diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7634 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 760,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.236748993396759,
+      "epoch": 0.013289036544850499,
+      "grad_norm": 0.7060697078704834,
+      "learning_rate": 0.0002,
+      "loss": 2.64042592048645,
+      "mean_token_accuracy": 0.5126412361860275,
+      "num_tokens": 77636.0,
+      "step": 1
+    },
+    {
+      "entropy": 1.2850274741649628,
+      "epoch": 0.026578073089700997,
+      "grad_norm": 0.41636043787002563,
+      "learning_rate": 0.0002,
+      "loss": 2.205017566680908,
+      "mean_token_accuracy": 0.551362007856369,
+      "num_tokens": 154070.0,
+      "step": 2
+    },
+    {
+      "entropy": 1.3720425963401794,
+      "epoch": 0.03986710963455149,
+      "grad_norm": 0.38282454013824463,
+      "learning_rate": 0.0002,
+      "loss": 1.861063003540039,
+      "mean_token_accuracy": 0.5767628997564316,
+      "num_tokens": 231088.0,
+      "step": 3
+    },
+    {
+      "entropy": 1.4496844112873077,
+      "epoch": 0.053156146179401995,
+      "grad_norm": 0.3939046561717987,
+      "learning_rate": 0.0002,
+      "loss": 1.642488718032837,
+      "mean_token_accuracy": 0.589305192232132,
+      "num_tokens": 307396.0,
+      "step": 4
+    },
+    {
+      "entropy": 1.4270202219486237,
+      "epoch": 0.0664451827242525,
+      "grad_norm": 0.24181145429611206,
+      "learning_rate": 0.0002,
+      "loss": 1.4825536012649536,
+      "mean_token_accuracy": 0.6070653051137924,
+      "num_tokens": 384317.0,
+      "step": 5
+    },
+    {
+      "entropy": 1.396634191274643,
+      "epoch": 0.07973421926910298,
+      "grad_norm": 0.16257749497890472,
+      "learning_rate": 0.0002,
+      "loss": 1.3350005149841309,
+      "mean_token_accuracy": 0.6365590989589691,
+      "num_tokens": 462862.0,
+      "step": 6
+    },
+    {
+      "entropy": 1.3969248235225677,
+      "epoch": 0.09302325581395349,
+      "grad_norm": 0.13336774706840515,
+      "learning_rate": 0.0002,
+      "loss": 1.285440444946289,
+      "mean_token_accuracy": 0.6306734532117844,
+      "num_tokens": 539431.0,
+      "step": 7
+    },
+    {
+      "entropy": 1.342583417892456,
+      "epoch": 0.10631229235880399,
+      "grad_norm": 0.1381940096616745,
+      "learning_rate": 0.0002,
+      "loss": 1.1918363571166992,
+      "mean_token_accuracy": 0.6458747684955597,
+      "num_tokens": 615886.0,
+      "step": 8
+    },
+    {
+      "entropy": 1.2815902531147003,
+      "epoch": 0.11960132890365449,
+      "grad_norm": 0.13844357430934906,
+      "learning_rate": 0.0002,
+      "loss": 1.1147487163543701,
+      "mean_token_accuracy": 0.656469076871872,
+      "num_tokens": 693016.0,
+      "step": 9
+    },
+    {
+      "entropy": 1.2026029527187347,
+      "epoch": 0.132890365448505,
+      "grad_norm": 0.15192069113254547,
+      "learning_rate": 0.0002,
+      "loss": 1.034226417541504,
+      "mean_token_accuracy": 0.6756248027086258,
+      "num_tokens": 770638.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.1153404414653778,
+      "epoch": 0.1461794019933555,
+      "grad_norm": 0.1124810054898262,
+      "learning_rate": 0.0002,
+      "loss": 0.9820771217346191,
+      "mean_token_accuracy": 0.6814178228378296,
+      "num_tokens": 847371.0,
+      "step": 11
+    },
+    {
+      "entropy": 1.016791969537735,
+      "epoch": 0.15946843853820597,
+      "grad_norm": 0.09400399029254913,
+      "learning_rate": 0.0002,
+      "loss": 0.9312019348144531,
+      "mean_token_accuracy": 0.6897322982549667,
+      "num_tokens": 923630.0,
+      "step": 12
+    },
+    {
+      "entropy": 0.9511139243841171,
+      "epoch": 0.17275747508305647,
+      "grad_norm": 0.09321432560682297,
+      "learning_rate": 0.0002,
+      "loss": 0.9056980609893799,
+      "mean_token_accuracy": 0.6875409334897995,
+      "num_tokens": 1000432.0,
+      "step": 13
+    },
+    {
+      "entropy": 0.8801429569721222,
+      "epoch": 0.18604651162790697,
+      "grad_norm": 0.09821294248104095,
+      "learning_rate": 0.0002,
+      "loss": 0.871140718460083,
+      "mean_token_accuracy": 0.692296713590622,
+      "num_tokens": 1077348.0,
+      "step": 14
+    },
+    {
+      "entropy": 0.8406370431184769,
+      "epoch": 0.19933554817275748,
+      "grad_norm": 0.09626225382089615,
+      "learning_rate": 0.0002,
+      "loss": 0.85018390417099,
+      "mean_token_accuracy": 0.6941163539886475,
+      "num_tokens": 1153872.0,
+      "step": 15
+    },
+    {
+      "entropy": 0.7903019487857819,
+      "epoch": 0.21262458471760798,
+      "grad_norm": 0.08542217314243317,
+      "learning_rate": 0.0002,
+      "loss": 0.7919514179229736,
+      "mean_token_accuracy": 0.7115834504365921,
+      "num_tokens": 1231939.0,
+      "step": 16
+    },
+    {
+      "entropy": 0.8033469468355179,
+      "epoch": 0.22591362126245848,
+      "grad_norm": 0.0788746327161789,
+      "learning_rate": 0.0002,
+      "loss": 0.7911616563796997,
+      "mean_token_accuracy": 0.7051160037517548,
+      "num_tokens": 1308393.0,
+      "step": 17
+    },
+    {
+      "entropy": 0.7930255830287933,
+      "epoch": 0.23920265780730898,
+      "grad_norm": 0.09115136414766312,
+      "learning_rate": 0.0002,
+      "loss": 0.7672959566116333,
+      "mean_token_accuracy": 0.7108412981033325,
+      "num_tokens": 1385266.0,
+      "step": 18
+    },
+    {
+      "entropy": 0.7890239208936691,
+      "epoch": 0.25249169435215946,
+      "grad_norm": 0.08325205743312836,
+      "learning_rate": 0.0002,
+      "loss": 0.7619231343269348,
+      "mean_token_accuracy": 0.7112641334533691,
+      "num_tokens": 1462104.0,
+      "step": 19
+    },
+    {
+      "entropy": 0.7456277012825012,
+      "epoch": 0.26578073089701,
+      "grad_norm": 0.0718981996178627,
+      "learning_rate": 0.0002,
+      "loss": 0.7284970283508301,
+      "mean_token_accuracy": 0.7222618609666824,
+      "num_tokens": 1540189.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.7269423454999924,
+      "epoch": 0.27906976744186046,
+      "grad_norm": 0.06821414083242416,
+      "learning_rate": 0.0002,
+      "loss": 0.7141512632369995,
+      "mean_token_accuracy": 0.7253042161464691,
+      "num_tokens": 1617227.0,
+      "step": 21
+    },
+    {
+      "entropy": 0.7203116118907928,
+      "epoch": 0.292358803986711,
+      "grad_norm": 0.08054713159799576,
+      "learning_rate": 0.0002,
+      "loss": 0.7158239483833313,
+      "mean_token_accuracy": 0.7215390652418137,
+      "num_tokens": 1694721.0,
+      "step": 22
+    },
+    {
+      "entropy": 0.7209072113037109,
+      "epoch": 0.30564784053156147,
+      "grad_norm": 0.07433964312076569,
+      "learning_rate": 0.0002,
+      "loss": 0.7111976146697998,
+      "mean_token_accuracy": 0.7223243564367294,
+      "num_tokens": 1772011.0,
+      "step": 23
+    },
+    {
+      "entropy": 0.6966264992952347,
+      "epoch": 0.31893687707641194,
+      "grad_norm": 0.062079496681690216,
+      "learning_rate": 0.0002,
+      "loss": 0.6817796230316162,
+      "mean_token_accuracy": 0.7308934032917023,
+      "num_tokens": 1849187.0,
+      "step": 24
+    },
+    {
+      "entropy": 0.6859240978956223,
+      "epoch": 0.33222591362126247,
+      "grad_norm": 0.05567821115255356,
+      "learning_rate": 0.0002,
+      "loss": 0.6765352487564087,
+      "mean_token_accuracy": 0.7338166683912277,
+      "num_tokens": 1926706.0,
+      "step": 25
+    },
+    {
+      "entropy": 0.6890190839767456,
+      "epoch": 0.34551495016611294,
+      "grad_norm": 0.061284177005290985,
+      "learning_rate": 0.0002,
+      "loss": 0.6843230128288269,
+      "mean_token_accuracy": 0.7304094433784485,
+      "num_tokens": 2003441.0,
+      "step": 26
+    },
+    {
+      "entropy": 0.6757776588201523,
+      "epoch": 0.3588039867109635,
+      "grad_norm": 0.0667528584599495,
+      "learning_rate": 0.0002,
+      "loss": 0.6786775588989258,
+      "mean_token_accuracy": 0.7338518500328064,
+      "num_tokens": 2079773.0,
+      "step": 27
+    },
+    {
+      "entropy": 0.6639043688774109,
+      "epoch": 0.37209302325581395,
+      "grad_norm": 0.06810814887285233,
+      "learning_rate": 0.0002,
+      "loss": 0.6645445823669434,
+      "mean_token_accuracy": 0.7401989549398422,
+      "num_tokens": 2156620.0,
+      "step": 28
+    },
+    {
+      "entropy": 0.6565608233213425,
+      "epoch": 0.3853820598006645,
+      "grad_norm": 0.08254855126142502,
+      "learning_rate": 0.0002,
+      "loss": 0.6565409898757935,
+      "mean_token_accuracy": 0.7424292862415314,
+      "num_tokens": 2234590.0,
+      "step": 29
+    },
+    {
+      "entropy": 0.6681215614080429,
+      "epoch": 0.39867109634551495,
+      "grad_norm": 0.055675048381090164,
+      "learning_rate": 0.0002,
+      "loss": 0.6622647047042847,
+      "mean_token_accuracy": 0.7392790168523788,
+      "num_tokens": 2310873.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6501797735691071,
+      "epoch": 0.4119601328903654,
+      "grad_norm": 0.06277249753475189,
+      "learning_rate": 0.0002,
+      "loss": 0.6405087113380432,
+      "mean_token_accuracy": 0.7475987821817398,
+      "num_tokens": 2387958.0,
+      "step": 31
+    },
+    {
+      "entropy": 0.6630964279174805,
+      "epoch": 0.42524916943521596,
+      "grad_norm": 0.04983459413051605,
+      "learning_rate": 0.0002,
+      "loss": 0.6560376882553101,
+      "mean_token_accuracy": 0.7404419034719467,
+      "num_tokens": 2464655.0,
+      "step": 32
+    },
+    {
+      "entropy": 0.6463229656219482,
+      "epoch": 0.43853820598006643,
+      "grad_norm": 0.057940058410167694,
+      "learning_rate": 0.0002,
+      "loss": 0.6427592039108276,
+      "mean_token_accuracy": 0.7448015511035919,
+      "num_tokens": 2542565.0,
+      "step": 33
+    },
+    {
+      "entropy": 0.6510194540023804,
+      "epoch": 0.45182724252491696,
+      "grad_norm": 0.051239825785160065,
+      "learning_rate": 0.0002,
+      "loss": 0.6500669717788696,
+      "mean_token_accuracy": 0.7419901490211487,
+      "num_tokens": 2619587.0,
+      "step": 34
+    },
+    {
+      "entropy": 0.6514755338430405,
+      "epoch": 0.46511627906976744,
+      "grad_norm": 0.047699857503175735,
+      "learning_rate": 0.0002,
+      "loss": 0.6523705720901489,
+      "mean_token_accuracy": 0.7411993145942688,
+      "num_tokens": 2697337.0,
+      "step": 35
+    },
+    {
+      "entropy": 0.6403260231018066,
+      "epoch": 0.47840531561461797,
+      "grad_norm": 0.05285811051726341,
+      "learning_rate": 0.0002,
+      "loss": 0.64084792137146,
+      "mean_token_accuracy": 0.7455829083919525,
+      "num_tokens": 2774952.0,
+      "step": 36
+    },
+    {
+      "entropy": 0.6452134400606155,
+      "epoch": 0.49169435215946844,
+      "grad_norm": 0.04835190623998642,
+      "learning_rate": 0.0002,
+      "loss": 0.6418968439102173,
+      "mean_token_accuracy": 0.7455920875072479,
+      "num_tokens": 2852362.0,
+      "step": 37
+    },
+    {
+      "entropy": 0.6381319016218185,
+      "epoch": 0.5049833887043189,
+      "grad_norm": 0.0392417386174202,
+      "learning_rate": 0.0002,
+      "loss": 0.6349405646324158,
+      "mean_token_accuracy": 0.7478068023920059,
+      "num_tokens": 2928832.0,
+      "step": 38
+    },
+    {
+      "entropy": 0.6351741403341293,
+      "epoch": 0.5182724252491694,
+      "grad_norm": 0.04395252466201782,
+      "learning_rate": 0.0002,
+      "loss": 0.6348677277565002,
+      "mean_token_accuracy": 0.7502338886260986,
+      "num_tokens": 3006905.0,
+      "step": 39
+    },
+    {
+      "entropy": 0.6367054581642151,
+      "epoch": 0.53156146179402,
+      "grad_norm": 0.03797860071063042,
+      "learning_rate": 0.0002,
+      "loss": 0.637786328792572,
+      "mean_token_accuracy": 0.7473134547472,
+      "num_tokens": 3083226.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6302400976419449,
+      "epoch": 0.5448504983388704,
+      "grad_norm": 0.04471680149435997,
+      "learning_rate": 0.0002,
+      "loss": 0.6324583888053894,
+      "mean_token_accuracy": 0.7488639950752258,
+      "num_tokens": 3159758.0,
+      "step": 41
+    },
+    {
+      "entropy": 0.6376523673534393,
+      "epoch": 0.5581395348837209,
+      "grad_norm": 0.03543337062001228,
+      "learning_rate": 0.0002,
+      "loss": 0.635400652885437,
+      "mean_token_accuracy": 0.7479510009288788,
+      "num_tokens": 3237257.0,
+      "step": 42
+    },
+    {
+      "entropy": 0.6355589032173157,
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.03427452594041824,
+      "learning_rate": 0.0002,
+      "loss": 0.6347056031227112,
+      "mean_token_accuracy": 0.7467037290334702,
+      "num_tokens": 3314343.0,
+      "step": 43
+    },
+    {
+      "entropy": 0.6208891570568085,
+      "epoch": 0.584717607973422,
+      "grad_norm": 0.033996518701314926,
+      "learning_rate": 0.0002,
+      "loss": 0.618794322013855,
+      "mean_token_accuracy": 0.7535620033740997,
+      "num_tokens": 3393198.0,
+      "step": 44
+    },
+    {
+      "entropy": 0.6390635222196579,
+      "epoch": 0.5980066445182725,
+      "grad_norm": 0.03224889189004898,
+      "learning_rate": 0.0002,
+      "loss": 0.6415340304374695,
+      "mean_token_accuracy": 0.7460517883300781,
+      "num_tokens": 3469250.0,
+      "step": 45
+    },
+    {
+      "entropy": 0.6261286735534668,
+      "epoch": 0.6112956810631229,
+      "grad_norm": 0.04010876268148422,
+      "learning_rate": 0.0002,
+      "loss": 0.621030330657959,
+      "mean_token_accuracy": 0.7536878883838654,
+      "num_tokens": 3546143.0,
+      "step": 46
+    },
+    {
+      "entropy": 0.6229505985975266,
+      "epoch": 0.6245847176079734,
+      "grad_norm": 0.024563085287809372,
+      "learning_rate": 0.0002,
+      "loss": 0.6213774681091309,
+      "mean_token_accuracy": 0.7527979761362076,
+      "num_tokens": 3622674.0,
+      "step": 47
+    },
+    {
+      "entropy": 0.6298186033964157,
+      "epoch": 0.6378737541528239,
+      "grad_norm": 0.038946483284235,
+      "learning_rate": 0.0002,
+      "loss": 0.6281483173370361,
+      "mean_token_accuracy": 0.7499598264694214,
+      "num_tokens": 3699454.0,
+      "step": 48
+    },
+    {
+      "entropy": 0.629898801445961,
+      "epoch": 0.6511627906976745,
+      "grad_norm": 0.029423469677567482,
+      "learning_rate": 0.0002,
+      "loss": 0.6269102096557617,
+      "mean_token_accuracy": 0.7491558194160461,
+      "num_tokens": 3775954.0,
+      "step": 49
+    },
+    {
+      "entropy": 0.6194917857646942,
+      "epoch": 0.6644518272425249,
+      "grad_norm": 0.03048616088926792,
+      "learning_rate": 0.0002,
+      "loss": 0.6176614761352539,
+      "mean_token_accuracy": 0.7542635649442673,
+      "num_tokens": 3853666.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.6288475841283798,
+      "epoch": 0.6777408637873754,
+      "grad_norm": 0.024089554324746132,
+      "learning_rate": 0.0002,
+      "loss": 0.6285361051559448,
+      "mean_token_accuracy": 0.7502647340297699,
+      "num_tokens": 3930580.0,
+      "step": 51
+    },
+    {
+      "entropy": 0.630060225725174,
+      "epoch": 0.6910299003322259,
+      "grad_norm": 0.02456744946539402,
+      "learning_rate": 0.0002,
+      "loss": 0.6316722631454468,
+      "mean_token_accuracy": 0.7478295862674713,
+      "num_tokens": 4008361.0,
+      "step": 52
+    },
+    {
+      "entropy": 0.616634801030159,
+      "epoch": 0.7043189368770764,
+      "grad_norm": 0.027101170271635056,
+      "learning_rate": 0.0002,
+      "loss": 0.6193246841430664,
+      "mean_token_accuracy": 0.7523616254329681,
+      "num_tokens": 4084978.0,
+      "step": 53
+    },
+    {
+      "entropy": 0.6331924647092819,
+      "epoch": 0.717607973421927,
+      "grad_norm": 0.024089930579066277,
+      "learning_rate": 0.0002,
+      "loss": 0.635392427444458,
+      "mean_token_accuracy": 0.7456516772508621,
+      "num_tokens": 4163011.0,
+      "step": 54
+    },
+    {
+      "entropy": 0.6317191421985626,
+      "epoch": 0.7308970099667774,
+      "grad_norm": 0.023995358496904373,
+      "learning_rate": 0.0002,
+      "loss": 0.6338692903518677,
+      "mean_token_accuracy": 0.7470207065343857,
+      "num_tokens": 4240203.0,
+      "step": 55
+    },
+    {
+      "entropy": 0.624017983675003,
+      "epoch": 0.7441860465116279,
+      "grad_norm": 0.02172154001891613,
+      "learning_rate": 0.0002,
+      "loss": 0.6213433742523193,
+      "mean_token_accuracy": 0.7516606599092484,
+      "num_tokens": 4317184.0,
+      "step": 56
+    },
+    {
+      "entropy": 0.6348051577806473,
+      "epoch": 0.7574750830564784,
+      "grad_norm": 0.020663734525442123,
+      "learning_rate": 0.0002,
+      "loss": 0.6317377686500549,
+      "mean_token_accuracy": 0.7475810945034027,
+      "num_tokens": 4394073.0,
+      "step": 57
+    },
+    {
+      "entropy": 0.6199813932180405,
+      "epoch": 0.770764119601329,
+      "grad_norm": 0.022987928241491318,
+      "learning_rate": 0.0002,
+      "loss": 0.6143967509269714,
+      "mean_token_accuracy": 0.7555971294641495,
+      "num_tokens": 4471745.0,
+      "step": 58
+    },
+    {
+      "entropy": 0.6262989342212677,
+      "epoch": 0.7840531561461794,
+      "grad_norm": 0.023222770541906357,
+      "learning_rate": 0.0002,
+      "loss": 0.6190211772918701,
+      "mean_token_accuracy": 0.752941370010376,
+      "num_tokens": 4548128.0,
+      "step": 59
+    },
+    {
+      "entropy": 0.6274468749761581,
+      "epoch": 0.7973421926910299,
+      "grad_norm": 0.02218114212155342,
+      "learning_rate": 0.0002,
+      "loss": 0.6246206760406494,
+      "mean_token_accuracy": 0.7506777495145798,
+      "num_tokens": 4625224.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.6210617572069168,
+      "epoch": 0.8106312292358804,
+      "grad_norm": 0.024545960128307343,
+      "learning_rate": 0.0002,
+      "loss": 0.6223740577697754,
+      "mean_token_accuracy": 0.7492077946662903,
+      "num_tokens": 4703042.0,
+      "step": 61
+    },
+    {
+      "entropy": 0.6245881468057632,
+      "epoch": 0.8239202657807309,
+      "grad_norm": 0.029528915882110596,
+      "learning_rate": 0.0002,
+      "loss": 0.6285632848739624,
+      "mean_token_accuracy": 0.7494110465049744,
+      "num_tokens": 4779244.0,
+      "step": 62
+    },
+    {
+      "entropy": 0.6198074072599411,
+      "epoch": 0.8372093023255814,
+      "grad_norm": 0.023333575576543808,
+      "learning_rate": 0.0002,
+      "loss": 0.6192362308502197,
+      "mean_token_accuracy": 0.7524196207523346,
+      "num_tokens": 4857457.0,
+      "step": 63
+    },
+    {
+      "entropy": 0.633359968662262,
+      "epoch": 0.8504983388704319,
+      "grad_norm": 0.0235088299959898,
+      "learning_rate": 0.0002,
+      "loss": 0.6320174932479858,
+      "mean_token_accuracy": 0.7467200756072998,
+      "num_tokens": 4933407.0,
+      "step": 64
+    },
+    {
+      "entropy": 0.6193127632141113,
+      "epoch": 0.8637873754152824,
+      "grad_norm": 0.02574855461716652,
+      "learning_rate": 0.0002,
+      "loss": 0.6140483021736145,
+      "mean_token_accuracy": 0.7537259012460709,
+      "num_tokens": 5009918.0,
+      "step": 65
+    },
+    {
+      "entropy": 0.6258151680231094,
+      "epoch": 0.8770764119601329,
+      "grad_norm": 0.022545455023646355,
+      "learning_rate": 0.0002,
+      "loss": 0.6209959387779236,
+      "mean_token_accuracy": 0.7510837912559509,
+      "num_tokens": 5087830.0,
+      "step": 66
+    },
+    {
+      "entropy": 0.6285541504621506,
+      "epoch": 0.8903654485049833,
+      "grad_norm": 0.024571102112531662,
+      "learning_rate": 0.0002,
+      "loss": 0.6246232986450195,
+      "mean_token_accuracy": 0.7501810193061829,
+      "num_tokens": 5164939.0,
+      "step": 67
+    },
+    {
+      "entropy": 0.6188823282718658,
+      "epoch": 0.9036544850498339,
+      "grad_norm": 0.025842731818556786,
+      "learning_rate": 0.0002,
+      "loss": 0.6188026666641235,
+      "mean_token_accuracy": 0.7524883896112442,
+      "num_tokens": 5242253.0,
+      "step": 68
+    },
+    {
+      "entropy": 0.6232005655765533,
+      "epoch": 0.9169435215946844,
+      "grad_norm": 0.02302289754152298,
+      "learning_rate": 0.0002,
+      "loss": 0.6259397268295288,
+      "mean_token_accuracy": 0.7492635697126389,
+      "num_tokens": 5318706.0,
+      "step": 69
+    },
+    {
+      "entropy": 0.6116026937961578,
+      "epoch": 0.9302325581395349,
+      "grad_norm": 0.02471403032541275,
+      "learning_rate": 0.0002,
+      "loss": 0.6138839721679688,
+      "mean_token_accuracy": 0.7528299242258072,
+      "num_tokens": 5395336.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.618821457028389,
+      "epoch": 0.9435215946843853,
+      "grad_norm": 0.023918448016047478,
+      "learning_rate": 0.0002,
+      "loss": 0.6226047277450562,
+      "mean_token_accuracy": 0.7493781894445419,
+      "num_tokens": 5471175.0,
+      "step": 71
+    },
+    {
+      "entropy": 0.6161183714866638,
+      "epoch": 0.9568106312292359,
+      "grad_norm": 0.024829547852277756,
+      "learning_rate": 0.0002,
+      "loss": 0.6122779846191406,
+      "mean_token_accuracy": 0.754673570394516,
+      "num_tokens": 5548409.0,
+      "step": 72
+    },
+    {
+      "entropy": 0.621030643582344,
+      "epoch": 0.9700996677740864,
+      "grad_norm": 0.024347985163331032,
+      "learning_rate": 0.0002,
+      "loss": 0.6146630048751831,
+      "mean_token_accuracy": 0.7522032558917999,
+      "num_tokens": 5625017.0,
+      "step": 73
+    },
+    {
+      "entropy": 0.6309276968240738,
+      "epoch": 0.9833887043189369,
+      "grad_norm": 0.022102857008576393,
+      "learning_rate": 0.0002,
+      "loss": 0.6264951229095459,
+      "mean_token_accuracy": 0.7490104138851166,
+      "num_tokens": 5702215.0,
+      "step": 74
+    },
+    {
+      "entropy": 0.6195333749055862,
+      "epoch": 0.9966777408637874,
+      "grad_norm": 0.021700864657759666,
+      "learning_rate": 0.0002,
+      "loss": 0.6143898963928223,
+      "mean_token_accuracy": 0.7532503306865692,
+      "num_tokens": 5780519.0,
+      "step": 75
+    },
+    {
+      "entropy": 0.6297195553779602,
+      "epoch": 1.0,
+      "grad_norm": 0.03480999171733856,
+      "learning_rate": 0.0002,
+      "loss": 0.6311229467391968,
+      "mean_token_accuracy": 0.745398759841919,
+      "num_tokens": 5799555.0,
+      "step": 76
+    },
+    {
+      "entropy": 0.6155448108911514,
+      "epoch": 1.0132890365448506,
+      "grad_norm": 0.023401184007525444,
+      "learning_rate": 0.0002,
+      "loss": 0.6173519492149353,
+      "mean_token_accuracy": 0.7522856891155243,
+      "num_tokens": 5876962.0,
+      "step": 77
+    },
+    {
+      "entropy": 0.6246323734521866,
+      "epoch": 1.026578073089701,
+      "grad_norm": 0.022852230817079544,
+      "learning_rate": 0.0002,
+      "loss": 0.6256575584411621,
+      "mean_token_accuracy": 0.7486253380775452,
+      "num_tokens": 5953753.0,
+      "step": 78
+    },
+    {
+      "entropy": 0.6294799596071243,
+      "epoch": 1.0398671096345515,
+      "grad_norm": 0.022332875058054924,
+      "learning_rate": 0.0002,
+      "loss": 0.627347469329834,
+      "mean_token_accuracy": 0.7484346181154251,
+      "num_tokens": 6031156.0,
+      "step": 79
+    },
+    {
+      "entropy": 0.6130478978157043,
+      "epoch": 1.053156146179402,
+      "grad_norm": 0.022138696163892746,
+      "learning_rate": 0.0002,
+      "loss": 0.6106399893760681,
+      "mean_token_accuracy": 0.755229189991951,
+      "num_tokens": 6109095.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.6223983317613602,
+      "epoch": 1.0664451827242525,
+      "grad_norm": 0.022859349846839905,
+      "learning_rate": 0.0002,
+      "loss": 0.6226997375488281,
+      "mean_token_accuracy": 0.7491207867860794,
+      "num_tokens": 6186352.0,
+      "step": 81
+    },
+    {
+      "entropy": 0.6232358813285828,
+      "epoch": 1.079734219269103,
+      "grad_norm": 0.01912062056362629,
+      "learning_rate": 0.0002,
+      "loss": 0.6226857900619507,
+      "mean_token_accuracy": 0.749756470322609,
+      "num_tokens": 6262924.0,
+      "step": 82
+    },
+    {
+      "entropy": 0.6186039000749588,
+      "epoch": 1.0930232558139534,
+      "grad_norm": 0.023234324529767036,
+      "learning_rate": 0.0002,
+      "loss": 0.6204067468643188,
+      "mean_token_accuracy": 0.7503676414489746,
+      "num_tokens": 6339829.0,
+      "step": 83
+    },
+    {
+      "entropy": 0.6187210232019424,
+      "epoch": 1.106312292358804,
+      "grad_norm": 0.01966855302453041,
+      "learning_rate": 0.0002,
+      "loss": 0.6205964684486389,
+      "mean_token_accuracy": 0.7499649524688721,
+      "num_tokens": 6417932.0,
+      "step": 84
+    },
+    {
+      "entropy": 0.6108027994632721,
+      "epoch": 1.1196013289036544,
+      "grad_norm": 0.02231299877166748,
+      "learning_rate": 0.0002,
+      "loss": 0.6092526912689209,
+      "mean_token_accuracy": 0.755171075463295,
+      "num_tokens": 6495236.0,
+      "step": 85
+    },
+    {
+      "entropy": 0.6252855509519577,
+      "epoch": 1.132890365448505,
+      "grad_norm": 0.02397831901907921,
+      "learning_rate": 0.0002,
+      "loss": 0.6228755712509155,
+      "mean_token_accuracy": 0.7505564838647842,
+      "num_tokens": 6571780.0,
+      "step": 86
+    },
+    {
+      "entropy": 0.6250486820936203,
+      "epoch": 1.1461794019933556,
+      "grad_norm": 0.021679693832993507,
+      "learning_rate": 0.0002,
+      "loss": 0.6200405955314636,
+      "mean_token_accuracy": 0.7508185654878616,
+      "num_tokens": 6648008.0,
+      "step": 87
+    },
+    {
+      "entropy": 0.6233238279819489,
+      "epoch": 1.159468438538206,
+      "grad_norm": 0.01941733807325363,
+      "learning_rate": 0.0002,
+      "loss": 0.6168190240859985,
+      "mean_token_accuracy": 0.7522764652967453,
+      "num_tokens": 6724624.0,
+      "step": 88
+    },
+    {
+      "entropy": 0.6189088374376297,
+      "epoch": 1.1727574750830565,
+      "grad_norm": 0.02111164852976799,
+      "learning_rate": 0.0002,
+      "loss": 0.6196113228797913,
+      "mean_token_accuracy": 0.7506090253591537,
+      "num_tokens": 6800925.0,
+      "step": 89
+    },
+    {
+      "entropy": 0.6136585921049118,
+      "epoch": 1.1860465116279069,
+      "grad_norm": 0.01873047649860382,
+      "learning_rate": 0.0002,
+      "loss": 0.6147369146347046,
+      "mean_token_accuracy": 0.753093421459198,
+      "num_tokens": 6877907.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.6133056879043579,
+      "epoch": 1.1993355481727574,
+      "grad_norm": 0.02167166955769062,
+      "learning_rate": 0.0002,
+      "loss": 0.6140564680099487,
+      "mean_token_accuracy": 0.7525176405906677,
+      "num_tokens": 6954865.0,
+      "step": 91
+    },
+    {
+      "entropy": 0.6283499300479889,
+      "epoch": 1.212624584717608,
+      "grad_norm": 0.021320724859833717,
+      "learning_rate": 0.0002,
+      "loss": 0.6233468651771545,
+      "mean_token_accuracy": 0.7499544322490692,
+      "num_tokens": 7031714.0,
+      "step": 92
+    },
+    {
+      "entropy": 0.6230917423963547,
+      "epoch": 1.2259136212624584,
+      "grad_norm": 0.020842526108026505,
+      "learning_rate": 0.0002,
+      "loss": 0.6210839748382568,
+      "mean_token_accuracy": 0.7498301565647125,
+      "num_tokens": 7107656.0,
+      "step": 93
+    },
+    {
+      "entropy": 0.6245012730360031,
+      "epoch": 1.239202657807309,
+      "grad_norm": 0.02319970726966858,
+      "learning_rate": 0.0002,
+      "loss": 0.6247711181640625,
+      "mean_token_accuracy": 0.7482653558254242,
+      "num_tokens": 7184588.0,
+      "step": 94
+    },
+    {
+      "entropy": 0.621661052107811,
+      "epoch": 1.2524916943521593,
+      "grad_norm": 0.020472005009651184,
+      "learning_rate": 0.0002,
+      "loss": 0.6205976605415344,
+      "mean_token_accuracy": 0.7511607855558395,
+      "num_tokens": 7262128.0,
+      "step": 95
+    },
+    {
