diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15833 @@
+{
+  "best_metric": 2.7007529735565186,
+  "best_model_checkpoint": "fat5-fr-small_v1/checkpoint-200000",
+  "epoch": 2.6882443076426785,
+  "eval_steps": 1000,
+  "global_step": 200000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0013441221538213392,
+      "grad_norm": 0.44315826892852783,
+      "learning_rate": 0.002512499999999994,
+      "loss": 7.5626,
+      "step": 100
+    },
+    {
+      "epoch": 0.0026882443076426785,
+      "grad_norm": 0.5437139272689819,
+      "learning_rate": 0.0025249999999999904,
+      "loss": 6.2859,
+      "step": 200
+    },
+    {
+      "epoch": 0.004032366461464018,
+      "grad_norm": 0.7252405881881714,
+      "learning_rate": 0.00253749999999999,
+      "loss": 5.8614,
+      "step": 300
+    },
+    {
+      "epoch": 0.005376488615285357,
+      "grad_norm": 1.1030696630477905,
+      "learning_rate": 0.0025499999999999885,
+      "loss": 5.6073,
+      "step": 400
+    },
+    {
+      "epoch": 0.006720610769106696,
+      "grad_norm": 0.8097702264785767,
+      "learning_rate": 0.0025624999999999862,
+      "loss": 5.4185,
+      "step": 500
+    },
+    {
+      "epoch": 0.008064732922928036,
+      "grad_norm": 0.9873973727226257,
+      "learning_rate": 0.0025749999999999827,
+      "loss": 5.2865,
+      "step": 600
+    },
+    {
+      "epoch": 0.009408855076749375,
+      "grad_norm": 1.1101303100585938,
+      "learning_rate": 0.0025874999999999813,
+      "loss": 5.1929,
+      "step": 700
+    },
+    {
+      "epoch": 0.010752977230570714,
+      "grad_norm": 1.1394418478012085,
+      "learning_rate": 0.0025999999999999795,
+      "loss": 5.1122,
+      "step": 800
+    },
+    {
+      "epoch": 0.012097099384392053,
+      "grad_norm": 1.2768878936767578,
+      "learning_rate": 0.0026124999999999764,
+      "loss": 5.0262,
+      "step": 900
+    },
+    {
+      "epoch": 0.013441221538213392,
+      "grad_norm": 1.238966464996338,
+      "learning_rate": 0.00262499999999997,
+      "loss": 4.9743,
+      "step": 1000
+    },
+    {
+      "epoch": 0.013441221538213392,
+      "eval_MaskedAccuracy": 0.19784819583436258,
+      "eval_loss": 5.1476359367370605,
+      "eval_runtime": 157.0766,
+      "eval_samples_per_second": 404.109,
+      "eval_steps_per_second": 1.579,
+      "step": 1000
+    },
+    {
+      "epoch": 0.014785343692034733,
+      "grad_norm": 1.1027828454971313,
+      "learning_rate": 0.002637499999999968,
+      "loss": 4.9126,
+      "step": 1100
+    },
+    {
+      "epoch": 0.016129465845856072,
+      "grad_norm": 1.021935224533081,
+      "learning_rate": 0.002649999999999965,
+      "loss": 4.855,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01747358799967741,
+      "grad_norm": 0.9193494319915771,
+      "learning_rate": 0.0026624999999999635,
+      "loss": 4.8046,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01881771015349875,
+      "grad_norm": 1.0063543319702148,
+      "learning_rate": 0.002674999999999961,
+      "loss": 4.7762,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02016183230732009,
+      "grad_norm": 1.034637689590454,
+      "learning_rate": 0.0026874999999999564,
+      "loss": 4.7222,
+      "step": 1500
+    },
+    {
+      "epoch": 0.021505954461141428,
+      "grad_norm": 0.9538853764533997,
+      "learning_rate": 0.002699999999999955,
+      "loss": 4.6804,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02285007661496277,
+      "grad_norm": 0.8716675043106079,
+      "learning_rate": 0.002712499999999955,
+      "loss": 4.6452,
+      "step": 1700
+    },
+    {
+      "epoch": 0.024194198768784106,
+      "grad_norm": 1.194484829902649,
+      "learning_rate": 0.0027249999999999514,
+      "loss": 4.6026,
+      "step": 1800
+    },
+    {
+      "epoch": 0.025538320922605447,
+      "grad_norm": 0.9794184565544128,
+      "learning_rate": 0.002737499999999953,
+      "loss": 4.5661,
+      "step": 1900
+    },
+    {
+      "epoch": 0.026882443076426784,
+      "grad_norm": 0.8896113038063049,
+      "learning_rate": 0.0027499999999999495,
+      "loss": 4.528,
+      "step": 2000
+    },
+    {
+      "epoch": 0.026882443076426784,
+      "eval_MaskedAccuracy": 0.22670220534969857,
+      "eval_loss": 4.742874622344971,
+      "eval_runtime": 155.1957,
+      "eval_samples_per_second": 409.006,
+      "eval_steps_per_second": 1.598,
+      "step": 2000
+    },
+    {
+      "epoch": 0.028226565230248125,
+      "grad_norm": 0.9724192023277283,
+      "learning_rate": 0.0027624999999999425,
+      "loss": 4.503,
+      "step": 2100
+    },
+    {
+      "epoch": 0.029570687384069465,
+      "grad_norm": 0.9325888752937317,
+      "learning_rate": 0.0027749999999999394,
+      "loss": 4.4762,
+      "step": 2200
+    },
+    {
+      "epoch": 0.030914809537890803,
+      "grad_norm": 0.8246894478797913,
+      "learning_rate": 0.0027874999999999363,
+      "loss": 4.4439,
+      "step": 2300
+    },
+    {
+      "epoch": 0.032258931691712144,
+      "grad_norm": 0.9386630654335022,
+      "learning_rate": 0.0027999999999999345,
+      "loss": 4.4196,
+      "step": 2400
+    },
+    {
+      "epoch": 0.033603053845533484,
+      "grad_norm": 0.7635268568992615,
+      "learning_rate": 0.0028124999999999322,
+      "loss": 4.3982,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03494717599935482,
+      "grad_norm": 0.905320405960083,
+      "learning_rate": 0.00282499999999993,
+      "loss": 4.3657,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03629129815317616,
+      "grad_norm": 0.8763816952705383,
+      "learning_rate": 0.0028374999999999286,
+      "loss": 4.3476,
+      "step": 2700
+    },
+    {
+      "epoch": 0.0376354203069975,
+      "grad_norm": 1.0410003662109375,
+      "learning_rate": 0.002849999999999924,
+      "loss": 4.3162,
+      "step": 2800
+    },
+    {
+      "epoch": 0.03897954246081884,
+      "grad_norm": 0.8011109232902527,
+      "learning_rate": 0.0028624999999999224,
+      "loss": 4.2953,
+      "step": 2900
+    },
+    {
+      "epoch": 0.04032366461464018,
+      "grad_norm": 0.9530174136161804,
+      "learning_rate": 0.002874999999999916,
+      "loss": 4.2607,
+      "step": 3000
+    },
+    {
+      "epoch": 0.04032366461464018,
+      "eval_MaskedAccuracy": 0.24486576013786512,
+      "eval_loss": 4.528601169586182,
+      "eval_runtime": 159.1854,
+      "eval_samples_per_second": 398.755,
+      "eval_steps_per_second": 1.558,
+      "step": 3000
+    },
+    {
+      "epoch": 0.041667786768461515,
+      "grad_norm": 0.9223951101303101,
+      "learning_rate": 0.0028874999999999123,
+      "loss": 4.2462,
+      "step": 3100
+    },
+    {
+      "epoch": 0.043011908922282856,
+      "grad_norm": 0.76118403673172,
+      "learning_rate": 0.0028999999999999113,
+      "loss": 4.2295,
+      "step": 3200
+    },
+    {
+      "epoch": 0.044356031076104196,
+      "grad_norm": 1.0855016708374023,
+      "learning_rate": 0.002912499999999909,
+      "loss": 4.1973,
+      "step": 3300
+    },
+    {
+      "epoch": 0.04570015322992554,
+      "grad_norm": 1.12669038772583,
+      "learning_rate": 0.0029249999999999042,
+      "loss": 4.1689,
+      "step": 3400
+    },
+    {
+      "epoch": 0.04704427538374688,
+      "grad_norm": 0.8598240613937378,
+      "learning_rate": 0.002937499999999899,
+      "loss": 4.1503,
+      "step": 3500
+    },
+    {
+      "epoch": 0.04838839753756821,
+      "grad_norm": 0.9049589037895203,
+      "learning_rate": 0.002949999999999894,
+      "loss": 4.1266,
+      "step": 3600
+    },
+    {
+      "epoch": 0.04973251969138955,
+      "grad_norm": 0.9112903475761414,
+      "learning_rate": 0.002962499999999894,
+      "loss": 4.0996,
+      "step": 3700
+    },
+    {
+      "epoch": 0.05107664184521089,
+      "grad_norm": 0.8487406969070435,
+      "learning_rate": 0.002974999999999889,
+      "loss": 4.0884,
+      "step": 3800
+    },
+    {
+      "epoch": 0.052420763999032234,
+      "grad_norm": 0.9482388496398926,
+      "learning_rate": 0.002987499999999884,
+      "loss": 4.0599,
+      "step": 3900
+    },
+    {
+      "epoch": 0.05376488615285357,
+      "grad_norm": 1.0483149290084839,
+      "learning_rate": 0.002999999999999876,
+      "loss": 4.039,
+      "step": 4000
+    },
+    {
+      "epoch": 0.05376488615285357,
+      "eval_MaskedAccuracy": 0.2591465652834645,
+      "eval_loss": 4.372995376586914,
+      "eval_runtime": 156.5453,
+      "eval_samples_per_second": 405.48,
+      "eval_steps_per_second": 1.584,
+      "step": 4000
+    },
+    {
+      "epoch": 0.05510900830667491,
+      "grad_norm": 0.885144054889679,
+      "learning_rate": 0.003012499999999872,
+      "loss": 4.0297,
+      "step": 4100
+    },
+    {
+      "epoch": 0.05645313046049625,
+      "grad_norm": 1.0170485973358154,
+      "learning_rate": 0.0030249999999998685,
+      "loss": 4.0089,
+      "step": 4200
+    },
+    {
+      "epoch": 0.05779725261431759,
+      "grad_norm": 0.9548616409301758,
+      "learning_rate": 0.003037499999999864,
+      "loss": 3.9864,
+      "step": 4300
+    },
+    {
+      "epoch": 0.05914137476813893,
+      "grad_norm": 1.0004748106002808,
+      "learning_rate": 0.0030499999999998627,
+      "loss": 3.9803,
+      "step": 4400
+    },
+    {
+      "epoch": 0.060485496921960265,
+      "grad_norm": 0.8399825096130371,
+      "learning_rate": 0.003062499999999858,
+      "loss": 3.9609,
+      "step": 4500
+    },
+    {
+      "epoch": 0.061829619075781606,
+      "grad_norm": 0.9576703906059265,
+      "learning_rate": 0.0030749999999998565,
+      "loss": 3.9492,
+      "step": 4600
+    },
+    {
+      "epoch": 0.06317374122960295,
+      "grad_norm": 0.8874163031578064,
+      "learning_rate": 0.0030874999999998525,
+      "loss": 3.9307,
+      "step": 4700
+    },
+    {
+      "epoch": 0.06451786338342429,
+      "grad_norm": 0.8954266309738159,
+      "learning_rate": 0.003099999999999849,
+      "loss": 3.917,
+      "step": 4800
+    },
+    {
+      "epoch": 0.06586198553724562,
+      "grad_norm": 1.2326369285583496,
+      "learning_rate": 0.0031124999999998476,
+      "loss": 3.9057,
+      "step": 4900
+    },
+    {
+      "epoch": 0.06720610769106697,
+      "grad_norm": 0.9583144783973694,
+      "learning_rate": 0.003124999999999845,
+      "loss": 3.8926,
+      "step": 5000
+    },
+    {
+      "epoch": 0.06720610769106697,
+      "eval_MaskedAccuracy": 0.272405235858602,
+      "eval_loss": 4.243067741394043,
+      "eval_runtime": 160.7599,
+      "eval_samples_per_second": 394.85,
+      "eval_steps_per_second": 1.543,
+      "step": 5000
+    },
+    {
+      "epoch": 0.0685502298448883,
+      "grad_norm": 1.063835620880127,
+      "learning_rate": 0.003137499999999841,
+      "loss": 3.8825,
+      "step": 5100
+    },
+    {
+      "epoch": 0.06989435199870964,
+      "grad_norm": 0.8451062440872192,
+      "learning_rate": 0.003149999999999838,
+      "loss": 3.8725,
+      "step": 5200
+    },
+    {
+      "epoch": 0.07123847415253098,
+      "grad_norm": 0.8295220732688904,
+      "learning_rate": 0.003162499999999835,
+      "loss": 3.8622,
+      "step": 5300
+    },
+    {
+      "epoch": 0.07258259630635232,
+      "grad_norm": 0.8434109091758728,
+      "learning_rate": 0.003174999999999833,
+      "loss": 3.8418,
+      "step": 5400
+    },
+    {
+      "epoch": 0.07392671846017367,
+      "grad_norm": 0.9849230051040649,
+      "learning_rate": 0.0031874999999998307,
+      "loss": 3.8338,
+      "step": 5500
+    },
+    {
+      "epoch": 0.075270840613995,
+      "grad_norm": 1.2400304079055786,
+      "learning_rate": 0.003199999999999828,
+      "loss": 3.8234,
+      "step": 5600
+    },
+    {
+      "epoch": 0.07661496276781633,
+      "grad_norm": 0.8637450933456421,
+      "learning_rate": 0.0032124999999998288,
+      "loss": 3.8194,
+      "step": 5700
+    },
+    {
+      "epoch": 0.07795908492163768,
+      "grad_norm": 0.7561401128768921,
+      "learning_rate": 0.0032249999999998287,
+      "loss": 3.805,
+      "step": 5800
+    },
+    {
+      "epoch": 0.07930320707545901,
+      "grad_norm": 0.822270929813385,
+      "learning_rate": 0.003237499999999828,
+      "loss": 3.7957,
+      "step": 5900
+    },
+    {
+      "epoch": 0.08064732922928036,
+      "grad_norm": 0.9090524315834045,
+      "learning_rate": 0.0032499999999998216,
+      "loss": 3.7777,
+      "step": 6000
+    },
+    {
+      "epoch": 0.08064732922928036,
+      "eval_MaskedAccuracy": 0.2802241909468907,
+      "eval_loss": 4.147753715515137,
+      "eval_runtime": 154.4313,
+      "eval_samples_per_second": 411.031,
+      "eval_steps_per_second": 1.606,
+      "step": 6000
+    },
+    {
+      "epoch": 0.0819914513831017,
+      "grad_norm": 1.0061832666397095,
+      "learning_rate": 0.0032624999999998194,
+      "loss": 3.7827,
+      "step": 6100
+    },
+    {
+      "epoch": 0.08333557353692303,
+      "grad_norm": 1.0628759860992432,
+      "learning_rate": 0.0032749999999998154,
+      "loss": 3.7667,
+      "step": 6200
+    },
+    {
+      "epoch": 0.08467969569074438,
+      "grad_norm": 1.1841171979904175,
+      "learning_rate": 0.003287499999999813,
+      "loss": 3.7672,
+      "step": 6300
+    },
+    {
+      "epoch": 0.08602381784456571,
+      "grad_norm": 0.9175704121589661,
+      "learning_rate": 0.003299999999999811,
+      "loss": 3.7556,
+      "step": 6400
+    },
+    {
+      "epoch": 0.08736793999838706,
+      "grad_norm": 0.9184996485710144,
+      "learning_rate": 0.0033124999999998082,
+      "loss": 3.7411,
+      "step": 6500
+    },
+    {
+      "epoch": 0.08871206215220839,
+      "grad_norm": 0.8513078093528748,
+      "learning_rate": 0.0033249999999998042,
+      "loss": 3.7332,
+      "step": 6600
+    },
+    {
+      "epoch": 0.09005618430602973,
+      "grad_norm": 0.9670765399932861,
+      "learning_rate": 0.0033374999999997994,
+      "loss": 3.7274,
+      "step": 6700
+    },
+    {
+      "epoch": 0.09140030645985107,
+      "grad_norm": 0.9749046564102173,
+      "learning_rate": 0.0033499999999997932,
+      "loss": 3.7223,
+      "step": 6800
+    },
+    {
+      "epoch": 0.09274442861367241,
+      "grad_norm": 0.8485523462295532,
+      "learning_rate": 0.003362499999999787,
+      "loss": 3.7125,
+      "step": 6900
+    },
+    {
+      "epoch": 0.09408855076749376,
+      "grad_norm": 0.84412682056427,
+      "learning_rate": 0.0033749999999997853,
+      "loss": 3.71,
+      "step": 7000
+    },
+    {
+      "epoch": 0.09408855076749376,
+      "eval_MaskedAccuracy": 0.2868874563852213,
+      "eval_loss": 4.067524433135986,
+      "eval_runtime": 154.2299,
+      "eval_samples_per_second": 411.567,
+      "eval_steps_per_second": 1.608,
+      "step": 7000
+    },
+    {
+      "epoch": 0.09543267292131509,
+      "grad_norm": 0.8195430040359497,
+      "learning_rate": 0.0033874999999997804,
+      "loss": 3.708,
+      "step": 7100
+    },
+    {
+      "epoch": 0.09677679507513642,
+      "grad_norm": 0.8684709668159485,
+      "learning_rate": 0.0033999999999997713,
+      "loss": 3.7003,
+      "step": 7200
+    },
+    {
+      "epoch": 0.09812091722895777,
+      "grad_norm": 0.8237408399581909,
+      "learning_rate": 0.0034124999999997647,
+      "loss": 3.676,
+      "step": 7300
+    },
+    {
+      "epoch": 0.0994650393827791,
+      "grad_norm": 0.8418594598770142,
+      "learning_rate": 0.003424999999999764,
+      "loss": 3.6748,
+      "step": 7400
+    },
+    {
+      "epoch": 0.10080916153660045,
+      "grad_norm": 0.900364100933075,
+      "learning_rate": 0.0034374999999997606,
+      "loss": 3.6822,
+      "step": 7500
+    },
+    {
+      "epoch": 0.10215328369042179,
+      "grad_norm": 0.9300949573516846,
+      "learning_rate": 0.003449999999999758,
+      "loss": 3.6657,
+      "step": 7600
+    },
+    {
+      "epoch": 0.10349740584424312,
+      "grad_norm": 0.8330152034759521,
+      "learning_rate": 0.003462499999999754,
+      "loss": 3.6635,
+      "step": 7700
+    },
+    {
+      "epoch": 0.10484152799806447,
+      "grad_norm": 0.8555970191955566,
+      "learning_rate": 0.003474999999999749,
+      "loss": 3.6461,
+      "step": 7800
+    },
+    {
+      "epoch": 0.1061856501518858,
+      "grad_norm": 0.9717864394187927,
+      "learning_rate": 0.003487499999999749,
+      "loss": 3.6402,
+      "step": 7900
+    },
+    {
+      "epoch": 0.10752977230570714,
+      "grad_norm": 1.3664884567260742,
+      "learning_rate": 0.003499999999999745,
+      "loss": 3.6341,
+      "step": 8000
+    },
+    {
+      "epoch": 0.10752977230570714,
+      "eval_MaskedAccuracy": 0.2957456999836483,
+      "eval_loss": 3.9853811264038086,
+      "eval_runtime": 154.2803,
+      "eval_samples_per_second": 411.433,
+      "eval_steps_per_second": 1.607,
+      "step": 8000
+    },
+    {
+      "epoch": 0.10887389445952848,
+      "grad_norm": 1.9192525148391724,
+      "learning_rate": 0.0035124999999997415,
+      "loss": 3.6234,
+      "step": 8100
+    },
+    {
+      "epoch": 0.11021801661334982,
+      "grad_norm": 0.875035285949707,
+      "learning_rate": 0.003524999999999741,
+      "loss": 3.6126,
+      "step": 8200
+    },
+    {
+      "epoch": 0.11156213876717117,
+      "grad_norm": 1.4383764266967773,
+      "learning_rate": 0.0035374999999997353,
+      "loss": 3.5893,
+      "step": 8300
+    },
+    {
+      "epoch": 0.1129062609209925,
+      "grad_norm": 1.0789637565612793,
+      "learning_rate": 0.0035499999999997313,
+      "loss": 3.5687,
+      "step": 8400
+    },
+    {
+      "epoch": 0.11425038307481383,
+      "grad_norm": 1.7937785387039185,
+      "learning_rate": 0.00356249999999973,
+      "loss": 3.5473,
+      "step": 8500
+    },
+    {
+      "epoch": 0.11559450522863518,
+      "grad_norm": 1.0099656581878662,
+      "learning_rate": 0.0035749999999997255,
+      "loss": 3.5311,
+      "step": 8600
+    },
+    {
+      "epoch": 0.11693862738245651,
+      "grad_norm": 0.876243531703949,
+      "learning_rate": 0.0035874999999997237,
+      "loss": 3.5129,
+      "step": 8700
+    },
+    {
+      "epoch": 0.11828274953627786,
+      "grad_norm": 1.4001803398132324,
+      "learning_rate": 0.0035999999999997236,
+      "loss": 3.5052,
+      "step": 8800
+    },
+    {
+      "epoch": 0.1196268716900992,
+      "grad_norm": 1.103613018989563,
+      "learning_rate": 0.003612499999999721,
+      "loss": 3.4943,
+      "step": 8900
+    },
+    {
+      "epoch": 0.12097099384392053,
+      "grad_norm": 0.8662921190261841,
+      "learning_rate": 0.00362499999999972,
+      "loss": 3.4864,
+      "step": 9000
+    },
+    {
+      "epoch": 0.12097099384392053,
+      "eval_MaskedAccuracy": 0.308084678333672,
+      "eval_loss": 3.829771041870117,
+      "eval_runtime": 161.1285,
+      "eval_samples_per_second": 393.946,
+      "eval_steps_per_second": 1.539,
+      "step": 9000
+    },
+    {
+      "epoch": 0.12231511599774188,
+      "grad_norm": 0.9641631841659546,
+      "learning_rate": 0.0036374999999997113,
+      "loss": 3.4636,
+      "step": 9100
+    },
+    {
+      "epoch": 0.12365923815156321,
+      "grad_norm": 0.8773326277732849,
+      "learning_rate": 0.0036499999999997073,
+      "loss": 3.4694,
+      "step": 9200
+    },
+    {
+      "epoch": 0.12500336030538456,
+      "grad_norm": 1.132733941078186,
+      "learning_rate": 0.003662499999999711,
+      "loss": 3.4452,
+      "step": 9300
+    },
+    {
+      "epoch": 0.1263474824592059,
+      "grad_norm": 1.1771531105041504,
+      "learning_rate": 0.0036749999999997097,
+      "loss": 3.4436,
+      "step": 9400
+    },
+    {
+      "epoch": 0.12769160461302723,
+      "grad_norm": 0.7729461193084717,
+      "learning_rate": 0.0036874999999997097,
+      "loss": 3.4241,
+      "step": 9500
+    },
+    {
+      "epoch": 0.12903572676684857,
+      "grad_norm": 1.1024218797683716,
+      "learning_rate": 0.0036999999999997118,
+      "loss": 3.4175,
+      "step": 9600
+    },
+    {
+      "epoch": 0.13037984892066992,
+      "grad_norm": 0.7504922747612,
+      "learning_rate": 0.0037124999999997074,
+      "loss": 3.3946,
+      "step": 9700
+    },
+    {
+      "epoch": 0.13172397107449124,
+      "grad_norm": 0.7903368473052979,
+      "learning_rate": 0.0037249999999997034,
+      "loss": 3.3848,
+      "step": 9800
+    },
+    {
+      "epoch": 0.1330680932283126,
+      "grad_norm": 0.8163586258888245,
+      "learning_rate": 0.0037374999999996972,
+      "loss": 3.3669,
+      "step": 9900
+    },
+    {
+      "epoch": 0.13441221538213394,
+      "grad_norm": 0.839239776134491,
+      "learning_rate": 0.003749999999999697,
+      "loss": 3.3447,
+      "step": 10000
+    },
+    {
+      "epoch": 0.13441221538213394,
+      "eval_MaskedAccuracy": 0.32656299271612027,
+      "eval_loss": 3.7040650844573975,
+      "eval_runtime": 161.5088,
+      "eval_samples_per_second": 393.019,
+      "eval_steps_per_second": 1.536,
+      "step": 10000
+    },
+    {
+      "epoch": 0.13575633753595526,
+      "grad_norm": 1.061814546585083,
+      "learning_rate": 0.003762499999999692,
+      "loss": 3.3364,
+      "step": 10100
+    },
+    {
+      "epoch": 0.1371004596897766,
+      "grad_norm": 0.7809873223304749,
+      "learning_rate": 0.0037749999999996914,
+      "loss": 3.3223,
+      "step": 10200
+    },
+    {
+      "epoch": 0.13844458184359795,
+      "grad_norm": 0.8472948670387268,
+      "learning_rate": 0.003787499999999695,
+      "loss": 3.3116,
+      "step": 10300
+    },
+    {
+      "epoch": 0.13978870399741927,
+      "grad_norm": 0.7124361991882324,
+      "learning_rate": 0.0037999999999996873,
+      "loss": 3.2889,
+      "step": 10400
+    },
+    {
+      "epoch": 0.14113282615124062,
+      "grad_norm": 1.7072255611419678,
+      "learning_rate": 0.003812499999999687,
+      "loss": 3.2982,
+      "step": 10500
+    },
+    {
+      "epoch": 0.14247694830506197,
+      "grad_norm": 2.472754716873169,
+      "learning_rate": 0.0038249999999996833,
+      "loss": 3.2711,
+      "step": 10600
+    },
+    {
+      "epoch": 0.14382107045888332,
+      "grad_norm": 0.9063215255737305,
+      "learning_rate": 0.0038374999999996823,
+      "loss": 3.2667,
+      "step": 10700
+    },
+    {
+      "epoch": 0.14516519261270464,
+      "grad_norm": 0.8283532857894897,
+      "learning_rate": 0.0038499999999996775,
+      "loss": 3.2545,
+      "step": 10800
+    },
+    {
+      "epoch": 0.14650931476652598,
+      "grad_norm": 0.8591430187225342,
+      "learning_rate": 0.0038624999999996735,
+      "loss": 3.248,
+      "step": 10900
+    },
+    {
+      "epoch": 0.14785343692034733,
+      "grad_norm": 0.971303403377533,
+      "learning_rate": 0.0038749999999996712,
+      "loss": 3.2447,
+      "step": 11000
+    },
+    {
+      "epoch": 0.14785343692034733,
+      "eval_MaskedAccuracy": 0.34290362452711665,
+      "eval_loss": 3.601362705230713,
+      "eval_runtime": 162.0306,
+      "eval_samples_per_second": 391.753,
+      "eval_steps_per_second": 1.531,
+      "step": 11000
+    },
+    {
+      "epoch": 0.14919755907416865,
+      "grad_norm": 1.0778330564498901,
+      "learning_rate": 0.0038874999999996686,
+      "loss": 3.2298,
+      "step": 11100
+    },
+    {
+      "epoch": 0.15054168122799,
+      "grad_norm": 0.7707659602165222,
+      "learning_rate": 0.003899999999999665,
+      "loss": 3.2199,
+      "step": 11200
+    },
+    {
+      "epoch": 0.15188580338181135,
+      "grad_norm": 1.2470498085021973,
+      "learning_rate": 0.003912499999999662,
+      "loss": 3.2137,
+      "step": 11300
+    },
+    {
+      "epoch": 0.15322992553563267,
+      "grad_norm": 0.8463084101676941,
+      "learning_rate": 0.003924999999999659,
+      "loss": 3.2123,
+      "step": 11400
+    },
+    {
+      "epoch": 0.154574047689454,
+      "grad_norm": 0.8666761517524719,
+      "learning_rate": 0.0039374999999996566,
+      "loss": 3.1978,
+      "step": 11500
+    },
+    {
+      "epoch": 0.15591816984327536,
+      "grad_norm": 0.899533748626709,
+      "learning_rate": 0.0039499999999996595,
+      "loss": 3.1932,
+      "step": 11600
+    },
+    {
+      "epoch": 0.1572622919970967,
+      "grad_norm": 0.8859838843345642,
+      "learning_rate": 0.00396249999999966,
+      "loss": 3.1799,
+      "step": 11700
+    },
+    {
+      "epoch": 0.15860641415091803,
+      "grad_norm": 1.3646000623703003,
+      "learning_rate": 0.003974999999999655,
+      "loss": 3.1739,
+      "step": 11800
+    },
+    {
+      "epoch": 0.15995053630473938,
+      "grad_norm": 0.7711676359176636,
+      "learning_rate": 0.003987499999999656,
+      "loss": 3.1746,
+      "step": 11900
+    },
+    {
+      "epoch": 0.16129465845856072,
+      "grad_norm": 0.9850155711174011,
+      "learning_rate": 0.003999999999999657,
+      "loss": 3.1749,
+      "step": 12000
+    },
+    {
+      "epoch": 0.16129465845856072,
+      "eval_MaskedAccuracy": 0.3521379038689249,
+      "eval_loss": 3.5329132080078125,
+      "eval_runtime": 154.2202,
+      "eval_samples_per_second": 411.593,
+      "eval_steps_per_second": 1.608,
+      "step": 12000
+    },
+    {
+      "epoch": 0.16263878061238204,
+      "grad_norm": 2.1732335090637207,
+      "learning_rate": 0.004012499999999649,
+      "loss": 3.1557,
+      "step": 12100
+    },
+    {
+      "epoch": 0.1639829027662034,
+      "grad_norm": 0.6864789724349976,
+      "learning_rate": 0.00402499999999965,
+      "loss": 3.1457,
+      "step": 12200
+    },
+    {
+      "epoch": 0.16532702492002474,
+      "grad_norm": 0.7963569760322571,
+      "learning_rate": 0.004037499999999652,
+      "loss": 3.1462,
+      "step": 12300
+    },
+    {
+      "epoch": 0.16667114707384606,
+      "grad_norm": 0.7270441651344299,
+      "learning_rate": 0.00404999999999965,
+      "loss": 3.1424,
+      "step": 12400
+    },
+    {
+      "epoch": 0.1680152692276674,
+      "grad_norm": 1.0645862817764282,
+      "learning_rate": 0.004062499999999646,
+      "loss": 3.1301,
+      "step": 12500
+    },
+    {
+      "epoch": 0.16935939138148876,
+      "grad_norm": 0.7641903758049011,
+      "learning_rate": 0.004074999999999641,
+      "loss": 3.1309,
+      "step": 12600
+    },
+    {
+      "epoch": 0.17070351353531008,
+      "grad_norm": 1.2100406885147095,
+      "learning_rate": 0.004087499999999641,
+      "loss": 3.1348,
+      "step": 12700
+    },
+    {
+      "epoch": 0.17204763568913142,
+      "grad_norm": 0.7301545143127441,
+      "learning_rate": 0.004099999999999645,
+      "loss": 3.1175,
+      "step": 12800
+    },
+    {
+      "epoch": 0.17339175784295277,
+      "grad_norm": 1.3868557214736938,
+      "learning_rate": 0.004112499999999646,
+      "loss": 3.1196,
+      "step": 12900
+    },
+    {
+      "epoch": 0.17473587999677412,
+      "grad_norm": 1.3212717771530151,
+      "learning_rate": 0.004124999999999645,
+      "loss": 3.1098,
+      "step": 13000
+    },
+    {
+      "epoch": 0.17473587999677412,
+      "eval_MaskedAccuracy": 0.35983040074292133,
+      "eval_loss": 3.4708173274993896,
+      "eval_runtime": 161.2353,
+      "eval_samples_per_second": 393.685,
+      "eval_steps_per_second": 1.538,
+      "step": 13000
+    },
+    {
+      "epoch": 0.17608000215059544,
+      "grad_norm": 0.6421147584915161,
+      "learning_rate": 0.0041374999999996415,
+      "loss": 3.1074,
+      "step": 13100
+    },
+    {
+      "epoch": 0.17742412430441679,
+      "grad_norm": 1.0816092491149902,
+      "learning_rate": 0.004149999999999641,
+      "loss": 3.0902,
+      "step": 13200
+    },
+    {
+      "epoch": 0.17876824645823813,
+      "grad_norm": 0.8088307976722717,
+      "learning_rate": 0.004162499999999636,
+      "loss": 3.0973,
+      "step": 13300
+    },
+    {
+      "epoch": 0.18011236861205945,
+      "grad_norm": 0.9084308743476868,
+      "learning_rate": 0.004174999999999633,
+      "loss": 3.0835,
+      "step": 13400
+    },
+    {
+      "epoch": 0.1814564907658808,
+      "grad_norm": 1.0027776956558228,
+      "learning_rate": 0.004187499999999628,
+      "loss": 3.0876,
+      "step": 13500
+    },
+    {
+      "epoch": 0.18280061291970215,
+      "grad_norm": 1.0761163234710693,
+      "learning_rate": 0.004199999999999622,
+      "loss": 3.0867,
+      "step": 13600
+    },
+    {
+      "epoch": 0.18414473507352347,
+      "grad_norm": 0.7274723052978516,
+      "learning_rate": 0.004212499999999617,
+      "loss": 3.0793,
+      "step": 13700
+    },
+    {
+      "epoch": 0.18548885722734482,
+      "grad_norm": 0.9493256211280823,
+      "learning_rate": 0.004224999999999617,
+      "loss": 3.0753,
+      "step": 13800
+    },
+    {
+      "epoch": 0.18683297938116616,
+      "grad_norm": 1.288206696510315,
+      "learning_rate": 0.004237499999999612,
+      "loss": 3.0625,
+      "step": 13900
+    },
+    {
+      "epoch": 0.1881771015349875,
+      "grad_norm": 0.8344607353210449,
+      "learning_rate": 0.004249999999999605,
+      "loss": 3.0665,
+      "step": 14000
+    },
+    {
+      "epoch": 0.1881771015349875,
+      "eval_MaskedAccuracy": 0.36507595159603434,
+      "eval_loss": 3.4270057678222656,
+      "eval_runtime": 161.5547,
+      "eval_samples_per_second": 392.907,
+      "eval_steps_per_second": 1.535,
+      "step": 14000
+    },
+    {
+      "epoch": 0.18952122368880883,
+      "grad_norm": 0.7024100422859192,
+      "learning_rate": 0.004262499999999602,
+      "loss": 3.0688,
+      "step": 14100
+    },
+    {
+      "epoch": 0.19086534584263018,
+      "grad_norm": 0.7407594323158264,
+      "learning_rate": 0.004274999999999597,
+      "loss": 3.0539,
+      "step": 14200
+    },
+    {
+      "epoch": 0.19220946799645153,
+      "grad_norm": 0.8240285515785217,
+      "learning_rate": 0.004287499999999591,
+      "loss": 3.0552,
+      "step": 14300
+    },
+    {
+      "epoch": 0.19355359015027285,
+      "grad_norm": 0.7373182773590088,
+      "learning_rate": 0.004299999999999582,
+      "loss": 3.0529,
+      "step": 14400
+    },
+    {
+      "epoch": 0.1948977123040942,
+      "grad_norm": 0.6955730319023132,
+      "learning_rate": 0.004312499999999577,
+      "loss": 3.0478,
+      "step": 14500
+    },
+    {
+      "epoch": 0.19624183445791554,
+      "grad_norm": 1.1683614253997803,
+      "learning_rate": 0.004324999999999574,
+      "loss": 3.0314,
+      "step": 14600
+    },
+    {
+      "epoch": 0.19758595661173686,
+      "grad_norm": 1.4355748891830444,
+      "learning_rate": 0.004337499999999567,
+      "loss": 3.0478,
+      "step": 14700
+    },
+    {
+      "epoch": 0.1989300787655582,
+      "grad_norm": 0.7114657163619995,
+      "learning_rate": 0.004349999999999561,
+      "loss": 3.0324,
+      "step": 14800
+    },
+    {
+      "epoch": 0.20027420091937956,
+      "grad_norm": 1.0332584381103516,
+      "learning_rate": 0.004362499999999559,
+      "loss": 3.0353,
+      "step": 14900
+    },
+    {
+      "epoch": 0.2016183230732009,
+      "grad_norm": 0.6959288716316223,
+      "learning_rate": 0.0043749999999995555,
+      "loss": 3.0278,
+      "step": 15000
+    },
+    {
+      "epoch": 0.2016183230732009,
+      "eval_MaskedAccuracy": 0.37069473714124845,
+      "eval_loss": 3.379894256591797,
+      "eval_runtime": 161.4047,
+      "eval_samples_per_second": 393.272,
+      "eval_steps_per_second": 1.537,
+      "step": 15000
+    },
+    {
+      "epoch": 0.20296244522702223,
+      "grad_norm": 1.1070080995559692,
+      "learning_rate": 0.004387499999999551,
+      "loss": 3.0205,
+      "step": 15100
+    },
+    {
+      "epoch": 0.20430656738084357,
+      "grad_norm": 0.8838359117507935,
+      "learning_rate": 0.004399999999999546,
+      "loss": 3.0299,
+      "step": 15200
+    },
+    {
+      "epoch": 0.20565068953466492,
+      "grad_norm": 0.8130790591239929,
+      "learning_rate": 0.004412499999999543,
+      "loss": 3.0231,
+      "step": 15300
+    },
+    {
+      "epoch": 0.20699481168848624,
+      "grad_norm": 0.6710375547409058,
+      "learning_rate": 0.004424999999999538,
+      "loss": 3.0161,
+      "step": 15400
+    },
+    {
+      "epoch": 0.2083389338423076,
+      "grad_norm": 0.7571612000465393,
+      "learning_rate": 0.004437499999999531,
+      "loss": 3.0166,
+      "step": 15500
+    },
+    {
+      "epoch": 0.20968305599612894,
+      "grad_norm": 0.658545196056366,
+      "learning_rate": 0.004449999999999525,
+      "loss": 3.0011,
+      "step": 15600
+    },
+    {
+      "epoch": 0.21102717814995026,
+      "grad_norm": 0.7161542177200317,
+      "learning_rate": 0.004462499999999522,
+      "loss": 3.0085,
+      "step": 15700
+    },
+    {
+      "epoch": 0.2123713003037716,
+      "grad_norm": 0.7213286757469177,
+      "learning_rate": 0.004474999999999518,
+      "loss": 3.0037,
+      "step": 15800
+    },
+    {
+      "epoch": 0.21371542245759295,
+      "grad_norm": 1.3048126697540283,
+      "learning_rate": 0.004487499999999511,
+      "loss": 2.9974,
+      "step": 15900
+    },
+    {
+      "epoch": 0.21505954461141427,
+      "grad_norm": 0.999495804309845,
+      "learning_rate": 0.00449999999999951,
+      "loss": 2.9973,
+      "step": 16000
+    },
+    {
+      "epoch": 0.21505954461141427,
+      "eval_MaskedAccuracy": 0.3738960824804252,
+      "eval_loss": 3.3454906940460205,
+      "eval_runtime": 157.1677,
+      "eval_samples_per_second": 403.874,
+      "eval_steps_per_second": 1.578,
+      "step": 16000
+    },
+    {
+      "epoch": 0.21640366676523562,
+      "grad_norm": 2.9935622215270996,
+      "learning_rate": 0.004512499999999503,
+      "loss": 3.0009,
+      "step": 16100
+    },
+    {
+      "epoch": 0.21774778891905697,
+      "grad_norm": 1.100763201713562,
+      "learning_rate": 0.004524999999999504,
+      "loss": 2.9812,
+      "step": 16200
+    },
+    {
+      "epoch": 0.21909191107287831,
+      "grad_norm": 0.7473815083503723,
+      "learning_rate": 0.004537499999999495,
+      "loss": 2.9833,
+      "step": 16300
+    },
+    {
+      "epoch": 0.22043603322669963,
+      "grad_norm": 0.8163867592811584,
+      "learning_rate": 0.004549999999999481,
+      "loss": 2.9867,
+      "step": 16400
+    },
+    {
+      "epoch": 0.22178015538052098,
+      "grad_norm": 4.292899131774902,
+      "learning_rate": 0.0045624999999994715,
+      "loss": 2.982,
+      "step": 16500
+    },
+    {
+      "epoch": 0.22312427753434233,
+      "grad_norm": 9.086414337158203,
+      "learning_rate": 0.0045749999999994675,
+      "loss": 2.9867,
+      "step": 16600
+    },
+    {
+      "epoch": 0.22446839968816365,
+      "grad_norm": 0.6632740497589111,
+      "learning_rate": 0.004587499999999466,
+      "loss": 2.972,
+      "step": 16700
+    },
+    {
+      "epoch": 0.225812521841985,
+      "grad_norm": 3.345919132232666,
+      "learning_rate": 0.004599999999999462,
+      "loss": 2.972,
+      "step": 16800
+    },
+    {
+      "epoch": 0.22715664399580635,
+      "grad_norm": 0.6550186276435852,
+      "learning_rate": 0.0046124999999994564,
+      "loss": 2.9672,
+      "step": 16900
+    },
+    {
+      "epoch": 0.22850076614962767,
+      "grad_norm": 0.6624706983566284,
+      "learning_rate": 0.004624999999999449,
+      "loss": 2.9731,
+      "step": 17000
+    },
+    {
+      "epoch": 0.22850076614962767,
+      "eval_MaskedAccuracy": 0.3775072955122549,
+      "eval_loss": 3.3128609657287598,
+      "eval_runtime": 154.9603,
+      "eval_samples_per_second": 409.628,
+      "eval_steps_per_second": 1.6,
+      "step": 17000
+    },
+    {
+      "epoch": 0.229844888303449,
+      "grad_norm": 1.0394245386123657,
+      "learning_rate": 0.004637499999999442,
+      "loss": 2.9749,
+      "step": 17100
+    },
+    {
+      "epoch": 0.23118901045727036,
+      "grad_norm": 0.6592782139778137,
+      "learning_rate": 0.00464999999999944,
+      "loss": 2.967,
+      "step": 17200
+    },
+    {
+      "epoch": 0.2325331326110917,
+      "grad_norm": 0.923537015914917,
+      "learning_rate": 0.00466249999999944,
+      "loss": 2.9607,
+      "step": 17300
