diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,9034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7598784194528876,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0007598784194528875,
+      "grad_norm": 11.767926216125488,
+      "learning_rate": 0.0,
+      "loss": 0.7937269806861877,
+      "mean_token_accuracy": 0.7822731137275696,
+      "num_tokens": 10507.0,
+      "step": 1
+    },
+    {
+      "epoch": 0.001519756838905775,
+      "grad_norm": 14.9199800491333,
+      "learning_rate": 2.5252525252525256e-08,
+      "loss": 0.7665389776229858,
+      "mean_token_accuracy": 0.8342233300209045,
+      "num_tokens": 14806.0,
+      "step": 2
+    },
+    {
+      "epoch": 0.0022796352583586625,
+      "grad_norm": 11.991217613220215,
+      "learning_rate": 5.050505050505051e-08,
+      "loss": 0.9597002267837524,
+      "mean_token_accuracy": 0.7054992318153381,
+      "num_tokens": 27170.0,
+      "step": 3
+    },
+    {
+      "epoch": 0.00303951367781155,
+      "grad_norm": 12.958333015441895,
+      "learning_rate": 7.575757575757576e-08,
+      "loss": 0.9971482753753662,
+      "mean_token_accuracy": 0.7261134386062622,
+      "num_tokens": 33729.0,
+      "step": 4
+    },
+    {
+      "epoch": 0.003799392097264438,
+      "grad_norm": 13.5665283203125,
+      "learning_rate": 1.0101010101010103e-07,
+      "loss": 0.9504883885383606,
+      "mean_token_accuracy": 0.745307445526123,
+      "num_tokens": 41174.0,
+      "step": 5
+    },
+    {
+      "epoch": 0.004559270516717325,
+      "grad_norm": 10.09444808959961,
+      "learning_rate": 1.2626262626262626e-07,
+      "loss": 0.759548008441925,
+      "mean_token_accuracy": 0.7842121124267578,
+      "num_tokens": 47943.0,
+      "step": 6
+    },
+    {
+      "epoch": 0.005319148936170213,
+      "grad_norm": 10.741650581359863,
+      "learning_rate": 1.5151515151515152e-07,
+      "loss": 0.8231598138809204,
+      "mean_token_accuracy": 0.7550969123840332,
+      "num_tokens": 56665.0,
+      "step": 7
+    },
+    {
+      "epoch": 0.0060790273556231,
+      "grad_norm": 12.250170707702637,
+      "learning_rate": 1.767676767676768e-07,
+      "loss": 0.8576581478118896,
+      "mean_token_accuracy": 0.7568671703338623,
+      "num_tokens": 67606.0,
+      "step": 8
+    },
+    {
+      "epoch": 0.006838905775075988,
+      "grad_norm": 12.828629493713379,
+      "learning_rate": 2.0202020202020205e-07,
+      "loss": 0.9886435866355896,
+      "mean_token_accuracy": 0.733400285243988,
+      "num_tokens": 74272.0,
+      "step": 9
+    },
+    {
+      "epoch": 0.007598784194528876,
+      "grad_norm": 15.966923713684082,
+      "learning_rate": 2.2727272727272729e-07,
+      "loss": 1.064985990524292,
+      "mean_token_accuracy": 0.7101132869720459,
+      "num_tokens": 80524.0,
+      "step": 10
+    },
+    {
+      "epoch": 0.008358662613981762,
+      "grad_norm": 10.864850044250488,
+      "learning_rate": 2.525252525252525e-07,
+      "loss": 0.8311550617218018,
+      "mean_token_accuracy": 0.7431639432907104,
+      "num_tokens": 96292.0,
+      "step": 11
+    },
+    {
+      "epoch": 0.00911854103343465,
+      "grad_norm": 16.438785552978516,
+      "learning_rate": 2.7777777777777776e-07,
+      "loss": 1.0579866170883179,
+      "mean_token_accuracy": 0.7222976684570312,
+      "num_tokens": 102992.0,
+      "step": 12
+    },
+    {
+      "epoch": 0.009878419452887538,
+      "grad_norm": 11.179214477539062,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 0.9816144704818726,
+      "mean_token_accuracy": 0.7206371426582336,
+      "num_tokens": 113571.0,
+      "step": 13
+    },
+    {
+      "epoch": 0.010638297872340425,
+      "grad_norm": 12.780299186706543,
+      "learning_rate": 3.2828282828282834e-07,
+      "loss": 0.847449004650116,
+      "mean_token_accuracy": 0.7826199531555176,
+      "num_tokens": 119568.0,
+      "step": 14
+    },
+    {
+      "epoch": 0.011398176291793313,
+      "grad_norm": 14.800421714782715,
+      "learning_rate": 3.535353535353536e-07,
+      "loss": 0.9275516271591187,
+      "mean_token_accuracy": 0.7655045986175537,
+      "num_tokens": 126258.0,
+      "step": 15
+    },
+    {
+      "epoch": 0.0121580547112462,
+      "grad_norm": 11.267602920532227,
+      "learning_rate": 3.787878787878788e-07,
+      "loss": 0.8464037179946899,
+      "mean_token_accuracy": 0.7606508731842041,
+      "num_tokens": 136831.0,
+      "step": 16
+    },
+    {
+      "epoch": 0.012917933130699088,
+      "grad_norm": 12.891013145446777,
+      "learning_rate": 4.040404040404041e-07,
+      "loss": 0.9903074502944946,
+      "mean_token_accuracy": 0.7247487306594849,
+      "num_tokens": 150434.0,
+      "step": 17
+    },
+    {
+      "epoch": 0.013677811550151976,
+      "grad_norm": 11.13957691192627,
+      "learning_rate": 4.2929292929292934e-07,
+      "loss": 0.8287211656570435,
+      "mean_token_accuracy": 0.7621913552284241,
+      "num_tokens": 158516.0,
+      "step": 18
+    },
+    {
+      "epoch": 0.014437689969604863,
+      "grad_norm": 18.39569664001465,
+      "learning_rate": 4.5454545454545457e-07,
+      "loss": 1.150015115737915,
+      "mean_token_accuracy": 0.7349498271942139,
+      "num_tokens": 162214.0,
+      "step": 19
+    },
+    {
+      "epoch": 0.015197568389057751,
+      "grad_norm": 9.353750228881836,
+      "learning_rate": 4.797979797979798e-07,
+      "loss": 0.7228299379348755,
+      "mean_token_accuracy": 0.7969573736190796,
+      "num_tokens": 173035.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.015957446808510637,
+      "grad_norm": 8.267163276672363,
+      "learning_rate": 5.05050505050505e-07,
+      "loss": 0.7358136177062988,
+      "mean_token_accuracy": 0.7903937101364136,
+      "num_tokens": 183568.0,
+      "step": 21
+    },
+    {
+      "epoch": 0.016717325227963525,
+      "grad_norm": 11.137128829956055,
+      "learning_rate": 5.303030303030304e-07,
+      "loss": 1.0075397491455078,
+      "mean_token_accuracy": 0.702807605266571,
+      "num_tokens": 192759.0,
+      "step": 22
+    },
+    {
+      "epoch": 0.017477203647416412,
+      "grad_norm": 10.734103202819824,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 0.8925919532775879,
+      "mean_token_accuracy": 0.7475671768188477,
+      "num_tokens": 201280.0,
+      "step": 23
+    },
+    {
+      "epoch": 0.0182370820668693,
+      "grad_norm": 11.945566177368164,
+      "learning_rate": 5.808080808080809e-07,
+      "loss": 0.7260514497756958,
+      "mean_token_accuracy": 0.7859152555465698,
+      "num_tokens": 218053.0,
+      "step": 24
+    },
+    {
+      "epoch": 0.018996960486322188,
+      "grad_norm": 18.610652923583984,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 0.8995465636253357,
+      "mean_token_accuracy": 0.7931990623474121,
+      "num_tokens": 220953.0,
+      "step": 25
+    },
+    {
+      "epoch": 0.019756838905775075,
+      "grad_norm": 10.51898193359375,
+      "learning_rate": 6.313131313131314e-07,
+      "loss": 0.9532671570777893,
+      "mean_token_accuracy": 0.7257645726203918,
+      "num_tokens": 231200.0,
+      "step": 26
+    },
+    {
+      "epoch": 0.020516717325227963,
+      "grad_norm": 9.581812858581543,
+      "learning_rate": 6.565656565656567e-07,
+      "loss": 0.9038010239601135,
+      "mean_token_accuracy": 0.7390379905700684,
+      "num_tokens": 237711.0,
+      "step": 27
+    },
+    {
+      "epoch": 0.02127659574468085,
+      "grad_norm": 12.297484397888184,
+      "learning_rate": 6.818181818181818e-07,
+      "loss": 1.048936367034912,
+      "mean_token_accuracy": 0.7175670862197876,
+      "num_tokens": 242503.0,
+      "step": 28
+    },
+    {
+      "epoch": 0.022036474164133738,
+      "grad_norm": 7.437953472137451,
+      "learning_rate": 7.070707070707071e-07,
+      "loss": 0.8308826684951782,
+      "mean_token_accuracy": 0.7415335774421692,
+      "num_tokens": 250842.0,
+      "step": 29
+    },
+    {
+      "epoch": 0.022796352583586626,
+      "grad_norm": 6.134475231170654,
+      "learning_rate": 7.323232323232324e-07,
+      "loss": 0.647913932800293,
+      "mean_token_accuracy": 0.8124054670333862,
+      "num_tokens": 267453.0,
+      "step": 30
+    },
+    {
+      "epoch": 0.023556231003039513,
+      "grad_norm": 6.678966045379639,
+      "learning_rate": 7.575757575757576e-07,
+      "loss": 0.7052810192108154,
+      "mean_token_accuracy": 0.7908754348754883,
+      "num_tokens": 284416.0,
+      "step": 31
+    },
+    {
+      "epoch": 0.0243161094224924,
+      "grad_norm": 7.42232084274292,
+      "learning_rate": 7.82828282828283e-07,
+      "loss": 1.022383213043213,
+      "mean_token_accuracy": 0.7053230404853821,
+      "num_tokens": 292073.0,
+      "step": 32
+    },
+    {
+      "epoch": 0.02507598784194529,
+      "grad_norm": 6.463219165802002,
+      "learning_rate": 8.080808080808082e-07,
+      "loss": 0.7603012323379517,
+      "mean_token_accuracy": 0.7728140354156494,
+      "num_tokens": 298550.0,
+      "step": 33
+    },
+    {
+      "epoch": 0.025835866261398176,
+      "grad_norm": 5.668411731719971,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 0.7707852721214294,
+      "mean_token_accuracy": 0.7827773094177246,
+      "num_tokens": 306683.0,
+      "step": 34
+    },
+    {
+      "epoch": 0.026595744680851064,
+      "grad_norm": 4.984964847564697,
+      "learning_rate": 8.585858585858587e-07,
+      "loss": 0.6317349672317505,
+      "mean_token_accuracy": 0.8106861114501953,
+      "num_tokens": 318842.0,
+      "step": 35
+    },
+    {
+      "epoch": 0.02735562310030395,
+      "grad_norm": 4.421732425689697,
+      "learning_rate": 8.838383838383839e-07,
+      "loss": 0.6228617429733276,
+      "mean_token_accuracy": 0.8023355603218079,
+      "num_tokens": 329850.0,
+      "step": 36
+    },
+    {
+      "epoch": 0.02811550151975684,
+      "grad_norm": 5.970808029174805,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.8443238139152527,
+      "mean_token_accuracy": 0.7462409734725952,
+      "num_tokens": 335844.0,
+      "step": 37
+    },
+    {
+      "epoch": 0.028875379939209727,
+      "grad_norm": 4.5389084815979,
+      "learning_rate": 9.343434343434345e-07,
+      "loss": 0.6976436376571655,
+      "mean_token_accuracy": 0.790410041809082,
+      "num_tokens": 348768.0,
+      "step": 38
+    },
+    {
+      "epoch": 0.029635258358662615,
+      "grad_norm": 4.116631507873535,
+      "learning_rate": 9.595959595959596e-07,
+      "loss": 0.6698519587516785,
+      "mean_token_accuracy": 0.7818127870559692,
+      "num_tokens": 355460.0,
+      "step": 39
+    },
+    {
+      "epoch": 0.030395136778115502,
+      "grad_norm": 3.3714773654937744,
+      "learning_rate": 9.84848484848485e-07,
+      "loss": 0.5723201036453247,
+      "mean_token_accuracy": 0.8100086450576782,
+      "num_tokens": 368507.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.03115501519756839,
+      "grad_norm": 4.4438347816467285,
+      "learning_rate": 1.01010101010101e-06,
+      "loss": 0.7508786916732788,
+      "mean_token_accuracy": 0.7711942791938782,
+      "num_tokens": 376467.0,
+      "step": 41
+    },
+    {
+      "epoch": 0.031914893617021274,
+      "grad_norm": 5.609974384307861,
+      "learning_rate": 1.0353535353535354e-06,
+      "loss": 0.566256046295166,
+      "mean_token_accuracy": 0.8319284319877625,
+      "num_tokens": 381399.0,
+      "step": 42
+    },
+    {
+      "epoch": 0.03267477203647416,
+      "grad_norm": 5.124386787414551,
+      "learning_rate": 1.0606060606060608e-06,
+      "loss": 0.8151067495346069,
+      "mean_token_accuracy": 0.7537785768508911,
+      "num_tokens": 387389.0,
+      "step": 43
+    },
+    {
+      "epoch": 0.03343465045592705,
+      "grad_norm": 3.6318116188049316,
+      "learning_rate": 1.085858585858586e-06,
+      "loss": 0.5989949107170105,
+      "mean_token_accuracy": 0.8129256963729858,
+      "num_tokens": 395302.0,
+      "step": 44
+    },
+    {
+      "epoch": 0.03419452887537994,
+      "grad_norm": 2.694424629211426,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 0.5831396579742432,
+      "mean_token_accuracy": 0.8056820631027222,
+      "num_tokens": 409920.0,
+      "step": 45
+    },
+    {
+      "epoch": 0.034954407294832825,
+      "grad_norm": 2.2949178218841553,
+      "learning_rate": 1.1363636363636364e-06,
+      "loss": 0.472550630569458,
+      "mean_token_accuracy": 0.8343006372451782,
+      "num_tokens": 428323.0,
+      "step": 46
+    },
+    {
+      "epoch": 0.03571428571428571,
+      "grad_norm": 3.3930575847625732,
+      "learning_rate": 1.1616161616161617e-06,
+      "loss": 0.6246505379676819,
+      "mean_token_accuracy": 0.783149003982544,
+      "num_tokens": 435889.0,
+      "step": 47
+    },
+    {
+      "epoch": 0.0364741641337386,
+      "grad_norm": 3.692598819732666,
+      "learning_rate": 1.186868686868687e-06,
+      "loss": 0.46132946014404297,
+      "mean_token_accuracy": 0.8583089113235474,
+      "num_tokens": 441192.0,
+      "step": 48
+    },
+    {
+      "epoch": 0.03723404255319149,
+      "grad_norm": 6.571533203125,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 0.9351121783256531,
+      "mean_token_accuracy": 0.7580878734588623,
+      "num_tokens": 444277.0,
+      "step": 49
+    },
+    {
+      "epoch": 0.037993920972644375,
+      "grad_norm": 5.029570579528809,
+      "learning_rate": 1.2373737373737375e-06,
+      "loss": 0.6921554803848267,
+      "mean_token_accuracy": 0.8131166100502014,
+      "num_tokens": 447646.0,
+      "step": 50
+    },
+    {
+      "epoch": 0.03875379939209726,
+      "grad_norm": 2.9174208641052246,
+      "learning_rate": 1.2626262626262629e-06,
+      "loss": 0.591706395149231,
+      "mean_token_accuracy": 0.8108617067337036,
+      "num_tokens": 461397.0,
+      "step": 51
+    },
+    {
+      "epoch": 0.03951367781155015,
+      "grad_norm": 4.315536022186279,
+      "learning_rate": 1.287878787878788e-06,
+      "loss": 0.6986310482025146,
+      "mean_token_accuracy": 0.7710754871368408,
+      "num_tokens": 472047.0,
+      "step": 52
+    },
+    {
+      "epoch": 0.04027355623100304,
+      "grad_norm": 2.6216275691986084,
+      "learning_rate": 1.3131313131313134e-06,
+      "loss": 0.5553690791130066,
+      "mean_token_accuracy": 0.8167896866798401,
+      "num_tokens": 482795.0,
+      "step": 53
+    },
+    {
+      "epoch": 0.041033434650455926,
+      "grad_norm": 3.0562477111816406,
+      "learning_rate": 1.3383838383838385e-06,
+      "loss": 0.6909202337265015,
+      "mean_token_accuracy": 0.7859863638877869,
+      "num_tokens": 494818.0,
+      "step": 54
+    },
+    {
+      "epoch": 0.04179331306990881,
+      "grad_norm": 2.1420412063598633,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 0.5415265560150146,
+      "mean_token_accuracy": 0.818886399269104,
+      "num_tokens": 513695.0,
+      "step": 55
+    },
+    {
+      "epoch": 0.0425531914893617,
+      "grad_norm": 2.9610488414764404,
+      "learning_rate": 1.3888888888888892e-06,
+      "loss": 0.6602212190628052,
+      "mean_token_accuracy": 0.7830734252929688,
+      "num_tokens": 523784.0,
+      "step": 56
+    },
+    {
+      "epoch": 0.04331306990881459,
+      "grad_norm": 2.511972665786743,
+      "learning_rate": 1.4141414141414143e-06,
+      "loss": 0.5717809796333313,
+      "mean_token_accuracy": 0.8053616285324097,
+      "num_tokens": 546308.0,
+      "step": 57
+    },
+    {
+      "epoch": 0.044072948328267476,
+      "grad_norm": 3.52642822265625,
+      "learning_rate": 1.4393939393939396e-06,
+      "loss": 0.6242594718933105,
+      "mean_token_accuracy": 0.8162082433700562,
+      "num_tokens": 552019.0,
+      "step": 58
+    },
+    {
+      "epoch": 0.044832826747720364,
+      "grad_norm": 3.02362322807312,
+      "learning_rate": 1.4646464646464648e-06,
+      "loss": 0.6634255647659302,
+      "mean_token_accuracy": 0.7682032585144043,
+      "num_tokens": 560009.0,
+      "step": 59
+    },
+    {
+      "epoch": 0.04559270516717325,
+      "grad_norm": 2.3910107612609863,
+      "learning_rate": 1.48989898989899e-06,
+      "loss": 0.5519146919250488,
+      "mean_token_accuracy": 0.8270269632339478,
+      "num_tokens": 571005.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.04635258358662614,
+      "grad_norm": 4.28154993057251,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 0.7437789440155029,
+      "mean_token_accuracy": 0.7782418131828308,
+      "num_tokens": 574950.0,
+      "step": 61
+    },
+    {
+      "epoch": 0.04711246200607903,
+      "grad_norm": 3.4078686237335205,
+      "learning_rate": 1.5404040404040404e-06,
+      "loss": 0.6345915198326111,
+      "mean_token_accuracy": 0.7903392314910889,
+      "num_tokens": 581657.0,
+      "step": 62
+    },
+    {
+      "epoch": 0.047872340425531915,
+      "grad_norm": 2.6834158897399902,
+      "learning_rate": 1.565656565656566e-06,
+      "loss": 0.5981127023696899,
+      "mean_token_accuracy": 0.7911489605903625,
+      "num_tokens": 591267.0,
+      "step": 63
+    },
+    {
+      "epoch": 0.0486322188449848,
+      "grad_norm": 2.1054461002349854,
+      "learning_rate": 1.590909090909091e-06,
+      "loss": 0.5523523688316345,
+      "mean_token_accuracy": 0.8194501399993896,
+      "num_tokens": 606787.0,
+      "step": 64
+    },
+    {
+      "epoch": 0.04939209726443769,
+      "grad_norm": 3.322596788406372,
+      "learning_rate": 1.6161616161616164e-06,
+      "loss": 0.48417025804519653,
+      "mean_token_accuracy": 0.8293706178665161,
+      "num_tokens": 611068.0,
+      "step": 65
+    },
+    {
+      "epoch": 0.05015197568389058,
+      "grad_norm": 2.302450180053711,
+      "learning_rate": 1.6414141414141415e-06,
+      "loss": 0.6498389840126038,
+      "mean_token_accuracy": 0.7728497385978699,
+      "num_tokens": 624452.0,
+      "step": 66
+    },
+    {
+      "epoch": 0.050911854103343465,
+      "grad_norm": 2.680191993713379,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.6347037553787231,
+      "mean_token_accuracy": 0.8108306527137756,
+      "num_tokens": 638049.0,
+      "step": 67
+    },
+    {
+      "epoch": 0.05167173252279635,
+      "grad_norm": 3.0297021865844727,
+      "learning_rate": 1.6919191919191922e-06,
+      "loss": 0.5344363451004028,
+      "mean_token_accuracy": 0.8113535046577454,
+      "num_tokens": 643892.0,
+      "step": 68
+    },
+    {
+      "epoch": 0.05243161094224924,
+      "grad_norm": 2.9283676147460938,
+      "learning_rate": 1.7171717171717173e-06,
+      "loss": 0.6999260187149048,
+      "mean_token_accuracy": 0.7782022356987,
+      "num_tokens": 654418.0,
+      "step": 69
+    },
+    {
+      "epoch": 0.05319148936170213,
+      "grad_norm": 3.4098572731018066,
+      "learning_rate": 1.7424242424242427e-06,
+      "loss": 0.6508946418762207,
+      "mean_token_accuracy": 0.7942900657653809,
+      "num_tokens": 659837.0,
+      "step": 70
+    },
+    {
+      "epoch": 0.053951367781155016,
+      "grad_norm": 2.6756019592285156,
+      "learning_rate": 1.7676767676767678e-06,
+      "loss": 0.603486180305481,
+      "mean_token_accuracy": 0.8015457391738892,
+      "num_tokens": 668361.0,
+      "step": 71
+    },
+    {
+      "epoch": 0.0547112462006079,
+      "grad_norm": 2.2630293369293213,
+      "learning_rate": 1.792929292929293e-06,
+      "loss": 0.6608274579048157,
+      "mean_token_accuracy": 0.7753809690475464,
+      "num_tokens": 679025.0,
+      "step": 72
+    },
+    {
+      "epoch": 0.05547112462006079,
+      "grad_norm": 2.123962879180908,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 0.4525482654571533,
+      "mean_token_accuracy": 0.8425612449645996,
+      "num_tokens": 688574.0,
+      "step": 73
+    },
+    {
+      "epoch": 0.05623100303951368,
+      "grad_norm": 7.90519905090332,
+      "learning_rate": 1.8434343434343434e-06,
+      "loss": 0.6507195830345154,
+      "mean_token_accuracy": 0.7714964151382446,
+      "num_tokens": 694534.0,
+      "step": 74
+    },
+    {
+      "epoch": 0.056990881458966566,
+      "grad_norm": 2.372203826904297,
+      "learning_rate": 1.868686868686869e-06,
+      "loss": 0.4458143413066864,
+      "mean_token_accuracy": 0.7991449236869812,
+      "num_tokens": 703114.0,
+      "step": 75
+    },
+    {
+      "epoch": 0.057750759878419454,
+      "grad_norm": 2.918677568435669,
+      "learning_rate": 1.8939393939393941e-06,
+      "loss": 0.5614339113235474,
+      "mean_token_accuracy": 0.8211464881896973,
+      "num_tokens": 709038.0,
+      "step": 76
+    },
+    {
+      "epoch": 0.05851063829787234,
+      "grad_norm": 1.6106709241867065,
+      "learning_rate": 1.9191919191919192e-06,
+      "loss": 0.5802098512649536,
+      "mean_token_accuracy": 0.8055065870285034,
+      "num_tokens": 730482.0,
+      "step": 77
+    },
+    {
+      "epoch": 0.05927051671732523,
+      "grad_norm": 2.8069989681243896,
+      "learning_rate": 1.944444444444445e-06,
+      "loss": 0.5709059238433838,
+      "mean_token_accuracy": 0.8024872541427612,
+      "num_tokens": 751817.0,
+      "step": 78
+    },
+    {
+      "epoch": 0.06003039513677812,
+      "grad_norm": 2.641667127609253,
+      "learning_rate": 1.96969696969697e-06,
+      "loss": 0.6480152606964111,
+      "mean_token_accuracy": 0.7912271618843079,
+      "num_tokens": 759236.0,
+      "step": 79
+    },
+    {
+      "epoch": 0.060790273556231005,
+      "grad_norm": 2.6034350395202637,
+      "learning_rate": 1.994949494949495e-06,
+      "loss": 0.5535176396369934,
+      "mean_token_accuracy": 0.7980542778968811,
+      "num_tokens": 766496.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.06155015197568389,
+      "grad_norm": 1.7095069885253906,
+      "learning_rate": 2.02020202020202e-06,
+      "loss": 0.4545496106147766,
+      "mean_token_accuracy": 0.8229660391807556,
+      "num_tokens": 780124.0,
+      "step": 81
+    },
+    {
+      "epoch": 0.06231003039513678,
+      "grad_norm": 3.788830518722534,
+      "learning_rate": 2.0454545454545457e-06,
+      "loss": 0.6679391264915466,
+      "mean_token_accuracy": 0.7942397594451904,
+      "num_tokens": 784555.0,
+      "step": 82
+    },
+    {
+      "epoch": 0.06306990881458967,
+      "grad_norm": 2.009831666946411,
+      "learning_rate": 2.070707070707071e-06,
+      "loss": 0.5067101120948792,
+      "mean_token_accuracy": 0.8276634216308594,
+      "num_tokens": 797459.0,
+      "step": 83
+    },
+    {
+      "epoch": 0.06382978723404255,
+      "grad_norm": 2.201627731323242,
+      "learning_rate": 2.095959595959596e-06,
+      "loss": 0.5012127161026001,
+      "mean_token_accuracy": 0.8432504534721375,
+      "num_tokens": 810817.0,
+      "step": 84
+    },
+    {
+      "epoch": 0.06458966565349544,
+      "grad_norm": 2.492568016052246,
+      "learning_rate": 2.1212121212121216e-06,
+      "loss": 0.6142797470092773,
+      "mean_token_accuracy": 0.8338661193847656,
+      "num_tokens": 818191.0,
+      "step": 85
+    },
+    {
+      "epoch": 0.06534954407294832,
+      "grad_norm": 2.8360862731933594,
+      "learning_rate": 2.1464646464646467e-06,
+      "loss": 0.5569300651550293,
+      "mean_token_accuracy": 0.8121030330657959,
+      "num_tokens": 825325.0,
+      "step": 86
+    },
+    {
+      "epoch": 0.06610942249240122,
+      "grad_norm": 2.407548427581787,
+      "learning_rate": 2.171717171717172e-06,
+      "loss": 0.6442930102348328,
+      "mean_token_accuracy": 0.792514443397522,
+      "num_tokens": 834439.0,
+      "step": 87
+    },
+    {
+      "epoch": 0.0668693009118541,
+      "grad_norm": 2.340728759765625,
+      "learning_rate": 2.196969696969697e-06,
+      "loss": 0.6494365930557251,
+      "mean_token_accuracy": 0.7746615409851074,
+      "num_tokens": 843078.0,
+      "step": 88
+    },
+    {
+      "epoch": 0.067629179331307,
+      "grad_norm": 1.7703697681427002,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 0.598991870880127,
+      "mean_token_accuracy": 0.7992157340049744,
+      "num_tokens": 860171.0,
+      "step": 89
+    },
+    {
+      "epoch": 0.06838905775075987,
+      "grad_norm": 2.5779271125793457,
+      "learning_rate": 2.2474747474747476e-06,
+      "loss": 0.5693082809448242,
+      "mean_token_accuracy": 0.8093700408935547,
+      "num_tokens": 866669.0,
+      "step": 90
+    },
+    {
+      "epoch": 0.06914893617021277,
+      "grad_norm": 2.014092206954956,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 0.5346695780754089,
+      "mean_token_accuracy": 0.8165590763092041,
+      "num_tokens": 876698.0,
+      "step": 91
+    },
+    {
+      "epoch": 0.06990881458966565,
+      "grad_norm": 1.7555919885635376,
+      "learning_rate": 2.2979797979797983e-06,
+      "loss": 0.5321458578109741,
+      "mean_token_accuracy": 0.8166656494140625,
+      "num_tokens": 889488.0,
+      "step": 92
+    },
+    {
+      "epoch": 0.07066869300911854,
+      "grad_norm": 1.8631824254989624,
+      "learning_rate": 2.3232323232323234e-06,
+      "loss": 0.5246532559394836,
+      "mean_token_accuracy": 0.8088107705116272,
+      "num_tokens": 901322.0,
+      "step": 93
+    },
+    {
+      "epoch": 0.07142857142857142,
+      "grad_norm": 3.2332139015197754,
+      "learning_rate": 2.348484848484849e-06,
+      "loss": 0.5141711235046387,
+      "mean_token_accuracy": 0.8382217884063721,
+      "num_tokens": 905792.0,
+      "step": 94
+    },
+    {
+      "epoch": 0.07218844984802432,
+      "grad_norm": 1.7806555032730103,
+      "learning_rate": 2.373737373737374e-06,
+      "loss": 0.5233149528503418,
+      "mean_token_accuracy": 0.8101529479026794,
+      "num_tokens": 917320.0,
+      "step": 95
+    },
+    {
+      "epoch": 0.0729483282674772,
+      "grad_norm": 1.8169859647750854,
+      "learning_rate": 2.3989898989898993e-06,
+      "loss": 0.578881561756134,
+      "mean_token_accuracy": 0.8044873476028442,
+      "num_tokens": 931062.0,
+      "step": 96
+    },
+    {
+      "epoch": 0.0737082066869301,
+      "grad_norm": 4.677402496337891,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 0.7842556238174438,
+      "mean_token_accuracy": 0.7579764127731323,
+      "num_tokens": 934712.0,
+      "step": 97
+    },
+    {
+      "epoch": 0.07446808510638298,
+      "grad_norm": 2.6987264156341553,
+      "learning_rate": 2.4494949494949495e-06,
+      "loss": 0.5669287443161011,
+      "mean_token_accuracy": 0.8186933994293213,
+      "num_tokens": 941058.0,
+      "step": 98
+    },
+    {
+      "epoch": 0.07522796352583587,
+      "grad_norm": 1.6906023025512695,
+      "learning_rate": 2.474747474747475e-06,
+      "loss": 0.4976363778114319,
+      "mean_token_accuracy": 0.8198553323745728,
+      "num_tokens": 956509.0,
+      "step": 99
+    },
+    {
+      "epoch": 0.07598784194528875,
+      "grad_norm": 2.7256152629852295,
+      "learning_rate": 2.5e-06,
+      "loss": 0.7138420343399048,
+      "mean_token_accuracy": 0.7752805948257446,
+      "num_tokens": 963920.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.07674772036474165,
+      "grad_norm": 2.174870491027832,
+      "learning_rate": 2.5252525252525258e-06,
+      "loss": 0.6733541488647461,
+      "mean_token_accuracy": 0.7745175361633301,
+      "num_tokens": 975268.0,
+      "step": 101
+    },
+    {
+      "epoch": 0.07750759878419453,
+      "grad_norm": 1.5587213039398193,
+      "learning_rate": 2.5505050505050505e-06,
+      "loss": 0.44223445653915405,
+      "mean_token_accuracy": 0.8278359174728394,
+      "num_tokens": 991837.0,
+      "step": 102
+    },
+    {
+      "epoch": 0.07826747720364742,
+      "grad_norm": 2.181840658187866,
+      "learning_rate": 2.575757575757576e-06,
+      "loss": 0.625128448009491,
+      "mean_token_accuracy": 0.7941786050796509,
+      "num_tokens": 1004325.0,
+      "step": 103
+    },
+    {
+      "epoch": 0.0790273556231003,
+      "grad_norm": 1.4986687898635864,
+      "learning_rate": 2.601010101010101e-06,
+      "loss": 0.39262527227401733,
+      "mean_token_accuracy": 0.8412648439407349,
+      "num_tokens": 1018331.0,
+      "step": 104
+    },
+    {
+      "epoch": 0.0797872340425532,
+      "grad_norm": 2.3416061401367188,
+      "learning_rate": 2.6262626262626267e-06,
+      "loss": 0.5495132803916931,
+      "mean_token_accuracy": 0.8193322420120239,
+      "num_tokens": 1026090.0,
+      "step": 105
+    },
+    {
+      "epoch": 0.08054711246200608,
+      "grad_norm": 3.8168859481811523,
+      "learning_rate": 2.6515151515151514e-06,
+      "loss": 0.4898706376552582,
+      "mean_token_accuracy": 0.8467956185340881,
+      "num_tokens": 1029955.0,
+      "step": 106
+    },
+    {
+      "epoch": 0.08130699088145897,
+      "grad_norm": 4.113908767700195,
+      "learning_rate": 2.676767676767677e-06,
+      "loss": 0.6189584732055664,
+      "mean_token_accuracy": 0.8019394278526306,
+      "num_tokens": 1033598.0,
+      "step": 107
+    },
+    {
+      "epoch": 0.08206686930091185,
+      "grad_norm": 2.50003981590271,
+      "learning_rate": 2.7020202020202025e-06,
+      "loss": 0.6479471921920776,
+      "mean_token_accuracy": 0.7790026664733887,
+      "num_tokens": 1042533.0,
+      "step": 108
+    },
+    {
+      "epoch": 0.08282674772036475,
+      "grad_norm": 1.408934473991394,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.3909248113632202,
+      "mean_token_accuracy": 0.8477586507797241,
+      "num_tokens": 1061755.0,
+      "step": 109
+    },
+    {
+      "epoch": 0.08358662613981763,
+      "grad_norm": 3.360633611679077,
+      "learning_rate": 2.7525252525252528e-06,
+      "loss": 0.6952459812164307,
+      "mean_token_accuracy": 0.777535080909729,
+      "num_tokens": 1067316.0,
+      "step": 110
+    },
+    {
+      "epoch": 0.08434650455927052,
+      "grad_norm": 1.8631696701049805,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 0.5420593023300171,
+      "mean_token_accuracy": 0.8157662749290466,
+      "num_tokens": 1079930.0,
+      "step": 111
+    },
+    {
+      "epoch": 0.0851063829787234,
+      "grad_norm": 2.4308314323425293,
+      "learning_rate": 2.803030303030303e-06,
+      "loss": 0.5863882303237915,
+      "mean_token_accuracy": 0.8206346035003662,
+      "num_tokens": 1088069.0,
+      "step": 112
+    },
+    {
+      "epoch": 0.0858662613981763,
+      "grad_norm": 2.922808885574341,
+      "learning_rate": 2.8282828282828286e-06,
+      "loss": 0.5217319130897522,
+      "mean_token_accuracy": 0.8253234028816223,
+      "num_tokens": 1093607.0,
+      "step": 113
+    },
+    {
+      "epoch": 0.08662613981762918,
+      "grad_norm": 2.3596107959747314,
+      "learning_rate": 2.8535353535353537e-06,
+      "loss": 0.5070714950561523,
+      "mean_token_accuracy": 0.8258323669433594,
+      "num_tokens": 1100405.0,
+      "step": 114
+    },
+    {
+      "epoch": 0.08738601823708207,
+      "grad_norm": 3.0853066444396973,
+      "learning_rate": 2.8787878787878793e-06,
+      "loss": 0.591964840888977,
+      "mean_token_accuracy": 0.8047322630882263,
+      "num_tokens": 1107535.0,
+      "step": 115
+    },
+    {
+      "epoch": 0.08814589665653495,
+      "grad_norm": 1.9251092672348022,
+      "learning_rate": 2.904040404040404e-06,
+      "loss": 0.5226191878318787,
+      "mean_token_accuracy": 0.8022720217704773,
+      "num_tokens": 1118716.0,
+      "step": 116
+    },
+    {
+      "epoch": 0.08890577507598785,
+      "grad_norm": 1.9692988395690918,
+      "learning_rate": 2.9292929292929295e-06,
+      "loss": 0.5462069511413574,
+      "mean_token_accuracy": 0.8157015442848206,
+      "num_tokens": 1131917.0,
+      "step": 117
+    },
+    {
+      "epoch": 0.08966565349544073,
+      "grad_norm": 1.4738909006118774,
+      "learning_rate": 2.954545454545455e-06,
+      "loss": 0.4564219117164612,
+      "mean_token_accuracy": 0.849632978439331,
+      "num_tokens": 1148534.0,
+      "step": 118
+    },
+    {
+      "epoch": 0.09042553191489362,
+      "grad_norm": 2.72646164894104,
+      "learning_rate": 2.97979797979798e-06,
+      "loss": 0.6654808521270752,
+      "mean_token_accuracy": 0.7752684354782104,
+      "num_tokens": 1155438.0,
+      "step": 119
+    },
+    {
+      "epoch": 0.0911854103343465,
+      "grad_norm": 2.7843852043151855,
+      "learning_rate": 3.0050505050505054e-06,
+      "loss": 0.5354680418968201,
+      "mean_token_accuracy": 0.8196378946304321,
+      "num_tokens": 1161815.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.0919452887537994,
+      "grad_norm": 2.8052573204040527,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.6366757154464722,
+      "mean_token_accuracy": 0.7967483997344971,
+      "num_tokens": 1168295.0,
+      "step": 121
+    },
+    {
+      "epoch": 0.09270516717325228,
+      "grad_norm": 2.7462735176086426,
+      "learning_rate": 3.055555555555556e-06,
+      "loss": 0.59470534324646,
+      "mean_token_accuracy": 0.8023771047592163,
+      "num_tokens": 1174502.0,
+      "step": 122
+    },
+    {
+      "epoch": 0.09346504559270517,
+      "grad_norm": 2.2743821144104004,
+      "learning_rate": 3.0808080808080807e-06,
+      "loss": 0.5720560550689697,
+      "mean_token_accuracy": 0.8162771463394165,
+      "num_tokens": 1183615.0,
+      "step": 123
+    },
+    {
+      "epoch": 0.09422492401215805,
+      "grad_norm": 1.8669533729553223,
+      "learning_rate": 3.1060606060606063e-06,
+      "loss": 0.4655378758907318,
+      "mean_token_accuracy": 0.8360732793807983,
+      "num_tokens": 1193761.0,
+      "step": 124
+    },
+    {
+      "epoch": 0.09498480243161095,
+      "grad_norm": 1.7666901350021362,
+      "learning_rate": 3.131313131313132e-06,
+      "loss": 0.5524153709411621,
+      "mean_token_accuracy": 0.8252713680267334,
+      "num_tokens": 1207870.0,
+      "step": 125
+    },
+    {
+      "epoch": 0.09574468085106383,
+      "grad_norm": 2.4720070362091064,
+      "learning_rate": 3.1565656565656566e-06,
