
+      "loss": 0.7435,
+      "mean_token_accuracy": 0.7790042400360108,
+      "num_tokens": 89692498.0,
+      "step": 26360
+    },
+    {
+      "epoch": 0.2646633746838492,
+      "learning_rate": 0.00014706933236982618,
+      "loss": 0.7808,
+      "mean_token_accuracy": 0.7702409267425537,
+      "num_tokens": 89726203.0,
+      "step": 26370
+    },
+    {
+      "epoch": 0.26476374001364966,
+      "learning_rate": 0.0001470492593038661,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7702608048915863,
+      "num_tokens": 89761020.0,
+      "step": 26380
+    },
+    {
+      "epoch": 0.2648641053434502,
+      "learning_rate": 0.00014702918623790597,
+      "loss": 0.7636,
+      "mean_token_accuracy": 0.7724583685398102,
+      "num_tokens": 89794899.0,
+      "step": 26390
+    },
+    {
+      "epoch": 0.26496447067325063,
+      "learning_rate": 0.0001470091131719459,
+      "loss": 0.7647,
+      "mean_token_accuracy": 0.7716863036155701,
+      "num_tokens": 89829349.0,
+      "step": 26400
+    },
+    {
+      "epoch": 0.2650648360030511,
+      "learning_rate": 0.0001469890401059858,
+      "loss": 0.7896,
+      "mean_token_accuracy": 0.7599710524082184,
+      "num_tokens": 89863346.0,
+      "step": 26410
+    },
+    {
+      "epoch": 0.2651652013328516,
+      "learning_rate": 0.0001469689670400257,
+      "loss": 0.7805,
+      "mean_token_accuracy": 0.7709419846534729,
+      "num_tokens": 89897702.0,
+      "step": 26420
+    },
+    {
+      "epoch": 0.26526556666265205,
+      "learning_rate": 0.0001469488939740656,
+      "loss": 0.7534,
+      "mean_token_accuracy": 0.7745054662227631,
+      "num_tokens": 89931936.0,
+      "step": 26430
+    },
+    {
+      "epoch": 0.2653659319924525,
+      "learning_rate": 0.00014692882090810552,
+      "loss": 0.773,
+      "mean_token_accuracy": 0.7691297650337219,
+      "num_tokens": 89965848.0,
+      "step": 26440
+    },
+    {
+      "epoch": 0.265466297322253,
+      "learning_rate": 0.0001469087478421454,
+      "loss": 0.7411,
+      "mean_token_accuracy": 0.7742989897727967,
+      "num_tokens": 90000381.0,
+      "step": 26450
+    },
+    {
+      "epoch": 0.2655666626520535,
+      "learning_rate": 0.00014688867477618533,
+      "loss": 0.7487,
+      "mean_token_accuracy": 0.7739397466182709,
+      "num_tokens": 90034983.0,
+      "step": 26460
+    },
+    {
+      "epoch": 0.26566702798185393,
+      "learning_rate": 0.00014686860171022524,
+      "loss": 0.7722,
+      "mean_token_accuracy": 0.7688474774360656,
+      "num_tokens": 90068902.0,
+      "step": 26470
+    },
+    {
+      "epoch": 0.26576739331165444,
+      "learning_rate": 0.00014684852864426512,
+      "loss": 0.7474,
+      "mean_token_accuracy": 0.7761471569538116,
+      "num_tokens": 90102872.0,
+      "step": 26480
+    },
+    {
+      "epoch": 0.2658677586414549,
+      "learning_rate": 0.00014682845557830506,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.7720959186553955,
+      "num_tokens": 90136397.0,
+      "step": 26490
+    },
+    {
+      "epoch": 0.26596812397125535,
+      "learning_rate": 0.00014680838251234494,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.769785875082016,
+      "num_tokens": 90169930.0,
+      "step": 26500
+    },
+    {
+      "epoch": 0.26606848930105587,
+      "learning_rate": 0.00014678830944638485,
+      "loss": 0.7813,
+      "mean_token_accuracy": 0.7711447656154633,
+      "num_tokens": 90204175.0,
+      "step": 26510
+    },
+    {
+      "epoch": 0.2661688546308563,
+      "learning_rate": 0.00014676823638042476,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7681370198726654,
+      "num_tokens": 90238114.0,
+      "step": 26520
+    },
+    {
+      "epoch": 0.2662692199606568,
+      "learning_rate": 0.00014674816331446467,
+      "loss": 0.7562,
+      "mean_token_accuracy": 0.7734508454799652,
+      "num_tokens": 90272322.0,
+      "step": 26530
+    },
+    {
+      "epoch": 0.2663695852904573,
+      "learning_rate": 0.00014672809024850455,
+      "loss": 0.7734,
+      "mean_token_accuracy": 0.770289134979248,
+      "num_tokens": 90305800.0,
+      "step": 26540
+    },
+    {
+      "epoch": 0.26646995062025775,
+      "learning_rate": 0.00014670801718254448,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.7708206474781036,
+      "num_tokens": 90339684.0,
+      "step": 26550
+    },
+    {
+      "epoch": 0.2665703159500582,
+      "learning_rate": 0.00014668794411658436,
+      "loss": 0.7423,
+      "mean_token_accuracy": 0.7786689579486847,
+      "num_tokens": 90372906.0,
+      "step": 26560
+    },
+    {
+      "epoch": 0.2666706812798587,
+      "learning_rate": 0.00014666787105062427,
+      "loss": 0.7369,
+      "mean_token_accuracy": 0.7815804660320282,
+      "num_tokens": 90406942.0,
+      "step": 26570
+    },
+    {
+      "epoch": 0.26677104660965917,
+      "learning_rate": 0.00014664779798466418,
+      "loss": 0.776,
+      "mean_token_accuracy": 0.7725221812725067,
+      "num_tokens": 90440698.0,
+      "step": 26580
+    },
+    {
+      "epoch": 0.2668714119394596,
+      "learning_rate": 0.0001466277249187041,
+      "loss": 0.7444,
+      "mean_token_accuracy": 0.7757733106613159,
+      "num_tokens": 90474987.0,
+      "step": 26590
+    },
+    {
+      "epoch": 0.2669717772692601,
+      "learning_rate": 0.000146607651852744,
+      "loss": 0.7831,
+      "mean_token_accuracy": 0.7649937748908997,
+      "num_tokens": 90508097.0,
+      "step": 26600
+    },
+    {
+      "epoch": 0.2670721425990606,
+      "learning_rate": 0.0001465875787867839,
+      "loss": 0.7617,
+      "mean_token_accuracy": 0.7734456777572631,
+      "num_tokens": 90542306.0,
+      "step": 26610
+    },
+    {
+      "epoch": 0.26717250792886105,
+      "learning_rate": 0.00014656750572082381,
+      "loss": 0.7369,
+      "mean_token_accuracy": 0.7760408341884613,
+      "num_tokens": 90576882.0,
+      "step": 26620
+    },
+    {
+      "epoch": 0.2672728732586615,
+      "learning_rate": 0.0001465474326548637,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.771179610490799,
+      "num_tokens": 90611481.0,
+      "step": 26630
+    },
+    {
+      "epoch": 0.267373238588462,
+      "learning_rate": 0.00014652735958890363,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.7678513467311859,
+      "num_tokens": 90645536.0,
+      "step": 26640
+    },
+    {
+      "epoch": 0.26747360391826247,
+      "learning_rate": 0.0001465072865229435,
+      "loss": 0.7648,
+      "mean_token_accuracy": 0.7721072733402252,
+      "num_tokens": 90677927.0,
+      "step": 26650
+    },
+    {
+      "epoch": 0.2675739692480629,
+      "learning_rate": 0.00014648721345698342,
+      "loss": 0.756,
+      "mean_token_accuracy": 0.7761224567890167,
+      "num_tokens": 90711841.0,
+      "step": 26660
+    },
+    {
+      "epoch": 0.26767433457786344,
+      "learning_rate": 0.00014646714039102333,
+      "loss": 0.798,
+      "mean_token_accuracy": 0.7591996788978577,
+      "num_tokens": 90745399.0,
+      "step": 26670
+    },
+    {
+      "epoch": 0.2677746999076639,
+      "learning_rate": 0.00014644706732506324,
+      "loss": 0.7148,
+      "mean_token_accuracy": 0.7797797799110413,
+      "num_tokens": 90780285.0,
+      "step": 26680
+    },
+    {
+      "epoch": 0.26787506523746435,
+      "learning_rate": 0.00014642699425910315,
+      "loss": 0.7872,
+      "mean_token_accuracy": 0.7621639013290405,
+      "num_tokens": 90814627.0,
+      "step": 26690
+    },
+    {
+      "epoch": 0.26797543056726486,
+      "learning_rate": 0.00014640692119314305,
+      "loss": 0.748,
+      "mean_token_accuracy": 0.7777003645896912,
+      "num_tokens": 90848678.0,
+      "step": 26700
+    },
+    {
+      "epoch": 0.2680757958970653,
+      "learning_rate": 0.00014638684812718296,
+      "loss": 0.7405,
+      "mean_token_accuracy": 0.7803672850131989,
+      "num_tokens": 90883023.0,
+      "step": 26710
+    },
+    {
+      "epoch": 0.2681761612268658,
+      "learning_rate": 0.00014636677506122284,
+      "loss": 0.7517,
+      "mean_token_accuracy": 0.776670640707016,
+      "num_tokens": 90916308.0,
+      "step": 26720
+    },
+    {
+      "epoch": 0.2682765265566663,
+      "learning_rate": 0.00014634670199526278,
+      "loss": 0.7811,
+      "mean_token_accuracy": 0.7678254127502442,
+      "num_tokens": 90950190.0,
+      "step": 26730
+    },
+    {
+      "epoch": 0.26837689188646674,
+      "learning_rate": 0.00014632662892930266,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.7694431900978088,
+      "num_tokens": 90984264.0,
+      "step": 26740
+    },
+    {
+      "epoch": 0.2684772572162672,
+      "learning_rate": 0.00014630655586334257,
+      "loss": 0.7571,
+      "mean_token_accuracy": 0.7706903934478759,
+      "num_tokens": 91018052.0,
+      "step": 26750
+    },
+    {
+      "epoch": 0.2685776225460677,
+      "learning_rate": 0.00014628648279738248,
+      "loss": 0.7584,
+      "mean_token_accuracy": 0.7746901094913483,
+      "num_tokens": 91051633.0,
+      "step": 26760
+    },
+    {
+      "epoch": 0.26867798787586816,
+      "learning_rate": 0.0001462664097314224,
+      "loss": 0.7367,
+      "mean_token_accuracy": 0.7790070295333862,
+      "num_tokens": 91085155.0,
+      "step": 26770
+    },
+    {
+      "epoch": 0.2687783532056686,
+      "learning_rate": 0.00014624633666546227,
+      "loss": 0.7598,
+      "mean_token_accuracy": 0.7718618929386138,
+      "num_tokens": 91119620.0,
+      "step": 26780
+    },
+    {
+      "epoch": 0.26887871853546913,
+      "learning_rate": 0.0001462262635995022,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7654790699481964,
+      "num_tokens": 91153999.0,
+      "step": 26790
+    },
+    {
+      "epoch": 0.2689790838652696,
+      "learning_rate": 0.00014620619053354208,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7685680568218232,
+      "num_tokens": 91188823.0,
+      "step": 26800
+    },
+    {
+      "epoch": 0.26907944919507004,
+      "learning_rate": 0.000146186117467582,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7814339876174927,
+      "num_tokens": 91222517.0,
+      "step": 26810
+    },
+    {
+      "epoch": 0.26917981452487055,
+      "learning_rate": 0.00014616604440162193,
+      "loss": 0.7612,
+      "mean_token_accuracy": 0.7738438963890075,
+      "num_tokens": 91255788.0,
+      "step": 26820
+    },
+    {
+      "epoch": 0.269280179854671,
+      "learning_rate": 0.0001461459713356618,
+      "loss": 0.7524,
+      "mean_token_accuracy": 0.7798373401165009,
+      "num_tokens": 91290426.0,
+      "step": 26830
+    },
+    {
+      "epoch": 0.26938054518447146,
+      "learning_rate": 0.00014612589826970172,
+      "loss": 0.7551,
+      "mean_token_accuracy": 0.7722236275672912,
+      "num_tokens": 91324006.0,
+      "step": 26840
+    },
+    {
+      "epoch": 0.269480910514272,
+      "learning_rate": 0.00014610582520374163,
+      "loss": 0.7644,
+      "mean_token_accuracy": 0.7733603954315186,
+      "num_tokens": 91358023.0,
+      "step": 26850
+    },
+    {
+      "epoch": 0.26958127584407243,
+      "learning_rate": 0.00014608575213778154,
+      "loss": 0.7467,
+      "mean_token_accuracy": 0.7735972940921784,
+      "num_tokens": 91391146.0,
+      "step": 26860
+    },
+    {
+      "epoch": 0.2696816411738729,
+      "learning_rate": 0.00014606567907182142,
+      "loss": 0.7633,
+      "mean_token_accuracy": 0.7709302246570587,
+      "num_tokens": 91424533.0,
+      "step": 26870
+    },
+    {
+      "epoch": 0.26978200650367334,
+      "learning_rate": 0.00014604560600586135,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7727507174015045,
+      "num_tokens": 91459307.0,
+      "step": 26880
+    },
+    {
+      "epoch": 0.26988237183347386,
+      "learning_rate": 0.00014602553293990123,
+      "loss": 0.791,
+      "mean_token_accuracy": 0.7658800482749939,
+      "num_tokens": 91492206.0,
+      "step": 26890
+    },
+    {
+      "epoch": 0.2699827371632743,
+      "learning_rate": 0.00014600545987394114,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7779639959335327,
+      "num_tokens": 91526738.0,
+      "step": 26900
+    },
+    {
+      "epoch": 0.27008310249307477,
+      "learning_rate": 0.00014598538680798105,
+      "loss": 0.8066,
+      "mean_token_accuracy": 0.7683506786823273,
+      "num_tokens": 91561481.0,
+      "step": 26910
+    },
+    {
+      "epoch": 0.2701834678228753,
+      "learning_rate": 0.00014596531374202096,
+      "loss": 0.7674,
+      "mean_token_accuracy": 0.7711017251014709,
+      "num_tokens": 91594524.0,
+      "step": 26920
+    },
+    {
+      "epoch": 0.27028383315267573,
+      "learning_rate": 0.00014594524067606087,
+      "loss": 0.7502,
+      "mean_token_accuracy": 0.7745255947113037,
+      "num_tokens": 91628790.0,
+      "step": 26930
+    },
+    {
+      "epoch": 0.2703841984824762,
+      "learning_rate": 0.00014592516761010078,
+      "loss": 0.7484,
+      "mean_token_accuracy": 0.7774655818939209,
+      "num_tokens": 91662405.0,
+      "step": 26940
+    },
+    {
+      "epoch": 0.2704845638122767,
+      "learning_rate": 0.00014590509454414068,
+      "loss": 0.7758,
+      "mean_token_accuracy": 0.7700693845748902,
+      "num_tokens": 91695910.0,
+      "step": 26950
+    },
+    {
+      "epoch": 0.27058492914207716,
+      "learning_rate": 0.0001458850214781806,
+      "loss": 0.7373,
+      "mean_token_accuracy": 0.7784343719482422,
+      "num_tokens": 91730220.0,
+      "step": 26960
+    },
+    {
+      "epoch": 0.2706852944718776,
+      "learning_rate": 0.0001458649484122205,
+      "loss": 0.77,
+      "mean_token_accuracy": 0.7710197031497955,
+      "num_tokens": 91764271.0,
+      "step": 26970
+    },
+    {
+      "epoch": 0.2707856598016781,
+      "learning_rate": 0.00014584487534626038,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7732560396194458,
+      "num_tokens": 91798289.0,
+      "step": 26980
+    },
+    {
+      "epoch": 0.2708860251314786,
+      "learning_rate": 0.00014582480228030032,
+      "loss": 0.804,
+      "mean_token_accuracy": 0.7688061237335205,
+      "num_tokens": 91833897.0,
+      "step": 26990
+    },
+    {
+      "epoch": 0.27098639046127904,
+      "learning_rate": 0.0001458047292143402,
+      "loss": 0.7586,
+      "mean_token_accuracy": 0.7720942556858063,
+      "num_tokens": 91866369.0,
+      "step": 27000
+    },
+    {
+      "epoch": 0.27108675579107955,
+      "learning_rate": 0.0001457846561483801,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.7738721609115601,
+      "num_tokens": 91900038.0,
+      "step": 27010
+    },
+    {
+      "epoch": 0.27118712112088,
+      "learning_rate": 0.00014576458308242002,
+      "loss": 0.7948,
+      "mean_token_accuracy": 0.766669362783432,
+      "num_tokens": 91933852.0,
+      "step": 27020
+    },
+    {
+      "epoch": 0.27128748645068046,
+      "learning_rate": 0.00014574451001645992,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7770777583122254,
+      "num_tokens": 91968304.0,
+      "step": 27030
+    },
+    {
+      "epoch": 0.27138785178048097,
+      "learning_rate": 0.00014572443695049983,
+      "loss": 0.7481,
+      "mean_token_accuracy": 0.7717696845531463,
+      "num_tokens": 92001663.0,
+      "step": 27040
+    },
+    {
+      "epoch": 0.2714882171102814,
+      "learning_rate": 0.00014570436388453974,
+      "loss": 0.7413,
+      "mean_token_accuracy": 0.776682448387146,
+      "num_tokens": 92035535.0,
+      "step": 27050
+    },
+    {
+      "epoch": 0.2715885824400819,
+      "learning_rate": 0.00014568429081857965,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.7750038385391236,
+      "num_tokens": 92069123.0,
+      "step": 27060
+    },
+    {
+      "epoch": 0.2716889477698824,
+      "learning_rate": 0.00014566421775261953,
+      "loss": 0.7796,
+      "mean_token_accuracy": 0.769258689880371,
+      "num_tokens": 92103429.0,
+      "step": 27070
+    },
+    {
+      "epoch": 0.27178931309968285,
+      "learning_rate": 0.00014564414468665947,
+      "loss": 0.7679,
+      "mean_token_accuracy": 0.7737447798252106,
+      "num_tokens": 92137066.0,
+      "step": 27080
+    },
+    {
+      "epoch": 0.2718896784294833,
+      "learning_rate": 0.00014562407162069935,
+      "loss": 0.7683,
+      "mean_token_accuracy": 0.7694413065910339,
+      "num_tokens": 92170835.0,
+      "step": 27090
+    },
+    {
+      "epoch": 0.2719900437592838,
+      "learning_rate": 0.00014560399855473926,
+      "loss": 0.7937,
+      "mean_token_accuracy": 0.7688502252101899,
+      "num_tokens": 92204605.0,
+      "step": 27100
+    },
+    {
+      "epoch": 0.2720904090890843,
+      "learning_rate": 0.00014558392548877917,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7689731538295745,
+      "num_tokens": 92238421.0,
+      "step": 27110
+    },
+    {
+      "epoch": 0.27219077441888473,
+      "learning_rate": 0.00014556385242281907,
+      "loss": 0.7851,
+      "mean_token_accuracy": 0.7678522109985352,
+      "num_tokens": 92271632.0,
+      "step": 27120
+    },
+    {
+      "epoch": 0.27229113974868524,
+      "learning_rate": 0.00014554377935685895,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7715808153152466,
+      "num_tokens": 92306359.0,
+      "step": 27130
+    },
+    {
+      "epoch": 0.2723915050784857,
+      "learning_rate": 0.0001455237062908989,
+      "loss": 0.749,
+      "mean_token_accuracy": 0.7793966174125672,
+      "num_tokens": 92340190.0,
+      "step": 27140
+    },
+    {
+      "epoch": 0.27249187040828615,
+      "learning_rate": 0.00014550363322493877,
+      "loss": 0.7427,
+      "mean_token_accuracy": 0.7762248814105988,
+      "num_tokens": 92374829.0,
+      "step": 27150
+    },
+    {
+      "epoch": 0.2725922357380866,
+      "learning_rate": 0.00014548356015897868,
+      "loss": 0.7611,
+      "mean_token_accuracy": 0.774640703201294,
+      "num_tokens": 92408164.0,
+      "step": 27160
+    },
+    {
+      "epoch": 0.2726926010678871,
+      "learning_rate": 0.00014546348709301862,
+      "loss": 0.7566,
+      "mean_token_accuracy": 0.769939798116684,
+      "num_tokens": 92441790.0,
+      "step": 27170
+    },
+    {
+      "epoch": 0.2727929663976876,
+      "learning_rate": 0.0001454434140270585,
+      "loss": 0.7516,
+      "mean_token_accuracy": 0.7719286739826202,
+      "num_tokens": 92475623.0,
+      "step": 27180
+    },
+    {
+      "epoch": 0.27289333172748803,
+      "learning_rate": 0.0001454233409610984,
+      "loss": 0.759,
+      "mean_token_accuracy": 0.7721605718135833,
+      "num_tokens": 92509903.0,
+      "step": 27190
+    },
+    {
+      "epoch": 0.27299369705728854,
+      "learning_rate": 0.00014540326789513831,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.769053041934967,
+      "num_tokens": 92543066.0,
+      "step": 27200
+    },
+    {
+      "epoch": 0.273094062387089,
+      "learning_rate": 0.00014538319482917822,
+      "loss": 0.7449,
+      "mean_token_accuracy": 0.7751676380634308,
+      "num_tokens": 92577254.0,
+      "step": 27210
+    },
+    {
+      "epoch": 0.27319442771688945,
+      "learning_rate": 0.0001453631217632181,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7739300549030304,
+      "num_tokens": 92611429.0,
+      "step": 27220
+    },
+    {
+      "epoch": 0.27329479304668997,
+      "learning_rate": 0.00014534304869725804,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7717872381210327,
+      "num_tokens": 92645621.0,
+      "step": 27230
+    },
+    {
+      "epoch": 0.2733951583764904,
+      "learning_rate": 0.00014532297563129792,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7736692726612091,
+      "num_tokens": 92679943.0,
+      "step": 27240
+    },
+    {
+      "epoch": 0.2734955237062909,
+      "learning_rate": 0.00014530290256533783,
+      "loss": 0.7526,
+      "mean_token_accuracy": 0.7769086182117462,
+      "num_tokens": 92714044.0,
+      "step": 27250
+    },
+    {
+      "epoch": 0.2735958890360914,
+      "learning_rate": 0.00014528282949937774,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.7706098198890686,
+      "num_tokens": 92747702.0,
+      "step": 27260
+    },
+    {
+      "epoch": 0.27369625436589184,
+      "learning_rate": 0.00014526275643341765,
+      "loss": 0.758,
+      "mean_token_accuracy": 0.7733682990074158,
+      "num_tokens": 92781472.0,
+      "step": 27270
+    },
+    {
+      "epoch": 0.2737966196956923,
+      "learning_rate": 0.00014524268336745755,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7743779301643372,
+      "num_tokens": 92815143.0,
+      "step": 27280
+    },
+    {
+      "epoch": 0.2738969850254928,
+      "learning_rate": 0.00014522261030149746,
+      "loss": 0.772,
+      "mean_token_accuracy": 0.7625142872333527,
+      "num_tokens": 92849414.0,
+      "step": 27290
+    },
+    {
+      "epoch": 0.27399735035529327,
+      "learning_rate": 0.00014520253723553737,
+      "loss": 0.7282,
+      "mean_token_accuracy": 0.7782218098640442,
+      "num_tokens": 92883309.0,
+      "step": 27300
+    },
+    {
+      "epoch": 0.2740977156850937,
+      "learning_rate": 0.00014518246416957725,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.7723683476448059,
+      "num_tokens": 92917119.0,
+      "step": 27310
+    },
+    {
+      "epoch": 0.27419808101489423,
+      "learning_rate": 0.0001451623911036172,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.7603408515453338,
+      "num_tokens": 92951054.0,
+      "step": 27320
+    },
+    {
+      "epoch": 0.2742984463446947,
+      "learning_rate": 0.00014514231803765707,
+      "loss": 0.7496,
+      "mean_token_accuracy": 0.7706218063831329,
+      "num_tokens": 92985246.0,
+      "step": 27330
+    },
+    {
+      "epoch": 0.27439881167449515,
+      "learning_rate": 0.00014512224497169698,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7647865056991577,
+      "num_tokens": 93019185.0,
+      "step": 27340
+    },
+    {
+      "epoch": 0.27449917700429566,
+      "learning_rate": 0.00014510217190573689,
+      "loss": 0.7869,
+      "mean_token_accuracy": 0.7622201561927795,
+      "num_tokens": 93052004.0,
+      "step": 27350
+    },
+    {
+      "epoch": 0.2745995423340961,
+      "learning_rate": 0.0001450820988397768,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7691390633583068,
+      "num_tokens": 93085775.0,
+      "step": 27360
+    },
+    {
+      "epoch": 0.27469990766389657,
+      "learning_rate": 0.0001450620257738167,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.7697662651538849,
+      "num_tokens": 93119339.0,
+      "step": 27370
+    },
+    {
+      "epoch": 0.2748002729936971,
+      "learning_rate": 0.0001450419527078566,
+      "loss": 0.7812,
+      "mean_token_accuracy": 0.7708326518535614,
+      "num_tokens": 93152624.0,
+      "step": 27380
+    },
+    {
+      "epoch": 0.27490063832349754,
+      "learning_rate": 0.00014502187964189652,
+      "loss": 0.7749,
+      "mean_token_accuracy": 0.773492568731308,
+      "num_tokens": 93187350.0,
+      "step": 27390
+    },
+    {
+      "epoch": 0.275001003653298,
+      "learning_rate": 0.0001450018065759364,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.7714549422264099,
+      "num_tokens": 93221486.0,
+      "step": 27400
+    },
+    {
+      "epoch": 0.2751013689830985,
+      "learning_rate": 0.00014498173350997634,
+      "loss": 0.7441,
+      "mean_token_accuracy": 0.7719537556171417,
+      "num_tokens": 93255173.0,
+      "step": 27410
+    },
+    {
+      "epoch": 0.27520173431289896,
+      "learning_rate": 0.00014496166044401622,
+      "loss": 0.7648,
+      "mean_token_accuracy": 0.7742453396320343,
+      "num_tokens": 93287585.0,
+      "step": 27420
+    },
+    {
+      "epoch": 0.2753020996426994,
+      "learning_rate": 0.00014494158737805613,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7616688489913941,
+      "num_tokens": 93321191.0,
+      "step": 27430
+    },
+    {
+      "epoch": 0.27540246497249987,
+      "learning_rate": 0.00014492151431209603,
+      "loss": 0.7625,
+      "mean_token_accuracy": 0.7683378040790558,
+      "num_tokens": 93354782.0,
+      "step": 27440
+    },
+    {
+      "epoch": 0.2755028303023004,
+      "learning_rate": 0.00014490144124613594,
+      "loss": 0.7866,
+      "mean_token_accuracy": 0.7675630867481231,
+      "num_tokens": 93388494.0,
+      "step": 27450
+    },
+    {
+      "epoch": 0.27560319563210084,
+      "learning_rate": 0.00014488136818017582,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7782629013061524,
+      "num_tokens": 93422181.0,
+      "step": 27460
+    },
+    {
+      "epoch": 0.2757035609619013,
+      "learning_rate": 0.00014486129511421576,
+      "loss": 0.7416,
+      "mean_token_accuracy": 0.7764272034168244,
+      "num_tokens": 93456328.0,
+      "step": 27470
+    },
+    {
+      "epoch": 0.2758039262917018,
+      "learning_rate": 0.00014484122204825564,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7699039518833161,
+      "num_tokens": 93490705.0,
+      "step": 27480
+    },
+    {
+      "epoch": 0.27590429162150226,
+      "learning_rate": 0.00014482114898229558,
+      "loss": 0.7709,
+      "mean_token_accuracy": 0.773164016008377,
+      "num_tokens": 93524210.0,
+      "step": 27490
+    },
+    {
+      "epoch": 0.2760046569513027,
+      "learning_rate": 0.00014480107591633546,
+      "loss": 0.7364,
+      "mean_token_accuracy": 0.7783370196819306,
+      "num_tokens": 93558061.0,
+      "step": 27500
+    },
+    {
+      "epoch": 0.27610502228110323,
+      "learning_rate": 0.00014478100285037537,
+      "loss": 0.7507,
+      "mean_token_accuracy": 0.7777692973613739,
+      "num_tokens": 93593033.0,
+      "step": 27510
+    },
+    {
+      "epoch": 0.2762053876109037,
+      "learning_rate": 0.0001447609297844153,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7742154121398925,
+      "num_tokens": 93626303.0,
+      "step": 27520
+    },
+    {
+      "epoch": 0.27630575294070414,
+      "learning_rate": 0.00014474085671845518,
+      "loss": 0.7917,
+      "mean_token_accuracy": 0.76452516913414,
+      "num_tokens": 93661039.0,
+      "step": 27530
+    },
+    {
+      "epoch": 0.27640611827050465,
+      "learning_rate": 0.0001447207836524951,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7712850332260132,
+      "num_tokens": 93694127.0,
+      "step": 27540
+    },
+    {
+      "epoch": 0.2765064836003051,
+      "learning_rate": 0.000144700710586535,
+      "loss": 0.7515,
+      "mean_token_accuracy": 0.7779252588748932,
+      "num_tokens": 93727570.0,
+      "step": 27550
+    },
+    {
+      "epoch": 0.27660684893010556,
+      "learning_rate": 0.0001446806375205749,
+      "loss": 0.7731,
+      "mean_token_accuracy": 0.7694986999034882,
+      "num_tokens": 93761211.0,
+      "step": 27560
+    },
+    {
+      "epoch": 0.2767072142599061,
+      "learning_rate": 0.0001446605644546148,
+      "loss": 0.7703,
+      "mean_token_accuracy": 0.7661393880844116,
+      "num_tokens": 93794623.0,
+      "step": 27570
+    },
+    {
+      "epoch": 0.27680757958970653,
+      "learning_rate": 0.00014464049138865473,
+      "loss": 0.7699,
+      "mean_token_accuracy": 0.7691900432109833,
+      "num_tokens": 93828423.0,
+      "step": 27580
+    },
+    {
+      "epoch": 0.276907944919507,
+      "learning_rate": 0.0001446204183226946,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7683387637138367,
+      "num_tokens": 93861875.0,
+      "step": 27590
+    },
+    {
+      "epoch": 0.2770083102493075,
+      "learning_rate": 0.00014460034525673452,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7742133915424347,
+      "num_tokens": 93895356.0,
+      "step": 27600
+    },
+    {
+      "epoch": 0.27710867557910795,
+      "learning_rate": 0.00014458027219077442,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.7763895392417908,
+      "num_tokens": 93929096.0,
+      "step": 27610
+    },
+    {
+      "epoch": 0.2772090409089084,
+      "learning_rate": 0.00014456019912481433,
+      "loss": 0.7544,
+      "mean_token_accuracy": 0.7730475723743438,
+      "num_tokens": 93962868.0,
+      "step": 27620
+    },
+    {
+      "epoch": 0.2773094062387089,
+      "learning_rate": 0.00014454012605885424,
+      "loss": 0.7662,
+      "mean_token_accuracy": 0.7731011271476745,
+      "num_tokens": 93997156.0,
+      "step": 27630
+    },
+    {
+      "epoch": 0.2774097715685094,
+      "learning_rate": 0.00014452005299289415,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7725311994552613,
+      "num_tokens": 94031395.0,
+      "step": 27640
+    },
+    {
+      "epoch": 0.27751013689830983,
+      "learning_rate": 0.00014449997992693406,
+      "loss": 0.7388,
+      "mean_token_accuracy": 0.7824570715427399,
+      "num_tokens": 94065614.0,
+      "step": 27650
+    },
+    {
+      "epoch": 0.27761050222811035,
+      "learning_rate": 0.00014447990686097394,
+      "loss": 0.735,
+      "mean_token_accuracy": 0.7755841255187989,
+      "num_tokens": 94099789.0,
+      "step": 27660
+    },
+    {
+      "epoch": 0.2777108675579108,
+      "learning_rate": 0.00014445983379501387,
+      "loss": 0.7411,
+      "mean_token_accuracy": 0.7816408157348633,
+      "num_tokens": 94133642.0,
+      "step": 27670
+    },
+    {
+      "epoch": 0.27781123288771126,
+      "learning_rate": 0.00014443976072905376,
+      "loss": 0.7383,
+      "mean_token_accuracy": 0.7797886252403259,
+      "num_tokens": 94167582.0,
+      "step": 27680
+    },
+    {
+      "epoch": 0.27791159821751177,
+      "learning_rate": 0.00014441968766309366,
+      "loss": 0.7522,
+      "mean_token_accuracy": 0.7724378883838654,
+      "num_tokens": 94202495.0,
+      "step": 27690
+    },
+    {
+      "epoch": 0.2780119635473122,
+      "learning_rate": 0.00014439961459713357,
+      "loss": 0.7833,
+      "mean_token_accuracy": 0.7701262414455414,
+      "num_tokens": 94236805.0,
+      "step": 27700
+    },
+    {
+      "epoch": 0.2781123288771127,
+      "learning_rate": 0.00014437954153117348,
+      "loss": 0.802,
+      "mean_token_accuracy": 0.7600609242916108,
+      "num_tokens": 94270099.0,
+      "step": 27710
+    },
+    {
+      "epoch": 0.2782126942069132,
+      "learning_rate": 0.0001443594684652134,
+      "loss": 0.7378,
+      "mean_token_accuracy": 0.7771346986293792,
+      "num_tokens": 94304587.0,
+      "step": 27720
+    },
+    {
+      "epoch": 0.27831305953671365,
+      "learning_rate": 0.0001443393953992533,
+      "loss": 0.7512,
+      "mean_token_accuracy": 0.7794571101665497,
+      "num_tokens": 94338925.0,
+      "step": 27730
+    },
+    {
+      "epoch": 0.2784134248665141,
+      "learning_rate": 0.0001443193223332932,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.7762793719768524,
+      "num_tokens": 94373156.0,
+      "step": 27740
+    },
+    {
+      "epoch": 0.27851379019631456,
+      "learning_rate": 0.0001442992492673331,
+      "loss": 0.7466,
+      "mean_token_accuracy": 0.7774362564086914,
+      "num_tokens": 94407745.0,
+      "step": 27750
+    },
+    {
+      "epoch": 0.27861415552611507,
+      "learning_rate": 0.00014427917620137302,
+      "loss": 0.7695,
+      "mean_token_accuracy": 0.7731296360492707,
+      "num_tokens": 94441357.0,
+      "step": 27760
+    },
+    {
+      "epoch": 0.2787145208559155,
+      "learning_rate": 0.0001442591031354129,
+      "loss": 0.7778,
+      "mean_token_accuracy": 0.772819846868515,
+      "num_tokens": 94474707.0,
+      "step": 27770
+    },
+    {
+      "epoch": 0.278814886185716,
+      "learning_rate": 0.0001442390300694528,
+      "loss": 0.7193,
+      "mean_token_accuracy": 0.7839652538299561,
+      "num_tokens": 94509157.0,
+      "step": 27780
+    },
+    {
+      "epoch": 0.2789152515155165,
+      "learning_rate": 0.00014421895700349272,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7695025742053986,
+      "num_tokens": 94542957.0,
+      "step": 27790
+    },
+    {
+      "epoch": 0.27901561684531695,
+      "learning_rate": 0.00014419888393753263,
+      "loss": 0.772,
+      "mean_token_accuracy": 0.7738290429115295,
+      "num_tokens": 94577156.0,
+      "step": 27800
+    },
+    {
+      "epoch": 0.2791159821751174,
+      "learning_rate": 0.0001441788108715725,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.779708981513977,
+      "num_tokens": 94610412.0,
+      "step": 27810
+    },
+    {
+      "epoch": 0.2792163475049179,
+      "learning_rate": 0.00014415873780561245,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7689424514770508,
+      "num_tokens": 94644339.0,
+      "step": 27820
+    },
+    {
+      "epoch": 0.2793167128347184,
+      "learning_rate": 0.00014413866473965233,
+      "loss": 0.7807,
+      "mean_token_accuracy": 0.7719169199466706,
+      "num_tokens": 94678212.0,
+      "step": 27830
+    },
+    {
+      "epoch": 0.27941707816451883,
+      "learning_rate": 0.00014411859167369224,
+      "loss": 0.8012,
+      "mean_token_accuracy": 0.7627018094062805,
+      "num_tokens": 94711572.0,
+      "step": 27840
+    },
+    {
+      "epoch": 0.27951744349431934,
+      "learning_rate": 0.00014409851860773215,
+      "loss": 0.7563,
+      "mean_token_accuracy": 0.7732181370258331,
+      "num_tokens": 94746210.0,
+      "step": 27850
+    },
+    {
+      "epoch": 0.2796178088241198,
+      "learning_rate": 0.00014407844554177205,
+      "loss": 0.7649,
+      "mean_token_accuracy": 0.77310009598732,
+      "num_tokens": 94780684.0,
+      "step": 27860
+    },
+    {
+      "epoch": 0.27971817415392025,
+      "learning_rate": 0.00014405837247581196,
+      "loss": 0.7802,
+      "mean_token_accuracy": 0.7678790330886841,
+      "num_tokens": 94814629.0,
+      "step": 27870
+    },
+    {
+      "epoch": 0.27981853948372076,
+      "learning_rate": 0.00014403829940985187,
+      "loss": 0.7699,
+      "mean_token_accuracy": 0.7725537121295929,
+      "num_tokens": 94848391.0,
+      "step": 27880
+    },
+    {
+      "epoch": 0.2799189048135212,
+      "learning_rate": 0.00014401822634389178,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7687770128250122,
+      "num_tokens": 94882450.0,
+      "step": 27890
+    },
+    {
+      "epoch": 0.2800192701433217,
+      "learning_rate": 0.00014399815327793166,
+      "loss": 0.7744,
+      "mean_token_accuracy": 0.7670266807079316,
+      "num_tokens": 94916523.0,
+      "step": 27900
+    },
+    {
+      "epoch": 0.2801196354731222,
+      "learning_rate": 0.0001439780802119716,
+      "loss": 0.7619,
+      "mean_token_accuracy": 0.7759920358657837,
+      "num_tokens": 94950511.0,
+      "step": 27910
+    },
+    {
+      "epoch": 0.28022000080292264,
+      "learning_rate": 0.00014395800714601148,
+      "loss": 0.7888,
+      "mean_token_accuracy": 0.770870441198349,
+      "num_tokens": 94984809.0,
+      "step": 27920
+    },
+    {
+      "epoch": 0.2803203661327231,
+      "learning_rate": 0.00014393793408005139,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.7784211814403534,
+      "num_tokens": 95018325.0,
+      "step": 27930
+    },
+    {
+      "epoch": 0.2804207314625236,
+      "learning_rate": 0.0001439178610140913,
+      "loss": 0.7914,
+      "mean_token_accuracy": 0.7643942356109619,
+      "num_tokens": 95052556.0,
+      "step": 27940
+    },
+    {
+      "epoch": 0.28052109679232407,
+      "learning_rate": 0.0001438977879481312,
+      "loss": 0.779,
+      "mean_token_accuracy": 0.7691850602626801,
+      "num_tokens": 95086818.0,
+      "step": 27950
+    },
+    {
+      "epoch": 0.2806214621221245,
+      "learning_rate": 0.0001438777148821711,
+      "loss": 0.78,
+      "mean_token_accuracy": 0.7732650637626648,
+      "num_tokens": 95121026.0,
+      "step": 27960
+    },
+    {
+      "epoch": 0.28072182745192503,
+      "learning_rate": 0.00014385764181621102,
+      "loss": 0.7607,
+      "mean_token_accuracy": 0.7753560781478882,
+      "num_tokens": 95154968.0,
+      "step": 27970
+    },
+    {
+      "epoch": 0.2808221927817255,
+      "learning_rate": 0.00014383756875025093,
+      "loss": 0.7465,
+      "mean_token_accuracy": 0.7759646117687226,
+      "num_tokens": 95187520.0,
+      "step": 27980
+    },
+    {
+      "epoch": 0.28092255811152594,
+      "learning_rate": 0.0001438174956842908,
+      "loss": 0.7325,
+      "mean_token_accuracy": 0.7718597769737243,
+      "num_tokens": 95221151.0,
+      "step": 27990
+    },
+    {
+      "epoch": 0.28102292344132646,
+      "learning_rate": 0.00014379742261833074,
+      "loss": 0.738,
+      "mean_token_accuracy": 0.7761099696159363,
+      "num_tokens": 95254573.0,
+      "step": 28000
+    },
+    {
+      "epoch": 0.2811232887711269,
+      "learning_rate": 0.00014377734955237063,
+      "loss": 0.7829,
+      "mean_token_accuracy": 0.7705293476581574,
+      "num_tokens": 95289084.0,
+      "step": 28010
+    },
+    {
+      "epoch": 0.28122365410092737,
+      "learning_rate": 0.00014375727648641053,
+      "loss": 0.7435,
+      "mean_token_accuracy": 0.775824499130249,
+      "num_tokens": 95323588.0,
+      "step": 28020
+    },
+    {
+      "epoch": 0.2813240194307278,
+      "learning_rate": 0.00014373720342045044,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7712218165397644,
+      "num_tokens": 95357591.0,
+      "step": 28030
+    },
+    {
+      "epoch": 0.28142438476052833,
+      "learning_rate": 0.00014371713035449035,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7757143259048462,
+      "num_tokens": 95391528.0,
+      "step": 28040
+    },
+    {
+      "epoch": 0.2815247500903288,
+      "learning_rate": 0.00014369705728853026,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7702911853790283,
+      "num_tokens": 95426526.0,
+      "step": 28050
+    },
+    {
+      "epoch": 0.28162511542012925,
+      "learning_rate": 0.00014367698422257017,
+      "loss": 0.7534,
+      "mean_token_accuracy": 0.7716615200042725,
+      "num_tokens": 95460919.0,
+      "step": 28060
+    },
+    {
+      "epoch": 0.28172548074992976,
+      "learning_rate": 0.00014365691115661008,
+      "loss": 0.7908,
+      "mean_token_accuracy": 0.7679134845733643,
+      "num_tokens": 95495023.0,
+      "step": 28070
+    },
+    {
+      "epoch": 0.2818258460797302,
+      "learning_rate": 0.00014363683809064999,
+      "loss": 0.7572,
+      "mean_token_accuracy": 0.7714826464653015,
+      "num_tokens": 95528809.0,
+      "step": 28080
+    },
+    {
+      "epoch": 0.28192621140953067,
+      "learning_rate": 0.0001436167650246899,
+      "loss": 0.7317,
+      "mean_token_accuracy": 0.7837158501148224,
+      "num_tokens": 95561932.0,
+      "step": 28090
+    },
+    {
+      "epoch": 0.2820265767393312,
+      "learning_rate": 0.00014359669195872977,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.7799531280994415,
+      "num_tokens": 95595828.0,
+      "step": 28100
+    },
+    {
+      "epoch": 0.28212694206913164,
+      "learning_rate": 0.0001435766188927697,
+      "loss": 0.7409,
+      "mean_token_accuracy": 0.7749620318412781,
+      "num_tokens": 95629727.0,
+      "step": 28110
+    },
+    {
+      "epoch": 0.2822273073989321,
+      "learning_rate": 0.0001435565458268096,
+      "loss": 0.7538,
+      "mean_token_accuracy": 0.7765342891216278,
+      "num_tokens": 95663466.0,
+      "step": 28120
+    },
+    {
+      "epoch": 0.2823276727287326,
+      "learning_rate": 0.0001435364727608495,
+      "loss": 0.7656,
+      "mean_token_accuracy": 0.7728769242763519,
+      "num_tokens": 95697934.0,
+      "step": 28130
+    },
+    {
+      "epoch": 0.28242803805853306,
+      "learning_rate": 0.0001435163996948894,
+      "loss": 0.774,
+      "mean_token_accuracy": 0.7711954891681672,
+      "num_tokens": 95731856.0,
+      "step": 28140
+    },
+    {
+      "epoch": 0.2825284033883335,
+      "learning_rate": 0.00014349632662892932,
+      "loss": 0.7526,
+      "mean_token_accuracy": 0.7742810428142548,
+      "num_tokens": 95766197.0,
+      "step": 28150
+    },
+    {
+      "epoch": 0.282628768718134,
+      "learning_rate": 0.0001434762535629692,
+      "loss": 0.7757,
+      "mean_token_accuracy": 0.7725955188274384,
+      "num_tokens": 95800956.0,
+      "step": 28160
+    },
+    {
+      "epoch": 0.2827291340479345,
+      "learning_rate": 0.00014345618049700913,
+      "loss": 0.7657,
+      "mean_token_accuracy": 0.7697532892227172,
+      "num_tokens": 95834567.0,
+      "step": 28170
+    },
+    {
+      "epoch": 0.28282949937773494,
+      "learning_rate": 0.00014343610743104902,
+      "loss": 0.7459,
+      "mean_token_accuracy": 0.7763911068439484,
+      "num_tokens": 95869011.0,
+      "step": 28180
+    },
+    {
+      "epoch": 0.28292986470753545,
+      "learning_rate": 0.00014341603436508892,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7635152816772461,
+      "num_tokens": 95903302.0,
+      "step": 28190
+    },
+    {
+      "epoch": 0.2830302300373359,
+      "learning_rate": 0.00014339596129912883,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7720817685127258,
+      "num_tokens": 95936762.0,
+      "step": 28200
+    },
+    {
+      "epoch": 0.28313059536713636,
+      "learning_rate": 0.00014337588823316874,
+      "loss": 0.7494,
+      "mean_token_accuracy": 0.7792890071868896,
+      "num_tokens": 95971551.0,
+      "step": 28210
+    },
+    {
+      "epoch": 0.2832309606969369,
+      "learning_rate": 0.00014335581516720865,
+      "loss": 0.7348,
+      "mean_token_accuracy": 0.7871302604675293,
+      "num_tokens": 96006501.0,
+      "step": 28220
+    },
+    {
+      "epoch": 0.28333132602673733,
+      "learning_rate": 0.00014333574210124856,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.7612696290016174,
+      "num_tokens": 96040332.0,
+      "step": 28230
+    },
+    {
+      "epoch": 0.2834316913565378,
+      "learning_rate": 0.00014331566903528847,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7700464427471161,
+      "num_tokens": 96075156.0,
+      "step": 28240
+    },
+    {
+      "epoch": 0.2835320566863383,
+      "learning_rate": 0.00014329559596932835,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7754449605941772,
+      "num_tokens": 96108056.0,
+      "step": 28250
+    },
+    {
+      "epoch": 0.28363242201613875,
+      "learning_rate": 0.00014327552290336828,
+      "loss": 0.8003,
+      "mean_token_accuracy": 0.7639413118362427,
+      "num_tokens": 96142646.0,
+      "step": 28260
+    },
+    {
+      "epoch": 0.2837327873459392,
+      "learning_rate": 0.00014325544983740816,
+      "loss": 0.7905,
+      "mean_token_accuracy": 0.7684699833393097,
+      "num_tokens": 96176728.0,
+      "step": 28270
+    },
+    {
+      "epoch": 0.2838331526757397,
+      "learning_rate": 0.00014323537677144807,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.7639523983001709,
+      "num_tokens": 96212312.0,
+      "step": 28280
+    },
+    {
+      "epoch": 0.2839335180055402,
+      "learning_rate": 0.00014321530370548798,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.7776244401931762,
+      "num_tokens": 96246642.0,
+      "step": 28290
+    },
+    {
+      "epoch": 0.28403388333534063,
+      "learning_rate": 0.0001431952306395279,
+      "loss": 0.7831,
+      "mean_token_accuracy": 0.7680910229682922,
+      "num_tokens": 96279819.0,
+      "step": 28300
+    },
+    {
+      "epoch": 0.2841342486651411,
+      "learning_rate": 0.0001431751575735678,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7736326217651367,
+      "num_tokens": 96313203.0,
+      "step": 28310
+    },
+    {
+      "epoch": 0.2842346139949416,
+      "learning_rate": 0.0001431550845076077,
+      "loss": 0.7262,
+      "mean_token_accuracy": 0.7782347440719605,
+      "num_tokens": 96347472.0,
+      "step": 28320
+    },
+    {
+      "epoch": 0.28433497932474205,
+      "learning_rate": 0.00014313501144164761,
+      "loss": 0.7408,
+      "mean_token_accuracy": 0.7803001344203949,
+      "num_tokens": 96381772.0,
+      "step": 28330
+    },
+    {
+      "epoch": 0.2844353446545425,
+      "learning_rate": 0.0001431149383756875,
+      "loss": 0.7964,
+      "mean_token_accuracy": 0.7637731015682221,
+      "num_tokens": 96415551.0,
+      "step": 28340
+    },
+    {
+      "epoch": 0.284535709984343,
+      "learning_rate": 0.00014309486530972743,
+      "loss": 0.786,
+      "mean_token_accuracy": 0.76785329580307,
+      "num_tokens": 96448445.0,
+      "step": 28350
+    },
+    {
+      "epoch": 0.2846360753141435,
+      "learning_rate": 0.0001430747922437673,
+      "loss": 0.7455,
+      "mean_token_accuracy": 0.7801205277442932,
+      "num_tokens": 96483007.0,
+      "step": 28360
+    },
+    {
+      "epoch": 0.28473644064394393,
+      "learning_rate": 0.00014305471917780722,
+      "loss": 0.7391,
+      "mean_token_accuracy": 0.7741820096969605,
+      "num_tokens": 96517006.0,
+      "step": 28370
+    },
+    {
+      "epoch": 0.28483680597374444,
+      "learning_rate": 0.00014303464611184713,
+      "loss": 0.7697,
+      "mean_token_accuracy": 0.7698454201221466,
+      "num_tokens": 96550991.0,
+      "step": 28380
+    },
+    {
+      "epoch": 0.2849371713035449,
+      "learning_rate": 0.00014301457304588704,
+      "loss": 0.7568,
+      "mean_token_accuracy": 0.7713454842567444,
+      "num_tokens": 96585415.0,
+      "step": 28390
+    },
+    {
+      "epoch": 0.28503753663334536,
+      "learning_rate": 0.00014299449997992692,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7706193685531616,
+      "num_tokens": 96620165.0,
+      "step": 28400
+    },
+    {
+      "epoch": 0.28513790196314587,
+      "learning_rate": 0.00014297442691396686,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7788212418556213,
+      "num_tokens": 96653482.0,
+      "step": 28410
+    },
+    {
+      "epoch": 0.2852382672929463,
+      "learning_rate": 0.00014295435384800676,
+      "loss": 0.7192,
+      "mean_token_accuracy": 0.7818759381771088,
+      "num_tokens": 96687630.0,
+      "step": 28420
+    },
+    {
+      "epoch": 0.2853386326227468,
+      "learning_rate": 0.00014293428078204664,
+      "loss": 0.7786,
+      "mean_token_accuracy": 0.7628415882587433,
+      "num_tokens": 96722395.0,
+      "step": 28430
+    },
+    {
+      "epoch": 0.2854389979525473,
+      "learning_rate": 0.00014291420771608658,
+      "loss": 0.7831,
+      "mean_token_accuracy": 0.7688349485397339,
+      "num_tokens": 96756837.0,
+      "step": 28440
+    },
+    {
+      "epoch": 0.28553936328234775,
+      "learning_rate": 0.00014289413465012646,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.7714909553527832,
+      "num_tokens": 96791139.0,
+      "step": 28450
+    },
+    {
+      "epoch": 0.2856397286121482,
+      "learning_rate": 0.00014287406158416637,
+      "loss": 0.7824,
+      "mean_token_accuracy": 0.7721018195152283,
+      "num_tokens": 96825362.0,
+      "step": 28460
+    },
+    {
+      "epoch": 0.2857400939419487,
+      "learning_rate": 0.00014285398851820628,
+      "loss": 0.7156,
+      "mean_token_accuracy": 0.7840331375598908,
+      "num_tokens": 96859324.0,
+      "step": 28470
+    },
+    {
+      "epoch": 0.28584045927174917,
+      "learning_rate": 0.0001428339154522462,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.773687344789505,
+      "num_tokens": 96892927.0,
+      "step": 28480
+    },
+    {
+      "epoch": 0.2859408246015496,
+      "learning_rate": 0.00014281384238628607,
+      "loss": 0.7696,
+      "mean_token_accuracy": 0.7728011250495911,
+      "num_tokens": 96926712.0,
+      "step": 28490
+    },
+    {
+      "epoch": 0.28604118993135014,
+      "learning_rate": 0.000142793769320326,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7809718906879425,
+      "num_tokens": 96960701.0,
+      "step": 28500
+    },
+    {
+      "epoch": 0.2861415552611506,
+      "learning_rate": 0.00014277369625436589,
+      "loss": 0.7744,
+      "mean_token_accuracy": 0.7736421167850495,
+      "num_tokens": 96995049.0,
+      "step": 28510
+    },
+    {
+      "epoch": 0.28624192059095105,
+      "learning_rate": 0.0001427536231884058,
+      "loss": 0.7453,
+      "mean_token_accuracy": 0.7792791187763214,
+      "num_tokens": 97028900.0,
+      "step": 28520
+    },
+    {
+      "epoch": 0.28634228592075156,
+      "learning_rate": 0.0001427335501224457,
+      "loss": 0.7736,
+      "mean_token_accuracy": 0.770075649023056,
+      "num_tokens": 97062735.0,
+      "step": 28530
+    },
+    {
+      "epoch": 0.286442651250552,
+      "learning_rate": 0.0001427134770564856,
+      "loss": 0.7898,
+      "mean_token_accuracy": 0.7700929760932922,
+      "num_tokens": 97097572.0,
+      "step": 28540
+    },
+    {
+      "epoch": 0.28654301658035247,
+      "learning_rate": 0.00014269340399052552,
+      "loss": 0.7687,
+      "mean_token_accuracy": 0.7669449508190155,
+      "num_tokens": 97131601.0,
+      "step": 28550
+    },
+    {
+      "epoch": 0.286643381910153,
+      "learning_rate": 0.00014267333092456543,
+      "loss": 0.7908,
+      "mean_token_accuracy": 0.7685792148113251,
+      "num_tokens": 97166226.0,
+      "step": 28560
+    },
+    {
+      "epoch": 0.28674374723995344,
+      "learning_rate": 0.00014265325785860534,
+      "loss": 0.7688,
+      "mean_token_accuracy": 0.7743170499801636,
+      "num_tokens": 97200126.0,
+      "step": 28570
+    },
+    {
+      "epoch": 0.2868441125697539,
+      "learning_rate": 0.00014263318479264522,
+      "loss": 0.7925,
+      "mean_token_accuracy": 0.7632998406887055,
+      "num_tokens": 97234143.0,
+      "step": 28580
+    },
+    {
+      "epoch": 0.28694447789955435,
+      "learning_rate": 0.00014261311172668515,
+      "loss": 0.811,
+      "mean_token_accuracy": 0.7661246061325073,
+      "num_tokens": 97268940.0,
+      "step": 28590
+    },
+    {
+      "epoch": 0.28704484322935486,
+      "learning_rate": 0.00014259303866072503,
+      "loss": 0.77,
+      "mean_token_accuracy": 0.7769502639770508,
+      "num_tokens": 97303733.0,
+      "step": 28600
+    },
+    {
+      "epoch": 0.2871452085591553,
+      "learning_rate": 0.00014257296559476497,
+      "loss": 0.7538,
+      "mean_token_accuracy": 0.7684979557991027,
+      "num_tokens": 97337759.0,
+      "step": 28610
+    },
+    {
+      "epoch": 0.2872455738889558,
+      "learning_rate": 0.00014255289252880485,
+      "loss": 0.7318,
+      "mean_token_accuracy": 0.7757490456104279,
+      "num_tokens": 97371306.0,
+      "step": 28620
+    },
+    {
+      "epoch": 0.2873459392187563,
+      "learning_rate": 0.00014253281946284476,
+      "loss": 0.7584,
+      "mean_token_accuracy": 0.7712860763072967,
+      "num_tokens": 97404592.0,
+      "step": 28630
+    },
+    {
+      "epoch": 0.28744630454855674,
+      "learning_rate": 0.00014251274639688467,
+      "loss": 0.7659,
+      "mean_token_accuracy": 0.7660297691822052,
+      "num_tokens": 97438537.0,
+      "step": 28640
+    },
+    {
+      "epoch": 0.2875466698783572,
+      "learning_rate": 0.00014249267333092458,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.7710791051387786,
+      "num_tokens": 97473717.0,
+      "step": 28650
+    },
+    {
+      "epoch": 0.2876470352081577,
+      "learning_rate": 0.00014247260026496448,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7716709733009338,
+      "num_tokens": 97508297.0,
+      "step": 28660
+    },
+    {
+      "epoch": 0.28774740053795816,
+      "learning_rate": 0.0001424525271990044,
+      "loss": 0.7825,
+      "mean_token_accuracy": 0.7744401037693024,
+      "num_tokens": 97541722.0,
+      "step": 28670
+    },
+    {
+      "epoch": 0.2878477658677586,
+      "learning_rate": 0.0001424324541330443,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7714649736881256,
+      "num_tokens": 97576189.0,
+      "step": 28680
+    },
+    {
+      "epoch": 0.28794813119755913,
+      "learning_rate": 0.00014241238106708418,
+      "loss": 0.7725,
+      "mean_token_accuracy": 0.7723775625228881,
+      "num_tokens": 97610364.0,
+      "step": 28690
+    },
+    {
+      "epoch": 0.2880484965273596,
+      "learning_rate": 0.00014239230800112412,
+      "loss": 0.7923,
+      "mean_token_accuracy": 0.7643298923969268,
+      "num_tokens": 97644164.0,
+      "step": 28700
+    },
+    {
+      "epoch": 0.28814886185716004,
+      "learning_rate": 0.000142372234935164,
+      "loss": 0.7296,
+      "mean_token_accuracy": 0.7828871130943298,
+      "num_tokens": 97677997.0,
+      "step": 28710
+    },
+    {
+      "epoch": 0.28824922718696055,
+      "learning_rate": 0.0001423521618692039,
+      "loss": 0.7556,
+      "mean_token_accuracy": 0.7748179018497467,
+      "num_tokens": 97712317.0,
+      "step": 28720
+    },
+    {
+      "epoch": 0.288349592516761,
+      "learning_rate": 0.00014233208880324382,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7652441620826721,
+      "num_tokens": 97746507.0,
+      "step": 28730
+    },
+    {
+      "epoch": 0.28844995784656147,
+      "learning_rate": 0.00014231201573728372,
+      "loss": 0.7736,
+      "mean_token_accuracy": 0.7693808436393738,
+      "num_tokens": 97780635.0,
+      "step": 28740
+    },
+    {
+      "epoch": 0.288550323176362,
+      "learning_rate": 0.0001422919426713236,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7710655629634857,
+      "num_tokens": 97813917.0,
+      "step": 28750
+    },
+    {
+      "epoch": 0.28865068850616243,
+      "learning_rate": 0.00014227186960536354,
+      "loss": 0.7773,
+      "mean_token_accuracy": 0.7683239877223969,
+      "num_tokens": 97847757.0,
+      "step": 28760
+    },
+    {
+      "epoch": 0.2887510538359629,
+      "learning_rate": 0.00014225179653940345,
+      "loss": 0.7899,
+      "mean_token_accuracy": 0.7675133168697357,
+      "num_tokens": 97881711.0,
+      "step": 28770
+    },
+    {
+      "epoch": 0.2888514191657634,
+      "learning_rate": 0.00014223172347344333,
+      "loss": 0.737,
+      "mean_token_accuracy": 0.7838814437389374,
+      "num_tokens": 97915644.0,
+      "step": 28780
+    },
+    {
+      "epoch": 0.28895178449556386,
+      "learning_rate": 0.00014221165040748327,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7744276881217956,
+      "num_tokens": 97950484.0,
+      "step": 28790
+    },
+    {
+      "epoch": 0.2890521498253643,
+      "learning_rate": 0.00014219157734152315,
+      "loss": 0.7635,
+      "mean_token_accuracy": 0.7697767615318298,
+      "num_tokens": 97985047.0,
+      "step": 28800
+    },
+    {
+      "epoch": 0.2891525151551648,
+      "learning_rate": 0.00014217150427556306,
+      "loss": 0.7477,
+      "mean_token_accuracy": 0.781688803434372,
+      "num_tokens": 98019137.0,
+      "step": 28810
+    },
+    {
+      "epoch": 0.2892528804849653,
+      "learning_rate": 0.00014215143120960297,
+      "loss": 0.7424,
+      "mean_token_accuracy": 0.7789560616016388,
+      "num_tokens": 98053377.0,
+      "step": 28820
+    },
+    {
+      "epoch": 0.28935324581476574,
+      "learning_rate": 0.00014213135814364287,
+      "loss": 0.7792,
+      "mean_token_accuracy": 0.7686887979507446,
+      "num_tokens": 98087491.0,
+      "step": 28830
+    },
+    {
+      "epoch": 0.28945361114456625,
+      "learning_rate": 0.00014211128507768276,
+      "loss": 0.7418,
+      "mean_token_accuracy": 0.7751915872097015,
+      "num_tokens": 98121681.0,
+      "step": 28840
+    },
+    {
+      "epoch": 0.2895539764743667,
+      "learning_rate": 0.0001420912120117227,
+      "loss": 0.7228,
+      "mean_token_accuracy": 0.7830632448196411,
+      "num_tokens": 98155558.0,
+      "step": 28850
+    },
+    {
+      "epoch": 0.28965434180416716,
+      "learning_rate": 0.00014207113894576257,
+      "loss": 0.7255,
+      "mean_token_accuracy": 0.7823713898658753,
+      "num_tokens": 98189916.0,
+      "step": 28860
+    },
+    {
+      "epoch": 0.2897547071339676,
+      "learning_rate": 0.00014205106587980248,
+      "loss": 0.7819,
+      "mean_token_accuracy": 0.7697534620761871,
+      "num_tokens": 98223477.0,
+      "step": 28870
+    },
+    {
+      "epoch": 0.2898550724637681,
+      "learning_rate": 0.0001420309928138424,
+      "loss": 0.7167,
+      "mean_token_accuracy": 0.781836849451065,
+      "num_tokens": 98257219.0,
+      "step": 28880
+    },
+    {
+      "epoch": 0.2899554377935686,
+      "learning_rate": 0.0001420109197478823,
+      "loss": 0.7227,
+      "mean_token_accuracy": 0.7811435163021088,
+      "num_tokens": 98291835.0,
+      "step": 28890
+    },
+    {
+      "epoch": 0.29005580312336904,
+      "learning_rate": 0.0001419908466819222,
+      "loss": 0.7484,
+      "mean_token_accuracy": 0.7724690556526184,
+      "num_tokens": 98325991.0,
+      "step": 28900
+    },
+    {
+      "epoch": 0.29015616845316955,
+      "learning_rate": 0.00014197077361596211,
+      "loss": 0.7015,
+      "mean_token_accuracy": 0.7852222204208374,
+      "num_tokens": 98360498.0,
+      "step": 28910
+    },
+    {
+      "epoch": 0.29025653378297,
+      "learning_rate": 0.00014195070055000202,
+      "loss": 0.7659,
+      "mean_token_accuracy": 0.7725250482559204,
+      "num_tokens": 98395355.0,
+      "step": 28920
+    },
+    {
+      "epoch": 0.29035689911277046,
+      "learning_rate": 0.0001419306274840419,
+      "loss": 0.7502,
+      "mean_token_accuracy": 0.781274002790451,
+      "num_tokens": 98428841.0,
+      "step": 28930
+    },
+    {
+      "epoch": 0.290457264442571,
+      "learning_rate": 0.00014191055441808184,
+      "loss": 0.7604,
+      "mean_token_accuracy": 0.7756985306739808,
+      "num_tokens": 98462247.0,
+      "step": 28940
+    },
+    {
+      "epoch": 0.29055762977237143,
+      "learning_rate": 0.00014189048135212172,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.7663555085659027,
+      "num_tokens": 98496044.0,
+      "step": 28950
+    },
+    {
+      "epoch": 0.2906579951021719,
+      "learning_rate": 0.00014187040828616163,
+      "loss": 0.7511,
+      "mean_token_accuracy": 0.7796390414237976,
+      "num_tokens": 98529092.0,
+      "step": 28960
+    },
+    {
+      "epoch": 0.2907583604319724,
+      "learning_rate": 0.00014185033522020154,
+      "loss": 0.7276,
+      "mean_token_accuracy": 0.7819804608821869,
+      "num_tokens": 98563656.0,
+      "step": 28970
+    },
+    {
+      "epoch": 0.29085872576177285,
+      "learning_rate": 0.00014183026215424145,
+      "loss": 0.78,
+      "mean_token_accuracy": 0.7645665585994721,
+      "num_tokens": 98596523.0,
+      "step": 28980
+    },
+    {
+      "epoch": 0.2909590910915733,
+      "learning_rate": 0.00014181018908828135,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.772081583738327,
+      "num_tokens": 98630427.0,
+      "step": 28990
+    },
+    {
+      "epoch": 0.2910594564213738,
+      "learning_rate": 0.00014179011602232126,
+      "loss": 0.7434,
+      "mean_token_accuracy": 0.7764184951782227,
+      "num_tokens": 98663817.0,
+      "step": 29000
+    },
+    {
+      "epoch": 0.2911598217511743,
+      "learning_rate": 0.00014177004295636117,
+      "loss": 0.7724,
+      "mean_token_accuracy": 0.7713199734687806,
+      "num_tokens": 98697541.0,
+      "step": 29010
+    },
+    {
+      "epoch": 0.29126018708097473,
+      "learning_rate": 0.00014174996989040105,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7737714588642121,
+      "num_tokens": 98731366.0,
+      "step": 29020
+    },
+    {
+      "epoch": 0.29136055241077524,
+      "learning_rate": 0.000141729896824441,
+      "loss": 0.728,
+      "mean_token_accuracy": 0.7822924375534057,
+      "num_tokens": 98765363.0,
+      "step": 29030
+    },
+    {
+      "epoch": 0.2914609177405757,
+      "learning_rate": 0.00014170982375848087,
+      "loss": 0.7612,
+      "mean_token_accuracy": 0.7754798293113708,
+      "num_tokens": 98799259.0,
+      "step": 29040
+    },
+    {
+      "epoch": 0.29156128307037615,
+      "learning_rate": 0.00014168975069252078,
+      "loss": 0.7352,
+      "mean_token_accuracy": 0.7805355131626129,
+      "num_tokens": 98832864.0,
+      "step": 29050
+    },
+    {
+      "epoch": 0.29166164840017667,
+      "learning_rate": 0.0001416696776265607,
+      "loss": 0.762,
+      "mean_token_accuracy": 0.7746479451656342,
+      "num_tokens": 98867332.0,
+      "step": 29060
+    },
+    {
+      "epoch": 0.2917620137299771,
+      "learning_rate": 0.0001416496045606006,
+      "loss": 0.797,
+      "mean_token_accuracy": 0.7642375528812408,
+      "num_tokens": 98901255.0,
+      "step": 29070
+    },
+    {
+      "epoch": 0.2918623790597776,
+      "learning_rate": 0.00014162953149464048,
+      "loss": 0.708,
+      "mean_token_accuracy": 0.7842754781246185,
+      "num_tokens": 98935116.0,
+      "step": 29080
+    },
+    {
+      "epoch": 0.2919627443895781,
+      "learning_rate": 0.0001416094584286804,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7736884295940399,
+      "num_tokens": 98969475.0,
+      "step": 29090
+    },
+    {
+      "epoch": 0.29206310971937854,
+      "learning_rate": 0.0001415893853627203,
+      "loss": 0.7473,
+      "mean_token_accuracy": 0.7742801606655121,
+      "num_tokens": 99003262.0,
+      "step": 29100
+    },
+    {
+      "epoch": 0.292163475049179,
+      "learning_rate": 0.0001415693122967602,
+      "loss": 0.7715,
+      "mean_token_accuracy": 0.7666668891906738,
+      "num_tokens": 99037582.0,
+      "step": 29110
+    },
+    {
+      "epoch": 0.2922638403789795,
+      "learning_rate": 0.00014154923923080014,
+      "loss": 0.7807,
+      "mean_token_accuracy": 0.770354300737381,
+      "num_tokens": 99071947.0,
+      "step": 29120
+    },
+    {
+      "epoch": 0.29236420570877997,
+      "learning_rate": 0.00014152916616484002,
+      "loss": 0.7485,
+      "mean_token_accuracy": 0.7715291023254395,
+      "num_tokens": 99106290.0,
+      "step": 29130
+    },
+    {
+      "epoch": 0.2924645710385804,
+      "learning_rate": 0.00014150909309887993,
+      "loss": 0.7623,
+      "mean_token_accuracy": 0.7719914376735687,
+      "num_tokens": 99141141.0,
+      "step": 29140
+    },
+    {
+      "epoch": 0.2925649363683809,
+      "learning_rate": 0.00014148902003291984,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.7699817597866059,
+      "num_tokens": 99175816.0,
+      "step": 29150
+    },
+    {
+      "epoch": 0.2926653016981814,
+      "learning_rate": 0.00014146894696695974,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.7794719636440277,
+      "num_tokens": 99210735.0,
+      "step": 29160
+    },
+    {
+      "epoch": 0.29276566702798185,
+      "learning_rate": 0.00014144887390099963,
+      "loss": 0.7402,
+      "mean_token_accuracy": 0.7780381441116333,
+      "num_tokens": 99244651.0,
+      "step": 29170
+    },
+    {
+      "epoch": 0.2928660323577823,
+      "learning_rate": 0.00014142880083503956,
+      "loss": 0.7765,
+      "mean_token_accuracy": 0.7682926177978515,
+      "num_tokens": 99278357.0,
+      "step": 29180
+    },
+    {
+      "epoch": 0.2929663976875828,
+      "learning_rate": 0.00014140872776907944,
+      "loss": 0.7646,
+      "mean_token_accuracy": 0.7708244085311889,
+      "num_tokens": 99311832.0,
+      "step": 29190
+    },
+    {
+      "epoch": 0.29306676301738327,
+      "learning_rate": 0.00014138865470311938,
+      "loss": 0.7416,
+      "mean_token_accuracy": 0.7788143336772919,
+      "num_tokens": 99346131.0,
+      "step": 29200
+    },
+    {
+      "epoch": 0.2931671283471837,
+      "learning_rate": 0.00014136858163715926,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7704828321933747,
+      "num_tokens": 99380558.0,
+      "step": 29210
+    },
+    {
+      "epoch": 0.29326749367698424,
+      "learning_rate": 0.00014134850857119917,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7756388902664184,
+      "num_tokens": 99414723.0,
+      "step": 29220
+    },
+    {
+      "epoch": 0.2933678590067847,
+      "learning_rate": 0.00014132843550523908,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7792263329029083,
+      "num_tokens": 99448669.0,
+      "step": 29230
+    },
+    {
+      "epoch": 0.29346822433658515,
+      "learning_rate": 0.00014130836243927898,
+      "loss": 0.7497,
+      "mean_token_accuracy": 0.7784649908542634,
+      "num_tokens": 99482972.0,
+      "step": 29240
+    },
+    {
+      "epoch": 0.29356858966638566,
+      "learning_rate": 0.0001412882893733189,
+      "loss": 0.7427,
+      "mean_token_accuracy": 0.7733777105808258,
+      "num_tokens": 99517517.0,
+      "step": 29250
+    },
+    {
+      "epoch": 0.2936689549961861,
+      "learning_rate": 0.0001412682163073588,
+      "loss": 0.7199,
+      "mean_token_accuracy": 0.7836706876754761,
+      "num_tokens": 99552197.0,
+      "step": 29260
+    },
+    {
+      "epoch": 0.29376932032598657,
+      "learning_rate": 0.0001412481432413987,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7785599768161774,
+      "num_tokens": 99585933.0,
+      "step": 29270
+    },
+    {
+      "epoch": 0.2938696856557871,
+      "learning_rate": 0.0001412280701754386,
+      "loss": 0.7682,
+      "mean_token_accuracy": 0.7654083967208862,
+      "num_tokens": 99619927.0,
+      "step": 29280
+    },
+    {
+      "epoch": 0.29397005098558754,
+      "learning_rate": 0.00014120799710947853,
+      "loss": 0.762,
+      "mean_token_accuracy": 0.769352251291275,
+      "num_tokens": 99654333.0,
+      "step": 29290
+    },
+    {
+      "epoch": 0.294070416315388,
+      "learning_rate": 0.0001411879240435184,
+      "loss": 0.7675,
+      "mean_token_accuracy": 0.7726259291172027,
+      "num_tokens": 99688347.0,
+      "step": 29300
+    },
+    {
+      "epoch": 0.2941707816451885,