+    },
+    {
+      "epoch": 0.23387725476491303,
+      "grad_norm": 0.878057599067688,
+      "learning_rate": 0.004674999999999435,
+      "loss": 2.9609,
+      "step": 17400
+    },
+    {
+      "epoch": 0.23522137691873438,
+      "grad_norm": 1.1436913013458252,
+      "learning_rate": 0.004687499999999431,
+      "loss": 2.9521,
+      "step": 17500
+    },
+    {
+      "epoch": 0.23656549907255572,
+      "grad_norm": 0.8196259140968323,
+      "learning_rate": 0.004699999999999424,
+      "loss": 2.9523,
+      "step": 17600
+    },
+    {
+      "epoch": 0.23790962122637704,
+      "grad_norm": 0.625901460647583,
+      "learning_rate": 0.004712499999999414,
+      "loss": 2.955,
+      "step": 17700
+    },
+    {
+      "epoch": 0.2392537433801984,
+      "grad_norm": 0.7023401856422424,
+      "learning_rate": 0.004724999999999409,
+      "loss": 2.949,
+      "step": 17800
+    },
+    {
+      "epoch": 0.24059786553401974,
+      "grad_norm": 0.7306541204452515,
+      "learning_rate": 0.0047374999999993984,
+      "loss": 2.9478,
+      "step": 17900
+    },
+    {
+      "epoch": 0.24194198768784106,
+      "grad_norm": 0.6469770073890686,
+      "learning_rate": 0.004749999999999393,
+      "loss": 2.9395,
+      "step": 18000
+    },
+    {
+      "epoch": 0.24194198768784106,
+      "eval_MaskedAccuracy": 0.3805818810631038,
+      "eval_loss": 3.2859790325164795,
+      "eval_runtime": 159.4762,
+      "eval_samples_per_second": 398.028,
+      "eval_steps_per_second": 1.555,
+      "step": 18000
+    },
+    {
+      "epoch": 0.2432861098416624,
+      "grad_norm": 0.6748953461647034,
+      "learning_rate": 0.004762499999999384,
+      "loss": 2.9299,
+      "step": 18100
+    },
+    {
+      "epoch": 0.24463023199548375,
+      "grad_norm": 1.1163654327392578,
+      "learning_rate": 0.004774999999999384,
+      "loss": 2.9415,
+      "step": 18200
+    },
+    {
+      "epoch": 0.2459743541493051,
+      "grad_norm": 1.0123517513275146,
+      "learning_rate": 0.00478749999999938,
+      "loss": 2.9301,
+      "step": 18300
+    },
+    {
+      "epoch": 0.24731847630312642,
+      "grad_norm": 3.0257534980773926,
+      "learning_rate": 0.004799999999999372,
+      "loss": 2.9342,
+      "step": 18400
+    },
+    {
+      "epoch": 0.24866259845694777,
+      "grad_norm": 1.3270440101623535,
+      "learning_rate": 0.004812499999999362,
+      "loss": 2.9325,
+      "step": 18500
+    },
+    {
+      "epoch": 0.2500067206107691,
+      "grad_norm": 1.0727241039276123,
+      "learning_rate": 0.004824999999999358,
+      "loss": 2.9221,
+      "step": 18600
+    },
+    {
+      "epoch": 0.25135084276459047,
+      "grad_norm": 0.7316584587097168,
+      "learning_rate": 0.004837499999999354,
+      "loss": 2.9235,
+      "step": 18700
+    },
+    {
+      "epoch": 0.2526949649184118,
+      "grad_norm": 2.4931344985961914,
+      "learning_rate": 0.00484999999999935,
+      "loss": 2.9264,
+      "step": 18800
+    },
+    {
+      "epoch": 0.2540390870722331,
+      "grad_norm": 2.006314277648926,
+      "learning_rate": 0.004862499999999345,
+      "loss": 2.9229,
+      "step": 18900
+    },
+    {
+      "epoch": 0.25538320922605445,
+      "grad_norm": 1.1321533918380737,
+      "learning_rate": 0.004874999999999335,
+      "loss": 2.917,
+      "step": 19000
+    },
+    {
+      "epoch": 0.25538320922605445,
+      "eval_MaskedAccuracy": 0.38268668848805826,
+      "eval_loss": 3.2650303840637207,
+      "eval_runtime": 154.9489,
+      "eval_samples_per_second": 409.658,
+      "eval_steps_per_second": 1.601,
+      "step": 19000
+    },
+    {
+      "epoch": 0.2567273313798758,
+      "grad_norm": 0.637344241142273,
+      "learning_rate": 0.004887499999999324,
+      "loss": 2.9099,
+      "step": 19100
+    },
+    {
+      "epoch": 0.25807145353369715,
+      "grad_norm": 3.686405897140503,
+      "learning_rate": 0.004899999999999312,
+      "loss": 2.9115,
+      "step": 19200
+    },
+    {
+      "epoch": 0.2594155756875185,
+      "grad_norm": 0.734024703502655,
+      "learning_rate": 0.0049124999999993054,
+      "loss": 2.9133,
+      "step": 19300
+    },
+    {
+      "epoch": 0.26075969784133984,
+      "grad_norm": 0.9141765236854553,
+      "learning_rate": 0.004924999999999296,
+      "loss": 2.9188,
+      "step": 19400
+    },
+    {
+      "epoch": 0.26210381999516114,
+      "grad_norm": 0.9035413265228271,
+      "learning_rate": 0.004937499999999295,
+      "loss": 2.9223,
+      "step": 19500
+    },
+    {
+      "epoch": 0.2634479421489825,
+      "grad_norm": 0.9160522222518921,
+      "learning_rate": 0.004949999999999293,
+      "loss": 2.9057,
+      "step": 19600
+    },
+    {
+      "epoch": 0.26479206430280383,
+      "grad_norm": 0.6431342363357544,
+      "learning_rate": 0.004962499999999282,
+      "loss": 2.8985,
+      "step": 19700
+    },
+    {
+      "epoch": 0.2661361864566252,
+      "grad_norm": 0.6499119997024536,
+      "learning_rate": 0.004974999999999273,
+      "loss": 2.8946,
+      "step": 19800
+    },
+    {
+      "epoch": 0.2674803086104465,
+      "grad_norm": 0.7949116826057434,
+      "learning_rate": 0.004987499999999261,
+      "loss": 2.9091,
+      "step": 19900
+    },
+    {
+      "epoch": 0.2688244307642679,
+      "grad_norm": 2.123286008834839,
+      "learning_rate": 0.005,
+      "loss": 2.8978,
+      "step": 20000
+    },
+    {
+      "epoch": 0.2688244307642679,
+      "eval_MaskedAccuracy": 0.3850105547815323,
+      "eval_loss": 3.2462847232818604,
+      "eval_runtime": 158.4819,
+      "eval_samples_per_second": 400.525,
+      "eval_steps_per_second": 1.565,
+      "step": 20000
+    },
+    {
+      "epoch": 0.2701685529180892,
+      "grad_norm": 0.6592645645141602,
+      "learning_rate": 0.004999999943789577,
+      "loss": 2.896,
+      "step": 20100
+    },
+    {
+      "epoch": 0.2715126750719105,
+      "grad_norm": 1.3373818397521973,
+      "learning_rate": 0.0049999997751583045,
+      "loss": 2.8877,
+      "step": 20200
+    },
+    {
+      "epoch": 0.27285679722573186,
+      "grad_norm": 0.5964152812957764,
+      "learning_rate": 0.004999999494106201,
+      "loss": 2.8969,
+      "step": 20300
+    },
+    {
+      "epoch": 0.2742009193795532,
+      "grad_norm": 1.4561213254928589,
+      "learning_rate": 0.004999999100633266,
+      "loss": 2.8942,
+      "step": 20400
+    },
+    {
+      "epoch": 0.27554504153337456,
+      "grad_norm": 1.0474812984466553,
+      "learning_rate": 0.004999998594739517,
+      "loss": 2.9026,
+      "step": 20500
+    },
+    {
+      "epoch": 0.2768891636871959,
+      "grad_norm": 0.5717761516571045,
+      "learning_rate": 0.004999997976424982,
+      "loss": 2.8946,
+      "step": 20600
+    },
+    {
+      "epoch": 0.27823328584101725,
+      "grad_norm": 1.6426117420196533,
+      "learning_rate": 0.004999997245689687,
+      "loss": 2.881,
+      "step": 20700
+    },
+    {
+      "epoch": 0.27957740799483854,
+      "grad_norm": 2.5096325874328613,
+      "learning_rate": 0.004999996402533671,
+      "loss": 2.8809,
+      "step": 20800
+    },
+    {
+      "epoch": 0.2809215301486599,
+      "grad_norm": 1.026366949081421,
+      "learning_rate": 0.0049999954469569655,
+      "loss": 2.8773,
+      "step": 20900
+    },
+    {
+      "epoch": 0.28226565230248124,
+      "grad_norm": 1.2996127605438232,
+      "learning_rate": 0.0049999943789596135,
+      "loss": 2.871,
+      "step": 21000
+    },
+    {
+      "epoch": 0.28226565230248124,
+      "eval_MaskedAccuracy": 0.3880521470318861,
+      "eval_loss": 3.224468469619751,
+      "eval_runtime": 161.317,
+      "eval_samples_per_second": 393.486,
+      "eval_steps_per_second": 1.537,
+      "step": 21000
+    },
+    {
+      "epoch": 0.2836097744563026,
+      "grad_norm": 0.6892908215522766,
+      "learning_rate": 0.004999993198541667,
+      "loss": 2.8798,
+      "step": 21100
+    },
+    {
+      "epoch": 0.28495389661012394,
+      "grad_norm": 0.8045306205749512,
+      "learning_rate": 0.00499999190570318,
+      "loss": 2.8639,
+      "step": 21200
+    },
+    {
+      "epoch": 0.2862980187639453,
+      "grad_norm": 1.282244324684143,
+      "learning_rate": 0.004999990500444207,
+      "loss": 2.8754,
+      "step": 21300
+    },
+    {
+      "epoch": 0.28764214091776663,
+      "grad_norm": 0.9008962512016296,
+      "learning_rate": 0.00499998898276481,
+      "loss": 2.8698,
+      "step": 21400
+    },
+    {
+      "epoch": 0.2889862630715879,
+      "grad_norm": 0.8007209897041321,
+      "learning_rate": 0.00499998735266506,
+      "loss": 2.8647,
+      "step": 21500
+    },
+    {
+      "epoch": 0.29033038522540927,
+      "grad_norm": 1.014004111289978,
+      "learning_rate": 0.004999985610145036,
+      "loss": 2.8576,
+      "step": 21600
+    },
+    {
+      "epoch": 0.2916745073792306,
+      "grad_norm": 0.891608476638794,
+      "learning_rate": 0.0049999837552048105,
+      "loss": 2.8551,
+      "step": 21700
+    },
+    {
+      "epoch": 0.29301862953305197,
+      "grad_norm": 1.4745222330093384,
+      "learning_rate": 0.004999981787844464,
+      "loss": 2.8651,
+      "step": 21800
+    },
+    {
+      "epoch": 0.2943627516868733,
+      "grad_norm": 1.727281928062439,
+      "learning_rate": 0.004999979708064089,
+      "loss": 2.8577,
+      "step": 21900
+    },
+    {
+      "epoch": 0.29570687384069466,
+      "grad_norm": 0.5889946818351746,
+      "learning_rate": 0.0049999775158637745,
+      "loss": 2.8501,
+      "step": 22000
+    },
+    {
+      "epoch": 0.29570687384069466,
+      "eval_MaskedAccuracy": 0.3906472199993126,
+      "eval_loss": 3.1981027126312256,
+      "eval_runtime": 157.5475,
+      "eval_samples_per_second": 402.901,
+      "eval_steps_per_second": 1.574,
+      "step": 22000
+    },
+    {
+      "epoch": 0.29705099599451595,
+      "grad_norm": 1.3106350898742676,
+      "learning_rate": 0.00499997521124363,
+      "loss": 2.8498,
+      "step": 22100
+    },
+    {
+      "epoch": 0.2983951181483373,
+      "grad_norm": 0.670660674571991,
+      "learning_rate": 0.0049999727942037515,
+      "loss": 2.8471,
+      "step": 22200
+    },
+    {
+      "epoch": 0.29973924030215865,
+      "grad_norm": 0.6043007969856262,
+      "learning_rate": 0.004999970264744249,
+      "loss": 2.8529,
+      "step": 22300
+    },
+    {
+      "epoch": 0.30108336245598,
+      "grad_norm": 0.654592752456665,
+      "learning_rate": 0.004999967622865233,
+      "loss": 2.855,
+      "step": 22400
+    },
+    {
+      "epoch": 0.30242748460980134,
+      "grad_norm": 0.7757537364959717,
+      "learning_rate": 0.004999964868566834,
+      "loss": 2.8442,
+      "step": 22500
+    },
+    {
+      "epoch": 0.3037716067636227,
+      "grad_norm": 0.6157627105712891,
+      "learning_rate": 0.004999962001849164,
+      "loss": 2.8504,
+      "step": 22600
+    },
+    {
+      "epoch": 0.30511572891744404,
+      "grad_norm": 0.5535680055618286,
+      "learning_rate": 0.004999959022712354,
+      "loss": 2.8465,
+      "step": 22700
+    },
+    {
+      "epoch": 0.30645985107126533,
+      "grad_norm": 1.0820826292037964,
+      "learning_rate": 0.0049999559311565385,
+      "loss": 2.8407,
+      "step": 22800
+    },
+    {
+      "epoch": 0.3078039732250867,
+      "grad_norm": 0.898114800453186,
+      "learning_rate": 0.004999952727181866,
+      "loss": 2.8461,
+      "step": 22900
+    },
+    {
+      "epoch": 0.309148095378908,
+      "grad_norm": 0.787267804145813,
+      "learning_rate": 0.0049999494107884745,
+      "loss": 2.8352,
+      "step": 23000
+    },
+    {
+      "epoch": 0.309148095378908,
+      "eval_MaskedAccuracy": 0.3932447713059814,
+      "eval_loss": 3.172161340713501,
+      "eval_runtime": 157.8175,
+      "eval_samples_per_second": 402.212,
+      "eval_steps_per_second": 1.571,
+      "step": 23000
+    },
+    {
+      "epoch": 0.3104922175327294,
+      "grad_norm": 0.5774521827697754,
+      "learning_rate": 0.004999945981976511,
+      "loss": 2.8263,
+      "step": 23100
+    },
+    {
+      "epoch": 0.3118363396865507,
+      "grad_norm": 0.8543257117271423,
+      "learning_rate": 0.004999942440746131,
+      "loss": 2.8307,
+      "step": 23200
+    },
+    {
+      "epoch": 0.31318046184037207,
+      "grad_norm": 2.0594680309295654,
+      "learning_rate": 0.004999938787097496,
+      "loss": 2.8367,
+      "step": 23300
+    },
+    {
+      "epoch": 0.3145245839941934,
+      "grad_norm": 0.874093770980835,
+      "learning_rate": 0.00499993502103077,
+      "loss": 2.8386,
+      "step": 23400
+    },
+    {
+      "epoch": 0.3158687061480147,
+      "grad_norm": 0.5475397706031799,
+      "learning_rate": 0.004999931142546117,
+      "loss": 2.8264,
+      "step": 23500
+    },
+    {
+      "epoch": 0.31721282830183606,
+      "grad_norm": 0.8593809008598328,
+      "learning_rate": 0.004999927151643723,
+      "loss": 2.829,
+      "step": 23600
+    },
+    {
+      "epoch": 0.3185569504556574,
+      "grad_norm": 0.8034321665763855,
+      "learning_rate": 0.0049999230483237615,
+      "loss": 2.8212,
+      "step": 23700
+    },
+    {
+      "epoch": 0.31990107260947875,
+      "grad_norm": 0.6283460855484009,
+      "learning_rate": 0.004999918832586416,
+      "loss": 2.8213,
+      "step": 23800
+    },
+    {
+      "epoch": 0.3212451947633001,
+      "grad_norm": 0.7373483180999756,
+      "learning_rate": 0.004999914504431884,
+      "loss": 2.8307,
+      "step": 23900
+    },
+    {
+      "epoch": 0.32258931691712145,
+      "grad_norm": 0.5492293834686279,
+      "learning_rate": 0.004999910063860352,
+      "loss": 2.8248,
+      "step": 24000
+    },
+    {
+      "epoch": 0.32258931691712145,
+      "eval_MaskedAccuracy": 0.39530929398224096,
+      "eval_loss": 3.154477596282959,
+      "eval_runtime": 154.5504,
+      "eval_samples_per_second": 410.714,
+      "eval_steps_per_second": 1.605,
+      "step": 24000
+    },
+    {
+      "epoch": 0.32393343907094274,
+      "grad_norm": 0.599720299243927,
+      "learning_rate": 0.004999905510872023,
+      "loss": 2.8143,
+      "step": 24100
+    },
+    {
+      "epoch": 0.3252775612247641,
+      "grad_norm": 1.1359796524047852,
+      "learning_rate": 0.004999900845467094,
+      "loss": 2.8142,
+      "step": 24200
+    },
+    {
+      "epoch": 0.32662168337858544,
+      "grad_norm": 1.4440075159072876,
+      "learning_rate": 0.004999896067645785,
+      "loss": 2.8057,
+      "step": 24300
+    },
+    {
+      "epoch": 0.3279658055324068,
+      "grad_norm": 1.1459801197052002,
+      "learning_rate": 0.004999891177408309,
+      "loss": 2.8099,
+      "step": 24400
+    },
+    {
+      "epoch": 0.32930992768622813,
+      "grad_norm": 1.4969666004180908,
+      "learning_rate": 0.004999886174754891,
+      "loss": 2.8065,
+      "step": 24500
+    },
+    {
+      "epoch": 0.3306540498400495,
+      "grad_norm": 5.83732271194458,
+      "learning_rate": 0.004999881059685752,
+      "loss": 2.8102,
+      "step": 24600
+    },
+    {
+      "epoch": 0.3319981719938708,
+      "grad_norm": 1.80928373336792,
+      "learning_rate": 0.004999875832201121,
+      "loss": 2.8057,
+      "step": 24700
+    },
+    {
+      "epoch": 0.3333422941476921,
+      "grad_norm": 0.7706397771835327,
+      "learning_rate": 0.004999870492301236,
+      "loss": 2.8147,
+      "step": 24800
+    },
+    {
+      "epoch": 0.33468641630151347,
+      "grad_norm": 1.311485767364502,
+      "learning_rate": 0.0049998650399863355,
+      "loss": 2.8115,
+      "step": 24900
+    },
+    {
+      "epoch": 0.3360305384553348,
+      "grad_norm": 2.0457215309143066,
+      "learning_rate": 0.004999859475256669,
+      "loss": 2.8031,
+      "step": 25000
+    },
+    {
+      "epoch": 0.3360305384553348,
+      "eval_MaskedAccuracy": 0.39694718273561536,
+      "eval_loss": 3.1437790393829346,
+      "eval_runtime": 153.8578,
+      "eval_samples_per_second": 412.563,
+      "eval_steps_per_second": 1.612,
+      "step": 25000
+    },
+    {
+      "epoch": 0.33737466060915616,
+      "grad_norm": 1.1708168983459473,
+      "learning_rate": 0.004999853798112488,
+      "loss": 2.8014,
+      "step": 25100
+    },
+    {
+      "epoch": 0.3387187827629775,
+      "grad_norm": 0.6717488169670105,
+      "learning_rate": 0.004999848008554047,
+      "loss": 2.8061,
+      "step": 25200
+    },
+    {
+      "epoch": 0.34006290491679886,
+      "grad_norm": 1.743415355682373,
+      "learning_rate": 0.004999842106581601,
+      "loss": 2.804,
+      "step": 25300
+    },
+    {
+      "epoch": 0.34140702707062015,
+      "grad_norm": 1.1857333183288574,
+      "learning_rate": 0.004999836092195418,
+      "loss": 2.8049,
+      "step": 25400
+    },
+    {
+      "epoch": 0.3427511492244415,
+      "grad_norm": 0.7652652263641357,
+      "learning_rate": 0.004999829965395771,
+      "loss": 2.7976,
+      "step": 25500
+    },
+    {
+      "epoch": 0.34409527137826285,
+      "grad_norm": 0.8573779463768005,
+      "learning_rate": 0.004999823726182935,
+      "loss": 2.7952,
+      "step": 25600
+    },
+    {
+      "epoch": 0.3454393935320842,
+      "grad_norm": 1.1807929277420044,
+      "learning_rate": 0.004999817374557191,
+      "loss": 2.8,
+      "step": 25700
+    },
+    {
+      "epoch": 0.34678351568590554,
+      "grad_norm": 0.968813419342041,
+      "learning_rate": 0.004999810910518828,
+      "loss": 2.7862,
+      "step": 25800
+    },
+    {
+      "epoch": 0.3481276378397269,
+      "grad_norm": 1.2410879135131836,
+      "learning_rate": 0.004999804334068139,
+      "loss": 2.792,
+      "step": 25900
+    },
+    {
+      "epoch": 0.34947175999354824,
+      "grad_norm": 1.0421373844146729,
+      "learning_rate": 0.004999797645205414,
+      "loss": 2.7928,
+      "step": 26000
+    },
+    {
+      "epoch": 0.34947175999354824,
+      "eval_MaskedAccuracy": 0.40042939663715493,
+      "eval_loss": 3.116644859313965,
+      "eval_runtime": 159.2355,
+      "eval_samples_per_second": 398.63,
+      "eval_steps_per_second": 1.557,
+      "step": 26000
+    },
+    {
+      "epoch": 0.35081588214736953,
+      "grad_norm": 1.1622250080108643,
+      "learning_rate": 0.004999790843930961,
+      "loss": 2.7914,
+      "step": 26100
+    },
+    {
+      "epoch": 0.3521600043011909,
+      "grad_norm": 0.9314205050468445,
+      "learning_rate": 0.004999783930245077,
+      "loss": 2.7895,
+      "step": 26200
+    },
+    {
+      "epoch": 0.3535041264550122,
+      "grad_norm": 2.6555447578430176,
+      "learning_rate": 0.004999776904148078,
+      "loss": 2.7872,
+      "step": 26300
+    },
+    {
+      "epoch": 0.35484824860883357,
+      "grad_norm": 1.0436155796051025,
+      "learning_rate": 0.004999769765640281,
+      "loss": 2.7887,
+      "step": 26400
+    },
+    {
+      "epoch": 0.3561923707626549,
+      "grad_norm": 0.901914656162262,
+      "learning_rate": 0.004999762514722008,
+      "loss": 2.788,
+      "step": 26500
+    },
+    {
+      "epoch": 0.35753649291647627,
+      "grad_norm": 1.7726536989212036,
+      "learning_rate": 0.004999755151393587,
+      "loss": 2.7863,
+      "step": 26600
+    },
+    {
+      "epoch": 0.3588806150702976,
+      "grad_norm": 0.90220046043396,
+      "learning_rate": 0.004999747675655358,
+      "loss": 2.7789,
+      "step": 26700
+    },
+    {
+      "epoch": 0.3602247372241189,
+      "grad_norm": 1.074455976486206,
+      "learning_rate": 0.004999740087507646,
+      "loss": 2.7786,
+      "step": 26800
+    },
+    {
+      "epoch": 0.36156885937794025,
+      "grad_norm": 1.2537977695465088,
+      "learning_rate": 0.004999732386950793,
+      "loss": 2.7792,
+      "step": 26900
+    },
+    {
+      "epoch": 0.3629129815317616,
+      "grad_norm": 1.4169613122940063,
+      "learning_rate": 0.004999724573985147,
+      "loss": 2.7773,
+      "step": 27000
+    },
+    {
+      "epoch": 0.3629129815317616,
+      "eval_MaskedAccuracy": 0.4007810600247028,
+      "eval_loss": 3.110098361968994,
+      "eval_runtime": 157.997,
+      "eval_samples_per_second": 401.754,
+      "eval_steps_per_second": 1.57,
+      "step": 27000
+    },
+    {
+      "epoch": 0.36425710368558295,
+      "grad_norm": 0.6324722170829773,
+      "learning_rate": 0.004999716648611063,
+      "loss": 2.7757,
+      "step": 27100
+    },
+    {
+      "epoch": 0.3656012258394043,
+      "grad_norm": 5.492055416107178,
+      "learning_rate": 0.004999708610828894,
+      "loss": 2.7697,
+      "step": 27200
+    },
+    {
+      "epoch": 0.36694534799322565,
+      "grad_norm": 0.8834023475646973,
+      "learning_rate": 0.004999700460639006,
+      "loss": 2.7714,
+      "step": 27300
+    },
+    {
+      "epoch": 0.36828947014704694,
+      "grad_norm": 0.8748111724853516,
+      "learning_rate": 0.004999692198041765,
+      "loss": 2.7794,
+      "step": 27400
+    },
+    {
+      "epoch": 0.3696335923008683,
+      "grad_norm": 1.2332714796066284,
+      "learning_rate": 0.004999683823037545,
+      "loss": 2.7656,
+      "step": 27500
+    },
+    {
+      "epoch": 0.37097771445468963,
+      "grad_norm": 0.7697482705116272,
+      "learning_rate": 0.0049996753356267255,
+      "loss": 2.7677,
+      "step": 27600
+    },
+    {
+      "epoch": 0.372321836608511,
+      "grad_norm": 0.9096680283546448,
+      "learning_rate": 0.004999666735809681,
+      "loss": 2.7641,
+      "step": 27700
+    },
+    {
+      "epoch": 0.37366595876233233,
+      "grad_norm": 0.5890783667564392,
+      "learning_rate": 0.004999658023586803,
+      "loss": 2.7621,
+      "step": 27800
+    },
+    {
+      "epoch": 0.3750100809161537,
+      "grad_norm": 0.7764729261398315,
+      "learning_rate": 0.0049996491989584835,
+      "loss": 2.7697,
+      "step": 27900
+    },
+    {
+      "epoch": 0.376354203069975,
+      "grad_norm": 2.1192266941070557,
+      "learning_rate": 0.004999640261925122,
+      "loss": 2.7622,
+      "step": 28000
+    },
+    {
+      "epoch": 0.376354203069975,
+      "eval_MaskedAccuracy": 0.4028772828245879,
+      "eval_loss": 3.092843532562256,
+      "eval_runtime": 159.6753,
+      "eval_samples_per_second": 397.532,
+      "eval_steps_per_second": 1.553,
+      "step": 28000
+    },
+    {
+      "epoch": 0.3776983252237963,
+      "grad_norm": 1.4101366996765137,
+      "learning_rate": 0.00499963121248712,
+      "loss": 2.7642,
+      "step": 28100
+    },
+    {
+      "epoch": 0.37904244737761766,
+      "grad_norm": 1.5181620121002197,
+      "learning_rate": 0.0049996220506448846,
+      "loss": 2.7512,
+      "step": 28200
+    },
+    {
+      "epoch": 0.380386569531439,
+      "grad_norm": 1.365559697151184,
+      "learning_rate": 0.004999612776398833,
+      "loss": 2.7688,
+      "step": 28300
+    },
+    {
+      "epoch": 0.38173069168526036,
+      "grad_norm": 1.040536880493164,
+      "learning_rate": 0.0049996033897493795,
+      "loss": 2.7537,
+      "step": 28400
+    },
+    {
+      "epoch": 0.3830748138390817,
+      "grad_norm": 1.0811822414398193,
+      "learning_rate": 0.004999593890696948,
+      "loss": 2.7639,
+      "step": 28500
+    },
+    {
+      "epoch": 0.38441893599290305,
+      "grad_norm": 1.0675724744796753,
+      "learning_rate": 0.004999584279241965,
+      "loss": 2.7556,
+      "step": 28600
+    },
+    {
+      "epoch": 0.38576305814672435,
+      "grad_norm": 0.6620686650276184,
+      "learning_rate": 0.0049995745553848645,
+      "loss": 2.7583,
+      "step": 28700
+    },
+    {
+      "epoch": 0.3871071803005457,
+      "grad_norm": 1.1735743284225464,
+      "learning_rate": 0.004999564719126081,
+      "loss": 2.7482,
+      "step": 28800
+    },
+    {
+      "epoch": 0.38845130245436704,
+      "grad_norm": 0.5812009572982788,
+      "learning_rate": 0.004999554770466058,
+      "loss": 2.7443,
+      "step": 28900
+    },
+    {
+      "epoch": 0.3897954246081884,
+      "grad_norm": 1.6214362382888794,
+      "learning_rate": 0.0049995447094052506,
+      "loss": 2.7488,
+      "step": 29000
+    },
+    {
+      "epoch": 0.3897954246081884,
+      "eval_MaskedAccuracy": 0.403793286520234,
+      "eval_loss": 3.0827648639678955,
+      "eval_runtime": 160.6528,
+      "eval_samples_per_second": 395.113,
+      "eval_steps_per_second": 1.544,
+      "step": 29000
+    },
+    {
+      "epoch": 0.39113954676200974,
+      "grad_norm": 1.382157564163208,
+      "learning_rate": 0.004999534535944106,
+      "loss": 2.753,
+      "step": 29100
+    },
+    {
+      "epoch": 0.3924836689158311,
+      "grad_norm": 1.1490193605422974,
+      "learning_rate": 0.00499952425008309,
+      "loss": 2.7411,
+      "step": 29200
+    },
+    {
+      "epoch": 0.39382779106965243,
+      "grad_norm": 0.5846394300460815,
+      "learning_rate": 0.00499951385182266,
+      "loss": 2.7502,
+      "step": 29300
+    },
+    {
+      "epoch": 0.3951719132234737,
+      "grad_norm": 1.563256025314331,
+      "learning_rate": 0.0049995033411632805,
+      "loss": 2.7484,
+      "step": 29400
+    },
+    {
+      "epoch": 0.3965160353772951,
+      "grad_norm": 0.510226309299469,
+      "learning_rate": 0.004999492718105435,
+      "loss": 2.7418,
+      "step": 29500
+    },
+    {
+      "epoch": 0.3978601575311164,
+      "grad_norm": 0.7409377694129944,
+      "learning_rate": 0.0049994819826496015,
+      "loss": 2.7455,
+      "step": 29600
+    },
+    {
+      "epoch": 0.39920427968493777,
+      "grad_norm": 1.132575511932373,
+      "learning_rate": 0.004999471134796253,
+      "loss": 2.7337,
+      "step": 29700
+    },
+    {
+      "epoch": 0.4005484018387591,
+      "grad_norm": 0.8493718504905701,
+      "learning_rate": 0.004999460174545891,
+      "loss": 2.7374,
+      "step": 29800
+    },
+    {
+      "epoch": 0.40189252399258046,
+      "grad_norm": 1.6549121141433716,
+      "learning_rate": 0.004999449101898994,
+      "loss": 2.7345,
+      "step": 29900
+    },
+    {
+      "epoch": 0.4032366461464018,
+      "grad_norm": 1.0378655195236206,
+      "learning_rate": 0.004999437916856077,
+      "loss": 2.7411,
+      "step": 30000
+    },
+    {
+      "epoch": 0.4032366461464018,
+      "eval_MaskedAccuracy": 0.4059693369791507,
+      "eval_loss": 3.067436456680298,
+      "eval_runtime": 349.401,
+      "eval_samples_per_second": 181.671,
+      "eval_steps_per_second": 0.71,
+      "step": 30000
+    },
+    {
+      "epoch": 0.4045807683002231,
+      "grad_norm": 3.7114546298980713,
+      "learning_rate": 0.004999426619417629,
+      "loss": 2.7291,
+      "step": 30100
+    },
+    {
+      "epoch": 0.40592489045404445,
+      "grad_norm": 0.8062450289726257,
+      "learning_rate": 0.0049994152095841704,
+      "loss": 2.7369,
+      "step": 30200
+    },
+    {
+      "epoch": 0.4072690126078658,
+      "grad_norm": 0.82357257604599,
+      "learning_rate": 0.004999403687356217,
+      "loss": 2.7361,
+      "step": 30300
+    },
+    {
+      "epoch": 0.40861313476168715,
+      "grad_norm": 0.6285329461097717,
+      "learning_rate": 0.004999392052734278,
+      "loss": 2.7343,
+      "step": 30400
+    },
+    {
+      "epoch": 0.4099572569155085,
+      "grad_norm": 0.9262251853942871,
+      "learning_rate": 0.0049993803057188785,
+      "loss": 2.7323,
+      "step": 30500
+    },
+    {
+      "epoch": 0.41130137906932984,
+      "grad_norm": 1.2343696355819702,
+      "learning_rate": 0.004999368446310558,
+      "loss": 2.7286,
+      "step": 30600
+    },
+    {
+      "epoch": 0.41264550122315113,
+      "grad_norm": 0.9753904342651367,
+      "learning_rate": 0.004999356474509844,
+      "loss": 2.7363,
+      "step": 30700
+    },
+    {
+      "epoch": 0.4139896233769725,
+      "grad_norm": 0.7958431839942932,
+      "learning_rate": 0.004999344390317274,
+      "loss": 2.7298,
+      "step": 30800
+    },
+    {
+      "epoch": 0.41533374553079383,
+      "grad_norm": 4.113492965698242,
+      "learning_rate": 0.004999332193733396,
+      "loss": 2.732,
+      "step": 30900
+    },
+    {
+      "epoch": 0.4166778676846152,
+      "grad_norm": 1.0215173959732056,
+      "learning_rate": 0.004999319884758765,
+      "loss": 2.7333,
+      "step": 31000
+    },
+    {
+      "epoch": 0.4166778676846152,
+      "eval_MaskedAccuracy": 0.40677681485969963,
+      "eval_loss": 3.0563271045684814,
+      "eval_runtime": 159.104,
+      "eval_samples_per_second": 398.959,
+      "eval_steps_per_second": 1.559,
+      "step": 31000
+    },
+    {
+      "epoch": 0.4180219898384365,
+      "grad_norm": 1.9819655418395996,
+      "learning_rate": 0.004999307463393925,
+      "loss": 2.7266,
+      "step": 31100
+    },
+    {
+      "epoch": 0.4193661119922579,
+      "grad_norm": 1.4452855587005615,
+      "learning_rate": 0.004999294929639439,
+      "loss": 2.7248,
+      "step": 31200
+    },
+    {
+      "epoch": 0.4207102341460792,
+      "grad_norm": 0.7133325934410095,
+      "learning_rate": 0.004999282283495872,
+      "loss": 2.7337,
+      "step": 31300
+    },
+    {
+      "epoch": 0.4220543562999005,
+      "grad_norm": 0.930253267288208,
+      "learning_rate": 0.004999269524963801,
+      "loss": 2.7284,
+      "step": 31400
+    },
+    {
+      "epoch": 0.42339847845372186,
+      "grad_norm": 2.160118818283081,
+      "learning_rate": 0.004999256654043792,
+      "loss": 2.7228,
+      "step": 31500
+    },
+    {
+      "epoch": 0.4247426006075432,
+      "grad_norm": 1.322357416152954,
+      "learning_rate": 0.004999243670736428,
+      "loss": 2.729,
+      "step": 31600
+    },
+    {
+      "epoch": 0.42608672276136456,
+      "grad_norm": 1.2306387424468994,
+      "learning_rate": 0.00499923057504229,
+      "loss": 2.7278,
+      "step": 31700
+    },
+    {
+      "epoch": 0.4274308449151859,
+      "grad_norm": 2.193079710006714,
+      "learning_rate": 0.004999217366961968,
+      "loss": 2.7169,
+      "step": 31800
+    },
+    {
+      "epoch": 0.42877496706900725,
+      "grad_norm": 0.698665976524353,
+      "learning_rate": 0.0049992040464960615,
+      "loss": 2.717,
+      "step": 31900
+    },
+    {
+      "epoch": 0.43011908922282854,
+      "grad_norm": 0.765285313129425,
+      "learning_rate": 0.004999190613645174,
+      "loss": 2.7152,
+      "step": 32000
+    },
+    {
+      "epoch": 0.43011908922282854,
+      "eval_MaskedAccuracy": 0.40871540481758456,
+      "eval_loss": 3.042783260345459,
+      "eval_runtime": 159.0761,
+      "eval_samples_per_second": 399.029,
+      "eval_steps_per_second": 1.559,
+      "step": 32000
+    },
+    {
+      "epoch": 0.4314632113766499,
+      "grad_norm": 0.7268723845481873,
+      "learning_rate": 0.004999177068409907,
+      "loss": 2.7235,
+      "step": 32100
+    },
+    {
+      "epoch": 0.43280733353047124,
+      "grad_norm": 0.5239226818084717,
+      "learning_rate": 0.004999163410790871,
+      "loss": 2.7151,
+      "step": 32200
+    },
+    {
+      "epoch": 0.4341514556842926,
+      "grad_norm": 1.3244593143463135,
+      "learning_rate": 0.004999149640788679,
+      "loss": 2.7141,
+      "step": 32300
+    },
+    {
+      "epoch": 0.43549557783811393,
+      "grad_norm": 0.9290921688079834,
+      "learning_rate": 0.004999135758403953,
+      "loss": 2.7156,
+      "step": 32400
+    },
+    {
+      "epoch": 0.4368396999919353,
+      "grad_norm": 0.5451181530952454,
+      "learning_rate": 0.004999121763637319,
+      "loss": 2.7102,
+      "step": 32500
+    },
+    {
+      "epoch": 0.43818382214575663,
+      "grad_norm": 0.9701553583145142,
+      "learning_rate": 0.004999107656489411,
+      "loss": 2.7198,
+      "step": 32600
+    },
+    {
+      "epoch": 0.4395279442995779,
+      "grad_norm": 5.558290004730225,
+      "learning_rate": 0.004999093436960863,
+      "loss": 2.7124,
+      "step": 32700
+    },
+    {
+      "epoch": 0.44087206645339927,
+      "grad_norm": 0.5010631084442139,
+      "learning_rate": 0.004999079105052311,
+      "loss": 2.7112,
+      "step": 32800
+    },
+    {
+      "epoch": 0.4422161886072206,
+      "grad_norm": 0.5793175101280212,
+      "learning_rate": 0.004999064660764402,
+      "loss": 2.7111,
+      "step": 32900
+    },
+    {
+      "epoch": 0.44356031076104196,
+      "grad_norm": 0.9704194664955139,
+      "learning_rate": 0.004999050104097792,
+      "loss": 2.7093,
+      "step": 33000
+    },
+    {
+      "epoch": 0.44356031076104196,
+      "eval_MaskedAccuracy": 0.4106415839014644,
+      "eval_loss": 3.032620906829834,
+      "eval_runtime": 161.401,
+      "eval_samples_per_second": 393.281,
+      "eval_steps_per_second": 1.537,
+      "step": 33000
+    },
+    {
+      "epoch": 0.4449044329148633,
+      "grad_norm": 2.713127613067627,
+      "learning_rate": 0.004999035435053133,
+      "loss": 2.7105,
+      "step": 33100
+    },
+    {
+      "epoch": 0.44624855506868466,
+      "grad_norm": 0.9831237196922302,
+      "learning_rate": 0.0049990206536310805,
+      "loss": 2.7106,
+      "step": 33200
+    },
+    {
+      "epoch": 0.447592677222506,
+      "grad_norm": 1.5991238355636597,
+      "learning_rate": 0.004999005759832306,
+      "loss": 2.6998,
+      "step": 33300
+    },
+    {
+      "epoch": 0.4489367993763273,
+      "grad_norm": 1.0113264322280884,
+      "learning_rate": 0.004998990753657483,
+      "loss": 2.7157,
+      "step": 33400
+    },
+    {
+      "epoch": 0.45028092153014865,
+      "grad_norm": 1.1059205532073975,
+      "learning_rate": 0.0049989756351072855,
+      "loss": 2.7127,
+      "step": 33500
+    },
+    {
+      "epoch": 0.45162504368397,
+      "grad_norm": 0.8544152975082397,
+      "learning_rate": 0.0049989604041823935,
+      "loss": 2.7107,
+      "step": 33600
+    },
+    {
+      "epoch": 0.45296916583779134,
+      "grad_norm": 0.6586983799934387,
+      "learning_rate": 0.004998945060883495,
+      "loss": 2.7205,
+      "step": 33700
+    },
+    {
+      "epoch": 0.4543132879916127,
+      "grad_norm": 0.7723863124847412,
+      "learning_rate": 0.00499892960521128,
+      "loss": 2.694,
+      "step": 33800
+    },
+    {
+      "epoch": 0.45565741014543404,
+      "grad_norm": 1.5881659984588623,
+      "learning_rate": 0.00499891403716645,
+      "loss": 2.7052,
+      "step": 33900
+    },
+    {
+      "epoch": 0.45700153229925533,
+      "grad_norm": 1.2848504781723022,
+      "learning_rate": 0.004998898356749702,
+      "loss": 2.703,
+      "step": 34000
+    },
+    {
+      "epoch": 0.45700153229925533,
+      "eval_MaskedAccuracy": 0.4106395221691058,
+      "eval_loss": 3.0283474922180176,
+      "eval_runtime": 258.5416,
+      "eval_samples_per_second": 245.516,
+      "eval_steps_per_second": 0.959,
+      "step": 34000
+    },
+    {
+      "epoch": 0.4583456544530767,
+      "grad_norm": 0.7230987548828125,
+      "learning_rate": 0.0049988825639617415,
+      "loss": 2.6976,
+      "step": 34100
+    },
+    {
+      "epoch": 0.459689776606898,
+      "grad_norm": 0.671057403087616,
+      "learning_rate": 0.004998866658803282,
+      "loss": 2.6931,
+      "step": 34200
+    },
+    {
+      "epoch": 0.4610338987607194,
+      "grad_norm": 1.4508076906204224,
+      "learning_rate": 0.004998850641275041,
+      "loss": 2.7036,
+      "step": 34300
+    },
+    {
+      "epoch": 0.4623780209145407,
+      "grad_norm": 1.325042724609375,
+      "learning_rate": 0.004998834511377738,
+      "loss": 2.6982,
+      "step": 34400
+    },
+    {
+      "epoch": 0.46372214306836207,
+      "grad_norm": 0.49313464760780334,
+      "learning_rate": 0.004998818269112104,
+      "loss": 2.6976,
+      "step": 34500
+    },
+    {
+      "epoch": 0.4650662652221834,
+      "grad_norm": 0.7200739979743958,
+      "learning_rate": 0.00499880191447886,
+      "loss": 2.6979,
+      "step": 34600
+    },
+    {
+      "epoch": 0.4664103873760047,
+      "grad_norm": 1.495227336883545,
+      "learning_rate": 0.004998785447478756,
+      "loss": 2.6831,
+      "step": 34700
+    },
+    {
+      "epoch": 0.46775450952982606,
+      "grad_norm": 0.5236303210258484,
+      "learning_rate": 0.00499876886811252,
+      "loss": 2.6932,