+      "loss": 0.5003011226654053,
+      "mean_token_accuracy": 0.8491042852401733,
+      "num_tokens": 1214603.0,
+      "step": 126
+    },
+    {
+      "epoch": 0.09650455927051672,
+      "grad_norm": 1.6500422954559326,
+      "learning_rate": 3.181818181818182e-06,
+      "loss": 0.5137069225311279,
+      "mean_token_accuracy": 0.8273531198501587,
+      "num_tokens": 1228717.0,
+      "step": 127
+    },
+    {
+      "epoch": 0.0972644376899696,
+      "grad_norm": 3.402543067932129,
+      "learning_rate": 3.2070707070707072e-06,
+      "loss": 0.708167552947998,
+      "mean_token_accuracy": 0.7705385684967041,
+      "num_tokens": 1234361.0,
+      "step": 128
+    },
+    {
+      "epoch": 0.0980243161094225,
+      "grad_norm": 2.547285795211792,
+      "learning_rate": 3.232323232323233e-06,
+      "loss": 0.6020137071609497,
+      "mean_token_accuracy": 0.7981340289115906,
+      "num_tokens": 1244169.0,
+      "step": 129
+    },
+    {
+      "epoch": 0.09878419452887538,
+      "grad_norm": 2.0578792095184326,
+      "learning_rate": 3.257575757575758e-06,
+      "loss": 0.4425000250339508,
+      "mean_token_accuracy": 0.8567807674407959,
+      "num_tokens": 1252709.0,
+      "step": 130
+    },
+    {
+      "epoch": 0.09954407294832827,
+      "grad_norm": 1.672614336013794,
+      "learning_rate": 3.282828282828283e-06,
+      "loss": 0.4860966205596924,
+      "mean_token_accuracy": 0.8393139243125916,
+      "num_tokens": 1265766.0,
+      "step": 131
+    },
+    {
+      "epoch": 0.10030395136778116,
+      "grad_norm": 3.2560198307037354,
+      "learning_rate": 3.3080808080808086e-06,
+      "loss": 0.624736487865448,
+      "mean_token_accuracy": 0.7875322699546814,
+      "num_tokens": 1270779.0,
+      "step": 132
+    },
+    {
+      "epoch": 0.10106382978723404,
+      "grad_norm": 2.4468185901641846,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.5062227249145508,
+      "mean_token_accuracy": 0.8217229843139648,
+      "num_tokens": 1277113.0,
+      "step": 133
+    },
+    {
+      "epoch": 0.10182370820668693,
+      "grad_norm": 2.6371328830718994,
+      "learning_rate": 3.358585858585859e-06,
+      "loss": 0.477113276720047,
+      "mean_token_accuracy": 0.8605583906173706,
+      "num_tokens": 1282514.0,
+      "step": 134
+    },
+    {
+      "epoch": 0.10258358662613981,
+      "grad_norm": 2.48421311378479,
+      "learning_rate": 3.3838383838383844e-06,
+      "loss": 0.40855684876441956,
+      "mean_token_accuracy": 0.864548921585083,
+      "num_tokens": 1287859.0,
+      "step": 135
+    },
+    {
+      "epoch": 0.1033434650455927,
+      "grad_norm": 1.993099331855774,
+      "learning_rate": 3.409090909090909e-06,
+      "loss": 0.5913145542144775,
+      "mean_token_accuracy": 0.8248485922813416,
+      "num_tokens": 1301074.0,
+      "step": 136
+    },
+    {
+      "epoch": 0.10410334346504559,
+      "grad_norm": 3.5947680473327637,
+      "learning_rate": 3.4343434343434347e-06,
+      "loss": 0.5028599500656128,
+      "mean_token_accuracy": 0.8367215394973755,
+      "num_tokens": 1305219.0,
+      "step": 137
+    },
+    {
+      "epoch": 0.10486322188449848,
+      "grad_norm": 2.5778582096099854,
+      "learning_rate": 3.45959595959596e-06,
+      "loss": 0.5297672748565674,
+      "mean_token_accuracy": 0.8232187032699585,
+      "num_tokens": 1312482.0,
+      "step": 138
+    },
+    {
+      "epoch": 0.10562310030395136,
+      "grad_norm": 1.8961588144302368,
+      "learning_rate": 3.4848484848484854e-06,
+      "loss": 0.39954107999801636,
+      "mean_token_accuracy": 0.8605833053588867,
+      "num_tokens": 1323404.0,
+      "step": 139
+    },
+    {
+      "epoch": 0.10638297872340426,
+      "grad_norm": 1.9687960147857666,
+      "learning_rate": 3.51010101010101e-06,
+      "loss": 0.48791587352752686,
+      "mean_token_accuracy": 0.8200347423553467,
+      "num_tokens": 1333027.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.10714285714285714,
+      "grad_norm": 2.520242691040039,
+      "learning_rate": 3.5353535353535356e-06,
+      "loss": 0.6106002330780029,
+      "mean_token_accuracy": 0.790692150592804,
+      "num_tokens": 1340999.0,
+      "step": 141
+    },
+    {
+      "epoch": 0.10790273556231003,
+      "grad_norm": 3.751617431640625,
+      "learning_rate": 3.560606060606061e-06,
+      "loss": 0.48141729831695557,
+      "mean_token_accuracy": 0.8421382904052734,
+      "num_tokens": 1344687.0,
+      "step": 142
+    },
+    {
+      "epoch": 0.10866261398176291,
+      "grad_norm": 2.7101709842681885,
+      "learning_rate": 3.585858585858586e-06,
+      "loss": 0.5375241637229919,
+      "mean_token_accuracy": 0.8061438202857971,
+      "num_tokens": 1350192.0,
+      "step": 143
+    },
+    {
+      "epoch": 0.1094224924012158,
+      "grad_norm": 2.583484411239624,
+      "learning_rate": 3.6111111111111115e-06,
+      "loss": 0.6492470502853394,
+      "mean_token_accuracy": 0.7863001823425293,
+      "num_tokens": 1358148.0,
+      "step": 144
+    },
+    {
+      "epoch": 0.11018237082066869,
+      "grad_norm": 1.792561650276184,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.48480600118637085,
+      "mean_token_accuracy": 0.8358709812164307,
+      "num_tokens": 1369519.0,
+      "step": 145
+    },
+    {
+      "epoch": 0.11094224924012158,
+      "grad_norm": 2.6480472087860107,
+      "learning_rate": 3.661616161616162e-06,
+      "loss": 0.5268933176994324,
+      "mean_token_accuracy": 0.8214013576507568,
+      "num_tokens": 1375862.0,
+      "step": 146
+    },
+    {
+      "epoch": 0.11170212765957446,
+      "grad_norm": 2.3174469470977783,
+      "learning_rate": 3.686868686868687e-06,
+      "loss": 0.42517897486686707,
+      "mean_token_accuracy": 0.8523461222648621,
+      "num_tokens": 1381546.0,
+      "step": 147
+    },
+    {
+      "epoch": 0.11246200607902736,
+      "grad_norm": 3.0090949535369873,
+      "learning_rate": 3.7121212121212124e-06,
+      "loss": 0.4042336940765381,
+      "mean_token_accuracy": 0.8670448064804077,
+      "num_tokens": 1385896.0,
+      "step": 148
+    },
+    {
+      "epoch": 0.11322188449848024,
+      "grad_norm": 2.4928104877471924,
+      "learning_rate": 3.737373737373738e-06,
+      "loss": 0.6498878598213196,
+      "mean_token_accuracy": 0.7967068552970886,
+      "num_tokens": 1394169.0,
+      "step": 149
+    },
+    {
+      "epoch": 0.11398176291793313,
+      "grad_norm": 1.5984913110733032,
+      "learning_rate": 3.7626262626262627e-06,
+      "loss": 0.546096920967102,
+      "mean_token_accuracy": 0.8035850524902344,
+      "num_tokens": 1408785.0,
+      "step": 150
+    },
+    {
+      "epoch": 0.11474164133738601,
+      "grad_norm": 2.3663532733917236,
+      "learning_rate": 3.7878787878787882e-06,
+      "loss": 0.6111721992492676,
+      "mean_token_accuracy": 0.8015355467796326,
+      "num_tokens": 1417510.0,
+      "step": 151
+    },
+    {
+      "epoch": 0.11550151975683891,
+      "grad_norm": 2.518932819366455,
+      "learning_rate": 3.8131313131313138e-06,
+      "loss": 0.5274964570999146,
+      "mean_token_accuracy": 0.8155480623245239,
+      "num_tokens": 1424186.0,
+      "step": 152
+    },
+    {
+      "epoch": 0.11626139817629179,
+      "grad_norm": 2.14353609085083,
+      "learning_rate": 3.8383838383838385e-06,
+      "loss": 0.5283297896385193,
+      "mean_token_accuracy": 0.8275758028030396,
+      "num_tokens": 1432630.0,
+      "step": 153
+    },
+    {
+      "epoch": 0.11702127659574468,
+      "grad_norm": 1.8243604898452759,
+      "learning_rate": 3.863636363636364e-06,
+      "loss": 0.41854870319366455,
+      "mean_token_accuracy": 0.8222295045852661,
+      "num_tokens": 1442691.0,
+      "step": 154
+    },
+    {
+      "epoch": 0.11778115501519756,
+      "grad_norm": 2.088212251663208,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 0.6062943339347839,
+      "mean_token_accuracy": 0.8009427785873413,
+      "num_tokens": 1456890.0,
+      "step": 155
+    },
+    {
+      "epoch": 0.11854103343465046,
+      "grad_norm": 1.3469511270523071,
+      "learning_rate": 3.914141414141415e-06,
+      "loss": 0.4390433728694916,
+      "mean_token_accuracy": 0.8436295986175537,
+      "num_tokens": 1475349.0,
+      "step": 156
+    },
+    {
+      "epoch": 0.11930091185410334,
+      "grad_norm": 3.247023105621338,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 0.6490433216094971,
+      "mean_token_accuracy": 0.8037861585617065,
+      "num_tokens": 1479952.0,
+      "step": 157
+    },
+    {
+      "epoch": 0.12006079027355623,
+      "grad_norm": 2.6610445976257324,
+      "learning_rate": 3.964646464646465e-06,
+      "loss": 0.6221826076507568,
+      "mean_token_accuracy": 0.7848749160766602,
+      "num_tokens": 1487306.0,
+      "step": 158
+    },
+    {
+      "epoch": 0.12082066869300911,
+      "grad_norm": 2.3060810565948486,
+      "learning_rate": 3.98989898989899e-06,
+      "loss": 0.5052388310432434,
+      "mean_token_accuracy": 0.8281195759773254,
+      "num_tokens": 1495367.0,
+      "step": 159
+    },
+    {
+      "epoch": 0.12158054711246201,
+      "grad_norm": 2.504448652267456,
+      "learning_rate": 4.015151515151515e-06,
+      "loss": 0.5005477666854858,
+      "mean_token_accuracy": 0.8408058881759644,
+      "num_tokens": 1502069.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.12234042553191489,
+      "grad_norm": 3.993938446044922,
+      "learning_rate": 4.04040404040404e-06,
+      "loss": 0.5569638013839722,
+      "mean_token_accuracy": 0.8095242977142334,
+      "num_tokens": 1510224.0,
+      "step": 161
+    },
+    {
+      "epoch": 0.12310030395136778,
+      "grad_norm": 2.2287683486938477,
+      "learning_rate": 4.065656565656566e-06,
+      "loss": 0.524042546749115,
+      "mean_token_accuracy": 0.8102203607559204,
+      "num_tokens": 1518364.0,
+      "step": 162
+    },
+    {
+      "epoch": 0.12386018237082067,
+      "grad_norm": 1.9531738758087158,
+      "learning_rate": 4.0909090909090915e-06,
+      "loss": 0.45794573426246643,
+      "mean_token_accuracy": 0.8560376167297363,
+      "num_tokens": 1528097.0,
+      "step": 163
+    },
+    {
+      "epoch": 0.12462006079027356,
+      "grad_norm": 1.5841206312179565,
+      "learning_rate": 4.116161616161617e-06,
+      "loss": 0.5420972108840942,
+      "mean_token_accuracy": 0.8092726469039917,
+      "num_tokens": 1544119.0,
+      "step": 164
+    },
+    {
+      "epoch": 0.12537993920972645,
+      "grad_norm": 1.7536218166351318,
+      "learning_rate": 4.141414141414142e-06,
+      "loss": 0.554668664932251,
+      "mean_token_accuracy": 0.8193825483322144,
+      "num_tokens": 1559140.0,
+      "step": 165
+    },
+    {
+      "epoch": 0.12613981762917933,
+      "grad_norm": 3.545454740524292,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.580947995185852,
+      "mean_token_accuracy": 0.8286383152008057,
+      "num_tokens": 1563625.0,
+      "step": 166
+    },
+    {
+      "epoch": 0.12689969604863222,
+      "grad_norm": 1.6608915328979492,
+      "learning_rate": 4.191919191919192e-06,
+      "loss": 0.5523324012756348,
+      "mean_token_accuracy": 0.8155215978622437,
+      "num_tokens": 1574945.0,
+      "step": 167
+    },
+    {
+      "epoch": 0.1276595744680851,
+      "grad_norm": 1.4832708835601807,
+      "learning_rate": 4.217171717171717e-06,
+      "loss": 0.5133191347122192,
+      "mean_token_accuracy": 0.8367571830749512,
+      "num_tokens": 1595865.0,
+      "step": 168
+    },
+    {
+      "epoch": 0.128419452887538,
+      "grad_norm": 1.7807520627975464,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 0.5131410360336304,
+      "mean_token_accuracy": 0.8129367232322693,
+      "num_tokens": 1608723.0,
+      "step": 169
+    },
+    {
+      "epoch": 0.12917933130699089,
+      "grad_norm": 2.707569122314453,
+      "learning_rate": 4.267676767676767e-06,
+      "loss": 0.6129013299942017,
+      "mean_token_accuracy": 0.7926048040390015,
+      "num_tokens": 1616136.0,
+      "step": 170
+    },
+    {
+      "epoch": 0.12993920972644377,
+      "grad_norm": 2.5831644535064697,
+      "learning_rate": 4.292929292929293e-06,
+      "loss": 0.6264227628707886,
+      "mean_token_accuracy": 0.8074911236763,
+      "num_tokens": 1624228.0,
+      "step": 171
+    },
+    {
+      "epoch": 0.13069908814589665,
+      "grad_norm": 3.1124250888824463,
+      "learning_rate": 4.3181818181818185e-06,
+      "loss": 0.41763827204704285,
+      "mean_token_accuracy": 0.8565453290939331,
+      "num_tokens": 1628098.0,
+      "step": 172
+    },
+    {
+      "epoch": 0.13145896656534956,
+      "grad_norm": 2.3214211463928223,
+      "learning_rate": 4.343434343434344e-06,
+      "loss": 0.421974778175354,
+      "mean_token_accuracy": 0.8391546010971069,
+      "num_tokens": 1634950.0,
+      "step": 173
+    },
+    {
+      "epoch": 0.13221884498480244,
+      "grad_norm": 2.1010327339172363,
+      "learning_rate": 4.368686868686869e-06,
+      "loss": 0.5307331681251526,
+      "mean_token_accuracy": 0.8139588236808777,
+      "num_tokens": 1644132.0,
+      "step": 174
+    },
+    {
+      "epoch": 0.13297872340425532,
+      "grad_norm": 2.533612012863159,
+      "learning_rate": 4.393939393939394e-06,
+      "loss": 0.5626664161682129,
+      "mean_token_accuracy": 0.8029808402061462,
+      "num_tokens": 1651637.0,
+      "step": 175
+    },
+    {
+      "epoch": 0.1337386018237082,
+      "grad_norm": 1.669508457183838,
+      "learning_rate": 4.41919191919192e-06,
+      "loss": 0.5351508259773254,
+      "mean_token_accuracy": 0.8281655311584473,
+      "num_tokens": 1666776.0,
+      "step": 176
+    },
+    {
+      "epoch": 0.1344984802431611,
+      "grad_norm": 1.7579659223556519,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.5235031247138977,
+      "mean_token_accuracy": 0.8143284320831299,
+      "num_tokens": 1679241.0,
+      "step": 177
+    },
+    {
+      "epoch": 0.135258358662614,
+      "grad_norm": 3.123563528060913,
+      "learning_rate": 4.46969696969697e-06,
+      "loss": 0.43051332235336304,
+      "mean_token_accuracy": 0.8518186211585999,
+      "num_tokens": 1683317.0,
+      "step": 178
+    },
+    {
+      "epoch": 0.13601823708206687,
+      "grad_norm": 2.2411575317382812,
+      "learning_rate": 4.494949494949495e-06,
+      "loss": 0.5471380949020386,
+      "mean_token_accuracy": 0.8267596960067749,
+      "num_tokens": 1691366.0,
+      "step": 179
+    },
+    {
+      "epoch": 0.13677811550151975,
+      "grad_norm": 2.621973991394043,
+      "learning_rate": 4.520202020202021e-06,
+      "loss": 0.5685839653015137,
+      "mean_token_accuracy": 0.8260642290115356,
+      "num_tokens": 1698148.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.13753799392097266,
+      "grad_norm": 2.1553852558135986,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.5703883171081543,
+      "mean_token_accuracy": 0.8219090700149536,
+      "num_tokens": 1707225.0,
+      "step": 181
+    },
+    {
+      "epoch": 0.13829787234042554,
+      "grad_norm": 5.1767897605896,
+      "learning_rate": 4.5707070707070715e-06,
+      "loss": 0.32704639434814453,
+      "mean_token_accuracy": 0.8754568099975586,
+      "num_tokens": 1712748.0,
+      "step": 182
+    },
+    {
+      "epoch": 0.13905775075987842,
+      "grad_norm": 2.609168291091919,
+      "learning_rate": 4.595959595959597e-06,
+      "loss": 0.5939987301826477,
+      "mean_token_accuracy": 0.8034975528717041,
+      "num_tokens": 1719932.0,
+      "step": 183
+    },
+    {
+      "epoch": 0.1398176291793313,
+      "grad_norm": 2.2059099674224854,
+      "learning_rate": 4.621212121212122e-06,
+      "loss": 0.5310720205307007,
+      "mean_token_accuracy": 0.8177368640899658,
+      "num_tokens": 1727640.0,
+      "step": 184
+    },
+    {
+      "epoch": 0.1405775075987842,
+      "grad_norm": 2.6367759704589844,
+      "learning_rate": 4.646464646464647e-06,
+      "loss": 0.522086501121521,
+      "mean_token_accuracy": 0.826233983039856,
+      "num_tokens": 1733609.0,
+      "step": 185
+    },
+    {
+      "epoch": 0.1413373860182371,
+      "grad_norm": 3.326732873916626,
+      "learning_rate": 4.671717171717172e-06,
+      "loss": 0.4127829074859619,
+      "mean_token_accuracy": 0.8551101684570312,
+      "num_tokens": 1737256.0,
+      "step": 186
+    },
+    {
+      "epoch": 0.14209726443768997,
+      "grad_norm": 1.828412413597107,
+      "learning_rate": 4.696969696969698e-06,
+      "loss": 0.5444269180297852,
+      "mean_token_accuracy": 0.8350818157196045,
+      "num_tokens": 1750196.0,
+      "step": 187
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 3.209203004837036,
+      "learning_rate": 4.722222222222222e-06,
+      "loss": 0.5087994933128357,
+      "mean_token_accuracy": 0.8349015712738037,
+      "num_tokens": 1754836.0,
+      "step": 188
+    },
+    {
+      "epoch": 0.14361702127659576,
+      "grad_norm": 1.7339166402816772,
+      "learning_rate": 4.747474747474748e-06,
+      "loss": 0.5151352286338806,
+      "mean_token_accuracy": 0.8321266174316406,
+      "num_tokens": 1766015.0,
+      "step": 189
+    },
+    {
+      "epoch": 0.14437689969604864,
+      "grad_norm": 2.699068069458008,
+      "learning_rate": 4.772727272727273e-06,
+      "loss": 0.4406203031539917,
+      "mean_token_accuracy": 0.8425000905990601,
+      "num_tokens": 1771684.0,
+      "step": 190
+    },
+    {
+      "epoch": 0.14513677811550152,
+      "grad_norm": 2.8117282390594482,
+      "learning_rate": 4.7979797979797985e-06,
+      "loss": 0.40428489446640015,
+      "mean_token_accuracy": 0.8654326796531677,
+      "num_tokens": 1776301.0,
+      "step": 191
+    },
+    {
+      "epoch": 0.1458966565349544,
+      "grad_norm": 2.9204647541046143,
+      "learning_rate": 4.823232323232324e-06,
+      "loss": 0.4191770553588867,
+      "mean_token_accuracy": 0.8574687242507935,
+      "num_tokens": 1781678.0,
+      "step": 192
+    },
+    {
+      "epoch": 0.1466565349544073,
+      "grad_norm": 2.1648988723754883,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 0.5839012861251831,
+      "mean_token_accuracy": 0.8053664565086365,
+      "num_tokens": 1792516.0,
+      "step": 193
+    },
+    {
+      "epoch": 0.1474164133738602,
+      "grad_norm": 2.3221631050109863,
+      "learning_rate": 4.873737373737374e-06,
+      "loss": 0.5037894248962402,
+      "mean_token_accuracy": 0.8427227139472961,
+      "num_tokens": 1800192.0,
+      "step": 194
+    },
+    {
+      "epoch": 0.14817629179331307,
+      "grad_norm": 2.4536430835723877,
+      "learning_rate": 4.898989898989899e-06,
+      "loss": 0.42326074838638306,
+      "mean_token_accuracy": 0.8510633111000061,
+      "num_tokens": 1806159.0,
+      "step": 195
+    },
+    {
+      "epoch": 0.14893617021276595,
+      "grad_norm": 2.4875805377960205,
+      "learning_rate": 4.924242424242425e-06,
+      "loss": 0.539531409740448,
+      "mean_token_accuracy": 0.8060250282287598,
+      "num_tokens": 1813392.0,
+      "step": 196
+    },
+    {
+      "epoch": 0.14969604863221886,
+      "grad_norm": 2.1664798259735107,
+      "learning_rate": 4.94949494949495e-06,
+      "loss": 0.42502015829086304,
+      "mean_token_accuracy": 0.8503251075744629,
+      "num_tokens": 1821424.0,
+      "step": 197
+    },
+    {
+      "epoch": 0.15045592705167174,
+      "grad_norm": 2.568808078765869,
+      "learning_rate": 4.974747474747475e-06,
+      "loss": 0.5025098323822021,
+      "mean_token_accuracy": 0.8182311058044434,
+      "num_tokens": 1827225.0,
+      "step": 198
+    },
+    {
+      "epoch": 0.15121580547112462,
+      "grad_norm": 1.9116802215576172,
+      "learning_rate": 5e-06,
+      "loss": 0.4907258450984955,
+      "mean_token_accuracy": 0.8310189843177795,
+      "num_tokens": 1836297.0,
+      "step": 199
+    },
+    {
+      "epoch": 0.1519756838905775,
+      "grad_norm": 3.150765895843506,
+      "learning_rate": 4.999999122701883e-06,
+      "loss": 0.390616774559021,
+      "mean_token_accuracy": 0.8626647591590881,
+      "num_tokens": 1839984.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.15273556231003038,
+      "grad_norm": 3.2229044437408447,
+      "learning_rate": 4.999996490808146e-06,
+      "loss": 0.48009657859802246,
+      "mean_token_accuracy": 0.825214147567749,
+      "num_tokens": 1844610.0,
+      "step": 201
+    },
+    {
+      "epoch": 0.1534954407294833,
+      "grad_norm": 1.4473289251327515,
+      "learning_rate": 4.9999921043206356e-06,
+      "loss": 0.40135183930397034,
+      "mean_token_accuracy": 0.8537827730178833,
+      "num_tokens": 1859573.0,
+      "step": 202
+    },
+    {
+      "epoch": 0.15425531914893617,
+      "grad_norm": 4.072319507598877,
+      "learning_rate": 4.999985963242432e-06,
+      "loss": 0.6158689260482788,
+      "mean_token_accuracy": 0.8075432777404785,
+      "num_tokens": 1863147.0,
+      "step": 203
+    },
+    {
+      "epoch": 0.15501519756838905,
+      "grad_norm": 3.15741229057312,
+      "learning_rate": 4.999978067577844e-06,
+      "loss": 0.4603108763694763,
+      "mean_token_accuracy": 0.8418779373168945,
+      "num_tokens": 1867201.0,
+      "step": 204
+    },
+    {
+      "epoch": 0.15577507598784193,
+      "grad_norm": 2.1925418376922607,
+      "learning_rate": 4.999968417332415e-06,
+      "loss": 0.5552488565444946,
+      "mean_token_accuracy": 0.8216016292572021,
+      "num_tokens": 1874837.0,
+      "step": 205
+    },
+    {
+      "epoch": 0.15653495440729484,
+      "grad_norm": 2.2518117427825928,
+      "learning_rate": 4.999957012512916e-06,
+      "loss": 0.4912569522857666,
+      "mean_token_accuracy": 0.8284667730331421,
+      "num_tokens": 1881842.0,
+      "step": 206
+    },
+    {
+      "epoch": 0.15729483282674772,
+      "grad_norm": 1.8223762512207031,
+      "learning_rate": 4.999943853127351e-06,
+      "loss": 0.47709137201309204,
+      "mean_token_accuracy": 0.8311659097671509,
+      "num_tokens": 1890805.0,
+      "step": 207
+    },
+    {
+      "epoch": 0.1580547112462006,
+      "grad_norm": 2.066499948501587,
+      "learning_rate": 4.999928939184958e-06,
+      "loss": 0.44794657826423645,
+      "mean_token_accuracy": 0.8513424396514893,
+      "num_tokens": 1898264.0,
+      "step": 208
+    },
+    {
+      "epoch": 0.15881458966565348,
+      "grad_norm": 3.53865909576416,
+      "learning_rate": 4.999912270696202e-06,
+      "loss": 0.5978270769119263,
+      "mean_token_accuracy": 0.8080137968063354,
+      "num_tokens": 1902435.0,
+      "step": 209
+    },
+    {
+      "epoch": 0.1595744680851064,
+      "grad_norm": 2.0760679244995117,
+      "learning_rate": 4.999893847672783e-06,
+      "loss": 0.5930601358413696,
+      "mean_token_accuracy": 0.8028650283813477,
+      "num_tokens": 1912252.0,
+      "step": 210
+    },
+    {
+      "epoch": 0.16033434650455927,
+      "grad_norm": 2.21551513671875,
+      "learning_rate": 4.99987367012763e-06,
+      "loss": 0.6336753964424133,
+      "mean_token_accuracy": 0.7902286648750305,
+      "num_tokens": 1922095.0,
+      "step": 211
+    },
+    {
+      "epoch": 0.16109422492401215,
+      "grad_norm": 1.7654480934143066,
+      "learning_rate": 4.999851738074904e-06,
+      "loss": 0.6373403668403625,
+      "mean_token_accuracy": 0.7802424430847168,
+      "num_tokens": 1938962.0,
+      "step": 212
+    },
+    {
+      "epoch": 0.16185410334346503,
+      "grad_norm": 2.852834701538086,
+      "learning_rate": 4.9998280515300006e-06,
+      "loss": 0.6418683528900146,
+      "mean_token_accuracy": 0.7895716428756714,
+      "num_tokens": 1944668.0,
+      "step": 213
+    },
+    {
+      "epoch": 0.16261398176291794,
+      "grad_norm": 3.4737212657928467,
+      "learning_rate": 4.999802610509541e-06,
+      "loss": 0.6323273181915283,
+      "mean_token_accuracy": 0.7982614636421204,
+      "num_tokens": 1949142.0,
+      "step": 214
+    },
+    {
+      "epoch": 0.16337386018237082,
+      "grad_norm": 3.0802664756774902,
+      "learning_rate": 4.999775415031381e-06,
+      "loss": 0.5929068326950073,
+      "mean_token_accuracy": 0.8112219572067261,
+      "num_tokens": 1954141.0,
+      "step": 215
+    },
+    {
+      "epoch": 0.1641337386018237,
+      "grad_norm": 2.9808855056762695,
+      "learning_rate": 4.999746465114609e-06,
+      "loss": 0.5556406378746033,
+      "mean_token_accuracy": 0.8117628693580627,
+      "num_tokens": 1959406.0,
+      "step": 216
+    },
+    {
+      "epoch": 0.16489361702127658,
+      "grad_norm": 1.7346166372299194,
+      "learning_rate": 4.999715760779541e-06,
+      "loss": 0.5122925043106079,
+      "mean_token_accuracy": 0.8040724992752075,
+      "num_tokens": 1971921.0,
+      "step": 217
+    },
+    {
+      "epoch": 0.1656534954407295,
+      "grad_norm": 1.4183907508850098,
+      "learning_rate": 4.999683302047729e-06,
+      "loss": 0.46471893787384033,
+      "mean_token_accuracy": 0.8381330966949463,
+      "num_tokens": 1988863.0,
+      "step": 218
+    },
+    {
+      "epoch": 0.16641337386018237,
+      "grad_norm": 1.6797802448272705,
+      "learning_rate": 4.999649088941951e-06,
+      "loss": 0.38348832726478577,
+      "mean_token_accuracy": 0.8344278931617737,
+      "num_tokens": 2000003.0,
+      "step": 219
+    },
+    {
+      "epoch": 0.16717325227963525,
+      "grad_norm": 3.036963939666748,
+      "learning_rate": 4.999613121486222e-06,
+      "loss": 0.6062780618667603,
+      "mean_token_accuracy": 0.8217900991439819,
+      "num_tokens": 2004813.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.16793313069908813,
+      "grad_norm": 2.0343217849731445,
+      "learning_rate": 4.999575399705782e-06,
+      "loss": 0.5052450895309448,
+      "mean_token_accuracy": 0.8368623852729797,
+      "num_tokens": 2013565.0,
+      "step": 221
+    },
+    {
+      "epoch": 0.16869300911854104,
+      "grad_norm": 2.1162009239196777,
+      "learning_rate": 4.9995359236271094e-06,
+      "loss": 0.5169756412506104,
+      "mean_token_accuracy": 0.8339958190917969,
+      "num_tokens": 2025763.0,
+      "step": 222
+    },
+    {
+      "epoch": 0.16945288753799392,
+      "grad_norm": 2.055333375930786,
+      "learning_rate": 4.9994946932779076e-06,
+      "loss": 0.6327048540115356,
+      "mean_token_accuracy": 0.8078711032867432,
+      "num_tokens": 2037005.0,
+      "step": 223
+    },
+    {
+      "epoch": 0.1702127659574468,
+      "grad_norm": 3.334620475769043,
+      "learning_rate": 4.999451708687114e-06,
+      "loss": 0.5688358545303345,
+      "mean_token_accuracy": 0.8015589714050293,
+      "num_tokens": 2041473.0,
+      "step": 224
+    },
+    {
+      "epoch": 0.17097264437689969,
+      "grad_norm": 2.3734676837921143,
+      "learning_rate": 4.999406969884897e-06,
+      "loss": 0.5673821568489075,
+      "mean_token_accuracy": 0.8054057359695435,
+      "num_tokens": 2049397.0,
+      "step": 225
+    },
+    {
+      "epoch": 0.1717325227963526,
+      "grad_norm": 1.807358980178833,
+      "learning_rate": 4.999360476902656e-06,
+      "loss": 0.4376158118247986,
+      "mean_token_accuracy": 0.8456039428710938,
+      "num_tokens": 2058721.0,
+      "step": 226
+    },
+    {
+      "epoch": 0.17249240121580547,
+      "grad_norm": 3.231638193130493,
+      "learning_rate": 4.999312229773022e-06,
+      "loss": 0.5592809915542603,
+      "mean_token_accuracy": 0.8170154094696045,
+      "num_tokens": 2063455.0,
+      "step": 227
+    },
+    {
+      "epoch": 0.17325227963525835,
+      "grad_norm": 2.2717151641845703,
+      "learning_rate": 4.999262228529855e-06,
+      "loss": 0.6144396066665649,
+      "mean_token_accuracy": 0.7948470115661621,
+      "num_tokens": 2071686.0,
+      "step": 228
+    },
+    {
+      "epoch": 0.17401215805471124,
+      "grad_norm": 1.4171342849731445,
+      "learning_rate": 4.99921047320825e-06,
+      "loss": 0.43680912256240845,
+      "mean_token_accuracy": 0.84850013256073,
+      "num_tokens": 2086999.0,
+      "step": 229
+    },
+    {
+      "epoch": 0.17477203647416414,
+      "grad_norm": 3.162736654281616,
+      "learning_rate": 4.99915696384453e-06,
+      "loss": 0.6025407910346985,
+      "mean_token_accuracy": 0.8042335510253906,
+      "num_tokens": 2092001.0,
+      "step": 230
+    },
+    {
+      "epoch": 0.17553191489361702,
+      "grad_norm": 1.8672804832458496,
+      "learning_rate": 4.99910170047625e-06,
+      "loss": 0.5843087434768677,
+      "mean_token_accuracy": 0.8016980886459351,
+      "num_tokens": 2103372.0,
+      "step": 231
+    },
+    {
+      "epoch": 0.1762917933130699,
+      "grad_norm": 2.967587471008301,
+      "learning_rate": 4.999044683142196e-06,
+      "loss": 0.5123642086982727,
+      "mean_token_accuracy": 0.8216149806976318,
+      "num_tokens": 2108008.0,
+      "step": 232
+    },
+    {
+      "epoch": 0.1770516717325228,
+      "grad_norm": 1.9651981592178345,
+      "learning_rate": 4.998985911882383e-06,
+      "loss": 0.5868178606033325,
+      "mean_token_accuracy": 0.7904198169708252,
+      "num_tokens": 2119009.0,
+      "step": 233
+    },
+    {
+      "epoch": 0.1778115501519757,
+      "grad_norm": 2.7785449028015137,
+      "learning_rate": 4.998925386738063e-06,
+      "loss": 0.5075510144233704,
+      "mean_token_accuracy": 0.8280210494995117,
+      "num_tokens": 2124915.0,
+      "step": 234
+    },
+    {
+      "epoch": 0.17857142857142858,
+      "grad_norm": 2.957470417022705,
+      "learning_rate": 4.998863107751711e-06,
+      "loss": 0.5351958274841309,
+      "mean_token_accuracy": 0.846825122833252,
+      "num_tokens": 2129905.0,
+      "step": 235
+    },
+    {
+      "epoch": 0.17933130699088146,
+      "grad_norm": 3.207671880722046,
+      "learning_rate": 4.99879907496704e-06,
+      "loss": 0.6209091544151306,
+      "mean_token_accuracy": 0.789960503578186,
+      "num_tokens": 2135027.0,
+      "step": 236
+    },
+    {
+      "epoch": 0.18009118541033434,
+      "grad_norm": 2.018953800201416,
+      "learning_rate": 4.998733288428987e-06,
+      "loss": 0.601510763168335,
+      "mean_token_accuracy": 0.8136930465698242,
+      "num_tokens": 2147016.0,
+      "step": 237
+    },
+    {
+      "epoch": 0.18085106382978725,
+      "grad_norm": 2.437281847000122,
+      "learning_rate": 4.998665748183727e-06,
+      "loss": 0.5813639163970947,
+      "mean_token_accuracy": 0.8116716146469116,
+      "num_tokens": 2155386.0,
+      "step": 238
+    },
+    {
+      "epoch": 0.18161094224924013,
+      "grad_norm": 1.5708180665969849,
+      "learning_rate": 4.998596454278661e-06,
+      "loss": 0.5252395272254944,
+      "mean_token_accuracy": 0.8193864822387695,
+      "num_tokens": 2170295.0,
+      "step": 239
+    },
+    {
+      "epoch": 0.182370820668693,
+      "grad_norm": 1.9921495914459229,
+      "learning_rate": 4.998525406762422e-06,
+      "loss": 0.5335029363632202,
+      "mean_token_accuracy": 0.8120872974395752,
+      "num_tokens": 2180012.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.1831306990881459,
+      "grad_norm": 2.6562681198120117,
+      "learning_rate": 4.998452605684874e-06,
+      "loss": 0.48021435737609863,
+      "mean_token_accuracy": 0.8388714790344238,
+      "num_tokens": 2185607.0,
+      "step": 241
+    },
+    {
+      "epoch": 0.1838905775075988,
+      "grad_norm": 2.2535853385925293,
+      "learning_rate": 4.998378051097111e-06,
+      "loss": 0.5747300386428833,
+      "mean_token_accuracy": 0.8004639148712158,
+      "num_tokens": 2194105.0,
+      "step": 242
+    },
+    {
+      "epoch": 0.18465045592705168,
+      "grad_norm": 1.6151788234710693,
+      "learning_rate": 4.998301743051459e-06,
+      "loss": 0.6190565824508667,
+      "mean_token_accuracy": 0.7816627621650696,
+      "num_tokens": 2210629.0,
+      "step": 243
+    },
+    {
+      "epoch": 0.18541033434650456,
+      "grad_norm": 2.1088173389434814,
+      "learning_rate": 4.9982236816014735e-06,
+      "loss": 0.4715560972690582,
+      "mean_token_accuracy": 0.8485721349716187,
+      "num_tokens": 2218958.0,
+      "step": 244
+    },
+    {
+      "epoch": 0.18617021276595744,
+      "grad_norm": 2.6168735027313232,
+      "learning_rate": 4.998143866801941e-06,
+      "loss": 0.6077103018760681,
+      "mean_token_accuracy": 0.8057924509048462,
+      "num_tokens": 2226368.0,
+      "step": 245
+    },
+    {
+      "epoch": 0.18693009118541035,
+      "grad_norm": 2.5988616943359375,
+      "learning_rate": 4.99806229870888e-06,
+      "loss": 0.5021637678146362,
+      "mean_token_accuracy": 0.8361666202545166,
+      "num_tokens": 2232485.0,
+      "step": 246
+    },
+    {
+      "epoch": 0.18768996960486323,
+      "grad_norm": 2.015887498855591,
+      "learning_rate": 4.9979789773795365e-06,
+      "loss": 0.4309737980365753,
+      "mean_token_accuracy": 0.8508044481277466,
+      "num_tokens": 2240819.0,
+      "step": 247
+    },
+    {
+      "epoch": 0.1884498480243161,
+      "grad_norm": 2.3115265369415283,
+      "learning_rate": 4.997893902872389e-06,
+      "loss": 0.5776500701904297,
+      "mean_token_accuracy": 0.8079549074172974,
+      "num_tokens": 2249460.0,
+      "step": 248
+    },
+    {
+      "epoch": 0.189209726443769,
+      "grad_norm": 1.7387021780014038,
+      "learning_rate": 4.997807075247147e-06,
+      "loss": 0.430944561958313,
+      "mean_token_accuracy": 0.8483544588088989,
+      "num_tokens": 2259124.0,
+      "step": 249
+    },
+    {
+      "epoch": 0.1899696048632219,
+      "grad_norm": 1.6378381252288818,
+      "learning_rate": 4.997718494564747e-06,
+      "loss": 0.4123363792896271,
+      "mean_token_accuracy": 0.8557409644126892,
+      "num_tokens": 2269899.0,
+      "step": 250
+    },
+    {
+      "epoch": 0.19072948328267478,
+      "grad_norm": 1.336282730102539,
+      "learning_rate": 4.997628160887361e-06,
+      "loss": 0.502329409122467,