+      "learning_rate": 0.00014116785097755832,
+      "loss": 0.7247,
+      "mean_token_accuracy": 0.7817895948886872,
+      "num_tokens": 99722710.0,
+      "step": 29310
+    },
+    {
+      "epoch": 0.29427114697498896,
+      "learning_rate": 0.00014114777791159822,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7785114288330078,
+      "num_tokens": 99756516.0,
+      "step": 29320
+    },
+    {
+      "epoch": 0.2943715123047894,
+      "learning_rate": 0.00014112770484563813,
+      "loss": 0.7895,
+      "mean_token_accuracy": 0.7688580214977264,
+      "num_tokens": 99790753.0,
+      "step": 29330
+    },
+    {
+      "epoch": 0.29447187763458993,
+      "learning_rate": 0.00014110763177967804,
+      "loss": 0.7288,
+      "mean_token_accuracy": 0.7836663603782654,
+      "num_tokens": 99824741.0,
+      "step": 29340
+    },
+    {
+      "epoch": 0.2945722429643904,
+      "learning_rate": 0.00014108755871371795,
+      "loss": 0.7933,
+      "mean_token_accuracy": 0.7669318795204163,
+      "num_tokens": 99858920.0,
+      "step": 29350
+    },
+    {
+      "epoch": 0.29467260829419084,
+      "learning_rate": 0.00014106748564775786,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.7750026464462281,
+      "num_tokens": 99892374.0,
+      "step": 29360
+    },
+    {
+      "epoch": 0.29477297362399135,
+      "learning_rate": 0.00014104741258179774,
+      "loss": 0.7601,
+      "mean_token_accuracy": 0.7714162588119506,
+      "num_tokens": 99927242.0,
+      "step": 29370
+    },
+    {
+      "epoch": 0.2948733389537918,
+      "learning_rate": 0.00014102733951583768,
+      "loss": 0.7441,
+      "mean_token_accuracy": 0.7752436697483063,
+      "num_tokens": 99962405.0,
+      "step": 29380
+    },
+    {
+      "epoch": 0.29497370428359226,
+      "learning_rate": 0.00014100726644987756,
+      "loss": 0.7622,
+      "mean_token_accuracy": 0.7726584315299988,
+      "num_tokens": 99995965.0,
+      "step": 29390
+    },
+    {
+      "epoch": 0.2950740696133928,
+      "learning_rate": 0.00014098719338391746,
+      "loss": 0.7422,
+      "mean_token_accuracy": 0.7753499388694763,
+      "num_tokens": 100029884.0,
+      "step": 29400
+    },
+    {
+      "epoch": 0.29517443494319323,
+      "learning_rate": 0.00014096712031795737,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7731671333312988,
+      "num_tokens": 100064293.0,
+      "step": 29410
+    },
+    {
+      "epoch": 0.2952748002729937,
+      "learning_rate": 0.00014094704725199728,
+      "loss": 0.7929,
+      "mean_token_accuracy": 0.769537627696991,
+      "num_tokens": 100098475.0,
+      "step": 29420
+    },
+    {
+      "epoch": 0.2953751656027942,
+      "learning_rate": 0.00014092697418603716,
+      "loss": 0.7344,
+      "mean_token_accuracy": 0.7775164842605591,
+      "num_tokens": 100133249.0,
+      "step": 29430
+    },
+    {
+      "epoch": 0.29547553093259465,
+      "learning_rate": 0.0001409069011200771,
+      "loss": 0.7214,
+      "mean_token_accuracy": 0.779714035987854,
+      "num_tokens": 100167512.0,
+      "step": 29440
+    },
+    {
+      "epoch": 0.2955758962623951,
+      "learning_rate": 0.00014088682805411698,
+      "loss": 0.7678,
+      "mean_token_accuracy": 0.7712717652320862,
+      "num_tokens": 100202601.0,
+      "step": 29450
+    },
+    {
+      "epoch": 0.29567626159219557,
+      "learning_rate": 0.0001408667549881569,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7723312854766846,
+      "num_tokens": 100236753.0,
+      "step": 29460
+    },
+    {
+      "epoch": 0.2957766269219961,
+      "learning_rate": 0.0001408466819221968,
+      "loss": 0.7199,
+      "mean_token_accuracy": 0.7800745785236358,
+      "num_tokens": 100271362.0,
+      "step": 29470
+    },
+    {
+      "epoch": 0.29587699225179653,
+      "learning_rate": 0.0001408266088562367,
+      "loss": 0.7353,
+      "mean_token_accuracy": 0.77441685795784,
+      "num_tokens": 100305577.0,
+      "step": 29480
+    },
+    {
+      "epoch": 0.295977357581597,
+      "learning_rate": 0.00014080653579027661,
+      "loss": 0.7574,
+      "mean_token_accuracy": 0.7717891931533813,
+      "num_tokens": 100339939.0,
+      "step": 29490
+    },
+    {
+      "epoch": 0.2960777229113975,
+      "learning_rate": 0.00014078646272431652,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7759120404720307,
+      "num_tokens": 100374633.0,
+      "step": 29500
+    },
+    {
+      "epoch": 0.29617808824119796,
+      "learning_rate": 0.00014076638965835643,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7745722115039826,
+      "num_tokens": 100409806.0,
+      "step": 29510
+    },
+    {
+      "epoch": 0.2962784535709984,
+      "learning_rate": 0.0001407463165923963,
+      "loss": 0.6996,
+      "mean_token_accuracy": 0.7888730943202973,
+      "num_tokens": 100443260.0,
+      "step": 29520
+    },
+    {
+      "epoch": 0.2963788189007989,
+      "learning_rate": 0.00014072624352643625,
+      "loss": 0.6948,
+      "mean_token_accuracy": 0.7858321964740753,
+      "num_tokens": 100477022.0,
+      "step": 29530
+    },
+    {
+      "epoch": 0.2964791842305994,
+      "learning_rate": 0.00014070617046047613,
+      "loss": 0.759,
+      "mean_token_accuracy": 0.771923303604126,
+      "num_tokens": 100511138.0,
+      "step": 29540
+    },
+    {
+      "epoch": 0.29657954956039984,
+      "learning_rate": 0.00014068609739451604,
+      "loss": 0.7333,
+      "mean_token_accuracy": 0.7735685169696808,
+      "num_tokens": 100545533.0,
+      "step": 29550
+    },
+    {
+      "epoch": 0.29667991489020035,
+      "learning_rate": 0.00014066602432855595,
+      "loss": 0.7196,
+      "mean_token_accuracy": 0.7803890228271484,
+      "num_tokens": 100579346.0,
+      "step": 29560
+    },
+    {
+      "epoch": 0.2967802802200008,
+      "learning_rate": 0.00014064595126259585,
+      "loss": 0.7514,
+      "mean_token_accuracy": 0.7723907589912414,
+      "num_tokens": 100613387.0,
+      "step": 29570
+    },
+    {
+      "epoch": 0.29688064554980126,
+      "learning_rate": 0.00014062587819663576,
+      "loss": 0.7193,
+      "mean_token_accuracy": 0.7758615255355835,
+      "num_tokens": 100647040.0,
+      "step": 29580
+    },
+    {
+      "epoch": 0.29698101087960177,
+      "learning_rate": 0.00014060580513067567,
+      "loss": 0.7482,
+      "mean_token_accuracy": 0.7730087757110595,
+      "num_tokens": 100680660.0,
+      "step": 29590
+    },
+    {
+      "epoch": 0.2970813762094022,
+      "learning_rate": 0.00014058573206471558,
+      "loss": 0.7461,
+      "mean_token_accuracy": 0.7728002905845642,
+      "num_tokens": 100714616.0,
+      "step": 29600
+    },
+    {
+      "epoch": 0.2971817415392027,
+      "learning_rate": 0.00014056565899875546,
+      "loss": 0.7692,
+      "mean_token_accuracy": 0.7765908062458038,
+      "num_tokens": 100748952.0,
+      "step": 29610
+    },
+    {
+      "epoch": 0.2972821068690032,
+      "learning_rate": 0.0001405455859327954,
+      "loss": 0.7341,
+      "mean_token_accuracy": 0.7805267155170441,
+      "num_tokens": 100782497.0,
+      "step": 29620
+    },
+    {
+      "epoch": 0.29738247219880365,
+      "learning_rate": 0.00014052551286683528,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7727090954780579,
+      "num_tokens": 100816702.0,
+      "step": 29630
+    },
+    {
+      "epoch": 0.2974828375286041,
+      "learning_rate": 0.00014050543980087519,
+      "loss": 0.7204,
+      "mean_token_accuracy": 0.7859593152999877,
+      "num_tokens": 100850951.0,
+      "step": 29640
+    },
+    {
+      "epoch": 0.2975832028584046,
+      "learning_rate": 0.0001404853667349151,
+      "loss": 0.7352,
+      "mean_token_accuracy": 0.7814213275909424,
+      "num_tokens": 100885330.0,
+      "step": 29650
+    },
+    {
+      "epoch": 0.29768356818820507,
+      "learning_rate": 0.000140465293668955,
+      "loss": 0.7652,
+      "mean_token_accuracy": 0.7715207993984222,
+      "num_tokens": 100919542.0,
+      "step": 29660
+    },
+    {
+      "epoch": 0.29778393351800553,
+      "learning_rate": 0.0001404452206029949,
+      "loss": 0.7558,
+      "mean_token_accuracy": 0.772058779001236,
+      "num_tokens": 100953203.0,
+      "step": 29670
+    },
+    {
+      "epoch": 0.29788429884780604,
+      "learning_rate": 0.00014042514753703482,
+      "loss": 0.7221,
+      "mean_token_accuracy": 0.781199312210083,
+      "num_tokens": 100987335.0,
+      "step": 29680
+    },
+    {
+      "epoch": 0.2979846641776065,
+      "learning_rate": 0.00014040507447107473,
+      "loss": 0.7492,
+      "mean_token_accuracy": 0.7747159957885742,
+      "num_tokens": 101022161.0,
+      "step": 29690
+    },
+    {
+      "epoch": 0.29808502950740695,
+      "learning_rate": 0.0001403850014051146,
+      "loss": 0.7362,
+      "mean_token_accuracy": 0.7758022427558899,
+      "num_tokens": 101056388.0,
+      "step": 29700
+    },
+    {
+      "epoch": 0.29818539483720746,
+      "learning_rate": 0.00014036492833915455,
+      "loss": 0.7661,
+      "mean_token_accuracy": 0.7724496126174927,
+      "num_tokens": 101090576.0,
+      "step": 29710
+    },
+    {
+      "epoch": 0.2982857601670079,
+      "learning_rate": 0.00014034485527319443,
+      "loss": 0.7649,
+      "mean_token_accuracy": 0.7721573054790497,
+      "num_tokens": 101124962.0,
+      "step": 29720
+    },
+    {
+      "epoch": 0.2983861254968084,
+      "learning_rate": 0.00014032478220723433,
+      "loss": 0.7833,
+      "mean_token_accuracy": 0.7669749796390534,
+      "num_tokens": 101159222.0,
+      "step": 29730
+    },
+    {
+      "epoch": 0.29848649082660883,
+      "learning_rate": 0.00014030470914127424,
+      "loss": 0.7649,
+      "mean_token_accuracy": 0.7715265154838562,
+      "num_tokens": 101192782.0,
+      "step": 29740
+    },
+    {
+      "epoch": 0.29858685615640934,
+      "learning_rate": 0.00014028463607531415,
+      "loss": 0.7386,
+      "mean_token_accuracy": 0.7763209939002991,
+      "num_tokens": 101227386.0,
+      "step": 29750
+    },
+    {
+      "epoch": 0.2986872214862098,
+      "learning_rate": 0.00014026456300935406,
+      "loss": 0.7645,
+      "mean_token_accuracy": 0.7729579210281372,
+      "num_tokens": 101261496.0,
+      "step": 29760
+    },
+    {
+      "epoch": 0.29878758681601025,
+      "learning_rate": 0.00014024448994339397,
+      "loss": 0.7807,
+      "mean_token_accuracy": 0.7649962842464447,
+      "num_tokens": 101294541.0,
+      "step": 29770
+    },
+    {
+      "epoch": 0.29888795214581076,
+      "learning_rate": 0.00014022441687743385,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7742062866687774,
+      "num_tokens": 101328027.0,
+      "step": 29780
+    },
+    {
+      "epoch": 0.2989883174756112,
+      "learning_rate": 0.00014020434381147379,
+      "loss": 0.7641,
+      "mean_token_accuracy": 0.7726384162902832,
+      "num_tokens": 101362470.0,
+      "step": 29790
+    },
+    {
+      "epoch": 0.2990886828054117,
+      "learning_rate": 0.00014018427074551367,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.7763361215591431,
+      "num_tokens": 101397922.0,
+      "step": 29800
+    },
+    {
+      "epoch": 0.2991890481352122,
+      "learning_rate": 0.00014016419767955358,
+      "loss": 0.744,
+      "mean_token_accuracy": 0.77735715508461,
+      "num_tokens": 101432261.0,
+      "step": 29810
+    },
+    {
+      "epoch": 0.29928941346501264,
+      "learning_rate": 0.00014014412461359348,
+      "loss": 0.7689,
+      "mean_token_accuracy": 0.768517005443573,
+      "num_tokens": 101466295.0,
+      "step": 29820
+    },
+    {
+      "epoch": 0.2993897787948131,
+      "learning_rate": 0.0001401240515476334,
+      "loss": 0.7406,
+      "mean_token_accuracy": 0.7753639459609986,
+      "num_tokens": 101499988.0,
+      "step": 29830
+    },
+    {
+      "epoch": 0.2994901441246136,
+      "learning_rate": 0.0001401039784816733,
+      "loss": 0.7789,
+      "mean_token_accuracy": 0.7696060717105866,
+      "num_tokens": 101534045.0,
+      "step": 29840
+    },
+    {
+      "epoch": 0.29959050945441407,
+      "learning_rate": 0.0001400839054157132,
+      "loss": 0.7292,
+      "mean_token_accuracy": 0.783731073141098,
+      "num_tokens": 101568206.0,
+      "step": 29850
+    },
+    {
+      "epoch": 0.2996908747842145,
+      "learning_rate": 0.00014006383234975312,
+      "loss": 0.762,
+      "mean_token_accuracy": 0.7774661183357239,
+      "num_tokens": 101602848.0,
+      "step": 29860
+    },
+    {
+      "epoch": 0.29979124011401503,
+      "learning_rate": 0.000140043759283793,
+      "loss": 0.7366,
+      "mean_token_accuracy": 0.7775753319263459,
+      "num_tokens": 101637417.0,
+      "step": 29870
+    },
+    {
+      "epoch": 0.2998916054438155,
+      "learning_rate": 0.00014002368621783293,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7773273289203644,
+      "num_tokens": 101671036.0,
+      "step": 29880
+    },
+    {
+      "epoch": 0.29999197077361595,
+      "learning_rate": 0.00014000361315187282,
+      "loss": 0.761,
+      "mean_token_accuracy": 0.7722024261951447,
+      "num_tokens": 101704395.0,
+      "step": 29890
+    },
+    {
+      "epoch": 0.30009233610341646,
+      "learning_rate": 0.00013998354008591272,
+      "loss": 0.7846,
+      "mean_token_accuracy": 0.7594796240329742,
+      "num_tokens": 101739177.0,
+      "step": 29900
+    },
+    {
+      "epoch": 0.3001927014332169,
+      "learning_rate": 0.00013996346701995263,
+      "loss": 0.7551,
+      "mean_token_accuracy": 0.7724564135074615,
+      "num_tokens": 101773232.0,
+      "step": 29910
+    },
+    {
+      "epoch": 0.30029306676301737,
+      "learning_rate": 0.00013994339395399254,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7685364961624146,
+      "num_tokens": 101806573.0,
+      "step": 29920
+    },
+    {
+      "epoch": 0.3003934320928179,
+      "learning_rate": 0.00013992332088803245,
+      "loss": 0.7848,
+      "mean_token_accuracy": 0.7614635407924653,
+      "num_tokens": 101840236.0,
+      "step": 29930
+    },
+    {
+      "epoch": 0.30049379742261834,
+      "learning_rate": 0.00013990324782207236,
+      "loss": 0.7807,
+      "mean_token_accuracy": 0.767466914653778,
+      "num_tokens": 101874586.0,
+      "step": 29940
+    },
+    {
+      "epoch": 0.3005941627524188,
+      "learning_rate": 0.00013988317475611227,
+      "loss": 0.7185,
+      "mean_token_accuracy": 0.7807386338710784,
+      "num_tokens": 101908606.0,
+      "step": 29950
+    },
+    {
+      "epoch": 0.3006945280822193,
+      "learning_rate": 0.00013986310169015215,
+      "loss": 0.76,
+      "mean_token_accuracy": 0.7769132673740387,
+      "num_tokens": 101942444.0,
+      "step": 29960
+    },
+    {
+      "epoch": 0.30079489341201976,
+      "learning_rate": 0.00013984302862419208,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.771494460105896,
+      "num_tokens": 101976158.0,
+      "step": 29970
+    },
+    {
+      "epoch": 0.3008952587418202,
+      "learning_rate": 0.00013982295555823196,
+      "loss": 0.7816,
+      "mean_token_accuracy": 0.7671621084213257,
+      "num_tokens": 102009545.0,
+      "step": 29980
+    },
+    {
+      "epoch": 0.3009956240716207,
+      "learning_rate": 0.00013980288249227187,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7769341826438904,
+      "num_tokens": 102044039.0,
+      "step": 29990
+    },
+    {
+      "epoch": 0.3010959894014212,
+      "learning_rate": 0.00013978280942631178,
+      "loss": 0.7284,
+      "mean_token_accuracy": 0.779939603805542,
+      "num_tokens": 102077900.0,
+      "step": 30000
+    },
+    {
+      "epoch": 0.30119635473122164,
+      "learning_rate": 0.0001397627363603517,
+      "loss": 0.7558,
+      "mean_token_accuracy": 0.7702864706516266,
+      "num_tokens": 102112192.0,
+      "step": 30010
+    },
+    {
+      "epoch": 0.3012967200610221,
+      "learning_rate": 0.0001397426632943916,
+      "loss": 0.742,
+      "mean_token_accuracy": 0.7804159045219421,
+      "num_tokens": 102145289.0,
+      "step": 30020
+    },
+    {
+      "epoch": 0.3013970853908226,
+      "learning_rate": 0.0001397225902284315,
+      "loss": 0.7393,
+      "mean_token_accuracy": 0.7782248795032501,
+      "num_tokens": 102179561.0,
+      "step": 30030
+    },
+    {
+      "epoch": 0.30149745072062306,
+      "learning_rate": 0.00013970251716247142,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7731903374195099,
+      "num_tokens": 102214258.0,
+      "step": 30040
+    },
+    {
+      "epoch": 0.3015978160504235,
+      "learning_rate": 0.0001396824440965113,
+      "loss": 0.7254,
+      "mean_token_accuracy": 0.7820830702781677,
+      "num_tokens": 102248720.0,
+      "step": 30050
+    },
+    {
+      "epoch": 0.30169818138022403,
+      "learning_rate": 0.00013966237103055123,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7698108375072479,
+      "num_tokens": 102282978.0,
+      "step": 30060
+    },
+    {
+      "epoch": 0.3017985467100245,
+      "learning_rate": 0.0001396422979645911,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7666373908519745,
+      "num_tokens": 102316762.0,
+      "step": 30070
+    },
+    {
+      "epoch": 0.30189891203982494,
+      "learning_rate": 0.00013962222489863102,
+      "loss": 0.7706,
+      "mean_token_accuracy": 0.7717369139194489,
+      "num_tokens": 102350881.0,
+      "step": 30080
+    },
+    {
+      "epoch": 0.30199927736962545,
+      "learning_rate": 0.00013960215183267093,
+      "loss": 0.7883,
+      "mean_token_accuracy": 0.7646639883518219,
+      "num_tokens": 102385177.0,
+      "step": 30090
+    },
+    {
+      "epoch": 0.3020996426994259,
+      "learning_rate": 0.00013958207876671084,
+      "loss": 0.729,
+      "mean_token_accuracy": 0.7823030114173889,
+      "num_tokens": 102418790.0,
+      "step": 30100
+    },
+    {
+      "epoch": 0.30220000802922636,
+      "learning_rate": 0.00013956200570075072,
+      "loss": 0.7357,
+      "mean_token_accuracy": 0.7797044217586517,
+      "num_tokens": 102453132.0,
+      "step": 30110
+    },
+    {
+      "epoch": 0.3023003733590269,
+      "learning_rate": 0.00013954193263479066,
+      "loss": 0.8085,
+      "mean_token_accuracy": 0.7604319095611572,
+      "num_tokens": 102486414.0,
+      "step": 30120
+    },
+    {
+      "epoch": 0.30240073868882733,
+      "learning_rate": 0.00013952185956883054,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7722374677658081,
+      "num_tokens": 102520116.0,
+      "step": 30130
+    },
+    {
+      "epoch": 0.3025011040186278,
+      "learning_rate": 0.00013950178650287045,
+      "loss": 0.7408,
+      "mean_token_accuracy": 0.7763288021087646,
+      "num_tokens": 102554004.0,
+      "step": 30140
+    },
+    {
+      "epoch": 0.3026014693484283,
+      "learning_rate": 0.00013948171343691035,
+      "loss": 0.8025,
+      "mean_token_accuracy": 0.7670863509178162,
+      "num_tokens": 102588513.0,
+      "step": 30150
+    },
+    {
+      "epoch": 0.30270183467822875,
+      "learning_rate": 0.00013946164037095026,
+      "loss": 0.7587,
+      "mean_token_accuracy": 0.7740037620067597,
+      "num_tokens": 102623041.0,
+      "step": 30160
+    },
+    {
+      "epoch": 0.3028022000080292,
+      "learning_rate": 0.00013944156730499017,
+      "loss": 0.7721,
+      "mean_token_accuracy": 0.7665294706821442,
+      "num_tokens": 102657413.0,
+      "step": 30170
+    },
+    {
+      "epoch": 0.3029025653378297,
+      "learning_rate": 0.00013942149423903008,
+      "loss": 0.7335,
+      "mean_token_accuracy": 0.7742194294929504,
+      "num_tokens": 102691460.0,
+      "step": 30180
+    },
+    {
+      "epoch": 0.3030029306676302,
+      "learning_rate": 0.00013940142117307,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.7743681907653809,
+      "num_tokens": 102725438.0,
+      "step": 30190
+    },
+    {
+      "epoch": 0.30310329599743063,
+      "learning_rate": 0.00013938134810710987,
+      "loss": 0.767,
+      "mean_token_accuracy": 0.7699396669864654,
+      "num_tokens": 102760347.0,
+      "step": 30200
+    },
+    {
+      "epoch": 0.30320366132723114,
+      "learning_rate": 0.0001393612750411498,
+      "loss": 0.7362,
+      "mean_token_accuracy": 0.7853947639465332,
+      "num_tokens": 102795404.0,
+      "step": 30210
+    },
+    {
+      "epoch": 0.3033040266570316,
+      "learning_rate": 0.00013934120197518969,
+      "loss": 0.743,
+      "mean_token_accuracy": 0.7780166029930115,
+      "num_tokens": 102828487.0,
+      "step": 30220
+    },
+    {
+      "epoch": 0.30340439198683206,
+      "learning_rate": 0.0001393211289092296,
+      "loss": 0.7794,
+      "mean_token_accuracy": 0.7706028580665588,
+      "num_tokens": 102863431.0,
+      "step": 30230
+    },
+    {
+      "epoch": 0.30350475731663257,
+      "learning_rate": 0.0001393010558432695,
+      "loss": 0.776,
+      "mean_token_accuracy": 0.7714719414710999,
+      "num_tokens": 102898248.0,
+      "step": 30240
+    },
+    {
+      "epoch": 0.303605122646433,
+      "learning_rate": 0.0001392809827773094,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7676347851753235,
+      "num_tokens": 102932695.0,
+      "step": 30250
+    },
+    {
+      "epoch": 0.3037054879762335,
+      "learning_rate": 0.00013926090971134932,
+      "loss": 0.7732,
+      "mean_token_accuracy": 0.7656772911548615,
+      "num_tokens": 102966717.0,
+      "step": 30260
+    },
+    {
+      "epoch": 0.303805853306034,
+      "learning_rate": 0.00013924083664538923,
+      "loss": 0.826,
+      "mean_token_accuracy": 0.7555188834667206,
+      "num_tokens": 102999671.0,
+      "step": 30270
+    },
+    {
+      "epoch": 0.30390621863583445,
+      "learning_rate": 0.00013922076357942914,
+      "loss": 0.7552,
+      "mean_token_accuracy": 0.7759094893932342,
+      "num_tokens": 103033895.0,
+      "step": 30280
+    },
+    {
+      "epoch": 0.3040065839656349,
+      "learning_rate": 0.00013920069051346902,
+      "loss": 0.7419,
+      "mean_token_accuracy": 0.775572144985199,
+      "num_tokens": 103067820.0,
+      "step": 30290
+    },
+    {
+      "epoch": 0.30410694929543536,
+      "learning_rate": 0.00013918061744750895,
+      "loss": 0.7392,
+      "mean_token_accuracy": 0.7811022341251374,
+      "num_tokens": 103102289.0,
+      "step": 30300
+    },
+    {
+      "epoch": 0.30420731462523587,
+      "learning_rate": 0.00013916054438154883,
+      "loss": 0.7273,
+      "mean_token_accuracy": 0.7775943994522094,
+      "num_tokens": 103136433.0,
+      "step": 30310
+    },
+    {
+      "epoch": 0.3043076799550363,
+      "learning_rate": 0.00013914047131558877,
+      "loss": 0.733,
+      "mean_token_accuracy": 0.7790721535682679,
+      "num_tokens": 103170299.0,
+      "step": 30320
+    },
+    {
+      "epoch": 0.3044080452848368,
+      "learning_rate": 0.00013912039824962865,
+      "loss": 0.7854,
+      "mean_token_accuracy": 0.7666976571083068,
+      "num_tokens": 103204040.0,
+      "step": 30330
+    },
+    {
+      "epoch": 0.3045084106146373,
+      "learning_rate": 0.00013910032518366856,
+      "loss": 0.7398,
+      "mean_token_accuracy": 0.7777034878730774,
+      "num_tokens": 103237977.0,
+      "step": 30340
+    },
+    {
+      "epoch": 0.30460877594443775,
+      "learning_rate": 0.00013908025211770847,
+      "loss": 0.7711,
+      "mean_token_accuracy": 0.7719676792621613,
+      "num_tokens": 103272465.0,
+      "step": 30350
+    },
+    {
+      "epoch": 0.3047091412742382,
+      "learning_rate": 0.00013906017905174838,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.7756696701049804,
+      "num_tokens": 103306302.0,
+      "step": 30360
+    },
+    {
+      "epoch": 0.3048095066040387,
+      "learning_rate": 0.00013904010598578828,
+      "loss": 0.7459,
+      "mean_token_accuracy": 0.7794725835323334,
+      "num_tokens": 103340295.0,
+      "step": 30370
+    },
+    {
+      "epoch": 0.30490987193383917,
+      "learning_rate": 0.0001390200329198282,
+      "loss": 0.7594,
+      "mean_token_accuracy": 0.7782514810562133,
+      "num_tokens": 103374872.0,
+      "step": 30380
+    },
+    {
+      "epoch": 0.3050102372636396,
+      "learning_rate": 0.0001389999598538681,
+      "loss": 0.7999,
+      "mean_token_accuracy": 0.7694885909557343,
+      "num_tokens": 103408841.0,
+      "step": 30390
+    },
+    {
+      "epoch": 0.30511060259344014,
+      "learning_rate": 0.00013897988678790798,
+      "loss": 0.7787,
+      "mean_token_accuracy": 0.7691689252853393,
+      "num_tokens": 103443503.0,
+      "step": 30400
+    },
+    {
+      "epoch": 0.3052109679232406,
+      "learning_rate": 0.00013895981372194792,
+      "loss": 0.746,
+      "mean_token_accuracy": 0.7696506142616272,
+      "num_tokens": 103477779.0,
+      "step": 30410
+    },
+    {
+      "epoch": 0.30531133325304105,
+      "learning_rate": 0.0001389397406559878,
+      "loss": 0.7598,
+      "mean_token_accuracy": 0.7774660527706146,
+      "num_tokens": 103511407.0,
+      "step": 30420
+    },
+    {
+      "epoch": 0.30541169858284156,
+      "learning_rate": 0.0001389196675900277,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7759681701660156,
+      "num_tokens": 103545543.0,
+      "step": 30430
+    },
+    {
+      "epoch": 0.305512063912642,
+      "learning_rate": 0.00013889959452406762,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7778256833553314,
+      "num_tokens": 103579468.0,
+      "step": 30440
+    },
+    {
+      "epoch": 0.3056124292424425,
+      "learning_rate": 0.00013887952145810753,
+      "loss": 0.7532,
+      "mean_token_accuracy": 0.7719829618930817,
+      "num_tokens": 103613942.0,
+      "step": 30450
+    },
+    {
+      "epoch": 0.305712794572243,
+      "learning_rate": 0.0001388594483921474,
+      "loss": 0.7545,
+      "mean_token_accuracy": 0.7741083145141602,
+      "num_tokens": 103647817.0,
+      "step": 30460
+    },
+    {
+      "epoch": 0.30581315990204344,
+      "learning_rate": 0.00013883937532618734,
+      "loss": 0.7698,
+      "mean_token_accuracy": 0.7674217760562897,
+      "num_tokens": 103681743.0,
+      "step": 30470
+    },
+    {
+      "epoch": 0.3059135252318439,
+      "learning_rate": 0.00013881930226022722,
+      "loss": 0.7816,
+      "mean_token_accuracy": 0.7681707978248596,
+      "num_tokens": 103715424.0,
+      "step": 30480
+    },
+    {
+      "epoch": 0.3060138905616444,
+      "learning_rate": 0.00013879922919426713,
+      "loss": 0.7705,
+      "mean_token_accuracy": 0.7729944229125977,
+      "num_tokens": 103749449.0,
+      "step": 30490
+    },
+    {
+      "epoch": 0.30611425589144486,
+      "learning_rate": 0.00013877915612830704,
+      "loss": 0.7347,
+      "mean_token_accuracy": 0.7809194803237915,
+      "num_tokens": 103784175.0,
+      "step": 30500
+    },
+    {
+      "epoch": 0.3062146212212453,
+      "learning_rate": 0.00013875908306234695,
+      "loss": 0.7738,
+      "mean_token_accuracy": 0.7669017791748047,
+      "num_tokens": 103818085.0,
+      "step": 30510
+    },
+    {
+      "epoch": 0.30631498655104583,
+      "learning_rate": 0.00013873900999638686,
+      "loss": 0.7881,
+      "mean_token_accuracy": 0.7695654451847076,
+      "num_tokens": 103851803.0,
+      "step": 30520
+    },
+    {
+      "epoch": 0.3064153518808463,
+      "learning_rate": 0.00013871893693042677,
+      "loss": 0.7747,
+      "mean_token_accuracy": 0.7680711686611176,
+      "num_tokens": 103886459.0,
+      "step": 30530
+    },
+    {
+      "epoch": 0.30651571721064674,
+      "learning_rate": 0.00013869886386446667,
+      "loss": 0.7481,
+      "mean_token_accuracy": 0.7746942222118378,
+      "num_tokens": 103920953.0,
+      "step": 30540
+    },
+    {
+      "epoch": 0.30661608254044725,
+      "learning_rate": 0.00013867879079850656,
+      "loss": 0.793,
+      "mean_token_accuracy": 0.763790887594223,
+      "num_tokens": 103954390.0,
+      "step": 30550
+    },
+    {
+      "epoch": 0.3067164478702477,
+      "learning_rate": 0.0001386587177325465,
+      "loss": 0.7995,
+      "mean_token_accuracy": 0.7638724386692047,
+      "num_tokens": 103988371.0,
+      "step": 30560
+    },
+    {
+      "epoch": 0.30681681320004817,
+      "learning_rate": 0.00013863864466658637,
+      "loss": 0.7561,
+      "mean_token_accuracy": 0.774150276184082,
+      "num_tokens": 104021981.0,
+      "step": 30570
+    },
+    {
+      "epoch": 0.3069171785298486,
+      "learning_rate": 0.00013861857160062628,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.7727494657039642,
+      "num_tokens": 104055285.0,
+      "step": 30580
+    },
+    {
+      "epoch": 0.30701754385964913,
+      "learning_rate": 0.0001385984985346662,
+      "loss": 0.7524,
+      "mean_token_accuracy": 0.7772042453289032,
+      "num_tokens": 104089263.0,
+      "step": 30590
+    },
+    {
+      "epoch": 0.3071179091894496,
+      "learning_rate": 0.0001385784254687061,
+      "loss": 0.765,
+      "mean_token_accuracy": 0.7683424353599548,
+      "num_tokens": 104123211.0,
+      "step": 30600
+    },
+    {
+      "epoch": 0.30721827451925005,
+      "learning_rate": 0.000138558352402746,
+      "loss": 0.7288,
+      "mean_token_accuracy": 0.7830067038536072,
+      "num_tokens": 104157222.0,
+      "step": 30610
+    },
+    {
+      "epoch": 0.30731863984905056,
+      "learning_rate": 0.00013853827933678591,
+      "loss": 0.7709,
+      "mean_token_accuracy": 0.7685743570327759,
+      "num_tokens": 104190522.0,
+      "step": 30620
+    },
+    {
+      "epoch": 0.307419005178851,
+      "learning_rate": 0.00013851820627082582,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7725608289241791,
+      "num_tokens": 104224983.0,
+      "step": 30630
+    },
+    {
+      "epoch": 0.30751937050865147,
+      "learning_rate": 0.0001384981332048657,
+      "loss": 0.7824,
+      "mean_token_accuracy": 0.7672821164131165,
+      "num_tokens": 104259059.0,
+      "step": 30640
+    },
+    {
+      "epoch": 0.307619735838452,
+      "learning_rate": 0.00013847806013890564,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.7788600921630859,
+      "num_tokens": 104292259.0,
+      "step": 30650
+    },
+    {
+      "epoch": 0.30772010116825244,
+      "learning_rate": 0.00013845798707294552,
+      "loss": 0.8073,
+      "mean_token_accuracy": 0.76075199842453,
+      "num_tokens": 104326183.0,
+      "step": 30660
+    },
+    {
+      "epoch": 0.3078204664980529,
+      "learning_rate": 0.00013843791400698543,
+      "loss": 0.8143,
+      "mean_token_accuracy": 0.7618832290172577,
+      "num_tokens": 104360667.0,
+      "step": 30670
+    },
+    {
+      "epoch": 0.3079208318278534,
+      "learning_rate": 0.00013841784094102534,
+      "loss": 0.7751,
+      "mean_token_accuracy": 0.7713720083236695,
+      "num_tokens": 104395892.0,
+      "step": 30680
+    },
+    {
+      "epoch": 0.30802119715765386,
+      "learning_rate": 0.00013839776787506525,
+      "loss": 0.747,
+      "mean_token_accuracy": 0.7763884246349335,
+      "num_tokens": 104429388.0,
+      "step": 30690
+    },
+    {
+      "epoch": 0.3081215624874543,
+      "learning_rate": 0.00013837769480910513,
+      "loss": 0.7482,
+      "mean_token_accuracy": 0.7765169620513916,
+      "num_tokens": 104463555.0,
+      "step": 30700
+    },
+    {
+      "epoch": 0.3082219278172548,
+      "learning_rate": 0.00013835762174314506,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.7701191127300262,
+      "num_tokens": 104497207.0,
+      "step": 30710
+    },
+    {
+      "epoch": 0.3083222931470553,
+      "learning_rate": 0.00013833754867718497,
+      "loss": 0.7945,
+      "mean_token_accuracy": 0.7636801123619079,
+      "num_tokens": 104531413.0,
+      "step": 30720
+    },
+    {
+      "epoch": 0.30842265847685574,
+      "learning_rate": 0.00013831747561122485,
+      "loss": 0.7868,
+      "mean_token_accuracy": 0.7641001880168915,
+      "num_tokens": 104565304.0,
+      "step": 30730
+    },
+    {
+      "epoch": 0.30852302380665625,
+      "learning_rate": 0.0001382974025452648,
+      "loss": 0.7358,
+      "mean_token_accuracy": 0.7806992888450622,
+      "num_tokens": 104599197.0,
+      "step": 30740
+    },
+    {
+      "epoch": 0.3086233891364567,
+      "learning_rate": 0.00013827732947930467,
+      "loss": 0.7734,
+      "mean_token_accuracy": 0.767440915107727,
+      "num_tokens": 104633950.0,
+      "step": 30750
+    },
+    {
+      "epoch": 0.30872375446625716,
+      "learning_rate": 0.00013825725641334458,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.7703138053417206,
+      "num_tokens": 104667636.0,
+      "step": 30760
+    },
+    {
+      "epoch": 0.3088241197960577,
+      "learning_rate": 0.0001382371833473845,
+      "loss": 0.7899,
+      "mean_token_accuracy": 0.7707507610321045,
+      "num_tokens": 104701907.0,
+      "step": 30770
+    },
+    {
+      "epoch": 0.30892448512585813,
+      "learning_rate": 0.0001382171102814244,
+      "loss": 0.7455,
+      "mean_token_accuracy": 0.7727533578872681,
+      "num_tokens": 104735953.0,
+      "step": 30780
+    },
+    {
+      "epoch": 0.3090248504556586,
+      "learning_rate": 0.00013819703721546428,
+      "loss": 0.7535,
+      "mean_token_accuracy": 0.7779163002967835,
+      "num_tokens": 104770245.0,
+      "step": 30790
+    },
+    {
+      "epoch": 0.3091252157854591,
+      "learning_rate": 0.0001381769641495042,
+      "loss": 0.7664,
+      "mean_token_accuracy": 0.7698115825653076,
+      "num_tokens": 104804162.0,
+      "step": 30800
+    },
+    {
+      "epoch": 0.30922558111525955,
+      "learning_rate": 0.0001381568910835441,
+      "loss": 0.7567,
+      "mean_token_accuracy": 0.7743724882602692,
+      "num_tokens": 104838125.0,
+      "step": 30810
+    },
+    {
+      "epoch": 0.30932594644506,
+      "learning_rate": 0.000138136818017584,
+      "loss": 0.7572,
+      "mean_token_accuracy": 0.7760859906673432,
+      "num_tokens": 104871914.0,
+      "step": 30820
+    },
+    {
+      "epoch": 0.3094263117748605,
+      "learning_rate": 0.0001381167449516239,
+      "loss": 0.7655,
+      "mean_token_accuracy": 0.7666561007499695,
+      "num_tokens": 104905686.0,
+      "step": 30830
+    },
+    {
+      "epoch": 0.309526677104661,
+      "learning_rate": 0.00013809667188566382,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7743497550487518,
+      "num_tokens": 104939031.0,
+      "step": 30840
+    },
+    {
+      "epoch": 0.30962704243446143,
+      "learning_rate": 0.00013807659881970373,
+      "loss": 0.754,
+      "mean_token_accuracy": 0.7771598994731903,
+      "num_tokens": 104972478.0,
+      "step": 30850
+    },
+    {
+      "epoch": 0.3097274077642619,
+      "learning_rate": 0.00013805652575374364,
+      "loss": 0.7577,
+      "mean_token_accuracy": 0.7708552062511445,
+      "num_tokens": 105006288.0,
+      "step": 30860
+    },
+    {
+      "epoch": 0.3098277730940624,
+      "learning_rate": 0.00013803645268778354,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.7731885015964508,
+      "num_tokens": 105040326.0,
+      "step": 30870
+    },
+    {
+      "epoch": 0.30992813842386285,
+      "learning_rate": 0.00013801637962182343,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7685849487781524,
+      "num_tokens": 105074608.0,
+      "step": 30880
+    },
+    {
+      "epoch": 0.3100285037536633,
+      "learning_rate": 0.00013799630655586336,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.768442940711975,
+      "num_tokens": 105108269.0,
+      "step": 30890
+    },
+    {
+      "epoch": 0.3101288690834638,
+      "learning_rate": 0.00013797623348990324,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7735085904598236,
+      "num_tokens": 105142180.0,
+      "step": 30900
+    },
+    {
+      "epoch": 0.3102292344132643,
+      "learning_rate": 0.00013795616042394318,
+      "loss": 0.7409,
+      "mean_token_accuracy": 0.7764650583267212,
+      "num_tokens": 105177116.0,
+      "step": 30910
+    },
+    {
+      "epoch": 0.31032959974306473,
+      "learning_rate": 0.00013793608735798306,
+      "loss": 0.7345,
+      "mean_token_accuracy": 0.777233612537384,
+      "num_tokens": 105211607.0,
+      "step": 30920
+    },
+    {
+      "epoch": 0.31042996507286524,
+      "learning_rate": 0.00013791601429202297,
+      "loss": 0.7477,
+      "mean_token_accuracy": 0.7734798491001129,
+      "num_tokens": 105245472.0,
+      "step": 30930
+    },
+    {
+      "epoch": 0.3105303304026657,
+      "learning_rate": 0.00013789594122606288,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7751196205615998,
+      "num_tokens": 105280064.0,
+      "step": 30940
+    },
+    {
+      "epoch": 0.31063069573246616,
+      "learning_rate": 0.00013787586816010278,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7729607403278351,
+      "num_tokens": 105313611.0,
+      "step": 30950
+    },
+    {
+      "epoch": 0.31073106106226667,
+      "learning_rate": 0.0001378557950941427,
+      "loss": 0.7604,
+      "mean_token_accuracy": 0.7732480943202973,
+      "num_tokens": 105347185.0,
+      "step": 30960
+    },
+    {
+      "epoch": 0.3108314263920671,
+      "learning_rate": 0.0001378357220281826,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7731426000595093,
+      "num_tokens": 105381308.0,
+      "step": 30970
+    },
+    {
+      "epoch": 0.3109317917218676,
+      "learning_rate": 0.0001378156489622225,
+      "loss": 0.7839,
+      "mean_token_accuracy": 0.7677311778068543,
+      "num_tokens": 105415510.0,
+      "step": 30980
+    },
+    {
+      "epoch": 0.3110321570516681,
+      "learning_rate": 0.0001377955758962624,
+      "loss": 0.7344,
+      "mean_token_accuracy": 0.7777604460716248,
+      "num_tokens": 105448862.0,
+      "step": 30990
+    },
+    {
+      "epoch": 0.31113252238146855,
+      "learning_rate": 0.00013777550283030233,
+      "loss": 0.7904,
+      "mean_token_accuracy": 0.7629985630512237,
+      "num_tokens": 105483254.0,
+      "step": 31000
+    },
+    {
+      "epoch": 0.311232887711269,
+      "learning_rate": 0.0001377554297643422,
+      "loss": 0.7454,
+      "mean_token_accuracy": 0.7740285396575928,
+      "num_tokens": 105516709.0,
+      "step": 31010
+    },
+    {
+      "epoch": 0.3113332530410695,
+      "learning_rate": 0.00013773535669838212,
+      "loss": 0.7731,
+      "mean_token_accuracy": 0.7723234176635743,
+      "num_tokens": 105551036.0,
+      "step": 31020
+    },
+    {
+      "epoch": 0.31143361837086997,
+      "learning_rate": 0.00013771528363242202,
+      "loss": 0.7744,
+      "mean_token_accuracy": 0.7675805270671845,
+      "num_tokens": 105585587.0,
+      "step": 31030
+    },
+    {
+      "epoch": 0.3115339837006704,
+      "learning_rate": 0.00013769521056646193,
+      "loss": 0.7659,
+      "mean_token_accuracy": 0.7708247542381287,
+      "num_tokens": 105619742.0,
+      "step": 31040
+    },
+    {
+      "epoch": 0.31163434903047094,
+      "learning_rate": 0.00013767513750050181,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7713421285152435,
+      "num_tokens": 105653638.0,
+      "step": 31050
+    },
+    {
+      "epoch": 0.3117347143602714,
+      "learning_rate": 0.00013765506443454175,
+      "loss": 0.7852,
+      "mean_token_accuracy": 0.7653680384159088,
+      "num_tokens": 105687649.0,
+      "step": 31060
+    },
+    {
+      "epoch": 0.31183507969007185,
+      "learning_rate": 0.00013763499136858163,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7732458889484406,
+      "num_tokens": 105721675.0,
+      "step": 31070
+    },
+    {
+      "epoch": 0.31193544501987236,
+      "learning_rate": 0.00013761491830262154,
+      "loss": 0.7867,
+      "mean_token_accuracy": 0.7689832389354706,
+      "num_tokens": 105755249.0,
+      "step": 31080
+    },
+    {
+      "epoch": 0.3120358103496728,
+      "learning_rate": 0.00013759484523666148,
+      "loss": 0.7403,
+      "mean_token_accuracy": 0.7799917042255402,
+      "num_tokens": 105788421.0,
+      "step": 31090
+    },
+    {
+      "epoch": 0.31213617567947327,
+      "learning_rate": 0.00013757477217070136,
+      "loss": 0.7231,
+      "mean_token_accuracy": 0.7802851498126984,
+      "num_tokens": 105822170.0,
+      "step": 31100
+    },
+    {
+      "epoch": 0.3122365410092738,
+      "learning_rate": 0.00013755469910474127,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.7696276247501374,
+      "num_tokens": 105856158.0,
+      "step": 31110
+    },
+    {
+      "epoch": 0.31233690633907424,
+      "learning_rate": 0.00013753462603878117,
+      "loss": 0.7574,
+      "mean_token_accuracy": 0.773821359872818,
+      "num_tokens": 105890198.0,
+      "step": 31120
+    },
+    {
+      "epoch": 0.3124372716688747,
+      "learning_rate": 0.00013751455297282108,
+      "loss": 0.7371,
+      "mean_token_accuracy": 0.7768973231315612,
+      "num_tokens": 105923800.0,
+      "step": 31130
+    },
+    {
+      "epoch": 0.31253763699867515,
+      "learning_rate": 0.00013749447990686096,
+      "loss": 0.7404,
+      "mean_token_accuracy": 0.7785003006458282,
+      "num_tokens": 105957717.0,
+      "step": 31140
+    },
+    {
+      "epoch": 0.31263800232847566,
+      "learning_rate": 0.0001374744068409009,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7759685933589935,
+      "num_tokens": 105992243.0,
+      "step": 31150
+    },
+    {
+      "epoch": 0.3127383676582761,
+      "learning_rate": 0.00013745433377494078,
+      "loss": 0.7903,
+      "mean_token_accuracy": 0.7654335021972656,
+      "num_tokens": 106026004.0,
+      "step": 31160
+    },
+    {
+      "epoch": 0.3128387329880766,
+      "learning_rate": 0.0001374342607089807,
+      "loss": 0.8104,
+      "mean_token_accuracy": 0.7664890110492706,
+      "num_tokens": 106059229.0,
+      "step": 31170
+    },
+    {
+      "epoch": 0.3129390983178771,
+      "learning_rate": 0.0001374141876430206,
+      "loss": 0.7665,
+      "mean_token_accuracy": 0.7725627541542053,
+      "num_tokens": 106093422.0,
+      "step": 31180
+    },
+    {
+      "epoch": 0.31303946364767754,
+      "learning_rate": 0.0001373941145770605,
+      "loss": 0.7356,
+      "mean_token_accuracy": 0.7773735344409942,
+      "num_tokens": 106127644.0,
+      "step": 31190
+    },
+    {
+      "epoch": 0.313139828977478,
+      "learning_rate": 0.00013737404151110041,
+      "loss": 0.7087,
+      "mean_token_accuracy": 0.7839349806308746,
+      "num_tokens": 106161610.0,
+      "step": 31200
+    },
+    {
+      "epoch": 0.3132401943072785,
+      "learning_rate": 0.00013735396844514032,
+      "loss": 0.7562,
+      "mean_token_accuracy": 0.7767717659473419,
+      "num_tokens": 106194593.0,
+      "step": 31210
+    },
+    {
+      "epoch": 0.31334055963707896,
+      "learning_rate": 0.00013733389537918023,
+      "loss": 0.731,
+      "mean_token_accuracy": 0.7789615154266357,
+      "num_tokens": 106228903.0,
+      "step": 31220
+    },
+    {
+      "epoch": 0.3134409249668794,
+      "learning_rate": 0.0001373138223132201,
+      "loss": 0.6995,
+      "mean_token_accuracy": 0.7882622420787812,
+      "num_tokens": 106262657.0,
+      "step": 31230
+    },
+    {
+      "epoch": 0.31354129029667993,
+      "learning_rate": 0.00013729374924726005,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7682585716247559,
+      "num_tokens": 106296495.0,
+      "step": 31240
+    },
+    {
+      "epoch": 0.3136416556264804,
+      "learning_rate": 0.00013727367618129993,
+      "loss": 0.796,
+      "mean_token_accuracy": 0.7617917835712433,
+      "num_tokens": 106330196.0,
+      "step": 31250
+    },
+    {
+      "epoch": 0.31374202095628084,
+      "learning_rate": 0.00013725360311533984,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7745411932468415,
+      "num_tokens": 106363984.0,
+      "step": 31260
+    },
+    {
+      "epoch": 0.31384238628608135,
+      "learning_rate": 0.00013723353004937975,
+      "loss": 0.754,
+      "mean_token_accuracy": 0.7728968620300293,
+      "num_tokens": 106398353.0,
+      "step": 31270
+    },
+    {
+      "epoch": 0.3139427516158818,
+      "learning_rate": 0.00013721345698341965,
+      "loss": 0.7627,
+      "mean_token_accuracy": 0.7766358733177186,
+      "num_tokens": 106432556.0,
+      "step": 31280
+    },
+    {
+      "epoch": 0.31404311694568227,
+      "learning_rate": 0.00013719338391745956,
+      "loss": 0.7577,
+      "mean_token_accuracy": 0.7688096821308136,
+      "num_tokens": 106466061.0,
+      "step": 31290
+    },
+    {
+      "epoch": 0.3141434822754828,
+      "learning_rate": 0.00013717331085149947,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7754148781299591,
+      "num_tokens": 106500236.0,
+      "step": 31300
+    },
+    {
+      "epoch": 0.31424384760528323,
+      "learning_rate": 0.00013715323778553938,
+      "loss": 0.7528,
+      "mean_token_accuracy": 0.7724867820739746,
+      "num_tokens": 106533982.0,
+      "step": 31310
+    },
+    {
+      "epoch": 0.3143442129350837,
+      "learning_rate": 0.00013713316471957926,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.7755430579185486,
+      "num_tokens": 106567998.0,
+      "step": 31320
+    },
+    {
+      "epoch": 0.3144445782648842,
+      "learning_rate": 0.0001371130916536192,
+      "loss": 0.7558,
+      "mean_token_accuracy": 0.7747138500213623,
+      "num_tokens": 106602651.0,
+      "step": 31330
+    },
+    {
+      "epoch": 0.31454494359468466,
+      "learning_rate": 0.00013709301858765908,
+      "loss": 0.71,
+      "mean_token_accuracy": 0.7766462087631225,
+      "num_tokens": 106635982.0,
+      "step": 31340
+    },
+    {
+      "epoch": 0.3146453089244851,
+      "learning_rate": 0.00013707294552169899,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7780038237571716,
+      "num_tokens": 106670128.0,
+      "step": 31350
+    },
+    {
+      "epoch": 0.3147456742542856,
+      "learning_rate": 0.0001370528724557389,
+      "loss": 0.7622,
+      "mean_token_accuracy": 0.7697994351387024,
+      "num_tokens": 106704363.0,
+      "step": 31360
+    },
+    {
+      "epoch": 0.3148460395840861,
+      "learning_rate": 0.0001370327993897788,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7660178899765014,
+      "num_tokens": 106738907.0,
+      "step": 31370
+    },
+    {
+      "epoch": 0.31494640491388654,
+      "learning_rate": 0.00013701272632381868,
+      "loss": 0.7937,
+      "mean_token_accuracy": 0.763887357711792,
+      "num_tokens": 106773262.0,
+      "step": 31380
+    },
+    {
+      "epoch": 0.31504677024368705,
+      "learning_rate": 0.00013699265325785862,
+      "loss": 0.7767,
+      "mean_token_accuracy": 0.7671867370605469,
+      "num_tokens": 106807856.0,
+      "step": 31390
+    },
+    {
+      "epoch": 0.3151471355734875,
+      "learning_rate": 0.0001369725801918985,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7766685962677002,
+      "num_tokens": 106842772.0,
+      "step": 31400
+    },
+    {
+      "epoch": 0.31524750090328796,
+      "learning_rate": 0.0001369525071259384,
+      "loss": 0.735,
+      "mean_token_accuracy": 0.7800912439823151,
+      "num_tokens": 106876777.0,
+      "step": 31410
+    },
+    {
+      "epoch": 0.31534786623308847,
+      "learning_rate": 0.00013693243405997832,
+      "loss": 0.7646,
+      "mean_token_accuracy": 0.7699158966541291,
+      "num_tokens": 106909812.0,
+      "step": 31420
+    },
+    {
+      "epoch": 0.3154482315628889,
+      "learning_rate": 0.00013691236099401823,
+      "loss": 0.7232,
+      "mean_token_accuracy": 0.7792066931724548,
+      "num_tokens": 106944204.0,
+      "step": 31430
+    },
+    {
+      "epoch": 0.3155485968926894,
+      "learning_rate": 0.00013689228792805814,
+      "loss": 0.7657,
+      "mean_token_accuracy": 0.7711667120456696,
+      "num_tokens": 106978215.0,
+      "step": 31440
+    },
+    {
+      "epoch": 0.31564896222248984,
+      "learning_rate": 0.00013687221486209804,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.7722912907600403,
+      "num_tokens": 107012101.0,
+      "step": 31450
+    },
+    {
+      "epoch": 0.31574932755229035,
+      "learning_rate": 0.00013685214179613795,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.7634316205978393,
+      "num_tokens": 107045614.0,
+      "step": 31460
+    },
+    {
+      "epoch": 0.3158496928820908,
+      "learning_rate": 0.00013683206873017786,
+      "loss": 0.7509,
+      "mean_token_accuracy": 0.7773948848247528,
+      "num_tokens": 107079266.0,
+      "step": 31470
+    },
+    {
+      "epoch": 0.31595005821189126,
+      "learning_rate": 0.00013681199566421777,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7704627335071563,
+      "num_tokens": 107113132.0,
+      "step": 31480
+    },
+    {
+      "epoch": 0.31605042354169177,
+      "learning_rate": 0.00013679192259825765,
+      "loss": 0.7554,
+      "mean_token_accuracy": 0.7745836615562439,
+      "num_tokens": 107147541.0,
+      "step": 31490
+    },
+    {
+      "epoch": 0.3161507888714922,
+      "learning_rate": 0.00013677184953229759,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.7730957865715027,
+      "num_tokens": 107180909.0,
+      "step": 31500
+    },
+    {
+      "epoch": 0.3162511542012927,
+      "learning_rate": 0.00013675177646633747,
+      "loss": 0.7467,
+      "mean_token_accuracy": 0.7751357734203339,
+      "num_tokens": 107214309.0,
+      "step": 31510
+    },
+    {
+      "epoch": 0.3163515195310932,
+      "learning_rate": 0.00013673170340037738,
+      "loss": 0.7334,
+      "mean_token_accuracy": 0.7759930789470673,
+      "num_tokens": 107247628.0,
+      "step": 31520
+    },
+    {
+      "epoch": 0.31645188486089365,
+      "learning_rate": 0.00013671163033441728,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7736383378505707,
+      "num_tokens": 107280991.0,
+      "step": 31530
+    },
+    {
+      "epoch": 0.3165522501906941,
+      "learning_rate": 0.0001366915572684572,
+      "loss": 0.76,
+      "mean_token_accuracy": 0.7731526851654053,
+      "num_tokens": 107314784.0,
+      "step": 31540
+    },
+    {
+      "epoch": 0.3166526155204946,
+      "learning_rate": 0.0001366714842024971,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.7669507205486298,
+      "num_tokens": 107348737.0,
+      "step": 31550
+    },
+    {
+      "epoch": 0.3167529808502951,
+      "learning_rate": 0.000136651411136537,
+      "loss": 0.7573,
+      "mean_token_accuracy": 0.7724651575088501,
+      "num_tokens": 107382728.0,
+      "step": 31560
+    },
+    {
+      "epoch": 0.31685334618009553,
+      "learning_rate": 0.00013663133807057692,
+      "loss": 0.7223,
+      "mean_token_accuracy": 0.7813074648380279,
+      "num_tokens": 107416169.0,
+      "step": 31570
+    },
+    {
+      "epoch": 0.31695371150989604,
+      "learning_rate": 0.0001366112650046168,
+      "loss": 0.7422,
+      "mean_token_accuracy": 0.7836302399635315,
+      "num_tokens": 107449718.0,
+      "step": 31580
+    },
+    {
+      "epoch": 0.3170540768396965,
+      "learning_rate": 0.00013659119193865673,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.7790060758590698,
+      "num_tokens": 107484770.0,
+      "step": 31590
+    },
+    {
+      "epoch": 0.31715444216949695,
+      "learning_rate": 0.00013657111887269662,
+      "loss": 0.8104,
+      "mean_token_accuracy": 0.7588859498500824,
+      "num_tokens": 107519390.0,
+      "step": 31600
+    },
+    {
+      "epoch": 0.31725480749929746,
+      "learning_rate": 0.00013655104580673652,
+      "loss": 0.7816,
+      "mean_token_accuracy": 0.7687376499176025,
+      "num_tokens": 107553086.0,
+      "step": 31610
+    },
+    {
+      "epoch": 0.3173551728290979,
+      "learning_rate": 0.00013653097274077643,
+      "loss": 0.7693,
+      "mean_token_accuracy": 0.77187077999115,
+      "num_tokens": 107587578.0,
+      "step": 31620
+    },
+    {
+      "epoch": 0.3174555381588984,
+      "learning_rate": 0.00013651089967481634,
+      "loss": 0.7184,
+      "mean_token_accuracy": 0.7835727810859681,
+      "num_tokens": 107620742.0,
+      "step": 31630
+    },
+    {
+      "epoch": 0.3175559034886989,
+      "learning_rate": 0.00013649082660885625,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7751154661178589,
+      "num_tokens": 107655422.0,
+      "step": 31640
+    },
+    {
+      "epoch": 0.31765626881849934,
+      "learning_rate": 0.00013647075354289616,
+      "loss": 0.7476,
+      "mean_token_accuracy": 0.7770822584629059,
+      "num_tokens": 107689751.0,
+      "step": 31650
+    },
+    {
+      "epoch": 0.3177566341482998,
+      "learning_rate": 0.00013645068047693607,
+      "loss": 0.7391,
+      "mean_token_accuracy": 0.7775422394275665,
+      "num_tokens": 107723417.0,
+      "step": 31660
+    },
+    {
+      "epoch": 0.3178569994781003,
+      "learning_rate": 0.00013643060741097595,
+      "loss": 0.7269,
+      "mean_token_accuracy": 0.7786263823509216,
+      "num_tokens": 107757114.0,
+      "step": 31670
+    },
+    {
+      "epoch": 0.31795736480790077,
+      "learning_rate": 0.00013641053434501588,
+      "loss": 0.7582,
+      "mean_token_accuracy": 0.773527330160141,
+      "num_tokens": 107791142.0,
+      "step": 31680
+    },
+    {
+      "epoch": 0.3180577301377012,
+      "learning_rate": 0.00013639046127905576,
+      "loss": 0.8128,
+      "mean_token_accuracy": 0.7656544923782349,
+      "num_tokens": 107825259.0,
+      "step": 31690
+    },
+    {
+      "epoch": 0.31815809546750173,
+      "learning_rate": 0.00013637038821309567,
+      "loss": 0.7108,
+      "mean_token_accuracy": 0.7856820106506348,
+      "num_tokens": 107859872.0,
+      "step": 31700
+    },
+    {
+      "epoch": 0.3182584607973022,
+      "learning_rate": 0.00013635031514713558,
+      "loss": 0.7415,
+      "mean_token_accuracy": 0.7775092601776123,
+      "num_tokens": 107894388.0,
+      "step": 31710
+    },
+    {
+      "epoch": 0.31835882612710265,
+      "learning_rate": 0.0001363302420811755,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.7725362718105316,
+      "num_tokens": 107928361.0,
+      "step": 31720
+    },
+    {
+      "epoch": 0.3184591914569031,
+      "learning_rate": 0.00013631016901521537,
+      "loss": 0.7419,
+      "mean_token_accuracy": 0.7755842924118042,
+      "num_tokens": 107961945.0,
+      "step": 31730
+    },
+    {
+      "epoch": 0.3185595567867036,
+      "learning_rate": 0.0001362900959492553,
+      "loss": 0.764,
+      "mean_token_accuracy": 0.7704953134059906,
+      "num_tokens": 107995563.0,
+      "step": 31740
+    },
+    {
+      "epoch": 0.31865992211650407,
+      "learning_rate": 0.0001362700228832952,
+      "loss": 0.7374,
+      "mean_token_accuracy": 0.7792326033115387,
+      "num_tokens": 108030023.0,
+      "step": 31750
+    },
+    {
+      "epoch": 0.3187602874463045,
+      "learning_rate": 0.0001362499498173351,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7738947212696076,
+      "num_tokens": 108064184.0,
+      "step": 31760
+    },
+    {
+      "epoch": 0.31886065277610504,
+      "learning_rate": 0.000136229876751375,
+      "loss": 0.7567,
+      "mean_token_accuracy": 0.7785474181175231,
+      "num_tokens": 108098008.0,
+      "step": 31770
+    },
+    {
+      "epoch": 0.3189610181059055,
+      "learning_rate": 0.00013620980368541491,
+      "loss": 0.7346,
+      "mean_token_accuracy": 0.7725403249263764,
+      "num_tokens": 108131284.0,
+      "step": 31780
+    },
+    {
+      "epoch": 0.31906138343570595,
+      "learning_rate": 0.00013618973061945482,
+      "loss": 0.7419,
+      "mean_token_accuracy": 0.777106761932373,
+      "num_tokens": 108165107.0,
+      "step": 31790
+    },
+    {
+      "epoch": 0.31916174876550646,
+      "learning_rate": 0.00013616965755349473,
+      "loss": 0.7396,
+      "mean_token_accuracy": 0.7769767999649048,
+      "num_tokens": 108199352.0,
+      "step": 31800
+    },
+    {
+      "epoch": 0.3192621140953069,
+      "learning_rate": 0.00013614958448753464,
+      "loss": 0.7566,
+      "mean_token_accuracy": 0.7712903320789337,
+      "num_tokens": 108234070.0,
+      "step": 31810
+    },
+    {
+      "epoch": 0.31936247942510737,
+      "learning_rate": 0.00013612951142157452,
+      "loss": 0.7499,
+      "mean_token_accuracy": 0.7742361009120942,
+      "num_tokens": 108267886.0,
+      "step": 31820
+    },
+    {
+      "epoch": 0.3194628447549079,
+      "learning_rate": 0.00013610943835561446,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.775706660747528,
+      "num_tokens": 108301250.0,
+      "step": 31830
+    },
+    {
+      "epoch": 0.31956321008470834,
+      "learning_rate": 0.00013608936528965434,
+      "loss": 0.7878,
+      "mean_token_accuracy": 0.7631179451942444,
+      "num_tokens": 108335568.0,
+      "step": 31840
+    },
+    {
+      "epoch": 0.3196635754145088,
+      "learning_rate": 0.00013606929222369425,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.7747395813465119,
+      "num_tokens": 108369565.0,
+      "step": 31850
+    },
+    {
+      "epoch": 0.3197639407443093,
+      "learning_rate": 0.00013604921915773415,
+      "loss": 0.7515,
+      "mean_token_accuracy": 0.7772541761398315,
+      "num_tokens": 108403911.0,
+      "step": 31860
+    },
+    {
+      "epoch": 0.31986430607410976,
+      "learning_rate": 0.00013602914609177406,
+      "loss": 0.7281,
+      "mean_token_accuracy": 0.7788921952247619,
+      "num_tokens": 108438749.0,
+      "step": 31870
+    },
+    {
+      "epoch": 0.3199646714039102,
+      "learning_rate": 0.00013600907302581397,
+      "loss": 0.7665,
+      "mean_token_accuracy": 0.7673018515110016,
+      "num_tokens": 108472615.0,
+      "step": 31880
+    },
+    {
+      "epoch": 0.32006503673371073,
+      "learning_rate": 0.00013598899995985388,
+      "loss": 0.7649,
+      "mean_token_accuracy": 0.7717642366886139,
+      "num_tokens": 108506568.0,
+      "step": 31890
+    },
+    {
+      "epoch": 0.3201654020635112,
+      "learning_rate": 0.0001359689268938938,
+      "loss": 0.7806,
+      "mean_token_accuracy": 0.766598004102707,
+      "num_tokens": 108540136.0,
+      "step": 31900
+    },
+    {
+      "epoch": 0.32026576739331164,
+      "learning_rate": 0.00013594885382793367,
+      "loss": 0.8041,
+      "mean_token_accuracy": 0.7628247916698456,
+      "num_tokens": 108574895.0,
+      "step": 31910
+    },
+    {
+      "epoch": 0.32036613272311215,
+      "learning_rate": 0.0001359287807619736,
+      "loss": 0.7557,
+      "mean_token_accuracy": 0.770247095823288,
+      "num_tokens": 108608655.0,
+      "step": 31920
+    },
+    {
+      "epoch": 0.3204664980529126,
+      "learning_rate": 0.00013590870769601349,
+      "loss": 0.7336,
+      "mean_token_accuracy": 0.7751041054725647,
+      "num_tokens": 108643349.0,
+      "step": 31930
+    },
+    {
+      "epoch": 0.32056686338271306,
+      "learning_rate": 0.0001358886346300534,
+      "loss": 0.7708,
+      "mean_token_accuracy": 0.7705505788326263,
+      "num_tokens": 108677426.0,
+      "step": 31940
+    },
+    {
+      "epoch": 0.3206672287125136,
+      "learning_rate": 0.0001358685615640933,
+      "loss": 0.7629,
+      "mean_token_accuracy": 0.7727705180644989,
+      "num_tokens": 108712090.0,
+      "step": 31950
+    },
+    {
+      "epoch": 0.32076759404231403,
+      "learning_rate": 0.0001358484884981332,
+      "loss": 0.7613,
+      "mean_token_accuracy": 0.7704655826091766,
+      "num_tokens": 108746220.0,
+      "step": 31960
+    },
+    {
+      "epoch": 0.3208679593721145,
+      "learning_rate": 0.00013582841543217312,
+      "loss": 0.7518,
+      "mean_token_accuracy": 0.7753160715103149,
+      "num_tokens": 108780439.0,
+      "step": 31970
+    },
+    {
+      "epoch": 0.320968324701915,
+      "learning_rate": 0.00013580834236621303,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7801798641681671,
+      "num_tokens": 108814220.0,
+      "step": 31980
+    },
+    {
+      "epoch": 0.32106869003171545,
+      "learning_rate": 0.00013578826930025294,
+      "loss": 0.7658,
+      "mean_token_accuracy": 0.7712485730648041,
+      "num_tokens": 108848599.0,
+      "step": 31990
+    },
+    {
+      "epoch": 0.3211690553615159,
+      "learning_rate": 0.00013576819623429282,
+      "loss": 0.7603,
+      "mean_token_accuracy": 0.7678033530712127,
+      "num_tokens": 108881767.0,
+      "step": 32000
+    },
+    {
+      "epoch": 0.32126942069131637,
+      "learning_rate": 0.00013574812316833275,
+      "loss": 0.7152,
+      "mean_token_accuracy": 0.78734210729599,
+      "num_tokens": 108916916.0,
+      "step": 32010
+    },
+    {
+      "epoch": 0.3213697860211169,
+      "learning_rate": 0.00013572805010237263,
+      "loss": 0.7649,
+      "mean_token_accuracy": 0.7759453594684601,
+      "num_tokens": 108950128.0,
+      "step": 32020
+    },
+    {
+      "epoch": 0.32147015135091733,
+      "learning_rate": 0.00013570797703641257,
+      "loss": 0.7364,
+      "mean_token_accuracy": 0.7787595689296722,
+      "num_tokens": 108984782.0,
+      "step": 32030
+    },
+    {
+      "epoch": 0.3215705166807178,
+      "learning_rate": 0.00013568790397045245,
+      "loss": 0.7478,
+      "mean_token_accuracy": 0.774041885137558,
+      "num_tokens": 109018053.0,
+      "step": 32040
+    },
+    {
+      "epoch": 0.3216708820105183,
+      "learning_rate": 0.00013566783090449236,
+      "loss": 0.8212,
+      "mean_token_accuracy": 0.7619028091430664,
+      "num_tokens": 109051791.0,
+      "step": 32050
+    },
+    {
+      "epoch": 0.32177124734031876,
+      "learning_rate": 0.00013564775783853227,
+      "loss": 0.7473,
+      "mean_token_accuracy": 0.7691491365432739,
+      "num_tokens": 109085863.0,
+      "step": 32060
+    },
+    {
+      "epoch": 0.3218716126701192,
+      "learning_rate": 0.00013562768477257218,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.7673633575439454,
+      "num_tokens": 109119871.0,
+      "step": 32070
+    },
+    {
+      "epoch": 0.3219719779999197,
+      "learning_rate": 0.00013560761170661206,
+      "loss": 0.7477,
+      "mean_token_accuracy": 0.7754402875900268,
+      "num_tokens": 109152691.0,
+      "step": 32080
+    },
+    {
+      "epoch": 0.3220723433297202,
+      "learning_rate": 0.000135587538640652,
+      "loss": 0.76,
+      "mean_token_accuracy": 0.7736308097839355,
+      "num_tokens": 109186279.0,
+      "step": 32090
+    },
+    {
+      "epoch": 0.32217270865952063,
+      "learning_rate": 0.00013556746557469188,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.7687101304531098,
+      "num_tokens": 109221336.0,
+      "step": 32100
+    },
+    {
+      "epoch": 0.32227307398932115,
+      "learning_rate": 0.00013554739250873178,
+      "loss": 0.7682,
+      "mean_token_accuracy": 0.7724032521247863,
+      "num_tokens": 109256245.0,
+      "step": 32110
+    },
+    {
+      "epoch": 0.3223734393191216,
+      "learning_rate": 0.0001355273194427717,
+      "loss": 0.737,
+      "mean_token_accuracy": 0.7786790549755096,
+      "num_tokens": 109290679.0,
+      "step": 32120
+    },
+    {
+      "epoch": 0.32247380464892206,
+      "learning_rate": 0.0001355072463768116,
+      "loss": 0.7931,
+      "mean_token_accuracy": 0.7702713489532471,
+      "num_tokens": 109324445.0,
+      "step": 32130
+    },
+    {
+      "epoch": 0.32257416997872257,
+      "learning_rate": 0.0001354871733108515,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.773723566532135,
+      "num_tokens": 109358539.0,
+      "step": 32140
+    },
+    {
+      "epoch": 0.322674535308523,
+      "learning_rate": 0.00013546710024489142,
+      "loss": 0.7328,
+      "mean_token_accuracy": 0.7815119981765747,
+      "num_tokens": 109393450.0,
+      "step": 32150
+    },
+    {
+      "epoch": 0.3227749006383235,
+      "learning_rate": 0.00013544702717893133,
+      "loss": 0.7804,
+      "mean_token_accuracy": 0.7684081137180329,
+      "num_tokens": 109427666.0,
+      "step": 32160
+    },
+    {
+      "epoch": 0.322875265968124,
+      "learning_rate": 0.0001354269541129712,
+      "loss": 0.7633,
+      "mean_token_accuracy": 0.7712672531604767,
+      "num_tokens": 109461562.0,
+      "step": 32170
+    },
+    {
+      "epoch": 0.32297563129792445,
+      "learning_rate": 0.00013540688104701114,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.7765189170837402,
+      "num_tokens": 109495100.0,
+      "step": 32180
+    },
+    {
+      "epoch": 0.3230759966277249,
+      "learning_rate": 0.00013538680798105102,
+      "loss": 0.7528,
+      "mean_token_accuracy": 0.7719919204711914,
+      "num_tokens": 109528647.0,
+      "step": 32190
+    },
+    {
+      "epoch": 0.3231763619575254,
+      "learning_rate": 0.00013536673491509093,
+      "loss": 0.7375,
+      "mean_token_accuracy": 0.7776155114173889,
+      "num_tokens": 109562188.0,
+      "step": 32200
+    },
+    {
+      "epoch": 0.32327672728732587,
+      "learning_rate": 0.00013534666184913084,
+      "loss": 0.7821,
+      "mean_token_accuracy": 0.7668117403984069,
+      "num_tokens": 109596286.0,
+      "step": 32210
+    },
+    {
+      "epoch": 0.3233770926171263,
+      "learning_rate": 0.00013532658878317075,
+      "loss": 0.7488,
+      "mean_token_accuracy": 0.777635246515274,
+      "num_tokens": 109630329.0,
+      "step": 32220
+    },
+    {
+      "epoch": 0.32347745794692684,
+      "learning_rate": 0.00013530651571721066,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.7749003648757935,
+      "num_tokens": 109664612.0,
+      "step": 32230
+    },
+    {
+      "epoch": 0.3235778232767273,
+      "learning_rate": 0.00013528644265125057,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.7638455092906952,