+      "step": 34800
+    },
+    {
+      "epoch": 0.4690986316836474,
+      "grad_norm": 1.4335426092147827,
+      "learning_rate": 0.004998752176380915,
+      "loss": 2.6969,
+      "step": 34900
+    },
+    {
+      "epoch": 0.47044275383746875,
+      "grad_norm": 2.9952592849731445,
+      "learning_rate": 0.004998735372284684,
+      "loss": 2.7029,
+      "step": 35000
+    },
+    {
+      "epoch": 0.47044275383746875,
+      "eval_MaskedAccuracy": 0.4116071260461691,
+      "eval_loss": 3.016385078430176,
+      "eval_runtime": 156.2897,
+      "eval_samples_per_second": 406.143,
+      "eval_steps_per_second": 1.587,
+      "step": 35000
+    },
+    {
+      "epoch": 0.4717868759912901,
+      "grad_norm": 0.6854963302612305,
+      "learning_rate": 0.004998718455824586,
+      "loss": 2.6894,
+      "step": 35100
+    },
+    {
+      "epoch": 0.47313099814511145,
+      "grad_norm": 2.010087251663208,
+      "learning_rate": 0.004998701427001374,
+      "loss": 2.691,
+      "step": 35200
+    },
+    {
+      "epoch": 0.47447512029893274,
+      "grad_norm": 0.6356353163719177,
+      "learning_rate": 0.004998684285815833,
+      "loss": 2.69,
+      "step": 35300
+    },
+    {
+      "epoch": 0.4758192424527541,
+      "grad_norm": 0.5063838958740234,
+      "learning_rate": 0.004998667032268725,
+      "loss": 2.6895,
+      "step": 35400
+    },
+    {
+      "epoch": 0.47716336460657544,
+      "grad_norm": 0.8183920979499817,
+      "learning_rate": 0.0049986496663608285,
+      "loss": 2.6951,
+      "step": 35500
+    },
+    {
+      "epoch": 0.4785074867603968,
+      "grad_norm": 0.9003168940544128,
+      "learning_rate": 0.004998632188092928,
+      "loss": 2.6822,
+      "step": 35600
+    },
+    {
+      "epoch": 0.47985160891421813,
+      "grad_norm": 0.8852120041847229,
+      "learning_rate": 0.004998614597465808,
+      "loss": 2.6806,
+      "step": 35700
+    },
+    {
+      "epoch": 0.4811957310680395,
+      "grad_norm": 0.7671552896499634,
+      "learning_rate": 0.004998596894480259,
+      "loss": 2.6789,
+      "step": 35800
+    },
+    {
+      "epoch": 0.4825398532218608,
+      "grad_norm": 1.260359287261963,
+      "learning_rate": 0.004998579079137085,
+      "loss": 2.679,
+      "step": 35900
+    },
+    {
+      "epoch": 0.4838839753756821,
+      "grad_norm": 0.5925015807151794,
+      "learning_rate": 0.004998561151437081,
+      "loss": 2.69,
+      "step": 36000
+    },
+    {
+      "epoch": 0.4838839753756821,
+      "eval_MaskedAccuracy": 0.41254465747964864,
+      "eval_loss": 3.0127551555633545,
+      "eval_runtime": 154.2797,
+      "eval_samples_per_second": 411.434,
+      "eval_steps_per_second": 1.607,
+      "step": 36000
+    },
+    {
+      "epoch": 0.48522809752950347,
+      "grad_norm": 0.7644244432449341,
+      "learning_rate": 0.004998543111381064,
+      "loss": 2.6828,
+      "step": 36100
+    },
+    {
+      "epoch": 0.4865722196833248,
+      "grad_norm": 0.8469159007072449,
+      "learning_rate": 0.004998524958969839,
+      "loss": 2.6878,
+      "step": 36200
+    },
+    {
+      "epoch": 0.48791634183714616,
+      "grad_norm": 0.847205638885498,
+      "learning_rate": 0.004998506694204228,
+      "loss": 2.6818,
+      "step": 36300
+    },
+    {
+      "epoch": 0.4892604639909675,
+      "grad_norm": 0.581739604473114,
+      "learning_rate": 0.004998488317085059,
+      "loss": 2.6741,
+      "step": 36400
+    },
+    {
+      "epoch": 0.49060458614478886,
+      "grad_norm": 2.133725881576538,
+      "learning_rate": 0.004998469827613156,
+      "loss": 2.6786,
+      "step": 36500
+    },
+    {
+      "epoch": 0.4919487082986102,
+      "grad_norm": 1.3259893655776978,
+      "learning_rate": 0.00499845122578935,
+      "loss": 2.6819,
+      "step": 36600
+    },
+    {
+      "epoch": 0.4932928304524315,
+      "grad_norm": 1.4917380809783936,
+      "learning_rate": 0.004998432511614469,
+      "loss": 2.6802,
+      "step": 36700
+    },
+    {
+      "epoch": 0.49463695260625284,
+      "grad_norm": 2.2140398025512695,
+      "learning_rate": 0.004998413685089377,
+      "loss": 2.6767,
+      "step": 36800
+    },
+    {
+      "epoch": 0.4959810747600742,
+      "grad_norm": 0.7825784683227539,
+      "learning_rate": 0.004998394746214914,
+      "loss": 2.6801,
+      "step": 36900
+    },
+    {
+      "epoch": 0.49732519691389554,
+      "grad_norm": 2.3250296115875244,
+      "learning_rate": 0.00499837569499193,
+      "loss": 2.6736,
+      "step": 37000
+    },
+    {
+      "epoch": 0.49732519691389554,
+      "eval_MaskedAccuracy": 0.413791745700555,
+      "eval_loss": 3.001739740371704,
+      "eval_runtime": 155.081,
+      "eval_samples_per_second": 409.309,
+      "eval_steps_per_second": 1.599,
+      "step": 37000
+    },
+    {
+      "epoch": 0.4986693190677169,
+      "grad_norm": 0.9509634971618652,
+      "learning_rate": 0.004998356531421287,
+      "loss": 2.6815,
+      "step": 37100
+    },
+    {
+      "epoch": 0.5000134412215382,
+      "grad_norm": 0.5249235033988953,
+      "learning_rate": 0.004998337255503845,
+      "loss": 2.6761,
+      "step": 37200
+    },
+    {
+      "epoch": 0.5013575633753595,
+      "grad_norm": 0.492450088262558,
+      "learning_rate": 0.004998317867240472,
+      "loss": 2.6713,
+      "step": 37300
+    },
+    {
+      "epoch": 0.5027016855291809,
+      "grad_norm": 0.5649069547653198,
+      "learning_rate": 0.00499829836663204,
+      "loss": 2.6718,
+      "step": 37400
+    },
+    {
+      "epoch": 0.5040458076830022,
+      "grad_norm": 1.028743028640747,
+      "learning_rate": 0.004998278753679443,
+      "loss": 2.6786,
+      "step": 37500
+    },
+    {
+      "epoch": 0.5053899298368236,
+      "grad_norm": 1.2656359672546387,
+      "learning_rate": 0.004998259028383549,
+      "loss": 2.666,
+      "step": 37600
+    },
+    {
+      "epoch": 0.5067340519906449,
+      "grad_norm": 1.0689085721969604,
+      "learning_rate": 0.004998239190745246,
+      "loss": 2.6705,
+      "step": 37700
+    },
+    {
+      "epoch": 0.5080781741444662,
+      "grad_norm": 1.323631763458252,
+      "learning_rate": 0.00499821924076543,
+      "loss": 2.6748,
+      "step": 37800
+    },
+    {
+      "epoch": 0.5094222962982876,
+      "grad_norm": 1.0238298177719116,
+      "learning_rate": 0.004998199178445008,
+      "loss": 2.6692,
+      "step": 37900
+    },
+    {
+      "epoch": 0.5107664184521089,
+      "grad_norm": 1.5579122304916382,
+      "learning_rate": 0.0049981790037848806,
+      "loss": 2.6692,
+      "step": 38000
+    },
+    {
+      "epoch": 0.5107664184521089,
+      "eval_MaskedAccuracy": 0.41454185677096544,
+      "eval_loss": 2.9944686889648438,
+      "eval_runtime": 154.5873,
+      "eval_samples_per_second": 410.616,
+      "eval_steps_per_second": 1.604,
+      "step": 38000
+    },
+    {
+      "epoch": 0.5121105406059303,
+      "grad_norm": 0.9145413041114807,
+      "learning_rate": 0.004998158716785951,
+      "loss": 2.6664,
+      "step": 38100
+    },
+    {
+      "epoch": 0.5134546627597516,
+      "grad_norm": 1.3004742860794067,
+      "learning_rate": 0.0049981383174491365,
+      "loss": 2.6749,
+      "step": 38200
+    },
+    {
+      "epoch": 0.5147987849135729,
+      "grad_norm": 1.8650057315826416,
+      "learning_rate": 0.0049981178057753595,
+      "loss": 2.6799,
+      "step": 38300
+    },
+    {
+      "epoch": 0.5161429070673943,
+      "grad_norm": 1.2665841579437256,
+      "learning_rate": 0.004998097181765546,
+      "loss": 2.6638,
+      "step": 38400
+    },
+    {
+      "epoch": 0.5174870292212156,
+      "grad_norm": 0.8943463563919067,
+      "learning_rate": 0.004998076445420617,
+      "loss": 2.6612,
+      "step": 38500
+    },
+    {
+      "epoch": 0.518831151375037,
+      "grad_norm": 1.6985851526260376,
+      "learning_rate": 0.004998055596741511,
+      "loss": 2.6653,
+      "step": 38600
+    },
+    {
+      "epoch": 0.5201752735288583,
+      "grad_norm": 2.1343910694122314,
+      "learning_rate": 0.004998034635729168,
+      "loss": 2.6642,
+      "step": 38700
+    },
+    {
+      "epoch": 0.5215193956826797,
+      "grad_norm": 1.4548419713974,
+      "learning_rate": 0.004998013562384535,
+      "loss": 2.6524,
+      "step": 38800
+    },
+    {
+      "epoch": 0.522863517836501,
+      "grad_norm": 2.0018160343170166,
+      "learning_rate": 0.004997992376708557,
+      "loss": 2.6574,
+      "step": 38900
+    },
+    {
+      "epoch": 0.5242076399903223,
+      "grad_norm": 1.4500235319137573,
+      "learning_rate": 0.004997971078702187,
+      "loss": 2.6614,
+      "step": 39000
+    },
+    {
+      "epoch": 0.5242076399903223,
+      "eval_MaskedAccuracy": 0.4153363544702115,
+      "eval_loss": 2.9864044189453125,
+      "eval_runtime": 154.3717,
+      "eval_samples_per_second": 411.189,
+      "eval_steps_per_second": 1.607,
+      "step": 39000
+    },
+    {
+      "epoch": 0.5255517621441437,
+      "grad_norm": 1.6406347751617432,
+      "learning_rate": 0.0049979496683663855,
+      "loss": 2.6609,
+      "step": 39100
+    },
+    {
+      "epoch": 0.526895884297965,
+      "grad_norm": 1.1125140190124512,
+      "learning_rate": 0.004997928145702119,
+      "loss": 2.6553,
+      "step": 39200
+    },
+    {
+      "epoch": 0.5282400064517864,
+      "grad_norm": 4.977079391479492,
+      "learning_rate": 0.004997906510710358,
+      "loss": 2.6627,
+      "step": 39300
+    },
+    {
+      "epoch": 0.5295841286056077,
+      "grad_norm": 0.5258601903915405,
+      "learning_rate": 0.004997884763392075,
+      "loss": 2.6664,
+      "step": 39400
+    },
+    {
+      "epoch": 0.5309282507594291,
+      "grad_norm": 0.5638339519500732,
+      "learning_rate": 0.004997862903748257,
+      "loss": 2.6565,
+      "step": 39500
+    },
+    {
+      "epoch": 0.5322723729132504,
+      "grad_norm": 0.5276058912277222,
+      "learning_rate": 0.00499784093177988,
+      "loss": 2.6583,
+      "step": 39600
+    },
+    {
+      "epoch": 0.5336164950670717,
+      "grad_norm": 0.7912238836288452,
+      "learning_rate": 0.004997818847487944,
+      "loss": 2.6595,
+      "step": 39700
+    },
+    {
+      "epoch": 0.534960617220893,
+      "grad_norm": 0.949805736541748,
+      "learning_rate": 0.004997796650873429,
+      "loss": 2.6698,
+      "step": 39800
+    },
+    {
+      "epoch": 0.5363047393747143,
+      "grad_norm": 1.7956501245498657,
+      "learning_rate": 0.004997774341937344,
+      "loss": 2.6617,
+      "step": 39900
+    },
+    {
+      "epoch": 0.5376488615285357,
+      "grad_norm": 0.5093005895614624,
+      "learning_rate": 0.004997751920680701,
+      "loss": 2.6498,
+      "step": 40000
+    },
+    {
+      "epoch": 0.5376488615285357,
+      "eval_MaskedAccuracy": 0.4163031669419539,
+      "eval_loss": 2.9814693927764893,
+      "eval_runtime": 153.8302,
+      "eval_samples_per_second": 412.637,
+      "eval_steps_per_second": 1.612,
+      "step": 40000
+    },
+    {
+      "epoch": 0.538992983682357,
+      "grad_norm": 1.8199607133865356,
+      "learning_rate": 0.004997729387104499,
+      "loss": 2.6542,
+      "step": 40100
+    },
+    {
+      "epoch": 0.5403371058361784,
+      "grad_norm": 1.2168132066726685,
+      "learning_rate": 0.004997706741209757,
+      "loss": 2.6484,
+      "step": 40200
+    },
+    {
+      "epoch": 0.5416812279899997,
+      "grad_norm": 1.862183928489685,
+      "learning_rate": 0.004997683982997499,
+      "loss": 2.6529,
+      "step": 40300
+    },
+    {
+      "epoch": 0.543025350143821,
+      "grad_norm": 1.591640830039978,
+      "learning_rate": 0.004997661112468743,
+      "loss": 2.6541,
+      "step": 40400
+    },
+    {
+      "epoch": 0.5443694722976424,
+      "grad_norm": 3.0422515869140625,
+      "learning_rate": 0.004997638129624528,
+      "loss": 2.6493,
+      "step": 40500
+    },
+    {
+      "epoch": 0.5457135944514637,
+      "grad_norm": 0.5764662027359009,
+      "learning_rate": 0.004997615034465888,
+      "loss": 2.6454,
+      "step": 40600
+    },
+    {
+      "epoch": 0.5470577166052851,
+      "grad_norm": 0.7897889018058777,
+      "learning_rate": 0.004997591826993861,
+      "loss": 2.6526,
+      "step": 40700
+    },
+    {
+      "epoch": 0.5484018387591064,
+      "grad_norm": 0.5583525896072388,
+      "learning_rate": 0.004997568507209492,
+      "loss": 2.6488,
+      "step": 40800
+    },
+    {
+      "epoch": 0.5497459609129277,
+      "grad_norm": 1.2979474067687988,
+      "learning_rate": 0.004997545075113834,
+      "loss": 2.6439,
+      "step": 40900
+    },
+    {
+      "epoch": 0.5510900830667491,
+      "grad_norm": 2.09676456451416,
+      "learning_rate": 0.0049975215307079355,
+      "loss": 2.656,
+      "step": 41000
+    },
+    {
+      "epoch": 0.5510900830667491,
+      "eval_MaskedAccuracy": 0.4157399235251504,
+      "eval_loss": 2.981074571609497,
+      "eval_runtime": 154.1369,
+      "eval_samples_per_second": 411.816,
+      "eval_steps_per_second": 1.609,
+      "step": 41000
+    },
+    {
+      "epoch": 0.5524342052205704,
+      "grad_norm": 3.9577839374542236,
+      "learning_rate": 0.0049974978739928694,
+      "loss": 2.6505,
+      "step": 41100
+    },
+    {
+      "epoch": 0.5537783273743918,
+      "grad_norm": 0.5468003153800964,
+      "learning_rate": 0.004997474104969696,
+      "loss": 2.6442,
+      "step": 41200
+    },
+    {
+      "epoch": 0.5551224495282131,
+      "grad_norm": 0.5739485621452332,
+      "learning_rate": 0.004997450223639483,
+      "loss": 2.6446,
+      "step": 41300
+    },
+    {
+      "epoch": 0.5564665716820345,
+      "grad_norm": 1.8400816917419434,
+      "learning_rate": 0.00499742623000331,
+      "loss": 2.6448,
+      "step": 41400
+    },
+    {
+      "epoch": 0.5578106938358558,
+      "grad_norm": 1.2535688877105713,
+      "learning_rate": 0.004997402124062256,
+      "loss": 2.6482,
+      "step": 41500
+    },
+    {
+      "epoch": 0.5591548159896771,
+      "grad_norm": 1.228569746017456,
+      "learning_rate": 0.0049973779058174106,
+      "loss": 2.6498,
+      "step": 41600
+    },
+    {
+      "epoch": 0.5604989381434985,
+      "grad_norm": 1.4011684656143188,
+      "learning_rate": 0.004997353575269862,
+      "loss": 2.643,
+      "step": 41700
+    },
+    {
+      "epoch": 0.5618430602973198,
+      "grad_norm": 0.5663438439369202,
+      "learning_rate": 0.004997329132420707,
+      "loss": 2.6515,
+      "step": 41800
+    },
+    {
+      "epoch": 0.5631871824511412,
+      "grad_norm": 0.5198083519935608,
+      "learning_rate": 0.004997304577271046,
+      "loss": 2.6529,
+      "step": 41900
+    },
+    {
+      "epoch": 0.5645313046049625,
+      "grad_norm": 3.9535000324249268,
+      "learning_rate": 0.004997279909821991,
+      "loss": 2.6393,
+      "step": 42000
+    },
+    {
+      "epoch": 0.5645313046049625,
+      "eval_MaskedAccuracy": 0.4173549594387652,
+      "eval_loss": 2.971799850463867,
+      "eval_runtime": 155.0862,
+      "eval_samples_per_second": 409.295,
+      "eval_steps_per_second": 1.599,
+      "step": 42000
+    },
+    {
+      "epoch": 0.5658754267587839,
+      "grad_norm": 2.1890416145324707,
+      "learning_rate": 0.004997255130074655,
+      "loss": 2.6495,
+      "step": 42100
+    },
+    {
+      "epoch": 0.5672195489126052,
+      "grad_norm": 1.686848521232605,
+      "learning_rate": 0.004997230238030147,
+      "loss": 2.6365,
+      "step": 42200
+    },
+    {
+      "epoch": 0.5685636710664265,
+      "grad_norm": 1.8088412284851074,
+      "learning_rate": 0.004997205233689578,
+      "loss": 2.6421,
+      "step": 42300
+    },
+    {
+      "epoch": 0.5699077932202479,
+      "grad_norm": 1.7008030414581299,
+      "learning_rate": 0.004997180117054094,
+      "loss": 2.6371,
+      "step": 42400
+    },
+    {
+      "epoch": 0.5712519153740692,
+      "grad_norm": 1.355776309967041,
+      "learning_rate": 0.0049971548881248195,
+      "loss": 2.6357,
+      "step": 42500
+    },
+    {
+      "epoch": 0.5725960375278906,
+      "grad_norm": 1.5508376359939575,
+      "learning_rate": 0.004997129546902893,
+      "loss": 2.6438,
+      "step": 42600
+    },
+    {
+      "epoch": 0.5739401596817119,
+      "grad_norm": 0.7327139377593994,
+      "learning_rate": 0.004997104093389453,
+      "loss": 2.6318,
+      "step": 42700
+    },
+    {
+      "epoch": 0.5752842818355333,
+      "grad_norm": 0.8198758959770203,
+      "learning_rate": 0.004997078527585654,
+      "loss": 2.6407,
+      "step": 42800
+    },
+    {
+      "epoch": 0.5766284039893546,
+      "grad_norm": 1.41826331615448,
+      "learning_rate": 0.004997052849492641,
+      "loss": 2.6364,
+      "step": 42900
+    },
+    {
+      "epoch": 0.5779725261431758,
+      "grad_norm": 0.8289880156517029,
+      "learning_rate": 0.0049970270591115735,
+      "loss": 2.6494,
+      "step": 43000
+    },
+    {
+      "epoch": 0.5779725261431758,
+      "eval_MaskedAccuracy": 0.41913729394217447,
+      "eval_loss": 2.955897092819214,
+      "eval_runtime": 154.7124,
+      "eval_samples_per_second": 410.284,
+      "eval_steps_per_second": 1.603,
+      "step": 43000
+    },
+    {
+      "epoch": 0.5793166482969972,
+      "grad_norm": 1.136483907699585,
+      "learning_rate": 0.0049970011564436055,
+      "loss": 2.6364,
+      "step": 43100
+    },
+    {
+      "epoch": 0.5806607704508185,
+      "grad_norm": 1.627422571182251,
+      "learning_rate": 0.004996975141489916,
+      "loss": 2.6424,
+      "step": 43200
+    },
+    {
+      "epoch": 0.58200489260464,
+      "grad_norm": 1.4853564500808716,
+      "learning_rate": 0.004996949014251676,
+      "loss": 2.6462,
+      "step": 43300
+    },
+    {
+      "epoch": 0.5833490147584612,
+      "grad_norm": 1.7882176637649536,
+      "learning_rate": 0.0049969227747300535,
+      "loss": 2.6278,
+      "step": 43400
+    },
+    {
+      "epoch": 0.5846931369122826,
+      "grad_norm": 2.825515031814575,
+      "learning_rate": 0.004996896422926242,
+      "loss": 2.6412,
+      "step": 43500
+    },
+    {
+      "epoch": 0.5860372590661039,
+      "grad_norm": 0.4745466709136963,
+      "learning_rate": 0.004996869958841418,
+      "loss": 2.6332,
+      "step": 43600
+    },
+    {
+      "epoch": 0.5873813812199252,
+      "grad_norm": 1.2661569118499756,
+      "learning_rate": 0.00499684338247678,
+      "loss": 2.6325,
+      "step": 43700
+    },
+    {
+      "epoch": 0.5887255033737466,
+      "grad_norm": 1.2799876928329468,
+      "learning_rate": 0.004996816693833525,
+      "loss": 2.6413,
+      "step": 43800
+    },
+    {
+      "epoch": 0.5900696255275679,
+      "grad_norm": 0.7999758720397949,
+      "learning_rate": 0.004996789892912855,
+      "loss": 2.6256,
+      "step": 43900
+    },
+    {
+      "epoch": 0.5914137476813893,
+      "grad_norm": 0.7396803498268127,
+      "learning_rate": 0.004996762979715977,
+      "loss": 2.6313,
+      "step": 44000
+    },
+    {
+      "epoch": 0.5914137476813893,
+      "eval_MaskedAccuracy": 0.41881557303141803,
+      "eval_loss": 2.9574387073516846,
+      "eval_runtime": 156.1034,
+      "eval_samples_per_second": 406.628,
+      "eval_steps_per_second": 1.589,
+      "step": 44000
+    },
+    {
+      "epoch": 0.5927578698352106,
+      "grad_norm": 1.06534743309021,
+      "learning_rate": 0.00499673595424411,
+      "loss": 2.6311,
+      "step": 44100
+    },
+    {
+      "epoch": 0.5941019919890319,
+      "grad_norm": 1.120554804801941,
+      "learning_rate": 0.004996708816498461,
+      "loss": 2.6289,
+      "step": 44200
+    },
+    {
+      "epoch": 0.5954461141428533,
+      "grad_norm": 0.5264089703559875,
+      "learning_rate": 0.004996681566480261,
+      "loss": 2.6324,
+      "step": 44300
+    },
+    {
+      "epoch": 0.5967902362966746,
+      "grad_norm": 2.0513694286346436,
+      "learning_rate": 0.004996654204190736,
+      "loss": 2.6346,
+      "step": 44400
+    },
+    {
+      "epoch": 0.598134358450496,
+      "grad_norm": 1.3531913757324219,
+      "learning_rate": 0.004996626729631115,
+      "loss": 2.6351,
+      "step": 44500
+    },
+    {
+      "epoch": 0.5994784806043173,
+      "grad_norm": 0.9898929595947266,
+      "learning_rate": 0.004996599142802638,
+      "loss": 2.6232,
+      "step": 44600
+    },
+    {
+      "epoch": 0.6008226027581387,
+      "grad_norm": 0.674278736114502,
+      "learning_rate": 0.00499657144370655,
+      "loss": 2.6227,
+      "step": 44700
+    },
+    {
+      "epoch": 0.60216672491196,
+      "grad_norm": 0.5685105323791504,
+      "learning_rate": 0.004996543632344098,
+      "loss": 2.6306,
+      "step": 44800
+    },
+    {
+      "epoch": 0.6035108470657813,
+      "grad_norm": 0.5396636724472046,
+      "learning_rate": 0.004996515708716534,
+      "loss": 2.6351,
+      "step": 44900
+    },
+    {
+      "epoch": 0.6048549692196027,
+      "grad_norm": 1.0353196859359741,
+      "learning_rate": 0.004996487672825116,
+      "loss": 2.6332,
+      "step": 45000
+    },
+    {
+      "epoch": 0.6048549692196027,
+      "eval_MaskedAccuracy": 0.4208831607484112,
+      "eval_loss": 2.9444289207458496,
+      "eval_runtime": 157.2902,
+      "eval_samples_per_second": 403.56,
+      "eval_steps_per_second": 1.577,
+      "step": 45000
+    },
+    {
+      "epoch": 0.606199091373424,
+      "grad_norm": 0.9090419411659241,
+      "learning_rate": 0.004996459524671107,
+      "loss": 2.6303,
+      "step": 45100
+    },
+    {
+      "epoch": 0.6075432135272454,
+      "grad_norm": 0.5557056069374084,
+      "learning_rate": 0.004996431264255778,
+      "loss": 2.6253,
+      "step": 45200
+    },
+    {
+      "epoch": 0.6088873356810667,
+      "grad_norm": 1.1961573362350464,
+      "learning_rate": 0.004996402891580404,
+      "loss": 2.6249,
+      "step": 45300
+    },
+    {
+      "epoch": 0.6102314578348881,
+      "grad_norm": 0.7062890529632568,
+      "learning_rate": 0.00499637440664626,
+      "loss": 2.6264,
+      "step": 45400
+    },
+    {
+      "epoch": 0.6115755799887094,
+      "grad_norm": 0.7188198566436768,
+      "learning_rate": 0.004996345809454634,
+      "loss": 2.63,
+      "step": 45500
+    },
+    {
+      "epoch": 0.6129197021425307,
+      "grad_norm": 1.2873228788375854,
+      "learning_rate": 0.0049963171000068085,
+      "loss": 2.6234,
+      "step": 45600
+    },
+    {
+      "epoch": 0.6142638242963521,
+      "grad_norm": 1.0937819480895996,
+      "learning_rate": 0.004996288278304078,
+      "loss": 2.6309,
+      "step": 45700
+    },
+    {
+      "epoch": 0.6156079464501734,
+      "grad_norm": 0.9512178897857666,
+      "learning_rate": 0.004996259344347743,
+      "loss": 2.6194,
+      "step": 45800
+    },
+    {
+      "epoch": 0.6169520686039948,
+      "grad_norm": 0.4616401791572571,
+      "learning_rate": 0.004996230298139107,
+      "loss": 2.6233,
+      "step": 45900
+    },
+    {
+      "epoch": 0.618296190757816,
+      "grad_norm": 0.6116599440574646,
+      "learning_rate": 0.004996201139679473,
+      "loss": 2.6205,
+      "step": 46000
+    },
+    {
+      "epoch": 0.618296190757816,
+      "eval_MaskedAccuracy": 0.4190083845851312,
+      "eval_loss": 2.9517083168029785,
+      "eval_runtime": 142.8409,
+      "eval_samples_per_second": 444.382,
+      "eval_steps_per_second": 1.736,
+      "step": 46000
+    },
+    {
+      "epoch": 0.6196403129116375,
+      "grad_norm": 1.2493271827697754,
+      "learning_rate": 0.004996171868970165,
+      "loss": 2.6293,
+      "step": 46100
+    },
+    {
+      "epoch": 0.6209844350654588,
+      "grad_norm": 1.9406765699386597,
+      "learning_rate": 0.004996142486012493,
+      "loss": 2.626,
+      "step": 46200
+    },
+    {
+      "epoch": 0.62232855721928,
+      "grad_norm": 0.6907238960266113,
+      "learning_rate": 0.004996112990807786,
+      "loss": 2.6141,
+      "step": 46300
+    },
+    {
+      "epoch": 0.6236726793731014,
+      "grad_norm": 0.5618690848350525,
+      "learning_rate": 0.0049960833833573755,
+      "loss": 2.6151,
+      "step": 46400
+    },
+    {
+      "epoch": 0.6250168015269227,
+      "grad_norm": 1.8456848859786987,
+      "learning_rate": 0.004996053663662593,
+      "loss": 2.6158,
+      "step": 46500
+    },
+    {
+      "epoch": 0.6263609236807441,
+      "grad_norm": 0.4822538495063782,
+      "learning_rate": 0.004996023831724768,
+      "loss": 2.6219,
+      "step": 46600
+    },
+    {
+      "epoch": 0.6277050458345654,
+      "grad_norm": 0.4805866777896881,
+      "learning_rate": 0.004995993887545261,
+      "loss": 2.6134,
+      "step": 46700
+    },
+    {
+      "epoch": 0.6290491679883868,
+      "grad_norm": 1.5017101764678955,
+      "learning_rate": 0.004995963831125411,
+      "loss": 2.6146,
+      "step": 46800
+    },
+    {
+      "epoch": 0.6303932901422081,
+      "grad_norm": 1.2550396919250488,
+      "learning_rate": 0.00499593366246658,
+      "loss": 2.6176,
+      "step": 46900
+    },
+    {
+      "epoch": 0.6317374122960294,
+      "grad_norm": 1.6678540706634521,
+      "learning_rate": 0.004995903381570121,
+      "loss": 2.6095,
+      "step": 47000
+    },
+    {
+      "epoch": 0.6317374122960294,
+      "eval_MaskedAccuracy": 0.42101418324632395,
+      "eval_loss": 2.939000129699707,
+      "eval_runtime": 157.4312,
+      "eval_samples_per_second": 403.198,
+      "eval_steps_per_second": 1.575,
+      "step": 47000
+    },
+    {
+      "epoch": 0.6330815344498508,
+      "grad_norm": 0.7519531846046448,
+      "learning_rate": 0.004995872988437397,
+      "loss": 2.6146,
+      "step": 47100
+    },
+    {
+      "epoch": 0.6344256566036721,
+      "grad_norm": 0.5543345808982849,
+      "learning_rate": 0.004995842483069782,
+      "loss": 2.6165,
+      "step": 47200
+    },
+    {
+      "epoch": 0.6357697787574935,
+      "grad_norm": 0.8774918913841248,
+      "learning_rate": 0.004995811865468641,
+      "loss": 2.617,
+      "step": 47300
+    },
+    {
+      "epoch": 0.6371139009113148,
+      "grad_norm": 0.5010842680931091,
+      "learning_rate": 0.004995781135635373,
+      "loss": 2.622,
+      "step": 47400
+    },
+    {
+      "epoch": 0.6384580230651361,
+      "grad_norm": 0.537598192691803,
+      "learning_rate": 0.0049957502935713445,
+      "loss": 2.6228,
+      "step": 47500
+    },
+    {
+      "epoch": 0.6398021452189575,
+      "grad_norm": 0.65438312292099,
+      "learning_rate": 0.004995719339277949,
+      "loss": 2.6154,
+      "step": 47600
+    },
+    {
+      "epoch": 0.6411462673727788,
+      "grad_norm": 0.726901113986969,
+      "learning_rate": 0.004995688272756588,
+      "loss": 2.6126,
+      "step": 47700
+    },
+    {
+      "epoch": 0.6424903895266002,
+      "grad_norm": 1.2524980306625366,
+      "learning_rate": 0.004995657094008657,
+      "loss": 2.6075,
+      "step": 47800
+    },
+    {
+      "epoch": 0.6438345116804215,
+      "grad_norm": 2.223083257675171,
+      "learning_rate": 0.004995625803035562,
+      "loss": 2.6113,
+      "step": 47900
+    },
+    {
+      "epoch": 0.6451786338342429,
+      "grad_norm": 0.7421071529388428,
+      "learning_rate": 0.004995594399838711,
+      "loss": 2.6088,
+      "step": 48000
+    },
+    {
+      "epoch": 0.6451786338342429,
+      "eval_MaskedAccuracy": 0.42257422566160907,
+      "eval_loss": 2.9292173385620117,
+      "eval_runtime": 154.814,
+      "eval_samples_per_second": 410.015,
+      "eval_steps_per_second": 1.602,
+      "step": 48000
+    },
+    {
+      "epoch": 0.6465227559880642,
+      "grad_norm": 0.6506995558738708,
+      "learning_rate": 0.004995562884419525,
+      "loss": 2.6105,
+      "step": 48100
+    },
+    {
+      "epoch": 0.6478668781418855,
+      "grad_norm": 1.0766329765319824,
+      "learning_rate": 0.004995531256779415,
+      "loss": 2.6139,
+      "step": 48200
+    },
+    {
+      "epoch": 0.6492110002957069,
+      "grad_norm": 1.1762914657592773,
+      "learning_rate": 0.00499549951691981,
+      "loss": 2.6078,
+      "step": 48300
+    },
+    {
+      "epoch": 0.6505551224495282,
+      "grad_norm": 2.2142832279205322,
+      "learning_rate": 0.004995467664842145,
+      "loss": 2.6093,
+      "step": 48400
+    },
+    {
+      "epoch": 0.6518992446033496,
+      "grad_norm": 1.1189358234405518,
+      "learning_rate": 0.00499543570054785,
+      "loss": 2.6141,
+      "step": 48500
+    },
+    {
+      "epoch": 0.6532433667571709,
+      "grad_norm": 0.7989232540130615,
+      "learning_rate": 0.004995403624038362,
+      "loss": 2.6155,
+      "step": 48600
+    },
+    {
+      "epoch": 0.6545874889109923,
+      "grad_norm": 1.5602754354476929,
+      "learning_rate": 0.004995371435315131,
+      "loss": 2.6137,
+      "step": 48700
+    },
+    {
+      "epoch": 0.6559316110648136,
+      "grad_norm": 0.7082908749580383,
+      "learning_rate": 0.004995339134379602,
+      "loss": 2.6138,
+      "step": 48800
+    },
+    {
+      "epoch": 0.6572757332186349,
+      "grad_norm": 1.3697164058685303,
+      "learning_rate": 0.004995306721233229,
+      "loss": 2.6094,
+      "step": 48900
+    },
+    {
+      "epoch": 0.6586198553724563,
+      "grad_norm": 0.5310668349266052,
+      "learning_rate": 0.004995274195877492,
+      "loss": 2.6069,
+      "step": 49000
+    },
+    {
+      "epoch": 0.6586198553724563,
+      "eval_MaskedAccuracy": 0.42236824728278766,
+      "eval_loss": 2.928412437438965,
+      "eval_runtime": 155.9626,
+      "eval_samples_per_second": 406.995,
+      "eval_steps_per_second": 1.59,
+      "step": 49000
+    },
+    {
+      "epoch": 0.6599639775262776,
+      "grad_norm": 1.284600019454956,
+      "learning_rate": 0.00499524155831384,
+      "loss": 2.6058,
+      "step": 49100
+    },
+    {
+      "epoch": 0.661308099680099,
+      "grad_norm": 0.44148510694503784,
+      "learning_rate": 0.004995208808543738,
+      "loss": 2.6046,
+      "step": 49200
+    },
+    {
+      "epoch": 0.6626522218339203,
+      "grad_norm": 1.5299873352050781,
+      "learning_rate": 0.004995175946568674,
+      "loss": 2.6057,
+      "step": 49300
+    },
+    {
+      "epoch": 0.6639963439877417,
+      "grad_norm": 0.8762273192405701,
+      "learning_rate": 0.00499514297239013,
+      "loss": 2.6049,
+      "step": 49400
+    },
+    {
+      "epoch": 0.665340466141563,
+      "grad_norm": 0.6929760575294495,
+      "learning_rate": 0.004995109886009586,
+      "loss": 2.6117,
+      "step": 49500
+    },
+    {
+      "epoch": 0.6666845882953842,
+      "grad_norm": 0.9134759306907654,
+      "learning_rate": 0.004995076687428528,
+      "loss": 2.6008,
+      "step": 49600
+    },
+    {
+      "epoch": 0.6680287104492056,
+      "grad_norm": 1.4550007581710815,
+      "learning_rate": 0.004995043376648459,
+      "loss": 2.6087,
+      "step": 49700
+    },
+    {
+      "epoch": 0.6693728326030269,
+      "grad_norm": 1.1609247922897339,
+      "learning_rate": 0.004995009953670882,
+      "loss": 2.6043,
+      "step": 49800
+    },
+    {
+      "epoch": 0.6707169547568483,
+      "grad_norm": 0.5234310030937195,
+      "learning_rate": 0.004994976418497298,
+      "loss": 2.6068,
+      "step": 49900
+    },
+    {
+      "epoch": 0.6720610769106696,
+      "grad_norm": 1.2967578172683716,
+      "learning_rate": 0.004994942771129218,
+      "loss": 2.6029,
+      "step": 50000
+    },
+    {
+      "epoch": 0.6720610769106696,
+      "eval_MaskedAccuracy": 0.4233113243940233,
+      "eval_loss": 2.9228169918060303,
+      "eval_runtime": 161.4288,
+      "eval_samples_per_second": 393.213,
+      "eval_steps_per_second": 1.536,
+      "step": 50000
+    },
+    {
+      "epoch": 0.673405199064491,
+      "grad_norm": 1.6317380666732788,
+      "learning_rate": 0.00499490901156816,
+      "loss": 2.6123,
+      "step": 50100
+    },
+    {
+      "epoch": 0.6747493212183123,
+      "grad_norm": 0.8617536425590515,
+      "learning_rate": 0.004994875139815648,
+      "loss": 2.5964,
+      "step": 50200
+    },
+    {
+      "epoch": 0.6760934433721336,
+      "grad_norm": 1.1220712661743164,
+      "learning_rate": 0.004994841155873204,
+      "loss": 2.5989,
+      "step": 50300
+    },
+    {
+      "epoch": 0.677437565525955,
+      "grad_norm": 0.5247998237609863,
+      "learning_rate": 0.004994807059742355,
+      "loss": 2.598,
+      "step": 50400
+    },
+    {
+      "epoch": 0.6787816876797763,
+      "grad_norm": 1.532121181488037,
+      "learning_rate": 0.004994772851424643,
+      "loss": 2.5946,
+      "step": 50500
+    },
+    {
+      "epoch": 0.6801258098335977,
+      "grad_norm": 1.0279977321624756,
+      "learning_rate": 0.004994738530921611,
+      "loss": 2.6086,
+      "step": 50600
+    },
+    {
+      "epoch": 0.681469931987419,
+      "grad_norm": 0.8515674471855164,
+      "learning_rate": 0.0049947040982348005,
+      "loss": 2.5934,
+      "step": 50700
+    },
+    {
+      "epoch": 0.6828140541412403,
+      "grad_norm": 0.958196222782135,
+      "learning_rate": 0.004994669553365766,
+      "loss": 2.597,
+      "step": 50800
+    },
+    {
+      "epoch": 0.6841581762950617,
+      "grad_norm": 0.795566737651825,
+      "learning_rate": 0.0049946348963160634,
+      "loss": 2.6028,
+      "step": 50900
+    },
+    {
+      "epoch": 0.685502298448883,
+      "grad_norm": 1.396265983581543,
+      "learning_rate": 0.0049946001270872575,
+      "loss": 2.5972,
+      "step": 51000
+    },
+    {
+      "epoch": 0.685502298448883,
+      "eval_MaskedAccuracy": 0.42380398111438183,
+      "eval_loss": 2.9168331623077393,
+      "eval_runtime": 155.1031,
+      "eval_samples_per_second": 409.25,
+      "eval_steps_per_second": 1.599,
+      "step": 51000
+    },
+    {
+      "epoch": 0.6868464206027044,
+      "grad_norm": 1.0055732727050781,
+      "learning_rate": 0.004994565245680906,
+      "loss": 2.5967,
+      "step": 51100
+    },
+    {
+      "epoch": 0.6881905427565257,
+      "grad_norm": 1.37259840965271,
+      "learning_rate": 0.004994530252098587,
+      "loss": 2.6036,
+      "step": 51200
+    },
+    {
+      "epoch": 0.6895346649103471,
+      "grad_norm": 0.4619656801223755,
+      "learning_rate": 0.00499449514634188,
+      "loss": 2.6019,
+      "step": 51300
+    },
+    {
+      "epoch": 0.6908787870641684,
+      "grad_norm": 0.7385687828063965,
+      "learning_rate": 0.004994459928412364,
+      "loss": 2.6004,
+      "step": 51400
+    },
+    {
+      "epoch": 0.6922229092179897,
+      "grad_norm": 1.73374342918396,
+      "learning_rate": 0.004994424598311631,
+      "loss": 2.592,
+      "step": 51500
+    },
+    {
+      "epoch": 0.6935670313718111,
+      "grad_norm": 1.9898829460144043,
+      "learning_rate": 0.00499438915604126,
+      "loss": 2.5961,
+      "step": 51600
+    },
+    {
+      "epoch": 0.6949111535256324,
+      "grad_norm": 1.4267821311950684,
+      "learning_rate": 0.00499435360160286,
+      "loss": 2.5896,
+      "step": 51700
+    },
+    {
+      "epoch": 0.6962552756794538,
+      "grad_norm": 1.4741652011871338,
+      "learning_rate": 0.004994317934998031,
+      "loss": 2.5957,
+      "step": 51800
+    },
+    {
+      "epoch": 0.6975993978332751,
+      "grad_norm": 1.60403573513031,
+      "learning_rate": 0.004994282156228371,
+      "loss": 2.5936,
+      "step": 51900
+    },
+    {
+      "epoch": 0.6989435199870965,
+      "grad_norm": 1.172230839729309,
+      "learning_rate": 0.004994246265295497,
+      "loss": 2.5973,
+      "step": 52000
+    },
+    {
+      "epoch": 0.6989435199870965,
+      "eval_MaskedAccuracy": 0.42523802061668947,
+      "eval_loss": 2.911010265350342,
+      "eval_runtime": 154.888,
+      "eval_samples_per_second": 409.819,
+      "eval_steps_per_second": 1.601,
+      "step": 52000
+    },
+    {
+      "epoch": 0.7002876421409178,
+      "grad_norm": 1.7461953163146973,
+      "learning_rate": 0.004994210262201032,
+      "loss": 2.5981,
+      "step": 52100
+    },
+    {
+      "epoch": 0.7016317642947391,
+      "grad_norm": 1.0520185232162476,
+      "learning_rate": 0.004994174146946593,