+      "mean_token_accuracy": 0.8186938166618347,
+      "num_tokens": 2292821.0,
+      "step": 251
+    },
+    {
+      "epoch": 0.19148936170212766,
+      "grad_norm": 3.3335583209991455,
+      "learning_rate": 4.997536074278388e-06,
+      "loss": 0.584446907043457,
+      "mean_token_accuracy": 0.8062717318534851,
+      "num_tokens": 2297175.0,
+      "step": 252
+    },
+    {
+      "epoch": 0.19224924012158054,
+      "grad_norm": 2.246727228164673,
+      "learning_rate": 4.9974422348024565e-06,
+      "loss": 0.5683060884475708,
+      "mean_token_accuracy": 0.8193703293800354,
+      "num_tokens": 2305456.0,
+      "step": 253
+    },
+    {
+      "epoch": 0.19300911854103345,
+      "grad_norm": 2.3520865440368652,
+      "learning_rate": 4.997346642525429e-06,
+      "loss": 0.4724946618080139,
+      "mean_token_accuracy": 0.8426719307899475,
+      "num_tokens": 2312241.0,
+      "step": 254
+    },
+    {
+      "epoch": 0.19376899696048633,
+      "grad_norm": 2.7115702629089355,
+      "learning_rate": 4.9972492975143936e-06,
+      "loss": 0.5019032955169678,
+      "mean_token_accuracy": 0.8253573179244995,
+      "num_tokens": 2318094.0,
+      "step": 255
+    },
+    {
+      "epoch": 0.1945288753799392,
+      "grad_norm": 1.705528974533081,
+      "learning_rate": 4.997150199837671e-06,
+      "loss": 0.45588475465774536,
+      "mean_token_accuracy": 0.836666464805603,
+      "num_tokens": 2329025.0,
+      "step": 256
+    },
+    {
+      "epoch": 0.1952887537993921,
+      "grad_norm": 2.161400318145752,
+      "learning_rate": 4.997049349564814e-06,
+      "loss": 0.5170183777809143,
+      "mean_token_accuracy": 0.8287534117698669,
+      "num_tokens": 2337448.0,
+      "step": 257
+    },
+    {
+      "epoch": 0.196048632218845,
+      "grad_norm": 2.629669189453125,
+      "learning_rate": 4.996946746766602e-06,
+      "loss": 0.44650501012802124,
+      "mean_token_accuracy": 0.850114107131958,
+      "num_tokens": 2343207.0,
+      "step": 258
+    },
+    {
+      "epoch": 0.19680851063829788,
+      "grad_norm": 1.6735503673553467,
+      "learning_rate": 4.996842391515045e-06,
+      "loss": 0.5247820019721985,
+      "mean_token_accuracy": 0.8285071849822998,
+      "num_tokens": 2356801.0,
+      "step": 259
+    },
+    {
+      "epoch": 0.19756838905775076,
+      "grad_norm": 1.2753115892410278,
+      "learning_rate": 4.996736283883382e-06,
+      "loss": 0.41870927810668945,
+      "mean_token_accuracy": 0.8448047637939453,
+      "num_tokens": 2377306.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.19832826747720364,
+      "grad_norm": 2.6947314739227295,
+      "learning_rate": 4.9966284239460875e-06,
+      "loss": 0.5059205889701843,
+      "mean_token_accuracy": 0.8430814743041992,
+      "num_tokens": 2383352.0,
+      "step": 261
+    },
+    {
+      "epoch": 0.19908814589665655,
+      "grad_norm": 2.0509963035583496,
+      "learning_rate": 4.996518811778858e-06,
+      "loss": 0.4565388560295105,
+      "mean_token_accuracy": 0.8453130722045898,
+      "num_tokens": 2391149.0,
+      "step": 262
+    },
+    {
+      "epoch": 0.19984802431610943,
+      "grad_norm": 2.1856348514556885,
+      "learning_rate": 4.996407447458626e-06,
+      "loss": 0.531380832195282,
+      "mean_token_accuracy": 0.8387004137039185,
+      "num_tokens": 2399875.0,
+      "step": 263
+    },
+    {
+      "epoch": 0.2006079027355623,
+      "grad_norm": 2.7348573207855225,
+      "learning_rate": 4.99629433106355e-06,
+      "loss": 0.5242817401885986,
+      "mean_token_accuracy": 0.8177423477172852,
+      "num_tokens": 2406586.0,
+      "step": 264
+    },
+    {
+      "epoch": 0.2013677811550152,
+      "grad_norm": 1.76587975025177,
+      "learning_rate": 4.99617946267302e-06,
+      "loss": 0.49298471212387085,
+      "mean_token_accuracy": 0.8271149396896362,
+      "num_tokens": 2418683.0,
+      "step": 265
+    },
+    {
+      "epoch": 0.20212765957446807,
+      "grad_norm": 2.8129730224609375,
+      "learning_rate": 4.996062842367655e-06,
+      "loss": 0.46420302987098694,
+      "mean_token_accuracy": 0.8453244566917419,
+      "num_tokens": 2422929.0,
+      "step": 266
+    },
+    {
+      "epoch": 0.20288753799392098,
+      "grad_norm": 2.575744152069092,
+      "learning_rate": 4.9959444702293025e-06,
+      "loss": 0.43208545446395874,
+      "mean_token_accuracy": 0.8494843244552612,
+      "num_tokens": 2429567.0,
+      "step": 267
+    },
+    {
+      "epoch": 0.20364741641337386,
+      "grad_norm": 2.7586750984191895,
+      "learning_rate": 4.995824346341041e-06,
+      "loss": 0.4390473961830139,
+      "mean_token_accuracy": 0.8348895311355591,
+      "num_tokens": 2434700.0,
+      "step": 268
+    },
+    {
+      "epoch": 0.20440729483282674,
+      "grad_norm": 1.972145438194275,
+      "learning_rate": 4.99570247078718e-06,
+      "loss": 0.6219544410705566,
+      "mean_token_accuracy": 0.7939999103546143,
+      "num_tokens": 2447007.0,
+      "step": 269
+    },
+    {
+      "epoch": 0.20516717325227962,
+      "grad_norm": 2.2963485717773438,
+      "learning_rate": 4.995578843653255e-06,
+      "loss": 0.5008970499038696,
+      "mean_token_accuracy": 0.8255308866500854,
+      "num_tokens": 2453936.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.20592705167173253,
+      "grad_norm": 1.8897721767425537,
+      "learning_rate": 4.995453465026033e-06,
+      "loss": 0.5436089038848877,
+      "mean_token_accuracy": 0.819086492061615,
+      "num_tokens": 2464494.0,
+      "step": 271
+    },
+    {
+      "epoch": 0.2066869300911854,
+      "grad_norm": 2.319728374481201,
+      "learning_rate": 4.995326334993508e-06,
+      "loss": 0.5136368870735168,
+      "mean_token_accuracy": 0.820817232131958,
+      "num_tokens": 2470938.0,
+      "step": 272
+    },
+    {
+      "epoch": 0.2074468085106383,
+      "grad_norm": 2.230414390563965,
+      "learning_rate": 4.9951974536449055e-06,
+      "loss": 0.5272846817970276,
+      "mean_token_accuracy": 0.8203279972076416,
+      "num_tokens": 2478629.0,
+      "step": 273
+    },
+    {
+      "epoch": 0.20820668693009117,
+      "grad_norm": 3.401937484741211,
+      "learning_rate": 4.9950668210706795e-06,
+      "loss": 0.4389592111110687,
+      "mean_token_accuracy": 0.8647899031639099,
+      "num_tokens": 2482193.0,
+      "step": 274
+    },
+    {
+      "epoch": 0.20896656534954408,
+      "grad_norm": 2.1278507709503174,
+      "learning_rate": 4.994934437362513e-06,
+      "loss": 0.598863422870636,
+      "mean_token_accuracy": 0.7945119738578796,
+      "num_tokens": 2492465.0,
+      "step": 275
+    },
+    {
+      "epoch": 0.20972644376899696,
+      "grad_norm": 1.9259960651397705,
+      "learning_rate": 4.994800302613318e-06,
+      "loss": 0.49520939588546753,
+      "mean_token_accuracy": 0.8371536135673523,
+      "num_tokens": 2500825.0,
+      "step": 276
+    },
+    {
+      "epoch": 0.21048632218844984,
+      "grad_norm": 2.346418857574463,
+      "learning_rate": 4.994664416917236e-06,
+      "loss": 0.5412614345550537,
+      "mean_token_accuracy": 0.810661792755127,
+      "num_tokens": 2509513.0,
+      "step": 277
+    },
+    {
+      "epoch": 0.21124620060790272,
+      "grad_norm": 1.3092039823532104,
+      "learning_rate": 4.994526780369636e-06,
+      "loss": 0.46305379271507263,
+      "mean_token_accuracy": 0.8358527421951294,
+      "num_tokens": 2531405.0,
+      "step": 278
+    },
+    {
+      "epoch": 0.21200607902735563,
+      "grad_norm": 2.924611806869507,
+      "learning_rate": 4.9943873930671175e-06,
+      "loss": 0.6134544610977173,
+      "mean_token_accuracy": 0.7947378754615784,
+      "num_tokens": 2536744.0,
+      "step": 279
+    },
+    {
+      "epoch": 0.2127659574468085,
+      "grad_norm": 2.8290598392486572,
+      "learning_rate": 4.994246255107506e-06,
+      "loss": 0.465520441532135,
+      "mean_token_accuracy": 0.8440108299255371,
+      "num_tokens": 2541184.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.2135258358662614,
+      "grad_norm": 3.8081259727478027,
+      "learning_rate": 4.994103366589859e-06,
+      "loss": 0.43394139409065247,
+      "mean_token_accuracy": 0.8579148054122925,
+      "num_tokens": 2545395.0,
+      "step": 281
+    },
+    {
+      "epoch": 0.21428571428571427,
+      "grad_norm": 1.7994529008865356,
+      "learning_rate": 4.993958727614462e-06,
+      "loss": 0.5076484680175781,
+      "mean_token_accuracy": 0.8270803093910217,
+      "num_tokens": 2556541.0,
+      "step": 282
+    },
+    {
+      "epoch": 0.21504559270516718,
+      "grad_norm": 2.5582659244537354,
+      "learning_rate": 4.993812338282826e-06,
+      "loss": 0.4453684389591217,
+      "mean_token_accuracy": 0.8488293886184692,
+      "num_tokens": 2562949.0,
+      "step": 283
+    },
+    {
+      "epoch": 0.21580547112462006,
+      "grad_norm": 1.6448938846588135,
+      "learning_rate": 4.993664198697694e-06,
+      "loss": 0.461971640586853,
+      "mean_token_accuracy": 0.824763298034668,
+      "num_tokens": 2576407.0,
+      "step": 284
+    },
+    {
+      "epoch": 0.21656534954407294,
+      "grad_norm": 2.1264469623565674,
+      "learning_rate": 4.993514308963037e-06,
+      "loss": 0.6241602897644043,
+      "mean_token_accuracy": 0.7916014790534973,
+      "num_tokens": 2585695.0,
+      "step": 285
+    },
+    {
+      "epoch": 0.21732522796352582,
+      "grad_norm": 3.629991292953491,
+      "learning_rate": 4.993362669184051e-06,
+      "loss": 0.610355019569397,
+      "mean_token_accuracy": 0.7847568988800049,
+      "num_tokens": 2589778.0,
+      "step": 286
+    },
+    {
+      "epoch": 0.21808510638297873,
+      "grad_norm": 1.9070756435394287,
+      "learning_rate": 4.993209279467164e-06,
+      "loss": 0.5513623952865601,
+      "mean_token_accuracy": 0.7911607027053833,
+      "num_tokens": 2600920.0,
+      "step": 287
+    },
+    {
+      "epoch": 0.2188449848024316,
+      "grad_norm": 1.761062741279602,
+      "learning_rate": 4.993054139920031e-06,
+      "loss": 0.4579957127571106,
+      "mean_token_accuracy": 0.8189530372619629,
+      "num_tokens": 2611856.0,
+      "step": 288
+    },
+    {
+      "epoch": 0.2196048632218845,
+      "grad_norm": 1.7264713048934937,
+      "learning_rate": 4.992897250651535e-06,
+      "loss": 0.5871305465698242,
+      "mean_token_accuracy": 0.7918527126312256,
+      "num_tokens": 2624730.0,
+      "step": 289
+    },
+    {
+      "epoch": 0.22036474164133737,
+      "grad_norm": 1.7455977201461792,
+      "learning_rate": 4.992738611771787e-06,
+      "loss": 0.5475119948387146,
+      "mean_token_accuracy": 0.8226917386054993,
+      "num_tokens": 2635705.0,
+      "step": 290
+    },
+    {
+      "epoch": 0.22112462006079028,
+      "grad_norm": 2.095095157623291,
+      "learning_rate": 4.992578223392124e-06,
+      "loss": 0.5952225923538208,
+      "mean_token_accuracy": 0.8078469038009644,
+      "num_tokens": 2643954.0,
+      "step": 291
+    },
+    {
+      "epoch": 0.22188449848024316,
+      "grad_norm": 2.994664192199707,
+      "learning_rate": 4.992416085625115e-06,
+      "loss": 0.5432442426681519,
+      "mean_token_accuracy": 0.8329008221626282,
+      "num_tokens": 2648800.0,
+      "step": 292
+    },
+    {
+      "epoch": 0.22264437689969604,
+      "grad_norm": 2.796790361404419,
+      "learning_rate": 4.992252198584554e-06,
+      "loss": 0.5168961882591248,
+      "mean_token_accuracy": 0.8393474817276001,
+      "num_tokens": 2653546.0,
+      "step": 293
+    },
+    {
+      "epoch": 0.22340425531914893,
+      "grad_norm": 1.8610522747039795,
+      "learning_rate": 4.992086562385462e-06,
+      "loss": 0.5728024244308472,
+      "mean_token_accuracy": 0.797406792640686,
+      "num_tokens": 2667483.0,
+      "step": 294
+    },
+    {
+      "epoch": 0.22416413373860183,
+      "grad_norm": 1.695472002029419,
+      "learning_rate": 4.9919191771440905e-06,
+      "loss": 0.5460028648376465,
+      "mean_token_accuracy": 0.8123016357421875,
+      "num_tokens": 2683574.0,
+      "step": 295
+    },
+    {
+      "epoch": 0.22492401215805471,
+      "grad_norm": 2.8627376556396484,
+      "learning_rate": 4.9917500429779165e-06,
+      "loss": 0.5566985011100769,
+      "mean_token_accuracy": 0.815531313419342,
+      "num_tokens": 2688985.0,
+      "step": 296
+    },
+    {
+      "epoch": 0.2256838905775076,
+      "grad_norm": 2.73323655128479,
+      "learning_rate": 4.991579160005644e-06,
+      "loss": 0.48197102546691895,
+      "mean_token_accuracy": 0.8471829295158386,
+      "num_tokens": 2694799.0,
+      "step": 297
+    },
+    {
+      "epoch": 0.22644376899696048,
+      "grad_norm": 1.8436161279678345,
+      "learning_rate": 4.991406528347206e-06,
+      "loss": 0.4528339207172394,
+      "mean_token_accuracy": 0.8603188395500183,
+      "num_tokens": 2707321.0,
+      "step": 298
+    },
+    {
+      "epoch": 0.22720364741641338,
+      "grad_norm": 2.6231515407562256,
+      "learning_rate": 4.9912321481237616e-06,
+      "loss": 0.5916541814804077,
+      "mean_token_accuracy": 0.8050242066383362,
+      "num_tokens": 2714233.0,
+      "step": 299
+    },
+    {
+      "epoch": 0.22796352583586627,
+      "grad_norm": 3.08776593208313,
+      "learning_rate": 4.991056019457697e-06,
+      "loss": 0.4860580563545227,
+      "mean_token_accuracy": 0.8464088439941406,
+      "num_tokens": 2718443.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.22872340425531915,
+      "grad_norm": 2.2537803649902344,
+      "learning_rate": 4.990878142472628e-06,
+      "loss": 0.5158311128616333,
+      "mean_token_accuracy": 0.824694812297821,
+      "num_tokens": 2726158.0,
+      "step": 301
+    },
+    {
+      "epoch": 0.22948328267477203,
+      "grad_norm": 2.1122705936431885,
+      "learning_rate": 4.990698517293394e-06,
+      "loss": 0.495265394449234,
+      "mean_token_accuracy": 0.8343238830566406,
+      "num_tokens": 2735022.0,
+      "step": 302
+    },
+    {
+      "epoch": 0.23024316109422494,
+      "grad_norm": 3.5503528118133545,
+      "learning_rate": 4.9905171440460645e-06,
+      "loss": 0.46063232421875,
+      "mean_token_accuracy": 0.8420047760009766,
+      "num_tokens": 2738550.0,
+      "step": 303
+    },
+    {
+      "epoch": 0.23100303951367782,
+      "grad_norm": 3.9858486652374268,
+      "learning_rate": 4.990334022857932e-06,
+      "loss": 0.5832710266113281,
+      "mean_token_accuracy": 0.8144199848175049,
+      "num_tokens": 2741720.0,
+      "step": 304
+    },
+    {
+      "epoch": 0.2317629179331307,
+      "grad_norm": 2.407231330871582,
+      "learning_rate": 4.990149153857519e-06,
+      "loss": 0.4692630171775818,
+      "mean_token_accuracy": 0.8429223299026489,
+      "num_tokens": 2748693.0,
+      "step": 305
+    },
+    {
+      "epoch": 0.23252279635258358,
+      "grad_norm": 1.6996397972106934,
+      "learning_rate": 4.989962537174573e-06,
+      "loss": 0.49143946170806885,
+      "mean_token_accuracy": 0.8340128064155579,
+      "num_tokens": 2761254.0,
+      "step": 306
+    },
+    {
+      "epoch": 0.23328267477203649,
+      "grad_norm": 3.746432065963745,
+      "learning_rate": 4.989774172940071e-06,
+      "loss": 0.6282026767730713,
+      "mean_token_accuracy": 0.775698184967041,
+      "num_tokens": 2765115.0,
+      "step": 307
+    },
+    {
+      "epoch": 0.23404255319148937,
+      "grad_norm": 2.212872266769409,
+      "learning_rate": 4.989584061286211e-06,
+      "loss": 0.5193763971328735,
+      "mean_token_accuracy": 0.8168246746063232,
+      "num_tokens": 2772345.0,
+      "step": 308
+    },
+    {
+      "epoch": 0.23480243161094225,
+      "grad_norm": 1.752297282218933,
+      "learning_rate": 4.989392202346423e-06,
+      "loss": 0.4437984824180603,
+      "mean_token_accuracy": 0.8451256155967712,
+      "num_tokens": 2783072.0,
+      "step": 309
+    },
+    {
+      "epoch": 0.23556231003039513,
+      "grad_norm": 2.386019706726074,
+      "learning_rate": 4.989198596255361e-06,
+      "loss": 0.4090752899646759,
+      "mean_token_accuracy": 0.8480085134506226,
+      "num_tokens": 2788757.0,
+      "step": 310
+    },
+    {
+      "epoch": 0.23632218844984804,
+      "grad_norm": 3.9981489181518555,
+      "learning_rate": 4.989003243148904e-06,
+      "loss": 0.5149132013320923,
+      "mean_token_accuracy": 0.8179056644439697,
+      "num_tokens": 2792096.0,
+      "step": 311
+    },
+    {
+      "epoch": 0.23708206686930092,
+      "grad_norm": 1.8723100423812866,
+      "learning_rate": 4.988806143164159e-06,
+      "loss": 0.4531487822532654,
+      "mean_token_accuracy": 0.8400167226791382,
+      "num_tokens": 2802210.0,
+      "step": 312
+    },
+    {
+      "epoch": 0.2378419452887538,
+      "grad_norm": 2.3415136337280273,
+      "learning_rate": 4.988607296439459e-06,
+      "loss": 0.5974439978599548,
+      "mean_token_accuracy": 0.8035976886749268,
+      "num_tokens": 2810088.0,
+      "step": 313
+    },
+    {
+      "epoch": 0.23860182370820668,
+      "grad_norm": 1.5317577123641968,
+      "learning_rate": 4.98840670311436e-06,
+      "loss": 0.49247145652770996,
+      "mean_token_accuracy": 0.8292540311813354,
+      "num_tokens": 2824005.0,
+      "step": 314
+    },
+    {
+      "epoch": 0.2393617021276596,
+      "grad_norm": 2.170772075653076,
+      "learning_rate": 4.988204363329648e-06,
+      "loss": 0.6359974145889282,
+      "mean_token_accuracy": 0.7785564661026001,
+      "num_tokens": 2834680.0,
+      "step": 315
+    },
+    {
+      "epoch": 0.24012158054711247,
+      "grad_norm": 3.2655932903289795,
+      "learning_rate": 4.988000277227334e-06,
+      "loss": 0.5080196857452393,
+      "mean_token_accuracy": 0.8295877575874329,
+      "num_tokens": 2838735.0,
+      "step": 316
+    },
+    {
+      "epoch": 0.24088145896656535,
+      "grad_norm": 3.406589984893799,
+      "learning_rate": 4.987794444950651e-06,
+      "loss": 0.3939085006713867,
+      "mean_token_accuracy": 0.8700719475746155,
+      "num_tokens": 2842127.0,
+      "step": 317
+    },
+    {
+      "epoch": 0.24164133738601823,
+      "grad_norm": 1.8211106061935425,
+      "learning_rate": 4.987586866644061e-06,
+      "loss": 0.5270540118217468,
+      "mean_token_accuracy": 0.826683521270752,
+      "num_tokens": 2853656.0,
+      "step": 318
+    },
+    {
+      "epoch": 0.24240121580547114,
+      "grad_norm": 1.8429969549179077,
+      "learning_rate": 4.9873775424532515e-06,
+      "loss": 0.4705049991607666,
+      "mean_token_accuracy": 0.8355701565742493,
+      "num_tokens": 2863513.0,
+      "step": 319
+    },
+    {
+      "epoch": 0.24316109422492402,
+      "grad_norm": 2.2425320148468018,
+      "learning_rate": 4.9871664725251314e-06,
+      "loss": 0.485736608505249,
+      "mean_token_accuracy": 0.835182785987854,
+      "num_tokens": 2871556.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.2439209726443769,
+      "grad_norm": 1.6202056407928467,
+      "learning_rate": 4.986953657007841e-06,
+      "loss": 0.4437887370586395,
+      "mean_token_accuracy": 0.8282591700553894,
+      "num_tokens": 2884335.0,
+      "step": 321
+    },
+    {
+      "epoch": 0.24468085106382978,
+      "grad_norm": 1.1027268171310425,
+      "learning_rate": 4.98673909605074e-06,
+      "loss": 0.3770800828933716,
+      "mean_token_accuracy": 0.8325437307357788,
+      "num_tokens": 2904286.0,
+      "step": 322
+    },
+    {
+      "epoch": 0.2454407294832827,
+      "grad_norm": 2.3239076137542725,
+      "learning_rate": 4.986522789804417e-06,
+      "loss": 0.5387254953384399,
+      "mean_token_accuracy": 0.806242823600769,
+      "num_tokens": 2910975.0,
+      "step": 323
+    },
+    {
+      "epoch": 0.24620060790273557,
+      "grad_norm": 2.243482828140259,
+      "learning_rate": 4.986304738420684e-06,
+      "loss": 0.4396553039550781,
+      "mean_token_accuracy": 0.8561904430389404,
+      "num_tokens": 2917087.0,
+      "step": 324
+    },
+    {
+      "epoch": 0.24696048632218845,
+      "grad_norm": 2.537264347076416,
+      "learning_rate": 4.986084942052577e-06,
+      "loss": 0.395110160112381,
+      "mean_token_accuracy": 0.8636915683746338,
+      "num_tokens": 2921887.0,
+      "step": 325
+    },
+    {
+      "epoch": 0.24772036474164133,
+      "grad_norm": 2.319399118423462,
+      "learning_rate": 4.9858634008543574e-06,
+      "loss": 0.581517219543457,
+      "mean_token_accuracy": 0.8157487511634827,
+      "num_tokens": 2928996.0,
+      "step": 326
+    },
+    {
+      "epoch": 0.24848024316109424,
+      "grad_norm": 1.9787474870681763,
+      "learning_rate": 4.985640114981513e-06,
+      "loss": 0.5084106922149658,
+      "mean_token_accuracy": 0.835221529006958,
+      "num_tokens": 2940302.0,
+      "step": 327
+    },
+    {
+      "epoch": 0.24924012158054712,
+      "grad_norm": 2.4783265590667725,
+      "learning_rate": 4.985415084590752e-06,
+      "loss": 0.6062222719192505,
+      "mean_token_accuracy": 0.7885516285896301,
+      "num_tokens": 2946386.0,
+      "step": 328
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4081411361694336,
+      "learning_rate": 4.985188309840012e-06,
+      "loss": 0.5079880356788635,
+      "mean_token_accuracy": 0.8313904404640198,
+      "num_tokens": 2952323.0,
+      "step": 329
+    },
+    {
+      "epoch": 0.2507598784194529,
+      "grad_norm": 2.64993953704834,
+      "learning_rate": 4.984959790888451e-06,
+      "loss": 0.5461447834968567,
+      "mean_token_accuracy": 0.8125468492507935,
+      "num_tokens": 2958119.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.25151975683890576,
+      "grad_norm": 2.549734115600586,
+      "learning_rate": 4.984729527896451e-06,
+      "loss": 0.5998573303222656,
+      "mean_token_accuracy": 0.8076666593551636,
+      "num_tokens": 2964947.0,
+      "step": 331
+    },
+    {
+      "epoch": 0.25227963525835867,
+      "grad_norm": 3.2185161113739014,
+      "learning_rate": 4.984497521025622e-06,
+      "loss": 0.4232945442199707,
+      "mean_token_accuracy": 0.8543803095817566,
+      "num_tokens": 2968598.0,
+      "step": 332
+    },
+    {
+      "epoch": 0.2530395136778115,
+      "grad_norm": 2.588994264602661,
+      "learning_rate": 4.984263770438793e-06,
+      "loss": 0.460967481136322,
+      "mean_token_accuracy": 0.8416207432746887,
+      "num_tokens": 2974510.0,
+      "step": 333
+    },
+    {
+      "epoch": 0.25379939209726443,
+      "grad_norm": 2.1373162269592285,
+      "learning_rate": 4.984028276300021e-06,
+      "loss": 0.49382102489471436,
+      "mean_token_accuracy": 0.8388048410415649,
+      "num_tokens": 2981632.0,
+      "step": 334
+    },
+    {
+      "epoch": 0.25455927051671734,
+      "grad_norm": 2.2524826526641846,
+      "learning_rate": 4.983791038774585e-06,
+      "loss": 0.4947671890258789,
+      "mean_token_accuracy": 0.8066365122795105,
+      "num_tokens": 2988736.0,
+      "step": 335
+    },
+    {
+      "epoch": 0.2553191489361702,
+      "grad_norm": 1.7244199514389038,
+      "learning_rate": 4.983552058028985e-06,
+      "loss": 0.48096776008605957,
+      "mean_token_accuracy": 0.830735445022583,
+      "num_tokens": 3003576.0,
+      "step": 336
+    },
+    {
+      "epoch": 0.2560790273556231,
+      "grad_norm": 3.0628933906555176,
+      "learning_rate": 4.9833113342309495e-06,
+      "loss": 0.6027032136917114,
+      "mean_token_accuracy": 0.8008694648742676,
+      "num_tokens": 3009549.0,
+      "step": 337
+    },
+    {
+      "epoch": 0.256838905775076,
+      "grad_norm": 2.438674211502075,
+      "learning_rate": 4.983068867549427e-06,
+      "loss": 0.517090916633606,
+      "mean_token_accuracy": 0.827893853187561,
+      "num_tokens": 3015236.0,
+      "step": 338
+    },
+    {
+      "epoch": 0.25759878419452886,
+      "grad_norm": 2.131535053253174,
+      "learning_rate": 4.982824658154589e-06,
+      "loss": 0.6656812429428101,
+      "mean_token_accuracy": 0.7772425413131714,
+      "num_tokens": 3028142.0,
+      "step": 339
+    },
+    {
+      "epoch": 0.25835866261398177,
+      "grad_norm": 2.3206584453582764,
+      "learning_rate": 4.9825787062178315e-06,
+      "loss": 0.5757625699043274,
+      "mean_token_accuracy": 0.8073873519897461,
+      "num_tokens": 3040996.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.2591185410334346,
+      "grad_norm": 1.3905521631240845,
+      "learning_rate": 4.982331011911774e-06,
+      "loss": 0.4193805456161499,
+      "mean_token_accuracy": 0.8399466872215271,
+      "num_tokens": 3061931.0,
+      "step": 341
+    },
+    {
+      "epoch": 0.25987841945288753,
+      "grad_norm": 2.184173345565796,
+      "learning_rate": 4.982081575410256e-06,
+      "loss": 0.4751223921775818,
+      "mean_token_accuracy": 0.8409271240234375,
+      "num_tokens": 3069081.0,
+      "step": 342
+    },
+    {
+      "epoch": 0.26063829787234044,
+      "grad_norm": 3.538764238357544,
+      "learning_rate": 4.9818303968883445e-06,
+      "loss": 0.8119601011276245,
+      "mean_token_accuracy": 0.7442739009857178,
+      "num_tokens": 3073628.0,
+      "step": 343
+    },
+    {
+      "epoch": 0.2613981762917933,
+      "grad_norm": 1.8063762187957764,
+      "learning_rate": 4.981577476522323e-06,
+      "loss": 0.5615730881690979,
+      "mean_token_accuracy": 0.8207751512527466,
+      "num_tokens": 3086596.0,
+      "step": 344
+    },
+    {
+      "epoch": 0.2621580547112462,
+      "grad_norm": 2.4346961975097656,
+      "learning_rate": 4.981322814489703e-06,
+      "loss": 0.5266709327697754,
+      "mean_token_accuracy": 0.8211277723312378,
+      "num_tokens": 3092631.0,
+      "step": 345
+    },
+    {
+      "epoch": 0.2629179331306991,
+      "grad_norm": 1.91289484500885,
+      "learning_rate": 4.981066410969215e-06,
+      "loss": 0.5047177672386169,
+      "mean_token_accuracy": 0.8356877565383911,
+      "num_tokens": 3101102.0,
+      "step": 346
+    },
+    {
+      "epoch": 0.26367781155015196,
+      "grad_norm": 2.1495707035064697,
+      "learning_rate": 4.980808266140813e-06,
+      "loss": 0.47876280546188354,
+      "mean_token_accuracy": 0.8364313244819641,
+      "num_tokens": 3107998.0,
+      "step": 347
+    },
+    {
+      "epoch": 0.26443768996960487,
+      "grad_norm": 2.5961992740631104,
+      "learning_rate": 4.9805483801856744e-06,
+      "loss": 0.5512958765029907,
+      "mean_token_accuracy": 0.8181467652320862,
+      "num_tokens": 3113848.0,
+      "step": 348
+    },
+    {
+      "epoch": 0.2651975683890577,
+      "grad_norm": 3.2828900814056396,
+      "learning_rate": 4.980286753286196e-06,
+      "loss": 0.4217945635318756,
+      "mean_token_accuracy": 0.8617103099822998,
+      "num_tokens": 3117652.0,
+      "step": 349
+    },
+    {
+      "epoch": 0.26595744680851063,
+      "grad_norm": 1.425554871559143,
+      "learning_rate": 4.980023385625996e-06,
+      "loss": 0.4042487144470215,
+      "mean_token_accuracy": 0.8492785692214966,
+      "num_tokens": 3132336.0,
+      "step": 350
+    },
+    {
+      "epoch": 0.26671732522796354,
+      "grad_norm": 2.933504104614258,
+      "learning_rate": 4.979758277389919e-06,
+      "loss": 0.5406704545021057,
+      "mean_token_accuracy": 0.8035423755645752,
+      "num_tokens": 3137544.0,
+      "step": 351
+    },
+    {
+      "epoch": 0.2674772036474164,
+      "grad_norm": 1.9958966970443726,
+      "learning_rate": 4.9794914287640264e-06,
+      "loss": 0.5857555270195007,
+      "mean_token_accuracy": 0.7965140342712402,
+      "num_tokens": 3149705.0,
+      "step": 352
+    },
+    {
+      "epoch": 0.2682370820668693,
+      "grad_norm": 2.467694044113159,
+      "learning_rate": 4.979222839935602e-06,
+      "loss": 0.6404043436050415,
+      "mean_token_accuracy": 0.7823755741119385,
+      "num_tokens": 3158353.0,
+      "step": 353
+    },
+    {
+      "epoch": 0.2689969604863222,
+      "grad_norm": 2.0102720260620117,
+      "learning_rate": 4.9789525110931545e-06,
+      "loss": 0.5681496858596802,
+      "mean_token_accuracy": 0.8108169436454773,
+      "num_tokens": 3167121.0,
+      "step": 354
+    },
+    {
+      "epoch": 0.26975683890577506,
+      "grad_norm": 2.6017866134643555,
+      "learning_rate": 4.978680442426409e-06,
+      "loss": 0.6309828162193298,
+      "mean_token_accuracy": 0.7742617130279541,
+      "num_tokens": 3175012.0,
+      "step": 355
+    },
+    {
+      "epoch": 0.270516717325228,
+      "grad_norm": 1.8799268007278442,
+      "learning_rate": 4.978406634126315e-06,
+      "loss": 0.524029016494751,
+      "mean_token_accuracy": 0.8317689895629883,
+      "num_tokens": 3185331.0,
+      "step": 356
+    },
+    {
+      "epoch": 0.2712765957446808,
+      "grad_norm": 1.508332371711731,
+      "learning_rate": 4.978131086385041e-06,
+      "loss": 0.46656402945518494,
+      "mean_token_accuracy": 0.8339117765426636,
+      "num_tokens": 3198813.0,
+      "step": 357
+    },
+    {
+      "epoch": 0.27203647416413373,
+      "grad_norm": 3.595707654953003,
+      "learning_rate": 4.977853799395976e-06,
+      "loss": 0.5101234912872314,
+      "mean_token_accuracy": 0.8251723051071167,
+      "num_tokens": 3206557.0,
+      "step": 358
+    },
+    {
+      "epoch": 0.27279635258358664,
+      "grad_norm": 3.5317916870117188,
+      "learning_rate": 4.977574773353732e-06,
+      "loss": 0.5684665441513062,
+      "mean_token_accuracy": 0.8124493360519409,
+      "num_tokens": 3210912.0,
+      "step": 359
+    },
+    {
+      "epoch": 0.2735562310030395,
+      "grad_norm": 2.8606204986572266,
+      "learning_rate": 4.97729400845414e-06,
+      "loss": 0.4746384620666504,
+      "mean_token_accuracy": 0.8195606470108032,
+      "num_tokens": 3215365.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.2743161094224924,
+      "grad_norm": 1.8214033842086792,
+      "learning_rate": 4.977011504894253e-06,
+      "loss": 0.4842769503593445,
+      "mean_token_accuracy": 0.82928866147995,
+      "num_tokens": 3224037.0,
+      "step": 361
+    },
+    {
+      "epoch": 0.2750759878419453,
+      "grad_norm": 1.628746509552002,
+      "learning_rate": 4.97672726287234e-06,
+      "loss": 0.4397493302822113,
+      "mean_token_accuracy": 0.8606528043746948,
+      "num_tokens": 3235589.0,
+      "step": 362
+    },
+    {
+      "epoch": 0.27583586626139817,
+      "grad_norm": 3.557973861694336,
+      "learning_rate": 4.976441282587894e-06,
+      "loss": 0.5732032060623169,
+      "mean_token_accuracy": 0.8041545748710632,
+      "num_tokens": 3239958.0,
+      "step": 363
+    },
+    {
+      "epoch": 0.2765957446808511,
+      "grad_norm": 1.3467901945114136,
+      "learning_rate": 4.9761535642416284e-06,
+      "loss": 0.4525323510169983,
+      "mean_token_accuracy": 0.8281061053276062,
+      "num_tokens": 3257703.0,
+      "step": 364
+    },
+    {
+      "epoch": 0.2773556231003039,
+      "grad_norm": 2.2649986743927,
+      "learning_rate": 4.9758641080354745e-06,
+      "loss": 0.5074734687805176,
+      "mean_token_accuracy": 0.8447474241256714,
+      "num_tokens": 3264334.0,
+      "step": 365
+    },
+    {
+      "epoch": 0.27811550151975684,
+      "grad_norm": 2.8667566776275635,
+      "learning_rate": 4.975572914172581e-06,
+      "loss": 0.5759559869766235,
+      "mean_token_accuracy": 0.7976793050765991,
+      "num_tokens": 3269314.0,
+      "step": 366
+    },
+    {
+      "epoch": 0.27887537993920974,
+      "grad_norm": 2.2514986991882324,
+      "learning_rate": 4.975279982857324e-06,
+      "loss": 0.5786465406417847,
+      "mean_token_accuracy": 0.8058781623840332,
+      "num_tokens": 3277324.0,
+      "step": 367
+    },
+    {
+      "epoch": 0.2796352583586626,
+      "grad_norm": 1.3826723098754883,
+      "learning_rate": 4.97498531429529e-06,
+      "loss": 0.40801727771759033,
+      "mean_token_accuracy": 0.8601310849189758,
+      "num_tokens": 3290530.0,
+      "step": 368
+    },
+    {
+      "epoch": 0.2803951367781155,
+      "grad_norm": 2.084092617034912,
+      "learning_rate": 4.97468890869329e-06,
+      "loss": 0.47076648473739624,
+      "mean_token_accuracy": 0.8310186862945557,
+      "num_tokens": 3298325.0,
+      "step": 369
+    },
+    {
+      "epoch": 0.2811550151975684,
+      "grad_norm": 1.3467998504638672,
+      "learning_rate": 4.974390766259353e-06,
+      "loss": 0.44668465852737427,
+      "mean_token_accuracy": 0.8275353908538818,
+      "num_tokens": 3314302.0,
+      "step": 370
+    },
+    {
+      "epoch": 0.28191489361702127,
+      "grad_norm": 2.5921075344085693,
+      "learning_rate": 4.974090887202726e-06,
+      "loss": 0.5343953967094421,
+      "mean_token_accuracy": 0.8110706806182861,
+      "num_tokens": 3320963.0,
+      "step": 371
+    },
+    {
+      "epoch": 0.2826747720364742,
+      "grad_norm": 2.042781352996826,
+      "learning_rate": 4.973789271733877e-06,
+      "loss": 0.6293343305587769,
+      "mean_token_accuracy": 0.7800243496894836,
+      "num_tokens": 3332742.0,
+      "step": 372
+    },
+    {
+      "epoch": 0.28343465045592703,
+      "grad_norm": 4.822193145751953,
+      "learning_rate": 4.973485920064491e-06,
+      "loss": 0.6256728768348694,
+      "mean_token_accuracy": 0.7962433099746704,
+      "num_tokens": 3335872.0,
+      "step": 373
+    },
+    {
+      "epoch": 0.28419452887537994,
+      "grad_norm": 1.260988473892212,
+      "learning_rate": 4.973180832407471e-06,
+      "loss": 0.38731223344802856,
+      "mean_token_accuracy": 0.8385066986083984,
+      "num_tokens": 3351884.0,
+      "step": 374
+    },
+    {
+      "epoch": 0.28495440729483285,
+      "grad_norm": 2.669966697692871,
+      "learning_rate": 4.97287400897694e-06,
+      "loss": 0.5594710111618042,
+      "mean_token_accuracy": 0.8097212314605713,
+      "num_tokens": 3358197.0,
+      "step": 375
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 3.0344486236572266,
+      "learning_rate": 4.972565449988238e-06,