+      "entropy": 0.6095613390207291,
+      "epoch": 1.26578073089701,
+      "grad_norm": 0.02212020941078663,
+      "learning_rate": 0.0002,
+      "loss": 0.6093997955322266,
+      "mean_token_accuracy": 0.7555146813392639,
+      "num_tokens": 7339113.0,
+      "step": 96
+    },
+    {
+      "entropy": 0.6145715415477753,
+      "epoch": 1.2790697674418605,
+      "grad_norm": 0.02165311761200428,
+      "learning_rate": 0.0002,
+      "loss": 0.6127168536186218,
+      "mean_token_accuracy": 0.7532989978790283,
+      "num_tokens": 7416501.0,
+      "step": 97
+    },
+    {
+      "entropy": 0.6266685277223587,
+      "epoch": 1.292358803986711,
+      "grad_norm": 0.02169322967529297,
+      "learning_rate": 0.0002,
+      "loss": 0.6243869662284851,
+      "mean_token_accuracy": 0.7484752535820007,
+      "num_tokens": 7493281.0,
+      "step": 98
+    },
+    {
+      "entropy": 0.6184320002794266,
+      "epoch": 1.3056478405315615,
+      "grad_norm": 0.0211771372705698,
+      "learning_rate": 0.0002,
+      "loss": 0.6157566905021667,
+      "mean_token_accuracy": 0.752418577671051,
+      "num_tokens": 7570464.0,
+      "step": 99
+    },
+    {
+      "entropy": 0.6127575635910034,
+      "epoch": 1.3189368770764118,
+      "grad_norm": 0.02210194058716297,
+      "learning_rate": 0.0002,
+      "loss": 0.6128020286560059,
+      "mean_token_accuracy": 0.7543350905179977,
+      "num_tokens": 7646649.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.612200528383255,
+      "epoch": 1.3322259136212624,
+      "grad_norm": 0.022352103143930435,
+      "learning_rate": 0.0002,
+      "loss": 0.6118579506874084,
+      "mean_token_accuracy": 0.7532858699560165,
+      "num_tokens": 7723836.0,
+      "step": 101
+    },
+    {
+      "entropy": 0.6052316725254059,
+      "epoch": 1.345514950166113,
+      "grad_norm": 0.02019244246184826,
+      "learning_rate": 0.0002,
+      "loss": 0.603766679763794,
+      "mean_token_accuracy": 0.7570021897554398,
+      "num_tokens": 7802209.0,
+      "step": 102
+    },
+    {
+      "entropy": 0.6166553497314453,
+      "epoch": 1.3588039867109636,
+      "grad_norm": 0.01797507330775261,
+      "learning_rate": 0.0002,
+      "loss": 0.616031289100647,
+      "mean_token_accuracy": 0.7518447786569595,
+      "num_tokens": 7880014.0,
+      "step": 103
+    },
+    {
+      "entropy": 0.6216948330402374,
+      "epoch": 1.372093023255814,
+      "grad_norm": 0.020605172961950302,
+      "learning_rate": 0.0002,
+      "loss": 0.6161950826644897,
+      "mean_token_accuracy": 0.7535870969295502,
+      "num_tokens": 7956944.0,
+      "step": 104
+    },
+    {
+      "entropy": 0.611840009689331,
+      "epoch": 1.3853820598006645,
+      "grad_norm": 0.021973272785544395,
+      "learning_rate": 0.0002,
+      "loss": 0.6133339405059814,
+      "mean_token_accuracy": 0.7527298778295517,
+      "num_tokens": 8033655.0,
+      "step": 105
+    },
+    {
+      "entropy": 0.6033324301242828,
+      "epoch": 1.398671096345515,
+      "grad_norm": 0.02020864374935627,
+      "learning_rate": 0.0002,
+      "loss": 0.6040066480636597,
+      "mean_token_accuracy": 0.7564088255167007,
+      "num_tokens": 8111825.0,
+      "step": 106
+    },
+    {
+      "entropy": 0.615772470831871,
+      "epoch": 1.4119601328903655,
+      "grad_norm": 0.020991913974285126,
+      "learning_rate": 0.0002,
+      "loss": 0.6151546239852905,
+      "mean_token_accuracy": 0.7534815222024918,
+      "num_tokens": 8189709.0,
+      "step": 107
+    },
+    {
+      "entropy": 0.6171981394290924,
+      "epoch": 1.425249169435216,
+      "grad_norm": 0.021498622372746468,
+      "learning_rate": 0.0002,
+      "loss": 0.6125850677490234,
+      "mean_token_accuracy": 0.7512624114751816,
+      "num_tokens": 8265892.0,
+      "step": 108
+    },
+    {
+      "entropy": 0.6139850616455078,
+      "epoch": 1.4385382059800664,
+      "grad_norm": 0.019243517890572548,
+      "learning_rate": 0.0002,
+      "loss": 0.6125526428222656,
+      "mean_token_accuracy": 0.754551038146019,
+      "num_tokens": 8343618.0,
+      "step": 109
+    },
+    {
+      "entropy": 0.6268000602722168,
+      "epoch": 1.451827242524917,
+      "grad_norm": 0.01758020557463169,
+      "learning_rate": 0.0002,
+      "loss": 0.6260548830032349,
+      "mean_token_accuracy": 0.7474812120199203,
+      "num_tokens": 8420843.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.6195523887872696,
+      "epoch": 1.4651162790697674,
+      "grad_norm": 0.019546709954738617,
+      "learning_rate": 0.0002,
+      "loss": 0.623106837272644,
+      "mean_token_accuracy": 0.7489201724529266,
+      "num_tokens": 8497047.0,
+      "step": 111
+    },
+    {
+      "entropy": 0.6040176898241043,
+      "epoch": 1.478405315614618,
+      "grad_norm": 0.02087543159723282,
+      "learning_rate": 0.0002,
+      "loss": 0.6014326810836792,
+      "mean_token_accuracy": 0.7569248229265213,
+      "num_tokens": 8575249.0,
+      "step": 112
+    },
+    {
+      "entropy": 0.6175996363162994,
+      "epoch": 1.4916943521594686,
+      "grad_norm": 0.019000006839632988,
+      "learning_rate": 0.0002,
+      "loss": 0.6130571961402893,
+      "mean_token_accuracy": 0.7541480958461761,
+      "num_tokens": 8653473.0,
+      "step": 113
+    },
+    {
+      "entropy": 0.6209008097648621,
+      "epoch": 1.504983388704319,
+      "grad_norm": 0.018268495798110962,
+      "learning_rate": 0.0002,
+      "loss": 0.6194300651550293,
+      "mean_token_accuracy": 0.7502760142087936,
+      "num_tokens": 8730987.0,
+      "step": 114
+    },
+    {
+      "entropy": 0.6245647221803665,
+      "epoch": 1.5182724252491693,
+      "grad_norm": 0.018849894404411316,
+      "learning_rate": 0.0002,
+      "loss": 0.6210106611251831,
+      "mean_token_accuracy": 0.7502076923847198,
+      "num_tokens": 8807717.0,
+      "step": 115
+    },
+    {
+      "entropy": 0.6183042079210281,
+      "epoch": 1.5315614617940199,
+      "grad_norm": 0.01850067265331745,
+      "learning_rate": 0.0002,
+      "loss": 0.6139713525772095,
+      "mean_token_accuracy": 0.7533506751060486,
+      "num_tokens": 8884648.0,
+      "step": 116
+    },
+    {
+      "entropy": 0.6245974451303482,
+      "epoch": 1.5448504983388704,
+      "grad_norm": 0.020130403339862823,
+      "learning_rate": 0.0002,
+      "loss": 0.6242485046386719,
+      "mean_token_accuracy": 0.748986542224884,
+      "num_tokens": 8961668.0,
+      "step": 117
+    },
+    {
+      "entropy": 0.6045528948307037,
+      "epoch": 1.558139534883721,
+      "grad_norm": 0.023585041984915733,
+      "learning_rate": 0.0002,
+      "loss": 0.6086483597755432,
+      "mean_token_accuracy": 0.754342183470726,
+      "num_tokens": 9038662.0,
+      "step": 118
+    },
+    {
+      "entropy": 0.6051427721977234,
+      "epoch": 1.5714285714285714,
+      "grad_norm": 0.020509056746959686,
+      "learning_rate": 0.0002,
+      "loss": 0.6038928031921387,
+      "mean_token_accuracy": 0.7558873742818832,
+      "num_tokens": 9115878.0,
+      "step": 119
+    },
+    {
+      "entropy": 0.6210778206586838,
+      "epoch": 1.584717607973422,
+      "grad_norm": 0.018983472138643265,
+      "learning_rate": 0.0002,
+      "loss": 0.6190018653869629,
+      "mean_token_accuracy": 0.7505859136581421,
+      "num_tokens": 9193137.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.6255708932876587,
+      "epoch": 1.5980066445182723,
+      "grad_norm": 0.02053634636104107,
+      "learning_rate": 0.0002,
+      "loss": 0.6212979555130005,
+      "mean_token_accuracy": 0.7490561902523041,
+      "num_tokens": 9269704.0,
+      "step": 121
+    },
+    {
+      "entropy": 0.6224274635314941,
+      "epoch": 1.611295681063123,
+      "grad_norm": 0.019471660256385803,
+      "learning_rate": 0.0002,
+      "loss": 0.6190239191055298,
+      "mean_token_accuracy": 0.750290259718895,
+      "num_tokens": 9347396.0,
+      "step": 122
+    },
+    {
+      "entropy": 0.6200169771909714,
+      "epoch": 1.6245847176079735,
+      "grad_norm": 0.01963883824646473,
+      "learning_rate": 0.0002,
+      "loss": 0.6168637871742249,
+      "mean_token_accuracy": 0.7503404319286346,
+      "num_tokens": 9424266.0,
+      "step": 123
+    },
+    {
+      "entropy": 0.6060867607593536,
+      "epoch": 1.6378737541528239,
+      "grad_norm": 0.020356209948658943,
+      "learning_rate": 0.0002,
+      "loss": 0.6069276332855225,
+      "mean_token_accuracy": 0.7539948225021362,
+      "num_tokens": 9501476.0,
+      "step": 124
+    },
+    {
+      "entropy": 0.6077988147735596,
+      "epoch": 1.6511627906976745,
+      "grad_norm": 0.020215313881635666,
+      "learning_rate": 0.0002,
+      "loss": 0.6037485599517822,
+      "mean_token_accuracy": 0.7572648674249649,
+      "num_tokens": 9579023.0,
+      "step": 125
+    },
+    {
+      "entropy": 0.6222731024026871,
+      "epoch": 1.6644518272425248,
+      "grad_norm": 0.021222807466983795,
+      "learning_rate": 0.0002,
+      "loss": 0.6259433031082153,
+      "mean_token_accuracy": 0.7491642236709595,
+      "num_tokens": 9656167.0,
+      "step": 126
+    },
+    {
+      "entropy": 0.6067023426294327,
+      "epoch": 1.6777408637873754,
+      "grad_norm": 0.01992393285036087,
+      "learning_rate": 0.0002,
+      "loss": 0.604471743106842,
+      "mean_token_accuracy": 0.7570877373218536,
+      "num_tokens": 9733224.0,
+      "step": 127
+    },
+    {
+      "entropy": 0.6153317838907242,
+      "epoch": 1.691029900332226,
+      "grad_norm": 0.018730230629444122,
+      "learning_rate": 0.0002,
+      "loss": 0.6140354871749878,
+      "mean_token_accuracy": 0.7516682893037796,
+      "num_tokens": 9810897.0,
+      "step": 128
+    },
+    {
+      "entropy": 0.6132084429264069,
+      "epoch": 1.7043189368770764,
+      "grad_norm": 0.02278308942914009,
+      "learning_rate": 0.0002,
+      "loss": 0.6116592884063721,
+      "mean_token_accuracy": 0.7542857527732849,
+      "num_tokens": 9888143.0,
+      "step": 129
+    },
+    {
+      "entropy": 0.6155333071947098,
+      "epoch": 1.717607973421927,
+      "grad_norm": 0.018389983102679253,
+      "learning_rate": 0.0002,
+      "loss": 0.6164166927337646,
+      "mean_token_accuracy": 0.7513966411352158,
+      "num_tokens": 9964562.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.6257398724555969,
+      "epoch": 1.7308970099667773,
+      "grad_norm": 0.0222679041326046,
+      "learning_rate": 0.0002,
+      "loss": 0.6244347095489502,
+      "mean_token_accuracy": 0.746750995516777,
+      "num_tokens": 10041889.0,
+      "step": 131
+    },
+    {
+      "entropy": 0.6176516264677048,
+      "epoch": 1.744186046511628,
+      "grad_norm": 0.023927317932248116,
+      "learning_rate": 0.0002,
+      "loss": 0.6160902380943298,
+      "mean_token_accuracy": 0.7516377419233322,
+      "num_tokens": 10119032.0,
+      "step": 132
+    },
+    {
+      "entropy": 0.6168064475059509,
+      "epoch": 1.7574750830564785,
+      "grad_norm": 0.022455649450421333,
+      "learning_rate": 0.0002,
+      "loss": 0.6156400442123413,
+      "mean_token_accuracy": 0.7509373724460602,
+      "num_tokens": 10195982.0,
+      "step": 133
+    },
+    {
+      "entropy": 0.6134222745895386,
+      "epoch": 1.770764119601329,
+      "grad_norm": 0.022489670664072037,
+      "learning_rate": 0.0002,
+      "loss": 0.609217643737793,
+      "mean_token_accuracy": 0.7540788650512695,
+      "num_tokens": 10273546.0,
+      "step": 134
+    },
+    {
+      "entropy": 0.6264749467372894,
+      "epoch": 1.7840531561461794,
+      "grad_norm": 0.020588450133800507,
+      "learning_rate": 0.0002,
+      "loss": 0.6276311874389648,
+      "mean_token_accuracy": 0.7467976957559586,
+      "num_tokens": 10350201.0,
+      "step": 135
+    },
+    {
+      "entropy": 0.617802083492279,
+      "epoch": 1.7973421926910298,
+      "grad_norm": 0.022901175543665886,
+      "learning_rate": 0.0002,
+      "loss": 0.6194970607757568,
+      "mean_token_accuracy": 0.748701885342598,
+      "num_tokens": 10426453.0,
+      "step": 136
+    },
+    {
+      "entropy": 0.6245028972625732,
+      "epoch": 1.8106312292358804,
+      "grad_norm": 0.023188933730125427,
+      "learning_rate": 0.0002,
+      "loss": 0.6228166222572327,
+      "mean_token_accuracy": 0.7493370622396469,
+      "num_tokens": 10504425.0,
+      "step": 137
+    },
+    {
+      "entropy": 0.6234703809022903,
+      "epoch": 1.823920265780731,
+      "grad_norm": 0.021458793431520462,
+      "learning_rate": 0.0002,
+      "loss": 0.6192694902420044,
+      "mean_token_accuracy": 0.7504332214593887,
+      "num_tokens": 10581329.0,
+      "step": 138
+    },
+    {
+      "entropy": 0.6237349808216095,
+      "epoch": 1.8372093023255816,
+      "grad_norm": 0.019868241623044014,
+      "learning_rate": 0.0002,
+      "loss": 0.6175291538238525,
+      "mean_token_accuracy": 0.75069360435009,
+      "num_tokens": 10657436.0,
+      "step": 139
+    },
+    {
+      "entropy": 0.6118619441986084,
+      "epoch": 1.850498338870432,
+      "grad_norm": 0.02374374121427536,
+      "learning_rate": 0.0002,
+      "loss": 0.612147331237793,
+      "mean_token_accuracy": 0.7540915310382843,
+      "num_tokens": 10734436.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.6119570136070251,
+      "epoch": 1.8637873754152823,
+      "grad_norm": 0.021657617762684822,
+      "learning_rate": 0.0002,
+      "loss": 0.613890528678894,
+      "mean_token_accuracy": 0.7516901344060898,
+      "num_tokens": 10811567.0,
+      "step": 141
+    },
+    {
+      "entropy": 0.6127304881811142,
+      "epoch": 1.8770764119601329,
+      "grad_norm": 0.02199709787964821,
+      "learning_rate": 0.0002,
+      "loss": 0.6144627332687378,
+      "mean_token_accuracy": 0.7525850534439087,
+      "num_tokens": 10889817.0,
+      "step": 142
+    },
+    {
+      "entropy": 0.6066692769527435,
+      "epoch": 1.8903654485049834,
+      "grad_norm": 0.021036647260189056,
+      "learning_rate": 0.0002,
+      "loss": 0.6056303977966309,
+      "mean_token_accuracy": 0.7554345726966858,
+      "num_tokens": 10965918.0,
+      "step": 143
+    },
+    {
+      "entropy": 0.6308126002550125,
+      "epoch": 1.903654485049834,
+      "grad_norm": 0.021853173151612282,
+      "learning_rate": 0.0002,
+      "loss": 0.6275626420974731,
+      "mean_token_accuracy": 0.7469644248485565,
+      "num_tokens": 11042656.0,
+      "step": 144
+    },
+    {
+      "entropy": 0.6248054802417755,
+      "epoch": 1.9169435215946844,
+      "grad_norm": 0.01832008920609951,
+      "learning_rate": 0.0002,
+      "loss": 0.6218788623809814,
+      "mean_token_accuracy": 0.7486702352762222,
+      "num_tokens": 11119636.0,
+      "step": 145
+    },
+    {
+      "entropy": 0.6077689826488495,
+      "epoch": 1.9302325581395348,
+      "grad_norm": 0.020590562373399734,
+      "learning_rate": 0.0002,
+      "loss": 0.6078977584838867,
+      "mean_token_accuracy": 0.753954142332077,
+      "num_tokens": 11197034.0,
+      "step": 146
+    },
+    {
+      "entropy": 0.6224581599235535,
+      "epoch": 1.9435215946843853,
+      "grad_norm": 0.02063332125544548,
+      "learning_rate": 0.0002,
+      "loss": 0.623686671257019,
+      "mean_token_accuracy": 0.7492604851722717,
+      "num_tokens": 11273724.0,
+      "step": 147
+    },
+    {
+      "entropy": 0.6076382547616959,
+      "epoch": 1.956810631229236,
+      "grad_norm": 0.023211153224110603,
+      "learning_rate": 0.0002,
+      "loss": 0.6072738170623779,
+      "mean_token_accuracy": 0.7551898807287216,
+      "num_tokens": 11351227.0,
+      "step": 148
+    },
+    {
+      "entropy": 0.6254193186759949,
+      "epoch": 1.9700996677740865,
+      "grad_norm": 0.019823260605335236,
+      "learning_rate": 0.0002,
+      "loss": 0.6203402280807495,
+      "mean_token_accuracy": 0.7478555142879486,
+      "num_tokens": 11426932.0,
+      "step": 149
+    },
+    {
+      "entropy": 0.6270859688520432,
+      "epoch": 1.9833887043189369,
+      "grad_norm": 0.02455417439341545,
+      "learning_rate": 0.0002,
+      "loss": 0.6208802461624146,
+      "mean_token_accuracy": 0.7491328865289688,
+      "num_tokens": 11504314.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.6147411018610001,
+      "epoch": 1.9966777408637872,
+      "grad_norm": 0.019483575597405434,
+      "learning_rate": 0.0002,
+      "loss": 0.6152125000953674,
+      "mean_token_accuracy": 0.7504801601171494,
+      "num_tokens": 11580048.0,
+      "step": 151
+    },
+    {
+      "entropy": 0.6170828342437744,
+      "epoch": 2.0,
+      "grad_norm": 0.033206261694431305,
+      "learning_rate": 0.0002,
+      "loss": 0.6186127662658691,
+      "mean_token_accuracy": 0.7514370679855347,
+      "num_tokens": 11598964.0,
+      "step": 152
+    },
+    {
+      "entropy": 0.6080913096666336,
+      "epoch": 2.0132890365448506,
+      "grad_norm": 0.024741163477301598,
+      "learning_rate": 0.0002,
+      "loss": 0.6090770363807678,
+      "mean_token_accuracy": 0.7537964433431625,
+      "num_tokens": 11675540.0,
+      "step": 153
+    },
+    {
+      "entropy": 0.6185880750417709,
+      "epoch": 2.026578073089701,
+      "grad_norm": 0.02093474380671978,
+      "learning_rate": 0.0002,
+      "loss": 0.6206169128417969,
+      "mean_token_accuracy": 0.7499336451292038,
+      "num_tokens": 11751286.0,
+      "step": 154
+    },
+    {
+      "entropy": 0.6125958114862442,
+      "epoch": 2.0398671096345513,
+      "grad_norm": 0.02114335261285305,
+      "learning_rate": 0.0002,
+      "loss": 0.6094862818717957,
+      "mean_token_accuracy": 0.7536913305521011,
+      "num_tokens": 11827770.0,
+      "step": 155
+    },
+    {
+      "entropy": 0.6116930991411209,
+      "epoch": 2.053156146179402,
+      "grad_norm": 0.02459721826016903,
+      "learning_rate": 0.0002,
+      "loss": 0.6092466115951538,
+      "mean_token_accuracy": 0.7562535107135773,
+      "num_tokens": 11904344.0,
+      "step": 156
+    },
+    {
+      "entropy": 0.6187087297439575,
+      "epoch": 2.0664451827242525,
+      "grad_norm": 0.022668739780783653,
+      "learning_rate": 0.0002,
+      "loss": 0.6161172986030579,
+      "mean_token_accuracy": 0.7519093155860901,
+      "num_tokens": 11980923.0,
+      "step": 157
+    },
+    {
+      "entropy": 0.6256630122661591,
+      "epoch": 2.079734219269103,
+      "grad_norm": 0.021594170480966568,
+      "learning_rate": 0.0002,
+      "loss": 0.6238419413566589,
+      "mean_token_accuracy": 0.7482917159795761,
+      "num_tokens": 12057499.0,
+      "step": 158
+    },
+    {
+      "entropy": 0.6037028580904007,
+      "epoch": 2.0930232558139537,
+      "grad_norm": 0.02383469231426716,
+      "learning_rate": 0.0002,
+      "loss": 0.6051764488220215,
+      "mean_token_accuracy": 0.7555714845657349,
+      "num_tokens": 12135243.0,
+      "step": 159
+    },
+    {
+      "entropy": 0.6120942682027817,
+      "epoch": 2.106312292358804,
+      "grad_norm": 0.02117961086332798,
+      "learning_rate": 0.0002,
+      "loss": 0.6129953861236572,
+      "mean_token_accuracy": 0.7540014684200287,
+      "num_tokens": 12212331.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.6064229160547256,
+      "epoch": 2.1196013289036544,
+      "grad_norm": 0.024708833545446396,
+      "learning_rate": 0.0002,
+      "loss": 0.6064392328262329,
+      "mean_token_accuracy": 0.7545509487390518,
+      "num_tokens": 12290324.0,
+      "step": 161
+    },
+    {
+      "entropy": 0.619470626115799,
+      "epoch": 2.132890365448505,
+      "grad_norm": 0.01938418298959732,
+      "learning_rate": 0.0002,
+      "loss": 0.6178134679794312,
+      "mean_token_accuracy": 0.7504110932350159,
+      "num_tokens": 12368376.0,
+      "step": 162
+    },
+    {
+      "entropy": 0.6222792118787766,
+      "epoch": 2.1461794019933556,
+      "grad_norm": 0.02473956160247326,
+      "learning_rate": 0.0002,
+      "loss": 0.6153433322906494,
+      "mean_token_accuracy": 0.7508186101913452,
+      "num_tokens": 12445346.0,
+      "step": 163
+    },
+    {
+      "entropy": 0.6072852462530136,
+      "epoch": 2.159468438538206,
+      "grad_norm": 0.02353556640446186,
+      "learning_rate": 0.0002,
+      "loss": 0.6025260090827942,
+      "mean_token_accuracy": 0.7573738545179367,
+      "num_tokens": 12522887.0,
+      "step": 164
+    },
+    {
+      "entropy": 0.619188591837883,
+      "epoch": 2.1727574750830563,
+      "grad_norm": 0.0245651975274086,
+      "learning_rate": 0.0002,
+      "loss": 0.6182605028152466,
+      "mean_token_accuracy": 0.7504746615886688,
+      "num_tokens": 12600607.0,
+      "step": 165
+    },
+    {
+      "entropy": 0.6105179488658905,
+      "epoch": 2.186046511627907,
+      "grad_norm": 0.024771464988589287,
+      "learning_rate": 0.0002,
+      "loss": 0.6139010787010193,
+      "mean_token_accuracy": 0.7523958832025528,
+      "num_tokens": 12677955.0,
+      "step": 166
+    },
+    {
+      "entropy": 0.6116905808448792,
+      "epoch": 2.1993355481727574,
+      "grad_norm": 0.01931098662316799,
+      "learning_rate": 0.0002,
+      "loss": 0.6103055477142334,
+      "mean_token_accuracy": 0.7530241310596466,
+      "num_tokens": 12755809.0,
+      "step": 167
+    },
+    {
+      "entropy": 0.6136868000030518,
+      "epoch": 2.212624584717608,
+      "grad_norm": 0.02252528816461563,
+      "learning_rate": 0.0002,
+      "loss": 0.6109723448753357,
+      "mean_token_accuracy": 0.7527836114168167,
+      "num_tokens": 12831666.0,
+      "step": 168
+    },
+    {
+      "entropy": 0.6242966800928116,
+      "epoch": 2.2259136212624586,
+      "grad_norm": 0.020864849910140038,
+      "learning_rate": 0.0002,
+      "loss": 0.6239097118377686,
+      "mean_token_accuracy": 0.749090164899826,
+      "num_tokens": 12908442.0,
+      "step": 169
+    },
+    {
+      "entropy": 0.601732075214386,
+      "epoch": 2.2392026578073088,
+      "grad_norm": 0.02746833860874176,
+      "learning_rate": 0.0002,
+      "loss": 0.6067190170288086,
+      "mean_token_accuracy": 0.7551953494548798,
+      "num_tokens": 12985427.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.6038936227560043,
+      "epoch": 2.2524916943521593,
+      "grad_norm": 0.019441375508904457,
+      "learning_rate": 0.0002,
+      "loss": 0.6017920970916748,
+      "mean_token_accuracy": 0.7557870447635651,
+      "num_tokens": 13062487.0,
+      "step": 171
+    },
+    {
+      "entropy": 0.6137015223503113,
+      "epoch": 2.26578073089701,
+      "grad_norm": 0.024601176381111145,
+      "learning_rate": 0.0002,
+      "loss": 0.6114258766174316,
+      "mean_token_accuracy": 0.7543772608041763,
+      "num_tokens": 13138627.0,
+      "step": 172
+    },
+    {
+      "entropy": 0.6104173511266708,
+      "epoch": 2.2790697674418605,
+      "grad_norm": 0.022791611030697823,
+      "learning_rate": 0.0002,
+      "loss": 0.6149581670761108,
+      "mean_token_accuracy": 0.7507553249597549,
+      "num_tokens": 13214261.0,
+      "step": 173
+    },
+    {
+      "entropy": 0.6144902855157852,
+      "epoch": 2.292358803986711,
+      "grad_norm": 0.018319284543395042,
+      "learning_rate": 0.0002,
+      "loss": 0.6141914129257202,
+      "mean_token_accuracy": 0.7522214949131012,
+      "num_tokens": 13291202.0,
+      "step": 174
+    },
+    {
+      "entropy": 0.6215455085039139,
+      "epoch": 2.3056478405315612,
+      "grad_norm": 0.025469396263360977,
+      "learning_rate": 0.0002,
+      "loss": 0.6204476356506348,
+      "mean_token_accuracy": 0.7488576024770737,
+      "num_tokens": 13368064.0,
+      "step": 175
+    },
+    {
+      "entropy": 0.6212230175733566,
+      "epoch": 2.318936877076412,
+      "grad_norm": 0.02412879653275013,
+      "learning_rate": 0.0002,
+      "loss": 0.6155945062637329,
+      "mean_token_accuracy": 0.7521145790815353,
+      "num_tokens": 13445653.0,
+      "step": 176
+    },
+    {
+      "entropy": 0.6221545487642288,
+      "epoch": 2.3322259136212624,
+      "grad_norm": 0.02200164645910263,
+      "learning_rate": 0.0002,
+      "loss": 0.6199354529380798,
+      "mean_token_accuracy": 0.7487577795982361,
+      "num_tokens": 13521614.0,
+      "step": 177
+    },
+    {
+      "entropy": 0.6135935038328171,
+      "epoch": 2.345514950166113,
+      "grad_norm": 0.022985126823186874,
+      "learning_rate": 0.0002,
+      "loss": 0.6104751825332642,
+      "mean_token_accuracy": 0.752921000123024,
+      "num_tokens": 13597755.0,
+      "step": 178
+    },
+    {
+      "entropy": 0.6188717782497406,
+      "epoch": 2.3588039867109636,
+      "grad_norm": 0.019079145044088364,
+      "learning_rate": 0.0002,
+      "loss": 0.6193943023681641,
+      "mean_token_accuracy": 0.7509453445672989,
+      "num_tokens": 13674275.0,
+      "step": 179
+    },
+    {
+      "entropy": 0.6063332110643387,
+      "epoch": 2.3720930232558137,
+      "grad_norm": 0.02187221497297287,
+      "learning_rate": 0.0002,
+      "loss": 0.6081176400184631,
+      "mean_token_accuracy": 0.7544866353273392,
+      "num_tokens": 13751926.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.6089528501033783,
+      "epoch": 2.3853820598006643,
+      "grad_norm": 0.022133421152830124,
+      "learning_rate": 0.0002,
+      "loss": 0.6117560863494873,
+      "mean_token_accuracy": 0.7531794607639313,
+      "num_tokens": 13828580.0,
+      "step": 181
+    },
+    {
+      "entropy": 0.6238999515771866,
+      "epoch": 2.398671096345515,
+      "grad_norm": 0.019620148465037346,
+      "learning_rate": 0.0002,
+      "loss": 0.6234750747680664,
+      "mean_token_accuracy": 0.747876450419426,
+      "num_tokens": 13904990.0,
+      "step": 182
+    },
+    {
+      "entropy": 0.6090637892484665,
+      "epoch": 2.4119601328903655,
+      "grad_norm": 0.021700916811823845,
+      "learning_rate": 0.0002,
+      "loss": 0.6034246683120728,
+      "mean_token_accuracy": 0.7570995390415192,
+      "num_tokens": 13983780.0,
+      "step": 183
+    },
+    {
+      "entropy": 0.619151696562767,
+      "epoch": 2.425249169435216,
+      "grad_norm": 0.020949846133589745,
+      "learning_rate": 0.0002,
+      "loss": 0.6169127225875854,
+      "mean_token_accuracy": 0.7503035366535187,
+      "num_tokens": 14060227.0,
+      "step": 184
+    },
+    {
+      "entropy": 0.5920789837837219,
+      "epoch": 2.438538205980066,
+      "grad_norm": 0.022497912868857384,
+      "learning_rate": 0.0002,
+      "loss": 0.5969008207321167,
+      "mean_token_accuracy": 0.7576948702335358,
+      "num_tokens": 14137774.0,
+      "step": 185
+    },
+    {
+      "entropy": 0.6010369509458542,
+      "epoch": 2.451827242524917,
+      "grad_norm": 0.02242516353726387,
+      "learning_rate": 0.0002,
+      "loss": 0.6066923141479492,
+      "mean_token_accuracy": 0.757031038403511,
+      "num_tokens": 14215235.0,
+      "step": 186
+    },
+    {
+      "entropy": 0.6210763901472092,
+      "epoch": 2.4651162790697674,
+      "grad_norm": 0.019126396626234055,
+      "learning_rate": 0.0002,
+      "loss": 0.61758953332901,
+      "mean_token_accuracy": 0.7510004043579102,
+      "num_tokens": 14292709.0,
+      "step": 187
+    },
+    {
+      "entropy": 0.6269999742507935,
+      "epoch": 2.478405315614618,
+      "grad_norm": 0.021928519010543823,
+      "learning_rate": 0.0002,
+      "loss": 0.6187852621078491,
+      "mean_token_accuracy": 0.7508761584758759,
+      "num_tokens": 14369455.0,
+      "step": 188
+    },
+    {
+      "entropy": 0.6171849220991135,
+      "epoch": 2.4916943521594686,
+      "grad_norm": 0.0193313080817461,
+      "learning_rate": 0.0002,
+      "loss": 0.6106323003768921,
+      "mean_token_accuracy": 0.7533420473337173,
+      "num_tokens": 14447299.0,
+      "step": 189
+    },
+    {
+      "entropy": 0.6126044690608978,
+      "epoch": 2.5049833887043187,
+      "grad_norm": 0.02693628892302513,
+      "learning_rate": 0.0002,
+      "loss": 0.6135596632957458,
+      "mean_token_accuracy": 0.7525096237659454,
+      "num_tokens": 14524325.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.6129912734031677,