+      "loss": 2.5931,
+      "step": 52200
+    },
+    {
+      "epoch": 0.7029758864485605,
+      "grad_norm": 0.6437071561813354,
+      "learning_rate": 0.0049941379195338085,
+      "loss": 2.5871,
+      "step": 52300
+    },
+    {
+      "epoch": 0.7043200086023818,
+      "grad_norm": 0.5879480838775635,
+      "learning_rate": 0.004994101579964312,
+      "loss": 2.5895,
+      "step": 52400
+    },
+    {
+      "epoch": 0.7056641307562032,
+      "grad_norm": 0.4580605626106262,
+      "learning_rate": 0.00499406512823974,
+      "loss": 2.5958,
+      "step": 52500
+    },
+    {
+      "epoch": 0.7070082529100244,
+      "grad_norm": 1.544168472290039,
+      "learning_rate": 0.004994028564361737,
+      "loss": 2.5952,
+      "step": 52600
+    },
+    {
+      "epoch": 0.7083523750638459,
+      "grad_norm": 2.1680054664611816,
+      "learning_rate": 0.004993991888331947,
+      "loss": 2.5939,
+      "step": 52700
+    },
+    {
+      "epoch": 0.7096964972176671,
+      "grad_norm": 0.9118250608444214,
+      "learning_rate": 0.004993955100152014,
+      "loss": 2.5873,
+      "step": 52800
+    },
+    {
+      "epoch": 0.7110406193714884,
+      "grad_norm": 0.6249663233757019,
+      "learning_rate": 0.004993918199823611,
+      "loss": 2.5945,
+      "step": 52900
+    },
+    {
+      "epoch": 0.7123847415253098,
+      "grad_norm": 0.5641465783119202,
+      "learning_rate": 0.004993881187348401,
+      "loss": 2.5875,
+      "step": 53000
+    },
+    {
+      "epoch": 0.7123847415253098,
+      "eval_MaskedAccuracy": 0.42511639864491074,
+      "eval_loss": 2.9080851078033447,
+      "eval_runtime": 155.0598,
+      "eval_samples_per_second": 409.365,
+      "eval_steps_per_second": 1.599,
+      "step": 53000
+    },
+    {
+      "epoch": 0.7137288636791311,
+      "grad_norm": 1.5355523824691772,
+      "learning_rate": 0.004993844062728042,
+      "loss": 2.586,
+      "step": 53100
+    },
+    {
+      "epoch": 0.7150729858329525,
+      "grad_norm": 0.9230164885520935,
+      "learning_rate": 0.004993806825964211,
+      "loss": 2.5855,
+      "step": 53200
+    },
+    {
+      "epoch": 0.7164171079867738,
+      "grad_norm": 1.638444185256958,
+      "learning_rate": 0.004993769477058582,
+      "loss": 2.5989,
+      "step": 53300
+    },
+    {
+      "epoch": 0.7177612301405952,
+      "grad_norm": 0.5577353835105896,
+      "learning_rate": 0.004993732016012843,
+      "loss": 2.5967,
+      "step": 53400
+    },
+    {
+      "epoch": 0.7191053522944165,
+      "grad_norm": 0.5781148672103882,
+      "learning_rate": 0.0049936944428286825,
+      "loss": 2.5923,
+      "step": 53500
+    },
+    {
+      "epoch": 0.7204494744482378,
+      "grad_norm": 1.6455879211425781,
+      "learning_rate": 0.00499365675750779,
+      "loss": 2.5868,
+      "step": 53600
+    },
+    {
+      "epoch": 0.7217935966020592,
+      "grad_norm": 0.6370722651481628,
+      "learning_rate": 0.004993618960051859,
+      "loss": 2.5879,
+      "step": 53700
+    },
+    {
+      "epoch": 0.7231377187558805,
+      "grad_norm": 1.1286998987197876,
+      "learning_rate": 0.004993581050462601,
+      "loss": 2.6016,
+      "step": 53800
+    },
+    {
+      "epoch": 0.7244818409097019,
+      "grad_norm": 1.282611608505249,
+      "learning_rate": 0.004993543028741726,
+      "loss": 2.5869,
+      "step": 53900
+    },
+    {
+      "epoch": 0.7258259630635232,
+      "grad_norm": 0.51445072889328,
+      "learning_rate": 0.004993504894890944,
+      "loss": 2.5856,
+      "step": 54000
+    },
+    {
+      "epoch": 0.7258259630635232,
+      "eval_MaskedAccuracy": 0.4248532700853216,
+      "eval_loss": 2.909213066101074,
+      "eval_runtime": 154.7862,
+      "eval_samples_per_second": 410.088,
+      "eval_steps_per_second": 1.602,
+      "step": 54000
+    },
+    {
+      "epoch": 0.7271700852173445,
+      "grad_norm": 1.3245320320129395,
+      "learning_rate": 0.004993466648911963,
+      "loss": 2.5915,
+      "step": 54100
+    },
+    {
+      "epoch": 0.7285142073711659,
+      "grad_norm": 1.215162754058838,
+      "learning_rate": 0.004993428290806523,
+      "loss": 2.5772,
+      "step": 54200
+    },
+    {
+      "epoch": 0.7298583295249872,
+      "grad_norm": 1.2432647943496704,
+      "learning_rate": 0.004993389820576344,
+      "loss": 2.5902,
+      "step": 54300
+    },
+    {
+      "epoch": 0.7312024516788086,
+      "grad_norm": 1.581764817237854,
+      "learning_rate": 0.0049933512382231565,
+      "loss": 2.5795,
+      "step": 54400
+    },
+    {
+      "epoch": 0.7325465738326299,
+      "grad_norm": 0.9531721472740173,
+      "learning_rate": 0.004993312543748701,
+      "loss": 2.5764,
+      "step": 54500
+    },
+    {
+      "epoch": 0.7338906959864513,
+      "grad_norm": 2.677894115447998,
+      "learning_rate": 0.004993273737154721,
+      "loss": 2.5922,
+      "step": 54600
+    },
+    {
+      "epoch": 0.7352348181402726,
+      "grad_norm": 0.793094277381897,
+      "learning_rate": 0.00499323481844297,
+      "loss": 2.5841,
+      "step": 54700
+    },
+    {
+      "epoch": 0.7365789402940939,
+      "grad_norm": 1.0504934787750244,
+      "learning_rate": 0.0049931957876151924,
+      "loss": 2.5831,
+      "step": 54800
+    },
+    {
+      "epoch": 0.7379230624479153,
+      "grad_norm": 1.0326224565505981,
+      "learning_rate": 0.004993156644673158,
+      "loss": 2.5774,
+      "step": 54900
+    },
+    {
+      "epoch": 0.7392671846017366,
+      "grad_norm": 0.7911434173583984,
+      "learning_rate": 0.004993117389618626,
+      "loss": 2.5821,
+      "step": 55000
+    },
+    {
+      "epoch": 0.7392671846017366,
+      "eval_MaskedAccuracy": 0.42632109483323904,
+      "eval_loss": 2.8994028568267822,
+      "eval_runtime": 160.7388,
+      "eval_samples_per_second": 394.902,
+      "eval_steps_per_second": 1.543,
+      "step": 55000
+    },
+    {
+      "epoch": 0.740611306755558,
+      "grad_norm": 0.5436133742332458,
+      "learning_rate": 0.004993078022453364,
+      "loss": 2.5816,
+      "step": 55100
+    },
+    {
+      "epoch": 0.7419554289093793,
+      "grad_norm": 0.8503451943397522,
+      "learning_rate": 0.004993038543179143,
+      "loss": 2.5722,
+      "step": 55200
+    },
+    {
+      "epoch": 0.7432995510632007,
+      "grad_norm": 1.585143804550171,
+      "learning_rate": 0.004992998951797748,
+      "loss": 2.5707,
+      "step": 55300
+    },
+    {
+      "epoch": 0.744643673217022,
+      "grad_norm": 2.5457680225372314,
+      "learning_rate": 0.0049929592483109585,
+      "loss": 2.5768,
+      "step": 55400
+    },
+    {
+      "epoch": 0.7459877953708433,
+      "grad_norm": 1.12667977809906,
+      "learning_rate": 0.004992919432720564,
+      "loss": 2.5813,
+      "step": 55500
+    },
+    {
+      "epoch": 0.7473319175246647,
+      "grad_norm": 0.685434103012085,
+      "learning_rate": 0.00499287950502836,
+      "loss": 2.5824,
+      "step": 55600
+    },
+    {
+      "epoch": 0.748676039678486,
+      "grad_norm": 0.5508724451065063,
+      "learning_rate": 0.0049928394652361445,
+      "loss": 2.5788,
+      "step": 55700
+    },
+    {
+      "epoch": 0.7500201618323074,
+      "grad_norm": 1.0351439714431763,
+      "learning_rate": 0.004992799313345724,
+      "loss": 2.5811,
+      "step": 55800
+    },
+    {
+      "epoch": 0.7513642839861286,
+      "grad_norm": 0.4489949345588684,
+      "learning_rate": 0.004992759049358903,
+      "loss": 2.5788,
+      "step": 55900
+    },
+    {
+      "epoch": 0.75270840613995,
+      "grad_norm": 0.7552504539489746,
+      "learning_rate": 0.0049927186732775025,
+      "loss": 2.5738,
+      "step": 56000
+    },
+    {
+      "epoch": 0.75270840613995,
+      "eval_MaskedAccuracy": 0.42679037653842317,
+      "eval_loss": 2.8953652381896973,
+      "eval_runtime": 154.4325,
+      "eval_samples_per_second": 411.027,
+      "eval_steps_per_second": 1.606,
+      "step": 56000
+    },
+    {
+      "epoch": 0.7540525282937713,
+      "grad_norm": 1.503426194190979,
+      "learning_rate": 0.0049926781851033376,
+      "loss": 2.5817,
+      "step": 56100
+    },
+    {
+      "epoch": 0.7553966504475926,
+      "grad_norm": 0.7938827872276306,
+      "learning_rate": 0.004992637584838231,
+      "loss": 2.5747,
+      "step": 56200
+    },
+    {
+      "epoch": 0.756740772601414,
+      "grad_norm": 0.6982958316802979,
+      "learning_rate": 0.004992596872484014,
+      "loss": 2.5803,
+      "step": 56300
+    },
+    {
+      "epoch": 0.7580848947552353,
+      "grad_norm": 0.5538859963417053,
+      "learning_rate": 0.004992556048042525,
+      "loss": 2.5873,
+      "step": 56400
+    },
+    {
+      "epoch": 0.7594290169090567,
+      "grad_norm": 1.1309832334518433,
+      "learning_rate": 0.004992515111515598,
+      "loss": 2.5732,
+      "step": 56500
+    },
+    {
+      "epoch": 0.760773139062878,
+      "grad_norm": 2.124687671661377,
+      "learning_rate": 0.00499247406290508,
+      "loss": 2.582,
+      "step": 56600
+    },
+    {
+      "epoch": 0.7621172612166994,
+      "grad_norm": 1.6379998922348022,
+      "learning_rate": 0.004992432902212811,
+      "loss": 2.5806,
+      "step": 56700
+    },
+    {
+      "epoch": 0.7634613833705207,
+      "grad_norm": 2.21696400642395,
+      "learning_rate": 0.004992391629440661,
+      "loss": 2.5809,
+      "step": 56800
+    },
+    {
+      "epoch": 0.764805505524342,
+      "grad_norm": 0.6755267977714539,
+      "learning_rate": 0.0049923502445904785,
+      "loss": 2.5759,
+      "step": 56900
+    },
+    {
+      "epoch": 0.7661496276781634,
+      "grad_norm": 0.5627455711364746,
+      "learning_rate": 0.004992308747664136,
+      "loss": 2.5784,
+      "step": 57000
+    },
+    {
+      "epoch": 0.7661496276781634,
+      "eval_MaskedAccuracy": 0.42715101212392975,
+      "eval_loss": 2.889655351638794,
+      "eval_runtime": 156.8451,
+      "eval_samples_per_second": 404.705,
+      "eval_steps_per_second": 1.581,
+      "step": 57000
+    },
+    {
+      "epoch": 0.7674937498319847,
+      "grad_norm": 0.4193578064441681,
+      "learning_rate": 0.004992267138663497,
+      "loss": 2.5768,
+      "step": 57100
+    },
+    {
+      "epoch": 0.7688378719858061,
+      "grad_norm": 0.4133063554763794,
+      "learning_rate": 0.004992225417590446,
+      "loss": 2.5712,
+      "step": 57200
+    },
+    {
+      "epoch": 0.7701819941396274,
+      "grad_norm": 1.6166553497314453,
+      "learning_rate": 0.004992183584446851,
+      "loss": 2.5751,
+      "step": 57300
+    },
+    {
+      "epoch": 0.7715261162934487,
+      "grad_norm": 1.9402251243591309,
+      "learning_rate": 0.0049921416392346,
+      "loss": 2.573,
+      "step": 57400
+    },
+    {
+      "epoch": 0.7728702384472701,
+      "grad_norm": 1.0368021726608276,
+      "learning_rate": 0.004992099581955586,
+      "loss": 2.5801,
+      "step": 57500
+    },
+    {
+      "epoch": 0.7742143606010914,
+      "grad_norm": 0.431902676820755,
+      "learning_rate": 0.004992057412611701,
+      "loss": 2.5776,
+      "step": 57600
+    },
+    {
+      "epoch": 0.7755584827549128,
+      "grad_norm": 0.4904190003871918,
+      "learning_rate": 0.004992015131204851,
+      "loss": 2.5792,
+      "step": 57700
+    },
+    {
+      "epoch": 0.7769026049087341,
+      "grad_norm": 1.3063716888427734,
+      "learning_rate": 0.004991972737736936,
+      "loss": 2.5654,
+      "step": 57800
+    },
+    {
+      "epoch": 0.7782467270625555,
+      "grad_norm": 1.1166818141937256,
+      "learning_rate": 0.004991930232209869,
+      "loss": 2.5694,
+      "step": 57900
+    },
+    {
+      "epoch": 0.7795908492163768,
+      "grad_norm": 0.6093880534172058,
+      "learning_rate": 0.004991887614625561,
+      "loss": 2.5768,
+      "step": 58000
+    },
+    {
+      "epoch": 0.7795908492163768,
+      "eval_MaskedAccuracy": 0.42766976413111096,
+      "eval_loss": 2.886188507080078,
+      "eval_runtime": 155.0432,
+      "eval_samples_per_second": 409.408,
+      "eval_steps_per_second": 1.6,
+      "step": 58000
+    },
+    {
+      "epoch": 0.7809349713701981,
+      "grad_norm": 1.4043593406677246,
+      "learning_rate": 0.00499184488498594,
+      "loss": 2.5692,
+      "step": 58100
+    },
+    {
+      "epoch": 0.7822790935240195,
+      "grad_norm": 1.9828555583953857,
+      "learning_rate": 0.0049918020432929274,
+      "loss": 2.5707,
+      "step": 58200
+    },
+    {
+      "epoch": 0.7836232156778408,
+      "grad_norm": 1.0599697828292847,
+      "learning_rate": 0.0049917590895484495,
+      "loss": 2.5715,
+      "step": 58300
+    },
+    {
+      "epoch": 0.7849673378316622,
+      "grad_norm": 0.5353257656097412,
+      "learning_rate": 0.00499171602375444,
+      "loss": 2.5601,
+      "step": 58400
+    },
+    {
+      "epoch": 0.7863114599854835,
+      "grad_norm": 0.45401671528816223,
+      "learning_rate": 0.004991672845912851,
+      "loss": 2.5764,
+      "step": 58500
+    },
+    {
+      "epoch": 0.7876555821393049,
+      "grad_norm": 1.2403732538223267,
+      "learning_rate": 0.00499162955602562,
+      "loss": 2.5637,
+      "step": 58600
+    },
+    {
+      "epoch": 0.7889997042931262,
+      "grad_norm": 1.7403950691223145,
+      "learning_rate": 0.004991586154094695,
+      "loss": 2.5634,
+      "step": 58700
+    },
+    {
+      "epoch": 0.7903438264469475,
+      "grad_norm": 0.8085100650787354,
+      "learning_rate": 0.004991542640122036,
+      "loss": 2.564,
+      "step": 58800
+    },
+    {
+      "epoch": 0.7916879486007689,
+      "grad_norm": 0.4583624005317688,
+      "learning_rate": 0.004991499014109603,
+      "loss": 2.5618,
+      "step": 58900
+    },
+    {
+      "epoch": 0.7930320707545901,
+      "grad_norm": 1.7312582731246948,
+      "learning_rate": 0.004991455276059368,
+      "loss": 2.5712,
+      "step": 59000
+    },
+    {
+      "epoch": 0.7930320707545901,
+      "eval_MaskedAccuracy": 0.4281439150410401,
+      "eval_loss": 2.884404420852661,
+      "eval_runtime": 154.3415,
+      "eval_samples_per_second": 411.27,
+      "eval_steps_per_second": 1.607,
+      "step": 59000
+    },
+    {
+      "epoch": 0.7943761929084115,
+      "grad_norm": 1.1742795705795288,
+      "learning_rate": 0.004991411425973289,
+      "loss": 2.5578,
+      "step": 59100
+    },
+    {
+      "epoch": 0.7957203150622328,
+      "grad_norm": 0.8881747722625732,
+      "learning_rate": 0.004991367463853345,
+      "loss": 2.5723,
+      "step": 59200
+    },
+    {
+      "epoch": 0.7970644372160542,
+      "grad_norm": 1.2915089130401611,
+      "learning_rate": 0.004991323389701527,
+      "loss": 2.565,
+      "step": 59300
+    },
+    {
+      "epoch": 0.7984085593698755,
+      "grad_norm": 0.6965966820716858,
+      "learning_rate": 0.004991279203519816,
+      "loss": 2.5619,
+      "step": 59400
+    },
+    {
+      "epoch": 0.7997526815236968,
+      "grad_norm": 2.0180699825286865,
+      "learning_rate": 0.004991234905310197,
+      "loss": 2.5639,
+      "step": 59500
+    },
+    {
+      "epoch": 0.8010968036775182,
+      "grad_norm": 0.8435118794441223,
+      "learning_rate": 0.004991190495074672,
+      "loss": 2.5578,
+      "step": 59600
+    },
+    {
+      "epoch": 0.8024409258313395,
+      "grad_norm": 1.3183516263961792,
+      "learning_rate": 0.004991145972815238,
+      "loss": 2.5658,
+      "step": 59700
+    },
+    {
+      "epoch": 0.8037850479851609,
+      "grad_norm": 0.4824967384338379,
+      "learning_rate": 0.004991101338533903,
+      "loss": 2.5641,
+      "step": 59800
+    },
+    {
+      "epoch": 0.8051291701389822,
+      "grad_norm": 0.4352557361125946,
+      "learning_rate": 0.004991056592232678,
+      "loss": 2.5634,
+      "step": 59900
+    },
+    {
+      "epoch": 0.8064732922928036,
+      "grad_norm": 0.6158259510993958,
+      "learning_rate": 0.004991011733913582,
+      "loss": 2.5596,
+      "step": 60000
+    },
+    {
+      "epoch": 0.8064732922928036,
+      "eval_MaskedAccuracy": 0.42826808328062316,
+      "eval_loss": 2.881223440170288,
+      "eval_runtime": 157.4024,
+      "eval_samples_per_second": 403.272,
+      "eval_steps_per_second": 1.576,
+      "step": 60000
+    },
+    {
+      "epoch": 0.8078174144466249,
+      "grad_norm": 1.2049239873886108,
+      "learning_rate": 0.004990966763578632,
+      "loss": 2.5647,
+      "step": 60100
+    },
+    {
+      "epoch": 0.8091615366004462,
+      "grad_norm": 1.9667131900787354,
+      "learning_rate": 0.0049909216812298604,
+      "loss": 2.5715,
+      "step": 60200
+    },
+    {
+      "epoch": 0.8105056587542676,
+      "grad_norm": 1.7864271402359009,
+      "learning_rate": 0.004990876486869296,
+      "loss": 2.5616,
+      "step": 60300
+    },
+    {
+      "epoch": 0.8118497809080889,
+      "grad_norm": 0.9698309302330017,
+      "learning_rate": 0.004990831180498973,
+      "loss": 2.5691,
+      "step": 60400
+    },
+    {
+      "epoch": 0.8131939030619103,
+      "grad_norm": 0.7363952398300171,
+      "learning_rate": 0.0049907857621209264,
+      "loss": 2.5709,
+      "step": 60500
+    },
+    {
+      "epoch": 0.8145380252157316,
+      "grad_norm": 0.45567846298217773,
+      "learning_rate": 0.004990740231737212,
+      "loss": 2.5688,
+      "step": 60600
+    },
+    {
+      "epoch": 0.8158821473695529,
+      "grad_norm": 1.8314270973205566,
+      "learning_rate": 0.004990694589349877,
+      "loss": 2.5694,
+      "step": 60700
+    },
+    {
+      "epoch": 0.8172262695233743,
+      "grad_norm": 0.7680712342262268,
+      "learning_rate": 0.004990648834960976,
+      "loss": 2.5678,
+      "step": 60800
+    },
+    {
+      "epoch": 0.8185703916771956,
+      "grad_norm": 0.8235045075416565,
+      "learning_rate": 0.004990602968572579,
+      "loss": 2.5641,
+      "step": 60900
+    },
+    {
+      "epoch": 0.819914513831017,
+      "grad_norm": 2.173553228378296,
+      "learning_rate": 0.004990556990186752,
+      "loss": 2.5631,
+      "step": 61000
+    },
+    {
+      "epoch": 0.819914513831017,
+      "eval_MaskedAccuracy": 0.42752038773182294,
+      "eval_loss": 2.8879265785217285,
+      "eval_runtime": 154.4174,
+      "eval_samples_per_second": 411.068,
+      "eval_steps_per_second": 1.606,
+      "step": 61000
+    },
+    {
+      "epoch": 0.8212586359848383,
+      "grad_norm": 1.7964410781860352,
+      "learning_rate": 0.004990510899805561,
+      "loss": 2.5645,
+      "step": 61100
+    },
+    {
+      "epoch": 0.8226027581386597,
+      "grad_norm": 0.6248696446418762,
+      "learning_rate": 0.004990464697431084,
+      "loss": 2.5554,
+      "step": 61200
+    },
+    {
+      "epoch": 0.823946880292481,
+      "grad_norm": 1.9335159063339233,
+      "learning_rate": 0.004990418383065405,
+      "loss": 2.5484,
+      "step": 61300
+    },
+    {
+      "epoch": 0.8252910024463023,
+      "grad_norm": 1.169259786605835,
+      "learning_rate": 0.004990371956710612,
+      "loss": 2.5604,
+      "step": 61400
+    },
+    {
+      "epoch": 0.8266351246001237,
+      "grad_norm": 1.7953951358795166,
+      "learning_rate": 0.004990325418368784,
+      "loss": 2.5482,
+      "step": 61500
+    },
+    {
+      "epoch": 0.827979246753945,
+      "grad_norm": 0.47882091999053955,
+      "learning_rate": 0.004990278768042037,
+      "loss": 2.5603,
+      "step": 61600
+    },
+    {
+      "epoch": 0.8293233689077664,
+      "grad_norm": 0.668745219707489,
+      "learning_rate": 0.004990232005732461,
+      "loss": 2.5587,
+      "step": 61700
+    },
+    {
+      "epoch": 0.8306674910615877,
+      "grad_norm": 2.7591805458068848,
+      "learning_rate": 0.0049901851314421665,
+      "loss": 2.5525,
+      "step": 61800
+    },
+    {
+      "epoch": 0.8320116132154091,
+      "grad_norm": 0.5808740258216858,
+      "learning_rate": 0.004990138145173262,
+      "loss": 2.5544,
+      "step": 61900
+    },
+    {
+      "epoch": 0.8333557353692304,
+      "grad_norm": 1.9806839227676392,
+      "learning_rate": 0.00499009104692787,
+      "loss": 2.5569,
+      "step": 62000
+    },
+    {
+      "epoch": 0.8333557353692304,
+      "eval_MaskedAccuracy": 0.4281775187262364,
+      "eval_loss": 2.8783788681030273,
+      "eval_runtime": 157.1845,
+      "eval_samples_per_second": 403.831,
+      "eval_steps_per_second": 1.578,
+      "step": 62000
+    },
+    {
+      "epoch": 0.8346998575230516,
+      "grad_norm": 0.4611744284629822,
+      "learning_rate": 0.004990043836708112,
+      "loss": 2.5569,
+      "step": 62100
+    },
+    {
+      "epoch": 0.836043979676873,
+      "grad_norm": 0.9632844924926758,
+      "learning_rate": 0.00498999651451611,
+      "loss": 2.549,
+      "step": 62200
+    },
+    {
+      "epoch": 0.8373881018306943,
+      "grad_norm": 0.49269410967826843,
+      "learning_rate": 0.004989949080353999,
+      "loss": 2.5543,
+      "step": 62300
+    },
+    {
+      "epoch": 0.8387322239845157,
+      "grad_norm": 1.0680642127990723,
+      "learning_rate": 0.0049899015342239166,
+      "loss": 2.558,
+      "step": 62400
+    },
+    {
+      "epoch": 0.840076346138337,
+      "grad_norm": 1.5859142541885376,
+      "learning_rate": 0.004989853876128006,
+      "loss": 2.5603,
+      "step": 62500
+    },
+    {
+      "epoch": 0.8414204682921584,
+      "grad_norm": 1.1798615455627441,
+      "learning_rate": 0.0049898061060684165,
+      "loss": 2.5592,
+      "step": 62600
+    },
+    {
+      "epoch": 0.8427645904459797,
+      "grad_norm": 1.3057469129562378,
+      "learning_rate": 0.004989758224047301,
+      "loss": 2.5618,
+      "step": 62700
+    },
+    {
+      "epoch": 0.844108712599801,
+      "grad_norm": 0.3981241285800934,
+      "learning_rate": 0.004989710230066817,
+      "loss": 2.5625,
+      "step": 62800
+    },
+    {
+      "epoch": 0.8454528347536224,
+      "grad_norm": 1.141242504119873,
+      "learning_rate": 0.00498966212412912,
+      "loss": 2.5575,
+      "step": 62900
+    },
+    {
+      "epoch": 0.8467969569074437,
+      "grad_norm": 1.986693024635315,
+      "learning_rate": 0.004989613906236384,
+      "loss": 2.5557,
+      "step": 63000
+    },
+    {
+      "epoch": 0.8467969569074437,
+      "eval_MaskedAccuracy": 0.4287118227276102,
+      "eval_loss": 2.8758063316345215,
+      "eval_runtime": 157.2196,
+      "eval_samples_per_second": 403.741,
+      "eval_steps_per_second": 1.577,
+      "step": 63000
+    },
+    {
+      "epoch": 0.8481410790612651,
+      "grad_norm": 0.42742684483528137,
+      "learning_rate": 0.0049895655763907785,
+      "loss": 2.5609,
+      "step": 63100
+    },
+    {
+      "epoch": 0.8494852012150864,
+      "grad_norm": 0.6005121469497681,
+      "learning_rate": 0.004989517134594481,
+      "loss": 2.5627,
+      "step": 63200
+    },
+    {
+      "epoch": 0.8508293233689078,
+      "grad_norm": 1.2548339366912842,
+      "learning_rate": 0.004989468580849676,
+      "loss": 2.5541,
+      "step": 63300
+    },
+    {
+      "epoch": 0.8521734455227291,
+      "grad_norm": 1.1956408023834229,
+      "learning_rate": 0.004989419915158556,
+      "loss": 2.5494,
+      "step": 63400
+    },
+    {
+      "epoch": 0.8535175676765504,
+      "grad_norm": 1.7097578048706055,
+      "learning_rate": 0.004989371137523307,
+      "loss": 2.5559,
+      "step": 63500
+    },
+    {
+      "epoch": 0.8548616898303718,
+      "grad_norm": 0.8344634175300598,
+      "learning_rate": 0.004989322247946124,
+      "loss": 2.5594,
+      "step": 63600
+    },
+    {
+      "epoch": 0.8562058119841931,
+      "grad_norm": 1.0530003309249878,
+      "learning_rate": 0.004989273246429215,
+      "loss": 2.548,
+      "step": 63700
+    },
+    {
+      "epoch": 0.8575499341380145,
+      "grad_norm": 1.2484129667282104,
+      "learning_rate": 0.004989224132974791,
+      "loss": 2.5514,
+      "step": 63800
+    },
+    {
+      "epoch": 0.8588940562918358,
+      "grad_norm": 0.48474428057670593,
+      "learning_rate": 0.004989174907585054,
+      "loss": 2.5607,
+      "step": 63900
+    },
+    {
+      "epoch": 0.8602381784456571,
+      "grad_norm": 0.7260443568229675,
+      "learning_rate": 0.004989125570262237,
+      "loss": 2.5607,
+      "step": 64000
+    },
+    {
+      "epoch": 0.8602381784456571,
+      "eval_MaskedAccuracy": 0.4298587031625389,
+      "eval_loss": 2.8685121536254883,
+      "eval_runtime": 156.7219,
+      "eval_samples_per_second": 405.023,
+      "eval_steps_per_second": 1.582,
+      "step": 64000
+    },
+    {
+      "epoch": 0.8615823005994785,
+      "grad_norm": 1.7127766609191895,
+      "learning_rate": 0.0049890761210085515,
+      "loss": 2.5457,
+      "step": 64100
+    },
+    {
+      "epoch": 0.8629264227532998,
+      "grad_norm": 0.9210508465766907,
+      "learning_rate": 0.004989026559826237,
+      "loss": 2.5567,
+      "step": 64200
+    },
+    {
+      "epoch": 0.8642705449071212,
+      "grad_norm": 0.5926476120948792,
+      "learning_rate": 0.004988976886717514,
+      "loss": 2.5502,
+      "step": 64300
+    },
+    {
+      "epoch": 0.8656146670609425,
+      "grad_norm": 0.8887967467308044,
+      "learning_rate": 0.0049889271016846275,
+      "loss": 2.5511,
+      "step": 64400
+    },
+    {
+      "epoch": 0.8669587892147639,
+      "grad_norm": 0.6354972124099731,
+      "learning_rate": 0.0049888772047298165,
+      "loss": 2.5549,
+      "step": 64500
+    },
+    {
+      "epoch": 0.8683029113685852,
+      "grad_norm": 0.45211437344551086,
+      "learning_rate": 0.004988827195855339,
+      "loss": 2.5541,
+      "step": 64600
+    },
+    {
+      "epoch": 0.8696470335224065,
+      "grad_norm": 1.1961520910263062,
+      "learning_rate": 0.004988777075063434,
+      "loss": 2.549,
+      "step": 64700
+    },
+    {
+      "epoch": 0.8709911556762279,
+      "grad_norm": 0.7773739099502563,
+      "learning_rate": 0.004988726842356369,
+      "loss": 2.5453,
+      "step": 64800
+    },
+    {
+      "epoch": 0.8723352778300492,
+      "grad_norm": 1.6650779247283936,
+      "learning_rate": 0.004988676497736411,
+      "loss": 2.5591,
+      "step": 64900
+    },
+    {
+      "epoch": 0.8736793999838706,
+      "grad_norm": 3.7934086322784424,
+      "learning_rate": 0.004988626041205816,
+      "loss": 2.5477,
+      "step": 65000
+    },
+    {
+      "epoch": 0.8736793999838706,
+      "eval_MaskedAccuracy": 0.42935040461530544,
+      "eval_loss": 2.8740768432617188,
+      "eval_runtime": 157.0461,
+      "eval_samples_per_second": 404.187,
+      "eval_steps_per_second": 1.579,
+      "step": 65000
+    },
+    {
+      "epoch": 0.8750235221376919,
+      "grad_norm": 0.37112095952033997,
+      "learning_rate": 0.004988575472766864,
+      "loss": 2.5442,
+      "step": 65100
+    },
+    {
+      "epoch": 0.8763676442915133,
+      "grad_norm": 0.4250340461730957,
+      "learning_rate": 0.0049885247924218385,
+      "loss": 2.5397,
+      "step": 65200
+    },
+    {
+      "epoch": 0.8777117664453346,
+      "grad_norm": 0.42517802119255066,
+      "learning_rate": 0.004988474000173024,
+      "loss": 2.5453,
+      "step": 65300
+    },
+    {
+      "epoch": 0.8790558885991558,
+      "grad_norm": 2.0652735233306885,
+      "learning_rate": 0.0049884230960227,
+      "loss": 2.5497,
+      "step": 65400
+    },
+    {
+      "epoch": 0.8804000107529772,
+      "grad_norm": 0.6342276930809021,
+      "learning_rate": 0.0049883720799731605,
+      "loss": 2.5378,
+      "step": 65500
+    },
+    {
+      "epoch": 0.8817441329067985,
+      "grad_norm": 1.0406826734542847,
+      "learning_rate": 0.0049883209520267075,
+      "loss": 2.5453,
+      "step": 65600
+    },
+    {
+      "epoch": 0.8830882550606199,
+      "grad_norm": 1.4055812358856201,
+      "learning_rate": 0.00498826971218565,
+      "loss": 2.5437,
+      "step": 65700
+    },
+    {
+      "epoch": 0.8844323772144412,
+      "grad_norm": 2.7914650440216064,
+      "learning_rate": 0.004988218360452281,
+      "loss": 2.5512,
+      "step": 65800
+    },
+    {
+      "epoch": 0.8857764993682626,
+      "grad_norm": 0.9103912711143494,
+      "learning_rate": 0.004988166896828934,
+      "loss": 2.5466,
+      "step": 65900
+    },
+    {
+      "epoch": 0.8871206215220839,
+      "grad_norm": 1.0117913484573364,
+      "learning_rate": 0.004988115321317916,
+      "loss": 2.5452,
+      "step": 66000
+    },
+    {
+      "epoch": 0.8871206215220839,
+      "eval_MaskedAccuracy": 0.4316935610840671,
+      "eval_loss": 2.8568239212036133,
+      "eval_runtime": 155.5022,
+      "eval_samples_per_second": 408.2,
+      "eval_steps_per_second": 1.595,
+      "step": 66000
+    },
+    {
+      "epoch": 0.8884647436759052,
+      "grad_norm": 0.934119701385498,
+      "learning_rate": 0.004988063633921554,
+      "loss": 2.5407,
+      "step": 66100
+    },
+    {
+      "epoch": 0.8898088658297266,
+      "grad_norm": 0.5075318217277527,
+      "learning_rate": 0.004988011834642176,
+      "loss": 2.5387,
+      "step": 66200
+    },
+    {
+      "epoch": 0.8911529879835479,
+      "grad_norm": 2.349472761154175,
+      "learning_rate": 0.004987959923482121,
+      "loss": 2.5453,
+      "step": 66300
+    },
+    {
+      "epoch": 0.8924971101373693,
+      "grad_norm": 1.2346925735473633,
+      "learning_rate": 0.004987907900443723,
+      "loss": 2.5433,
+      "step": 66400
+    },
+    {
+      "epoch": 0.8938412322911906,
+      "grad_norm": 0.9349079132080078,
+      "learning_rate": 0.004987855765529321,
+      "loss": 2.5386,
+      "step": 66500
+    },
+    {
+      "epoch": 0.895185354445012,
+      "grad_norm": 2.3298845291137695,
+      "learning_rate": 0.00498780351874127,
+      "loss": 2.5579,
+      "step": 66600
+    },
+    {
+      "epoch": 0.8965294765988333,
+      "grad_norm": 0.7276595830917358,
+      "learning_rate": 0.0049877511600819255,
+      "loss": 2.547,
+      "step": 66700
+    },
+    {
+      "epoch": 0.8978735987526546,
+      "grad_norm": 0.9799304604530334,
+      "learning_rate": 0.004987698689553648,
+      "loss": 2.5479,
+      "step": 66800
+    },
+    {
+      "epoch": 0.899217720906476,
+      "grad_norm": 4.606154918670654,
+      "learning_rate": 0.0049876461071588,
+      "loss": 2.5486,
+      "step": 66900
+    },
+    {
+      "epoch": 0.9005618430602973,
+      "grad_norm": 0.8625683784484863,
+      "learning_rate": 0.004987593412899746,
+      "loss": 2.5448,
+      "step": 67000
+    },
+    {
+      "epoch": 0.9005618430602973,
+      "eval_MaskedAccuracy": 0.431047719967855,
+      "eval_loss": 2.8602442741394043,
+      "eval_runtime": 154.394,
+      "eval_samples_per_second": 411.13,
+      "eval_steps_per_second": 1.606,
+      "step": 67000
+    },
+    {
+      "epoch": 0.9019059652141187,
+      "grad_norm": 0.6021907925605774,
+      "learning_rate": 0.0049875406067788645,
+      "loss": 2.5473,
+      "step": 67100
+    },
+    {
+      "epoch": 0.90325008736794,
+      "grad_norm": 1.0424331426620483,
+      "learning_rate": 0.004987487688798529,
+      "loss": 2.5401,
+      "step": 67200
+    },
+    {
+      "epoch": 0.9045942095217613,
+      "grad_norm": 0.7000874280929565,
+      "learning_rate": 0.004987434658961136,
+      "loss": 2.5403,
+      "step": 67300
+    },
+    {
+      "epoch": 0.9059383316755827,
+      "grad_norm": 0.437077134847641,
+      "learning_rate": 0.004987381517269073,
+      "loss": 2.5457,
+      "step": 67400
+    },
+    {
+      "epoch": 0.907282453829404,
+      "grad_norm": 0.5734995007514954,
+      "learning_rate": 0.004987328263724719,
+      "loss": 2.5496,
+      "step": 67500
+    },
+    {
+      "epoch": 0.9086265759832254,
+      "grad_norm": 0.9793594479560852,
+      "learning_rate": 0.00498727489833049,
+      "loss": 2.5374,
+      "step": 67600
+    },
+    {
+      "epoch": 0.9099706981370467,
+      "grad_norm": 1.336166262626648,
+      "learning_rate": 0.004987221421088781,
+      "loss": 2.5348,
+      "step": 67700
+    },
+    {
+      "epoch": 0.9113148202908681,
+      "grad_norm": 0.8772963881492615,
+      "learning_rate": 0.00498716783200201,
+      "loss": 2.5453,
+      "step": 67800
+    },
+    {
+      "epoch": 0.9126589424446894,
+      "grad_norm": 1.0549404621124268,
+      "learning_rate": 0.0049871141310725816,
+      "loss": 2.5394,
+      "step": 67900
+    },
+    {
+      "epoch": 0.9140030645985107,
+      "grad_norm": 0.6671266555786133,
+      "learning_rate": 0.004987060318302928,
+      "loss": 2.5408,
+      "step": 68000
+    },
+    {
+      "epoch": 0.9140030645985107,
+      "eval_MaskedAccuracy": 0.43042410070670717,
+      "eval_loss": 2.864036798477173,
+      "eval_runtime": 158.2908,
+      "eval_samples_per_second": 401.009,
+      "eval_steps_per_second": 1.567,
+      "step": 68000
+    },
+    {
+      "epoch": 0.9153471867523321,
+      "grad_norm": 1.8076815605163574,
+      "learning_rate": 0.004987006393695469,
+      "loss": 2.5349,
+      "step": 68100
+    },
+    {
+      "epoch": 0.9166913089061534,
+      "grad_norm": 0.9832874536514282,
+      "learning_rate": 0.004986952357252627,
+      "loss": 2.5372,
+      "step": 68200
+    },
+    {
+      "epoch": 0.9180354310599748,
+      "grad_norm": 0.6802740693092346,
+      "learning_rate": 0.004986898208976844,
+      "loss": 2.538,
+      "step": 68300
+    },
+    {
+      "epoch": 0.919379553213796,
+      "grad_norm": 1.3755671977996826,
+      "learning_rate": 0.004986843948870563,
+      "loss": 2.5427,
+      "step": 68400
+    },
+    {
+      "epoch": 0.9207236753676175,
+      "grad_norm": 0.6860373020172119,
+      "learning_rate": 0.0049867895769362194,
+      "loss": 2.5456,
+      "step": 68500
+    },
+    {
+      "epoch": 0.9220677975214387,
+      "grad_norm": 0.37920355796813965,
+      "learning_rate": 0.004986735093176264,
+      "loss": 2.5388,
+      "step": 68600
+    },
+    {
+      "epoch": 0.92341191967526,
+      "grad_norm": 0.48161619901657104,
+      "learning_rate": 0.004986680497593163,
+      "loss": 2.5482,
+      "step": 68700
+    },
+    {
+      "epoch": 0.9247560418290814,
+      "grad_norm": 1.5831481218338013,
+      "learning_rate": 0.0049866257901893625,
+      "loss": 2.533,
+      "step": 68800
+    },
+    {
+      "epoch": 0.9261001639829027,
+      "grad_norm": 1.874329924583435,
+      "learning_rate": 0.004986570970967331,
+      "loss": 2.5432,
+      "step": 68900
+    },
+    {
+      "epoch": 0.9274442861367241,
+      "grad_norm": 0.6668677926063538,
+      "learning_rate": 0.004986516039929541,
+      "loss": 2.5458,
+      "step": 69000
+    },
+    {
+      "epoch": 0.9274442861367241,
+      "eval_MaskedAccuracy": 0.4320338287315044,
+      "eval_loss": 2.851020336151123,
+      "eval_runtime": 154.6645,
+      "eval_samples_per_second": 410.411,
+      "eval_steps_per_second": 1.603,
+      "step": 69000
+    },
+    {
+      "epoch": 0.9287884082905454,
+      "grad_norm": 0.850222647190094,
+      "learning_rate": 0.004986460997078469,
+      "loss": 2.5411,
+      "step": 69100
+    },
+    {
+      "epoch": 0.9301325304443668,
+      "grad_norm": 1.386954665184021,
+      "learning_rate": 0.004986405842416592,
+      "loss": 2.5424,
+      "step": 69200
+    },
+    {
+      "epoch": 0.9314766525981881,
+      "grad_norm": 0.7696434259414673,
+      "learning_rate": 0.0049863505759464,
+      "loss": 2.5336,
+      "step": 69300
+    },
+    {
+      "epoch": 0.9328207747520094,
+      "grad_norm": 0.8896027207374573,
+      "learning_rate": 0.004986295197670379,
+      "loss": 2.5325,
+      "step": 69400
+    },
+    {
+      "epoch": 0.9341648969058308,
+      "grad_norm": 1.7881613969802856,
+      "learning_rate": 0.004986239707591028,
+      "loss": 2.5402,
+      "step": 69500
+    },
+    {
+      "epoch": 0.9355090190596521,
+      "grad_norm": 0.5906102061271667,
+      "learning_rate": 0.0049861841057108405,
+      "loss": 2.542,
+      "step": 69600
+    },
+    {
+      "epoch": 0.9368531412134735,
+      "grad_norm": 0.7546370625495911,
+      "learning_rate": 0.004986128392032327,
+      "loss": 2.5392,