+      "loss": 0.34449583292007446,
+      "mean_token_accuracy": 0.8813316822052002,
+      "num_tokens": 3362133.0,
+      "step": 376
+    },
+    {
+      "epoch": 0.2864741641337386,
+      "grad_norm": 2.562251091003418,
+      "learning_rate": 4.972255155657925e-06,
+      "loss": 0.5331522822380066,
+      "mean_token_accuracy": 0.8212941288948059,
+      "num_tokens": 3370346.0,
+      "step": 377
+    },
+    {
+      "epoch": 0.2872340425531915,
+      "grad_norm": 2.7083740234375,
+      "learning_rate": 4.9719431262037755e-06,
+      "loss": 0.5403046011924744,
+      "mean_token_accuracy": 0.8108335733413696,
+      "num_tokens": 3375588.0,
+      "step": 378
+    },
+    {
+      "epoch": 0.28799392097264437,
+      "grad_norm": 1.396430492401123,
+      "learning_rate": 4.971629361844785e-06,
+      "loss": 0.4041529893875122,
+      "mean_token_accuracy": 0.8588063716888428,
+      "num_tokens": 3390749.0,
+      "step": 379
+    },
+    {
+      "epoch": 0.2887537993920973,
+      "grad_norm": 1.9872784614562988,
+      "learning_rate": 4.971313862801166e-06,
+      "loss": 0.4336993098258972,
+      "mean_token_accuracy": 0.8511303663253784,
+      "num_tokens": 3399064.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.28951367781155013,
+      "grad_norm": 1.9652575254440308,
+      "learning_rate": 4.9709966292943455e-06,
+      "loss": 0.4578358232975006,
+      "mean_token_accuracy": 0.8229440450668335,
+      "num_tokens": 3407229.0,
+      "step": 381
+    },
+    {
+      "epoch": 0.29027355623100304,
+      "grad_norm": 1.6626898050308228,
+      "learning_rate": 4.970677661546972e-06,
+      "loss": 0.5427594184875488,
+      "mean_token_accuracy": 0.815427303314209,
+      "num_tokens": 3422321.0,
+      "step": 382
+    },
+    {
+      "epoch": 0.29103343465045595,
+      "grad_norm": 3.5265562534332275,
+      "learning_rate": 4.970356959782909e-06,
+      "loss": 0.6661460995674133,
+      "mean_token_accuracy": 0.7856965065002441,
+      "num_tokens": 3427442.0,
+      "step": 383
+    },
+    {
+      "epoch": 0.2917933130699088,
+      "grad_norm": 1.667205572128296,
+      "learning_rate": 4.970034524227239e-06,
+      "loss": 0.36256325244903564,
+      "mean_token_accuracy": 0.8711205720901489,
+      "num_tokens": 3436662.0,
+      "step": 384
+    },
+    {
+      "epoch": 0.2925531914893617,
+      "grad_norm": 1.3389486074447632,
+      "learning_rate": 4.969710355106256e-06,
+      "loss": 0.4282698631286621,
+      "mean_token_accuracy": 0.838951587677002,
+      "num_tokens": 3450060.0,
+      "step": 385
+    },
+    {
+      "epoch": 0.2933130699088146,
+      "grad_norm": 2.5163397789001465,
+      "learning_rate": 4.969384452647477e-06,
+      "loss": 0.5176984071731567,
+      "mean_token_accuracy": 0.8235267996788025,
+      "num_tokens": 3456990.0,
+      "step": 386
+    },
+    {
+      "epoch": 0.29407294832826747,
+      "grad_norm": 1.7588495016098022,
+      "learning_rate": 4.969056817079633e-06,
+      "loss": 0.49710947275161743,
+      "mean_token_accuracy": 0.818520724773407,
+      "num_tokens": 3468098.0,
+      "step": 387
+    },
+    {
+      "epoch": 0.2948328267477204,
+      "grad_norm": 2.6381046772003174,
+      "learning_rate": 4.968727448632669e-06,
+      "loss": 0.4425308108329773,
+      "mean_token_accuracy": 0.8451643586158752,
+      "num_tokens": 3472899.0,
+      "step": 388
+    },
+    {
+      "epoch": 0.29559270516717323,
+      "grad_norm": 1.6345038414001465,
+      "learning_rate": 4.968396347537751e-06,
+      "loss": 0.4177059829235077,
+      "mean_token_accuracy": 0.8498886227607727,
+      "num_tokens": 3484826.0,
+      "step": 389
+    },
+    {
+      "epoch": 0.29635258358662614,
+      "grad_norm": 3.0466468334198,
+      "learning_rate": 4.968063514027258e-06,
+      "loss": 0.4274463951587677,
+      "mean_token_accuracy": 0.8387278318405151,
+      "num_tokens": 3488610.0,
+      "step": 390
+    },
+    {
+      "epoch": 0.29711246200607905,
+      "grad_norm": 2.6509406566619873,
+      "learning_rate": 4.967728948334784e-06,
+      "loss": 0.5401753783226013,
+      "mean_token_accuracy": 0.8252490162849426,
+      "num_tokens": 3493657.0,
+      "step": 391
+    },
+    {
+      "epoch": 0.2978723404255319,
+      "grad_norm": 1.6372219324111938,
+      "learning_rate": 4.967392650695141e-06,
+      "loss": 0.3862472176551819,
+      "mean_token_accuracy": 0.8555525541305542,
+      "num_tokens": 3505588.0,
+      "step": 392
+    },
+    {
+      "epoch": 0.2986322188449848,
+      "grad_norm": 2.1615452766418457,
+      "learning_rate": 4.967054621344356e-06,
+      "loss": 0.57850581407547,
+      "mean_token_accuracy": 0.8222678899765015,
+      "num_tokens": 3514396.0,
+      "step": 393
+    },
+    {
+      "epoch": 0.2993920972644377,
+      "grad_norm": 1.8610916137695312,
+      "learning_rate": 4.96671486051967e-06,
+      "loss": 0.5440595149993896,
+      "mean_token_accuracy": 0.8196715116500854,
+      "num_tokens": 3523604.0,
+      "step": 394
+    },
+    {
+      "epoch": 0.30015197568389057,
+      "grad_norm": 2.9585862159729004,
+      "learning_rate": 4.966373368459542e-06,
+      "loss": 0.6921588182449341,
+      "mean_token_accuracy": 0.7816659808158875,
+      "num_tokens": 3529849.0,
+      "step": 395
+    },
+    {
+      "epoch": 0.3009118541033435,
+      "grad_norm": 1.9374035596847534,
+      "learning_rate": 4.966030145403642e-06,
+      "loss": 0.5494055151939392,
+      "mean_token_accuracy": 0.8126792907714844,
+      "num_tokens": 3539529.0,
+      "step": 396
+    },
+    {
+      "epoch": 0.30167173252279633,
+      "grad_norm": 1.730530023574829,
+      "learning_rate": 4.965685191592859e-06,
+      "loss": 0.4271572232246399,
+      "mean_token_accuracy": 0.8383668661117554,
+      "num_tokens": 3550972.0,
+      "step": 397
+    },
+    {
+      "epoch": 0.30243161094224924,
+      "grad_norm": 3.9635560512542725,
+      "learning_rate": 4.9653385072692935e-06,
+      "loss": 0.5576210021972656,
+      "mean_token_accuracy": 0.799404501914978,
+      "num_tokens": 3554147.0,
+      "step": 398
+    },
+    {
+      "epoch": 0.30319148936170215,
+      "grad_norm": 2.5731968879699707,
+      "learning_rate": 4.964990092676263e-06,
+      "loss": 0.5478942394256592,
+      "mean_token_accuracy": 0.8220961093902588,
+      "num_tokens": 3559972.0,
+      "step": 399
+    },
+    {
+      "epoch": 0.303951367781155,
+      "grad_norm": 2.2096588611602783,
+      "learning_rate": 4.964639948058297e-06,
+      "loss": 0.35461270809173584,
+      "mean_token_accuracy": 0.8640927076339722,
+      "num_tokens": 3565770.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.3047112462006079,
+      "grad_norm": 1.7874189615249634,
+      "learning_rate": 4.964288073661142e-06,
+      "loss": 0.38849619030952454,
+      "mean_token_accuracy": 0.8443037271499634,
+      "num_tokens": 3574514.0,
+      "step": 401
+    },
+    {
+      "epoch": 0.30547112462006076,
+      "grad_norm": 1.5583146810531616,
+      "learning_rate": 4.963934469731756e-06,
+      "loss": 0.48909449577331543,
+      "mean_token_accuracy": 0.8429768681526184,
+      "num_tokens": 3585877.0,
+      "step": 402
+    },
+    {
+      "epoch": 0.30623100303951367,
+      "grad_norm": 3.026599645614624,
+      "learning_rate": 4.963579136518312e-06,
+      "loss": 0.5138992071151733,
+      "mean_token_accuracy": 0.8283728361129761,
+      "num_tokens": 3590412.0,
+      "step": 403
+    },
+    {
+      "epoch": 0.3069908814589666,
+      "grad_norm": 2.777505874633789,
+      "learning_rate": 4.963222074270197e-06,
+      "loss": 0.6241534948348999,
+      "mean_token_accuracy": 0.8130464553833008,
+      "num_tokens": 3596246.0,
+      "step": 404
+    },
+    {
+      "epoch": 0.30775075987841943,
+      "grad_norm": 2.4772839546203613,
+      "learning_rate": 4.962863283238011e-06,
+      "loss": 0.5930814146995544,
+      "mean_token_accuracy": 0.8036394715309143,
+      "num_tokens": 3602878.0,
+      "step": 405
+    },
+    {
+      "epoch": 0.30851063829787234,
+      "grad_norm": 1.5049982070922852,
+      "learning_rate": 4.962502763673566e-06,
+      "loss": 0.4903082549571991,
+      "mean_token_accuracy": 0.8184912204742432,
+      "num_tokens": 3617018.0,
+      "step": 406
+    },
+    {
+      "epoch": 0.30927051671732525,
+      "grad_norm": 2.453155040740967,
+      "learning_rate": 4.96214051582989e-06,
+      "loss": 0.5138067603111267,
+      "mean_token_accuracy": 0.8336835503578186,
+      "num_tokens": 3624188.0,
+      "step": 407
+    },
+    {
+      "epoch": 0.3100303951367781,
+      "grad_norm": 2.4038336277008057,
+      "learning_rate": 4.961776539961222e-06,
+      "loss": 0.5752760171890259,
+      "mean_token_accuracy": 0.8054730892181396,
+      "num_tokens": 3634152.0,
+      "step": 408
+    },
+    {
+      "epoch": 0.310790273556231,
+      "grad_norm": 2.629068374633789,
+      "learning_rate": 4.961410836323014e-06,
+      "loss": 0.5580606460571289,
+      "mean_token_accuracy": 0.8121089935302734,
+      "num_tokens": 3639528.0,
+      "step": 409
+    },
+    {
+      "epoch": 0.31155015197568386,
+      "grad_norm": 1.4245928525924683,
+      "learning_rate": 4.961043405171931e-06,
+      "loss": 0.5399882793426514,
+      "mean_token_accuracy": 0.812280535697937,
+      "num_tokens": 3655744.0,
+      "step": 410
+    },
+    {
+      "epoch": 0.3123100303951368,
+      "grad_norm": 1.5236459970474243,
+      "learning_rate": 4.9606742467658505e-06,
+      "loss": 0.5234690308570862,
+      "mean_token_accuracy": 0.8188928365707397,
+      "num_tokens": 3675010.0,
+      "step": 411
+    },
+    {
+      "epoch": 0.3130699088145897,
+      "grad_norm": 2.27961802482605,
+      "learning_rate": 4.960303361363863e-06,
+      "loss": 0.5502505898475647,
+      "mean_token_accuracy": 0.8161963224411011,
+      "num_tokens": 3682328.0,
+      "step": 412
+    },
+    {
+      "epoch": 0.31382978723404253,
+      "grad_norm": 1.554518222808838,
+      "learning_rate": 4.959930749226269e-06,
+      "loss": 0.420867919921875,
+      "mean_token_accuracy": 0.8499157428741455,
+      "num_tokens": 3694980.0,
+      "step": 413
+    },
+    {
+      "epoch": 0.31458966565349544,
+      "grad_norm": 2.609218120574951,
+      "learning_rate": 4.9595564106145825e-06,
+      "loss": 0.4706704318523407,
+      "mean_token_accuracy": 0.8412490487098694,
+      "num_tokens": 3700033.0,
+      "step": 414
+    },
+    {
+      "epoch": 0.31534954407294835,
+      "grad_norm": 1.5303231477737427,
+      "learning_rate": 4.959180345791528e-06,
+      "loss": 0.4668654799461365,
+      "mean_token_accuracy": 0.8125015497207642,
+      "num_tokens": 3715012.0,
+      "step": 415
+    },
+    {
+      "epoch": 0.3161094224924012,
+      "grad_norm": 1.2774665355682373,
+      "learning_rate": 4.958802555021042e-06,
+      "loss": 0.4339369237422943,
+      "mean_token_accuracy": 0.8442851901054382,
+      "num_tokens": 3733928.0,
+      "step": 416
+    },
+    {
+      "epoch": 0.3168693009118541,
+      "grad_norm": 2.1240181922912598,
+      "learning_rate": 4.958423038568274e-06,
+      "loss": 0.4029104709625244,
+      "mean_token_accuracy": 0.8627674579620361,
+      "num_tokens": 3740202.0,
+      "step": 417
+    },
+    {
+      "epoch": 0.31762917933130697,
+      "grad_norm": 2.00538969039917,
+      "learning_rate": 4.958041796699583e-06,
+      "loss": 0.5229607820510864,
+      "mean_token_accuracy": 0.8282366394996643,
+      "num_tokens": 3749308.0,
+      "step": 418
+    },
+    {
+      "epoch": 0.3183890577507599,
+      "grad_norm": 2.6555092334747314,
+      "learning_rate": 4.957658829682539e-06,
+      "loss": 0.5344101190567017,
+      "mean_token_accuracy": 0.8183202743530273,
+      "num_tokens": 3754595.0,
+      "step": 419
+    },
+    {
+      "epoch": 0.3191489361702128,
+      "grad_norm": 1.7468839883804321,
+      "learning_rate": 4.9572741377859225e-06,
+      "loss": 0.5667245984077454,
+      "mean_token_accuracy": 0.8080123662948608,
+      "num_tokens": 3765761.0,
+      "step": 420
+    },
+    {
+      "epoch": 0.31990881458966564,
+      "grad_norm": 2.9612457752227783,
+      "learning_rate": 4.956887721279726e-06,
+      "loss": 0.5389559864997864,
+      "mean_token_accuracy": 0.8019476532936096,
+      "num_tokens": 3770844.0,
+      "step": 421
+    },
+    {
+      "epoch": 0.32066869300911854,
+      "grad_norm": 1.842403769493103,
+      "learning_rate": 4.95649958043515e-06,
+      "loss": 0.38279837369918823,
+      "mean_token_accuracy": 0.858866810798645,
+      "num_tokens": 3778094.0,
+      "step": 422
+    },
+    {
+      "epoch": 0.32142857142857145,
+      "grad_norm": 2.3108131885528564,
+      "learning_rate": 4.956109715524609e-06,
+      "loss": 0.5453893542289734,
+      "mean_token_accuracy": 0.8085013031959534,
+      "num_tokens": 3785015.0,
+      "step": 423
+    },
+    {
+      "epoch": 0.3221884498480243,
+      "grad_norm": 3.0326945781707764,
+      "learning_rate": 4.9557181268217225e-06,
+      "loss": 0.5550523400306702,
+      "mean_token_accuracy": 0.8125876188278198,
+      "num_tokens": 3789830.0,
+      "step": 424
+    },
+    {
+      "epoch": 0.3229483282674772,
+      "grad_norm": 1.8851977586746216,
+      "learning_rate": 4.955324814601324e-06,
+      "loss": 0.4902324974536896,
+      "mean_token_accuracy": 0.8205406665802002,
+      "num_tokens": 3799862.0,
+      "step": 425
+    },
+    {
+      "epoch": 0.32370820668693007,
+      "grad_norm": 2.6018171310424805,
+      "learning_rate": 4.954929779139455e-06,
+      "loss": 0.5920133590698242,
+      "mean_token_accuracy": 0.8340690732002258,
+      "num_tokens": 3806617.0,
+      "step": 426
+    },
+    {
+      "epoch": 0.324468085106383,
+      "grad_norm": 2.4283878803253174,
+      "learning_rate": 4.954533020713367e-06,
+      "loss": 0.5305854082107544,
+      "mean_token_accuracy": 0.8137468099594116,
+      "num_tokens": 3813843.0,
+      "step": 427
+    },
+    {
+      "epoch": 0.3252279635258359,
+      "grad_norm": 2.667978525161743,
+      "learning_rate": 4.954134539601519e-06,
+      "loss": 0.5333638787269592,
+      "mean_token_accuracy": 0.8402629494667053,
+      "num_tokens": 3819450.0,
+      "step": 428
+    },
+    {
+      "epoch": 0.32598784194528874,
+      "grad_norm": 1.7302523851394653,
+      "learning_rate": 4.953734336083582e-06,
+      "loss": 0.422895610332489,
+      "mean_token_accuracy": 0.8709704875946045,
+      "num_tokens": 3831027.0,
+      "step": 429
+    },
+    {
+      "epoch": 0.32674772036474165,
+      "grad_norm": 2.427192211151123,
+      "learning_rate": 4.953332410440434e-06,
+      "loss": 0.6334598064422607,
+      "mean_token_accuracy": 0.7817479968070984,
+      "num_tokens": 3841776.0,
+      "step": 430
+    },
+    {
+      "epoch": 0.32750759878419455,
+      "grad_norm": 1.460949182510376,
+      "learning_rate": 4.952928762954161e-06,
+      "loss": 0.3654777705669403,
+      "mean_token_accuracy": 0.8780122995376587,
+      "num_tokens": 3852213.0,
+      "step": 431
+    },
+    {
+      "epoch": 0.3282674772036474,
+      "grad_norm": 1.9855005741119385,
+      "learning_rate": 4.952523393908059e-06,
+      "loss": 0.5117089748382568,
+      "mean_token_accuracy": 0.811911404132843,
+      "num_tokens": 3861176.0,
+      "step": 432
+    },
+    {
+      "epoch": 0.3290273556231003,
+      "grad_norm": 2.2653207778930664,
+      "learning_rate": 4.952116303586631e-06,
+      "loss": 0.42514950037002563,
+      "mean_token_accuracy": 0.8448518514633179,
+      "num_tokens": 3867164.0,
+      "step": 433
+    },
+    {
+      "epoch": 0.32978723404255317,
+      "grad_norm": 1.9780964851379395,
+      "learning_rate": 4.951707492275589e-06,
+      "loss": 0.5095293521881104,
+      "mean_token_accuracy": 0.8262748718261719,
+      "num_tokens": 3876406.0,
+      "step": 434
+    },
+    {
+      "epoch": 0.3305471124620061,
+      "grad_norm": 2.9480233192443848,
+      "learning_rate": 4.951296960261853e-06,
+      "loss": 0.3494448959827423,
+      "mean_token_accuracy": 0.8781307935714722,
+      "num_tokens": 3880298.0,
+      "step": 435
+    },
+    {
+      "epoch": 0.331306990881459,
+      "grad_norm": 2.335571527481079,
+      "learning_rate": 4.95088470783355e-06,
+      "loss": 0.5456914901733398,
+      "mean_token_accuracy": 0.816297173500061,
+      "num_tokens": 3886487.0,
+      "step": 436
+    },
+    {
+      "epoch": 0.33206686930091184,
+      "grad_norm": 2.3046419620513916,
+      "learning_rate": 4.950470735280013e-06,
+      "loss": 0.4835948944091797,
+      "mean_token_accuracy": 0.8539175391197205,
+      "num_tokens": 3892706.0,
+      "step": 437
+    },
+    {
+      "epoch": 0.33282674772036475,
+      "grad_norm": 2.44047474861145,
+      "learning_rate": 4.950055042891786e-06,
+      "loss": 0.5154092907905579,
+      "mean_token_accuracy": 0.8579919338226318,
+      "num_tokens": 3899532.0,
+      "step": 438
+    },
+    {
+      "epoch": 0.33358662613981765,
+      "grad_norm": 4.826764106750488,
+      "learning_rate": 4.949637630960618e-06,
+      "loss": 0.5270259976387024,
+      "mean_token_accuracy": 0.8172192573547363,
+      "num_tokens": 3902260.0,
+      "step": 439
+    },
+    {
+      "epoch": 0.3343465045592705,
+      "grad_norm": 2.001574754714966,
+      "learning_rate": 4.949218499779462e-06,
+      "loss": 0.5413002967834473,
+      "mean_token_accuracy": 0.8162837028503418,
+      "num_tokens": 3911706.0,
+      "step": 440
+    },
+    {
+      "epoch": 0.3351063829787234,
+      "grad_norm": 1.7998944520950317,
+      "learning_rate": 4.948797649642484e-06,
+      "loss": 0.5131614208221436,
+      "mean_token_accuracy": 0.8367440700531006,
+      "num_tokens": 3923490.0,
+      "step": 441
+    },
+    {
+      "epoch": 0.33586626139817627,
+      "grad_norm": 3.4566173553466797,
+      "learning_rate": 4.94837508084505e-06,
+      "loss": 0.7258909940719604,
+      "mean_token_accuracy": 0.771377444267273,
+      "num_tokens": 3928099.0,
+      "step": 442
+    },
+    {
+      "epoch": 0.3366261398176292,
+      "grad_norm": 2.0040442943573,
+      "learning_rate": 4.9479507936837364e-06,
+      "loss": 0.482135534286499,
+      "mean_token_accuracy": 0.8339327573776245,
+      "num_tokens": 3937328.0,
+      "step": 443
+    },
+    {
+      "epoch": 0.3373860182370821,
+      "grad_norm": 2.949502944946289,
+      "learning_rate": 4.947524788456325e-06,
+      "loss": 0.6474795341491699,
+      "mean_token_accuracy": 0.7951677441596985,
+      "num_tokens": 3942529.0,
+      "step": 444
+    },
+    {
+      "epoch": 0.33814589665653494,
+      "grad_norm": 1.5528364181518555,
+      "learning_rate": 4.947097065461801e-06,
+      "loss": 0.48791584372520447,
+      "mean_token_accuracy": 0.8425545692443848,
+      "num_tokens": 3955200.0,
+      "step": 445
+    },
+    {
+      "epoch": 0.33890577507598785,
+      "grad_norm": 1.8813284635543823,
+      "learning_rate": 4.946667625000358e-06,
+      "loss": 0.45922309160232544,
+      "mean_token_accuracy": 0.8206527233123779,
+      "num_tokens": 3962975.0,
+      "step": 446
+    },
+    {
+      "epoch": 0.33966565349544076,
+      "grad_norm": 1.7157847881317139,
+      "learning_rate": 4.946236467373392e-06,
+      "loss": 0.5454182028770447,
+      "mean_token_accuracy": 0.8049604892730713,
+      "num_tokens": 3973956.0,
+      "step": 447
+    },
+    {
+      "epoch": 0.3404255319148936,
+      "grad_norm": 2.008857250213623,
+      "learning_rate": 4.945803592883509e-06,
+      "loss": 0.5151860117912292,
+      "mean_token_accuracy": 0.8262045383453369,
+      "num_tokens": 3982853.0,
+      "step": 448
+    },
+    {
+      "epoch": 0.3411854103343465,
+      "grad_norm": 1.6632496118545532,
+      "learning_rate": 4.9453690018345144e-06,
+      "loss": 0.42710691690444946,
+      "mean_token_accuracy": 0.8521314859390259,
+      "num_tokens": 3993838.0,
+      "step": 449
+    },
+    {
+      "epoch": 0.34194528875379937,
+      "grad_norm": 1.365234375,
+      "learning_rate": 4.944932694531423e-06,
+      "loss": 0.5172526836395264,
+      "mean_token_accuracy": 0.8277045488357544,
+      "num_tokens": 4014179.0,
+      "step": 450
+    },
+    {
+      "epoch": 0.3427051671732523,
+      "grad_norm": 1.7610243558883667,
+      "learning_rate": 4.94449467128045e-06,
+      "loss": 0.42104798555374146,
+      "mean_token_accuracy": 0.8552065491676331,
+      "num_tokens": 4023663.0,
+      "step": 451
+    },
+    {
+      "epoch": 0.3434650455927052,
+      "grad_norm": 2.3732354640960693,
+      "learning_rate": 4.944054932389018e-06,
+      "loss": 0.5471175909042358,
+      "mean_token_accuracy": 0.8487317562103271,
+      "num_tokens": 4030100.0,
+      "step": 452
+    },
+    {
+      "epoch": 0.34422492401215804,
+      "grad_norm": 1.5973623991012573,
+      "learning_rate": 4.943613478165753e-06,
+      "loss": 0.419813871383667,
+      "mean_token_accuracy": 0.8484025001525879,
+      "num_tokens": 4041124.0,
+      "step": 453
+    },
+    {
+      "epoch": 0.34498480243161095,
+      "grad_norm": 2.966381549835205,
+      "learning_rate": 4.943170308920484e-06,
+      "loss": 0.5370652675628662,
+      "mean_token_accuracy": 0.8439491987228394,
+      "num_tokens": 4045675.0,
+      "step": 454
+    },
+    {
+      "epoch": 0.34574468085106386,
+      "grad_norm": 2.5097248554229736,
+      "learning_rate": 4.9427254249642445e-06,
+      "loss": 0.5776349306106567,
+      "mean_token_accuracy": 0.8060523867607117,
+      "num_tokens": 4053250.0,
+      "step": 455
+    },
+    {
+      "epoch": 0.3465045592705167,
+      "grad_norm": 1.6779125928878784,
+      "learning_rate": 4.942278826609272e-06,
+      "loss": 0.5245476961135864,
+      "mean_token_accuracy": 0.8168526887893677,
+      "num_tokens": 4064106.0,
+      "step": 456
+    },
+    {
+      "epoch": 0.3472644376899696,
+      "grad_norm": 1.5945546627044678,
+      "learning_rate": 4.9418305141690045e-06,
+      "loss": 0.4972047209739685,
+      "mean_token_accuracy": 0.8257735967636108,
+      "num_tokens": 4077687.0,
+      "step": 457
+    },
+    {
+      "epoch": 0.34802431610942247,
+      "grad_norm": 2.864778757095337,
+      "learning_rate": 4.9413804879580865e-06,
+      "loss": 0.5372499823570251,
+      "mean_token_accuracy": 0.8423776626586914,
+      "num_tokens": 4082632.0,
+      "step": 458
+    },
+    {
+      "epoch": 0.3487841945288754,
+      "grad_norm": 1.4797078371047974,
+      "learning_rate": 4.940928748292363e-06,
+      "loss": 0.5903409719467163,
+      "mean_token_accuracy": 0.8061295747756958,
+      "num_tokens": 4104218.0,
+      "step": 459
+    },
+    {
+      "epoch": 0.3495440729483283,
+      "grad_norm": 2.4376983642578125,
+      "learning_rate": 4.940475295488882e-06,
+      "loss": 0.4534894824028015,
+      "mean_token_accuracy": 0.8395825028419495,
+      "num_tokens": 4110530.0,
+      "step": 460
+    },
+    {
+      "epoch": 0.35030395136778114,
+      "grad_norm": 1.2955626249313354,
+      "learning_rate": 4.940020129865895e-06,
+      "loss": 0.47155818343162537,
+      "mean_token_accuracy": 0.8253582715988159,
+      "num_tokens": 4128398.0,
+      "step": 461
+    },
+    {
+      "epoch": 0.35106382978723405,
+      "grad_norm": 2.066575527191162,
+      "learning_rate": 4.9395632517428546e-06,
+      "loss": 0.5555641651153564,
+      "mean_token_accuracy": 0.814624547958374,
+      "num_tokens": 4137623.0,
+      "step": 462
+    },
+    {
+      "epoch": 0.3518237082066869,
+      "grad_norm": 1.6407525539398193,
+      "learning_rate": 4.939104661440415e-06,
+      "loss": 0.4361790418624878,
+      "mean_token_accuracy": 0.8544459342956543,
+      "num_tokens": 4152803.0,
+      "step": 463
+    },
+    {
+      "epoch": 0.3525835866261398,
+      "grad_norm": 2.1685116291046143,
+      "learning_rate": 4.938644359280433e-06,
+      "loss": 0.5347012877464294,
+      "mean_token_accuracy": 0.853853702545166,
+      "num_tokens": 4160778.0,
+      "step": 464
+    },
+    {
+      "epoch": 0.3533434650455927,
+      "grad_norm": 1.8824869394302368,
+      "learning_rate": 4.938182345585967e-06,
+      "loss": 0.5512481927871704,
+      "mean_token_accuracy": 0.7985891699790955,
+      "num_tokens": 4170380.0,
+      "step": 465
+    },
+    {
+      "epoch": 0.3541033434650456,
+      "grad_norm": 2.2229504585266113,
+      "learning_rate": 4.937718620681273e-06,
+      "loss": 0.516828179359436,
+      "mean_token_accuracy": 0.8265621066093445,
+      "num_tokens": 4178179.0,
+      "step": 466
+    },
+    {
+      "epoch": 0.3548632218844985,
+      "grad_norm": 1.955990195274353,
+      "learning_rate": 4.9372531848918145e-06,
+      "loss": 0.5586158037185669,
+      "mean_token_accuracy": 0.8367916345596313,
+      "num_tokens": 4188626.0,
+      "step": 467
+    },
+    {
+      "epoch": 0.3556231003039514,
+      "grad_norm": 1.9687023162841797,
+      "learning_rate": 4.936786038544251e-06,
+      "loss": 0.5517531633377075,
+      "mean_token_accuracy": 0.8134098052978516,
+      "num_tokens": 4198144.0,
+      "step": 468
+    },
+    {
+      "epoch": 0.35638297872340424,
+      "grad_norm": 1.405516505241394,
+      "learning_rate": 4.9363171819664434e-06,
+      "loss": 0.5305492877960205,
+      "mean_token_accuracy": 0.8014427423477173,
+      "num_tokens": 4222818.0,
+      "step": 469
+    },
+    {
+      "epoch": 0.35714285714285715,
+      "grad_norm": 2.6355695724487305,
+      "learning_rate": 4.9358466154874535e-06,
+      "loss": 0.5303391218185425,
+      "mean_token_accuracy": 0.8028861284255981,
+      "num_tokens": 4228318.0,
+      "step": 470
+    },
+    {
+      "epoch": 0.35790273556231,
+      "grad_norm": 1.5133824348449707,
+      "learning_rate": 4.935374339437543e-06,
+      "loss": 0.5329189300537109,
+      "mean_token_accuracy": 0.8479441404342651,
+      "num_tokens": 4244527.0,
+      "step": 471
+    },
+    {
+      "epoch": 0.3586626139817629,
+      "grad_norm": 3.4356725215911865,
+      "learning_rate": 4.934900354148173e-06,
+      "loss": 0.5431582927703857,
+      "mean_token_accuracy": 0.8328983783721924,
+      "num_tokens": 4248034.0,
+      "step": 472
+    },
+    {
+      "epoch": 0.3594224924012158,
+      "grad_norm": 2.5789499282836914,
+      "learning_rate": 4.934424659952006e-06,
+      "loss": 0.4141455292701721,
+      "mean_token_accuracy": 0.8658635020256042,
+      "num_tokens": 4252953.0,
+      "step": 473
+    },
+    {
+      "epoch": 0.3601823708206687,
+      "grad_norm": 1.145262598991394,
+      "learning_rate": 4.933947257182901e-06,
+      "loss": 0.40294092893600464,
+      "mean_token_accuracy": 0.8565847277641296,
+      "num_tokens": 4277813.0,
+      "step": 474
+    },
+    {
+      "epoch": 0.3609422492401216,
+      "grad_norm": 1.7242133617401123,
+      "learning_rate": 4.933468146175918e-06,
+      "loss": 0.6036738753318787,
+      "mean_token_accuracy": 0.8072597980499268,
+      "num_tokens": 4291088.0,
+      "step": 475
+    },
+    {
+      "epoch": 0.3617021276595745,
+      "grad_norm": 2.3490941524505615,
+      "learning_rate": 4.932987327267317e-06,
+      "loss": 0.49456146359443665,
+      "mean_token_accuracy": 0.8372673988342285,
+      "num_tokens": 4297376.0,
+      "step": 476
+    },
+    {
+      "epoch": 0.36246200607902734,
+      "grad_norm": 1.3605526685714722,
+      "learning_rate": 4.932504800794553e-06,
+      "loss": 0.43595948815345764,
+      "mean_token_accuracy": 0.8415953516960144,
+      "num_tokens": 4312054.0,
+      "step": 477
+    },
+    {
+      "epoch": 0.36322188449848025,
+      "grad_norm": 1.4525885581970215,
+      "learning_rate": 4.9320205670962815e-06,
+      "loss": 0.5390371680259705,
+      "mean_token_accuracy": 0.8101649284362793,
+      "num_tokens": 4328701.0,
+      "step": 478
+    },
+    {
+      "epoch": 0.3639817629179331,
+      "grad_norm": 1.9862419366836548,
+      "learning_rate": 4.931534626512359e-06,
+      "loss": 0.45436930656433105,
+      "mean_token_accuracy": 0.8352861404418945,
+      "num_tokens": 4338372.0,
+      "step": 479
+    },
+    {
+      "epoch": 0.364741641337386,
+      "grad_norm": 1.7804961204528809,
+      "learning_rate": 4.931046979383836e-06,
+      "loss": 0.4677754044532776,
+      "mean_token_accuracy": 0.840467095375061,
+      "num_tokens": 4347897.0,
+      "step": 480
+    },
+    {
+      "epoch": 0.3655015197568389,
+      "grad_norm": 2.066632032394409,
+      "learning_rate": 4.930557626052961e-06,
+      "loss": 0.42418140172958374,
+      "mean_token_accuracy": 0.8528275489807129,
+      "num_tokens": 4354061.0,
+      "step": 481
+    },
+    {
+      "epoch": 0.3662613981762918,
+      "grad_norm": 1.6155282258987427,
+      "learning_rate": 4.930066566863182e-06,
+      "loss": 0.5424284934997559,
+      "mean_token_accuracy": 0.825040876865387,
+      "num_tokens": 4370400.0,
+      "step": 482
+    },
+    {
+      "epoch": 0.3670212765957447,
+      "grad_norm": 2.1452953815460205,
+      "learning_rate": 4.929573802159143e-06,
+      "loss": 0.5105804204940796,
+      "mean_token_accuracy": 0.8284053802490234,
+      "num_tokens": 4377579.0,
+      "step": 483
+    },
+    {
+      "epoch": 0.3677811550151976,
+      "grad_norm": 1.8940945863723755,
+      "learning_rate": 4.929079332286685e-06,
+      "loss": 0.43478304147720337,
+      "mean_token_accuracy": 0.8505665063858032,
+      "num_tokens": 4385686.0,
+      "step": 484
+    },
+    {
+      "epoch": 0.36854103343465044,
+      "grad_norm": 1.6785860061645508,
+      "learning_rate": 4.928583157592846e-06,
+      "loss": 0.40227848291397095,
+      "mean_token_accuracy": 0.8623573780059814,
+      "num_tokens": 4396128.0,
+      "step": 485
+    },
+    {
+      "epoch": 0.36930091185410335,
+      "grad_norm": 1.6416733264923096,
+      "learning_rate": 4.928085278425862e-06,
+      "loss": 0.526267409324646,
+      "mean_token_accuracy": 0.8284667730331421,
+      "num_tokens": 4407963.0,
+      "step": 486
+    },
+    {
+      "epoch": 0.3700607902735562,
+      "grad_norm": 1.8882389068603516,
+      "learning_rate": 4.927585695135162e-06,
+      "loss": 0.5555213093757629,
+      "mean_token_accuracy": 0.8115293979644775,
+      "num_tokens": 4418057.0,
+      "step": 487
+    },
+    {
+      "epoch": 0.3708206686930091,
+      "grad_norm": 2.300248384475708,
+      "learning_rate": 4.9270844080713735e-06,
+      "loss": 0.5812339186668396,
+      "mean_token_accuracy": 0.800270676612854,
+      "num_tokens": 4425358.0,
+      "step": 488
+    },
+    {
+      "epoch": 0.371580547112462,
+      "grad_norm": 1.6802922487258911,
+      "learning_rate": 4.926581417586319e-06,
+      "loss": 0.5134941935539246,
+      "mean_token_accuracy": 0.8247408866882324,
+      "num_tokens": 4437702.0,
+      "step": 489
+    },
+    {
+      "epoch": 0.3723404255319149,
+      "grad_norm": 1.7620291709899902,
+      "learning_rate": 4.926076724033016e-06,
+      "loss": 0.5233973264694214,
+      "mean_token_accuracy": 0.8102161884307861,
+      "num_tokens": 4448584.0,
+      "step": 490
+    },
+    {
+      "epoch": 0.3731003039513678,
+      "grad_norm": 1.6911998987197876,
+      "learning_rate": 4.925570327765678e-06,
+      "loss": 0.5337274074554443,
+      "mean_token_accuracy": 0.845306396484375,
+      "num_tokens": 4462651.0,
+      "step": 491
+    },
+    {
+      "epoch": 0.3738601823708207,
+      "grad_norm": 1.7991242408752441,
+      "learning_rate": 4.9250622291397144e-06,
+      "loss": 0.31018948554992676,
+      "mean_token_accuracy": 0.8857606053352356,
+      "num_tokens": 4469971.0,
+      "step": 492
+    },
+    {
+      "epoch": 0.37462006079027355,
+      "grad_norm": 4.9776835441589355,
+      "learning_rate": 4.924552428511727e-06,
+      "loss": 0.44114983081817627,
+      "mean_token_accuracy": 0.8429906368255615,
+      "num_tokens": 4478275.0,
+      "step": 493
+    },
+    {
+      "epoch": 0.37537993920972645,
+      "grad_norm": 1.8007272481918335,
+      "learning_rate": 4.924040926239515e-06,
+      "loss": 0.574328601360321,
+      "mean_token_accuracy": 0.7669196128845215,
+      "num_tokens": 4491551.0,
+      "step": 494
+    },
+    {
+      "epoch": 0.3761398176291793,
+      "grad_norm": 2.021300792694092,
+      "learning_rate": 4.92352772268207e-06,
+      "loss": 0.45636120438575745,
+      "mean_token_accuracy": 0.840438723564148,
+      "num_tokens": 4498658.0,
+      "step": 495
+    },
+    {
+      "epoch": 0.3768996960486322,
+      "grad_norm": 2.369748592376709,
+      "learning_rate": 4.923012818199576e-06,
+      "loss": 0.5206376910209656,
+      "mean_token_accuracy": 0.8521823287010193,
+      "num_tokens": 4504648.0,
+      "step": 496
+    },
+    {
+      "epoch": 0.3776595744680851,
+      "grad_norm": 2.733485221862793,
+      "learning_rate": 4.922496213153416e-06,
+      "loss": 0.5067723989486694,
+      "mean_token_accuracy": 0.8168281316757202,
+      "num_tokens": 4509990.0,
+      "step": 497
+    },
+    {
+      "epoch": 0.378419452887538,
+      "grad_norm": 2.3751676082611084,
+      "learning_rate": 4.921977907906161e-06,
+      "loss": 0.49757206439971924,
+      "mean_token_accuracy": 0.8325017690658569,
+      "num_tokens": 4518373.0,
+      "step": 498
+    },
+    {
+      "epoch": 0.3791793313069909,
+      "grad_norm": 2.1672775745391846,
+      "learning_rate": 4.921457902821578e-06,
+      "loss": 0.4237566590309143,
+      "mean_token_accuracy": 0.8404698371887207,
+      "num_tokens": 4524338.0,
+      "step": 499
+    },
+    {
+      "epoch": 0.3799392097264438,
+      "grad_norm": 1.8374360799789429,
+      "learning_rate": 4.9209361982646275e-06,
+      "loss": 0.4995468854904175,
+      "mean_token_accuracy": 0.8299649953842163,
+      "num_tokens": 4533396.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.38069908814589665,
+      "grad_norm": 2.083967924118042,