+      "num_tokens": 109698578.0,
+      "step": 32240
+    },
+    {
+      "epoch": 0.32367818860652775,
+      "learning_rate": 0.00013526636958529047,
+      "loss": 0.7846,
+      "mean_token_accuracy": 0.7743920683860779,
+      "num_tokens": 109731731.0,
+      "step": 32250
+    },
+    {
+      "epoch": 0.32377855393632826,
+      "learning_rate": 0.00013524629651933036,
+      "loss": 0.7861,
+      "mean_token_accuracy": 0.7681153833866119,
+      "num_tokens": 109765201.0,
+      "step": 32260
+    },
+    {
+      "epoch": 0.3238789192661287,
+      "learning_rate": 0.0001352262234533703,
+      "loss": 0.7314,
+      "mean_token_accuracy": 0.7845546007156372,
+      "num_tokens": 109799147.0,
+      "step": 32270
+    },
+    {
+      "epoch": 0.3239792845959292,
+      "learning_rate": 0.00013520615038741017,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7764472961425781,
+      "num_tokens": 109832615.0,
+      "step": 32280
+    },
+    {
+      "epoch": 0.32407964992572963,
+      "learning_rate": 0.00013518607732145008,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.77084321975708,
+      "num_tokens": 109867074.0,
+      "step": 32290
+    },
+    {
+      "epoch": 0.32418001525553014,
+      "learning_rate": 0.00013516600425549,
+      "loss": 0.752,
+      "mean_token_accuracy": 0.7727360308170319,
+      "num_tokens": 109900583.0,
+      "step": 32300
+    },
+    {
+      "epoch": 0.3242803805853306,
+      "learning_rate": 0.0001351459311895299,
+      "loss": 0.7216,
+      "mean_token_accuracy": 0.7850228488445282,
+      "num_tokens": 109934782.0,
+      "step": 32310
+    },
+    {
+      "epoch": 0.32438074591513105,
+      "learning_rate": 0.00013512585812356978,
+      "loss": 0.7754,
+      "mean_token_accuracy": 0.7770364701747894,
+      "num_tokens": 109968322.0,
+      "step": 32320
+    },
+    {
+      "epoch": 0.32448111124493156,
+      "learning_rate": 0.00013510578505760971,
+      "loss": 0.7411,
+      "mean_token_accuracy": 0.7774547398090362,
+      "num_tokens": 110002369.0,
+      "step": 32330
+    },
+    {
+      "epoch": 0.324581476574732,
+      "learning_rate": 0.00013508571199164962,
+      "loss": 0.7426,
+      "mean_token_accuracy": 0.7777173161506653,
+      "num_tokens": 110036212.0,
+      "step": 32340
+    },
+    {
+      "epoch": 0.3246818419045325,
+      "learning_rate": 0.0001350656389256895,
+      "loss": 0.748,
+      "mean_token_accuracy": 0.7730215549468994,
+      "num_tokens": 110070000.0,
+      "step": 32350
+    },
+    {
+      "epoch": 0.324782207234333,
+      "learning_rate": 0.00013504556585972944,
+      "loss": 0.7533,
+      "mean_token_accuracy": 0.7774468004703522,
+      "num_tokens": 110103539.0,
+      "step": 32360
+    },
+    {
+      "epoch": 0.32488257256413344,
+      "learning_rate": 0.00013502549279376932,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.7731855750083924,
+      "num_tokens": 110136345.0,
+      "step": 32370
+    },
+    {
+      "epoch": 0.3249829378939339,
+      "learning_rate": 0.00013500541972780923,
+      "loss": 0.7251,
+      "mean_token_accuracy": 0.7809814929962158,
+      "num_tokens": 110170090.0,
+      "step": 32380
+    },
+    {
+      "epoch": 0.3250833032237344,
+      "learning_rate": 0.00013498534666184914,
+      "loss": 0.7673,
+      "mean_token_accuracy": 0.7671895265579224,
+      "num_tokens": 110204187.0,
+      "step": 32390
+    },
+    {
+      "epoch": 0.32518366855353487,
+      "learning_rate": 0.00013496527359588905,
+      "loss": 0.779,
+      "mean_token_accuracy": 0.7657867550849915,
+      "num_tokens": 110237628.0,
+      "step": 32400
+    },
+    {
+      "epoch": 0.3252840338833353,
+      "learning_rate": 0.00013494520052992893,
+      "loss": 0.7715,
+      "mean_token_accuracy": 0.7649196982383728,
+      "num_tokens": 110271972.0,
+      "step": 32410
+    },
+    {
+      "epoch": 0.32538439921313583,
+      "learning_rate": 0.00013492512746396886,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7747393310070038,
+      "num_tokens": 110305474.0,
+      "step": 32420
+    },
+    {
+      "epoch": 0.3254847645429363,
+      "learning_rate": 0.00013490505439800875,
+      "loss": 0.7726,
+      "mean_token_accuracy": 0.7721804261207581,
+      "num_tokens": 110338273.0,
+      "step": 32430
+    },
+    {
+      "epoch": 0.32558512987273674,
+      "learning_rate": 0.00013488498133204865,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.7693785786628723,
+      "num_tokens": 110371668.0,
+      "step": 32440
+    },
+    {
+      "epoch": 0.32568549520253726,
+      "learning_rate": 0.00013486490826608856,
+      "loss": 0.733,
+      "mean_token_accuracy": 0.7804516375064849,
+      "num_tokens": 110405971.0,
+      "step": 32450
+    },
+    {
+      "epoch": 0.3257858605323377,
+      "learning_rate": 0.00013484483520012847,
+      "loss": 0.7488,
+      "mean_token_accuracy": 0.7758582949638366,
+      "num_tokens": 110439381.0,
+      "step": 32460
+    },
+    {
+      "epoch": 0.32588622586213817,
+      "learning_rate": 0.00013482476213416838,
+      "loss": 0.7303,
+      "mean_token_accuracy": 0.7835233032703399,
+      "num_tokens": 110473526.0,
+      "step": 32470
+    },
+    {
+      "epoch": 0.3259865911919387,
+      "learning_rate": 0.0001348046890682083,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.7742891490459443,
+      "num_tokens": 110507850.0,
+      "step": 32480
+    },
+    {
+      "epoch": 0.32608695652173914,
+      "learning_rate": 0.0001347846160022482,
+      "loss": 0.7535,
+      "mean_token_accuracy": 0.7720338225364685,
+      "num_tokens": 110541759.0,
+      "step": 32490
+    },
+    {
+      "epoch": 0.3261873218515396,
+      "learning_rate": 0.00013476454293628808,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7763753592967987,
+      "num_tokens": 110576055.0,
+      "step": 32500
+    },
+    {
+      "epoch": 0.3262876871813401,
+      "learning_rate": 0.000134744469870328,
+      "loss": 0.7564,
+      "mean_token_accuracy": 0.7785202860832214,
+      "num_tokens": 110610513.0,
+      "step": 32510
+    },
+    {
+      "epoch": 0.32638805251114056,
+      "learning_rate": 0.0001347243968043679,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.779188472032547,
+      "num_tokens": 110644185.0,
+      "step": 32520
+    },
+    {
+      "epoch": 0.326488417840941,
+      "learning_rate": 0.0001347043237384078,
+      "loss": 0.761,
+      "mean_token_accuracy": 0.7707403719425201,
+      "num_tokens": 110678583.0,
+      "step": 32530
+    },
+    {
+      "epoch": 0.3265887831707415,
+      "learning_rate": 0.0001346842506724477,
+      "loss": 0.8044,
+      "mean_token_accuracy": 0.7612372517585755,
+      "num_tokens": 110712884.0,
+      "step": 32540
+    },
+    {
+      "epoch": 0.326689148500542,
+      "learning_rate": 0.00013466417760648762,
+      "loss": 0.7689,
+      "mean_token_accuracy": 0.7703858196735383,
+      "num_tokens": 110747062.0,
+      "step": 32550
+    },
+    {
+      "epoch": 0.32678951383034244,
+      "learning_rate": 0.00013464410454052753,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7742425918579101,
+      "num_tokens": 110780826.0,
+      "step": 32560
+    },
+    {
+      "epoch": 0.3268898791601429,
+      "learning_rate": 0.00013462403147456744,
+      "loss": 0.7633,
+      "mean_token_accuracy": 0.7720452129840851,
+      "num_tokens": 110814562.0,
+      "step": 32570
+    },
+    {
+      "epoch": 0.3269902444899434,
+      "learning_rate": 0.00013460395840860734,
+      "loss": 0.7455,
+      "mean_token_accuracy": 0.7741398751735687,
+      "num_tokens": 110848154.0,
+      "step": 32580
+    },
+    {
+      "epoch": 0.32709060981974386,
+      "learning_rate": 0.00013458388534264725,
+      "loss": 0.7573,
+      "mean_token_accuracy": 0.7721172392368316,
+      "num_tokens": 110881459.0,
+      "step": 32590
+    },
+    {
+      "epoch": 0.3271909751495443,
+      "learning_rate": 0.00013456381227668716,
+      "loss": 0.7939,
+      "mean_token_accuracy": 0.7620940685272217,
+      "num_tokens": 110914397.0,
+      "step": 32600
+    },
+    {
+      "epoch": 0.32729134047934483,
+      "learning_rate": 0.00013454373921072704,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7706295728683472,
+      "num_tokens": 110947918.0,
+      "step": 32610
+    },
+    {
+      "epoch": 0.3273917058091453,
+      "learning_rate": 0.00013452366614476698,
+      "loss": 0.7222,
+      "mean_token_accuracy": 0.781663328409195,
+      "num_tokens": 110981979.0,
+      "step": 32620
+    },
+    {
+      "epoch": 0.32749207113894574,
+      "learning_rate": 0.00013450359307880686,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.7725927770137787,
+      "num_tokens": 111015820.0,
+      "step": 32630
+    },
+    {
+      "epoch": 0.32759243646874625,
+      "learning_rate": 0.00013448352001284677,
+      "loss": 0.7562,
+      "mean_token_accuracy": 0.7748718917369842,
+      "num_tokens": 111049692.0,
+      "step": 32640
+    },
+    {
+      "epoch": 0.3276928017985467,
+      "learning_rate": 0.00013446344694688668,
+      "loss": 0.7309,
+      "mean_token_accuracy": 0.7789419114589691,
+      "num_tokens": 111083911.0,
+      "step": 32650
+    },
+    {
+      "epoch": 0.32779316712834716,
+      "learning_rate": 0.00013444337388092658,
+      "loss": 0.7848,
+      "mean_token_accuracy": 0.7680621981620789,
+      "num_tokens": 111117609.0,
+      "step": 32660
+    },
+    {
+      "epoch": 0.3278935324581477,
+      "learning_rate": 0.00013442330081496647,
+      "loss": 0.7443,
+      "mean_token_accuracy": 0.7761209666728973,
+      "num_tokens": 111151150.0,
+      "step": 32670
+    },
+    {
+      "epoch": 0.32799389778794813,
+      "learning_rate": 0.0001344032277490064,
+      "loss": 0.7367,
+      "mean_token_accuracy": 0.7787201941013336,
+      "num_tokens": 111185526.0,
+      "step": 32680
+    },
+    {
+      "epoch": 0.3280942631177486,
+      "learning_rate": 0.0001343831546830463,
+      "loss": 0.7413,
+      "mean_token_accuracy": 0.7789070963859558,
+      "num_tokens": 111219612.0,
+      "step": 32690
+    },
+    {
+      "epoch": 0.3281946284475491,
+      "learning_rate": 0.0001343630816170862,
+      "loss": 0.7764,
+      "mean_token_accuracy": 0.7700829148292542,
+      "num_tokens": 111253256.0,
+      "step": 32700
+    },
+    {
+      "epoch": 0.32829499377734955,
+      "learning_rate": 0.00013434300855112613,
+      "loss": 0.7816,
+      "mean_token_accuracy": 0.7685696482658386,
+      "num_tokens": 111287245.0,
+      "step": 32710
+    },
+    {
+      "epoch": 0.32839535910715,
+      "learning_rate": 0.000134322935485166,
+      "loss": 0.7604,
+      "mean_token_accuracy": 0.7750800907611847,
+      "num_tokens": 111321100.0,
+      "step": 32720
+    },
+    {
+      "epoch": 0.3284957244369505,
+      "learning_rate": 0.00013430286241920592,
+      "loss": 0.7915,
+      "mean_token_accuracy": 0.7680218696594239,
+      "num_tokens": 111355057.0,
+      "step": 32730
+    },
+    {
+      "epoch": 0.328596089766751,
+      "learning_rate": 0.00013428278935324583,
+      "loss": 0.7215,
+      "mean_token_accuracy": 0.7815347969532013,
+      "num_tokens": 111388575.0,
+      "step": 32740
+    },
+    {
+      "epoch": 0.32869645509655143,
+      "learning_rate": 0.00013426271628728573,
+      "loss": 0.7809,
+      "mean_token_accuracy": 0.76324343085289,
+      "num_tokens": 111422438.0,
+      "step": 32750
+    },
+    {
+      "epoch": 0.32879682042635194,
+      "learning_rate": 0.00013424264322132561,
+      "loss": 0.7351,
+      "mean_token_accuracy": 0.7771598517894744,
+      "num_tokens": 111456396.0,
+      "step": 32760
+    },
+    {
+      "epoch": 0.3288971857561524,
+      "learning_rate": 0.00013422257015536555,
+      "loss": 0.7734,
+      "mean_token_accuracy": 0.770964401960373,
+      "num_tokens": 111490311.0,
+      "step": 32770
+    },
+    {
+      "epoch": 0.32899755108595286,
+      "learning_rate": 0.00013420249708940543,
+      "loss": 0.7446,
+      "mean_token_accuracy": 0.7775906801223755,
+      "num_tokens": 111524145.0,
+      "step": 32780
+    },
+    {
+      "epoch": 0.32909791641575337,
+      "learning_rate": 0.00013418242402344534,
+      "loss": 0.7306,
+      "mean_token_accuracy": 0.7793404579162597,
+      "num_tokens": 111557884.0,
+      "step": 32790
+    },
+    {
+      "epoch": 0.3291982817455538,
+      "learning_rate": 0.00013416235095748525,
+      "loss": 0.7758,
+      "mean_token_accuracy": 0.7654808759689331,
+      "num_tokens": 111592317.0,
+      "step": 32800
+    },
+    {
+      "epoch": 0.3292986470753543,
+      "learning_rate": 0.00013414227789152516,
+      "loss": 0.7803,
+      "mean_token_accuracy": 0.7690751791000366,
+      "num_tokens": 111626167.0,
+      "step": 32810
+    },
+    {
+      "epoch": 0.3293990124051548,
+      "learning_rate": 0.00013412220482556507,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7735684156417847,
+      "num_tokens": 111660400.0,
+      "step": 32820
+    },
+    {
+      "epoch": 0.32949937773495525,
+      "learning_rate": 0.00013410213175960497,
+      "loss": 0.7404,
+      "mean_token_accuracy": 0.7744576811790467,
+      "num_tokens": 111693706.0,
+      "step": 32830
+    },
+    {
+      "epoch": 0.3295997430647557,
+      "learning_rate": 0.00013408205869364488,
+      "loss": 0.7158,
+      "mean_token_accuracy": 0.7870796263217926,
+      "num_tokens": 111728784.0,
+      "step": 32840
+    },
+    {
+      "epoch": 0.32970010839455616,
+      "learning_rate": 0.00013406198562768476,
+      "loss": 0.7483,
+      "mean_token_accuracy": 0.7753908455371856,
+      "num_tokens": 111762190.0,
+      "step": 32850
+    },
+    {
+      "epoch": 0.32980047372435667,
+      "learning_rate": 0.0001340419125617247,
+      "loss": 0.7715,
+      "mean_token_accuracy": 0.7680457413196564,
+      "num_tokens": 111795924.0,
+      "step": 32860
+    },
+    {
+      "epoch": 0.3299008390541571,
+      "learning_rate": 0.00013402183949576458,
+      "loss": 0.7497,
+      "mean_token_accuracy": 0.780243182182312,
+      "num_tokens": 111830676.0,
+      "step": 32870
+    },
+    {
+      "epoch": 0.3300012043839576,
+      "learning_rate": 0.0001340017664298045,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.7785914659500122,
+      "num_tokens": 111865386.0,
+      "step": 32880
+    },
+    {
+      "epoch": 0.3301015697137581,
+      "learning_rate": 0.0001339816933638444,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7758556783199311,
+      "num_tokens": 111900025.0,
+      "step": 32890
+    },
+    {
+      "epoch": 0.33020193504355855,
+      "learning_rate": 0.0001339616202978843,
+      "loss": 0.7167,
+      "mean_token_accuracy": 0.7822706520557403,
+      "num_tokens": 111934397.0,
+      "step": 32900
+    },
+    {
+      "epoch": 0.330302300373359,
+      "learning_rate": 0.00013394154723192421,
+      "loss": 0.7619,
+      "mean_token_accuracy": 0.7716590762138367,
+      "num_tokens": 111968689.0,
+      "step": 32910
+    },
+    {
+      "epoch": 0.3304026657031595,
+      "learning_rate": 0.00013392147416596412,
+      "loss": 0.7765,
+      "mean_token_accuracy": 0.7701333463191986,
+      "num_tokens": 112002218.0,
+      "step": 32920
+    },
+    {
+      "epoch": 0.33050303103295997,
+      "learning_rate": 0.00013390140110000403,
+      "loss": 0.7771,
+      "mean_token_accuracy": 0.7687518417835235,
+      "num_tokens": 112036797.0,
+      "step": 32930
+    },
+    {
+      "epoch": 0.3306033963627604,
+      "learning_rate": 0.0001338813280340439,
+      "loss": 0.713,
+      "mean_token_accuracy": 0.7801484882831573,
+      "num_tokens": 112070215.0,
+      "step": 32940
+    },
+    {
+      "epoch": 0.33070376169256094,
+      "learning_rate": 0.00013386125496808385,
+      "loss": 0.7612,
+      "mean_token_accuracy": 0.7749812602996826,
+      "num_tokens": 112103638.0,
+      "step": 32950
+    },
+    {
+      "epoch": 0.3308041270223614,
+      "learning_rate": 0.00013384118190212373,
+      "loss": 0.7673,
+      "mean_token_accuracy": 0.7733631312847138,
+      "num_tokens": 112138067.0,
+      "step": 32960
+    },
+    {
+      "epoch": 0.33090449235216185,
+      "learning_rate": 0.00013382110883616364,
+      "loss": 0.7533,
+      "mean_token_accuracy": 0.7756132185459137,
+      "num_tokens": 112172079.0,
+      "step": 32970
+    },
+    {
+      "epoch": 0.33100485768196236,
+      "learning_rate": 0.00013380103577020355,
+      "loss": 0.7388,
+      "mean_token_accuracy": 0.7763786613941193,
+      "num_tokens": 112206025.0,
+      "step": 32980
+    },
+    {
+      "epoch": 0.3311052230117628,
+      "learning_rate": 0.00013378096270424345,
+      "loss": 0.7757,
+      "mean_token_accuracy": 0.7680737316608429,
+      "num_tokens": 112240165.0,
+      "step": 32990
+    },
+    {
+      "epoch": 0.3312055883415633,
+      "learning_rate": 0.00013376088963828334,
+      "loss": 0.735,
+      "mean_token_accuracy": 0.7775642335414886,
+      "num_tokens": 112275631.0,
+      "step": 33000
+    },
+    {
+      "epoch": 0.3313059536713638,
+      "learning_rate": 0.00013374081657232327,
+      "loss": 0.7499,
+      "mean_token_accuracy": 0.7708619713783265,
+      "num_tokens": 112309657.0,
+      "step": 33010
+    },
+    {
+      "epoch": 0.33140631900116424,
+      "learning_rate": 0.00013372074350636315,
+      "loss": 0.7433,
+      "mean_token_accuracy": 0.7745214879512787,
+      "num_tokens": 112343525.0,
+      "step": 33020
+    },
+    {
+      "epoch": 0.3315066843309647,
+      "learning_rate": 0.00013370067044040306,
+      "loss": 0.7469,
+      "mean_token_accuracy": 0.7738293886184693,
+      "num_tokens": 112377528.0,
+      "step": 33030
+    },
+    {
+      "epoch": 0.3316070496607652,
+      "learning_rate": 0.000133680597374443,
+      "loss": 0.7807,
+      "mean_token_accuracy": 0.7650661706924439,
+      "num_tokens": 112411905.0,
+      "step": 33040
+    },
+    {
+      "epoch": 0.33170741499056566,
+      "learning_rate": 0.00013366052430848288,
+      "loss": 0.731,
+      "mean_token_accuracy": 0.786162942647934,
+      "num_tokens": 112445630.0,
+      "step": 33050
+    },
+    {
+      "epoch": 0.3318077803203661,
+      "learning_rate": 0.0001336404512425228,
+      "loss": 0.77,
+      "mean_token_accuracy": 0.772778207063675,
+      "num_tokens": 112479626.0,
+      "step": 33060
+    },
+    {
+      "epoch": 0.33190814565016663,
+      "learning_rate": 0.0001336203781765627,
+      "loss": 0.7362,
+      "mean_token_accuracy": 0.779965078830719,
+      "num_tokens": 112514031.0,
+      "step": 33070
+    },
+    {
+      "epoch": 0.3320085109799671,
+      "learning_rate": 0.0001336003051106026,
+      "loss": 0.7623,
+      "mean_token_accuracy": 0.7643588244915008,
+      "num_tokens": 112547695.0,
+      "step": 33080
+    },
+    {
+      "epoch": 0.33210887630976754,
+      "learning_rate": 0.00013358023204464248,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.76376793384552,
+      "num_tokens": 112581138.0,
+      "step": 33090
+    },
+    {
+      "epoch": 0.33220924163956805,
+      "learning_rate": 0.00013356015897868242,
+      "loss": 0.7519,
+      "mean_token_accuracy": 0.7737384855747222,
+      "num_tokens": 112614594.0,
+      "step": 33100
+    },
+    {
+      "epoch": 0.3323096069693685,
+      "learning_rate": 0.0001335400859127223,
+      "loss": 0.7219,
+      "mean_token_accuracy": 0.7840776026248932,
+      "num_tokens": 112649149.0,
+      "step": 33110
+    },
+    {
+      "epoch": 0.33240997229916897,
+      "learning_rate": 0.0001335200128467622,
+      "loss": 0.7044,
+      "mean_token_accuracy": 0.785427987575531,
+      "num_tokens": 112683810.0,
+      "step": 33120
+    },
+    {
+      "epoch": 0.3325103376289694,
+      "learning_rate": 0.00013349993978080212,
+      "loss": 0.7675,
+      "mean_token_accuracy": 0.7721398591995239,
+      "num_tokens": 112718643.0,
+      "step": 33130
+    },
+    {
+      "epoch": 0.33261070295876993,
+      "learning_rate": 0.00013347986671484203,
+      "loss": 0.7445,
+      "mean_token_accuracy": 0.7793696880340576,
+      "num_tokens": 112753716.0,
+      "step": 33140
+    },
+    {
+      "epoch": 0.3327110682885704,
+      "learning_rate": 0.00013345979364888194,
+      "loss": 0.738,
+      "mean_token_accuracy": 0.7804455637931824,
+      "num_tokens": 112787918.0,
+      "step": 33150
+    },
+    {
+      "epoch": 0.33281143361837084,
+      "learning_rate": 0.00013343972058292184,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.774154680967331,
+      "num_tokens": 112821656.0,
+      "step": 33160
+    },
+    {
+      "epoch": 0.33291179894817136,
+      "learning_rate": 0.00013341964751696175,
+      "loss": 0.7533,
+      "mean_token_accuracy": 0.7727014660835266,
+      "num_tokens": 112854785.0,
+      "step": 33170
+    },
+    {
+      "epoch": 0.3330121642779718,
+      "learning_rate": 0.00013339957445100166,
+      "loss": 0.786,
+      "mean_token_accuracy": 0.7679924726486206,
+      "num_tokens": 112888695.0,
+      "step": 33180
+    },
+    {
+      "epoch": 0.33311252960777227,
+      "learning_rate": 0.00013337950138504157,
+      "loss": 0.7842,
+      "mean_token_accuracy": 0.7727905869483948,
+      "num_tokens": 112922612.0,
+      "step": 33190
+    },
+    {
+      "epoch": 0.3332128949375728,
+      "learning_rate": 0.00013335942831908145,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7724378049373627,
+      "num_tokens": 112956776.0,
+      "step": 33200
+    },
+    {
+      "epoch": 0.33331326026737323,
+      "learning_rate": 0.00013333935525312139,
+      "loss": 0.747,
+      "mean_token_accuracy": 0.7820969820022583,
+      "num_tokens": 112991034.0,
+      "step": 33210
+    },
+    {
+      "epoch": 0.3334136255971737,
+      "learning_rate": 0.00013331928218716127,
+      "loss": 0.8043,
+      "mean_token_accuracy": 0.7656574070453643,
+      "num_tokens": 113025181.0,
+      "step": 33220
+    },
+    {
+      "epoch": 0.3335139909269742,
+      "learning_rate": 0.00013329920912120118,
+      "loss": 0.7941,
+      "mean_token_accuracy": 0.7625795304775238,
+      "num_tokens": 113059321.0,
+      "step": 33230
+    },
+    {
+      "epoch": 0.33361435625677466,
+      "learning_rate": 0.00013327913605524108,
+      "loss": 0.7509,
+      "mean_token_accuracy": 0.7745759189128876,
+      "num_tokens": 113093678.0,
+      "step": 33240
+    },
+    {
+      "epoch": 0.3337147215865751,
+      "learning_rate": 0.000133259062989281,
+      "loss": 0.77,
+      "mean_token_accuracy": 0.7712205052375793,
+      "num_tokens": 113128563.0,
+      "step": 33250
+    },
+    {
+      "epoch": 0.3338150869163756,
+      "learning_rate": 0.0001332389899233209,
+      "loss": 0.7575,
+      "mean_token_accuracy": 0.7760208666324615,
+      "num_tokens": 113161900.0,
+      "step": 33260
+    },
+    {
+      "epoch": 0.3339154522461761,
+      "learning_rate": 0.0001332189168573608,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.7747590005397796,
+      "num_tokens": 113195796.0,
+      "step": 33270
+    },
+    {
+      "epoch": 0.33401581757597654,
+      "learning_rate": 0.00013319884379140072,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.7751935005187989,
+      "num_tokens": 113230217.0,
+      "step": 33280
+    },
+    {
+      "epoch": 0.33411618290577705,
+      "learning_rate": 0.0001331787707254406,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.7739264309406281,
+      "num_tokens": 113264363.0,
+      "step": 33290
+    },
+    {
+      "epoch": 0.3342165482355775,
+      "learning_rate": 0.00013315869765948053,
+      "loss": 0.7416,
+      "mean_token_accuracy": 0.777640187740326,
+      "num_tokens": 113297975.0,
+      "step": 33300
+    },
+    {
+      "epoch": 0.33431691356537796,
+      "learning_rate": 0.00013313862459352042,
+      "loss": 0.7687,
+      "mean_token_accuracy": 0.7725668013095855,
+      "num_tokens": 113331956.0,
+      "step": 33310
+    },
+    {
+      "epoch": 0.33441727889517847,
+      "learning_rate": 0.00013311855152756032,
+      "loss": 0.7507,
+      "mean_token_accuracy": 0.7736175775527954,
+      "num_tokens": 113365602.0,
+      "step": 33320
+    },
+    {
+      "epoch": 0.3345176442249789,
+      "learning_rate": 0.00013309847846160023,
+      "loss": 0.7624,
+      "mean_token_accuracy": 0.7726190626621247,
+      "num_tokens": 113400072.0,
+      "step": 33330
+    },
+    {
+      "epoch": 0.3346180095547794,
+      "learning_rate": 0.00013307840539564014,
+      "loss": 0.7673,
+      "mean_token_accuracy": 0.7765252709388732,
+      "num_tokens": 113434391.0,
+      "step": 33340
+    },
+    {
+      "epoch": 0.3347183748845799,
+      "learning_rate": 0.00013305833232968002,
+      "loss": 0.7335,
+      "mean_token_accuracy": 0.7800052285194397,
+      "num_tokens": 113468908.0,
+      "step": 33350
+    },
+    {
+      "epoch": 0.33481874021438035,
+      "learning_rate": 0.00013303825926371996,
+      "loss": 0.7814,
+      "mean_token_accuracy": 0.7701805233955383,
+      "num_tokens": 113502358.0,
+      "step": 33360
+    },
+    {
+      "epoch": 0.3349191055441808,
+      "learning_rate": 0.00013301818619775984,
+      "loss": 0.7517,
+      "mean_token_accuracy": 0.7761437237262726,
+      "num_tokens": 113536314.0,
+      "step": 33370
+    },
+    {
+      "epoch": 0.3350194708739813,
+      "learning_rate": 0.00013299811313179975,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.775888878107071,
+      "num_tokens": 113570309.0,
+      "step": 33380
+    },
+    {
+      "epoch": 0.3351198362037818,
+      "learning_rate": 0.00013297804006583968,
+      "loss": 0.7892,
+      "mean_token_accuracy": 0.765402227640152,
+      "num_tokens": 113604455.0,
+      "step": 33390
+    },
+    {
+      "epoch": 0.33522020153358223,
+      "learning_rate": 0.00013295796699987957,
+      "loss": 0.779,
+      "mean_token_accuracy": 0.7697222113609314,
+      "num_tokens": 113638269.0,
+      "step": 33400
+    },
+    {
+      "epoch": 0.33532056686338274,
+      "learning_rate": 0.00013293789393391947,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.7712249636650086,
+      "num_tokens": 113671611.0,
+      "step": 33410
+    },
+    {
+      "epoch": 0.3354209321931832,
+      "learning_rate": 0.00013291782086795938,
+      "loss": 0.7975,
+      "mean_token_accuracy": 0.7598975419998169,
+      "num_tokens": 113705370.0,
+      "step": 33420
+    },
+    {
+      "epoch": 0.33552129752298365,
+      "learning_rate": 0.0001328977478019993,
+      "loss": 0.7537,
+      "mean_token_accuracy": 0.7766621589660645,
+      "num_tokens": 113739413.0,
+      "step": 33430
+    },
+    {
+      "epoch": 0.3356216628527841,
+      "learning_rate": 0.00013287767473603917,
+      "loss": 0.7412,
+      "mean_token_accuracy": 0.7762495934963226,
+      "num_tokens": 113773122.0,
+      "step": 33440
+    },
+    {
+      "epoch": 0.3357220281825846,
+      "learning_rate": 0.0001328576016700791,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7701814830303192,
+      "num_tokens": 113806750.0,
+      "step": 33450
+    },
+    {
+      "epoch": 0.3358223935123851,
+      "learning_rate": 0.000132837528604119,
+      "loss": 0.7169,
+      "mean_token_accuracy": 0.7808758616447449,
+      "num_tokens": 113840318.0,
+      "step": 33460
+    },
+    {
+      "epoch": 0.33592275884218553,
+      "learning_rate": 0.0001328174555381589,
+      "loss": 0.7436,
+      "mean_token_accuracy": 0.7756459295749665,
+      "num_tokens": 113874432.0,
+      "step": 33470
+    },
+    {
+      "epoch": 0.33602312417198604,
+      "learning_rate": 0.0001327973824721988,
+      "loss": 0.7119,
+      "mean_token_accuracy": 0.7874850273132324,
+      "num_tokens": 113908087.0,
+      "step": 33480
+    },
+    {
+      "epoch": 0.3361234895017865,
+      "learning_rate": 0.00013277730940623871,
+      "loss": 0.7261,
+      "mean_token_accuracy": 0.7868964076042175,
+      "num_tokens": 113942239.0,
+      "step": 33490
+    },
+    {
+      "epoch": 0.33622385483158695,
+      "learning_rate": 0.00013275723634027862,
+      "loss": 0.7738,
+      "mean_token_accuracy": 0.7751582086086273,
+      "num_tokens": 113977213.0,
+      "step": 33500
+    },
+    {
+      "epoch": 0.33632422016138747,
+      "learning_rate": 0.00013273716327431853,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7670100629329681,
+      "num_tokens": 114011306.0,
+      "step": 33510
+    },
+    {
+      "epoch": 0.3364245854911879,
+      "learning_rate": 0.00013271709020835844,
+      "loss": 0.7309,
+      "mean_token_accuracy": 0.7738402605056762,
+      "num_tokens": 114044696.0,
+      "step": 33520
+    },
+    {
+      "epoch": 0.3365249508209884,
+      "learning_rate": 0.00013269701714239832,
+      "loss": 0.7194,
+      "mean_token_accuracy": 0.7780163168907166,
+      "num_tokens": 114079123.0,
+      "step": 33530
+    },
+    {
+      "epoch": 0.3366253161507889,
+      "learning_rate": 0.00013267694407643826,
+      "loss": 0.7331,
+      "mean_token_accuracy": 0.7774795651435852,
+      "num_tokens": 114112400.0,
+      "step": 33540
+    },
+    {
+      "epoch": 0.33672568148058935,
+      "learning_rate": 0.00013265687101047814,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.7680506944656372,
+      "num_tokens": 114146960.0,
+      "step": 33550
+    },
+    {
+      "epoch": 0.3368260468103898,
+      "learning_rate": 0.00013263679794451805,
+      "loss": 0.7855,
+      "mean_token_accuracy": 0.7739035665988923,
+      "num_tokens": 114181005.0,
+      "step": 33560
+    },
+    {
+      "epoch": 0.3369264121401903,
+      "learning_rate": 0.00013261672487855795,
+      "loss": 0.7608,
+      "mean_token_accuracy": 0.7679487466812134,
+      "num_tokens": 114215045.0,
+      "step": 33570
+    },
+    {
+      "epoch": 0.33702677746999077,
+      "learning_rate": 0.00013259665181259786,
+      "loss": 0.7793,
+      "mean_token_accuracy": 0.7649344563484192,
+      "num_tokens": 114249721.0,
+      "step": 33580
+    },
+    {
+      "epoch": 0.3371271427997912,
+      "learning_rate": 0.00013257657874663777,
+      "loss": 0.756,
+      "mean_token_accuracy": 0.7732738375663757,
+      "num_tokens": 114283046.0,
+      "step": 33590
+    },
+    {
+      "epoch": 0.33722750812959174,
+      "learning_rate": 0.00013255650568067768,
+      "loss": 0.7317,
+      "mean_token_accuracy": 0.7844031035900116,
+      "num_tokens": 114316849.0,
+      "step": 33600
+    },
+    {
+      "epoch": 0.3373278734593922,
+      "learning_rate": 0.0001325364326147176,
+      "loss": 0.7988,
+      "mean_token_accuracy": 0.7640057265758514,
+      "num_tokens": 114351571.0,
+      "step": 33610
+    },
+    {
+      "epoch": 0.33742823878919265,
+      "learning_rate": 0.00013251635954875747,
+      "loss": 0.7519,
+      "mean_token_accuracy": 0.7793439984321594,
+      "num_tokens": 114386001.0,
+      "step": 33620
+    },
+    {
+      "epoch": 0.33752860411899316,
+      "learning_rate": 0.0001324962864827974,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.7815295815467834,
+      "num_tokens": 114419334.0,
+      "step": 33630
+    },
+    {
+      "epoch": 0.3376289694487936,
+      "learning_rate": 0.00013247621341683729,
+      "loss": 0.7377,
+      "mean_token_accuracy": 0.7777803003787994,
+      "num_tokens": 114453671.0,
+      "step": 33640
+    },
+    {
+      "epoch": 0.33772933477859407,
+      "learning_rate": 0.0001324561403508772,
+      "loss": 0.7362,
+      "mean_token_accuracy": 0.7769293785095215,
+      "num_tokens": 114487910.0,
+      "step": 33650
+    },
+    {
+      "epoch": 0.3378297001083946,
+      "learning_rate": 0.0001324360672849171,
+      "loss": 0.773,
+      "mean_token_accuracy": 0.7708508253097535,
+      "num_tokens": 114521896.0,
+      "step": 33660
+    },
+    {
+      "epoch": 0.33793006543819504,
+      "learning_rate": 0.000132415994218957,
+      "loss": 0.7422,
+      "mean_token_accuracy": 0.7837636888027191,
+      "num_tokens": 114555562.0,
+      "step": 33670
+    },
+    {
+      "epoch": 0.3380304307679955,
+      "learning_rate": 0.0001323959211529969,
+      "loss": 0.7765,
+      "mean_token_accuracy": 0.7641663014888763,
+      "num_tokens": 114589536.0,
+      "step": 33680
+    },
+    {
+      "epoch": 0.338130796097796,
+      "learning_rate": 0.00013237584808703683,
+      "loss": 0.7351,
+      "mean_token_accuracy": 0.7833424389362336,
+      "num_tokens": 114624010.0,
+      "step": 33690
+    },
+    {
+      "epoch": 0.33823116142759646,
+      "learning_rate": 0.0001323557750210767,
+      "loss": 0.7415,
+      "mean_token_accuracy": 0.7737758994102478,
+      "num_tokens": 114659004.0,
+      "step": 33700
+    },
+    {
+      "epoch": 0.3383315267573969,
+      "learning_rate": 0.00013233570195511662,
+      "loss": 0.7369,
+      "mean_token_accuracy": 0.7746557354927063,
+      "num_tokens": 114693217.0,
+      "step": 33710
+    },
+    {
+      "epoch": 0.3384318920871974,
+      "learning_rate": 0.00013231562888915653,
+      "loss": 0.758,
+      "mean_token_accuracy": 0.7718977212905884,
+      "num_tokens": 114727453.0,
+      "step": 33720
+    },
+    {
+      "epoch": 0.3385322574169979,
+      "learning_rate": 0.00013229555582319644,
+      "loss": 0.7768,
+      "mean_token_accuracy": 0.7677544355392456,
+      "num_tokens": 114761416.0,
+      "step": 33730
+    },
+    {
+      "epoch": 0.33863262274679834,
+      "learning_rate": 0.00013227548275723637,
+      "loss": 0.7648,
+      "mean_token_accuracy": 0.7792239546775818,
+      "num_tokens": 114796186.0,
+      "step": 33740
+    },
+    {
+      "epoch": 0.3387329880765988,
+      "learning_rate": 0.00013225540969127625,
+      "loss": 0.7497,
+      "mean_token_accuracy": 0.777522736787796,
+      "num_tokens": 114830535.0,
+      "step": 33750
+    },
+    {
+      "epoch": 0.3388333534063993,
+      "learning_rate": 0.00013223533662531616,
+      "loss": 0.7905,
+      "mean_token_accuracy": 0.7705022990703583,
+      "num_tokens": 114864785.0,
+      "step": 33760
+    },
+    {
+      "epoch": 0.33893371873619976,
+      "learning_rate": 0.00013221526355935607,
+      "loss": 0.7456,
+      "mean_token_accuracy": 0.7735327541828155,
+      "num_tokens": 114899124.0,
+      "step": 33770
+    },
+    {
+      "epoch": 0.3390340840660002,
+      "learning_rate": 0.00013219519049339598,
+      "loss": 0.738,
+      "mean_token_accuracy": 0.7773309469223022,
+      "num_tokens": 114933255.0,
+      "step": 33780
+    },
+    {
+      "epoch": 0.33913444939580073,
+      "learning_rate": 0.00013217511742743586,
+      "loss": 0.7447,
+      "mean_token_accuracy": 0.7761066138744355,
+      "num_tokens": 114966444.0,
+      "step": 33790
+    },
+    {
+      "epoch": 0.3392348147256012,
+      "learning_rate": 0.0001321550443614758,
+      "loss": 0.7671,
+      "mean_token_accuracy": 0.7687490105628967,
+      "num_tokens": 115000794.0,
+      "step": 33800
+    },
+    {
+      "epoch": 0.33933518005540164,
+      "learning_rate": 0.00013213497129551568,
+      "loss": 0.8074,
+      "mean_token_accuracy": 0.7668362140655518,
+      "num_tokens": 115034685.0,
+      "step": 33810
+    },
+    {
+      "epoch": 0.33943554538520215,
+      "learning_rate": 0.00013211489822955558,
+      "loss": 0.7391,
+      "mean_token_accuracy": 0.776666647195816,
+      "num_tokens": 115068788.0,
+      "step": 33820
+    },
+    {
+      "epoch": 0.3395359107150026,
+      "learning_rate": 0.0001320948251635955,
+      "loss": 0.7724,
+      "mean_token_accuracy": 0.7716356873512268,
+      "num_tokens": 115103290.0,
+      "step": 33830
+    },
+    {
+      "epoch": 0.33963627604480306,
+      "learning_rate": 0.0001320747520976354,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.7730439364910126,
+      "num_tokens": 115137765.0,
+      "step": 33840
+    },
+    {
+      "epoch": 0.3397366413746036,
+      "learning_rate": 0.0001320546790316753,
+      "loss": 0.6979,
+      "mean_token_accuracy": 0.7836260735988617,
+      "num_tokens": 115172081.0,
+      "step": 33850
+    },
+    {
+      "epoch": 0.33983700670440403,
+      "learning_rate": 0.00013203460596571522,
+      "loss": 0.7441,
+      "mean_token_accuracy": 0.7787925243377686,
+      "num_tokens": 115206076.0,
+      "step": 33860
+    },
+    {
+      "epoch": 0.3399373720342045,
+      "learning_rate": 0.00013201453289975513,
+      "loss": 0.7535,
+      "mean_token_accuracy": 0.7702636957168579,
+      "num_tokens": 115240393.0,
+      "step": 33870
+    },
+    {
+      "epoch": 0.340037737364005,
+      "learning_rate": 0.000131994459833795,
+      "loss": 0.7475,
+      "mean_token_accuracy": 0.7734738945960998,
+      "num_tokens": 115274738.0,
+      "step": 33880
+    },
+    {
+      "epoch": 0.34013810269380546,
+      "learning_rate": 0.00013197438676783494,
+      "loss": 0.7848,
+      "mean_token_accuracy": 0.7629347026348114,
+      "num_tokens": 115309232.0,
+      "step": 33890
+    },
+    {
+      "epoch": 0.3402384680236059,
+      "learning_rate": 0.00013195431370187482,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.7716972589492798,
+      "num_tokens": 115343459.0,
+      "step": 33900
+    },
+    {
+      "epoch": 0.3403388333534064,
+      "learning_rate": 0.00013193424063591473,
+      "loss": 0.7236,
+      "mean_token_accuracy": 0.7831964731216431,
+      "num_tokens": 115376601.0,
+      "step": 33910
+    },
+    {
+      "epoch": 0.3404391986832069,
+      "learning_rate": 0.00013191416756995464,
+      "loss": 0.7618,
+      "mean_token_accuracy": 0.7727328360080719,
+      "num_tokens": 115411652.0,
+      "step": 33920
+    },
+    {
+      "epoch": 0.34053956401300733,
+      "learning_rate": 0.00013189409450399455,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.7645107567310333,
+      "num_tokens": 115445195.0,
+      "step": 33930
+    },
+    {
+      "epoch": 0.34063992934280785,
+      "learning_rate": 0.00013187402143803446,
+      "loss": 0.7518,
+      "mean_token_accuracy": 0.776811945438385,
+      "num_tokens": 115480020.0,
+      "step": 33940
+    },
+    {
+      "epoch": 0.3407402946726083,
+      "learning_rate": 0.00013185394837207437,
+      "loss": 0.7629,
+      "mean_token_accuracy": 0.7730513215065002,
+      "num_tokens": 115513375.0,
+      "step": 33950
+    },
+    {
+      "epoch": 0.34084066000240876,
+      "learning_rate": 0.00013183387530611427,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.7708856523036957,
+      "num_tokens": 115547440.0,
+      "step": 33960
+    },
+    {
+      "epoch": 0.34094102533220927,
+      "learning_rate": 0.00013181380224015416,
+      "loss": 0.7721,
+      "mean_token_accuracy": 0.7715115666389465,
+      "num_tokens": 115581768.0,
+      "step": 33970
+    },
+    {
+      "epoch": 0.3410413906620097,
+      "learning_rate": 0.0001317937291741941,
+      "loss": 0.7532,
+      "mean_token_accuracy": 0.7767409980297089,
+      "num_tokens": 115615429.0,
+      "step": 33980
+    },
+    {
+      "epoch": 0.3411417559918102,
+      "learning_rate": 0.00013177365610823397,
+      "loss": 0.77,
+      "mean_token_accuracy": 0.7690534293651581,
+      "num_tokens": 115649748.0,
+      "step": 33990
+    },
+    {
+      "epoch": 0.34124212132161064,
+      "learning_rate": 0.00013175358304227388,
+      "loss": 0.7296,
+      "mean_token_accuracy": 0.7769282817840576,
+      "num_tokens": 115683651.0,
+      "step": 34000
+    },
+    {
+      "epoch": 0.34134248665141115,
+      "learning_rate": 0.0001317335099763138,
+      "loss": 0.7216,
+      "mean_token_accuracy": 0.7827814757823944,
+      "num_tokens": 115716730.0,
+      "step": 34010
+    },
+    {
+      "epoch": 0.3414428519812116,
+      "learning_rate": 0.0001317134369103537,
+      "loss": 0.7357,
+      "mean_token_accuracy": 0.7795480132102967,
+      "num_tokens": 115751124.0,
+      "step": 34020
+    },
+    {
+      "epoch": 0.34154321731101206,
+      "learning_rate": 0.00013169336384439358,
+      "loss": 0.7566,
+      "mean_token_accuracy": 0.7710654735565186,
+      "num_tokens": 115785214.0,
+      "step": 34030
+    },
+    {
+      "epoch": 0.34164358264081257,
+      "learning_rate": 0.00013167329077843352,
+      "loss": 0.7452,
+      "mean_token_accuracy": 0.7726763129234314,
+      "num_tokens": 115820126.0,
+      "step": 34040
+    },
+    {
+      "epoch": 0.341743947970613,
+      "learning_rate": 0.0001316532177124734,
+      "loss": 0.7341,
+      "mean_token_accuracy": 0.7802583396434783,
+      "num_tokens": 115853698.0,
+      "step": 34050
+    },
+    {
+      "epoch": 0.3418443133004135,
+      "learning_rate": 0.0001316331446465133,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.7817862033843994,
+      "num_tokens": 115888144.0,
+      "step": 34060
+    },
+    {
+      "epoch": 0.341944678630214,
+      "learning_rate": 0.0001316130715805532,
+      "loss": 0.766,
+      "mean_token_accuracy": 0.770089042186737,
+      "num_tokens": 115921786.0,
+      "step": 34070
+    },
+    {
+      "epoch": 0.34204504396001445,
+      "learning_rate": 0.00013159299851459312,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7753535389900208,
+      "num_tokens": 115955388.0,
+      "step": 34080
+    },
+    {
+      "epoch": 0.3421454092898149,
+      "learning_rate": 0.00013157292544863303,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7741257071495056,
+      "num_tokens": 115989914.0,
+      "step": 34090
+    },
+    {
+      "epoch": 0.3422457746196154,
+      "learning_rate": 0.00013155285238267294,
+      "loss": 0.7196,
+      "mean_token_accuracy": 0.7788474857807159,
+      "num_tokens": 116024304.0,
+      "step": 34100
+    },
+    {
+      "epoch": 0.3423461399494159,
+      "learning_rate": 0.00013153277931671285,
+      "loss": 0.7428,
+      "mean_token_accuracy": 0.7727693736553192,
+      "num_tokens": 116058515.0,
+      "step": 34110
+    },
+    {
+      "epoch": 0.34244650527921633,
+      "learning_rate": 0.00013151270625075273,
+      "loss": 0.7728,
+      "mean_token_accuracy": 0.770117312669754,
+      "num_tokens": 116092672.0,
+      "step": 34120
+    },
+    {
+      "epoch": 0.34254687060901684,
+      "learning_rate": 0.00013149263318479266,
+      "loss": 0.7581,
+      "mean_token_accuracy": 0.7690131008625031,
+      "num_tokens": 116125713.0,
+      "step": 34130
+    },
+    {
+      "epoch": 0.3426472359388173,
+      "learning_rate": 0.00013147256011883255,
+      "loss": 0.7164,
+      "mean_token_accuracy": 0.7797726035118103,
+      "num_tokens": 116159272.0,
+      "step": 34140
+    },
+    {
+      "epoch": 0.34274760126861775,
+      "learning_rate": 0.00013145248705287245,
+      "loss": 0.7331,
+      "mean_token_accuracy": 0.7755063772201538,
+      "num_tokens": 116193734.0,
+      "step": 34150
+    },
+    {
+      "epoch": 0.34284796659841826,
+      "learning_rate": 0.00013143241398691236,
+      "loss": 0.7498,
+      "mean_token_accuracy": 0.7762515246868134,
+      "num_tokens": 116227360.0,
+      "step": 34160
+    },
+    {
+      "epoch": 0.3429483319282187,
+      "learning_rate": 0.00013141234092095227,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.7699919462203979,
+      "num_tokens": 116261561.0,
+      "step": 34170
+    },
+    {
+      "epoch": 0.3430486972580192,
+      "learning_rate": 0.00013139226785499218,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.7696660220623016,
+      "num_tokens": 116295154.0,
+      "step": 34180
+    },
+    {
+      "epoch": 0.3431490625878197,
+      "learning_rate": 0.0001313721947890321,
+      "loss": 0.7273,
+      "mean_token_accuracy": 0.7759283125400543,
+      "num_tokens": 116329446.0,
+      "step": 34190
+    },
+    {
+      "epoch": 0.34324942791762014,
+      "learning_rate": 0.000131352121723072,
+      "loss": 0.7253,
+      "mean_token_accuracy": 0.780819696187973,
+      "num_tokens": 116362873.0,
+      "step": 34200
+    },
+    {
+      "epoch": 0.3433497932474206,
+      "learning_rate": 0.00013133204865711188,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7736681044101715,
+      "num_tokens": 116397957.0,
+      "step": 34210
+    },
+    {
+      "epoch": 0.3434501585772211,
+      "learning_rate": 0.0001313119755911518,
+      "loss": 0.7398,
+      "mean_token_accuracy": 0.776227742433548,
+      "num_tokens": 116431767.0,
+      "step": 34220
+    },
+    {
+      "epoch": 0.34355052390702157,
+      "learning_rate": 0.0001312919025251917,
+      "loss": 0.7586,
+      "mean_token_accuracy": 0.7764150321483612,
+      "num_tokens": 116465353.0,
+      "step": 34230
+    },
+    {
+      "epoch": 0.343650889236822,
+      "learning_rate": 0.0001312718294592316,
+      "loss": 0.7457,
+      "mean_token_accuracy": 0.778573626279831,
+      "num_tokens": 116499430.0,
+      "step": 34240
+    },
+    {
+      "epoch": 0.34375125456662253,
+      "learning_rate": 0.0001312517563932715,
+      "loss": 0.7607,
+      "mean_token_accuracy": 0.7732001185417176,
+      "num_tokens": 116533055.0,
+      "step": 34250
+    },
+    {
+      "epoch": 0.343851619896423,
+      "learning_rate": 0.00013123168332731142,
+      "loss": 0.7104,
+      "mean_token_accuracy": 0.7862925827503204,
+      "num_tokens": 116566761.0,
+      "step": 34260
+    },
+    {
+      "epoch": 0.34395198522622344,
+      "learning_rate": 0.0001312116102613513,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.7757634878158569,
+      "num_tokens": 116601230.0,
+      "step": 34270
+    },
+    {
+      "epoch": 0.3440523505560239,
+      "learning_rate": 0.00013119153719539124,
+      "loss": 0.7143,
+      "mean_token_accuracy": 0.7833714783191681,
+      "num_tokens": 116635485.0,
+      "step": 34280
+    },
+    {
+      "epoch": 0.3441527158858244,
+      "learning_rate": 0.00013117146412943114,
+      "loss": 0.7714,
+      "mean_token_accuracy": 0.7680694222450256,
+      "num_tokens": 116669000.0,
+      "step": 34290
+    },
+    {
+      "epoch": 0.34425308121562487,
+      "learning_rate": 0.00013115139106347105,
+      "loss": 0.7681,
+      "mean_token_accuracy": 0.7697263181209564,
+      "num_tokens": 116702279.0,
+      "step": 34300
+    },
+    {
+      "epoch": 0.3443534465454253,
+      "learning_rate": 0.00013113131799751096,
+      "loss": 0.7651,
+      "mean_token_accuracy": 0.7704175889492035,
+      "num_tokens": 116737439.0,
+      "step": 34310
+    },
+    {
+      "epoch": 0.34445381187522583,
+      "learning_rate": 0.00013111124493155084,
+      "loss": 0.7408,
+      "mean_token_accuracy": 0.772112762928009,
+      "num_tokens": 116771416.0,
+      "step": 34320
+    },
+    {
+      "epoch": 0.3445541772050263,
+      "learning_rate": 0.00013109117186559078,
+      "loss": 0.7741,
+      "mean_token_accuracy": 0.7686093211174011,
+      "num_tokens": 116805417.0,
+      "step": 34330
+    },
+    {
+      "epoch": 0.34465454253482675,
+      "learning_rate": 0.00013107109879963066,
+      "loss": 0.7347,
+      "mean_token_accuracy": 0.7795671701431275,
+      "num_tokens": 116839763.0,
+      "step": 34340
+    },
+    {
+      "epoch": 0.34475490786462726,
+      "learning_rate": 0.00013105102573367057,
+      "loss": 0.7435,
+      "mean_token_accuracy": 0.7816221535205841,
+      "num_tokens": 116873743.0,
+      "step": 34350
+    },
+    {
+      "epoch": 0.3448552731944277,
+      "learning_rate": 0.00013103095266771048,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7747231245040893,
+      "num_tokens": 116907959.0,
+      "step": 34360
+    },
+    {
+      "epoch": 0.34495563852422817,
+      "learning_rate": 0.00013101087960175039,
+      "loss": 0.776,
+      "mean_token_accuracy": 0.7699971199035645,
+      "num_tokens": 116943038.0,
+      "step": 34370
+    },
+    {
+      "epoch": 0.3450560038540287,
+      "learning_rate": 0.00013099080653579027,
+      "loss": 0.7639,
+      "mean_token_accuracy": 0.771040403842926,
+      "num_tokens": 116977229.0,
+      "step": 34380
+    },
+    {
+      "epoch": 0.34515636918382914,
+      "learning_rate": 0.0001309707334698302,
+      "loss": 0.7505,
+      "mean_token_accuracy": 0.776633882522583,
+      "num_tokens": 117010447.0,
+      "step": 34390
+    },
+    {
+      "epoch": 0.3452567345136296,
+      "learning_rate": 0.00013095066040387008,
+      "loss": 0.7568,
+      "mean_token_accuracy": 0.7718197226524353,
+      "num_tokens": 117044615.0,
+      "step": 34400
+    },
+    {
+      "epoch": 0.3453570998434301,
+      "learning_rate": 0.00013093058733791,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7776704132556915,
+      "num_tokens": 117078330.0,
+      "step": 34410
+    },
+    {
+      "epoch": 0.34545746517323056,
+      "learning_rate": 0.0001309105142719499,
+      "loss": 0.7387,
+      "mean_token_accuracy": 0.7776882827281952,
+      "num_tokens": 117112490.0,
+      "step": 34420
+    },
+    {
+      "epoch": 0.345557830503031,
+      "learning_rate": 0.0001308904412059898,
+      "loss": 0.7586,
+      "mean_token_accuracy": 0.7724784135818481,
+      "num_tokens": 117146721.0,
+      "step": 34430
+    },
+    {
+      "epoch": 0.3456581958328315,
+      "learning_rate": 0.00013087036814002972,
+      "loss": 0.8169,
+      "mean_token_accuracy": 0.7559751987457275,
+      "num_tokens": 117180219.0,
+      "step": 34440
+    },
+    {
+      "epoch": 0.345758561162632,
+      "learning_rate": 0.00013085029507406963,
+      "loss": 0.7619,
+      "mean_token_accuracy": 0.7739854872226715,
+      "num_tokens": 117214541.0,
+      "step": 34450
+    },
+    {
+      "epoch": 0.34585892649243244,
+      "learning_rate": 0.00013083022200810953,
+      "loss": 0.751,
+      "mean_token_accuracy": 0.7724648296833039,
+      "num_tokens": 117248602.0,
+      "step": 34460
+    },
+    {
+      "epoch": 0.34595929182223295,
+      "learning_rate": 0.00013081014894214942,
+      "loss": 0.7364,
+      "mean_token_accuracy": 0.7811800599098205,
+      "num_tokens": 117282082.0,
+      "step": 34470
+    },
+    {
+      "epoch": 0.3460596571520334,
+      "learning_rate": 0.00013079007587618935,
+      "loss": 0.7703,
+      "mean_token_accuracy": 0.771990841627121,
+      "num_tokens": 117315963.0,
+      "step": 34480
+    },
+    {
+      "epoch": 0.34616002248183386,
+      "learning_rate": 0.00013077000281022923,
+      "loss": 0.7498,
+      "mean_token_accuracy": 0.7722388863563537,
+      "num_tokens": 117350226.0,
+      "step": 34490
+    },
+    {
+      "epoch": 0.3462603878116344,
+      "learning_rate": 0.00013074992974426914,
+      "loss": 0.7521,
+      "mean_token_accuracy": 0.7755684375762939,
+      "num_tokens": 117383725.0,
+      "step": 34500
+    },
+    {
+      "epoch": 0.34636075314143483,
+      "learning_rate": 0.00013072985667830905,
+      "loss": 0.7562,
+      "mean_token_accuracy": 0.7758636116981507,
+      "num_tokens": 117417001.0,
+      "step": 34510
+    },
+    {
+      "epoch": 0.3464611184712353,
+      "learning_rate": 0.00013070978361234896,
+      "loss": 0.7435,
+      "mean_token_accuracy": 0.7790121793746948,
+      "num_tokens": 117451132.0,
+      "step": 34520
+    },
+    {
+      "epoch": 0.3465614838010358,
+      "learning_rate": 0.00013068971054638887,
+      "loss": 0.7731,
+      "mean_token_accuracy": 0.7720594286918641,
+      "num_tokens": 117485231.0,
+      "step": 34530
+    },
+    {
+      "epoch": 0.34666184913083625,
+      "learning_rate": 0.00013066963748042877,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7742433428764344,
+      "num_tokens": 117519021.0,
+      "step": 34540
+    },
+    {
+      "epoch": 0.3467622144606367,
+      "learning_rate": 0.00013064956441446868,
+      "loss": 0.7545,
+      "mean_token_accuracy": 0.7752372026443481,
+      "num_tokens": 117553157.0,
+      "step": 34550
+    },
+    {
+      "epoch": 0.34686257979043716,
+      "learning_rate": 0.00013062949134850856,
+      "loss": 0.7887,
+      "mean_token_accuracy": 0.7634439051151276,
+      "num_tokens": 117586881.0,
+      "step": 34560
+    },
+    {
+      "epoch": 0.3469629451202377,
+      "learning_rate": 0.0001306094182825485,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.771806401014328,
+      "num_tokens": 117621407.0,
+      "step": 34570
+    },
+    {
+      "epoch": 0.34706331045003813,
+      "learning_rate": 0.00013058934521658838,
+      "loss": 0.7535,
+      "mean_token_accuracy": 0.7751614928245545,
+      "num_tokens": 117655538.0,
+      "step": 34580
+    },
+    {
+      "epoch": 0.3471636757798386,
+      "learning_rate": 0.0001305692721506283,
+      "loss": 0.7037,
+      "mean_token_accuracy": 0.7887448370456696,
+      "num_tokens": 117689169.0,
+      "step": 34590
+    },
+    {
+      "epoch": 0.3472640411096391,
+      "learning_rate": 0.0001305491990846682,
+      "loss": 0.7226,
+      "mean_token_accuracy": 0.7777604997158051,
+      "num_tokens": 117723065.0,
+      "step": 34600
+    },
+    {
+      "epoch": 0.34736440643943955,
+      "learning_rate": 0.0001305291260187081,
+      "loss": 0.7218,
+      "mean_token_accuracy": 0.7842798173427582,
+      "num_tokens": 117757752.0,
+      "step": 34610
+    },
+    {
+      "epoch": 0.34746477176924,
+      "learning_rate": 0.000130509052952748,
+      "loss": 0.7401,
+      "mean_token_accuracy": 0.7753164947032929,
+      "num_tokens": 117791530.0,
+      "step": 34620
+    },
+    {
+      "epoch": 0.3475651370990405,
+      "learning_rate": 0.00013048897988678792,
+      "loss": 0.7484,
+      "mean_token_accuracy": 0.7727838516235351,
+      "num_tokens": 117825861.0,
+      "step": 34630
+    },
+    {
+      "epoch": 0.347665502428841,
+      "learning_rate": 0.00013046890682082783,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.7652743220329284,
+      "num_tokens": 117859863.0,
+      "step": 34640
+    },
+    {
+      "epoch": 0.34776586775864143,
+      "learning_rate": 0.0001304488337548677,
+      "loss": 0.7521,
+      "mean_token_accuracy": 0.778465723991394,
+      "num_tokens": 117893913.0,
+      "step": 34650
+    },
+    {
+      "epoch": 0.34786623308844195,
+      "learning_rate": 0.00013042876068890765,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.7738387942314148,
+      "num_tokens": 117927990.0,
+      "step": 34660
+    },
+    {
+      "epoch": 0.3479665984182424,
+      "learning_rate": 0.00013040868762294753,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7747078657150268,
+      "num_tokens": 117962835.0,
+      "step": 34670
+    },
+    {
+      "epoch": 0.34806696374804286,
+      "learning_rate": 0.00013038861455698744,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7754677116870881,
+      "num_tokens": 117996774.0,
+      "step": 34680
+    },
+    {
+      "epoch": 0.34816732907784337,
+      "learning_rate": 0.00013036854149102735,
+      "loss": 0.7348,
+      "mean_token_accuracy": 0.7804238975048066,
+      "num_tokens": 118031436.0,
+      "step": 34690
+    },
+    {
+      "epoch": 0.3482676944076438,
+      "learning_rate": 0.00013034846842506726,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7717331767082214,
+      "num_tokens": 118065332.0,
+      "step": 34700
+    },
+    {
+      "epoch": 0.3483680597374443,
+      "learning_rate": 0.00013032839535910714,
+      "loss": 0.7345,
+      "mean_token_accuracy": 0.7782244682312012,
+      "num_tokens": 118099396.0,
+      "step": 34710
+    },
+    {
+      "epoch": 0.3484684250672448,
+      "learning_rate": 0.00013030832229314707,
+      "loss": 0.7393,
+      "mean_token_accuracy": 0.7732568264007569,
+      "num_tokens": 118133559.0,
+      "step": 34720
+    },
+    {
+      "epoch": 0.34856879039704525,
+      "learning_rate": 0.00013028824922718695,
+      "loss": 0.7557,
+      "mean_token_accuracy": 0.7746400594711303,
+      "num_tokens": 118167606.0,
+      "step": 34730
+    },
+    {
+      "epoch": 0.3486691557268457,
+      "learning_rate": 0.00013026817616122686,
+      "loss": 0.7452,
+      "mean_token_accuracy": 0.7743682622909546,
+      "num_tokens": 118201977.0,
+      "step": 34740
+    },
+    {
+      "epoch": 0.3487695210566462,
+      "learning_rate": 0.00013024810309526677,
+      "loss": 0.7286,
+      "mean_token_accuracy": 0.7818284869194031,
+      "num_tokens": 118236029.0,
+      "step": 34750
+    },
+    {
+      "epoch": 0.34886988638644667,
+      "learning_rate": 0.00013022803002930668,
+      "loss": 0.7231,
+      "mean_token_accuracy": 0.7850748658180237,
+      "num_tokens": 118269975.0,
+      "step": 34760
+    },
+    {
+      "epoch": 0.3489702517162471,
+      "learning_rate": 0.0001302079569633466,
+      "loss": 0.7516,
+      "mean_token_accuracy": 0.7747095584869385,
+      "num_tokens": 118304176.0,
+      "step": 34770
+    },
+    {
+      "epoch": 0.34907061704604764,
+      "learning_rate": 0.0001301878838973865,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7725032269954681,
+      "num_tokens": 118338089.0,
+      "step": 34780
+    },
+    {
+      "epoch": 0.3491709823758481,
+      "learning_rate": 0.0001301678108314264,
+      "loss": 0.7505,
+      "mean_token_accuracy": 0.7774797141551971,
+      "num_tokens": 118371871.0,
+      "step": 34790
+    },
+    {
+      "epoch": 0.34927134770564855,
+      "learning_rate": 0.00013014773776546629,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.7733864784240723,
+      "num_tokens": 118406554.0,
+      "step": 34800
+    },
+    {
+      "epoch": 0.34937171303544906,
+      "learning_rate": 0.00013012766469950622,
+      "loss": 0.7157,
+      "mean_token_accuracy": 0.7768620789051056,
+      "num_tokens": 118440454.0,
+      "step": 34810
+    },
+    {
+      "epoch": 0.3494720783652495,
+      "learning_rate": 0.0001301075916335461,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7789655983448028,
+      "num_tokens": 118473740.0,
+      "step": 34820
+    },
+    {
+      "epoch": 0.34957244369505,
+      "learning_rate": 0.000130087518567586,
+      "loss": 0.7277,
+      "mean_token_accuracy": 0.7804815113544464,
+      "num_tokens": 118507591.0,
+      "step": 34830
+    },
+    {
+      "epoch": 0.34967280902485043,
+      "learning_rate": 0.00013006744550162592,
+      "loss": 0.7067,
+      "mean_token_accuracy": 0.7835338473320007,
+      "num_tokens": 118542423.0,
+      "step": 34840
+    },
+    {
+      "epoch": 0.34977317435465094,
+      "learning_rate": 0.00013004737243566583,
+      "loss": 0.7683,
+      "mean_token_accuracy": 0.7751769661903382,
+      "num_tokens": 118576057.0,
+      "step": 34850
+    },
+    {
+      "epoch": 0.3498735396844514,
+      "learning_rate": 0.00013002729936970574,
+      "loss": 0.7503,
+      "mean_token_accuracy": 0.7780750274658204,
+      "num_tokens": 118610283.0,
+      "step": 34860
+    },
+    {
+      "epoch": 0.34997390501425185,
+      "learning_rate": 0.00013000722630374564,
+      "loss": 0.7419,
+      "mean_token_accuracy": 0.7775280714035034,
+      "num_tokens": 118645136.0,
+      "step": 34870
+    },
+    {
+      "epoch": 0.35007427034405236,
+      "learning_rate": 0.00012998715323778555,
+      "loss": 0.7446,
+      "mean_token_accuracy": 0.7766178011894226,
+      "num_tokens": 118679708.0,
+      "step": 34880
+    },
+    {
+      "epoch": 0.3501746356738528,
+      "learning_rate": 0.00012996708017182546,
+      "loss": 0.745,
+      "mean_token_accuracy": 0.7750972747802735,
+      "num_tokens": 118713556.0,
+      "step": 34890
+    },
+    {
+      "epoch": 0.3502750010036533,
+      "learning_rate": 0.00012994700710586537,
+      "loss": 0.7508,
+      "mean_token_accuracy": 0.7772462069988251,
+      "num_tokens": 118747915.0,
+      "step": 34900
+    },
+    {
+      "epoch": 0.3503753663334538,
+      "learning_rate": 0.00012992693403990525,
+      "loss": 0.7364,
+      "mean_token_accuracy": 0.7773713529109955,
+      "num_tokens": 118780943.0,
+      "step": 34910
+    },
+    {
+      "epoch": 0.35047573166325424,
+      "learning_rate": 0.0001299068609739452,
+      "loss": 0.7935,
+      "mean_token_accuracy": 0.7650506019592285,
+      "num_tokens": 118815612.0,
+      "step": 34920
+    },
+    {
+      "epoch": 0.3505760969930547,
+      "learning_rate": 0.00012988678790798507,
+      "loss": 0.7442,
+      "mean_token_accuracy": 0.7729040741920471,
+      "num_tokens": 118849356.0,
+      "step": 34930
+    },
+    {
+      "epoch": 0.3506764623228552,
+      "learning_rate": 0.00012986671484202498,
+      "loss": 0.7363,
+      "mean_token_accuracy": 0.7809243083000184,
+      "num_tokens": 118883002.0,
+      "step": 34940
+    },
+    {
+      "epoch": 0.35077682765265567,
+      "learning_rate": 0.00012984664177606488,
+      "loss": 0.7398,
+      "mean_token_accuracy": 0.7733363807201385,
+      "num_tokens": 118916859.0,
+      "step": 34950
+    },
+    {
+      "epoch": 0.3508771929824561,
+      "learning_rate": 0.0001298265687101048,
+      "loss": 0.7857,
+      "mean_token_accuracy": 0.7694664359092712,
+      "num_tokens": 118949953.0,
+      "step": 34960
+    },
+    {
+      "epoch": 0.35097755831225663,
+      "learning_rate": 0.00012980649564414467,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7718825221061707,
+      "num_tokens": 118983543.0,
+      "step": 34970
+    },
+    {
+      "epoch": 0.3510779236420571,
+      "learning_rate": 0.0001297864225781846,
+      "loss": 0.7233,
+      "mean_token_accuracy": 0.7823235094547272,
+      "num_tokens": 119017302.0,
+      "step": 34980
+    },
+    {
+      "epoch": 0.35117828897185754,
+      "learning_rate": 0.00012976634951222452,
+      "loss": 0.7734,
+      "mean_token_accuracy": 0.766890674829483,
+      "num_tokens": 119051990.0,
+      "step": 34990
+    },
+    {
+      "epoch": 0.35127865430165806,
+      "learning_rate": 0.0001297462764462644,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7784459352493286,
+      "num_tokens": 119086337.0,
+      "step": 35000
+    },
+    {
+      "epoch": 0.3513790196314585,
+      "learning_rate": 0.00012972620338030434,
+      "loss": 0.7317,
+      "mean_token_accuracy": 0.7820458710193634,
+      "num_tokens": 119120358.0,
+      "step": 35010
+    },
+    {
+      "epoch": 0.35147938496125897,
+      "learning_rate": 0.00012970613031434422,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.7753281235694885,
+      "num_tokens": 119154366.0,
+      "step": 35020
+    },
+    {
+      "epoch": 0.3515797502910595,
+      "learning_rate": 0.00012968605724838413,
+      "loss": 0.7549,
+      "mean_token_accuracy": 0.7730974972248077,
+      "num_tokens": 119189498.0,
+      "step": 35030
+    },
+    {
+      "epoch": 0.35168011562085993,
+      "learning_rate": 0.00012966598418242403,
+      "loss": 0.7693,
+      "mean_token_accuracy": 0.7727134704589844,
+      "num_tokens": 119223230.0,
+      "step": 35040
+    },
+    {
+      "epoch": 0.3517804809506604,
+      "learning_rate": 0.00012964591111646394,
+      "loss": 0.7698,
+      "mean_token_accuracy": 0.7701881527900696,
+      "num_tokens": 119256910.0,
+      "step": 35050
+    },
+    {
+      "epoch": 0.3518808462804609,
+      "learning_rate": 0.00012962583805050382,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7781509280204773,
+      "num_tokens": 119290293.0,
+      "step": 35060
+    },
+    {
+      "epoch": 0.35198121161026136,
+      "learning_rate": 0.00012960576498454376,
+      "loss": 0.7561,
+      "mean_token_accuracy": 0.772142744064331,
+      "num_tokens": 119323527.0,
+      "step": 35070
+    },
+    {
+      "epoch": 0.3520815769400618,
+      "learning_rate": 0.00012958569191858364,
+      "loss": 0.7348,
+      "mean_token_accuracy": 0.776472669839859,
+      "num_tokens": 119357956.0,
+      "step": 35080
+    },
+    {
+      "epoch": 0.3521819422698623,
+      "learning_rate": 0.00012956561885262355,
+      "loss": 0.7893,
+      "mean_token_accuracy": 0.7657097697257995,
+      "num_tokens": 119391878.0,
+      "step": 35090
+    },
+    {
+      "epoch": 0.3522823075996628,
+      "learning_rate": 0.00012954554578666346,
+      "loss": 0.7827,
+      "mean_token_accuracy": 0.7680422246456147,
+      "num_tokens": 119425021.0,
+      "step": 35100
+    },
+    {
+      "epoch": 0.35238267292946324,
+      "learning_rate": 0.00012952547272070337,
+      "loss": 0.7347,
+      "mean_token_accuracy": 0.7755476057529449,
+      "num_tokens": 119458978.0,
+      "step": 35110
+    },
+    {
+      "epoch": 0.3524830382592637,
+      "learning_rate": 0.00012950539965474327,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7770498335361481,
+      "num_tokens": 119492774.0,
+      "step": 35120
+    },
+    {
+      "epoch": 0.3525834035890642,
+      "learning_rate": 0.00012948532658878318,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7735799431800843,
+      "num_tokens": 119526989.0,
+      "step": 35130
+    },
+    {
+      "epoch": 0.35268376891886466,
+      "learning_rate": 0.0001294652535228231,
+      "loss": 0.7584,
+      "mean_token_accuracy": 0.7750583529472351,
+      "num_tokens": 119560782.0,
+      "step": 35140
+    },
+    {
+      "epoch": 0.3527841342486651,
+      "learning_rate": 0.00012944518045686297,
+      "loss": 0.7641,
+      "mean_token_accuracy": 0.7715739667415619,
+      "num_tokens": 119594978.0,
+      "step": 35150
+    },
+    {
+      "epoch": 0.3528844995784656,
+      "learning_rate": 0.0001294251073909029,
+      "loss": 0.7301,
+      "mean_token_accuracy": 0.7819527506828308,
+      "num_tokens": 119629725.0,
+      "step": 35160
+    },
+    {
+      "epoch": 0.3529848649082661,
+      "learning_rate": 0.0001294050343249428,
+      "loss": 0.7613,
+      "mean_token_accuracy": 0.7737753927707672,
+      "num_tokens": 119662809.0,
+      "step": 35170