+      "step": 69700
+    },
+    {
+      "epoch": 0.9381972633672948,
+      "grad_norm": 0.7366216778755188,
+      "learning_rate": 0.004986072566557999,
+      "loss": 2.5316,
+      "step": 69800
+    },
+    {
+      "epoch": 0.9395413855211162,
+      "grad_norm": 1.4292287826538086,
+      "learning_rate": 0.00498601662929037,
+      "loss": 2.5371,
+      "step": 69900
+    },
+    {
+      "epoch": 0.9408855076749375,
+      "grad_norm": 0.9893214106559753,
+      "learning_rate": 0.00498596058023197,
+      "loss": 2.5313,
+      "step": 70000
+    },
+    {
+      "epoch": 0.9408855076749375,
+      "eval_MaskedAccuracy": 0.4319902861219465,
+      "eval_loss": 2.853639602661133,
+      "eval_runtime": 157.3354,
+      "eval_samples_per_second": 403.444,
+      "eval_steps_per_second": 1.576,
+      "step": 70000
+    },
+    {
+      "epoch": 0.9422296298287588,
+      "grad_norm": 1.4988006353378296,
+      "learning_rate": 0.004985904419385302,
+      "loss": 2.5369,
+      "step": 70100
+    },
+    {
+      "epoch": 0.9435737519825802,
+      "grad_norm": 1.3903696537017822,
+      "learning_rate": 0.004985848146752917,
+      "loss": 2.5312,
+      "step": 70200
+    },
+    {
+      "epoch": 0.9449178741364015,
+      "grad_norm": 1.741083025932312,
+      "learning_rate": 0.004985791762337335,
+      "loss": 2.5343,
+      "step": 70300
+    },
+    {
+      "epoch": 0.9462619962902229,
+      "grad_norm": 0.43283969163894653,
+      "learning_rate": 0.00498573526614111,
+      "loss": 2.5308,
+      "step": 70400
+    },
+    {
+      "epoch": 0.9476061184440442,
+      "grad_norm": 0.6915990114212036,
+      "learning_rate": 0.004985678658166784,
+      "loss": 2.5316,
+      "step": 70500
+    },
+    {
+      "epoch": 0.9489502405978655,
+      "grad_norm": 3.1753523349761963,
+      "learning_rate": 0.004985621938416906,
+      "loss": 2.5376,
+      "step": 70600
+    },
+    {
+      "epoch": 0.9502943627516869,
+      "grad_norm": 1.0071120262145996,
+      "learning_rate": 0.0049855651068940335,
+      "loss": 2.5395,
+      "step": 70700
+    },
+    {
+      "epoch": 0.9516384849055082,
+      "grad_norm": 3.022740125656128,
+      "learning_rate": 0.004985508163600716,
+      "loss": 2.5267,
+      "step": 70800
+    },
+    {
+      "epoch": 0.9529826070593296,
+      "grad_norm": 2.2657601833343506,
+      "learning_rate": 0.004985451108539533,
+      "loss": 2.5279,
+      "step": 70900
+    },
+    {
+      "epoch": 0.9543267292131509,
+      "grad_norm": 1.0798835754394531,
+      "learning_rate": 0.004985393941713058,
+      "loss": 2.5346,
+      "step": 71000
+    },
+    {
+      "epoch": 0.9543267292131509,
+      "eval_MaskedAccuracy": 0.43291381575543303,
+      "eval_loss": 2.847073554992676,
+      "eval_runtime": 157.3815,
+      "eval_samples_per_second": 403.326,
+      "eval_steps_per_second": 1.576,
+      "step": 71000
+    },
+    {
+      "epoch": 0.9556708513669723,
+      "grad_norm": 0.8452098369598389,
+      "learning_rate": 0.004985336663123854,
+      "loss": 2.5344,
+      "step": 71100
+    },
+    {
+      "epoch": 0.9570149735207936,
+      "grad_norm": 1.136425256729126,
+      "learning_rate": 0.0049852792727745135,
+      "loss": 2.5373,
+      "step": 71200
+    },
+    {
+      "epoch": 0.9583590956746149,
+      "grad_norm": 0.421425461769104,
+      "learning_rate": 0.004985221770667615,
+      "loss": 2.5365,
+      "step": 71300
+    },
+    {
+      "epoch": 0.9597032178284363,
+      "grad_norm": 1.0452524423599243,
+      "learning_rate": 0.004985164156805749,
+      "loss": 2.5405,
+      "step": 71400
+    },
+    {
+      "epoch": 0.9610473399822576,
+      "grad_norm": 0.6131758689880371,
+      "learning_rate": 0.004985106431191513,
+      "loss": 2.5379,
+      "step": 71500
+    },
+    {
+      "epoch": 0.962391462136079,
+      "grad_norm": 0.5693802833557129,
+      "learning_rate": 0.004985048593827507,
+      "loss": 2.5415,
+      "step": 71600
+    },
+    {
+      "epoch": 0.9637355842899002,
+      "grad_norm": 0.576070249080658,
+      "learning_rate": 0.004984990644716343,
+      "loss": 2.5258,
+      "step": 71700
+    },
+    {
+      "epoch": 0.9650797064437217,
+      "grad_norm": 0.8983374834060669,
+      "learning_rate": 0.004984932583860627,
+      "loss": 2.5222,
+      "step": 71800
+    },
+    {
+      "epoch": 0.9664238285975429,
+      "grad_norm": 1.3723649978637695,
+      "learning_rate": 0.004984874411262972,
+      "loss": 2.5328,
+      "step": 71900
+    },
+    {
+      "epoch": 0.9677679507513642,
+      "grad_norm": 0.638632595539093,
+      "learning_rate": 0.004984816126926011,
+      "loss": 2.5291,
+      "step": 72000
+    },
+    {
+      "epoch": 0.9677679507513642,
+      "eval_MaskedAccuracy": 0.4332407392153184,
+      "eval_loss": 2.8442609310150146,
+      "eval_runtime": 157.5641,
+      "eval_samples_per_second": 402.858,
+      "eval_steps_per_second": 1.574,
+      "step": 72000
+    },
+    {
+      "epoch": 0.9691120729051856,
+      "grad_norm": 0.999791145324707,
+      "learning_rate": 0.0049847577308523554,
+      "loss": 2.5289,
+      "step": 72100
+    },
+    {
+      "epoch": 0.9704561950590069,
+      "grad_norm": 0.9468411803245544,
+      "learning_rate": 0.0049846992230446465,
+      "loss": 2.5244,
+      "step": 72200
+    },
+    {
+      "epoch": 0.9718003172128283,
+      "grad_norm": 1.2006036043167114,
+      "learning_rate": 0.0049846406035055114,
+      "loss": 2.5294,
+      "step": 72300
+    },
+    {
+      "epoch": 0.9731444393666496,
+      "grad_norm": 1.1461031436920166,
+      "learning_rate": 0.0049845818722376025,
+      "loss": 2.5251,
+      "step": 72400
+    },
+    {
+      "epoch": 0.974488561520471,
+      "grad_norm": 1.3583283424377441,
+      "learning_rate": 0.004984523029243558,
+      "loss": 2.5401,
+      "step": 72500
+    },
+    {
+      "epoch": 0.9758326836742923,
+      "grad_norm": 0.8848266005516052,
+      "learning_rate": 0.004984464074526038,
+      "loss": 2.534,
+      "step": 72600
+    },
+    {
+      "epoch": 0.9771768058281136,
+      "grad_norm": 0.8786698579788208,
+      "learning_rate": 0.004984405008087696,
+      "loss": 2.5268,
+      "step": 72700
+    },
+    {
+      "epoch": 0.978520927981935,
+      "grad_norm": 1.2354623079299927,
+      "learning_rate": 0.004984345829931181,
+      "loss": 2.5235,
+      "step": 72800
+    },
+    {
+      "epoch": 0.9798650501357563,
+      "grad_norm": 0.44590994715690613,
+      "learning_rate": 0.004984286540059167,
+      "loss": 2.5249,
+      "step": 72900
+    },
+    {
+      "epoch": 0.9812091722895777,
+      "grad_norm": 1.5958350896835327,
+      "learning_rate": 0.004984227138474335,
+      "loss": 2.5252,
+      "step": 73000
+    },
+    {
+      "epoch": 0.9812091722895777,
+      "eval_MaskedAccuracy": 0.43342773457995387,
+      "eval_loss": 2.8439483642578125,
+      "eval_runtime": 156.2562,
+      "eval_samples_per_second": 406.23,
+      "eval_steps_per_second": 1.587,
+      "step": 73000
+    },
+    {
+      "epoch": 0.982553294443399,
+      "grad_norm": 0.9275982975959778,
+      "learning_rate": 0.004984167625179348,
+      "loss": 2.5338,
+      "step": 73100
+    },
+    {
+      "epoch": 0.9838974165972204,
+      "grad_norm": 1.3247750997543335,
+      "learning_rate": 0.0049841080001768985,
+      "loss": 2.5258,
+      "step": 73200
+    },
+    {
+      "epoch": 0.9852415387510417,
+      "grad_norm": 0.8629142642021179,
+      "learning_rate": 0.004984048263469671,
+      "loss": 2.5277,
+      "step": 73300
+    },
+    {
+      "epoch": 0.986585660904863,
+      "grad_norm": 1.2178809642791748,
+      "learning_rate": 0.004983988415060354,
+      "loss": 2.5345,
+      "step": 73400
+    },
+    {
+      "epoch": 0.9879297830586844,
+      "grad_norm": 1.3900634050369263,
+      "learning_rate": 0.004983928454951644,
+      "loss": 2.5164,
+      "step": 73500
+    },
+    {
+      "epoch": 0.9892739052125057,
+      "grad_norm": 1.1817967891693115,
+      "learning_rate": 0.00498386838314624,
+      "loss": 2.5267,
+      "step": 73600
+    },
+    {
+      "epoch": 0.9906180273663271,
+      "grad_norm": 0.8701839447021484,
+      "learning_rate": 0.004983808199646853,
+      "loss": 2.5238,
+      "step": 73700
+    },
+    {
+      "epoch": 0.9919621495201484,
+      "grad_norm": 1.1527352333068848,
+      "learning_rate": 0.0049837479044561985,
+      "loss": 2.5293,
+      "step": 73800
+    },
+    {
+      "epoch": 0.9933062716739697,
+      "grad_norm": 1.1712994575500488,
+      "learning_rate": 0.004983687497576983,
+      "loss": 2.5292,
+      "step": 73900
+    },
+    {
+      "epoch": 0.9946503938277911,
+      "grad_norm": 1.6270473003387451,
+      "learning_rate": 0.004983626979011935,
+      "loss": 2.5288,
+      "step": 74000
+    },
+    {
+      "epoch": 0.9946503938277911,
+      "eval_MaskedAccuracy": 0.432422431765529,
+      "eval_loss": 2.8484365940093994,
+      "eval_runtime": 154.8637,
+      "eval_samples_per_second": 409.883,
+      "eval_steps_per_second": 1.601,
+      "step": 74000
+    },
+    {
+      "epoch": 0.9959945159816124,
+      "grad_norm": 0.4700554609298706,
+      "learning_rate": 0.0049835663487637835,
+      "loss": 2.5225,
+      "step": 74100
+    },
+    {
+      "epoch": 0.9973386381354338,
+      "grad_norm": 5.065925121307373,
+      "learning_rate": 0.004983505606835256,
+      "loss": 2.5303,
+      "step": 74200
+    },
+    {
+      "epoch": 0.9986827602892551,
+      "grad_norm": 0.3857990503311157,
+      "learning_rate": 0.004983444753229089,
+      "loss": 2.5351,
+      "step": 74300
+    },
+    {
+      "epoch": 1.0000268824430765,
+      "grad_norm": 3.9997522830963135,
+      "learning_rate": 0.00498338378794803,
+      "loss": 2.5212,
+      "step": 74400
+    },
+    {
+      "epoch": 1.0013710045968978,
+      "grad_norm": 1.400756597518921,
+      "learning_rate": 0.00498332271099482,
+      "loss": 2.52,
+      "step": 74500
+    },
+    {
+      "epoch": 1.002715126750719,
+      "grad_norm": 0.6085138916969299,
+      "learning_rate": 0.004983261522372212,
+      "loss": 2.526,
+      "step": 74600
+    },
+    {
+      "epoch": 1.0040592489045403,
+      "grad_norm": 1.8259930610656738,
+      "learning_rate": 0.004983200222082968,
+      "loss": 2.5176,
+      "step": 74700
+    },
+    {
+      "epoch": 1.0054033710583619,
+      "grad_norm": 2.1580281257629395,
+      "learning_rate": 0.004983138810129848,
+      "loss": 2.5273,
+      "step": 74800
+    },
+    {
+      "epoch": 1.0067474932121832,
+      "grad_norm": 0.8419142365455627,
+      "learning_rate": 0.004983077286515609,
+      "loss": 2.5232,
+      "step": 74900
+    },
+    {
+      "epoch": 1.0080916153660044,
+      "grad_norm": 1.1031572818756104,
+      "learning_rate": 0.004983015651243041,
+      "loss": 2.5381,
+      "step": 75000
+    },
+    {
+      "epoch": 1.0080916153660044,
+      "eval_MaskedAccuracy": 0.43352066355967,
+      "eval_loss": 2.838632822036743,
+      "eval_runtime": 154.6683,
+      "eval_samples_per_second": 410.401,
+      "eval_steps_per_second": 1.603,
+      "step": 75000
+    },
+    {
+      "epoch": 1.0094357375198257,
+      "grad_norm": 0.4656729996204376,
+      "learning_rate": 0.004982953904314914,
+      "loss": 2.5295,
+      "step": 75100
+    },
+    {
+      "epoch": 1.0107798596736473,
+      "grad_norm": 0.7429810762405396,
+      "learning_rate": 0.004982892045734002,
+      "loss": 2.5279,
+      "step": 75200
+    },
+    {
+      "epoch": 1.0121239818274685,
+      "grad_norm": 0.36424189805984497,
+      "learning_rate": 0.004982830075503102,
+      "loss": 2.5211,
+      "step": 75300
+    },
+    {
+      "epoch": 1.0134681039812898,
+      "grad_norm": 0.45235106348991394,
+      "learning_rate": 0.004982767993625003,
+      "loss": 2.5254,
+      "step": 75400
+    },
+    {
+      "epoch": 1.0148122261351111,
+      "grad_norm": 0.6336104273796082,
+      "learning_rate": 0.0049827058001024995,
+      "loss": 2.5318,
+      "step": 75500
+    },
+    {
+      "epoch": 1.0161563482889324,
+      "grad_norm": 0.5706799030303955,
+      "learning_rate": 0.004982643494938397,
+      "loss": 2.5267,
+      "step": 75600
+    },
+    {
+      "epoch": 1.017500470442754,
+      "grad_norm": 0.5101211667060852,
+      "learning_rate": 0.004982581078135498,
+      "loss": 2.5239,
+      "step": 75700
+    },
+    {
+      "epoch": 1.0188445925965752,
+      "grad_norm": 0.6169567108154297,
+      "learning_rate": 0.004982518549696623,
+      "loss": 2.5309,
+      "step": 75800
+    },
+    {
+      "epoch": 1.0201887147503965,
+      "grad_norm": 1.594337821006775,
+      "learning_rate": 0.004982455909624589,
+      "loss": 2.5253,
+      "step": 75900
+    },
+    {
+      "epoch": 1.0215328369042178,
+      "grad_norm": 1.7339344024658203,
+      "learning_rate": 0.004982393157922211,
+      "loss": 2.5248,
+      "step": 76000
+    },
+    {
+      "epoch": 1.0215328369042178,
+      "eval_MaskedAccuracy": 0.43333633554622425,
+      "eval_loss": 2.8395068645477295,
+      "eval_runtime": 155.9239,
+      "eval_samples_per_second": 407.096,
+      "eval_steps_per_second": 1.591,
+      "step": 76000
+    },
+    {
+      "epoch": 1.022876959058039,
+      "grad_norm": 0.3945859968662262,
+      "learning_rate": 0.00498233029459232,
+      "loss": 2.52,
+      "step": 76100
+    },
+    {
+      "epoch": 1.0242210812118606,
+      "grad_norm": 0.9986801743507385,
+      "learning_rate": 0.00498226731963775,
+      "loss": 2.5088,
+      "step": 76200
+    },
+    {
+      "epoch": 1.025565203365682,
+      "grad_norm": 0.9021244049072266,
+      "learning_rate": 0.0049822042330613416,
+      "loss": 2.5183,
+      "step": 76300
+    },
+    {
+      "epoch": 1.0269093255195032,
+      "grad_norm": 0.3702966272830963,
+      "learning_rate": 0.004982141034865935,
+      "loss": 2.5185,
+      "step": 76400
+    },
+    {
+      "epoch": 1.0282534476733245,
+      "grad_norm": 1.010032057762146,
+      "learning_rate": 0.004982077725054371,
+      "loss": 2.5323,
+      "step": 76500
+    },
+    {
+      "epoch": 1.0295975698271458,
+      "grad_norm": 3.7542576789855957,
+      "learning_rate": 0.004982014303629503,
+      "loss": 2.5153,
+      "step": 76600
+    },
+    {
+      "epoch": 1.0309416919809673,
+      "grad_norm": 0.729006826877594,
+      "learning_rate": 0.0049819507705941984,
+      "loss": 2.5259,
+      "step": 76700
+    },
+    {
+      "epoch": 1.0322858141347886,
+      "grad_norm": 0.475752592086792,
+      "learning_rate": 0.0049818871259513136,
+      "loss": 2.5191,
+      "step": 76800
+    },
+    {
+      "epoch": 1.0336299362886099,
+      "grad_norm": 1.1220066547393799,
+      "learning_rate": 0.004981823369703716,
+      "loss": 2.5177,
+      "step": 76900
+    },
+    {
+      "epoch": 1.0349740584424312,
+      "grad_norm": 0.5770747661590576,
+      "learning_rate": 0.004981759501854277,
+      "loss": 2.5298,
+      "step": 77000
+    },
+    {
+      "epoch": 1.0349740584424312,
+      "eval_MaskedAccuracy": 0.4340570336735209,
+      "eval_loss": 2.8367176055908203,
+      "eval_runtime": 156.2265,
+      "eval_samples_per_second": 406.307,
+      "eval_steps_per_second": 1.587,
+      "step": 77000
+    },
+    {
+      "epoch": 1.0363181805962527,
+      "grad_norm": 0.4791591763496399,
+      "learning_rate": 0.004981695522405884,
+      "loss": 2.5189,
+      "step": 77100
+    },
+    {
+      "epoch": 1.037662302750074,
+      "grad_norm": 1.2197085618972778,
+      "learning_rate": 0.00498163143136141,
+      "loss": 2.5181,
+      "step": 77200
+    },
+    {
+      "epoch": 1.0390064249038953,
+      "grad_norm": 1.1560014486312866,
+      "learning_rate": 0.004981567228723745,
+      "loss": 2.517,
+      "step": 77300
+    },
+    {
+      "epoch": 1.0403505470577166,
+      "grad_norm": 0.5016676783561707,
+      "learning_rate": 0.004981502914495785,
+      "loss": 2.5196,
+      "step": 77400
+    },
+    {
+      "epoch": 1.0416946692115379,
+      "grad_norm": 1.4357668161392212,
+      "learning_rate": 0.004981438488680427,
+      "loss": 2.5121,
+      "step": 77500
+    },
+    {
+      "epoch": 1.0430387913653594,
+      "grad_norm": 1.0016825199127197,
+      "learning_rate": 0.004981373951280573,
+      "loss": 2.5197,
+      "step": 77600
+    },
+    {
+      "epoch": 1.0443829135191807,
+      "grad_norm": 5.918942451477051,
+      "learning_rate": 0.004981309302299129,
+      "loss": 2.5243,
+      "step": 77700
+    },
+    {
+      "epoch": 1.045727035673002,
+      "grad_norm": 1.207068681716919,
+      "learning_rate": 0.004981244541739009,
+      "loss": 2.5257,
+      "step": 77800
+    },
+    {
+      "epoch": 1.0470711578268233,
+      "grad_norm": 0.649711012840271,
+      "learning_rate": 0.004981179669603129,
+      "loss": 2.5274,
+      "step": 77900
+    },
+    {
+      "epoch": 1.0484152799806445,
+      "grad_norm": 1.0089892148971558,
+      "learning_rate": 0.004981114685894418,
+      "loss": 2.5166,
+      "step": 78000
+    },
+    {
+      "epoch": 1.0484152799806445,
+      "eval_MaskedAccuracy": 0.4343209398198679,
+      "eval_loss": 2.835813045501709,
+      "eval_runtime": 154.6977,
+      "eval_samples_per_second": 410.323,
+      "eval_steps_per_second": 1.603,
+      "step": 78000
+    },
+    {
+      "epoch": 1.049759402134466,
+      "grad_norm": 0.8486099243164062,
+      "learning_rate": 0.0049810495906158,
+      "loss": 2.5232,
+      "step": 78100
+    },
+    {
+      "epoch": 1.0511035242882873,
+      "grad_norm": 0.6109094023704529,
+      "learning_rate": 0.004980984383770206,
+      "loss": 2.5207,
+      "step": 78200
+    },
+    {
+      "epoch": 1.0524476464421086,
+      "grad_norm": 0.463224858045578,
+      "learning_rate": 0.004980919065360583,
+      "loss": 2.5198,
+      "step": 78300
+    },
+    {
+      "epoch": 1.05379176859593,
+      "grad_norm": 0.5557699799537659,
+      "learning_rate": 0.004980853635389861,
+      "loss": 2.5235,
+      "step": 78400
+    },
+    {
+      "epoch": 1.0551358907497512,
+      "grad_norm": 4.43305778503418,
+      "learning_rate": 0.004980788093860997,
+      "loss": 2.5133,
+      "step": 78500
+    },
+    {
+      "epoch": 1.0564800129035727,
+      "grad_norm": 0.5371536612510681,
+      "learning_rate": 0.0049807224407769435,
+      "loss": 2.5208,
+      "step": 78600
+    },
+    {
+      "epoch": 1.057824135057394,
+      "grad_norm": 1.074479341506958,
+      "learning_rate": 0.004980656676140655,
+      "loss": 2.5064,
+      "step": 78700
+    },
+    {
+      "epoch": 1.0591682572112153,
+      "grad_norm": 2.650197744369507,
+      "learning_rate": 0.004980590799955096,
+      "loss": 2.523,
+      "step": 78800
+    },
+    {
+      "epoch": 1.0605123793650366,
+      "grad_norm": 2.4881539344787598,
+      "learning_rate": 0.004980524812223236,
+      "loss": 2.5181,
+      "step": 78900
+    },
+    {
+      "epoch": 1.0618565015188581,
+      "grad_norm": 1.3180015087127686,
+      "learning_rate": 0.004980458712948045,
+      "loss": 2.5122,
+      "step": 79000
+    },
+    {
+      "epoch": 1.0618565015188581,
+      "eval_MaskedAccuracy": 0.4343943324864396,
+      "eval_loss": 2.8312034606933594,
+      "eval_runtime": 155.8402,
+      "eval_samples_per_second": 407.315,
+      "eval_steps_per_second": 1.591,
+      "step": 79000
+    },
+    {
+      "epoch": 1.0632006236726794,
+      "grad_norm": 0.38148483633995056,
+      "learning_rate": 0.004980392502132505,
+      "loss": 2.5165,
+      "step": 79100
+    },
+    {
+      "epoch": 1.0645447458265007,
+      "grad_norm": 0.5590552091598511,
+      "learning_rate": 0.004980326179779604,
+      "loss": 2.5068,
+      "step": 79200
+    },
+    {
+      "epoch": 1.065888867980322,
+      "grad_norm": 0.4895831346511841,
+      "learning_rate": 0.004980259745892328,
+      "loss": 2.5226,
+      "step": 79300
+    },
+    {
+      "epoch": 1.0672329901341433,
+      "grad_norm": 0.5289179086685181,
+      "learning_rate": 0.004980193200473662,
+      "loss": 2.5117,
+      "step": 79400
+    },
+    {
+      "epoch": 1.0685771122879648,
+      "grad_norm": 1.0307695865631104,
+      "learning_rate": 0.004980126543526612,
+      "loss": 2.5128,
+      "step": 79500
+    },
+    {
+      "epoch": 1.069921234441786,
+      "grad_norm": 1.9092541933059692,
+      "learning_rate": 0.004980059775054183,
+      "loss": 2.5241,
+      "step": 79600
+    },
+    {
+      "epoch": 1.0712653565956074,
+      "grad_norm": 0.8294230103492737,
+      "learning_rate": 0.004979992895059378,
+      "loss": 2.5195,
+      "step": 79700
+    },
+    {
+      "epoch": 1.0726094787494287,
+      "grad_norm": 3.4225094318389893,
+      "learning_rate": 0.004979925903545211,
+      "loss": 2.517,
+      "step": 79800
+    },
+    {
+      "epoch": 1.0739536009032502,
+      "grad_norm": 0.4758341312408447,
+      "learning_rate": 0.004979858800514705,
+      "loss": 2.5176,
+      "step": 79900
+    },
+    {
+      "epoch": 1.0752977230570715,
+      "grad_norm": 0.5281692147254944,
+      "learning_rate": 0.00497979158597088,
+      "loss": 2.522,
+      "step": 80000
+    },
+    {
+      "epoch": 1.0752977230570715,
+      "eval_MaskedAccuracy": 0.4351065125485484,
+      "eval_loss": 2.828768491744995,
+      "eval_runtime": 155.597,
+      "eval_samples_per_second": 407.951,
+      "eval_steps_per_second": 1.594,
+      "step": 80000
+    },
+    {
+      "epoch": 1.0766418452108928,
+      "grad_norm": 0.611584484577179,
+      "learning_rate": 0.004979724259916765,
+      "loss": 2.5146,
+      "step": 80100
+    },
+    {
+      "epoch": 1.077985967364714,
+      "grad_norm": 0.5003161430358887,
+      "learning_rate": 0.0049796568223553946,
+      "loss": 2.519,
+      "step": 80200
+    },
+    {
+      "epoch": 1.0793300895185354,
+      "grad_norm": 0.8842087984085083,
+      "learning_rate": 0.004979589273289808,
+      "loss": 2.5081,
+      "step": 80300
+    },
+    {
+      "epoch": 1.0806742116723567,
+      "grad_norm": 0.86569744348526,
+      "learning_rate": 0.004979521612723048,
+      "loss": 2.5062,
+      "step": 80400
+    },
+    {
+      "epoch": 1.0820183338261782,
+      "grad_norm": 0.7057967782020569,
+      "learning_rate": 0.004979453840658162,
+      "loss": 2.5149,
+      "step": 80500
+    },
+    {
+      "epoch": 1.0833624559799995,
+      "grad_norm": 0.7422287464141846,
+      "learning_rate": 0.004979385957098205,
+      "loss": 2.5153,
+      "step": 80600
+    },
+    {
+      "epoch": 1.0847065781338208,
+      "grad_norm": 0.6432771682739258,
+      "learning_rate": 0.004979317962046237,
+      "loss": 2.5157,
+      "step": 80700
+    },
+    {
+      "epoch": 1.086050700287642,
+      "grad_norm": 0.5962923169136047,
+      "learning_rate": 0.004979249855505322,
+      "loss": 2.5068,
+      "step": 80800
+    },
+    {
+      "epoch": 1.0873948224414636,
+      "grad_norm": 0.939208984375,
+      "learning_rate": 0.004979181637478523,
+      "loss": 2.5066,
+      "step": 80900
+    },
+    {
+      "epoch": 1.0887389445952849,
+      "grad_norm": 0.3996613025665283,
+      "learning_rate": 0.004979113307968915,
+      "loss": 2.5086,
+      "step": 81000
+    },
+    {
+      "epoch": 1.0887389445952849,
+      "eval_MaskedAccuracy": 0.4360474147201799,
+      "eval_loss": 2.824847936630249,
+      "eval_runtime": 156.0397,
+      "eval_samples_per_second": 406.794,
+      "eval_steps_per_second": 1.589,
+      "step": 81000
+    },
+    {
+      "epoch": 1.0900830667491062,
+      "grad_norm": 0.3728293478488922,
+      "learning_rate": 0.004979044866979578,
+      "loss": 2.5134,
+      "step": 81100
+    },
+    {
+      "epoch": 1.0914271889029274,
+      "grad_norm": 1.5539531707763672,
+      "learning_rate": 0.0049789763145136005,
+      "loss": 2.5235,
+      "step": 81200
+    },
+    {
+      "epoch": 1.0927713110567487,
+      "grad_norm": 1.7607640027999878,
+      "learning_rate": 0.004978907650574065,
+      "loss": 2.5174,
+      "step": 81300
+    },
+    {
+      "epoch": 1.0941154332105703,
+      "grad_norm": 1.1798290014266968,
+      "learning_rate": 0.004978838875164072,
+      "loss": 2.5081,
+      "step": 81400
+    },
+    {
+      "epoch": 1.0954595553643915,
+      "grad_norm": 0.4024448096752167,
+      "learning_rate": 0.00497876998828672,
+      "loss": 2.5172,
+      "step": 81500
+    },
+    {
+      "epoch": 1.0968036775182128,
+      "grad_norm": 0.5295103788375854,
+      "learning_rate": 0.004978700989945107,
+      "loss": 2.5015,
+      "step": 81600
+    },
+    {
+      "epoch": 1.0981477996720341,
+      "grad_norm": 0.8088193535804749,
+      "learning_rate": 0.0049786318801423425,
+      "loss": 2.508,
+      "step": 81700
+    },
+    {
+      "epoch": 1.0994919218258556,
+      "grad_norm": 0.8756518363952637,
+      "learning_rate": 0.004978562658881545,
+      "loss": 2.5063,
+      "step": 81800
+    },
+    {
+      "epoch": 1.100836043979677,
+      "grad_norm": 4.078980922698975,
+      "learning_rate": 0.004978493326165831,
+      "loss": 2.5063,
+      "step": 81900
+    },
+    {
+      "epoch": 1.1021801661334982,
+      "grad_norm": 1.339497685432434,
+      "learning_rate": 0.0049784238819983295,
+      "loss": 2.5074,
+      "step": 82000
+    },
+    {
+      "epoch": 1.1021801661334982,
+      "eval_MaskedAccuracy": 0.43618356871003144,
+      "eval_loss": 2.82234525680542,
+      "eval_runtime": 156.3995,
+      "eval_samples_per_second": 405.858,
+      "eval_steps_per_second": 1.586,
+      "step": 82000
+    },
+    {
+      "epoch": 1.1035242882873195,
+      "grad_norm": 1.6204960346221924,
+      "learning_rate": 0.004978354326382158,
+      "loss": 2.5152,
+      "step": 82100
+    },
+    {
+      "epoch": 1.1048684104411408,
+      "grad_norm": 0.4489379823207855,
+      "learning_rate": 0.00497828465932046,
+      "loss": 2.5102,
+      "step": 82200
+    },
+    {
+      "epoch": 1.1062125325949623,
+      "grad_norm": 1.3593122959136963,
+      "learning_rate": 0.004978214880816373,
+      "loss": 2.5025,
+      "step": 82300
+    },
+    {
+      "epoch": 1.1075566547487836,
+      "grad_norm": 0.4765743017196655,
+      "learning_rate": 0.004978144990873039,
+      "loss": 2.5071,
+      "step": 82400
+    },
+    {
+      "epoch": 1.108900776902605,
+      "grad_norm": 1.4956125020980835,
+      "learning_rate": 0.004978074989493609,
+      "loss": 2.5068,
+      "step": 82500
+    },
+    {
+      "epoch": 1.1102448990564262,
+      "grad_norm": 0.820233941078186,
+      "learning_rate": 0.004978004876681244,
+      "loss": 2.5143,
+      "step": 82600
+    },
+    {
+      "epoch": 1.1115890212102475,
+      "grad_norm": 0.39537087082862854,
+      "learning_rate": 0.004977934652439093,
+      "loss": 2.5172,
+      "step": 82700
+    },
+    {
+      "epoch": 1.112933143364069,
+      "grad_norm": 2.4342026710510254,
+      "learning_rate": 0.004977864316770319,
+      "loss": 2.5119,
+      "step": 82800
+    },
+    {
+      "epoch": 1.1142772655178903,
+      "grad_norm": 1.4095677137374878,
+      "learning_rate": 0.004977793869678101,
+      "loss": 2.5082,
+      "step": 82900
+    },
+    {
+      "epoch": 1.1156213876717116,
+      "grad_norm": 0.6153007745742798,
+      "learning_rate": 0.004977723311165601,
+      "loss": 2.5059,
+      "step": 83000
+    },
+    {
+      "epoch": 1.1156213876717116,
+      "eval_MaskedAccuracy": 0.43704451908857084,
+      "eval_loss": 2.815913200378418,
+      "eval_runtime": 155.698,
+      "eval_samples_per_second": 407.687,
+      "eval_steps_per_second": 1.593,
+      "step": 83000
+    },
+    {
+      "epoch": 1.1169655098255329,
+      "grad_norm": 0.7392873764038086,
+      "learning_rate": 0.004977652641236012,
+      "loss": 2.5019,
+      "step": 83100
+    },
+    {
+      "epoch": 1.1183096319793542,
+      "grad_norm": 2.7315380573272705,
+      "learning_rate": 0.004977581859892506,
+      "loss": 2.5046,
+      "step": 83200
+    },
+    {
+      "epoch": 1.1196537541331757,
+      "grad_norm": 0.5851826667785645,
+      "learning_rate": 0.0049775109671382715,
+      "loss": 2.5031,
+      "step": 83300
+    },
+    {
+      "epoch": 1.120997876286997,
+      "grad_norm": 0.34558480978012085,
+      "learning_rate": 0.004977439962976512,
+      "loss": 2.5129,
+      "step": 83400
+    },
+    {
+      "epoch": 1.1223419984408183,
+      "grad_norm": 0.9053849577903748,
+      "learning_rate": 0.004977368847410425,
+      "loss": 2.5075,
+      "step": 83500
+    },
+    {
+      "epoch": 1.1236861205946396,
+      "grad_norm": 1.7211493253707886,
+      "learning_rate": 0.004977297620443216,
+      "loss": 2.5035,
+      "step": 83600
+    },
+    {
+      "epoch": 1.125030242748461,
+      "grad_norm": 0.7423088550567627,
+      "learning_rate": 0.004977226282078084,
+      "loss": 2.5118,
+      "step": 83700
+    },
+    {
+      "epoch": 1.1263743649022824,
+      "grad_norm": 1.4103665351867676,
+      "learning_rate": 0.004977154832318255,
+      "loss": 2.5172,
+      "step": 83800
+    },
+    {
+      "epoch": 1.1277184870561037,
+      "grad_norm": 2.3322620391845703,
+      "learning_rate": 0.0049770832711669415,
+      "loss": 2.4982,
+      "step": 83900
+    },
+    {
+      "epoch": 1.129062609209925,
+      "grad_norm": 0.7170735001564026,
+      "learning_rate": 0.004977011598627376,
+      "loss": 2.5056,
+      "step": 84000
+    },
+    {
+      "epoch": 1.129062609209925,
+      "eval_MaskedAccuracy": 0.4369403158372308,
+      "eval_loss": 2.8143200874328613,
+      "eval_runtime": 156.8035,
+      "eval_samples_per_second": 404.812,
+      "eval_steps_per_second": 1.582,
+      "step": 84000
+    },
+    {
+      "epoch": 1.1304067313637463,
+      "grad_norm": 0.4916757643222809,
+      "learning_rate": 0.004976939814702779,
+      "loss": 2.5041,
+      "step": 84100
+    },
+    {
+      "epoch": 1.1317508535175678,
+      "grad_norm": 1.2962486743927002,
+      "learning_rate": 0.004976867919396393,
+      "loss": 2.5092,
+      "step": 84200
+    },
+    {
+      "epoch": 1.133094975671389,
+      "grad_norm": 0.47311854362487793,
+      "learning_rate": 0.004976795912711456,
+      "loss": 2.5074,
+      "step": 84300
+    },
+    {
+      "epoch": 1.1344390978252104,
+      "grad_norm": 0.4036632180213928,
+      "learning_rate": 0.004976723794651204,
+      "loss": 2.5089,
+      "step": 84400
+    },
+    {
+      "epoch": 1.1357832199790316,
+      "grad_norm": 1.4147541522979736,
+      "learning_rate": 0.004976651565218892,
+      "loss": 2.5074,
+      "step": 84500
+    },
+    {
+      "epoch": 1.1371273421328532,
+      "grad_norm": 1.4491769075393677,
+      "learning_rate": 0.004976579224417779,
+      "loss": 2.5051,
+      "step": 84600
+    },
+    {
+      "epoch": 1.1384714642866745,
+      "grad_norm": 1.161087155342102,
+      "learning_rate": 0.0049765067722511176,
+      "loss": 2.5125,
+      "step": 84700
+    },
+    {
+      "epoch": 1.1398155864404957,
+      "grad_norm": 0.37587714195251465,
+      "learning_rate": 0.004976434208722176,
+      "loss": 2.5046,
+      "step": 84800
+    },
+    {
+      "epoch": 1.141159708594317,
+      "grad_norm": 0.6457163095474243,
+      "learning_rate": 0.004976361533834221,
+      "loss": 2.5107,
+      "step": 84900
+    },
+    {
+      "epoch": 1.1425038307481383,
+      "grad_norm": 0.40892913937568665,
+      "learning_rate": 0.004976288747590531,
+      "loss": 2.5095,
+      "step": 85000
+    },
+    {
+      "epoch": 1.1425038307481383,
+      "eval_MaskedAccuracy": 0.4367847754877898,
+      "eval_loss": 2.817359685897827,
+      "eval_runtime": 155.4685,
+      "eval_samples_per_second": 408.288,
+      "eval_steps_per_second": 1.595,
+      "step": 85000
+    },
+    {
+      "epoch": 1.1438479529019596,
+      "grad_norm": 0.6328030228614807,
+      "learning_rate": 0.00497621584999438,
+      "loss": 2.5027,
+      "step": 85100
+    },
+    {
+      "epoch": 1.1451920750557811,
+      "grad_norm": 0.8241310119628906,
+      "learning_rate": 0.004976142841049059,
+      "loss": 2.5109,
+      "step": 85200
+    },
+    {
+      "epoch": 1.1465361972096024,
+      "grad_norm": 0.5292043089866638,
+      "learning_rate": 0.0049760697207578515,
+      "loss": 2.5051,
+      "step": 85300
+    },
+    {
+      "epoch": 1.1478803193634237,
+      "grad_norm": 0.6570155620574951,
+      "learning_rate": 0.004975996489124052,
+      "loss": 2.4996,
+      "step": 85400
+    },
+    {
+      "epoch": 1.149224441517245,
+      "grad_norm": 0.3944413959980011,
+      "learning_rate": 0.0049759231461509726,
+      "loss": 2.5018,
+      "step": 85500
+    },
+    {
+      "epoch": 1.1505685636710665,
+      "grad_norm": 0.7858182787895203,
+      "learning_rate": 0.00497584969184191,
+      "loss": 2.4994,
+      "step": 85600
+    },
+    {
+      "epoch": 1.1519126858248878,
+      "grad_norm": 0.354902058839798,
+      "learning_rate": 0.004975776126200173,
+      "loss": 2.5109,
+      "step": 85700
+    },
+    {
+      "epoch": 1.153256807978709,
+      "grad_norm": 1.0036402940750122,
+      "learning_rate": 0.004975702449229073,
+      "loss": 2.5043,
+      "step": 85800
+    },
+    {
+      "epoch": 1.1546009301325304,
+      "grad_norm": 0.9507333636283875,
+      "learning_rate": 0.004975628660931937,
+      "loss": 2.502,
+      "step": 85900
+    },
+    {
+      "epoch": 1.1559450522863517,
+      "grad_norm": 0.5167057514190674,
+      "learning_rate": 0.004975554761312083,
+      "loss": 2.5026,
+      "step": 86000
+    },
+    {
+      "epoch": 1.1559450522863517,
+      "eval_MaskedAccuracy": 0.43768769620070525,
+      "eval_loss": 2.8092353343963623,
+      "eval_runtime": 156.668,
+      "eval_samples_per_second": 405.162,
+      "eval_steps_per_second": 1.583,
+      "step": 86000
+    },
+    {
+      "epoch": 1.1572891744401732,
+      "grad_norm": 0.5097525715827942,
+      "learning_rate": 0.004975480750372851,
+      "loss": 2.5018,
+      "step": 86100
+    },
+    {
+      "epoch": 1.1586332965939945,
+      "grad_norm": 0.35351938009262085,
+      "learning_rate": 0.0049754066281175605,
+      "loss": 2.501,
+      "step": 86200
+    },
+    {
+      "epoch": 1.1599774187478158,
+      "grad_norm": 0.8449835181236267,
+      "learning_rate": 0.004975332394549564,
+      "loss": 2.5024,
+      "step": 86300
+    },
+    {
+      "epoch": 1.161321540901637,
+      "grad_norm": 63.61347579956055,
+      "learning_rate": 0.004975258049672201,
+      "loss": 2.5018,
+      "step": 86400
+    },
+    {
+      "epoch": 1.1626656630554586,
+      "grad_norm": 0.5342414975166321,
+      "learning_rate": 0.004975183593488831,
+      "loss": 2.6116,
+      "step": 86500
+    },
+    {
+      "epoch": 1.16400978520928,
+      "grad_norm": 0.47309231758117676,
+      "learning_rate": 0.0049751090260028,
+      "loss": 2.5154,
+      "step": 86600
+    },
+    {
+      "epoch": 1.1653539073631012,
+      "grad_norm": 1.168167233467102,
+      "learning_rate": 0.004975034347217465,
+      "loss": 2.5076,
+      "step": 86700
+    },
+    {
+      "epoch": 1.1666980295169225,
+      "grad_norm": 0.49217191338539124,
+      "learning_rate": 0.004974959557136195,
+      "loss": 2.5049,
+      "step": 86800
+    },
+    {
+      "epoch": 1.1680421516707438,
+      "grad_norm": 2.020880937576294,
+      "learning_rate": 0.004974884655762357,
+      "loss": 2.5017,
+      "step": 86900
+    },
+    {
+      "epoch": 1.169386273824565,
+      "grad_norm": 1.273708701133728,
+      "learning_rate": 0.004974809643099329,
+      "loss": 2.503,
+      "step": 87000
+    },
+    {
+      "epoch": 1.169386273824565,
+      "eval_MaskedAccuracy": 0.4362821121410326,
+      "eval_loss": 2.820596218109131,
+      "eval_runtime": 155.9859,
+      "eval_samples_per_second": 406.934,
+      "eval_steps_per_second": 1.59,
+      "step": 87000
+    },
+    {
+      "epoch": 1.1707303959783866,
+      "grad_norm": 0.9947301745414734,