+      "learning_rate": 4.920412794601461e-06,
+      "loss": 0.489935040473938,
+      "mean_token_accuracy": 0.8315291404724121,
+      "num_tokens": 4540941.0,
+      "step": 501
+    },
+    {
+      "epoch": 0.38145896656534956,
+      "grad_norm": 2.2075610160827637,
+      "learning_rate": 4.919887692199423e-06,
+      "loss": 0.5233147740364075,
+      "mean_token_accuracy": 0.804171085357666,
+      "num_tokens": 4548215.0,
+      "step": 502
+    },
+    {
+      "epoch": 0.3822188449848024,
+      "grad_norm": 2.076775312423706,
+      "learning_rate": 4.9193608914270515e-06,
+      "loss": 0.5785550475120544,
+      "mean_token_accuracy": 0.7993186116218567,
+      "num_tokens": 4558204.0,
+      "step": 503
+    },
+    {
+      "epoch": 0.3829787234042553,
+      "grad_norm": 2.238546133041382,
+      "learning_rate": 4.918832392654075e-06,
+      "loss": 0.5287384390830994,
+      "mean_token_accuracy": 0.8214945793151855,
+      "num_tokens": 4565407.0,
+      "step": 504
+    },
+    {
+      "epoch": 0.3837386018237082,
+      "grad_norm": 1.6783074140548706,
+      "learning_rate": 4.9183021962514145e-06,
+      "loss": 0.6063359379768372,
+      "mean_token_accuracy": 0.7914625406265259,
+      "num_tokens": 4580991.0,
+      "step": 505
+    },
+    {
+      "epoch": 0.3844984802431611,
+      "grad_norm": 1.6287449598312378,
+      "learning_rate": 4.917770302591183e-06,
+      "loss": 0.3598247766494751,
+      "mean_token_accuracy": 0.8706809878349304,
+      "num_tokens": 4590579.0,
+      "step": 506
+    },
+    {
+      "epoch": 0.385258358662614,
+      "grad_norm": 1.5432041883468628,
+      "learning_rate": 4.917236712046682e-06,
+      "loss": 0.5267890095710754,
+      "mean_token_accuracy": 0.8032117486000061,
+      "num_tokens": 4608380.0,
+      "step": 507
+    },
+    {
+      "epoch": 0.3860182370820669,
+      "grad_norm": 1.7664037942886353,
+      "learning_rate": 4.9167014249924075e-06,
+      "loss": 0.3552354574203491,
+      "mean_token_accuracy": 0.8569793701171875,
+      "num_tokens": 4616426.0,
+      "step": 508
+    },
+    {
+      "epoch": 0.38677811550151975,
+      "grad_norm": 2.1147472858428955,
+      "learning_rate": 4.916164441804044e-06,
+      "loss": 0.5212404727935791,
+      "mean_token_accuracy": 0.8196578025817871,
+      "num_tokens": 4623908.0,
+      "step": 509
+    },
+    {
+      "epoch": 0.38753799392097266,
+      "grad_norm": 2.1092333793640137,
+      "learning_rate": 4.915625762858467e-06,
+      "loss": 0.5197038650512695,
+      "mean_token_accuracy": 0.8245604634284973,
+      "num_tokens": 4630956.0,
+      "step": 510
+    },
+    {
+      "epoch": 0.3882978723404255,
+      "grad_norm": 1.23331880569458,
+      "learning_rate": 4.915085388533743e-06,
+      "loss": 0.4759839177131653,
+      "mean_token_accuracy": 0.8192248344421387,
+      "num_tokens": 4651269.0,
+      "step": 511
+    },
+    {
+      "epoch": 0.3890577507598784,
+      "grad_norm": 2.424199104309082,
+      "learning_rate": 4.914543319209126e-06,
+      "loss": 0.5576270818710327,
+      "mean_token_accuracy": 0.8203302621841431,
+      "num_tokens": 4657296.0,
+      "step": 512
+    },
+    {
+      "epoch": 0.3898176291793313,
+      "grad_norm": 2.725156307220459,
+      "learning_rate": 4.913999555265062e-06,
+      "loss": 0.4337949752807617,
+      "mean_token_accuracy": 0.8382406234741211,
+      "num_tokens": 4661850.0,
+      "step": 513
+    },
+    {
+      "epoch": 0.3905775075987842,
+      "grad_norm": 2.3120534420013428,
+      "learning_rate": 4.913454097083185e-06,
+      "loss": 0.4941597580909729,
+      "mean_token_accuracy": 0.8302834033966064,
+      "num_tokens": 4667769.0,
+      "step": 514
+    },
+    {
+      "epoch": 0.3913373860182371,
+      "grad_norm": 2.3111207485198975,
+      "learning_rate": 4.912906945046319e-06,
+      "loss": 0.5253715515136719,
+      "mean_token_accuracy": 0.84515380859375,
+      "num_tokens": 4674537.0,
+      "step": 515
+    },
+    {
+      "epoch": 0.39209726443769,
+      "grad_norm": 1.4117841720581055,
+      "learning_rate": 4.912358099538476e-06,
+      "loss": 0.4521017074584961,
+      "mean_token_accuracy": 0.8208256959915161,
+      "num_tokens": 4690605.0,
+      "step": 516
+    },
+    {
+      "epoch": 0.39285714285714285,
+      "grad_norm": 2.3742799758911133,
+      "learning_rate": 4.911807560944858e-06,
+      "loss": 0.41572901606559753,
+      "mean_token_accuracy": 0.8550551533699036,
+      "num_tokens": 4706437.0,
+      "step": 517
+    },
+    {
+      "epoch": 0.39361702127659576,
+      "grad_norm": 2.4052202701568604,
+      "learning_rate": 4.911255329651852e-06,
+      "loss": 0.6003736257553101,
+      "mean_token_accuracy": 0.8247885704040527,
+      "num_tokens": 4712746.0,
+      "step": 518
+    },
+    {
+      "epoch": 0.3943768996960486,
+      "grad_norm": 1.9335490465164185,
+      "learning_rate": 4.910701406047037e-06,
+      "loss": 0.5457713603973389,
+      "mean_token_accuracy": 0.787429690361023,
+      "num_tokens": 4731937.0,
+      "step": 519
+    },
+    {
+      "epoch": 0.3951367781155015,
+      "grad_norm": 2.257706880569458,
+      "learning_rate": 4.910145790519177e-06,
+      "loss": 0.5300652980804443,
+      "mean_token_accuracy": 0.8192912936210632,
+      "num_tokens": 4739422.0,
+      "step": 520
+    },
+    {
+      "epoch": 0.3958966565349544,
+      "grad_norm": 1.2099462747573853,
+      "learning_rate": 4.9095884834582256e-06,
+      "loss": 0.45872747898101807,
+      "mean_token_accuracy": 0.8362667560577393,
+      "num_tokens": 4757113.0,
+      "step": 521
+    },
+    {
+      "epoch": 0.3966565349544073,
+      "grad_norm": 2.7991135120391846,
+      "learning_rate": 4.909029485255321e-06,
+      "loss": 0.49039560556411743,
+      "mean_token_accuracy": 0.8260016441345215,
+      "num_tokens": 4761709.0,
+      "step": 522
+    },
+    {
+      "epoch": 0.3974164133738602,
+      "grad_norm": 2.2360129356384277,
+      "learning_rate": 4.90846879630279e-06,
+      "loss": 0.49556830525398254,
+      "mean_token_accuracy": 0.827864408493042,
+      "num_tokens": 4769048.0,
+      "step": 523
+    },
+    {
+      "epoch": 0.3981762917933131,
+      "grad_norm": 2.5953688621520996,
+      "learning_rate": 4.907906416994146e-06,
+      "loss": 0.387208491563797,
+      "mean_token_accuracy": 0.8467001914978027,
+      "num_tokens": 4774637.0,
+      "step": 524
+    },
+    {
+      "epoch": 0.39893617021276595,
+      "grad_norm": 2.1046814918518066,
+      "learning_rate": 4.907342347724088e-06,
+      "loss": 0.5477259755134583,
+      "mean_token_accuracy": 0.8060322999954224,
+      "num_tokens": 4782774.0,
+      "step": 525
+    },
+    {
+      "epoch": 0.39969604863221886,
+      "grad_norm": 2.5622646808624268,
+      "learning_rate": 4.906776588888502e-06,
+      "loss": 0.5684159398078918,
+      "mean_token_accuracy": 0.8095303177833557,
+      "num_tokens": 4788766.0,
+      "step": 526
+    },
+    {
+      "epoch": 0.4004559270516717,
+      "grad_norm": 1.9027913808822632,
+      "learning_rate": 4.906209140884459e-06,
+      "loss": 0.535524845123291,
+      "mean_token_accuracy": 0.815237820148468,
+      "num_tokens": 4798492.0,
+      "step": 527
+    },
+    {
+      "epoch": 0.4012158054711246,
+      "grad_norm": 2.1447622776031494,
+      "learning_rate": 4.905640004110216e-06,
+      "loss": 0.5628632307052612,
+      "mean_token_accuracy": 0.8085395097732544,
+      "num_tokens": 4805737.0,
+      "step": 528
+    },
+    {
+      "epoch": 0.40197568389057753,
+      "grad_norm": 1.6754741668701172,
+      "learning_rate": 4.905069178965215e-06,
+      "loss": 0.5046736598014832,
+      "mean_token_accuracy": 0.8247535228729248,
+      "num_tokens": 4816912.0,
+      "step": 529
+    },
+    {
+      "epoch": 0.4027355623100304,
+      "grad_norm": 2.271230459213257,
+      "learning_rate": 4.904496665850083e-06,
+      "loss": 0.6086187958717346,
+      "mean_token_accuracy": 0.7935276627540588,
+      "num_tokens": 4824577.0,
+      "step": 530
+    },
+    {
+      "epoch": 0.4034954407294833,
+      "grad_norm": 2.107595205307007,
+      "learning_rate": 4.903922465166633e-06,
+      "loss": 0.5431341528892517,
+      "mean_token_accuracy": 0.8129537105560303,
+      "num_tokens": 4831772.0,
+      "step": 531
+    },
+    {
+      "epoch": 0.40425531914893614,
+      "grad_norm": 1.3860732316970825,
+      "learning_rate": 4.903346577317859e-06,
+      "loss": 0.45816320180892944,
+      "mean_token_accuracy": 0.8328287601470947,
+      "num_tokens": 4850302.0,
+      "step": 532
+    },
+    {
+      "epoch": 0.40501519756838905,
+      "grad_norm": 1.9186837673187256,
+      "learning_rate": 4.902769002707942e-06,
+      "loss": 0.3294633626937866,
+      "mean_token_accuracy": 0.8853933811187744,
+      "num_tokens": 4856624.0,
+      "step": 533
+    },
+    {
+      "epoch": 0.40577507598784196,
+      "grad_norm": 1.516194462776184,
+      "learning_rate": 4.902189741742247e-06,
+      "loss": 0.45482105016708374,
+      "mean_token_accuracy": 0.8370342254638672,
+      "num_tokens": 4870395.0,
+      "step": 534
+    },
+    {
+      "epoch": 0.4065349544072948,
+      "grad_norm": 2.3235628604888916,
+      "learning_rate": 4.901608794827321e-06,
+      "loss": 0.40688639879226685,
+      "mean_token_accuracy": 0.8643521666526794,
+      "num_tokens": 4875645.0,
+      "step": 535
+    },
+    {
+      "epoch": 0.4072948328267477,
+      "grad_norm": 2.29286527633667,
+      "learning_rate": 4.9010261623708945e-06,
+      "loss": 0.45482826232910156,
+      "mean_token_accuracy": 0.8429383039474487,
+      "num_tokens": 4881772.0,
+      "step": 536
+    },
+    {
+      "epoch": 0.40805471124620063,
+      "grad_norm": 1.5907070636749268,
+      "learning_rate": 4.900441844781882e-06,
+      "loss": 0.5266948342323303,
+      "mean_token_accuracy": 0.8348641395568848,
+      "num_tokens": 4894289.0,
+      "step": 537
+    },
+    {
+      "epoch": 0.4088145896656535,
+      "grad_norm": 2.1816294193267822,
+      "learning_rate": 4.89985584247038e-06,
+      "loss": 0.4797617793083191,
+      "mean_token_accuracy": 0.8549500703811646,
+      "num_tokens": 4901106.0,
+      "step": 538
+    },
+    {
+      "epoch": 0.4095744680851064,
+      "grad_norm": 1.7347146272659302,
+      "learning_rate": 4.899268155847667e-06,
+      "loss": 0.4754739999771118,
+      "mean_token_accuracy": 0.8278418183326721,
+      "num_tokens": 4912131.0,
+      "step": 539
+    },
+    {
+      "epoch": 0.41033434650455924,
+      "grad_norm": 2.0694527626037598,
+      "learning_rate": 4.898678785326205e-06,
+      "loss": 0.5071008801460266,
+      "mean_token_accuracy": 0.8157946467399597,
+      "num_tokens": 4921141.0,
+      "step": 540
+    },
+    {
+      "epoch": 0.41109422492401215,
+      "grad_norm": 2.570047616958618,
+      "learning_rate": 4.898087731319637e-06,
+      "loss": 0.43639278411865234,
+      "mean_token_accuracy": 0.8682913780212402,
+      "num_tokens": 4926182.0,
+      "step": 541
+    },
+    {
+      "epoch": 0.41185410334346506,
+      "grad_norm": 4.064006805419922,
+      "learning_rate": 4.8974949942427854e-06,
+      "loss": 0.539260745048523,
+      "mean_token_accuracy": 0.8225528001785278,
+      "num_tokens": 4929449.0,
+      "step": 542
+    },
+    {
+      "epoch": 0.4126139817629179,
+      "grad_norm": 1.7644332647323608,
+      "learning_rate": 4.896900574511657e-06,
+      "loss": 0.472618043422699,
+      "mean_token_accuracy": 0.8332902193069458,
+      "num_tokens": 4939443.0,
+      "step": 543
+    },
+    {
+      "epoch": 0.4133738601823708,
+      "grad_norm": 2.879918336868286,
+      "learning_rate": 4.89630447254344e-06,
+      "loss": 0.6360667943954468,
+      "mean_token_accuracy": 0.8215296268463135,
+      "num_tokens": 4950838.0,
+      "step": 544
+    },
+    {
+      "epoch": 0.41413373860182373,
+      "grad_norm": 1.4575570821762085,
+      "learning_rate": 4.8957066887565005e-06,
+      "loss": 0.45617997646331787,
+      "mean_token_accuracy": 0.8373187184333801,
+      "num_tokens": 4965222.0,
+      "step": 545
+    },
+    {
+      "epoch": 0.4148936170212766,
+      "grad_norm": 2.4829535484313965,
+      "learning_rate": 4.895107223570386e-06,
+      "loss": 0.42285341024398804,
+      "mean_token_accuracy": 0.8686380386352539,
+      "num_tokens": 4970724.0,
+      "step": 546
+    },
+    {
+      "epoch": 0.4156534954407295,
+      "grad_norm": 2.639474630355835,
+      "learning_rate": 4.894506077405824e-06,
+      "loss": 0.5906289219856262,
+      "mean_token_accuracy": 0.8174435496330261,
+      "num_tokens": 4976766.0,
+      "step": 547
+    },
+    {
+      "epoch": 0.41641337386018235,
+      "grad_norm": 2.7960562705993652,
+      "learning_rate": 4.893903250684723e-06,
+      "loss": 0.4518949091434479,
+      "mean_token_accuracy": 0.8387585282325745,
+      "num_tokens": 4980991.0,
+      "step": 548
+    },
+    {
+      "epoch": 0.41717325227963525,
+      "grad_norm": 2.184176206588745,
+      "learning_rate": 4.893298743830168e-06,
+      "loss": 0.5223842859268188,
+      "mean_token_accuracy": 0.8170937299728394,
+      "num_tokens": 4987781.0,
+      "step": 549
+    },
+    {
+      "epoch": 0.41793313069908816,
+      "grad_norm": 2.2393438816070557,
+      "learning_rate": 4.892692557266429e-06,
+      "loss": 0.5238431692123413,
+      "mean_token_accuracy": 0.8217905759811401,
+      "num_tokens": 4994321.0,
+      "step": 550
+    },
+    {
+      "epoch": 0.418693009118541,
+      "grad_norm": 3.579047441482544,
+      "learning_rate": 4.8920846914189465e-06,
+      "loss": 0.5367584228515625,
+      "mean_token_accuracy": 0.8312011361122131,
+      "num_tokens": 4997951.0,
+      "step": 551
+    },
+    {
+      "epoch": 0.4194528875379939,
+      "grad_norm": 1.6330240964889526,
+      "learning_rate": 4.891475146714348e-06,
+      "loss": 0.6054705381393433,
+      "mean_token_accuracy": 0.7938206791877747,
+      "num_tokens": 5012726.0,
+      "step": 552
+    },
+    {
+      "epoch": 0.42021276595744683,
+      "grad_norm": 1.5775716304779053,
+      "learning_rate": 4.8908639235804324e-06,
+      "loss": 0.4774656891822815,
+      "mean_token_accuracy": 0.828762948513031,
+      "num_tokens": 5026751.0,
+      "step": 553
+    },
+    {
+      "epoch": 0.4209726443768997,
+      "grad_norm": 1.5719101428985596,
+      "learning_rate": 4.890251022446181e-06,
+      "loss": 0.549429178237915,
+      "mean_token_accuracy": 0.8110791444778442,
+      "num_tokens": 5041861.0,
+      "step": 554
+    },
+    {
+      "epoch": 0.4217325227963526,
+      "grad_norm": 1.8585275411605835,
+      "learning_rate": 4.889636443741752e-06,
+      "loss": 0.4448118805885315,
+      "mean_token_accuracy": 0.8462690711021423,
+      "num_tokens": 5052690.0,
+      "step": 555
+    },
+    {
+      "epoch": 0.42249240121580545,
+      "grad_norm": 2.189202070236206,
+      "learning_rate": 4.88902018789848e-06,
+      "loss": 0.4296762943267822,
+      "mean_token_accuracy": 0.8488791584968567,
+      "num_tokens": 5058964.0,
+      "step": 556
+    },
+    {
+      "epoch": 0.42325227963525835,
+      "grad_norm": 1.9328460693359375,
+      "learning_rate": 4.888402255348877e-06,
+      "loss": 0.5369474291801453,
+      "mean_token_accuracy": 0.8184729814529419,
+      "num_tokens": 5068465.0,
+      "step": 557
+    },
+    {
+      "epoch": 0.42401215805471126,
+      "grad_norm": 1.6233323812484741,
+      "learning_rate": 4.887782646526631e-06,
+      "loss": 0.5284391641616821,
+      "mean_token_accuracy": 0.8276044726371765,
+      "num_tokens": 5081052.0,
+      "step": 558
+    },
+    {
+      "epoch": 0.4247720364741641,
+      "grad_norm": 2.222813844680786,
+      "learning_rate": 4.887161361866608e-06,
+      "loss": 0.5679137706756592,
+      "mean_token_accuracy": 0.8012375831604004,
+      "num_tokens": 5090001.0,
+      "step": 559
+    },
+    {
+      "epoch": 0.425531914893617,
+      "grad_norm": 2.1062207221984863,
+      "learning_rate": 4.8865384018048494e-06,
+      "loss": 0.5554201602935791,
+      "mean_token_accuracy": 0.8128066062927246,
+      "num_tokens": 5097644.0,
+      "step": 560
+    },
+    {
+      "epoch": 0.42629179331306993,
+      "grad_norm": 1.5380984544754028,
+      "learning_rate": 4.8859137667785735e-06,
+      "loss": 0.4948265850543976,
+      "mean_token_accuracy": 0.8258291482925415,
+      "num_tokens": 5110069.0,
+      "step": 561
+    },
+    {
+      "epoch": 0.4270516717325228,
+      "grad_norm": 2.0290257930755615,
+      "learning_rate": 4.8852874572261715e-06,
+      "loss": 0.4969530403614044,
+      "mean_token_accuracy": 0.8297134637832642,
+      "num_tokens": 5117452.0,
+      "step": 562
+    },
+    {
+      "epoch": 0.4278115501519757,
+      "grad_norm": 1.5651452541351318,
+      "learning_rate": 4.884659473587213e-06,
+      "loss": 0.5353102087974548,
+      "mean_token_accuracy": 0.8161719441413879,
+      "num_tokens": 5133756.0,
+      "step": 563
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 2.2470998764038086,
+      "learning_rate": 4.884029816302441e-06,
+      "loss": 0.5104288458824158,
+      "mean_token_accuracy": 0.8081635236740112,
+      "num_tokens": 5140278.0,
+      "step": 564
+    },
+    {
+      "epoch": 0.42933130699088146,
+      "grad_norm": 1.726891279220581,
+      "learning_rate": 4.883398485813772e-06,
+      "loss": 0.4508771002292633,
+      "mean_token_accuracy": 0.8548800349235535,
+      "num_tokens": 5150115.0,
+      "step": 565
+    },
+    {
+      "epoch": 0.43009118541033436,
+      "grad_norm": 1.4779289960861206,
+      "learning_rate": 4.8827654825642984e-06,
+      "loss": 0.46861088275909424,
+      "mean_token_accuracy": 0.8209476470947266,
+      "num_tokens": 5163225.0,
+      "step": 566
+    },
+    {
+      "epoch": 0.4308510638297872,
+      "grad_norm": 1.2361034154891968,
+      "learning_rate": 4.882130806998287e-06,
+      "loss": 0.4591076672077179,
+      "mean_token_accuracy": 0.803041934967041,
+      "num_tokens": 5180342.0,
+      "step": 567
+    },
+    {
+      "epoch": 0.4316109422492401,
+      "grad_norm": 1.882467269897461,
+      "learning_rate": 4.881494459561177e-06,
+      "loss": 0.579258143901825,
+      "mean_token_accuracy": 0.8007112741470337,
+      "num_tokens": 5189595.0,
+      "step": 568
+    },
+    {
+      "epoch": 0.43237082066869303,
+      "grad_norm": 1.095462441444397,
+      "learning_rate": 4.880856440699582e-06,
+      "loss": 0.3806574046611786,
+      "mean_token_accuracy": 0.8650111556053162,
+      "num_tokens": 5211642.0,
+      "step": 569
+    },
+    {
+      "epoch": 0.4331306990881459,
+      "grad_norm": 1.6469846963882446,
+      "learning_rate": 4.880216750861288e-06,
+      "loss": 0.544589638710022,
+      "mean_token_accuracy": 0.8060122728347778,
+      "num_tokens": 5224137.0,
+      "step": 570
+    },
+    {
+      "epoch": 0.4338905775075988,
+      "grad_norm": 1.8561251163482666,
+      "learning_rate": 4.879575390495254e-06,
+      "loss": 0.4094924330711365,
+      "mean_token_accuracy": 0.8591406345367432,
+      "num_tokens": 5231588.0,
+      "step": 571
+    },
+    {
+      "epoch": 0.43465045592705165,
+      "grad_norm": 3.01326847076416,
+      "learning_rate": 4.878932360051611e-06,
+      "loss": 0.6139192581176758,
+      "mean_token_accuracy": 0.8108739852905273,
+      "num_tokens": 5236853.0,
+      "step": 572
+    },
+    {
+      "epoch": 0.43541033434650456,
+      "grad_norm": 2.1753034591674805,
+      "learning_rate": 4.878287659981663e-06,
+      "loss": 0.49082931876182556,
+      "mean_token_accuracy": 0.862828254699707,
+      "num_tokens": 5243264.0,
+      "step": 573
+    },
+    {
+      "epoch": 0.43617021276595747,
+      "grad_norm": 1.4437755346298218,
+      "learning_rate": 4.8776412907378845e-06,
+      "loss": 0.5608728528022766,
+      "mean_token_accuracy": 0.8271626234054565,
+      "num_tokens": 5261757.0,
+      "step": 574
+    },
+    {
+      "epoch": 0.4369300911854103,
+      "grad_norm": 1.786683440208435,
+      "learning_rate": 4.876993252773923e-06,
+      "loss": 0.4377627968788147,
+      "mean_token_accuracy": 0.844936192035675,
+      "num_tokens": 5271038.0,
+      "step": 575
+    },
+    {
+      "epoch": 0.4376899696048632,
+      "grad_norm": 1.3425915241241455,
+      "learning_rate": 4.876343546544596e-06,
+      "loss": 0.44762521982192993,
+      "mean_token_accuracy": 0.8397793769836426,
+      "num_tokens": 5285555.0,
+      "step": 576
+    },
+    {
+      "epoch": 0.43844984802431614,
+      "grad_norm": 2.1549675464630127,
+      "learning_rate": 4.8756921725058935e-06,
+      "loss": 0.5332942008972168,
+      "mean_token_accuracy": 0.820149302482605,
+      "num_tokens": 5294595.0,
+      "step": 577
+    },
+    {
+      "epoch": 0.439209726443769,
+      "grad_norm": 1.5254042148590088,
+      "learning_rate": 4.875039131114975e-06,
+      "loss": 0.3646543622016907,
+      "mean_token_accuracy": 0.8442583084106445,
+      "num_tokens": 5304955.0,
+      "step": 578
+    },
+    {
+      "epoch": 0.4399696048632219,
+      "grad_norm": 1.5751557350158691,
+      "learning_rate": 4.8743844228301676e-06,
+      "loss": 0.4854734539985657,
+      "mean_token_accuracy": 0.8317523002624512,
+      "num_tokens": 5317351.0,
+      "step": 579
+    },
+    {
+      "epoch": 0.44072948328267475,
+      "grad_norm": 1.6950466632843018,
+      "learning_rate": 4.873728048110973e-06,
+      "loss": 0.5907570719718933,
+      "mean_token_accuracy": 0.7946986556053162,
+      "num_tokens": 5332542.0,
+      "step": 580
+    },
+    {
+      "epoch": 0.44148936170212766,
+      "grad_norm": 2.1180708408355713,
+      "learning_rate": 4.873070007418059e-06,
+      "loss": 0.5220296382904053,
+      "mean_token_accuracy": 0.8037363290786743,
+      "num_tokens": 5341722.0,
+      "step": 581
+    },
+    {
+      "epoch": 0.44224924012158057,
+      "grad_norm": 1.3643816709518433,
+      "learning_rate": 4.872410301213265e-06,
+      "loss": 0.4865502417087555,
+      "mean_token_accuracy": 0.8377852439880371,
+      "num_tokens": 5359359.0,
+      "step": 582
+    },
+    {
+      "epoch": 0.4430091185410334,
+      "grad_norm": 1.483280897140503,
+      "learning_rate": 4.871748929959598e-06,
+      "loss": 0.36856764554977417,
+      "mean_token_accuracy": 0.8709549903869629,
+      "num_tokens": 5369749.0,
+      "step": 583
+    },
+    {
+      "epoch": 0.44376899696048633,
+      "grad_norm": 1.6891541481018066,
+      "learning_rate": 4.871085894121234e-06,
+      "loss": 0.5768930912017822,
+      "mean_token_accuracy": 0.8030461668968201,
+      "num_tokens": 5383912.0,
+      "step": 584
+    },
+    {
+      "epoch": 0.44452887537993924,
+      "grad_norm": 2.1318740844726562,
+      "learning_rate": 4.870421194163515e-06,
+      "loss": 0.4337100386619568,
+      "mean_token_accuracy": 0.8562518358230591,
+      "num_tokens": 5389412.0,
+      "step": 585
+    },
+    {
+      "epoch": 0.4452887537993921,
+      "grad_norm": 2.540255546569824,
+      "learning_rate": 4.869754830552956e-06,
+      "loss": 0.4708256125450134,
+      "mean_token_accuracy": 0.8446552753448486,
+      "num_tokens": 5394762.0,
+      "step": 586
+    },
+    {
+      "epoch": 0.446048632218845,
+      "grad_norm": 2.048015594482422,
+      "learning_rate": 4.869086803757235e-06,
+      "loss": 0.5265800952911377,
+      "mean_token_accuracy": 0.8181137442588806,
+      "num_tokens": 5402379.0,
+      "step": 587
+    },
+    {
+      "epoch": 0.44680851063829785,
+      "grad_norm": 2.9821012020111084,
+      "learning_rate": 4.868417114245199e-06,
+      "loss": 0.6299797296524048,
+      "mean_token_accuracy": 0.8237329125404358,
+      "num_tokens": 5408229.0,
+      "step": 588
+    },
+    {
+      "epoch": 0.44756838905775076,
+      "grad_norm": 1.7807202339172363,
+      "learning_rate": 4.867745762486862e-06,
+      "loss": 0.5176759958267212,
+      "mean_token_accuracy": 0.8184244632720947,
+      "num_tokens": 5418383.0,
+      "step": 589
+    },
+    {
+      "epoch": 0.44832826747720367,
+      "grad_norm": 1.5466399192810059,
+      "learning_rate": 4.8670727489534035e-06,
+      "loss": 0.5137228965759277,
+      "mean_token_accuracy": 0.8365053534507751,
+      "num_tokens": 5432127.0,
+      "step": 590
+    },
+    {
+      "epoch": 0.4490881458966565,
+      "grad_norm": 2.9521141052246094,
+      "learning_rate": 4.866398074117173e-06,
+      "loss": 0.4056887924671173,
+      "mean_token_accuracy": 0.8561501502990723,
+      "num_tokens": 5436062.0,
+      "step": 591
+    },
+    {
+      "epoch": 0.44984802431610943,
+      "grad_norm": 2.058743953704834,
+      "learning_rate": 4.86572173845168e-06,
+      "loss": 0.6124799251556396,
+      "mean_token_accuracy": 0.8007957339286804,
+      "num_tokens": 5444989.0,
+      "step": 592
+    },
+    {
+      "epoch": 0.4506079027355623,
+      "grad_norm": 2.1243767738342285,
+      "learning_rate": 4.865043742431605e-06,
+      "loss": 0.5659694671630859,
+      "mean_token_accuracy": 0.8084750175476074,
+      "num_tokens": 5453865.0,
+      "step": 593
+    },
+    {
+      "epoch": 0.4513677811550152,
+      "grad_norm": 1.6732314825057983,
+      "learning_rate": 4.864364086532792e-06,
+      "loss": 0.47879064083099365,
+      "mean_token_accuracy": 0.8346436023712158,
+      "num_tokens": 5466398.0,
+      "step": 594
+    },
+    {
+      "epoch": 0.4521276595744681,
+      "grad_norm": 1.3793858289718628,
+      "learning_rate": 4.863682771232249e-06,
+      "loss": 0.45989373326301575,
+      "mean_token_accuracy": 0.8254791498184204,
+      "num_tokens": 5482121.0,
+      "step": 595
+    },
+    {
+      "epoch": 0.45288753799392095,
+      "grad_norm": 1.9812315702438354,
+      "learning_rate": 4.862999797008149e-06,
+      "loss": 0.5778874754905701,
+      "mean_token_accuracy": 0.8041508197784424,
+      "num_tokens": 5493000.0,
+      "step": 596
+    },
+    {
+      "epoch": 0.45364741641337386,
+      "grad_norm": 3.3065083026885986,
+      "learning_rate": 4.862315164339829e-06,
+      "loss": 0.4623975157737732,
+      "mean_token_accuracy": 0.8426318168640137,
+      "num_tokens": 5496723.0,
+      "step": 597
+    },
+    {
+      "epoch": 0.45440729483282677,
+      "grad_norm": 3.167119026184082,
+      "learning_rate": 4.861628873707792e-06,
+      "loss": 0.6984533667564392,
+      "mean_token_accuracy": 0.772136926651001,
+      "num_tokens": 5501161.0,
+      "step": 598
+    },
+    {
+      "epoch": 0.4551671732522796,
+      "grad_norm": 2.2130985260009766,
+      "learning_rate": 4.860940925593703e-06,
+      "loss": 0.4823192059993744,
+      "mean_token_accuracy": 0.8462972640991211,
+      "num_tokens": 5509544.0,
+      "step": 599
+    },
+    {
+      "epoch": 0.45592705167173253,
+      "grad_norm": 3.029191732406616,
+      "learning_rate": 4.86025132048039e-06,
+      "loss": 0.523664116859436,
+      "mean_token_accuracy": 0.8229140043258667,
+      "num_tokens": 5514586.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.4566869300911854,
+      "grad_norm": 1.6983962059020996,
+      "learning_rate": 4.859560058851844e-06,
+      "loss": 0.4832698106765747,
+      "mean_token_accuracy": 0.8403248190879822,
+      "num_tokens": 5525773.0,
+      "step": 601
+    },
+    {
+      "epoch": 0.4574468085106383,
+      "grad_norm": 3.0504038333892822,
+      "learning_rate": 4.8588671411932195e-06,
+      "loss": 0.5158926248550415,
+      "mean_token_accuracy": 0.8098392486572266,
+      "num_tokens": 5529739.0,
+      "step": 602
+    },
+    {
+      "epoch": 0.4582066869300912,
+      "grad_norm": 2.584836483001709,
+      "learning_rate": 4.858172567990832e-06,
+      "loss": 0.5724587440490723,
+      "mean_token_accuracy": 0.8128519058227539,
+      "num_tokens": 5535763.0,
+      "step": 603
+    },
+    {
+      "epoch": 0.45896656534954405,
+      "grad_norm": 2.0514042377471924,
+      "learning_rate": 4.857476339732162e-06,
+      "loss": 0.4337679445743561,
+      "mean_token_accuracy": 0.8405929207801819,
+      "num_tokens": 5543075.0,
+      "step": 604
+    },
+    {
+      "epoch": 0.45972644376899696,
+      "grad_norm": 2.2949347496032715,
+      "learning_rate": 4.856778456905846e-06,
+      "loss": 0.46532145142555237,
+      "mean_token_accuracy": 0.8345137238502502,
+      "num_tokens": 5549035.0,
+      "step": 605
+    },
+    {
+      "epoch": 0.46048632218844987,
+      "grad_norm": 2.2067551612854004,
+      "learning_rate": 4.856078920001689e-06,
+      "loss": 0.5855136513710022,
+      "mean_token_accuracy": 0.8043795228004456,
+      "num_tokens": 5555545.0,
+      "step": 606
+    },
+    {
+      "epoch": 0.4612462006079027,
+      "grad_norm": 2.101945161819458,
+      "learning_rate": 4.855377729510648e-06,
+      "loss": 0.6071814298629761,
+      "mean_token_accuracy": 0.7973253130912781,
+      "num_tokens": 5563615.0,
+      "step": 607
+    },
+    {
+      "epoch": 0.46200607902735563,
+      "grad_norm": 2.5958821773529053,
+      "learning_rate": 4.8546748859248504e-06,
+      "loss": 0.6278061866760254,
+      "mean_token_accuracy": 0.7864972352981567,
+      "num_tokens": 5570078.0,
+      "step": 608
+    },
+    {
+      "epoch": 0.4627659574468085,
+      "grad_norm": 2.778101921081543,
+      "learning_rate": 4.853970389737576e-06,
+      "loss": 0.35521194338798523,
+      "mean_token_accuracy": 0.8752605319023132,
+      "num_tokens": 5573995.0,
+      "step": 609
+    },
+    {
+      "epoch": 0.4635258358662614,
+      "grad_norm": 2.600534677505493,
+      "learning_rate": 4.8532642414432675e-06,
+      "loss": 0.6541563868522644,
+      "mean_token_accuracy": 0.7843613028526306,
+      "num_tokens": 5580333.0,
+      "step": 610
+    },
+    {
+      "epoch": 0.4642857142857143,
+      "grad_norm": 1.778337836265564,
+      "learning_rate": 4.852556441537528e-06,
+      "loss": 0.3561405837535858,
+      "mean_token_accuracy": 0.8579353094100952,
+      "num_tokens": 5588430.0,
+      "step": 611
+    },
+    {
+      "epoch": 0.46504559270516715,
+      "grad_norm": 1.5653862953186035,
+      "learning_rate": 4.851846990517118e-06,
+      "loss": 0.6067906618118286,
+      "mean_token_accuracy": 0.7919317483901978,
+      "num_tokens": 5601700.0,
+      "step": 612
+    },
+    {
+      "epoch": 0.46580547112462006,
+      "grad_norm": 1.6097723245620728,
+      "learning_rate": 4.851135888879958e-06,
+      "loss": 0.446664422750473,
+      "mean_token_accuracy": 0.8441969156265259,
+      "num_tokens": 5612063.0,
+      "step": 613
+    },
+    {
+      "epoch": 0.46656534954407297,
+      "grad_norm": 1.961207389831543,
+      "learning_rate": 4.850423137125126e-06,
+      "loss": 0.5508605241775513,
+      "mean_token_accuracy": 0.8240450024604797,
+      "num_tokens": 5620245.0,
+      "step": 614
+    },
+    {
+      "epoch": 0.4673252279635258,
+      "grad_norm": 2.2189085483551025,
+      "learning_rate": 4.8497087357528585e-06,
+      "loss": 0.6805076599121094,
+      "mean_token_accuracy": 0.771978497505188,
+      "num_tokens": 5629590.0,
+      "step": 615
+    },
+    {
+      "epoch": 0.46808510638297873,
+      "grad_norm": 2.5176279544830322,
+      "learning_rate": 4.8489926852645505e-06,
+      "loss": 0.4512156844139099,
+      "mean_token_accuracy": 0.836459755897522,
+      "num_tokens": 5635259.0,
+      "step": 616
+    },
+    {
+      "epoch": 0.4688449848024316,
+      "grad_norm": 1.5327287912368774,
+      "learning_rate": 4.848274986162754e-06,
+      "loss": 0.4884302616119385,
+      "mean_token_accuracy": 0.8194037079811096,
+      "num_tokens": 5649993.0,
+      "step": 617
+    },
+    {
+      "epoch": 0.4696048632218845,
+      "grad_norm": 2.184554100036621,
+      "learning_rate": 4.847555638951177e-06,
+      "loss": 0.5141451358795166,
+      "mean_token_accuracy": 0.8245922327041626,
+      "num_tokens": 5657375.0,
+      "step": 618
+    },
+    {
+      "epoch": 0.4703647416413374,
+      "grad_norm": 1.6143407821655273,
+      "learning_rate": 4.846834644134686e-06,
+      "loss": 0.4276641607284546,
+      "mean_token_accuracy": 0.8481845855712891,
+      "num_tokens": 5667941.0,
+      "step": 619
+    },
+    {
+      "epoch": 0.47112462006079026,
+      "grad_norm": 2.3747270107269287,
+      "learning_rate": 4.846112002219301e-06,
+      "loss": 0.5608246922492981,
+      "mean_token_accuracy": 0.8073011040687561,
+      "num_tokens": 5675042.0,
+      "step": 620
+    },
+    {
+      "epoch": 0.47188449848024316,
+      "grad_norm": 2.390404224395752,
+      "learning_rate": 4.845387713712203e-06,
+      "loss": 0.46616724133491516,
+      "mean_token_accuracy": 0.8468319177627563,
+      "num_tokens": 5680207.0,
+      "step": 621
+    },
+    {
+      "epoch": 0.4726443768996961,
+      "grad_norm": 1.7245099544525146,
+      "learning_rate": 4.844661779121723e-06,
+      "loss": 0.5652435421943665,
+      "mean_token_accuracy": 0.8010749816894531,
+      "num_tokens": 5693759.0,
+      "step": 622
+    },
+    {
+      "epoch": 0.4734042553191489,
+      "grad_norm": 2.6923108100891113,
+      "learning_rate": 4.843934198957351e-06,
+      "loss": 0.6254661679267883,
+      "mean_token_accuracy": 0.8236024975776672,
+      "num_tokens": 5699916.0,
+      "step": 623
+    },
+    {
+      "epoch": 0.47416413373860183,
+      "grad_norm": 2.516901969909668,
+      "learning_rate": 4.84320497372973e-06,
+      "loss": 0.6334252953529358,
+      "mean_token_accuracy": 0.7803834676742554,
+      "num_tokens": 5706554.0,
+      "step": 624
+    },
+    {
+      "epoch": 0.4749240121580547,
+      "grad_norm": 2.3744447231292725,
+      "learning_rate": 4.842474103950658e-06,
+      "loss": 0.4221811890602112,
+      "mean_token_accuracy": 0.8639545440673828,
+      "num_tokens": 5711756.0,
+      "step": 625
+    },
+    {
+      "epoch": 0.4756838905775076,