+    },
+    {
+      "epoch": 0.35308523023806654,
+      "learning_rate": 0.0001293849612589827,
+      "loss": 0.7381,
+      "mean_token_accuracy": 0.7770018994808197,
+      "num_tokens": 119696635.0,
+      "step": 35180
+    },
+    {
+      "epoch": 0.35318559556786705,
+      "learning_rate": 0.0001293648881930226,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7756446003913879,
+      "num_tokens": 119730825.0,
+      "step": 35190
+    },
+    {
+      "epoch": 0.3532859608976675,
+      "learning_rate": 0.00012934481512706251,
+      "loss": 0.7447,
+      "mean_token_accuracy": 0.7755038917064667,
+      "num_tokens": 119765070.0,
+      "step": 35200
+    },
+    {
+      "epoch": 0.35338632622746796,
+      "learning_rate": 0.00012932474206110242,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7713152647018433,
+      "num_tokens": 119799584.0,
+      "step": 35210
+    },
+    {
+      "epoch": 0.3534866915572685,
+      "learning_rate": 0.00012930466899514233,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7718316853046417,
+      "num_tokens": 119833470.0,
+      "step": 35220
+    },
+    {
+      "epoch": 0.35358705688706893,
+      "learning_rate": 0.00012928459592918224,
+      "loss": 0.814,
+      "mean_token_accuracy": 0.7650098145008087,
+      "num_tokens": 119867558.0,
+      "step": 35230
+    },
+    {
+      "epoch": 0.3536874222168694,
+      "learning_rate": 0.00012926452286322212,
+      "loss": 0.7499,
+      "mean_token_accuracy": 0.7746418297290802,
+      "num_tokens": 119900867.0,
+      "step": 35240
+    },
+    {
+      "epoch": 0.3537877875466699,
+      "learning_rate": 0.00012924444979726206,
+      "loss": 0.7478,
+      "mean_token_accuracy": 0.7810865759849548,
+      "num_tokens": 119935275.0,
+      "step": 35250
+    },
+    {
+      "epoch": 0.35388815287647035,
+      "learning_rate": 0.00012922437673130194,
+      "loss": 0.7718,
+      "mean_token_accuracy": 0.7729858040809632,
+      "num_tokens": 119969548.0,
+      "step": 35260
+    },
+    {
+      "epoch": 0.3539885182062708,
+      "learning_rate": 0.00012920430366534185,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.7698695123195648,
+      "num_tokens": 120003401.0,
+      "step": 35270
+    },
+    {
+      "epoch": 0.3540888835360713,
+      "learning_rate": 0.00012918423059938175,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.7710729897022247,
+      "num_tokens": 120038126.0,
+      "step": 35280
+    },
+    {
+      "epoch": 0.3541892488658718,
+      "learning_rate": 0.00012916415753342166,
+      "loss": 0.74,
+      "mean_token_accuracy": 0.7778858602046966,
+      "num_tokens": 120072359.0,
+      "step": 35290
+    },
+    {
+      "epoch": 0.35428961419567223,
+      "learning_rate": 0.00012914408446746154,
+      "loss": 0.7641,
+      "mean_token_accuracy": 0.7736245393753052,
+      "num_tokens": 120106331.0,
+      "step": 35300
+    },
+    {
+      "epoch": 0.35438997952547274,
+      "learning_rate": 0.00012912401140150148,
+      "loss": 0.7479,
+      "mean_token_accuracy": 0.7788862705230712,
+      "num_tokens": 120140422.0,
+      "step": 35310
+    },
+    {
+      "epoch": 0.3544903448552732,
+      "learning_rate": 0.00012910393833554136,
+      "loss": 0.7651,
+      "mean_token_accuracy": 0.7646833181381225,
+      "num_tokens": 120173897.0,
+      "step": 35320
+    },
+    {
+      "epoch": 0.35459071018507365,
+      "learning_rate": 0.00012908386526958127,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7729146480560303,
+      "num_tokens": 120207831.0,
+      "step": 35330
+    },
+    {
+      "epoch": 0.35469107551487417,
+      "learning_rate": 0.0001290637922036212,
+      "loss": 0.7664,
+      "mean_token_accuracy": 0.7723476946353912,
+      "num_tokens": 120241925.0,
+      "step": 35340
+    },
+    {
+      "epoch": 0.3547914408446746,
+      "learning_rate": 0.0001290437191376611,
+      "loss": 0.7642,
+      "mean_token_accuracy": 0.772037923336029,
+      "num_tokens": 120277005.0,
+      "step": 35350
+    },
+    {
+      "epoch": 0.3548918061744751,
+      "learning_rate": 0.000129023646071701,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7775270521640778,
+      "num_tokens": 120310403.0,
+      "step": 35360
+    },
+    {
+      "epoch": 0.3549921715042756,
+      "learning_rate": 0.0001290035730057409,
+      "loss": 0.7591,
+      "mean_token_accuracy": 0.7727593421936035,
+      "num_tokens": 120344465.0,
+      "step": 35370
+    },
+    {
+      "epoch": 0.35509253683407604,
+      "learning_rate": 0.0001289834999397808,
+      "loss": 0.7851,
+      "mean_token_accuracy": 0.767771178483963,
+      "num_tokens": 120377939.0,
+      "step": 35380
+    },
+    {
+      "epoch": 0.3551929021638765,
+      "learning_rate": 0.0001289634268738207,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7773448050022125,
+      "num_tokens": 120411248.0,
+      "step": 35390
+    },
+    {
+      "epoch": 0.355293267493677,
+      "learning_rate": 0.00012894335380786063,
+      "loss": 0.7235,
+      "mean_token_accuracy": 0.7817453920841217,
+      "num_tokens": 120443872.0,
+      "step": 35400
+    },
+    {
+      "epoch": 0.35539363282347747,
+      "learning_rate": 0.0001289232807419005,
+      "loss": 0.7481,
+      "mean_token_accuracy": 0.7752549231052399,
+      "num_tokens": 120477526.0,
+      "step": 35410
+    },
+    {
+      "epoch": 0.3554939981532779,
+      "learning_rate": 0.00012890320767594042,
+      "loss": 0.7439,
+      "mean_token_accuracy": 0.7809197425842285,
+      "num_tokens": 120511817.0,
+      "step": 35420
+    },
+    {
+      "epoch": 0.3555943634830784,
+      "learning_rate": 0.00012888313460998033,
+      "loss": 0.743,
+      "mean_token_accuracy": 0.7750954687595367,
+      "num_tokens": 120545772.0,
+      "step": 35430
+    },
+    {
+      "epoch": 0.3556947288128789,
+      "learning_rate": 0.00012886306154402024,
+      "loss": 0.7735,
+      "mean_token_accuracy": 0.769282442331314,
+      "num_tokens": 120579574.0,
+      "step": 35440
+    },
+    {
+      "epoch": 0.35579509414267935,
+      "learning_rate": 0.00012884298847806014,
+      "loss": 0.7494,
+      "mean_token_accuracy": 0.7742872178554535,
+      "num_tokens": 120613386.0,
+      "step": 35450
+    },
+    {
+      "epoch": 0.3558954594724798,
+      "learning_rate": 0.00012882291541210005,
+      "loss": 0.7359,
+      "mean_token_accuracy": 0.7799999475479126,
+      "num_tokens": 120647508.0,
+      "step": 35460
+    },
+    {
+      "epoch": 0.3559958248022803,
+      "learning_rate": 0.00012880284234613996,
+      "loss": 0.7635,
+      "mean_token_accuracy": 0.7773070394992828,
+      "num_tokens": 120680975.0,
+      "step": 35470
+    },
+    {
+      "epoch": 0.35609619013208077,
+      "learning_rate": 0.00012878276928017987,
+      "loss": 0.7352,
+      "mean_token_accuracy": 0.7768107950687408,
+      "num_tokens": 120714574.0,
+      "step": 35480
+    },
+    {
+      "epoch": 0.3561965554618812,
+      "learning_rate": 0.00012876269621421978,
+      "loss": 0.7722,
+      "mean_token_accuracy": 0.7694811820983887,
+      "num_tokens": 120748991.0,
+      "step": 35490
+    },
+    {
+      "epoch": 0.35629692079168174,
+      "learning_rate": 0.00012874262314825966,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.7733526527881622,
+      "num_tokens": 120782737.0,
+      "step": 35500
+    },
+    {
+      "epoch": 0.3563972861214822,
+      "learning_rate": 0.0001287225500822996,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.7753716170787811,
+      "num_tokens": 120817244.0,
+      "step": 35510
+    },
+    {
+      "epoch": 0.35649765145128265,
+      "learning_rate": 0.00012870247701633948,
+      "loss": 0.7179,
+      "mean_token_accuracy": 0.7856552004814148,
+      "num_tokens": 120850616.0,
+      "step": 35520
+    },
+    {
+      "epoch": 0.35659801678108316,
+      "learning_rate": 0.00012868240395037938,
+      "loss": 0.7295,
+      "mean_token_accuracy": 0.7802593469619751,
+      "num_tokens": 120884428.0,
+      "step": 35530
+    },
+    {
+      "epoch": 0.3566983821108836,
+      "learning_rate": 0.0001286623308844193,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7753828525543213,
+      "num_tokens": 120918222.0,
+      "step": 35540
+    },
+    {
+      "epoch": 0.35679874744068407,
+      "learning_rate": 0.0001286422578184592,
+      "loss": 0.763,
+      "mean_token_accuracy": 0.7744974434375763,
+      "num_tokens": 120952602.0,
+      "step": 35550
+    },
+    {
+      "epoch": 0.3568991127704846,
+      "learning_rate": 0.0001286221847524991,
+      "loss": 0.7488,
+      "mean_token_accuracy": 0.7765093147754669,
+      "num_tokens": 120985794.0,
+      "step": 35560
+    },
+    {
+      "epoch": 0.35699947810028504,
+      "learning_rate": 0.00012860211168653902,
+      "loss": 0.7375,
+      "mean_token_accuracy": 0.7780269742012024,
+      "num_tokens": 121020158.0,
+      "step": 35570
+    },
+    {
+      "epoch": 0.3570998434300855,
+      "learning_rate": 0.00012858203862057893,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.770986270904541,
+      "num_tokens": 121054426.0,
+      "step": 35580
+    },
+    {
+      "epoch": 0.357200208759886,
+      "learning_rate": 0.0001285619655546188,
+      "loss": 0.7816,
+      "mean_token_accuracy": 0.7665948092937469,
+      "num_tokens": 121087924.0,
+      "step": 35590
+    },
+    {
+      "epoch": 0.35730057408968646,
+      "learning_rate": 0.00012854189248865874,
+      "loss": 0.782,
+      "mean_token_accuracy": 0.7741276800632477,
+      "num_tokens": 121121311.0,
+      "step": 35600
+    },
+    {
+      "epoch": 0.3574009394194869,
+      "learning_rate": 0.00012852181942269862,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7732280850410461,
+      "num_tokens": 121155020.0,
+      "step": 35610
+    },
+    {
+      "epoch": 0.35750130474928743,
+      "learning_rate": 0.00012850174635673853,
+      "loss": 0.7487,
+      "mean_token_accuracy": 0.7780355155467987,
+      "num_tokens": 121189723.0,
+      "step": 35620
+    },
+    {
+      "epoch": 0.3576016700790879,
+      "learning_rate": 0.00012848167329077844,
+      "loss": 0.7381,
+      "mean_token_accuracy": 0.774751079082489,
+      "num_tokens": 121223810.0,
+      "step": 35630
+    },
+    {
+      "epoch": 0.35770203540888834,
+      "learning_rate": 0.00012846160022481835,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7648861527442932,
+      "num_tokens": 121258285.0,
+      "step": 35640
+    },
+    {
+      "epoch": 0.35780240073868885,
+      "learning_rate": 0.00012844152715885823,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.782308691740036,
+      "num_tokens": 121292731.0,
+      "step": 35650
+    },
+    {
+      "epoch": 0.3579027660684893,
+      "learning_rate": 0.00012842145409289817,
+      "loss": 0.7618,
+      "mean_token_accuracy": 0.7717048048973083,
+      "num_tokens": 121326456.0,
+      "step": 35660
+    },
+    {
+      "epoch": 0.35800313139828976,
+      "learning_rate": 0.00012840138102693805,
+      "loss": 0.7591,
+      "mean_token_accuracy": 0.771227502822876,
+      "num_tokens": 121361074.0,
+      "step": 35670
+    },
+    {
+      "epoch": 0.3581034967280903,
+      "learning_rate": 0.00012838130796097796,
+      "loss": 0.7428,
+      "mean_token_accuracy": 0.7824000954627991,
+      "num_tokens": 121394728.0,
+      "step": 35680
+    },
+    {
+      "epoch": 0.35820386205789073,
+      "learning_rate": 0.0001283612348950179,
+      "loss": 0.7315,
+      "mean_token_accuracy": 0.7799839854240418,
+      "num_tokens": 121429478.0,
+      "step": 35690
+    },
+    {
+      "epoch": 0.3583042273876912,
+      "learning_rate": 0.00012834116182905777,
+      "loss": 0.7511,
+      "mean_token_accuracy": 0.7744730293750763,
+      "num_tokens": 121462687.0,
+      "step": 35700
+    },
+    {
+      "epoch": 0.35840459271749164,
+      "learning_rate": 0.00012832108876309768,
+      "loss": 0.7324,
+      "mean_token_accuracy": 0.783033686876297,
+      "num_tokens": 121496030.0,
+      "step": 35710
+    },
+    {
+      "epoch": 0.35850495804729215,
+      "learning_rate": 0.0001283010156971376,
+      "loss": 0.8035,
+      "mean_token_accuracy": 0.7618011116981507,
+      "num_tokens": 121530474.0,
+      "step": 35720
+    },
+    {
+      "epoch": 0.3586053233770926,
+      "learning_rate": 0.0001282809426311775,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7772661507129669,
+      "num_tokens": 121564440.0,
+      "step": 35730
+    },
+    {
+      "epoch": 0.35870568870689307,
+      "learning_rate": 0.00012826086956521738,
+      "loss": 0.737,
+      "mean_token_accuracy": 0.7823009550571441,
+      "num_tokens": 121597786.0,
+      "step": 35740
+    },
+    {
+      "epoch": 0.3588060540366936,
+      "learning_rate": 0.00012824079649925732,
+      "loss": 0.7905,
+      "mean_token_accuracy": 0.7676131546497345,
+      "num_tokens": 121632887.0,
+      "step": 35750
+    },
+    {
+      "epoch": 0.35890641936649403,
+      "learning_rate": 0.0001282207234332972,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.7723377823829651,
+      "num_tokens": 121667257.0,
+      "step": 35760
+    },
+    {
+      "epoch": 0.3590067846962945,
+      "learning_rate": 0.0001282006503673371,
+      "loss": 0.7635,
+      "mean_token_accuracy": 0.7732644557952881,
+      "num_tokens": 121701581.0,
+      "step": 35770
+    },
+    {
+      "epoch": 0.359107150026095,
+      "learning_rate": 0.00012818057730137701,
+      "loss": 0.7566,
+      "mean_token_accuracy": 0.7727309167385101,
+      "num_tokens": 121735125.0,
+      "step": 35780
+    },
+    {
+      "epoch": 0.35920751535589546,
+      "learning_rate": 0.00012816050423541692,
+      "loss": 0.7271,
+      "mean_token_accuracy": 0.7802690982818603,
+      "num_tokens": 121768958.0,
+      "step": 35790
+    },
+    {
+      "epoch": 0.3593078806856959,
+      "learning_rate": 0.00012814043116945683,
+      "loss": 0.7563,
+      "mean_token_accuracy": 0.7734170138835907,
+      "num_tokens": 121802223.0,
+      "step": 35800
+    },
+    {
+      "epoch": 0.3594082460154964,
+      "learning_rate": 0.00012812035810349674,
+      "loss": 0.7542,
+      "mean_token_accuracy": 0.7750784277915954,
+      "num_tokens": 121836128.0,
+      "step": 35810
+    },
+    {
+      "epoch": 0.3595086113452969,
+      "learning_rate": 0.00012810028503753665,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.7743003606796265,
+      "num_tokens": 121870033.0,
+      "step": 35820
+    },
+    {
+      "epoch": 0.35960897667509734,
+      "learning_rate": 0.00012808021197157653,
+      "loss": 0.7568,
+      "mean_token_accuracy": 0.775709331035614,
+      "num_tokens": 121904555.0,
+      "step": 35830
+    },
+    {
+      "epoch": 0.35970934200489785,
+      "learning_rate": 0.00012806013890561646,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7707355499267579,
+      "num_tokens": 121938328.0,
+      "step": 35840
+    },
+    {
+      "epoch": 0.3598097073346983,
+      "learning_rate": 0.00012804006583965635,
+      "loss": 0.7381,
+      "mean_token_accuracy": 0.7774211645126343,
+      "num_tokens": 121973012.0,
+      "step": 35850
+    },
+    {
+      "epoch": 0.35991007266449876,
+      "learning_rate": 0.00012801999277369625,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7730853080749511,
+      "num_tokens": 122007119.0,
+      "step": 35860
+    },
+    {
+      "epoch": 0.36001043799429927,
+      "learning_rate": 0.00012799991970773616,
+      "loss": 0.7423,
+      "mean_token_accuracy": 0.7800334930419922,
+      "num_tokens": 122040525.0,
+      "step": 35870
+    },
+    {
+      "epoch": 0.3601108033240997,
+      "learning_rate": 0.00012797984664177607,
+      "loss": 0.747,
+      "mean_token_accuracy": 0.7766052901744842,
+      "num_tokens": 122075351.0,
+      "step": 35880
+    },
+    {
+      "epoch": 0.3602111686539002,
+      "learning_rate": 0.00012795977357581598,
+      "loss": 0.7497,
+      "mean_token_accuracy": 0.7744327545166015,
+      "num_tokens": 122109400.0,
+      "step": 35890
+    },
+    {
+      "epoch": 0.3603115339837007,
+      "learning_rate": 0.0001279397005098559,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.773150897026062,
+      "num_tokens": 122143863.0,
+      "step": 35900
+    },
+    {
+      "epoch": 0.36041189931350115,
+      "learning_rate": 0.0001279196274438958,
+      "loss": 0.7214,
+      "mean_token_accuracy": 0.7819910287857056,
+      "num_tokens": 122176909.0,
+      "step": 35910
+    },
+    {
+      "epoch": 0.3605122646433016,
+      "learning_rate": 0.00012789955437793568,
+      "loss": 0.721,
+      "mean_token_accuracy": 0.7813787996768952,
+      "num_tokens": 122210184.0,
+      "step": 35920
+    },
+    {
+      "epoch": 0.3606126299731021,
+      "learning_rate": 0.0001278794813119756,
+      "loss": 0.7653,
+      "mean_token_accuracy": 0.7672504305839538,
+      "num_tokens": 122244535.0,
+      "step": 35930
+    },
+    {
+      "epoch": 0.3607129953029026,
+      "learning_rate": 0.0001278594082460155,
+      "loss": 0.7697,
+      "mean_token_accuracy": 0.7666830718517303,
+      "num_tokens": 122278573.0,
+      "step": 35940
+    },
+    {
+      "epoch": 0.36081336063270303,
+      "learning_rate": 0.0001278393351800554,
+      "loss": 0.7392,
+      "mean_token_accuracy": 0.7782879769802094,
+      "num_tokens": 122312826.0,
+      "step": 35950
+    },
+    {
+      "epoch": 0.36091372596250354,
+      "learning_rate": 0.0001278192621140953,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.7697879791259765,
+      "num_tokens": 122347337.0,
+      "step": 35960
+    },
+    {
+      "epoch": 0.361014091292304,
+      "learning_rate": 0.00012779918904813522,
+      "loss": 0.7556,
+      "mean_token_accuracy": 0.7757453262805939,
+      "num_tokens": 122380709.0,
+      "step": 35970
+    },
+    {
+      "epoch": 0.36111445662210445,
+      "learning_rate": 0.0001277791159821751,
+      "loss": 0.7042,
+      "mean_token_accuracy": 0.7824141263961792,
+      "num_tokens": 122415641.0,
+      "step": 35980
+    },
+    {
+      "epoch": 0.3612148219519049,
+      "learning_rate": 0.00012775904291621504,
+      "loss": 0.7718,
+      "mean_token_accuracy": 0.7685666620731354,
+      "num_tokens": 122449788.0,
+      "step": 35990
+    },
+    {
+      "epoch": 0.3613151872817054,
+      "learning_rate": 0.00012773896985025492,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.7693254113197326,
+      "num_tokens": 122484571.0,
+      "step": 36000
+    },
+    {
+      "epoch": 0.3614155526115059,
+      "learning_rate": 0.00012771889678429485,
+      "loss": 0.7638,
+      "mean_token_accuracy": 0.7732280671596528,
+      "num_tokens": 122518313.0,
+      "step": 36010
+    },
+    {
+      "epoch": 0.36151591794130633,
+      "learning_rate": 0.00012769882371833473,
+      "loss": 0.7635,
+      "mean_token_accuracy": 0.7754628539085389,
+      "num_tokens": 122552576.0,
+      "step": 36020
+    },
+    {
+      "epoch": 0.36161628327110684,
+      "learning_rate": 0.00012767875065237464,
+      "loss": 0.7739,
+      "mean_token_accuracy": 0.7710064649581909,
+      "num_tokens": 122586626.0,
+      "step": 36030
+    },
+    {
+      "epoch": 0.3617166486009073,
+      "learning_rate": 0.00012765867758641458,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.7669225811958313,
+      "num_tokens": 122620849.0,
+      "step": 36040
+    },
+    {
+      "epoch": 0.36181701393070775,
+      "learning_rate": 0.00012763860452045446,
+      "loss": 0.78,
+      "mean_token_accuracy": 0.7726854979991913,
+      "num_tokens": 122654206.0,
+      "step": 36050
+    },
+    {
+      "epoch": 0.36191737926050827,
+      "learning_rate": 0.00012761853145449437,
+      "loss": 0.7336,
+      "mean_token_accuracy": 0.7790078163146973,
+      "num_tokens": 122688152.0,
+      "step": 36060
+    },
+    {
+      "epoch": 0.3620177445903087,
+      "learning_rate": 0.00012759845838853428,
+      "loss": 0.7666,
+      "mean_token_accuracy": 0.7728044927120209,
+      "num_tokens": 122721656.0,
+      "step": 36070
+    },
+    {
+      "epoch": 0.3621181099201092,
+      "learning_rate": 0.00012757838532257419,
+      "loss": 0.7433,
+      "mean_token_accuracy": 0.7780297935009003,
+      "num_tokens": 122755914.0,
+      "step": 36080
+    },
+    {
+      "epoch": 0.3622184752499097,
+      "learning_rate": 0.00012755831225661407,
+      "loss": 0.7392,
+      "mean_token_accuracy": 0.7752133071422577,
+      "num_tokens": 122790201.0,
+      "step": 36090
+    },
+    {
+      "epoch": 0.36231884057971014,
+      "learning_rate": 0.000127538239190654,
+      "loss": 0.7132,
+      "mean_token_accuracy": 0.7838562309741974,
+      "num_tokens": 122824911.0,
+      "step": 36100
+    },
+    {
+      "epoch": 0.3624192059095106,
+      "learning_rate": 0.00012751816612469388,
+      "loss": 0.7333,
+      "mean_token_accuracy": 0.7770668208599091,
+      "num_tokens": 122858503.0,
+      "step": 36110
+    },
+    {
+      "epoch": 0.3625195712393111,
+      "learning_rate": 0.0001274980930587338,
+      "loss": 0.7346,
+      "mean_token_accuracy": 0.7744692265987396,
+      "num_tokens": 122892736.0,
+      "step": 36120
+    },
+    {
+      "epoch": 0.36261993656911157,
+      "learning_rate": 0.0001274780199927737,
+      "loss": 0.786,
+      "mean_token_accuracy": 0.7676080942153931,
+      "num_tokens": 122926498.0,
+      "step": 36130
+    },
+    {
+      "epoch": 0.362720301898912,
+      "learning_rate": 0.0001274579469268136,
+      "loss": 0.7725,
+      "mean_token_accuracy": 0.7723757565021515,
+      "num_tokens": 122960969.0,
+      "step": 36140
+    },
+    {
+      "epoch": 0.36282066722871253,
+      "learning_rate": 0.00012743787386085352,
+      "loss": 0.7768,
+      "mean_token_accuracy": 0.7706885755062103,
+      "num_tokens": 122994995.0,
+      "step": 36150
+    },
+    {
+      "epoch": 0.362921032558513,
+      "learning_rate": 0.00012741780079489343,
+      "loss": 0.775,
+      "mean_token_accuracy": 0.7717813134193421,
+      "num_tokens": 123028786.0,
+      "step": 36160
+    },
+    {
+      "epoch": 0.36302139788831345,
+      "learning_rate": 0.00012739772772893333,
+      "loss": 0.74,
+      "mean_token_accuracy": 0.7784698903560638,
+      "num_tokens": 123063918.0,
+      "step": 36170
+    },
+    {
+      "epoch": 0.36312176321811396,
+      "learning_rate": 0.00012737765466297322,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7742575347423554,
+      "num_tokens": 123097707.0,
+      "step": 36180
+    },
+    {
+      "epoch": 0.3632221285479144,
+      "learning_rate": 0.00012735758159701315,
+      "loss": 0.7498,
+      "mean_token_accuracy": 0.7733630478382111,
+      "num_tokens": 123132416.0,
+      "step": 36190
+    },
+    {
+      "epoch": 0.36332249387771487,
+      "learning_rate": 0.00012733750853105303,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7720991015434265,
+      "num_tokens": 123165899.0,
+      "step": 36200
+    },
+    {
+      "epoch": 0.3634228592075154,
+      "learning_rate": 0.00012731743546509294,
+      "loss": 0.7876,
+      "mean_token_accuracy": 0.7656482338905335,
+      "num_tokens": 123199786.0,
+      "step": 36210
+    },
+    {
+      "epoch": 0.36352322453731584,
+      "learning_rate": 0.00012729736239913285,
+      "loss": 0.7835,
+      "mean_token_accuracy": 0.7608466207981109,
+      "num_tokens": 123233311.0,
+      "step": 36220
+    },
+    {
+      "epoch": 0.3636235898671163,
+      "learning_rate": 0.00012727728933317276,
+      "loss": 0.7825,
+      "mean_token_accuracy": 0.7665168404579162,
+      "num_tokens": 123267317.0,
+      "step": 36230
+    },
+    {
+      "epoch": 0.3637239551969168,
+      "learning_rate": 0.00012725721626721267,
+      "loss": 0.7422,
+      "mean_token_accuracy": 0.775238174200058,
+      "num_tokens": 123301977.0,
+      "step": 36240
+    },
+    {
+      "epoch": 0.36382432052671726,
+      "learning_rate": 0.00012723714320125257,
+      "loss": 0.7442,
+      "mean_token_accuracy": 0.7765831291675568,
+      "num_tokens": 123335894.0,
+      "step": 36250
+    },
+    {
+      "epoch": 0.3639246858565177,
+      "learning_rate": 0.00012721707013529248,
+      "loss": 0.7848,
+      "mean_token_accuracy": 0.7693867802619934,
+      "num_tokens": 123369608.0,
+      "step": 36260
+    },
+    {
+      "epoch": 0.36402505118631817,
+      "learning_rate": 0.00012719699706933236,
+      "loss": 0.7299,
+      "mean_token_accuracy": 0.7803408145904541,
+      "num_tokens": 123404677.0,
+      "step": 36270
+    },
+    {
+      "epoch": 0.3641254165161187,
+      "learning_rate": 0.0001271769240033723,
+      "loss": 0.7166,
+      "mean_token_accuracy": 0.786678260564804,
+      "num_tokens": 123439082.0,
+      "step": 36280
+    },
+    {
+      "epoch": 0.36422578184591914,
+      "learning_rate": 0.00012715685093741218,
+      "loss": 0.7497,
+      "mean_token_accuracy": 0.7783904135227203,
+      "num_tokens": 123473545.0,
+      "step": 36290
+    },
+    {
+      "epoch": 0.3643261471757196,
+      "learning_rate": 0.0001271367778714521,
+      "loss": 0.7603,
+      "mean_token_accuracy": 0.773338520526886,
+      "num_tokens": 123507677.0,
+      "step": 36300
+    },
+    {
+      "epoch": 0.3644265125055201,
+      "learning_rate": 0.000127116704805492,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7701963305473327,
+      "num_tokens": 123541560.0,
+      "step": 36310
+    },
+    {
+      "epoch": 0.36452687783532056,
+      "learning_rate": 0.0001270966317395319,
+      "loss": 0.7538,
+      "mean_token_accuracy": 0.7746260046958924,
+      "num_tokens": 123574876.0,
+      "step": 36320
+    },
+    {
+      "epoch": 0.364627243165121,
+      "learning_rate": 0.0001270765586735718,
+      "loss": 0.7334,
+      "mean_token_accuracy": 0.7801806092262268,
+      "num_tokens": 123608802.0,
+      "step": 36330
+    },
+    {
+      "epoch": 0.36472760849492153,
+      "learning_rate": 0.00012705648560761172,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.7709551453590393,
+      "num_tokens": 123641850.0,
+      "step": 36340
+    },
+    {
+      "epoch": 0.364827973824722,
+      "learning_rate": 0.0001270364125416516,
+      "loss": 0.7155,
+      "mean_token_accuracy": 0.785818237066269,
+      "num_tokens": 123676054.0,
+      "step": 36350
+    },
+    {
+      "epoch": 0.36492833915452244,
+      "learning_rate": 0.0001270163394756915,
+      "loss": 0.7449,
+      "mean_token_accuracy": 0.7793585658073425,
+      "num_tokens": 123710310.0,
+      "step": 36360
+    },
+    {
+      "epoch": 0.36502870448432295,
+      "learning_rate": 0.00012699626640973142,
+      "loss": 0.7147,
+      "mean_token_accuracy": 0.7818877160549164,
+      "num_tokens": 123744525.0,
+      "step": 36370
+    },
+    {
+      "epoch": 0.3651290698141234,
+      "learning_rate": 0.00012697619334377133,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7787317574024201,
+      "num_tokens": 123779054.0,
+      "step": 36380
+    },
+    {
+      "epoch": 0.36522943514392386,
+      "learning_rate": 0.00012695612027781124,
+      "loss": 0.7748,
+      "mean_token_accuracy": 0.7739263355731965,
+      "num_tokens": 123813024.0,
+      "step": 36390
+    },
+    {
+      "epoch": 0.3653298004737244,
+      "learning_rate": 0.00012693604721185115,
+      "loss": 0.7489,
+      "mean_token_accuracy": 0.7733473181724548,
+      "num_tokens": 123846524.0,
+      "step": 36400
+    },
+    {
+      "epoch": 0.36543016580352483,
+      "learning_rate": 0.00012691597414589106,
+      "loss": 0.7646,
+      "mean_token_accuracy": 0.7685162484645843,
+      "num_tokens": 123879497.0,
+      "step": 36410
+    },
+    {
+      "epoch": 0.3655305311333253,
+      "learning_rate": 0.00012689590107993094,
+      "loss": 0.7519,
+      "mean_token_accuracy": 0.775278776884079,
+      "num_tokens": 123913015.0,
+      "step": 36420
+    },
+    {
+      "epoch": 0.3656308964631258,
+      "learning_rate": 0.00012687582801397087,
+      "loss": 0.7495,
+      "mean_token_accuracy": 0.7701024949550629,
+      "num_tokens": 123946886.0,
+      "step": 36430
+    },
+    {
+      "epoch": 0.36573126179292625,
+      "learning_rate": 0.00012685575494801075,
+      "loss": 0.7326,
+      "mean_token_accuracy": 0.7799596846103668,
+      "num_tokens": 123981595.0,
+      "step": 36440
+    },
+    {
+      "epoch": 0.3658316271227267,
+      "learning_rate": 0.00012683568188205066,
+      "loss": 0.7393,
+      "mean_token_accuracy": 0.7762674391269684,
+      "num_tokens": 124015129.0,
+      "step": 36450
+    },
+    {
+      "epoch": 0.3659319924525272,
+      "learning_rate": 0.00012681560881609057,
+      "loss": 0.7475,
+      "mean_token_accuracy": 0.776815903186798,
+      "num_tokens": 124049488.0,
+      "step": 36460
+    },
+    {
+      "epoch": 0.3660323577823277,
+      "learning_rate": 0.00012679553575013048,
+      "loss": 0.731,
+      "mean_token_accuracy": 0.7786135077476501,
+      "num_tokens": 124084321.0,
+      "step": 36470
+    },
+    {
+      "epoch": 0.36613272311212813,
+      "learning_rate": 0.0001267754626841704,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7756665587425232,
+      "num_tokens": 124118416.0,
+      "step": 36480
+    },
+    {
+      "epoch": 0.36623308844192864,
+      "learning_rate": 0.0001267553896182103,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7719801664352417,
+      "num_tokens": 124152553.0,
+      "step": 36490
+    },
+    {
+      "epoch": 0.3663334537717291,
+      "learning_rate": 0.0001267353165522502,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7742100894451142,
+      "num_tokens": 124186663.0,
+      "step": 36500
+    },
+    {
+      "epoch": 0.36643381910152956,
+      "learning_rate": 0.00012671524348629009,
+      "loss": 0.7253,
+      "mean_token_accuracy": 0.7852876901626586,
+      "num_tokens": 124220256.0,
+      "step": 36510
+    },
+    {
+      "epoch": 0.36653418443133007,
+      "learning_rate": 0.00012669517042033002,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.7759455144405365,
+      "num_tokens": 124253958.0,
+      "step": 36520
+    },
+    {
+      "epoch": 0.3666345497611305,
+      "learning_rate": 0.0001266750973543699,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7705866992473602,
+      "num_tokens": 124288559.0,
+      "step": 36530
+    },
+    {
+      "epoch": 0.366734915090931,
+      "learning_rate": 0.0001266550242884098,
+      "loss": 0.7348,
+      "mean_token_accuracy": 0.7789439380168914,
+      "num_tokens": 124322607.0,
+      "step": 36540
+    },
+    {
+      "epoch": 0.36683528042073144,
+      "learning_rate": 0.00012663495122244972,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7821188330650329,
+      "num_tokens": 124356366.0,
+      "step": 36550
+    },
+    {
+      "epoch": 0.36693564575053195,
+      "learning_rate": 0.00012661487815648963,
+      "loss": 0.7654,
+      "mean_token_accuracy": 0.7741107821464539,
+      "num_tokens": 124390334.0,
+      "step": 36560
+    },
+    {
+      "epoch": 0.3670360110803324,
+      "learning_rate": 0.00012659480509052954,
+      "loss": 0.779,
+      "mean_token_accuracy": 0.7723045289516449,
+      "num_tokens": 124423962.0,
+      "step": 36570
+    },
+    {
+      "epoch": 0.36713637641013286,
+      "learning_rate": 0.00012657473202456944,
+      "loss": 0.7574,
+      "mean_token_accuracy": 0.7718424618244171,
+      "num_tokens": 124458136.0,
+      "step": 36580
+    },
+    {
+      "epoch": 0.36723674173993337,
+      "learning_rate": 0.00012655465895860935,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7768797993659973,
+      "num_tokens": 124492658.0,
+      "step": 36590
+    },
+    {
+      "epoch": 0.3673371070697338,
+      "learning_rate": 0.00012653458589264926,
+      "loss": 0.7618,
+      "mean_token_accuracy": 0.7682752072811126,
+      "num_tokens": 124526025.0,
+      "step": 36600
+    },
+    {
+      "epoch": 0.3674374723995343,
+      "learning_rate": 0.00012651451282668917,
+      "loss": 0.7153,
+      "mean_token_accuracy": 0.7839506208896637,
+      "num_tokens": 124559705.0,
+      "step": 36610
+    },
+    {
+      "epoch": 0.3675378377293348,
+      "learning_rate": 0.00012649443976072905,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.7706827342510223,
+      "num_tokens": 124593705.0,
+      "step": 36620
+    },
+    {
+      "epoch": 0.36763820305913525,
+      "learning_rate": 0.000126474366694769,
+      "loss": 0.79,
+      "mean_token_accuracy": 0.7627219080924987,
+      "num_tokens": 124627356.0,
+      "step": 36630
+    },
+    {
+      "epoch": 0.3677385683889357,
+      "learning_rate": 0.00012645429362880887,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.7729246199131012,
+      "num_tokens": 124661844.0,
+      "step": 36640
+    },
+    {
+      "epoch": 0.3678389337187362,
+      "learning_rate": 0.00012643422056284878,
+      "loss": 0.742,
+      "mean_token_accuracy": 0.7788021087646484,
+      "num_tokens": 124695749.0,
+      "step": 36650
+    },
+    {
+      "epoch": 0.36793929904853667,
+      "learning_rate": 0.00012641414749688869,
+      "loss": 0.786,
+      "mean_token_accuracy": 0.7673965930938721,
+      "num_tokens": 124730002.0,
+      "step": 36660
+    },
+    {
+      "epoch": 0.36803966437833713,
+      "learning_rate": 0.0001263940744309286,
+      "loss": 0.7331,
+      "mean_token_accuracy": 0.7784696877002716,
+      "num_tokens": 124764025.0,
+      "step": 36670
+    },
+    {
+      "epoch": 0.36814002970813764,
+      "learning_rate": 0.00012637400136496847,
+      "loss": 0.772,
+      "mean_token_accuracy": 0.7756337761878968,
+      "num_tokens": 124797840.0,
+      "step": 36680
+    },
+    {
+      "epoch": 0.3682403950379381,
+      "learning_rate": 0.0001263539282990084,
+      "loss": 0.7283,
+      "mean_token_accuracy": 0.7852561414241791,
+      "num_tokens": 124831965.0,
+      "step": 36690
+    },
+    {
+      "epoch": 0.36834076036773855,
+      "learning_rate": 0.0001263338552330483,
+      "loss": 0.7324,
+      "mean_token_accuracy": 0.7768737018108368,
+      "num_tokens": 124865812.0,
+      "step": 36700
+    },
+    {
+      "epoch": 0.36844112569753906,
+      "learning_rate": 0.0001263137821670882,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.7810425579547882,
+      "num_tokens": 124899291.0,
+      "step": 36710
+    },
+    {
+      "epoch": 0.3685414910273395,
+      "learning_rate": 0.0001262937091011281,
+      "loss": 0.7568,
+      "mean_token_accuracy": 0.7733590781688691,
+      "num_tokens": 124933442.0,
+      "step": 36720
+    },
+    {
+      "epoch": 0.36864185635714,
+      "learning_rate": 0.00012627363603516802,
+      "loss": 0.7473,
+      "mean_token_accuracy": 0.7781011044979096,
+      "num_tokens": 124966912.0,
+      "step": 36730
+    },
+    {
+      "epoch": 0.3687422216869405,
+      "learning_rate": 0.00012625356296920793,
+      "loss": 0.7422,
+      "mean_token_accuracy": 0.7781444549560547,
+      "num_tokens": 125000514.0,
+      "step": 36740
+    },
+    {
+      "epoch": 0.36884258701674094,
+      "learning_rate": 0.00012623348990324783,
+      "loss": 0.7532,
+      "mean_token_accuracy": 0.7719095110893249,
+      "num_tokens": 125034191.0,
+      "step": 36750
+    },
+    {
+      "epoch": 0.3689429523465414,
+      "learning_rate": 0.00012621341683728774,
+      "loss": 0.7706,
+      "mean_token_accuracy": 0.769522351026535,
+      "num_tokens": 125068318.0,
+      "step": 36760
+    },
+    {
+      "epoch": 0.3690433176763419,
+      "learning_rate": 0.00012619334377132762,
+      "loss": 0.7397,
+      "mean_token_accuracy": 0.7740334391593933,
+      "num_tokens": 125102526.0,
+      "step": 36770
+    },
+    {
+      "epoch": 0.36914368300614236,
+      "learning_rate": 0.00012617327070536756,
+      "loss": 0.7409,
+      "mean_token_accuracy": 0.7786827027797699,
+      "num_tokens": 125137056.0,
+      "step": 36780
+    },
+    {
+      "epoch": 0.3692440483359428,
+      "learning_rate": 0.00012615319763940744,
+      "loss": 0.7508,
+      "mean_token_accuracy": 0.7754316866397858,
+      "num_tokens": 125172222.0,
+      "step": 36790
+    },
+    {
+      "epoch": 0.36934441366574333,
+      "learning_rate": 0.00012613312457344735,
+      "loss": 0.7331,
+      "mean_token_accuracy": 0.7798702359199524,
+      "num_tokens": 125206714.0,
+      "step": 36800
+    },
+    {
+      "epoch": 0.3694447789955438,
+      "learning_rate": 0.00012611305150748726,
+      "loss": 0.753,
+      "mean_token_accuracy": 0.7793877422809601,
+      "num_tokens": 125240778.0,
+      "step": 36810
+    },
+    {
+      "epoch": 0.36954514432534424,
+      "learning_rate": 0.00012609297844152717,
+      "loss": 0.7496,
+      "mean_token_accuracy": 0.7778722882270813,
+      "num_tokens": 125274674.0,
+      "step": 36820
+    },
+    {
+      "epoch": 0.3696455096551447,
+      "learning_rate": 0.00012607290537556707,
+      "loss": 0.7401,
+      "mean_token_accuracy": 0.7754093527793884,
+      "num_tokens": 125308296.0,
+      "step": 36830
+    },
+    {
+      "epoch": 0.3697458749849452,
+      "learning_rate": 0.00012605283230960698,
+      "loss": 0.7658,
+      "mean_token_accuracy": 0.7728193759918213,
+      "num_tokens": 125342130.0,
+      "step": 36840
+    },
+    {
+      "epoch": 0.36984624031474567,
+      "learning_rate": 0.0001260327592436469,
+      "loss": 0.7711,
+      "mean_token_accuracy": 0.7682547867298126,
+      "num_tokens": 125375303.0,
+      "step": 36850
+    },
+    {
+      "epoch": 0.3699466056445461,
+      "learning_rate": 0.00012601268617768677,
+      "loss": 0.7434,
+      "mean_token_accuracy": 0.7757311046123505,
+      "num_tokens": 125408713.0,
+      "step": 36860
+    },
+    {
+      "epoch": 0.37004697097434663,
+      "learning_rate": 0.0001259926131117267,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.7736545741558075,
+      "num_tokens": 125442807.0,
+      "step": 36870
+    },
+    {
+      "epoch": 0.3701473363041471,
+      "learning_rate": 0.0001259725400457666,
+      "loss": 0.7238,
+      "mean_token_accuracy": 0.7788444459438324,
+      "num_tokens": 125476682.0,
+      "step": 36880
+    },
+    {
+      "epoch": 0.37024770163394755,
+      "learning_rate": 0.0001259524669798065,
+      "loss": 0.7333,
+      "mean_token_accuracy": 0.7783869564533233,
+      "num_tokens": 125510944.0,
+      "step": 36890
+    },
+    {
+      "epoch": 0.37034806696374806,
+      "learning_rate": 0.0001259323939138464,
+      "loss": 0.7818,
+      "mean_token_accuracy": 0.7712550461292267,
+      "num_tokens": 125544681.0,
+      "step": 36900
+    },
+    {
+      "epoch": 0.3704484322935485,
+      "learning_rate": 0.00012591232084788631,
+      "loss": 0.7148,
+      "mean_token_accuracy": 0.7841877639293671,
+      "num_tokens": 125579749.0,
+      "step": 36910
+    },
+    {
+      "epoch": 0.37054879762334897,
+      "learning_rate": 0.0001258922477819262,
+      "loss": 0.7792,
+      "mean_token_accuracy": 0.773759263753891,
+      "num_tokens": 125613366.0,
+      "step": 36920
+    },
+    {
+      "epoch": 0.3706491629531495,
+      "learning_rate": 0.00012587217471596613,
+      "loss": 0.7691,
+      "mean_token_accuracy": 0.7693368136882782,
+      "num_tokens": 125647466.0,
+      "step": 36930
+    },
+    {
+      "epoch": 0.37074952828294994,
+      "learning_rate": 0.00012585210165000604,
+      "loss": 0.7572,
+      "mean_token_accuracy": 0.7806031584739686,
+      "num_tokens": 125682089.0,
+      "step": 36940
+    },
+    {
+      "epoch": 0.3708498936127504,
+      "learning_rate": 0.00012583202858404592,
+      "loss": 0.7364,
+      "mean_token_accuracy": 0.7828087329864502,
+      "num_tokens": 125715999.0,
+      "step": 36950
+    },
+    {
+      "epoch": 0.3709502589425509,
+      "learning_rate": 0.00012581195551808586,
+      "loss": 0.7653,
+      "mean_token_accuracy": 0.7720663547515869,
+      "num_tokens": 125750224.0,
+      "step": 36960
+    },
+    {
+      "epoch": 0.37105062427235136,
+      "learning_rate": 0.00012579188245212574,
+      "loss": 0.7514,
+      "mean_token_accuracy": 0.7769839286804199,
+      "num_tokens": 125784255.0,
+      "step": 36970
+    },
+    {
+      "epoch": 0.3711509896021518,
+      "learning_rate": 0.00012577180938616565,
+      "loss": 0.7622,
+      "mean_token_accuracy": 0.7722682952880859,
+      "num_tokens": 125818557.0,
+      "step": 36980
+    },
+    {
+      "epoch": 0.3712513549319523,
+      "learning_rate": 0.00012575173632020556,
+      "loss": 0.7481,
+      "mean_token_accuracy": 0.7747491598129272,
+      "num_tokens": 125852593.0,
+      "step": 36990
+    },
+    {
+      "epoch": 0.3713517202617528,
+      "learning_rate": 0.00012573166325424546,
+      "loss": 0.7312,
+      "mean_token_accuracy": 0.7797623395919799,
+      "num_tokens": 125886279.0,
+      "step": 37000
+    },
+    {
+      "epoch": 0.37145208559155324,
+      "learning_rate": 0.00012571159018828534,
+      "loss": 0.7521,
+      "mean_token_accuracy": 0.775166916847229,
+      "num_tokens": 125920013.0,
+      "step": 37010
+    },
+    {
+      "epoch": 0.37155245092135375,
+      "learning_rate": 0.00012569151712232528,
+      "loss": 0.7385,
+      "mean_token_accuracy": 0.7798730194568634,
+      "num_tokens": 125954378.0,
+      "step": 37020
+    },
+    {
+      "epoch": 0.3716528162511542,
+      "learning_rate": 0.00012567144405636516,
+      "loss": 0.7952,
+      "mean_token_accuracy": 0.7619469940662384,
+      "num_tokens": 125988314.0,
+      "step": 37030
+    },
+    {
+      "epoch": 0.37175318158095466,
+      "learning_rate": 0.00012565137099040507,
+      "loss": 0.7297,
+      "mean_token_accuracy": 0.7822118043899536,
+      "num_tokens": 126021084.0,
+      "step": 37040
+    },
+    {
+      "epoch": 0.3718535469107552,
+      "learning_rate": 0.00012563129792444498,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.7743511915206909,
+      "num_tokens": 126054971.0,
+      "step": 37050
+    },
+    {
+      "epoch": 0.37195391224055563,
+      "learning_rate": 0.0001256112248584849,
+      "loss": 0.7542,
+      "mean_token_accuracy": 0.7733421862125397,
+      "num_tokens": 126089252.0,
+      "step": 37060
+    },
+    {
+      "epoch": 0.3720542775703561,
+      "learning_rate": 0.0001255911517925248,
+      "loss": 0.739,
+      "mean_token_accuracy": 0.7727400422096252,
+      "num_tokens": 126123276.0,
+      "step": 37070
+    },
+    {
+      "epoch": 0.3721546429001566,
+      "learning_rate": 0.0001255710787265647,
+      "loss": 0.7305,
+      "mean_token_accuracy": 0.773516982793808,
+      "num_tokens": 126157140.0,
+      "step": 37080
+    },
+    {
+      "epoch": 0.37225500822995705,
+      "learning_rate": 0.0001255510056606046,
+      "loss": 0.7482,
+      "mean_token_accuracy": 0.7750555634498596,
+      "num_tokens": 126190891.0,
+      "step": 37090
+    },
+    {
+      "epoch": 0.3723553735597575,
+      "learning_rate": 0.0001255309325946445,
+      "loss": 0.779,
+      "mean_token_accuracy": 0.7678870260715485,
+      "num_tokens": 126224600.0,
+      "step": 37100
+    },
+    {
+      "epoch": 0.37245573888955796,
+      "learning_rate": 0.00012551085952868443,
+      "loss": 0.7706,
+      "mean_token_accuracy": 0.7689167320728302,
+      "num_tokens": 126259173.0,
+      "step": 37110
+    },
+    {
+      "epoch": 0.3725561042193585,
+      "learning_rate": 0.0001254907864627243,
+      "loss": 0.7361,
+      "mean_token_accuracy": 0.7755484044551849,
+      "num_tokens": 126293443.0,
+      "step": 37120
+    },
+    {
+      "epoch": 0.37265646954915893,
+      "learning_rate": 0.00012547071339676422,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7669099032878876,
+      "num_tokens": 126326676.0,
+      "step": 37130
+    },
+    {
+      "epoch": 0.3727568348789594,
+      "learning_rate": 0.00012545064033080413,
+      "loss": 0.7481,
+      "mean_token_accuracy": 0.7756665706634521,
+      "num_tokens": 126360511.0,
+      "step": 37140
+    },
+    {
+      "epoch": 0.3728572002087599,
+      "learning_rate": 0.00012543056726484404,
+      "loss": 0.7877,
+      "mean_token_accuracy": 0.7703516781330109,
+      "num_tokens": 126394643.0,
+      "step": 37150
+    },
+    {
+      "epoch": 0.37295756553856035,
+      "learning_rate": 0.00012541049419888394,
+      "loss": 0.7954,
+      "mean_token_accuracy": 0.7689454019069671,
+      "num_tokens": 126429303.0,
+      "step": 37160
+    },
+    {
+      "epoch": 0.3730579308683608,
+      "learning_rate": 0.00012539042113292385,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7690429866313935,
+      "num_tokens": 126462979.0,
+      "step": 37170
+    },
+    {
+      "epoch": 0.3731582961981613,
+      "learning_rate": 0.00012537034806696376,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.7736899137496949,
+      "num_tokens": 126497695.0,
+      "step": 37180
+    },
+    {
+      "epoch": 0.3732586615279618,
+      "learning_rate": 0.00012535027500100367,
+      "loss": 0.7522,
+      "mean_token_accuracy": 0.7704955279827118,
+      "num_tokens": 126531858.0,
+      "step": 37190
+    },
+    {
+      "epoch": 0.37335902685776223,
+      "learning_rate": 0.00012533020193504358,
+      "loss": 0.7859,
+      "mean_token_accuracy": 0.765929788351059,
+      "num_tokens": 126565500.0,
+      "step": 37200
+    },
+    {
+      "epoch": 0.37345939218756274,
+      "learning_rate": 0.00012531012886908346,
+      "loss": 0.7629,
+      "mean_token_accuracy": 0.7764375567436218,
+      "num_tokens": 126598901.0,
+      "step": 37210
+    },
+    {
+      "epoch": 0.3735597575173632,
+      "learning_rate": 0.0001252900558031234,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7729214072227478,
+      "num_tokens": 126632624.0,
+      "step": 37220
+    },
+    {
+      "epoch": 0.37366012284716366,
+      "learning_rate": 0.00012526998273716328,
+      "loss": 0.7678,
+      "mean_token_accuracy": 0.7708661913871765,
+      "num_tokens": 126666181.0,
+      "step": 37230
+    },
+    {
+      "epoch": 0.37376048817696417,
+      "learning_rate": 0.00012524990967120318,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.7777769088745117,
+      "num_tokens": 126700232.0,
+      "step": 37240
+    },
+    {
+      "epoch": 0.3738608535067646,
+      "learning_rate": 0.0001252298366052431,
+      "loss": 0.7397,
+      "mean_token_accuracy": 0.7761428117752075,
+      "num_tokens": 126734328.0,
+      "step": 37250
+    },
+    {
+      "epoch": 0.3739612188365651,
+      "learning_rate": 0.000125209763539283,
+      "loss": 0.7419,
+      "mean_token_accuracy": 0.7773242831230164,
+      "num_tokens": 126767999.0,
+      "step": 37260
+    },
+    {
+      "epoch": 0.3740615841663656,
+      "learning_rate": 0.00012518969047332288,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.7692415654659271,
+      "num_tokens": 126801550.0,
+      "step": 37270
+    },
+    {
+      "epoch": 0.37416194949616605,
+      "learning_rate": 0.00012516961740736282,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.7795442044734955,
+      "num_tokens": 126835838.0,
+      "step": 37280
+    },
+    {
+      "epoch": 0.3742623148259665,
+      "learning_rate": 0.00012514954434140273,
+      "loss": 0.7749,
+      "mean_token_accuracy": 0.7730243325233459,
+      "num_tokens": 126869550.0,
+      "step": 37290
+    },
+    {
+      "epoch": 0.374362680155767,
+      "learning_rate": 0.0001251294712754426,
+      "loss": 0.7701,
+      "mean_token_accuracy": 0.773316603899002,
+      "num_tokens": 126903801.0,
+      "step": 37300
+    },
+    {
+      "epoch": 0.37446304548556747,
+      "learning_rate": 0.00012510939820948254,
+      "loss": 0.7607,
+      "mean_token_accuracy": 0.7711721777915954,
+      "num_tokens": 126937835.0,
+      "step": 37310
+    },
+    {
+      "epoch": 0.3745634108153679,
+      "learning_rate": 0.00012508932514352242,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7751730561256409,
+      "num_tokens": 126972570.0,
+      "step": 37320
+    },
+    {
+      "epoch": 0.37466377614516844,
+      "learning_rate": 0.00012506925207756233,
+      "loss": 0.7466,
+      "mean_token_accuracy": 0.7792658150196076,
+      "num_tokens": 127005529.0,
+      "step": 37330
+    },
+    {
+      "epoch": 0.3747641414749689,
+      "learning_rate": 0.00012504917901160224,
+      "loss": 0.7662,
+      "mean_token_accuracy": 0.7746534407138824,
+      "num_tokens": 127039792.0,
+      "step": 37340
+    },
+    {
+      "epoch": 0.37486450680476935,
+      "learning_rate": 0.00012502910594564215,
+      "loss": 0.7619,
+      "mean_token_accuracy": 0.769725251197815,
+      "num_tokens": 127074021.0,
+      "step": 37350
+    },
+    {
+      "epoch": 0.37496487213456986,
+      "learning_rate": 0.00012500903287968203,
+      "loss": 0.7462,
+      "mean_token_accuracy": 0.7742079257965088,
+      "num_tokens": 127107609.0,
+      "step": 37360
+    },
+    {
+      "epoch": 0.3750652374643703,
+      "learning_rate": 0.00012498895981372197,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.774655556678772,
+      "num_tokens": 127141443.0,
+      "step": 37370
+    },
+    {
+      "epoch": 0.37516560279417077,
+      "learning_rate": 0.00012496888674776185,
+      "loss": 0.7144,
+      "mean_token_accuracy": 0.7826993942260743,
+      "num_tokens": 127175539.0,
+      "step": 37380
+    },
+    {
+      "epoch": 0.3752659681239713,
+      "learning_rate": 0.00012494881368180176,
+      "loss": 0.771,
+      "mean_token_accuracy": 0.7703517079353333,
+      "num_tokens": 127209830.0,
+      "step": 37390
+    },
+    {
+      "epoch": 0.37536633345377174,
+      "learning_rate": 0.00012492874061584167,
+      "loss": 0.7498,
+      "mean_token_accuracy": 0.7781586468219757,
+      "num_tokens": 127244945.0,
+      "step": 37400
+    },
+    {
+      "epoch": 0.3754666987835722,
+      "learning_rate": 0.00012490866754988157,
+      "loss": 0.7556,
+      "mean_token_accuracy": 0.7736250162124634,
+      "num_tokens": 127278901.0,
+      "step": 37410
+    },
+    {
+      "epoch": 0.37556706411337265,
+      "learning_rate": 0.00012488859448392148,
+      "loss": 0.76,
+      "mean_token_accuracy": 0.7787852048873901,
+      "num_tokens": 127313577.0,
+      "step": 37420
+    },
+    {
+      "epoch": 0.37566742944317316,
+      "learning_rate": 0.0001248685214179614,
+      "loss": 0.7328,
+      "mean_token_accuracy": 0.7778769314289093,
+      "num_tokens": 127346907.0,
+      "step": 37430
+    },
+    {
+      "epoch": 0.3757677947729736,
+      "learning_rate": 0.0001248484483520013,
+      "loss": 0.7852,
+      "mean_token_accuracy": 0.7669208109378814,
+      "num_tokens": 127381209.0,
+      "step": 37440
+    },
+    {
+      "epoch": 0.3758681601027741,
+      "learning_rate": 0.00012482837528604118,
+      "loss": 0.7606,
+      "mean_token_accuracy": 0.7742077112197876,
+      "num_tokens": 127415900.0,
+      "step": 37450
+    },
+    {
+      "epoch": 0.3759685254325746,
+      "learning_rate": 0.00012480830222008112,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.7786307811737061,
+      "num_tokens": 127449805.0,
+      "step": 37460
+    },
+    {
+      "epoch": 0.37606889076237504,
+      "learning_rate": 0.000124788229154121,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7758641123771668,
+      "num_tokens": 127484249.0,
+      "step": 37470
+    },
+    {
+      "epoch": 0.3761692560921755,
+      "learning_rate": 0.0001247681560881609,
+      "loss": 0.7584,
+      "mean_token_accuracy": 0.7752665996551513,
+      "num_tokens": 127518912.0,
+      "step": 37480
+    },
+    {
+      "epoch": 0.376269621421976,
+      "learning_rate": 0.00012474808302220081,
+      "loss": 0.7446,
+      "mean_token_accuracy": 0.7730976998806,
+      "num_tokens": 127553336.0,
+      "step": 37490
+    },
+    {
+      "epoch": 0.37636998675177646,
+      "learning_rate": 0.00012472800995624072,
+      "loss": 0.7123,
+      "mean_token_accuracy": 0.779695737361908,
+      "num_tokens": 127587317.0,
+      "step": 37500
+    },
+    {
+      "epoch": 0.3764703520815769,
+      "learning_rate": 0.00012470793689028063,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7722173571586609,
+      "num_tokens": 127620720.0,
+      "step": 37510
+    },
+    {
+      "epoch": 0.37657071741137743,
+      "learning_rate": 0.00012468786382432054,
+      "loss": 0.7075,
+      "mean_token_accuracy": 0.783935296535492,
+      "num_tokens": 127655610.0,
+      "step": 37520
+    },
+    {
+      "epoch": 0.3766710827411779,
+      "learning_rate": 0.00012466779075836045,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.7720625638961792,
+      "num_tokens": 127689079.0,
+      "step": 37530
+    },
+    {
+      "epoch": 0.37677144807097834,
+      "learning_rate": 0.00012464771769240033,
+      "loss": 0.7648,
+      "mean_token_accuracy": 0.7687886416912079,
+      "num_tokens": 127723638.0,
+      "step": 37540
+    },
+    {
+      "epoch": 0.37687181340077885,
+      "learning_rate": 0.00012462764462644026,
+      "loss": 0.7929,
+      "mean_token_accuracy": 0.766310304403305,
+      "num_tokens": 127757394.0,
+      "step": 37550
+    },
+    {
+      "epoch": 0.3769721787305793,
+      "learning_rate": 0.00012460757156048015,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7758144676685333,
+      "num_tokens": 127791887.0,
+      "step": 37560
+    },
+    {
+      "epoch": 0.37707254406037977,
+      "learning_rate": 0.00012458749849452005,
+      "loss": 0.7444,
+      "mean_token_accuracy": 0.7755796909332275,
+      "num_tokens": 127827347.0,
+      "step": 37570
+    },
+    {
+      "epoch": 0.3771729093901803,
+      "learning_rate": 0.00012456742542855996,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7779051125049591,
+      "num_tokens": 127861561.0,
+      "step": 37580
+    },
+    {
+      "epoch": 0.37727327471998073,
+      "learning_rate": 0.00012454735236259987,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7814350962638855,
+      "num_tokens": 127895629.0,
+      "step": 37590
+    },
+    {
+      "epoch": 0.3773736400497812,
+      "learning_rate": 0.00012452727929663975,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7693463325500488,
+      "num_tokens": 127930957.0,
+      "step": 37600
+    },
+    {
+      "epoch": 0.3774740053795817,
+      "learning_rate": 0.0001245072062306797,
+      "loss": 0.74,
+      "mean_token_accuracy": 0.7794726371765137,
+      "num_tokens": 127964539.0,
+      "step": 37610
+    },
+    {
+      "epoch": 0.37757437070938216,
+      "learning_rate": 0.00012448713316471957,
+      "loss": 0.7354,
+      "mean_token_accuracy": 0.7787318110466004,
+      "num_tokens": 127997862.0,
+      "step": 37620
+    },
+    {
+      "epoch": 0.3776747360391826,
+      "learning_rate": 0.00012446706009875948,
+      "loss": 0.7348,
+      "mean_token_accuracy": 0.7819286227226258,
+      "num_tokens": 128031151.0,
+      "step": 37630
+    },
+    {
+      "epoch": 0.3777751013689831,
+      "learning_rate": 0.00012444698703279941,
+      "loss": 0.7738,
+      "mean_token_accuracy": 0.7697048485279083,
+      "num_tokens": 128065071.0,
+      "step": 37640
+    },
+    {
+      "epoch": 0.3778754666987836,
+      "learning_rate": 0.0001244269139668393,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7777126967906952,
+      "num_tokens": 128098837.0,
+      "step": 37650
+    },
+    {
+      "epoch": 0.37797583202858404,
+      "learning_rate": 0.0001244068409008792,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7643070816993713,
+      "num_tokens": 128132289.0,
+      "step": 37660
+    },
+    {
+      "epoch": 0.37807619735838455,
+      "learning_rate": 0.0001243867678349191,
+      "loss": 0.7416,
+      "mean_token_accuracy": 0.7768400371074676,
+      "num_tokens": 128166670.0,
+      "step": 37670
+    },
+    {
+      "epoch": 0.378176562688185,
+      "learning_rate": 0.00012436669476895902,
+      "loss": 0.7536,
+      "mean_token_accuracy": 0.7739063024520874,
+      "num_tokens": 128200677.0,
+      "step": 37680
+    },
+    {
+      "epoch": 0.37827692801798546,
+      "learning_rate": 0.0001243466217029989,
+      "loss": 0.7402,
+      "mean_token_accuracy": 0.772838932275772,
+      "num_tokens": 128234446.0,
+      "step": 37690
+    },
+    {
+      "epoch": 0.3783772933477859,
+      "learning_rate": 0.00012432654863703884,
+      "loss": 0.7492,
+      "mean_token_accuracy": 0.7724959373474121,
+      "num_tokens": 128268474.0,
+      "step": 37700
+    },
+    {
+      "epoch": 0.3784776586775864,
+      "learning_rate": 0.00012430647557107872,
+      "loss": 0.7496,
+      "mean_token_accuracy": 0.7706626117229461,
+      "num_tokens": 128302859.0,
+      "step": 37710
+    },
+    {
+      "epoch": 0.3785780240073869,
+      "learning_rate": 0.00012428640250511865,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.771821403503418,
+      "num_tokens": 128336585.0,
+      "step": 37720
+    },
+    {
+      "epoch": 0.37867838933718734,
+      "learning_rate": 0.00012426632943915854,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.7749598860740662,
+      "num_tokens": 128370302.0,
+      "step": 37730
+    },
+    {
+      "epoch": 0.37877875466698785,
+      "learning_rate": 0.00012424625637319844,
+      "loss": 0.733,
+      "mean_token_accuracy": 0.781812310218811,
+      "num_tokens": 128404138.0,
+      "step": 37740
+    },
+    {
+      "epoch": 0.3788791199967883,
+      "learning_rate": 0.00012422618330723835,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.7723971903324127,
+      "num_tokens": 128437981.0,
+      "step": 37750
+    },
+    {
+      "epoch": 0.37897948532658876,
+      "learning_rate": 0.00012420611024127826,
+      "loss": 0.779,
+      "mean_token_accuracy": 0.7666724562644959,
+      "num_tokens": 128472955.0,
+      "step": 37760
+    },
+    {
+      "epoch": 0.3790798506563893,
+      "learning_rate": 0.00012418603717531817,
+      "loss": 0.7549,
+      "mean_token_accuracy": 0.7720897555351257,
+      "num_tokens": 128506598.0,
+      "step": 37770
+    },
+    {
+      "epoch": 0.37918021598618973,
+      "learning_rate": 0.00012416596410935808,
+      "loss": 0.7518,
+      "mean_token_accuracy": 0.7744605660438537,
+      "num_tokens": 128540548.0,
+      "step": 37780
+    },
+    {
+      "epoch": 0.3792805813159902,
+      "learning_rate": 0.00012414589104339799,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7729527950286865,
+      "num_tokens": 128574041.0,
+      "step": 37790
+    },
+    {
+      "epoch": 0.3793809466457907,
+      "learning_rate": 0.00012412581797743787,
+      "loss": 0.764,
+      "mean_token_accuracy": 0.7709854006767273,
+      "num_tokens": 128607976.0,
+      "step": 37800
+    },
+    {
+      "epoch": 0.37948131197559115,
+      "learning_rate": 0.0001241057449114778,
+      "loss": 0.7566,
+      "mean_token_accuracy": 0.775441586971283,
+      "num_tokens": 128641925.0,
+      "step": 37810
+    },
+    {
+      "epoch": 0.3795816773053916,
+      "learning_rate": 0.00012408567184551768,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.7733296632766724,
+      "num_tokens": 128675818.0,
+      "step": 37820
+    },
+    {
+      "epoch": 0.3796820426351921,
+      "learning_rate": 0.0001240655987795576,
+      "loss": 0.7503,
+      "mean_token_accuracy": 0.7772371828556061,
+      "num_tokens": 128710322.0,
+      "step": 37830
+    },
+    {
+      "epoch": 0.3797824079649926,
+      "learning_rate": 0.0001240455257135975,
+      "loss": 0.7498,
+      "mean_token_accuracy": 0.7743576824665069,
+      "num_tokens": 128743458.0,
+      "step": 37840
+    },
+    {
+      "epoch": 0.37988277329479303,
+      "learning_rate": 0.0001240254526476374,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7745215237140656,
+      "num_tokens": 128776880.0,
+      "step": 37850
+    },
+    {
+      "epoch": 0.37998313862459354,
+      "learning_rate": 0.00012400537958167732,
+      "loss": 0.758,
+      "mean_token_accuracy": 0.7736277103424072,
+      "num_tokens": 128810765.0,
+      "step": 37860
+    },
+    {
+      "epoch": 0.380083503954394,
+      "learning_rate": 0.00012398530651571723,
+      "loss": 0.7683,
+      "mean_token_accuracy": 0.7720694422721863,
+      "num_tokens": 128843795.0,
+      "step": 37870
+    },
+    {
+      "epoch": 0.38018386928419445,
+      "learning_rate": 0.00012396523344975713,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.7744679272174835,
+      "num_tokens": 128878547.0,
+      "step": 37880
+    },
+    {
+      "epoch": 0.38028423461399496,
+      "learning_rate": 0.00012394516038379702,
+      "loss": 0.7611,
+      "mean_token_accuracy": 0.772064471244812,
+      "num_tokens": 128912492.0,
+      "step": 37890
+    },
+    {
+      "epoch": 0.3803845999437954,
+      "learning_rate": 0.00012392508731783695,
+      "loss": 0.7212,
+      "mean_token_accuracy": 0.7858875036239624,
+      "num_tokens": 128946006.0,
+      "step": 37900
+    },
+    {
+      "epoch": 0.3804849652735959,
+      "learning_rate": 0.00012390501425187683,
+      "loss": 0.7402,
+      "mean_token_accuracy": 0.7749140918254852,
+      "num_tokens": 128980245.0,
+      "step": 37910
+    },
+    {
+      "epoch": 0.3805853306033964,
+      "learning_rate": 0.00012388494118591674,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7678647816181183,
+      "num_tokens": 129013458.0,
+      "step": 37920
+    },
+    {
+      "epoch": 0.38068569593319684,
+      "learning_rate": 0.00012386486811995665,
+      "loss": 0.7517,
+      "mean_token_accuracy": 0.7781884849071503,
+      "num_tokens": 129047531.0,
+      "step": 37930
+    },
+    {
+      "epoch": 0.3807860612629973,
+      "learning_rate": 0.00012384479505399656,
+      "loss": 0.7385,
+      "mean_token_accuracy": 0.7778566002845764,
+      "num_tokens": 129081531.0,
+      "step": 37940
+    },
+    {
+      "epoch": 0.3808864265927978,
+      "learning_rate": 0.00012382472198803644,
+      "loss": 0.767,
+      "mean_token_accuracy": 0.7704708695411682,
+      "num_tokens": 129115716.0,
+      "step": 37950
+    },
+    {
+      "epoch": 0.38098679192259827,
+      "learning_rate": 0.00012380464892207638,
+      "loss": 0.7781,
+      "mean_token_accuracy": 0.7695282697677612,
+      "num_tokens": 129150803.0,
+      "step": 37960
+    },
+    {
+      "epoch": 0.3810871572523987,
+      "learning_rate": 0.00012378457585611626,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.7646220207214356,
+      "num_tokens": 129184192.0,
+      "step": 37970
+    },
+    {
+      "epoch": 0.3811875225821992,
+      "learning_rate": 0.00012376450279015616,
+      "loss": 0.7709,
+      "mean_token_accuracy": 0.7719972848892211,
+      "num_tokens": 129217864.0,
+      "step": 37980
+    },
+    {
+      "epoch": 0.3812878879119997,
+      "learning_rate": 0.0001237444297241961,
+      "loss": 0.7496,
+      "mean_token_accuracy": 0.7696025133132934,
+      "num_tokens": 129251618.0,
+      "step": 37990
+    },
+    {
+      "epoch": 0.38138825324180015,
+      "learning_rate": 0.00012372435665823598,
+      "loss": 0.7725,
+      "mean_token_accuracy": 0.7679459154605865,
+      "num_tokens": 129285137.0,
+      "step": 38000
+    },
+    {
+      "epoch": 0.3814886185716006,
+      "learning_rate": 0.0001237042835922759,
+      "loss": 0.7738,
+      "mean_token_accuracy": 0.7686336696147918,
+      "num_tokens": 129318834.0,
+      "step": 38010
+    },
+    {
+      "epoch": 0.3815889839014011,
+      "learning_rate": 0.0001236842105263158,
+      "loss": 0.7478,
+      "mean_token_accuracy": 0.7723007798194885,
+      "num_tokens": 129353198.0,
+      "step": 38020
+    },
+    {
+      "epoch": 0.38168934923120157,
+      "learning_rate": 0.0001236641374603557,
+      "loss": 0.7817,
+      "mean_token_accuracy": 0.7713904976844788,
+      "num_tokens": 129387340.0,
+      "step": 38030
+    },
+    {
+      "epoch": 0.381789714561002,
+      "learning_rate": 0.0001236440643943956,
+      "loss": 0.7343,
+      "mean_token_accuracy": 0.7806975066661834,
+      "num_tokens": 129421396.0,
+      "step": 38040
+    },
+    {
+      "epoch": 0.38189007989080254,
+      "learning_rate": 0.00012362399132843552,
+      "loss": 0.7543,
+      "mean_token_accuracy": 0.7753277361392975,
+      "num_tokens": 129454695.0,
+      "step": 38050
+    },
+    {
+      "epoch": 0.381990445220603,
+      "learning_rate": 0.0001236039182624754,
+      "loss": 0.7086,
+      "mean_token_accuracy": 0.787426209449768,
+      "num_tokens": 129487747.0,
+      "step": 38060
+    },
+    {
+      "epoch": 0.38209081055040345,
+      "learning_rate": 0.00012358384519651531,
+      "loss": 0.7587,
+      "mean_token_accuracy": 0.7715207815170289,
+      "num_tokens": 129521523.0,
+      "step": 38070
+    },
+    {
+      "epoch": 0.38219117588020396,
+      "learning_rate": 0.00012356377213055522,
+      "loss": 0.7447,
+      "mean_token_accuracy": 0.7790843069553375,
+      "num_tokens": 129555619.0,
+      "step": 38080
+    },
+    {
+      "epoch": 0.3822915412100044,
+      "learning_rate": 0.00012354369906459513,
+      "loss": 0.7683,
+      "mean_token_accuracy": 0.770205271244049,
+      "num_tokens": 129590842.0,
+      "step": 38090
+    },
+    {
+      "epoch": 0.38239190653980487,
+      "learning_rate": 0.00012352362599863504,
+      "loss": 0.7362,
+      "mean_token_accuracy": 0.7805061757564544,
+      "num_tokens": 129625176.0,
+      "step": 38100
+    },
+    {
+      "epoch": 0.3824922718696054,
+      "learning_rate": 0.00012350355293267495,