+      "epoch": 2.5182724252491693,
+      "grad_norm": 0.018071722239255905,
+      "learning_rate": 0.0002,
+      "loss": 0.6142337322235107,
+      "mean_token_accuracy": 0.7519230097532272,
+      "num_tokens": 14600878.0,
+      "step": 191
+    },
+    {
+      "entropy": 0.6005150228738785,
+      "epoch": 2.53156146179402,
+      "grad_norm": 0.020812232047319412,
+      "learning_rate": 0.0002,
+      "loss": 0.6019083857536316,
+      "mean_token_accuracy": 0.7572149932384491,
+      "num_tokens": 14678986.0,
+      "step": 192
+    },
+    {
+      "entropy": 0.6029624193906784,
+      "epoch": 2.5448504983388704,
+      "grad_norm": 0.01937534287571907,
+      "learning_rate": 0.0002,
+      "loss": 0.6034685969352722,
+      "mean_token_accuracy": 0.7564429342746735,
+      "num_tokens": 14755672.0,
+      "step": 193
+    },
+    {
+      "entropy": 0.6214014291763306,
+      "epoch": 2.558139534883721,
+      "grad_norm": 0.018799029290676117,
+      "learning_rate": 0.0002,
+      "loss": 0.6193259954452515,
+      "mean_token_accuracy": 0.7494612336158752,
+      "num_tokens": 14834347.0,
+      "step": 194
+    },
+    {
+      "entropy": 0.601834237575531,
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.020581917837262154,
+      "learning_rate": 0.0002,
+      "loss": 0.6011050939559937,
+      "mean_token_accuracy": 0.7573233097791672,
+      "num_tokens": 14911834.0,
+      "step": 195
+    },
+    {
+      "entropy": 0.6118612587451935,
+      "epoch": 2.584717607973422,
+      "grad_norm": 0.01760011911392212,
+      "learning_rate": 0.0002,
+      "loss": 0.6075634956359863,
+      "mean_token_accuracy": 0.7539022862911224,
+      "num_tokens": 14989335.0,
+      "step": 196
+    },
+    {
+      "entropy": 0.6093568801879883,
+      "epoch": 2.5980066445182723,
+      "grad_norm": 0.019189173355698586,
+      "learning_rate": 0.0002,
+      "loss": 0.6092637181282043,
+      "mean_token_accuracy": 0.7539743185043335,
+      "num_tokens": 15066227.0,
+      "step": 197
+    },
+    {
+      "entropy": 0.6058226972818375,
+      "epoch": 2.611295681063123,
+      "grad_norm": 0.02091783657670021,
+      "learning_rate": 0.0002,
+      "loss": 0.6096549034118652,
+      "mean_token_accuracy": 0.7535166591405869,
+      "num_tokens": 15144896.0,
+      "step": 198
+    },
+    {
+      "entropy": 0.6119845509529114,
+      "epoch": 2.6245847176079735,
+      "grad_norm": 0.0155878197401762,
+      "learning_rate": 0.0002,
+      "loss": 0.6110509037971497,
+      "mean_token_accuracy": 0.7528887540102005,
+      "num_tokens": 15222726.0,
+      "step": 199
+    },
+    {
+      "entropy": 0.6130461543798447,
+      "epoch": 2.6378737541528237,
+      "grad_norm": 0.019653214141726494,
+      "learning_rate": 0.0002,
+      "loss": 0.6072310209274292,
+      "mean_token_accuracy": 0.7544162422418594,
+      "num_tokens": 15299909.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6181019693613052,
+      "epoch": 2.6511627906976747,
+      "grad_norm": 0.01918438822031021,
+      "learning_rate": 0.0002,
+      "loss": 0.6127801537513733,
+      "mean_token_accuracy": 0.7529577761888504,
+      "num_tokens": 15377612.0,
+      "step": 201
+    },
+    {
+      "entropy": 0.612848162651062,
+      "epoch": 2.664451827242525,
+      "grad_norm": 0.017989927902817726,
+      "learning_rate": 0.0002,
+      "loss": 0.6118600964546204,
+      "mean_token_accuracy": 0.7521955966949463,
+      "num_tokens": 15455592.0,
+      "step": 202
+    },
+    {
+      "entropy": 0.6041008532047272,
+      "epoch": 2.6777408637873754,
+      "grad_norm": 0.020594241097569466,
+      "learning_rate": 0.0002,
+      "loss": 0.6040845513343811,
+      "mean_token_accuracy": 0.7570705264806747,
+      "num_tokens": 15533049.0,
+      "step": 203
+    },
+    {
+      "entropy": 0.6044898629188538,
+      "epoch": 2.691029900332226,
+      "grad_norm": 0.019329074770212173,
+      "learning_rate": 0.0002,
+      "loss": 0.6082852482795715,
+      "mean_token_accuracy": 0.7537263184785843,
+      "num_tokens": 15610279.0,
+      "step": 204
+    },
+    {
+      "entropy": 0.6139257401227951,
+      "epoch": 2.704318936877076,
+      "grad_norm": 0.02171601541340351,
+      "learning_rate": 0.0002,
+      "loss": 0.6151280999183655,
+      "mean_token_accuracy": 0.7505405843257904,
+      "num_tokens": 15687157.0,
+      "step": 205
+    },
+    {
+      "entropy": 0.6186133027076721,
+      "epoch": 2.717607973421927,
+      "grad_norm": 0.021530168130993843,
+      "learning_rate": 0.0002,
+      "loss": 0.6109325885772705,
+      "mean_token_accuracy": 0.7531622499227524,
+      "num_tokens": 15764579.0,
+      "step": 206
+    },
+    {
+      "entropy": 0.632555365562439,
+      "epoch": 2.7308970099667773,
+      "grad_norm": 0.021901527419686317,
+      "learning_rate": 0.0002,
+      "loss": 0.6271617412567139,
+      "mean_token_accuracy": 0.7462858557701111,
+      "num_tokens": 15841481.0,
+      "step": 207
+    },
+    {
+      "entropy": 0.6050974130630493,
+      "epoch": 2.744186046511628,
+      "grad_norm": 0.01687164232134819,
+      "learning_rate": 0.0002,
+      "loss": 0.6053076386451721,
+      "mean_token_accuracy": 0.7553501129150391,
+      "num_tokens": 15918604.0,
+      "step": 208
+    },
+    {
+      "entropy": 0.6087679266929626,
+      "epoch": 2.7574750830564785,
+      "grad_norm": 0.022793259471654892,
+      "learning_rate": 0.0002,
+      "loss": 0.6142212748527527,
+      "mean_token_accuracy": 0.751279816031456,
+      "num_tokens": 15994660.0,
+      "step": 209
+    },
+    {
+      "entropy": 0.6259231269359589,
+      "epoch": 2.770764119601329,
+      "grad_norm": 0.02084580808877945,
+      "learning_rate": 0.0002,
+      "loss": 0.6297311782836914,
+      "mean_token_accuracy": 0.7452670186758041,
+      "num_tokens": 16071309.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.6174523383378983,
+      "epoch": 2.7840531561461797,
+      "grad_norm": 0.01989596337080002,
+      "learning_rate": 0.0002,
+      "loss": 0.6135072708129883,
+      "mean_token_accuracy": 0.7530443668365479,
+      "num_tokens": 16148017.0,
+      "step": 211
+    },
+    {
+      "entropy": 0.6220860630273819,
+      "epoch": 2.79734219269103,
+      "grad_norm": 0.02036948874592781,
+      "learning_rate": 0.0002,
+      "loss": 0.6163336038589478,
+      "mean_token_accuracy": 0.7497790604829788,
+      "num_tokens": 16224889.0,
+      "step": 212
+    },
+    {
+      "entropy": 0.6255825906991959,
+      "epoch": 2.8106312292358804,
+      "grad_norm": 0.018022220581769943,
+      "learning_rate": 0.0002,
+      "loss": 0.6198499202728271,
+      "mean_token_accuracy": 0.7492111325263977,
+      "num_tokens": 16302189.0,
+      "step": 213
+    },
+    {
+      "entropy": 0.6192898452281952,
+      "epoch": 2.823920265780731,
+      "grad_norm": 0.025126567110419273,
+      "learning_rate": 0.0002,
+      "loss": 0.6222077012062073,
+      "mean_token_accuracy": 0.749178871512413,
+      "num_tokens": 16378282.0,
+      "step": 214
+    },
+    {
+      "entropy": 0.6129559427499771,
+      "epoch": 2.8372093023255816,
+      "grad_norm": 0.02307634986937046,
+      "learning_rate": 0.0002,
+      "loss": 0.6161656975746155,
+      "mean_token_accuracy": 0.7505893558263779,
+      "num_tokens": 16454604.0,
+      "step": 215
+    },
+    {
+      "entropy": 0.6182385087013245,
+      "epoch": 2.850498338870432,
+      "grad_norm": 0.024152085185050964,
+      "learning_rate": 0.0002,
+      "loss": 0.6123607158660889,
+      "mean_token_accuracy": 0.7522745132446289,
+      "num_tokens": 16531409.0,
+      "step": 216
+    },
+    {
+      "entropy": 0.6112461090087891,
+      "epoch": 2.8637873754152823,
+      "grad_norm": 0.0176935363560915,
+      "learning_rate": 0.0002,
+      "loss": 0.6054509878158569,
+      "mean_token_accuracy": 0.7557886838912964,
+      "num_tokens": 16609113.0,
+      "step": 217
+    },
+    {
+      "entropy": 0.613155335187912,
+      "epoch": 2.877076411960133,
+      "grad_norm": 0.017970658838748932,
+      "learning_rate": 0.0002,
+      "loss": 0.6126760244369507,
+      "mean_token_accuracy": 0.7521080523729324,
+      "num_tokens": 16686355.0,
+      "step": 218
+    },
+    {
+      "entropy": 0.6032315939664841,
+      "epoch": 2.8903654485049834,
+      "grad_norm": 0.019020307809114456,
+      "learning_rate": 0.0002,
+      "loss": 0.6042345762252808,
+      "mean_token_accuracy": 0.7560944706201553,
+      "num_tokens": 16763608.0,
+      "step": 219
+    },
+    {
+      "entropy": 0.6077173799276352,
+      "epoch": 2.903654485049834,
+      "grad_norm": 0.018304336816072464,
+      "learning_rate": 0.0002,
+      "loss": 0.6100585460662842,
+      "mean_token_accuracy": 0.752144530415535,
+      "num_tokens": 16840993.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.6215966045856476,
+      "epoch": 2.9169435215946846,
+      "grad_norm": 0.017233379185199738,
+      "learning_rate": 0.0002,
+      "loss": 0.6201040744781494,
+      "mean_token_accuracy": 0.7491229772567749,
+      "num_tokens": 16918681.0,
+      "step": 221
+    },
+    {
+      "entropy": 0.6101587861776352,
+      "epoch": 2.9302325581395348,
+      "grad_norm": 0.017897607758641243,
+      "learning_rate": 0.0002,
+      "loss": 0.6113307476043701,
+      "mean_token_accuracy": 0.7526025176048279,
+      "num_tokens": 16995481.0,
+      "step": 222
+    },
+    {
+      "entropy": 0.6191541105508804,
+      "epoch": 2.9435215946843853,
+      "grad_norm": 0.017590591683983803,
+      "learning_rate": 0.0002,
+      "loss": 0.6188543438911438,
+      "mean_token_accuracy": 0.7504222095012665,
+      "num_tokens": 17071026.0,
+      "step": 223
+    },
+    {
+      "entropy": 0.6130050867795944,
+      "epoch": 2.956810631229236,
+      "grad_norm": 0.017167285084724426,
+      "learning_rate": 0.0002,
+      "loss": 0.6105394959449768,
+      "mean_token_accuracy": 0.7534568905830383,
+      "num_tokens": 17148098.0,
+      "step": 224
+    },
+    {
+      "entropy": 0.6177979409694672,
+      "epoch": 2.9700996677740865,
+      "grad_norm": 0.018956758081912994,
+      "learning_rate": 0.0002,
+      "loss": 0.6166577339172363,
+      "mean_token_accuracy": 0.7518346011638641,
+      "num_tokens": 17225177.0,
+      "step": 225
+    },
+    {
+      "entropy": 0.6153401881456375,
+      "epoch": 2.983388704318937,
+      "grad_norm": 0.01750190183520317,
+      "learning_rate": 0.0002,
+      "loss": 0.6157947778701782,
+      "mean_token_accuracy": 0.7507430613040924,
+      "num_tokens": 17302294.0,
+      "step": 226
+    },
+    {
+      "entropy": 0.6117560118436813,
+      "epoch": 2.9966777408637872,
+      "grad_norm": 0.019476674497127533,
+      "learning_rate": 0.0002,
+      "loss": 0.6088637113571167,
+      "mean_token_accuracy": 0.7535500228404999,
+      "num_tokens": 17379608.0,
+      "step": 227
+    },
+    {
+      "entropy": 0.6316746473312378,
+      "epoch": 3.0,
+      "grad_norm": 0.03286419063806534,
+      "learning_rate": 0.0002,
+      "loss": 0.6236717104911804,
+      "mean_token_accuracy": 0.7476998567581177,
+      "num_tokens": 17398539.0,
+      "step": 228
+    },
+    {
+      "entropy": 0.6095534861087799,
+      "epoch": 3.0132890365448506,
+      "grad_norm": 0.015681935474276543,
+      "learning_rate": 0.0002,
+      "loss": 0.6087620854377747,
+      "mean_token_accuracy": 0.7521023005247116,
+      "num_tokens": 17475557.0,
+      "step": 229
+    },
+    {
+      "entropy": 0.5973314344882965,
+      "epoch": 3.026578073089701,
+      "grad_norm": 0.022470587864518166,
+      "learning_rate": 0.0002,
+      "loss": 0.5996940732002258,
+      "mean_token_accuracy": 0.7573732435703278,
+      "num_tokens": 17553481.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.6088445037603378,
+      "epoch": 3.0398671096345513,
+      "grad_norm": 0.018227772787213326,
+      "learning_rate": 0.0002,
+      "loss": 0.6091829538345337,
+      "mean_token_accuracy": 0.7527692764997482,
+      "num_tokens": 17630948.0,
+      "step": 231
+    },
+    {
+      "entropy": 0.6049168854951859,
+      "epoch": 3.053156146179402,
+      "grad_norm": 0.017723534256219864,
+      "learning_rate": 0.0002,
+      "loss": 0.6032490730285645,
+      "mean_token_accuracy": 0.7568238973617554,
+      "num_tokens": 17708071.0,
+      "step": 232
+    },
+    {
+      "entropy": 0.6116957366466522,
+      "epoch": 3.0664451827242525,
+      "grad_norm": 0.019640754908323288,
+      "learning_rate": 0.0002,
+      "loss": 0.6129127144813538,
+      "mean_token_accuracy": 0.7514138072729111,
+      "num_tokens": 17785914.0,
+      "step": 233
+    },
+    {
+      "entropy": 0.6018583327531815,
+      "epoch": 3.079734219269103,
+      "grad_norm": 0.01823512278497219,
+      "learning_rate": 0.0002,
+      "loss": 0.6023464202880859,
+      "mean_token_accuracy": 0.7567893713712692,
+      "num_tokens": 17863068.0,
+      "step": 234
+    },
+    {
+      "entropy": 0.6079549640417099,
+      "epoch": 3.0930232558139537,
+      "grad_norm": 0.019149288535118103,
+      "learning_rate": 0.0002,
+      "loss": 0.6062753200531006,
+      "mean_token_accuracy": 0.7556976824998856,
+      "num_tokens": 17940519.0,
+      "step": 235
+    },
+    {
+      "entropy": 0.6101312786340714,
+      "epoch": 3.106312292358804,
+      "grad_norm": 0.01750970259308815,
+      "learning_rate": 0.0002,
+      "loss": 0.6074748039245605,
+      "mean_token_accuracy": 0.7548793107271194,
+      "num_tokens": 18016965.0,
+      "step": 236
+    },
+    {
+      "entropy": 0.5994770526885986,
+      "epoch": 3.1196013289036544,
+      "grad_norm": 0.016457732766866684,
+      "learning_rate": 0.0002,
+      "loss": 0.5967164039611816,
+      "mean_token_accuracy": 0.7584220916032791,
+      "num_tokens": 18093780.0,
+      "step": 237
+    },
+    {
+      "entropy": 0.6183150410652161,
+      "epoch": 3.132890365448505,
+      "grad_norm": 0.018564699217677116,
+      "learning_rate": 0.0002,
+      "loss": 0.6204038858413696,
+      "mean_token_accuracy": 0.7490763515233994,
+      "num_tokens": 18170368.0,
+      "step": 238
+    },
+    {
+      "entropy": 0.6149475276470184,
+      "epoch": 3.1461794019933556,
+      "grad_norm": 0.020549966022372246,
+      "learning_rate": 0.0002,
+      "loss": 0.6142029762268066,
+      "mean_token_accuracy": 0.7511983513832092,
+      "num_tokens": 18247503.0,
+      "step": 239
+    },
+    {
+      "entropy": 0.6074440479278564,
+      "epoch": 3.159468438538206,
+      "grad_norm": 0.017675703391432762,
+      "learning_rate": 0.0002,
+      "loss": 0.6047203540802002,
+      "mean_token_accuracy": 0.7551320791244507,
+      "num_tokens": 18324699.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.6145879477262497,
+      "epoch": 3.1727574750830563,
+      "grad_norm": 0.01775139756500721,
+      "learning_rate": 0.0002,
+      "loss": 0.6116558313369751,
+      "mean_token_accuracy": 0.7518284618854523,
+      "num_tokens": 18401605.0,
+      "step": 241
+    },
+    {
+      "entropy": 0.6134417057037354,
+      "epoch": 3.186046511627907,
+      "grad_norm": 0.018705733120441437,
+      "learning_rate": 0.0002,
+      "loss": 0.6137744188308716,
+      "mean_token_accuracy": 0.7512363493442535,
+      "num_tokens": 18477835.0,
+      "step": 242
+    },
+    {
+      "entropy": 0.618760734796524,
+      "epoch": 3.1993355481727574,
+      "grad_norm": 0.017789587378501892,
+      "learning_rate": 0.0002,
+      "loss": 0.6183280348777771,
+      "mean_token_accuracy": 0.7504749894142151,
+      "num_tokens": 18554598.0,
+      "step": 243
+    },
+    {
+      "entropy": 0.6092619597911835,
+      "epoch": 3.212624584717608,
+      "grad_norm": 0.01854095049202442,
+      "learning_rate": 0.0002,
+      "loss": 0.6108678579330444,
+      "mean_token_accuracy": 0.7533981502056122,
+      "num_tokens": 18631379.0,
+      "step": 244
+    },
+    {
+      "entropy": 0.6182709038257599,
+      "epoch": 3.2259136212624586,
+      "grad_norm": 0.019728144630789757,
+      "learning_rate": 0.0002,
+      "loss": 0.6169389486312866,
+      "mean_token_accuracy": 0.7508316785097122,
+      "num_tokens": 18707923.0,
+      "step": 245
+    },
+    {
+      "entropy": 0.6175456345081329,
+      "epoch": 3.2392026578073088,
+      "grad_norm": 0.018149685114622116,
+      "learning_rate": 0.0002,
+      "loss": 0.6160181760787964,
+      "mean_token_accuracy": 0.7498097270727158,
+      "num_tokens": 18784899.0,
+      "step": 246
+    },
+    {
+      "entropy": 0.6098749786615372,
+      "epoch": 3.2524916943521593,
+      "grad_norm": 0.022258635610342026,
+      "learning_rate": 0.0002,
+      "loss": 0.606079638004303,
+      "mean_token_accuracy": 0.7557233422994614,
+      "num_tokens": 18862479.0,
+      "step": 247
+    },
+    {
+      "entropy": 0.6224018037319183,
+      "epoch": 3.26578073089701,
+      "grad_norm": 0.01621960662305355,
+      "learning_rate": 0.0002,
+      "loss": 0.618665337562561,
+      "mean_token_accuracy": 0.7498545050621033,
+      "num_tokens": 18940087.0,
+      "step": 248
+    },
+    {
+      "entropy": 0.6083466410636902,
+      "epoch": 3.2790697674418605,
+      "grad_norm": 0.019777534529566765,
+      "learning_rate": 0.0002,
+      "loss": 0.6086328029632568,
+      "mean_token_accuracy": 0.7534854263067245,
+      "num_tokens": 19016696.0,
+      "step": 249
+    },
+    {
+      "entropy": 0.6064578592777252,
+      "epoch": 3.292358803986711,
+      "grad_norm": 0.01872546598315239,
+      "learning_rate": 0.0002,
+      "loss": 0.6035436987876892,
+      "mean_token_accuracy": 0.755558580160141,
+      "num_tokens": 19094418.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6000941842794418,
+      "epoch": 3.3056478405315612,
+      "grad_norm": 0.018199468031525612,
+      "learning_rate": 0.0002,
+      "loss": 0.5997027158737183,
+      "mean_token_accuracy": 0.7565109431743622,
+      "num_tokens": 19171347.0,
+      "step": 251
+    },
+    {
+      "entropy": 0.6095773726701736,
+      "epoch": 3.318936877076412,
+      "grad_norm": 0.019168905913829803,
+      "learning_rate": 0.0002,
+      "loss": 0.6119214296340942,
+      "mean_token_accuracy": 0.7526026219129562,
+      "num_tokens": 19248370.0,
+      "step": 252
+    },
+    {
+      "entropy": 0.6261710673570633,
+      "epoch": 3.3322259136212624,
+      "grad_norm": 0.01938040554523468,
+      "learning_rate": 0.0002,
+      "loss": 0.6250886917114258,
+      "mean_token_accuracy": 0.7475594282150269,
+      "num_tokens": 19324234.0,
+      "step": 253
+    },
+    {
+      "entropy": 0.6130911856889725,
+      "epoch": 3.345514950166113,
+      "grad_norm": 0.018235258758068085,
+      "learning_rate": 0.0002,
+      "loss": 0.6104439496994019,
+      "mean_token_accuracy": 0.7527066767215729,
+      "num_tokens": 19401967.0,
+      "step": 254
+    },
+    {
+      "entropy": 0.6052329689264297,
+      "epoch": 3.3588039867109636,
+      "grad_norm": 0.01986231654882431,
+      "learning_rate": 0.0002,
+      "loss": 0.6020320653915405,
+      "mean_token_accuracy": 0.7550270408391953,
+      "num_tokens": 19478743.0,
+      "step": 255
+    },
+    {
+      "entropy": 0.6114888489246368,
+      "epoch": 3.3720930232558137,
+      "grad_norm": 0.020737677812576294,
+      "learning_rate": 0.0002,
+      "loss": 0.6141330003738403,
+      "mean_token_accuracy": 0.7509262412786484,
+      "num_tokens": 19554557.0,
+      "step": 256
+    },
+    {
+      "entropy": 0.6046124845743179,
+      "epoch": 3.3853820598006643,
+      "grad_norm": 0.017336487770080566,
+      "learning_rate": 0.0002,
+      "loss": 0.6059973239898682,
+      "mean_token_accuracy": 0.7553645223379135,
+      "num_tokens": 19631324.0,
+      "step": 257
+    },
+    {
+      "entropy": 0.6068678796291351,
+      "epoch": 3.398671096345515,
+      "grad_norm": 0.020718904212117195,
+      "learning_rate": 0.0002,
+      "loss": 0.6086840629577637,
+      "mean_token_accuracy": 0.7532851994037628,
+      "num_tokens": 19708325.0,
+      "step": 258
+    },
+    {
+      "entropy": 0.6072741150856018,
+      "epoch": 3.4119601328903655,
+      "grad_norm": 0.01954406313598156,
+      "learning_rate": 0.0002,
+      "loss": 0.6017572283744812,
+      "mean_token_accuracy": 0.755853459239006,
+      "num_tokens": 19786447.0,
+      "step": 259
+    },
+    {
+      "entropy": 0.6193552911281586,
+      "epoch": 3.425249169435216,
+      "grad_norm": 0.020680461078882217,
+      "learning_rate": 0.0002,
+      "loss": 0.6165962219238281,
+      "mean_token_accuracy": 0.7497520446777344,
+      "num_tokens": 19863145.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.6086042821407318,
+      "epoch": 3.438538205980066,
+      "grad_norm": 0.020301401615142822,
+      "learning_rate": 0.0002,
+      "loss": 0.6113531589508057,
+      "mean_token_accuracy": 0.7527741938829422,
+      "num_tokens": 19941270.0,
+      "step": 261
+    },
+    {
+      "entropy": 0.5985937118530273,
+      "epoch": 3.451827242524917,
+      "grad_norm": 0.01930115558207035,
+      "learning_rate": 0.0002,
+      "loss": 0.5998382568359375,
+      "mean_token_accuracy": 0.7559645771980286,
+      "num_tokens": 20017437.0,
+      "step": 262
+    },
+    {
+      "entropy": 0.6116500794887543,
+      "epoch": 3.4651162790697674,
+      "grad_norm": 0.022508280351758003,
+      "learning_rate": 0.0002,
+      "loss": 0.6121885180473328,
+      "mean_token_accuracy": 0.7508966326713562,
+      "num_tokens": 20094666.0,
+      "step": 263
+    },
+    {
+      "entropy": 0.6296782493591309,
+      "epoch": 3.478405315614618,
+      "grad_norm": 0.020131876692175865,
+      "learning_rate": 0.0002,
+      "loss": 0.6230026483535767,
+      "mean_token_accuracy": 0.7477834820747375,
+      "num_tokens": 20171204.0,
+      "step": 264
+    },
+    {
+      "entropy": 0.6045475602149963,
+      "epoch": 3.4916943521594686,
+      "grad_norm": 0.022214079275727272,
+      "learning_rate": 0.0002,
+      "loss": 0.6007340550422668,
+      "mean_token_accuracy": 0.7573538422584534,
+      "num_tokens": 20248428.0,
+      "step": 265
+    },
+    {
+      "entropy": 0.6215962320566177,
+      "epoch": 3.5049833887043187,
+      "grad_norm": 0.017833082005381584,
+      "learning_rate": 0.0002,
+      "loss": 0.6178830862045288,
+      "mean_token_accuracy": 0.7502043694257736,
+      "num_tokens": 20325757.0,
+      "step": 266
+    },
+    {
+      "entropy": 0.6136264950037003,
+      "epoch": 3.5182724252491693,
+      "grad_norm": 0.021755537018179893,
+      "learning_rate": 0.0002,
+      "loss": 0.6184818744659424,
+      "mean_token_accuracy": 0.7506276816129684,
+      "num_tokens": 20402903.0,
+      "step": 267
+    },
+    {
+      "entropy": 0.6180002838373184,
+      "epoch": 3.53156146179402,
+      "grad_norm": 0.021835198625922203,
+      "learning_rate": 0.0002,
+      "loss": 0.6181987524032593,
+      "mean_token_accuracy": 0.7490601688623428,
+      "num_tokens": 20478943.0,
+      "step": 268
+    },
+    {
+      "entropy": 0.6150650382041931,
+      "epoch": 3.5448504983388704,
+      "grad_norm": 0.018519792705774307,
+      "learning_rate": 0.0002,
+      "loss": 0.6126790046691895,
+      "mean_token_accuracy": 0.7521289139986038,
+      "num_tokens": 20556519.0,
+      "step": 269
+    },
+    {
+      "entropy": 0.6134590953588486,
+      "epoch": 3.558139534883721,
+      "grad_norm": 0.02214435115456581,
+      "learning_rate": 0.0002,
+      "loss": 0.6102150082588196,
+      "mean_token_accuracy": 0.7517832219600677,
+      "num_tokens": 20633265.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.6049845069646835,
+      "epoch": 3.571428571428571,
+      "grad_norm": 0.020151155069470406,
+      "learning_rate": 0.0002,
+      "loss": 0.6048744916915894,
+      "mean_token_accuracy": 0.753717228770256,
+      "num_tokens": 20710107.0,
+      "step": 271
+    },
+    {
+      "entropy": 0.6087531298398972,
+      "epoch": 3.584717607973422,
+      "grad_norm": 0.020998802036046982,
+      "learning_rate": 0.0002,
+      "loss": 0.6117347478866577,
+      "mean_token_accuracy": 0.7524616122245789,
+      "num_tokens": 20786881.0,
+      "step": 272
+    },
+    {
+      "entropy": 0.6079637855291367,
+      "epoch": 3.5980066445182723,
+      "grad_norm": 0.028278745710849762,
+      "learning_rate": 0.0002,
+      "loss": 0.6119199395179749,
+      "mean_token_accuracy": 0.7529398500919342,
+      "num_tokens": 20865202.0,
+      "step": 273
+    },
+    {
+      "entropy": 0.6151505410671234,
+      "epoch": 3.611295681063123,
+      "grad_norm": 0.021120117977261543,
+      "learning_rate": 0.0002,
+      "loss": 0.6089338064193726,
+      "mean_token_accuracy": 0.7535832971334457,
+      "num_tokens": 20942503.0,
+      "step": 274
+    },
+    {
+      "entropy": 0.610705703496933,
+      "epoch": 3.6245847176079735,
+      "grad_norm": 0.027490589767694473,
+      "learning_rate": 0.0002,
+      "loss": 0.6012994050979614,
+      "mean_token_accuracy": 0.7582906633615494,
+      "num_tokens": 21020344.0,
+      "step": 275
+    },
+    {
+      "entropy": 0.60928113758564,
+      "epoch": 3.6378737541528237,
+      "grad_norm": 0.01714142970740795,
+      "learning_rate": 0.0002,
+      "loss": 0.6091657876968384,
+      "mean_token_accuracy": 0.753533348441124,
+      "num_tokens": 21096173.0,
+      "step": 276
+    },
+    {
+      "entropy": 0.6081078201532364,
+      "epoch": 3.6511627906976747,
+      "grad_norm": 0.02431553602218628,
+      "learning_rate": 0.0002,
+      "loss": 0.6148205995559692,
+      "mean_token_accuracy": 0.7495276033878326,
+      "num_tokens": 21172590.0,
+      "step": 277
+    },
+    {
+      "entropy": 0.6118222177028656,
+      "epoch": 3.664451827242525,
+      "grad_norm": 0.021395722404122353,
+      "learning_rate": 0.0002,
+      "loss": 0.6128278970718384,
+      "mean_token_accuracy": 0.7514612227678299,
+      "num_tokens": 21249938.0,
+      "step": 278
+    },
+    {
+      "entropy": 0.6015144437551498,
+      "epoch": 3.6777408637873754,
+      "grad_norm": 0.020723391324281693,
+      "learning_rate": 0.0002,
+      "loss": 0.5997154712677002,
+      "mean_token_accuracy": 0.7576779872179031,
+      "num_tokens": 21326172.0,
+      "step": 279
+    },
+    {
+      "entropy": 0.6092483848333359,
+      "epoch": 3.691029900332226,
+      "grad_norm": 0.021157050505280495,
+      "learning_rate": 0.0002,
+      "loss": 0.6041587591171265,
+      "mean_token_accuracy": 0.7551498264074326,
+      "num_tokens": 21404038.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.612848699092865,
+      "epoch": 3.704318936877076,
+      "grad_norm": 0.01788610778748989,
+      "learning_rate": 0.0002,
+      "loss": 0.612815260887146,
+      "mean_token_accuracy": 0.7505994439125061,
+      "num_tokens": 21480184.0,
+      "step": 281
+    },
+    {
+      "entropy": 0.6070238351821899,
+      "epoch": 3.717607973421927,
+      "grad_norm": 0.019672557711601257,
+      "learning_rate": 0.0002,
+      "loss": 0.605423092842102,
+      "mean_token_accuracy": 0.7557143568992615,
+      "num_tokens": 21558365.0,
+      "step": 282
+    },
+    {
+      "entropy": 0.6032809317111969,
+      "epoch": 3.7308970099667773,
+      "grad_norm": 0.0190837811678648,
+      "learning_rate": 0.0002,
+      "loss": 0.6031093001365662,
+      "mean_token_accuracy": 0.7566817253828049,
+      "num_tokens": 21635728.0,
+      "step": 283
+    },
+    {
+      "entropy": 0.6072507947683334,
+      "epoch": 3.744186046511628,
+      "grad_norm": 0.019605537876486778,
+      "learning_rate": 0.0002,
+      "loss": 0.6065791249275208,
+      "mean_token_accuracy": 0.7529509514570236,
+      "num_tokens": 21713107.0,
+      "step": 284
+    },
+    {
+      "entropy": 0.6070082485675812,
+      "epoch": 3.7574750830564785,
+      "grad_norm": 0.024298321455717087,
+      "learning_rate": 0.0002,
+      "loss": 0.6076079607009888,
+      "mean_token_accuracy": 0.7544230818748474,
+      "num_tokens": 21791423.0,
+      "step": 285