+      "learning_rate": 0.004974734519150486,
+      "loss": 2.5085,
+      "step": 87100
+    },
+    {
+      "epoch": 1.1720745181322079,
+      "grad_norm": 0.36364173889160156,
+      "learning_rate": 0.004974659283919222,
+      "loss": 2.52,
+      "step": 87200
+    },
+    {
+      "epoch": 1.1734186402860292,
+      "grad_norm": 1.1245728731155396,
+      "learning_rate": 0.0049745839374089225,
+      "loss": 2.5178,
+      "step": 87300
+    },
+    {
+      "epoch": 1.1747627624398504,
+      "grad_norm": 0.8172126412391663,
+      "learning_rate": 0.004974508479622976,
+      "loss": 2.5063,
+      "step": 87400
+    },
+    {
+      "epoch": 1.176106884593672,
+      "grad_norm": 0.5378974676132202,
+      "learning_rate": 0.004974432910564789,
+      "loss": 2.5085,
+      "step": 87500
+    },
+    {
+      "epoch": 1.1774510067474933,
+      "grad_norm": 1.354956030845642,
+      "learning_rate": 0.00497435723023777,
+      "loss": 2.4993,
+      "step": 87600
+    },
+    {
+      "epoch": 1.1787951289013145,
+      "grad_norm": 1.2725735902786255,
+      "learning_rate": 0.00497428143864532,
+      "loss": 2.4983,
+      "step": 87700
+    },
+    {
+      "epoch": 1.1801392510551358,
+      "grad_norm": 1.188934326171875,
+      "learning_rate": 0.004974205535790863,
+      "loss": 2.4999,
+      "step": 87800
+    },
+    {
+      "epoch": 1.1814833732089571,
+      "grad_norm": 1.1253455877304077,
+      "learning_rate": 0.004974129521677812,
+      "loss": 2.496,
+      "step": 87900
+    },
+    {
+      "epoch": 1.1828274953627786,
+      "grad_norm": 0.6546825170516968,
+      "learning_rate": 0.004974053396309598,
+      "loss": 2.5013,
+      "step": 88000
+    },
+    {
+      "epoch": 1.1828274953627786,
+      "eval_MaskedAccuracy": 0.43612094406632873,
+      "eval_loss": 2.817842721939087,
+      "eval_runtime": 155.3942,
+      "eval_samples_per_second": 408.484,
+      "eval_steps_per_second": 1.596,
+      "step": 88000
+    },
+    {
+      "epoch": 1.1841716175166,
+      "grad_norm": 0.5742254257202148,
+      "learning_rate": 0.004973977159689648,
+      "loss": 2.4927,
+      "step": 88100
+    },
+    {
+      "epoch": 1.1855157396704212,
+      "grad_norm": 0.9239218831062317,
+      "learning_rate": 0.0049739008118213955,
+      "loss": 2.4991,
+      "step": 88200
+    },
+    {
+      "epoch": 1.1868598618242425,
+      "grad_norm": 0.8331120014190674,
+      "learning_rate": 0.004973824352708291,
+      "loss": 2.5034,
+      "step": 88300
+    },
+    {
+      "epoch": 1.188203983978064,
+      "grad_norm": 0.5568376779556274,
+      "learning_rate": 0.004973747782353763,
+      "loss": 2.4976,
+      "step": 88400
+    },
+    {
+      "epoch": 1.1895481061318853,
+      "grad_norm": 0.7204086184501648,
+      "learning_rate": 0.004973671100761278,
+      "loss": 2.4977,
+      "step": 88500
+    },
+    {
+      "epoch": 1.1908922282857066,
+      "grad_norm": 0.5486904382705688,
+      "learning_rate": 0.004973594307934272,
+      "loss": 2.5017,
+      "step": 88600
+    },
+    {
+      "epoch": 1.192236350439528,
+      "grad_norm": 0.8633172512054443,
+      "learning_rate": 0.0049735174038762235,
+      "loss": 2.5038,
+      "step": 88700
+    },
+    {
+      "epoch": 1.1935804725933492,
+      "grad_norm": 0.5683305859565735,
+      "learning_rate": 0.0049734403885905885,
+      "loss": 2.5001,
+      "step": 88800
+    },
+    {
+      "epoch": 1.1949245947471705,
+      "grad_norm": 0.5062243342399597,
+      "learning_rate": 0.004973363262080839,
+      "loss": 2.5065,
+      "step": 88900
+    },
+    {
+      "epoch": 1.196268716900992,
+      "grad_norm": 3.755258321762085,
+      "learning_rate": 0.004973286024350451,
+      "loss": 2.491,
+      "step": 89000
+    },
+    {
+      "epoch": 1.196268716900992,
+      "eval_MaskedAccuracy": 0.4371673629048697,
+      "eval_loss": 2.814087390899658,
+      "eval_runtime": 156.1272,
+      "eval_samples_per_second": 406.566,
+      "eval_steps_per_second": 1.588,
+      "step": 89000
+    },
+    {
+      "epoch": 1.1976128390548133,
+      "grad_norm": 0.8349953293800354,
+      "learning_rate": 0.004973208675402902,
+      "loss": 2.5118,
+      "step": 89100
+    },
+    {
+      "epoch": 1.1989569612086346,
+      "grad_norm": 1.188719391822815,
+      "learning_rate": 0.004973131215241677,
+      "loss": 2.5103,
+      "step": 89200
+    },
+    {
+      "epoch": 1.200301083362456,
+      "grad_norm": 1.0278277397155762,
+      "learning_rate": 0.004973053643870272,
+      "loss": 2.4956,
+      "step": 89300
+    },
+    {
+      "epoch": 1.2016452055162774,
+      "grad_norm": 1.174872636795044,
+      "learning_rate": 0.004972975961292177,
+      "loss": 2.4981,
+      "step": 89400
+    },
+    {
+      "epoch": 1.2029893276700987,
+      "grad_norm": 1.0490007400512695,
+      "learning_rate": 0.004972898167510894,
+      "loss": 2.5007,
+      "step": 89500
+    },
+    {
+      "epoch": 1.20433344982392,
+      "grad_norm": 1.1336147785186768,
+      "learning_rate": 0.004972820262529928,
+      "loss": 2.5084,
+      "step": 89600
+    },
+    {
+      "epoch": 1.2056775719777413,
+      "grad_norm": 0.770378589630127,
+      "learning_rate": 0.004972742246352788,
+      "loss": 2.4995,
+      "step": 89700
+    },
+    {
+      "epoch": 1.2070216941315626,
+      "grad_norm": 1.1456156969070435,
+      "learning_rate": 0.00497266411898299,
+      "loss": 2.4966,
+      "step": 89800
+    },
+    {
+      "epoch": 1.208365816285384,
+      "grad_norm": 2.0087265968322754,
+      "learning_rate": 0.004972585880424056,
+      "loss": 2.4936,
+      "step": 89900
+    },
+    {
+      "epoch": 1.2097099384392054,
+      "grad_norm": 1.6110386848449707,
+      "learning_rate": 0.0049725075306795135,
+      "loss": 2.4972,
+      "step": 90000
+    },
+    {
+      "epoch": 1.2097099384392054,
+      "eval_MaskedAccuracy": 0.4370350689102483,
+      "eval_loss": 2.812059164047241,
+      "eval_runtime": 155.8421,
+      "eval_samples_per_second": 407.31,
+      "eval_steps_per_second": 1.591,
+      "step": 90000
+    },
+    {
+      "epoch": 1.2110540605930267,
+      "grad_norm": 0.35183417797088623,
+      "learning_rate": 0.004972429069752887,
+      "loss": 2.4858,
+      "step": 90100
+    },
+    {
+      "epoch": 1.212398182746848,
+      "grad_norm": 0.4271332621574402,
+      "learning_rate": 0.004972350497647718,
+      "loss": 2.4989,
+      "step": 90200
+    },
+    {
+      "epoch": 1.2137423049006695,
+      "grad_norm": 3.801957845687866,
+      "learning_rate": 0.004972271814367547,
+      "loss": 2.4984,
+      "step": 90300
+    },
+    {
+      "epoch": 1.2150864270544908,
+      "grad_norm": 2.1145997047424316,
+      "learning_rate": 0.004972193019915915,
+      "loss": 2.4964,
+      "step": 90400
+    },
+    {
+      "epoch": 1.216430549208312,
+      "grad_norm": 0.3601829707622528,
+      "learning_rate": 0.004972114114296368,
+      "loss": 2.5001,
+      "step": 90500
+    },
+    {
+      "epoch": 1.2177746713621334,
+      "grad_norm": 0.31775668263435364,
+      "learning_rate": 0.004972035097512468,
+      "loss": 2.4985,
+      "step": 90600
+    },
+    {
+      "epoch": 1.2191187935159546,
+      "grad_norm": 0.4949300289154053,
+      "learning_rate": 0.004971955969567773,
+      "loss": 2.4992,
+      "step": 90700
+    },
+    {
+      "epoch": 1.2204629156697762,
+      "grad_norm": 0.5731763243675232,
+      "learning_rate": 0.004971876730465848,
+      "loss": 2.4985,
+      "step": 90800
+    },
+    {
+      "epoch": 1.2218070378235975,
+      "grad_norm": 0.6746828556060791,
+      "learning_rate": 0.004971797380210261,
+      "loss": 2.4902,
+      "step": 90900
+    },
+    {
+      "epoch": 1.2231511599774187,
+      "grad_norm": 2.1242833137512207,
+      "learning_rate": 0.004971717918804597,
+      "loss": 2.5026,
+      "step": 91000
+    },
+    {
+      "epoch": 1.2231511599774187,
+      "eval_MaskedAccuracy": 0.4372796066068638,
+      "eval_loss": 2.8106234073638916,
+      "eval_runtime": 155.6365,
+      "eval_samples_per_second": 407.848,
+      "eval_steps_per_second": 1.593,
+      "step": 91000
+    },
+    {
+      "epoch": 1.22449528213124,
+      "grad_norm": 1.0051345825195312,
+      "learning_rate": 0.004971638346252431,
+      "loss": 2.484,
+      "step": 91100
+    },
+    {
+      "epoch": 1.2258394042850613,
+      "grad_norm": 0.33799469470977783,
+      "learning_rate": 0.0049715586625573484,
+      "loss": 2.4978,
+      "step": 91200
+    },
+    {
+      "epoch": 1.2271835264388828,
+      "grad_norm": 0.37209373712539673,
+      "learning_rate": 0.004971478867722932,
+      "loss": 2.5002,
+      "step": 91300
+    },
+    {
+      "epoch": 1.2285276485927041,
+      "grad_norm": 0.9765738248825073,
+      "learning_rate": 0.004971398961752787,
+      "loss": 2.4962,
+      "step": 91400
+    },
+    {
+      "epoch": 1.2298717707465254,
+      "grad_norm": 1.2135403156280518,
+      "learning_rate": 0.004971318944650509,
+      "loss": 2.4967,
+      "step": 91500
+    },
+    {
+      "epoch": 1.2312158929003467,
+      "grad_norm": 0.776421070098877,
+      "learning_rate": 0.004971238816419708,
+      "loss": 2.4915,
+      "step": 91600
+    },
+    {
+      "epoch": 1.232560015054168,
+      "grad_norm": 0.8908390998840332,
+      "learning_rate": 0.004971158577063992,
+      "loss": 2.5007,
+      "step": 91700
+    },
+    {
+      "epoch": 1.2339041372079895,
+      "grad_norm": 0.3629126250743866,
+      "learning_rate": 0.004971078226586978,
+      "loss": 2.4831,
+      "step": 91800
+    },
+    {
+      "epoch": 1.2352482593618108,
+      "grad_norm": 0.5708677172660828,
+      "learning_rate": 0.004970997764992286,
+      "loss": 2.488,
+      "step": 91900
+    },
+    {
+      "epoch": 1.236592381515632,
+      "grad_norm": 1.317496418952942,
+      "learning_rate": 0.004970917192283536,
+      "loss": 2.4921,
+      "step": 92000
+    },
+    {
+      "epoch": 1.236592381515632,
+      "eval_MaskedAccuracy": 0.43781475066312364,
+      "eval_loss": 2.807178020477295,
+      "eval_runtime": 155.1214,
+      "eval_samples_per_second": 409.202,
+      "eval_steps_per_second": 1.599,
+      "step": 92000
+    },
+    {
+      "epoch": 1.2379365036694534,
+      "grad_norm": 1.5067718029022217,
+      "learning_rate": 0.004970836508464369,
+      "loss": 2.4944,
+      "step": 92100
+    },
+    {
+      "epoch": 1.239280625823275,
+      "grad_norm": 0.6655193567276001,
+      "learning_rate": 0.004970755713538412,
+      "loss": 2.4849,
+      "step": 92200
+    },
+    {
+      "epoch": 1.2406247479770962,
+      "grad_norm": 0.9877846837043762,
+      "learning_rate": 0.004970674807509303,
+      "loss": 2.4908,
+      "step": 92300
+    },
+    {
+      "epoch": 1.2419688701309175,
+      "grad_norm": 0.47736746072769165,
+      "learning_rate": 0.004970593790380696,
+      "loss": 2.498,
+      "step": 92400
+    },
+    {
+      "epoch": 1.2433129922847388,
+      "grad_norm": 0.6927477121353149,
+      "learning_rate": 0.004970512662156239,
+      "loss": 2.4849,
+      "step": 92500
+    },
+    {
+      "epoch": 1.24465711443856,
+      "grad_norm": 0.5318709015846252,
+      "learning_rate": 0.004970431422839584,
+      "loss": 2.4965,
+      "step": 92600
+    },
+    {
+      "epoch": 1.2460012365923816,
+      "grad_norm": 3.21836519241333,
+      "learning_rate": 0.004970350072434393,
+      "loss": 2.5004,
+      "step": 92700
+    },
+    {
+      "epoch": 1.247345358746203,
+      "grad_norm": 0.5743933320045471,
+      "learning_rate": 0.004970268610944332,
+      "loss": 2.4935,
+      "step": 92800
+    },
+    {
+      "epoch": 1.2486894809000242,
+      "grad_norm": 0.4667443335056305,
+      "learning_rate": 0.004970187038373073,
+      "loss": 2.4883,
+      "step": 92900
+    },
+    {
+      "epoch": 1.2500336030538455,
+      "grad_norm": 0.6037049889564514,
+      "learning_rate": 0.00497010535472429,
+      "loss": 2.4922,
+      "step": 93000
+    },
+    {
+      "epoch": 1.2500336030538455,
+      "eval_MaskedAccuracy": 0.4379076740559415,
+      "eval_loss": 2.804267644882202,
+      "eval_runtime": 157.0744,
+      "eval_samples_per_second": 404.114,
+      "eval_steps_per_second": 1.579,
+      "step": 93000
+    },
+    {
+      "epoch": 1.251377725207667,
+      "grad_norm": 0.6085057854652405,
+      "learning_rate": 0.004970023560001663,
+      "loss": 2.4916,
+      "step": 93100
+    },
+    {
+      "epoch": 1.2527218473614883,
+      "grad_norm": 0.8414444923400879,
+      "learning_rate": 0.00496994165420888,
+      "loss": 2.4915,
+      "step": 93200
+    },
+    {
+      "epoch": 1.2540659695153096,
+      "grad_norm": 0.962984561920166,
+      "learning_rate": 0.004969859637349628,
+      "loss": 2.493,
+      "step": 93300
+    },
+    {
+      "epoch": 1.2554100916691309,
+      "grad_norm": 0.9756949543952942,
+      "learning_rate": 0.0049697775094276085,
+      "loss": 2.496,
+      "step": 93400
+    },
+    {
+      "epoch": 1.2567542138229522,
+      "grad_norm": 1.2727997303009033,
+      "learning_rate": 0.004969695270446518,
+      "loss": 2.4915,
+      "step": 93500
+    },
+    {
+      "epoch": 1.2580983359767735,
+      "grad_norm": 1.2825567722320557,
+      "learning_rate": 0.004969612920410054,
+      "loss": 2.4862,
+      "step": 93600
+    },
+    {
+      "epoch": 1.259442458130595,
+      "grad_norm": 0.7585506439208984,
+      "learning_rate": 0.004969530459321946,
+      "loss": 2.494,
+      "step": 93700
+    },
+    {
+      "epoch": 1.2607865802844163,
+      "grad_norm": 0.6428466439247131,
+      "learning_rate": 0.004969447887185897,
+      "loss": 2.4782,
+      "step": 93800
+    },
+    {
+      "epoch": 1.2621307024382376,
+      "grad_norm": 0.8756228089332581,
+      "learning_rate": 0.004969365204005632,
+      "loss": 2.4977,
+      "step": 93900
+    },
+    {
+      "epoch": 1.263474824592059,
+      "grad_norm": 0.36692315340042114,
+      "learning_rate": 0.004969282409784867,
+      "loss": 2.4931,
+      "step": 94000
+    },
+    {
+      "epoch": 1.263474824592059,
+      "eval_MaskedAccuracy": 0.43933527822037677,
+      "eval_loss": 2.797823429107666,
+      "eval_runtime": 155.1464,
+      "eval_samples_per_second": 409.136,
+      "eval_steps_per_second": 1.598,
+      "step": 94000
+    },
+    {
+      "epoch": 1.2648189467458804,
+      "grad_norm": 2.2874348163604736,
+      "learning_rate": 0.004969199504527345,
+      "loss": 2.4854,
+      "step": 94100
+    },
+    {
+      "epoch": 1.2661630688997016,
+      "grad_norm": 1.3956974744796753,
+      "learning_rate": 0.0049691164882367895,
+      "loss": 2.4901,
+      "step": 94200
+    },
+    {
+      "epoch": 1.267507191053523,
+      "grad_norm": 0.6008788347244263,
+      "learning_rate": 0.004969033360916954,
+      "loss": 2.4849,
+      "step": 94300
+    },
+    {
+      "epoch": 1.2688513132073442,
+      "grad_norm": 1.0495715141296387,
+      "learning_rate": 0.004968950122571579,
+      "loss": 2.4919,
+      "step": 94400
+    },
+    {
+      "epoch": 1.2701954353611655,
+      "grad_norm": 1.057627558708191,
+      "learning_rate": 0.00496886677320441,
+      "loss": 2.4909,
+      "step": 94500
+    },
+    {
+      "epoch": 1.271539557514987,
+      "grad_norm": 0.4978176951408386,
+      "learning_rate": 0.0049687833128192055,
+      "loss": 2.4938,
+      "step": 94600
+    },
+    {
+      "epoch": 1.2728836796688083,
+      "grad_norm": 0.5044616460800171,
+      "learning_rate": 0.004968699741419729,
+      "loss": 2.4907,
+      "step": 94700
+    },
+    {
+      "epoch": 1.2742278018226296,
+      "grad_norm": 0.37560752034187317,
+      "learning_rate": 0.00496861605900974,
+      "loss": 2.4896,
+      "step": 94800
+    },
+    {
+      "epoch": 1.275571923976451,
+      "grad_norm": 0.877223789691925,
+      "learning_rate": 0.004968532265593022,
+      "loss": 2.4882,
+      "step": 94900
+    },
+    {
+      "epoch": 1.2769160461302724,
+      "grad_norm": 1.2958379983901978,
+      "learning_rate": 0.004968448361173342,
+      "loss": 2.4856,
+      "step": 95000
+    },
+    {
+      "epoch": 1.2769160461302724,
+      "eval_MaskedAccuracy": 0.4387117447553094,
+      "eval_loss": 2.799860954284668,
+      "eval_runtime": 154.6014,
+      "eval_samples_per_second": 410.578,
+      "eval_steps_per_second": 1.604,
+      "step": 95000
+    },
+    {
+      "epoch": 1.2782601682840937,
+      "grad_norm": 0.3757873773574829,
+      "learning_rate": 0.004968364345754483,
+      "loss": 2.4836,
+      "step": 95100
+    },
+    {
+      "epoch": 1.279604290437915,
+      "grad_norm": 0.9294460415840149,
+      "learning_rate": 0.004968280219340219,
+      "loss": 2.4937,
+      "step": 95200
+    },
+    {
+      "epoch": 1.2809484125917363,
+      "grad_norm": 0.3245985805988312,
+      "learning_rate": 0.004968195981934356,
+      "loss": 2.4949,
+      "step": 95300
+    },
+    {
+      "epoch": 1.2822925347455576,
+      "grad_norm": 0.342230886220932,
+      "learning_rate": 0.004968111633540681,
+      "loss": 2.492,
+      "step": 95400
+    },
+    {
+      "epoch": 1.283636656899379,
+      "grad_norm": 1.0859248638153076,
+      "learning_rate": 0.004968027174162995,
+      "loss": 2.4855,
+      "step": 95500
+    },
+    {
+      "epoch": 1.2849807790532004,
+      "grad_norm": 0.45761579275131226,
+      "learning_rate": 0.0049679426038051104,
+      "loss": 2.4863,
+      "step": 95600
+    },
+    {
+      "epoch": 1.2863249012070217,
+      "grad_norm": 0.49789857864379883,
+      "learning_rate": 0.004967857922470824,
+      "loss": 2.4931,
+      "step": 95700
+    },
+    {
+      "epoch": 1.287669023360843,
+      "grad_norm": 0.39404183626174927,
+      "learning_rate": 0.004967773130163965,
+      "loss": 2.4774,
+      "step": 95800
+    },
+    {
+      "epoch": 1.2890131455146645,
+      "grad_norm": 0.3045353293418884,
+      "learning_rate": 0.0049676882268883515,
+      "loss": 2.4826,
+      "step": 95900
+    },
+    {
+      "epoch": 1.2903572676684858,
+      "grad_norm": 0.34618687629699707,
+      "learning_rate": 0.004967603212647805,
+      "loss": 2.4788,
+      "step": 96000
+    },
+    {
+      "epoch": 1.2903572676684858,
+      "eval_MaskedAccuracy": 0.43972966453745477,
+      "eval_loss": 2.7956392765045166,
+      "eval_runtime": 154.9374,
+      "eval_samples_per_second": 409.688,
+      "eval_steps_per_second": 1.601,
+      "step": 96000
+    },
+    {
+      "epoch": 1.291701389822307,
+      "grad_norm": 1.395040512084961,
+      "learning_rate": 0.004967518087446164,
+      "loss": 2.4897,
+      "step": 96100
+    },
+    {
+      "epoch": 1.2930455119761284,
+      "grad_norm": 0.49540796875953674,
+      "learning_rate": 0.0049674328512872585,
+      "loss": 2.4829,
+      "step": 96200
+    },
+    {
+      "epoch": 1.2943896341299497,
+      "grad_norm": 1.1871672868728638,
+      "learning_rate": 0.00496734750417493,
+      "loss": 2.4834,
+      "step": 96300
+    },
+    {
+      "epoch": 1.295733756283771,
+      "grad_norm": 0.9125702977180481,
+      "learning_rate": 0.004967262046113016,
+      "loss": 2.4926,
+      "step": 96400
+    },
+    {
+      "epoch": 1.2970778784375925,
+      "grad_norm": 0.8528000712394714,
+      "learning_rate": 0.004967176477105385,
+      "loss": 2.4924,
+      "step": 96500
+    },
+    {
+      "epoch": 1.2984220005914138,
+      "grad_norm": 0.3273971378803253,
+      "learning_rate": 0.0049670907971558756,
+      "loss": 2.481,
+      "step": 96600
+    },
+    {
+      "epoch": 1.299766122745235,
+      "grad_norm": 1.2552522420883179,
+      "learning_rate": 0.004967005006268358,
+      "loss": 2.4869,
+      "step": 96700
+    },
+    {
+      "epoch": 1.3011102448990564,
+      "grad_norm": 0.8510228991508484,
+      "learning_rate": 0.004966919104446687,
+      "loss": 2.4875,
+      "step": 96800
+    },
+    {
+      "epoch": 1.3024543670528779,
+      "grad_norm": 0.6518368124961853,
+      "learning_rate": 0.004966833091694741,
+      "loss": 2.4836,
+      "step": 96900
+    },
+    {
+      "epoch": 1.3037984892066992,
+      "grad_norm": 1.7670749425888062,
+      "learning_rate": 0.004966746968016394,
+      "loss": 2.4831,
+      "step": 97000
+    },
+    {
+      "epoch": 1.3037984892066992,
+      "eval_MaskedAccuracy": 0.4384087863650507,
+      "eval_loss": 2.801335573196411,
+      "eval_runtime": 156.4566,
+      "eval_samples_per_second": 405.71,
+      "eval_steps_per_second": 1.585,
+      "step": 97000
+    },
+    {
+      "epoch": 1.3051426113605205,
+      "grad_norm": 0.37767720222473145,
+      "learning_rate": 0.004966660733415522,
+      "loss": 2.4896,
+      "step": 97100
+    },
+    {
+      "epoch": 1.3064867335143417,
+      "grad_norm": 1.352961540222168,
+      "learning_rate": 0.004966574387896025,
+      "loss": 2.4821,
+      "step": 97200
+    },
+    {
+      "epoch": 1.307830855668163,
+      "grad_norm": 0.5152901411056519,
+      "learning_rate": 0.004966487931461782,
+      "loss": 2.4872,
+      "step": 97300
+    },
+    {
+      "epoch": 1.3091749778219843,
+      "grad_norm": 0.6184568405151367,
+      "learning_rate": 0.00496640136411669,
+      "loss": 2.4917,
+      "step": 97400
+    },
+    {
+      "epoch": 1.3105190999758058,
+      "grad_norm": 0.8382551074028015,
+      "learning_rate": 0.004966314685864655,
+      "loss": 2.4785,
+      "step": 97500
+    },
+    {
+      "epoch": 1.3118632221296271,
+      "grad_norm": 0.9581834077835083,
+      "learning_rate": 0.004966227896709579,
+      "loss": 2.4839,
+      "step": 97600
+    },
+    {
+      "epoch": 1.3132073442834484,
+      "grad_norm": 0.33205175399780273,
+      "learning_rate": 0.004966140996655376,
+      "loss": 2.4767,
+      "step": 97700
+    },
+    {
+      "epoch": 1.31455146643727,
+      "grad_norm": 1.0447673797607422,
+      "learning_rate": 0.0049660539857059515,
+      "loss": 2.491,
+      "step": 97800
+    },
+    {
+      "epoch": 1.3158955885910912,
+      "grad_norm": 1.3554537296295166,
+      "learning_rate": 0.0049659668638652316,
+      "loss": 2.4827,
+      "step": 97900
+    },
+    {
+      "epoch": 1.3172397107449125,
+      "grad_norm": 1.2488070726394653,
+      "learning_rate": 0.00496587963113714,
+      "loss": 2.4825,
+      "step": 98000
+    },
+    {
+      "epoch": 1.3172397107449125,
+      "eval_MaskedAccuracy": 0.439554404426727,
+      "eval_loss": 2.793813943862915,
+      "eval_runtime": 158.828,
+      "eval_samples_per_second": 399.652,
+      "eval_steps_per_second": 1.561,
+      "step": 98000
+    },
+    {
+      "epoch": 1.3185838328987338,
+      "grad_norm": 0.7793965935707092,
+      "learning_rate": 0.004965792287525609,
+      "loss": 2.4838,
+      "step": 98100
+    },
+    {
+      "epoch": 1.3199279550525551,
+      "grad_norm": 2.033632755279541,
+      "learning_rate": 0.004965704833034586,
+      "loss": 2.4777,
+      "step": 98200
+    },
+    {
+      "epoch": 1.3212720772063764,
+      "grad_norm": 0.8052381277084351,
+      "learning_rate": 0.0049656172676679905,
+      "loss": 2.4794,
+      "step": 98300
+    },
+    {
+      "epoch": 1.322616199360198,
+      "grad_norm": 0.6361316442489624,
+      "learning_rate": 0.0049655295914297825,
+      "loss": 2.4872,
+      "step": 98400
+    },
+    {
+      "epoch": 1.3239603215140192,
+      "grad_norm": 0.8865070343017578,
+      "learning_rate": 0.004965441804323918,
+      "loss": 2.4818,
+      "step": 98500
+    },
+    {
+      "epoch": 1.3253044436678405,
+      "grad_norm": 0.9113361239433289,
+      "learning_rate": 0.004965353906354336,
+      "loss": 2.4834,
+      "step": 98600
+    },
+    {
+      "epoch": 1.3266485658216618,
+      "grad_norm": 1.3836218118667603,
+      "learning_rate": 0.004965265897525005,
+      "loss": 2.4795,
+      "step": 98700
+    },
+    {
+      "epoch": 1.3279926879754833,
+      "grad_norm": 0.7738713026046753,
+      "learning_rate": 0.0049651777778398805,
+      "loss": 2.4869,
+      "step": 98800
+    },
+    {
+      "epoch": 1.3293368101293046,
+      "grad_norm": 1.1943604946136475,
+      "learning_rate": 0.004965089547302944,
+      "loss": 2.489,
+      "step": 98900
+    },
+    {
+      "epoch": 1.330680932283126,
+      "grad_norm": 0.4339142441749573,
+      "learning_rate": 0.004965001205918176,
+      "loss": 2.4801,
+      "step": 99000
+    },
+    {
+      "epoch": 1.330680932283126,
+      "eval_MaskedAccuracy": 0.4402919161127212,
+      "eval_loss": 2.789841651916504,
+      "eval_runtime": 155.2442,
+      "eval_samples_per_second": 408.878,
+      "eval_steps_per_second": 1.597,
+      "step": 99000
+    },
+    {
+      "epoch": 1.3320250544369472,
+      "grad_norm": 0.5905622243881226,
+      "learning_rate": 0.00496491275368955,
+      "loss": 2.4818,
+      "step": 99100
+    },
+    {
+      "epoch": 1.3333691765907685,
+      "grad_norm": 0.8844038844108582,
+      "learning_rate": 0.004964824190621052,
+      "loss": 2.4847,
+      "step": 99200
+    },
+    {
+      "epoch": 1.3347132987445898,
+      "grad_norm": 0.4743984639644623,
+      "learning_rate": 0.004964735516716671,
+      "loss": 2.4873,
+      "step": 99300
+    },
+    {
+      "epoch": 1.3360574208984113,
+      "grad_norm": 0.4455043077468872,
+      "learning_rate": 0.0049646467319804,
+      "loss": 2.4894,
+      "step": 99400
+    },
+    {
+      "epoch": 1.3374015430522326,
+      "grad_norm": 0.33678489923477173,
+      "learning_rate": 0.004964557836416248,
+      "loss": 2.4707,
+      "step": 99500
+    },
+    {
+      "epoch": 1.3387456652060539,
+      "grad_norm": 0.557610809803009,
+      "learning_rate": 0.004964468830028214,
+      "loss": 2.4761,
+      "step": 99600
+    },
+    {
+      "epoch": 1.3400897873598754,
+      "grad_norm": 1.0966076850891113,
+      "learning_rate": 0.00496437971282031,
+      "loss": 2.4776,
+      "step": 99700
+    },
+    {
+      "epoch": 1.3414339095136967,
+      "grad_norm": 1.7193931341171265,
+      "learning_rate": 0.00496429048479655,
+      "loss": 2.5095,
+      "step": 99800
+    },
+    {
+      "epoch": 1.342778031667518,
+      "grad_norm": 1.0982494354248047,
+      "learning_rate": 0.004964201145960957,
+      "loss": 2.4835,
+      "step": 99900
+    },
+    {
+      "epoch": 1.3441221538213393,
+      "grad_norm": 0.7176316976547241,
+      "learning_rate": 0.004964111696317556,
+      "loss": 2.4881,
+      "step": 100000
+    },
+    {
+      "epoch": 1.3441221538213393,
+      "eval_MaskedAccuracy": 0.439665552172269,
+      "eval_loss": 2.7934842109680176,
+      "eval_runtime": 156.2627,
+      "eval_samples_per_second": 406.213,
+      "eval_steps_per_second": 1.587,
+      "step": 100000
+    },
+    {
+      "epoch": 1.3454662759751606,
+      "grad_norm": 0.6092787384986877,
+      "learning_rate": 0.00496402213587037,
+      "loss": 2.4923,
+      "step": 100100
+    },
+    {
+      "epoch": 1.3468103981289818,
+      "grad_norm": 0.48341110348701477,
+      "learning_rate": 0.004963932464623445,
+      "loss": 2.4847,
+      "step": 100200
+    },
+    {
+      "epoch": 1.3481545202828034,
+      "grad_norm": 1.8525594472885132,
+      "learning_rate": 0.004963842682580817,
+      "loss": 2.4742,
+      "step": 100300
+    },
+    {
+      "epoch": 1.3494986424366247,
+      "grad_norm": 1.4322495460510254,
+      "learning_rate": 0.004963752789746537,
+      "loss": 2.4793,
+      "step": 100400
+    },
+    {
+      "epoch": 1.350842764590446,
+      "grad_norm": 0.881243884563446,
+      "learning_rate": 0.004963662786124655,
+      "loss": 2.4773,
+      "step": 100500
+    },
+    {
+      "epoch": 1.3521868867442675,
+      "grad_norm": 1.2258617877960205,
+      "learning_rate": 0.0049635726717192205,
+      "loss": 2.4833,
+      "step": 100600
+    },
+    {
+      "epoch": 1.3535310088980887,
+      "grad_norm": 0.9638917446136475,
+      "learning_rate": 0.004963482446534289,
+      "loss": 2.4804,
+      "step": 100700
+    },
+    {
+      "epoch": 1.35487513105191,
+      "grad_norm": 0.9644854068756104,
+      "learning_rate": 0.004963392110573937,
+      "loss": 2.4719,
+      "step": 100800
+    },
+    {
+      "epoch": 1.3562192532057313,
+      "grad_norm": 0.689267635345459,
+      "learning_rate": 0.0049633016638422275,
+      "loss": 2.4756,
+      "step": 100900
+    },
+    {
+      "epoch": 1.3575633753595526,
+      "grad_norm": 1.2867717742919922,
+      "learning_rate": 0.004963211106343239,
+      "loss": 2.4812,
+      "step": 101000
+    },
+    {
+      "epoch": 1.3575633753595526,
+      "eval_MaskedAccuracy": 0.4396971572510662,
+      "eval_loss": 2.793617010116577,
+      "eval_runtime": 157.0355,
+      "eval_samples_per_second": 404.214,
+      "eval_steps_per_second": 1.579,
+      "step": 101000
+    },
+    {
+      "epoch": 1.358907497513374,
+      "grad_norm": 1.8802392482757568,
+      "learning_rate": 0.00496312043808105,
+      "loss": 2.485,
+      "step": 101100
+    },
+    {
+      "epoch": 1.3602516196671954,
+      "grad_norm": 0.35604092478752136,
+      "learning_rate": 0.004963029659059752,
+      "loss": 2.4812,
+      "step": 101200
+    },
+    {
+      "epoch": 1.3615957418210167,
+      "grad_norm": 1.0150797367095947,
+      "learning_rate": 0.004962938769283431,
+      "loss": 2.4875,
+      "step": 101300
+    },
+    {
+      "epoch": 1.362939863974838,
+      "grad_norm": 1.3340040445327759,
+      "learning_rate": 0.004962847768756179,
+      "loss": 2.4684,
+      "step": 101400
+    },
+    {
+      "epoch": 1.3642839861286593,
+      "grad_norm": 0.38257646560668945,
+      "learning_rate": 0.004962756657482102,
+      "loss": 2.4866,
+      "step": 101500
+    },
+    {
+      "epoch": 1.3656281082824808,
+      "grad_norm": 0.6464138031005859,
+      "learning_rate": 0.0049626654354653025,
+      "loss": 2.4844,
+      "step": 101600
+    },
+    {
+      "epoch": 1.3669722304363021,
+      "grad_norm": 0.41035962104797363,
+      "learning_rate": 0.004962574102709892,
+      "loss": 2.4817,
+      "step": 101700
+    },
+    {
+      "epoch": 1.3683163525901234,
+      "grad_norm": 0.7176283001899719,
+      "learning_rate": 0.004962482659219983,
+      "loss": 2.4751,
+      "step": 101800
+    },
+    {
+      "epoch": 1.3696604747439447,
+      "grad_norm": 0.7560930848121643,
+      "learning_rate": 0.004962391104999701,
+      "loss": 2.48,
+      "step": 101900
+    },
+    {
+      "epoch": 1.371004596897766,
+      "grad_norm": 0.8050186038017273,
+      "learning_rate": 0.004962299440053163,
+      "loss": 2.4691,
+      "step": 102000
+    },
+    {
+      "epoch": 1.371004596897766,
+      "eval_MaskedAccuracy": 0.44033106657058535,
+      "eval_loss": 2.7889299392700195,
+      "eval_runtime": 155.2645,
+      "eval_samples_per_second": 408.825,
+      "eval_steps_per_second": 1.597,
+      "step": 102000
+    },
+    {
+      "epoch": 1.3723487190515873,
+      "grad_norm": 0.39613771438598633,
+      "learning_rate": 0.004962207664384512,
+      "loss": 2.4773,
+      "step": 102100
+    },
+    {
+      "epoch": 1.3736928412054088,
+      "grad_norm": 2.5852887630462646,
+      "learning_rate": 0.004962115777997869,
+      "loss": 2.479,
+      "step": 102200
+    },
+    {
+      "epoch": 1.37503696335923,
+      "grad_norm": 1.9417086839675903,
+      "learning_rate": 0.004962023780897379,
+      "loss": 2.479,
+      "step": 102300
+    },
+    {
+      "epoch": 1.3763810855130514,
+      "grad_norm": 1.185256004333496,
+      "learning_rate": 0.004961931673087198,
+      "loss": 2.4713,
+      "step": 102400
+    },
+    {
+      "epoch": 1.377725207666873,
+      "grad_norm": 0.5902850031852722,
+      "learning_rate": 0.004961839454571464,
+      "loss": 2.4805,
+      "step": 102500
+    },
+    {
+      "epoch": 1.3790693298206942,
+      "grad_norm": 0.61485755443573,
+      "learning_rate": 0.0049617471253543335,
+      "loss": 2.478,
+      "step": 102600
+    },
+    {
+      "epoch": 1.3804134519745155,
+      "grad_norm": 0.3328634798526764,
+      "learning_rate": 0.004961654685439971,
+      "loss": 2.4705,
+      "step": 102700
+    },
+    {
+      "epoch": 1.3817575741283368,
+      "grad_norm": 1.5581071376800537,
+      "learning_rate": 0.004961562134832535,
+      "loss": 2.468,
+      "step": 102800
+    },
+    {
+      "epoch": 1.383101696282158,
+      "grad_norm": 0.8161963820457458,
+      "learning_rate": 0.004961469473536204,
+      "loss": 2.4773,
+      "step": 102900
+    },
+    {
+      "epoch": 1.3844458184359794,
+      "grad_norm": 0.4553765654563904,
+      "learning_rate": 0.004961376701555151,
+      "loss": 2.4774,
+      "step": 103000
+    },
+    {
+      "epoch": 1.3844458184359794,
+      "eval_MaskedAccuracy": 0.44020060249449044,
+      "eval_loss": 2.7893545627593994,
+      "eval_runtime": 157.1906,
+      "eval_samples_per_second": 403.815,
+      "eval_steps_per_second": 1.578,
+      "step": 103000
+    },
+    {
+      "epoch": 1.3857899405898009,
+      "grad_norm": 0.7315437197685242,
+      "learning_rate": 0.004961283818893556,
+      "loss": 2.4893,
+      "step": 103100
+    },
+    {
+      "epoch": 1.3871340627436222,
+      "grad_norm": 1.7851052284240723,
+      "learning_rate": 0.004961190825555604,
+      "loss": 2.4825,
+      "step": 103200
+    },
+    {
+      "epoch": 1.3884781848974435,
+      "grad_norm": 0.7662883400917053,
+      "learning_rate": 0.004961097721545482,
+      "loss": 2.4694,
+      "step": 103300
+    },
+    {
+      "epoch": 1.3898223070512647,
+      "grad_norm": 0.3333396315574646,
+      "learning_rate": 0.0049610045068673855,
+      "loss": 2.4772,
+      "step": 103400
+    },
+    {
+      "epoch": 1.3911664292050863,
+      "grad_norm": 0.4882153570652008,
+      "learning_rate": 0.004960911181525515,
+      "loss": 2.4764,
+      "step": 103500
+    },
+    {
+      "epoch": 1.3925105513589076,
+      "grad_norm": 1.4237163066864014,
+      "learning_rate": 0.0049608177455240775,
+      "loss": 2.4822,
+      "step": 103600
+    },
+    {
+      "epoch": 1.3938546735127288,
+      "grad_norm": 0.5641297101974487,
+      "learning_rate": 0.004960724198867284,
+      "loss": 2.4779,
+      "step": 103700
+    },
+    {
+      "epoch": 1.3951987956665501,
+      "grad_norm": 0.6121580600738525,
+      "learning_rate": 0.00496063054155935,
+      "loss": 2.4663,
+      "step": 103800
+    },
+    {
+      "epoch": 1.3965429178203714,
+      "grad_norm": 0.6751658320426941,
+      "learning_rate": 0.004960536773604492,
+      "loss": 2.4718,
+      "step": 103900
+    },
+    {
+      "epoch": 1.3978870399741927,
+      "grad_norm": 0.9351543188095093,
+      "learning_rate": 0.004960442895006929,
+      "loss": 2.4855,
+      "step": 104000
+    },
+    {
+      "epoch": 1.3978870399741927,
+      "eval_MaskedAccuracy": 0.44113983122276385,
+      "eval_loss": 2.7834458351135254,
+      "eval_runtime": 156.6243,
+      "eval_samples_per_second": 405.276,
+      "eval_steps_per_second": 1.583,
+      "step": 104000
+    },
+    {
+      "epoch": 1.3992311621280142,
+      "grad_norm": 0.44563743472099304,
+      "learning_rate": 0.004960348905770906,
+      "loss": 2.478,
+      "step": 104100
+    },
+    {
+      "epoch": 1.4005752842818355,
+      "grad_norm": 0.7417739629745483,
+      "learning_rate": 0.004960254805900649,
+      "loss": 2.4782,
+      "step": 104200
+    },
+    {
+      "epoch": 1.4019194064356568,
+      "grad_norm": 0.38673731684684753,
+      "learning_rate": 0.004960160595400398,
+      "loss": 2.4703,
+      "step": 104300
+    },
+    {
+      "epoch": 1.4032635285894783,
+      "grad_norm": 0.41961851716041565,
+      "learning_rate": 0.004960066274274401,
+      "loss": 2.4758,
+      "step": 104400
+    },
+    {
+      "epoch": 1.4046076507432996,
+      "grad_norm": 1.0936461687088013,
+      "learning_rate": 0.00495997184252691,
+      "loss": 2.4711,
+      "step": 104500
+    },
+    {
+      "epoch": 1.405951772897121,
+      "grad_norm": 1.4237301349639893,
+      "learning_rate": 0.004959877300162172,