+      "loss": 0.7554,
+      "mean_token_accuracy": 0.77535160779953,
+      "num_tokens": 129660292.0,
+      "step": 38110
+    },
+    {
+      "epoch": 0.38259263719940584,
+      "learning_rate": 0.00012348347986671486,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7780167102813721,
+      "num_tokens": 129693764.0,
+      "step": 38120
+    },
+    {
+      "epoch": 0.3826930025292063,
+      "learning_rate": 0.00012346340680075474,
+      "loss": 0.7535,
+      "mean_token_accuracy": 0.7745382368564606,
+      "num_tokens": 129728851.0,
+      "step": 38130
+    },
+    {
+      "epoch": 0.3827933678590068,
+      "learning_rate": 0.00012344333373479467,
+      "loss": 0.7453,
+      "mean_token_accuracy": 0.7712686598300934,
+      "num_tokens": 129762943.0,
+      "step": 38140
+    },
+    {
+      "epoch": 0.38289373318880726,
+      "learning_rate": 0.00012342326066883455,
+      "loss": 0.7849,
+      "mean_token_accuracy": 0.7647489547729492,
+      "num_tokens": 129797045.0,
+      "step": 38150
+    },
+    {
+      "epoch": 0.3829940985186077,
+      "learning_rate": 0.00012340318760287446,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.7758142232894898,
+      "num_tokens": 129831329.0,
+      "step": 38160
+    },
+    {
+      "epoch": 0.38309446384840823,
+      "learning_rate": 0.00012338311453691437,
+      "loss": 0.771,
+      "mean_token_accuracy": 0.7669499278068542,
+      "num_tokens": 129865473.0,
+      "step": 38170
+    },
+    {
+      "epoch": 0.3831948291782087,
+      "learning_rate": 0.00012336304147095428,
+      "loss": 0.7539,
+      "mean_token_accuracy": 0.776168817281723,
+      "num_tokens": 129899116.0,
+      "step": 38180
+    },
+    {
+      "epoch": 0.38329519450800914,
+      "learning_rate": 0.0001233429684049942,
+      "loss": 0.7457,
+      "mean_token_accuracy": 0.776677405834198,
+      "num_tokens": 129932521.0,
+      "step": 38190
+    },
+    {
+      "epoch": 0.38339555983780965,
+      "learning_rate": 0.0001233228953390341,
+      "loss": 0.6868,
+      "mean_token_accuracy": 0.7909096240997314,
+      "num_tokens": 129967257.0,
+      "step": 38200
+    },
+    {
+      "epoch": 0.3834959251676101,
+      "learning_rate": 0.000123302822273074,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7738026797771453,
+      "num_tokens": 130000891.0,
+      "step": 38210
+    },
+    {
+      "epoch": 0.38359629049741056,
+      "learning_rate": 0.00012328274920711389,
+      "loss": 0.7661,
+      "mean_token_accuracy": 0.774511742591858,
+      "num_tokens": 130035102.0,
+      "step": 38220
+    },
+    {
+      "epoch": 0.3836966558272111,
+      "learning_rate": 0.00012326267614115382,
+      "loss": 0.7562,
+      "mean_token_accuracy": 0.7748418092727661,
+      "num_tokens": 130068809.0,
+      "step": 38230
+    },
+    {
+      "epoch": 0.38379702115701153,
+      "learning_rate": 0.0001232426030751937,
+      "loss": 0.749,
+      "mean_token_accuracy": 0.7768145143985749,
+      "num_tokens": 130102140.0,
+      "step": 38240
+    },
+    {
+      "epoch": 0.383897386486812,
+      "learning_rate": 0.0001232225300092336,
+      "loss": 0.7685,
+      "mean_token_accuracy": 0.7733086585998535,
+      "num_tokens": 130136159.0,
+      "step": 38250
+    },
+    {
+      "epoch": 0.38399775181661244,
+      "learning_rate": 0.00012320245694327352,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7655957579612732,
+      "num_tokens": 130171165.0,
+      "step": 38260
+    },
+    {
+      "epoch": 0.38409811714641295,
+      "learning_rate": 0.00012318238387731343,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.7753102958202363,
+      "num_tokens": 130204632.0,
+      "step": 38270
+    },
+    {
+      "epoch": 0.3841984824762134,
+      "learning_rate": 0.00012316231081135334,
+      "loss": 0.7387,
+      "mean_token_accuracy": 0.7776745617389679,
+      "num_tokens": 130238758.0,
+      "step": 38280
+    },
+    {
+      "epoch": 0.38429884780601387,
+      "learning_rate": 0.00012314223774539325,
+      "loss": 0.7489,
+      "mean_token_accuracy": 0.7758101284503937,
+      "num_tokens": 130272703.0,
+      "step": 38290
+    },
+    {
+      "epoch": 0.3843992131358144,
+      "learning_rate": 0.00012312216467943313,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.7725667238235474,
+      "num_tokens": 130305097.0,
+      "step": 38300
+    },
+    {
+      "epoch": 0.38449957846561483,
+      "learning_rate": 0.00012310209161347306,
+      "loss": 0.748,
+      "mean_token_accuracy": 0.774202698469162,
+      "num_tokens": 130338847.0,
+      "step": 38310
+    },
+    {
+      "epoch": 0.3845999437954153,
+      "learning_rate": 0.00012308201854751294,
+      "loss": 0.7947,
+      "mean_token_accuracy": 0.7686450958251954,
+      "num_tokens": 130372560.0,
+      "step": 38320
+    },
+    {
+      "epoch": 0.3847003091252158,
+      "learning_rate": 0.00012306194548155285,
+      "loss": 0.7729,
+      "mean_token_accuracy": 0.7694132089614868,
+      "num_tokens": 130406376.0,
+      "step": 38330
+    },
+    {
+      "epoch": 0.38480067445501626,
+      "learning_rate": 0.0001230418724155928,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.7745043933391571,
+      "num_tokens": 130440324.0,
+      "step": 38340
+    },
+    {
+      "epoch": 0.3849010397848167,
+      "learning_rate": 0.00012302179934963267,
+      "loss": 0.7386,
+      "mean_token_accuracy": 0.7757987082004547,
+      "num_tokens": 130474536.0,
+      "step": 38350
+    },
+    {
+      "epoch": 0.3850014051146172,
+      "learning_rate": 0.00012300172628367258,
+      "loss": 0.7236,
+      "mean_token_accuracy": 0.7806441009044647,
+      "num_tokens": 130508465.0,
+      "step": 38360
+    },
+    {
+      "epoch": 0.3851017704444177,
+      "learning_rate": 0.00012298165321771249,
+      "loss": 0.7572,
+      "mean_token_accuracy": 0.7715339839458466,
+      "num_tokens": 130542995.0,
+      "step": 38370
+    },
+    {
+      "epoch": 0.38520213577421814,
+      "learning_rate": 0.0001229615801517524,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7743283450603485,
+      "num_tokens": 130577429.0,
+      "step": 38380
+    },
+    {
+      "epoch": 0.38530250110401865,
+      "learning_rate": 0.00012294150708579228,
+      "loss": 0.7389,
+      "mean_token_accuracy": 0.7832254827022552,
+      "num_tokens": 130611545.0,
+      "step": 38390
+    },
+    {
+      "epoch": 0.3854028664338191,
+      "learning_rate": 0.0001229214340198322,
+      "loss": 0.7191,
+      "mean_token_accuracy": 0.7876890063285827,
+      "num_tokens": 130646019.0,
+      "step": 38400
+    },
+    {
+      "epoch": 0.38550323176361956,
+      "learning_rate": 0.0001229013609538721,
+      "loss": 0.7649,
+      "mean_token_accuracy": 0.7704071640968323,
+      "num_tokens": 130680793.0,
+      "step": 38410
+    },
+    {
+      "epoch": 0.38560359709342007,
+      "learning_rate": 0.000122881287887912,
+      "loss": 0.7601,
+      "mean_token_accuracy": 0.7702291071414947,
+      "num_tokens": 130713834.0,
+      "step": 38420
+    },
+    {
+      "epoch": 0.3857039624232205,
+      "learning_rate": 0.0001228612148219519,
+      "loss": 0.7368,
+      "mean_token_accuracy": 0.7808869123458863,
+      "num_tokens": 130747439.0,
+      "step": 38430
+    },
+    {
+      "epoch": 0.385804327753021,
+      "learning_rate": 0.00012284114175599182,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.775489890575409,
+      "num_tokens": 130780557.0,
+      "step": 38440
+    },
+    {
+      "epoch": 0.3859046930828215,
+      "learning_rate": 0.00012282106869003173,
+      "loss": 0.7447,
+      "mean_token_accuracy": 0.7734454572200775,
+      "num_tokens": 130814569.0,
+      "step": 38450
+    },
+    {
+      "epoch": 0.38600505841262195,
+      "learning_rate": 0.00012280099562407163,
+      "loss": 0.7276,
+      "mean_token_accuracy": 0.7792356193065644,
+      "num_tokens": 130848026.0,
+      "step": 38460
+    },
+    {
+      "epoch": 0.3861054237424224,
+      "learning_rate": 0.00012278092255811154,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.778665965795517,
+      "num_tokens": 130882409.0,
+      "step": 38470
+    },
+    {
+      "epoch": 0.3862057890722229,
+      "learning_rate": 0.00012276084949215142,
+      "loss": 0.7307,
+      "mean_token_accuracy": 0.781080013513565,
+      "num_tokens": 130916451.0,
+      "step": 38480
+    },
+    {
+      "epoch": 0.38630615440202337,
+      "learning_rate": 0.00012274077642619136,
+      "loss": 0.776,
+      "mean_token_accuracy": 0.7673209249973297,
+      "num_tokens": 130950498.0,
+      "step": 38490
+    },
+    {
+      "epoch": 0.3864065197318238,
+      "learning_rate": 0.00012272070336023124,
+      "loss": 0.7404,
+      "mean_token_accuracy": 0.7765831172466278,
+      "num_tokens": 130984362.0,
+      "step": 38500
+    },
+    {
+      "epoch": 0.38650688506162434,
+      "learning_rate": 0.00012270063029427115,
+      "loss": 0.7406,
+      "mean_token_accuracy": 0.7788306057453156,
+      "num_tokens": 131019472.0,
+      "step": 38510
+    },
+    {
+      "epoch": 0.3866072503914248,
+      "learning_rate": 0.00012268055722831106,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.7686613500118256,
+      "num_tokens": 131053639.0,
+      "step": 38520
+    },
+    {
+      "epoch": 0.38670761572122525,
+      "learning_rate": 0.00012266048416235097,
+      "loss": 0.789,
+      "mean_token_accuracy": 0.769978278875351,
+      "num_tokens": 131087555.0,
+      "step": 38530
+    },
+    {
+      "epoch": 0.3868079810510257,
+      "learning_rate": 0.00012264041109639087,
+      "loss": 0.7668,
+      "mean_token_accuracy": 0.7697659850120544,
+      "num_tokens": 131121449.0,
+      "step": 38540
+    },
+    {
+      "epoch": 0.3869083463808262,
+      "learning_rate": 0.00012262033803043078,
+      "loss": 0.7858,
+      "mean_token_accuracy": 0.7668357491493225,
+      "num_tokens": 131155915.0,
+      "step": 38550
+    },
+    {
+      "epoch": 0.3870087117106267,
+      "learning_rate": 0.0001226002649644707,
+      "loss": 0.7612,
+      "mean_token_accuracy": 0.7744127333164215,
+      "num_tokens": 131189730.0,
+      "step": 38560
+    },
+    {
+      "epoch": 0.38710907704042713,
+      "learning_rate": 0.00012258019189851057,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.775093162059784,
+      "num_tokens": 131223537.0,
+      "step": 38570
+    },
+    {
+      "epoch": 0.38720944237022764,
+      "learning_rate": 0.0001225601188325505,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.7749661922454834,
+      "num_tokens": 131257383.0,
+      "step": 38580
+    },
+    {
+      "epoch": 0.3873098077000281,
+      "learning_rate": 0.0001225400457665904,
+      "loss": 0.7249,
+      "mean_token_accuracy": 0.7785895526409149,
+      "num_tokens": 131291828.0,
+      "step": 38590
+    },
+    {
+      "epoch": 0.38741017302982855,
+      "learning_rate": 0.0001225199727006303,
+      "loss": 0.7819,
+      "mean_token_accuracy": 0.7721785426139831,
+      "num_tokens": 131325991.0,
+      "step": 38600
+    },
+    {
+      "epoch": 0.38751053835962906,
+      "learning_rate": 0.0001224998996346702,
+      "loss": 0.7303,
+      "mean_token_accuracy": 0.7802187919616699,
+      "num_tokens": 131359877.0,
+      "step": 38610
+    },
+    {
+      "epoch": 0.3876109036894295,
+      "learning_rate": 0.00012247982656871011,
+      "loss": 0.7423,
+      "mean_token_accuracy": 0.7740880131721497,
+      "num_tokens": 131393686.0,
+      "step": 38620
+    },
+    {
+      "epoch": 0.38771126901923,
+      "learning_rate": 0.00012245975350275,
+      "loss": 0.7778,
+      "mean_token_accuracy": 0.7720871508121491,
+      "num_tokens": 131427396.0,
+      "step": 38630
+    },
+    {
+      "epoch": 0.3878116343490305,
+      "learning_rate": 0.00012243968043678993,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7709931194782257,
+      "num_tokens": 131462104.0,
+      "step": 38640
+    },
+    {
+      "epoch": 0.38791199967883094,
+      "learning_rate": 0.0001224196073708298,
+      "loss": 0.7991,
+      "mean_token_accuracy": 0.7669247448444366,
+      "num_tokens": 131495929.0,
+      "step": 38650
+    },
+    {
+      "epoch": 0.3880123650086314,
+      "learning_rate": 0.00012239953430486972,
+      "loss": 0.7608,
+      "mean_token_accuracy": 0.7727199912071228,
+      "num_tokens": 131530577.0,
+      "step": 38660
+    },
+    {
+      "epoch": 0.3881127303384319,
+      "learning_rate": 0.00012237946123890963,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7702851474285126,
+      "num_tokens": 131564373.0,
+      "step": 38670
+    },
+    {
+      "epoch": 0.38821309566823237,
+      "learning_rate": 0.00012235938817294954,
+      "loss": 0.7544,
+      "mean_token_accuracy": 0.777365791797638,
+      "num_tokens": 131598366.0,
+      "step": 38680
+    },
+    {
+      "epoch": 0.3883134609980328,
+      "learning_rate": 0.00012233931510698945,
+      "loss": 0.691,
+      "mean_token_accuracy": 0.7855935454368591,
+      "num_tokens": 131632672.0,
+      "step": 38690
+    },
+    {
+      "epoch": 0.38841382632783333,
+      "learning_rate": 0.00012231924204102936,
+      "loss": 0.7812,
+      "mean_token_accuracy": 0.767137199640274,
+      "num_tokens": 131666443.0,
+      "step": 38700
+    },
+    {
+      "epoch": 0.3885141916576338,
+      "learning_rate": 0.00012229916897506926,
+      "loss": 0.7284,
+      "mean_token_accuracy": 0.7804687798023224,
+      "num_tokens": 131700246.0,
+      "step": 38710
+    },
+    {
+      "epoch": 0.38861455698743425,
+      "learning_rate": 0.00012227909590910915,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7715389132499695,
+      "num_tokens": 131734412.0,
+      "step": 38720
+    },
+    {
+      "epoch": 0.38871492231723476,
+      "learning_rate": 0.00012225902284314908,
+      "loss": 0.7426,
+      "mean_token_accuracy": 0.7703909456729889,
+      "num_tokens": 131768375.0,
+      "step": 38730
+    },
+    {
+      "epoch": 0.3888152876470352,
+      "learning_rate": 0.00012223894977718896,
+      "loss": 0.729,
+      "mean_token_accuracy": 0.7757822453975678,
+      "num_tokens": 131802755.0,
+      "step": 38740
+    },
+    {
+      "epoch": 0.38891565297683567,
+      "learning_rate": 0.00012221887671122887,
+      "loss": 0.7558,
+      "mean_token_accuracy": 0.7776188731193543,
+      "num_tokens": 131836704.0,
+      "step": 38750
+    },
+    {
+      "epoch": 0.3890160183066362,
+      "learning_rate": 0.00012219880364526878,
+      "loss": 0.7869,
+      "mean_token_accuracy": 0.7667187333106995,
+      "num_tokens": 131869974.0,
+      "step": 38760
+    },
+    {
+      "epoch": 0.38911638363643664,
+      "learning_rate": 0.0001221787305793087,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.7685837090015412,
+      "num_tokens": 131904038.0,
+      "step": 38770
+    },
+    {
+      "epoch": 0.3892167489662371,
+      "learning_rate": 0.0001221586575133486,
+      "loss": 0.7767,
+      "mean_token_accuracy": 0.7679896771907806,
+      "num_tokens": 131937738.0,
+      "step": 38780
+    },
+    {
+      "epoch": 0.3893171142960376,
+      "learning_rate": 0.0001221385844473885,
+      "loss": 0.7818,
+      "mean_token_accuracy": 0.7681105613708497,
+      "num_tokens": 131971935.0,
+      "step": 38790
+    },
+    {
+      "epoch": 0.38941747962583806,
+      "learning_rate": 0.0001221185113814284,
+      "loss": 0.7328,
+      "mean_token_accuracy": 0.7784608542919159,
+      "num_tokens": 132006571.0,
+      "step": 38800
+    },
+    {
+      "epoch": 0.3895178449556385,
+      "learning_rate": 0.0001220984383154683,
+      "loss": 0.7608,
+      "mean_token_accuracy": 0.7722057044506073,
+      "num_tokens": 132040572.0,
+      "step": 38810
+    },
+    {
+      "epoch": 0.38961821028543897,
+      "learning_rate": 0.00012207836524950823,
+      "loss": 0.7398,
+      "mean_token_accuracy": 0.7809996604919434,
+      "num_tokens": 132074680.0,
+      "step": 38820
+    },
+    {
+      "epoch": 0.3897185756152395,
+      "learning_rate": 0.00012205829218354811,
+      "loss": 0.7421,
+      "mean_token_accuracy": 0.7776970565319061,
+      "num_tokens": 132108596.0,
+      "step": 38830
+    },
+    {
+      "epoch": 0.38981894094503994,
+      "learning_rate": 0.00012203821911758802,
+      "loss": 0.7506,
+      "mean_token_accuracy": 0.7727792620658874,
+      "num_tokens": 132142672.0,
+      "step": 38840
+    },
+    {
+      "epoch": 0.3899193062748404,
+      "learning_rate": 0.00012201814605162794,
+      "loss": 0.7424,
+      "mean_token_accuracy": 0.7815340638160706,
+      "num_tokens": 132176880.0,
+      "step": 38850
+    },
+    {
+      "epoch": 0.3900196716046409,
+      "learning_rate": 0.00012199807298566784,
+      "loss": 0.7745,
+      "mean_token_accuracy": 0.7714296519756317,
+      "num_tokens": 132210013.0,
+      "step": 38860
+    },
+    {
+      "epoch": 0.39012003693444136,
+      "learning_rate": 0.00012197799991970776,
+      "loss": 0.7401,
+      "mean_token_accuracy": 0.7776513814926147,
+      "num_tokens": 132244647.0,
+      "step": 38870
+    },
+    {
+      "epoch": 0.3902204022642418,
+      "learning_rate": 0.00012195792685374765,
+      "loss": 0.7731,
+      "mean_token_accuracy": 0.7630275428295136,
+      "num_tokens": 132279247.0,
+      "step": 38880
+    },
+    {
+      "epoch": 0.39032076759404233,
+      "learning_rate": 0.00012193785378778755,
+      "loss": 0.7405,
+      "mean_token_accuracy": 0.7797455608844757,
+      "num_tokens": 132314044.0,
+      "step": 38890
+    },
+    {
+      "epoch": 0.3904211329238428,
+      "learning_rate": 0.00012191778072182747,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.765515285730362,
+      "num_tokens": 132347953.0,
+      "step": 38900
+    },
+    {
+      "epoch": 0.39052149825364324,
+      "learning_rate": 0.00012189770765586736,
+      "loss": 0.7662,
+      "mean_token_accuracy": 0.7708156883716584,
+      "num_tokens": 132382564.0,
+      "step": 38910
+    },
+    {
+      "epoch": 0.39062186358344375,
+      "learning_rate": 0.00012187763458990726,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.7723477423191071,
+      "num_tokens": 132416001.0,
+      "step": 38920
+    },
+    {
+      "epoch": 0.3907222289132442,
+      "learning_rate": 0.00012185756152394718,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7714898586273193,
+      "num_tokens": 132450365.0,
+      "step": 38930
+    },
+    {
+      "epoch": 0.39082259424304466,
+      "learning_rate": 0.00012183748845798708,
+      "loss": 0.7166,
+      "mean_token_accuracy": 0.7790085256099701,
+      "num_tokens": 132484146.0,
+      "step": 38940
+    },
+    {
+      "epoch": 0.3909229595728452,
+      "learning_rate": 0.00012181741539202698,
+      "loss": 0.7511,
+      "mean_token_accuracy": 0.7746637761592865,
+      "num_tokens": 132518411.0,
+      "step": 38950
+    },
+    {
+      "epoch": 0.39102332490264563,
+      "learning_rate": 0.0001217973423260669,
+      "loss": 0.7379,
+      "mean_token_accuracy": 0.7813819944858551,
+      "num_tokens": 132552330.0,
+      "step": 38960
+    },
+    {
+      "epoch": 0.3911236902324461,
+      "learning_rate": 0.0001217772692601068,
+      "loss": 0.746,
+      "mean_token_accuracy": 0.7709066867828369,
+      "num_tokens": 132586964.0,
+      "step": 38970
+    },
+    {
+      "epoch": 0.3912240555622466,
+      "learning_rate": 0.0001217571961941467,
+      "loss": 0.7537,
+      "mean_token_accuracy": 0.7693379759788513,
+      "num_tokens": 132620807.0,
+      "step": 38980
+    },
+    {
+      "epoch": 0.39132442089204705,
+      "learning_rate": 0.00012173712312818662,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7710390269756318,
+      "num_tokens": 132655281.0,
+      "step": 38990
+    },
+    {
+      "epoch": 0.3914247862218475,
+      "learning_rate": 0.00012171705006222651,
+      "loss": 0.7218,
+      "mean_token_accuracy": 0.7797894060611725,
+      "num_tokens": 132689630.0,
+      "step": 39000
+    },
+    {
+      "epoch": 0.391525151551648,
+      "learning_rate": 0.00012169697699626641,
+      "loss": 0.7749,
+      "mean_token_accuracy": 0.7722760915756226,
+      "num_tokens": 132724265.0,
+      "step": 39010
+    },
+    {
+      "epoch": 0.3916255168814485,
+      "learning_rate": 0.00012167690393030633,
+      "loss": 0.7282,
+      "mean_token_accuracy": 0.7828829348087311,
+      "num_tokens": 132758592.0,
+      "step": 39020
+    },
+    {
+      "epoch": 0.39172588221124893,
+      "learning_rate": 0.00012165683086434623,
+      "loss": 0.7601,
+      "mean_token_accuracy": 0.7701749205589294,
+      "num_tokens": 132791836.0,
+      "step": 39030
+    },
+    {
+      "epoch": 0.39182624754104944,
+      "learning_rate": 0.00012163675779838612,
+      "loss": 0.7761,
+      "mean_token_accuracy": 0.7680363118648529,
+      "num_tokens": 132825788.0,
+      "step": 39040
+    },
+    {
+      "epoch": 0.3919266128708499,
+      "learning_rate": 0.00012161668473242604,
+      "loss": 0.7336,
+      "mean_token_accuracy": 0.7725139141082764,
+      "num_tokens": 132860034.0,
+      "step": 39050
+    },
+    {
+      "epoch": 0.39202697820065036,
+      "learning_rate": 0.00012159661166646594,
+      "loss": 0.7484,
+      "mean_token_accuracy": 0.7744539499282836,
+      "num_tokens": 132893613.0,
+      "step": 39060
+    },
+    {
+      "epoch": 0.39212734353045087,
+      "learning_rate": 0.00012157653860050585,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.7795374810695648,
+      "num_tokens": 132927565.0,
+      "step": 39070
+    },
+    {
+      "epoch": 0.3922277088602513,
+      "learning_rate": 0.00012155646553454575,
+      "loss": 0.752,
+      "mean_token_accuracy": 0.7768670856952667,
+      "num_tokens": 132962180.0,
+      "step": 39080
+    },
+    {
+      "epoch": 0.3923280741900518,
+      "learning_rate": 0.00012153639246858566,
+      "loss": 0.7512,
+      "mean_token_accuracy": 0.7783297538757324,
+      "num_tokens": 132996930.0,
+      "step": 39090
+    },
+    {
+      "epoch": 0.39242843951985223,
+      "learning_rate": 0.00012151631940262556,
+      "loss": 0.7484,
+      "mean_token_accuracy": 0.7781550347805023,
+      "num_tokens": 133031217.0,
+      "step": 39100
+    },
+    {
+      "epoch": 0.39252880484965275,
+      "learning_rate": 0.00012149624633666548,
+      "loss": 0.7325,
+      "mean_token_accuracy": 0.7798486769199371,
+      "num_tokens": 133065608.0,
+      "step": 39110
+    },
+    {
+      "epoch": 0.3926291701794532,
+      "learning_rate": 0.00012147617327070537,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7634629428386688,
+      "num_tokens": 133099837.0,
+      "step": 39120
+    },
+    {
+      "epoch": 0.39272953550925366,
+      "learning_rate": 0.00012145610020474527,
+      "loss": 0.7603,
+      "mean_token_accuracy": 0.7756829202175141,
+      "num_tokens": 133132890.0,
+      "step": 39130
+    },
+    {
+      "epoch": 0.39282990083905417,
+      "learning_rate": 0.00012143602713878519,
+      "loss": 0.7536,
+      "mean_token_accuracy": 0.7724991142749786,
+      "num_tokens": 133166951.0,
+      "step": 39140
+    },
+    {
+      "epoch": 0.3929302661688546,
+      "learning_rate": 0.00012141595407282509,
+      "loss": 0.7772,
+      "mean_token_accuracy": 0.7673039019107819,
+      "num_tokens": 133200373.0,
+      "step": 39150
+    },
+    {
+      "epoch": 0.3930306314986551,
+      "learning_rate": 0.00012139588100686498,
+      "loss": 0.7418,
+      "mean_token_accuracy": 0.7788579583168029,
+      "num_tokens": 133234595.0,
+      "step": 39160
+    },
+    {
+      "epoch": 0.3931309968284556,
+      "learning_rate": 0.0001213758079409049,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7733618021011353,
+      "num_tokens": 133268711.0,
+      "step": 39170
+    },
+    {
+      "epoch": 0.39323136215825605,
+      "learning_rate": 0.0001213557348749448,
+      "loss": 0.7494,
+      "mean_token_accuracy": 0.7766754031181335,
+      "num_tokens": 133302600.0,
+      "step": 39180
+    },
+    {
+      "epoch": 0.3933317274880565,
+      "learning_rate": 0.0001213356618089847,
+      "loss": 0.7122,
+      "mean_token_accuracy": 0.7791902840137481,
+      "num_tokens": 133336578.0,
+      "step": 39190
+    },
+    {
+      "epoch": 0.393432092817857,
+      "learning_rate": 0.00012131558874302463,
+      "loss": 0.7321,
+      "mean_token_accuracy": 0.780260443687439,
+      "num_tokens": 133371034.0,
+      "step": 39200
+    },
+    {
+      "epoch": 0.39353245814765747,
+      "learning_rate": 0.00012129551567706452,
+      "loss": 0.7377,
+      "mean_token_accuracy": 0.7821172118186951,
+      "num_tokens": 133405477.0,
+      "step": 39210
+    },
+    {
+      "epoch": 0.3936328234774579,
+      "learning_rate": 0.00012127544261110442,
+      "loss": 0.7296,
+      "mean_token_accuracy": 0.7814566135406494,
+      "num_tokens": 133438630.0,
+      "step": 39220
+    },
+    {
+      "epoch": 0.39373318880725844,
+      "learning_rate": 0.00012125536954514434,
+      "loss": 0.7748,
+      "mean_token_accuracy": 0.7685065865516663,
+      "num_tokens": 133472765.0,
+      "step": 39230
+    },
+    {
+      "epoch": 0.3938335541370589,
+      "learning_rate": 0.00012123529647918423,
+      "loss": 0.7582,
+      "mean_token_accuracy": 0.7735672891139984,
+      "num_tokens": 133507592.0,
+      "step": 39240
+    },
+    {
+      "epoch": 0.39393391946685935,
+      "learning_rate": 0.00012121522341322413,
+      "loss": 0.7449,
+      "mean_token_accuracy": 0.7802322506904602,
+      "num_tokens": 133541214.0,
+      "step": 39250
+    },
+    {
+      "epoch": 0.39403428479665986,
+      "learning_rate": 0.00012119515034726405,
+      "loss": 0.7601,
+      "mean_token_accuracy": 0.7737901687622071,
+      "num_tokens": 133575816.0,
+      "step": 39260
+    },
+    {
+      "epoch": 0.3941346501264603,
+      "learning_rate": 0.00012117507728130395,
+      "loss": 0.7735,
+      "mean_token_accuracy": 0.7727476596832276,
+      "num_tokens": 133608864.0,
+      "step": 39270
+    },
+    {
+      "epoch": 0.3942350154562608,
+      "learning_rate": 0.00012115500421534384,
+      "loss": 0.7278,
+      "mean_token_accuracy": 0.7765253186225891,
+      "num_tokens": 133642999.0,
+      "step": 39280
+    },
+    {
+      "epoch": 0.3943353807860613,
+      "learning_rate": 0.00012113493114938376,
+      "loss": 0.7584,
+      "mean_token_accuracy": 0.7723966479301453,
+      "num_tokens": 133677280.0,
+      "step": 39290
+    },
+    {
+      "epoch": 0.39443574611586174,
+      "learning_rate": 0.00012111485808342367,
+      "loss": 0.7446,
+      "mean_token_accuracy": 0.7744259595870971,
+      "num_tokens": 133711493.0,
+      "step": 39300
+    },
+    {
+      "epoch": 0.3945361114456622,
+      "learning_rate": 0.00012109478501746357,
+      "loss": 0.7543,
+      "mean_token_accuracy": 0.7753329038619995,
+      "num_tokens": 133745072.0,
+      "step": 39310
+    },
+    {
+      "epoch": 0.3946364767754627,
+      "learning_rate": 0.00012107471195150349,
+      "loss": 0.7849,
+      "mean_token_accuracy": 0.7633687257766724,
+      "num_tokens": 133779839.0,
+      "step": 39320
+    },
+    {
+      "epoch": 0.39473684210526316,
+      "learning_rate": 0.00012105463888554338,
+      "loss": 0.7682,
+      "mean_token_accuracy": 0.7730082750320435,
+      "num_tokens": 133813760.0,
+      "step": 39330
+    },
+    {
+      "epoch": 0.3948372074350636,
+      "learning_rate": 0.00012103456581958328,
+      "loss": 0.7456,
+      "mean_token_accuracy": 0.7697203040122986,
+      "num_tokens": 133848294.0,
+      "step": 39340
+    },
+    {
+      "epoch": 0.39493757276486413,
+      "learning_rate": 0.0001210144927536232,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.7708020269870758,
+      "num_tokens": 133881934.0,
+      "step": 39350
+    },
+    {
+      "epoch": 0.3950379380946646,
+      "learning_rate": 0.0001209944196876631,
+      "loss": 0.7521,
+      "mean_token_accuracy": 0.7752474427223206,
+      "num_tokens": 133915804.0,
+      "step": 39360
+    },
+    {
+      "epoch": 0.39513830342446504,
+      "learning_rate": 0.00012097434662170299,
+      "loss": 0.7526,
+      "mean_token_accuracy": 0.7766995847225189,
+      "num_tokens": 133950193.0,
+      "step": 39370
+    },
+    {
+      "epoch": 0.39523866875426555,
+      "learning_rate": 0.00012095427355574291,
+      "loss": 0.7838,
+      "mean_token_accuracy": 0.7682909846305848,
+      "num_tokens": 133984092.0,
+      "step": 39380
+    },
+    {
+      "epoch": 0.395339034084066,
+      "learning_rate": 0.00012093420048978281,
+      "loss": 0.7449,
+      "mean_token_accuracy": 0.780186551809311,
+      "num_tokens": 134018007.0,
+      "step": 39390
+    },
+    {
+      "epoch": 0.39543939941386647,
+      "learning_rate": 0.00012091412742382272,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7751919984817505,
+      "num_tokens": 134052068.0,
+      "step": 39400
+    },
+    {
+      "epoch": 0.3955397647436669,
+      "learning_rate": 0.00012089405435786262,
+      "loss": 0.7894,
+      "mean_token_accuracy": 0.7675959348678589,
+      "num_tokens": 134086871.0,
+      "step": 39410
+    },
+    {
+      "epoch": 0.39564013007346743,
+      "learning_rate": 0.00012087398129190253,
+      "loss": 0.7402,
+      "mean_token_accuracy": 0.7770104825496673,
+      "num_tokens": 134121099.0,
+      "step": 39420
+    },
+    {
+      "epoch": 0.3957404954032679,
+      "learning_rate": 0.00012085390822594244,
+      "loss": 0.7466,
+      "mean_token_accuracy": 0.7756132900714874,
+      "num_tokens": 134154883.0,
+      "step": 39430
+    },
+    {
+      "epoch": 0.39584086073306834,
+      "learning_rate": 0.00012083383515998235,
+      "loss": 0.7214,
+      "mean_token_accuracy": 0.7835146069526673,
+      "num_tokens": 134188767.0,
+      "step": 39440
+    },
+    {
+      "epoch": 0.39594122606286886,
+      "learning_rate": 0.00012081376209402224,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.7679894447326661,
+      "num_tokens": 134222891.0,
+      "step": 39450
+    },
+    {
+      "epoch": 0.3960415913926693,
+      "learning_rate": 0.00012079368902806217,
+      "loss": 0.7329,
+      "mean_token_accuracy": 0.7798572301864624,
+      "num_tokens": 134257281.0,
+      "step": 39460
+    },
+    {
+      "epoch": 0.39614195672246977,
+      "learning_rate": 0.00012077361596210206,
+      "loss": 0.7267,
+      "mean_token_accuracy": 0.7835379421710968,
+      "num_tokens": 134291092.0,
+      "step": 39470
+    },
+    {
+      "epoch": 0.3962423220522703,
+      "learning_rate": 0.00012075354289614196,
+      "loss": 0.7802,
+      "mean_token_accuracy": 0.7687345266342163,
+      "num_tokens": 134325484.0,
+      "step": 39480
+    },
+    {
+      "epoch": 0.39634268738207074,
+      "learning_rate": 0.00012073346983018188,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7736931264400482,
+      "num_tokens": 134360034.0,
+      "step": 39490
+    },
+    {
+      "epoch": 0.3964430527118712,
+      "learning_rate": 0.00012071339676422177,
+      "loss": 0.7585,
+      "mean_token_accuracy": 0.7733866035938263,
+      "num_tokens": 134393906.0,
+      "step": 39500
+    },
+    {
+      "epoch": 0.3965434180416717,
+      "learning_rate": 0.00012069332369826167,
+      "loss": 0.7836,
+      "mean_token_accuracy": 0.7700192868709564,
+      "num_tokens": 134428403.0,
+      "step": 39510
+    },
+    {
+      "epoch": 0.39664378337147216,
+      "learning_rate": 0.00012067325063230159,
+      "loss": 0.7374,
+      "mean_token_accuracy": 0.7759145796298981,
+      "num_tokens": 134462462.0,
+      "step": 39520
+    },
+    {
+      "epoch": 0.3967441487012726,
+      "learning_rate": 0.00012065317756634148,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7735425055027008,
+      "num_tokens": 134496437.0,
+      "step": 39530
+    },
+    {
+      "epoch": 0.3968445140310731,
+      "learning_rate": 0.00012063310450038139,
+      "loss": 0.7377,
+      "mean_token_accuracy": 0.7802790820598602,
+      "num_tokens": 134530426.0,
+      "step": 39540
+    },
+    {
+      "epoch": 0.3969448793608736,
+      "learning_rate": 0.00012061303143442131,
+      "loss": 0.744,
+      "mean_token_accuracy": 0.7735237538814544,
+      "num_tokens": 134564435.0,
+      "step": 39550
+    },
+    {
+      "epoch": 0.39704524469067404,
+      "learning_rate": 0.00012059295836846121,
+      "loss": 0.7358,
+      "mean_token_accuracy": 0.7738739252090454,
+      "num_tokens": 134598618.0,
+      "step": 39560
+    },
+    {
+      "epoch": 0.39714561002047455,
+      "learning_rate": 0.0001205728853025011,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7694804608821869,
+      "num_tokens": 134632116.0,
+      "step": 39570
+    },
+    {
+      "epoch": 0.397245975350275,
+      "learning_rate": 0.00012055281223654103,
+      "loss": 0.7379,
+      "mean_token_accuracy": 0.7793560028076172,
+      "num_tokens": 134666154.0,
+      "step": 39580
+    },
+    {
+      "epoch": 0.39734634068007546,
+      "learning_rate": 0.00012053273917058092,
+      "loss": 0.7504,
+      "mean_token_accuracy": 0.7775729656219482,
+      "num_tokens": 134701294.0,
+      "step": 39590
+    },
+    {
+      "epoch": 0.39744670600987597,
+      "learning_rate": 0.00012051266610462082,
+      "loss": 0.7706,
+      "mean_token_accuracy": 0.7736427843570709,
+      "num_tokens": 134735055.0,
+      "step": 39600
+    },
+    {
+      "epoch": 0.3975470713396764,
+      "learning_rate": 0.00012049259303866074,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7698934495449066,
+      "num_tokens": 134768993.0,
+      "step": 39610
+    },
+    {
+      "epoch": 0.3976474366694769,
+      "learning_rate": 0.00012047251997270063,
+      "loss": 0.7671,
+      "mean_token_accuracy": 0.7665896236896514,
+      "num_tokens": 134802020.0,
+      "step": 39620
+    },
+    {
+      "epoch": 0.3977478019992774,
+      "learning_rate": 0.00012045244690674053,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7705266535282135,
+      "num_tokens": 134835584.0,
+      "step": 39630
+    },
+    {
+      "epoch": 0.39784816732907785,
+      "learning_rate": 0.00012043237384078045,
+      "loss": 0.7747,
+      "mean_token_accuracy": 0.7658677756786346,
+      "num_tokens": 134869863.0,
+      "step": 39640
+    },
+    {
+      "epoch": 0.3979485326588783,
+      "learning_rate": 0.00012041230077482036,
+      "loss": 0.7431,
+      "mean_token_accuracy": 0.7791848838329315,
+      "num_tokens": 134904118.0,
+      "step": 39650
+    },
+    {
+      "epoch": 0.3980488979886788,
+      "learning_rate": 0.00012039222770886025,
+      "loss": 0.7636,
+      "mean_token_accuracy": 0.7708900988101959,
+      "num_tokens": 134938038.0,
+      "step": 39660
+    },
+    {
+      "epoch": 0.3981492633184793,
+      "learning_rate": 0.00012037215464290018,
+      "loss": 0.7603,
+      "mean_token_accuracy": 0.7700263500213623,
+      "num_tokens": 134972463.0,
+      "step": 39670
+    },
+    {
+      "epoch": 0.39824962864827973,
+      "learning_rate": 0.00012035208157694007,
+      "loss": 0.7457,
+      "mean_token_accuracy": 0.7756147027015686,
+      "num_tokens": 135005835.0,
+      "step": 39680
+    },
+    {
+      "epoch": 0.3983499939780802,
+      "learning_rate": 0.00012033200851097997,
+      "loss": 0.7457,
+      "mean_token_accuracy": 0.7762618720531463,
+      "num_tokens": 135040050.0,
+      "step": 39690
+    },
+    {
+      "epoch": 0.3984503593078807,
+      "learning_rate": 0.00012031193544501989,
+      "loss": 0.7786,
+      "mean_token_accuracy": 0.7641853630542755,
+      "num_tokens": 135073690.0,
+      "step": 39700
+    },
+    {
+      "epoch": 0.39855072463768115,
+      "learning_rate": 0.00012029186237905978,
+      "loss": 0.7627,
+      "mean_token_accuracy": 0.7746669232845307,
+      "num_tokens": 135106550.0,
+      "step": 39710
+    },
+    {
+      "epoch": 0.3986510899674816,
+      "learning_rate": 0.00012027178931309968,
+      "loss": 0.7691,
+      "mean_token_accuracy": 0.7719944953918457,
+      "num_tokens": 135140110.0,
+      "step": 39720
+    },
+    {
+      "epoch": 0.3987514552972821,
+      "learning_rate": 0.0001202517162471396,
+      "loss": 0.7505,
+      "mean_token_accuracy": 0.7685466587543488,
+      "num_tokens": 135174141.0,
+      "step": 39730
+    },
+    {
+      "epoch": 0.3988518206270826,
+      "learning_rate": 0.0001202316431811795,
+      "loss": 0.7435,
+      "mean_token_accuracy": 0.7785994470119476,
+      "num_tokens": 135207965.0,
+      "step": 39740
+    },
+    {
+      "epoch": 0.39895218595688303,
+      "learning_rate": 0.0001202115701152194,
+      "loss": 0.7542,
+      "mean_token_accuracy": 0.7713268518447876,
+      "num_tokens": 135242410.0,
+      "step": 39750
+    },
+    {
+      "epoch": 0.39905255128668354,
+      "learning_rate": 0.00012019149704925931,
+      "loss": 0.7346,
+      "mean_token_accuracy": 0.7775916159152985,
+      "num_tokens": 135276320.0,
+      "step": 39760
+    },
+    {
+      "epoch": 0.399152916616484,
+      "learning_rate": 0.00012017142398329922,
+      "loss": 0.7261,
+      "mean_token_accuracy": 0.7829974949359894,
+      "num_tokens": 135310970.0,
+      "step": 39770
+    },
+    {
+      "epoch": 0.39925328194628446,
+      "learning_rate": 0.00012015135091733911,
+      "loss": 0.7623,
+      "mean_token_accuracy": 0.7746733903884888,
+      "num_tokens": 135345360.0,
+      "step": 39780
+    },
+    {
+      "epoch": 0.39935364727608497,
+      "learning_rate": 0.00012013127785137904,
+      "loss": 0.7433,
+      "mean_token_accuracy": 0.7795181810855866,
+      "num_tokens": 135379554.0,
+      "step": 39790
+    },
+    {
+      "epoch": 0.3994540126058854,
+      "learning_rate": 0.00012011120478541893,
+      "loss": 0.7886,
+      "mean_token_accuracy": 0.7656014263629913,
+      "num_tokens": 135413554.0,
+      "step": 39800
+    },
+    {
+      "epoch": 0.3995543779356859,
+      "learning_rate": 0.00012009113171945883,
+      "loss": 0.7741,
+      "mean_token_accuracy": 0.7673084974288941,
+      "num_tokens": 135447136.0,
+      "step": 39810
+    },
+    {
+      "epoch": 0.3996547432654864,
+      "learning_rate": 0.00012007105865349875,
+      "loss": 0.7456,
+      "mean_token_accuracy": 0.7771454095840454,
+      "num_tokens": 135481422.0,
+      "step": 39820
+    },
+    {
+      "epoch": 0.39975510859528685,
+      "learning_rate": 0.00012005098558753864,
+      "loss": 0.7483,
+      "mean_token_accuracy": 0.7750567555427551,
+      "num_tokens": 135516082.0,
+      "step": 39830
+    },
+    {
+      "epoch": 0.3998554739250873,
+      "learning_rate": 0.00012003091252157854,
+      "loss": 0.7421,
+      "mean_token_accuracy": 0.7777432143688202,
+      "num_tokens": 135550461.0,
+      "step": 39840
+    },
+    {
+      "epoch": 0.3999558392548878,
+      "learning_rate": 0.00012001083945561846,
+      "loss": 0.7864,
+      "mean_token_accuracy": 0.767791497707367,
+      "num_tokens": 135584405.0,
+      "step": 39850
+    },
+    {
+      "epoch": 0.40005620458468827,
+      "learning_rate": 0.00011999076638965835,
+      "loss": 0.7232,
+      "mean_token_accuracy": 0.7791144847869873,
+      "num_tokens": 135618466.0,
+      "step": 39860
+    },
+    {
+      "epoch": 0.4001565699144887,
+      "learning_rate": 0.00011997069332369826,
+      "loss": 0.7816,
+      "mean_token_accuracy": 0.7730731844902039,
+      "num_tokens": 135652625.0,
+      "step": 39870
+    },
+    {
+      "epoch": 0.40025693524428924,
+      "learning_rate": 0.00011995062025773817,
+      "loss": 0.7491,
+      "mean_token_accuracy": 0.7685365915298462,
+      "num_tokens": 135685999.0,
+      "step": 39880
+    },
+    {
+      "epoch": 0.4003573005740897,
+      "learning_rate": 0.00011993054719177808,
+      "loss": 0.7417,
+      "mean_token_accuracy": 0.7769519567489624,
+      "num_tokens": 135719647.0,
+      "step": 39890
+    },
+    {
+      "epoch": 0.40045766590389015,
+      "learning_rate": 0.00011991047412581797,
+      "loss": 0.7416,
+      "mean_token_accuracy": 0.7781095504760742,
+      "num_tokens": 135753435.0,
+      "step": 39900
+    },
+    {
+      "epoch": 0.40055803123369066,
+      "learning_rate": 0.0001198904010598579,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.7716181695461273,
+      "num_tokens": 135787223.0,
+      "step": 39910
+    },
+    {
+      "epoch": 0.4006583965634911,
+      "learning_rate": 0.00011987032799389779,
+      "loss": 0.7417,
+      "mean_token_accuracy": 0.7748284220695496,
+      "num_tokens": 135821663.0,
+      "step": 39920
+    },
+    {
+      "epoch": 0.40075876189329157,
+      "learning_rate": 0.00011985025492793769,
+      "loss": 0.7541,
+      "mean_token_accuracy": 0.7745453119277954,
+      "num_tokens": 135855973.0,
+      "step": 39930
+    },
+    {
+      "epoch": 0.4008591272230921,
+      "learning_rate": 0.00011983018186197761,
+      "loss": 0.7401,
+      "mean_token_accuracy": 0.778628796339035,
+      "num_tokens": 135889965.0,
+      "step": 39940
+    },
+    {
+      "epoch": 0.40095949255289254,
+      "learning_rate": 0.0001198101087960175,
+      "loss": 0.756,
+      "mean_token_accuracy": 0.7736190974712371,
+      "num_tokens": 135923490.0,
+      "step": 39950
+    },
+    {
+      "epoch": 0.401059857882693,
+      "learning_rate": 0.0001197900357300574,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.7710761725902557,
+      "num_tokens": 135957766.0,
+      "step": 39960
+    },
+    {
+      "epoch": 0.40116022321249345,
+      "learning_rate": 0.00011976996266409732,
+      "loss": 0.7434,
+      "mean_token_accuracy": 0.7756464302539825,
+      "num_tokens": 135991673.0,
+      "step": 39970
+    },
+    {
+      "epoch": 0.40126058854229396,
+      "learning_rate": 0.00011974988959813721,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7727165937423706,
+      "num_tokens": 136025278.0,
+      "step": 39980
+    },
+    {
+      "epoch": 0.4013609538720944,
+      "learning_rate": 0.00011972981653217714,
+      "loss": 0.7375,
+      "mean_token_accuracy": 0.7782612979412079,
+      "num_tokens": 136059283.0,
+      "step": 39990
+    },
+    {
+      "epoch": 0.4014613192018949,
+      "learning_rate": 0.00011970974346621705,
+      "loss": 0.7679,
+      "mean_token_accuracy": 0.7702317774295807,
+      "num_tokens": 136093114.0,
+      "step": 40000
+    },
+    {
+      "epoch": 0.4015616845316954,
+      "learning_rate": 0.00011968967040025694,
+      "loss": 0.7292,
+      "mean_token_accuracy": 0.7799321055412293,
+      "num_tokens": 136127593.0,
+      "step": 40010
+    },
+    {
+      "epoch": 0.40166204986149584,
+      "learning_rate": 0.00011966959733429686,
+      "loss": 0.7762,
+      "mean_token_accuracy": 0.7717219829559326,
+      "num_tokens": 136161130.0,
+      "step": 40020
+    },
+    {
+      "epoch": 0.4017624151912963,
+      "learning_rate": 0.00011964952426833676,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7801750481128693,
+      "num_tokens": 136194702.0,
+      "step": 40030
+    },
+    {
+      "epoch": 0.4018627805210968,
+      "learning_rate": 0.00011962945120237665,
+      "loss": 0.7717,
+      "mean_token_accuracy": 0.7715950429439544,
+      "num_tokens": 136229077.0,
+      "step": 40040
+    },
+    {
+      "epoch": 0.40196314585089726,
+      "learning_rate": 0.00011960937813641657,
+      "loss": 0.7583,
+      "mean_token_accuracy": 0.7762810230255127,
+      "num_tokens": 136263119.0,
+      "step": 40050
+    },
+    {
+      "epoch": 0.4020635111806977,
+      "learning_rate": 0.00011958930507045647,
+      "loss": 0.763,
+      "mean_token_accuracy": 0.7726292669773102,
+      "num_tokens": 136297003.0,
+      "step": 40060
+    },
+    {
+      "epoch": 0.40216387651049823,
+      "learning_rate": 0.00011956923200449636,
+      "loss": 0.8073,
+      "mean_token_accuracy": 0.7658033907413483,
+      "num_tokens": 136331043.0,
+      "step": 40070
+    },
+    {
+      "epoch": 0.4022642418402987,
+      "learning_rate": 0.00011954915893853629,
+      "loss": 0.7467,
+      "mean_token_accuracy": 0.7768625020980835,
+      "num_tokens": 136364615.0,
+      "step": 40080
+    },
+    {
+      "epoch": 0.40236460717009914,
+      "learning_rate": 0.00011952908587257618,
+      "loss": 0.7678,
+      "mean_token_accuracy": 0.7703774511814118,
+      "num_tokens": 136398721.0,
+      "step": 40090
+    },
+    {
+      "epoch": 0.40246497249989965,
+      "learning_rate": 0.00011950901280661609,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7728688240051269,
+      "num_tokens": 136432969.0,
+      "step": 40100
+    },
+    {
+      "epoch": 0.4025653378297001,
+      "learning_rate": 0.000119488939740656,
+      "loss": 0.78,
+      "mean_token_accuracy": 0.771339625120163,
+      "num_tokens": 136467593.0,
+      "step": 40110
+    },
+    {
+      "epoch": 0.40266570315950057,
+      "learning_rate": 0.0001194688666746959,
+      "loss": 0.7264,
+      "mean_token_accuracy": 0.777644407749176,
+      "num_tokens": 136501159.0,
+      "step": 40120
+    },
+    {
+      "epoch": 0.4027660684893011,
+      "learning_rate": 0.0001194487936087358,
+      "loss": 0.7387,
+      "mean_token_accuracy": 0.7751397848129272,
+      "num_tokens": 136535752.0,
+      "step": 40130
+    },
+    {
+      "epoch": 0.40286643381910153,
+      "learning_rate": 0.00011942872054277572,
+      "loss": 0.7665,
+      "mean_token_accuracy": 0.7697070360183715,
+      "num_tokens": 136570297.0,
+      "step": 40140
+    },
+    {
+      "epoch": 0.402966799148902,
+      "learning_rate": 0.00011940864747681562,
+      "loss": 0.7268,
+      "mean_token_accuracy": 0.7820415377616883,
+      "num_tokens": 136604191.0,
+      "step": 40150
+    },
+    {
+      "epoch": 0.4030671644787025,
+      "learning_rate": 0.00011938857441085551,
+      "loss": 0.7938,
+      "mean_token_accuracy": 0.7644527673721313,
+      "num_tokens": 136638222.0,
+      "step": 40160
+    },
+    {
+      "epoch": 0.40316752980850296,
+      "learning_rate": 0.00011936850134489543,
+      "loss": 0.7641,
+      "mean_token_accuracy": 0.7696475744247436,
+      "num_tokens": 136672379.0,
+      "step": 40170
+    },
+    {
+      "epoch": 0.4032678951383034,
+      "learning_rate": 0.00011934842827893533,
+      "loss": 0.7164,
+      "mean_token_accuracy": 0.7865774869918823,
+      "num_tokens": 136706577.0,
+      "step": 40180
+    },
+    {
+      "epoch": 0.4033682604681039,
+      "learning_rate": 0.00011932835521297522,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.7706251263618469,
+      "num_tokens": 136740564.0,
+      "step": 40190
+    },
+    {
+      "epoch": 0.4034686257979044,
+      "learning_rate": 0.00011930828214701515,
+      "loss": 0.746,
+      "mean_token_accuracy": 0.7773006856441498,
+      "num_tokens": 136774055.0,
+      "step": 40200
+    },
+    {
+      "epoch": 0.40356899112770483,
+      "learning_rate": 0.00011928820908105504,
+      "loss": 0.7551,
+      "mean_token_accuracy": 0.7801610410213471,
+      "num_tokens": 136808108.0,
+      "step": 40210
+    },
+    {
+      "epoch": 0.40366935645750535,
+      "learning_rate": 0.00011926813601509495,
+      "loss": 0.7513,
+      "mean_token_accuracy": 0.7703775525093078,
+      "num_tokens": 136842157.0,
+      "step": 40220
+    },
+    {
+      "epoch": 0.4037697217873058,
+      "learning_rate": 0.00011924806294913486,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7736020088195801,
+      "num_tokens": 136877028.0,
+      "step": 40230
+    },
+    {
+      "epoch": 0.40387008711710626,
+      "learning_rate": 0.00011922798988317477,
+      "loss": 0.7598,
+      "mean_token_accuracy": 0.7710412681102753,
+      "num_tokens": 136911143.0,
+      "step": 40240
+    },
+    {
+      "epoch": 0.4039704524469067,
+      "learning_rate": 0.00011920791681721466,
+      "loss": 0.7682,
+      "mean_token_accuracy": 0.7682018101215362,
+      "num_tokens": 136944985.0,
+      "step": 40250
+    },
+    {
+      "epoch": 0.4040708177767072,
+      "learning_rate": 0.00011918784375125458,
+      "loss": 0.7538,
+      "mean_token_accuracy": 0.7723013639450074,
+      "num_tokens": 136979089.0,
+      "step": 40260
+    },
+    {
+      "epoch": 0.4041711831065077,
+      "learning_rate": 0.00011916777068529448,
+      "loss": 0.7589,
+      "mean_token_accuracy": 0.7686575293540955,
+      "num_tokens": 137013646.0,
+      "step": 40270
+    },
+    {
+      "epoch": 0.40427154843630814,
+      "learning_rate": 0.00011914769761933437,
+      "loss": 0.7563,
+      "mean_token_accuracy": 0.7762044131755829,
+      "num_tokens": 137047982.0,
+      "step": 40280
+    },
+    {
+      "epoch": 0.40437191376610865,
+      "learning_rate": 0.0001191276245533743,
+      "loss": 0.7594,
+      "mean_token_accuracy": 0.7722022533416748,
+      "num_tokens": 137082041.0,
+      "step": 40290
+    },
+    {
+      "epoch": 0.4044722790959091,
+      "learning_rate": 0.00011910755148741419,
+      "loss": 0.7481,
+      "mean_token_accuracy": 0.7727467536926269,
+      "num_tokens": 137115503.0,
+      "step": 40300
+    },
+    {
+      "epoch": 0.40457264442570956,
+      "learning_rate": 0.00011908747842145408,
+      "loss": 0.7504,
+      "mean_token_accuracy": 0.7831682920455932,
+      "num_tokens": 137149409.0,
+      "step": 40310
+    },
+    {
+      "epoch": 0.40467300975551007,
+      "learning_rate": 0.00011906740535549401,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7700968325138092,
+      "num_tokens": 137182906.0,
+      "step": 40320
+    },
+    {
+      "epoch": 0.4047733750853105,
+      "learning_rate": 0.0001190473322895339,
+      "loss": 0.7391,
+      "mean_token_accuracy": 0.7785880506038666,
+      "num_tokens": 137216828.0,
+      "step": 40330
+    },
+    {
+      "epoch": 0.404873740415111,
+      "learning_rate": 0.00011902725922357381,
+      "loss": 0.7575,
+      "mean_token_accuracy": 0.7717571794986725,
+      "num_tokens": 137251433.0,
+      "step": 40340
+    },
+    {
+      "epoch": 0.4049741057449115,
+      "learning_rate": 0.00011900718615761373,
+      "loss": 0.7589,
+      "mean_token_accuracy": 0.7681475281715393,
+      "num_tokens": 137285818.0,
+      "step": 40350
+    },
+    {
+      "epoch": 0.40507447107471195,
+      "learning_rate": 0.00011898711309165363,
+      "loss": 0.7513,
+      "mean_token_accuracy": 0.7803711235523224,
+      "num_tokens": 137320567.0,
+      "step": 40360
+    },
+    {
+      "epoch": 0.4051748364045124,
+      "learning_rate": 0.00011896704002569352,
+      "loss": 0.7594,
+      "mean_token_accuracy": 0.7707147121429443,
+      "num_tokens": 137354244.0,
+      "step": 40370
+    },
+    {
+      "epoch": 0.4052752017343129,
+      "learning_rate": 0.00011894696695973344,
+      "loss": 0.7487,
+      "mean_token_accuracy": 0.7731487333774567,
+      "num_tokens": 137387920.0,
+      "step": 40380
+    },
+    {
+      "epoch": 0.4053755670641134,
+      "learning_rate": 0.00011892689389377334,
+      "loss": 0.7619,
+      "mean_token_accuracy": 0.7695264756679535,
+      "num_tokens": 137422308.0,
+      "step": 40390
+    },
+    {
+      "epoch": 0.40547593239391383,
+      "learning_rate": 0.00011890682082781323,
+      "loss": 0.7775,
+      "mean_token_accuracy": 0.7670171678066253,
+      "num_tokens": 137456020.0,
+      "step": 40400
+    },
+    {
+      "epoch": 0.40557629772371434,
+      "learning_rate": 0.00011888674776185316,
+      "loss": 0.747,
+      "mean_token_accuracy": 0.7790325105190277,
+      "num_tokens": 137490571.0,
+      "step": 40410
+    },
+    {
+      "epoch": 0.4056766630535148,
+      "learning_rate": 0.00011886667469589305,
+      "loss": 0.7544,
+      "mean_token_accuracy": 0.7761958301067352,
+      "num_tokens": 137525331.0,
+      "step": 40420
+    },
+    {
+      "epoch": 0.40577702838331525,
+      "learning_rate": 0.00011884660162993295,
+      "loss": 0.7316,
+      "mean_token_accuracy": 0.7793192148208619,
+      "num_tokens": 137559540.0,
+      "step": 40430
+    },
+    {
+      "epoch": 0.40587739371311576,
+      "learning_rate": 0.00011882652856397287,
+      "loss": 0.7504,
+      "mean_token_accuracy": 0.770765197277069,
+      "num_tokens": 137593641.0,
+      "step": 40440
+    },
+    {
+      "epoch": 0.4059777590429162,
+      "learning_rate": 0.00011880645549801278,
+      "loss": 0.753,
+      "mean_token_accuracy": 0.7740166127681732,
+      "num_tokens": 137627269.0,
+      "step": 40450
+    },
+    {
+      "epoch": 0.4060781243727167,
+      "learning_rate": 0.00011878638243205267,
+      "loss": 0.7394,
+      "mean_token_accuracy": 0.7752980709075927,
+      "num_tokens": 137661338.0,
+      "step": 40460
+    },
+    {
+      "epoch": 0.4061784897025172,
+      "learning_rate": 0.00011876630936609259,
+      "loss": 0.7388,
+      "mean_token_accuracy": 0.7813899517059326,
+      "num_tokens": 137695734.0,
+      "step": 40470
+    },
+    {
+      "epoch": 0.40627885503231764,
+      "learning_rate": 0.00011874623630013249,
+      "loss": 0.7262,
+      "mean_token_accuracy": 0.7795300543308258,
+      "num_tokens": 137729585.0,
+      "step": 40480
+    },
+    {
+      "epoch": 0.4063792203621181,
+      "learning_rate": 0.00011872616323417238,
+      "loss": 0.7736,
+      "mean_token_accuracy": 0.7737720370292663,
+      "num_tokens": 137763600.0,
+      "step": 40490
+    },
+    {
+      "epoch": 0.4064795856919186,
+      "learning_rate": 0.0001187060901682123,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7761506915092469,
+      "num_tokens": 137797753.0,
+      "step": 40500
+    },
+    {
+      "epoch": 0.40657995102171907,
+      "learning_rate": 0.0001186860171022522,
+      "loss": 0.7748,
+      "mean_token_accuracy": 0.7699367046356201,
+      "num_tokens": 137831717.0,
+      "step": 40510
+    },
+    {
+      "epoch": 0.4066803163515195,
+      "learning_rate": 0.0001186659440362921,
+      "loss": 0.7444,
+      "mean_token_accuracy": 0.7797233581542968,
+      "num_tokens": 137865259.0,
+      "step": 40520
+    },
+    {
+      "epoch": 0.40678068168132,
+      "learning_rate": 0.00011864587097033202,
+      "loss": 0.774,
+      "mean_token_accuracy": 0.770091712474823,
+      "num_tokens": 137898475.0,
+      "step": 40530
+    },
+    {
+      "epoch": 0.4068810470111205,
+      "learning_rate": 0.00011862579790437191,
+      "loss": 0.7355,
+      "mean_token_accuracy": 0.7780645191669464,
+      "num_tokens": 137931714.0,
+      "step": 40540
+    },
+    {
+      "epoch": 0.40698141234092094,
+      "learning_rate": 0.00011860572483841183,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7742593169212342,
+      "num_tokens": 137965370.0,
+      "step": 40550
+    },
+    {
+      "epoch": 0.4070817776707214,
+      "learning_rate": 0.00011858565177245173,
+      "loss": 0.7656,
+      "mean_token_accuracy": 0.7720534443855286,
+      "num_tokens": 138000108.0,
+      "step": 40560
+    },
+    {
+      "epoch": 0.4071821430005219,
+      "learning_rate": 0.00011856557870649164,
+      "loss": 0.778,
+      "mean_token_accuracy": 0.7693661630153656,
+      "num_tokens": 138033795.0,
+      "step": 40570
+    },
+    {
+      "epoch": 0.40728250833032237,
+      "learning_rate": 0.00011854550564053154,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.7773890793323517,
+      "num_tokens": 138067536.0,
+      "step": 40580
+    },
+    {
+      "epoch": 0.4073828736601228,
+      "learning_rate": 0.00011852543257457145,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7715484380722046,
+      "num_tokens": 138101544.0,
+      "step": 40590
+    },
+    {
+      "epoch": 0.40748323898992334,
+      "learning_rate": 0.00011850535950861135,
+      "loss": 0.7064,
+      "mean_token_accuracy": 0.7851023733615875,
+      "num_tokens": 138135285.0,
+      "step": 40600
+    },
+    {
+      "epoch": 0.4075836043197238,
+      "learning_rate": 0.00011848528644265127,
+      "loss": 0.7476,
+      "mean_token_accuracy": 0.7710342526435852,
+      "num_tokens": 138170031.0,
+      "step": 40610
+    },
+    {
+      "epoch": 0.40768396964952425,
+      "learning_rate": 0.00011846521337669117,
+      "loss": 0.7515,
+      "mean_token_accuracy": 0.7754992127418519,
+      "num_tokens": 138204338.0,
+      "step": 40620
+    },
+    {
+      "epoch": 0.40778433497932476,
+      "learning_rate": 0.00011844514031073106,
+      "loss": 0.7358,
+      "mean_token_accuracy": 0.7735447943210602,
+      "num_tokens": 138238926.0,
+      "step": 40630
+    },
+    {
+      "epoch": 0.4078847003091252,
+      "learning_rate": 0.00011842506724477098,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7683779120445251,
+      "num_tokens": 138273083.0,
+      "step": 40640
+    },
+    {
+      "epoch": 0.40798506563892567,
+      "learning_rate": 0.00011840499417881088,
+      "loss": 0.7102,
+      "mean_token_accuracy": 0.7821719765663147,
+      "num_tokens": 138307340.0,
+      "step": 40650
+    },
+    {
+      "epoch": 0.4080854309687262,
+      "learning_rate": 0.00011838492111285077,
+      "loss": 0.7406,
+      "mean_token_accuracy": 0.7764710009098053,
+      "num_tokens": 138341334.0,
+      "step": 40660
+    },
+    {
+      "epoch": 0.40818579629852664,
+      "learning_rate": 0.0001183648480468907,
+      "loss": 0.7235,
+      "mean_token_accuracy": 0.7790306329727172,
+      "num_tokens": 138375579.0,
+      "step": 40670
+    },
+    {
+      "epoch": 0.4082861616283271,
+      "learning_rate": 0.00011834477498093059,
+      "loss": 0.7573,
+      "mean_token_accuracy": 0.7704456508159637,
+      "num_tokens": 138409339.0,
+      "step": 40680
+    },
+    {
+      "epoch": 0.4083865269581276,
+      "learning_rate": 0.0001183247019149705,
+      "loss": 0.7631,
+      "mean_token_accuracy": 0.7724515259265899,
+      "num_tokens": 138443032.0,
+      "step": 40690
+    },
+    {
+      "epoch": 0.40848689228792806,
+      "learning_rate": 0.00011830462884901042,
+      "loss": 0.7234,
+      "mean_token_accuracy": 0.7780615448951721,
+      "num_tokens": 138477905.0,
+      "step": 40700
+    },
+    {
+      "epoch": 0.4085872576177285,
+      "learning_rate": 0.00011828455578305031,
+      "loss": 0.7823,
+      "mean_token_accuracy": 0.7687849044799805,
+      "num_tokens": 138512803.0,
+      "step": 40710
+    },
+    {
+      "epoch": 0.40868762294752903,
+      "learning_rate": 0.00011826448271709021,
+      "loss": 0.7545,
+      "mean_token_accuracy": 0.7672878324985504,
+      "num_tokens": 138547653.0,
+      "step": 40720
+    },
+    {
+      "epoch": 0.4087879882773295,
+      "learning_rate": 0.00011824440965113013,
+      "loss": 0.7322,
+      "mean_token_accuracy": 0.7838719069957734,
+      "num_tokens": 138582169.0,
+      "step": 40730
+    },
+    {
+      "epoch": 0.40888835360712994,
+      "learning_rate": 0.00011822433658517003,
+      "loss": 0.7204,
+      "mean_token_accuracy": 0.7826182544231415,
+      "num_tokens": 138616066.0,
+      "step": 40740
+    },
+    {
+      "epoch": 0.40898871893693045,
+      "learning_rate": 0.00011820426351920992,
+      "loss": 0.776,
+      "mean_token_accuracy": 0.7676548540592194,
+      "num_tokens": 138650305.0,
+      "step": 40750
+    },
+    {
+      "epoch": 0.4090890842667309,
+      "learning_rate": 0.00011818419045324984,
+      "loss": 0.7671,
+      "mean_token_accuracy": 0.7706285357475281,
+      "num_tokens": 138684542.0,
+      "step": 40760
+    },
+    {
+      "epoch": 0.40918944959653136,
+      "learning_rate": 0.00011816411738728974,
+      "loss": 0.7867,
+      "mean_token_accuracy": 0.7677037358283997,
+      "num_tokens": 138718245.0,
+      "step": 40770
+    },
+    {
+      "epoch": 0.4092898149263319,
+      "learning_rate": 0.00011814404432132963,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7797528266906738,
+      "num_tokens": 138752264.0,
+      "step": 40780
+    },
+    {
+      "epoch": 0.40939018025613233,
+      "learning_rate": 0.00011812397125536955,
+      "loss": 0.7189,
+      "mean_token_accuracy": 0.7785382807254791,
+      "num_tokens": 138786166.0,
+      "step": 40790
+    },
+    {
+      "epoch": 0.4094905455859328,
+      "learning_rate": 0.00011810389818940946,
+      "loss": 0.7317,
+      "mean_token_accuracy": 0.7766336321830749,
+      "num_tokens": 138819757.0,
+      "step": 40800
+    },
+    {
+      "epoch": 0.40959091091573324,
+      "learning_rate": 0.00011808382512344936,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.773804223537445,
+      "num_tokens": 138853929.0,
+      "step": 40810
+    },
+    {
+      "epoch": 0.40969127624553375,
+      "learning_rate": 0.00011806375205748928,
+      "loss": 0.7371,
+      "mean_token_accuracy": 0.7774279773235321,
+      "num_tokens": 138888189.0,
+      "step": 40820
+    },
+    {
+      "epoch": 0.4097916415753342,
+      "learning_rate": 0.00011804367899152917,
+      "loss": 0.7831,
+      "mean_token_accuracy": 0.7649127542972565,
+      "num_tokens": 138921873.0,
+      "step": 40830
+    },
+    {
+      "epoch": 0.40989200690513466,
+      "learning_rate": 0.00011802360592556907,
+      "loss": 0.7893,
+      "mean_token_accuracy": 0.7690079867839813,
+      "num_tokens": 138955520.0,
+      "step": 40840
+    },
+    {
+      "epoch": 0.4099923722349352,
+      "learning_rate": 0.00011800353285960899,
+      "loss": 0.7455,
+      "mean_token_accuracy": 0.7748494923114777,
+      "num_tokens": 138989147.0,
+      "step": 40850
+    },
+    {
+      "epoch": 0.41009273756473563,
+      "learning_rate": 0.00011798345979364889,
+      "loss": 0.7563,
+      "mean_token_accuracy": 0.77107834815979,
+      "num_tokens": 139022662.0,
+      "step": 40860
+    },
+    {
+      "epoch": 0.4101931028945361,
+      "learning_rate": 0.00011796338672768878,
+      "loss": 0.7072,
+      "mean_token_accuracy": 0.7833503663539887,
+      "num_tokens": 139056757.0,
+      "step": 40870
+    },
+    {
+      "epoch": 0.4102934682243366,
+      "learning_rate": 0.0001179433136617287,
+      "loss": 0.7281,
+      "mean_token_accuracy": 0.7813907444477082,
+      "num_tokens": 139090864.0,
+      "step": 40880
+    },
+    {
+      "epoch": 0.41039383355413706,
+      "learning_rate": 0.0001179232405957686,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7724337577819824,
+      "num_tokens": 139124678.0,
+      "step": 40890
+    },
+    {
+      "epoch": 0.4104941988839375,
+      "learning_rate": 0.0001179031675298085,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7724938094615936,
+      "num_tokens": 139158464.0,
+      "step": 40900
+    },
+    {
+      "epoch": 0.410594564213738,
+      "learning_rate": 0.00011788309446384841,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.7783440172672271,
+      "num_tokens": 139192885.0,
+      "step": 40910
+    },
+    {
+      "epoch": 0.4106949295435385,
+      "learning_rate": 0.00011786302139788832,
+      "loss": 0.7126,
+      "mean_token_accuracy": 0.7847510516643524,
+      "num_tokens": 139227834.0,
+      "step": 40920
+    },
+    {
+      "epoch": 0.41079529487333893,
+      "learning_rate": 0.00011784294833192822,
+      "loss": 0.759,
+      "mean_token_accuracy": 0.7764720797538758,
+      "num_tokens": 139261774.0,
+      "step": 40930
+    },
+    {
+      "epoch": 0.41089566020313945,
+      "learning_rate": 0.00011782287526596814,
+      "loss": 0.7234,
+      "mean_token_accuracy": 0.7811742126941681,
+      "num_tokens": 139295678.0,
+      "step": 40940
+    },
+    {
+      "epoch": 0.4109960255329399,
+      "learning_rate": 0.00011780280220000804,
+      "loss": 0.7572,
+      "mean_token_accuracy": 0.7743043065071106,
+      "num_tokens": 139329327.0,
+      "step": 40950
+    },
+    {
+      "epoch": 0.41109639086274036,
+      "learning_rate": 0.00011778272913404793,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7678183138370513,
+      "num_tokens": 139363631.0,
+      "step": 40960
+    },
+    {
+      "epoch": 0.41119675619254087,
+      "learning_rate": 0.00011776265606808785,
+      "loss": 0.7533,
+      "mean_token_accuracy": 0.77596116065979,
+      "num_tokens": 139397490.0,
+      "step": 40970
+    },
+    {
+      "epoch": 0.4112971215223413,
+      "learning_rate": 0.00011774258300212775,
+      "loss": 0.7351,
+      "mean_token_accuracy": 0.7769738376140595,
+      "num_tokens": 139431700.0,
+      "step": 40980
+    },
+    {
+      "epoch": 0.4113974868521418,
+      "learning_rate": 0.00011772250993616764,
+      "loss": 0.748,
+      "mean_token_accuracy": 0.7776111900806427,
+      "num_tokens": 139465321.0,
+      "step": 40990
+    },
+    {
+      "epoch": 0.4114978521819423,
+      "learning_rate": 0.00011770243687020756,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7742827475070954,
+      "num_tokens": 139499535.0,
+      "step": 41000
+    },
+    {
+      "epoch": 0.41159821751174275,
+      "learning_rate": 0.00011768236380424746,
+      "loss": 0.7195,
+      "mean_token_accuracy": 0.7819054663181305,
+      "num_tokens": 139533933.0,
+      "step": 41010
+    },
+    {
+      "epoch": 0.4116985828415432,
+      "learning_rate": 0.00011766229073828737,
+      "loss": 0.7695,
+      "mean_token_accuracy": 0.767099529504776,
+      "num_tokens": 139568333.0,
+      "step": 41020
+    },
+    {
+      "epoch": 0.4117989481713437,
+      "learning_rate": 0.00011764221767232728,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7762628257274627,
+      "num_tokens": 139602665.0,
+      "step": 41030
+    },
+    {
+      "epoch": 0.41189931350114417,