+    },
+    {
+      "entropy": 0.6056810468435287,
+      "epoch": 3.770764119601329,
+      "grad_norm": 0.018431881442666054,
+      "learning_rate": 0.0002,
+      "loss": 0.6028933525085449,
+      "mean_token_accuracy": 0.7561915963888168,
+      "num_tokens": 21868420.0,
+      "step": 286
+    },
+    {
+      "entropy": 0.6198001354932785,
+      "epoch": 3.7840531561461797,
+      "grad_norm": 0.017862267792224884,
+      "learning_rate": 0.0002,
+      "loss": 0.6161085963249207,
+      "mean_token_accuracy": 0.750722274184227,
+      "num_tokens": 21944982.0,
+      "step": 287
+    },
+    {
+      "entropy": 0.6113523244857788,
+      "epoch": 3.79734219269103,
+      "grad_norm": 0.018622959032654762,
+      "learning_rate": 0.0002,
+      "loss": 0.6093748211860657,
+      "mean_token_accuracy": 0.752977579832077,
+      "num_tokens": 22021863.0,
+      "step": 288
+    },
+    {
+      "entropy": 0.6059132963418961,
+      "epoch": 3.8106312292358804,
+      "grad_norm": 0.019381843507289886,
+      "learning_rate": 0.0002,
+      "loss": 0.6066963076591492,
+      "mean_token_accuracy": 0.7544227838516235,
+      "num_tokens": 22098383.0,
+      "step": 289
+    },
+    {
+      "entropy": 0.6028926521539688,
+      "epoch": 3.823920265780731,
+      "grad_norm": 0.019265756011009216,
+      "learning_rate": 0.0002,
+      "loss": 0.6037235260009766,
+      "mean_token_accuracy": 0.755391463637352,
+      "num_tokens": 22176002.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.6092864274978638,
+      "epoch": 3.8372093023255816,
+      "grad_norm": 0.017850292846560478,
+      "learning_rate": 0.0002,
+      "loss": 0.6102504730224609,
+      "mean_token_accuracy": 0.7526959478855133,
+      "num_tokens": 22252147.0,
+      "step": 291
+    },
+    {
+      "entropy": 0.6057061702013016,
+      "epoch": 3.850498338870432,
+      "grad_norm": 0.01840970665216446,
+      "learning_rate": 0.0002,
+      "loss": 0.602671205997467,
+      "mean_token_accuracy": 0.7557494938373566,
+      "num_tokens": 22329209.0,
+      "step": 292
+    },
+    {
+      "entropy": 0.6129421442747116,
+      "epoch": 3.8637873754152823,
+      "grad_norm": 0.0204786229878664,
+      "learning_rate": 0.0002,
+      "loss": 0.607947826385498,
+      "mean_token_accuracy": 0.7542832642793655,
+      "num_tokens": 22407009.0,
+      "step": 293
+    },
+    {
+      "entropy": 0.6123421639204025,
+      "epoch": 3.877076411960133,
+      "grad_norm": 0.018487131223082542,
+      "learning_rate": 0.0002,
+      "loss": 0.6074858903884888,
+      "mean_token_accuracy": 0.7547514736652374,
+      "num_tokens": 22484212.0,
+      "step": 294
+    },
+    {
+      "entropy": 0.5983436703681946,
+      "epoch": 3.8903654485049834,
+      "grad_norm": 0.02583969570696354,
+      "learning_rate": 0.0002,
+      "loss": 0.6036494970321655,
+      "mean_token_accuracy": 0.7553357183933258,
+      "num_tokens": 22561298.0,
+      "step": 295
+    },
+    {
+      "entropy": 0.6064856052398682,
+      "epoch": 3.903654485049834,
+      "grad_norm": 0.018397798761725426,
+      "learning_rate": 0.0002,
+      "loss": 0.6074536442756653,
+      "mean_token_accuracy": 0.7535169720649719,
+      "num_tokens": 22637575.0,
+      "step": 296
+    },
+    {
+      "entropy": 0.616366982460022,
+      "epoch": 3.9169435215946846,
+      "grad_norm": 0.02177377976477146,
+      "learning_rate": 0.0002,
+      "loss": 0.6141782999038696,
+      "mean_token_accuracy": 0.7520369738340378,
+      "num_tokens": 22715485.0,
+      "step": 297
+    },
+    {
+      "entropy": 0.6121602952480316,
+      "epoch": 3.9302325581395348,
+      "grad_norm": 0.02346772328019142,
+      "learning_rate": 0.0002,
+      "loss": 0.6075172424316406,
+      "mean_token_accuracy": 0.7541345357894897,
+      "num_tokens": 22793378.0,
+      "step": 298
+    },
+    {
+      "entropy": 0.6224130690097809,
+      "epoch": 3.9435215946843853,
+      "grad_norm": 0.022563831880688667,
+      "learning_rate": 0.0002,
+      "loss": 0.6254736185073853,
+      "mean_token_accuracy": 0.7476143836975098,
+      "num_tokens": 22870065.0,
+      "step": 299
+    },
+    {
+      "entropy": 0.6144284307956696,
+      "epoch": 3.956810631229236,
+      "grad_norm": 0.023518579080700874,
+      "learning_rate": 0.0002,
+      "loss": 0.6168658137321472,
+      "mean_token_accuracy": 0.7495181858539581,
+      "num_tokens": 22947597.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6144127398729324,
+      "epoch": 3.9700996677740865,
+      "grad_norm": 0.019872847944498062,
+      "learning_rate": 0.0002,
+      "loss": 0.6118189692497253,
+      "mean_token_accuracy": 0.7515655010938644,
+      "num_tokens": 23024459.0,
+      "step": 301
+    },
+    {
+      "entropy": 0.6152849346399307,
+      "epoch": 3.983388704318937,
+      "grad_norm": 0.02049420215189457,
+      "learning_rate": 0.0002,
+      "loss": 0.6128252148628235,
+      "mean_token_accuracy": 0.752276748418808,
+      "num_tokens": 23101344.0,
+      "step": 302
+    },
+    {
+      "entropy": 0.6097502261400223,
+      "epoch": 3.9966777408637872,
+      "grad_norm": 0.01893959939479828,
+      "learning_rate": 0.0002,
+      "loss": 0.6084878444671631,
+      "mean_token_accuracy": 0.7544471174478531,
+      "num_tokens": 23178869.0,
+      "step": 303
+    },
+    {
+      "entropy": 0.6282399892807007,
+      "epoch": 4.0,
+      "grad_norm": 0.029287271201610565,
+      "learning_rate": 0.0002,
+      "loss": 0.6299281120300293,
+      "mean_token_accuracy": 0.7467920184135437,
+      "num_tokens": 23197934.0,
+      "step": 304
+    },
+    {
+      "entropy": 0.6078995913267136,
+      "epoch": 4.01328903654485,
+      "grad_norm": 0.019241703674197197,
+      "learning_rate": 0.0002,
+      "loss": 0.6031018495559692,
+      "mean_token_accuracy": 0.7556031346321106,
+      "num_tokens": 23275130.0,
+      "step": 305
+    },
+    {
+      "entropy": 0.6076922863721848,
+      "epoch": 4.026578073089701,
+      "grad_norm": 0.01899712160229683,
+      "learning_rate": 0.0002,
+      "loss": 0.6067972779273987,
+      "mean_token_accuracy": 0.7541791796684265,
+      "num_tokens": 23352058.0,
+      "step": 306
+    },
+    {
+      "entropy": 0.600939005613327,
+      "epoch": 4.039867109634551,
+      "grad_norm": 0.021121008321642876,
+      "learning_rate": 0.0002,
+      "loss": 0.5987560153007507,
+      "mean_token_accuracy": 0.7575933039188385,
+      "num_tokens": 23428667.0,
+      "step": 307
+    },
+    {
+      "entropy": 0.5970161706209183,
+      "epoch": 4.053156146179402,
+      "grad_norm": 0.018559519201517105,
+      "learning_rate": 0.0002,
+      "loss": 0.5973871350288391,
+      "mean_token_accuracy": 0.7576845437288284,
+      "num_tokens": 23506688.0,
+      "step": 308
+    },
+    {
+      "entropy": 0.6015695184469223,
+      "epoch": 4.0664451827242525,
+      "grad_norm": 0.02038990519940853,
+      "learning_rate": 0.0002,
+      "loss": 0.6018570065498352,
+      "mean_token_accuracy": 0.7563586086034775,
+      "num_tokens": 23583756.0,
+      "step": 309
+    },
+    {
+      "entropy": 0.6081244051456451,
+      "epoch": 4.079734219269103,
+      "grad_norm": 0.018375808373093605,
+      "learning_rate": 0.0002,
+      "loss": 0.6059246063232422,
+      "mean_token_accuracy": 0.7543869763612747,
+      "num_tokens": 23660956.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.6051978170871735,
+      "epoch": 4.093023255813954,
+      "grad_norm": 0.020722337067127228,
+      "learning_rate": 0.0002,
+      "loss": 0.603718638420105,
+      "mean_token_accuracy": 0.7539190798997879,
+      "num_tokens": 23737537.0,
+      "step": 311
+    },
+    {
+      "entropy": 0.6093511432409286,
+      "epoch": 4.106312292358804,
+      "grad_norm": 0.023197615519165993,
+      "learning_rate": 0.0002,
+      "loss": 0.6051844358444214,
+      "mean_token_accuracy": 0.7553325593471527,
+      "num_tokens": 23814068.0,
+      "step": 312
+    },
+    {
+      "entropy": 0.6088575571775436,
+      "epoch": 4.119601328903655,
+      "grad_norm": 0.02205248922109604,
+      "learning_rate": 0.0002,
+      "loss": 0.6104272603988647,
+      "mean_token_accuracy": 0.7520817220211029,
+      "num_tokens": 23890784.0,
+      "step": 313
+    },
+    {
+      "entropy": 0.6190015077590942,
+      "epoch": 4.132890365448505,
+      "grad_norm": 0.025014696642756462,
+      "learning_rate": 0.0002,
+      "loss": 0.620580792427063,
+      "mean_token_accuracy": 0.748920276761055,
+      "num_tokens": 23967606.0,
+      "step": 314
+    },
+    {
+      "entropy": 0.6102283596992493,
+      "epoch": 4.146179401993355,
+      "grad_norm": 0.020540745928883553,
+      "learning_rate": 0.0002,
+      "loss": 0.6080856323242188,
+      "mean_token_accuracy": 0.75365149974823,
+      "num_tokens": 24045059.0,
+      "step": 315
+    },
+    {
+      "entropy": 0.6060895025730133,
+      "epoch": 4.159468438538206,
+      "grad_norm": 0.024182360619306564,
+      "learning_rate": 0.0002,
+      "loss": 0.602268636226654,
+      "mean_token_accuracy": 0.7558044046163559,
+      "num_tokens": 24122544.0,
+      "step": 316
+    },
+    {
+      "entropy": 0.5959110260009766,
+      "epoch": 4.172757475083056,
+      "grad_norm": 0.018297569826245308,
+      "learning_rate": 0.0002,
+      "loss": 0.5969607830047607,
+      "mean_token_accuracy": 0.7568109482526779,
+      "num_tokens": 24200551.0,
+      "step": 317
+    },
+    {
+      "entropy": 0.5999089032411575,
+      "epoch": 4.186046511627907,
+      "grad_norm": 0.02408657781779766,
+      "learning_rate": 0.0002,
+      "loss": 0.6017069816589355,
+      "mean_token_accuracy": 0.7562790066003799,
+      "num_tokens": 24278083.0,
+      "step": 318
+    },
+    {
+      "entropy": 0.6096744239330292,
+      "epoch": 4.1993355481727574,
+      "grad_norm": 0.019474782049655914,
+      "learning_rate": 0.0002,
+      "loss": 0.6083636283874512,
+      "mean_token_accuracy": 0.7534827589988708,
+      "num_tokens": 24354655.0,
+      "step": 319
+    },
+    {
+      "entropy": 0.6241651922464371,
+      "epoch": 4.212624584717608,
+      "grad_norm": 0.020875826478004456,
+      "learning_rate": 0.0002,
+      "loss": 0.6206015348434448,
+      "mean_token_accuracy": 0.7486886084079742,
+      "num_tokens": 24431355.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.610044926404953,
+      "epoch": 4.225913621262459,
+      "grad_norm": 0.02331143245100975,
+      "learning_rate": 0.0002,
+      "loss": 0.6060656309127808,
+      "mean_token_accuracy": 0.7543890476226807,
+      "num_tokens": 24508973.0,
+      "step": 321
+    },
+    {
+      "entropy": 0.6080483794212341,
+      "epoch": 4.239202657807309,
+      "grad_norm": 0.020508069545030594,
+      "learning_rate": 0.0002,
+      "loss": 0.6093933582305908,
+      "mean_token_accuracy": 0.7528631091117859,
+      "num_tokens": 24586431.0,
+      "step": 322
+    },
+    {
+      "entropy": 0.6133122891187668,
+      "epoch": 4.25249169435216,
+      "grad_norm": 0.028193378821015358,
+      "learning_rate": 0.0002,
+      "loss": 0.6164911389350891,
+      "mean_token_accuracy": 0.7494592070579529,
+      "num_tokens": 24662931.0,
+      "step": 323
+    },
+    {
+      "entropy": 0.6075657159090042,
+      "epoch": 4.26578073089701,
+      "grad_norm": 0.022722216323018074,
+      "learning_rate": 0.0002,
+      "loss": 0.600710391998291,
+      "mean_token_accuracy": 0.7569031566381454,
+      "num_tokens": 24741559.0,
+      "step": 324
+    },
+    {
+      "entropy": 0.615620344877243,
+      "epoch": 4.27906976744186,
+      "grad_norm": 0.025660408660769463,
+      "learning_rate": 0.0002,
+      "loss": 0.6100776195526123,
+      "mean_token_accuracy": 0.7526432275772095,
+      "num_tokens": 24818237.0,
+      "step": 325
+    },
+    {
+      "entropy": 0.5883046835660934,
+      "epoch": 4.292358803986711,
+      "grad_norm": 0.018514303490519524,
+      "learning_rate": 0.0002,
+      "loss": 0.5883811712265015,
+      "mean_token_accuracy": 0.7614999413490295,
+      "num_tokens": 24896271.0,
+      "step": 326
+    },
+    {
+      "entropy": 0.6124174892902374,
+      "epoch": 4.305647840531561,
+      "grad_norm": 0.02814031019806862,
+      "learning_rate": 0.0002,
+      "loss": 0.6168115139007568,
+      "mean_token_accuracy": 0.7499870508909225,
+      "num_tokens": 24972812.0,
+      "step": 327
+    },
+    {
+      "entropy": 0.6104706823825836,
+      "epoch": 4.318936877076412,
+      "grad_norm": 0.017883731052279472,
+      "learning_rate": 0.0002,
+      "loss": 0.6079275608062744,
+      "mean_token_accuracy": 0.7537025660276413,
+      "num_tokens": 25049472.0,
+      "step": 328
+    },
+    {
+      "entropy": 0.6112405806779861,
+      "epoch": 4.332225913621262,
+      "grad_norm": 0.025313958525657654,
+      "learning_rate": 0.0002,
+      "loss": 0.6067090630531311,
+      "mean_token_accuracy": 0.7538552135229111,
+      "num_tokens": 25127082.0,
+      "step": 329
+    },
+    {
+      "entropy": 0.6116072982549667,
+      "epoch": 4.3455149501661126,
+      "grad_norm": 0.019383935257792473,
+      "learning_rate": 0.0002,
+      "loss": 0.6074379682540894,
+      "mean_token_accuracy": 0.7537506520748138,
+      "num_tokens": 25204323.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.6128946393728256,
+      "epoch": 4.358803986710964,
+      "grad_norm": 0.023746777325868607,
+      "learning_rate": 0.0002,
+      "loss": 0.617768406867981,
+      "mean_token_accuracy": 0.7481751292943954,
+      "num_tokens": 25280019.0,
+      "step": 331
+    },
+    {
+      "entropy": 0.6033320724964142,
+      "epoch": 4.372093023255814,
+      "grad_norm": 0.020396512001752853,
+      "learning_rate": 0.0002,
+      "loss": 0.606078028678894,
+      "mean_token_accuracy": 0.7542640864849091,
+      "num_tokens": 25356315.0,
+      "step": 332
+    },
+    {
+      "entropy": 0.6129639148712158,
+      "epoch": 4.385382059800665,
+      "grad_norm": 0.018004735931754112,
+      "learning_rate": 0.0002,
+      "loss": 0.6129087209701538,
+      "mean_token_accuracy": 0.7506652474403381,
+      "num_tokens": 25433383.0,
+      "step": 333
+    },
+    {
+      "entropy": 0.6221887469291687,
+      "epoch": 4.398671096345515,
+      "grad_norm": 0.022268855944275856,
+      "learning_rate": 0.0002,
+      "loss": 0.6149997711181641,
+      "mean_token_accuracy": 0.749396800994873,
+      "num_tokens": 25510141.0,
+      "step": 334
+    },
+    {
+      "entropy": 0.6167555749416351,
+      "epoch": 4.411960132890365,
+      "grad_norm": 0.01839759759604931,
+      "learning_rate": 0.0002,
+      "loss": 0.6128036975860596,
+      "mean_token_accuracy": 0.751603752374649,
+      "num_tokens": 25588417.0,
+      "step": 335
+    },
+    {
+      "entropy": 0.6068084985017776,
+      "epoch": 4.425249169435216,
+      "grad_norm": 0.023206762969493866,
+      "learning_rate": 0.0002,
+      "loss": 0.6072171926498413,
+      "mean_token_accuracy": 0.7537854909896851,
+      "num_tokens": 25665457.0,
+      "step": 336
+    },
+    {
+      "entropy": 0.6054465770721436,
+      "epoch": 4.438538205980066,
+      "grad_norm": 0.022475862875580788,
+      "learning_rate": 0.0002,
+      "loss": 0.6104304790496826,
+      "mean_token_accuracy": 0.752402052283287,
+      "num_tokens": 25742100.0,
+      "step": 337
+    },
+    {
+      "entropy": 0.6154294013977051,
+      "epoch": 4.451827242524917,
+      "grad_norm": 0.01877933368086815,
+      "learning_rate": 0.0002,
+      "loss": 0.6110727787017822,
+      "mean_token_accuracy": 0.7519305050373077,
+      "num_tokens": 25819237.0,
+      "step": 338
+    },
+    {
+      "entropy": 0.6140685379505157,
+      "epoch": 4.465116279069767,
+      "grad_norm": 0.022017061710357666,
+      "learning_rate": 0.0002,
+      "loss": 0.6092947721481323,
+      "mean_token_accuracy": 0.752203956246376,
+      "num_tokens": 25896673.0,
+      "step": 339
+    },
+    {
+      "entropy": 0.6095300614833832,
+      "epoch": 4.4784053156146175,
+      "grad_norm": 0.018494611606001854,
+      "learning_rate": 0.0002,
+      "loss": 0.606695830821991,
+      "mean_token_accuracy": 0.7532522529363632,
+      "num_tokens": 25973039.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.5996545255184174,
+      "epoch": 4.4916943521594686,
+      "grad_norm": 0.029419148340821266,
+      "learning_rate": 0.0002,
+      "loss": 0.6072653532028198,
+      "mean_token_accuracy": 0.7545234113931656,
+      "num_tokens": 26049966.0,
+      "step": 341
+    },
+    {
+      "entropy": 0.6011916100978851,
+      "epoch": 4.504983388704319,
+      "grad_norm": 0.018680213019251823,
+      "learning_rate": 0.0002,
+      "loss": 0.6019576191902161,
+      "mean_token_accuracy": 0.7563180327415466,
+      "num_tokens": 26126850.0,
+      "step": 342
+    },
+    {
+      "entropy": 0.6171986162662506,
+      "epoch": 4.51827242524917,
+      "grad_norm": 0.025131572037935257,
+      "learning_rate": 0.0002,
+      "loss": 0.61095130443573,
+      "mean_token_accuracy": 0.7528986483812332,
+      "num_tokens": 26202932.0,
+      "step": 343
+    },
+    {
+      "entropy": 0.6055793315172195,
+      "epoch": 4.53156146179402,
+      "grad_norm": 0.02243235521018505,
+      "learning_rate": 0.0002,
+      "loss": 0.600302517414093,
+      "mean_token_accuracy": 0.756937175989151,
+      "num_tokens": 26279555.0,
+      "step": 344
+    },
+    {
+      "entropy": 0.602609395980835,
+      "epoch": 4.544850498338871,
+      "grad_norm": 0.02473926544189453,
+      "learning_rate": 0.0002,
+      "loss": 0.6063109636306763,
+      "mean_token_accuracy": 0.7552246153354645,
+      "num_tokens": 26355532.0,
+      "step": 345
+    },
+    {
+      "entropy": 0.599660649895668,
+      "epoch": 4.558139534883721,
+      "grad_norm": 0.026357337832450867,
+      "learning_rate": 0.0002,
+      "loss": 0.6082413196563721,
+      "mean_token_accuracy": 0.7543327510356903,
+      "num_tokens": 26432597.0,
+      "step": 346
+    },
+    {
+      "entropy": 0.6072948724031448,
+      "epoch": 4.571428571428571,
+      "grad_norm": 0.022914156317710876,
+      "learning_rate": 0.0002,
+      "loss": 0.6029258370399475,
+      "mean_token_accuracy": 0.7551669776439667,
+      "num_tokens": 26509425.0,
+      "step": 347
+    },
+    {
+      "entropy": 0.6152079552412033,
+      "epoch": 4.584717607973422,
+      "grad_norm": 0.025272825732827187,
+      "learning_rate": 0.0002,
+      "loss": 0.6049219369888306,
+      "mean_token_accuracy": 0.7542230039834976,
+      "num_tokens": 26586972.0,
+      "step": 348
+    },
+    {
+      "entropy": 0.614995539188385,
+      "epoch": 4.598006644518272,
+      "grad_norm": 0.020252224057912827,
+      "learning_rate": 0.0002,
+      "loss": 0.6108348369598389,
+      "mean_token_accuracy": 0.752870187163353,
+      "num_tokens": 26664480.0,
+      "step": 349
+    },
+    {
+      "entropy": 0.6094898730516434,
+      "epoch": 4.6112956810631225,
+      "grad_norm": 0.030948877334594727,
+      "learning_rate": 0.0002,
+      "loss": 0.6137002110481262,
+      "mean_token_accuracy": 0.7513982206583023,
+      "num_tokens": 26742003.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.5978283584117889,
+      "epoch": 4.6245847176079735,
+      "grad_norm": 0.01938704401254654,
+      "learning_rate": 0.0002,
+      "loss": 0.6012290716171265,
+      "mean_token_accuracy": 0.7540938407182693,
+      "num_tokens": 26818026.0,
+      "step": 351
+    },
+    {
+      "entropy": 0.6089412868022919,
+      "epoch": 4.637873754152824,
+      "grad_norm": 0.0248025581240654,
+      "learning_rate": 0.0002,
+      "loss": 0.6055797338485718,
+      "mean_token_accuracy": 0.754187747836113,
+      "num_tokens": 26894165.0,
+      "step": 352
+    },
+    {
+      "entropy": 0.6032460033893585,
+      "epoch": 4.651162790697675,
+      "grad_norm": 0.0183709766715765,
+      "learning_rate": 0.0002,
+      "loss": 0.5996910333633423,
+      "mean_token_accuracy": 0.7573108524084091,
+      "num_tokens": 26971844.0,
+      "step": 353
+    },
+    {
+      "entropy": 0.6083274334669113,
+      "epoch": 4.664451827242525,
+      "grad_norm": 0.02345374971628189,
+      "learning_rate": 0.0002,
+      "loss": 0.610886812210083,
+      "mean_token_accuracy": 0.7522048950195312,
+      "num_tokens": 27048769.0,
+      "step": 354
+    },
+    {
+      "entropy": 0.6071414351463318,
+      "epoch": 4.677740863787376,
+      "grad_norm": 0.01983507163822651,
+      "learning_rate": 0.0002,
+      "loss": 0.6074281930923462,
+      "mean_token_accuracy": 0.7520459145307541,
+      "num_tokens": 27125617.0,
+      "step": 355
+    },
+    {
+      "entropy": 0.6114618331193924,
+      "epoch": 4.691029900332226,
+      "grad_norm": 0.019038749858736992,
+      "learning_rate": 0.0002,
+      "loss": 0.60816890001297,
+      "mean_token_accuracy": 0.7530447542667389,
+      "num_tokens": 27202362.0,
+      "step": 356
+    },
+    {
+      "entropy": 0.612061083316803,
+      "epoch": 4.704318936877076,
+      "grad_norm": 0.020625028759241104,
+      "learning_rate": 0.0002,
+      "loss": 0.6064684987068176,
+      "mean_token_accuracy": 0.7540236860513687,
+      "num_tokens": 27279536.0,
+      "step": 357
+    },
+    {
+      "entropy": 0.5989892184734344,
+      "epoch": 4.717607973421927,
+      "grad_norm": 0.022561894729733467,
+      "learning_rate": 0.0002,
+      "loss": 0.6045775413513184,
+      "mean_token_accuracy": 0.7552504390478134,
+      "num_tokens": 27355777.0,
+      "step": 358
+    },
+    {
+      "entropy": 0.6044065952301025,
+      "epoch": 4.730897009966777,
+      "grad_norm": 0.02225874364376068,
+      "learning_rate": 0.0002,
+      "loss": 0.6069464087486267,
+      "mean_token_accuracy": 0.7543323785066605,
+      "num_tokens": 27433236.0,
+      "step": 359
+    },
+    {
+      "entropy": 0.5972677916288376,
+      "epoch": 4.7441860465116275,
+      "grad_norm": 0.02093445137143135,
+      "learning_rate": 0.0002,
+      "loss": 0.5954780578613281,
+      "mean_token_accuracy": 0.7601328641176224,
+      "num_tokens": 27510986.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.6022268831729889,
+      "epoch": 4.7574750830564785,
+      "grad_norm": 0.021075721830129623,
+      "learning_rate": 0.0002,
+      "loss": 0.5991562604904175,
+      "mean_token_accuracy": 0.7573902904987335,
+      "num_tokens": 27588512.0,
+      "step": 361
+    },
+    {
+      "entropy": 0.6112276464700699,
+      "epoch": 4.770764119601329,
+      "grad_norm": 0.02440483868122101,
+      "learning_rate": 0.0002,
+      "loss": 0.6071096062660217,
+      "mean_token_accuracy": 0.7541078478097916,
+      "num_tokens": 27665251.0,
+      "step": 362
+    },
+    {
+      "entropy": 0.5961045920848846,
+      "epoch": 4.78405315614618,
+      "grad_norm": 0.022120723500847816,
+      "learning_rate": 0.0002,
+      "loss": 0.5972861051559448,
+      "mean_token_accuracy": 0.759039893746376,
+      "num_tokens": 27743007.0,
+      "step": 363
+    },
+    {
+      "entropy": 0.6066542267799377,
+      "epoch": 4.79734219269103,
+      "grad_norm": 0.023716753348708153,
+      "learning_rate": 0.0002,
+      "loss": 0.6147413849830627,
+      "mean_token_accuracy": 0.7499552518129349,
+      "num_tokens": 27820021.0,
+      "step": 364
+    },
+    {
+      "entropy": 0.5942088961601257,
+      "epoch": 4.810631229235881,
+      "grad_norm": 0.02047412469983101,
+      "learning_rate": 0.0002,
+      "loss": 0.5935304164886475,
+      "mean_token_accuracy": 0.7592828571796417,
+      "num_tokens": 27899202.0,
+      "step": 365
+    },
+    {
+      "entropy": 0.6142376959323883,
+      "epoch": 4.823920265780731,
+      "grad_norm": 0.019972192123532295,
+      "learning_rate": 0.0002,
+      "loss": 0.6080060601234436,
+      "mean_token_accuracy": 0.7540056705474854,
+      "num_tokens": 27975745.0,
+      "step": 366
+    },
+    {
+      "entropy": 0.6211676597595215,
+      "epoch": 4.837209302325581,
+      "grad_norm": 0.01891135610640049,
+      "learning_rate": 0.0002,
+      "loss": 0.6165120601654053,
+      "mean_token_accuracy": 0.7498104870319366,
+      "num_tokens": 28052581.0,
+      "step": 367
+    },
+    {
+      "entropy": 0.6221711933612823,
+      "epoch": 4.850498338870432,
+      "grad_norm": 0.021222159266471863,
+      "learning_rate": 0.0002,
+      "loss": 0.6228503584861755,
+      "mean_token_accuracy": 0.7477325350046158,
+      "num_tokens": 28130356.0,
+      "step": 368
+    },
+    {
+      "entropy": 0.6133296489715576,
+      "epoch": 4.863787375415282,
+      "grad_norm": 0.01770627126097679,
+      "learning_rate": 0.0002,
+      "loss": 0.6078294515609741,
+      "mean_token_accuracy": 0.75361567735672,
+      "num_tokens": 28209070.0,
+      "step": 369
+    },
+    {
+      "entropy": 0.6115191429853439,
+      "epoch": 4.877076411960132,
+      "grad_norm": 0.0216510072350502,
+      "learning_rate": 0.0002,
+      "loss": 0.6120319366455078,
+      "mean_token_accuracy": 0.7516486942768097,
+      "num_tokens": 28285924.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.6047016680240631,
+      "epoch": 4.8903654485049834,
+      "grad_norm": 0.01667115092277527,
+      "learning_rate": 0.0002,
+      "loss": 0.6051798462867737,
+      "mean_token_accuracy": 0.7555322796106339,
+      "num_tokens": 28363320.0,
+      "step": 371
+    },
+    {
+      "entropy": 0.5979461371898651,
+      "epoch": 4.903654485049834,
+      "grad_norm": 0.019139152020215988,
+      "learning_rate": 0.0002,
+      "loss": 0.5993932485580444,
+      "mean_token_accuracy": 0.7574638575315475,
+      "num_tokens": 28439806.0,
+      "step": 372
+    },
+    {
+      "entropy": 0.6109994500875473,
+      "epoch": 4.916943521594685,
+      "grad_norm": 0.01768781803548336,
+      "learning_rate": 0.0002,
+      "loss": 0.6106141805648804,
+      "mean_token_accuracy": 0.7535261958837509,
+      "num_tokens": 28515862.0,
+      "step": 373
+    },
+    {
+      "entropy": 0.6056813150644302,
+      "epoch": 4.930232558139535,
+      "grad_norm": 0.018335502594709396,
+      "learning_rate": 0.0002,
+      "loss": 0.6071027517318726,
+      "mean_token_accuracy": 0.7535324543714523,
+      "num_tokens": 28593556.0,
+      "step": 374
+    },
+    {
+      "entropy": 0.6102660298347473,
+      "epoch": 4.943521594684386,
+      "grad_norm": 0.021193454042077065,
+      "learning_rate": 0.0002,
+      "loss": 0.6112580299377441,
+      "mean_token_accuracy": 0.7507810145616531,
+      "num_tokens": 28669898.0,
+      "step": 375
+    },
+    {
+      "entropy": 0.6051753759384155,
+      "epoch": 4.956810631229236,
+      "grad_norm": 0.017476912587881088,
+      "learning_rate": 0.0002,
+      "loss": 0.6048047542572021,
+      "mean_token_accuracy": 0.7546576112508774,
+      "num_tokens": 28747381.0,
+      "step": 376
+    },
+    {
+      "entropy": 0.6031095385551453,
+      "epoch": 4.970099667774086,
+      "grad_norm": 0.02010265551507473,
+      "learning_rate": 0.0002,
+      "loss": 0.6020570993423462,
+      "mean_token_accuracy": 0.7562482059001923,
+      "num_tokens": 28824682.0,
+      "step": 377
+    },
+    {
+      "entropy": 0.6015651375055313,
+      "epoch": 4.983388704318937,
+      "grad_norm": 0.017594128847122192,
+      "learning_rate": 0.0002,
+      "loss": 0.5969172120094299,
+      "mean_token_accuracy": 0.7577322870492935,
+      "num_tokens": 28902597.0,
+      "step": 378
+    },
+    {
+      "entropy": 0.6108541786670685,
+      "epoch": 4.996677740863787,
+      "grad_norm": 0.017801743000745773,
+      "learning_rate": 0.0002,
+      "loss": 0.6123338937759399,
+      "mean_token_accuracy": 0.7520528733730316,
+      "num_tokens": 28978655.0,
+      "step": 379
+    },
+    {
+      "entropy": 0.6161795854568481,
+      "epoch": 5.0,
+      "grad_norm": 0.03153039887547493,
+      "learning_rate": 0.0002,
+      "loss": 0.615233302116394,
+      "mean_token_accuracy": 0.747843325138092,
+      "num_tokens": 28997446.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.5995142012834549,