+      "grad_norm": 3.2373476028442383,
+      "learning_rate": 4.841741590133089e-06,
+      "loss": 0.6637828946113586,
+      "mean_token_accuracy": 0.7968347072601318,
+      "num_tokens": 5716458.0,
+      "step": 626
+    },
+    {
+      "epoch": 0.4764437689969605,
+      "grad_norm": 2.153888463973999,
+      "learning_rate": 4.841007432791129e-06,
+      "loss": 0.4877486228942871,
+      "mean_token_accuracy": 0.8345249891281128,
+      "num_tokens": 5723155.0,
+      "step": 627
+    },
+    {
+      "epoch": 0.47720364741641336,
+      "grad_norm": 2.120497703552246,
+      "learning_rate": 4.8402716324400375e-06,
+      "loss": 0.37323033809661865,
+      "mean_token_accuracy": 0.8734050393104553,
+      "num_tokens": 5729171.0,
+      "step": 628
+    },
+    {
+      "epoch": 0.47796352583586627,
+      "grad_norm": 1.5294172763824463,
+      "learning_rate": 4.839534189596228e-06,
+      "loss": 0.4057067334651947,
+      "mean_token_accuracy": 0.8523319959640503,
+      "num_tokens": 5740112.0,
+      "step": 629
+    },
+    {
+      "epoch": 0.4787234042553192,
+      "grad_norm": 2.1913886070251465,
+      "learning_rate": 4.8387951047772656e-06,
+      "loss": 0.4835960865020752,
+      "mean_token_accuracy": 0.8438145518302917,
+      "num_tokens": 5746838.0,
+      "step": 630
+    },
+    {
+      "epoch": 0.479483282674772,
+      "grad_norm": 1.482897162437439,
+      "learning_rate": 4.838054378501868e-06,
+      "loss": 0.46967992186546326,
+      "mean_token_accuracy": 0.8315759897232056,
+      "num_tokens": 5760428.0,
+      "step": 631
+    },
+    {
+      "epoch": 0.48024316109422494,
+      "grad_norm": 1.38850998878479,
+      "learning_rate": 4.837312011289907e-06,
+      "loss": 0.41845446825027466,
+      "mean_token_accuracy": 0.8557186126708984,
+      "num_tokens": 5773437.0,
+      "step": 632
+    },
+    {
+      "epoch": 0.4810030395136778,
+      "grad_norm": 3.8337457180023193,
+      "learning_rate": 4.836568003662403e-06,
+      "loss": 0.5102912187576294,
+      "mean_token_accuracy": 0.830644965171814,
+      "num_tokens": 5776367.0,
+      "step": 633
+    },
+    {
+      "epoch": 0.4817629179331307,
+      "grad_norm": 1.2084007263183594,
+      "learning_rate": 4.8358223561415304e-06,
+      "loss": 0.3835333585739136,
+      "mean_token_accuracy": 0.8639016151428223,
+      "num_tokens": 5792246.0,
+      "step": 634
+    },
+    {
+      "epoch": 0.4825227963525836,
+      "grad_norm": 1.939408540725708,
+      "learning_rate": 4.835075069250613e-06,
+      "loss": 0.4044850468635559,
+      "mean_token_accuracy": 0.8488376140594482,
+      "num_tokens": 5799853.0,
+      "step": 635
+    },
+    {
+      "epoch": 0.48328267477203646,
+      "grad_norm": 1.345870852470398,
+      "learning_rate": 4.8343261435141245e-06,
+      "loss": 0.46660199761390686,
+      "mean_token_accuracy": 0.8371681571006775,
+      "num_tokens": 5817478.0,
+      "step": 636
+    },
+    {
+      "epoch": 0.48404255319148937,
+      "grad_norm": 1.6531339883804321,
+      "learning_rate": 4.833575579457691e-06,
+      "loss": 0.3886989951133728,
+      "mean_token_accuracy": 0.8763507008552551,
+      "num_tokens": 5825739.0,
+      "step": 637
+    },
+    {
+      "epoch": 0.4848024316109423,
+      "grad_norm": 1.6443969011306763,
+      "learning_rate": 4.832823377608088e-06,
+      "loss": 0.4070289731025696,
+      "mean_token_accuracy": 0.8586630821228027,
+      "num_tokens": 5837917.0,
+      "step": 638
+    },
+    {
+      "epoch": 0.48556231003039513,
+      "grad_norm": 2.005136013031006,
+      "learning_rate": 4.832069538493237e-06,
+      "loss": 0.40616685152053833,
+      "mean_token_accuracy": 0.8571510314941406,
+      "num_tokens": 5845250.0,
+      "step": 639
+    },
+    {
+      "epoch": 0.48632218844984804,
+      "grad_norm": 1.5244266986846924,
+      "learning_rate": 4.831314062642213e-06,
+      "loss": 0.49530288577079773,
+      "mean_token_accuracy": 0.8328841924667358,
+      "num_tokens": 5857407.0,
+      "step": 640
+    },
+    {
+      "epoch": 0.4870820668693009,
+      "grad_norm": 1.9876971244812012,
+      "learning_rate": 4.830556950585239e-06,
+      "loss": 0.4583776593208313,
+      "mean_token_accuracy": 0.8427221179008484,
+      "num_tokens": 5865391.0,
+      "step": 641
+    },
+    {
+      "epoch": 0.4878419452887538,
+      "grad_norm": 3.023336172103882,
+      "learning_rate": 4.829798202853683e-06,
+      "loss": 0.6134771108627319,
+      "mean_token_accuracy": 0.7981935739517212,
+      "num_tokens": 5870729.0,
+      "step": 642
+    },
+    {
+      "epoch": 0.4886018237082067,
+      "grad_norm": 1.8889515399932861,
+      "learning_rate": 4.829037819980065e-06,
+      "loss": 0.4420135021209717,
+      "mean_token_accuracy": 0.8480775356292725,
+      "num_tokens": 5878982.0,
+      "step": 643
+    },
+    {
+      "epoch": 0.48936170212765956,
+      "grad_norm": 2.2408435344696045,
+      "learning_rate": 4.828275802498051e-06,
+      "loss": 0.525706946849823,
+      "mean_token_accuracy": 0.8271557092666626,
+      "num_tokens": 5885097.0,
+      "step": 644
+    },
+    {
+      "epoch": 0.49012158054711247,
+      "grad_norm": 1.9734224081039429,
+      "learning_rate": 4.827512150942454e-06,
+      "loss": 0.44246578216552734,
+      "mean_token_accuracy": 0.8456668257713318,
+      "num_tokens": 5893941.0,
+      "step": 645
+    },
+    {
+      "epoch": 0.4908814589665654,
+      "grad_norm": 1.9618173837661743,
+      "learning_rate": 4.8267468658492335e-06,
+      "loss": 0.5119768381118774,
+      "mean_token_accuracy": 0.8355510830879211,
+      "num_tokens": 5902829.0,
+      "step": 646
+    },
+    {
+      "epoch": 0.49164133738601823,
+      "grad_norm": 1.7181587219238281,
+      "learning_rate": 4.825979947755496e-06,
+      "loss": 0.5666520595550537,
+      "mean_token_accuracy": 0.7951971888542175,
+      "num_tokens": 5915212.0,
+      "step": 647
+    },
+    {
+      "epoch": 0.49240121580547114,
+      "grad_norm": 3.0121164321899414,
+      "learning_rate": 4.8252113971994955e-06,
+      "loss": 0.628632128238678,
+      "mean_token_accuracy": 0.8041050434112549,
+      "num_tokens": 5921410.0,
+      "step": 648
+    },
+    {
+      "epoch": 0.493161094224924,
+      "grad_norm": 2.9980475902557373,
+      "learning_rate": 4.824441214720629e-06,
+      "loss": 0.4507424831390381,
+      "mean_token_accuracy": 0.8636263608932495,
+      "num_tokens": 5925179.0,
+      "step": 649
+    },
+    {
+      "epoch": 0.4939209726443769,
+      "grad_norm": 2.0096445083618164,
+      "learning_rate": 4.823669400859441e-06,
+      "loss": 0.602759838104248,
+      "mean_token_accuracy": 0.8104915618896484,
+      "num_tokens": 5934160.0,
+      "step": 650
+    },
+    {
+      "epoch": 0.4946808510638298,
+      "grad_norm": 1.1186442375183105,
+      "learning_rate": 4.8228959561576195e-06,
+      "loss": 0.41168469190597534,
+      "mean_token_accuracy": 0.8461419939994812,
+      "num_tokens": 5954163.0,
+      "step": 651
+    },
+    {
+      "epoch": 0.49544072948328266,
+      "grad_norm": 1.855465054512024,
+      "learning_rate": 4.822120881157998e-06,
+      "loss": 0.5049735307693481,
+      "mean_token_accuracy": 0.8225747346878052,
+      "num_tokens": 5963840.0,
+      "step": 652
+    },
+    {
+      "epoch": 0.49620060790273557,
+      "grad_norm": 3.550563335418701,
+      "learning_rate": 4.821344176404554e-06,
+      "loss": 0.49025264382362366,
+      "mean_token_accuracy": 0.8265978693962097,
+      "num_tokens": 5967358.0,
+      "step": 653
+    },
+    {
+      "epoch": 0.4969604863221885,
+      "grad_norm": 3.063910484313965,
+      "learning_rate": 4.820565842442408e-06,
+      "loss": 0.5652767419815063,
+      "mean_token_accuracy": 0.811700701713562,
+      "num_tokens": 5971858.0,
+      "step": 654
+    },
+    {
+      "epoch": 0.49772036474164133,
+      "grad_norm": 2.4613308906555176,
+      "learning_rate": 4.819785879817827e-06,
+      "loss": 0.5296125411987305,
+      "mean_token_accuracy": 0.8336488008499146,
+      "num_tokens": 5977442.0,
+      "step": 655
+    },
+    {
+      "epoch": 0.49848024316109424,
+      "grad_norm": 2.342519760131836,
+      "learning_rate": 4.819004289078217e-06,
+      "loss": 0.5753380060195923,
+      "mean_token_accuracy": 0.7922406792640686,
+      "num_tokens": 5984531.0,
+      "step": 656
+    },
+    {
+      "epoch": 0.4992401215805471,
+      "grad_norm": 2.0410680770874023,
+      "learning_rate": 4.818221070772129e-06,
+      "loss": 0.5433275699615479,
+      "mean_token_accuracy": 0.8043830990791321,
+      "num_tokens": 5992642.0,
+      "step": 657
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.4999698400497437,
+      "learning_rate": 4.8174362254492555e-06,
+      "loss": 0.5248899459838867,
+      "mean_token_accuracy": 0.8107168674468994,
+      "num_tokens": 6005543.0,
+      "step": 658
+    },
+    {
+      "epoch": 0.5007598784194529,
+      "grad_norm": 1.9494401216506958,
+      "learning_rate": 4.816649753660431e-06,
+      "loss": 0.41291385889053345,
+      "mean_token_accuracy": 0.8650569915771484,
+      "num_tokens": 6012185.0,
+      "step": 659
+    },
+    {
+      "epoch": 0.5015197568389058,
+      "grad_norm": 2.7514095306396484,
+      "learning_rate": 4.815861655957632e-06,
+      "loss": 0.4244142770767212,
+      "mean_token_accuracy": 0.8485112190246582,
+      "num_tokens": 6016809.0,
+      "step": 660
+    },
+    {
+      "epoch": 0.5022796352583586,
+      "grad_norm": 1.4354928731918335,
+      "learning_rate": 4.815071932893976e-06,
+      "loss": 0.4332060217857361,
+      "mean_token_accuracy": 0.8386815786361694,
+      "num_tokens": 6034795.0,
+      "step": 661
+    },
+    {
+      "epoch": 0.5030395136778115,
+      "grad_norm": 1.3113417625427246,
+      "learning_rate": 4.81428058502372e-06,
+      "loss": 0.5415540933609009,
+      "mean_token_accuracy": 0.8115285038948059,
+      "num_tokens": 6053624.0,
+      "step": 662
+    },
+    {
+      "epoch": 0.5037993920972644,
+      "grad_norm": 1.820868730545044,
+      "learning_rate": 4.813487612902265e-06,
+      "loss": 0.5360245108604431,
+      "mean_token_accuracy": 0.8313555717468262,
+      "num_tokens": 6063399.0,
+      "step": 663
+    },
+    {
+      "epoch": 0.5045592705167173,
+      "grad_norm": 2.347001552581787,
+      "learning_rate": 4.812693017086145e-06,
+      "loss": 0.4926982820034027,
+      "mean_token_accuracy": 0.8137006759643555,
+      "num_tokens": 6070111.0,
+      "step": 664
+    },
+    {
+      "epoch": 0.5053191489361702,
+      "grad_norm": 1.8830888271331787,
+      "learning_rate": 4.811896798133042e-06,
+      "loss": 0.5419014692306519,
+      "mean_token_accuracy": 0.8027454614639282,
+      "num_tokens": 6081090.0,
+      "step": 665
+    },
+    {
+      "epoch": 0.506079027355623,
+      "grad_norm": 2.3258056640625,
+      "learning_rate": 4.811098956601772e-06,
+      "loss": 0.4629337787628174,
+      "mean_token_accuracy": 0.8416580557823181,
+      "num_tokens": 6087921.0,
+      "step": 666
+    },
+    {
+      "epoch": 0.506838905775076,
+      "grad_norm": 1.9578291177749634,
+      "learning_rate": 4.810299493052289e-06,
+      "loss": 0.40305402874946594,
+      "mean_token_accuracy": 0.8529061079025269,
+      "num_tokens": 6100034.0,
+      "step": 667
+    },
+    {
+      "epoch": 0.5075987841945289,
+      "grad_norm": 2.800635576248169,
+      "learning_rate": 4.809498408045691e-06,
+      "loss": 0.5087342262268066,
+      "mean_token_accuracy": 0.8214689493179321,
+      "num_tokens": 6104742.0,
+      "step": 668
+    },
+    {
+      "epoch": 0.5083586626139818,
+      "grad_norm": 1.5318149328231812,
+      "learning_rate": 4.808695702144206e-06,
+      "loss": 0.4733222723007202,
+      "mean_token_accuracy": 0.837577223777771,
+      "num_tokens": 6117242.0,
+      "step": 669
+    },
+    {
+      "epoch": 0.5091185410334347,
+      "grad_norm": 1.2368661165237427,
+      "learning_rate": 4.807891375911207e-06,
+      "loss": 0.3929097056388855,
+      "mean_token_accuracy": 0.8331400752067566,
+      "num_tokens": 6133509.0,
+      "step": 670
+    },
+    {
+      "epoch": 0.5098784194528876,
+      "grad_norm": 2.4711415767669678,
+      "learning_rate": 4.8070854299112e-06,
+      "loss": 0.6294851303100586,
+      "mean_token_accuracy": 0.7956781983375549,
+      "num_tokens": 6140294.0,
+      "step": 671
+    },
+    {
+      "epoch": 0.5106382978723404,
+      "grad_norm": 2.590961217880249,
+      "learning_rate": 4.806277864709828e-06,
+      "loss": 0.580160915851593,
+      "mean_token_accuracy": 0.809589684009552,
+      "num_tokens": 6145803.0,
+      "step": 672
+    },
+    {
+      "epoch": 0.5113981762917933,
+      "grad_norm": 2.4653842449188232,
+      "learning_rate": 4.805468680873874e-06,
+      "loss": 0.5262120366096497,
+      "mean_token_accuracy": 0.822458803653717,
+      "num_tokens": 6151236.0,
+      "step": 673
+    },
+    {
+      "epoch": 0.5121580547112462,
+      "grad_norm": 2.860720157623291,
+      "learning_rate": 4.804657878971252e-06,
+      "loss": 0.4007391035556793,
+      "mean_token_accuracy": 0.8637382984161377,
+      "num_tokens": 6155310.0,
+      "step": 674
+    },
+    {
+      "epoch": 0.5129179331306991,
+      "grad_norm": 2.520282030105591,
+      "learning_rate": 4.803845459571014e-06,
+      "loss": 0.45798182487487793,
+      "mean_token_accuracy": 0.8270114660263062,
+      "num_tokens": 6160326.0,
+      "step": 675
+    },
+    {
+      "epoch": 0.513677811550152,
+      "grad_norm": 2.7290921211242676,
+      "learning_rate": 4.803031423243349e-06,
+      "loss": 0.5745848417282104,
+      "mean_token_accuracy": 0.8401234745979309,
+      "num_tokens": 6165709.0,
+      "step": 676
+    },
+    {
+      "epoch": 0.5144376899696048,
+      "grad_norm": 1.6678650379180908,
+      "learning_rate": 4.802215770559578e-06,
+      "loss": 0.5257721543312073,
+      "mean_token_accuracy": 0.8241991996765137,
+      "num_tokens": 6177875.0,
+      "step": 677
+    },
+    {
+      "epoch": 0.5151975683890577,
+      "grad_norm": 2.1720468997955322,
+      "learning_rate": 4.801398502092156e-06,
+      "loss": 0.45342206954956055,
+      "mean_token_accuracy": 0.8463799953460693,
+      "num_tokens": 6185415.0,
+      "step": 678
+    },
+    {
+      "epoch": 0.5159574468085106,
+      "grad_norm": 2.282259702682495,
+      "learning_rate": 4.800579618414677e-06,
+      "loss": 0.4864169955253601,
+      "mean_token_accuracy": 0.8300632238388062,
+      "num_tokens": 6191832.0,
+      "step": 679
+    },
+    {
+      "epoch": 0.5167173252279635,
+      "grad_norm": 2.0092248916625977,
+      "learning_rate": 4.799759120101861e-06,
+      "loss": 0.5781463980674744,
+      "mean_token_accuracy": 0.8267031908035278,
+      "num_tokens": 6199440.0,
+      "step": 680
+    },
+    {
+      "epoch": 0.5174772036474165,
+      "grad_norm": 1.396580696105957,
+      "learning_rate": 4.798937007729568e-06,
+      "loss": 0.49689239263534546,
+      "mean_token_accuracy": 0.8257499933242798,
+      "num_tokens": 6213840.0,
+      "step": 681
+    },
+    {
+      "epoch": 0.5182370820668692,
+      "grad_norm": 1.9060769081115723,
+      "learning_rate": 4.798113281874788e-06,
+      "loss": 0.48969539999961853,
+      "mean_token_accuracy": 0.8171790838241577,
+      "num_tokens": 6223006.0,
+      "step": 682
+    },
+    {
+      "epoch": 0.5189969604863222,
+      "grad_norm": 1.6255282163619995,
+      "learning_rate": 4.797287943115642e-06,
+      "loss": 0.5532330870628357,
+      "mean_token_accuracy": 0.8173393607139587,
+      "num_tokens": 6234857.0,
+      "step": 683
+    },
+    {
+      "epoch": 0.5197568389057751,
+      "grad_norm": 1.6923905611038208,
+      "learning_rate": 4.796460992031386e-06,
+      "loss": 0.4880887269973755,
+      "mean_token_accuracy": 0.834983229637146,
+      "num_tokens": 6245252.0,
+      "step": 684
+    },
+    {
+      "epoch": 0.520516717325228,
+      "grad_norm": 2.13161301612854,
+      "learning_rate": 4.7956324292024045e-06,
+      "loss": 0.5687593817710876,
+      "mean_token_accuracy": 0.7996571063995361,
+      "num_tokens": 6253726.0,
+      "step": 685
+    },
+    {
+      "epoch": 0.5212765957446809,
+      "grad_norm": 2.509375810623169,
+      "learning_rate": 4.794802255210217e-06,
+      "loss": 0.5396929979324341,
+      "mean_token_accuracy": 0.8007107973098755,
+      "num_tokens": 6259238.0,
+      "step": 686
+    },
+    {
+      "epoch": 0.5220364741641338,
+      "grad_norm": 2.393710136413574,
+      "learning_rate": 4.793970470637469e-06,
+      "loss": 0.6165191531181335,
+      "mean_token_accuracy": 0.7891418933868408,
+      "num_tokens": 6266325.0,
+      "step": 687
+    },
+    {
+      "epoch": 0.5227963525835866,
+      "grad_norm": 1.511647343635559,
+      "learning_rate": 4.7931370760679415e-06,
+      "loss": 0.4773876965045929,
+      "mean_token_accuracy": 0.8381044864654541,
+      "num_tokens": 6277447.0,
+      "step": 688
+    },
+    {
+      "epoch": 0.5235562310030395,
+      "grad_norm": 2.206587314605713,
+      "learning_rate": 4.792302072086542e-06,
+      "loss": 0.5482058525085449,
+      "mean_token_accuracy": 0.8239108920097351,
+      "num_tokens": 6285163.0,
+      "step": 689
+    },
+    {
+      "epoch": 0.5243161094224924,
+      "grad_norm": 3.018146514892578,
+      "learning_rate": 4.7914654592793065e-06,
+      "loss": 0.4880615472793579,
+      "mean_token_accuracy": 0.8361308574676514,
+      "num_tokens": 6289386.0,
+      "step": 690
+    },
+    {
+      "epoch": 0.5250759878419453,
+      "grad_norm": 1.6469231843948364,
+      "learning_rate": 4.790627238233405e-06,
+      "loss": 0.4164774715900421,
+      "mean_token_accuracy": 0.8496290445327759,
+      "num_tokens": 6298915.0,
+      "step": 691
+    },
+    {
+      "epoch": 0.5258358662613982,
+      "grad_norm": 2.352505922317505,
+      "learning_rate": 4.789787409537131e-06,
+      "loss": 0.5366303324699402,
+      "mean_token_accuracy": 0.8350417613983154,
+      "num_tokens": 6306130.0,
+      "step": 692
+    },
+    {
+      "epoch": 0.526595744680851,
+      "grad_norm": 1.7463021278381348,
+      "learning_rate": 4.7889459737799105e-06,
+      "loss": 0.4389137923717499,
+      "mean_token_accuracy": 0.8463300466537476,
+      "num_tokens": 6315503.0,
+      "step": 693
+    },
+    {
+      "epoch": 0.5273556231003039,
+      "grad_norm": 2.257706642150879,
+      "learning_rate": 4.788102931552294e-06,
+      "loss": 0.5309344530105591,
+      "mean_token_accuracy": 0.8164352178573608,
+      "num_tokens": 6321852.0,
+      "step": 694
+    },
+    {
+      "epoch": 0.5281155015197568,
+      "grad_norm": 2.392732620239258,
+      "learning_rate": 4.787258283445962e-06,
+      "loss": 0.3956204056739807,
+      "mean_token_accuracy": 0.8671456575393677,
+      "num_tokens": 6327380.0,
+      "step": 695
+    },
+    {
+      "epoch": 0.5288753799392097,
+      "grad_norm": 2.210514545440674,
+      "learning_rate": 4.786412030053721e-06,
+      "loss": 0.4842875003814697,
+      "mean_token_accuracy": 0.8508446216583252,
+      "num_tokens": 6334898.0,
+      "step": 696
+    },
+    {
+      "epoch": 0.5296352583586627,
+      "grad_norm": 1.8678946495056152,
+      "learning_rate": 4.785564171969503e-06,
+      "loss": 0.47399595379829407,
+      "mean_token_accuracy": 0.8514996767044067,
+      "num_tokens": 6346374.0,
+      "step": 697
+    },
+    {
+      "epoch": 0.5303951367781155,
+      "grad_norm": 2.604079484939575,
+      "learning_rate": 4.784714709788368e-06,
+      "loss": 0.5950228571891785,
+      "mean_token_accuracy": 0.7983481884002686,
+      "num_tokens": 6351648.0,
+      "step": 698
+    },
+    {
+      "epoch": 0.5311550151975684,
+      "grad_norm": 1.662381649017334,
+      "learning_rate": 4.783863644106502e-06,
+      "loss": 0.41616758704185486,
+      "mean_token_accuracy": 0.8554803133010864,
+      "num_tokens": 6360506.0,
+      "step": 699
+    },
+    {
+      "epoch": 0.5319148936170213,
+      "grad_norm": 1.6300342082977295,
+      "learning_rate": 4.783010975521216e-06,
+      "loss": 0.43029269576072693,
+      "mean_token_accuracy": 0.8443028926849365,
+      "num_tokens": 6370675.0,
+      "step": 700
+    },
+    {
+      "epoch": 0.5326747720364742,
+      "grad_norm": 1.731873869895935,
+      "learning_rate": 4.782156704630944e-06,
+      "loss": 0.4383814334869385,
+      "mean_token_accuracy": 0.8443183898925781,
+      "num_tokens": 6381803.0,
+      "step": 701
+    },
+    {
+      "epoch": 0.5334346504559271,
+      "grad_norm": 3.1788413524627686,
+      "learning_rate": 4.7813008320352475e-06,
+      "loss": 0.32194480299949646,
+      "mean_token_accuracy": 0.8870962858200073,
+      "num_tokens": 6389263.0,
+      "step": 702
+    },
+    {
+      "epoch": 0.53419452887538,
+      "grad_norm": 2.099513530731201,
+      "learning_rate": 4.78044335833481e-06,
+      "loss": 0.36962923407554626,
+      "mean_token_accuracy": 0.8661133646965027,
+      "num_tokens": 6395589.0,
+      "step": 703
+    },
+    {
+      "epoch": 0.5349544072948328,
+      "grad_norm": 1.4859435558319092,
+      "learning_rate": 4.77958428413144e-06,
+      "loss": 0.4619954824447632,
+      "mean_token_accuracy": 0.8438555002212524,
+      "num_tokens": 6407470.0,
+      "step": 704
+    },
+    {
+      "epoch": 0.5357142857142857,
+      "grad_norm": 1.2561073303222656,
+      "learning_rate": 4.7787236100280685e-06,
+      "loss": 0.3770977258682251,
+      "mean_token_accuracy": 0.8515733480453491,
+      "num_tokens": 6422888.0,
+      "step": 705
+    },
+    {
+      "epoch": 0.5364741641337386,
+      "grad_norm": 1.4455817937850952,
+      "learning_rate": 4.777861336628751e-06,
+      "loss": 0.46481069922447205,
+      "mean_token_accuracy": 0.8502002954483032,
+      "num_tokens": 6441266.0,
+      "step": 706
+    },
+    {
+      "epoch": 0.5372340425531915,
+      "grad_norm": 1.1387295722961426,
+      "learning_rate": 4.7769974645386616e-06,
+      "loss": 0.36964765191078186,
+      "mean_token_accuracy": 0.8719524145126343,
+      "num_tokens": 6463686.0,
+      "step": 707
+    },
+    {
+      "epoch": 0.5379939209726444,
+      "grad_norm": 1.7179663181304932,
+      "learning_rate": 4.776131994364102e-06,
+      "loss": 0.4231719970703125,
+      "mean_token_accuracy": 0.8416585922241211,
+      "num_tokens": 6472956.0,
+      "step": 708
+    },
+    {
+      "epoch": 0.5387537993920972,
+      "grad_norm": 1.6328502893447876,
+      "learning_rate": 4.775264926712489e-06,
+      "loss": 0.5836569666862488,
+      "mean_token_accuracy": 0.8039724230766296,
+      "num_tokens": 6485773.0,
+      "step": 709
+    },
+    {
+      "epoch": 0.5395136778115501,
+      "grad_norm": 1.8515360355377197,
+      "learning_rate": 4.774396262192368e-06,
+      "loss": 0.5477553009986877,
+      "mean_token_accuracy": 0.8136521577835083,
+      "num_tokens": 6496379.0,
+      "step": 710
+    },
+    {
+      "epoch": 0.540273556231003,
+      "grad_norm": 1.741858959197998,
+      "learning_rate": 4.7735260014133986e-06,
+      "loss": 0.4663267731666565,
+      "mean_token_accuracy": 0.8473691940307617,
+      "num_tokens": 6507652.0,
+      "step": 711
+    },
+    {
+      "epoch": 0.541033434650456,
+      "grad_norm": 1.7516659498214722,
+      "learning_rate": 4.772654144986364e-06,
+      "loss": 0.374914288520813,
+      "mean_token_accuracy": 0.8600220680236816,
+      "num_tokens": 6519030.0,
+      "step": 712
+    },
+    {
+      "epoch": 0.5417933130699089,
+      "grad_norm": 2.662343978881836,
+      "learning_rate": 4.7717806935231665e-06,
+      "loss": 0.4206875264644623,
+      "mean_token_accuracy": 0.8544126749038696,
+      "num_tokens": 6523669.0,
+      "step": 713
+    },
+    {
+      "epoch": 0.5425531914893617,
+      "grad_norm": 1.4088834524154663,
+      "learning_rate": 4.770905647636828e-06,
+      "loss": 0.5824331045150757,
+      "mean_token_accuracy": 0.7857901453971863,
+      "num_tokens": 6540560.0,
+      "step": 714
+    },
+    {
+      "epoch": 0.5433130699088146,
+      "grad_norm": 2.173656940460205,
+      "learning_rate": 4.77002900794149e-06,
+      "loss": 0.555023729801178,
+      "mean_token_accuracy": 0.8067290782928467,
+      "num_tokens": 6548946.0,
+      "step": 715
+    },
+    {
+      "epoch": 0.5440729483282675,
+      "grad_norm": 2.121018648147583,
+      "learning_rate": 4.769150775052411e-06,
+      "loss": 0.559730052947998,
+      "mean_token_accuracy": 0.8166372776031494,
+      "num_tokens": 6556065.0,
+      "step": 716
+    },
+    {
+      "epoch": 0.5448328267477204,
+      "grad_norm": 3.335866928100586,
+      "learning_rate": 4.768270949585968e-06,
+      "loss": 0.6442267894744873,
+      "mean_token_accuracy": 0.7858607769012451,
+      "num_tokens": 6560615.0,
+      "step": 717
+    },
+    {
+      "epoch": 0.5455927051671733,
+      "grad_norm": 2.3813695907592773,
+      "learning_rate": 4.767389532159659e-06,
+      "loss": 0.4027421474456787,
+      "mean_token_accuracy": 0.8635619282722473,
+      "num_tokens": 6565841.0,
+      "step": 718
+    },
+    {
+      "epoch": 0.5463525835866262,
+      "grad_norm": 2.0657708644866943,
+      "learning_rate": 4.766506523392095e-06,
+      "loss": 0.38899827003479004,
+      "mean_token_accuracy": 0.8660480380058289,
+      "num_tokens": 6572362.0,
+      "step": 719
+    },
+    {
+      "epoch": 0.547112462006079,
+      "grad_norm": 1.093705415725708,
+      "learning_rate": 4.765621923903005e-06,
+      "loss": 0.45967352390289307,
+      "mean_token_accuracy": 0.8338102102279663,
+      "num_tokens": 6595998.0,
+      "step": 720
+    },
+    {
+      "epoch": 0.5478723404255319,
+      "grad_norm": 2.942065954208374,
+      "learning_rate": 4.764735734313236e-06,
+      "loss": 0.42910510301589966,
+      "mean_token_accuracy": 0.8406122922897339,
+      "num_tokens": 6601075.0,
+      "step": 721
+    },
+    {
+      "epoch": 0.5486322188449848,
+      "grad_norm": 2.049011707305908,
+      "learning_rate": 4.763847955244749e-06,
+      "loss": 0.5584231615066528,
+      "mean_token_accuracy": 0.8171684741973877,
+      "num_tokens": 6609310.0,
+      "step": 722
+    },
+    {
+      "epoch": 0.5493920972644377,
+      "grad_norm": 2.485543966293335,
+      "learning_rate": 4.762958587320623e-06,
+      "loss": 0.5396170020103455,
+      "mean_token_accuracy": 0.8158525824546814,
+      "num_tokens": 6616185.0,
+      "step": 723
+    },
+    {
+      "epoch": 0.5501519756838906,
+      "grad_norm": 1.87015962600708,
+      "learning_rate": 4.762067631165049e-06,
+      "loss": 0.49739527702331543,
+      "mean_token_accuracy": 0.8303765654563904,
+      "num_tokens": 6625629.0,
+      "step": 724
+    },
+    {
+      "epoch": 0.5509118541033434,
+      "grad_norm": 4.239654541015625,
+      "learning_rate": 4.761175087403336e-06,
+      "loss": 0.6029239296913147,
+      "mean_token_accuracy": 0.8123486042022705,
+      "num_tokens": 6629194.0,
+      "step": 725
+    },
+    {
+      "epoch": 0.5516717325227963,
+      "grad_norm": 2.0134730339050293,
+      "learning_rate": 4.760280956661904e-06,
+      "loss": 0.4777873754501343,
+      "mean_token_accuracy": 0.8283513784408569,
+      "num_tokens": 6636929.0,
+      "step": 726
+    },
+    {
+      "epoch": 0.5524316109422492,
+      "grad_norm": 1.991780400276184,
+      "learning_rate": 4.75938523956829e-06,
+      "loss": 0.4631248116493225,
+      "mean_token_accuracy": 0.8275107741355896,
+      "num_tokens": 6645135.0,
+      "step": 727
+    },
+    {
+      "epoch": 0.5531914893617021,
+      "grad_norm": 1.423792839050293,
+      "learning_rate": 4.75848793675114e-06,
+      "loss": 0.49630722403526306,
+      "mean_token_accuracy": 0.8388000130653381,
+      "num_tokens": 6662690.0,
+      "step": 728
+    },
+    {
+      "epoch": 0.5539513677811551,
+      "grad_norm": 2.345294952392578,
+      "learning_rate": 4.757589048840219e-06,
+      "loss": 0.37830638885498047,
+      "mean_token_accuracy": 0.8782080411911011,
+      "num_tokens": 6667285.0,
+      "step": 729
+    },
+    {
+      "epoch": 0.5547112462006079,
+      "grad_norm": 2.7452144622802734,
+      "learning_rate": 4.756688576466398e-06,
+      "loss": 0.51595538854599,
+      "mean_token_accuracy": 0.8441770672798157,
+      "num_tokens": 6672324.0,
+      "step": 730
+    },
+    {
+      "epoch": 0.5554711246200608,
+      "grad_norm": 1.5247859954833984,
+      "learning_rate": 4.755786520261666e-06,
+      "loss": 0.48365193605422974,
+      "mean_token_accuracy": 0.8276445269584656,
+      "num_tokens": 6685296.0,
+      "step": 731
+    },
+    {
+      "epoch": 0.5562310030395137,
+      "grad_norm": 1.4018276929855347,
+      "learning_rate": 4.75488288085912e-06,
+      "loss": 0.3876481354236603,
+      "mean_token_accuracy": 0.8612343072891235,
+      "num_tokens": 6697515.0,
+      "step": 732
+    },
+    {
+      "epoch": 0.5569908814589666,
+      "grad_norm": 2.9570324420928955,
+      "learning_rate": 4.753977658892967e-06,
+      "loss": 0.5468149185180664,
+      "mean_token_accuracy": 0.8054271340370178,
+      "num_tokens": 6702194.0,
+      "step": 733
+    },
+    {
+      "epoch": 0.5577507598784195,
+      "grad_norm": 1.9282715320587158,
+      "learning_rate": 4.753070854998529e-06,
+      "loss": 0.4758574962615967,
+      "mean_token_accuracy": 0.8379775285720825,
+      "num_tokens": 6709938.0,
+      "step": 734
+    },
+    {
+      "epoch": 0.5585106382978723,
+      "grad_norm": 1.981264591217041,
+      "learning_rate": 4.752162469812234e-06,
+      "loss": 0.48461222648620605,
+      "mean_token_accuracy": 0.833509087562561,
+      "num_tokens": 6718125.0,
+      "step": 735
+    },
+    {
+      "epoch": 0.5592705167173252,
+      "grad_norm": 1.1643427610397339,
+      "learning_rate": 4.751252503971624e-06,
+      "loss": 0.410121887922287,
+      "mean_token_accuracy": 0.8221402764320374,
+      "num_tokens": 6735125.0,
+      "step": 736
+    },
+    {
+      "epoch": 0.5600303951367781,
+      "grad_norm": 1.786566972732544,
+      "learning_rate": 4.750340958115346e-06,
+      "loss": 0.5964341163635254,
+      "mean_token_accuracy": 0.8038164377212524,
+      "num_tokens": 6747369.0,
+      "step": 737
+    },
+    {
+      "epoch": 0.560790273556231,
+      "grad_norm": 1.7256991863250732,
+      "learning_rate": 4.749427832883158e-06,
+      "loss": 0.48737066984176636,
+      "mean_token_accuracy": 0.830894947052002,
+      "num_tokens": 6758115.0,
+      "step": 738
+    },
+    {
+      "epoch": 0.5615501519756839,
+      "grad_norm": 1.997747540473938,
+      "learning_rate": 4.748513128915928e-06,
+      "loss": 0.5238886475563049,
+      "mean_token_accuracy": 0.8066858053207397,
+      "num_tokens": 6766111.0,
+      "step": 739
+    },
+    {
+      "epoch": 0.5623100303951368,
+      "grad_norm": 2.127016305923462,
+      "learning_rate": 4.747596846855629e-06,
+      "loss": 0.5045586228370667,
+      "mean_token_accuracy": 0.821424126625061,
+      "num_tokens": 6772893.0,
+      "step": 740
+    },
+    {
+      "epoch": 0.5630699088145896,
+      "grad_norm": 1.7664796113967896,
+      "learning_rate": 4.7466789873453446e-06,
+      "loss": 0.42954835295677185,
+      "mean_token_accuracy": 0.8533384799957275,
+      "num_tokens": 6785133.0,
+      "step": 741
+    },
+    {
+      "epoch": 0.5638297872340425,
+      "grad_norm": 1.4987404346466064,
+      "learning_rate": 4.7457595510292615e-06,
+      "loss": 0.5378558039665222,
+      "mean_token_accuracy": 0.8184819221496582,
+      "num_tokens": 6799563.0,
+      "step": 742
+    },
+    {
+      "epoch": 0.5645896656534954,
+      "grad_norm": 1.4444655179977417,
+      "learning_rate": 4.744838538552678e-06,
+      "loss": 0.42193782329559326,
+      "mean_token_accuracy": 0.837514340877533,
+      "num_tokens": 6812470.0,
+      "step": 743
+    },
+    {
+      "epoch": 0.5653495440729484,
+      "grad_norm": 3.867751121520996,
+      "learning_rate": 4.7439159505619946e-06,
+      "loss": 0.4457814693450928,
+      "mean_token_accuracy": 0.8630104660987854,
+      "num_tokens": 6815652.0,
+      "step": 744
+    },
+    {
+      "epoch": 0.5661094224924013,
+      "grad_norm": 2.1250710487365723,
+      "learning_rate": 4.74299178770472e-06,
+      "loss": 0.5638922452926636,
+      "mean_token_accuracy": 0.7969781160354614,
+      "num_tokens": 6824566.0,
+      "step": 745
+    },
+    {
+      "epoch": 0.5668693009118541,
+      "grad_norm": 2.547072410583496,
+      "learning_rate": 4.742066050629465e-06,
+      "loss": 0.5516207814216614,
+      "mean_token_accuracy": 0.8160669803619385,
+      "num_tokens": 6830589.0,
+      "step": 746
+    },
+    {
+      "epoch": 0.567629179331307,
+      "grad_norm": 1.2975233793258667,
+      "learning_rate": 4.741138739985951e-06,
+      "loss": 0.3823344111442566,
+      "mean_token_accuracy": 0.8668368458747864,
+      "num_tokens": 6842707.0,
+      "step": 747
+    },
+    {
+      "epoch": 0.5683890577507599,
+      "grad_norm": 1.3410450220108032,
+      "learning_rate": 4.740209856424998e-06,
+      "loss": 0.5148671269416809,
+      "mean_token_accuracy": 0.8188045024871826,
+      "num_tokens": 6857624.0,
+      "step": 748
+    },
+    {
+      "epoch": 0.5691489361702128,
+      "grad_norm": 1.219467282295227,
+      "learning_rate": 4.7392794005985324e-06,
+      "loss": 0.3998957872390747,
+      "mean_token_accuracy": 0.855175256729126,
+      "num_tokens": 6875064.0,
+      "step": 749
+    },
+    {
+      "epoch": 0.5699088145896657,
+      "grad_norm": 1.3530343770980835,
+      "learning_rate": 4.738347373159585e-06,
+      "loss": 0.5359633564949036,
+      "mean_token_accuracy": 0.8178457021713257,
+      "num_tokens": 6890911.0,
+      "step": 750
+    },
+    {
+      "epoch": 0.5706686930091185,
+      "grad_norm": 2.146988868713379,