+      "learning_rate": 0.00011762214460636718,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7746689438819885,
+      "num_tokens": 139636128.0,
+      "step": 41040
+    },
+    {
+      "epoch": 0.4119996788309446,
+      "learning_rate": 0.00011760207154040708,
+      "loss": 0.7558,
+      "mean_token_accuracy": 0.7738821566104889,
+      "num_tokens": 139669982.0,
+      "step": 41050
+    },
+    {
+      "epoch": 0.41210004416074514,
+      "learning_rate": 0.000117581998474447,
+      "loss": 0.7689,
+      "mean_token_accuracy": 0.7646086633205413,
+      "num_tokens": 139703713.0,
+      "step": 41060
+    },
+    {
+      "epoch": 0.4122004094905456,
+      "learning_rate": 0.0001175619254084869,
+      "loss": 0.7692,
+      "mean_token_accuracy": 0.7704572319984436,
+      "num_tokens": 139736886.0,
+      "step": 41070
+    },
+    {
+      "epoch": 0.41230077482034605,
+      "learning_rate": 0.00011754185234252679,
+      "loss": 0.7346,
+      "mean_token_accuracy": 0.7775911629199982,
+      "num_tokens": 139771723.0,
+      "step": 41080
+    },
+    {
+      "epoch": 0.4124011401501465,
+      "learning_rate": 0.00011752177927656671,
+      "loss": 0.7569,
+      "mean_token_accuracy": 0.7772452890872955,
+      "num_tokens": 139805498.0,
+      "step": 41090
+    },
+    {
+      "epoch": 0.412501505479947,
+      "learning_rate": 0.00011750170621060661,
+      "loss": 0.7485,
+      "mean_token_accuracy": 0.7740998506546021,
+      "num_tokens": 139839525.0,
+      "step": 41100
+    },
+    {
+      "epoch": 0.4126018708097475,
+      "learning_rate": 0.0001174816331446465,
+      "loss": 0.7457,
+      "mean_token_accuracy": 0.780066853761673,
+      "num_tokens": 139872842.0,
+      "step": 41110
+    },
+    {
+      "epoch": 0.41270223613954793,
+      "learning_rate": 0.00011746156007868642,
+      "loss": 0.7552,
+      "mean_token_accuracy": 0.7751128315925598,
+      "num_tokens": 139907535.0,
+      "step": 41120
+    },
+    {
+      "epoch": 0.41280260146934844,
+      "learning_rate": 0.00011744148701272632,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7722532749176025,
+      "num_tokens": 139941345.0,
+      "step": 41130
+    },
+    {
+      "epoch": 0.4129029667991489,
+      "learning_rate": 0.00011742141394676624,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7710931181907654,
+      "num_tokens": 139975219.0,
+      "step": 41140
+    },
+    {
+      "epoch": 0.41300333212894935,
+      "learning_rate": 0.00011740134088080615,
+      "loss": 0.735,
+      "mean_token_accuracy": 0.7752216398715973,
+      "num_tokens": 140008585.0,
+      "step": 41150
+    },
+    {
+      "epoch": 0.41310369745874986,
+      "learning_rate": 0.00011738126781484604,
+      "loss": 0.7485,
+      "mean_token_accuracy": 0.7801789045333862,
+      "num_tokens": 140042289.0,
+      "step": 41160
+    },
+    {
+      "epoch": 0.4132040627885503,
+      "learning_rate": 0.00011736119474888597,
+      "loss": 0.7721,
+      "mean_token_accuracy": 0.7664551675319672,
+      "num_tokens": 140076349.0,
+      "step": 41170
+    },
+    {
+      "epoch": 0.4133044281183508,
+      "learning_rate": 0.00011734112168292586,
+      "loss": 0.7676,
+      "mean_token_accuracy": 0.770530503988266,
+      "num_tokens": 140109969.0,
+      "step": 41180
+    },
+    {
+      "epoch": 0.4134047934481513,
+      "learning_rate": 0.00011732104861696576,
+      "loss": 0.7579,
+      "mean_token_accuracy": 0.770295524597168,
+      "num_tokens": 140143764.0,
+      "step": 41190
+    },
+    {
+      "epoch": 0.41350515877795174,
+      "learning_rate": 0.00011730097555100568,
+      "loss": 0.7432,
+      "mean_token_accuracy": 0.7698943912982941,
+      "num_tokens": 140178335.0,
+      "step": 41200
+    },
+    {
+      "epoch": 0.4136055241077522,
+      "learning_rate": 0.00011728090248504557,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7745707631111145,
+      "num_tokens": 140212049.0,
+      "step": 41210
+    },
+    {
+      "epoch": 0.4137058894375527,
+      "learning_rate": 0.00011726082941908547,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.7812729001045227,
+      "num_tokens": 140246142.0,
+      "step": 41220
+    },
+    {
+      "epoch": 0.41380625476735317,
+      "learning_rate": 0.00011724075635312539,
+      "loss": 0.7532,
+      "mean_token_accuracy": 0.7762118577957153,
+      "num_tokens": 140280003.0,
+      "step": 41230
+    },
+    {
+      "epoch": 0.4139066200971536,
+      "learning_rate": 0.00011722068328716528,
+      "loss": 0.7666,
+      "mean_token_accuracy": 0.7742089807987214,
+      "num_tokens": 140314210.0,
+      "step": 41240
+    },
+    {
+      "epoch": 0.41400698542695413,
+      "learning_rate": 0.0001172006102212052,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.7702260375022888,
+      "num_tokens": 140347757.0,
+      "step": 41250
+    },
+    {
+      "epoch": 0.4141073507567546,
+      "learning_rate": 0.0001171805371552451,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7715366303920745,
+      "num_tokens": 140382109.0,
+      "step": 41260
+    },
+    {
+      "epoch": 0.41420771608655504,
+      "learning_rate": 0.00011716046408928501,
+      "loss": 0.7936,
+      "mean_token_accuracy": 0.7636014819145203,
+      "num_tokens": 140415812.0,
+      "step": 41270
+    },
+    {
+      "epoch": 0.41430808141635556,
+      "learning_rate": 0.0001171403910233249,
+      "loss": 0.7159,
+      "mean_token_accuracy": 0.7836873412132264,
+      "num_tokens": 140450053.0,
+      "step": 41280
+    },
+    {
+      "epoch": 0.414408446746156,
+      "learning_rate": 0.00011712031795736483,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.7716418683528901,
+      "num_tokens": 140484340.0,
+      "step": 41290
+    },
+    {
+      "epoch": 0.41450881207595647,
+      "learning_rate": 0.00011710024489140472,
+      "loss": 0.7392,
+      "mean_token_accuracy": 0.7766879677772522,
+      "num_tokens": 140518537.0,
+      "step": 41300
+    },
+    {
+      "epoch": 0.414609177405757,
+      "learning_rate": 0.00011708017182544462,
+      "loss": 0.778,
+      "mean_token_accuracy": 0.7727922677993775,
+      "num_tokens": 140552468.0,
+      "step": 41310
+    },
+    {
+      "epoch": 0.41470954273555743,
+      "learning_rate": 0.00011706009875948454,
+      "loss": 0.727,
+      "mean_token_accuracy": 0.7849180281162262,
+      "num_tokens": 140586845.0,
+      "step": 41320
+    },
+    {
+      "epoch": 0.4148099080653579,
+      "learning_rate": 0.00011704002569352443,
+      "loss": 0.7367,
+      "mean_token_accuracy": 0.780201518535614,
+      "num_tokens": 140620594.0,
+      "step": 41330
+    },
+    {
+      "epoch": 0.4149102733951584,
+      "learning_rate": 0.00011701995262756433,
+      "loss": 0.774,
+      "mean_token_accuracy": 0.7677059173583984,
+      "num_tokens": 140655350.0,
+      "step": 41340
+    },
+    {
+      "epoch": 0.41501063872495886,
+      "learning_rate": 0.00011699987956160425,
+      "loss": 0.7329,
+      "mean_token_accuracy": 0.7847771227359772,
+      "num_tokens": 140689492.0,
+      "step": 41350
+    },
+    {
+      "epoch": 0.4151110040547593,
+      "learning_rate": 0.00011697980649564415,
+      "loss": 0.7356,
+      "mean_token_accuracy": 0.7767213582992554,
+      "num_tokens": 140723377.0,
+      "step": 41360
+    },
+    {
+      "epoch": 0.4152113693845598,
+      "learning_rate": 0.00011695973342968405,
+      "loss": 0.7261,
+      "mean_token_accuracy": 0.7769228696823121,
+      "num_tokens": 140757454.0,
+      "step": 41370
+    },
+    {
+      "epoch": 0.4153117347143603,
+      "learning_rate": 0.00011693966036372396,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.772264164686203,
+      "num_tokens": 140791467.0,
+      "step": 41380
+    },
+    {
+      "epoch": 0.41541210004416074,
+      "learning_rate": 0.00011691958729776387,
+      "loss": 0.7376,
+      "mean_token_accuracy": 0.7786181330680847,
+      "num_tokens": 140826132.0,
+      "step": 41390
+    },
+    {
+      "epoch": 0.4155124653739612,
+      "learning_rate": 0.00011689951423180377,
+      "loss": 0.7479,
+      "mean_token_accuracy": 0.7768079102039337,
+      "num_tokens": 140860422.0,
+      "step": 41400
+    },
+    {
+      "epoch": 0.4156128307037617,
+      "learning_rate": 0.00011687944116584369,
+      "loss": 0.7605,
+      "mean_token_accuracy": 0.7711485028266907,
+      "num_tokens": 140894125.0,
+      "step": 41410
+    },
+    {
+      "epoch": 0.41571319603356216,
+      "learning_rate": 0.00011685936809988358,
+      "loss": 0.7507,
+      "mean_token_accuracy": 0.7740625619888306,
+      "num_tokens": 140927887.0,
+      "step": 41420
+    },
+    {
+      "epoch": 0.4158135613633626,
+      "learning_rate": 0.00011683929503392348,
+      "loss": 0.7538,
+      "mean_token_accuracy": 0.7745043277740479,
+      "num_tokens": 140962183.0,
+      "step": 41430
+    },
+    {
+      "epoch": 0.4159139266931631,
+      "learning_rate": 0.0001168192219679634,
+      "loss": 0.7283,
+      "mean_token_accuracy": 0.7818750143051147,
+      "num_tokens": 140996523.0,
+      "step": 41440
+    },
+    {
+      "epoch": 0.4160142920229636,
+      "learning_rate": 0.0001167991489020033,
+      "loss": 0.7458,
+      "mean_token_accuracy": 0.7724709391593934,
+      "num_tokens": 141030741.0,
+      "step": 41450
+    },
+    {
+      "epoch": 0.41611465735276404,
+      "learning_rate": 0.00011677907583604319,
+      "loss": 0.7401,
+      "mean_token_accuracy": 0.7825262606143951,
+      "num_tokens": 141065439.0,
+      "step": 41460
+    },
+    {
+      "epoch": 0.41621502268256455,
+      "learning_rate": 0.00011675900277008311,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7768517851829528,
+      "num_tokens": 141098789.0,
+      "step": 41470
+    },
+    {
+      "epoch": 0.416315388012365,
+      "learning_rate": 0.000116738929704123,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.7819433808326721,
+      "num_tokens": 141132998.0,
+      "step": 41480
+    },
+    {
+      "epoch": 0.41641575334216546,
+      "learning_rate": 0.00011671885663816291,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7756395041942596,
+      "num_tokens": 141167153.0,
+      "step": 41490
+    },
+    {
+      "epoch": 0.416516118671966,
+      "learning_rate": 0.00011669878357220284,
+      "loss": 0.7353,
+      "mean_token_accuracy": 0.7808833479881286,
+      "num_tokens": 141201802.0,
+      "step": 41500
+    },
+    {
+      "epoch": 0.41661648400176643,
+      "learning_rate": 0.00011667871050624273,
+      "loss": 0.7476,
+      "mean_token_accuracy": 0.7750447750091553,
+      "num_tokens": 141235160.0,
+      "step": 41510
+    },
+    {
+      "epoch": 0.4167168493315669,
+      "learning_rate": 0.00011665863744028263,
+      "loss": 0.7626,
+      "mean_token_accuracy": 0.7666607677936554,
+      "num_tokens": 141269775.0,
+      "step": 41520
+    },
+    {
+      "epoch": 0.4168172146613674,
+      "learning_rate": 0.00011663856437432255,
+      "loss": 0.7793,
+      "mean_token_accuracy": 0.7676746845245361,
+      "num_tokens": 141304175.0,
+      "step": 41530
+    },
+    {
+      "epoch": 0.41691757999116785,
+      "learning_rate": 0.00011661849130836244,
+      "loss": 0.7443,
+      "mean_token_accuracy": 0.7779941499233246,
+      "num_tokens": 141338306.0,
+      "step": 41540
+    },
+    {
+      "epoch": 0.4170179453209683,
+      "learning_rate": 0.00011659841824240234,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7654629826545716,
+      "num_tokens": 141371972.0,
+      "step": 41550
+    },
+    {
+      "epoch": 0.4171183106507688,
+      "learning_rate": 0.00011657834517644226,
+      "loss": 0.7736,
+      "mean_token_accuracy": 0.7669499337673187,
+      "num_tokens": 141405702.0,
+      "step": 41560
+    },
+    {
+      "epoch": 0.4172186759805693,
+      "learning_rate": 0.00011655827211048215,
+      "loss": 0.7295,
+      "mean_token_accuracy": 0.7826376080513,
+      "num_tokens": 141439029.0,
+      "step": 41570
+    },
+    {
+      "epoch": 0.41731904131036973,
+      "learning_rate": 0.00011653819904452205,
+      "loss": 0.7786,
+      "mean_token_accuracy": 0.7668355464935303,
+      "num_tokens": 141473522.0,
+      "step": 41580
+    },
+    {
+      "epoch": 0.41741940664017024,
+      "learning_rate": 0.00011651812597856197,
+      "loss": 0.7544,
+      "mean_token_accuracy": 0.7743890285491943,
+      "num_tokens": 141507059.0,
+      "step": 41590
+    },
+    {
+      "epoch": 0.4175197719699707,
+      "learning_rate": 0.00011649805291260188,
+      "loss": 0.7769,
+      "mean_token_accuracy": 0.7657489776611328,
+      "num_tokens": 141541134.0,
+      "step": 41600
+    },
+    {
+      "epoch": 0.41762013729977115,
+      "learning_rate": 0.00011647797984664177,
+      "loss": 0.7359,
+      "mean_token_accuracy": 0.7797170877456665,
+      "num_tokens": 141575496.0,
+      "step": 41610
+    },
+    {
+      "epoch": 0.41772050262957167,
+      "learning_rate": 0.0001164579067806817,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7683320343494415,
+      "num_tokens": 141610575.0,
+      "step": 41620
+    },
+    {
+      "epoch": 0.4178208679593721,
+      "learning_rate": 0.00011643783371472159,
+      "loss": 0.7502,
+      "mean_token_accuracy": 0.7801962435245514,
+      "num_tokens": 141645224.0,
+      "step": 41630
+    },
+    {
+      "epoch": 0.4179212332891726,
+      "learning_rate": 0.00011641776064876149,
+      "loss": 0.7639,
+      "mean_token_accuracy": 0.770879864692688,
+      "num_tokens": 141679721.0,
+      "step": 41640
+    },
+    {
+      "epoch": 0.4180215986189731,
+      "learning_rate": 0.00011639768758280141,
+      "loss": 0.7467,
+      "mean_token_accuracy": 0.779033613204956,
+      "num_tokens": 141713432.0,
+      "step": 41650
+    },
+    {
+      "epoch": 0.41812196394877355,
+      "learning_rate": 0.0001163776145168413,
+      "loss": 0.7842,
+      "mean_token_accuracy": 0.7664332866668702,
+      "num_tokens": 141747915.0,
+      "step": 41660
+    },
+    {
+      "epoch": 0.418222329278574,
+      "learning_rate": 0.0001163575414508812,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.7792607843875885,
+      "num_tokens": 141781878.0,
+      "step": 41670
+    },
+    {
+      "epoch": 0.41832269460837446,
+      "learning_rate": 0.00011633746838492112,
+      "loss": 0.7458,
+      "mean_token_accuracy": 0.7750950932502747,
+      "num_tokens": 141815215.0,
+      "step": 41680
+    },
+    {
+      "epoch": 0.41842305993817497,
+      "learning_rate": 0.00011631739531896102,
+      "loss": 0.7032,
+      "mean_token_accuracy": 0.7843857049942017,
+      "num_tokens": 141849495.0,
+      "step": 41690
+    },
+    {
+      "epoch": 0.4185234252679754,
+      "learning_rate": 0.00011629732225300094,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.776856005191803,
+      "num_tokens": 141883598.0,
+      "step": 41700
+    },
+    {
+      "epoch": 0.4186237905977759,
+      "learning_rate": 0.00011627724918704083,
+      "loss": 0.7578,
+      "mean_token_accuracy": 0.7720408916473389,
+      "num_tokens": 141917319.0,
+      "step": 41710
+    },
+    {
+      "epoch": 0.4187241559275764,
+      "learning_rate": 0.00011625717612108074,
+      "loss": 0.77,
+      "mean_token_accuracy": 0.775605422258377,
+      "num_tokens": 141951575.0,
+      "step": 41720
+    },
+    {
+      "epoch": 0.41882452125737685,
+      "learning_rate": 0.00011623710305512065,
+      "loss": 0.7508,
+      "mean_token_accuracy": 0.7744235992431641,
+      "num_tokens": 141985525.0,
+      "step": 41730
+    },
+    {
+      "epoch": 0.4189248865871773,
+      "learning_rate": 0.00011621702998916056,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.7771193206310272,
+      "num_tokens": 142018386.0,
+      "step": 41740
+    },
+    {
+      "epoch": 0.4190252519169778,
+      "learning_rate": 0.00011619695692320045,
+      "loss": 0.7218,
+      "mean_token_accuracy": 0.7799361944198608,
+      "num_tokens": 142053318.0,
+      "step": 41750
+    },
+    {
+      "epoch": 0.41912561724677827,
+      "learning_rate": 0.00011617688385724037,
+      "loss": 0.7374,
+      "mean_token_accuracy": 0.7763052105903625,
+      "num_tokens": 142087398.0,
+      "step": 41760
+    },
+    {
+      "epoch": 0.4192259825765787,
+      "learning_rate": 0.00011615681079128027,
+      "loss": 0.7316,
+      "mean_token_accuracy": 0.7780951380729675,
+      "num_tokens": 142122626.0,
+      "step": 41770
+    },
+    {
+      "epoch": 0.41932634790637924,
+      "learning_rate": 0.00011613673772532016,
+      "loss": 0.7343,
+      "mean_token_accuracy": 0.7779183864593506,
+      "num_tokens": 142156233.0,
+      "step": 41780
+    },
+    {
+      "epoch": 0.4194267132361797,
+      "learning_rate": 0.00011611666465936009,
+      "loss": 0.7527,
+      "mean_token_accuracy": 0.7760959804058075,
+      "num_tokens": 142190636.0,
+      "step": 41790
+    },
+    {
+      "epoch": 0.41952707856598015,
+      "learning_rate": 0.00011609659159339998,
+      "loss": 0.7405,
+      "mean_token_accuracy": 0.7813480198383331,
+      "num_tokens": 142224812.0,
+      "step": 41800
+    },
+    {
+      "epoch": 0.41962744389578066,
+      "learning_rate": 0.00011607651852743988,
+      "loss": 0.7321,
+      "mean_token_accuracy": 0.7838366448879241,
+      "num_tokens": 142258952.0,
+      "step": 41810
+    },
+    {
+      "epoch": 0.4197278092255811,
+      "learning_rate": 0.0001160564454614798,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7750254154205323,
+      "num_tokens": 142292930.0,
+      "step": 41820
+    },
+    {
+      "epoch": 0.4198281745553816,
+      "learning_rate": 0.00011603637239551969,
+      "loss": 0.7439,
+      "mean_token_accuracy": 0.7807756662368774,
+      "num_tokens": 142326719.0,
+      "step": 41830
+    },
+    {
+      "epoch": 0.4199285398851821,
+      "learning_rate": 0.0001160162993295596,
+      "loss": 0.7823,
+      "mean_token_accuracy": 0.7710380792617798,
+      "num_tokens": 142359477.0,
+      "step": 41840
+    },
+    {
+      "epoch": 0.42002890521498254,
+      "learning_rate": 0.00011599622626359952,
+      "loss": 0.749,
+      "mean_token_accuracy": 0.777097511291504,
+      "num_tokens": 142393997.0,
+      "step": 41850
+    },
+    {
+      "epoch": 0.420129270544783,
+      "learning_rate": 0.00011597615319763942,
+      "loss": 0.7514,
+      "mean_token_accuracy": 0.7785601139068603,
+      "num_tokens": 142427956.0,
+      "step": 41860
+    },
+    {
+      "epoch": 0.4202296358745835,
+      "learning_rate": 0.00011595608013167931,
+      "loss": 0.789,
+      "mean_token_accuracy": 0.7634051084518433,
+      "num_tokens": 142461618.0,
+      "step": 41870
+    },
+    {
+      "epoch": 0.42033000120438396,
+      "learning_rate": 0.00011593600706571923,
+      "loss": 0.759,
+      "mean_token_accuracy": 0.7732995927333832,
+      "num_tokens": 142495121.0,
+      "step": 41880
+    },
+    {
+      "epoch": 0.4204303665341844,
+      "learning_rate": 0.00011591593399975913,
+      "loss": 0.7259,
+      "mean_token_accuracy": 0.7791109800338745,
+      "num_tokens": 142529111.0,
+      "step": 41890
+    },
+    {
+      "epoch": 0.42053073186398493,
+      "learning_rate": 0.00011589586093379902,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.7657994270324707,
+      "num_tokens": 142562708.0,
+      "step": 41900
+    },
+    {
+      "epoch": 0.4206310971937854,
+      "learning_rate": 0.00011587578786783895,
+      "loss": 0.731,
+      "mean_token_accuracy": 0.7786415398120881,
+      "num_tokens": 142597009.0,
+      "step": 41910
+    },
+    {
+      "epoch": 0.42073146252358584,
+      "learning_rate": 0.00011585571480187884,
+      "loss": 0.7478,
+      "mean_token_accuracy": 0.7720061779022217,
+      "num_tokens": 142629882.0,
+      "step": 41920
+    },
+    {
+      "epoch": 0.42083182785338635,
+      "learning_rate": 0.00011583564173591874,
+      "loss": 0.728,
+      "mean_token_accuracy": 0.7804020464420318,
+      "num_tokens": 142663262.0,
+      "step": 41930
+    },
+    {
+      "epoch": 0.4209321931831868,
+      "learning_rate": 0.00011581556866995866,
+      "loss": 0.7389,
+      "mean_token_accuracy": 0.7790018320083618,
+      "num_tokens": 142697552.0,
+      "step": 41940
+    },
+    {
+      "epoch": 0.42103255851298726,
+      "learning_rate": 0.00011579549560399857,
+      "loss": 0.7374,
+      "mean_token_accuracy": 0.780879122018814,
+      "num_tokens": 142731175.0,
+      "step": 41950
+    },
+    {
+      "epoch": 0.4211329238427877,
+      "learning_rate": 0.00011577542253803846,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.7732527136802674,
+      "num_tokens": 142766072.0,
+      "step": 41960
+    },
+    {
+      "epoch": 0.42123328917258823,
+      "learning_rate": 0.00011575534947207838,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7731917262077331,
+      "num_tokens": 142800031.0,
+      "step": 41970
+    },
+    {
+      "epoch": 0.4213336545023887,
+      "learning_rate": 0.00011573527640611828,
+      "loss": 0.7428,
+      "mean_token_accuracy": 0.7734259366989136,
+      "num_tokens": 142832827.0,
+      "step": 41980
+    },
+    {
+      "epoch": 0.42143401983218914,
+      "learning_rate": 0.00011571520334015817,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7777356564998626,
+      "num_tokens": 142866975.0,
+      "step": 41990
+    },
+    {
+      "epoch": 0.42153438516198966,
+      "learning_rate": 0.0001156951302741981,
+      "loss": 0.7223,
+      "mean_token_accuracy": 0.7811121463775634,
+      "num_tokens": 142900069.0,
+      "step": 42000
+    },
+    {
+      "epoch": 0.4216347504917901,
+      "learning_rate": 0.00011567505720823799,
+      "loss": 0.7568,
+      "mean_token_accuracy": 0.7768263280391693,
+      "num_tokens": 142933803.0,
+      "step": 42010
+    },
+    {
+      "epoch": 0.42173511582159057,
+      "learning_rate": 0.00011565498414227789,
+      "loss": 0.7739,
+      "mean_token_accuracy": 0.7687878906726837,
+      "num_tokens": 142968125.0,
+      "step": 42020
+    },
+    {
+      "epoch": 0.4218354811513911,
+      "learning_rate": 0.00011563491107631781,
+      "loss": 0.7476,
+      "mean_token_accuracy": 0.7753892004489898,
+      "num_tokens": 143002152.0,
+      "step": 42030
+    },
+    {
+      "epoch": 0.42193584648119153,
+      "learning_rate": 0.0001156148380103577,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.768572723865509,
+      "num_tokens": 143036434.0,
+      "step": 42040
+    },
+    {
+      "epoch": 0.422036211810992,
+      "learning_rate": 0.00011559476494439761,
+      "loss": 0.7378,
+      "mean_token_accuracy": 0.7769821345806122,
+      "num_tokens": 143070746.0,
+      "step": 42050
+    },
+    {
+      "epoch": 0.4221365771407925,
+      "learning_rate": 0.00011557469187843752,
+      "loss": 0.7258,
+      "mean_token_accuracy": 0.7808934569358825,
+      "num_tokens": 143103651.0,
+      "step": 42060
+    },
+    {
+      "epoch": 0.42223694247059296,
+      "learning_rate": 0.00011555461881247743,
+      "loss": 0.7374,
+      "mean_token_accuracy": 0.7748523354530334,
+      "num_tokens": 143137910.0,
+      "step": 42070
+    },
+    {
+      "epoch": 0.4223373078003934,
+      "learning_rate": 0.00011553454574651732,
+      "loss": 0.7479,
+      "mean_token_accuracy": 0.7740685045719147,
+      "num_tokens": 143171565.0,
+      "step": 42080
+    },
+    {
+      "epoch": 0.4224376731301939,
+      "learning_rate": 0.00011551447268055724,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.7703144967555999,
+      "num_tokens": 143205989.0,
+      "step": 42090
+    },
+    {
+      "epoch": 0.4225380384599944,
+      "learning_rate": 0.00011549439961459714,
+      "loss": 0.7799,
+      "mean_token_accuracy": 0.7680784404277802,
+      "num_tokens": 143240141.0,
+      "step": 42100
+    },
+    {
+      "epoch": 0.42263840378979484,
+      "learning_rate": 0.00011547432654863703,
+      "loss": 0.7363,
+      "mean_token_accuracy": 0.777568644285202,
+      "num_tokens": 143275019.0,
+      "step": 42110
+    },
+    {
+      "epoch": 0.42273876911959535,
+      "learning_rate": 0.00011545425348267696,
+      "loss": 0.7591,
+      "mean_token_accuracy": 0.7781234681606293,
+      "num_tokens": 143309493.0,
+      "step": 42120
+    },
+    {
+      "epoch": 0.4228391344493958,
+      "learning_rate": 0.00011543418041671685,
+      "loss": 0.7544,
+      "mean_token_accuracy": 0.7739103138446808,
+      "num_tokens": 143343264.0,
+      "step": 42130
+    },
+    {
+      "epoch": 0.42293949977919626,
+      "learning_rate": 0.00011541410735075675,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7781742632389068,
+      "num_tokens": 143376970.0,
+      "step": 42140
+    },
+    {
+      "epoch": 0.42303986510899677,
+      "learning_rate": 0.00011539403428479667,
+      "loss": 0.7552,
+      "mean_token_accuracy": 0.7751582503318787,
+      "num_tokens": 143410449.0,
+      "step": 42150
+    },
+    {
+      "epoch": 0.4231402304387972,
+      "learning_rate": 0.00011537396121883656,
+      "loss": 0.7557,
+      "mean_token_accuracy": 0.7700955033302307,
+      "num_tokens": 143444118.0,
+      "step": 42160
+    },
+    {
+      "epoch": 0.4232405957685977,
+      "learning_rate": 0.00011535388815287647,
+      "loss": 0.7375,
+      "mean_token_accuracy": 0.771321439743042,
+      "num_tokens": 143477535.0,
+      "step": 42170
+    },
+    {
+      "epoch": 0.4233409610983982,
+      "learning_rate": 0.00011533381508691638,
+      "loss": 0.7311,
+      "mean_token_accuracy": 0.777759313583374,
+      "num_tokens": 143511652.0,
+      "step": 42180
+    },
+    {
+      "epoch": 0.42344132642819865,
+      "learning_rate": 0.00011531374202095629,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.7791213870048523,
+      "num_tokens": 143545525.0,
+      "step": 42190
+    },
+    {
+      "epoch": 0.4235416917579991,
+      "learning_rate": 0.00011529366895499618,
+      "loss": 0.7633,
+      "mean_token_accuracy": 0.7726755797863006,
+      "num_tokens": 143579727.0,
+      "step": 42200
+    },
+    {
+      "epoch": 0.4236420570877996,
+      "learning_rate": 0.0001152735958890361,
+      "loss": 0.7438,
+      "mean_token_accuracy": 0.7750461578369141,
+      "num_tokens": 143614240.0,
+      "step": 42210
+    },
+    {
+      "epoch": 0.4237424224176001,
+      "learning_rate": 0.000115253522823076,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7775866985321045,
+      "num_tokens": 143647665.0,
+      "step": 42220
+    },
+    {
+      "epoch": 0.42384278774740053,
+      "learning_rate": 0.0001152334497571159,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7775680303573609,
+      "num_tokens": 143681430.0,
+      "step": 42230
+    },
+    {
+      "epoch": 0.423943153077201,
+      "learning_rate": 0.00011521337669115582,
+      "loss": 0.7568,
+      "mean_token_accuracy": 0.77844118475914,
+      "num_tokens": 143715338.0,
+      "step": 42240
+    },
+    {
+      "epoch": 0.4240435184070015,
+      "learning_rate": 0.00011519330362519571,
+      "loss": 0.782,
+      "mean_token_accuracy": 0.7664641857147216,
+      "num_tokens": 143750014.0,
+      "step": 42250
+    },
+    {
+      "epoch": 0.42414388373680195,
+      "learning_rate": 0.00011517323055923563,
+      "loss": 0.7461,
+      "mean_token_accuracy": 0.7792369246482849,
+      "num_tokens": 143784996.0,
+      "step": 42260
+    },
+    {
+      "epoch": 0.4242442490666024,
+      "learning_rate": 0.00011515315749327553,
+      "loss": 0.7664,
+      "mean_token_accuracy": 0.7747955799102784,
+      "num_tokens": 143819719.0,
+      "step": 42270
+    },
+    {
+      "epoch": 0.4243446143964029,
+      "learning_rate": 0.00011513308442731542,
+      "loss": 0.7531,
+      "mean_token_accuracy": 0.7744141519069672,
+      "num_tokens": 143853169.0,
+      "step": 42280
+    },
+    {
+      "epoch": 0.4244449797262034,
+      "learning_rate": 0.00011511301136135535,
+      "loss": 0.7568,
+      "mean_token_accuracy": 0.7710319399833679,
+      "num_tokens": 143886625.0,
+      "step": 42290
+    },
+    {
+      "epoch": 0.42454534505600383,
+      "learning_rate": 0.00011509293829539525,
+      "loss": 0.7635,
+      "mean_token_accuracy": 0.7699625313282012,
+      "num_tokens": 143920842.0,
+      "step": 42300
+    },
+    {
+      "epoch": 0.42464571038580434,
+      "learning_rate": 0.00011507286522943515,
+      "loss": 0.7649,
+      "mean_token_accuracy": 0.769263744354248,
+      "num_tokens": 143954610.0,
+      "step": 42310
+    },
+    {
+      "epoch": 0.4247460757156048,
+      "learning_rate": 0.00011505279216347507,
+      "loss": 0.7878,
+      "mean_token_accuracy": 0.7658296942710876,
+      "num_tokens": 143988125.0,
+      "step": 42320
+    },
+    {
+      "epoch": 0.42484644104540525,
+      "learning_rate": 0.00011503271909751497,
+      "loss": 0.7302,
+      "mean_token_accuracy": 0.7820160806179046,
+      "num_tokens": 144022118.0,
+      "step": 42330
+    },
+    {
+      "epoch": 0.42494680637520577,
+      "learning_rate": 0.00011501264603155486,
+      "loss": 0.768,
+      "mean_token_accuracy": 0.7747110068798065,
+      "num_tokens": 144057330.0,
+      "step": 42340
+    },
+    {
+      "epoch": 0.4250471717050062,
+      "learning_rate": 0.00011499257296559478,
+      "loss": 0.7311,
+      "mean_token_accuracy": 0.7816978096961975,
+      "num_tokens": 144091710.0,
+      "step": 42350
+    },
+    {
+      "epoch": 0.4251475370348067,
+      "learning_rate": 0.00011497249989963468,
+      "loss": 0.7477,
+      "mean_token_accuracy": 0.7742366671562195,
+      "num_tokens": 144125285.0,
+      "step": 42360
+    },
+    {
+      "epoch": 0.4252479023646072,
+      "learning_rate": 0.00011495242683367457,
+      "loss": 0.7863,
+      "mean_token_accuracy": 0.769205504655838,
+      "num_tokens": 144159023.0,
+      "step": 42370
+    },
+    {
+      "epoch": 0.42534826769440764,
+      "learning_rate": 0.0001149323537677145,
+      "loss": 0.7348,
+      "mean_token_accuracy": 0.7731286525726319,
+      "num_tokens": 144193543.0,
+      "step": 42380
+    },
+    {
+      "epoch": 0.4254486330242081,
+      "learning_rate": 0.00011491228070175439,
+      "loss": 0.752,
+      "mean_token_accuracy": 0.7733675420284272,
+      "num_tokens": 144227794.0,
+      "step": 42390
+    },
+    {
+      "epoch": 0.4255489983540086,
+      "learning_rate": 0.0001148922076357943,
+      "loss": 0.7608,
+      "mean_token_accuracy": 0.7742749333381653,
+      "num_tokens": 144261510.0,
+      "step": 42400
+    },
+    {
+      "epoch": 0.42564936368380907,
+      "learning_rate": 0.0001148721345698342,
+      "loss": 0.7561,
+      "mean_token_accuracy": 0.7709795594215393,
+      "num_tokens": 144295409.0,
+      "step": 42410
+    },
+    {
+      "epoch": 0.4257497290136095,
+      "learning_rate": 0.00011485206150387411,
+      "loss": 0.7287,
+      "mean_token_accuracy": 0.7776812016963959,
+      "num_tokens": 144328163.0,
+      "step": 42420
+    },
+    {
+      "epoch": 0.42585009434341003,
+      "learning_rate": 0.00011483198843791401,
+      "loss": 0.791,
+      "mean_token_accuracy": 0.7648392498493195,
+      "num_tokens": 144362159.0,
+      "step": 42430
+    },
+    {
+      "epoch": 0.4259504596732105,
+      "learning_rate": 0.00011481191537195393,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7748533010482788,
+      "num_tokens": 144396163.0,
+      "step": 42440
+    },
+    {
+      "epoch": 0.42605082500301095,
+      "learning_rate": 0.00011479184230599383,
+      "loss": 0.7505,
+      "mean_token_accuracy": 0.7726286232471467,
+      "num_tokens": 144429723.0,
+      "step": 42450
+    },
+    {
+      "epoch": 0.42615119033281146,
+      "learning_rate": 0.00011477176924003372,
+      "loss": 0.7231,
+      "mean_token_accuracy": 0.7830985367298127,
+      "num_tokens": 144463855.0,
+      "step": 42460
+    },
+    {
+      "epoch": 0.4262515556626119,
+      "learning_rate": 0.00011475169617407364,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.777648288011551,
+      "num_tokens": 144497794.0,
+      "step": 42470
+    },
+    {
+      "epoch": 0.42635192099241237,
+      "learning_rate": 0.00011473162310811354,
+      "loss": 0.7313,
+      "mean_token_accuracy": 0.7775257229804993,
+      "num_tokens": 144531243.0,
+      "step": 42480
+    },
+    {
+      "epoch": 0.4264522863222129,
+      "learning_rate": 0.00011471155004215343,
+      "loss": 0.762,
+      "mean_token_accuracy": 0.7685877799987793,
+      "num_tokens": 144564993.0,
+      "step": 42490
+    },
+    {
+      "epoch": 0.42655265165201334,
+      "learning_rate": 0.00011469147697619335,
+      "loss": 0.7743,
+      "mean_token_accuracy": 0.7730763196945191,
+      "num_tokens": 144598872.0,
+      "step": 42500
+    },
+    {
+      "epoch": 0.4266530169818138,
+      "learning_rate": 0.00011467140391023325,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.7757616639137268,
+      "num_tokens": 144633256.0,
+      "step": 42510
+    },
+    {
+      "epoch": 0.42675338231161425,
+      "learning_rate": 0.00011465133084427316,
+      "loss": 0.7642,
+      "mean_token_accuracy": 0.7709073722362518,
+      "num_tokens": 144667867.0,
+      "step": 42520
+    },
+    {
+      "epoch": 0.42685374764141476,
+      "learning_rate": 0.00011463125777831307,
+      "loss": 0.7775,
+      "mean_token_accuracy": 0.7658672034740448,
+      "num_tokens": 144701722.0,
+      "step": 42530
+    },
+    {
+      "epoch": 0.4269541129712152,
+      "learning_rate": 0.00011461118471235297,
+      "loss": 0.7499,
+      "mean_token_accuracy": 0.7720082879066468,
+      "num_tokens": 144735854.0,
+      "step": 42540
+    },
+    {
+      "epoch": 0.42705447830101567,
+      "learning_rate": 0.00011459111164639287,
+      "loss": 0.7797,
+      "mean_token_accuracy": 0.7656648099422455,
+      "num_tokens": 144769258.0,
+      "step": 42550
+    },
+    {
+      "epoch": 0.4271548436308162,
+      "learning_rate": 0.00011457103858043279,
+      "loss": 0.7564,
+      "mean_token_accuracy": 0.7687288165092468,
+      "num_tokens": 144802998.0,
+      "step": 42560
+    },
+    {
+      "epoch": 0.42725520896061664,
+      "learning_rate": 0.00011455096551447269,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.769432681798935,
+      "num_tokens": 144837811.0,
+      "step": 42570
+    },
+    {
+      "epoch": 0.4273555742904171,
+      "learning_rate": 0.00011453089244851258,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.774343591928482,
+      "num_tokens": 144871769.0,
+      "step": 42580
+    },
+    {
+      "epoch": 0.4274559396202176,
+      "learning_rate": 0.0001145108193825525,
+      "loss": 0.7432,
+      "mean_token_accuracy": 0.7731312811374664,
+      "num_tokens": 144905705.0,
+      "step": 42590
+    },
+    {
+      "epoch": 0.42755630495001806,
+      "learning_rate": 0.0001144907463165924,
+      "loss": 0.7228,
+      "mean_token_accuracy": 0.7832659125328064,
+      "num_tokens": 144939609.0,
+      "step": 42600
+    },
+    {
+      "epoch": 0.4276566702798185,
+      "learning_rate": 0.0001144706732506323,
+      "loss": 0.7175,
+      "mean_token_accuracy": 0.7817785501480102,
+      "num_tokens": 144974461.0,
+      "step": 42610
+    },
+    {
+      "epoch": 0.42775703560961903,
+      "learning_rate": 0.00011445060018467222,
+      "loss": 0.7643,
+      "mean_token_accuracy": 0.7768852293491364,
+      "num_tokens": 145008987.0,
+      "step": 42620
+    },
+    {
+      "epoch": 0.4278574009394195,
+      "learning_rate": 0.00011443052711871211,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.772595363855362,
+      "num_tokens": 145043081.0,
+      "step": 42630
+    },
+    {
+      "epoch": 0.42795776626921994,
+      "learning_rate": 0.00011441045405275202,
+      "loss": 0.7788,
+      "mean_token_accuracy": 0.7669863939285279,
+      "num_tokens": 145076597.0,
+      "step": 42640
+    },
+    {
+      "epoch": 0.42805813159902045,
+      "learning_rate": 0.00011439038098679194,
+      "loss": 0.7572,
+      "mean_token_accuracy": 0.7708061218261719,
+      "num_tokens": 145110220.0,
+      "step": 42650
+    },
+    {
+      "epoch": 0.4281584969288209,
+      "learning_rate": 0.00011437030792083184,
+      "loss": 0.7696,
+      "mean_token_accuracy": 0.7687025308609009,
+      "num_tokens": 145144375.0,
+      "step": 42660
+    },
+    {
+      "epoch": 0.42825886225862136,
+      "learning_rate": 0.00011435023485487173,
+      "loss": 0.7492,
+      "mean_token_accuracy": 0.7752351701259613,
+      "num_tokens": 145178694.0,
+      "step": 42670
+    },
+    {
+      "epoch": 0.4283592275884219,
+      "learning_rate": 0.00011433016178891165,
+      "loss": 0.746,
+      "mean_token_accuracy": 0.7760239541530609,
+      "num_tokens": 145212679.0,
+      "step": 42680
+    },
+    {
+      "epoch": 0.42845959291822233,
+      "learning_rate": 0.00011431008872295155,
+      "loss": 0.7762,
+      "mean_token_accuracy": 0.7718257665634155,
+      "num_tokens": 145246193.0,
+      "step": 42690
+    },
+    {
+      "epoch": 0.4285599582480228,
+      "learning_rate": 0.00011429001565699144,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7723347902297973,
+      "num_tokens": 145279682.0,
+      "step": 42700
+    },
+    {
+      "epoch": 0.4286603235778233,
+      "learning_rate": 0.00011426994259103136,
+      "loss": 0.7197,
+      "mean_token_accuracy": 0.7853658556938171,
+      "num_tokens": 145313919.0,
+      "step": 42710
+    },
+    {
+      "epoch": 0.42876068890762375,
+      "learning_rate": 0.00011424986952507126,
+      "loss": 0.765,
+      "mean_token_accuracy": 0.7730284512043,
+      "num_tokens": 145347406.0,
+      "step": 42720
+    },
+    {
+      "epoch": 0.4288610542374242,
+      "learning_rate": 0.00011422979645911115,
+      "loss": 0.7662,
+      "mean_token_accuracy": 0.7666055858135223,
+      "num_tokens": 145381587.0,
+      "step": 42730
+    },
+    {
+      "epoch": 0.4289614195672247,
+      "learning_rate": 0.00011420972339315108,
+      "loss": 0.7112,
+      "mean_token_accuracy": 0.7838475942611695,
+      "num_tokens": 145416177.0,
+      "step": 42740
+    },
+    {
+      "epoch": 0.4290617848970252,
+      "learning_rate": 0.00011418965032719098,
+      "loss": 0.7887,
+      "mean_token_accuracy": 0.7718412697315216,
+      "num_tokens": 145449761.0,
+      "step": 42750
+    },
+    {
+      "epoch": 0.42916215022682563,
+      "learning_rate": 0.00011416957726123088,
+      "loss": 0.7185,
+      "mean_token_accuracy": 0.7836577832698822,
+      "num_tokens": 145483524.0,
+      "step": 42760
+    },
+    {
+      "epoch": 0.42926251555662615,
+      "learning_rate": 0.0001141495041952708,
+      "loss": 0.7263,
+      "mean_token_accuracy": 0.7785545229911804,
+      "num_tokens": 145517639.0,
+      "step": 42770
+    },
+    {
+      "epoch": 0.4293628808864266,
+      "learning_rate": 0.0001141294311293107,
+      "loss": 0.7382,
+      "mean_token_accuracy": 0.7751895666122437,
+      "num_tokens": 145552476.0,
+      "step": 42780
+    },
+    {
+      "epoch": 0.42946324621622706,
+      "learning_rate": 0.00011410935806335059,
+      "loss": 0.7589,
+      "mean_token_accuracy": 0.7743195414543151,
+      "num_tokens": 145585953.0,
+      "step": 42790
+    },
+    {
+      "epoch": 0.4295636115460275,
+      "learning_rate": 0.00011408928499739051,
+      "loss": 0.7299,
+      "mean_token_accuracy": 0.78284832239151,
+      "num_tokens": 145619705.0,
+      "step": 42800
+    },
+    {
+      "epoch": 0.429663976875828,
+      "learning_rate": 0.00011406921193143041,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.7725806474685669,
+      "num_tokens": 145653006.0,
+      "step": 42810
+    },
+    {
+      "epoch": 0.4297643422056285,
+      "learning_rate": 0.0001140491388654703,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7755830824375153,
+      "num_tokens": 145687064.0,
+      "step": 42820
+    },
+    {
+      "epoch": 0.42986470753542894,
+      "learning_rate": 0.00011402906579951022,
+      "loss": 0.7784,
+      "mean_token_accuracy": 0.7719202518463135,
+      "num_tokens": 145720957.0,
+      "step": 42830
+    },
+    {
+      "epoch": 0.42996507286522945,
+      "learning_rate": 0.00011400899273355012,
+      "loss": 0.7606,
+      "mean_token_accuracy": 0.7707092642784119,
+      "num_tokens": 145754449.0,
+      "step": 42840
+    },
+    {
+      "epoch": 0.4300654381950299,
+      "learning_rate": 0.00011398891966759004,
+      "loss": 0.7515,
+      "mean_token_accuracy": 0.7736448407173157,
+      "num_tokens": 145787727.0,
+      "step": 42850
+    },
+    {
+      "epoch": 0.43016580352483036,
+      "learning_rate": 0.00011396884660162994,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.7771518647670745,
+      "num_tokens": 145821506.0,
+      "step": 42860
+    },
+    {
+      "epoch": 0.43026616885463087,
+      "learning_rate": 0.00011394877353566984,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.7683781266212464,
+      "num_tokens": 145854865.0,
+      "step": 42870
+    },
+    {
+      "epoch": 0.4303665341844313,
+      "learning_rate": 0.00011392870046970975,
+      "loss": 0.7702,
+      "mean_token_accuracy": 0.7732557237148285,
+      "num_tokens": 145889360.0,
+      "step": 42880
+    },
+    {
+      "epoch": 0.4304668995142318,
+      "learning_rate": 0.00011390862740374966,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7752172887325287,
+      "num_tokens": 145923478.0,
+      "step": 42890
+    },
+    {
+      "epoch": 0.4305672648440323,
+      "learning_rate": 0.00011388855433778956,
+      "loss": 0.7983,
+      "mean_token_accuracy": 0.762095820903778,
+      "num_tokens": 145958344.0,
+      "step": 42900
+    },
+    {
+      "epoch": 0.43066763017383275,
+      "learning_rate": 0.00011386848127182948,
+      "loss": 0.766,
+      "mean_token_accuracy": 0.7728649318218231,
+      "num_tokens": 145992565.0,
+      "step": 42910
+    },
+    {
+      "epoch": 0.4307679955036332,
+      "learning_rate": 0.00011384840820586937,
+      "loss": 0.7023,
+      "mean_token_accuracy": 0.7814755737781525,
+      "num_tokens": 146025837.0,
+      "step": 42920
+    },
+    {
+      "epoch": 0.4308683608334337,
+      "learning_rate": 0.00011382833513990927,
+      "loss": 0.7496,
+      "mean_token_accuracy": 0.7717016398906708,
+      "num_tokens": 146060082.0,
+      "step": 42930
+    },
+    {
+      "epoch": 0.4309687261632342,
+      "learning_rate": 0.00011380826207394919,
+      "loss": 0.7506,
+      "mean_token_accuracy": 0.7746443390846253,
+      "num_tokens": 146093673.0,
+      "step": 42940
+    },
+    {
+      "epoch": 0.43106909149303463,
+      "learning_rate": 0.00011378818900798909,
+      "loss": 0.7262,
+      "mean_token_accuracy": 0.7759982824325562,
+      "num_tokens": 146127839.0,
+      "step": 42950
+    },
+    {
+      "epoch": 0.43116945682283514,
+      "learning_rate": 0.00011376811594202898,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.7662631928920746,
+      "num_tokens": 146161483.0,
+      "step": 42960
+    },
+    {
+      "epoch": 0.4312698221526356,
+      "learning_rate": 0.0001137480428760689,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.7674487829208374,
+      "num_tokens": 146195755.0,
+      "step": 42970
+    },
+    {
+      "epoch": 0.43137018748243605,
+      "learning_rate": 0.0001137279698101088,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7768530964851379,
+      "num_tokens": 146229186.0,
+      "step": 42980
+    },
+    {
+      "epoch": 0.43147055281223656,
+      "learning_rate": 0.0001137078967441487,
+      "loss": 0.7498,
+      "mean_token_accuracy": 0.7748568475246429,
+      "num_tokens": 146263112.0,
+      "step": 42990
+    },
+    {
+      "epoch": 0.431570918142037,
+      "learning_rate": 0.00011368782367818863,
+      "loss": 0.7451,
+      "mean_token_accuracy": 0.7756645023822785,
+      "num_tokens": 146296685.0,
+      "step": 43000
+    },
+    {
+      "epoch": 0.4316712834718375,
+      "learning_rate": 0.00011366775061222852,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7742310643196106,
+      "num_tokens": 146330720.0,
+      "step": 43010
+    },
+    {
+      "epoch": 0.431771648801638,
+      "learning_rate": 0.00011364767754626842,
+      "loss": 0.7228,
+      "mean_token_accuracy": 0.780841052532196,
+      "num_tokens": 146365615.0,
+      "step": 43020
+    },
+    {
+      "epoch": 0.43187201413143844,
+      "learning_rate": 0.00011362760448030834,
+      "loss": 0.7349,
+      "mean_token_accuracy": 0.7785764336585999,
+      "num_tokens": 146399285.0,
+      "step": 43030
+    },
+    {
+      "epoch": 0.4319723794612389,
+      "learning_rate": 0.00011360753141434823,
+      "loss": 0.7332,
+      "mean_token_accuracy": 0.7761741042137146,
+      "num_tokens": 146433625.0,
+      "step": 43040
+    },
+    {
+      "epoch": 0.4320727447910394,
+      "learning_rate": 0.00011358745834838813,
+      "loss": 0.7322,
+      "mean_token_accuracy": 0.781286609172821,
+      "num_tokens": 146466849.0,
+      "step": 43050
+    },
+    {
+      "epoch": 0.43217311012083987,
+      "learning_rate": 0.00011356738528242805,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7679896533489228,
+      "num_tokens": 146500899.0,
+      "step": 43060
+    },
+    {
+      "epoch": 0.4322734754506403,
+      "learning_rate": 0.00011354731221646795,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.7758651614189148,
+      "num_tokens": 146535913.0,
+      "step": 43070
+    },
+    {
+      "epoch": 0.4323738407804408,
+      "learning_rate": 0.00011352723915050784,
+      "loss": 0.7364,
+      "mean_token_accuracy": 0.7757045984268188,
+      "num_tokens": 146570949.0,
+      "step": 43080
+    },
+    {
+      "epoch": 0.4324742061102413,
+      "learning_rate": 0.00011350716608454776,
+      "loss": 0.7528,
+      "mean_token_accuracy": 0.7761816620826721,
+      "num_tokens": 146605146.0,
+      "step": 43090
+    },
+    {
+      "epoch": 0.43257457144004174,
+      "learning_rate": 0.00011348709301858767,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.7720991194248199,
+      "num_tokens": 146638703.0,
+      "step": 43100
+    },
+    {
+      "epoch": 0.4326749367698422,
+      "learning_rate": 0.00011346701995262757,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7714218497276306,
+      "num_tokens": 146673208.0,
+      "step": 43110
+    },
+    {
+      "epoch": 0.4327753020996427,
+      "learning_rate": 0.00011344694688666749,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.768765789270401,
+      "num_tokens": 146706635.0,
+      "step": 43120
+    },
+    {
+      "epoch": 0.43287566742944317,
+      "learning_rate": 0.00011342687382070738,
+      "loss": 0.721,
+      "mean_token_accuracy": 0.7781619131565094,
+      "num_tokens": 146740682.0,
+      "step": 43130
+    },
+    {
+      "epoch": 0.4329760327592436,
+      "learning_rate": 0.00011340680075474728,
+      "loss": 0.7711,
+      "mean_token_accuracy": 0.7729579508304596,
+      "num_tokens": 146774162.0,
+      "step": 43140
+    },
+    {
+      "epoch": 0.43307639808904413,
+      "learning_rate": 0.0001133867276887872,
+      "loss": 0.7362,
+      "mean_token_accuracy": 0.7753993928432464,
+      "num_tokens": 146808242.0,
+      "step": 43150
+    },
+    {
+      "epoch": 0.4331767634188446,
+      "learning_rate": 0.0001133666546228271,
+      "loss": 0.7897,
+      "mean_token_accuracy": 0.7667720377445221,
+      "num_tokens": 146842093.0,
+      "step": 43160
+    },
+    {
+      "epoch": 0.43327712874864505,
+      "learning_rate": 0.00011334658155686699,
+      "loss": 0.7774,
+      "mean_token_accuracy": 0.7674240648746491,
+      "num_tokens": 146875708.0,
+      "step": 43170
+    },
+    {
+      "epoch": 0.43337749407844556,
+      "learning_rate": 0.00011332650849090691,
+      "loss": 0.7324,
+      "mean_token_accuracy": 0.7821752607822419,
+      "num_tokens": 146909808.0,
+      "step": 43180
+    },
+    {
+      "epoch": 0.433477859408246,
+      "learning_rate": 0.0001133064354249468,
+      "loss": 0.7317,
+      "mean_token_accuracy": 0.776114422082901,
+      "num_tokens": 146943677.0,
+      "step": 43190
+    },
+    {
+      "epoch": 0.43357822473804647,
+      "learning_rate": 0.00011328636235898671,
+      "loss": 0.7313,
+      "mean_token_accuracy": 0.7763771116733551,
+      "num_tokens": 146977457.0,
+      "step": 43200
+    },
+    {
+      "epoch": 0.433678590067847,
+      "learning_rate": 0.00011326628929302662,
+      "loss": 0.7248,
+      "mean_token_accuracy": 0.786783367395401,
+      "num_tokens": 147011235.0,
+      "step": 43210
+    },
+    {
+      "epoch": 0.43377895539764744,
+      "learning_rate": 0.00011324621622706653,
+      "loss": 0.7602,
+      "mean_token_accuracy": 0.7720573008060455,
+      "num_tokens": 147044449.0,
+      "step": 43220
+    },
+    {
+      "epoch": 0.4338793207274479,
+      "learning_rate": 0.00011322614316110643,
+      "loss": 0.7151,
+      "mean_token_accuracy": 0.7858515202999115,
+      "num_tokens": 147078592.0,
+      "step": 43230
+    },
+    {
+      "epoch": 0.4339796860572484,
+      "learning_rate": 0.00011320607009514635,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.7693959653377533,
+      "num_tokens": 147112695.0,
+      "step": 43240
+    },
+    {
+      "epoch": 0.43408005138704886,
+      "learning_rate": 0.00011318599702918624,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.7777252614498138,
+      "num_tokens": 147147202.0,
+      "step": 43250
+    },
+    {
+      "epoch": 0.4341804167168493,
+      "learning_rate": 0.00011316592396322614,
+      "loss": 0.7755,
+      "mean_token_accuracy": 0.7707893550395966,
+      "num_tokens": 147181151.0,
+      "step": 43260
+    },
+    {
+      "epoch": 0.4342807820466498,
+      "learning_rate": 0.00011314585089726606,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.7738515973091126,
+      "num_tokens": 147215481.0,
+      "step": 43270
+    },
+    {
+      "epoch": 0.4343811473764503,
+      "learning_rate": 0.00011312577783130596,
+      "loss": 0.766,
+      "mean_token_accuracy": 0.7724314451217651,
+      "num_tokens": 147248850.0,
+      "step": 43280
+    },
+    {
+      "epoch": 0.43448151270625074,
+      "learning_rate": 0.00011310570476534585,
+      "loss": 0.764,
+      "mean_token_accuracy": 0.7727940082550049,
+      "num_tokens": 147282441.0,
+      "step": 43290
+    },
+    {
+      "epoch": 0.43458187803605125,
+      "learning_rate": 0.00011308563169938577,
+      "loss": 0.7686,
+      "mean_token_accuracy": 0.7741721034049988,
+      "num_tokens": 147316660.0,
+      "step": 43300
+    },
+    {
+      "epoch": 0.4346822433658517,
+      "learning_rate": 0.00011306555863342567,
+      "loss": 0.7157,
+      "mean_token_accuracy": 0.7825799405574798,
+      "num_tokens": 147351036.0,
+      "step": 43310
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "learning_rate": 0.00011304548556746558,
+      "loss": 0.7505,
+      "mean_token_accuracy": 0.7806974291801453,
+      "num_tokens": 147384464.0,
+      "step": 43320
+    },
+    {
+      "epoch": 0.4348829740254527,
+      "learning_rate": 0.00011302541250150548,
+      "loss": 0.7794,
+      "mean_token_accuracy": 0.7736136972904205,
+      "num_tokens": 147418810.0,
+      "step": 43330
+    },
+    {
+      "epoch": 0.43498333935525313,
+      "learning_rate": 0.00011300533943554539,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7703942358493805,
+      "num_tokens": 147452464.0,
+      "step": 43340
+    },
+    {
+      "epoch": 0.4350837046850536,
+      "learning_rate": 0.00011298526636958529,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7802060544490814,
+      "num_tokens": 147486019.0,
+      "step": 43350
+    },
+    {
+      "epoch": 0.4351840700148541,
+      "learning_rate": 0.00011296519330362521,
+      "loss": 0.7777,
+      "mean_token_accuracy": 0.7689915657043457,
+      "num_tokens": 147519125.0,
+      "step": 43360
+    },
+    {
+      "epoch": 0.43528443534465455,
+      "learning_rate": 0.0001129451202376651,
+      "loss": 0.7309,
+      "mean_token_accuracy": 0.7763258755207062,
+      "num_tokens": 147552872.0,
+      "step": 43370
+    },
+    {
+      "epoch": 0.435384800674455,
+      "learning_rate": 0.000112925047171705,
+      "loss": 0.7458,
+      "mean_token_accuracy": 0.7751060128211975,
+      "num_tokens": 147586885.0,
+      "step": 43380
+    },
+    {
+      "epoch": 0.43548516600425546,
+      "learning_rate": 0.00011290497410574492,
+      "loss": 0.7917,
+      "mean_token_accuracy": 0.7648883879184722,
+      "num_tokens": 147621258.0,
+      "step": 43390
+    },
+    {
+      "epoch": 0.435585531334056,
+      "learning_rate": 0.00011288490103978482,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.7755088090896607,
+      "num_tokens": 147654561.0,
+      "step": 43400
+    },
+    {
+      "epoch": 0.43568589666385643,
+      "learning_rate": 0.00011286482797382474,
+      "loss": 0.7507,
+      "mean_token_accuracy": 0.7761448502540589,
+      "num_tokens": 147688265.0,
+      "step": 43410
+    },
+    {
+      "epoch": 0.4357862619936569,
+      "learning_rate": 0.00011284475490786463,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7711888194084168,
+      "num_tokens": 147722114.0,
+      "step": 43420
+    },
+    {
+      "epoch": 0.4358866273234574,
+      "learning_rate": 0.00011282468184190453,
+      "loss": 0.777,
+      "mean_token_accuracy": 0.7686375737190246,
+      "num_tokens": 147755892.0,
+      "step": 43430
+    },
+    {
+      "epoch": 0.43598699265325785,
+      "learning_rate": 0.00011280460877594445,
+      "loss": 0.778,
+      "mean_token_accuracy": 0.7742932558059692,
+      "num_tokens": 147789319.0,
+      "step": 43440
+    },
+    {
+      "epoch": 0.4360873579830583,
+      "learning_rate": 0.00011278453570998436,
+      "loss": 0.7649,
+      "mean_token_accuracy": 0.7734306454658508,
+      "num_tokens": 147822474.0,
+      "step": 43450
+    },
+    {
+      "epoch": 0.4361877233128588,
+      "learning_rate": 0.00011276446264402425,
+      "loss": 0.7415,
+      "mean_token_accuracy": 0.7750584840774536,
+      "num_tokens": 147856600.0,
+      "step": 43460
+    },
+    {
+      "epoch": 0.4362880886426593,
+      "learning_rate": 0.00011274438957806417,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.7757338583469391,
+      "num_tokens": 147889611.0,
+      "step": 43470
+    },
+    {
+      "epoch": 0.43638845397245973,
+      "learning_rate": 0.00011272431651210407,
+      "loss": 0.7761,
+      "mean_token_accuracy": 0.7676257133483887,
+      "num_tokens": 147924358.0,
+      "step": 43480
+    },
+    {
+      "epoch": 0.43648881930226024,
+      "learning_rate": 0.00011270424344614396,
+      "loss": 0.7474,
+      "mean_token_accuracy": 0.7697301208972931,
+      "num_tokens": 147958186.0,
+      "step": 43490
+    },
+    {
+      "epoch": 0.4365891846320607,
+      "learning_rate": 0.00011268417038018389,
+      "loss": 0.7327,
+      "mean_token_accuracy": 0.7785479128360748,
+      "num_tokens": 147992440.0,
+      "step": 43500
+    },
+    {
+      "epoch": 0.43668954996186116,
+      "learning_rate": 0.00011266409731422378,
+      "loss": 0.7158,
+      "mean_token_accuracy": 0.7759631454944611,
+      "num_tokens": 148026123.0,
+      "step": 43510
+    },
+    {
+      "epoch": 0.43678991529166167,
+      "learning_rate": 0.00011264402424826368,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.775789487361908,
+      "num_tokens": 148060170.0,
+      "step": 43520
+    },
+    {
+      "epoch": 0.4368902806214621,
+      "learning_rate": 0.0001126239511823036,
+      "loss": 0.7443,
+      "mean_token_accuracy": 0.7734920859336853,
+      "num_tokens": 148094738.0,
+      "step": 43530
+    },
+    {
+      "epoch": 0.4369906459512626,
+      "learning_rate": 0.00011260387811634349,
+      "loss": 0.7617,
+      "mean_token_accuracy": 0.7684929430484772,
+      "num_tokens": 148128157.0,
+      "step": 43540
+    },
+    {
+      "epoch": 0.4370910112810631,
+      "learning_rate": 0.0001125838050503834,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.776504111289978,
+      "num_tokens": 148163114.0,
+      "step": 43550
+    },
+    {
+      "epoch": 0.43719137661086355,
+      "learning_rate": 0.00011256373198442331,
+      "loss": 0.7764,
+      "mean_token_accuracy": 0.7684085130691528,
+      "num_tokens": 148196998.0,
+      "step": 43560
+    },
+    {
+      "epoch": 0.437291741940664,
+      "learning_rate": 0.00011254365891846322,
+      "loss": 0.7363,
+      "mean_token_accuracy": 0.7734168648719788,
+      "num_tokens": 148230922.0,
+      "step": 43570
+    },
+    {
+      "epoch": 0.4373921072704645,
+      "learning_rate": 0.00011252358585250311,
+      "loss": 0.7672,
+      "mean_token_accuracy": 0.7757761478424072,
+      "num_tokens": 148265034.0,
+      "step": 43580
+    },
+    {
+      "epoch": 0.43749247260026497,
+      "learning_rate": 0.00011250351278654304,
+      "loss": 0.7228,
+      "mean_token_accuracy": 0.780917638540268,
+      "num_tokens": 148299100.0,
+      "step": 43590
+    },
+    {
+      "epoch": 0.4375928379300654,
+      "learning_rate": 0.00011248343972058293,
+      "loss": 0.739,
+      "mean_token_accuracy": 0.7782020151615143,
+      "num_tokens": 148333913.0,
+      "step": 43600
+    },
+    {
+      "epoch": 0.43769320325986594,
+      "learning_rate": 0.00011246336665462283,
+      "loss": 0.763,
+      "mean_token_accuracy": 0.7715815305709839,
+      "num_tokens": 148367724.0,
+      "step": 43610
+    },
+    {
+      "epoch": 0.4377935685896664,
+      "learning_rate": 0.00011244329358866275,
+      "loss": 0.7528,
+      "mean_token_accuracy": 0.7737765967845917,
+      "num_tokens": 148401807.0,
+      "step": 43620
+    },
+    {
+      "epoch": 0.43789393391946685,
+      "learning_rate": 0.00011242322052270264,
+      "loss": 0.7465,
+      "mean_token_accuracy": 0.7716784477233887,
+      "num_tokens": 148435545.0,
+      "step": 43630
+    },
+    {
+      "epoch": 0.43799429924926736,
+      "learning_rate": 0.00011240314745674254,
+      "loss": 0.7552,
+      "mean_token_accuracy": 0.776634031534195,
+      "num_tokens": 148469623.0,
+      "step": 43640
+    },
+    {
+      "epoch": 0.4380946645790678,
+      "learning_rate": 0.00011238307439078246,
+      "loss": 0.7294,
+      "mean_token_accuracy": 0.7753891468048095,
+      "num_tokens": 148502671.0,
+      "step": 43650
+    },
+    {
+      "epoch": 0.43819502990886827,
+      "learning_rate": 0.00011236300132482235,
+      "loss": 0.7266,
+      "mean_token_accuracy": 0.779669314622879,
+      "num_tokens": 148536980.0,
+      "step": 43660
+    },
+    {
+      "epoch": 0.43829539523866873,
+      "learning_rate": 0.00011234292825886226,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7682388186454773,
+      "num_tokens": 148570675.0,
+      "step": 43670
+    },
+    {
+      "epoch": 0.43839576056846924,
+      "learning_rate": 0.00011232285519290217,
+      "loss": 0.7646,
+      "mean_token_accuracy": 0.7727665424346923,
+      "num_tokens": 148604730.0,
+      "step": 43680
+    },
+    {
+      "epoch": 0.4384961258982697,
+      "learning_rate": 0.00011230278212694208,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.7830365002155304,
+      "num_tokens": 148638983.0,
+      "step": 43690
+    },
+    {
+      "epoch": 0.43859649122807015,
+      "learning_rate": 0.00011228270906098197,
+      "loss": 0.7491,
+      "mean_token_accuracy": 0.7743676781654358,
+      "num_tokens": 148672923.0,
+      "step": 43700
+    },
+    {
+      "epoch": 0.43869685655787066,
+      "learning_rate": 0.0001122626359950219,
+      "loss": 0.7255,
+      "mean_token_accuracy": 0.7819778323173523,
+      "num_tokens": 148707032.0,
+      "step": 43710
+    },
+    {
+      "epoch": 0.4387972218876711,
+      "learning_rate": 0.00011224256292906179,
+      "loss": 0.7355,
+      "mean_token_accuracy": 0.7821864724159241,
+      "num_tokens": 148740780.0,
+      "step": 43720
+    },
+    {
+      "epoch": 0.4388975872174716,
+      "learning_rate": 0.00011222248986310169,
+      "loss": 0.7204,
+      "mean_token_accuracy": 0.7825488030910492,
+      "num_tokens": 148774642.0,
+      "step": 43730
+    },
+    {
+      "epoch": 0.4389979525472721,
+      "learning_rate": 0.00011220241679714161,
+      "loss": 0.7429,
+      "mean_token_accuracy": 0.7777960121631622,
+      "num_tokens": 148808512.0,
+      "step": 43740
+    },
+    {
+      "epoch": 0.43909831787707254,
+      "learning_rate": 0.0001121823437311815,
+      "loss": 0.7159,
+      "mean_token_accuracy": 0.7805725157260894,
+      "num_tokens": 148842492.0,
+      "step": 43750
+    },
+    {
+      "epoch": 0.439198683206873,
+      "learning_rate": 0.0001121622706652214,
+      "loss": 0.7439,
+      "mean_token_accuracy": 0.7803267896175384,
+      "num_tokens": 148876804.0,
+      "step": 43760
+    },
+    {
+      "epoch": 0.4392990485366735,
+      "learning_rate": 0.00011214219759926132,
+      "loss": 0.7601,
+      "mean_token_accuracy": 0.7764488101005554,
+      "num_tokens": 148910864.0,
+      "step": 43770
+    },
+    {
+      "epoch": 0.43939941386647396,
+      "learning_rate": 0.00011212212453330121,
+      "loss": 0.7685,
+      "mean_token_accuracy": 0.7743898093700409,
+      "num_tokens": 148945189.0,
+      "step": 43780
+    },
+    {
+      "epoch": 0.4394997791962744,
+      "learning_rate": 0.00011210205146734112,
+      "loss": 0.7247,
+      "mean_token_accuracy": 0.7831371545791626,
+      "num_tokens": 148980389.0,
+      "step": 43790
+    },
+    {
+      "epoch": 0.43960014452607493,
+      "learning_rate": 0.00011208197840138104,
+      "loss": 0.7657,
+      "mean_token_accuracy": 0.7732807636260987,
+      "num_tokens": 149014489.0,
+      "step": 43800
+    },
+    {
+      "epoch": 0.4397005098558754,
+      "learning_rate": 0.00011206190533542094,
+      "loss": 0.7768,
+      "mean_token_accuracy": 0.7680534720420837,
+      "num_tokens": 149048386.0,
+      "step": 43810
+    },
+    {
+      "epoch": 0.43980087518567584,
+      "learning_rate": 0.00011204183226946083,
+      "loss": 0.7605,
+      "mean_token_accuracy": 0.7759904325008392,
+      "num_tokens": 149083384.0,
+      "step": 43820
+    },
+    {
+      "epoch": 0.43990124051547635,
+      "learning_rate": 0.00011202175920350076,
+      "loss": 0.7334,
+      "mean_token_accuracy": 0.782331782579422,
+      "num_tokens": 149117597.0,
+      "step": 43830
+    },
+    {
+      "epoch": 0.4400016058452768,
+      "learning_rate": 0.00011200168613754065,
+      "loss": 0.7647,
+      "mean_token_accuracy": 0.7740571975708008,
+      "num_tokens": 149151403.0,
+      "step": 43840
+    },
+    {
+      "epoch": 0.44010197117507727,
+      "learning_rate": 0.00011198161307158055,
+      "loss": 0.7756,
+      "mean_token_accuracy": 0.7720715939998627,
+      "num_tokens": 149184826.0,
+      "step": 43850
+    },
+    {
+      "epoch": 0.4402023365048778,
+      "learning_rate": 0.00011196154000562047,
+      "loss": 0.7698,
+      "mean_token_accuracy": 0.7729233264923095,
+      "num_tokens": 149219488.0,
+      "step": 43860
+    },
+    {
+      "epoch": 0.44030270183467823,
+      "learning_rate": 0.00011194146693966036,
+      "loss": 0.7215,
+      "mean_token_accuracy": 0.7808591425418854,
+      "num_tokens": 149253761.0,
+      "step": 43870
+    },
+    {
+      "epoch": 0.4404030671644787,
+      "learning_rate": 0.00011192139387370026,
+      "loss": 0.7472,
+      "mean_token_accuracy": 0.7746274948120118,
+      "num_tokens": 149288748.0,
+      "step": 43880
+    },
+    {
+      "epoch": 0.4405034324942792,
+      "learning_rate": 0.00011190132080774018,
+      "loss": 0.7425,
+      "mean_token_accuracy": 0.7758005857467651,
+      "num_tokens": 149322532.0,
+      "step": 43890
+    },
+    {
+      "epoch": 0.44060379782407966,
+      "learning_rate": 0.00011188124774178009,
+      "loss": 0.7444,
+      "mean_token_accuracy": 0.778124886751175,
+      "num_tokens": 149357154.0,
+      "step": 43900
+    },
+    {
+      "epoch": 0.4407041631538801,
+      "learning_rate": 0.00011186117467581998,
+      "loss": 0.7631,
+      "mean_token_accuracy": 0.7752729535102845,
+      "num_tokens": 149391861.0,
+      "step": 43910
+    },
+    {
+      "epoch": 0.4408045284836806,
+      "learning_rate": 0.0001118411016098599,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.770001494884491,
+      "num_tokens": 149425672.0,
+      "step": 43920
+    },
+    {
+      "epoch": 0.4409048938134811,
+      "learning_rate": 0.0001118210285438998,
+      "loss": 0.7421,
+      "mean_token_accuracy": 0.7809359133243561,
+      "num_tokens": 149460223.0,
+      "step": 43930
+    },
+    {
+      "epoch": 0.44100525914328154,
+      "learning_rate": 0.0001118009554779397,
+      "loss": 0.7204,
+      "mean_token_accuracy": 0.7850530624389649,
+      "num_tokens": 149493781.0,
+      "step": 43940
+    },
+    {
+      "epoch": 0.441105624473082,
+      "learning_rate": 0.00011178088241197962,
+      "loss": 0.7252,
+      "mean_token_accuracy": 0.7797057867050171,
+      "num_tokens": 149528669.0,
+      "step": 43950
+    },
+    {
+      "epoch": 0.4412059898028825,
+      "learning_rate": 0.00011176080934601951,
+      "loss": 0.7568,
+      "mean_token_accuracy": 0.7771291673183441,
+      "num_tokens": 149562236.0,
+      "step": 43960
+    },
+    {
+      "epoch": 0.44130635513268296,
+      "learning_rate": 0.00011174073628005943,
+      "loss": 0.7623,