+      "epoch": 5.01328903654485,
+      "grad_norm": 0.026796070858836174,
+      "learning_rate": 0.0002,
+      "loss": 0.5935805439949036,
+      "mean_token_accuracy": 0.7586488872766495,
+      "num_tokens": 29076201.0,
+      "step": 381
+    },
+    {
+      "entropy": 0.601159080862999,
+      "epoch": 5.026578073089701,
+      "grad_norm": 0.018563998863101006,
+      "learning_rate": 0.0002,
+      "loss": 0.5995253324508667,
+      "mean_token_accuracy": 0.7558069676160812,
+      "num_tokens": 29153928.0,
+      "step": 382
+    },
+    {
+      "entropy": 0.6012768000364304,
+      "epoch": 5.039867109634551,
+      "grad_norm": 0.02645065449178219,
+      "learning_rate": 0.0002,
+      "loss": 0.6047909259796143,
+      "mean_token_accuracy": 0.7550944089889526,
+      "num_tokens": 29229990.0,
+      "step": 383
+    },
+    {
+      "entropy": 0.6111899763345718,
+      "epoch": 5.053156146179402,
+      "grad_norm": 0.018297409638762474,
+      "learning_rate": 0.0002,
+      "loss": 0.6090985536575317,
+      "mean_token_accuracy": 0.7526888698339462,
+      "num_tokens": 29306722.0,
+      "step": 384
+    },
+    {
+      "entropy": 0.6029917150735855,
+      "epoch": 5.0664451827242525,
+      "grad_norm": 0.020064735785126686,
+      "learning_rate": 0.0002,
+      "loss": 0.6002858877182007,
+      "mean_token_accuracy": 0.7555706948041916,
+      "num_tokens": 29383481.0,
+      "step": 385
+    },
+    {
+      "entropy": 0.6103447079658508,
+      "epoch": 5.079734219269103,
+      "grad_norm": 0.020298415794968605,
+      "learning_rate": 0.0002,
+      "loss": 0.6126797795295715,
+      "mean_token_accuracy": 0.750050887465477,
+      "num_tokens": 29459538.0,
+      "step": 386
+    },
+    {
+      "entropy": 0.5905372053384781,
+      "epoch": 5.093023255813954,
+      "grad_norm": 0.02092077024281025,
+      "learning_rate": 0.0002,
+      "loss": 0.5949006676673889,
+      "mean_token_accuracy": 0.7592107206583023,
+      "num_tokens": 29537367.0,
+      "step": 387
+    },
+    {
+      "entropy": 0.6050836890935898,
+      "epoch": 5.106312292358804,
+      "grad_norm": 0.01800566352903843,
+      "learning_rate": 0.0002,
+      "loss": 0.6042923331260681,
+      "mean_token_accuracy": 0.7541112899780273,
+      "num_tokens": 29615052.0,
+      "step": 388
+    },
+    {
+      "entropy": 0.6001386642456055,
+      "epoch": 5.119601328903655,
+      "grad_norm": 0.018633391708135605,
+      "learning_rate": 0.0002,
+      "loss": 0.5960696935653687,
+      "mean_token_accuracy": 0.7574971616268158,
+      "num_tokens": 29692726.0,
+      "step": 389
+    },
+    {
+      "entropy": 0.6089471280574799,
+      "epoch": 5.132890365448505,
+      "grad_norm": 0.01885078474879265,
+      "learning_rate": 0.0002,
+      "loss": 0.6077396869659424,
+      "mean_token_accuracy": 0.7536272257566452,
+      "num_tokens": 29769618.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.6024085730314255,
+      "epoch": 5.146179401993355,
+      "grad_norm": 0.018104583024978638,
+      "learning_rate": 0.0002,
+      "loss": 0.600939154624939,
+      "mean_token_accuracy": 0.7558382600545883,
+      "num_tokens": 29846694.0,
+      "step": 391
+    },
+    {
+      "entropy": 0.6064421534538269,
+      "epoch": 5.159468438538206,
+      "grad_norm": 0.02171364240348339,
+      "learning_rate": 0.0002,
+      "loss": 0.6077273488044739,
+      "mean_token_accuracy": 0.7519265413284302,
+      "num_tokens": 29924010.0,
+      "step": 392
+    },
+    {
+      "entropy": 0.6087502092123032,
+      "epoch": 5.172757475083056,
+      "grad_norm": 0.01920831762254238,
+      "learning_rate": 0.0002,
+      "loss": 0.6057215929031372,
+      "mean_token_accuracy": 0.7533555775880814,
+      "num_tokens": 30000259.0,
+      "step": 393
+    },
+    {
+      "entropy": 0.6092672497034073,
+      "epoch": 5.186046511627907,
+      "grad_norm": 0.022611141204833984,
+      "learning_rate": 0.0002,
+      "loss": 0.6060106754302979,
+      "mean_token_accuracy": 0.7521137148141861,
+      "num_tokens": 30076312.0,
+      "step": 394
+    },
+    {
+      "entropy": 0.6014163643121719,
+      "epoch": 5.1993355481727574,
+      "grad_norm": 0.020460529252886772,
+      "learning_rate": 0.0002,
+      "loss": 0.6025012731552124,
+      "mean_token_accuracy": 0.7566916793584824,
+      "num_tokens": 30152994.0,
+      "step": 395
+    },
+    {
+      "entropy": 0.6095530837774277,
+      "epoch": 5.212624584717608,
+      "grad_norm": 0.025972452014684677,
+      "learning_rate": 0.0002,
+      "loss": 0.6138255596160889,
+      "mean_token_accuracy": 0.7497290223836899,
+      "num_tokens": 30230278.0,
+      "step": 396
+    },
+    {
+      "entropy": 0.6058608740568161,
+      "epoch": 5.225913621262459,
+      "grad_norm": 0.022098204120993614,
+      "learning_rate": 0.0002,
+      "loss": 0.6035760641098022,
+      "mean_token_accuracy": 0.7541337609291077,
+      "num_tokens": 30307279.0,
+      "step": 397
+    },
+    {
+      "entropy": 0.6170176565647125,
+      "epoch": 5.239202657807309,
+      "grad_norm": 0.023136481642723083,
+      "learning_rate": 0.0002,
+      "loss": 0.6125384569168091,
+      "mean_token_accuracy": 0.7498274445533752,
+      "num_tokens": 30383739.0,
+      "step": 398
+    },
+    {
+      "entropy": 0.6107608377933502,
+      "epoch": 5.25249169435216,
+      "grad_norm": 0.02006644383072853,
+      "learning_rate": 0.0002,
+      "loss": 0.6078338623046875,
+      "mean_token_accuracy": 0.7535765767097473,
+      "num_tokens": 30461994.0,
+      "step": 399
+    },
+    {
+      "entropy": 0.5954063683748245,
+      "epoch": 5.26578073089701,
+      "grad_norm": 0.0214250311255455,
+      "learning_rate": 0.0002,
+      "loss": 0.5987457633018494,
+      "mean_token_accuracy": 0.7550365626811981,
+      "num_tokens": 30539252.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.5920732617378235,
+      "epoch": 5.27906976744186,
+      "grad_norm": 0.023511109873652458,
+      "learning_rate": 0.0002,
+      "loss": 0.5931231379508972,
+      "mean_token_accuracy": 0.7596782594919205,
+      "num_tokens": 30616185.0,
+      "step": 401
+    },
+    {
+      "entropy": 0.6125746220350266,
+      "epoch": 5.292358803986711,
+      "grad_norm": 0.021425219252705574,
+      "learning_rate": 0.0002,
+      "loss": 0.6082620024681091,
+      "mean_token_accuracy": 0.7525101006031036,
+      "num_tokens": 30694587.0,
+      "step": 402
+    },
+    {
+      "entropy": 0.6051463186740875,
+      "epoch": 5.305647840531561,
+      "grad_norm": 0.02150619402527809,
+      "learning_rate": 0.0002,
+      "loss": 0.6045330762863159,
+      "mean_token_accuracy": 0.7547799497842789,
+      "num_tokens": 30770628.0,
+      "step": 403
+    },
+    {
+      "entropy": 0.6073088049888611,
+      "epoch": 5.318936877076412,
+      "grad_norm": 0.02284197509288788,
+      "learning_rate": 0.0002,
+      "loss": 0.6082872152328491,
+      "mean_token_accuracy": 0.7527147084474564,
+      "num_tokens": 30848582.0,
+      "step": 404
+    },
+    {
+      "entropy": 0.6121865510940552,
+      "epoch": 5.332225913621262,
+      "grad_norm": 0.020565764978528023,
+      "learning_rate": 0.0002,
+      "loss": 0.6121588945388794,
+      "mean_token_accuracy": 0.7523892968893051,
+      "num_tokens": 30924512.0,
+      "step": 405
+    },
+    {
+      "entropy": 0.6082987636327744,
+      "epoch": 5.3455149501661126,
+      "grad_norm": 0.022758690640330315,
+      "learning_rate": 0.0002,
+      "loss": 0.6041635274887085,
+      "mean_token_accuracy": 0.753796249628067,
+      "num_tokens": 31001346.0,
+      "step": 406
+    },
+    {
+      "entropy": 0.5950210988521576,
+      "epoch": 5.358803986710964,
+      "grad_norm": 0.0222301185131073,
+      "learning_rate": 0.0002,
+      "loss": 0.59589684009552,
+      "mean_token_accuracy": 0.7587986141443253,
+      "num_tokens": 31078724.0,
+      "step": 407
+    },
+    {
+      "entropy": 0.6093401461839676,
+      "epoch": 5.372093023255814,
+      "grad_norm": 0.02389850653707981,
+      "learning_rate": 0.0002,
+      "loss": 0.6132184863090515,
+      "mean_token_accuracy": 0.7504485845565796,
+      "num_tokens": 31155347.0,
+      "step": 408
+    },
+    {
+      "entropy": 0.608631044626236,
+      "epoch": 5.385382059800665,
+      "grad_norm": 0.022204812616109848,
+      "learning_rate": 0.0002,
+      "loss": 0.6059432029724121,
+      "mean_token_accuracy": 0.7542549520730972,
+      "num_tokens": 31231904.0,
+      "step": 409
+    },
+    {
+      "entropy": 0.6069070398807526,
+      "epoch": 5.398671096345515,
+      "grad_norm": 0.023435460403561592,
+      "learning_rate": 0.0002,
+      "loss": 0.6022323966026306,
+      "mean_token_accuracy": 0.7551652491092682,
+      "num_tokens": 31308737.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.5989623218774796,
+      "epoch": 5.411960132890365,
+      "grad_norm": 0.02185729704797268,
+      "learning_rate": 0.0002,
+      "loss": 0.6007922887802124,
+      "mean_token_accuracy": 0.7555037587881088,
+      "num_tokens": 31386653.0,
+      "step": 411
+    },
+    {
+      "entropy": 0.5904630720615387,
+      "epoch": 5.425249169435216,
+      "grad_norm": 0.022226441651582718,
+      "learning_rate": 0.0002,
+      "loss": 0.5928571224212646,
+      "mean_token_accuracy": 0.760121151804924,
+      "num_tokens": 31463650.0,
+      "step": 412
+    },
+    {
+      "entropy": 0.6058047413825989,
+      "epoch": 5.438538205980066,
+      "grad_norm": 0.023930639028549194,
+      "learning_rate": 0.0002,
+      "loss": 0.6058739423751831,
+      "mean_token_accuracy": 0.7543499618768692,
+      "num_tokens": 31540496.0,
+      "step": 413
+    },
+    {
+      "entropy": 0.6111002117395401,
+      "epoch": 5.451827242524917,
+      "grad_norm": 0.025340192019939423,
+      "learning_rate": 0.0002,
+      "loss": 0.6081179976463318,
+      "mean_token_accuracy": 0.7525225430727005,
+      "num_tokens": 31618008.0,
+      "step": 414
+    },
+    {
+      "entropy": 0.60789455473423,
+      "epoch": 5.465116279069767,
+      "grad_norm": 0.026184657588601112,
+      "learning_rate": 0.0002,
+      "loss": 0.6058934926986694,
+      "mean_token_accuracy": 0.7536182105541229,
+      "num_tokens": 31694894.0,
+      "step": 415
+    },
+    {
+      "entropy": 0.6068686991930008,
+      "epoch": 5.4784053156146175,
+      "grad_norm": 0.025053244084119797,
+      "learning_rate": 0.0002,
+      "loss": 0.6021319627761841,
+      "mean_token_accuracy": 0.7556831687688828,
+      "num_tokens": 31771491.0,
+      "step": 416
+    },
+    {
+      "entropy": 0.5977303683757782,
+      "epoch": 5.4916943521594686,
+      "grad_norm": 0.027724595740437508,
+      "learning_rate": 0.0002,
+      "loss": 0.5973233580589294,
+      "mean_token_accuracy": 0.7579130828380585,
+      "num_tokens": 31848344.0,
+      "step": 417
+    },
+    {
+      "entropy": 0.5988890677690506,
+      "epoch": 5.504983388704319,
+      "grad_norm": 0.024887511506676674,
+      "learning_rate": 0.0002,
+      "loss": 0.5990090370178223,
+      "mean_token_accuracy": 0.7568196505308151,
+      "num_tokens": 31924995.0,
+      "step": 418
+    },
+    {
+      "entropy": 0.604000598192215,
+      "epoch": 5.51827242524917,
+      "grad_norm": 0.022586684674024582,
+      "learning_rate": 0.0002,
+      "loss": 0.603076159954071,
+      "mean_token_accuracy": 0.7565865963697433,
+      "num_tokens": 32002254.0,
+      "step": 419
+    },
+    {
+      "entropy": 0.615940272808075,
+      "epoch": 5.53156146179402,
+      "grad_norm": 0.022485308349132538,
+      "learning_rate": 0.0002,
+      "loss": 0.6086373329162598,
+      "mean_token_accuracy": 0.7531540840864182,
+      "num_tokens": 32079285.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.6028707474470139,
+      "epoch": 5.544850498338871,
+      "grad_norm": 0.020232876762747765,
+      "learning_rate": 0.0002,
+      "loss": 0.601585865020752,
+      "mean_token_accuracy": 0.7566490918397903,
+      "num_tokens": 32156864.0,
+      "step": 421
+    },
+    {
+      "entropy": 0.6076253652572632,
+      "epoch": 5.558139534883721,
+      "grad_norm": 0.02617458626627922,
+      "learning_rate": 0.0002,
+      "loss": 0.6145195960998535,
+      "mean_token_accuracy": 0.7500647902488708,
+      "num_tokens": 32232846.0,
+      "step": 422
+    },
+    {
+      "entropy": 0.6028987616300583,
+      "epoch": 5.571428571428571,
+      "grad_norm": 0.02070811204612255,
+      "learning_rate": 0.0002,
+      "loss": 0.6059699654579163,
+      "mean_token_accuracy": 0.7536882907152176,
+      "num_tokens": 32310436.0,
+      "step": 423
+    },
+    {
+      "entropy": 0.5917141884565353,
+      "epoch": 5.584717607973422,
+      "grad_norm": 0.023316729813814163,
+      "learning_rate": 0.0002,
+      "loss": 0.5854599475860596,
+      "mean_token_accuracy": 0.7629671841859818,
+      "num_tokens": 32388971.0,
+      "step": 424
+    },
+    {
+      "entropy": 0.6023311167955399,
+      "epoch": 5.598006644518272,
+      "grad_norm": 0.020066890865564346,
+      "learning_rate": 0.0002,
+      "loss": 0.5990334749221802,
+      "mean_token_accuracy": 0.7573604434728622,
+      "num_tokens": 32467120.0,
+      "step": 425
+    },
+    {
+      "entropy": 0.5993306934833527,
+      "epoch": 5.6112956810631225,
+      "grad_norm": 0.023425690829753876,
+      "learning_rate": 0.0002,
+      "loss": 0.6018670797348022,
+      "mean_token_accuracy": 0.7574460357427597,
+      "num_tokens": 32543753.0,
+      "step": 426
+    },
+    {
+      "entropy": 0.6028042286634445,
+      "epoch": 5.6245847176079735,
+      "grad_norm": 0.02132745273411274,
+      "learning_rate": 0.0002,
+      "loss": 0.6041662693023682,
+      "mean_token_accuracy": 0.754668191075325,
+      "num_tokens": 32621427.0,
+      "step": 427
+    },
+    {
+      "entropy": 0.604213684797287,
+      "epoch": 5.637873754152824,
+      "grad_norm": 0.018084127455949783,
+      "learning_rate": 0.0002,
+      "loss": 0.6012581586837769,
+      "mean_token_accuracy": 0.7556634992361069,
+      "num_tokens": 32699682.0,
+      "step": 428
+    },
+    {
+      "entropy": 0.6124961525201797,
+      "epoch": 5.651162790697675,
+      "grad_norm": 0.02083444967865944,
+      "learning_rate": 0.0002,
+      "loss": 0.6100834608078003,
+      "mean_token_accuracy": 0.7531006336212158,
+      "num_tokens": 32775962.0,
+      "step": 429
+    },
+    {
+      "entropy": 0.5948987603187561,
+      "epoch": 5.664451827242525,
+      "grad_norm": 0.021068530157208443,
+      "learning_rate": 0.0002,
+      "loss": 0.5865616798400879,
+      "mean_token_accuracy": 0.7617778778076172,
+      "num_tokens": 32854001.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.5962036401033401,
+      "epoch": 5.677740863787376,
+      "grad_norm": 0.018948592245578766,
+      "learning_rate": 0.0002,
+      "loss": 0.5950455069541931,
+      "mean_token_accuracy": 0.7579799294471741,
+      "num_tokens": 32931336.0,
+      "step": 431
+    },
+    {
+      "entropy": 0.5971734374761581,
+      "epoch": 5.691029900332226,
+      "grad_norm": 0.020609233528375626,
+      "learning_rate": 0.0002,
+      "loss": 0.5992528796195984,
+      "mean_token_accuracy": 0.7560512125492096,
+      "num_tokens": 33007780.0,
+      "step": 432
+    },
+    {
+      "entropy": 0.600962370634079,
+      "epoch": 5.704318936877076,
+      "grad_norm": 0.02043914794921875,
+      "learning_rate": 0.0002,
+      "loss": 0.6053571701049805,
+      "mean_token_accuracy": 0.7534149289131165,
+      "num_tokens": 33083865.0,
+      "step": 433
+    },
+    {
+      "entropy": 0.6037783920764923,
+      "epoch": 5.717607973421927,
+      "grad_norm": 0.02007639966905117,
+      "learning_rate": 0.0002,
+      "loss": 0.6004524230957031,
+      "mean_token_accuracy": 0.7554492056369781,
+      "num_tokens": 33160697.0,
+      "step": 434
+    },
+    {
+      "entropy": 0.6123286634683609,
+      "epoch": 5.730897009966777,
+      "grad_norm": 0.022184055298566818,
+      "learning_rate": 0.0002,
+      "loss": 0.6058014631271362,
+      "mean_token_accuracy": 0.7532261461019516,
+      "num_tokens": 33237480.0,
+      "step": 435
+    },
+    {
+      "entropy": 0.6141447126865387,
+      "epoch": 5.7441860465116275,
+      "grad_norm": 0.02083767205476761,
+      "learning_rate": 0.0002,
+      "loss": 0.610468864440918,
+      "mean_token_accuracy": 0.7515555173158646,
+      "num_tokens": 33314013.0,
+      "step": 436
+    },
+    {
+      "entropy": 0.6035348027944565,
+      "epoch": 5.7574750830564785,
+      "grad_norm": 0.024816464632749557,
+      "learning_rate": 0.0002,
+      "loss": 0.6087946891784668,
+      "mean_token_accuracy": 0.754722535610199,
+      "num_tokens": 33390218.0,
+      "step": 437
+    },
+    {
+      "entropy": 0.5946770757436752,
+      "epoch": 5.770764119601329,
+      "grad_norm": 0.022571545094251633,
+      "learning_rate": 0.0002,
+      "loss": 0.596878170967102,
+      "mean_token_accuracy": 0.7579701691865921,
+      "num_tokens": 33468134.0,
+      "step": 438
+    },
+    {
+      "entropy": 0.6164768636226654,
+      "epoch": 5.78405315614618,
+      "grad_norm": 0.024533748626708984,
+      "learning_rate": 0.0002,
+      "loss": 0.6126947402954102,
+      "mean_token_accuracy": 0.7517720609903336,
+      "num_tokens": 33543916.0,
+      "step": 439
+    },
+    {
+      "entropy": 0.6056448072195053,
+      "epoch": 5.79734219269103,
+      "grad_norm": 0.020430127158761024,
+      "learning_rate": 0.0002,
+      "loss": 0.6013854742050171,
+      "mean_token_accuracy": 0.756991907954216,
+      "num_tokens": 33620951.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.5991196930408478,
+      "epoch": 5.810631229235881,
+      "grad_norm": 0.021590569987893105,
+      "learning_rate": 0.0002,
+      "loss": 0.5991401672363281,
+      "mean_token_accuracy": 0.7567178606987,
+      "num_tokens": 33698477.0,
+      "step": 441
+    },
+    {
+      "entropy": 0.5983557999134064,
+      "epoch": 5.823920265780731,
+      "grad_norm": 0.02526337280869484,
+      "learning_rate": 0.0002,
+      "loss": 0.6028835773468018,
+      "mean_token_accuracy": 0.7568080127239227,
+      "num_tokens": 33776296.0,
+      "step": 442
+    },
+    {
+      "entropy": 0.6031540036201477,
+      "epoch": 5.837209302325581,
+      "grad_norm": 0.02267344854772091,
+      "learning_rate": 0.0002,
+      "loss": 0.6034103631973267,
+      "mean_token_accuracy": 0.754448652267456,
+      "num_tokens": 33853842.0,
+      "step": 443
+    },
+    {
+      "entropy": 0.6104815006256104,
+      "epoch": 5.850498338870432,
+      "grad_norm": 0.020782535895705223,
+      "learning_rate": 0.0002,
+      "loss": 0.6049699783325195,
+      "mean_token_accuracy": 0.7535883188247681,
+      "num_tokens": 33930886.0,
+      "step": 444
+    },
+    {
+      "entropy": 0.5981206297874451,
+      "epoch": 5.863787375415282,
+      "grad_norm": 0.023540036752820015,
+      "learning_rate": 0.0002,
+      "loss": 0.5923410058021545,
+      "mean_token_accuracy": 0.7591070085763931,
+      "num_tokens": 34007350.0,
+      "step": 445
+    },
+    {
+      "entropy": 0.6036737859249115,
+      "epoch": 5.877076411960132,
+      "grad_norm": 0.02499704249203205,
+      "learning_rate": 0.0002,
+      "loss": 0.6046547889709473,
+      "mean_token_accuracy": 0.7538949996232986,
+      "num_tokens": 34084012.0,
+      "step": 446
+    },
+    {
+      "entropy": 0.5917275696992874,
+      "epoch": 5.8903654485049834,
+      "grad_norm": 0.025873389095067978,
+      "learning_rate": 0.0002,
+      "loss": 0.5952780246734619,
+      "mean_token_accuracy": 0.7577290534973145,
+      "num_tokens": 34160998.0,
+      "step": 447
+    },
+    {
+      "entropy": 0.6122117638587952,
+      "epoch": 5.903654485049834,
+      "grad_norm": 0.028246693313121796,
+      "learning_rate": 0.0002,
+      "loss": 0.6137705445289612,
+      "mean_token_accuracy": 0.7516694515943527,
+      "num_tokens": 34238302.0,
+      "step": 448
+    },
+    {
+      "entropy": 0.6051677614450455,
+      "epoch": 5.916943521594685,
+      "grad_norm": 0.031234685331583023,
+      "learning_rate": 0.0002,
+      "loss": 0.5974487066268921,
+      "mean_token_accuracy": 0.7570804208517075,
+      "num_tokens": 34315312.0,
+      "step": 449
+    },
+    {
+      "entropy": 0.6137758642435074,
+      "epoch": 5.930232558139535,
+      "grad_norm": 0.03217904269695282,
+      "learning_rate": 0.0002,
+      "loss": 0.6115902066230774,
+      "mean_token_accuracy": 0.752965971827507,
+      "num_tokens": 34392319.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.6072733849287033,
+      "epoch": 5.943521594684386,
+      "grad_norm": 0.03081611543893814,
+      "learning_rate": 0.0002,
+      "loss": 0.6136135458946228,
+      "mean_token_accuracy": 0.7503504902124405,
+      "num_tokens": 34468277.0,
+      "step": 451
+    },
+    {
+      "entropy": 0.6001786440610886,
+      "epoch": 5.956810631229236,
+      "grad_norm": 0.02687067538499832,
+      "learning_rate": 0.0002,
+      "loss": 0.5990195274353027,
+      "mean_token_accuracy": 0.7577933222055435,
+      "num_tokens": 34545839.0,
+      "step": 452
+    },
+    {
+      "entropy": 0.612625390291214,
+      "epoch": 5.970099667774086,
+      "grad_norm": 0.03187267854809761,
+      "learning_rate": 0.0002,
+      "loss": 0.6082343459129333,
+      "mean_token_accuracy": 0.7538764029741287,
+      "num_tokens": 34623128.0,
+      "step": 453
+    },
+    {
+      "entropy": 0.6080211997032166,
+      "epoch": 5.983388704318937,
+      "grad_norm": 0.028988594189286232,
+      "learning_rate": 0.0002,
+      "loss": 0.6047749519348145,
+      "mean_token_accuracy": 0.7554045468568802,
+      "num_tokens": 34700832.0,
+      "step": 454
+    },
+    {
+      "entropy": 0.5958880931138992,
+      "epoch": 5.996677740863787,
+      "grad_norm": 0.027844693511724472,
+      "learning_rate": 0.0002,
+      "loss": 0.5956987142562866,
+      "mean_token_accuracy": 0.7574048638343811,
+      "num_tokens": 34777605.0,
+      "step": 455
+    },
+    {
+      "entropy": 0.5998430252075195,
+      "epoch": 6.0,
+      "grad_norm": 0.04323003441095352,
+      "learning_rate": 0.0002,
+      "loss": 0.6032893657684326,
+      "mean_token_accuracy": 0.7568454146385193,
+      "num_tokens": 34796943.0,
+      "step": 456
+    },
+    {
+      "entropy": 0.6120886653661728,
+      "epoch": 6.01328903654485,
+      "grad_norm": 0.025641195476055145,
+      "learning_rate": 0.0002,
+      "loss": 0.6044338345527649,
+      "mean_token_accuracy": 0.7545018643140793,
+      "num_tokens": 34873973.0,
+      "step": 457
+    },
+    {
+      "entropy": 0.600526973605156,
+      "epoch": 6.026578073089701,
+      "grad_norm": 0.027644747868180275,
+      "learning_rate": 0.0002,
+      "loss": 0.5954704284667969,
+      "mean_token_accuracy": 0.7591830790042877,
+      "num_tokens": 34950550.0,
+      "step": 458
+    },
+    {
+      "entropy": 0.5978350341320038,
+      "epoch": 6.039867109634551,
+      "grad_norm": 0.022663088515400887,
+      "learning_rate": 0.0002,
+      "loss": 0.5966336727142334,
+      "mean_token_accuracy": 0.7568895518779755,
+      "num_tokens": 35027626.0,
+      "step": 459
+    },
+    {
+      "entropy": 0.5985796898603439,
+      "epoch": 6.053156146179402,
+      "grad_norm": 0.027262212708592415,
+      "learning_rate": 0.0002,
+      "loss": 0.6055795550346375,
+      "mean_token_accuracy": 0.7541183084249496,
+      "num_tokens": 35104160.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.606357529759407,
+      "epoch": 6.0664451827242525,
+      "grad_norm": 0.021243030205368996,
+      "learning_rate": 0.0002,
+      "loss": 0.6038681268692017,
+      "mean_token_accuracy": 0.7561346888542175,
+      "num_tokens": 35180440.0,
+      "step": 461
+    },
+    {
+      "entropy": 0.5993641018867493,
+      "epoch": 6.079734219269103,
+      "grad_norm": 0.02454681135714054,
+      "learning_rate": 0.0002,
+      "loss": 0.5979468822479248,
+      "mean_token_accuracy": 0.7567647844552994,
+      "num_tokens": 35257736.0,
+      "step": 462
+    },
+    {
+      "entropy": 0.5848215073347092,
+      "epoch": 6.093023255813954,
+      "grad_norm": 0.022167058661580086,
+      "learning_rate": 0.0002,
+      "loss": 0.585566520690918,
+      "mean_token_accuracy": 0.762741819024086,
+      "num_tokens": 35335235.0,
+      "step": 463
+    },
+    {
+      "entropy": 0.5985981374979019,
+      "epoch": 6.106312292358804,
+      "grad_norm": 0.024154378101229668,
+      "learning_rate": 0.0002,
+      "loss": 0.6016069650650024,
+      "mean_token_accuracy": 0.7554454803466797,
+      "num_tokens": 35412861.0,
+      "step": 464
+    },
+    {
+      "entropy": 0.6006908863782883,
+      "epoch": 6.119601328903655,
+      "grad_norm": 0.022460851818323135,
+      "learning_rate": 0.0002,
+      "loss": 0.6020187139511108,
+      "mean_token_accuracy": 0.7560145705938339,
+      "num_tokens": 35489552.0,
+      "step": 465
+    },
+    {
+      "entropy": 0.5931369364261627,
+      "epoch": 6.132890365448505,
+      "grad_norm": 0.02682335674762726,
+      "learning_rate": 0.0002,
+      "loss": 0.5886173248291016,
+      "mean_token_accuracy": 0.7600135207176208,
+      "num_tokens": 35566976.0,
+      "step": 466
+    },
+    {
+      "entropy": 0.606135219335556,
+      "epoch": 6.146179401993355,
+      "grad_norm": 0.022801680490374565,
+      "learning_rate": 0.0002,
+      "loss": 0.6008643507957458,
+      "mean_token_accuracy": 0.7537801265716553,
+      "num_tokens": 35644809.0,
+      "step": 467
+    },
+    {
+      "entropy": 0.5942219793796539,
+      "epoch": 6.159468438538206,
+      "grad_norm": 0.023526513949036598,
+      "learning_rate": 0.0002,
+      "loss": 0.5949859619140625,
+      "mean_token_accuracy": 0.7583627551794052,
+      "num_tokens": 35722320.0,
+      "step": 468
+    },
+    {
+      "entropy": 0.600348949432373,
+      "epoch": 6.172757475083056,
+      "grad_norm": 0.02136663906276226,
+      "learning_rate": 0.0002,
+      "loss": 0.6002007722854614,
+      "mean_token_accuracy": 0.7563091218471527,
+      "num_tokens": 35799676.0,
+      "step": 469
+    },
+    {
+      "entropy": 0.5929514020681381,
+      "epoch": 6.186046511627907,
+      "grad_norm": 0.020419945940375328,
+      "learning_rate": 0.0002,
+      "loss": 0.5907375812530518,
+      "mean_token_accuracy": 0.758481040596962,
+      "num_tokens": 35876561.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6010562926530838,
+      "epoch": 6.1993355481727574,
+      "grad_norm": 0.021938690915703773,
+      "learning_rate": 0.0002,
+      "loss": 0.596697211265564,
+      "mean_token_accuracy": 0.7577521502971649,
+      "num_tokens": 35954003.0,
+      "step": 471
+    },
+    {
+      "entropy": 0.5978391021490097,
+      "epoch": 6.212624584717608,
+      "grad_norm": 0.02111574076116085,
+      "learning_rate": 0.0002,
+      "loss": 0.5977692008018494,
+      "mean_token_accuracy": 0.7567316740751266,
+      "num_tokens": 36030974.0,
+      "step": 472
+    },
+    {
+      "entropy": 0.6060666143894196,
+      "epoch": 6.225913621262459,
+      "grad_norm": 0.020165806636214256,
+      "learning_rate": 0.0002,
+      "loss": 0.6070280075073242,
+      "mean_token_accuracy": 0.753100574016571,
+      "num_tokens": 36108601.0,
+      "step": 473
+    },
+    {
+      "entropy": 0.5992994010448456,
+      "epoch": 6.239202657807309,
+      "grad_norm": 0.020219672471284866,
+      "learning_rate": 0.0002,
+      "loss": 0.5985912680625916,
+      "mean_token_accuracy": 0.7557387799024582,
+      "num_tokens": 36185763.0,
+      "step": 474
+    },
+    {
+      "entropy": 0.6000545620918274,
+      "epoch": 6.25249169435216,
+      "grad_norm": 0.02075883187353611,
+      "learning_rate": 0.0002,
+      "loss": 0.6000256538391113,
+      "mean_token_accuracy": 0.756579726934433,
+      "num_tokens": 36262158.0,
+      "step": 475
+    },
+    {
+      "entropy": 0.5993035733699799,