+      "loss": 2.476,
+      "step": 104600
+    },
+    {
+      "epoch": 1.4072958950509422,
+      "grad_norm": 0.5586555600166321,
+      "learning_rate": 0.004959782647184455,
+      "loss": 2.4744,
+      "step": 104700
+    },
+    {
+      "epoch": 1.4086400172047635,
+      "grad_norm": 0.5998302102088928,
+      "learning_rate": 0.0049596878835980115,
+      "loss": 2.481,
+      "step": 104800
+    },
+    {
+      "epoch": 1.4099841393585848,
+      "grad_norm": 0.6141105890274048,
+      "learning_rate": 0.004959593009407127,
+      "loss": 2.4683,
+      "step": 104900
+    },
+    {
+      "epoch": 1.4113282615124063,
+      "grad_norm": 0.6551980376243591,
+      "learning_rate": 0.00495949802461607,
+      "loss": 2.4758,
+      "step": 105000
+    },
+    {
+      "epoch": 1.4113282615124063,
+      "eval_MaskedAccuracy": 0.4408968621448549,
+      "eval_loss": 2.78601336479187,
+      "eval_runtime": 159.2788,
+      "eval_samples_per_second": 398.521,
+      "eval_steps_per_second": 1.557,
+      "step": 105000
+    },
+    {
+      "epoch": 1.4126723836662276,
+      "grad_norm": 0.6265648603439331,
+      "learning_rate": 0.00495940292922911,
+      "loss": 2.4772,
+      "step": 105100
+    },
+    {
+      "epoch": 1.414016505820049,
+      "grad_norm": 1.0922765731811523,
+      "learning_rate": 0.004959307723250553,
+      "loss": 2.4649,
+      "step": 105200
+    },
+    {
+      "epoch": 1.4153606279738702,
+      "grad_norm": 1.4056020975112915,
+      "learning_rate": 0.004959212406684676,
+      "loss": 2.4811,
+      "step": 105300
+    },
+    {
+      "epoch": 1.4167047501276917,
+      "grad_norm": 1.5905888080596924,
+      "learning_rate": 0.004959116979535776,
+      "loss": 2.4769,
+      "step": 105400
+    },
+    {
+      "epoch": 1.418048872281513,
+      "grad_norm": 0.448127418756485,
+      "learning_rate": 0.004959021441808149,
+      "loss": 2.4706,
+      "step": 105500
+    },
+    {
+      "epoch": 1.4193929944353343,
+      "grad_norm": 1.482865571975708,
+      "learning_rate": 0.004958925793506108,
+      "loss": 2.4795,
+      "step": 105600
+    },
+    {
+      "epoch": 1.4207371165891556,
+      "grad_norm": 0.37079110741615295,
+      "learning_rate": 0.004958830034633952,
+      "loss": 2.4688,
+      "step": 105700
+    },
+    {
+      "epoch": 1.4220812387429769,
+      "grad_norm": 0.398387610912323,
+      "learning_rate": 0.004958734165196007,
+      "loss": 2.473,
+      "step": 105800
+    },
+    {
+      "epoch": 1.4234253608967982,
+      "grad_norm": 0.49912792444229126,
+      "learning_rate": 0.004958638185196587,
+      "loss": 2.4701,
+      "step": 105900
+    },
+    {
+      "epoch": 1.4247694830506197,
+      "grad_norm": 0.8730640411376953,
+      "learning_rate": 0.004958542094640022,
+      "loss": 2.474,
+      "step": 106000
+    },
+    {
+      "epoch": 1.4247694830506197,
+      "eval_MaskedAccuracy": 0.44114819868603417,
+      "eval_loss": 2.7835192680358887,
+      "eval_runtime": 154.9227,
+      "eval_samples_per_second": 409.727,
+      "eval_steps_per_second": 1.601,
+      "step": 106000
+    },
+    {
+      "epoch": 1.426113605204441,
+      "grad_norm": 0.4877232313156128,
+      "learning_rate": 0.004958445893530628,
+      "loss": 2.4685,
+      "step": 106100
+    },
+    {
+      "epoch": 1.4274577273582623,
+      "grad_norm": 0.6099728345870972,
+      "learning_rate": 0.004958349581872748,
+      "loss": 2.4724,
+      "step": 106200
+    },
+    {
+      "epoch": 1.4288018495120838,
+      "grad_norm": 0.4108321964740753,
+      "learning_rate": 0.004958253159670722,
+      "loss": 2.4686,
+      "step": 106300
+    },
+    {
+      "epoch": 1.430145971665905,
+      "grad_norm": 0.5882393717765808,
+      "learning_rate": 0.004958156626928895,
+      "loss": 2.4763,
+      "step": 106400
+    },
+    {
+      "epoch": 1.4314900938197264,
+      "grad_norm": 0.7170044779777527,
+      "learning_rate": 0.004958059983651618,
+      "loss": 2.4681,
+      "step": 106500
+    },
+    {
+      "epoch": 1.4328342159735477,
+      "grad_norm": 0.8796346783638,
+      "learning_rate": 0.0049579632298432435,
+      "loss": 2.4734,
+      "step": 106600
+    },
+    {
+      "epoch": 1.434178338127369,
+      "grad_norm": 0.3585628271102905,
+      "learning_rate": 0.0049578663655081385,
+      "loss": 2.4599,
+      "step": 106700
+    },
+    {
+      "epoch": 1.4355224602811902,
+      "grad_norm": 1.9920889139175415,
+      "learning_rate": 0.004957769390650657,
+      "loss": 2.4726,
+      "step": 106800
+    },
+    {
+      "epoch": 1.4368665824350118,
+      "grad_norm": 1.0133873224258423,
+      "learning_rate": 0.004957672305275166,
+      "loss": 2.4684,
+      "step": 106900
+    },
+    {
+      "epoch": 1.438210704588833,
+      "grad_norm": 1.567334771156311,
+      "learning_rate": 0.004957575109386042,
+      "loss": 2.4643,
+      "step": 107000
+    },
+    {
+      "epoch": 1.438210704588833,
+      "eval_MaskedAccuracy": 0.4405992607707991,
+      "eval_loss": 2.786257743835449,
+      "eval_runtime": 154.0026,
+      "eval_samples_per_second": 412.175,
+      "eval_steps_per_second": 1.61,
+      "step": 107000
+    },
+    {
+      "epoch": 1.4395548267426543,
+      "grad_norm": 1.3851370811462402,
+      "learning_rate": 0.004957477802987676,
+      "loss": 2.4719,
+      "step": 107100
+    },
+    {
+      "epoch": 1.4408989488964756,
+      "grad_norm": 0.3829158842563629,
+      "learning_rate": 0.0049573803860844465,
+      "loss": 2.4618,
+      "step": 107200
+    },
+    {
+      "epoch": 1.4422430710502971,
+      "grad_norm": 0.8415204882621765,
+      "learning_rate": 0.004957282858680741,
+      "loss": 2.4691,
+      "step": 107300
+    },
+    {
+      "epoch": 1.4435871932041184,
+      "grad_norm": 1.198485016822815,
+      "learning_rate": 0.00495718522078095,
+      "loss": 2.4661,
+      "step": 107400
+    },
+    {
+      "epoch": 1.4449313153579397,
+      "grad_norm": 2.3275532722473145,
+      "learning_rate": 0.00495708747238948,
+      "loss": 2.4734,
+      "step": 107500
+    },
+    {
+      "epoch": 1.446275437511761,
+      "grad_norm": 0.5716771483421326,
+      "learning_rate": 0.0049569896135107355,
+      "loss": 2.4612,
+      "step": 107600
+    },
+    {
+      "epoch": 1.4476195596655823,
+      "grad_norm": 1.2711353302001953,
+      "learning_rate": 0.00495689164414912,
+      "loss": 2.4738,
+      "step": 107700
+    },
+    {
+      "epoch": 1.4489636818194038,
+      "grad_norm": 0.5984246730804443,
+      "learning_rate": 0.004956793564309049,
+      "loss": 2.4793,
+      "step": 107800
+    },
+    {
+      "epoch": 1.4503078039732251,
+      "grad_norm": 0.992817223072052,
+      "learning_rate": 0.004956695373994945,
+      "loss": 2.4736,
+      "step": 107900
+    },
+    {
+      "epoch": 1.4516519261270464,
+      "grad_norm": 0.7644572257995605,
+      "learning_rate": 0.004956597073211232,
+      "loss": 2.4834,
+      "step": 108000
+    },
+    {
+      "epoch": 1.4516519261270464,
+      "eval_MaskedAccuracy": 0.4405933158523938,
+      "eval_loss": 2.7867140769958496,
+      "eval_runtime": 155.9739,
+      "eval_samples_per_second": 406.966,
+      "eval_steps_per_second": 1.59,
+      "step": 108000
+    },
+    {
+      "epoch": 1.4529960482808677,
+      "grad_norm": 0.39693641662597656,
+      "learning_rate": 0.004956498661962332,
+      "loss": 2.4675,
+      "step": 108100
+    },
+    {
+      "epoch": 1.4543401704346892,
+      "grad_norm": 1.5463731288909912,
+      "learning_rate": 0.004956400140252694,
+      "loss": 2.4737,
+      "step": 108200
+    },
+    {
+      "epoch": 1.4556842925885105,
+      "grad_norm": 1.4359521865844727,
+      "learning_rate": 0.004956301508086743,
+      "loss": 2.475,
+      "step": 108300
+    },
+    {
+      "epoch": 1.4570284147423318,
+      "grad_norm": 1.5656912326812744,
+      "learning_rate": 0.004956202765468931,
+      "loss": 2.4734,
+      "step": 108400
+    },
+    {
+      "epoch": 1.458372536896153,
+      "grad_norm": 0.3067225515842438,
+      "learning_rate": 0.004956103912403708,
+      "loss": 2.4703,
+      "step": 108500
+    },
+    {
+      "epoch": 1.4597166590499744,
+      "grad_norm": 1.173917293548584,
+      "learning_rate": 0.0049560049488955255,
+      "loss": 2.4626,
+      "step": 108600
+    },
+    {
+      "epoch": 1.4610607812037957,
+      "grad_norm": 0.5422664284706116,
+      "learning_rate": 0.004955905874948838,
+      "loss": 2.4634,
+      "step": 108700
+    },
+    {
+      "epoch": 1.4624049033576172,
+      "grad_norm": 1.207939863204956,
+      "learning_rate": 0.004955806690568119,
+      "loss": 2.4676,
+      "step": 108800
+    },
+    {
+      "epoch": 1.4637490255114385,
+      "grad_norm": 1.1675505638122559,
+      "learning_rate": 0.0049557073957578355,
+      "loss": 2.4668,
+      "step": 108900
+    },
+    {
+      "epoch": 1.4650931476652598,
+      "grad_norm": 0.4084634482860565,
+      "learning_rate": 0.004955607990522452,
+      "loss": 2.4674,
+      "step": 109000
+    },
+    {
+      "epoch": 1.4650931476652598,
+      "eval_MaskedAccuracy": 0.44142586982077486,
+      "eval_loss": 2.781691551208496,
+      "eval_runtime": 154.5643,
+      "eval_samples_per_second": 410.677,
+      "eval_steps_per_second": 1.605,
+      "step": 109000
+    },
+    {
+      "epoch": 1.4664372698190813,
+      "grad_norm": 0.33248618245124817,
+      "learning_rate": 0.004955508474866459,
+      "loss": 2.47,
+      "step": 109100
+    },
+    {
+      "epoch": 1.4677813919729026,
+      "grad_norm": 1.3263065814971924,
+      "learning_rate": 0.004955408848794333,
+      "loss": 2.4666,
+      "step": 109200
+    },
+    {
+      "epoch": 1.4691255141267239,
+      "grad_norm": 0.5887484550476074,
+      "learning_rate": 0.004955309112310564,
+      "loss": 2.4736,
+      "step": 109300
+    },
+    {
+      "epoch": 1.4704696362805452,
+      "grad_norm": 0.5707719326019287,
+      "learning_rate": 0.004955209265419648,
+      "loss": 2.4723,
+      "step": 109400
+    },
+    {
+      "epoch": 1.4718137584343665,
+      "grad_norm": 1.564012885093689,
+      "learning_rate": 0.004955109308126084,
+      "loss": 2.4677,
+      "step": 109500
+    },
+    {
+      "epoch": 1.4731578805881878,
+      "grad_norm": 0.5303487181663513,
+      "learning_rate": 0.004955009240434372,
+      "loss": 2.4748,
+      "step": 109600
+    },
+    {
+      "epoch": 1.4745020027420093,
+      "grad_norm": 0.8532441258430481,
+      "learning_rate": 0.0049549090623490265,
+      "loss": 2.4605,
+      "step": 109700
+    },
+    {
+      "epoch": 1.4758461248958306,
+      "grad_norm": 1.2264505624771118,
+      "learning_rate": 0.004954808773874565,
+      "loss": 2.4663,
+      "step": 109800
+    },
+    {
+      "epoch": 1.4771902470496519,
+      "grad_norm": 0.4173908233642578,
+      "learning_rate": 0.0049547083750154965,
+      "loss": 2.4677,
+      "step": 109900
+    },
+    {
+      "epoch": 1.4785343692034731,
+      "grad_norm": 1.8136452436447144,
+      "learning_rate": 0.004954607865776349,
+      "loss": 2.4652,
+      "step": 110000
+    },
+    {
+      "epoch": 1.4785343692034731,
+      "eval_MaskedAccuracy": 0.44079433809670326,
+      "eval_loss": 2.7854974269866943,
+      "eval_runtime": 155.1144,
+      "eval_samples_per_second": 409.221,
+      "eval_steps_per_second": 1.599,
+      "step": 110000
+    },
+    {
+      "epoch": 1.4798784913572947,
+      "grad_norm": 1.8435359001159668,
+      "learning_rate": 0.004954507246161656,
+      "loss": 2.4611,
+      "step": 110100
+    },
+    {
+      "epoch": 1.481222613511116,
+      "grad_norm": 0.33811017870903015,
+      "learning_rate": 0.004954406516175938,
+      "loss": 2.4659,
+      "step": 110200
+    },
+    {
+      "epoch": 1.4825667356649372,
+      "grad_norm": 1.902899146080017,
+      "learning_rate": 0.00495430567582375,
+      "loss": 2.4629,
+      "step": 110300
+    },
+    {
+      "epoch": 1.4839108578187585,
+      "grad_norm": 2.60919451713562,
+      "learning_rate": 0.004954204725109627,
+      "loss": 2.4783,
+      "step": 110400
+    },
+    {
+      "epoch": 1.4852549799725798,
+      "grad_norm": 0.7571696639060974,
+      "learning_rate": 0.0049541036640381175,
+      "loss": 2.4551,
+      "step": 110500
+    },
+    {
+      "epoch": 1.4865991021264011,
+      "grad_norm": 0.4986100196838379,
+      "learning_rate": 0.004954002492613781,
+      "loss": 2.4625,
+      "step": 110600
+    },
+    {
+      "epoch": 1.4879432242802226,
+      "grad_norm": 1.091294765472412,
+      "learning_rate": 0.0049539012108411684,
+      "loss": 2.4686,
+      "step": 110700
+    },
+    {
+      "epoch": 1.489287346434044,
+      "grad_norm": 0.8807032704353333,
+      "learning_rate": 0.004953799818724848,
+      "loss": 2.4764,
+      "step": 110800
+    },
+    {
+      "epoch": 1.4906314685878652,
+      "grad_norm": 1.00475013256073,
+      "learning_rate": 0.0049536983162693895,
+      "loss": 2.4637,
+      "step": 110900
+    },
+    {
+      "epoch": 1.4919755907416867,
+      "grad_norm": 0.33167359232902527,
+      "learning_rate": 0.004953596703479359,
+      "loss": 2.4678,
+      "step": 111000
+    },
+    {
+      "epoch": 1.4919755907416867,
+      "eval_MaskedAccuracy": 0.44179725878685755,
+      "eval_loss": 2.776883602142334,
+      "eval_runtime": 154.4158,
+      "eval_samples_per_second": 411.072,
+      "eval_steps_per_second": 1.606,
+      "step": 111000
+    },
+    {
+      "epoch": 1.493319712895508,
+      "grad_norm": 1.6402376890182495,
+      "learning_rate": 0.004953494980359343,
+      "loss": 2.4685,
+      "step": 111100
+    },
+    {
+      "epoch": 1.4946638350493293,
+      "grad_norm": 0.7305079698562622,
+      "learning_rate": 0.004953393146913917,
+      "loss": 2.4643,
+      "step": 111200
+    },
+    {
+      "epoch": 1.4960079572031506,
+      "grad_norm": 0.560355544090271,
+      "learning_rate": 0.0049532912031476735,
+      "loss": 2.4659,
+      "step": 111300
+    },
+    {
+      "epoch": 1.497352079356972,
+      "grad_norm": 0.9456803202629089,
+      "learning_rate": 0.004953189149065217,
+      "loss": 2.4618,
+      "step": 111400
+    },
+    {
+      "epoch": 1.4986962015107932,
+      "grad_norm": 1.5349628925323486,
+      "learning_rate": 0.004953086984671131,
+      "loss": 2.4661,
+      "step": 111500
+    },
+    {
+      "epoch": 1.5000403236646145,
+      "grad_norm": 1.062753438949585,
+      "learning_rate": 0.004952984709970025,
+      "loss": 2.4618,
+      "step": 111600
+    },
+    {
+      "epoch": 1.501384445818436,
+      "grad_norm": 0.45811429619789124,
+      "learning_rate": 0.0049528823249665056,
+      "loss": 2.4586,
+      "step": 111700
+    },
+    {
+      "epoch": 1.5027285679722573,
+      "grad_norm": 0.5497962832450867,
+      "learning_rate": 0.004952779829665184,
+      "loss": 2.4693,
+      "step": 111800
+    },
+    {
+      "epoch": 1.5040726901260788,
+      "grad_norm": 0.5482380986213684,
+      "learning_rate": 0.004952677224070681,
+      "loss": 2.4623,
+      "step": 111900
+    },
+    {
+      "epoch": 1.5054168122799,
+      "grad_norm": 1.2586387395858765,
+      "learning_rate": 0.0049525745081876165,
+      "loss": 2.4669,
+      "step": 112000
+    },
+    {
+      "epoch": 1.5054168122799,
+      "eval_MaskedAccuracy": 0.44191414103263826,
+      "eval_loss": 2.776205539703369,
+      "eval_runtime": 153.7585,
+      "eval_samples_per_second": 412.829,
+      "eval_steps_per_second": 1.613,
+      "step": 112000
+    },
+    {
+      "epoch": 1.5067609344337214,
+      "grad_norm": 2.1461663246154785,
+      "learning_rate": 0.004952471682020625,
+      "loss": 2.4569,
+      "step": 112100
+    },
+    {
+      "epoch": 1.5081050565875427,
+      "grad_norm": 3.052988290786743,
+      "learning_rate": 0.00495236874557434,
+      "loss": 2.4583,
+      "step": 112200
+    },
+    {
+      "epoch": 1.509449178741364,
+      "grad_norm": 0.3190886676311493,
+      "learning_rate": 0.00495226569885339,
+      "loss": 2.4577,
+      "step": 112300
+    },
+    {
+      "epoch": 1.5107933008951853,
+      "grad_norm": 0.390209436416626,
+      "learning_rate": 0.0049521625418624295,
+      "loss": 2.4577,
+      "step": 112400
+    },
+    {
+      "epoch": 1.5121374230490066,
+      "grad_norm": 1.6563820838928223,
+      "learning_rate": 0.004952059274606098,
+      "loss": 2.4577,
+      "step": 112500
+    },
+    {
+      "epoch": 1.513481545202828,
+      "grad_norm": 0.9080681204795837,
+      "learning_rate": 0.004951955897089055,
+      "loss": 2.4618,
+      "step": 112600
+    },
+    {
+      "epoch": 1.5148256673566494,
+      "grad_norm": 0.5969460010528564,
+      "learning_rate": 0.0049518524093159514,
+      "loss": 2.472,
+      "step": 112700
+    },
+    {
+      "epoch": 1.5161697895104709,
+      "grad_norm": 0.358441025018692,
+      "learning_rate": 0.004951748811291452,
+      "loss": 2.4619,
+      "step": 112800
+    },
+    {
+      "epoch": 1.5175139116642922,
+      "grad_norm": 0.6230987906455994,
+      "learning_rate": 0.004951645103020229,
+      "loss": 2.4626,
+      "step": 112900
+    },
+    {
+      "epoch": 1.5188580338181135,
+      "grad_norm": 0.5158119797706604,
+      "learning_rate": 0.00495154128450696,
+      "loss": 2.4671,
+      "step": 113000
+    },
+    {
+      "epoch": 1.5188580338181135,
+      "eval_MaskedAccuracy": 0.4421825141264522,
+      "eval_loss": 2.7729415893554688,
+      "eval_runtime": 153.7445,
+      "eval_samples_per_second": 412.867,
+      "eval_steps_per_second": 1.613,
+      "step": 113000
+    },
+    {
+      "epoch": 1.5202021559719348,
+      "grad_norm": 1.2907798290252686,
+      "learning_rate": 0.004951437355756309,
+      "loss": 2.4693,
+      "step": 113100
+    },
+    {
+      "epoch": 1.521546278125756,
+      "grad_norm": 1.6166244745254517,
+      "learning_rate": 0.004951333316772972,
+      "loss": 2.4585,
+      "step": 113200
+    },
+    {
+      "epoch": 1.5228904002795773,
+      "grad_norm": 0.5393241047859192,
+      "learning_rate": 0.004951229167561628,
+      "loss": 2.4643,
+      "step": 113300
+    },
+    {
+      "epoch": 1.5242345224333986,
+      "grad_norm": 0.7464740872383118,
+      "learning_rate": 0.004951124908126974,
+      "loss": 2.4576,
+      "step": 113400
+    },
+    {
+      "epoch": 1.52557864458722,
+      "grad_norm": 0.47540420293807983,
+      "learning_rate": 0.004951020538473705,
+      "loss": 2.4567,
+      "step": 113500
+    },
+    {
+      "epoch": 1.5269227667410414,
+      "grad_norm": 0.3154013454914093,
+      "learning_rate": 0.00495091605860653,
+      "loss": 2.4664,
+      "step": 113600
+    },
+    {
+      "epoch": 1.5282668888948627,
+      "grad_norm": 0.5142553448677063,
+      "learning_rate": 0.0049508114685301455,
+      "loss": 2.4582,
+      "step": 113700
+    },
+    {
+      "epoch": 1.5296110110486842,
+      "grad_norm": 1.193684697151184,
+      "learning_rate": 0.004950706768249272,
+      "loss": 2.4598,
+      "step": 113800
+    },
+    {
+      "epoch": 1.5309551332025055,
+      "grad_norm": 0.3604363203048706,
+      "learning_rate": 0.0049506019577686316,
+      "loss": 2.4589,
+      "step": 113900
+    },
+    {
+      "epoch": 1.5322992553563268,
+      "grad_norm": 0.5039530396461487,
+      "learning_rate": 0.004950497037092934,
+      "loss": 2.466,
+      "step": 114000
+    },
+    {
+      "epoch": 1.5322992553563268,
+      "eval_MaskedAccuracy": 0.4417917832476142,
+      "eval_loss": 2.775747299194336,
+      "eval_runtime": 154.7989,
+      "eval_samples_per_second": 410.055,
+      "eval_steps_per_second": 1.602,
+      "step": 114000
+    },
+    {
+      "epoch": 1.5336433775101481,
+      "grad_norm": 0.35583773255348206,
+      "learning_rate": 0.004950392006226925,
+      "loss": 2.4673,
+      "step": 114100
+    },
+    {
+      "epoch": 1.5349874996639694,
+      "grad_norm": 0.35848045349121094,
+      "learning_rate": 0.0049502868651753185,
+      "loss": 2.4615,
+      "step": 114200
+    },
+    {
+      "epoch": 1.5363316218177907,
+      "grad_norm": 1.3079990148544312,
+      "learning_rate": 0.004950181613942862,
+      "loss": 2.4687,
+      "step": 114300
+    },
+    {
+      "epoch": 1.537675743971612,
+      "grad_norm": 1.409440279006958,
+      "learning_rate": 0.004950076252534293,
+      "loss": 2.4501,
+      "step": 114400
+    },
+    {
+      "epoch": 1.5390198661254335,
+      "grad_norm": 2.4961087703704834,
+      "learning_rate": 0.0049499707809543565,
+      "loss": 2.461,
+      "step": 114500
+    },
+    {
+      "epoch": 1.5403639882792548,
+      "grad_norm": 0.69174724817276,
+      "learning_rate": 0.0049498651992078155,
+      "loss": 2.4543,
+      "step": 114600
+    },
+    {
+      "epoch": 1.5417081104330763,
+      "grad_norm": 1.2737400531768799,
+      "learning_rate": 0.004949759507299418,
+      "loss": 2.4584,
+      "step": 114700
+    },
+    {
+      "epoch": 1.5430522325868976,
+      "grad_norm": 1.2695642709732056,
+      "learning_rate": 0.00494965370523393,
+      "loss": 2.4593,
+      "step": 114800
+    },
+    {
+      "epoch": 1.544396354740719,
+      "grad_norm": 0.31750351190567017,
+      "learning_rate": 0.004949547793016122,
+      "loss": 2.4551,
+      "step": 114900
+    },
+    {
+      "epoch": 1.5457404768945402,
+      "grad_norm": 0.38432586193084717,
+      "learning_rate": 0.004949441770650766,
+      "loss": 2.4576,
+      "step": 115000
+    },
+    {
+      "epoch": 1.5457404768945402,
+      "eval_MaskedAccuracy": 0.4425237323528715,
+      "eval_loss": 2.774029493331909,
+      "eval_runtime": 156.0215,
+      "eval_samples_per_second": 406.841,
+      "eval_steps_per_second": 1.59,
+      "step": 115000
+    },
+    {
+      "epoch": 1.5470845990483615,
+      "grad_norm": 0.34672319889068604,
+      "learning_rate": 0.00494933563814263,
+      "loss": 2.4499,
+      "step": 115100
+    },
+    {
+      "epoch": 1.5484287212021828,
+      "grad_norm": 0.6219214200973511,
+      "learning_rate": 0.004949229395496506,
+      "loss": 2.4494,
+      "step": 115200
+    },
+    {
+      "epoch": 1.549772843356004,
+      "grad_norm": 3.128021240234375,
+      "learning_rate": 0.004949123042717173,
+      "loss": 2.4652,
+      "step": 115300
+    },
+    {
+      "epoch": 1.5511169655098256,
+      "grad_norm": 0.5915595889091492,
+      "learning_rate": 0.004949016579809432,
+      "loss": 2.4696,
+      "step": 115400
+    },
+    {
+      "epoch": 1.5524610876636469,
+      "grad_norm": 1.1795876026153564,
+      "learning_rate": 0.004948910006778072,
+      "loss": 2.4595,
+      "step": 115500
+    },
+    {
+      "epoch": 1.5538052098174682,
+      "grad_norm": 0.6927655339241028,
+      "learning_rate": 0.004948803323627903,
+      "loss": 2.4517,
+      "step": 115600
+    },
+    {
+      "epoch": 1.5551493319712897,
+      "grad_norm": 0.35303354263305664,
+      "learning_rate": 0.004948696530363729,
+      "loss": 2.4593,
+      "step": 115700
+    },
+    {
+      "epoch": 1.556493454125111,
+      "grad_norm": 2.31588077545166,
+      "learning_rate": 0.004948589626990361,
+      "loss": 2.4566,
+      "step": 115800
+    },
+    {
+      "epoch": 1.5578375762789323,
+      "grad_norm": 0.37357497215270996,
+      "learning_rate": 0.004948482613512615,
+      "loss": 2.4553,
+      "step": 115900
+    },
+    {
+      "epoch": 1.5591816984327536,
+      "grad_norm": 1.7756373882293701,
+      "learning_rate": 0.0049483754899353125,
+      "loss": 2.4547,
+      "step": 116000
+    },
+    {
+      "epoch": 1.5591816984327536,
+      "eval_MaskedAccuracy": 0.4417388896389656,
+      "eval_loss": 2.7752127647399902,
+      "eval_runtime": 155.7219,
+      "eval_samples_per_second": 407.624,
+      "eval_steps_per_second": 1.593,
+      "step": 116000
+    },
+    {
+      "epoch": 1.5605258205865749,
+      "grad_norm": 1.2487781047821045,
+      "learning_rate": 0.004948268256263274,
+      "loss": 2.4604,
+      "step": 116100
+    },
+    {
+      "epoch": 1.5618699427403961,
+      "grad_norm": 0.5920694470405579,
+      "learning_rate": 0.0049481609125013384,
+      "loss": 2.4585,
+      "step": 116200
+    },
+    {
+      "epoch": 1.5632140648942174,
+      "grad_norm": 1.373819351196289,
+      "learning_rate": 0.004948053458654348,
+      "loss": 2.4526,
+      "step": 116300
+    },
+    {
+      "epoch": 1.564558187048039,
+      "grad_norm": 1.0958904027938843,
+      "learning_rate": 0.004947945894727141,
+      "loss": 2.4591,
+      "step": 116400
+    },
+    {
+      "epoch": 1.5659023092018602,
+      "grad_norm": 4.045643329620361,
+      "learning_rate": 0.004947838220724564,
+      "loss": 2.4613,
+      "step": 116500
+    },
+    {
+      "epoch": 1.5672464313556818,
+      "grad_norm": 1.2977932691574097,
+      "learning_rate": 0.004947730436651466,
+      "loss": 2.4564,
+      "step": 116600
+    },
+    {
+      "epoch": 1.568590553509503,
+      "grad_norm": 1.1461297273635864,
+      "learning_rate": 0.004947622542512704,
+      "loss": 2.4542,
+      "step": 116700
+    },
+    {
+      "epoch": 1.5699346756633243,
+      "grad_norm": 2.07126522064209,
+      "learning_rate": 0.004947514538313141,
+      "loss": 2.4584,
+      "step": 116800
+    },
+    {
+      "epoch": 1.5712787978171456,
+      "grad_norm": 0.6101120114326477,
+      "learning_rate": 0.004947406424057641,
+      "loss": 2.4596,
+      "step": 116900
+    },
+    {
+      "epoch": 1.572622919970967,
+      "grad_norm": 1.1047863960266113,
+      "learning_rate": 0.0049472981997510735,
+      "loss": 2.4563,
+      "step": 117000
+    },
+    {
+      "epoch": 1.572622919970967,
+      "eval_MaskedAccuracy": 0.44274268801439726,
+      "eval_loss": 2.769578218460083,
+      "eval_runtime": 156.3839,
+      "eval_samples_per_second": 405.899,
+      "eval_steps_per_second": 1.586,
+      "step": 117000
+    },
+    {
+      "epoch": 1.5739670421247882,
+      "grad_norm": 0.3559087812900543,
+      "learning_rate": 0.004947189865398322,
+      "loss": 2.4558,
+      "step": 117100
+    },
+    {
+      "epoch": 1.5753111642786095,
+      "grad_norm": 1.209089994430542,
+      "learning_rate": 0.004947081421004263,
+      "loss": 2.4592,
+      "step": 117200
+    },
+    {
+      "epoch": 1.576655286432431,
+      "grad_norm": 1.5720434188842773,
+      "learning_rate": 0.004946972866573786,
+      "loss": 2.4598,
+      "step": 117300
+    },
+    {
+      "epoch": 1.5779994085862523,
+      "grad_norm": 0.33877983689308167,
+      "learning_rate": 0.004946864202111778,
+      "loss": 2.4539,
+      "step": 117400
+    },
+    {
+      "epoch": 1.5793435307400736,
+      "grad_norm": 1.1109495162963867,
+      "learning_rate": 0.004946755427623137,
+      "loss": 2.4621,
+      "step": 117500
+    },
+    {
+      "epoch": 1.5806876528938951,
+      "grad_norm": 0.32806700468063354,
+      "learning_rate": 0.0049466465431127656,
+      "loss": 2.4553,
+      "step": 117600
+    },
+    {
+      "epoch": 1.5820317750477164,
+      "grad_norm": 1.0989540815353394,
+      "learning_rate": 0.004946537548585571,
+      "loss": 2.4632,
+      "step": 117700
+    },
+    {
+      "epoch": 1.5833758972015377,
+      "grad_norm": 0.3961617350578308,
+      "learning_rate": 0.004946428444046461,
+      "loss": 2.4529,
+      "step": 117800
+    },
+    {
+      "epoch": 1.584720019355359,
+      "grad_norm": 2.489666700363159,
+      "learning_rate": 0.004946319229500352,
+      "loss": 2.4506,
+      "step": 117900
+    },
+    {
+      "epoch": 1.5860641415091803,
+      "grad_norm": 0.2874549627304077,
+      "learning_rate": 0.004946209904952169,
+      "loss": 2.451,
+      "step": 118000
+    },
+    {
+      "epoch": 1.5860641415091803,
+      "eval_MaskedAccuracy": 0.44264748991328356,
+      "eval_loss": 2.770142078399658,
+      "eval_runtime": 153.5901,
+      "eval_samples_per_second": 413.282,
+      "eval_steps_per_second": 1.615,
+      "step": 118000
+    },
+    {
+      "epoch": 1.5874082636630016,
+      "grad_norm": 2.0315134525299072,
+      "learning_rate": 0.004946100470406832,
+      "loss": 2.4472,
+      "step": 118100
+    },
+    {
+      "epoch": 1.5887523858168229,
+      "grad_norm": 0.7130675911903381,
+      "learning_rate": 0.004945990925869273,
+      "loss": 2.4548,
+      "step": 118200
+    },
+    {
+      "epoch": 1.5900965079706444,
+      "grad_norm": 0.9596794843673706,
+      "learning_rate": 0.0049458812713444305,
+      "loss": 2.462,
+      "step": 118300
+    },
+    {
+      "epoch": 1.5914406301244657,
+      "grad_norm": 1.4742804765701294,
+      "learning_rate": 0.004945771506837244,
+      "loss": 2.4549,
+      "step": 118400
+    },
+    {
+      "epoch": 1.5927847522782872,
+      "grad_norm": 1.3615772724151611,
+      "learning_rate": 0.004945661632352659,
+      "loss": 2.4574,
+      "step": 118500
+    },
+    {
+      "epoch": 1.5941288744321085,
+      "grad_norm": 0.30037587881088257,
+      "learning_rate": 0.004945551647895632,
+      "loss": 2.4514,
+      "step": 118600
+    },
+    {
+      "epoch": 1.5954729965859298,
+      "grad_norm": 1.7664852142333984,
+      "learning_rate": 0.004945441553471114,
+      "loss": 2.452,
+      "step": 118700
+    },
+    {
+      "epoch": 1.596817118739751,
+      "grad_norm": 1.8963963985443115,
+      "learning_rate": 0.004945331349084064,
+      "loss": 2.4552,
+      "step": 118800
+    },
+    {
+      "epoch": 1.5981612408935724,
+      "grad_norm": 0.5631555318832397,
+      "learning_rate": 0.004945221034739453,
+      "loss": 2.4525,
+      "step": 118900
+    },
+    {
+      "epoch": 1.5995053630473937,
+      "grad_norm": 0.28932493925094604,
+      "learning_rate": 0.00494511061044224,
+      "loss": 2.4576,
+      "step": 119000
+    },
+    {
+      "epoch": 1.5995053630473937,
+      "eval_MaskedAccuracy": 0.44307573744928136,
+      "eval_loss": 2.768948793411255,
+      "eval_runtime": 153.3984,
+      "eval_samples_per_second": 413.798,
+      "eval_steps_per_second": 1.617,
+      "step": 119000
+    },
+    {
+      "epoch": 1.600849485201215,
+      "grad_norm": 1.037090539932251,
+      "learning_rate": 0.0049450000761974144,
+      "loss": 2.4576,
+      "step": 119100
+    },
+    {
+      "epoch": 1.6021936073550365,
+      "grad_norm": 4.012909412384033,
+      "learning_rate": 0.004944889432009946,
+      "loss": 2.4638,
+      "step": 119200
+    },
+    {
+      "epoch": 1.6035377295088578,
+      "grad_norm": 0.358288437128067,
+      "learning_rate": 0.004944778677884832,
+      "loss": 2.4572,
+      "step": 119300
+    },
+    {
+      "epoch": 1.6048818516626793,
+      "grad_norm": 0.7761825323104858,
+      "learning_rate": 0.004944667813827054,
+      "loss": 2.4585,
+      "step": 119400
+    },
+    {
+      "epoch": 1.6062259738165006,
+      "grad_norm": 1.2888187170028687,
+      "learning_rate": 0.004944556839841606,
+      "loss": 2.4593,
+      "step": 119500
+    },
+    {
+      "epoch": 1.6075700959703219,
+      "grad_norm": 0.6099095344543457,
+      "learning_rate": 0.0049444457559334925,
+      "loss": 2.4618,
+      "step": 119600
+    },
+    {
+      "epoch": 1.6089142181241431,
+      "grad_norm": 0.30497902631759644,
+      "learning_rate": 0.004944334562107721,
+      "loss": 2.4596,
+      "step": 119700
+    },
+    {
+      "epoch": 1.6102583402779644,
+      "grad_norm": 3.6389715671539307,
+      "learning_rate": 0.004944223258369297,
+      "loss": 2.4551,
+      "step": 119800
+    },
+    {
+      "epoch": 1.6116024624317857,
+      "grad_norm": 0.3001348078250885,
+      "learning_rate": 0.004944111844723238,
+      "loss": 2.4522,
+      "step": 119900
+    },
+    {
+      "epoch": 1.612946584585607,
+      "grad_norm": 0.6880463361740112,
+      "learning_rate": 0.004944000321174563,
+      "loss": 2.4455,
+      "step": 120000
+    },
+    {
+      "epoch": 1.612946584585607,
+      "eval_MaskedAccuracy": 0.44322096964619634,
+      "eval_loss": 2.7688114643096924,
+      "eval_runtime": 153.4713,
+      "eval_samples_per_second": 413.602,
+      "eval_steps_per_second": 1.616,
+      "step": 120000
+    },
+    {
+      "epoch": 1.6142907067394283,
+      "grad_norm": 0.5405184626579285,
+      "learning_rate": 0.0049438886877282975,
+      "loss": 2.4582,
+      "step": 120100
+    },
+    {
+      "epoch": 1.6156348288932498,
+      "grad_norm": 1.32143235206604,
+      "learning_rate": 0.004943776944389477,
+      "loss": 2.4501,
+      "step": 120200
+    },
+    {
+      "epoch": 1.6169789510470711,
+      "grad_norm": 0.532461941242218,
+      "learning_rate": 0.004943665091163127,
+      "loss": 2.4512,
+      "step": 120300
+    },
+    {
+      "epoch": 1.6183230732008926,
+      "grad_norm": 1.1819446086883545,
+      "learning_rate": 0.004943553128054288,
+      "loss": 2.4623,
+      "step": 120400
+    },
+    {
+      "epoch": 1.619667195354714,
+      "grad_norm": 1.5699783563613892,
+      "learning_rate": 0.004943441055068009,
+      "loss": 2.4504,
+      "step": 120500
+    },
+    {
+      "epoch": 1.6210113175085352,
+      "grad_norm": 1.6754220724105835,
+      "learning_rate": 0.004943328872209337,
+      "loss": 2.45,
+      "step": 120600
+    },
+    {
+      "epoch": 1.6223554396623565,
+      "grad_norm": 0.35783451795578003,
+      "learning_rate": 0.004943216579483333,
+      "loss": 2.4411,
+      "step": 120700
+    },
+    {
+      "epoch": 1.6236995618161778,
+      "grad_norm": 2.1771366596221924,
+      "learning_rate": 0.004943104176895054,
+      "loss": 2.4516,
+      "step": 120800
+    },
+    {
+      "epoch": 1.625043683969999,
+      "grad_norm": 1.1872481107711792,
+      "learning_rate": 0.0049429916644495656,
+      "loss": 2.4464,
+      "step": 120900
+    },
+    {
+      "epoch": 1.6263878061238204,
+      "grad_norm": 1.4642506837844849,
+      "learning_rate": 0.004942879042151933,
+      "loss": 2.461,
+      "step": 121000
+    },
+    {
+      "epoch": 1.6263878061238204,
+      "eval_MaskedAccuracy": 0.4426167755231782,
+      "eval_loss": 2.7695846557617188,
+      "eval_runtime": 153.853,
+      "eval_samples_per_second": 412.576,
+      "eval_steps_per_second": 1.612,
+      "step": 121000
+    },
+    {
+      "epoch": 1.627731928277642,
+      "grad_norm": 0.3797900080680847,
+      "learning_rate": 0.004942766310007232,
+      "loss": 2.4495,
+      "step": 121100
+    },
+    {
+      "epoch": 1.6290760504314632,
+      "grad_norm": 0.4284350872039795,
+      "learning_rate": 0.004942653468020548,
+      "loss": 2.4434,
+      "step": 121200
+    },
+    {
+      "epoch": 1.6304201725852847,
+      "grad_norm": 1.1178627014160156,
+      "learning_rate": 0.004942540516196953,
+      "loss": 2.4532,
+      "step": 121300
+    },
+    {
+      "epoch": 1.631764294739106,
+      "grad_norm": 1.3800808191299438,
+      "learning_rate": 0.0049424274545415425,
+      "loss": 2.4582,
+      "step": 121400
+    },
+    {
+      "epoch": 1.6331084168929273,
+      "grad_norm": 0.3553604781627655,
+      "learning_rate": 0.0049423142830594185,
+      "loss": 2.4546,
+      "step": 121500
+    },
+    {
+      "epoch": 1.6344525390467486,
+      "grad_norm": 0.29325976967811584,
+      "learning_rate": 0.004942201001755672,
+      "loss": 2.449,
+      "step": 121600
+    },
+    {
+      "epoch": 1.6357966612005699,
+      "grad_norm": 0.28023549914360046,
+      "learning_rate": 0.004942087610635415,
+      "loss": 2.4559,
+      "step": 121700
+    },
+    {
+      "epoch": 1.6371407833543912,
+      "grad_norm": 0.6937592029571533,
+      "learning_rate": 0.0049419741097037435,
+      "loss": 2.458,
+      "step": 121800
+    },
+    {
+      "epoch": 1.6384849055082125,
+      "grad_norm": 0.3913755416870117,
+      "learning_rate": 0.004941860498965778,
+      "loss": 2.4511,
+      "step": 121900
+    },
+    {
+      "epoch": 1.639829027662034,
+      "grad_norm": 0.817623496055603,
+      "learning_rate": 0.004941746778426644,