+      "mean_token_accuracy": 0.7755987465381622,
+      "num_tokens": 149596168.0,
+      "step": 43970
+    },
+    {
+      "epoch": 0.4414067204624834,
+      "learning_rate": 0.00011172066321409933,
+      "loss": 0.7263,
+      "mean_token_accuracy": 0.7827870607376098,
+      "num_tokens": 149630105.0,
+      "step": 43980
+    },
+    {
+      "epoch": 0.4415070857922839,
+      "learning_rate": 0.00011170059014813922,
+      "loss": 0.7481,
+      "mean_token_accuracy": 0.7742818355560303,
+      "num_tokens": 149664931.0,
+      "step": 43990
+    },
+    {
+      "epoch": 0.4416074511220844,
+      "learning_rate": 0.00011168051708217915,
+      "loss": 0.7752,
+      "mean_token_accuracy": 0.770356547832489,
+      "num_tokens": 149697876.0,
+      "step": 44000
+    },
+    {
+      "epoch": 0.44170781645188484,
+      "learning_rate": 0.00011166044401621904,
+      "loss": 0.7507,
+      "mean_token_accuracy": 0.7784092545509338,
+      "num_tokens": 149731738.0,
+      "step": 44010
+    },
+    {
+      "epoch": 0.44180818178168535,
+      "learning_rate": 0.00011164037095025895,
+      "loss": 0.7059,
+      "mean_token_accuracy": 0.7859094977378845,
+      "num_tokens": 149765648.0,
+      "step": 44020
+    },
+    {
+      "epoch": 0.4419085471114858,
+      "learning_rate": 0.00011162029788429886,
+      "loss": 0.762,
+      "mean_token_accuracy": 0.774725592136383,
+      "num_tokens": 149800114.0,
+      "step": 44030
+    },
+    {
+      "epoch": 0.44200891244128626,
+      "learning_rate": 0.00011160022481833877,
+      "loss": 0.731,
+      "mean_token_accuracy": 0.7784102320671081,
+      "num_tokens": 149834841.0,
+      "step": 44040
+    },
+    {
+      "epoch": 0.4421092777710868,
+      "learning_rate": 0.00011158015175237866,
+      "loss": 0.7157,
+      "mean_token_accuracy": 0.7827341020107269,
+      "num_tokens": 149868802.0,
+      "step": 44050
+    },
+    {
+      "epoch": 0.44220964310088723,
+      "learning_rate": 0.00011156007868641858,
+      "loss": 0.7534,
+      "mean_token_accuracy": 0.7758202314376831,
+      "num_tokens": 149902453.0,
+      "step": 44060
+    },
+    {
+      "epoch": 0.4423100084306877,
+      "learning_rate": 0.00011154000562045848,
+      "loss": 0.7614,
+      "mean_token_accuracy": 0.7681388676166534,
+      "num_tokens": 149937141.0,
+      "step": 44070
+    },
+    {
+      "epoch": 0.4424103737604882,
+      "learning_rate": 0.00011151993255449837,
+      "loss": 0.783,
+      "mean_token_accuracy": 0.7707764804363251,
+      "num_tokens": 149971447.0,
+      "step": 44080
+    },
+    {
+      "epoch": 0.44251073909028865,
+      "learning_rate": 0.0001114998594885383,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.7753174424171447,
+      "num_tokens": 150005271.0,
+      "step": 44090
+    },
+    {
+      "epoch": 0.4426111044200891,
+      "learning_rate": 0.00011147978642257819,
+      "loss": 0.7534,
+      "mean_token_accuracy": 0.7698142409324646,
+      "num_tokens": 150039632.0,
+      "step": 44100
+    },
+    {
+      "epoch": 0.4427114697498896,
+      "learning_rate": 0.00011145971335661808,
+      "loss": 0.7691,
+      "mean_token_accuracy": 0.7704128801822663,
+      "num_tokens": 150074429.0,
+      "step": 44110
+    },
+    {
+      "epoch": 0.4428118350796901,
+      "learning_rate": 0.000111439640290658,
+      "loss": 0.754,
+      "mean_token_accuracy": 0.767986536026001,
+      "num_tokens": 150108714.0,
+      "step": 44120
+    },
+    {
+      "epoch": 0.44291220040949053,
+      "learning_rate": 0.0001114195672246979,
+      "loss": 0.7313,
+      "mean_token_accuracy": 0.7801444411277771,
+      "num_tokens": 150143349.0,
+      "step": 44130
+    },
+    {
+      "epoch": 0.44301256573929104,
+      "learning_rate": 0.00011139949415873781,
+      "loss": 0.769,
+      "mean_token_accuracy": 0.7680265784263611,
+      "num_tokens": 150177716.0,
+      "step": 44140
+    },
+    {
+      "epoch": 0.4431129310690915,
+      "learning_rate": 0.00011137942109277773,
+      "loss": 0.7555,
+      "mean_token_accuracy": 0.7723772168159485,
+      "num_tokens": 150212615.0,
+      "step": 44150
+    },
+    {
+      "epoch": 0.44321329639889195,
+      "learning_rate": 0.00011135934802681763,
+      "loss": 0.7439,
+      "mean_token_accuracy": 0.7749141871929168,
+      "num_tokens": 150246173.0,
+      "step": 44160
+    },
+    {
+      "epoch": 0.44331366172869247,
+      "learning_rate": 0.00011133927496085752,
+      "loss": 0.7371,
+      "mean_token_accuracy": 0.7740356266498566,
+      "num_tokens": 150280009.0,
+      "step": 44170
+    },
+    {
+      "epoch": 0.4434140270584929,
+      "learning_rate": 0.00011131920189489744,
+      "loss": 0.7833,
+      "mean_token_accuracy": 0.7707145631313324,
+      "num_tokens": 150315302.0,
+      "step": 44180
+    },
+    {
+      "epoch": 0.4435143923882934,
+      "learning_rate": 0.00011129912882893734,
+      "loss": 0.7599,
+      "mean_token_accuracy": 0.772494101524353,
+      "num_tokens": 150349697.0,
+      "step": 44190
+    },
+    {
+      "epoch": 0.4436147577180939,
+      "learning_rate": 0.00011127905576297723,
+      "loss": 0.7576,
+      "mean_token_accuracy": 0.7743092060089112,
+      "num_tokens": 150384328.0,
+      "step": 44200
+    },
+    {
+      "epoch": 0.44371512304789434,
+      "learning_rate": 0.00011125898269701716,
+      "loss": 0.7171,
+      "mean_token_accuracy": 0.782273119688034,
+      "num_tokens": 150417859.0,
+      "step": 44210
+    },
+    {
+      "epoch": 0.4438154883776948,
+      "learning_rate": 0.00011123890963105705,
+      "loss": 0.7779,
+      "mean_token_accuracy": 0.7740826606750488,
+      "num_tokens": 150451836.0,
+      "step": 44220
+    },
+    {
+      "epoch": 0.44391585370749526,
+      "learning_rate": 0.00011121883656509694,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.7781951189041137,
+      "num_tokens": 150486201.0,
+      "step": 44230
+    },
+    {
+      "epoch": 0.44401621903729577,
+      "learning_rate": 0.00011119876349913687,
+      "loss": 0.7301,
+      "mean_token_accuracy": 0.7788292706012726,
+      "num_tokens": 150519669.0,
+      "step": 44240
+    },
+    {
+      "epoch": 0.4441165843670962,
+      "learning_rate": 0.00011117869043317678,
+      "loss": 0.7575,
+      "mean_token_accuracy": 0.7742524087429047,
+      "num_tokens": 150553112.0,
+      "step": 44250
+    },
+    {
+      "epoch": 0.4442169496968967,
+      "learning_rate": 0.00011115861736721667,
+      "loss": 0.7408,
+      "mean_token_accuracy": 0.7746887743473053,
+      "num_tokens": 150587479.0,
+      "step": 44260
+    },
+    {
+      "epoch": 0.4443173150266972,
+      "learning_rate": 0.00011113854430125659,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.7738109111785889,
+      "num_tokens": 150621557.0,
+      "step": 44270
+    },
+    {
+      "epoch": 0.44441768035649765,
+      "learning_rate": 0.00011111847123529649,
+      "loss": 0.7206,
+      "mean_token_accuracy": 0.7809482932090759,
+      "num_tokens": 150656884.0,
+      "step": 44280
+    },
+    {
+      "epoch": 0.4445180456862981,
+      "learning_rate": 0.00011109839816933638,
+      "loss": 0.7546,
+      "mean_token_accuracy": 0.775887793302536,
+      "num_tokens": 150690433.0,
+      "step": 44290
+    },
+    {
+      "epoch": 0.4446184110160986,
+      "learning_rate": 0.0001110783251033763,
+      "loss": 0.7631,
+      "mean_token_accuracy": 0.7731802225112915,
+      "num_tokens": 150724169.0,
+      "step": 44300
+    },
+    {
+      "epoch": 0.44471877634589907,
+      "learning_rate": 0.0001110582520374162,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7732235372066498,
+      "num_tokens": 150759354.0,
+      "step": 44310
+    },
+    {
+      "epoch": 0.4448191416756995,
+      "learning_rate": 0.0001110381789714561,
+      "loss": 0.7327,
+      "mean_token_accuracy": 0.7866859495639801,
+      "num_tokens": 150792697.0,
+      "step": 44320
+    },
+    {
+      "epoch": 0.44491950700550004,
+      "learning_rate": 0.00011101810590549602,
+      "loss": 0.7328,
+      "mean_token_accuracy": 0.7779637038707733,
+      "num_tokens": 150825950.0,
+      "step": 44330
+    },
+    {
+      "epoch": 0.4450198723353005,
+      "learning_rate": 0.00011099803283953591,
+      "loss": 0.714,
+      "mean_token_accuracy": 0.7852058053016663,
+      "num_tokens": 150859211.0,
+      "step": 44340
+    },
+    {
+      "epoch": 0.44512023766510095,
+      "learning_rate": 0.00011097795977357582,
+      "loss": 0.7416,
+      "mean_token_accuracy": 0.7797508060932159,
+      "num_tokens": 150892624.0,
+      "step": 44350
+    },
+    {
+      "epoch": 0.44522060299490146,
+      "learning_rate": 0.00011095788670761573,
+      "loss": 0.7918,
+      "mean_token_accuracy": 0.7642795860767364,
+      "num_tokens": 150926497.0,
+      "step": 44360
+    },
+    {
+      "epoch": 0.4453209683247019,
+      "learning_rate": 0.00011093781364165564,
+      "loss": 0.7533,
+      "mean_token_accuracy": 0.7732256591320038,
+      "num_tokens": 150960100.0,
+      "step": 44370
+    },
+    {
+      "epoch": 0.44542133365450237,
+      "learning_rate": 0.00011091774057569553,
+      "loss": 0.7976,
+      "mean_token_accuracy": 0.7641650021076203,
+      "num_tokens": 150994591.0,
+      "step": 44380
+    },
+    {
+      "epoch": 0.4455216989843029,
+      "learning_rate": 0.00011089766750973545,
+      "loss": 0.7487,
+      "mean_token_accuracy": 0.7758909225463867,
+      "num_tokens": 151028977.0,
+      "step": 44390
+    },
+    {
+      "epoch": 0.44562206431410334,
+      "learning_rate": 0.00011087759444377535,
+      "loss": 0.7403,
+      "mean_token_accuracy": 0.7733930587768555,
+      "num_tokens": 151063183.0,
+      "step": 44400
+    },
+    {
+      "epoch": 0.4457224296439038,
+      "learning_rate": 0.00011085752137781524,
+      "loss": 0.7305,
+      "mean_token_accuracy": 0.7791752755641937,
+      "num_tokens": 151097491.0,
+      "step": 44410
+    },
+    {
+      "epoch": 0.4458227949737043,
+      "learning_rate": 0.00011083744831185516,
+      "loss": 0.7367,
+      "mean_token_accuracy": 0.7794057905673981,
+      "num_tokens": 151131641.0,
+      "step": 44420
+    },
+    {
+      "epoch": 0.44592316030350476,
+      "learning_rate": 0.00011081737524589506,
+      "loss": 0.7263,
+      "mean_token_accuracy": 0.7804259061813354,
+      "num_tokens": 151166489.0,
+      "step": 44430
+    },
+    {
+      "epoch": 0.4460235256333052,
+      "learning_rate": 0.00011079730217993495,
+      "loss": 0.7809,
+      "mean_token_accuracy": 0.7714639127254486,
+      "num_tokens": 151200008.0,
+      "step": 44440
+    },
+    {
+      "epoch": 0.44612389096310573,
+      "learning_rate": 0.00011077722911397488,
+      "loss": 0.7866,
+      "mean_token_accuracy": 0.7731386125087738,
+      "num_tokens": 151233927.0,
+      "step": 44450
+    },
+    {
+      "epoch": 0.4462242562929062,
+      "learning_rate": 0.00011075715604801477,
+      "loss": 0.7067,
+      "mean_token_accuracy": 0.7901102244853974,
+      "num_tokens": 151267858.0,
+      "step": 44460
+    },
+    {
+      "epoch": 0.44632462162270664,
+      "learning_rate": 0.00011073708298205468,
+      "loss": 0.7218,
+      "mean_token_accuracy": 0.7798720598220825,
+      "num_tokens": 151302475.0,
+      "step": 44470
+    },
+    {
+      "epoch": 0.44642498695250715,
+      "learning_rate": 0.00011071700991609459,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7675409078598022,
+      "num_tokens": 151337152.0,
+      "step": 44480
+    },
+    {
+      "epoch": 0.4465253522823076,
+      "learning_rate": 0.0001106969368501345,
+      "loss": 0.7652,
+      "mean_token_accuracy": 0.7725721955299377,
+      "num_tokens": 151371174.0,
+      "step": 44490
+    },
+    {
+      "epoch": 0.44662571761210806,
+      "learning_rate": 0.00011067686378417439,
+      "loss": 0.7406,
+      "mean_token_accuracy": 0.779172706604004,
+      "num_tokens": 151405625.0,
+      "step": 44500
+    },
+    {
+      "epoch": 0.4467260829419085,
+      "learning_rate": 0.00011065679071821431,
+      "loss": 0.7582,
+      "mean_token_accuracy": 0.7733696103096008,
+      "num_tokens": 151439959.0,
+      "step": 44510
+    },
+    {
+      "epoch": 0.44682644827170903,
+      "learning_rate": 0.00011063671765225421,
+      "loss": 0.7573,
+      "mean_token_accuracy": 0.7659881412982941,
+      "num_tokens": 151473539.0,
+      "step": 44520
+    },
+    {
+      "epoch": 0.4469268136015095,
+      "learning_rate": 0.0001106166445862941,
+      "loss": 0.7334,
+      "mean_token_accuracy": 0.7809433043003082,
+      "num_tokens": 151507369.0,
+      "step": 44530
+    },
+    {
+      "epoch": 0.44702717893130994,
+      "learning_rate": 0.00011059657152033402,
+      "loss": 0.7186,
+      "mean_token_accuracy": 0.7804858088493347,
+      "num_tokens": 151541480.0,
+      "step": 44540
+    },
+    {
+      "epoch": 0.44712754426111045,
+      "learning_rate": 0.00011057649845437392,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7767960965633393,
+      "num_tokens": 151575833.0,
+      "step": 44550
+    },
+    {
+      "epoch": 0.4472279095909109,
+      "learning_rate": 0.00011055642538841384,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.7782451093196869,
+      "num_tokens": 151609860.0,
+      "step": 44560
+    },
+    {
+      "epoch": 0.44732827492071137,
+      "learning_rate": 0.00011053635232245374,
+      "loss": 0.7038,
+      "mean_token_accuracy": 0.7875931859016418,
+      "num_tokens": 151644156.0,
+      "step": 44570
+    },
+    {
+      "epoch": 0.4474286402505119,
+      "learning_rate": 0.00011051627925649363,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.7787350296974183,
+      "num_tokens": 151677964.0,
+      "step": 44580
+    },
+    {
+      "epoch": 0.44752900558031233,
+      "learning_rate": 0.00011049620619053355,
+      "loss": 0.7525,
+      "mean_token_accuracy": 0.7707826018333435,
+      "num_tokens": 151711574.0,
+      "step": 44590
+    },
+    {
+      "epoch": 0.4476293709101128,
+      "learning_rate": 0.00011047613312457346,
+      "loss": 0.7491,
+      "mean_token_accuracy": 0.7768722116947174,
+      "num_tokens": 151745582.0,
+      "step": 44600
+    },
+    {
+      "epoch": 0.4477297362399133,
+      "learning_rate": 0.00011045606005861336,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.7713538646697998,
+      "num_tokens": 151780065.0,
+      "step": 44610
+    },
+    {
+      "epoch": 0.44783010156971376,
+      "learning_rate": 0.00011043598699265328,
+      "loss": 0.7473,
+      "mean_token_accuracy": 0.776011997461319,
+      "num_tokens": 151814621.0,
+      "step": 44620
+    },
+    {
+      "epoch": 0.4479304668995142,
+      "learning_rate": 0.00011041591392669317,
+      "loss": 0.7217,
+      "mean_token_accuracy": 0.7846306204795838,
+      "num_tokens": 151848477.0,
+      "step": 44630
+    },
+    {
+      "epoch": 0.4480308322293147,
+      "learning_rate": 0.00011039584086073307,
+      "loss": 0.7295,
+      "mean_token_accuracy": 0.7849527895450592,
+      "num_tokens": 151883337.0,
+      "step": 44640
+    },
+    {
+      "epoch": 0.4481311975591152,
+      "learning_rate": 0.00011037576779477299,
+      "loss": 0.7574,
+      "mean_token_accuracy": 0.7724159836769104,
+      "num_tokens": 151916760.0,
+      "step": 44650
+    },
+    {
+      "epoch": 0.44823156288891564,
+      "learning_rate": 0.00011035569472881289,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.7737322509288788,
+      "num_tokens": 151950685.0,
+      "step": 44660
+    },
+    {
+      "epoch": 0.44833192821871615,
+      "learning_rate": 0.00011033562166285278,
+      "loss": 0.7194,
+      "mean_token_accuracy": 0.7801818192005158,
+      "num_tokens": 151984859.0,
+      "step": 44670
+    },
+    {
+      "epoch": 0.4484322935485166,
+      "learning_rate": 0.0001103155485968927,
+      "loss": 0.7338,
+      "mean_token_accuracy": 0.781706178188324,
+      "num_tokens": 152019151.0,
+      "step": 44680
+    },
+    {
+      "epoch": 0.44853265887831706,
+      "learning_rate": 0.0001102954755309326,
+      "loss": 0.7529,
+      "mean_token_accuracy": 0.7799510359764099,
+      "num_tokens": 152053767.0,
+      "step": 44690
+    },
+    {
+      "epoch": 0.44863302420811757,
+      "learning_rate": 0.0001102754024649725,
+      "loss": 0.7218,
+      "mean_token_accuracy": 0.7807285010814666,
+      "num_tokens": 152088255.0,
+      "step": 44700
+    },
+    {
+      "epoch": 0.448733389537918,
+      "learning_rate": 0.00011025532939901241,
+      "loss": 0.7885,
+      "mean_token_accuracy": 0.7673042953014374,
+      "num_tokens": 152122375.0,
+      "step": 44710
+    },
+    {
+      "epoch": 0.4488337548677185,
+      "learning_rate": 0.00011023525633305232,
+      "loss": 0.7403,
+      "mean_token_accuracy": 0.7771869957447052,
+      "num_tokens": 152156587.0,
+      "step": 44720
+    },
+    {
+      "epoch": 0.448934120197519,
+      "learning_rate": 0.00011021518326709222,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.7746523737907409,
+      "num_tokens": 152190688.0,
+      "step": 44730
+    },
+    {
+      "epoch": 0.44903448552731945,
+      "learning_rate": 0.00011019511020113214,
+      "loss": 0.7294,
+      "mean_token_accuracy": 0.7841888248920441,
+      "num_tokens": 152225354.0,
+      "step": 44740
+    },
+    {
+      "epoch": 0.4491348508571199,
+      "learning_rate": 0.00011017503713517203,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7752269029617309,
+      "num_tokens": 152259841.0,
+      "step": 44750
+    },
+    {
+      "epoch": 0.4492352161869204,
+      "learning_rate": 0.00011015496406921193,
+      "loss": 0.709,
+      "mean_token_accuracy": 0.7895790874958039,
+      "num_tokens": 152294626.0,
+      "step": 44760
+    },
+    {
+      "epoch": 0.44933558151672087,
+      "learning_rate": 0.00011013489100325185,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.7733225226402283,
+      "num_tokens": 152328799.0,
+      "step": 44770
+    },
+    {
+      "epoch": 0.44943594684652133,
+      "learning_rate": 0.00011011481793729175,
+      "loss": 0.7224,
+      "mean_token_accuracy": 0.7841477155685425,
+      "num_tokens": 152363147.0,
+      "step": 44780
+    },
+    {
+      "epoch": 0.4495363121763218,
+      "learning_rate": 0.00011009474487133164,
+      "loss": 0.7523,
+      "mean_token_accuracy": 0.7750907003879547,
+      "num_tokens": 152397871.0,
+      "step": 44790
+    },
+    {
+      "epoch": 0.4496366775061223,
+      "learning_rate": 0.00011007467180537156,
+      "loss": 0.7183,
+      "mean_token_accuracy": 0.7833089828491211,
+      "num_tokens": 152432334.0,
+      "step": 44800
+    },
+    {
+      "epoch": 0.44973704283592275,
+      "learning_rate": 0.00011005459873941146,
+      "loss": 0.7488,
+      "mean_token_accuracy": 0.7749288022518158,
+      "num_tokens": 152465909.0,
+      "step": 44810
+    },
+    {
+      "epoch": 0.4498374081657232,
+      "learning_rate": 0.00011003452567345137,
+      "loss": 0.7312,
+      "mean_token_accuracy": 0.7770084142684937,
+      "num_tokens": 152499684.0,
+      "step": 44820
+    },
+    {
+      "epoch": 0.4499377734955237,
+      "learning_rate": 0.00011001445260749127,
+      "loss": 0.7403,
+      "mean_token_accuracy": 0.7816311717033386,
+      "num_tokens": 152534067.0,
+      "step": 44830
+    },
+    {
+      "epoch": 0.4500381388253242,
+      "learning_rate": 0.00010999437954153118,
+      "loss": 0.7358,
+      "mean_token_accuracy": 0.783948940038681,
+      "num_tokens": 152567928.0,
+      "step": 44840
+    },
+    {
+      "epoch": 0.45013850415512463,
+      "learning_rate": 0.00010997430647557108,
+      "loss": 0.7558,
+      "mean_token_accuracy": 0.7711122810840607,
+      "num_tokens": 152602037.0,
+      "step": 44850
+    },
+    {
+      "epoch": 0.45023886948492514,
+      "learning_rate": 0.000109954233409611,
+      "loss": 0.742,
+      "mean_token_accuracy": 0.77590092420578,
+      "num_tokens": 152636816.0,
+      "step": 44860
+    },
+    {
+      "epoch": 0.4503392348147256,
+      "learning_rate": 0.0001099341603436509,
+      "loss": 0.7564,
+      "mean_token_accuracy": 0.772537213563919,
+      "num_tokens": 152671468.0,
+      "step": 44870
+    },
+    {
+      "epoch": 0.45043960014452605,
+      "learning_rate": 0.00010991408727769079,
+      "loss": 0.7897,
+      "mean_token_accuracy": 0.767767709493637,
+      "num_tokens": 152705343.0,
+      "step": 44880
+    },
+    {
+      "epoch": 0.45053996547432656,
+      "learning_rate": 0.00010989401421173071,
+      "loss": 0.7162,
+      "mean_token_accuracy": 0.7875835716724395,
+      "num_tokens": 152739295.0,
+      "step": 44890
+    },
+    {
+      "epoch": 0.450640330804127,
+      "learning_rate": 0.0001098739411457706,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.7732517123222351,
+      "num_tokens": 152773289.0,
+      "step": 44900
+    },
+    {
+      "epoch": 0.4507406961339275,
+      "learning_rate": 0.0001098538680798105,
+      "loss": 0.7365,
+      "mean_token_accuracy": 0.7796446919441223,
+      "num_tokens": 152806787.0,
+      "step": 44910
+    },
+    {
+      "epoch": 0.450841061463728,
+      "learning_rate": 0.00010983379501385042,
+      "loss": 0.7393,
+      "mean_token_accuracy": 0.782460218667984,
+      "num_tokens": 152841397.0,
+      "step": 44920
+    },
+    {
+      "epoch": 0.45094142679352844,
+      "learning_rate": 0.00010981372194789032,
+      "loss": 0.7203,
+      "mean_token_accuracy": 0.7838031113147735,
+      "num_tokens": 152876756.0,
+      "step": 44930
+    },
+    {
+      "epoch": 0.4510417921233289,
+      "learning_rate": 0.00010979364888193023,
+      "loss": 0.7636,
+      "mean_token_accuracy": 0.7748829603195191,
+      "num_tokens": 152910930.0,
+      "step": 44940
+    },
+    {
+      "epoch": 0.4511421574531294,
+      "learning_rate": 0.00010977357581597015,
+      "loss": 0.721,
+      "mean_token_accuracy": 0.7819502830505372,
+      "num_tokens": 152944540.0,
+      "step": 44950
+    },
+    {
+      "epoch": 0.45124252278292987,
+      "learning_rate": 0.00010975350275001004,
+      "loss": 0.7315,
+      "mean_token_accuracy": 0.7765802860260009,
+      "num_tokens": 152979407.0,
+      "step": 44960
+    },
+    {
+      "epoch": 0.4513428881127303,
+      "learning_rate": 0.00010973342968404994,
+      "loss": 0.7499,
+      "mean_token_accuracy": 0.7792832970619201,
+      "num_tokens": 153013721.0,
+      "step": 44970
+    },
+    {
+      "epoch": 0.45144325344253083,
+      "learning_rate": 0.00010971335661808986,
+      "loss": 0.7203,
+      "mean_token_accuracy": 0.7842995822429657,
+      "num_tokens": 153049466.0,
+      "step": 44980
+    },
+    {
+      "epoch": 0.4515436187723313,
+      "learning_rate": 0.00010969328355212976,
+      "loss": 0.7149,
+      "mean_token_accuracy": 0.7853386759757995,
+      "num_tokens": 153084689.0,
+      "step": 44990
+    },
+    {
+      "epoch": 0.45164398410213175,
+      "learning_rate": 0.00010967321048616965,
+      "loss": 0.7136,
+      "mean_token_accuracy": 0.7828909814357757,
+      "num_tokens": 153120308.0,
+      "step": 45000
+    },
+    {
+      "epoch": 0.45174434943193226,
+      "learning_rate": 0.00010965313742020957,
+      "loss": 0.7786,
+      "mean_token_accuracy": 0.7707717061042786,
+      "num_tokens": 153153962.0,
+      "step": 45010
+    },
+    {
+      "epoch": 0.4518447147617327,
+      "learning_rate": 0.00010963306435424947,
+      "loss": 0.7105,
+      "mean_token_accuracy": 0.7827567040920258,
+      "num_tokens": 153188249.0,
+      "step": 45020
+    },
+    {
+      "epoch": 0.45194508009153317,
+      "learning_rate": 0.00010961299128828936,
+      "loss": 0.7274,
+      "mean_token_accuracy": 0.7786150217056275,
+      "num_tokens": 153223850.0,
+      "step": 45030
+    },
+    {
+      "epoch": 0.4520454454213337,
+      "learning_rate": 0.00010959291822232928,
+      "loss": 0.7495,
+      "mean_token_accuracy": 0.7740147352218628,
+      "num_tokens": 153259052.0,
+      "step": 45040
+    },
+    {
+      "epoch": 0.45214581075113414,
+      "learning_rate": 0.00010957284515636919,
+      "loss": 0.7279,
+      "mean_token_accuracy": 0.7824159443378449,
+      "num_tokens": 153294295.0,
+      "step": 45050
+    },
+    {
+      "epoch": 0.4522461760809346,
+      "learning_rate": 0.00010955277209040909,
+      "loss": 0.7455,
+      "mean_token_accuracy": 0.7790737450122833,
+      "num_tokens": 153329893.0,
+      "step": 45060
+    },
+    {
+      "epoch": 0.45234654141073505,
+      "learning_rate": 0.00010953269902444901,
+      "loss": 0.7302,
+      "mean_token_accuracy": 0.7857704102993012,
+      "num_tokens": 153364620.0,
+      "step": 45070
+    },
+    {
+      "epoch": 0.45244690674053556,
+      "learning_rate": 0.0001095126259584889,
+      "loss": 0.7335,
+      "mean_token_accuracy": 0.7841358959674836,
+      "num_tokens": 153400195.0,
+      "step": 45080
+    },
+    {
+      "epoch": 0.452547272070336,
+      "learning_rate": 0.0001094925528925288,
+      "loss": 0.7118,
+      "mean_token_accuracy": 0.7873798191547394,
+      "num_tokens": 153436065.0,
+      "step": 45090
+    },
+    {
+      "epoch": 0.45264763740013647,
+      "learning_rate": 0.00010947247982656872,
+      "loss": 0.7097,
+      "mean_token_accuracy": 0.7894898891448975,
+      "num_tokens": 153470537.0,
+      "step": 45100
+    },
+    {
+      "epoch": 0.452748002729937,
+      "learning_rate": 0.00010945240676060862,
+      "loss": 0.6934,
+      "mean_token_accuracy": 0.7880873382091522,
+      "num_tokens": 153505241.0,
+      "step": 45110
+    },
+    {
+      "epoch": 0.45284836805973744,
+      "learning_rate": 0.00010943233369464854,
+      "loss": 0.7378,
+      "mean_token_accuracy": 0.7773111522197723,
+      "num_tokens": 153540342.0,
+      "step": 45120
+    },
+    {
+      "epoch": 0.4529487333895379,
+      "learning_rate": 0.00010941226062868843,
+      "loss": 0.7293,
+      "mean_token_accuracy": 0.7838761627674102,
+      "num_tokens": 153574356.0,
+      "step": 45130
+    },
+    {
+      "epoch": 0.4530490987193384,
+      "learning_rate": 0.00010939218756272833,
+      "loss": 0.7543,
+      "mean_token_accuracy": 0.7750289261341095,
+      "num_tokens": 153609263.0,
+      "step": 45140
+    },
+    {
+      "epoch": 0.45314946404913886,
+      "learning_rate": 0.00010937211449676825,
+      "loss": 0.757,
+      "mean_token_accuracy": 0.771094697713852,
+      "num_tokens": 153644077.0,
+      "step": 45150
+    },
+    {
+      "epoch": 0.4532498293789393,
+      "learning_rate": 0.00010935204143080814,
+      "loss": 0.6772,
+      "mean_token_accuracy": 0.7960553526878357,
+      "num_tokens": 153678370.0,
+      "step": 45160
+    },
+    {
+      "epoch": 0.45335019470873983,
+      "learning_rate": 0.00010933196836484805,
+      "loss": 0.7351,
+      "mean_token_accuracy": 0.7795609951019287,
+      "num_tokens": 153713627.0,
+      "step": 45170
+    },
+    {
+      "epoch": 0.4534505600385403,
+      "learning_rate": 0.00010931189529888796,
+      "loss": 0.7204,
+      "mean_token_accuracy": 0.7858222723007202,
+      "num_tokens": 153749165.0,
+      "step": 45180
+    },
+    {
+      "epoch": 0.45355092536834074,
+      "learning_rate": 0.00010929182223292787,
+      "loss": 0.7185,
+      "mean_token_accuracy": 0.7809831619262695,
+      "num_tokens": 153783844.0,
+      "step": 45190
+    },
+    {
+      "epoch": 0.45365129069814125,
+      "learning_rate": 0.00010927174916696776,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.7743128657341003,
+      "num_tokens": 153819545.0,
+      "step": 45200
+    },
+    {
+      "epoch": 0.4537516560279417,
+      "learning_rate": 0.00010925167610100769,
+      "loss": 0.7398,
+      "mean_token_accuracy": 0.7764536440372467,
+      "num_tokens": 153854909.0,
+      "step": 45210
+    },
+    {
+      "epoch": 0.45385202135774216,
+      "learning_rate": 0.00010923160303504758,
+      "loss": 0.7376,
+      "mean_token_accuracy": 0.7779523968696594,
+      "num_tokens": 153888924.0,
+      "step": 45220
+    },
+    {
+      "epoch": 0.4539523866875427,
+      "learning_rate": 0.00010921152996908748,
+      "loss": 0.7594,
+      "mean_token_accuracy": 0.7741849064826966,
+      "num_tokens": 153923677.0,
+      "step": 45230
+    },
+    {
+      "epoch": 0.45405275201734313,
+      "learning_rate": 0.0001091914569031274,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7729076683521271,
+      "num_tokens": 153958052.0,
+      "step": 45240
+    },
+    {
+      "epoch": 0.4541531173471436,
+      "learning_rate": 0.0001091713838371673,
+      "loss": 0.7289,
+      "mean_token_accuracy": 0.7829089462757111,
+      "num_tokens": 153993469.0,
+      "step": 45250
+    },
+    {
+      "epoch": 0.4542534826769441,
+      "learning_rate": 0.00010915131077120719,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.776902949810028,
+      "num_tokens": 154027804.0,
+      "step": 45260
+    },
+    {
+      "epoch": 0.45435384800674455,
+      "learning_rate": 0.00010913123770524711,
+      "loss": 0.7839,
+      "mean_token_accuracy": 0.7743165254592895,
+      "num_tokens": 154062293.0,
+      "step": 45270
+    },
+    {
+      "epoch": 0.454454213336545,
+      "learning_rate": 0.000109111164639287,
+      "loss": 0.7439,
+      "mean_token_accuracy": 0.7767428517341614,
+      "num_tokens": 154096430.0,
+      "step": 45280
+    },
+    {
+      "epoch": 0.4545545786663455,
+      "learning_rate": 0.00010909109157332691,
+      "loss": 0.7317,
+      "mean_token_accuracy": 0.7799388825893402,
+      "num_tokens": 154130176.0,
+      "step": 45290
+    },
+    {
+      "epoch": 0.454654943996146,
+      "learning_rate": 0.00010907101850736684,
+      "loss": 0.7365,
+      "mean_token_accuracy": 0.7814435422420501,
+      "num_tokens": 154165050.0,
+      "step": 45300
+    },
+    {
+      "epoch": 0.45475530932594643,
+      "learning_rate": 0.00010905094544140673,
+      "loss": 0.7404,
+      "mean_token_accuracy": 0.7786262392997741,
+      "num_tokens": 154201074.0,
+      "step": 45310
+    },
+    {
+      "epoch": 0.45485567465574694,
+      "learning_rate": 0.00010903087237544663,
+      "loss": 0.7584,
+      "mean_token_accuracy": 0.7741450905799866,
+      "num_tokens": 154235624.0,
+      "step": 45320
+    },
+    {
+      "epoch": 0.4549560399855474,
+      "learning_rate": 0.00010901079930948655,
+      "loss": 0.7304,
+      "mean_token_accuracy": 0.7823881149291992,
+      "num_tokens": 154270374.0,
+      "step": 45330
+    },
+    {
+      "epoch": 0.45505640531534786,
+      "learning_rate": 0.00010899072624352644,
+      "loss": 0.7514,
+      "mean_token_accuracy": 0.7742278933525085,
+      "num_tokens": 154305065.0,
+      "step": 45340
+    },
+    {
+      "epoch": 0.45515677064514837,
+      "learning_rate": 0.00010897065317756634,
+      "loss": 0.754,
+      "mean_token_accuracy": 0.7744691073894501,
+      "num_tokens": 154339332.0,
+      "step": 45350
+    },
+    {
+      "epoch": 0.4552571359749488,
+      "learning_rate": 0.00010895058011160626,
+      "loss": 0.7119,
+      "mean_token_accuracy": 0.785189688205719,
+      "num_tokens": 154374046.0,
+      "step": 45360
+    },
+    {
+      "epoch": 0.4553575013047493,
+      "learning_rate": 0.00010893050704564615,
+      "loss": 0.7178,
+      "mean_token_accuracy": 0.7814117550849915,
+      "num_tokens": 154409918.0,
+      "step": 45370
+    },
+    {
+      "epoch": 0.45545786663454974,
+      "learning_rate": 0.00010891043397968605,
+      "loss": 0.7421,
+      "mean_token_accuracy": 0.7787565946578979,
+      "num_tokens": 154444201.0,
+      "step": 45380
+    },
+    {
+      "epoch": 0.45555823196435025,
+      "learning_rate": 0.00010889036091372597,
+      "loss": 0.7259,
+      "mean_token_accuracy": 0.7825199484825134,
+      "num_tokens": 154479637.0,
+      "step": 45390
+    },
+    {
+      "epoch": 0.4556585972941507,
+      "learning_rate": 0.00010887028784776588,
+      "loss": 0.6876,
+      "mean_token_accuracy": 0.7944839775562287,
+      "num_tokens": 154514378.0,
+      "step": 45400
+    },
+    {
+      "epoch": 0.45575896262395116,
+      "learning_rate": 0.00010885021478180577,
+      "loss": 0.6979,
+      "mean_token_accuracy": 0.7917682349681854,
+      "num_tokens": 154549440.0,
+      "step": 45410
+    },
+    {
+      "epoch": 0.45585932795375167,
+      "learning_rate": 0.0001088301417158457,
+      "loss": 0.6868,
+      "mean_token_accuracy": 0.7883335769176483,
+      "num_tokens": 154583459.0,
+      "step": 45420
+    },
+    {
+      "epoch": 0.4559596932835521,
+      "learning_rate": 0.00010881006864988559,
+      "loss": 0.7167,
+      "mean_token_accuracy": 0.7852450430393219,
+      "num_tokens": 154618681.0,
+      "step": 45430
+    },
+    {
+      "epoch": 0.4560600586133526,
+      "learning_rate": 0.00010878999558392549,
+      "loss": 0.7006,
+      "mean_token_accuracy": 0.7889388144016266,
+      "num_tokens": 154655079.0,
+      "step": 45440
+    },
+    {
+      "epoch": 0.4561604239431531,
+      "learning_rate": 0.00010876992251796541,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.7862299740314483,
+      "num_tokens": 154689987.0,
+      "step": 45450
+    },
+    {
+      "epoch": 0.45626078927295355,
+      "learning_rate": 0.0001087498494520053,
+      "loss": 0.7543,
+      "mean_token_accuracy": 0.7773892939090729,
+      "num_tokens": 154725531.0,
+      "step": 45460
+    },
+    {
+      "epoch": 0.456361154602754,
+      "learning_rate": 0.0001087297763860452,
+      "loss": 0.6791,
+      "mean_token_accuracy": 0.79625204205513,
+      "num_tokens": 154760949.0,
+      "step": 45470
+    },
+    {
+      "epoch": 0.4564615199325545,
+      "learning_rate": 0.00010870970332008512,
+      "loss": 0.7394,
+      "mean_token_accuracy": 0.7802935123443604,
+      "num_tokens": 154795596.0,
+      "step": 45480
+    },
+    {
+      "epoch": 0.45656188526235497,
+      "learning_rate": 0.00010868963025412501,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.7932131230831146,
+      "num_tokens": 154831966.0,
+      "step": 45490
+    },
+    {
+      "epoch": 0.4566622505921554,
+      "learning_rate": 0.00010866955718816492,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.781095540523529,
+      "num_tokens": 154867253.0,
+      "step": 45500
+    },
+    {
+      "epoch": 0.45676261592195594,
+      "learning_rate": 0.00010864948412220483,
+      "loss": 0.7215,
+      "mean_token_accuracy": 0.7840093016624451,
+      "num_tokens": 154900976.0,
+      "step": 45510
+    },
+    {
+      "epoch": 0.4568629812517564,
+      "learning_rate": 0.00010862941105624474,
+      "loss": 0.6961,
+      "mean_token_accuracy": 0.7899511754512787,
+      "num_tokens": 154935158.0,
+      "step": 45520
+    },
+    {
+      "epoch": 0.45696334658155685,
+      "learning_rate": 0.00010860933799028463,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.7798848330974579,
+      "num_tokens": 154970636.0,
+      "step": 45530
+    },
+    {
+      "epoch": 0.45706371191135736,
+      "learning_rate": 0.00010858926492432456,
+      "loss": 0.7331,
+      "mean_token_accuracy": 0.7831038177013397,
+      "num_tokens": 155006127.0,
+      "step": 45540
+    },
+    {
+      "epoch": 0.4571640772411578,
+      "learning_rate": 0.00010856919185836445,
+      "loss": 0.7258,
+      "mean_token_accuracy": 0.783970981836319,
+      "num_tokens": 155040627.0,
+      "step": 45550
+    },
+    {
+      "epoch": 0.4572644425709583,
+      "learning_rate": 0.00010854911879240435,
+      "loss": 0.677,
+      "mean_token_accuracy": 0.797710633277893,
+      "num_tokens": 155075643.0,
+      "step": 45560
+    },
+    {
+      "epoch": 0.4573648079007588,
+      "learning_rate": 0.00010852904572644427,
+      "loss": 0.7518,
+      "mean_token_accuracy": 0.7771606564521789,
+      "num_tokens": 155110559.0,
+      "step": 45570
+    },
+    {
+      "epoch": 0.45746517323055924,
+      "learning_rate": 0.00010850897266048416,
+      "loss": 0.7317,
+      "mean_token_accuracy": 0.7806562066078186,
+      "num_tokens": 155145920.0,
+      "step": 45580
+    },
+    {
+      "epoch": 0.4575655385603597,
+      "learning_rate": 0.00010848889959452406,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.7896435856819153,
+      "num_tokens": 155181887.0,
+      "step": 45590
+    },
+    {
+      "epoch": 0.4576659038901602,
+      "learning_rate": 0.00010846882652856398,
+      "loss": 0.7561,
+      "mean_token_accuracy": 0.7798638761043548,
+      "num_tokens": 155216768.0,
+      "step": 45600
+    },
+    {
+      "epoch": 0.45776626921996066,
+      "learning_rate": 0.00010844875346260388,
+      "loss": 0.719,
+      "mean_token_accuracy": 0.7846980154514313,
+      "num_tokens": 155250378.0,
+      "step": 45610
+    },
+    {
+      "epoch": 0.4578666345497611,
+      "learning_rate": 0.00010842868039664378,
+      "loss": 0.7079,
+      "mean_token_accuracy": 0.7912899911403656,
+      "num_tokens": 155285650.0,
+      "step": 45620
+    },
+    {
+      "epoch": 0.45796699987956163,
+      "learning_rate": 0.00010840860733068369,
+      "loss": 0.6993,
+      "mean_token_accuracy": 0.7907711029052734,
+      "num_tokens": 155319932.0,
+      "step": 45630
+    },
+    {
+      "epoch": 0.4580673652093621,
+      "learning_rate": 0.0001083885342647236,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.7919665277004242,
+      "num_tokens": 155354891.0,
+      "step": 45640
+    },
+    {
+      "epoch": 0.45816773053916254,
+      "learning_rate": 0.0001083684611987635,
+      "loss": 0.7217,
+      "mean_token_accuracy": 0.7810214936733246,
+      "num_tokens": 155390599.0,
+      "step": 45650
+    },
+    {
+      "epoch": 0.458268095868963,
+      "learning_rate": 0.00010834838813280342,
+      "loss": 0.7328,
+      "mean_token_accuracy": 0.7812167346477509,
+      "num_tokens": 155426956.0,
+      "step": 45660
+    },
+    {
+      "epoch": 0.4583684611987635,
+      "learning_rate": 0.00010832831506684331,
+      "loss": 0.7415,
+      "mean_token_accuracy": 0.775298398733139,
+      "num_tokens": 155461457.0,
+      "step": 45670
+    },
+    {
+      "epoch": 0.45846882652856397,
+      "learning_rate": 0.00010830824200088323,
+      "loss": 0.7573,
+      "mean_token_accuracy": 0.7747935891151428,
+      "num_tokens": 155496523.0,
+      "step": 45680
+    },
+    {
+      "epoch": 0.4585691918583644,
+      "learning_rate": 0.00010828816893492313,
+      "loss": 0.7088,
+      "mean_token_accuracy": 0.7924448072910308,
+      "num_tokens": 155531567.0,
+      "step": 45690
+    },
+    {
+      "epoch": 0.45866955718816493,
+      "learning_rate": 0.00010826809586896302,
+      "loss": 0.7197,
+      "mean_token_accuracy": 0.7861683249473572,
+      "num_tokens": 155567009.0,
+      "step": 45700
+    },
+    {
+      "epoch": 0.4587699225179654,
+      "learning_rate": 0.00010824802280300295,
+      "loss": 0.7209,
+      "mean_token_accuracy": 0.7828059613704681,
+      "num_tokens": 155601500.0,
+      "step": 45710
+    },
+    {
+      "epoch": 0.45887028784776585,
+      "learning_rate": 0.00010822794973704284,
+      "loss": 0.6994,
+      "mean_token_accuracy": 0.7851306378841401,
+      "num_tokens": 155637849.0,
+      "step": 45720
+    },
+    {
+      "epoch": 0.45897065317756636,
+      "learning_rate": 0.00010820787667108274,
+      "loss": 0.7675,
+      "mean_token_accuracy": 0.7704609274864197,
+      "num_tokens": 155672443.0,
+      "step": 45730
+    },
+    {
+      "epoch": 0.4590710185073668,
+      "learning_rate": 0.00010818780360512266,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.7772091090679168,
+      "num_tokens": 155707804.0,
+      "step": 45740
+    },
+    {
+      "epoch": 0.45917138383716727,
+      "learning_rate": 0.00010816773053916257,
+      "loss": 0.7221,
+      "mean_token_accuracy": 0.7843359291553498,
+      "num_tokens": 155742610.0,
+      "step": 45750
+    },
+    {
+      "epoch": 0.4592717491669678,
+      "learning_rate": 0.00010814765747320246,
+      "loss": 0.7159,
+      "mean_token_accuracy": 0.78268381357193,
+      "num_tokens": 155777374.0,
+      "step": 45760
+    },
+    {
+      "epoch": 0.45937211449676824,
+      "learning_rate": 0.00010812758440724238,
+      "loss": 0.7114,
+      "mean_token_accuracy": 0.7857536256313324,
+      "num_tokens": 155812932.0,
+      "step": 45770
+    },
+    {
+      "epoch": 0.4594724798265687,
+      "learning_rate": 0.00010810751134128228,
+      "loss": 0.712,
+      "mean_token_accuracy": 0.7849782645702362,
+      "num_tokens": 155848264.0,
+      "step": 45780
+    },
+    {
+      "epoch": 0.4595728451563692,
+      "learning_rate": 0.00010808743827532217,
+      "loss": 0.7135,
+      "mean_token_accuracy": 0.7836347043514251,
+      "num_tokens": 155883947.0,
+      "step": 45790
+    },
+    {
+      "epoch": 0.45967321048616966,
+      "learning_rate": 0.0001080673652093621,
+      "loss": 0.7009,
+      "mean_token_accuracy": 0.7856990694999695,
+      "num_tokens": 155919399.0,
+      "step": 45800
+    },
+    {
+      "epoch": 0.4597735758159701,
+      "learning_rate": 0.00010804729214340199,
+      "loss": 0.6993,
+      "mean_token_accuracy": 0.7913999378681182,
+      "num_tokens": 155954288.0,
+      "step": 45810
+    },
+    {
+      "epoch": 0.4598739411457706,
+      "learning_rate": 0.00010802721907744188,
+      "loss": 0.7121,
+      "mean_token_accuracy": 0.7861483752727508,
+      "num_tokens": 155989524.0,
+      "step": 45820
+    },
+    {
+      "epoch": 0.4599743064755711,
+      "learning_rate": 0.0001080071460114818,
+      "loss": 0.7216,
+      "mean_token_accuracy": 0.7861035943031311,
+      "num_tokens": 156024902.0,
+      "step": 45830
+    },
+    {
+      "epoch": 0.46007467180537154,
+      "learning_rate": 0.0001079870729455217,
+      "loss": 0.678,
+      "mean_token_accuracy": 0.796373850107193,
+      "num_tokens": 156060776.0,
+      "step": 45840
+    },
+    {
+      "epoch": 0.46017503713517205,
+      "learning_rate": 0.00010796699987956161,
+      "loss": 0.7376,
+      "mean_token_accuracy": 0.7753163158893586,
+      "num_tokens": 156094988.0,
+      "step": 45850
+    },
+    {
+      "epoch": 0.4602754024649725,
+      "learning_rate": 0.00010794692681360152,
+      "loss": 0.7073,
+      "mean_token_accuracy": 0.7867257773876191,
+      "num_tokens": 156130118.0,
+      "step": 45860
+    },
+    {
+      "epoch": 0.46037576779477296,
+      "learning_rate": 0.00010792685374764143,
+      "loss": 0.714,
+      "mean_token_accuracy": 0.7838221549987793,
+      "num_tokens": 156164745.0,
+      "step": 45870
+    },
+    {
+      "epoch": 0.4604761331245735,
+      "learning_rate": 0.00010790678068168132,
+      "loss": 0.705,
+      "mean_token_accuracy": 0.7828870356082916,
+      "num_tokens": 156200350.0,
+      "step": 45880
+    },
+    {
+      "epoch": 0.46057649845437393,
+      "learning_rate": 0.00010788670761572124,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.7931342422962189,
+      "num_tokens": 156236015.0,
+      "step": 45890
+    },
+    {
+      "epoch": 0.4606768637841744,
+      "learning_rate": 0.00010786663454976114,
+      "loss": 0.7284,
+      "mean_token_accuracy": 0.7817682504653931,
+      "num_tokens": 156272743.0,
+      "step": 45900
+    },
+    {
+      "epoch": 0.4607772291139749,
+      "learning_rate": 0.00010784656148380103,
+      "loss": 0.7142,
+      "mean_token_accuracy": 0.7831542611122131,
+      "num_tokens": 156308985.0,
+      "step": 45910
+    },
+    {
+      "epoch": 0.46087759444377535,
+      "learning_rate": 0.00010782648841784096,
+      "loss": 0.7159,
+      "mean_token_accuracy": 0.7871647775173187,
+      "num_tokens": 156343738.0,
+      "step": 45920
+    },
+    {
+      "epoch": 0.4609779597735758,
+      "learning_rate": 0.00010780641535188085,
+      "loss": 0.7162,
+      "mean_token_accuracy": 0.7845615029335022,
+      "num_tokens": 156379077.0,
+      "step": 45930
+    },
+    {
+      "epoch": 0.46107832510337626,
+      "learning_rate": 0.00010778634228592075,
+      "loss": 0.7375,
+      "mean_token_accuracy": 0.7808549463748932,
+      "num_tokens": 156414109.0,
+      "step": 45940
+    },
+    {
+      "epoch": 0.4611786904331768,
+      "learning_rate": 0.00010776626921996067,
+      "loss": 0.708,
+      "mean_token_accuracy": 0.7884536445140838,
+      "num_tokens": 156449434.0,
+      "step": 45950
+    },
+    {
+      "epoch": 0.46127905576297723,
+      "learning_rate": 0.00010774619615400056,
+      "loss": 0.6897,
+      "mean_token_accuracy": 0.7883106231689453,
+      "num_tokens": 156484881.0,
+      "step": 45960
+    },
+    {
+      "epoch": 0.4613794210927777,
+      "learning_rate": 0.00010772612308804047,
+      "loss": 0.72,
+      "mean_token_accuracy": 0.7828457951545715,
+      "num_tokens": 156520355.0,
+      "step": 45970
+    },
+    {
+      "epoch": 0.4614797864225782,
+      "learning_rate": 0.00010770605002208038,
+      "loss": 0.6935,
+      "mean_token_accuracy": 0.7907245039939881,
+      "num_tokens": 156555662.0,
+      "step": 45980
+    },
+    {
+      "epoch": 0.46158015175237865,
+      "learning_rate": 0.00010768597695612029,
+      "loss": 0.7392,
+      "mean_token_accuracy": 0.7876706302165986,
+      "num_tokens": 156590255.0,
+      "step": 45990
+    },
+    {
+      "epoch": 0.4616805170821791,
+      "learning_rate": 0.00010766590389016018,
+      "loss": 0.7175,
+      "mean_token_accuracy": 0.790300065279007,
+      "num_tokens": 156625458.0,
+      "step": 46000
+    },
+    {
+      "epoch": 0.4617808824119796,
+      "learning_rate": 0.0001076458308242001,
+      "loss": 0.7247,
+      "mean_token_accuracy": 0.784890878200531,
+      "num_tokens": 156662011.0,
+      "step": 46010
+    },
+    {
+      "epoch": 0.4618812477417801,
+      "learning_rate": 0.00010762575775824,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.7865753948688508,
+      "num_tokens": 156697122.0,
+      "step": 46020
+    },
+    {
+      "epoch": 0.46198161307158053,
+      "learning_rate": 0.0001076056846922799,
+      "loss": 0.7179,
+      "mean_token_accuracy": 0.7825608730316163,
+      "num_tokens": 156732175.0,
+      "step": 46030
+    },
+    {
+      "epoch": 0.46208197840138104,
+      "learning_rate": 0.00010758561162631982,
+      "loss": 0.705,
+      "mean_token_accuracy": 0.7832401931285858,
+      "num_tokens": 156767635.0,
+      "step": 46040
+    },
+    {
+      "epoch": 0.4621823437311815,
+      "learning_rate": 0.00010756553856035971,
+      "loss": 0.7308,
+      "mean_token_accuracy": 0.780849301815033,
+      "num_tokens": 156801943.0,
+      "step": 46050
+    },
+    {
+      "epoch": 0.46228270906098196,
+      "learning_rate": 0.0001075454654943996,
+      "loss": 0.7358,
+      "mean_token_accuracy": 0.7800325095653534,
+      "num_tokens": 156836748.0,
+      "step": 46060
+    },
+    {
+      "epoch": 0.46238307439078247,
+      "learning_rate": 0.00010752539242843953,
+      "loss": 0.6611,
+      "mean_token_accuracy": 0.7982037842273713,
+      "num_tokens": 156872539.0,
+      "step": 46070
+    },
+    {
+      "epoch": 0.4624834397205829,
+      "learning_rate": 0.00010750531936247942,
+      "loss": 0.72,
+      "mean_token_accuracy": 0.7860180675983429,
+      "num_tokens": 156907083.0,
+      "step": 46080
+    },
+    {
+      "epoch": 0.4625838050503834,
+      "learning_rate": 0.00010748524629651933,
+      "loss": 0.7332,
+      "mean_token_accuracy": 0.7771995067596436,
+      "num_tokens": 156942262.0,
+      "step": 46090
+    },
+    {
+      "epoch": 0.4626841703801839,
+      "learning_rate": 0.00010746517323055925,
+      "loss": 0.7231,
+      "mean_token_accuracy": 0.7848806202411651,
+      "num_tokens": 156978630.0,
+      "step": 46100
+    },
+    {
+      "epoch": 0.46278453570998435,
+      "learning_rate": 0.00010744510016459915,
+      "loss": 0.7335,
+      "mean_token_accuracy": 0.7821138381958008,
+      "num_tokens": 157013098.0,
+      "step": 46110
+    },
+    {
+      "epoch": 0.4628849010397848,
+      "learning_rate": 0.00010742502709863904,
+      "loss": 0.7254,
+      "mean_token_accuracy": 0.7811341166496277,
+      "num_tokens": 157047915.0,
+      "step": 46120
+    },
+    {
+      "epoch": 0.4629852663695853,
+      "learning_rate": 0.00010740495403267896,
+      "loss": 0.6864,
+      "mean_token_accuracy": 0.7955152571201325,
+      "num_tokens": 157083690.0,
+      "step": 46130
+    },
+    {
+      "epoch": 0.46308563169938577,
+      "learning_rate": 0.00010738488096671886,
+      "loss": 0.705,
+      "mean_token_accuracy": 0.7844863951206207,
+      "num_tokens": 157118204.0,
+      "step": 46140
+    },
+    {
+      "epoch": 0.4631859970291862,
+      "learning_rate": 0.00010736480790075875,
+      "loss": 0.6849,
+      "mean_token_accuracy": 0.7934133410453796,
+      "num_tokens": 157152228.0,
+      "step": 46150
+    },
+    {
+      "epoch": 0.46328636235898674,
+      "learning_rate": 0.00010734473483479868,
+      "loss": 0.7248,
+      "mean_token_accuracy": 0.7831818222999573,
+      "num_tokens": 157188068.0,
+      "step": 46160
+    },
+    {
+      "epoch": 0.4633867276887872,
+      "learning_rate": 0.00010732466176883857,
+      "loss": 0.6677,
+      "mean_token_accuracy": 0.7984673023223877,
+      "num_tokens": 157223352.0,
+      "step": 46170
+    },
+    {
+      "epoch": 0.46348709301858765,
+      "learning_rate": 0.00010730458870287847,
+      "loss": 0.6803,
+      "mean_token_accuracy": 0.796294242143631,
+      "num_tokens": 157258284.0,
+      "step": 46180
+    },
+    {
+      "epoch": 0.46358745834838816,
+      "learning_rate": 0.00010728451563691839,
+      "loss": 0.6924,
+      "mean_token_accuracy": 0.7861761391162873,
+      "num_tokens": 157293501.0,
+      "step": 46190
+    },
+    {
+      "epoch": 0.4636878236781886,
+      "learning_rate": 0.0001072644425709583,
+      "loss": 0.688,
+      "mean_token_accuracy": 0.7971710205078125,
+      "num_tokens": 157330040.0,
+      "step": 46200
+    },
+    {
+      "epoch": 0.46378818900798907,
+      "learning_rate": 0.00010724436950499819,
+      "loss": 0.6885,
+      "mean_token_accuracy": 0.7873896420001983,
+      "num_tokens": 157365148.0,
+      "step": 46210
+    },
+    {
+      "epoch": 0.4638885543377895,
+      "learning_rate": 0.00010722429643903811,
+      "loss": 0.7051,
+      "mean_token_accuracy": 0.7853258192539215,
+      "num_tokens": 157401536.0,
+      "step": 46220
+    },
+    {
+      "epoch": 0.46398891966759004,
+      "learning_rate": 0.00010720422337307801,
+      "loss": 0.7102,
+      "mean_token_accuracy": 0.7846766710281372,
+      "num_tokens": 157436736.0,
+      "step": 46230
+    },
+    {
+      "epoch": 0.4640892849973905,
+      "learning_rate": 0.00010718415030711793,
+      "loss": 0.7403,
+      "mean_token_accuracy": 0.7822953999042511,
+      "num_tokens": 157471551.0,
+      "step": 46240
+    },
+    {
+      "epoch": 0.46418965032719095,
+      "learning_rate": 0.00010716407724115783,
+      "loss": 0.6797,
+      "mean_token_accuracy": 0.7962810456752777,
+      "num_tokens": 157507143.0,
+      "step": 46250
+    },
+    {
+      "epoch": 0.46429001565699146,
+      "learning_rate": 0.00010714400417519772,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.7917849063873291,
+      "num_tokens": 157543095.0,
+      "step": 46260
+    },
+    {
+      "epoch": 0.4643903809867919,
+      "learning_rate": 0.00010712393110923764,
+      "loss": 0.7074,
+      "mean_token_accuracy": 0.7860041975975036,
+      "num_tokens": 157578818.0,
+      "step": 46270
+    },
+    {
+      "epoch": 0.4644907463165924,
+      "learning_rate": 0.00010710385804327754,
+      "loss": 0.7194,
+      "mean_token_accuracy": 0.7866369724273682,
+      "num_tokens": 157613086.0,
+      "step": 46280
+    },
+    {
+      "epoch": 0.4645911116463929,
+      "learning_rate": 0.00010708378497731743,
+      "loss": 0.7323,
+      "mean_token_accuracy": 0.7846727252006531,
+      "num_tokens": 157647792.0,
+      "step": 46290
+    },
+    {
+      "epoch": 0.46469147697619334,
+      "learning_rate": 0.00010706371191135735,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.7749805808067322,
+      "num_tokens": 157683453.0,
+      "step": 46300
+    },
+    {
+      "epoch": 0.4647918423059938,
+      "learning_rate": 0.00010704363884539725,
+      "loss": 0.681,
+      "mean_token_accuracy": 0.7935085654258728,
+      "num_tokens": 157719305.0,
+      "step": 46310
+    },
+    {
+      "epoch": 0.4648922076357943,
+      "learning_rate": 0.00010702356577943716,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.79261554479599,
+      "num_tokens": 157754832.0,
+      "step": 46320
+    },
+    {
+      "epoch": 0.46499257296559476,
+      "learning_rate": 0.00010700349271347707,
+      "loss": 0.673,
+      "mean_token_accuracy": 0.7986871778964997,
+      "num_tokens": 157789464.0,
+      "step": 46330
+    },
+    {
+      "epoch": 0.4650929382953952,
+      "learning_rate": 0.00010698341964751697,
+      "loss": 0.687,
+      "mean_token_accuracy": 0.7916176676750183,
+      "num_tokens": 157824600.0,
+      "step": 46340
+    },
+    {
+      "epoch": 0.46519330362519573,
+      "learning_rate": 0.00010696334658155687,
+      "loss": 0.7182,
+      "mean_token_accuracy": 0.7853668570518494,
+      "num_tokens": 157858964.0,
+      "step": 46350
+    },
+    {
+      "epoch": 0.4652936689549962,
+      "learning_rate": 0.00010694327351559679,
+      "loss": 0.7119,
+      "mean_token_accuracy": 0.7871406614780426,
+      "num_tokens": 157894188.0,
+      "step": 46360
+    },
+    {
+      "epoch": 0.46539403428479664,
+      "learning_rate": 0.00010692320044963669,
+      "loss": 0.7194,
+      "mean_token_accuracy": 0.786019378900528,
+      "num_tokens": 157929823.0,
+      "step": 46370
+    },
+    {
+      "epoch": 0.46549439961459715,
+      "learning_rate": 0.00010690312738367658,
+      "loss": 0.6977,
+      "mean_token_accuracy": 0.7925242602825164,
+      "num_tokens": 157964698.0,
+      "step": 46380
+    },
+    {
+      "epoch": 0.4655947649443976,
+      "learning_rate": 0.0001068830543177165,
+      "loss": 0.7354,
+      "mean_token_accuracy": 0.784424090385437,
+      "num_tokens": 157999845.0,
+      "step": 46390
+    },
+    {
+      "epoch": 0.46569513027419807,
+      "learning_rate": 0.0001068629812517564,
+      "loss": 0.7034,
+      "mean_token_accuracy": 0.7869425654411316,
+      "num_tokens": 158035108.0,
+      "step": 46400
+    },
+    {
+      "epoch": 0.4657954956039986,
+      "learning_rate": 0.00010684290818579629,
+      "loss": 0.703,
+      "mean_token_accuracy": 0.7890412390232087,
+      "num_tokens": 158070592.0,
+      "step": 46410
+    },
+    {
+      "epoch": 0.46589586093379903,
+      "learning_rate": 0.00010682283511983621,
+      "loss": 0.6805,
+      "mean_token_accuracy": 0.7949465453624726,
+      "num_tokens": 158105994.0,
+      "step": 46420
+    },
+    {
+      "epoch": 0.4659962262635995,
+      "learning_rate": 0.00010680276205387611,
+      "loss": 0.7034,
+      "mean_token_accuracy": 0.7892135560512543,
+      "num_tokens": 158142003.0,
+      "step": 46430
+    },
+    {
+      "epoch": 0.4660965915934,
+      "learning_rate": 0.00010678268898791602,
+      "loss": 0.6919,
+      "mean_token_accuracy": 0.7986708045005798,
+      "num_tokens": 158177831.0,
+      "step": 46440
+    },
+    {
+      "epoch": 0.46619695692320046,
+      "learning_rate": 0.00010676261592195593,
+      "loss": 0.7129,
+      "mean_token_accuracy": 0.7889884889125824,
+      "num_tokens": 158212578.0,
+      "step": 46450
+    },
+    {
+      "epoch": 0.4662973222530009,
+      "learning_rate": 0.00010674254285599583,
+      "loss": 0.7223,
+      "mean_token_accuracy": 0.7806691288948059,
+      "num_tokens": 158247415.0,
+      "step": 46460
+    },
+    {
+      "epoch": 0.4663976875828014,
+      "learning_rate": 0.00010672246979003573,
+      "loss": 0.73,
+      "mean_token_accuracy": 0.7802925348281861,
+      "num_tokens": 158283404.0,
+      "step": 46470
+    },
+    {
+      "epoch": 0.4664980529126019,
+      "learning_rate": 0.00010670239672407565,
+      "loss": 0.7161,
+      "mean_token_accuracy": 0.7881710827350616,
+      "num_tokens": 158318305.0,
+      "step": 46480
+    },
+    {
+      "epoch": 0.46659841824240234,
+      "learning_rate": 0.00010668232365811555,
+      "loss": 0.6878,
+      "mean_token_accuracy": 0.7952527344226837,
+      "num_tokens": 158353990.0,
+      "step": 46490
+    },
+    {
+      "epoch": 0.4666987835722028,
+      "learning_rate": 0.00010666225059215544,
+      "loss": 0.6749,
+      "mean_token_accuracy": 0.7952859818935394,
+      "num_tokens": 158389377.0,
+      "step": 46500
+    },
+    {
+      "epoch": 0.4667991489020033,
+      "learning_rate": 0.00010664217752619536,
+      "loss": 0.6919,
+      "mean_token_accuracy": 0.790702360868454,
+      "num_tokens": 158423925.0,
+      "step": 46510
+    },
+    {
+      "epoch": 0.46689951423180376,
+      "learning_rate": 0.00010662210446023526,
+      "loss": 0.6766,
+      "mean_token_accuracy": 0.7953005731105804,
+      "num_tokens": 158459573.0,
+      "step": 46520
+    },
+    {
+      "epoch": 0.4669998795616042,
+      "learning_rate": 0.00010660203139427515,
+      "loss": 0.7178,
+      "mean_token_accuracy": 0.7834354937076569,
+      "num_tokens": 158494715.0,
+      "step": 46530
+    },
+    {
+      "epoch": 0.4671002448914047,
+      "learning_rate": 0.00010658195832831508,
+      "loss": 0.7163,
+      "mean_token_accuracy": 0.7839456260204315,
+      "num_tokens": 158530318.0,
+      "step": 46540
+    },
+    {
+      "epoch": 0.4672006102212052,
+      "learning_rate": 0.00010656188526235498,
+      "loss": 0.6858,
+      "mean_token_accuracy": 0.795786327123642,
+      "num_tokens": 158566561.0,
+      "step": 46550
+    },
+    {
+      "epoch": 0.46730097555100564,
+      "learning_rate": 0.00010654181219639488,
+      "loss": 0.7185,
+      "mean_token_accuracy": 0.7875998079776764,
+      "num_tokens": 158602989.0,
+      "step": 46560
+    },
+    {
+      "epoch": 0.46740134088080615,
+      "learning_rate": 0.0001065217391304348,
+      "loss": 0.7199,
+      "mean_token_accuracy": 0.782318115234375,
+      "num_tokens": 158637538.0,
+      "step": 46570
+    },
+    {
+      "epoch": 0.4675017062106066,
+      "learning_rate": 0.0001065016660644747,
+      "loss": 0.7044,
+      "mean_token_accuracy": 0.7850953221321106,
+      "num_tokens": 158672562.0,
+      "step": 46580
+    },
+    {
+      "epoch": 0.46760207154040706,
+      "learning_rate": 0.00010648159299851459,
+      "loss": 0.6886,
+      "mean_token_accuracy": 0.7909720003604889,
+      "num_tokens": 158708099.0,
+      "step": 46590
+    },
+    {
+      "epoch": 0.46770243687020757,
+      "learning_rate": 0.00010646151993255451,
+      "loss": 0.7244,
+      "mean_token_accuracy": 0.7835875153541565,
+      "num_tokens": 158744023.0,
+      "step": 46600
+    },
+    {
+      "epoch": 0.467802802200008,
+      "learning_rate": 0.00010644144686659441,
+      "loss": 0.6923,
+      "mean_token_accuracy": 0.7904627203941346,
+      "num_tokens": 158780913.0,
+      "step": 46610
+    },
+    {
+      "epoch": 0.4679031675298085,
+      "learning_rate": 0.0001064213738006343,
+      "loss": 0.7,
+      "mean_token_accuracy": 0.7869254350662231,
+      "num_tokens": 158815378.0,
+      "step": 46620
+    },
+    {
+      "epoch": 0.468003532859609,
+      "learning_rate": 0.00010640130073467422,
+      "loss": 0.698,
+      "mean_token_accuracy": 0.7847469210624695,
+      "num_tokens": 158850652.0,
+      "step": 46630
+    },
+    {
+      "epoch": 0.46810389818940945,
+      "learning_rate": 0.00010638122766871412,
+      "loss": 0.729,
+      "mean_token_accuracy": 0.7812280833721161,
+      "num_tokens": 158885127.0,
+      "step": 46640
+    },
+    {
+      "epoch": 0.4682042635192099,
+      "learning_rate": 0.00010636115460275403,
+      "loss": 0.7087,
+      "mean_token_accuracy": 0.7840165257453918,
+      "num_tokens": 158920160.0,
+      "step": 46650
+    },
+    {
+      "epoch": 0.4683046288490104,
+      "learning_rate": 0.00010634108153679394,
+      "loss": 0.6999,
+      "mean_token_accuracy": 0.7896547734737396,
+      "num_tokens": 158956129.0,
+      "step": 46660
+    },
+    {
+      "epoch": 0.4684049941788109,
+      "learning_rate": 0.00010632100847083384,
+      "loss": 0.6897,
+      "mean_token_accuracy": 0.7955524504184723,
+      "num_tokens": 158992607.0,
+      "step": 46670
+    },
+    {
+      "epoch": 0.46850535950861133,
+      "learning_rate": 0.00010630093540487374,
+      "loss": 0.688,
+      "mean_token_accuracy": 0.791974002122879,
+      "num_tokens": 159028574.0,
+      "step": 46680
+    },
+    {
+      "epoch": 0.46860572483841184,
+      "learning_rate": 0.00010628086233891366,
+      "loss": 0.6904,
+      "mean_token_accuracy": 0.7901484429836273,
+      "num_tokens": 159063310.0,
+      "step": 46690
+    },
+    {
+      "epoch": 0.4687060901682123,
+      "learning_rate": 0.00010626078927295356,
+      "loss": 0.6924,
+      "mean_token_accuracy": 0.7883610904216767,
+      "num_tokens": 159098102.0,
+      "step": 46700
+    },
+    {
+      "epoch": 0.46880645549801275,
+      "learning_rate": 0.00010624071620699345,
+      "loss": 0.7042,
+      "mean_token_accuracy": 0.7841535389423371,
+      "num_tokens": 159133043.0,
+      "step": 46710
+    },
+    {
+      "epoch": 0.46890682082781326,
+      "learning_rate": 0.00010622064314103337,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7765646338462829,
+      "num_tokens": 159168109.0,
+      "step": 46720
+    },
+    {
+      "epoch": 0.4690071861576137,
+      "learning_rate": 0.00010620057007507327,
+      "loss": 0.6987,
+      "mean_token_accuracy": 0.7861594557762146,
+      "num_tokens": 159203514.0,
+      "step": 46730
+    },
+    {
+      "epoch": 0.4691075514874142,
+      "learning_rate": 0.00010618049700911316,
+      "loss": 0.6795,
+      "mean_token_accuracy": 0.7951503574848175,
+      "num_tokens": 159239365.0,
+      "step": 46740
+    },
+    {
+      "epoch": 0.4692079168172147,
+      "learning_rate": 0.00010616042394315308,
+      "loss": 0.7216,
+      "mean_token_accuracy": 0.7838666141033173,
+      "num_tokens": 159275307.0,
+      "step": 46750
+    },
+    {
+      "epoch": 0.46930828214701514,
+      "learning_rate": 0.00010614035087719298,
+      "loss": 0.707,
+      "mean_token_accuracy": 0.7920927166938782,
+      "num_tokens": 159310176.0,
+      "step": 46760
+    },
+    {
+      "epoch": 0.4694086474768156,
+      "learning_rate": 0.00010612027781123289,
+      "loss": 0.7161,
+      "mean_token_accuracy": 0.785585218667984,
+      "num_tokens": 159345957.0,
+      "step": 46770
+    },
+    {
+      "epoch": 0.46950901280661606,
+      "learning_rate": 0.0001061002047452728,
+      "loss": 0.7228,
+      "mean_token_accuracy": 0.7791460871696472,
+      "num_tokens": 159380901.0,
+      "step": 46780
+    },
+    {
+      "epoch": 0.46960937813641657,
+      "learning_rate": 0.0001060801316793127,
+      "loss": 0.7145,
+      "mean_token_accuracy": 0.7856329619884491,
+      "num_tokens": 159415532.0,
+      "step": 46790
+    },
+    {
+      "epoch": 0.469709743466217,
+      "learning_rate": 0.0001060600586133526,
+      "loss": 0.6804,
+      "mean_token_accuracy": 0.7934350728988647,
+      "num_tokens": 159451370.0,
+      "step": 46800
+    },
+    {
+      "epoch": 0.4698101087960175,
+      "learning_rate": 0.00010603998554739252,
+      "loss": 0.7475,
+      "mean_token_accuracy": 0.7766209721565247,
+      "num_tokens": 159486342.0,
+      "step": 46810
+    },
+    {
+      "epoch": 0.469910474125818,
+      "learning_rate": 0.00010601991248143242,
+      "loss": 0.7309,
+      "mean_token_accuracy": 0.7836766362190246,
+      "num_tokens": 159521532.0,
+      "step": 46820
+    },
+    {
+      "epoch": 0.47001083945561845,
+      "learning_rate": 0.00010599983941547234,
+      "loss": 0.7163,
+      "mean_token_accuracy": 0.7804810166358948,
+      "num_tokens": 159557383.0,
+      "step": 46830
+    },
+    {
+      "epoch": 0.4701112047854189,
+      "learning_rate": 0.00010597976634951223,
+      "loss": 0.7222,
+      "mean_token_accuracy": 0.7816921055316925,
+      "num_tokens": 159592790.0,
+      "step": 46840
+    },
+    {
+      "epoch": 0.4702115701152194,
+      "learning_rate": 0.00010595969328355213,
+      "loss": 0.7286,
+      "mean_token_accuracy": 0.783270001411438,
+      "num_tokens": 159629078.0,
+      "step": 46850
+    },
+    {
+      "epoch": 0.47031193544501987,
+      "learning_rate": 0.00010593962021759205,
+      "loss": 0.7284,
+      "mean_token_accuracy": 0.7808925092220307,
+      "num_tokens": 159665237.0,
+      "step": 46860
+    },
+    {
+      "epoch": 0.4704123007748203,
+      "learning_rate": 0.00010591954715163195,
+      "loss": 0.6923,
+      "mean_token_accuracy": 0.7907767117023468,
+      "num_tokens": 159700427.0,
+      "step": 46870
+    },
+    {
+      "epoch": 0.47051266610462084,
+      "learning_rate": 0.00010589947408567184,
+      "loss": 0.7066,
+      "mean_token_accuracy": 0.787431925535202,
+      "num_tokens": 159735443.0,
+      "step": 46880
+    },
+    {
+      "epoch": 0.4706130314344213,
+      "learning_rate": 0.00010587940101971176,
+      "loss": 0.7247,
+      "mean_token_accuracy": 0.7834123134613037,
+      "num_tokens": 159771058.0,
+      "step": 46890
+    },
+    {
+      "epoch": 0.47071339676422175,
+      "learning_rate": 0.00010585932795375167,