+      "epoch": 6.26578073089701,
+      "grad_norm": 0.021321367472410202,
+      "learning_rate": 0.0002,
+      "loss": 0.59999680519104,
+      "mean_token_accuracy": 0.7556854337453842,
+      "num_tokens": 36338496.0,
+      "step": 476
+    },
+    {
+      "entropy": 0.5979324728250504,
+      "epoch": 6.27906976744186,
+      "grad_norm": 0.024508945643901825,
+      "learning_rate": 0.0002,
+      "loss": 0.5982510447502136,
+      "mean_token_accuracy": 0.7566314190626144,
+      "num_tokens": 36415527.0,
+      "step": 477
+    },
+    {
+      "entropy": 0.6013223379850388,
+      "epoch": 6.292358803986711,
+      "grad_norm": 0.02557053230702877,
+      "learning_rate": 0.0002,
+      "loss": 0.5949776768684387,
+      "mean_token_accuracy": 0.7575513571500778,
+      "num_tokens": 36492514.0,
+      "step": 478
+    },
+    {
+      "entropy": 0.5949330031871796,
+      "epoch": 6.305647840531561,
+      "grad_norm": 0.02301526628434658,
+      "learning_rate": 0.0002,
+      "loss": 0.5910919904708862,
+      "mean_token_accuracy": 0.7598573863506317,
+      "num_tokens": 36570074.0,
+      "step": 479
+    },
+    {
+      "entropy": 0.5883470773696899,
+      "epoch": 6.318936877076412,
+      "grad_norm": 0.025473203510046005,
+      "learning_rate": 0.0002,
+      "loss": 0.5894684791564941,
+      "mean_token_accuracy": 0.758475586771965,
+      "num_tokens": 36646855.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.6115803718566895,
+      "epoch": 6.332225913621262,
+      "grad_norm": 0.027968522161245346,
+      "learning_rate": 0.0002,
+      "loss": 0.6107940673828125,
+      "mean_token_accuracy": 0.751880869269371,
+      "num_tokens": 36723088.0,
+      "step": 481
+    },
+    {
+      "entropy": 0.5995512008666992,
+      "epoch": 6.3455149501661126,
+      "grad_norm": 0.023139847442507744,
+      "learning_rate": 0.0002,
+      "loss": 0.5955934524536133,
+      "mean_token_accuracy": 0.758063018321991,
+      "num_tokens": 36800377.0,
+      "step": 482
+    },
+    {
+      "entropy": 0.6044798940420151,
+      "epoch": 6.358803986710964,
+      "grad_norm": 0.024419818073511124,
+      "learning_rate": 0.0002,
+      "loss": 0.6051414608955383,
+      "mean_token_accuracy": 0.7533292770385742,
+      "num_tokens": 36876791.0,
+      "step": 483
+    },
+    {
+      "entropy": 0.590464860200882,
+      "epoch": 6.372093023255814,
+      "grad_norm": 0.024506734684109688,
+      "learning_rate": 0.0002,
+      "loss": 0.5884811282157898,
+      "mean_token_accuracy": 0.7608850747346878,
+      "num_tokens": 36955168.0,
+      "step": 484
+    },
+    {
+      "entropy": 0.5885339975357056,
+      "epoch": 6.385382059800665,
+      "grad_norm": 0.02895374596118927,
+      "learning_rate": 0.0002,
+      "loss": 0.5923447608947754,
+      "mean_token_accuracy": 0.7594189345836639,
+      "num_tokens": 37032024.0,
+      "step": 485
+    },
+    {
+      "entropy": 0.5949011445045471,
+      "epoch": 6.398671096345515,
+      "grad_norm": 0.02935759536921978,
+      "learning_rate": 0.0002,
+      "loss": 0.5962386727333069,
+      "mean_token_accuracy": 0.7577309012413025,
+      "num_tokens": 37108607.0,
+      "step": 486
+    },
+    {
+      "entropy": 0.606280654668808,
+      "epoch": 6.411960132890365,
+      "grad_norm": 0.027602598071098328,
+      "learning_rate": 0.0002,
+      "loss": 0.6015469431877136,
+      "mean_token_accuracy": 0.7563475072383881,
+      "num_tokens": 37185963.0,
+      "step": 487
+    },
+    {
+      "entropy": 0.6054241806268692,
+      "epoch": 6.425249169435216,
+      "grad_norm": 0.03342454135417938,
+      "learning_rate": 0.0002,
+      "loss": 0.6038094758987427,
+      "mean_token_accuracy": 0.7535726577043533,
+      "num_tokens": 37262688.0,
+      "step": 488
+    },
+    {
+      "entropy": 0.6115904003381729,
+      "epoch": 6.438538205980066,
+      "grad_norm": 0.024562479928135872,
+      "learning_rate": 0.0002,
+      "loss": 0.6105020046234131,
+      "mean_token_accuracy": 0.752171978354454,
+      "num_tokens": 37339918.0,
+      "step": 489
+    },
+    {
+      "entropy": 0.6080169379711151,
+      "epoch": 6.451827242524917,
+      "grad_norm": 0.033320341259241104,
+      "learning_rate": 0.0002,
+      "loss": 0.6115527153015137,
+      "mean_token_accuracy": 0.7520752102136612,
+      "num_tokens": 37416200.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.6040970087051392,
+      "epoch": 6.465116279069767,
+      "grad_norm": 0.02561773732304573,
+      "learning_rate": 0.0002,
+      "loss": 0.5990139842033386,
+      "mean_token_accuracy": 0.7566020637750626,
+      "num_tokens": 37494068.0,
+      "step": 491
+    },
+    {
+      "entropy": 0.5961304754018784,
+      "epoch": 6.4784053156146175,
+      "grad_norm": 0.030982809141278267,
+      "learning_rate": 0.0002,
+      "loss": 0.5901075601577759,
+      "mean_token_accuracy": 0.7607396245002747,
+      "num_tokens": 37571533.0,
+      "step": 492
+    },
+    {
+      "entropy": 0.5945215076208115,
+      "epoch": 6.4916943521594686,
+      "grad_norm": 0.024695293977856636,
+      "learning_rate": 0.0002,
+      "loss": 0.5971148014068604,
+      "mean_token_accuracy": 0.7563796937465668,
+      "num_tokens": 37648973.0,
+      "step": 493
+    },
+    {
+      "entropy": 0.5900432914495468,
+      "epoch": 6.504983388704319,
+      "grad_norm": 0.03201661631464958,
+      "learning_rate": 0.0002,
+      "loss": 0.5974127054214478,
+      "mean_token_accuracy": 0.7566387355327606,
+      "num_tokens": 37726459.0,
+      "step": 494
+    },
+    {
+      "entropy": 0.6024730056524277,
+      "epoch": 6.51827242524917,
+      "grad_norm": 0.024753453209996223,
+      "learning_rate": 0.0002,
+      "loss": 0.60154789686203,
+      "mean_token_accuracy": 0.7547302693128586,
+      "num_tokens": 37803467.0,
+      "step": 495
+    },
+    {
+      "entropy": 0.6144786328077316,
+      "epoch": 6.53156146179402,
+      "grad_norm": 0.028725173324346542,
+      "learning_rate": 0.0002,
+      "loss": 0.6057987213134766,
+      "mean_token_accuracy": 0.7548320293426514,
+      "num_tokens": 37880199.0,
+      "step": 496
+    },
+    {
+      "entropy": 0.6062168329954147,
+      "epoch": 6.544850498338871,
+      "grad_norm": 0.02327624335885048,
+      "learning_rate": 0.0002,
+      "loss": 0.6026209592819214,
+      "mean_token_accuracy": 0.7539927959442139,
+      "num_tokens": 37957412.0,
+      "step": 497
+    },
+    {
+      "entropy": 0.6021516770124435,
+      "epoch": 6.558139534883721,
+      "grad_norm": 0.03300193324685097,
+      "learning_rate": 0.0002,
+      "loss": 0.6078506112098694,
+      "mean_token_accuracy": 0.7522564381361008,
+      "num_tokens": 38034181.0,
+      "step": 498
+    },
+    {
+      "entropy": 0.5943173170089722,
+      "epoch": 6.571428571428571,
+      "grad_norm": 0.021800996735692024,
+      "learning_rate": 0.0002,
+      "loss": 0.5896530747413635,
+      "mean_token_accuracy": 0.7614179700613022,
+      "num_tokens": 38112342.0,
+      "step": 499
+    },
+    {
+      "entropy": 0.6014736741781235,
+      "epoch": 6.584717607973422,
+      "grad_norm": 0.029887963086366653,
+      "learning_rate": 0.0002,
+      "loss": 0.5958458185195923,
+      "mean_token_accuracy": 0.7576052695512772,
+      "num_tokens": 38189802.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.6024068295955658,
+      "epoch": 6.598006644518272,
+      "grad_norm": 0.025521252304315567,
+      "learning_rate": 0.0002,
+      "loss": 0.6060903072357178,
+      "mean_token_accuracy": 0.753528967499733,
+      "num_tokens": 38266306.0,
+      "step": 501
+    },
+    {
+      "entropy": 0.5893438309431076,
+      "epoch": 6.6112956810631225,
+      "grad_norm": 0.0318656824529171,
+      "learning_rate": 0.0002,
+      "loss": 0.5963033437728882,
+      "mean_token_accuracy": 0.7568179666996002,
+      "num_tokens": 38343848.0,
+      "step": 502
+    },
+    {
+      "entropy": 0.6008411794900894,
+      "epoch": 6.6245847176079735,
+      "grad_norm": 0.024844203144311905,
+      "learning_rate": 0.0002,
+      "loss": 0.600379467010498,
+      "mean_token_accuracy": 0.7562743723392487,
+      "num_tokens": 38421065.0,
+      "step": 503
+    },
+    {
+      "entropy": 0.6024555712938309,
+      "epoch": 6.637873754152824,
+      "grad_norm": 0.02915586344897747,
+      "learning_rate": 0.0002,
+      "loss": 0.5952292084693909,
+      "mean_token_accuracy": 0.7569020986557007,
+      "num_tokens": 38497792.0,
+      "step": 504
+    },
+    {
+      "entropy": 0.6041858941316605,
+      "epoch": 6.651162790697675,
+      "grad_norm": 0.027332143858075142,
+      "learning_rate": 0.0002,
+      "loss": 0.5974034070968628,
+      "mean_token_accuracy": 0.7567848265171051,
+      "num_tokens": 38574161.0,
+      "step": 505
+    },
+    {
+      "entropy": 0.6042163372039795,
+      "epoch": 6.664451827242525,
+      "grad_norm": 0.027624888345599174,
+      "learning_rate": 0.0002,
+      "loss": 0.6048606634140015,
+      "mean_token_accuracy": 0.7530855089426041,
+      "num_tokens": 38650349.0,
+      "step": 506
+    },
+    {
+      "entropy": 0.59311243891716,
+      "epoch": 6.677740863787376,
+      "grad_norm": 0.028052296489477158,
+      "learning_rate": 0.0002,
+      "loss": 0.5949767827987671,
+      "mean_token_accuracy": 0.7584637403488159,
+      "num_tokens": 38727788.0,
+      "step": 507
+    },
+    {
+      "entropy": 0.5970088392496109,
+      "epoch": 6.691029900332226,
+      "grad_norm": 0.027488745748996735,
+      "learning_rate": 0.0002,
+      "loss": 0.5996758937835693,
+      "mean_token_accuracy": 0.7576368153095245,
+      "num_tokens": 38804821.0,
+      "step": 508
+    },
+    {
+      "entropy": 0.598954051733017,
+      "epoch": 6.704318936877076,
+      "grad_norm": 0.020494971424341202,
+      "learning_rate": 0.0002,
+      "loss": 0.5968008637428284,
+      "mean_token_accuracy": 0.7577658295631409,
+      "num_tokens": 38882149.0,
+      "step": 509
+    },
+    {
+      "entropy": 0.6018869280815125,
+      "epoch": 6.717607973421927,
+      "grad_norm": 0.025988485664129257,
+      "learning_rate": 0.0002,
+      "loss": 0.5987365245819092,
+      "mean_token_accuracy": 0.7569222897291183,
+      "num_tokens": 38958745.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.5928902477025986,
+      "epoch": 6.730897009966777,
+      "grad_norm": 0.024826789274811745,
+      "learning_rate": 0.0002,
+      "loss": 0.5931652784347534,
+      "mean_token_accuracy": 0.760022446513176,
+      "num_tokens": 39036088.0,
+      "step": 511
+    },
+    {
+      "entropy": 0.600498303771019,
+      "epoch": 6.7441860465116275,
+      "grad_norm": 0.02464020438492298,
+      "learning_rate": 0.0002,
+      "loss": 0.5990219116210938,
+      "mean_token_accuracy": 0.7561146169900894,
+      "num_tokens": 39112787.0,
+      "step": 512
+    },
+    {
+      "entropy": 0.5943007320165634,
+      "epoch": 6.7574750830564785,
+      "grad_norm": 0.022298917174339294,
+      "learning_rate": 0.0002,
+      "loss": 0.5947438478469849,
+      "mean_token_accuracy": 0.7584742158651352,
+      "num_tokens": 39189290.0,
+      "step": 513
+    },
+    {
+      "entropy": 0.6023274958133698,
+      "epoch": 6.770764119601329,
+      "grad_norm": 0.02406785823404789,
+      "learning_rate": 0.0002,
+      "loss": 0.601665735244751,
+      "mean_token_accuracy": 0.754782423377037,
+      "num_tokens": 39266176.0,
+      "step": 514
+    },
+    {
+      "entropy": 0.6048448234796524,
+      "epoch": 6.78405315614618,
+      "grad_norm": 0.026156462728977203,
+      "learning_rate": 0.0002,
+      "loss": 0.6030054092407227,
+      "mean_token_accuracy": 0.7556321769952774,
+      "num_tokens": 39344181.0,
+      "step": 515
+    },
+    {
+      "entropy": 0.6049788296222687,
+      "epoch": 6.79734219269103,
+      "grad_norm": 0.02489808201789856,
+      "learning_rate": 0.0002,
+      "loss": 0.6032531261444092,
+      "mean_token_accuracy": 0.7542305439710617,
+      "num_tokens": 39421015.0,
+      "step": 516
+    },
+    {
+      "entropy": 0.597603052854538,
+      "epoch": 6.810631229235881,
+      "grad_norm": 0.023581495508551598,
+      "learning_rate": 0.0002,
+      "loss": 0.5995160341262817,
+      "mean_token_accuracy": 0.7571886032819748,
+      "num_tokens": 39497442.0,
+      "step": 517
+    },
+    {
+      "entropy": 0.6044445037841797,
+      "epoch": 6.823920265780731,
+      "grad_norm": 0.02690713107585907,
+      "learning_rate": 0.0002,
+      "loss": 0.6078607439994812,
+      "mean_token_accuracy": 0.7533780336380005,
+      "num_tokens": 39573687.0,
+      "step": 518
+    },
+    {
+      "entropy": 0.5997501760721207,
+      "epoch": 6.837209302325581,
+      "grad_norm": 0.025829756632447243,
+      "learning_rate": 0.0002,
+      "loss": 0.5987580418586731,
+      "mean_token_accuracy": 0.7570158541202545,
+      "num_tokens": 39651340.0,
+      "step": 519
+    },
+    {
+      "entropy": 0.6061921864748001,
+      "epoch": 6.850498338870432,
+      "grad_norm": 0.0223165825009346,
+      "learning_rate": 0.0002,
+      "loss": 0.6045077443122864,
+      "mean_token_accuracy": 0.7532154619693756,
+      "num_tokens": 39728030.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.5982785224914551,
+      "epoch": 6.863787375415282,
+      "grad_norm": 0.028076212853193283,
+      "learning_rate": 0.0002,
+      "loss": 0.5973300337791443,
+      "mean_token_accuracy": 0.7568475604057312,
+      "num_tokens": 39805292.0,
+      "step": 521
+    },
+    {
+      "entropy": 0.595564216375351,
+      "epoch": 6.877076411960132,
+      "grad_norm": 0.02324676886200905,
+      "learning_rate": 0.0002,
+      "loss": 0.5942668914794922,
+      "mean_token_accuracy": 0.7570454329252243,
+      "num_tokens": 39882311.0,
+      "step": 522
+    },
+    {
+      "entropy": 0.5887253731489182,
+      "epoch": 6.8903654485049834,
+      "grad_norm": 0.03060692921280861,
+      "learning_rate": 0.0002,
+      "loss": 0.592413067817688,
+      "mean_token_accuracy": 0.7588198632001877,
+      "num_tokens": 39959797.0,
+      "step": 523
+    },
+    {
+      "entropy": 0.6044618636369705,
+      "epoch": 6.903654485049834,
+      "grad_norm": 0.023472849279642105,
+      "learning_rate": 0.0002,
+      "loss": 0.6028954982757568,
+      "mean_token_accuracy": 0.7550192028284073,
+      "num_tokens": 40036861.0,
+      "step": 524
+    },
+    {
+      "entropy": 0.6093322485685349,
+      "epoch": 6.916943521594685,
+      "grad_norm": 0.02798568457365036,
+      "learning_rate": 0.0002,
+      "loss": 0.6010950207710266,
+      "mean_token_accuracy": 0.756156861782074,
+      "num_tokens": 40114416.0,
+      "step": 525
+    },
+    {
+      "entropy": 0.608117938041687,
+      "epoch": 6.930232558139535,
+      "grad_norm": 0.026745513081550598,
+      "learning_rate": 0.0002,
+      "loss": 0.6025017499923706,
+      "mean_token_accuracy": 0.7551372200250626,
+      "num_tokens": 40191275.0,
+      "step": 526
+    },
+    {
+      "entropy": 0.5973224937915802,
+      "epoch": 6.943521594684386,
+      "grad_norm": 0.02841629832983017,
+      "learning_rate": 0.0002,
+      "loss": 0.6008439064025879,
+      "mean_token_accuracy": 0.7560015767812729,
+      "num_tokens": 40268155.0,
+      "step": 527
+    },
+    {
+      "entropy": 0.5969992578029633,
+      "epoch": 6.956810631229236,
+      "grad_norm": 0.038494374603033066,
+      "learning_rate": 0.0002,
+      "loss": 0.6079155206680298,
+      "mean_token_accuracy": 0.7526611685752869,
+      "num_tokens": 40344778.0,
+      "step": 528
+    },
+    {
+      "entropy": 0.6128715723752975,
+      "epoch": 6.970099667774086,
+      "grad_norm": 0.022489842027425766,
+      "learning_rate": 0.0002,
+      "loss": 0.6073602437973022,
+      "mean_token_accuracy": 0.7529550343751907,
+      "num_tokens": 40421973.0,
+      "step": 529
+    },
+    {
+      "entropy": 0.610251396894455,
+      "epoch": 6.983388704318937,
+      "grad_norm": 0.033306654542684555,
+      "learning_rate": 0.0002,
+      "loss": 0.5994991064071655,
+      "mean_token_accuracy": 0.7565908879041672,
+      "num_tokens": 40499293.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.5960075706243515,
+      "epoch": 6.996677740863787,
+      "grad_norm": 0.019969960674643517,
+      "learning_rate": 0.0002,
+      "loss": 0.5940055847167969,
+      "mean_token_accuracy": 0.7587366998195648,
+      "num_tokens": 40577102.0,
+      "step": 531
+    },
+    {
+      "entropy": 0.6263917684555054,
+      "epoch": 7.0,
+      "grad_norm": 0.04845428094267845,
+      "learning_rate": 0.0002,
+      "loss": 0.6320151090621948,
+      "mean_token_accuracy": 0.7457926273345947,
+      "num_tokens": 40596482.0,
+      "step": 532
+    },
+    {
+      "entropy": 0.60090272128582,
+      "epoch": 7.01328903654485,
+      "grad_norm": 0.020749559625983238,
+      "learning_rate": 0.0002,
+      "loss": 0.5972373485565186,
+      "mean_token_accuracy": 0.7571604698896408,
+      "num_tokens": 40674334.0,
+      "step": 533
+    },
+    {
+      "entropy": 0.6017645597457886,
+      "epoch": 7.026578073089701,
+      "grad_norm": 0.030549099668860435,
+      "learning_rate": 0.0002,
+      "loss": 0.5974453687667847,
+      "mean_token_accuracy": 0.7568349391222,
+      "num_tokens": 40751117.0,
+      "step": 534
+    },
+    {
+      "entropy": 0.5874526649713516,
+      "epoch": 7.039867109634551,
+      "grad_norm": 0.02841581031680107,
+      "learning_rate": 0.0002,
+      "loss": 0.5857692360877991,
+      "mean_token_accuracy": 0.7610917538404465,
+      "num_tokens": 40827603.0,
+      "step": 535
+    },
+    {
+      "entropy": 0.5865802317857742,
+      "epoch": 7.053156146179402,
+      "grad_norm": 0.036679480224847794,
+      "learning_rate": 0.0002,
+      "loss": 0.5936608910560608,
+      "mean_token_accuracy": 0.7579147815704346,
+      "num_tokens": 40904279.0,
+      "step": 536
+    },
+    {
+      "entropy": 0.5945817977190018,
+      "epoch": 7.0664451827242525,
+      "grad_norm": 0.026053892448544502,
+      "learning_rate": 0.0002,
+      "loss": 0.5957525968551636,
+      "mean_token_accuracy": 0.7568493187427521,
+      "num_tokens": 40981735.0,
+      "step": 537
+    },
+    {
+      "entropy": 0.6033788025379181,
+      "epoch": 7.079734219269103,
+      "grad_norm": 0.02980688028037548,
+      "learning_rate": 0.0002,
+      "loss": 0.5996099710464478,
+      "mean_token_accuracy": 0.7563089579343796,
+      "num_tokens": 41058383.0,
+      "step": 538
+    },
+    {
+      "entropy": 0.5973725020885468,
+      "epoch": 7.093023255813954,
+      "grad_norm": 0.029874414205551147,
+      "learning_rate": 0.0002,
+      "loss": 0.5903159976005554,
+      "mean_token_accuracy": 0.7587748616933823,
+      "num_tokens": 41135909.0,
+      "step": 539
+    },
+    {
+      "entropy": 0.6045922935009003,
+      "epoch": 7.106312292358804,
+      "grad_norm": 0.033622536808252335,
+      "learning_rate": 0.0002,
+      "loss": 0.6066581606864929,
+      "mean_token_accuracy": 0.7533113658428192,
+      "num_tokens": 41212236.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.5878050029277802,
+      "epoch": 7.119601328903655,
+      "grad_norm": 0.024177011102437973,
+      "learning_rate": 0.0002,
+      "loss": 0.5889069437980652,
+      "mean_token_accuracy": 0.7598751783370972,
+      "num_tokens": 41289316.0,
+      "step": 541
+    },
+    {
+      "entropy": 0.5878520309925079,
+      "epoch": 7.132890365448505,
+      "grad_norm": 0.030242595821619034,
+      "learning_rate": 0.0002,
+      "loss": 0.5862522125244141,
+      "mean_token_accuracy": 0.7602282017469406,
+      "num_tokens": 41365544.0,
+      "step": 542
+    },
+    {
+      "entropy": 0.580777645111084,
+      "epoch": 7.146179401993355,
+      "grad_norm": 0.025827951729297638,
+      "learning_rate": 0.0002,
+      "loss": 0.5805131196975708,
+      "mean_token_accuracy": 0.7632949352264404,
+      "num_tokens": 41442375.0,
+      "step": 543
+    },
+    {
+      "entropy": 0.5955946892499924,
+      "epoch": 7.159468438538206,
+      "grad_norm": 0.03238573297858238,
+      "learning_rate": 0.0002,
+      "loss": 0.5978742837905884,
+      "mean_token_accuracy": 0.756095290184021,
+      "num_tokens": 41518639.0,
+      "step": 544
+    },
+    {
+      "entropy": 0.5933609753847122,
+      "epoch": 7.172757475083056,
+      "grad_norm": 0.026595359668135643,
+      "learning_rate": 0.0002,
+      "loss": 0.5913410782814026,
+      "mean_token_accuracy": 0.75886270403862,
+      "num_tokens": 41595728.0,
+      "step": 545
+    },
+    {
+      "entropy": 0.5813855677843094,
+      "epoch": 7.186046511627907,
+      "grad_norm": 0.03226481005549431,
+      "learning_rate": 0.0002,
+      "loss": 0.5804150104522705,
+      "mean_token_accuracy": 0.762812003493309,
+      "num_tokens": 41672700.0,
+      "step": 546
+    },
+    {
+      "entropy": 0.5884519517421722,
+      "epoch": 7.1993355481727574,
+      "grad_norm": 0.03133225440979004,
+      "learning_rate": 0.0002,
+      "loss": 0.587933361530304,
+      "mean_token_accuracy": 0.759622797369957,
+      "num_tokens": 41749957.0,
+      "step": 547
+    },
+    {
+      "entropy": 0.5912621170282364,
+      "epoch": 7.212624584717608,
+      "grad_norm": 0.02699204906821251,
+      "learning_rate": 0.0002,
+      "loss": 0.5909606218338013,
+      "mean_token_accuracy": 0.7599730789661407,
+      "num_tokens": 41827181.0,
+      "step": 548
+    },
+    {
+      "entropy": 0.5834774374961853,
+      "epoch": 7.225913621262459,
+      "grad_norm": 0.031116921454668045,
+      "learning_rate": 0.0002,
+      "loss": 0.5860615968704224,
+      "mean_token_accuracy": 0.7615607529878616,
+      "num_tokens": 41904276.0,
+      "step": 549
+    },
+    {
+      "entropy": 0.6060900539159775,
+      "epoch": 7.239202657807309,
+      "grad_norm": 0.03337494656443596,
+      "learning_rate": 0.0002,
+      "loss": 0.6010215282440186,
+      "mean_token_accuracy": 0.7554879635572433,
+      "num_tokens": 41981777.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.5843900144100189,
+      "epoch": 7.25249169435216,
+      "grad_norm": 0.0280764177441597,
+      "learning_rate": 0.0002,
+      "loss": 0.5816437005996704,
+      "mean_token_accuracy": 0.7622091621160507,
+      "num_tokens": 42058406.0,
+      "step": 551
+    },
+    {
+      "entropy": 0.5719930976629257,
+      "epoch": 7.26578073089701,
+      "grad_norm": 0.04221399873495102,
+      "learning_rate": 0.0002,
+      "loss": 0.5766112208366394,
+      "mean_token_accuracy": 0.7648352682590485,
+      "num_tokens": 42137283.0,
+      "step": 552
+    },
+    {
+      "entropy": 0.5938063710927963,
+      "epoch": 7.27906976744186,
+      "grad_norm": 0.04285285621881485,
+      "learning_rate": 0.0002,
+      "loss": 0.5963371396064758,
+      "mean_token_accuracy": 0.7569244503974915,
+      "num_tokens": 42213577.0,
+      "step": 553
+    },
+    {
+      "entropy": 0.5992718636989594,
+      "epoch": 7.292358803986711,
+      "grad_norm": 0.03192923218011856,
+      "learning_rate": 0.0002,
+      "loss": 0.5919658541679382,
+      "mean_token_accuracy": 0.7584283500909805,
+      "num_tokens": 42290765.0,
+      "step": 554
+    },
+    {
+      "entropy": 0.598827674984932,
+      "epoch": 7.305647840531561,
+      "grad_norm": 0.03549162670969963,
+      "learning_rate": 0.0002,
+      "loss": 0.5959571003913879,
+      "mean_token_accuracy": 0.7582818567752838,
+      "num_tokens": 42367554.0,
+      "step": 555
+    },
+    {
+      "entropy": 0.5883237272500992,
+      "epoch": 7.318936877076412,
+      "grad_norm": 0.03574231639504433,
+      "learning_rate": 0.0002,
+      "loss": 0.5865534543991089,
+      "mean_token_accuracy": 0.7601163238286972,
+      "num_tokens": 42443934.0,
+      "step": 556
+    },
+    {
+      "entropy": 0.5952439904212952,
+      "epoch": 7.332225913621262,
+      "grad_norm": 0.03858065605163574,
+      "learning_rate": 0.0002,
+      "loss": 0.5947918891906738,
+      "mean_token_accuracy": 0.7584473639726639,
+      "num_tokens": 42522384.0,
+      "step": 557
+    },
+    {
+      "entropy": 0.5895354747772217,
+      "epoch": 7.3455149501661126,
+      "grad_norm": 0.04516797512769699,
+      "learning_rate": 0.0002,
+      "loss": 0.5862763524055481,
+      "mean_token_accuracy": 0.7603979557752609,
+      "num_tokens": 42600686.0,
+      "step": 558
+    },
+    {
+      "entropy": 0.5977571457624435,
+      "epoch": 7.358803986710964,
+      "grad_norm": 0.030313201248645782,
+      "learning_rate": 0.0002,
+      "loss": 0.5932716131210327,
+      "mean_token_accuracy": 0.757668063044548,
+      "num_tokens": 42678840.0,
+      "step": 559
+    },
+    {
+      "entropy": 0.5848244279623032,
+      "epoch": 7.372093023255814,
+      "grad_norm": 0.04240503907203674,
+      "learning_rate": 0.0002,
+      "loss": 0.5870546698570251,
+      "mean_token_accuracy": 0.7618324756622314,
+      "num_tokens": 42755954.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.5977549999952316,
+      "epoch": 7.385382059800665,
+      "grad_norm": 0.031350430101156235,
+      "learning_rate": 0.0002,
+      "loss": 0.6006947755813599,
+      "mean_token_accuracy": 0.7568647265434265,
+      "num_tokens": 42832956.0,
+      "step": 561
+    },
+    {
+      "entropy": 0.5922767370939255,
+      "epoch": 7.398671096345515,
+      "grad_norm": 0.0464928075671196,
+      "learning_rate": 0.0002,
+      "loss": 0.5898110866546631,
+      "mean_token_accuracy": 0.761441171169281,
+      "num_tokens": 42910049.0,
+      "step": 562
+    },
+    {
+      "entropy": 0.5975309163331985,
+      "epoch": 7.411960132890365,
+      "grad_norm": 0.029810184612870216,
+      "learning_rate": 0.0002,
+      "loss": 0.5963293313980103,
+      "mean_token_accuracy": 0.7576682418584824,
+      "num_tokens": 42987206.0,
+      "step": 563
+    },
+    {
+      "entropy": 0.6032759100198746,
+      "epoch": 7.425249169435216,
+      "grad_norm": 0.03480512276291847,
+      "learning_rate": 0.0002,
+      "loss": 0.6015341281890869,
+      "mean_token_accuracy": 0.7556877881288528,
+      "num_tokens": 43062979.0,
+      "step": 564
+    },
+    {
+      "entropy": 0.5968350321054459,
+      "epoch": 7.438538205980066,
+      "grad_norm": 0.034472063183784485,
+      "learning_rate": 0.0002,
+      "loss": 0.6002466082572937,
+      "mean_token_accuracy": 0.7546430677175522,
+      "num_tokens": 43139685.0,
+      "step": 565
+    },
+    {
+      "entropy": 0.5969191342592239,
+      "epoch": 7.451827242524917,
+      "grad_norm": 0.030943354591727257,
+      "learning_rate": 0.0002,
+      "loss": 0.5945968627929688,
+      "mean_token_accuracy": 0.7584541887044907,
+      "num_tokens": 43217473.0,
+      "step": 566
+    },
+    {
+      "entropy": 0.6047678291797638,
+      "epoch": 7.465116279069767,
+      "grad_norm": 0.03206373378634453,
+      "learning_rate": 0.0002,
+      "loss": 0.6000797748565674,
+      "mean_token_accuracy": 0.7554351985454559,
+      "num_tokens": 43294484.0,
+      "step": 567
+    },
+    {
+      "entropy": 0.5903201103210449,
+      "epoch": 7.4784053156146175,
+      "grad_norm": 0.029840532690286636,
+      "learning_rate": 0.0002,
+      "loss": 0.5920034050941467,
+      "mean_token_accuracy": 0.7583808600902557,
+      "num_tokens": 43371808.0,
+      "step": 568
+    },
+    {
+      "entropy": 0.5891302227973938,
+      "epoch": 7.4916943521594686,
+      "grad_norm": 0.032370489090681076,
+      "learning_rate": 0.0002,
+      "loss": 0.5905359983444214,
+      "mean_token_accuracy": 0.7588883638381958,
+      "num_tokens": 43448977.0,
+      "step": 569
+    },
+    {
+      "entropy": 0.5911503434181213,
+      "epoch": 7.504983388704319,
+      "grad_norm": 0.024661490693688393,
+      "learning_rate": 0.0002,
+      "loss": 0.589153528213501,
+      "mean_token_accuracy": 0.7603337615728378,
+      "num_tokens": 43525211.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.5936204791069031,
+      "epoch": 7.51827242524917,