+      "learning_rate": 4.737413774762287e-06,
+      "loss": 0.4460008144378662,
+      "mean_token_accuracy": 0.8172903060913086,
+      "num_tokens": 6896959.0,
+      "step": 751
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 1.456023097038269,
+      "learning_rate": 4.736478606061876e-06,
+      "loss": 0.43616920709609985,
+      "mean_token_accuracy": 0.8465108871459961,
+      "num_tokens": 6908904.0,
+      "step": 752
+    },
+    {
+      "epoch": 0.5721884498480243,
+      "grad_norm": 2.9696967601776123,
+      "learning_rate": 4.735541867714687e-06,
+      "loss": 0.43464532494544983,
+      "mean_token_accuracy": 0.8608652353286743,
+      "num_tokens": 6913026.0,
+      "step": 753
+    },
+    {
+      "epoch": 0.5729483282674772,
+      "grad_norm": 2.2990667819976807,
+      "learning_rate": 4.73460356037816e-06,
+      "loss": 0.6619116067886353,
+      "mean_token_accuracy": 0.7821142673492432,
+      "num_tokens": 6920588.0,
+      "step": 754
+    },
+    {
+      "epoch": 0.5737082066869301,
+      "grad_norm": 2.054746389389038,
+      "learning_rate": 4.733663684710835e-06,
+      "loss": 0.5304250717163086,
+      "mean_token_accuracy": 0.8265531063079834,
+      "num_tokens": 6928910.0,
+      "step": 755
+    },
+    {
+      "epoch": 0.574468085106383,
+      "grad_norm": 2.0050594806671143,
+      "learning_rate": 4.732722241372354e-06,
+      "loss": 0.6393026113510132,
+      "mean_token_accuracy": 0.796819806098938,
+      "num_tokens": 6940217.0,
+      "step": 756
+    },
+    {
+      "epoch": 0.5752279635258358,
+      "grad_norm": 1.4285320043563843,
+      "learning_rate": 4.731779231023456e-06,
+      "loss": 0.5432837009429932,
+      "mean_token_accuracy": 0.8104778528213501,
+      "num_tokens": 6959101.0,
+      "step": 757
+    },
+    {
+      "epoch": 0.5759878419452887,
+      "grad_norm": 2.3941943645477295,
+      "learning_rate": 4.730834654325984e-06,
+      "loss": 0.46550673246383667,
+      "mean_token_accuracy": 0.8444503545761108,
+      "num_tokens": 6965036.0,
+      "step": 758
+    },
+    {
+      "epoch": 0.5767477203647416,
+      "grad_norm": 2.3850574493408203,
+      "learning_rate": 4.729888511942877e-06,
+      "loss": 0.4916389584541321,
+      "mean_token_accuracy": 0.8228527307510376,
+      "num_tokens": 6971184.0,
+      "step": 759
+    },
+    {
+      "epoch": 0.5775075987841946,
+      "grad_norm": 1.627480149269104,
+      "learning_rate": 4.728940804538176e-06,
+      "loss": 0.5863215923309326,
+      "mean_token_accuracy": 0.7995302677154541,
+      "num_tokens": 6982569.0,
+      "step": 760
+    },
+    {
+      "epoch": 0.5782674772036475,
+      "grad_norm": 1.1723195314407349,
+      "learning_rate": 4.727991532777016e-06,
+      "loss": 0.36908864974975586,
+      "mean_token_accuracy": 0.8355655670166016,
+      "num_tokens": 6998659.0,
+      "step": 761
+    },
+    {
+      "epoch": 0.5790273556231003,
+      "grad_norm": 1.5324925184249878,
+      "learning_rate": 4.727040697325634e-06,
+      "loss": 0.557658851146698,
+      "mean_token_accuracy": 0.8141458034515381,
+      "num_tokens": 7012969.0,
+      "step": 762
+    },
+    {
+      "epoch": 0.5797872340425532,
+      "grad_norm": 2.4106390476226807,
+      "learning_rate": 4.726088298851362e-06,
+      "loss": 0.5004243850708008,
+      "mean_token_accuracy": 0.8376860618591309,
+      "num_tokens": 7018301.0,
+      "step": 763
+    },
+    {
+      "epoch": 0.5805471124620061,
+      "grad_norm": 2.2594921588897705,
+      "learning_rate": 4.725134338022631e-06,
+      "loss": 0.6067016124725342,
+      "mean_token_accuracy": 0.8100241422653198,
+      "num_tokens": 7025201.0,
+      "step": 764
+    },
+    {
+      "epoch": 0.581306990881459,
+      "grad_norm": 1.4649826288223267,
+      "learning_rate": 4.724178815508967e-06,
+      "loss": 0.36200693249702454,
+      "mean_token_accuracy": 0.8621826171875,
+      "num_tokens": 7035112.0,
+      "step": 765
+    },
+    {
+      "epoch": 0.5820668693009119,
+      "grad_norm": 2.3634560108184814,
+      "learning_rate": 4.723221731980993e-06,
+      "loss": 0.41862213611602783,
+      "mean_token_accuracy": 0.8541463613510132,
+      "num_tokens": 7040339.0,
+      "step": 766
+    },
+    {
+      "epoch": 0.5828267477203647,
+      "grad_norm": 2.7798104286193848,
+      "learning_rate": 4.722263088110426e-06,
+      "loss": 0.4647108018398285,
+      "mean_token_accuracy": 0.8505672216415405,
+      "num_tokens": 7044880.0,
+      "step": 767
+    },
+    {
+      "epoch": 0.5835866261398176,
+      "grad_norm": 2.070528507232666,
+      "learning_rate": 4.721302884570079e-06,
+      "loss": 0.5147565007209778,
+      "mean_token_accuracy": 0.8113877773284912,
+      "num_tokens": 7052433.0,
+      "step": 768
+    },
+    {
+      "epoch": 0.5843465045592705,
+      "grad_norm": 2.1953284740448,
+      "learning_rate": 4.720341122033862e-06,
+      "loss": 0.5075466632843018,
+      "mean_token_accuracy": 0.8474211096763611,
+      "num_tokens": 7058686.0,
+      "step": 769
+    },
+    {
+      "epoch": 0.5851063829787234,
+      "grad_norm": 1.9287755489349365,
+      "learning_rate": 4.719377801176774e-06,
+      "loss": 0.5382202863693237,
+      "mean_token_accuracy": 0.8148090243339539,
+      "num_tokens": 7067538.0,
+      "step": 770
+    },
+    {
+      "epoch": 0.5858662613981763,
+      "grad_norm": 1.5574456453323364,
+      "learning_rate": 4.718412922674913e-06,
+      "loss": 0.43406790494918823,
+      "mean_token_accuracy": 0.8477081060409546,
+      "num_tokens": 7077853.0,
+      "step": 771
+    },
+    {
+      "epoch": 0.5866261398176292,
+      "grad_norm": 1.5490336418151855,
+      "learning_rate": 4.717446487205466e-06,
+      "loss": 0.43164271116256714,
+      "mean_token_accuracy": 0.8504570126533508,
+      "num_tokens": 7091728.0,
+      "step": 772
+    },
+    {
+      "epoch": 0.587386018237082,
+      "grad_norm": 1.6945984363555908,
+      "learning_rate": 4.716478495446717e-06,
+      "loss": 0.5153743624687195,
+      "mean_token_accuracy": 0.8213579058647156,
+      "num_tokens": 7108680.0,
+      "step": 773
+    },
+    {
+      "epoch": 0.5881458966565349,
+      "grad_norm": 2.2633883953094482,
+      "learning_rate": 4.715508948078037e-06,
+      "loss": 0.45254790782928467,
+      "mean_token_accuracy": 0.8392219543457031,
+      "num_tokens": 7115546.0,
+      "step": 774
+    },
+    {
+      "epoch": 0.5889057750759878,
+      "grad_norm": 1.5731090307235718,
+      "learning_rate": 4.714537845779894e-06,
+      "loss": 0.38678881525993347,
+      "mean_token_accuracy": 0.8800252676010132,
+      "num_tokens": 7126360.0,
+      "step": 775
+    },
+    {
+      "epoch": 0.5896656534954408,
+      "grad_norm": 2.4873392581939697,
+      "learning_rate": 4.7135651892338445e-06,
+      "loss": 0.5190927386283875,
+      "mean_token_accuracy": 0.8145407438278198,
+      "num_tokens": 7135705.0,
+      "step": 776
+    },
+    {
+      "epoch": 0.5904255319148937,
+      "grad_norm": 1.2931004762649536,
+      "learning_rate": 4.712590979122534e-06,
+      "loss": 0.3686544895172119,
+      "mean_token_accuracy": 0.8720537424087524,
+      "num_tokens": 7150688.0,
+      "step": 777
+    },
+    {
+      "epoch": 0.5911854103343465,
+      "grad_norm": 1.6353671550750732,
+      "learning_rate": 4.7116152161297045e-06,
+      "loss": 0.49065062403678894,
+      "mean_token_accuracy": 0.8203760385513306,
+      "num_tokens": 7161040.0,
+      "step": 778
+    },
+    {
+      "epoch": 0.5919452887537994,
+      "grad_norm": 1.2345483303070068,
+      "learning_rate": 4.710637900940181e-06,
+      "loss": 0.4004976451396942,
+      "mean_token_accuracy": 0.8302007913589478,
+      "num_tokens": 7178074.0,
+      "step": 779
+    },
+    {
+      "epoch": 0.5927051671732523,
+      "grad_norm": 2.2506837844848633,
+      "learning_rate": 4.7096590342398825e-06,
+      "loss": 0.45142874121665955,
+      "mean_token_accuracy": 0.8481036424636841,
+      "num_tokens": 7184153.0,
+      "step": 780
+    },
+    {
+      "epoch": 0.5934650455927052,
+      "grad_norm": 1.420479416847229,
+      "learning_rate": 4.708678616715815e-06,
+      "loss": 0.4802100360393524,
+      "mean_token_accuracy": 0.8586992025375366,
+      "num_tokens": 7202810.0,
+      "step": 781
+    },
+    {
+      "epoch": 0.5942249240121581,
+      "grad_norm": 3.457632303237915,
+      "learning_rate": 4.707696649056073e-06,
+      "loss": 0.5265094041824341,
+      "mean_token_accuracy": 0.8260114192962646,
+      "num_tokens": 7206396.0,
+      "step": 782
+    },
+    {
+      "epoch": 0.5949848024316109,
+      "grad_norm": 1.1592093706130981,
+      "learning_rate": 4.706713131949839e-06,
+      "loss": 0.3708173632621765,
+      "mean_token_accuracy": 0.8476542234420776,
+      "num_tokens": 7225034.0,
+      "step": 783
+    },
+    {
+      "epoch": 0.5957446808510638,
+      "grad_norm": 1.6761400699615479,
+      "learning_rate": 4.705728066087384e-06,
+      "loss": 0.4137252867221832,
+      "mean_token_accuracy": 0.8462049961090088,
+      "num_tokens": 7237101.0,
+      "step": 784
+    },
+    {
+      "epoch": 0.5965045592705167,
+      "grad_norm": 2.320185422897339,
+      "learning_rate": 4.704741452160064e-06,
+      "loss": 0.5157154202461243,
+      "mean_token_accuracy": 0.8391785621643066,
+      "num_tokens": 7243826.0,
+      "step": 785
+    },
+    {
+      "epoch": 0.5972644376899696,
+      "grad_norm": 2.079423427581787,
+      "learning_rate": 4.703753290860323e-06,
+      "loss": 0.4734993278980255,
+      "mean_token_accuracy": 0.8353281021118164,
+      "num_tokens": 7250175.0,
+      "step": 786
+    },
+    {
+      "epoch": 0.5980243161094225,
+      "grad_norm": 1.8215159177780151,
+      "learning_rate": 4.702763582881692e-06,
+      "loss": 0.520193338394165,
+      "mean_token_accuracy": 0.844062864780426,
+      "num_tokens": 7258868.0,
+      "step": 787
+    },
+    {
+      "epoch": 0.5987841945288754,
+      "grad_norm": 1.3823071718215942,
+      "learning_rate": 4.701772328918784e-06,
+      "loss": 0.4177844822406769,
+      "mean_token_accuracy": 0.8363165259361267,
+      "num_tokens": 7271744.0,
+      "step": 788
+    },
+    {
+      "epoch": 0.5995440729483282,
+      "grad_norm": 2.4749298095703125,
+      "learning_rate": 4.700779529667301e-06,
+      "loss": 0.5115069150924683,
+      "mean_token_accuracy": 0.8473520278930664,
+      "num_tokens": 7277040.0,
+      "step": 789
+    },
+    {
+      "epoch": 0.6003039513677811,
+      "grad_norm": 1.7072296142578125,
+      "learning_rate": 4.699785185824026e-06,
+      "loss": 0.5265800952911377,
+      "mean_token_accuracy": 0.8161447048187256,
+      "num_tokens": 7288288.0,
+      "step": 790
+    },
+    {
+      "epoch": 0.601063829787234,
+      "grad_norm": 1.6479384899139404,
+      "learning_rate": 4.69878929808683e-06,
+      "loss": 0.4445168972015381,
+      "mean_token_accuracy": 0.8381255865097046,
+      "num_tokens": 7298640.0,
+      "step": 791
+    },
+    {
+      "epoch": 0.601823708206687,
+      "grad_norm": 1.9095896482467651,
+      "learning_rate": 4.6977918671546635e-06,
+      "loss": 0.5841238498687744,
+      "mean_token_accuracy": 0.7971454858779907,
+      "num_tokens": 7307220.0,
+      "step": 792
+    },
+    {
+      "epoch": 0.6025835866261399,
+      "grad_norm": 1.9614146947860718,
+      "learning_rate": 4.696792893727562e-06,
+      "loss": 0.34684082865715027,
+      "mean_token_accuracy": 0.8739526271820068,
+      "num_tokens": 7313875.0,
+      "step": 793
+    },
+    {
+      "epoch": 0.6033434650455927,
+      "grad_norm": 2.015570640563965,
+      "learning_rate": 4.695792378506645e-06,
+      "loss": 0.42779117822647095,
+      "mean_token_accuracy": 0.8625012636184692,
+      "num_tokens": 7321439.0,
+      "step": 794
+    },
+    {
+      "epoch": 0.6041033434650456,
+      "grad_norm": 2.8581228256225586,
+      "learning_rate": 4.694790322194111e-06,
+      "loss": 0.6519991159439087,
+      "mean_token_accuracy": 0.7629562616348267,
+      "num_tokens": 7326916.0,
+      "step": 795
+    },
+    {
+      "epoch": 0.6048632218844985,
+      "grad_norm": 2.482715368270874,
+      "learning_rate": 4.693786725493242e-06,
+      "loss": 0.532963216304779,
+      "mean_token_accuracy": 0.832184910774231,
+      "num_tokens": 7333311.0,
+      "step": 796
+    },
+    {
+      "epoch": 0.6056231003039514,
+      "grad_norm": 1.6076741218566895,
+      "learning_rate": 4.692781589108402e-06,
+      "loss": 0.43381205201148987,
+      "mean_token_accuracy": 0.8402494192123413,
+      "num_tokens": 7343731.0,
+      "step": 797
+    },
+    {
+      "epoch": 0.6063829787234043,
+      "grad_norm": 2.2133216857910156,
+      "learning_rate": 4.691774913745033e-06,
+      "loss": 0.4380851089954376,
+      "mean_token_accuracy": 0.8600908517837524,
+      "num_tokens": 7350224.0,
+      "step": 798
+    },
+    {
+      "epoch": 0.6071428571428571,
+      "grad_norm": 2.046280860900879,
+      "learning_rate": 4.690766700109659e-06,
+      "loss": 0.3821919560432434,
+      "mean_token_accuracy": 0.8691814541816711,
+      "num_tokens": 7356717.0,
+      "step": 799
+    },
+    {
+      "epoch": 0.60790273556231,
+      "grad_norm": 1.8482693433761597,
+      "learning_rate": 4.689756948909884e-06,
+      "loss": 0.5217651128768921,
+      "mean_token_accuracy": 0.803473711013794,
+      "num_tokens": 7365806.0,
+      "step": 800
+    },
+    {
+      "epoch": 0.6086626139817629,
+      "grad_norm": 2.192134141921997,
+      "learning_rate": 4.688745660854388e-06,
+      "loss": 0.573980987071991,
+      "mean_token_accuracy": 0.8198676109313965,
+      "num_tokens": 7380281.0,
+      "step": 801
+    },
+    {
+      "epoch": 0.6094224924012158,
+      "grad_norm": 2.363626718521118,
+      "learning_rate": 4.687732836652935e-06,
+      "loss": 0.5204599499702454,
+      "mean_token_accuracy": 0.8373252153396606,
+      "num_tokens": 7386938.0,
+      "step": 802
+    },
+    {
+      "epoch": 0.6101823708206687,
+      "grad_norm": 1.9320523738861084,
+      "learning_rate": 4.686718477016361e-06,
+      "loss": 0.47316622734069824,
+      "mean_token_accuracy": 0.830596923828125,
+      "num_tokens": 7395069.0,
+      "step": 803
+    },
+    {
+      "epoch": 0.6109422492401215,
+      "grad_norm": 2.6573057174682617,
+      "learning_rate": 4.6857025826565845e-06,
+      "loss": 0.5495861768722534,
+      "mean_token_accuracy": 0.8187421560287476,
+      "num_tokens": 7400563.0,
+      "step": 804
+    },
+    {
+      "epoch": 0.6117021276595744,
+      "grad_norm": 2.0893123149871826,
+      "learning_rate": 4.684685154286599e-06,
+      "loss": 0.5362675786018372,
+      "mean_token_accuracy": 0.8394701480865479,
+      "num_tokens": 7406973.0,
+      "step": 805
+    },
+    {
+      "epoch": 0.6124620060790273,
+      "grad_norm": 2.455130100250244,
+      "learning_rate": 4.683666192620474e-06,
+      "loss": 0.5405995845794678,
+      "mean_token_accuracy": 0.8079100847244263,
+      "num_tokens": 7412931.0,
+      "step": 806
+    },
+    {
+      "epoch": 0.6132218844984803,
+      "grad_norm": 2.311915636062622,
+      "learning_rate": 4.682645698373357e-06,
+      "loss": 0.5395106077194214,
+      "mean_token_accuracy": 0.8156260251998901,
+      "num_tokens": 7419699.0,
+      "step": 807
+    },
+    {
+      "epoch": 0.6139817629179332,
+      "grad_norm": 1.686838984489441,
+      "learning_rate": 4.6816236722614694e-06,
+      "loss": 0.6034521460533142,
+      "mean_token_accuracy": 0.7855954170227051,
+      "num_tokens": 7431899.0,
+      "step": 808
+    },
+    {
+      "epoch": 0.6147416413373861,
+      "grad_norm": 1.682759165763855,
+      "learning_rate": 4.680600115002109e-06,
+      "loss": 0.48593831062316895,
+      "mean_token_accuracy": 0.8229435682296753,
+      "num_tokens": 7443187.0,
+      "step": 809
+    },
+    {
+      "epoch": 0.6155015197568389,
+      "grad_norm": 2.064589738845825,
+      "learning_rate": 4.679575027313649e-06,
+      "loss": 0.5098468661308289,
+      "mean_token_accuracy": 0.8234638571739197,
+      "num_tokens": 7450868.0,
+      "step": 810
+    },
+    {
+      "epoch": 0.6162613981762918,
+      "grad_norm": 2.2063486576080322,
+      "learning_rate": 4.6785484099155324e-06,
+      "loss": 0.5138497352600098,
+      "mean_token_accuracy": 0.8152111172676086,
+      "num_tokens": 7457176.0,
+      "step": 811
+    },
+    {
+      "epoch": 0.6170212765957447,
+      "grad_norm": 1.6258726119995117,
+      "learning_rate": 4.67752026352828e-06,
+      "loss": 0.4064181447029114,
+      "mean_token_accuracy": 0.8720619678497314,
+      "num_tokens": 7466557.0,
+      "step": 812
+    },
+    {
+      "epoch": 0.6177811550151976,
+      "grad_norm": 2.3309383392333984,
+      "learning_rate": 4.676490588873486e-06,
+      "loss": 0.5180112719535828,
+      "mean_token_accuracy": 0.8233879804611206,
+      "num_tokens": 7472650.0,
+      "step": 813
+    },
+    {
+      "epoch": 0.6185410334346505,
+      "grad_norm": 1.4545246362686157,
+      "learning_rate": 4.675459386673815e-06,
+      "loss": 0.37917959690093994,
+      "mean_token_accuracy": 0.8598103523254395,
+      "num_tokens": 7485171.0,
+      "step": 814
+    },
+    {
+      "epoch": 0.6193009118541033,
+      "grad_norm": 2.654231071472168,
+      "learning_rate": 4.674426657653003e-06,
+      "loss": 0.554074227809906,
+      "mean_token_accuracy": 0.8026446104049683,
+      "num_tokens": 7490787.0,
+      "step": 815
+    },
+    {
+      "epoch": 0.6200607902735562,
+      "grad_norm": 1.5543994903564453,
+      "learning_rate": 4.67339240253586e-06,
+      "loss": 0.6335440278053284,
+      "mean_token_accuracy": 0.783241868019104,
+      "num_tokens": 7505975.0,
+      "step": 816
+    },
+    {
+      "epoch": 0.6208206686930091,
+      "grad_norm": 2.079998016357422,
+      "learning_rate": 4.672356622048266e-06,
+      "loss": 0.5169394016265869,
+      "mean_token_accuracy": 0.8088761568069458,
+      "num_tokens": 7513470.0,
+      "step": 817
+    },
+    {
+      "epoch": 0.621580547112462,
+      "grad_norm": 1.5971896648406982,
+      "learning_rate": 4.671319316917172e-06,
+      "loss": 0.44588586688041687,
+      "mean_token_accuracy": 0.8518649339675903,
+      "num_tokens": 7524352.0,
+      "step": 818
+    },
+    {
+      "epoch": 0.6223404255319149,
+      "grad_norm": 2.477579116821289,
+      "learning_rate": 4.670280487870599e-06,
+      "loss": 0.5713893175125122,
+      "mean_token_accuracy": 0.8116940259933472,
+      "num_tokens": 7530359.0,
+      "step": 819
+    },
+    {
+      "epoch": 0.6231003039513677,
+      "grad_norm": 2.066211700439453,
+      "learning_rate": 4.669240135637635e-06,
+      "loss": 0.5295331478118896,
+      "mean_token_accuracy": 0.819536566734314,
+      "num_tokens": 7536963.0,
+      "step": 820
+    },
+    {
+      "epoch": 0.6238601823708206,
+      "grad_norm": 2.1217997074127197,
+      "learning_rate": 4.668198260948442e-06,
+      "loss": 0.6146406531333923,
+      "mean_token_accuracy": 0.7932635545730591,
+      "num_tokens": 7545800.0,
+      "step": 821
+    },
+    {
+      "epoch": 0.6246200607902735,
+      "grad_norm": 2.0173542499542236,
+      "learning_rate": 4.667154864534245e-06,
+      "loss": 0.6240535974502563,
+      "mean_token_accuracy": 0.7883644104003906,
+      "num_tokens": 7556165.0,
+      "step": 822
+    },
+    {
+      "epoch": 0.6253799392097265,
+      "grad_norm": 2.014526128768921,
+      "learning_rate": 4.666109947127343e-06,
+      "loss": 0.40367332100868225,
+      "mean_token_accuracy": 0.8653522729873657,
+      "num_tokens": 7562665.0,
+      "step": 823
+    },
+    {
+      "epoch": 0.6261398176291794,
+      "grad_norm": 2.5078861713409424,
+      "learning_rate": 4.665063509461098e-06,
+      "loss": 0.5903617739677429,
+      "mean_token_accuracy": 0.7902897596359253,
+      "num_tokens": 7568922.0,
+      "step": 824
+    },
+    {
+      "epoch": 0.6268996960486323,
+      "grad_norm": 2.454622745513916,
+      "learning_rate": 4.664015552269938e-06,
+      "loss": 0.5238361358642578,
+      "mean_token_accuracy": 0.838546872138977,
+      "num_tokens": 7575965.0,
+      "step": 825
+    },
+    {
+      "epoch": 0.6276595744680851,
+      "grad_norm": 2.920919418334961,
+      "learning_rate": 4.662966076289363e-06,
+      "loss": 0.5028782486915588,
+      "mean_token_accuracy": 0.8311152458190918,
+      "num_tokens": 7580193.0,
+      "step": 826
+    },
+    {
+      "epoch": 0.628419452887538,
+      "grad_norm": 1.545382022857666,
+      "learning_rate": 4.661915082255932e-06,
+      "loss": 0.4817378520965576,
+      "mean_token_accuracy": 0.8373227119445801,
+      "num_tokens": 7593024.0,
+      "step": 827
+    },
+    {
+      "epoch": 0.6291793313069909,
+      "grad_norm": 1.5152469873428345,
+      "learning_rate": 4.6608625709072766e-06,
+      "loss": 0.4693033695220947,
+      "mean_token_accuracy": 0.8150848150253296,
+      "num_tokens": 7606459.0,
+      "step": 828
+    },
+    {
+      "epoch": 0.6299392097264438,
+      "grad_norm": 2.1310224533081055,
+      "learning_rate": 4.659808542982089e-06,
+      "loss": 0.4653395414352417,
+      "mean_token_accuracy": 0.8286294341087341,
+      "num_tokens": 7613036.0,
+      "step": 829
+    },
+    {
+      "epoch": 0.6306990881458967,
+      "grad_norm": 2.1949679851531982,
+      "learning_rate": 4.658752999220125e-06,
+      "loss": 0.3698633909225464,
+      "mean_token_accuracy": 0.871590793132782,
+      "num_tokens": 7618527.0,
+      "step": 830
+    },
+    {
+      "epoch": 0.6314589665653495,
+      "grad_norm": 2.2770416736602783,
+      "learning_rate": 4.657695940362207e-06,
+      "loss": 0.5202419757843018,
+      "mean_token_accuracy": 0.817577600479126,
+      "num_tokens": 7624459.0,
+      "step": 831
+    },
+    {
+      "epoch": 0.6322188449848024,
+      "grad_norm": 1.402042269706726,
+      "learning_rate": 4.65663736715022e-06,
+      "loss": 0.51531583070755,
+      "mean_token_accuracy": 0.8228116631507874,
+      "num_tokens": 7639371.0,
+      "step": 832
+    },
+    {
+      "epoch": 0.6329787234042553,
+      "grad_norm": 3.3554883003234863,
+      "learning_rate": 4.65557728032711e-06,
+      "loss": 0.6771188378334045,
+      "mean_token_accuracy": 0.7880028486251831,
+      "num_tokens": 7643924.0,
+      "step": 833
+    },
+    {
+      "epoch": 0.6337386018237082,
+      "grad_norm": 2.081040143966675,
+      "learning_rate": 4.654515680636888e-06,
+      "loss": 0.5712796449661255,
+      "mean_token_accuracy": 0.8177868127822876,
+      "num_tokens": 7651881.0,
+      "step": 834
+    },
+    {
+      "epoch": 0.6344984802431611,
+      "grad_norm": 0.9128716588020325,
+      "learning_rate": 4.653452568824625e-06,
+      "loss": 0.3423936069011688,
+      "mean_token_accuracy": 0.8782886266708374,
+      "num_tokens": 7677829.0,
+      "step": 835
+    },
+    {
+      "epoch": 0.6352583586626139,
+      "grad_norm": 3.49015736579895,
+      "learning_rate": 4.652387945636454e-06,
+      "loss": 0.34657734632492065,
+      "mean_token_accuracy": 0.8770567178726196,
+      "num_tokens": 7680796.0,
+      "step": 836
+    },
+    {
+      "epoch": 0.6360182370820668,
+      "grad_norm": 2.026247501373291,
+      "learning_rate": 4.651321811819568e-06,
+      "loss": 0.5098431706428528,
+      "mean_token_accuracy": 0.8216961622238159,
+      "num_tokens": 7688746.0,
+      "step": 837
+    },
+    {
+      "epoch": 0.6367781155015197,
+      "grad_norm": 2.444343090057373,
+      "learning_rate": 4.650254168122222e-06,
+      "loss": 0.5490090250968933,
+      "mean_token_accuracy": 0.8092857599258423,
+      "num_tokens": 7695220.0,
+      "step": 838
+    },
+    {
+      "epoch": 0.6375379939209727,
+      "grad_norm": 2.0171122550964355,
+      "learning_rate": 4.649185015293728e-06,
+      "loss": 0.47221142053604126,
+      "mean_token_accuracy": 0.8514408469200134,
+      "num_tokens": 7702759.0,
+      "step": 839
+    },
+    {
+      "epoch": 0.6382978723404256,
+      "grad_norm": 1.9800984859466553,
+      "learning_rate": 4.64811435408446e-06,
+      "loss": 0.5238803625106812,
+      "mean_token_accuracy": 0.8479194641113281,
+      "num_tokens": 7714017.0,
+      "step": 840
+    },
+    {
+      "epoch": 0.6390577507598785,
+      "grad_norm": 3.0674357414245605,
+      "learning_rate": 4.647042185245848e-06,
+      "loss": 0.4668245315551758,
+      "mean_token_accuracy": 0.8381714820861816,
+      "num_tokens": 7717801.0,
+      "step": 841
+    },
+    {
+      "epoch": 0.6398176291793313,
+      "grad_norm": 1.5672820806503296,
+      "learning_rate": 4.645968509530381e-06,
+      "loss": 0.4428741931915283,
+      "mean_token_accuracy": 0.8416479825973511,
+      "num_tokens": 7728342.0,
+      "step": 842
+    },
+    {
+      "epoch": 0.6405775075987842,
+      "grad_norm": 2.3042354583740234,
+      "learning_rate": 4.644893327691608e-06,
+      "loss": 0.49937760829925537,
+      "mean_token_accuracy": 0.827070951461792,
+      "num_tokens": 7734576.0,
+      "step": 843
+    },
+    {
+      "epoch": 0.6413373860182371,
+      "grad_norm": 2.057772159576416,
+      "learning_rate": 4.6438166404841316e-06,
+      "loss": 0.5912986993789673,
+      "mean_token_accuracy": 0.805509090423584,
+      "num_tokens": 7742481.0,
+      "step": 844
+    },
+    {
+      "epoch": 0.64209726443769,
+      "grad_norm": 1.9688186645507812,
+      "learning_rate": 4.6427384486636115e-06,
+      "loss": 0.482401967048645,
+      "mean_token_accuracy": 0.8358086347579956,
+      "num_tokens": 7750002.0,
+      "step": 845
+    },
+    {
+      "epoch": 0.6428571428571429,
+      "grad_norm": 2.6852948665618896,
+      "learning_rate": 4.6416587529867665e-06,
+      "loss": 0.5479315519332886,
+      "mean_token_accuracy": 0.8091106414794922,
+      "num_tokens": 7755578.0,
+      "step": 846
+    },
+    {
+      "epoch": 0.6436170212765957,
+      "grad_norm": 2.0547337532043457,
+      "learning_rate": 4.640577554211366e-06,
+      "loss": 0.5327274203300476,
+      "mean_token_accuracy": 0.8280376195907593,
+      "num_tokens": 7763513.0,
+      "step": 847
+    },
+    {
+      "epoch": 0.6443768996960486,
+      "grad_norm": 2.0328633785247803,
+      "learning_rate": 4.63949485309624e-06,
+      "loss": 0.4814409613609314,
+      "mean_token_accuracy": 0.8527672290802002,
+      "num_tokens": 7771131.0,
+      "step": 848
+    },
+    {
+      "epoch": 0.6451367781155015,
+      "grad_norm": 1.5892863273620605,
+      "learning_rate": 4.638410650401267e-06,
+      "loss": 0.4492785334587097,
+      "mean_token_accuracy": 0.846997857093811,
+      "num_tokens": 7781572.0,
+      "step": 849
+    },
+    {
+      "epoch": 0.6458966565349544,
+      "grad_norm": 1.8295910358428955,
+      "learning_rate": 4.637324946887384e-06,
+      "loss": 0.37088239192962646,
+      "mean_token_accuracy": 0.8616628646850586,
+      "num_tokens": 7788604.0,
+      "step": 850
+    },
+    {
+      "epoch": 0.6466565349544073,
+      "grad_norm": 3.380040168762207,
+      "learning_rate": 4.636237743316578e-06,
+      "loss": 0.4737280607223511,
+      "mean_token_accuracy": 0.855940580368042,
+      "num_tokens": 7792504.0,
+      "step": 851
+    },
+    {
+      "epoch": 0.6474164133738601,
+      "grad_norm": 2.8790009021759033,
+      "learning_rate": 4.635149040451891e-06,
+      "loss": 0.39790448546409607,
+      "mean_token_accuracy": 0.8710698485374451,
+      "num_tokens": 7796333.0,
+      "step": 852
+    },
+    {
+      "epoch": 0.648176291793313,
+      "grad_norm": 1.914914608001709,
+      "learning_rate": 4.634058839057417e-06,
+      "loss": 0.2954312562942505,
+      "mean_token_accuracy": 0.8880234956741333,
+      "num_tokens": 7802456.0,
+      "step": 853
+    },
+    {
+      "epoch": 0.648936170212766,
+      "grad_norm": 1.3709120750427246,
+      "learning_rate": 4.632967139898301e-06,
+      "loss": 0.43224576115608215,
+      "mean_token_accuracy": 0.8446190357208252,
+      "num_tokens": 7816770.0,
+      "step": 854
+    },
+    {
+      "epoch": 0.6496960486322189,
+      "grad_norm": 1.6579312086105347,
+      "learning_rate": 4.63187394374074e-06,
+      "loss": 0.3535553514957428,
+      "mean_token_accuracy": 0.8738704919815063,
+      "num_tokens": 7824963.0,
+      "step": 855
+    },
+    {
+      "epoch": 0.6504559270516718,
+      "grad_norm": 2.4055678844451904,
+      "learning_rate": 4.63077925135198e-06,
+      "loss": 0.5078744292259216,
+      "mean_token_accuracy": 0.8430874347686768,
+      "num_tokens": 7830962.0,
+      "step": 856
+    },
+    {
+      "epoch": 0.6512158054711246,
+      "grad_norm": 2.5171499252319336,
+      "learning_rate": 4.629683063500319e-06,
+      "loss": 0.5172419548034668,
+      "mean_token_accuracy": 0.8087141513824463,
+      "num_tokens": 7836638.0,
+      "step": 857
+    },
+    {
+      "epoch": 0.6519756838905775,
+      "grad_norm": 1.7588486671447754,
+      "learning_rate": 4.628585380955104e-06,
+      "loss": 0.5759496092796326,
+      "mean_token_accuracy": 0.8043236136436462,
+      "num_tokens": 7844654.0,
+      "step": 858
+    },
+    {
+      "epoch": 0.6527355623100304,
+      "grad_norm": 1.5887070894241333,
+      "learning_rate": 4.62748620448673e-06,
+      "loss": 0.41849038004875183,
+      "mean_token_accuracy": 0.8556643724441528,
+      "num_tokens": 7855642.0,
+      "step": 859
+    },
+    {
+      "epoch": 0.6534954407294833,
+      "grad_norm": 3.227942705154419,
+      "learning_rate": 4.626385534866642e-06,
+      "loss": 0.5279449224472046,
+      "mean_token_accuracy": 0.8250958323478699,
+      "num_tokens": 7859890.0,
+      "step": 860
+    },
+    {
+      "epoch": 0.6542553191489362,
+      "grad_norm": 2.440467119216919,
+      "learning_rate": 4.625283372867333e-06,
+      "loss": 0.5294933319091797,
+      "mean_token_accuracy": 0.8235013484954834,
+      "num_tokens": 7866766.0,
+      "step": 861
+    },
+    {
+      "epoch": 0.6550151975683891,
+      "grad_norm": 2.4106903076171875,
+      "learning_rate": 4.624179719262342e-06,
+      "loss": 0.5662813186645508,
+      "mean_token_accuracy": 0.8061668872833252,
+      "num_tokens": 7872809.0,
+      "step": 862
+    },
+    {
+      "epoch": 0.6557750759878419,
+      "grad_norm": 3.5151145458221436,
+      "learning_rate": 4.623074574826254e-06,
+      "loss": 0.5471097230911255,
+      "mean_token_accuracy": 0.8220691084861755,
+      "num_tokens": 7876136.0,
+      "step": 863
+    },
+    {
+      "epoch": 0.6565349544072948,
+      "grad_norm": 1.5319840908050537,
+      "learning_rate": 4.621967940334705e-06,
+      "loss": 0.4178982377052307,
+      "mean_token_accuracy": 0.8517135977745056,
+      "num_tokens": 7886113.0,
+      "step": 864
+    },
+    {
+      "epoch": 0.6572948328267477,
+      "grad_norm": 1.63701331615448,
+      "learning_rate": 4.620859816564371e-06,
+      "loss": 0.4666512608528137,
+      "mean_token_accuracy": 0.8223508596420288,
+      "num_tokens": 7897982.0,
+      "step": 865
+    },
+    {
+      "epoch": 0.6580547112462006,
+      "grad_norm": 2.1515414714813232,
+      "learning_rate": 4.619750204292978e-06,
+      "loss": 0.5359305143356323,
+      "mean_token_accuracy": 0.8192868232727051,
+      "num_tokens": 7904947.0,
+      "step": 866
+    },
+    {
+      "epoch": 0.6588145896656535,
+      "grad_norm": 2.2140955924987793,
+      "learning_rate": 4.618639104299294e-06,
+      "loss": 0.5275633931159973,
+      "mean_token_accuracy": 0.8120715618133545,
+      "num_tokens": 7913913.0,
+      "step": 867
+    },
+    {
+      "epoch": 0.6595744680851063,
+      "grad_norm": 1.3956893682479858,
+      "learning_rate": 4.6175265173631304e-06,
+      "loss": 0.4378768503665924,
+      "mean_token_accuracy": 0.8479125499725342,
+      "num_tokens": 7927979.0,
+      "step": 868
+    },
+    {
+      "epoch": 0.6603343465045592,
+      "grad_norm": 2.98103928565979,
+      "learning_rate": 4.616412444265344e-06,
+      "loss": 0.42614591121673584,
+      "mean_token_accuracy": 0.8595094680786133,
+      "num_tokens": 7934293.0,
+      "step": 869
+    },
+    {
+      "epoch": 0.6610942249240122,
+      "grad_norm": 2.554845094680786,
+      "learning_rate": 4.6152968857878365e-06,
+      "loss": 0.3698030412197113,
+      "mean_token_accuracy": 0.8717041015625,
+      "num_tokens": 7938547.0,
+      "step": 870
+    },
+    {
+      "epoch": 0.6618541033434651,
+      "grad_norm": 3.0901825428009033,
+      "learning_rate": 4.6141798427135475e-06,
+      "loss": 0.5037497282028198,
+      "mean_token_accuracy": 0.8354041576385498,
+      "num_tokens": 7942829.0,
+      "step": 871
+    },
+    {
+      "epoch": 0.662613981762918,
+      "grad_norm": 2.8692073822021484,
+      "learning_rate": 4.6130613158264605e-06,
+      "loss": 0.5418164134025574,
+      "mean_token_accuracy": 0.8298909664154053,
+      "num_tokens": 7949303.0,
+      "step": 872
+    },
+    {
+      "epoch": 0.6633738601823708,
+      "grad_norm": 3.960404396057129,
+      "learning_rate": 4.611941305911602e-06,
+      "loss": 0.6284480094909668,
+      "mean_token_accuracy": 0.837495744228363,
+      "num_tokens": 7952486.0,
+      "step": 873
+    },
+    {
+      "epoch": 0.6641337386018237,
+      "grad_norm": 2.6690115928649902,
+      "learning_rate": 4.610819813755038e-06,
+      "loss": 0.5214360952377319,
+      "mean_token_accuracy": 0.8213508129119873,
+      "num_tokens": 7957559.0,
+      "step": 874
+    },
+    {
+      "epoch": 0.6648936170212766,
+      "grad_norm": 2.3376171588897705,
+      "learning_rate": 4.609696840143875e-06,
+      "loss": 0.46887528896331787,
+      "mean_token_accuracy": 0.8438819646835327,
+      "num_tokens": 7962826.0,
+      "step": 875
+    },
+    {
+      "epoch": 0.6656534954407295,
+      "grad_norm": 2.2222683429718018,