+      "loss": 0.7399,
+      "mean_token_accuracy": 0.7818175792694092,
+      "num_tokens": 159805741.0,
+      "step": 46900
+    },
+    {
+      "epoch": 0.47081376209402226,
+      "learning_rate": 0.00010583925488779157,
+      "loss": 0.6757,
+      "mean_token_accuracy": 0.7935092628002167,
+      "num_tokens": 159840654.0,
+      "step": 46910
+    },
+    {
+      "epoch": 0.4709141274238227,
+      "learning_rate": 0.00010581918182183149,
+      "loss": 0.7174,
+      "mean_token_accuracy": 0.781839793920517,
+      "num_tokens": 159875540.0,
+      "step": 46920
+    },
+    {
+      "epoch": 0.47101449275362317,
+      "learning_rate": 0.00010579910875587138,
+      "loss": 0.7264,
+      "mean_token_accuracy": 0.7855076014995575,
+      "num_tokens": 159911077.0,
+      "step": 46930
+    },
+    {
+      "epoch": 0.4711148580834237,
+      "learning_rate": 0.00010577903568991128,
+      "loss": 0.7224,
+      "mean_token_accuracy": 0.7857255399227142,
+      "num_tokens": 159946051.0,
+      "step": 46940
+    },
+    {
+      "epoch": 0.47121522341322414,
+      "learning_rate": 0.0001057589626239512,
+      "loss": 0.7131,
+      "mean_token_accuracy": 0.7874164462089539,
+      "num_tokens": 159980596.0,
+      "step": 46950
+    },
+    {
+      "epoch": 0.4713155887430246,
+      "learning_rate": 0.0001057388895579911,
+      "loss": 0.697,
+      "mean_token_accuracy": 0.7895408928394317,
+      "num_tokens": 160015468.0,
+      "step": 46960
+    },
+    {
+      "epoch": 0.4714159540728251,
+      "learning_rate": 0.00010571881649203099,
+      "loss": 0.6808,
+      "mean_token_accuracy": 0.794541311264038,
+      "num_tokens": 160051232.0,
+      "step": 46970
+    },
+    {
+      "epoch": 0.47151631940262556,
+      "learning_rate": 0.00010569874342607091,
+      "loss": 0.6809,
+      "mean_token_accuracy": 0.7936924457550049,
+      "num_tokens": 160086443.0,
+      "step": 46980
+    },
+    {
+      "epoch": 0.471616684732426,
+      "learning_rate": 0.0001056786703601108,
+      "loss": 0.7042,
+      "mean_token_accuracy": 0.7852703154087066,
+      "num_tokens": 160121781.0,
+      "step": 46990
+    },
+    {
+      "epoch": 0.47171705006222653,
+      "learning_rate": 0.00010565859729415071,
+      "loss": 0.7128,
+      "mean_token_accuracy": 0.7857403755187988,
+      "num_tokens": 160157537.0,
+      "step": 47000
+    },
+    {
+      "epoch": 0.471817415392027,
+      "learning_rate": 0.00010563852422819062,
+      "loss": 0.701,
+      "mean_token_accuracy": 0.7863259792327881,
+      "num_tokens": 160192578.0,
+      "step": 47010
+    },
+    {
+      "epoch": 0.47191778072182744,
+      "learning_rate": 0.00010561845116223053,
+      "loss": 0.7109,
+      "mean_token_accuracy": 0.790937626361847,
+      "num_tokens": 160227880.0,
+      "step": 47020
+    },
+    {
+      "epoch": 0.47201814605162795,
+      "learning_rate": 0.00010559837809627043,
+      "loss": 0.6927,
+      "mean_token_accuracy": 0.7863008499145507,
+      "num_tokens": 160263157.0,
+      "step": 47030
+    },
+    {
+      "epoch": 0.4721185113814284,
+      "learning_rate": 0.00010557830503031035,
+      "loss": 0.6976,
+      "mean_token_accuracy": 0.7901015520095825,
+      "num_tokens": 160298414.0,
+      "step": 47040
+    },
+    {
+      "epoch": 0.47221887671122886,
+      "learning_rate": 0.00010555823196435024,
+      "loss": 0.7208,
+      "mean_token_accuracy": 0.7872840464115143,
+      "num_tokens": 160333241.0,
+      "step": 47050
+    },
+    {
+      "epoch": 0.4723192420410293,
+      "learning_rate": 0.00010553815889839014,
+      "loss": 0.6925,
+      "mean_token_accuracy": 0.7923729300498963,
+      "num_tokens": 160368384.0,
+      "step": 47060
+    },
+    {
+      "epoch": 0.47241960737082983,
+      "learning_rate": 0.00010551808583243006,
+      "loss": 0.6982,
+      "mean_token_accuracy": 0.7913429319858551,
+      "num_tokens": 160404066.0,
+      "step": 47070
+    },
+    {
+      "epoch": 0.4725199727006303,
+      "learning_rate": 0.00010549801276646995,
+      "loss": 0.6821,
+      "mean_token_accuracy": 0.7913277566432952,
+      "num_tokens": 160439108.0,
+      "step": 47080
+    },
+    {
+      "epoch": 0.47262033803043074,
+      "learning_rate": 0.00010547793970050985,
+      "loss": 0.7312,
+      "mean_token_accuracy": 0.7841997802257538,
+      "num_tokens": 160474451.0,
+      "step": 47090
+    },
+    {
+      "epoch": 0.47272070336023125,
+      "learning_rate": 0.00010545786663454977,
+      "loss": 0.6899,
+      "mean_token_accuracy": 0.7895389914512634,
+      "num_tokens": 160511167.0,
+      "step": 47100
+    },
+    {
+      "epoch": 0.4728210686900317,
+      "learning_rate": 0.00010543779356858967,
+      "loss": 0.685,
+      "mean_token_accuracy": 0.7937025189399719,
+      "num_tokens": 160546665.0,
+      "step": 47110
+    },
+    {
+      "epoch": 0.47292143401983217,
+      "learning_rate": 0.00010541772050262957,
+      "loss": 0.6752,
+      "mean_token_accuracy": 0.7924473464488984,
+      "num_tokens": 160581892.0,
+      "step": 47120
+    },
+    {
+      "epoch": 0.4730217993496327,
+      "learning_rate": 0.00010539764743666948,
+      "loss": 0.6958,
+      "mean_token_accuracy": 0.7883609414100647,
+      "num_tokens": 160617929.0,
+      "step": 47130
+    },
+    {
+      "epoch": 0.47312216467943313,
+      "learning_rate": 0.00010537757437070939,
+      "loss": 0.6845,
+      "mean_token_accuracy": 0.7942955136299134,
+      "num_tokens": 160653461.0,
+      "step": 47140
+    },
+    {
+      "epoch": 0.4732225300092336,
+      "learning_rate": 0.00010535750130474929,
+      "loss": 0.6831,
+      "mean_token_accuracy": 0.7937300324440002,
+      "num_tokens": 160690001.0,
+      "step": 47150
+    },
+    {
+      "epoch": 0.4733228953390341,
+      "learning_rate": 0.00010533742823878921,
+      "loss": 0.6631,
+      "mean_token_accuracy": 0.8001127362251281,
+      "num_tokens": 160725785.0,
+      "step": 47160
+    },
+    {
+      "epoch": 0.47342326066883456,
+      "learning_rate": 0.0001053173551728291,
+      "loss": 0.7243,
+      "mean_token_accuracy": 0.7867645204067231,
+      "num_tokens": 160762595.0,
+      "step": 47170
+    },
+    {
+      "epoch": 0.473523625998635,
+      "learning_rate": 0.000105297282106869,
+      "loss": 0.6938,
+      "mean_token_accuracy": 0.7910487055778503,
+      "num_tokens": 160797614.0,
+      "step": 47180
+    },
+    {
+      "epoch": 0.4736239913284355,
+      "learning_rate": 0.00010527720904090892,
+      "loss": 0.6979,
+      "mean_token_accuracy": 0.7920831203460693,
+      "num_tokens": 160833634.0,
+      "step": 47190
+    },
+    {
+      "epoch": 0.473724356658236,
+      "learning_rate": 0.00010525713597494881,
+      "loss": 0.7532,
+      "mean_token_accuracy": 0.7816699028015137,
+      "num_tokens": 160869280.0,
+      "step": 47200
+    },
+    {
+      "epoch": 0.47382472198803643,
+      "learning_rate": 0.00010523706290898871,
+      "loss": 0.6871,
+      "mean_token_accuracy": 0.7951245188713074,
+      "num_tokens": 160905819.0,
+      "step": 47210
+    },
+    {
+      "epoch": 0.47392508731783695,
+      "learning_rate": 0.00010521698984302863,
+      "loss": 0.6751,
+      "mean_token_accuracy": 0.7951266944408417,
+      "num_tokens": 160942206.0,
+      "step": 47220
+    },
+    {
+      "epoch": 0.4740254526476374,
+      "learning_rate": 0.00010519691677706853,
+      "loss": 0.7116,
+      "mean_token_accuracy": 0.7847038805484772,
+      "num_tokens": 160978577.0,
+      "step": 47230
+    },
+    {
+      "epoch": 0.47412581797743786,
+      "learning_rate": 0.00010517684371110844,
+      "loss": 0.6958,
+      "mean_token_accuracy": 0.7886900484561921,
+      "num_tokens": 161013465.0,
+      "step": 47240
+    },
+    {
+      "epoch": 0.47422618330723837,
+      "learning_rate": 0.00010515677064514834,
+      "loss": 0.6786,
+      "mean_token_accuracy": 0.7906596243381501,
+      "num_tokens": 161049466.0,
+      "step": 47250
+    },
+    {
+      "epoch": 0.4743265486370388,
+      "learning_rate": 0.00010513669757918825,
+      "loss": 0.6674,
+      "mean_token_accuracy": 0.7982393562793731,
+      "num_tokens": 161085007.0,
+      "step": 47260
+    },
+    {
+      "epoch": 0.4744269139668393,
+      "learning_rate": 0.00010511662451322815,
+      "loss": 0.6767,
+      "mean_token_accuracy": 0.7966917872428894,
+      "num_tokens": 161121639.0,
+      "step": 47270
+    },
+    {
+      "epoch": 0.4745272792966398,
+      "learning_rate": 0.00010509655144726807,
+      "loss": 0.71,
+      "mean_token_accuracy": 0.7855958700180053,
+      "num_tokens": 161158615.0,
+      "step": 47280
+    },
+    {
+      "epoch": 0.47462764462644025,
+      "learning_rate": 0.00010507647838130796,
+      "loss": 0.6909,
+      "mean_token_accuracy": 0.7912957191467285,
+      "num_tokens": 161194038.0,
+      "step": 47290
+    },
+    {
+      "epoch": 0.4747280099562407,
+      "learning_rate": 0.00010505640531534786,
+      "loss": 0.6953,
+      "mean_token_accuracy": 0.7897424101829529,
+      "num_tokens": 161229845.0,
+      "step": 47300
+    },
+    {
+      "epoch": 0.4748283752860412,
+      "learning_rate": 0.00010503633224938778,
+      "loss": 0.7494,
+      "mean_token_accuracy": 0.773971438407898,
+      "num_tokens": 161266003.0,
+      "step": 47310
+    },
+    {
+      "epoch": 0.47492874061584167,
+      "learning_rate": 0.00010501625918342768,
+      "loss": 0.7151,
+      "mean_token_accuracy": 0.7860597670078278,
+      "num_tokens": 161301194.0,
+      "step": 47320
+    },
+    {
+      "epoch": 0.4750291059456421,
+      "learning_rate": 0.00010499618611746757,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8060030400753021,
+      "num_tokens": 161337294.0,
+      "step": 47330
+    },
+    {
+      "epoch": 0.47512947127544264,
+      "learning_rate": 0.00010497611305150749,
+      "loss": 0.6808,
+      "mean_token_accuracy": 0.7956064879894257,
+      "num_tokens": 161373107.0,
+      "step": 47340
+    },
+    {
+      "epoch": 0.4752298366052431,
+      "learning_rate": 0.0001049560399855474,
+      "loss": 0.7048,
+      "mean_token_accuracy": 0.7940467596054077,
+      "num_tokens": 161408814.0,
+      "step": 47350
+    },
+    {
+      "epoch": 0.47533020193504355,
+      "learning_rate": 0.0001049359669195873,
+      "loss": 0.7486,
+      "mean_token_accuracy": 0.7760293006896972,
+      "num_tokens": 161443917.0,
+      "step": 47360
+    },
+    {
+      "epoch": 0.475430567264844,
+      "learning_rate": 0.00010491589385362722,
+      "loss": 0.7265,
+      "mean_token_accuracy": 0.7832448840141296,
+      "num_tokens": 161479605.0,
+      "step": 47370
+    },
+    {
+      "epoch": 0.4755309325946445,
+      "learning_rate": 0.00010489582078766711,
+      "loss": 0.7102,
+      "mean_token_accuracy": 0.7841754913330078,
+      "num_tokens": 161515285.0,
+      "step": 47380
+    },
+    {
+      "epoch": 0.475631297924445,
+      "learning_rate": 0.00010487574772170703,
+      "loss": 0.6772,
+      "mean_token_accuracy": 0.7950868487358094,
+      "num_tokens": 161550831.0,
+      "step": 47390
+    },
+    {
+      "epoch": 0.47573166325424543,
+      "learning_rate": 0.00010485567465574693,
+      "loss": 0.7081,
+      "mean_token_accuracy": 0.7866260468959808,
+      "num_tokens": 161586881.0,
+      "step": 47400
+    },
+    {
+      "epoch": 0.47583202858404594,
+      "learning_rate": 0.00010483560158978682,
+      "loss": 0.6949,
+      "mean_token_accuracy": 0.7889688670635223,
+      "num_tokens": 161622634.0,
+      "step": 47410
+    },
+    {
+      "epoch": 0.4759323939138464,
+      "learning_rate": 0.00010481552852382675,
+      "loss": 0.7066,
+      "mean_token_accuracy": 0.7932239294052124,
+      "num_tokens": 161659136.0,
+      "step": 47420
+    },
+    {
+      "epoch": 0.47603275924364685,
+      "learning_rate": 0.00010479545545786664,
+      "loss": 0.7152,
+      "mean_token_accuracy": 0.7909041404724121,
+      "num_tokens": 161694868.0,
+      "step": 47430
+    },
+    {
+      "epoch": 0.47613312457344736,
+      "learning_rate": 0.00010477538239190654,
+      "loss": 0.712,
+      "mean_token_accuracy": 0.7855082929134369,
+      "num_tokens": 161730614.0,
+      "step": 47440
+    },
+    {
+      "epoch": 0.4762334899032478,
+      "learning_rate": 0.00010475530932594646,
+      "loss": 0.6949,
+      "mean_token_accuracy": 0.7938051342964172,
+      "num_tokens": 161767012.0,
+      "step": 47450
+    },
+    {
+      "epoch": 0.4763338552330483,
+      "learning_rate": 0.00010473523625998635,
+      "loss": 0.6633,
+      "mean_token_accuracy": 0.796594899892807,
+      "num_tokens": 161803108.0,
+      "step": 47460
+    },
+    {
+      "epoch": 0.4764342205628488,
+      "learning_rate": 0.00010471516319402626,
+      "loss": 0.7105,
+      "mean_token_accuracy": 0.782571405172348,
+      "num_tokens": 161838645.0,
+      "step": 47470
+    },
+    {
+      "epoch": 0.47653458589264924,
+      "learning_rate": 0.00010469509012806617,
+      "loss": 0.7094,
+      "mean_token_accuracy": 0.7875165998935699,
+      "num_tokens": 161874286.0,
+      "step": 47480
+    },
+    {
+      "epoch": 0.4766349512224497,
+      "learning_rate": 0.00010467501706210608,
+      "loss": 0.7128,
+      "mean_token_accuracy": 0.7863965630531311,
+      "num_tokens": 161909907.0,
+      "step": 47490
+    },
+    {
+      "epoch": 0.4767353165522502,
+      "learning_rate": 0.00010465494399614597,
+      "loss": 0.7182,
+      "mean_token_accuracy": 0.7878428757190704,
+      "num_tokens": 161944935.0,
+      "step": 47500
+    },
+    {
+      "epoch": 0.47683568188205067,
+      "learning_rate": 0.0001046348709301859,
+      "loss": 0.6971,
+      "mean_token_accuracy": 0.794540810585022,
+      "num_tokens": 161981262.0,
+      "step": 47510
+    },
+    {
+      "epoch": 0.4769360472118511,
+      "learning_rate": 0.00010461479786422579,
+      "loss": 0.708,
+      "mean_token_accuracy": 0.7874243080615997,
+      "num_tokens": 162016280.0,
+      "step": 47520
+    },
+    {
+      "epoch": 0.47703641254165163,
+      "learning_rate": 0.00010459472479826568,
+      "loss": 0.6525,
+      "mean_token_accuracy": 0.8007276117801666,
+      "num_tokens": 162051253.0,
+      "step": 47530
+    },
+    {
+      "epoch": 0.4771367778714521,
+      "learning_rate": 0.00010457465173230561,
+      "loss": 0.7023,
+      "mean_token_accuracy": 0.7896904945373535,
+      "num_tokens": 162087552.0,
+      "step": 47540
+    },
+    {
+      "epoch": 0.47723714320125254,
+      "learning_rate": 0.0001045545786663455,
+      "loss": 0.6924,
+      "mean_token_accuracy": 0.7869489848613739,
+      "num_tokens": 162122832.0,
+      "step": 47550
+    },
+    {
+      "epoch": 0.47733750853105306,
+      "learning_rate": 0.0001045345056003854,
+      "loss": 0.6936,
+      "mean_token_accuracy": 0.7942219018936157,
+      "num_tokens": 162158731.0,
+      "step": 47560
+    },
+    {
+      "epoch": 0.4774378738608535,
+      "learning_rate": 0.00010451443253442532,
+      "loss": 0.6841,
+      "mean_token_accuracy": 0.7955293655395508,
+      "num_tokens": 162194422.0,
+      "step": 47570
+    },
+    {
+      "epoch": 0.47753823919065397,
+      "learning_rate": 0.00010449435946846521,
+      "loss": 0.7177,
+      "mean_token_accuracy": 0.7842896938323974,
+      "num_tokens": 162230673.0,
+      "step": 47580
+    },
+    {
+      "epoch": 0.4776386045204545,
+      "learning_rate": 0.00010447428640250512,
+      "loss": 0.7382,
+      "mean_token_accuracy": 0.7757864713668823,
+      "num_tokens": 162266350.0,
+      "step": 47590
+    },
+    {
+      "epoch": 0.47773896985025494,
+      "learning_rate": 0.00010445421333654503,
+      "loss": 0.7236,
+      "mean_token_accuracy": 0.7828563153743744,
+      "num_tokens": 162301576.0,
+      "step": 47600
+    },
+    {
+      "epoch": 0.4778393351800554,
+      "learning_rate": 0.00010443414027058494,
+      "loss": 0.6793,
+      "mean_token_accuracy": 0.7924531519412994,
+      "num_tokens": 162337268.0,
+      "step": 47610
+    },
+    {
+      "epoch": 0.4779397005098559,
+      "learning_rate": 0.00010441406720462483,
+      "loss": 0.6919,
+      "mean_token_accuracy": 0.7864301800727844,
+      "num_tokens": 162372409.0,
+      "step": 47620
+    },
+    {
+      "epoch": 0.47804006583965636,
+      "learning_rate": 0.00010439399413866476,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.7930449903011322,
+      "num_tokens": 162408319.0,
+      "step": 47630
+    },
+    {
+      "epoch": 0.4781404311694568,
+      "learning_rate": 0.00010437392107270465,
+      "loss": 0.6903,
+      "mean_token_accuracy": 0.79172642827034,
+      "num_tokens": 162444898.0,
+      "step": 47640
+    },
+    {
+      "epoch": 0.47824079649925727,
+      "learning_rate": 0.00010435384800674455,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.7925096869468689,
+      "num_tokens": 162480532.0,
+      "step": 47650
+    },
+    {
+      "epoch": 0.4783411618290578,
+      "learning_rate": 0.00010433377494078447,
+      "loss": 0.687,
+      "mean_token_accuracy": 0.7912269413471222,
+      "num_tokens": 162516434.0,
+      "step": 47660
+    },
+    {
+      "epoch": 0.47844152715885824,
+      "learning_rate": 0.00010431370187482436,
+      "loss": 0.7054,
+      "mean_token_accuracy": 0.7908867537975312,
+      "num_tokens": 162552621.0,
+      "step": 47670
+    },
+    {
+      "epoch": 0.4785418924886587,
+      "learning_rate": 0.00010429362880886426,
+      "loss": 0.6796,
+      "mean_token_accuracy": 0.7933435320854187,
+      "num_tokens": 162588410.0,
+      "step": 47680
+    },
+    {
+      "epoch": 0.4786422578184592,
+      "learning_rate": 0.00010427355574290418,
+      "loss": 0.6498,
+      "mean_token_accuracy": 0.8042173683643341,
+      "num_tokens": 162625530.0,
+      "step": 47690
+    },
+    {
+      "epoch": 0.47874262314825966,
+      "learning_rate": 0.00010425348267694407,
+      "loss": 0.6755,
+      "mean_token_accuracy": 0.7967062354087829,
+      "num_tokens": 162661367.0,
+      "step": 47700
+    },
+    {
+      "epoch": 0.4788429884780601,
+      "learning_rate": 0.00010423340961098398,
+      "loss": 0.7098,
+      "mean_token_accuracy": 0.7858090043067932,
+      "num_tokens": 162696852.0,
+      "step": 47710
+    },
+    {
+      "epoch": 0.47894335380786063,
+      "learning_rate": 0.0001042133365450239,
+      "loss": 0.6726,
+      "mean_token_accuracy": 0.7962167382240295,
+      "num_tokens": 162732782.0,
+      "step": 47720
+    },
+    {
+      "epoch": 0.4790437191376611,
+      "learning_rate": 0.0001041932634790638,
+      "loss": 0.6759,
+      "mean_token_accuracy": 0.7944247961044312,
+      "num_tokens": 162768668.0,
+      "step": 47730
+    },
+    {
+      "epoch": 0.47914408446746154,
+      "learning_rate": 0.0001041731904131037,
+      "loss": 0.7139,
+      "mean_token_accuracy": 0.785028612613678,
+      "num_tokens": 162803623.0,
+      "step": 47740
+    },
+    {
+      "epoch": 0.47924444979726205,
+      "learning_rate": 0.00010415311734714362,
+      "loss": 0.6542,
+      "mean_token_accuracy": 0.800380527973175,
+      "num_tokens": 162839575.0,
+      "step": 47750
+    },
+    {
+      "epoch": 0.4793448151270625,
+      "learning_rate": 0.00010413304428118351,
+      "loss": 0.6459,
+      "mean_token_accuracy": 0.7978033125400543,
+      "num_tokens": 162876112.0,
+      "step": 47760
+    },
+    {
+      "epoch": 0.47944518045686296,
+      "learning_rate": 0.0001041129712152234,
+      "loss": 0.7162,
+      "mean_token_accuracy": 0.7856225728988647,
+      "num_tokens": 162911288.0,
+      "step": 47770
+    },
+    {
+      "epoch": 0.4795455457866635,
+      "learning_rate": 0.00010409289814926333,
+      "loss": 0.6821,
+      "mean_token_accuracy": 0.7910649538040161,
+      "num_tokens": 162946297.0,
+      "step": 47780
+    },
+    {
+      "epoch": 0.47964591111646393,
+      "learning_rate": 0.00010407282508330322,
+      "loss": 0.72,
+      "mean_token_accuracy": 0.7854718923568725,
+      "num_tokens": 162981256.0,
+      "step": 47790
+    },
+    {
+      "epoch": 0.4797462764462644,
+      "learning_rate": 0.00010405275201734313,
+      "loss": 0.6887,
+      "mean_token_accuracy": 0.7937243342399597,
+      "num_tokens": 163016659.0,
+      "step": 47800
+    },
+    {
+      "epoch": 0.4798466417760649,
+      "learning_rate": 0.00010403267895138304,
+      "loss": 0.6769,
+      "mean_token_accuracy": 0.7947654485702514,
+      "num_tokens": 163052921.0,
+      "step": 47810
+    },
+    {
+      "epoch": 0.47994700710586535,
+      "learning_rate": 0.00010401260588542295,
+      "loss": 0.6673,
+      "mean_token_accuracy": 0.799593985080719,
+      "num_tokens": 163089902.0,
+      "step": 47820
+    },
+    {
+      "epoch": 0.4800473724356658,
+      "learning_rate": 0.00010399253281946284,
+      "loss": 0.684,
+      "mean_token_accuracy": 0.7932468891143799,
+      "num_tokens": 163125318.0,
+      "step": 47830
+    },
+    {
+      "epoch": 0.4801477377654663,
+      "learning_rate": 0.00010397245975350277,
+      "loss": 0.742,
+      "mean_token_accuracy": 0.7809097528457641,
+      "num_tokens": 163160809.0,
+      "step": 47840
+    },
+    {
+      "epoch": 0.4802481030952668,
+      "learning_rate": 0.00010395238668754266,
+      "loss": 0.6726,
+      "mean_token_accuracy": 0.7962616145610809,
+      "num_tokens": 163195946.0,
+      "step": 47850
+    },
+    {
+      "epoch": 0.48034846842506723,
+      "learning_rate": 0.00010393231362158255,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.7961674690246582,
+      "num_tokens": 163231686.0,
+      "step": 47860
+    },
+    {
+      "epoch": 0.48044883375486774,
+      "learning_rate": 0.00010391224055562248,
+      "loss": 0.6822,
+      "mean_token_accuracy": 0.7897140741348266,
+      "num_tokens": 163268030.0,
+      "step": 47870
+    },
+    {
+      "epoch": 0.4805491990846682,
+      "learning_rate": 0.00010389216748966237,
+      "loss": 0.692,
+      "mean_token_accuracy": 0.7926612317562103,
+      "num_tokens": 163303833.0,
+      "step": 47880
+    },
+    {
+      "epoch": 0.48064956441446866,
+      "learning_rate": 0.00010387209442370227,
+      "loss": 0.7397,
+      "mean_token_accuracy": 0.7783870279788971,
+      "num_tokens": 163338956.0,
+      "step": 47890
+    },
+    {
+      "epoch": 0.48074992974426917,
+      "learning_rate": 0.00010385202135774219,
+      "loss": 0.6783,
+      "mean_token_accuracy": 0.7934011518955231,
+      "num_tokens": 163375923.0,
+      "step": 47900
+    },
+    {
+      "epoch": 0.4808502950740696,
+      "learning_rate": 0.00010383194829178208,
+      "loss": 0.6754,
+      "mean_token_accuracy": 0.7942137479782104,
+      "num_tokens": 163410955.0,
+      "step": 47910
+    },
+    {
+      "epoch": 0.4809506604038701,
+      "learning_rate": 0.00010381187522582199,
+      "loss": 0.6776,
+      "mean_token_accuracy": 0.7968982100486756,
+      "num_tokens": 163445528.0,
+      "step": 47920
+    },
+    {
+      "epoch": 0.48105102573367053,
+      "learning_rate": 0.0001037918021598619,
+      "loss": 0.6751,
+      "mean_token_accuracy": 0.7926284074783325,
+      "num_tokens": 163481903.0,
+      "step": 47930
+    },
+    {
+      "epoch": 0.48115139106347105,
+      "learning_rate": 0.00010377172909390181,
+      "loss": 0.6576,
+      "mean_token_accuracy": 0.7997497975826263,
+      "num_tokens": 163518436.0,
+      "step": 47940
+    },
+    {
+      "epoch": 0.4812517563932715,
+      "learning_rate": 0.00010375165602794172,
+      "loss": 0.7369,
+      "mean_token_accuracy": 0.7813829720020294,
+      "num_tokens": 163554326.0,
+      "step": 47950
+    },
+    {
+      "epoch": 0.48135212172307196,
+      "learning_rate": 0.00010373158296198163,
+      "loss": 0.7072,
+      "mean_token_accuracy": 0.7888287842273712,
+      "num_tokens": 163590652.0,
+      "step": 47960
+    },
+    {
+      "epoch": 0.48145248705287247,
+      "learning_rate": 0.00010371150989602152,
+      "loss": 0.6511,
+      "mean_token_accuracy": 0.7944359362125397,
+      "num_tokens": 163626494.0,
+      "step": 47970
+    },
+    {
+      "epoch": 0.4815528523826729,
+      "learning_rate": 0.00010369143683006144,
+      "loss": 0.6718,
+      "mean_token_accuracy": 0.7970012128353119,
+      "num_tokens": 163661563.0,
+      "step": 47980
+    },
+    {
+      "epoch": 0.4816532177124734,
+      "learning_rate": 0.00010367136376410134,
+      "loss": 0.6784,
+      "mean_token_accuracy": 0.7969679594039917,
+      "num_tokens": 163697040.0,
+      "step": 47990
+    },
+    {
+      "epoch": 0.4817535830422739,
+      "learning_rate": 0.00010365129069814123,
+      "loss": 0.681,
+      "mean_token_accuracy": 0.7943235516548157,
+      "num_tokens": 163732537.0,
+      "step": 48000
+    },
+    {
+      "epoch": 0.48185394837207435,
+      "learning_rate": 0.00010363121763218115,
+      "loss": 0.6951,
+      "mean_token_accuracy": 0.7890613734722137,
+      "num_tokens": 163768361.0,
+      "step": 48010
+    },
+    {
+      "epoch": 0.4819543137018748,
+      "learning_rate": 0.00010361114456622105,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.7916137874126434,
+      "num_tokens": 163803475.0,
+      "step": 48020
+    },
+    {
+      "epoch": 0.4820546790316753,
+      "learning_rate": 0.00010359107150026094,
+      "loss": 0.6598,
+      "mean_token_accuracy": 0.8004237949848175,
+      "num_tokens": 163839338.0,
+      "step": 48030
+    },
+    {
+      "epoch": 0.48215504436147577,
+      "learning_rate": 0.00010357099843430087,
+      "loss": 0.6912,
+      "mean_token_accuracy": 0.7930728733539582,
+      "num_tokens": 163874577.0,
+      "step": 48040
+    },
+    {
+      "epoch": 0.4822554096912762,
+      "learning_rate": 0.00010355092536834076,
+      "loss": 0.7043,
+      "mean_token_accuracy": 0.789654529094696,
+      "num_tokens": 163910646.0,
+      "step": 48050
+    },
+    {
+      "epoch": 0.48235577502107674,
+      "learning_rate": 0.00010353085230238067,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.7975339889526367,
+      "num_tokens": 163946744.0,
+      "step": 48060
+    },
+    {
+      "epoch": 0.4824561403508772,
+      "learning_rate": 0.00010351077923642059,
+      "loss": 0.6915,
+      "mean_token_accuracy": 0.789144366979599,
+      "num_tokens": 163982922.0,
+      "step": 48070
+    },
+    {
+      "epoch": 0.48255650568067765,
+      "learning_rate": 0.00010349070617046049,
+      "loss": 0.6586,
+      "mean_token_accuracy": 0.8003958880901336,
+      "num_tokens": 164018490.0,
+      "step": 48080
+    },
+    {
+      "epoch": 0.48265687101047816,
+      "learning_rate": 0.00010347063310450038,
+      "loss": 0.6936,
+      "mean_token_accuracy": 0.7923797130584717,
+      "num_tokens": 164053312.0,
+      "step": 48090
+    },
+    {
+      "epoch": 0.4827572363402786,
+      "learning_rate": 0.0001034505600385403,
+      "loss": 0.7184,
+      "mean_token_accuracy": 0.7866088509559631,
+      "num_tokens": 164088516.0,
+      "step": 48100
+    },
+    {
+      "epoch": 0.4828576016700791,
+      "learning_rate": 0.0001034304869725802,
+      "loss": 0.6624,
+      "mean_token_accuracy": 0.7987448096275329,
+      "num_tokens": 164125498.0,
+      "step": 48110
+    },
+    {
+      "epoch": 0.4829579669998796,
+      "learning_rate": 0.00010341041390662009,
+      "loss": 0.6403,
+      "mean_token_accuracy": 0.804091477394104,
+      "num_tokens": 164161279.0,
+      "step": 48120
+    },
+    {
+      "epoch": 0.48305833232968004,
+      "learning_rate": 0.00010339034084066001,
+      "loss": 0.727,
+      "mean_token_accuracy": 0.781208735704422,
+      "num_tokens": 164196114.0,
+      "step": 48130
+    },
+    {
+      "epoch": 0.4831586976594805,
+      "learning_rate": 0.00010337026777469991,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.7808577239513397,
+      "num_tokens": 164231733.0,
+      "step": 48140
+    },
+    {
+      "epoch": 0.483259062989281,
+      "learning_rate": 0.00010335019470873982,
+      "loss": 0.673,
+      "mean_token_accuracy": 0.7946267187595367,
+      "num_tokens": 164267460.0,
+      "step": 48150
+    },
+    {
+      "epoch": 0.48335942831908146,
+      "learning_rate": 0.00010333012164277973,
+      "loss": 0.679,
+      "mean_token_accuracy": 0.795447838306427,
+      "num_tokens": 164302051.0,
+      "step": 48160
+    },
+    {
+      "epoch": 0.4834597936488819,
+      "learning_rate": 0.00010331004857681964,
+      "loss": 0.6521,
+      "mean_token_accuracy": 0.7975180566310882,
+      "num_tokens": 164337925.0,
+      "step": 48170
+    },
+    {
+      "epoch": 0.48356015897868243,
+      "learning_rate": 0.00010328997551085953,
+      "loss": 0.6621,
+      "mean_token_accuracy": 0.8010074675083161,
+      "num_tokens": 164373647.0,
+      "step": 48180
+    },
+    {
+      "epoch": 0.4836605243084829,
+      "learning_rate": 0.00010326990244489945,
+      "loss": 0.6718,
+      "mean_token_accuracy": 0.7939490079879761,
+      "num_tokens": 164409199.0,
+      "step": 48190
+    },
+    {
+      "epoch": 0.48376088963828334,
+      "learning_rate": 0.00010324982937893935,
+      "loss": 0.6579,
+      "mean_token_accuracy": 0.7961607813835144,
+      "num_tokens": 164445741.0,
+      "step": 48200
+    },
+    {
+      "epoch": 0.4838612549680838,
+      "learning_rate": 0.00010322975631297924,
+      "loss": 0.6565,
+      "mean_token_accuracy": 0.7984263598918915,
+      "num_tokens": 164482421.0,
+      "step": 48210
+    },
+    {
+      "epoch": 0.4839616202978843,
+      "learning_rate": 0.00010320968324701916,
+      "loss": 0.6909,
+      "mean_token_accuracy": 0.7874985694885254,
+      "num_tokens": 164517951.0,
+      "step": 48220
+    },
+    {
+      "epoch": 0.48406198562768477,
+      "learning_rate": 0.00010318961018105906,
+      "loss": 0.6705,
+      "mean_token_accuracy": 0.797230851650238,
+      "num_tokens": 164554573.0,
+      "step": 48230
+    },
+    {
+      "epoch": 0.4841623509574852,
+      "learning_rate": 0.00010316953711509895,
+      "loss": 0.6821,
+      "mean_token_accuracy": 0.795600700378418,
+      "num_tokens": 164590873.0,
+      "step": 48240
+    },
+    {
+      "epoch": 0.48426271628728573,
+      "learning_rate": 0.00010314946404913888,
+      "loss": 0.7072,
+      "mean_token_accuracy": 0.7906558811664581,
+      "num_tokens": 164627073.0,
+      "step": 48250
+    },
+    {
+      "epoch": 0.4843630816170862,
+      "learning_rate": 0.00010312939098317877,
+      "loss": 0.6872,
+      "mean_token_accuracy": 0.7920962035655975,
+      "num_tokens": 164662493.0,
+      "step": 48260
+    },
+    {
+      "epoch": 0.48446344694688664,
+      "learning_rate": 0.00010310931791721868,
+      "loss": 0.6774,
+      "mean_token_accuracy": 0.7944299578666687,
+      "num_tokens": 164698299.0,
+      "step": 48270
+    },
+    {
+      "epoch": 0.48456381227668716,
+      "learning_rate": 0.00010308924485125859,
+      "loss": 0.7223,
+      "mean_token_accuracy": 0.7846730351448059,
+      "num_tokens": 164734390.0,
+      "step": 48280
+    },
+    {
+      "epoch": 0.4846641776064876,
+      "learning_rate": 0.0001030691717852985,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8061159372329711,
+      "num_tokens": 164770787.0,
+      "step": 48290
+    },
+    {
+      "epoch": 0.48476454293628807,
+      "learning_rate": 0.00010304909871933839,
+      "loss": 0.664,
+      "mean_token_accuracy": 0.7991957008838654,
+      "num_tokens": 164806877.0,
+      "step": 48300
+    },
+    {
+      "epoch": 0.4848649082660886,
+      "learning_rate": 0.00010302902565337831,
+      "loss": 0.6992,
+      "mean_token_accuracy": 0.7914974689483643,
+      "num_tokens": 164842674.0,
+      "step": 48310
+    },
+    {
+      "epoch": 0.48496527359588903,
+      "learning_rate": 0.00010300895258741821,
+      "loss": 0.6354,
+      "mean_token_accuracy": 0.806444126367569,
+      "num_tokens": 164878725.0,
+      "step": 48320
+    },
+    {
+      "epoch": 0.4850656389256895,
+      "learning_rate": 0.0001029888795214581,
+      "loss": 0.7072,
+      "mean_token_accuracy": 0.7881014108657837,
+      "num_tokens": 164914354.0,
+      "step": 48330
+    },
+    {
+      "epoch": 0.48516600425549,
+      "learning_rate": 0.00010296880645549802,
+      "loss": 0.6931,
+      "mean_token_accuracy": 0.7914582073688508,
+      "num_tokens": 164951175.0,
+      "step": 48340
+    },
+    {
+      "epoch": 0.48526636958529046,
+      "learning_rate": 0.00010294873338953792,
+      "loss": 0.6771,
+      "mean_token_accuracy": 0.7916079461574554,
+      "num_tokens": 164988217.0,
+      "step": 48350
+    },
+    {
+      "epoch": 0.4853667349150909,
+      "learning_rate": 0.00010292866032357781,
+      "loss": 0.6617,
+      "mean_token_accuracy": 0.7961135387420655,
+      "num_tokens": 165023908.0,
+      "step": 48360
+    },
+    {
+      "epoch": 0.4854671002448914,
+      "learning_rate": 0.00010290858725761774,
+      "loss": 0.6929,
+      "mean_token_accuracy": 0.7928700387477875,
+      "num_tokens": 165059954.0,
+      "step": 48370
+    },
+    {
+      "epoch": 0.4855674655746919,
+      "learning_rate": 0.00010288851419165763,
+      "loss": 0.6912,
+      "mean_token_accuracy": 0.7901743710041046,
+      "num_tokens": 165096806.0,
+      "step": 48380
+    },
+    {
+      "epoch": 0.48566783090449234,
+      "learning_rate": 0.00010286844112569754,
+      "loss": 0.694,
+      "mean_token_accuracy": 0.7921668708324432,
+      "num_tokens": 165132310.0,
+      "step": 48390
+    },
+    {
+      "epoch": 0.48576819623429285,
+      "learning_rate": 0.00010284836805973745,
+      "loss": 0.6701,
+      "mean_token_accuracy": 0.7965425670146942,
+      "num_tokens": 165168260.0,
+      "step": 48400
+    },
+    {
+      "epoch": 0.4858685615640933,
+      "learning_rate": 0.00010282829499377736,
+      "loss": 0.6903,
+      "mean_token_accuracy": 0.793122661113739,
+      "num_tokens": 165204508.0,
+      "step": 48410
+    },
+    {
+      "epoch": 0.48596892689389376,
+      "learning_rate": 0.00010280822192781725,
+      "loss": 0.6573,
+      "mean_token_accuracy": 0.800396591424942,
+      "num_tokens": 165240999.0,
+      "step": 48420
+    },
+    {
+      "epoch": 0.48606929222369427,
+      "learning_rate": 0.00010278814886185717,
+      "loss": 0.6445,
+      "mean_token_accuracy": 0.800655859708786,
+      "num_tokens": 165277706.0,
+      "step": 48430
+    },
+    {
+      "epoch": 0.4861696575534947,
+      "learning_rate": 0.00010276807579589707,
+      "loss": 0.6958,
+      "mean_token_accuracy": 0.7916760146617889,
+      "num_tokens": 165312787.0,
+      "step": 48440
+    },
+    {
+      "epoch": 0.4862700228832952,
+      "learning_rate": 0.00010274800272993696,
+      "loss": 0.6804,
+      "mean_token_accuracy": 0.7930470466613769,
+      "num_tokens": 165348317.0,
+      "step": 48450
+    },
+    {
+      "epoch": 0.4863703882130957,
+      "learning_rate": 0.00010272792966397688,
+      "loss": 0.6506,
+      "mean_token_accuracy": 0.8018568575382232,
+      "num_tokens": 165385040.0,
+      "step": 48460
+    },
+    {
+      "epoch": 0.48647075354289615,
+      "learning_rate": 0.00010270785659801678,
+      "loss": 0.6931,
+      "mean_token_accuracy": 0.783112621307373,
+      "num_tokens": 165421786.0,
+      "step": 48470
+    },
+    {
+      "epoch": 0.4865711188726966,
+      "learning_rate": 0.00010268778353205667,
+      "loss": 0.6831,
+      "mean_token_accuracy": 0.7922792792320251,
+      "num_tokens": 165457647.0,
+      "step": 48480
+    },
+    {
+      "epoch": 0.48667148420249706,
+      "learning_rate": 0.0001026677104660966,
+      "loss": 0.6987,
+      "mean_token_accuracy": 0.7867967963218689,
+      "num_tokens": 165493599.0,
+      "step": 48490
+    },
+    {
+      "epoch": 0.4867718495322976,
+      "learning_rate": 0.00010264763740013649,
+      "loss": 0.6617,
+      "mean_token_accuracy": 0.7973835945129395,
+      "num_tokens": 165529231.0,
+      "step": 48500
+    },
+    {
+      "epoch": 0.48687221486209803,
+      "learning_rate": 0.0001026275643341764,
+      "loss": 0.6587,
+      "mean_token_accuracy": 0.7987306237220764,
+      "num_tokens": 165566117.0,
+      "step": 48510
+    },
+    {
+      "epoch": 0.4869725801918985,
+      "learning_rate": 0.00010260749126821632,
+      "loss": 0.6776,
+      "mean_token_accuracy": 0.7935501039028168,
+      "num_tokens": 165602360.0,
+      "step": 48520
+    },
+    {
+      "epoch": 0.487072945521699,
+      "learning_rate": 0.00010258741820225622,
+      "loss": 0.6813,
+      "mean_token_accuracy": 0.7961971998214722,
+      "num_tokens": 165639131.0,
+      "step": 48530
+    },
+    {
+      "epoch": 0.48717331085149945,
+      "learning_rate": 0.00010256734513629614,
+      "loss": 0.6497,
+      "mean_token_accuracy": 0.8006950974464416,
+      "num_tokens": 165676654.0,
+      "step": 48540
+    },
+    {
+      "epoch": 0.4872736761812999,
+      "learning_rate": 0.00010254727207033603,
+      "loss": 0.6629,
+      "mean_token_accuracy": 0.7984619259834289,
+      "num_tokens": 165713192.0,
+      "step": 48550
+    },
+    {
+      "epoch": 0.4873740415111004,
+      "learning_rate": 0.00010252719900437593,
+      "loss": 0.647,
+      "mean_token_accuracy": 0.8039807558059693,
+      "num_tokens": 165749737.0,
+      "step": 48560
+    },
+    {
+      "epoch": 0.4874744068409009,
+      "learning_rate": 0.00010250712593841585,
+      "loss": 0.6746,
+      "mean_token_accuracy": 0.7973509192466736,
+      "num_tokens": 165785171.0,
+      "step": 48570
+    },
+    {
+      "epoch": 0.48757477217070133,
+      "learning_rate": 0.00010248705287245575,
+      "loss": 0.6998,
+      "mean_token_accuracy": 0.7900561451911926,
+      "num_tokens": 165821362.0,
+      "step": 48580
+    },
+    {
+      "epoch": 0.48767513750050184,
+      "learning_rate": 0.00010246697980649564,
+      "loss": 0.6383,
+      "mean_token_accuracy": 0.8029297113418579,
+      "num_tokens": 165858520.0,
+      "step": 48590
+    },
+    {
+      "epoch": 0.4877755028303023,
+      "learning_rate": 0.00010244690674053556,
+      "loss": 0.6647,
+      "mean_token_accuracy": 0.7972795307636261,
+      "num_tokens": 165895360.0,
+      "step": 48600
+    },
+    {
+      "epoch": 0.48787586816010275,
+      "learning_rate": 0.00010242683367457546,
+      "loss": 0.6649,
+      "mean_token_accuracy": 0.7969582557678223,
+      "num_tokens": 165932580.0,
+      "step": 48610
+    },
+    {
+      "epoch": 0.48797623348990327,
+      "learning_rate": 0.00010240676060861537,
+      "loss": 0.6778,
+      "mean_token_accuracy": 0.7921225070953369,
+      "num_tokens": 165968971.0,
+      "step": 48620
+    },
+    {
+      "epoch": 0.4880765988197037,
+      "learning_rate": 0.00010238668754265527,
+      "loss": 0.6896,
+      "mean_token_accuracy": 0.7959330201148986,
+      "num_tokens": 166004902.0,
+      "step": 48630
+    },
+    {
+      "epoch": 0.4881769641495042,
+      "learning_rate": 0.00010236661447669518,
+      "loss": 0.692,
+      "mean_token_accuracy": 0.7915273070335388,
+      "num_tokens": 166040884.0,
+      "step": 48640
+    },
+    {
+      "epoch": 0.4882773294793047,
+      "learning_rate": 0.00010234654141073508,
+      "loss": 0.6529,
+      "mean_token_accuracy": 0.8000566363334656,
+      "num_tokens": 166077731.0,
+      "step": 48650
+    },
+    {
+      "epoch": 0.48837769480910515,
+      "learning_rate": 0.000102326468344775,
+      "loss": 0.6744,
+      "mean_token_accuracy": 0.7959813833236694,
+      "num_tokens": 166114314.0,
+      "step": 48660
+    },
+    {
+      "epoch": 0.4884780601389056,
+      "learning_rate": 0.0001023063952788149,
+      "loss": 0.6817,
+      "mean_token_accuracy": 0.7944016873836517,
+      "num_tokens": 166149915.0,
+      "step": 48670
+    },
+    {
+      "epoch": 0.4885784254687061,
+      "learning_rate": 0.00010228632221285479,
+      "loss": 0.6778,
+      "mean_token_accuracy": 0.7906705737113953,
+      "num_tokens": 166184851.0,
+      "step": 48680
+    },
+    {
+      "epoch": 0.48867879079850657,
+      "learning_rate": 0.00010226624914689471,
+      "loss": 0.7067,
+      "mean_token_accuracy": 0.7922101736068725,
+      "num_tokens": 166220345.0,
+      "step": 48690
+    },
+    {
+      "epoch": 0.488779156128307,
+      "learning_rate": 0.0001022461760809346,
+      "loss": 0.662,
+      "mean_token_accuracy": 0.8018300950527191,
+      "num_tokens": 166256077.0,
+      "step": 48700
+    },
+    {
+      "epoch": 0.48887952145810754,
+      "learning_rate": 0.0001022261030149745,
+      "loss": 0.6916,
+      "mean_token_accuracy": 0.7912471473217011,
+      "num_tokens": 166292475.0,
+      "step": 48710
+    },
+    {
+      "epoch": 0.488979886787908,
+      "learning_rate": 0.00010220602994901442,
+      "loss": 0.6423,
+      "mean_token_accuracy": 0.7997991442680359,
+      "num_tokens": 166329031.0,
+      "step": 48720
+    },
+    {
+      "epoch": 0.48908025211770845,
+      "learning_rate": 0.00010218595688305432,
+      "loss": 0.6528,
+      "mean_token_accuracy": 0.7999334573745728,
+      "num_tokens": 166363954.0,
+      "step": 48730
+    },
+    {
+      "epoch": 0.48918061744750896,
+      "learning_rate": 0.00010216588381709423,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.7921888053417205,
+      "num_tokens": 166400684.0,
+      "step": 48740
+    },
+    {
+      "epoch": 0.4892809827773094,
+      "learning_rate": 0.00010214581075113413,
+      "loss": 0.666,
+      "mean_token_accuracy": 0.7992009162902832,
+      "num_tokens": 166436290.0,
+      "step": 48750
+    },
+    {
+      "epoch": 0.48938134810710987,
+      "learning_rate": 0.00010212573768517404,
+      "loss": 0.6451,
+      "mean_token_accuracy": 0.8084811270236969,
+      "num_tokens": 166472149.0,
+      "step": 48760
+    },
+    {
+      "epoch": 0.4894817134369103,
+      "learning_rate": 0.00010210566461921394,
+      "loss": 0.6563,
+      "mean_token_accuracy": 0.8018660724163056,
+      "num_tokens": 166508884.0,
+      "step": 48770
+    },
+    {
+      "epoch": 0.48958207876671084,
+      "learning_rate": 0.00010208559155325386,
+      "loss": 0.6501,
+      "mean_token_accuracy": 0.8009030163288117,
+      "num_tokens": 166545244.0,
+      "step": 48780
+    },
+    {
+      "epoch": 0.4896824440965113,
+      "learning_rate": 0.00010206551848729375,
+      "loss": 0.678,
+      "mean_token_accuracy": 0.7926466882228851,
+      "num_tokens": 166581823.0,
+      "step": 48790
+    },
+    {
+      "epoch": 0.48978280942631175,
+      "learning_rate": 0.00010204544542133365,
+      "loss": 0.6618,
+      "mean_token_accuracy": 0.8020177721977234,
+      "num_tokens": 166618826.0,
+      "step": 48800
+    },
+    {
+      "epoch": 0.48988317475611226,
+      "learning_rate": 0.00010202537235537357,
+      "loss": 0.6815,
+      "mean_token_accuracy": 0.7936260044574738,
+      "num_tokens": 166654734.0,
+      "step": 48810
+    },
+    {
+      "epoch": 0.4899835400859127,
+      "learning_rate": 0.00010200529928941347,
+      "loss": 0.6418,
+      "mean_token_accuracy": 0.8066985130310058,
+      "num_tokens": 166691645.0,
+      "step": 48820
+    },
+    {
+      "epoch": 0.4900839054157132,
+      "learning_rate": 0.00010198522622345336,
+      "loss": 0.6597,
+      "mean_token_accuracy": 0.7996945679187775,
+      "num_tokens": 166727650.0,
+      "step": 48830
+    },
+    {
+      "epoch": 0.4901842707455137,
+      "learning_rate": 0.00010196515315749328,
+      "loss": 0.6809,
+      "mean_token_accuracy": 0.792582529783249,
+      "num_tokens": 166763886.0,
+      "step": 48840
+    },
+    {
+      "epoch": 0.49028463607531414,
+      "learning_rate": 0.00010194508009153318,
+      "loss": 0.7033,
+      "mean_token_accuracy": 0.786809754371643,
+      "num_tokens": 166799672.0,
+      "step": 48850
+    },
+    {
+      "epoch": 0.4903850014051146,
+      "learning_rate": 0.00010192500702557309,
+      "loss": 0.6783,
+      "mean_token_accuracy": 0.7955593287944793,
+      "num_tokens": 166834849.0,
+      "step": 48860
+    },
+    {
+      "epoch": 0.4904853667349151,
+      "learning_rate": 0.00010190493395961301,
+      "loss": 0.647,
+      "mean_token_accuracy": 0.8022721767425537,
+      "num_tokens": 166870259.0,
+      "step": 48870
+    },
+    {
+      "epoch": 0.49058573206471556,
+      "learning_rate": 0.0001018848608936529,
+      "loss": 0.6751,
+      "mean_token_accuracy": 0.8003849744796753,
+      "num_tokens": 166906060.0,
+      "step": 48880
+    },
+    {
+      "epoch": 0.490686097394516,
+      "learning_rate": 0.0001018647878276928,
+      "loss": 0.6758,
+      "mean_token_accuracy": 0.7969516217708588,
+      "num_tokens": 166942371.0,
+      "step": 48890
+    },
+    {
+      "epoch": 0.49078646272431653,
+      "learning_rate": 0.00010184471476173272,
+      "loss": 0.6725,
+      "mean_token_accuracy": 0.7958665549755096,
+      "num_tokens": 166978893.0,
+      "step": 48900
+    },
+    {
+      "epoch": 0.490886828054117,
+      "learning_rate": 0.00010182464169577262,
+      "loss": 0.6459,
+      "mean_token_accuracy": 0.8034192264080048,
+      "num_tokens": 167015286.0,
+      "step": 48910
+    },
+    {
+      "epoch": 0.49098719338391744,
+      "learning_rate": 0.00010180456862981251,
+      "loss": 0.648,
+      "mean_token_accuracy": 0.802105039358139,
+      "num_tokens": 167051732.0,
+      "step": 48920
+    },
+    {
+      "epoch": 0.49108755871371795,
+      "learning_rate": 0.00010178449556385243,
+      "loss": 0.682,
+      "mean_token_accuracy": 0.7966529786586761,
+      "num_tokens": 167086827.0,
+      "step": 48930
+    },
+    {
+      "epoch": 0.4911879240435184,
+      "learning_rate": 0.00010176442249789233,
+      "loss": 0.6815,
+      "mean_token_accuracy": 0.7956272065639496,
+      "num_tokens": 167122570.0,
+      "step": 48940
+    },
+    {
+      "epoch": 0.49128828937331886,
+      "learning_rate": 0.00010174434943193224,
+      "loss": 0.6448,
+      "mean_token_accuracy": 0.8040194809436798,
+      "num_tokens": 167158267.0,
+      "step": 48950
+    },
+    {
+      "epoch": 0.4913886547031194,
+      "learning_rate": 0.00010172427636597214,
+      "loss": 0.6736,
+      "mean_token_accuracy": 0.791230309009552,
+      "num_tokens": 167194290.0,
+      "step": 48960
+    },
+    {
+      "epoch": 0.49148902003291983,
+      "learning_rate": 0.00010170420330001205,
+      "loss": 0.6891,
+      "mean_token_accuracy": 0.7936301529407501,
+      "num_tokens": 167229898.0,
+      "step": 48970
+    },
+    {
+      "epoch": 0.4915893853627203,
+      "learning_rate": 0.00010168413023405195,
+      "loss": 0.6653,
+      "mean_token_accuracy": 0.8004696249961853,
+      "num_tokens": 167266071.0,
+      "step": 48980
+    },
+    {
+      "epoch": 0.4916897506925208,
+      "learning_rate": 0.00010166405716809187,
+      "loss": 0.6598,
+      "mean_token_accuracy": 0.7968671381473541,
+      "num_tokens": 167303234.0,
+      "step": 48990
+    },
+    {
+      "epoch": 0.49179011602232126,
+      "learning_rate": 0.00010164398410213176,
+      "loss": 0.6603,
+      "mean_token_accuracy": 0.8036011934280396,
+      "num_tokens": 167339447.0,
+      "step": 49000
+    },
+    {
+      "epoch": 0.4918904813521217,
+      "learning_rate": 0.00010162391103617166,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.7965591430664063,
+      "num_tokens": 167375969.0,
+      "step": 49010
+    },
+    {
+      "epoch": 0.4919908466819222,
+      "learning_rate": 0.00010160383797021158,
+      "loss": 0.6534,
+      "mean_token_accuracy": 0.795904415845871,
+      "num_tokens": 167412465.0,
+      "step": 49020
+    },
+    {
+      "epoch": 0.4920912120117227,
+      "learning_rate": 0.00010158376490425148,
+      "loss": 0.6913,
+      "mean_token_accuracy": 0.7899755477905274,
+      "num_tokens": 167448800.0,
+      "step": 49030
+    },
+    {
+      "epoch": 0.49219157734152313,
+      "learning_rate": 0.00010156369183829137,
+      "loss": 0.658,
+      "mean_token_accuracy": 0.7980289340019227,
+      "num_tokens": 167485083.0,
+      "step": 49040
+    },
+    {
+      "epoch": 0.49229194267132365,
+      "learning_rate": 0.00010154361877233129,
+      "loss": 0.6702,
+      "mean_token_accuracy": 0.7956407785415649,
+      "num_tokens": 167522992.0,
+      "step": 49050
+    },
+    {
+      "epoch": 0.4923923080011241,
+      "learning_rate": 0.00010152354570637119,
+      "loss": 0.6795,
+      "mean_token_accuracy": 0.7919821202754974,
+      "num_tokens": 167559245.0,
+      "step": 49060
+    },
+    {
+      "epoch": 0.49249267333092456,
+      "learning_rate": 0.0001015034726404111,
+      "loss": 0.7002,
+      "mean_token_accuracy": 0.788819408416748,
+      "num_tokens": 167596616.0,
+      "step": 49070
+    },
+    {
+      "epoch": 0.492593038660725,
+      "learning_rate": 0.000101483399574451,
+      "loss": 0.6537,
+      "mean_token_accuracy": 0.7997047662734985,
+      "num_tokens": 167631223.0,
+      "step": 49080
+    },
+    {
+      "epoch": 0.4926934039905255,
+      "learning_rate": 0.00010146332650849091,
+      "loss": 0.6921,
+      "mean_token_accuracy": 0.7928080320358276,
+      "num_tokens": 167667315.0,
+      "step": 49090
+    },
+    {
+      "epoch": 0.492793769320326,
+      "learning_rate": 0.00010144325344253082,
+      "loss": 0.633,
+      "mean_token_accuracy": 0.8074764609336853,
+      "num_tokens": 167702728.0,
+      "step": 49100
+    },
+    {
+      "epoch": 0.49289413465012644,
+      "learning_rate": 0.00010142318037657073,
+      "loss": 0.6803,
+      "mean_token_accuracy": 0.7898491978645324,
+      "num_tokens": 167738498.0,
+      "step": 49110
+    },
+    {
+      "epoch": 0.49299449997992695,
+      "learning_rate": 0.00010140310731061062,
+      "loss": 0.667,
+      "mean_token_accuracy": 0.799743378162384,
+      "num_tokens": 167774600.0,
+      "step": 49120
+    },
+    {
+      "epoch": 0.4930948653097274,
+      "learning_rate": 0.00010138303424465055,
+      "loss": 0.6774,
+      "mean_token_accuracy": 0.7991834044456482,
+      "num_tokens": 167810392.0,
+      "step": 49130
+    },
+    {
+      "epoch": 0.49319523063952786,
+      "learning_rate": 0.00010136296117869044,
+      "loss": 0.6419,
+      "mean_token_accuracy": 0.80216743350029,
+      "num_tokens": 167846423.0,
+      "step": 49140
+    },
+    {
+      "epoch": 0.49329559596932837,
+      "learning_rate": 0.00010134288811273034,
+      "loss": 0.6904,
+      "mean_token_accuracy": 0.7967431306838989,
+      "num_tokens": 167882764.0,
+      "step": 49150
+    },
+    {
+      "epoch": 0.4933959612991288,
+      "learning_rate": 0.00010132281504677026,
+      "loss": 0.6737,
+      "mean_token_accuracy": 0.7951765596866608,
+      "num_tokens": 167918193.0,
+      "step": 49160
+    },
+    {
+      "epoch": 0.4934963266289293,
+      "learning_rate": 0.00010130274198081015,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8014727294445038,
+      "num_tokens": 167953522.0,
+      "step": 49170
+    },
+    {
+      "epoch": 0.4935966919587298,
+      "learning_rate": 0.00010128266891485005,
+      "loss": 0.6846,
+      "mean_token_accuracy": 0.7900352478027344,
+      "num_tokens": 167989788.0,
+      "step": 49180
+    },
+    {
+      "epoch": 0.49369705728853025,
+      "learning_rate": 0.00010126259584888997,
+      "loss": 0.653,
+      "mean_token_accuracy": 0.8036088228225708,
+      "num_tokens": 168026492.0,
+      "step": 49190
+    },
+    {
+      "epoch": 0.4937974226183307,
+      "learning_rate": 0.00010124252278292987,
+      "loss": 0.6212,
+      "mean_token_accuracy": 0.8075027704238892,
+      "num_tokens": 168063125.0,
+      "step": 49200
+    },
+    {
+      "epoch": 0.4938977879481312,
+      "learning_rate": 0.00010122244971696977,
+      "loss": 0.6641,
+      "mean_token_accuracy": 0.8012927532196045,
+      "num_tokens": 168100114.0,
+      "step": 49210
+    },
+    {
+      "epoch": 0.4939981532779317,
+      "learning_rate": 0.0001012023766510097,
+      "loss": 0.6206,
+      "mean_token_accuracy": 0.8106712639331818,
+      "num_tokens": 168137049.0,
+      "step": 49220
+    },
+    {
+      "epoch": 0.49409851860773213,
+      "learning_rate": 0.00010118230358504959,
+      "loss": 0.6676,
+      "mean_token_accuracy": 0.8016164481639863,
+      "num_tokens": 168173489.0,
+      "step": 49230
+    },
+    {
+      "epoch": 0.49419888393753264,
+      "learning_rate": 0.00010116223051908949,
+      "loss": 0.6707,
+      "mean_token_accuracy": 0.7990468502044678,
+      "num_tokens": 168210405.0,
+      "step": 49240
+    },
+    {
+      "epoch": 0.4942992492673331,
+      "learning_rate": 0.00010114215745312941,
+      "loss": 0.6819,
+      "mean_token_accuracy": 0.7885425746440887,
+      "num_tokens": 168246121.0,
+      "step": 49250
+    },
+    {
+      "epoch": 0.49439961459713355,
+      "learning_rate": 0.0001011220843871693,
+      "loss": 0.6384,
+      "mean_token_accuracy": 0.8060955345630646,
+      "num_tokens": 168282885.0,
+      "step": 49260
+    },
+    {
+      "epoch": 0.49449997992693406,
+      "learning_rate": 0.0001011020113212092,
+      "loss": 0.6329,
+      "mean_token_accuracy": 0.8048595309257507,
+      "num_tokens": 168320252.0,
+      "step": 49270
+    },
+    {
+      "epoch": 0.4946003452567345,
+      "learning_rate": 0.00010108193825524912,
+      "loss": 0.6712,
+      "mean_token_accuracy": 0.7946819722652435,
+      "num_tokens": 168356206.0,
+      "step": 49280
+    },
+    {
+      "epoch": 0.494700710586535,
+      "learning_rate": 0.00010106186518928901,
+      "loss": 0.6733,
+      "mean_token_accuracy": 0.7937299966812134,
+      "num_tokens": 168391953.0,
+      "step": 49290
+    },
+    {
+      "epoch": 0.4948010759163355,
+      "learning_rate": 0.00010104179212332891,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8032675206661224,
+      "num_tokens": 168428933.0,
+      "step": 49300
+    },
+    {
+      "epoch": 0.49490144124613594,
+      "learning_rate": 0.00010102171905736883,
+      "loss": 0.6546,
+      "mean_token_accuracy": 0.8021373987197876,
+      "num_tokens": 168465598.0,
+      "step": 49310
+    },
+    {
+      "epoch": 0.4950018065759364,
+      "learning_rate": 0.00010100164599140874,
+      "loss": 0.6679,
+      "mean_token_accuracy": 0.7983918905258178,
+      "num_tokens": 168501674.0,
+      "step": 49320
+    },
+    {
+      "epoch": 0.4951021719057369,
+      "learning_rate": 0.00010098157292544863,
+      "loss": 0.6692,
+      "mean_token_accuracy": 0.7963124752044678,
+      "num_tokens": 168538275.0,
+      "step": 49330
+    },
+    {
+      "epoch": 0.49520253723553737,
+      "learning_rate": 0.00010096149985948856,
+      "loss": 0.6484,
+      "mean_token_accuracy": 0.8078245878219604,
+      "num_tokens": 168574324.0,
+      "step": 49340
+    },
+    {
+      "epoch": 0.4953029025653378,
+      "learning_rate": 0.00010094142679352845,
+      "loss": 0.6205,
+      "mean_token_accuracy": 0.8130067169666291,
+      "num_tokens": 168612080.0,
+      "step": 49350
+    },
+    {
+      "epoch": 0.4954032678951383,
+      "learning_rate": 0.00010092135372756835,
+      "loss": 0.6383,
+      "mean_token_accuracy": 0.8053999781608582,
+      "num_tokens": 168647904.0,
+      "step": 49360
+    },
+    {
+      "epoch": 0.4955036332249388,
+      "learning_rate": 0.00010090128066160827,
+      "loss": 0.6892,
+      "mean_token_accuracy": 0.792946869134903,
+      "num_tokens": 168684673.0,
+      "step": 49370
+    },
+    {
+      "epoch": 0.49560399855473924,
+      "learning_rate": 0.00010088120759564816,
+      "loss": 0.6418,
+      "mean_token_accuracy": 0.8067975401878357,
+      "num_tokens": 168721319.0,
+      "step": 49380
+    },
+    {
+      "epoch": 0.4957043638845397,
+      "learning_rate": 0.00010086113452968806,
+      "loss": 0.6532,
+      "mean_token_accuracy": 0.8007255733013153,
+      "num_tokens": 168756977.0,
+      "step": 49390
+    },
+    {
+      "epoch": 0.4958047292143402,
+      "learning_rate": 0.00010084106146372798,
+      "loss": 0.6465,
+      "mean_token_accuracy": 0.8025450706481934,
+      "num_tokens": 168793421.0,
+      "step": 49400
+    },
+    {
+      "epoch": 0.49590509454414067,
+      "learning_rate": 0.00010082098839776787,
+      "loss": 0.6299,
+      "mean_token_accuracy": 0.8060990929603576,
+      "num_tokens": 168830890.0,
+      "step": 49410
+    },
+    {
+      "epoch": 0.4960054598739411,
+      "learning_rate": 0.00010080091533180778,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8004979431629181,
+      "num_tokens": 168867582.0,
+      "step": 49420
+    },
+    {
+      "epoch": 0.49610582520374163,
+      "learning_rate": 0.00010078084226584769,
+      "loss": 0.6628,
+      "mean_token_accuracy": 0.8030794143676758,
+      "num_tokens": 168904001.0,
+      "step": 49430
+    },
+    {
+      "epoch": 0.4962061905335421,
+      "learning_rate": 0.0001007607691998876,
+      "loss": 0.6462,
+      "mean_token_accuracy": 0.8015866696834564,
+      "num_tokens": 168940290.0,
+      "step": 49440
+    },
+    {
+      "epoch": 0.49630655586334255,
+      "learning_rate": 0.0001007406961339275,
+      "loss": 0.6191,
+      "mean_token_accuracy": 0.8119408249855041,
+      "num_tokens": 168976309.0,
+      "step": 49450
+    },
+    {
+      "epoch": 0.49640692119314306,
+      "learning_rate": 0.00010072062306796742,
+      "loss": 0.6673,
+      "mean_token_accuracy": 0.8010443806648254,
+      "num_tokens": 169012558.0,
+      "step": 49460
+    },
+    {
+      "epoch": 0.4965072865229435,
+      "learning_rate": 0.00010070055000200731,
+      "loss": 0.6195,
+      "mean_token_accuracy": 0.8097262918949127,
+      "num_tokens": 169049535.0,
+      "step": 49470
+    },
+    {
+      "epoch": 0.49660765185274397,
+      "learning_rate": 0.0001006804769360472,
+      "loss": 0.6852,
+      "mean_token_accuracy": 0.7921403408050537,
+      "num_tokens": 169085900.0,
+      "step": 49480
+    },
+    {
+      "epoch": 0.4967080171825445,
+      "learning_rate": 0.00010066040387008713,
+      "loss": 0.6463,
+      "mean_token_accuracy": 0.803951519727707,
+      "num_tokens": 169122709.0,
+      "step": 49490
+    },
+    {
+      "epoch": 0.49680838251234494,
+      "learning_rate": 0.00010064033080412702,
+      "loss": 0.6329,
+      "mean_token_accuracy": 0.8065069198608399,
+      "num_tokens": 169159716.0,
+      "step": 49500
+    },
+    {
+      "epoch": 0.4969087478421454,
+      "learning_rate": 0.00010062025773816692,
+      "loss": 0.6487,
+      "mean_token_accuracy": 0.805515992641449,
+      "num_tokens": 169196873.0,
+      "step": 49510
+    },
+    {
+      "epoch": 0.4970091131719459,
+      "learning_rate": 0.00010060018467220684,
+      "loss": 0.6463,
+      "mean_token_accuracy": 0.8037728667259216,
+      "num_tokens": 169233220.0,
+      "step": 49520
+    },
+    {
+      "epoch": 0.49710947850174636,
+      "learning_rate": 0.00010058011160624674,
+      "loss": 0.6525,
+      "mean_token_accuracy": 0.8033924281597138,
+      "num_tokens": 169268598.0,
+      "step": 49530
+    },
+    {
+      "epoch": 0.4972098438315468,
+      "learning_rate": 0.00010056003854028664,
+      "loss": 0.6387,
+      "mean_token_accuracy": 0.8102546036243439,
+      "num_tokens": 169305958.0,
+      "step": 49540
+    },
+    {
+      "epoch": 0.4973102091613473,
+      "learning_rate": 0.00010053996547432655,
+      "loss": 0.6351,
+      "mean_token_accuracy": 0.8053695142269135,
+      "num_tokens": 169343030.0,
+      "step": 49550
+    },
+    {
+      "epoch": 0.4974105744911478,
+      "learning_rate": 0.00010051989240836646,
+      "loss": 0.6684,
+      "mean_token_accuracy": 0.7995011508464813,
+      "num_tokens": 169379356.0,
+      "step": 49560
+    },
+    {
+      "epoch": 0.49751093982094824,
+      "learning_rate": 0.00010049981934240636,
+      "loss": 0.6539,
+      "mean_token_accuracy": 0.8013176560401917,
+      "num_tokens": 169416109.0,
+      "step": 49570
+    },
+    {
+      "epoch": 0.49761130515074875,
+      "learning_rate": 0.00010047974627644628,
+      "loss": 0.6802,
+      "mean_token_accuracy": 0.7919177412986755,
+      "num_tokens": 169451622.0,
+      "step": 49580
+    },
+    {
+      "epoch": 0.4977116704805492,
+      "learning_rate": 0.00010045967321048617,
+      "loss": 0.6383,
+      "mean_token_accuracy": 0.8054239809513092,
+      "num_tokens": 169488330.0,
+      "step": 49590
+    },
+    {
+      "epoch": 0.49781203581034966,
+      "learning_rate": 0.00010043960014452607,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.802841168642044,
+      "num_tokens": 169525885.0,
+      "step": 49600
+    },
+    {
+      "epoch": 0.4979124011401502,
+      "learning_rate": 0.00010041952707856599,
+      "loss": 0.6379,
+      "mean_token_accuracy": 0.8047163426876068,
+      "num_tokens": 169562296.0,
+      "step": 49610
+    },
+    {
+      "epoch": 0.49801276646995063,
+      "learning_rate": 0.00010039945401260588,
+      "loss": 0.6268,
+      "mean_token_accuracy": 0.8083502173423767,
+      "num_tokens": 169599402.0,
+      "step": 49620
+    },
+    {
+      "epoch": 0.4981131317997511,
+      "learning_rate": 0.00010037938094664578,
+      "loss": 0.6333,
+      "mean_token_accuracy": 0.803332257270813,
+      "num_tokens": 169636381.0,
+      "step": 49630
+    },
+    {
+      "epoch": 0.49821349712955154,
+      "learning_rate": 0.0001003593078806857,
+      "loss": 0.6426,
+      "mean_token_accuracy": 0.8038518130779266,
+      "num_tokens": 169672825.0,
+      "step": 49640
+    },
+    {
+      "epoch": 0.49831386245935205,
+      "learning_rate": 0.0001003392348147256,
+      "loss": 0.6638,
+      "mean_token_accuracy": 0.8027487874031067,
+      "num_tokens": 169709047.0,
+      "step": 49650
+    },
+    {
+      "epoch": 0.4984142277891525,
+      "learning_rate": 0.00010031916174876552,
+      "loss": 0.6689,
+      "mean_token_accuracy": 0.8009410977363587,
+      "num_tokens": 169745629.0,
+      "step": 49660
+    },
+    {
+      "epoch": 0.49851459311895296,
+      "learning_rate": 0.00010029908868280543,
+      "loss": 0.6241,
+      "mean_token_accuracy": 0.8074738621711731,
+      "num_tokens": 169782734.0,
+      "step": 49670
+    },
+    {
+      "epoch": 0.4986149584487535,
+      "learning_rate": 0.00010027901561684532,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8067907512187957,
+      "num_tokens": 169818752.0,
+      "step": 49680
+    },
+    {
+      "epoch": 0.49871532377855393,
+      "learning_rate": 0.00010025894255088524,
+      "loss": 0.6743,
+      "mean_token_accuracy": 0.7934290409088135,
+      "num_tokens": 169855523.0,
+      "step": 49690
+    },
+    {
+      "epoch": 0.4988156891083544,
+      "learning_rate": 0.00010023886948492514,
+      "loss": 0.6636,
+      "mean_token_accuracy": 0.7973025441169739,
+      "num_tokens": 169891446.0,
+      "step": 49700
+    },
+    {
+      "epoch": 0.4989160544381549,
+      "learning_rate": 0.00010021879641896503,
+      "loss": 0.6419,
+      "mean_token_accuracy": 0.803942346572876,
+      "num_tokens": 169928179.0,
+      "step": 49710
+    },
+    {
+      "epoch": 0.49901641976795535,
+      "learning_rate": 0.00010019872335300495,
+      "loss": 0.6543,
+      "mean_token_accuracy": 0.8052600800991059,
+      "num_tokens": 169965747.0,
+      "step": 49720
+    },
+    {
+      "epoch": 0.4991167850977558,
+      "learning_rate": 0.00010017865028704485,
+      "loss": 0.6674,
+      "mean_token_accuracy": 0.7950818300247192,
+      "num_tokens": 170002866.0,
+      "step": 49730
+    },
+    {
+      "epoch": 0.4992171504275563,
+      "learning_rate": 0.00010015857722108474,
+      "loss": 0.6541,
+      "mean_token_accuracy": 0.8007044196128845,
+      "num_tokens": 170038593.0,
+      "step": 49740
+    },
+    {
+      "epoch": 0.4993175157573568,
+      "learning_rate": 0.00010013850415512467,
+      "loss": 0.6492,
+      "mean_token_accuracy": 0.7999036848545075,
+      "num_tokens": 170074815.0,
+      "step": 49750
+    },
+    {
+      "epoch": 0.49941788108715723,
+      "learning_rate": 0.00010011843108916456,
+      "loss": 0.6641,
+      "mean_token_accuracy": 0.8008026003837585,
+      "num_tokens": 170111348.0,
+      "step": 49760
+    },
+    {
+      "epoch": 0.49951824641695775,
+      "learning_rate": 0.00010009835802320447,
+      "loss": 0.6804,
+      "mean_token_accuracy": 0.7891123533248902,
+      "num_tokens": 170148266.0,
+      "step": 49770
+    },
+    {
+      "epoch": 0.4996186117467582,
+      "learning_rate": 0.00010007828495724438,
+      "loss": 0.6088,
+      "mean_token_accuracy": 0.8154561638832092,
+      "num_tokens": 170185136.0,
+      "step": 49780
+    },
+    {
+      "epoch": 0.49971897707655866,
+      "learning_rate": 0.00010005821189128429,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.8048878014087677,
+      "num_tokens": 170221700.0,
+      "step": 49790
+    },
+    {
+      "epoch": 0.49981934240635917,
+      "learning_rate": 0.00010003813882532418,
+      "loss": 0.6362,
+      "mean_token_accuracy": 0.8061830341815949,
+      "num_tokens": 170257157.0,
+      "step": 49800
+    },
+    {
+      "epoch": 0.4999197077361596,
+      "learning_rate": 0.0001000180657593641,
+      "loss": 0.6683,
+      "mean_token_accuracy": 0.8000795781612396,
+      "num_tokens": 170294589.0,
+      "step": 49810
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 99636,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 6227,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0383456402076877e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}