+      "grad_norm": 0.033602483570575714,
+      "learning_rate": 0.0002,
+      "loss": 0.5895155668258667,
+      "mean_token_accuracy": 0.7598973661661148,
+      "num_tokens": 43601774.0,
+      "step": 571
+    },
+    {
+      "entropy": 0.5936924070119858,
+      "epoch": 7.53156146179402,
+      "grad_norm": 0.028339127078652382,
+      "learning_rate": 0.0002,
+      "loss": 0.5915228128433228,
+      "mean_token_accuracy": 0.7601437717676163,
+      "num_tokens": 43679558.0,
+      "step": 572
+    },
+    {
+      "entropy": 0.5924441665410995,
+      "epoch": 7.544850498338871,
+      "grad_norm": 0.03638297691941261,
+      "learning_rate": 0.0002,
+      "loss": 0.5932373404502869,
+      "mean_token_accuracy": 0.7579649388790131,
+      "num_tokens": 43758060.0,
+      "step": 573
+    },
+    {
+      "entropy": 0.5993626266717911,
+      "epoch": 7.558139534883721,
+      "grad_norm": 0.028465403243899345,
+      "learning_rate": 0.0002,
+      "loss": 0.5999648571014404,
+      "mean_token_accuracy": 0.7560222744941711,
+      "num_tokens": 43834817.0,
+      "step": 574
+    },
+    {
+      "entropy": 0.5959221720695496,
+      "epoch": 7.571428571428571,
+      "grad_norm": 0.035040296614170074,
+      "learning_rate": 0.0002,
+      "loss": 0.592374324798584,
+      "mean_token_accuracy": 0.7590507417917252,
+      "num_tokens": 43911957.0,
+      "step": 575
+    },
+    {
+      "entropy": 0.5967639237642288,
+      "epoch": 7.584717607973422,
+      "grad_norm": 0.030499596148729324,
+      "learning_rate": 0.0002,
+      "loss": 0.5920624136924744,
+      "mean_token_accuracy": 0.7570383697748184,
+      "num_tokens": 43989718.0,
+      "step": 576
+    },
+    {
+      "entropy": 0.595067635178566,
+      "epoch": 7.598006644518272,
+      "grad_norm": 0.03523232042789459,
+      "learning_rate": 0.0002,
+      "loss": 0.597061812877655,
+      "mean_token_accuracy": 0.7577027082443237,
+      "num_tokens": 44066657.0,
+      "step": 577
+    },
+    {
+      "entropy": 0.5887541025876999,
+      "epoch": 7.6112956810631225,
+      "grad_norm": 0.03475109487771988,
+      "learning_rate": 0.0002,
+      "loss": 0.5964131951332092,
+      "mean_token_accuracy": 0.7577889561653137,
+      "num_tokens": 44143767.0,
+      "step": 578
+    },
+    {
+      "entropy": 0.592231884598732,
+      "epoch": 7.6245847176079735,
+      "grad_norm": 0.03178296983242035,
+      "learning_rate": 0.0002,
+      "loss": 0.5877656936645508,
+      "mean_token_accuracy": 0.7602957934141159,
+      "num_tokens": 44220541.0,
+      "step": 579
+    },
+    {
+      "entropy": 0.5901675373315811,
+      "epoch": 7.637873754152824,
+      "grad_norm": 0.028729397803544998,
+      "learning_rate": 0.0002,
+      "loss": 0.5811964273452759,
+      "mean_token_accuracy": 0.7640891075134277,
+      "num_tokens": 44298202.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.6007028222084045,
+      "epoch": 7.651162790697675,
+      "grad_norm": 0.03181817755103111,
+      "learning_rate": 0.0002,
+      "loss": 0.5992697477340698,
+      "mean_token_accuracy": 0.7546785622835159,
+      "num_tokens": 44375044.0,
+      "step": 581
+    },
+    {
+      "entropy": 0.5993082076311111,
+      "epoch": 7.664451827242525,
+      "grad_norm": 0.032050300389528275,
+      "learning_rate": 0.0002,
+      "loss": 0.6053569912910461,
+      "mean_token_accuracy": 0.7536642551422119,
+      "num_tokens": 44451384.0,
+      "step": 582
+    },
+    {
+      "entropy": 0.6015539616346359,
+      "epoch": 7.677740863787376,
+      "grad_norm": 0.03009367734193802,
+      "learning_rate": 0.0002,
+      "loss": 0.5985173583030701,
+      "mean_token_accuracy": 0.7561749666929245,
+      "num_tokens": 44528766.0,
+      "step": 583
+    },
+    {
+      "entropy": 0.6066677421331406,
+      "epoch": 7.691029900332226,
+      "grad_norm": 0.02756679244339466,
+      "learning_rate": 0.0002,
+      "loss": 0.604720950126648,
+      "mean_token_accuracy": 0.7532592117786407,
+      "num_tokens": 44605549.0,
+      "step": 584
+    },
+    {
+      "entropy": 0.5922277718782425,
+      "epoch": 7.704318936877076,
+      "grad_norm": 0.027669459581375122,
+      "learning_rate": 0.0002,
+      "loss": 0.5902705192565918,
+      "mean_token_accuracy": 0.7600447237491608,
+      "num_tokens": 44682878.0,
+      "step": 585
+    },
+    {
+      "entropy": 0.5769661366939545,
+      "epoch": 7.717607973421927,
+      "grad_norm": 0.028020597994327545,
+      "learning_rate": 0.0002,
+      "loss": 0.5795618891716003,
+      "mean_token_accuracy": 0.7647197842597961,
+      "num_tokens": 44759899.0,
+      "step": 586
+    },
+    {
+      "entropy": 0.5908134877681732,
+      "epoch": 7.730897009966777,
+      "grad_norm": 0.029171260073781013,
+      "learning_rate": 0.0002,
+      "loss": 0.5940332412719727,
+      "mean_token_accuracy": 0.7575349360704422,
+      "num_tokens": 44836446.0,
+      "step": 587
+    },
+    {
+      "entropy": 0.6053617596626282,
+      "epoch": 7.7441860465116275,
+      "grad_norm": 0.029087018221616745,
+      "learning_rate": 0.0002,
+      "loss": 0.6014258861541748,
+      "mean_token_accuracy": 0.7543994039297104,
+      "num_tokens": 44912843.0,
+      "step": 588
+    },
+    {
+      "entropy": 0.6038393527269363,
+      "epoch": 7.7574750830564785,
+      "grad_norm": 0.027699144557118416,
+      "learning_rate": 0.0002,
+      "loss": 0.5987313985824585,
+      "mean_token_accuracy": 0.7549974322319031,
+      "num_tokens": 44990430.0,
+      "step": 589
+    },
+    {
+      "entropy": 0.5965250134468079,
+      "epoch": 7.770764119601329,
+      "grad_norm": 0.030259517952799797,
+      "learning_rate": 0.0002,
+      "loss": 0.5975747108459473,
+      "mean_token_accuracy": 0.7567678689956665,
+      "num_tokens": 45068612.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.5944515317678452,
+      "epoch": 7.78405315614618,
+      "grad_norm": 0.026580356061458588,
+      "learning_rate": 0.0002,
+      "loss": 0.5963719487190247,
+      "mean_token_accuracy": 0.7574294954538345,
+      "num_tokens": 45146637.0,
+      "step": 591
+    },
+    {
+      "entropy": 0.6014003604650497,
+      "epoch": 7.79734219269103,
+      "grad_norm": 0.023800505325198174,
+      "learning_rate": 0.0002,
+      "loss": 0.599528431892395,
+      "mean_token_accuracy": 0.7546277940273285,
+      "num_tokens": 45222652.0,
+      "step": 592
+    },
+    {
+      "entropy": 0.6073395758867264,
+      "epoch": 7.810631229235881,
+      "grad_norm": 0.025431310757994652,
+      "learning_rate": 0.0002,
+      "loss": 0.6014832258224487,
+      "mean_token_accuracy": 0.7544971704483032,
+      "num_tokens": 45299371.0,
+      "step": 593
+    },
+    {
+      "entropy": 0.6004442870616913,
+      "epoch": 7.823920265780731,
+      "grad_norm": 0.025382395833730698,
+      "learning_rate": 0.0002,
+      "loss": 0.5996565818786621,
+      "mean_token_accuracy": 0.7561800926923752,
+      "num_tokens": 45375872.0,
+      "step": 594
+    },
+    {
+      "entropy": 0.597792387008667,
+      "epoch": 7.837209302325581,
+      "grad_norm": 0.02821579948067665,
+      "learning_rate": 0.0002,
+      "loss": 0.6011848449707031,
+      "mean_token_accuracy": 0.7544896304607391,
+      "num_tokens": 45452764.0,
+      "step": 595
+    },
+    {
+      "entropy": 0.6014498323202133,
+      "epoch": 7.850498338870432,
+      "grad_norm": 0.025945568457245827,
+      "learning_rate": 0.0002,
+      "loss": 0.5997246503829956,
+      "mean_token_accuracy": 0.756523534655571,
+      "num_tokens": 45529755.0,
+      "step": 596
+    },
+    {
+      "entropy": 0.602736622095108,
+      "epoch": 7.863787375415282,
+      "grad_norm": 0.023420821875333786,
+      "learning_rate": 0.0002,
+      "loss": 0.601398229598999,
+      "mean_token_accuracy": 0.7540724575519562,
+      "num_tokens": 45606356.0,
+      "step": 597
+    },
+    {
+      "entropy": 0.5954908579587936,
+      "epoch": 7.877076411960132,
+      "grad_norm": 0.028988869860768318,
+      "learning_rate": 0.0002,
+      "loss": 0.5935001373291016,
+      "mean_token_accuracy": 0.7580942809581757,
+      "num_tokens": 45683622.0,
+      "step": 598
+    },
+    {
+      "entropy": 0.5893382877111435,
+      "epoch": 7.8903654485049834,
+      "grad_norm": 0.02311522886157036,
+      "learning_rate": 0.0002,
+      "loss": 0.5844165086746216,
+      "mean_token_accuracy": 0.7620206922292709,
+      "num_tokens": 45761740.0,
+      "step": 599
+    },
+    {
+      "entropy": 0.6034763604402542,
+      "epoch": 7.903654485049834,
+      "grad_norm": 0.03306819126009941,
+      "learning_rate": 0.0002,
+      "loss": 0.6084754467010498,
+      "mean_token_accuracy": 0.7525132596492767,
+      "num_tokens": 45836951.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.5898222327232361,
+      "epoch": 7.916943521594685,
+      "grad_norm": 0.027422117069363594,
+      "learning_rate": 0.0002,
+      "loss": 0.5919756889343262,
+      "mean_token_accuracy": 0.7595413774251938,
+      "num_tokens": 45914305.0,
+      "step": 601
+    },
+    {
+      "entropy": 0.5984858423471451,
+      "epoch": 7.930232558139535,
+      "grad_norm": 0.030068816617131233,
+      "learning_rate": 0.0002,
+      "loss": 0.5928657054901123,
+      "mean_token_accuracy": 0.7575226724147797,
+      "num_tokens": 45990931.0,
+      "step": 602
+    },
+    {
+      "entropy": 0.5966551899909973,
+      "epoch": 7.943521594684386,
+      "grad_norm": 0.02550889365375042,
+      "learning_rate": 0.0002,
+      "loss": 0.5950169563293457,
+      "mean_token_accuracy": 0.7567807137966156,
+      "num_tokens": 46068210.0,
+      "step": 603
+    },
+    {
+      "entropy": 0.5984641313552856,
+      "epoch": 7.956810631229236,
+      "grad_norm": 0.027936860918998718,
+      "learning_rate": 0.0002,
+      "loss": 0.5978131294250488,
+      "mean_token_accuracy": 0.7564933449029922,
+      "num_tokens": 46145168.0,
+      "step": 604
+    },
+    {
+      "entropy": 0.5898311734199524,
+      "epoch": 7.970099667774086,
+      "grad_norm": 0.028104180470108986,
+      "learning_rate": 0.0002,
+      "loss": 0.5902965664863586,
+      "mean_token_accuracy": 0.7601361870765686,
+      "num_tokens": 46222214.0,
+      "step": 605
+    },
+    {
+      "entropy": 0.5912729799747467,
+      "epoch": 7.983388704318937,
+      "grad_norm": 0.022967718541622162,
+      "learning_rate": 0.0002,
+      "loss": 0.5875831842422485,
+      "mean_token_accuracy": 0.7609845697879791,
+      "num_tokens": 46299649.0,
+      "step": 606
+    },
+    {
+      "entropy": 0.5996886044740677,
+      "epoch": 7.996677740863787,
+      "grad_norm": 0.02525177225470543,
+      "learning_rate": 0.0002,
+      "loss": 0.5975291728973389,
+      "mean_token_accuracy": 0.7567523270845413,
+      "num_tokens": 46376767.0,
+      "step": 607
+    },
+    {
+      "entropy": 0.6052418351173401,
+      "epoch": 8.0,
+      "grad_norm": 0.03844604641199112,
+      "learning_rate": 0.0002,
+      "loss": 0.6005574464797974,
+      "mean_token_accuracy": 0.7535819411277771,
+      "num_tokens": 46396228.0,
+      "step": 608
+    },
+    {
+      "entropy": 0.5849415957927704,
+      "epoch": 8.013289036544851,
+      "grad_norm": 0.026697669178247452,
+      "learning_rate": 0.0002,
+      "loss": 0.5827944278717041,
+      "mean_token_accuracy": 0.763024628162384,
+      "num_tokens": 46472432.0,
+      "step": 609
+    },
+    {
+      "entropy": 0.5896167308092117,
+      "epoch": 8.0265780730897,
+      "grad_norm": 0.026246508583426476,
+      "learning_rate": 0.0002,
+      "loss": 0.5901464819908142,
+      "mean_token_accuracy": 0.7580210715532303,
+      "num_tokens": 46548667.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.5954991430044174,
+      "epoch": 8.039867109634551,
+      "grad_norm": 0.03505807742476463,
+      "learning_rate": 0.0002,
+      "loss": 0.5982897281646729,
+      "mean_token_accuracy": 0.7570490837097168,
+      "num_tokens": 46625714.0,
+      "step": 611
+    },
+    {
+      "entropy": 0.5948772430419922,
+      "epoch": 8.053156146179402,
+      "grad_norm": 0.03613806515932083,
+      "learning_rate": 0.0002,
+      "loss": 0.5921391844749451,
+      "mean_token_accuracy": 0.7585531324148178,
+      "num_tokens": 46702173.0,
+      "step": 612
+    },
+    {
+      "entropy": 0.6061599105596542,
+      "epoch": 8.066445182724253,
+      "grad_norm": 0.03219094127416611,
+      "learning_rate": 0.0002,
+      "loss": 0.6035000085830688,
+      "mean_token_accuracy": 0.7535400837659836,
+      "num_tokens": 46779368.0,
+      "step": 613
+    },
+    {
+      "entropy": 0.5738667398691177,
+      "epoch": 8.079734219269103,
+      "grad_norm": 0.03252333030104637,
+      "learning_rate": 0.0002,
+      "loss": 0.5781555771827698,
+      "mean_token_accuracy": 0.7640644162893295,
+      "num_tokens": 46856996.0,
+      "step": 614
+    },
+    {
+      "entropy": 0.5789659023284912,
+      "epoch": 8.093023255813954,
+      "grad_norm": 0.031124601140618324,
+      "learning_rate": 0.0002,
+      "loss": 0.5793648958206177,
+      "mean_token_accuracy": 0.7631232887506485,
+      "num_tokens": 46934246.0,
+      "step": 615
+    },
+    {
+      "entropy": 0.5891662240028381,
+      "epoch": 8.106312292358805,
+      "grad_norm": 0.037481456995010376,
+      "learning_rate": 0.0002,
+      "loss": 0.5809412598609924,
+      "mean_token_accuracy": 0.7630375772714615,
+      "num_tokens": 47011246.0,
+      "step": 616
+    },
+    {
+      "entropy": 0.5797739326953888,
+      "epoch": 8.119601328903654,
+      "grad_norm": 0.03575124964118004,
+      "learning_rate": 0.0002,
+      "loss": 0.5810019969940186,
+      "mean_token_accuracy": 0.7621732503175735,
+      "num_tokens": 47088374.0,
+      "step": 617
+    },
+    {
+      "entropy": 0.5935344249010086,
+      "epoch": 8.132890365448505,
+      "grad_norm": 0.03413181006908417,
+      "learning_rate": 0.0002,
+      "loss": 0.5933163166046143,
+      "mean_token_accuracy": 0.7567701190710068,
+      "num_tokens": 47165072.0,
+      "step": 618
+    },
+    {
+      "entropy": 0.5915709435939789,
+      "epoch": 8.146179401993356,
+      "grad_norm": 0.0331747867166996,
+      "learning_rate": 0.0002,
+      "loss": 0.5865128636360168,
+      "mean_token_accuracy": 0.7611654847860336,
+      "num_tokens": 47241162.0,
+      "step": 619
+    },
+    {
+      "entropy": 0.5944840461015701,
+      "epoch": 8.159468438538205,
+      "grad_norm": 0.03591196611523628,
+      "learning_rate": 0.0002,
+      "loss": 0.5911703109741211,
+      "mean_token_accuracy": 0.7580583393573761,
+      "num_tokens": 47317281.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.5692449063062668,
+      "epoch": 8.172757475083056,
+      "grad_norm": 0.03591502457857132,
+      "learning_rate": 0.0002,
+      "loss": 0.5714719891548157,
+      "mean_token_accuracy": 0.7654573917388916,
+      "num_tokens": 47394033.0,
+      "step": 621
+    },
+    {
+      "entropy": 0.5919756591320038,
+      "epoch": 8.186046511627907,
+      "grad_norm": 0.03331475332379341,
+      "learning_rate": 0.0002,
+      "loss": 0.5916996598243713,
+      "mean_token_accuracy": 0.7590637356042862,
+      "num_tokens": 47471503.0,
+      "step": 622
+    },
+    {
+      "entropy": 0.5817323327064514,
+      "epoch": 8.199335548172758,
+      "grad_norm": 0.03191077336668968,
+      "learning_rate": 0.0002,
+      "loss": 0.5777145624160767,
+      "mean_token_accuracy": 0.7650918811559677,
+      "num_tokens": 47548641.0,
+      "step": 623
+    },
+    {
+      "entropy": 0.5874795615673065,
+      "epoch": 8.212624584717608,
+      "grad_norm": 0.031134847551584244,
+      "learning_rate": 0.0002,
+      "loss": 0.5871837139129639,
+      "mean_token_accuracy": 0.7600111067295074,
+      "num_tokens": 47625739.0,
+      "step": 624
+    },
+    {
+      "entropy": 0.5865576267242432,
+      "epoch": 8.225913621262459,
+      "grad_norm": 0.02946043387055397,
+      "learning_rate": 0.0002,
+      "loss": 0.5841938257217407,
+      "mean_token_accuracy": 0.7618899345397949,
+      "num_tokens": 47703269.0,
+      "step": 625
+    },
+    {
+      "entropy": 0.5737685710191727,
+      "epoch": 8.23920265780731,
+      "grad_norm": 0.03301022946834564,
+      "learning_rate": 0.0002,
+      "loss": 0.5730248689651489,
+      "mean_token_accuracy": 0.7667439430952072,
+      "num_tokens": 47780176.0,
+      "step": 626
+    },
+    {
+      "entropy": 0.5927734076976776,
+      "epoch": 8.252491694352159,
+      "grad_norm": 0.030395688489079475,
+      "learning_rate": 0.0002,
+      "loss": 0.5933593511581421,
+      "mean_token_accuracy": 0.7567963600158691,
+      "num_tokens": 47855549.0,
+      "step": 627
+    },
+    {
+      "entropy": 0.5866550207138062,
+      "epoch": 8.26578073089701,
+      "grad_norm": 0.03712524101138115,
+      "learning_rate": 0.0002,
+      "loss": 0.5855146646499634,
+      "mean_token_accuracy": 0.7610810846090317,
+      "num_tokens": 47933667.0,
+      "step": 628
+    },
+    {
+      "entropy": 0.5913572609424591,
+      "epoch": 8.279069767441861,
+      "grad_norm": 0.038173478096723557,
+      "learning_rate": 0.0002,
+      "loss": 0.5929369330406189,
+      "mean_token_accuracy": 0.7585016936063766,
+      "num_tokens": 48010150.0,
+      "step": 629
+    },
+    {
+      "entropy": 0.5938078314065933,
+      "epoch": 8.29235880398671,
+      "grad_norm": 0.030125852674245834,
+      "learning_rate": 0.0002,
+      "loss": 0.5897150039672852,
+      "mean_token_accuracy": 0.758837953209877,
+      "num_tokens": 48087617.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.5898828655481339,
+      "epoch": 8.305647840531561,
+      "grad_norm": 0.03530623018741608,
+      "learning_rate": 0.0002,
+      "loss": 0.5891162753105164,
+      "mean_token_accuracy": 0.7587278485298157,
+      "num_tokens": 48164647.0,
+      "step": 631
+    },
+    {
+      "entropy": 0.5901964008808136,
+      "epoch": 8.318936877076412,
+      "grad_norm": 0.04116959869861603,
+      "learning_rate": 0.0002,
+      "loss": 0.5871089696884155,
+      "mean_token_accuracy": 0.7609442174434662,
+      "num_tokens": 48242570.0,
+      "step": 632
+    },
+    {
+      "entropy": 0.5975192934274673,
+      "epoch": 8.332225913621263,
+      "grad_norm": 0.033791229128837585,
+      "learning_rate": 0.0002,
+      "loss": 0.5978984832763672,
+      "mean_token_accuracy": 0.7558712214231491,
+      "num_tokens": 48318831.0,
+      "step": 633
+    },
+    {
+      "entropy": 0.5898595005273819,
+      "epoch": 8.345514950166113,
+      "grad_norm": 0.03173010051250458,
+      "learning_rate": 0.0002,
+      "loss": 0.5908152461051941,
+      "mean_token_accuracy": 0.7584023624658585,
+      "num_tokens": 48396014.0,
+      "step": 634
+    },
+    {
+      "entropy": 0.5868335217237473,
+      "epoch": 8.358803986710964,
+      "grad_norm": 0.03656291961669922,
+      "learning_rate": 0.0002,
+      "loss": 0.5912126302719116,
+      "mean_token_accuracy": 0.7578298896551132,
+      "num_tokens": 48472501.0,
+      "step": 635
+    },
+    {
+      "entropy": 0.5803808122873306,
+      "epoch": 8.372093023255815,
+      "grad_norm": 0.029666099697351456,
+      "learning_rate": 0.0002,
+      "loss": 0.5778638124465942,
+      "mean_token_accuracy": 0.7648247480392456,
+      "num_tokens": 48549876.0,
+      "step": 636
+    },
+    {
+      "entropy": 0.5892583876848221,
+      "epoch": 8.385382059800664,
+      "grad_norm": 0.036448702216148376,
+      "learning_rate": 0.0002,
+      "loss": 0.5853127241134644,
+      "mean_token_accuracy": 0.7601494789123535,
+      "num_tokens": 48627162.0,
+      "step": 637
+    },
+    {
+      "entropy": 0.6033164262771606,
+      "epoch": 8.398671096345515,
+      "grad_norm": 0.041355833411216736,
+      "learning_rate": 0.0002,
+      "loss": 0.6030725240707397,
+      "mean_token_accuracy": 0.7540614157915115,
+      "num_tokens": 48704614.0,
+      "step": 638
+    },
+    {
+      "entropy": 0.5965287089347839,
+      "epoch": 8.411960132890366,
+      "grad_norm": 0.03134165331721306,
+      "learning_rate": 0.0002,
+      "loss": 0.591283917427063,
+      "mean_token_accuracy": 0.7578448504209518,
+      "num_tokens": 48782332.0,
+      "step": 639
+    },
+    {
+      "entropy": 0.5925293862819672,
+      "epoch": 8.425249169435215,
+      "grad_norm": 0.0464448481798172,
+      "learning_rate": 0.0002,
+      "loss": 0.5897094011306763,
+      "mean_token_accuracy": 0.7594808042049408,
+      "num_tokens": 48860168.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.58897465467453,
+      "epoch": 8.438538205980066,
+      "grad_norm": 0.04942963272333145,
+      "learning_rate": 0.0002,
+      "loss": 0.591568648815155,
+      "mean_token_accuracy": 0.7598203122615814,
+      "num_tokens": 48937560.0,
+      "step": 641
+    },
+    {
+      "entropy": 0.5860904455184937,
+      "epoch": 8.451827242524917,
+      "grad_norm": 0.03515082225203514,
+      "learning_rate": 0.0002,
+      "loss": 0.5869107246398926,
+      "mean_token_accuracy": 0.7615831941366196,
+      "num_tokens": 49014115.0,
+      "step": 642
+    },
+    {
+      "entropy": 0.5864955633878708,
+      "epoch": 8.465116279069768,
+      "grad_norm": 0.04708025977015495,
+      "learning_rate": 0.0002,
+      "loss": 0.5874521732330322,
+      "mean_token_accuracy": 0.7600408792495728,
+      "num_tokens": 49090510.0,
+      "step": 643
+    },
+    {
+      "entropy": 0.5912910401821136,
+      "epoch": 8.478405315614618,
+      "grad_norm": 0.04989413917064667,
+      "learning_rate": 0.0002,
+      "loss": 0.5909178853034973,
+      "mean_token_accuracy": 0.7597139179706573,
+      "num_tokens": 49167949.0,
+      "step": 644
+    },
+    {
+      "entropy": 0.5893669277429581,
+      "epoch": 8.491694352159469,
+      "grad_norm": 0.02918706089258194,
+      "learning_rate": 0.0002,
+      "loss": 0.5860124230384827,
+      "mean_token_accuracy": 0.7605527341365814,
+      "num_tokens": 49245468.0,
+      "step": 645
+    },
+    {
+      "entropy": 0.587207704782486,
+      "epoch": 8.50498338870432,
+      "grad_norm": 0.03909502178430557,
+      "learning_rate": 0.0002,
+      "loss": 0.5876778364181519,
+      "mean_token_accuracy": 0.7586503326892853,
+      "num_tokens": 49322083.0,
+      "step": 646
+    },
+    {
+      "entropy": 0.5928920805454254,
+      "epoch": 8.518272425249169,
+      "grad_norm": 0.03155143931508064,
+      "learning_rate": 0.0002,
+      "loss": 0.5916935205459595,
+      "mean_token_accuracy": 0.7580639719963074,
+      "num_tokens": 49398151.0,
+      "step": 647
+    },
+    {
+      "entropy": 0.5820858925580978,
+      "epoch": 8.53156146179402,
+      "grad_norm": 0.03789787366986275,
+      "learning_rate": 0.0002,
+      "loss": 0.5806615352630615,
+      "mean_token_accuracy": 0.7624358385801315,
+      "num_tokens": 49474932.0,
+      "step": 648
+    },
+    {
+      "entropy": 0.594434067606926,
+      "epoch": 8.544850498338871,
+      "grad_norm": 0.035347383469343185,
+      "learning_rate": 0.0002,
+      "loss": 0.593879222869873,
+      "mean_token_accuracy": 0.7570043802261353,
+      "num_tokens": 49552035.0,
+      "step": 649
+    },
+    {
+      "entropy": 0.5807432979345322,
+      "epoch": 8.55813953488372,
+      "grad_norm": 0.03238654136657715,
+      "learning_rate": 0.0002,
+      "loss": 0.5827980637550354,
+      "mean_token_accuracy": 0.76199671626091,
+      "num_tokens": 49628511.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.5857634395360947,
+      "epoch": 8.571428571428571,
+      "grad_norm": 0.04061926156282425,
+      "learning_rate": 0.0002,
+      "loss": 0.5857123136520386,
+      "mean_token_accuracy": 0.7616678923368454,
+      "num_tokens": 49705806.0,
+      "step": 651
+    },
+    {
+      "entropy": 0.5947589576244354,
+      "epoch": 8.584717607973422,
+      "grad_norm": 0.03115455061197281,
+      "learning_rate": 0.0002,
+      "loss": 0.5889809727668762,
+      "mean_token_accuracy": 0.7592536062002182,
+      "num_tokens": 49782950.0,
+      "step": 652
+    },
+    {
+      "entropy": 0.5894382148981094,
+      "epoch": 8.598006644518273,
+      "grad_norm": 0.04479070007801056,
+      "learning_rate": 0.0002,
+      "loss": 0.5914884805679321,
+      "mean_token_accuracy": 0.7576984465122223,
+      "num_tokens": 49859768.0,
+      "step": 653
+    },
+    {
+      "entropy": 0.6050767302513123,
+      "epoch": 8.611295681063122,
+      "grad_norm": 0.03744325041770935,
+      "learning_rate": 0.0002,
+      "loss": 0.6043350696563721,
+      "mean_token_accuracy": 0.7538948655128479,
+      "num_tokens": 49935768.0,
+      "step": 654
+    },
+    {
+      "entropy": 0.5913153141736984,
+      "epoch": 8.624584717607974,
+      "grad_norm": 0.03681696206331253,
+      "learning_rate": 0.0002,
+      "loss": 0.5855178236961365,
+      "mean_token_accuracy": 0.7591262012720108,
+      "num_tokens": 50013624.0,
+      "step": 655
+    },
+    {
+      "entropy": 0.579682931303978,
+      "epoch": 8.637873754152825,
+      "grad_norm": 0.03786518797278404,
+      "learning_rate": 0.0002,
+      "loss": 0.5794950723648071,
+      "mean_token_accuracy": 0.7616621106863022,
+      "num_tokens": 50091066.0,
+      "step": 656
+    },
+    {
+      "entropy": 0.5738009661436081,
+      "epoch": 8.651162790697674,
+      "grad_norm": 0.0365300215780735,
+      "learning_rate": 0.0002,
+      "loss": 0.579700767993927,
+      "mean_token_accuracy": 0.7637236416339874,
+      "num_tokens": 50168260.0,
+      "step": 657
+    },
+    {
+      "entropy": 0.5828498899936676,
+      "epoch": 8.664451827242525,
+      "grad_norm": 0.039035145193338394,
+      "learning_rate": 0.0002,
+      "loss": 0.583361029624939,
+      "mean_token_accuracy": 0.7617398202419281,
+      "num_tokens": 50246062.0,
+      "step": 658
+    },
+    {
+      "entropy": 0.5916730761528015,
+      "epoch": 8.677740863787376,
+      "grad_norm": 0.037739627063274384,
+      "learning_rate": 0.0002,
+      "loss": 0.5868880748748779,
+      "mean_token_accuracy": 0.7602831721305847,
+      "num_tokens": 50323537.0,
+      "step": 659
+    },
+    {
+      "entropy": 0.5911594331264496,
+      "epoch": 8.691029900332225,
+      "grad_norm": 0.03326238691806793,
+      "learning_rate": 0.0002,
+      "loss": 0.5849337577819824,
+      "mean_token_accuracy": 0.7604767978191376,
+      "num_tokens": 50401316.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.5687000751495361,
+      "epoch": 8.704318936877076,
+      "grad_norm": 0.041488006711006165,
+      "learning_rate": 0.0002,
+      "loss": 0.5731285810470581,
+      "mean_token_accuracy": 0.7658777087926865,
+      "num_tokens": 50478547.0,
+      "step": 661
+    },
+    {
+      "entropy": 0.5819157212972641,
+      "epoch": 8.717607973421927,
+      "grad_norm": 0.04120723903179169,
+      "learning_rate": 0.0002,
+      "loss": 0.5868937969207764,
+      "mean_token_accuracy": 0.7604488730430603,
+      "num_tokens": 50554825.0,
+      "step": 662
+    },
+    {
+      "entropy": 0.6036993712186813,
+      "epoch": 8.730897009966778,
+      "grad_norm": 0.03696327656507492,
+      "learning_rate": 0.0002,
+      "loss": 0.5956385731697083,
+      "mean_token_accuracy": 0.7566656321287155,
+      "num_tokens": 50631462.0,
+      "step": 663
+    },
+    {
+      "entropy": 0.5896592438220978,
+      "epoch": 8.744186046511627,
+      "grad_norm": 0.03166646137833595,
+      "learning_rate": 0.0002,
+      "loss": 0.5841723680496216,
+      "mean_token_accuracy": 0.7621931433677673,
+      "num_tokens": 50708446.0,
+      "step": 664
+    },
+    {
+      "entropy": 0.5567464232444763,
+      "epoch": 8.757475083056478,
+      "grad_norm": 0.03862195834517479,
+      "learning_rate": 0.0002,
+      "loss": 0.5601667165756226,
+      "mean_token_accuracy": 0.7705302089452744,
+      "num_tokens": 50787365.0,
+      "step": 665
+    },
+    {
+      "entropy": 0.5759493261575699,
+      "epoch": 8.77076411960133,
+      "grad_norm": 0.03633789345622063,