+      "learning_rate": 4.6085723858662575e-06,
+      "loss": 0.5607719421386719,
+      "mean_token_accuracy": 0.8128405809402466,
+      "num_tokens": 7970131.0,
+      "step": 876
+    },
+    {
+      "epoch": 0.6664133738601824,
+      "grad_norm": 2.069091558456421,
+      "learning_rate": 4.607446451711372e-06,
+      "loss": 0.506301760673523,
+      "mean_token_accuracy": 0.8256827592849731,
+      "num_tokens": 7977524.0,
+      "step": 877
+    },
+    {
+      "epoch": 0.6671732522796353,
+      "grad_norm": 1.3724967241287231,
+      "learning_rate": 4.606319038469443e-06,
+      "loss": 0.43285101652145386,
+      "mean_token_accuracy": 0.8525032997131348,
+      "num_tokens": 7989174.0,
+      "step": 878
+    },
+    {
+      "epoch": 0.6679331306990881,
+      "grad_norm": 2.278205156326294,
+      "learning_rate": 4.605190146931731e-06,
+      "loss": 0.4845905303955078,
+      "mean_token_accuracy": 0.8284652829170227,
+      "num_tokens": 7998524.0,
+      "step": 879
+    },
+    {
+      "epoch": 0.668693009118541,
+      "grad_norm": 1.3871766328811646,
+      "learning_rate": 4.604059777890537e-06,
+      "loss": 0.5736679434776306,
+      "mean_token_accuracy": 0.8223285675048828,
+      "num_tokens": 8015776.0,
+      "step": 880
+    },
+    {
+      "epoch": 0.6694528875379939,
+      "grad_norm": 1.926164984703064,
+      "learning_rate": 4.602927932139197e-06,
+      "loss": 0.4133230447769165,
+      "mean_token_accuracy": 0.8653768301010132,
+      "num_tokens": 8022979.0,
+      "step": 881
+    },
+    {
+      "epoch": 0.6702127659574468,
+      "grad_norm": 2.109272003173828,
+      "learning_rate": 4.601794610472083e-06,
+      "loss": 0.7005600929260254,
+      "mean_token_accuracy": 0.7777010202407837,
+      "num_tokens": 8032618.0,
+      "step": 882
+    },
+    {
+      "epoch": 0.6709726443768997,
+      "grad_norm": 2.077977418899536,
+      "learning_rate": 4.6006598136846056e-06,
+      "loss": 0.5278208255767822,
+      "mean_token_accuracy": 0.8230358958244324,
+      "num_tokens": 8040534.0,
+      "step": 883
+    },
+    {
+      "epoch": 0.6717325227963525,
+      "grad_norm": 1.678581714630127,
+      "learning_rate": 4.599523542573207e-06,
+      "loss": 0.4955351650714874,
+      "mean_token_accuracy": 0.8270003795623779,
+      "num_tokens": 8052249.0,
+      "step": 884
+    },
+    {
+      "epoch": 0.6724924012158054,
+      "grad_norm": 2.0751662254333496,
+      "learning_rate": 4.598385797935368e-06,
+      "loss": 0.5266247987747192,
+      "mean_token_accuracy": 0.8263581991195679,
+      "num_tokens": 8060600.0,
+      "step": 885
+    },
+    {
+      "epoch": 0.6732522796352584,
+      "grad_norm": 2.418405771255493,
+      "learning_rate": 4.5972465805696e-06,
+      "loss": 0.4481425881385803,
+      "mean_token_accuracy": 0.846164345741272,
+      "num_tokens": 8066025.0,
+      "step": 886
+    },
+    {
+      "epoch": 0.6740121580547113,
+      "grad_norm": 2.3936474323272705,
+      "learning_rate": 4.596105891275449e-06,
+      "loss": 0.4553404450416565,
+      "mean_token_accuracy": 0.8412896394729614,
+      "num_tokens": 8071544.0,
+      "step": 887
+    },
+    {
+      "epoch": 0.6747720364741642,
+      "grad_norm": 2.2024407386779785,
+      "learning_rate": 4.594963730853497e-06,
+      "loss": 0.6218541860580444,
+      "mean_token_accuracy": 0.7890232801437378,
+      "num_tokens": 8079061.0,
+      "step": 888
+    },
+    {
+      "epoch": 0.675531914893617,
+      "grad_norm": 2.51015567779541,
+      "learning_rate": 4.593820100105355e-06,
+      "loss": 0.5149124264717102,
+      "mean_token_accuracy": 0.8241918087005615,
+      "num_tokens": 8084293.0,
+      "step": 889
+    },
+    {
+      "epoch": 0.6762917933130699,
+      "grad_norm": 1.8748939037322998,
+      "learning_rate": 4.5926749998336665e-06,
+      "loss": 0.50836181640625,
+      "mean_token_accuracy": 0.8067223429679871,
+      "num_tokens": 8092511.0,
+      "step": 890
+    },
+    {
+      "epoch": 0.6770516717325228,
+      "grad_norm": 1.801193118095398,
+      "learning_rate": 4.5915284308421075e-06,
+      "loss": 0.4372861683368683,
+      "mean_token_accuracy": 0.8510604500770569,
+      "num_tokens": 8101174.0,
+      "step": 891
+    },
+    {
+      "epoch": 0.6778115501519757,
+      "grad_norm": 2.6476457118988037,
+      "learning_rate": 4.590380393935383e-06,
+      "loss": 0.38700711727142334,
+      "mean_token_accuracy": 0.8659796714782715,
+      "num_tokens": 8105398.0,
+      "step": 892
+    },
+    {
+      "epoch": 0.6785714285714286,
+      "grad_norm": 1.1147183179855347,
+      "learning_rate": 4.589230889919232e-06,
+      "loss": 0.38546115159988403,
+      "mean_token_accuracy": 0.8570581674575806,
+      "num_tokens": 8127394.0,
+      "step": 893
+    },
+    {
+      "epoch": 0.6793313069908815,
+      "grad_norm": 2.908905506134033,
+      "learning_rate": 4.588079919600419e-06,
+      "loss": 0.5108504295349121,
+      "mean_token_accuracy": 0.8121406435966492,
+      "num_tokens": 8131801.0,
+      "step": 894
+    },
+    {
+      "epoch": 0.6800911854103343,
+      "grad_norm": 3.1522326469421387,
+      "learning_rate": 4.586927483786739e-06,
+      "loss": 0.44059112668037415,
+      "mean_token_accuracy": 0.8448011875152588,
+      "num_tokens": 8154416.0,
+      "step": 895
+    },
+    {
+      "epoch": 0.6808510638297872,
+      "grad_norm": 1.5142440795898438,
+      "learning_rate": 4.585773583287017e-06,
+      "loss": 0.513217568397522,
+      "mean_token_accuracy": 0.8386049270629883,
+      "num_tokens": 8171156.0,
+      "step": 896
+    },
+    {
+      "epoch": 0.6816109422492401,
+      "grad_norm": 2.597881317138672,
+      "learning_rate": 4.584618218911104e-06,
+      "loss": 0.4937712550163269,
+      "mean_token_accuracy": 0.8223681449890137,
+      "num_tokens": 8176124.0,
+      "step": 897
+    },
+    {
+      "epoch": 0.682370820668693,
+      "grad_norm": 1.8185619115829468,
+      "learning_rate": 4.583461391469879e-06,
+      "loss": 0.519811749458313,
+      "mean_token_accuracy": 0.8169777393341064,
+      "num_tokens": 8185136.0,
+      "step": 898
+    },
+    {
+      "epoch": 0.6831306990881459,
+      "grad_norm": 3.2061994075775146,
+      "learning_rate": 4.582303101775249e-06,
+      "loss": 0.4655115008354187,
+      "mean_token_accuracy": 0.8425977230072021,
+      "num_tokens": 8188864.0,
+      "step": 899
+    },
+    {
+      "epoch": 0.6838905775075987,
+      "grad_norm": 1.3485229015350342,
+      "learning_rate": 4.581143350640146e-06,
+      "loss": 0.5014470815658569,
+      "mean_token_accuracy": 0.8273109197616577,
+      "num_tokens": 8203460.0,
+      "step": 900
+    },
+    {
+      "epoch": 0.6846504559270516,
+      "grad_norm": 1.3264713287353516,
+      "learning_rate": 4.579982138878527e-06,
+      "loss": 0.5073703527450562,
+      "mean_token_accuracy": 0.8259357213973999,
+      "num_tokens": 8219348.0,
+      "step": 901
+    },
+    {
+      "epoch": 0.6854103343465046,
+      "grad_norm": 2.4436347484588623,
+      "learning_rate": 4.578819467305375e-06,
+      "loss": 0.47020310163497925,
+      "mean_token_accuracy": 0.8567265272140503,
+      "num_tokens": 8224427.0,
+      "step": 902
+    },
+    {
+      "epoch": 0.6861702127659575,
+      "grad_norm": 1.921749234199524,
+      "learning_rate": 4.5776553367367e-06,
+      "loss": 0.622514009475708,
+      "mean_token_accuracy": 0.7863982319831848,
+      "num_tokens": 8233151.0,
+      "step": 903
+    },
+    {
+      "epoch": 0.6869300911854104,
+      "grad_norm": 1.8815616369247437,
+      "learning_rate": 4.576489747989532e-06,
+      "loss": 0.4910545349121094,
+      "mean_token_accuracy": 0.8147122859954834,
+      "num_tokens": 8240762.0,
+      "step": 904
+    },
+    {
+      "epoch": 0.6876899696048632,
+      "grad_norm": 1.2366989850997925,
+      "learning_rate": 4.575322701881926e-06,
+      "loss": 0.3947566747665405,
+      "mean_token_accuracy": 0.873993992805481,
+      "num_tokens": 8259381.0,
+      "step": 905
+    },
+    {
+      "epoch": 0.6884498480243161,
+      "grad_norm": 1.5767735242843628,
+      "learning_rate": 4.57415419923296e-06,
+      "loss": 0.57136070728302,
+      "mean_token_accuracy": 0.8028088808059692,
+      "num_tokens": 8273296.0,
+      "step": 906
+    },
+    {
+      "epoch": 0.689209726443769,
+      "grad_norm": 2.378675699234009,
+      "learning_rate": 4.572984240862733e-06,
+      "loss": 0.5894849896430969,
+      "mean_token_accuracy": 0.7977708578109741,
+      "num_tokens": 8280083.0,
+      "step": 907
+    },
+    {
+      "epoch": 0.6899696048632219,
+      "grad_norm": 2.0401132106781006,
+      "learning_rate": 4.57181282759237e-06,
+      "loss": 0.5524613261222839,
+      "mean_token_accuracy": 0.8138598203659058,
+      "num_tokens": 8288236.0,
+      "step": 908
+    },
+    {
+      "epoch": 0.6907294832826748,
+      "grad_norm": 2.293701648712158,
+      "learning_rate": 4.570639960244011e-06,
+      "loss": 0.5154546499252319,
+      "mean_token_accuracy": 0.8234660625457764,
+      "num_tokens": 8294493.0,
+      "step": 909
+    },
+    {
+      "epoch": 0.6914893617021277,
+      "grad_norm": 1.9286527633666992,
+      "learning_rate": 4.56946563964082e-06,
+      "loss": 0.5364264845848083,
+      "mean_token_accuracy": 0.8147368431091309,
+      "num_tokens": 8303441.0,
+      "step": 910
+    },
+    {
+      "epoch": 0.6922492401215805,
+      "grad_norm": 1.2571251392364502,
+      "learning_rate": 4.5682898666069815e-06,
+      "loss": 0.43535223603248596,
+      "mean_token_accuracy": 0.859239935874939,
+      "num_tokens": 8321548.0,
+      "step": 911
+    },
+    {
+      "epoch": 0.6930091185410334,
+      "grad_norm": 1.2224860191345215,
+      "learning_rate": 4.567112641967697e-06,
+      "loss": 0.40205076336860657,
+      "mean_token_accuracy": 0.8724711537361145,
+      "num_tokens": 8335205.0,
+      "step": 912
+    },
+    {
+      "epoch": 0.6937689969604863,
+      "grad_norm": 1.2064491510391235,
+      "learning_rate": 4.5659339665491894e-06,
+      "loss": 0.37790587544441223,
+      "mean_token_accuracy": 0.8464339971542358,
+      "num_tokens": 8350926.0,
+      "step": 913
+    },
+    {
+      "epoch": 0.6945288753799392,
+      "grad_norm": 2.1755270957946777,
+      "learning_rate": 4.5647538411786965e-06,
+      "loss": 0.42034298181533813,
+      "mean_token_accuracy": 0.84148108959198,
+      "num_tokens": 8356739.0,
+      "step": 914
+    },
+    {
+      "epoch": 0.6952887537993921,
+      "grad_norm": 1.234864592552185,
+      "learning_rate": 4.563572266684478e-06,
+      "loss": 0.5062938332557678,
+      "mean_token_accuracy": 0.8132052421569824,
+      "num_tokens": 8373660.0,
+      "step": 915
+    },
+    {
+      "epoch": 0.6960486322188449,
+      "grad_norm": 2.4250621795654297,
+      "learning_rate": 4.562389243895807e-06,
+      "loss": 0.4907791018486023,
+      "mean_token_accuracy": 0.8337979912757874,
+      "num_tokens": 8378661.0,
+      "step": 916
+    },
+    {
+      "epoch": 0.6968085106382979,
+      "grad_norm": 1.5018314123153687,
+      "learning_rate": 4.561204773642974e-06,
+      "loss": 0.41041281819343567,
+      "mean_token_accuracy": 0.8569784164428711,
+      "num_tokens": 8390322.0,
+      "step": 917
+    },
+    {
+      "epoch": 0.6975683890577508,
+      "grad_norm": 2.797269344329834,
+      "learning_rate": 4.5600188567572874e-06,
+      "loss": 0.3146931529045105,
+      "mean_token_accuracy": 0.8913302421569824,
+      "num_tokens": 8393567.0,
+      "step": 918
+    },
+    {
+      "epoch": 0.6983282674772037,
+      "grad_norm": 1.4002827405929565,
+      "learning_rate": 4.558831494071069e-06,
+      "loss": 0.4275597333908081,
+      "mean_token_accuracy": 0.8504893779754639,
+      "num_tokens": 8407119.0,
+      "step": 919
+    },
+    {
+      "epoch": 0.6990881458966566,
+      "grad_norm": 1.7045831680297852,
+      "learning_rate": 4.557642686417654e-06,
+      "loss": 0.49593430757522583,
+      "mean_token_accuracy": 0.8185091018676758,
+      "num_tokens": 8417408.0,
+      "step": 920
+    },
+    {
+      "epoch": 0.6998480243161094,
+      "grad_norm": 2.8818066120147705,
+      "learning_rate": 4.556452434631396e-06,
+      "loss": 0.637908935546875,
+      "mean_token_accuracy": 0.7883946895599365,
+      "num_tokens": 8422319.0,
+      "step": 921
+    },
+    {
+      "epoch": 0.7006079027355623,
+      "grad_norm": 2.3587265014648438,
+      "learning_rate": 4.555260739547657e-06,
+      "loss": 0.38749319314956665,
+      "mean_token_accuracy": 0.8774704933166504,
+      "num_tokens": 8427315.0,
+      "step": 922
+    },
+    {
+      "epoch": 0.7013677811550152,
+      "grad_norm": 1.6648749113082886,
+      "learning_rate": 4.554067602002815e-06,
+      "loss": 0.4044865369796753,
+      "mean_token_accuracy": 0.8524141311645508,
+      "num_tokens": 8438662.0,
+      "step": 923
+    },
+    {
+      "epoch": 0.7021276595744681,
+      "grad_norm": 3.467787742614746,
+      "learning_rate": 4.55287302283426e-06,
+      "loss": 0.591016411781311,
+      "mean_token_accuracy": 0.81184983253479,
+      "num_tokens": 8442237.0,
+      "step": 924
+    },
+    {
+      "epoch": 0.702887537993921,
+      "grad_norm": 2.1458635330200195,
+      "learning_rate": 4.551677002880395e-06,
+      "loss": 0.5017476677894592,
+      "mean_token_accuracy": 0.822914183139801,
+      "num_tokens": 8449494.0,
+      "step": 925
+    },
+    {
+      "epoch": 0.7036474164133738,
+      "grad_norm": 2.521714448928833,
+      "learning_rate": 4.550479542980632e-06,
+      "loss": 0.531912088394165,
+      "mean_token_accuracy": 0.8225687742233276,
+      "num_tokens": 8454983.0,
+      "step": 926
+    },
+    {
+      "epoch": 0.7044072948328267,
+      "grad_norm": 3.5248100757598877,
+      "learning_rate": 4.549280643975394e-06,
+      "loss": 0.4631815254688263,
+      "mean_token_accuracy": 0.8443771600723267,
+      "num_tokens": 8458504.0,
+      "step": 927
+    },
+    {
+      "epoch": 0.7051671732522796,
+      "grad_norm": 2.5105819702148438,
+      "learning_rate": 4.548080306706114e-06,
+      "loss": 0.30487123131752014,
+      "mean_token_accuracy": 0.9018767476081848,
+      "num_tokens": 8462589.0,
+      "step": 928
+    },
+    {
+      "epoch": 0.7059270516717325,
+      "grad_norm": 1.3367713689804077,
+      "learning_rate": 4.5468785320152365e-06,
+      "loss": 0.4355026185512543,
+      "mean_token_accuracy": 0.8323584794998169,
+      "num_tokens": 8478450.0,
+      "step": 929
+    },
+    {
+      "epoch": 0.7066869300911854,
+      "grad_norm": 2.2506282329559326,
+      "learning_rate": 4.545675320746212e-06,
+      "loss": 0.5082957744598389,
+      "mean_token_accuracy": 0.823430597782135,
+      "num_tokens": 8485991.0,
+      "step": 930
+    },
+    {
+      "epoch": 0.7074468085106383,
+      "grad_norm": 1.7164632081985474,
+      "learning_rate": 4.544470673743502e-06,
+      "loss": 0.3960164785385132,
+      "mean_token_accuracy": 0.8592486381530762,
+      "num_tokens": 8495217.0,
+      "step": 931
+    },
+    {
+      "epoch": 0.7082066869300911,
+      "grad_norm": 1.5864969491958618,
+      "learning_rate": 4.543264591852572e-06,
+      "loss": 0.49114471673965454,
+      "mean_token_accuracy": 0.8330780267715454,
+      "num_tokens": 8508904.0,
+      "step": 932
+    },
+    {
+      "epoch": 0.708966565349544,
+      "grad_norm": 2.1707003116607666,
+      "learning_rate": 4.542057075919898e-06,
+      "loss": 0.49895772337913513,
+      "mean_token_accuracy": 0.8327431082725525,
+      "num_tokens": 8515792.0,
+      "step": 933
+    },
+    {
+      "epoch": 0.709726443768997,
+      "grad_norm": 1.9002083539962769,
+      "learning_rate": 4.54084812679296e-06,
+      "loss": 0.4548531472682953,
+      "mean_token_accuracy": 0.834532618522644,
+      "num_tokens": 8524006.0,
+      "step": 934
+    },
+    {
+      "epoch": 0.7104863221884499,
+      "grad_norm": 1.8505141735076904,
+      "learning_rate": 4.539637745320247e-06,
+      "loss": 0.35716521739959717,
+      "mean_token_accuracy": 0.872222900390625,
+      "num_tokens": 8533647.0,
+      "step": 935
+    },
+    {
+      "epoch": 0.7112462006079028,
+      "grad_norm": 2.092620849609375,
+      "learning_rate": 4.53842593235125e-06,
+      "loss": 0.4673694372177124,
+      "mean_token_accuracy": 0.8460999131202698,
+      "num_tokens": 8540734.0,
+      "step": 936
+    },
+    {
+      "epoch": 0.7120060790273556,
+      "grad_norm": 2.689514636993408,
+      "learning_rate": 4.537212688736466e-06,
+      "loss": 0.45461273193359375,
+      "mean_token_accuracy": 0.8450704216957092,
+      "num_tokens": 8544948.0,
+      "step": 937
+    },
+    {
+      "epoch": 0.7127659574468085,
+      "grad_norm": 2.4507734775543213,
+      "learning_rate": 4.535998015327396e-06,
+      "loss": 0.4571906626224518,
+      "mean_token_accuracy": 0.8429360389709473,
+      "num_tokens": 8550445.0,
+      "step": 938
+    },
+    {
+      "epoch": 0.7135258358662614,
+      "grad_norm": 1.8960013389587402,
+      "learning_rate": 4.534781912976546e-06,
+      "loss": 0.4461391568183899,
+      "mean_token_accuracy": 0.8487973213195801,
+      "num_tokens": 8557630.0,
+      "step": 939
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 1.602611780166626,
+      "learning_rate": 4.533564382537421e-06,
+      "loss": 0.5277102589607239,
+      "mean_token_accuracy": 0.8330916166305542,
+      "num_tokens": 8570397.0,
+      "step": 940
+    },
+    {
+      "epoch": 0.7150455927051672,
+      "grad_norm": 1.8936395645141602,
+      "learning_rate": 4.532345424864533e-06,
+      "loss": 0.38619571924209595,
+      "mean_token_accuracy": 0.8514572381973267,
+      "num_tokens": 8582673.0,
+      "step": 941
+    },
+    {
+      "epoch": 0.71580547112462,
+      "grad_norm": 1.3898619413375854,
+      "learning_rate": 4.531125040813392e-06,
+      "loss": 0.4825032949447632,
+      "mean_token_accuracy": 0.833012580871582,
+      "num_tokens": 8597239.0,
+      "step": 942
+    },
+    {
+      "epoch": 0.7165653495440729,
+      "grad_norm": 2.128230571746826,
+      "learning_rate": 4.529903231240511e-06,
+      "loss": 0.4862118065357208,
+      "mean_token_accuracy": 0.8210917711257935,
+      "num_tokens": 8605877.0,
+      "step": 943
+    },
+    {
+      "epoch": 0.7173252279635258,
+      "grad_norm": 1.6552259922027588,
+      "learning_rate": 4.528679997003403e-06,
+      "loss": 0.5092059373855591,
+      "mean_token_accuracy": 0.8247389793395996,
+      "num_tokens": 8617060.0,
+      "step": 944
+    },
+    {
+      "epoch": 0.7180851063829787,
+      "grad_norm": 2.1174771785736084,
+      "learning_rate": 4.52745533896058e-06,
+      "loss": 0.39110174775123596,
+      "mean_token_accuracy": 0.8672944903373718,
+      "num_tokens": 8623306.0,
+      "step": 945
+    },
+    {
+      "epoch": 0.7188449848024316,
+      "grad_norm": 2.8648383617401123,
+      "learning_rate": 4.526229257971556e-06,
+      "loss": 0.49864327907562256,
+      "mean_token_accuracy": 0.8305130004882812,
+      "num_tokens": 8627466.0,
+      "step": 946
+    },
+    {
+      "epoch": 0.7196048632218845,
+      "grad_norm": 2.155514717102051,
+      "learning_rate": 4.52500175489684e-06,
+      "loss": 0.5070191025733948,
+      "mean_token_accuracy": 0.8311188817024231,
+      "num_tokens": 8634759.0,
+      "step": 947
+    },
+    {
+      "epoch": 0.7203647416413373,
+      "grad_norm": 1.8432683944702148,
+      "learning_rate": 4.523772830597942e-06,
+      "loss": 0.5569252371788025,
+      "mean_token_accuracy": 0.8070821762084961,
+      "num_tokens": 8644160.0,
+      "step": 948
+    },
+    {
+      "epoch": 0.7211246200607903,
+      "grad_norm": 2.8912241458892822,
+      "learning_rate": 4.522542485937369e-06,
+      "loss": 0.4799427390098572,
+      "mean_token_accuracy": 0.8443552851676941,
+      "num_tokens": 8648377.0,
+      "step": 949
+    },
+    {
+      "epoch": 0.7218844984802432,
+      "grad_norm": 3.3449625968933105,
+      "learning_rate": 4.521310721778622e-06,
+      "loss": 0.44043463468551636,
+      "mean_token_accuracy": 0.8521315455436707,
+      "num_tokens": 8651846.0,
+      "step": 950
+    },
+    {
+      "epoch": 0.7226443768996961,
+      "grad_norm": 1.4127917289733887,
+      "learning_rate": 4.520077538986203e-06,
+      "loss": 0.4700999855995178,
+      "mean_token_accuracy": 0.8377952575683594,
+      "num_tokens": 8665199.0,
+      "step": 951
+    },
+    {
+      "epoch": 0.723404255319149,
+      "grad_norm": 2.1607301235198975,
+      "learning_rate": 4.518842938425606e-06,
+      "loss": 0.4374256730079651,
+      "mean_token_accuracy": 0.8448896408081055,
+      "num_tokens": 8672158.0,
+      "step": 952
+    },
+    {
+      "epoch": 0.7241641337386018,
+      "grad_norm": 1.3442779779434204,
+      "learning_rate": 4.51760692096332e-06,
+      "loss": 0.38948923349380493,
+      "mean_token_accuracy": 0.8598923683166504,
+      "num_tokens": 8684532.0,
+      "step": 953
+    },
+    {
+      "epoch": 0.7249240121580547,
+      "grad_norm": 2.0003178119659424,
+      "learning_rate": 4.516369487466832e-06,
+      "loss": 0.3797217011451721,
+      "mean_token_accuracy": 0.8652102947235107,
+      "num_tokens": 8691460.0,
+      "step": 954
+    },
+    {
+      "epoch": 0.7256838905775076,
+      "grad_norm": 1.8196535110473633,
+      "learning_rate": 4.5151306388046175e-06,
+      "loss": 0.5676811933517456,
+      "mean_token_accuracy": 0.818500816822052,
+      "num_tokens": 8701624.0,
+      "step": 955
+    },
+    {
+      "epoch": 0.7264437689969605,
+      "grad_norm": 2.1962296962738037,
+      "learning_rate": 4.513890375846152e-06,
+      "loss": 0.45399484038352966,
+      "mean_token_accuracy": 0.8463879227638245,
+      "num_tokens": 8707410.0,
+      "step": 956
+    },
+    {
+      "epoch": 0.7272036474164134,
+      "grad_norm": 1.8798872232437134,
+      "learning_rate": 4.512648699461897e-06,
+      "loss": 0.5679811239242554,
+      "mean_token_accuracy": 0.8089900016784668,
+      "num_tokens": 8715630.0,
+      "step": 957
+    },
+    {
+      "epoch": 0.7279635258358662,
+      "grad_norm": 2.3540258407592773,
+      "learning_rate": 4.511405610523309e-06,
+      "loss": 0.5282865762710571,
+      "mean_token_accuracy": 0.8196114301681519,
+      "num_tokens": 8721934.0,
+      "step": 958
+    },
+    {
+      "epoch": 0.7287234042553191,
+      "grad_norm": 2.5630908012390137,
+      "learning_rate": 4.510161109902837e-06,
+      "loss": 0.39442378282546997,
+      "mean_token_accuracy": 0.8400980830192566,
+      "num_tokens": 8726511.0,
+      "step": 959
+    },
+    {
+      "epoch": 0.729483282674772,
+      "grad_norm": 1.9829226732254028,
+      "learning_rate": 4.508915198473919e-06,
+      "loss": 0.4611976742744446,
+      "mean_token_accuracy": 0.8439624309539795,
+      "num_tokens": 8733460.0,
+      "step": 960
+    },
+    {
+      "epoch": 0.7302431610942249,
+      "grad_norm": 3.0291950702667236,
+      "learning_rate": 4.507667877110982e-06,
+      "loss": 0.5158340930938721,
+      "mean_token_accuracy": 0.8300060033798218,
+      "num_tokens": 8737629.0,
+      "step": 961
+    },
+    {
+      "epoch": 0.7310030395136778,
+      "grad_norm": 1.9208252429962158,
+      "learning_rate": 4.506419146689445e-06,
+      "loss": 0.3807099163532257,
+      "mean_token_accuracy": 0.871469259262085,
+      "num_tokens": 8744615.0,
+      "step": 962
+    },
+    {
+      "epoch": 0.7317629179331308,
+      "grad_norm": 3.051565408706665,
+      "learning_rate": 4.505169008085717e-06,
+      "loss": 0.38461726903915405,
+      "mean_token_accuracy": 0.874465823173523,
+      "num_tokens": 8748154.0,
+      "step": 963
+    },
+    {
+      "epoch": 0.7325227963525835,
+      "grad_norm": 1.375466227531433,
+      "learning_rate": 4.503917462177192e-06,
+      "loss": 0.42490679025650024,
+      "mean_token_accuracy": 0.8457326889038086,
+      "num_tokens": 8760965.0,
+      "step": 964
+    },
+    {
+      "epoch": 0.7332826747720365,
+      "grad_norm": 2.216681957244873,
+      "learning_rate": 4.5026645098422515e-06,
+      "loss": 0.43149900436401367,
+      "mean_token_accuracy": 0.8527278900146484,
+      "num_tokens": 8766996.0,
+      "step": 965
+    },
+    {
+      "epoch": 0.7340425531914894,
+      "grad_norm": 1.9422595500946045,
+      "learning_rate": 4.5014101519602684e-06,
+      "loss": 0.4964504539966583,
+      "mean_token_accuracy": 0.8137556314468384,
+      "num_tokens": 8774411.0,
+      "step": 966
+    },
+    {
+      "epoch": 0.7348024316109423,
+      "grad_norm": 2.058887004852295,
+      "learning_rate": 4.500154389411598e-06,
+      "loss": 0.4977570176124573,
+      "mean_token_accuracy": 0.8254626989364624,
+      "num_tokens": 8782220.0,
+      "step": 967
+    },
+    {
+      "epoch": 0.7355623100303952,
+      "grad_norm": 2.9977786540985107,
+      "learning_rate": 4.498897223077582e-06,
+      "loss": 0.4061415195465088,
+      "mean_token_accuracy": 0.8752427101135254,
+      "num_tokens": 8786120.0,
+      "step": 968
+    },
+    {
+      "epoch": 0.736322188449848,
+      "grad_norm": 2.2636303901672363,
+      "learning_rate": 4.49763865384055e-06,
+      "loss": 0.5062161087989807,
+      "mean_token_accuracy": 0.8171653747558594,
+      "num_tokens": 8792459.0,
+      "step": 969
+    },
+    {
+      "epoch": 0.7370820668693009,
+      "grad_norm": 1.8850842714309692,
+      "learning_rate": 4.496378682583813e-06,
+      "loss": 0.5014280676841736,
+      "mean_token_accuracy": 0.8547511100769043,
+      "num_tokens": 8800675.0,
+      "step": 970
+    },
+    {
+      "epoch": 0.7378419452887538,
+      "grad_norm": 1.191985011100769,
+      "learning_rate": 4.495117310191667e-06,
+      "loss": 0.4713883101940155,
+      "mean_token_accuracy": 0.8213596343994141,
+      "num_tokens": 8820740.0,
+      "step": 971
+    },
+    {
+      "epoch": 0.7386018237082067,
+      "grad_norm": 1.823000192642212,
+      "learning_rate": 4.493854537549393e-06,
+      "loss": 0.46332645416259766,
+      "mean_token_accuracy": 0.8359860777854919,
+      "num_tokens": 8828884.0,
+      "step": 972
+    },
+    {
+      "epoch": 0.7393617021276596,
+      "grad_norm": 2.590446949005127,
+      "learning_rate": 4.492590365543253e-06,
+      "loss": 0.49074703454971313,
+      "mean_token_accuracy": 0.8433758020401001,
+      "num_tokens": 8833859.0,
+      "step": 973
+    },
+    {
+      "epoch": 0.7401215805471124,
+      "grad_norm": 2.2762670516967773,
+      "learning_rate": 4.491324795060491e-06,
+      "loss": 0.39465656876564026,
+      "mean_token_accuracy": 0.8734766244888306,
+      "num_tokens": 8839350.0,
+      "step": 974
+    },
+    {
+      "epoch": 0.7408814589665653,
+      "grad_norm": 2.698725461959839,
+      "learning_rate": 4.490057826989333e-06,
+      "loss": 0.5552085041999817,
+      "mean_token_accuracy": 0.8132266998291016,
+      "num_tokens": 8844373.0,
+      "step": 975
+    },
+    {
+      "epoch": 0.7416413373860182,
+      "grad_norm": 2.704606294631958,
+      "learning_rate": 4.488789462218988e-06,
+      "loss": 0.3447791635990143,
+      "mean_token_accuracy": 0.8736170530319214,
+      "num_tokens": 8848236.0,
+      "step": 976
+    },
+    {
+      "epoch": 0.7424012158054711,
+      "grad_norm": 3.1260716915130615,
+      "learning_rate": 4.487519701639641e-06,
+      "loss": 0.5945233702659607,
+      "mean_token_accuracy": 0.7997599840164185,
+      "num_tokens": 8852935.0,
+      "step": 977
+    },
+    {
+      "epoch": 0.743161094224924,
+      "grad_norm": 1.6895452737808228,
+      "learning_rate": 4.486248546142459e-06,
+      "loss": 0.4823892116546631,
+      "mean_token_accuracy": 0.8279662132263184,
+      "num_tokens": 8861743.0,
+      "step": 978
+    },
+    {
+      "epoch": 0.743920972644377,
+      "grad_norm": 1.9161452054977417,
+      "learning_rate": 4.4849759966195885e-06,
+      "loss": 0.5266581773757935,
+      "mean_token_accuracy": 0.8218623399734497,
+      "num_tokens": 8870601.0,
+      "step": 979
+    },
+    {
+      "epoch": 0.7446808510638298,
+      "grad_norm": 1.6894301176071167,
+      "learning_rate": 4.483702053964154e-06,
+      "loss": 0.4186219573020935,
+      "mean_token_accuracy": 0.8471781015396118,
+      "num_tokens": 8885617.0,
+      "step": 980
+    },
+    {
+      "epoch": 0.7454407294832827,
+      "grad_norm": 1.6319992542266846,
+      "learning_rate": 4.482426719070258e-06,
+      "loss": 0.541317880153656,
+      "mean_token_accuracy": 0.8216162323951721,
+      "num_tokens": 8897595.0,
+      "step": 981
+    },
+    {
+      "epoch": 0.7462006079027356,
+      "grad_norm": 5.102413177490234,
+      "learning_rate": 4.4811499928329775e-06,
+      "loss": 0.3928517699241638,
+      "mean_token_accuracy": 0.858033299446106,
+      "num_tokens": 8901682.0,
+      "step": 982
+    },
+    {
+      "epoch": 0.7469604863221885,
+      "grad_norm": 2.213860273361206,
+      "learning_rate": 4.479871876148368e-06,
+      "loss": 0.4276347756385803,
+      "mean_token_accuracy": 0.8529798984527588,
+      "num_tokens": 8908088.0,
+      "step": 983
+    },
+    {
+      "epoch": 0.7477203647416414,
+      "grad_norm": 1.2180038690567017,
+      "learning_rate": 4.478592369913464e-06,
+      "loss": 0.3941590189933777,
+      "mean_token_accuracy": 0.8608149290084839,
+      "num_tokens": 8925876.0,
+      "step": 984
+    },
+    {
+      "epoch": 0.7484802431610942,
+      "grad_norm": 2.849802255630493,
+      "learning_rate": 4.477311475026271e-06,
+      "loss": 0.42190325260162354,
+      "mean_token_accuracy": 0.860505223274231,
+      "num_tokens": 8930190.0,
+      "step": 985
+    },
+    {
+      "epoch": 0.7492401215805471,
+      "grad_norm": 1.704128384590149,
+      "learning_rate": 4.476029192385769e-06,
+      "loss": 0.4786282777786255,
+      "mean_token_accuracy": 0.8302322626113892,
+      "num_tokens": 8938340.0,
+      "step": 986
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.06322979927063,
+      "learning_rate": 4.474745522891915e-06,
+      "loss": 0.4648786187171936,
+      "mean_token_accuracy": 0.8366481065750122,
+      "num_tokens": 8944633.0,
+      "step": 987
+    },
+    {
+      "epoch": 0.7507598784194529,
+      "grad_norm": 2.0745396614074707,
+      "learning_rate": 4.473460467445637e-06,
+      "loss": 0.5744885206222534,
+      "mean_token_accuracy": 0.8357284069061279,
+      "num_tokens": 8954457.0,
+      "step": 988
+    },
+    {
+      "epoch": 0.7515197568389058,
+      "grad_norm": 1.9281407594680786,
+      "learning_rate": 4.472174026948836e-06,
+      "loss": 0.528974175453186,
+      "mean_token_accuracy": 0.8083580732345581,
+      "num_tokens": 8962701.0,
+      "step": 989
+    },
+    {
+      "epoch": 0.7522796352583586,
+      "grad_norm": 3.012381076812744,
+      "learning_rate": 4.470886202304385e-06,
+      "loss": 0.48754751682281494,
+      "mean_token_accuracy": 0.8368391990661621,
+      "num_tokens": 8967272.0,
+      "step": 990
+    },
+    {
+      "epoch": 0.7530395136778115,
+      "grad_norm": 1.691826581954956,
+      "learning_rate": 4.469596994416131e-06,
+      "loss": 0.484740674495697,
+      "mean_token_accuracy": 0.8500643968582153,
+      "num_tokens": 8976615.0,
+      "step": 991
+    },
+    {
+      "epoch": 0.7537993920972644,
+      "grad_norm": 2.4961965084075928,
+      "learning_rate": 4.468306404188887e-06,
+      "loss": 0.50777268409729,
+      "mean_token_accuracy": 0.8168395757675171,
+      "num_tokens": 8983235.0,
+      "step": 992
+    },
+    {
+      "epoch": 0.7545592705167173,
+      "grad_norm": 1.512007713317871,
+      "learning_rate": 4.467014432528441e-06,
+      "loss": 0.4583340287208557,
+      "mean_token_accuracy": 0.8465162515640259,
+      "num_tokens": 8993815.0,
+      "step": 993
+    },
+    {
+      "epoch": 0.7553191489361702,
+      "grad_norm": 1.9362257719039917,
+      "learning_rate": 4.465721080341547e-06,
+      "loss": 0.6027892827987671,
+      "mean_token_accuracy": 0.8052380084991455,
+      "num_tokens": 9002697.0,
+      "step": 994
+    },
+    {
+      "epoch": 0.756079027355623,
+      "grad_norm": 2.473632335662842,
+      "learning_rate": 4.4644263485359316e-06,
+      "loss": 0.5394320487976074,
+      "mean_token_accuracy": 0.834665834903717,
+      "num_tokens": 9007428.0,
+      "step": 995
+    },
+    {
+      "epoch": 0.756838905775076,
+      "grad_norm": 2.2527434825897217,
+      "learning_rate": 4.463130238020284e-06,
+      "loss": 0.5485198497772217,
+      "mean_token_accuracy": 0.8090173006057739,
+      "num_tokens": 9013570.0,
+      "step": 996
+    },
+    {
+      "epoch": 0.7575987841945289,
+      "grad_norm": 1.4130940437316895,
+      "learning_rate": 4.4618327497042676e-06,
+      "loss": 0.37994423508644104,
+      "mean_token_accuracy": 0.8625167012214661,
+      "num_tokens": 9025485.0,
+      "step": 997
+    },
+    {
+      "epoch": 0.7583586626139818,
+      "grad_norm": 2.685115098953247,
+      "learning_rate": 4.460533884498509e-06,
+      "loss": 0.447973370552063,
+      "mean_token_accuracy": 0.8564165234565735,
+      "num_tokens": 9030355.0,
+      "step": 998
+    },
+    {
+      "epoch": 0.7591185410334347,
+      "grad_norm": 3.2743139266967773,
+      "learning_rate": 4.4592336433146e-06,
+      "loss": 0.45275989174842834,
+      "mean_token_accuracy": 0.8462578058242798,
+      "num_tokens": 9034406.0,
+      "step": 999
+    },
+    {
+      "epoch": 0.7598784194528876,
+      "grad_norm": 1.9383049011230469,
+      "learning_rate": 4.457932027065102e-06,
+      "loss": 0.5387729406356812,
+      "mean_token_accuracy": 0.8357330560684204,
+      "num_tokens": 9041502.0,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3948,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.855721706985882e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}