+      "learning_rate": 0.0002,
+      "loss": 0.5765956044197083,
+      "mean_token_accuracy": 0.7658006548881531,
+      "num_tokens": 50864646.0,
+      "step": 666
+    },
+    {
+      "entropy": 0.5924972295761108,
+      "epoch": 8.784053156146179,
+      "grad_norm": 0.030330289155244827,
+      "learning_rate": 0.0002,
+      "loss": 0.5899439454078674,
+      "mean_token_accuracy": 0.7596959173679352,
+      "num_tokens": 50942183.0,
+      "step": 667
+    },
+    {
+      "entropy": 0.5888451188802719,
+      "epoch": 8.79734219269103,
+      "grad_norm": 0.03190000355243683,
+      "learning_rate": 0.0002,
+      "loss": 0.5857913494110107,
+      "mean_token_accuracy": 0.7602512836456299,
+      "num_tokens": 51019060.0,
+      "step": 668
+    },
+    {
+      "entropy": 0.5876550525426865,
+      "epoch": 8.81063122923588,
+      "grad_norm": 0.03718399628996849,
+      "learning_rate": 0.0002,
+      "loss": 0.5891491174697876,
+      "mean_token_accuracy": 0.7600254416465759,
+      "num_tokens": 51097559.0,
+      "step": 669
+    },
+    {
+      "entropy": 0.5892974138259888,
+      "epoch": 8.82392026578073,
+      "grad_norm": 0.032171525061130524,
+      "learning_rate": 0.0002,
+      "loss": 0.5908107757568359,
+      "mean_token_accuracy": 0.7585913240909576,
+      "num_tokens": 51174991.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.584599032998085,
+      "epoch": 8.837209302325581,
+      "grad_norm": 0.03337855264544487,
+      "learning_rate": 0.0002,
+      "loss": 0.5803552865982056,
+      "mean_token_accuracy": 0.7630793899297714,
+      "num_tokens": 51252440.0,
+      "step": 671
+    },
+    {
+      "entropy": 0.5799373984336853,
+      "epoch": 8.850498338870432,
+      "grad_norm": 0.029419418424367905,
+      "learning_rate": 0.0002,
+      "loss": 0.5741597414016724,
+      "mean_token_accuracy": 0.7663457095623016,
+      "num_tokens": 51329340.0,
+      "step": 672
+    },
+    {
+      "entropy": 0.5873212218284607,
+      "epoch": 8.863787375415283,
+      "grad_norm": 0.031486447900533676,
+      "learning_rate": 0.0002,
+      "loss": 0.5863950252532959,
+      "mean_token_accuracy": 0.7606247216463089,
+      "num_tokens": 51407310.0,
+      "step": 673
+    },
+    {
+      "entropy": 0.5943780243396759,
+      "epoch": 8.877076411960132,
+      "grad_norm": 0.028674740344285965,
+      "learning_rate": 0.0002,
+      "loss": 0.5923240184783936,
+      "mean_token_accuracy": 0.7589106559753418,
+      "num_tokens": 51484477.0,
+      "step": 674
+    },
+    {
+      "entropy": 0.5996403247117996,
+      "epoch": 8.890365448504983,
+      "grad_norm": 0.031019171699881554,
+      "learning_rate": 0.0002,
+      "loss": 0.5965366363525391,
+      "mean_token_accuracy": 0.75755475461483,
+      "num_tokens": 51561266.0,
+      "step": 675
+    },
+    {
+      "entropy": 0.5805972665548325,
+      "epoch": 8.903654485049834,
+      "grad_norm": 0.03063480742275715,
+      "learning_rate": 0.0002,
+      "loss": 0.5774473547935486,
+      "mean_token_accuracy": 0.7652737647294998,
+      "num_tokens": 51638934.0,
+      "step": 676
+    },
+    {
+      "entropy": 0.5837110131978989,
+      "epoch": 8.916943521594684,
+      "grad_norm": 0.031470976769924164,
+      "learning_rate": 0.0002,
+      "loss": 0.588325560092926,
+      "mean_token_accuracy": 0.7599637508392334,
+      "num_tokens": 51715847.0,
+      "step": 677
+    },
+    {
+      "entropy": 0.5817466825246811,
+      "epoch": 8.930232558139535,
+      "grad_norm": 0.03200995922088623,
+      "learning_rate": 0.0002,
+      "loss": 0.5843440294265747,
+      "mean_token_accuracy": 0.7613124251365662,
+      "num_tokens": 51792332.0,
+      "step": 678
+    },
+    {
+      "entropy": 0.5915301889181137,
+      "epoch": 8.943521594684386,
+      "grad_norm": 0.03387102484703064,
+      "learning_rate": 0.0002,
+      "loss": 0.5847914218902588,
+      "mean_token_accuracy": 0.7612261474132538,
+      "num_tokens": 51869698.0,
+      "step": 679
+    },
+    {
+      "entropy": 0.5942060202360153,
+      "epoch": 8.956810631229235,
+      "grad_norm": 0.028463192284107208,
+      "learning_rate": 0.0002,
+      "loss": 0.5879127383232117,
+      "mean_token_accuracy": 0.7602778673171997,
+      "num_tokens": 51946396.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.587485060095787,
+      "epoch": 8.970099667774086,
+      "grad_norm": 0.036710530519485474,
+      "learning_rate": 0.0002,
+      "loss": 0.5914723873138428,
+      "mean_token_accuracy": 0.7598218619823456,
+      "num_tokens": 52023301.0,
+      "step": 681
+    },
+    {
+      "entropy": 0.5879035592079163,
+      "epoch": 8.983388704318937,
+      "grad_norm": 0.029370330274105072,
+      "learning_rate": 0.0002,
+      "loss": 0.5872666835784912,
+      "mean_token_accuracy": 0.7601160258054733,
+      "num_tokens": 52100376.0,
+      "step": 682
+    },
+    {
+      "entropy": 0.5898729115724564,
+      "epoch": 8.996677740863788,
+      "grad_norm": 0.028376396745443344,
+      "learning_rate": 0.0002,
+      "loss": 0.5859116315841675,
+      "mean_token_accuracy": 0.7616438567638397,
+      "num_tokens": 52176733.0,
+      "step": 683
+    },
+    {
+      "entropy": 0.5918471813201904,
+      "epoch": 9.0,
+      "grad_norm": 0.05119534581899643,
+      "learning_rate": 0.0002,
+      "loss": 0.5821671485900879,
+      "mean_token_accuracy": 0.761242687702179,
+      "num_tokens": 52195562.0,
+      "step": 684
+    },
+    {
+      "entropy": 0.5841783732175827,
+      "epoch": 9.013289036544851,
+      "grad_norm": 0.032985758036375046,
+      "learning_rate": 0.0002,
+      "loss": 0.5801397562026978,
+      "mean_token_accuracy": 0.761728972196579,
+      "num_tokens": 52272860.0,
+      "step": 685
+    },
+    {
+      "entropy": 0.5715355575084686,
+      "epoch": 9.0265780730897,
+      "grad_norm": 0.038291025906801224,
+      "learning_rate": 0.0002,
+      "loss": 0.5748767256736755,
+      "mean_token_accuracy": 0.7646484971046448,
+      "num_tokens": 52350111.0,
+      "step": 686
+    },
+    {
+      "entropy": 0.5795238018035889,
+      "epoch": 9.039867109634551,
+      "grad_norm": 0.038716547191143036,
+      "learning_rate": 0.0002,
+      "loss": 0.5806439518928528,
+      "mean_token_accuracy": 0.7631524950265884,
+      "num_tokens": 52427007.0,
+      "step": 687
+    },
+    {
+      "entropy": 0.5912983864545822,
+      "epoch": 9.053156146179402,
+      "grad_norm": 0.041771478950977325,
+      "learning_rate": 0.0002,
+      "loss": 0.5837312340736389,
+      "mean_token_accuracy": 0.7606383711099625,
+      "num_tokens": 52503166.0,
+      "step": 688
+    },
+    {
+      "entropy": 0.5750531256198883,
+      "epoch": 9.066445182724253,
+      "grad_norm": 0.04263452813029289,
+      "learning_rate": 0.0002,
+      "loss": 0.5768004655838013,
+      "mean_token_accuracy": 0.7641178071498871,
+      "num_tokens": 52581240.0,
+      "step": 689
+    },
+    {
+      "entropy": 0.5842915773391724,
+      "epoch": 9.079734219269103,
+      "grad_norm": 0.04072727635502815,
+      "learning_rate": 0.0002,
+      "loss": 0.5862613916397095,
+      "mean_token_accuracy": 0.7600968182086945,
+      "num_tokens": 52657288.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.5826019048690796,
+      "epoch": 9.093023255813954,
+      "grad_norm": 0.042537905275821686,
+      "learning_rate": 0.0002,
+      "loss": 0.5760419368743896,
+      "mean_token_accuracy": 0.7641205936670303,
+      "num_tokens": 52733779.0,
+      "step": 691
+    },
+    {
+      "entropy": 0.5833656191825867,
+      "epoch": 9.106312292358805,
+      "grad_norm": 0.039307933300733566,
+      "learning_rate": 0.0002,
+      "loss": 0.5780174136161804,
+      "mean_token_accuracy": 0.7621660232543945,
+      "num_tokens": 52809877.0,
+      "step": 692
+    },
+    {
+      "entropy": 0.5833144932985306,
+      "epoch": 9.119601328903654,
+      "grad_norm": 0.03763529285788536,
+      "learning_rate": 0.0002,
+      "loss": 0.5828508734703064,
+      "mean_token_accuracy": 0.7619175165891647,
+      "num_tokens": 52887809.0,
+      "step": 693
+    },
+    {
+      "entropy": 0.5802912563085556,
+      "epoch": 9.132890365448505,
+      "grad_norm": 0.042694512754678726,
+      "learning_rate": 0.0002,
+      "loss": 0.5785671472549438,
+      "mean_token_accuracy": 0.7632732540369034,
+      "num_tokens": 52964832.0,
+      "step": 694
+    },
+    {
+      "entropy": 0.5879446417093277,
+      "epoch": 9.146179401993356,
+      "grad_norm": 0.045146629214286804,
+      "learning_rate": 0.0002,
+      "loss": 0.5837844014167786,
+      "mean_token_accuracy": 0.7614665478467941,
+      "num_tokens": 53041546.0,
+      "step": 695
+    },
+    {
+      "entropy": 0.5769313871860504,
+      "epoch": 9.159468438538205,
+      "grad_norm": 0.04538911208510399,
+      "learning_rate": 0.0002,
+      "loss": 0.578600287437439,
+      "mean_token_accuracy": 0.7636725157499313,
+      "num_tokens": 53118699.0,
+      "step": 696
+    },
+    {
+      "entropy": 0.5738919973373413,
+      "epoch": 9.172757475083056,
+      "grad_norm": 0.0365930013358593,
+      "learning_rate": 0.0002,
+      "loss": 0.5768018960952759,
+      "mean_token_accuracy": 0.7647374421358109,
+      "num_tokens": 53194552.0,
+      "step": 697
+    },
+    {
+      "entropy": 0.5699289292097092,
+      "epoch": 9.186046511627907,
+      "grad_norm": 0.03911687061190605,
+      "learning_rate": 0.0002,
+      "loss": 0.5651339888572693,
+      "mean_token_accuracy": 0.768623873591423,
+      "num_tokens": 53272699.0,
+      "step": 698
+    },
+    {
+      "entropy": 0.5740307569503784,
+      "epoch": 9.199335548172758,
+      "grad_norm": 0.042123015969991684,
+      "learning_rate": 0.0002,
+      "loss": 0.573491096496582,
+      "mean_token_accuracy": 0.7646935433149338,
+      "num_tokens": 53350054.0,
+      "step": 699
+    },
+    {
+      "entropy": 0.579895555973053,
+      "epoch": 9.212624584717608,
+      "grad_norm": 0.050487667322158813,
+      "learning_rate": 0.0002,
+      "loss": 0.5794432163238525,
+      "mean_token_accuracy": 0.7630578577518463,
+      "num_tokens": 53425787.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.5742232501506805,
+      "epoch": 9.225913621262459,
+      "grad_norm": 0.039948951452970505,
+      "learning_rate": 0.0002,
+      "loss": 0.5666079521179199,
+      "mean_token_accuracy": 0.7682702094316483,
+      "num_tokens": 53503176.0,
+      "step": 701
+    },
+    {
+      "entropy": 0.5729279220104218,
+      "epoch": 9.23920265780731,
+      "grad_norm": 0.05120493844151497,
+      "learning_rate": 0.0002,
+      "loss": 0.5697590112686157,
+      "mean_token_accuracy": 0.7677944153547287,
+      "num_tokens": 53581943.0,
+      "step": 702
+    },
+    {
+      "entropy": 0.5736203342676163,
+      "epoch": 9.252491694352159,
+      "grad_norm": 0.035830773413181305,
+      "learning_rate": 0.0002,
+      "loss": 0.5681569576263428,
+      "mean_token_accuracy": 0.7680141031742096,
+      "num_tokens": 53659572.0,
+      "step": 703
+    },
+    {
+      "entropy": 0.5700356960296631,
+      "epoch": 9.26578073089701,
+      "grad_norm": 0.04752933606505394,
+      "learning_rate": 0.0002,
+      "loss": 0.5745888948440552,
+      "mean_token_accuracy": 0.7650690525770187,
+      "num_tokens": 53736316.0,
+      "step": 704
+    },
+    {
+      "entropy": 0.5778879225254059,
+      "epoch": 9.279069767441861,
+      "grad_norm": 0.050968557596206665,
+      "learning_rate": 0.0002,
+      "loss": 0.5786650776863098,
+      "mean_token_accuracy": 0.7637975066900253,
+      "num_tokens": 53812375.0,
+      "step": 705
+    },
+    {
+      "entropy": 0.5845597684383392,
+      "epoch": 9.29235880398671,
+      "grad_norm": 0.04049290716648102,
+      "learning_rate": 0.0002,
+      "loss": 0.5771433115005493,
+      "mean_token_accuracy": 0.7640694826841354,
+      "num_tokens": 53889373.0,
+      "step": 706
+    },
+    {
+      "entropy": 0.5710542052984238,
+      "epoch": 9.305647840531561,
+      "grad_norm": 0.049213092774152756,
+      "learning_rate": 0.0002,
+      "loss": 0.5740491151809692,
+      "mean_token_accuracy": 0.7647717893123627,
+      "num_tokens": 53966336.0,
+      "step": 707
+    },
+    {
+      "entropy": 0.5792699754238129,
+      "epoch": 9.318936877076412,
+      "grad_norm": 0.04844093322753906,
+      "learning_rate": 0.0002,
+      "loss": 0.5842309594154358,
+      "mean_token_accuracy": 0.7619152516126633,
+      "num_tokens": 54042331.0,
+      "step": 708
+    },
+    {
+      "entropy": 0.5852373540401459,
+      "epoch": 9.332225913621263,
+      "grad_norm": 0.04324189946055412,
+      "learning_rate": 0.0002,
+      "loss": 0.5743598937988281,
+      "mean_token_accuracy": 0.7650860399007797,
+      "num_tokens": 54120476.0,
+      "step": 709
+    },
+    {
+      "entropy": 0.5748791694641113,
+      "epoch": 9.345514950166113,
+      "grad_norm": 0.03747578710317612,
+      "learning_rate": 0.0002,
+      "loss": 0.5679724216461182,
+      "mean_token_accuracy": 0.7683089375495911,
+      "num_tokens": 54198246.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.5661651790142059,
+      "epoch": 9.358803986710964,
+      "grad_norm": 0.05258210748434067,
+      "learning_rate": 0.0002,
+      "loss": 0.576251745223999,
+      "mean_token_accuracy": 0.7638748437166214,
+      "num_tokens": 54275752.0,
+      "step": 711
+    },
+    {
+      "entropy": 0.5675118267536163,
+      "epoch": 9.372093023255815,
+      "grad_norm": 0.03817306086421013,
+      "learning_rate": 0.0002,
+      "loss": 0.5666477680206299,
+      "mean_token_accuracy": 0.7682253867387772,
+      "num_tokens": 54352088.0,
+      "step": 712
+    },
+    {
+      "entropy": 0.591354101896286,
+      "epoch": 9.385382059800664,
+      "grad_norm": 0.03948526084423065,
+      "learning_rate": 0.0002,
+      "loss": 0.5834983587265015,
+      "mean_token_accuracy": 0.7615845501422882,
+      "num_tokens": 54429552.0,
+      "step": 713
+    },
+    {
+      "entropy": 0.5859814137220383,
+      "epoch": 9.398671096345515,
+      "grad_norm": 0.03413126617670059,
+      "learning_rate": 0.0002,
+      "loss": 0.5804739594459534,
+      "mean_token_accuracy": 0.7621511518955231,
+      "num_tokens": 54507607.0,
+      "step": 714
+    },
+    {
+      "entropy": 0.5741334706544876,
+      "epoch": 9.411960132890366,
+      "grad_norm": 0.041531626135110855,
+      "learning_rate": 0.0002,
+      "loss": 0.5779021978378296,
+      "mean_token_accuracy": 0.7632759809494019,
+      "num_tokens": 54583653.0,
+      "step": 715
+    },
+    {
+      "entropy": 0.5825967490673065,
+      "epoch": 9.425249169435215,
+      "grad_norm": 0.04578561708331108,
+      "learning_rate": 0.0002,
+      "loss": 0.582665205001831,
+      "mean_token_accuracy": 0.7611964493989944,
+      "num_tokens": 54660880.0,
+      "step": 716
+    },
+    {
+      "entropy": 0.5794167369604111,
+      "epoch": 9.438538205980066,
+      "grad_norm": 0.049054164439439774,
+      "learning_rate": 0.0002,
+      "loss": 0.5784138441085815,
+      "mean_token_accuracy": 0.7645507901906967,
+      "num_tokens": 54737070.0,
+      "step": 717
+    },
+    {
+      "entropy": 0.5805319249629974,
+      "epoch": 9.451827242524917,
+      "grad_norm": 0.0353560745716095,
+      "learning_rate": 0.0002,
+      "loss": 0.5795561075210571,
+      "mean_token_accuracy": 0.7641272395849228,
+      "num_tokens": 54813163.0,
+      "step": 718
+    },
+    {
+      "entropy": 0.5851545929908752,
+      "epoch": 9.465116279069768,
+      "grad_norm": 0.04045746475458145,
+      "learning_rate": 0.0002,
+      "loss": 0.5845059752464294,
+      "mean_token_accuracy": 0.7607411295175552,
+      "num_tokens": 54889927.0,
+      "step": 719
+    },
+    {
+      "entropy": 0.579479768872261,
+      "epoch": 9.478405315614618,
+      "grad_norm": 0.03297734633088112,
+      "learning_rate": 0.0002,
+      "loss": 0.5776249170303345,
+      "mean_token_accuracy": 0.7639060616493225,
+      "num_tokens": 54967357.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.5843139588832855,
+      "epoch": 9.491694352159469,
+      "grad_norm": 0.04577886313199997,
+      "learning_rate": 0.0002,
+      "loss": 0.5836153030395508,
+      "mean_token_accuracy": 0.7609659135341644,
+      "num_tokens": 55045666.0,
+      "step": 721
+    },
+    {
+      "entropy": 0.5866421908140182,
+      "epoch": 9.50498338870432,
+      "grad_norm": 0.034662481397390366,
+      "learning_rate": 0.0002,
+      "loss": 0.5821191668510437,
+      "mean_token_accuracy": 0.7624327689409256,
+      "num_tokens": 55123419.0,
+      "step": 722
+    },
+    {
+      "entropy": 0.5771774500608444,
+      "epoch": 9.518272425249169,
+      "grad_norm": 0.03873879462480545,
+      "learning_rate": 0.0002,
+      "loss": 0.5764944553375244,
+      "mean_token_accuracy": 0.7636460214853287,
+      "num_tokens": 55200111.0,
+      "step": 723
+    },
+    {
+      "entropy": 0.5907417982816696,
+      "epoch": 9.53156146179402,
+      "grad_norm": 0.036843083798885345,
+      "learning_rate": 0.0002,
+      "loss": 0.5917870998382568,
+      "mean_token_accuracy": 0.7568515092134476,
+      "num_tokens": 55276410.0,
+      "step": 724
+    },
+    {
+      "entropy": 0.5745353251695633,
+      "epoch": 9.544850498338871,
+      "grad_norm": 0.041196517646312714,
+      "learning_rate": 0.0002,
+      "loss": 0.5695568323135376,
+      "mean_token_accuracy": 0.7673688530921936,
+      "num_tokens": 55354912.0,
+      "step": 725
+    },
+    {
+      "entropy": 0.5847089290618896,
+      "epoch": 9.55813953488372,
+      "grad_norm": 0.045686203986406326,
+      "learning_rate": 0.0002,
+      "loss": 0.5830461382865906,
+      "mean_token_accuracy": 0.7621099352836609,
+      "num_tokens": 55432664.0,
+      "step": 726
+    },
+    {
+      "entropy": 0.5793291479349136,
+      "epoch": 9.571428571428571,
+      "grad_norm": 0.037047456949949265,
+      "learning_rate": 0.0002,
+      "loss": 0.5785675048828125,
+      "mean_token_accuracy": 0.7639460265636444,
+      "num_tokens": 55509362.0,
+      "step": 727
+    },
+    {
+      "entropy": 0.5835801213979721,
+      "epoch": 9.584717607973422,
+      "grad_norm": 0.03607526794075966,
+      "learning_rate": 0.0002,
+      "loss": 0.582047700881958,
+      "mean_token_accuracy": 0.7607121020555496,
+      "num_tokens": 55586114.0,
+      "step": 728
+    },
+    {
+      "entropy": 0.5782389640808105,
+      "epoch": 9.598006644518273,
+      "grad_norm": 0.04172277823090553,
+      "learning_rate": 0.0002,
+      "loss": 0.5722988843917847,
+      "mean_token_accuracy": 0.7641795426607132,
+      "num_tokens": 55662923.0,
+      "step": 729
+    },
+    {
+      "entropy": 0.5846338421106339,
+      "epoch": 9.611295681063122,
+      "grad_norm": 0.037988658994436264,
+      "learning_rate": 0.0002,
+      "loss": 0.582650899887085,
+      "mean_token_accuracy": 0.7619550377130508,
+      "num_tokens": 55740517.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.5653723329305649,
+      "epoch": 9.624584717607974,
+      "grad_norm": 0.04472494497895241,
+      "learning_rate": 0.0002,
+      "loss": 0.5709476470947266,
+      "mean_token_accuracy": 0.7665883451700211,
+      "num_tokens": 55817432.0,
+      "step": 731
+    },
+    {
+      "entropy": 0.5787388980388641,
+      "epoch": 9.637873754152825,
+      "grad_norm": 0.04359956458210945,
+      "learning_rate": 0.0002,
+      "loss": 0.5829640626907349,
+      "mean_token_accuracy": 0.760590597987175,
+      "num_tokens": 55894580.0,
+      "step": 732
+    },
+    {
+      "entropy": 0.5837588608264923,
+      "epoch": 9.651162790697674,
+      "grad_norm": 0.037198908627033234,
+      "learning_rate": 0.0002,
+      "loss": 0.5767493844032288,
+      "mean_token_accuracy": 0.7640175223350525,
+      "num_tokens": 55972067.0,
+      "step": 733
+    },
+    {
+      "entropy": 0.5896154344081879,
+      "epoch": 9.664451827242525,
+      "grad_norm": 0.04088831692934036,
+      "learning_rate": 0.0002,
+      "loss": 0.5833977460861206,
+      "mean_token_accuracy": 0.7615736275911331,
+      "num_tokens": 56049153.0,
+      "step": 734
+    },
+    {
+      "entropy": 0.5845619142055511,
+      "epoch": 9.677740863787376,
+      "grad_norm": 0.0368829071521759,
+      "learning_rate": 0.0002,
+      "loss": 0.5855330228805542,
+      "mean_token_accuracy": 0.760240375995636,
+      "num_tokens": 56126404.0,
+      "step": 735
+    },
+    {
+      "entropy": 0.5753927528858185,
+      "epoch": 9.691029900332225,
+      "grad_norm": 0.03727378696203232,
+      "learning_rate": 0.0002,
+      "loss": 0.5770326852798462,
+      "mean_token_accuracy": 0.7630214542150497,
+      "num_tokens": 56203791.0,
+      "step": 736
+    },
+    {
+      "entropy": 0.5791745483875275,
+      "epoch": 9.704318936877076,
+      "grad_norm": 0.041700221598148346,
+      "learning_rate": 0.0002,
+      "loss": 0.5799466371536255,
+      "mean_token_accuracy": 0.7615901082754135,
+      "num_tokens": 56281724.0,
+      "step": 737
+    },
+    {
+      "entropy": 0.577799379825592,
+      "epoch": 9.717607973421927,
+      "grad_norm": 0.04183577001094818,
+      "learning_rate": 0.0002,
+      "loss": 0.5686119198799133,
+      "mean_token_accuracy": 0.7689932882785797,
+      "num_tokens": 56359768.0,
+      "step": 738
+    },
+    {
+      "entropy": 0.5778573751449585,
+      "epoch": 9.730897009966778,
+      "grad_norm": 0.03849725425243378,
+      "learning_rate": 0.0002,
+      "loss": 0.5799794793128967,
+      "mean_token_accuracy": 0.763456404209137,
+      "num_tokens": 56436568.0,
+      "step": 739
+    },
+    {
+      "entropy": 0.5874462425708771,
+      "epoch": 9.744186046511627,
+      "grad_norm": 0.04311611130833626,
+      "learning_rate": 0.0002,
+      "loss": 0.5923245549201965,
+      "mean_token_accuracy": 0.7570117563009262,
+      "num_tokens": 56514244.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.5856891423463821,
+      "epoch": 9.757475083056478,
+      "grad_norm": 0.03763333708047867,
+      "learning_rate": 0.0002,
+      "loss": 0.5764776468276978,
+      "mean_token_accuracy": 0.7640483975410461,
+      "num_tokens": 56590874.0,
+      "step": 741
+    },
+    {
+      "entropy": 0.5816144794225693,
+      "epoch": 9.77076411960133,
+      "grad_norm": 0.03455667570233345,
+      "learning_rate": 0.0002,
+      "loss": 0.577386736869812,
+      "mean_token_accuracy": 0.7633946985006332,
+      "num_tokens": 56668236.0,
+      "step": 742
+    },
+    {
+      "entropy": 0.5627869218587875,
+      "epoch": 9.784053156146179,
+      "grad_norm": 0.03804216906428337,
+      "learning_rate": 0.0002,
+      "loss": 0.5637524127960205,
+      "mean_token_accuracy": 0.7708553522825241,
+      "num_tokens": 56746118.0,
+      "step": 743
+    },
+    {
+      "entropy": 0.5718859136104584,
+      "epoch": 9.79734219269103,
+      "grad_norm": 0.0355333648622036,
+      "learning_rate": 0.0002,
+      "loss": 0.5740309357643127,
+      "mean_token_accuracy": 0.7656346410512924,
+      "num_tokens": 56822519.0,
+      "step": 744
+    },
+    {
+      "entropy": 0.591578796505928,
+      "epoch": 9.81063122923588,
+      "grad_norm": 0.03415751829743385,
+      "learning_rate": 0.0002,
+      "loss": 0.5866076946258545,
+      "mean_token_accuracy": 0.7593905031681061,
+      "num_tokens": 56898668.0,
+      "step": 745
+    },
+    {
+      "entropy": 0.574189156293869,
+      "epoch": 9.82392026578073,
+      "grad_norm": 0.03239757567644119,
+      "learning_rate": 0.0002,
+      "loss": 0.5689048767089844,
+      "mean_token_accuracy": 0.7667425870895386,
+      "num_tokens": 56975460.0,
+      "step": 746
+    },
+    {
+      "entropy": 0.5808965563774109,
+      "epoch": 9.837209302325581,
+      "grad_norm": 0.04112805798649788,
+      "learning_rate": 0.0002,
+      "loss": 0.5850220322608948,
+      "mean_token_accuracy": 0.7613527029752731,
+      "num_tokens": 57052463.0,
+      "step": 747
+    },
+    {
+      "entropy": 0.5827837735414505,
+      "epoch": 9.850498338870432,
+      "grad_norm": 0.04451862722635269,
+      "learning_rate": 0.0002,
+      "loss": 0.5784448385238647,
+      "mean_token_accuracy": 0.7636194676160812,
+      "num_tokens": 57128736.0,
+      "step": 748
+    },
+    {
+      "entropy": 0.5830895751714706,
+      "epoch": 9.863787375415283,
+      "grad_norm": 0.04024534672498703,
+      "learning_rate": 0.0002,
+      "loss": 0.5784624814987183,
+      "mean_token_accuracy": 0.763356551527977,
+      "num_tokens": 57205186.0,
+      "step": 749
+    },
+    {
+      "entropy": 0.5814227908849716,
+      "epoch": 9.877076411960132,
+      "grad_norm": 0.03781008720397949,
+      "learning_rate": 0.0002,
+      "loss": 0.5814759731292725,
+      "mean_token_accuracy": 0.7635377496480942,
+      "num_tokens": 57281898.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.5727444589138031,
+      "epoch": 9.890365448504983,
+      "grad_norm": 0.053311847150325775,
+      "learning_rate": 0.0002,
+      "loss": 0.5782246589660645,
+      "mean_token_accuracy": 0.7641912996768951,
+      "num_tokens": 57358985.0,
+      "step": 751
+    },
+    {
+      "entropy": 0.581045925617218,
+      "epoch": 9.903654485049834,
+      "grad_norm": 0.035100169479846954,
+      "learning_rate": 0.0002,
+      "loss": 0.5772634744644165,
+      "mean_token_accuracy": 0.7643590718507767,
+      "num_tokens": 57436207.0,
+      "step": 752
+    },
+    {
+      "entropy": 0.5704544335603714,
+      "epoch": 9.916943521594684,
+      "grad_norm": 0.03945276141166687,
+      "learning_rate": 0.0002,
+      "loss": 0.5681596994400024,
+      "mean_token_accuracy": 0.768840491771698,
+      "num_tokens": 57513512.0,
+      "step": 753
+    },
+    {
+      "entropy": 0.571413591504097,
+      "epoch": 9.930232558139535,
+      "grad_norm": 0.03285612910985947,
+      "learning_rate": 0.0002,
+      "loss": 0.5690507292747498,
+      "mean_token_accuracy": 0.767002746462822,
+      "num_tokens": 57591421.0,
+      "step": 754
+    },
+    {
+      "entropy": 0.5804395973682404,
+      "epoch": 9.943521594684386,
+      "grad_norm": 0.03729914873838425,
+      "learning_rate": 0.0002,
+      "loss": 0.5794394016265869,
+      "mean_token_accuracy": 0.7638011276721954,
+      "num_tokens": 57668937.0,
+      "step": 755
+    },
+    {
+      "entropy": 0.5725217163562775,
+      "epoch": 9.956810631229235,
+      "grad_norm": 0.03362439572811127,
+      "learning_rate": 0.0002,
+      "loss": 0.5742719173431396,
+      "mean_token_accuracy": 0.7654726803302765,
+      "num_tokens": 57745669.0,
+      "step": 756
+    },
+    {
+      "entropy": 0.5836540162563324,
+      "epoch": 9.970099667774086,
+      "grad_norm": 0.035550713539123535,
+      "learning_rate": 0.0002,
+      "loss": 0.5818772912025452,
+      "mean_token_accuracy": 0.763662725687027,
+      "num_tokens": 57822380.0,
+      "step": 757
+    },
+    {
+      "entropy": 0.5891828089952469,
+      "epoch": 9.983388704318937,
+      "grad_norm": 0.03719017282128334,
+      "learning_rate": 0.0002,
+      "loss": 0.5829074382781982,
+      "mean_token_accuracy": 0.7616835683584213,
+      "num_tokens": 57898798.0,
+      "step": 758
+    },
+    {
+      "entropy": 0.5854817181825638,
+      "epoch": 9.996677740863788,
+      "grad_norm": 0.03322125971317291,
+      "learning_rate": 0.0002,
+      "loss": 0.5807455778121948,
+      "mean_token_accuracy": 0.7624249011278152,
+      "num_tokens": 57975797.0,
+      "step": 759
+    },
+    {
+      "entropy": 0.596860408782959,
+      "epoch": 10.0,
+      "grad_norm": 0.0526103712618351,
+      "learning_rate": 0.0002,
+      "loss": 0.5905860662460327,
+      "mean_token_accuracy": 0.7580848336219788,
+      "num_tokens": 57994973.0,
+      "step": 760
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 760,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.4745117510960415e+18,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}