+      "loss": 2.4551,
+      "step": 122000
+    },
+    {
+      "epoch": 1.639829027662034,
+      "eval_MaskedAccuracy": 0.44339662189399426,
+      "eval_loss": 2.764580249786377,
+      "eval_runtime": 157.5643,
+      "eval_samples_per_second": 402.858,
+      "eval_steps_per_second": 1.574,
+      "step": 122000
+    },
+    {
+      "epoch": 1.6411731498158553,
+      "grad_norm": 0.36455515027046204,
+      "learning_rate": 0.004941632948091455,
+      "loss": 2.4556,
+      "step": 122100
+    },
+    {
+      "epoch": 1.6425172719696766,
+      "grad_norm": 0.5823009014129639,
+      "learning_rate": 0.004941519007965352,
+      "loss": 2.4511,
+      "step": 122200
+    },
+    {
+      "epoch": 1.643861394123498,
+      "grad_norm": 0.7032744884490967,
+      "learning_rate": 0.004941404958053464,
+      "loss": 2.4545,
+      "step": 122300
+    },
+    {
+      "epoch": 1.6452055162773194,
+      "grad_norm": 2.976278781890869,
+      "learning_rate": 0.004941290798360926,
+      "loss": 2.45,
+      "step": 122400
+    },
+    {
+      "epoch": 1.6465496384311407,
+      "grad_norm": 0.37186741828918457,
+      "learning_rate": 0.004941176528892885,
+      "loss": 2.4524,
+      "step": 122500
+    },
+    {
+      "epoch": 1.647893760584962,
+      "grad_norm": 1.0369256734848022,
+      "learning_rate": 0.004941062149654489,
+      "loss": 2.4522,
+      "step": 122600
+    },
+    {
+      "epoch": 1.6492378827387832,
+      "grad_norm": 0.8359701037406921,
+      "learning_rate": 0.004940947660650888,
+      "loss": 2.4618,
+      "step": 122700
+    },
+    {
+      "epoch": 1.6505820048926045,
+      "grad_norm": 0.3227209150791168,
+      "learning_rate": 0.004940833061887247,
+      "loss": 2.4419,
+      "step": 122800
+    },
+    {
+      "epoch": 1.6519261270464258,
+      "grad_norm": 0.44361066818237305,
+      "learning_rate": 0.00494071835336873,
+      "loss": 2.4504,
+      "step": 122900
+    },
+    {
+      "epoch": 1.6532702492002473,
+      "grad_norm": 0.9630382657051086,
+      "learning_rate": 0.004940603535100503,
+      "loss": 2.4494,
+      "step": 123000
+    },
+    {
+      "epoch": 1.6532702492002473,
+      "eval_MaskedAccuracy": 0.4438552018371928,
+      "eval_loss": 2.762709856033325,
+      "eval_runtime": 153.7406,
+      "eval_samples_per_second": 412.877,
+      "eval_steps_per_second": 1.613,
+      "step": 123000
+    },
+    {
+      "epoch": 1.6546143713540686,
+      "grad_norm": 1.0358442068099976,
+      "learning_rate": 0.004940488607087739,
+      "loss": 2.4472,
+      "step": 123100
+    },
+    {
+      "epoch": 1.6559584935078902,
+      "grad_norm": 1.2041329145431519,
+      "learning_rate": 0.004940373569335615,
+      "loss": 2.4454,
+      "step": 123200
+    },
+    {
+      "epoch": 1.6573026156617114,
+      "grad_norm": 1.5662907361984253,
+      "learning_rate": 0.004940258421849318,
+      "loss": 2.4495,
+      "step": 123300
+    },
+    {
+      "epoch": 1.6586467378155327,
+      "grad_norm": 0.6521201133728027,
+      "learning_rate": 0.004940143164634034,
+      "loss": 2.4478,
+      "step": 123400
+    },
+    {
+      "epoch": 1.659990859969354,
+      "grad_norm": 0.7364070415496826,
+      "learning_rate": 0.004940027797694953,
+      "loss": 2.4472,
+      "step": 123500
+    },
+    {
+      "epoch": 1.6613349821231753,
+      "grad_norm": 0.5109328031539917,
+      "learning_rate": 0.004939912321037278,
+      "loss": 2.4537,
+      "step": 123600
+    },
+    {
+      "epoch": 1.6626791042769966,
+      "grad_norm": 0.6922867298126221,
+      "learning_rate": 0.004939796734666213,
+      "loss": 2.456,
+      "step": 123700
+    },
+    {
+      "epoch": 1.664023226430818,
+      "grad_norm": 0.8259190320968628,
+      "learning_rate": 0.004939681038586966,
+      "loss": 2.4501,
+      "step": 123800
+    },
+    {
+      "epoch": 1.6653673485846394,
+      "grad_norm": 0.7042433619499207,
+      "learning_rate": 0.004939565232804747,
+      "loss": 2.4456,
+      "step": 123900
+    },
+    {
+      "epoch": 1.6667114707384607,
+      "grad_norm": 1.328395128250122,
+      "learning_rate": 0.0049394493173247804,
+      "loss": 2.4599,
+      "step": 124000
+    },
+    {
+      "epoch": 1.6667114707384607,
+      "eval_MaskedAccuracy": 0.44327534756472914,
+      "eval_loss": 2.7654707431793213,
+      "eval_runtime": 156.5562,
+      "eval_samples_per_second": 405.452,
+      "eval_steps_per_second": 1.584,
+      "step": 124000
+    },
+    {
+      "epoch": 1.668055592892282,
+      "grad_norm": 1.040299892425537,
+      "learning_rate": 0.0049393332921522795,
+      "loss": 2.4475,
+      "step": 124100
+    },
+    {
+      "epoch": 1.6693997150461035,
+      "grad_norm": 0.447637677192688,
+      "learning_rate": 0.004939217157292479,
+      "loss": 2.4522,
+      "step": 124200
+    },
+    {
+      "epoch": 1.6707438371999248,
+      "grad_norm": 0.3728015422821045,
+      "learning_rate": 0.004939100912750606,
+      "loss": 2.4541,
+      "step": 124300
+    },
+    {
+      "epoch": 1.672087959353746,
+      "grad_norm": 0.4562872350215912,
+      "learning_rate": 0.0049389845585319095,
+      "loss": 2.452,
+      "step": 124400
+    },
+    {
+      "epoch": 1.6734320815075674,
+      "grad_norm": 1.0070672035217285,
+      "learning_rate": 0.0049388680946416234,
+      "loss": 2.4478,
+      "step": 124500
+    },
+    {
+      "epoch": 1.6747762036613887,
+      "grad_norm": 0.3675072193145752,
+      "learning_rate": 0.004938751521084995,
+      "loss": 2.4464,
+      "step": 124600
+    },
+    {
+      "epoch": 1.67612032581521,
+      "grad_norm": 1.3940047025680542,
+      "learning_rate": 0.004938634837867282,
+      "loss": 2.44,
+      "step": 124700
+    },
+    {
+      "epoch": 1.6774644479690313,
+      "grad_norm": 0.6820449233055115,
+      "learning_rate": 0.004938518044993735,
+      "loss": 2.4448,
+      "step": 124800
+    },
+    {
+      "epoch": 1.6788085701228528,
+      "grad_norm": 0.2972992956638336,
+      "learning_rate": 0.004938401142469626,
+      "loss": 2.4407,
+      "step": 124900
+    },
+    {
+      "epoch": 1.680152692276674,
+      "grad_norm": 1.0912327766418457,
+      "learning_rate": 0.004938284130300211,
+      "loss": 2.4543,
+      "step": 125000
+    },
+    {
+      "epoch": 1.680152692276674,
+      "eval_MaskedAccuracy": 0.4438537899485854,
+      "eval_loss": 2.7616376876831055,
+      "eval_runtime": 153.4081,
+      "eval_samples_per_second": 413.772,
+      "eval_steps_per_second": 1.617,
+      "step": 125000
+    },
+    {
+      "epoch": 1.6814968144304956,
+      "grad_norm": 0.8189194798469543,
+      "learning_rate": 0.004938167008490767,
+      "loss": 2.4444,
+      "step": 125100
+    },
+    {
+      "epoch": 1.6828409365843169,
+      "grad_norm": 1.2573928833007812,
+      "learning_rate": 0.004938049777046575,
+      "loss": 2.4477,
+      "step": 125200
+    },
+    {
+      "epoch": 1.6841850587381382,
+      "grad_norm": 0.27490001916885376,
+      "learning_rate": 0.004937932435972912,
+      "loss": 2.4494,
+      "step": 125300
+    },
+    {
+      "epoch": 1.6855291808919595,
+      "grad_norm": 2.5110983848571777,
+      "learning_rate": 0.0049378149852750694,
+      "loss": 2.4453,
+      "step": 125400
+    },
+    {
+      "epoch": 1.6868733030457808,
+      "grad_norm": 0.5286669135093689,
+      "learning_rate": 0.004937697424958338,
+      "loss": 2.4499,
+      "step": 125500
+    },
+    {
+      "epoch": 1.688217425199602,
+      "grad_norm": 1.0157794952392578,
+      "learning_rate": 0.004937579755028025,
+      "loss": 2.4499,
+      "step": 125600
+    },
+    {
+      "epoch": 1.6895615473534233,
+      "grad_norm": 2.4944801330566406,
+      "learning_rate": 0.004937461975489416,
+      "loss": 2.4454,
+      "step": 125700
+    },
+    {
+      "epoch": 1.6909056695072449,
+      "grad_norm": 0.9877827763557434,
+      "learning_rate": 0.0049373440863478175,
+      "loss": 2.4435,
+      "step": 125800
+    },
+    {
+      "epoch": 1.6922497916610661,
+      "grad_norm": 0.28307971358299255,
+      "learning_rate": 0.004937226087608559,
+      "loss": 2.4502,
+      "step": 125900
+    },
+    {
+      "epoch": 1.6935939138148874,
+      "grad_norm": 0.6857683658599854,
+      "learning_rate": 0.0049371079792769395,
+      "loss": 2.4459,
+      "step": 126000
+    },
+    {
+      "epoch": 1.6935939138148874,
+      "eval_MaskedAccuracy": 0.4442026999946834,
+      "eval_loss": 2.760291576385498,
+      "eval_runtime": 154.5053,
+      "eval_samples_per_second": 410.834,
+      "eval_steps_per_second": 1.605,
+      "step": 126000
+    },
+    {
+      "epoch": 1.694938035968709,
+      "grad_norm": 0.3602958023548126,
+      "learning_rate": 0.0049369897613582864,
+      "loss": 2.4459,
+      "step": 126100
+    },
+    {
+      "epoch": 1.6962821581225302,
+      "grad_norm": 1.104148507118225,
+      "learning_rate": 0.004936871433857933,
+      "loss": 2.452,
+      "step": 126200
+    },
+    {
+      "epoch": 1.6976262802763515,
+      "grad_norm": 1.409397840499878,
+      "learning_rate": 0.004936752996781202,
+      "loss": 2.4469,
+      "step": 126300
+    },
+    {
+      "epoch": 1.6989704024301728,
+      "grad_norm": 1.9952348470687866,
+      "learning_rate": 0.004936634450133437,
+      "loss": 2.4584,
+      "step": 126400
+    },
+    {
+      "epoch": 1.7003145245839941,
+      "grad_norm": 1.7229875326156616,
+      "learning_rate": 0.004936515793919977,
+      "loss": 2.4491,
+      "step": 126500
+    },
+    {
+      "epoch": 1.7016586467378154,
+      "grad_norm": 0.6407601237297058,
+      "learning_rate": 0.00493639702814616,
+      "loss": 2.4525,
+      "step": 126600
+    },
+    {
+      "epoch": 1.7030027688916367,
+      "grad_norm": 1.1715171337127686,
+      "learning_rate": 0.004936278152817347,
+      "loss": 2.4403,
+      "step": 126700
+    },
+    {
+      "epoch": 1.7043468910454582,
+      "grad_norm": 0.6220793128013611,
+      "learning_rate": 0.004936159167938901,
+      "loss": 2.4487,
+      "step": 126800
+    },
+    {
+      "epoch": 1.7056910131992795,
+      "grad_norm": 0.5494009256362915,
+      "learning_rate": 0.004936040073516175,
+      "loss": 2.4331,
+      "step": 126900
+    },
+    {
+      "epoch": 1.707035135353101,
+      "grad_norm": 1.780874252319336,
+      "learning_rate": 0.004935920869554534,
+      "loss": 2.4391,
+      "step": 127000
+    },
+    {
+      "epoch": 1.707035135353101,
+      "eval_MaskedAccuracy": 0.4440303666034177,
+      "eval_loss": 2.7617318630218506,
+      "eval_runtime": 156.1849,
+      "eval_samples_per_second": 406.416,
+      "eval_steps_per_second": 1.588,
+      "step": 127000
+    },
+    {
+      "epoch": 1.7083792575069223,
+      "grad_norm": 1.470224142074585,
+      "learning_rate": 0.004935801556059347,
+      "loss": 2.4416,
+      "step": 127100
+    },
+    {
+      "epoch": 1.7097233796607436,
+      "grad_norm": 0.2865951359272003,
+      "learning_rate": 0.00493568213303599,
+      "loss": 2.4505,
+      "step": 127200
+    },
+    {
+      "epoch": 1.711067501814565,
+      "grad_norm": 1.1740427017211914,
+      "learning_rate": 0.004935562600489849,
+      "loss": 2.4505,
+      "step": 127300
+    },
+    {
+      "epoch": 1.7124116239683862,
+      "grad_norm": 0.6032086610794067,
+      "learning_rate": 0.0049354429584263105,
+      "loss": 2.4482,
+      "step": 127400
+    },
+    {
+      "epoch": 1.7137557461222075,
+      "grad_norm": 1.0448052883148193,
+      "learning_rate": 0.0049353232068507615,
+      "loss": 2.45,
+      "step": 127500
+    },
+    {
+      "epoch": 1.7150998682760288,
+      "grad_norm": 0.433826208114624,
+      "learning_rate": 0.004935203345768603,
+      "loss": 2.4493,
+      "step": 127600
+    },
+    {
+      "epoch": 1.7164439904298503,
+      "grad_norm": 1.4232683181762695,
+      "learning_rate": 0.004935083375185232,
+      "loss": 2.4463,
+      "step": 127700
+    },
+    {
+      "epoch": 1.7177881125836716,
+      "grad_norm": 1.5276943445205688,
+      "learning_rate": 0.004934963295106055,
+      "loss": 2.4501,
+      "step": 127800
+    },
+    {
+      "epoch": 1.719132234737493,
+      "grad_norm": 1.1934447288513184,
+      "learning_rate": 0.0049348431055364775,
+      "loss": 2.4529,
+      "step": 127900
+    },
+    {
+      "epoch": 1.7204763568913144,
+      "grad_norm": 0.4801059663295746,
+      "learning_rate": 0.004934722806481924,
+      "loss": 2.4412,
+      "step": 128000
+    },
+    {
+      "epoch": 1.7204763568913144,
+      "eval_MaskedAccuracy": 0.4440363882564416,
+      "eval_loss": 2.758544683456421,
+      "eval_runtime": 156.4492,
+      "eval_samples_per_second": 405.729,
+      "eval_steps_per_second": 1.585,
+      "step": 128000
+    },
+    {
+      "epoch": 1.7218204790451357,
+      "grad_norm": 0.24669547379016876,
+      "learning_rate": 0.004934602397947807,
+      "loss": 2.4503,
+      "step": 128100
+    },
+    {
+      "epoch": 1.723164601198957,
+      "grad_norm": 0.8044118285179138,
+      "learning_rate": 0.00493448187993956,
+      "loss": 2.4397,
+      "step": 128200
+    },
+    {
+      "epoch": 1.7245087233527783,
+      "grad_norm": 0.3931942582130432,
+      "learning_rate": 0.004934361252462605,
+      "loss": 2.4423,
+      "step": 128300
+    },
+    {
+      "epoch": 1.7258528455065996,
+      "grad_norm": 0.3876301944255829,
+      "learning_rate": 0.004934240515522382,
+      "loss": 2.4393,
+      "step": 128400
+    },
+    {
+      "epoch": 1.7271969676604209,
+      "grad_norm": 1.6018636226654053,
+      "learning_rate": 0.004934119669124325,
+      "loss": 2.4476,
+      "step": 128500
+    },
+    {
+      "epoch": 1.7285410898142424,
+      "grad_norm": 2.163212776184082,
+      "learning_rate": 0.004933998713273891,
+      "loss": 2.4496,
+      "step": 128600
+    },
+    {
+      "epoch": 1.7298852119680637,
+      "grad_norm": 0.32517024874687195,
+      "learning_rate": 0.004933877647976517,
+      "loss": 2.4492,
+      "step": 128700
+    },
+    {
+      "epoch": 1.731229334121885,
+      "grad_norm": 0.843076765537262,
+      "learning_rate": 0.00493375647323767,
+      "loss": 2.4452,
+      "step": 128800
+    },
+    {
+      "epoch": 1.7325734562757065,
+      "grad_norm": 0.7287679314613342,
+      "learning_rate": 0.004933635189062801,
+      "loss": 2.4411,
+      "step": 128900
+    },
+    {
+      "epoch": 1.7339175784295278,
+      "grad_norm": 0.561202347278595,
+      "learning_rate": 0.004933513795457381,
+      "loss": 2.437,
+      "step": 129000
+    },
+    {
+      "epoch": 1.7339175784295278,
+      "eval_MaskedAccuracy": 0.4451868773268175,
+      "eval_loss": 2.7529821395874023,
+      "eval_runtime": 154.118,
+      "eval_samples_per_second": 411.866,
+      "eval_steps_per_second": 1.609,
+      "step": 129000
+    },
+    {
+      "epoch": 1.735261700583349,
+      "grad_norm": 1.201848030090332,
+      "learning_rate": 0.0049333922924268804,
+      "loss": 2.4469,
+      "step": 129100
+    },
+    {
+      "epoch": 1.7366058227371703,
+      "grad_norm": 0.7610087990760803,
+      "learning_rate": 0.004933270679976775,
+      "loss": 2.4427,
+      "step": 129200
+    },
+    {
+      "epoch": 1.7379499448909916,
+      "grad_norm": 0.4432719051837921,
+      "learning_rate": 0.004933148958112536,
+      "loss": 2.4431,
+      "step": 129300
+    },
+    {
+      "epoch": 1.739294067044813,
+      "grad_norm": 0.4039577841758728,
+      "learning_rate": 0.004933027126839647,
+      "loss": 2.4416,
+      "step": 129400
+    },
+    {
+      "epoch": 1.7406381891986342,
+      "grad_norm": 0.47328197956085205,
+      "learning_rate": 0.004932905186163602,
+      "loss": 2.4409,
+      "step": 129500
+    },
+    {
+      "epoch": 1.7419823113524557,
+      "grad_norm": 0.9691961407661438,
+      "learning_rate": 0.004932783136089899,
+      "loss": 2.4415,
+      "step": 129600
+    },
+    {
+      "epoch": 1.743326433506277,
+      "grad_norm": 0.8540052771568298,
+      "learning_rate": 0.004932660976624034,
+      "loss": 2.444,
+      "step": 129700
+    },
+    {
+      "epoch": 1.7446705556600985,
+      "grad_norm": 1.1006224155426025,
+      "learning_rate": 0.004932538707771508,
+      "loss": 2.4509,
+      "step": 129800
+    },
+    {
+      "epoch": 1.7460146778139198,
+      "grad_norm": 2.2226502895355225,
+      "learning_rate": 0.004932416329537838,
+      "loss": 2.4423,
+      "step": 129900
+    },
+    {
+      "epoch": 1.7473587999677411,
+      "grad_norm": 0.42202070355415344,
+      "learning_rate": 0.004932293841928537,
+      "loss": 2.4394,
+      "step": 130000
+    },
+    {
+      "epoch": 1.7473587999677411,
+      "eval_MaskedAccuracy": 0.44459195138244084,
+      "eval_loss": 2.756187915802002,
+      "eval_runtime": 153.5748,
+      "eval_samples_per_second": 413.323,
+      "eval_steps_per_second": 1.615,
+      "step": 130000
+    },
+    {
+      "epoch": 1.7487029221215624,
+      "grad_norm": 0.47480854392051697,
+      "learning_rate": 0.004932171244949114,
+      "loss": 2.4463,
+      "step": 130100
+    },
+    {
+      "epoch": 1.7500470442753837,
+      "grad_norm": 0.9166473150253296,
+      "learning_rate": 0.004932048538605108,
+      "loss": 2.4467,
+      "step": 130200
+    },
+    {
+      "epoch": 1.751391166429205,
+      "grad_norm": 0.4468994438648224,
+      "learning_rate": 0.0049319257229020344,
+      "loss": 2.4422,
+      "step": 130300
+    },
+    {
+      "epoch": 1.7527352885830263,
+      "grad_norm": 2.049225091934204,
+      "learning_rate": 0.004931802797845434,
+      "loss": 2.4315,
+      "step": 130400
+    },
+    {
+      "epoch": 1.7540794107368478,
+      "grad_norm": 0.9961231350898743,
+      "learning_rate": 0.004931679763440849,
+      "loss": 2.4394,
+      "step": 130500
+    },
+    {
+      "epoch": 1.755423532890669,
+      "grad_norm": 0.8406715989112854,
+      "learning_rate": 0.00493155661969382,
+      "loss": 2.4436,
+      "step": 130600
+    },
+    {
+      "epoch": 1.7567676550444904,
+      "grad_norm": 0.6432211995124817,
+      "learning_rate": 0.004931433366609882,
+      "loss": 2.4352,
+      "step": 130700
+    },
+    {
+      "epoch": 1.758111777198312,
+      "grad_norm": 0.34998586773872375,
+      "learning_rate": 0.004931310004194613,
+      "loss": 2.4438,
+      "step": 130800
+    },
+    {
+      "epoch": 1.7594558993521332,
+      "grad_norm": 0.8097212910652161,
+      "learning_rate": 0.00493118653245356,
+      "loss": 2.4447,
+      "step": 130900
+    },
+    {
+      "epoch": 1.7608000215059545,
+      "grad_norm": 0.43594908714294434,
+      "learning_rate": 0.004931062951392285,
+      "loss": 2.4391,
+      "step": 131000
+    },
+    {
+      "epoch": 1.7608000215059545,
+      "eval_MaskedAccuracy": 0.4449512847956205,
+      "eval_loss": 2.7537848949432373,
+      "eval_runtime": 153.3786,
+      "eval_samples_per_second": 413.852,
+      "eval_steps_per_second": 1.617,
+      "step": 131000
+    },
+    {
+      "epoch": 1.7621441436597758,
+      "grad_norm": 0.555332362651825,
+      "learning_rate": 0.004930939261016357,
+      "loss": 2.439,
+      "step": 131100
+    },
+    {
+      "epoch": 1.763488265813597,
+      "grad_norm": 0.6408087611198425,
+      "learning_rate": 0.00493081546133135,
+      "loss": 2.4447,
+      "step": 131200
+    },
+    {
+      "epoch": 1.7648323879674184,
+      "grad_norm": 0.2732897996902466,
+      "learning_rate": 0.0049306915523428425,
+      "loss": 2.445,
+      "step": 131300
+    },
+    {
+      "epoch": 1.7661765101212397,
+      "grad_norm": 0.7091703414916992,
+      "learning_rate": 0.0049305675340564105,
+      "loss": 2.4377,
+      "step": 131400
+    },
+    {
+      "epoch": 1.7675206322750612,
+      "grad_norm": 0.3678547739982605,
+      "learning_rate": 0.004930443406477648,
+      "loss": 2.4239,
+      "step": 131500
+    },
+    {
+      "epoch": 1.7688647544288825,
+      "grad_norm": 0.7927549481391907,
+      "learning_rate": 0.004930319169612157,
+      "loss": 2.4298,
+      "step": 131600
+    },
+    {
+      "epoch": 1.770208876582704,
+      "grad_norm": 0.8828220963478088,
+      "learning_rate": 0.004930194823465527,
+      "loss": 2.4391,
+      "step": 131700
+    },
+    {
+      "epoch": 1.7715529987365253,
+      "grad_norm": 0.35567620396614075,
+      "learning_rate": 0.00493007036804336,
+      "loss": 2.4399,
+      "step": 131800
+    },
+    {
+      "epoch": 1.7728971208903466,
+      "grad_norm": 0.5886101722717285,
+      "learning_rate": 0.0049299458033512655,
+      "loss": 2.4472,
+      "step": 131900
+    },
+    {
+      "epoch": 1.7742412430441679,
+      "grad_norm": 0.3277896046638489,
+      "learning_rate": 0.0049298211293948594,
+      "loss": 2.4406,
+      "step": 132000
+    },
+    {
+      "epoch": 1.7742412430441679,
+      "eval_MaskedAccuracy": 0.44539411736195783,
+      "eval_loss": 2.7525792121887207,
+      "eval_runtime": 154.9926,
+      "eval_samples_per_second": 409.542,
+      "eval_steps_per_second": 1.6,
+      "step": 132000
+    },
+    {
+      "epoch": 1.7755853651979892,
+      "grad_norm": 0.967954695224762,
+      "learning_rate": 0.004929696346179746,
+      "loss": 2.445,
+      "step": 132100
+    },
+    {
+      "epoch": 1.7769294873518104,
+      "grad_norm": 0.38069549202919006,
+      "learning_rate": 0.004929571453711563,
+      "loss": 2.4397,
+      "step": 132200
+    },
+    {
+      "epoch": 1.7782736095056317,
+      "grad_norm": 0.7181752920150757,
+      "learning_rate": 0.004929446451995928,
+      "loss": 2.4411,
+      "step": 132300
+    },
+    {
+      "epoch": 1.7796177316594533,
+      "grad_norm": 0.7660533785820007,
+      "learning_rate": 0.004929321341038483,
+      "loss": 2.4344,
+      "step": 132400
+    },
+    {
+      "epoch": 1.7809618538132745,
+      "grad_norm": 1.3645398616790771,
+      "learning_rate": 0.004929196120844856,
+      "loss": 2.4429,
+      "step": 132500
+    },
+    {
+      "epoch": 1.7823059759670958,
+      "grad_norm": 0.40024033188819885,
+      "learning_rate": 0.004929070791420691,
+      "loss": 2.4525,
+      "step": 132600
+    },
+    {
+      "epoch": 1.7836500981209173,
+      "grad_norm": 2.2568159103393555,
+      "learning_rate": 0.004928945352771635,
+      "loss": 2.443,
+      "step": 132700
+    },
+    {
+      "epoch": 1.7849942202747386,
+      "grad_norm": 0.7219163775444031,
+      "learning_rate": 0.004928819804903343,
+      "loss": 2.4398,
+      "step": 132800
+    },
+    {
+      "epoch": 1.78633834242856,
+      "grad_norm": 1.4869344234466553,
+      "learning_rate": 0.004928694147821474,
+      "loss": 2.4397,
+      "step": 132900
+    },
+    {
+      "epoch": 1.7876824645823812,
+      "grad_norm": 0.3790728747844696,
+      "learning_rate": 0.0049285683815316825,
+      "loss": 2.4328,
+      "step": 133000
+    },
+    {
+      "epoch": 1.7876824645823812,
+      "eval_MaskedAccuracy": 0.44502501594055116,
+      "eval_loss": 2.751193046569824,
+      "eval_runtime": 154.0912,
+      "eval_samples_per_second": 411.938,
+      "eval_steps_per_second": 1.609,
+      "step": 133000
+    },
+    {
+      "epoch": 1.7890265867362025,
+      "grad_norm": 0.836524486541748,
+      "learning_rate": 0.00492844250603964,
+      "loss": 2.4433,
+      "step": 133100
+    },
+    {
+      "epoch": 1.7903707088900238,
+      "grad_norm": 0.29098621010780334,
+      "learning_rate": 0.004928316521351023,
+      "loss": 2.4376,
+      "step": 133200
+    },
+    {
+      "epoch": 1.791714831043845,
+      "grad_norm": 0.4187617003917694,
+      "learning_rate": 0.0049281904274715025,
+      "loss": 2.4439,
+      "step": 133300
+    },
+    {
+      "epoch": 1.7930589531976666,
+      "grad_norm": 1.8081835508346558,
+      "learning_rate": 0.004928064224406764,
+      "loss": 2.4372,
+      "step": 133400
+    },
+    {
+      "epoch": 1.794403075351488,
+      "grad_norm": 2.1965842247009277,
+      "learning_rate": 0.0049279379121624925,
+      "loss": 2.4302,
+      "step": 133500
+    },
+    {
+      "epoch": 1.7957471975053094,
+      "grad_norm": 1.1956915855407715,
+      "learning_rate": 0.004927811490744374,
+      "loss": 2.4379,
+      "step": 133600
+    },
+    {
+      "epoch": 1.7970913196591307,
+      "grad_norm": 0.7906240820884705,
+      "learning_rate": 0.004927684960158106,
+      "loss": 2.4423,
+      "step": 133700
+    },
+    {
+      "epoch": 1.798435441812952,
+      "grad_norm": 0.24601203203201294,
+      "learning_rate": 0.004927558320409392,
+      "loss": 2.4497,
+      "step": 133800
+    },
+    {
+      "epoch": 1.7997795639667733,
+      "grad_norm": 1.7259650230407715,
+      "learning_rate": 0.004927431571503945,
+      "loss": 2.4301,
+      "step": 133900
+    },
+    {
+      "epoch": 1.8011236861205946,
+      "grad_norm": 0.3489820957183838,
+      "learning_rate": 0.004927304713447465,
+      "loss": 2.4494,
+      "step": 134000
+    },
+    {
+      "epoch": 1.8011236861205946,
+      "eval_MaskedAccuracy": 0.4454901993570299,
+      "eval_loss": 2.751655101776123,
+      "eval_runtime": 155.8316,
+      "eval_samples_per_second": 407.337,
+      "eval_steps_per_second": 1.591,
+      "step": 134000
+    },
+    {
+      "epoch": 1.8024678082744159,
+      "grad_norm": 0.9807569980621338,
+      "learning_rate": 0.004927177746245671,
+      "loss": 2.4397,
+      "step": 134100
+    },
+    {
+      "epoch": 1.8038119304282372,
+      "grad_norm": 1.5453386306762695,
+      "learning_rate": 0.004927050669904284,
+      "loss": 2.4467,
+      "step": 134200
+    },
+    {
+      "epoch": 1.8051560525820587,
+      "grad_norm": 0.9938260912895203,
+      "learning_rate": 0.004926923484429032,
+      "loss": 2.4361,
+      "step": 134300
+    },
+    {
+      "epoch": 1.80650017473588,
+      "grad_norm": 1.328628420829773,
+      "learning_rate": 0.004926796189825648,
+      "loss": 2.4473,
+      "step": 134400
+    },
+    {
+      "epoch": 1.8078442968897015,
+      "grad_norm": 0.565514087677002,
+      "learning_rate": 0.004926668786099871,
+      "loss": 2.4383,
+      "step": 134500
+    },
+    {
+      "epoch": 1.8091884190435228,
+      "grad_norm": 3.7751681804656982,
+      "learning_rate": 0.004926541273257434,
+      "loss": 2.4386,
+      "step": 134600
+    },
+    {
+      "epoch": 1.810532541197344,
+      "grad_norm": 1.7504175901412964,
+      "learning_rate": 0.004926413651304083,
+      "loss": 2.4384,
+      "step": 134700
+    },
+    {
+      "epoch": 1.8118766633511654,
+      "grad_norm": 0.6130387783050537,
+      "learning_rate": 0.004926285920245568,
+      "loss": 2.437,
+      "step": 134800
+    },
+    {
+      "epoch": 1.8132207855049867,
+      "grad_norm": 0.5119666457176208,
+      "learning_rate": 0.004926158080087647,
+      "loss": 2.4349,
+      "step": 134900
+    },
+    {
+      "epoch": 1.814564907658808,
+      "grad_norm": 1.4035404920578003,
+      "learning_rate": 0.004926030130836081,
+      "loss": 2.4344,
+      "step": 135000
+    },
+    {
+      "epoch": 1.814564907658808,
+      "eval_MaskedAccuracy": 0.44536970526231284,
+      "eval_loss": 2.7522668838500977,
+      "eval_runtime": 154.4591,
+      "eval_samples_per_second": 410.957,
+      "eval_steps_per_second": 1.606,
+      "step": 135000
+    },
+    {
+      "epoch": 1.8159090298126292,
+      "grad_norm": 1.2096890211105347,
+      "learning_rate": 0.00492590207249663,
+      "loss": 2.4333,
+      "step": 135100
+    },
+    {
+      "epoch": 1.8172531519664508,
+      "grad_norm": 0.8168133497238159,
+      "learning_rate": 0.004925773905075064,
+      "loss": 2.4354,
+      "step": 135200
+    },
+    {
+      "epoch": 1.818597274120272,
+      "grad_norm": 0.46834301948547363,
+      "learning_rate": 0.004925645628577168,
+      "loss": 2.4345,
+      "step": 135300
+    },
+    {
+      "epoch": 1.8199413962740933,
+      "grad_norm": 0.7720035910606384,
+      "learning_rate": 0.004925517243008721,
+      "loss": 2.447,
+      "step": 135400
+    },
+    {
+      "epoch": 1.8212855184279149,
+      "grad_norm": 1.1966851949691772,
+      "learning_rate": 0.0049253887483755,
+      "loss": 2.4282,
+      "step": 135500
+    },
+    {
+      "epoch": 1.8226296405817362,
+      "grad_norm": 1.4665566682815552,
+      "learning_rate": 0.004925260144683294,
+      "loss": 2.4311,
+      "step": 135600
+    },
+    {
+      "epoch": 1.8239737627355574,
+      "grad_norm": 0.6458213925361633,
+      "learning_rate": 0.004925131431937897,
+      "loss": 2.4453,
+      "step": 135700
+    },
+    {
+      "epoch": 1.8253178848893787,
+      "grad_norm": 0.864764392375946,
+      "learning_rate": 0.004925002610145116,
+      "loss": 2.4315,
+      "step": 135800
+    },
+    {
+      "epoch": 1.8266620070432,
+      "grad_norm": 1.5035104751586914,
+      "learning_rate": 0.004924873679310756,
+      "loss": 2.435,
+      "step": 135900
+    },
+    {
+      "epoch": 1.8280061291970213,
+      "grad_norm": 0.5938438773155212,
+      "learning_rate": 0.004924744639440619,
+      "loss": 2.4461,
+      "step": 136000
+    },
+    {
+      "epoch": 1.8280061291970213,
+      "eval_MaskedAccuracy": 0.44473731371748054,
+      "eval_loss": 2.752194881439209,
+      "eval_runtime": 153.2163,
+      "eval_samples_per_second": 414.29,
+      "eval_steps_per_second": 1.619,
+      "step": 136000
+    },
+    {
+      "epoch": 1.8293502513508426,
+      "grad_norm": 1.69126558303833,
+      "learning_rate": 0.00492461549054052,
+      "loss": 2.4375,
+      "step": 136100
+    },
+    {
+      "epoch": 1.8306943735046641,
+      "grad_norm": 1.1451964378356934,
+      "learning_rate": 0.004924486232616281,
+      "loss": 2.4311,
+      "step": 136200
+    },
+    {
+      "epoch": 1.8320384956584854,
+      "grad_norm": 0.33867311477661133,
+      "learning_rate": 0.004924356865673731,
+      "loss": 2.4365,
+      "step": 136300
+    },
+    {
+      "epoch": 1.833382617812307,
+      "grad_norm": 0.7080395221710205,
+      "learning_rate": 0.0049242273897186955,
+      "loss": 2.438,
+      "step": 136400
+    },
+    {
+      "epoch": 1.8347267399661282,
+      "grad_norm": 0.8327525854110718,
+      "learning_rate": 0.0049240978047570055,
+      "loss": 2.4409,
+      "step": 136500
+    },
+    {
+      "epoch": 1.8360708621199495,
+      "grad_norm": 0.296032190322876,
+      "learning_rate": 0.004923968110794505,
+      "loss": 2.4377,
+      "step": 136600
+    },
+    {
+      "epoch": 1.8374149842737708,
+      "grad_norm": 0.8659167289733887,
+      "learning_rate": 0.00492383830783703,
+      "loss": 2.4317,
+      "step": 136700
+    },
+    {
+      "epoch": 1.838759106427592,
+      "grad_norm": 0.5200401544570923,
+      "learning_rate": 0.004923708395890436,
+      "loss": 2.445,
+      "step": 136800
+    },
+    {
+      "epoch": 1.8401032285814134,
+      "grad_norm": 1.0861247777938843,
+      "learning_rate": 0.004923578374960567,
+      "loss": 2.4332,
+      "step": 136900
+    },
+    {
+      "epoch": 1.8414473507352347,
+      "grad_norm": 0.3223403990268707,
+      "learning_rate": 0.004923448245053295,
+      "loss": 2.4422,
+      "step": 137000
+    },
+    {
+      "epoch": 1.8414473507352347,
+      "eval_MaskedAccuracy": 0.4446580648304679,
+      "eval_loss": 2.7543158531188965,
+      "eval_runtime": 153.5392,
+      "eval_samples_per_second": 413.419,
+      "eval_steps_per_second": 1.615,
+      "step": 137000
+    },
+    {
+      "epoch": 1.8427914728890562,
+      "grad_norm": 0.5863522887229919,
+      "learning_rate": 0.004923318006174473,
+      "loss": 2.4385,
+      "step": 137100
+    },
+    {
+      "epoch": 1.8441355950428775,
+      "grad_norm": 0.2514398694038391,
+      "learning_rate": 0.004923187658329979,
+      "loss": 2.4398,
+      "step": 137200
+    },
+    {
+      "epoch": 1.8454797171966988,
+      "grad_norm": 1.2516361474990845,
+      "learning_rate": 0.004923057201525681,
+      "loss": 2.4374,
+      "step": 137300
+    },
+    {
+      "epoch": 1.8468238393505203,
+      "grad_norm": 0.6415001153945923,
+      "learning_rate": 0.004922926635767458,
+      "loss": 2.4277,
+      "step": 137400
+    },
+    {
+      "epoch": 1.8481679615043416,
+      "grad_norm": 0.34172287583351135,
+      "learning_rate": 0.004922795961061186,
+      "loss": 2.425,
+      "step": 137500
+    },
+    {
+      "epoch": 1.8495120836581629,
+      "grad_norm": 2.2421274185180664,
+      "learning_rate": 0.004922665177412763,
+      "loss": 2.4364,
+      "step": 137600
+    },
+    {
+      "epoch": 1.8508562058119842,
+      "grad_norm": 0.48479756712913513,
+      "learning_rate": 0.00492253428482808,
+      "loss": 2.4314,
+      "step": 137700
+    },
+    {
+      "epoch": 1.8522003279658055,
+      "grad_norm": 0.8086810111999512,
+      "learning_rate": 0.00492240328331303,
+      "loss": 2.4392,
+      "step": 137800
+    },
+    {
+      "epoch": 1.8535444501196268,
+      "grad_norm": 0.8768898844718933,
+      "learning_rate": 0.004922272172873519,
+      "loss": 2.44,
+      "step": 137900
+    },
+    {
+      "epoch": 1.854888572273448,
+      "grad_norm": 0.7095423936843872,
+      "learning_rate": 0.004922140953515459,
+      "loss": 2.4383,
+      "step": 138000
+    },
+    {
+      "epoch": 1.854888572273448,
+      "eval_MaskedAccuracy": 0.44575168545661686,
+      "eval_loss": 2.747347116470337,
+      "eval_runtime": 154.9017,
+      "eval_samples_per_second": 409.782,
+      "eval_steps_per_second": 1.601,
+      "step": 138000
+    },
+    {
+      "epoch": 1.8562326944272696,
+      "grad_norm": 0.3725392818450928,
+      "learning_rate": 0.004922009625244753,
+      "loss": 2.4307,
+      "step": 138100
+    },
+    {
+      "epoch": 1.8575768165810909,
+      "grad_norm": 0.3521710932254791,
+      "learning_rate": 0.004921878188067327,
+      "loss": 2.4272,
+      "step": 138200
+    },
+    {
+      "epoch": 1.8589209387349124,
+      "grad_norm": 1.7538436651229858,
+      "learning_rate": 0.004921746641989098,
+      "loss": 2.4312,
+      "step": 138300
+    },
+    {
+      "epoch": 1.8602650608887337,
+      "grad_norm": 0.7190144658088684,
+      "learning_rate": 0.004921614987015991,
+      "loss": 2.4334,
+      "step": 138400
+    },
+    {
+      "epoch": 1.861609183042555,
+      "grad_norm": 0.3224917948246002,
+      "learning_rate": 0.004921483223153946,
+      "loss": 2.4315,
+      "step": 138500
+    },
+    {
+      "epoch": 1.8629533051963763,
+      "grad_norm": 0.6861767172813416,
+      "learning_rate": 0.004921351350408901,
+      "loss": 2.432,
+      "step": 138600
+    },
+    {
+      "epoch": 1.8642974273501975,
+      "grad_norm": 0.32877910137176514,
+      "learning_rate": 0.004921219368786787,
+      "loss": 2.4436,
+      "step": 138700
+    },
+    {
+      "epoch": 1.8656415495040188,
+      "grad_norm": 0.8731430172920227,
+      "learning_rate": 0.004921087278293557,
+      "loss": 2.4346,
+      "step": 138800
+    },
+    {
+      "epoch": 1.8669856716578401,
+      "grad_norm": 0.31697648763656616,
+      "learning_rate": 0.0049209550789351625,
+      "loss": 2.4353,
+      "step": 138900
+    },
+    {
+      "epoch": 1.8683297938116616,
+      "grad_norm": 0.30971816182136536,
+      "learning_rate": 0.004920822770717562,
+      "loss": 2.4242,
+      "step": 139000
+    },
+    {
+      "epoch": 1.8683297938116616,
+      "eval_MaskedAccuracy": 0.4455395990497761,
+      "eval_loss": 2.749335289001465,
+      "eval_runtime": 153.4613,
+      "eval_samples_per_second": 413.629,
+      "eval_steps_per_second": 1.616,
+      "step": 139000
+    },
+    {
+      "epoch": 1.869673915965483,
+      "grad_norm": 0.32947519421577454,
+      "learning_rate": 0.004920690353646715,
+      "loss": 2.4308,
+      "step": 139100
+    },
+    {
+      "epoch": 1.8710180381193042,
+      "grad_norm": 3.411869525909424,
+      "learning_rate": 0.004920557827728589,
+      "loss": 2.426,
+      "step": 139200
+    },
+    {
+      "epoch": 1.8723621602731257,
+      "grad_norm": 0.3665344715118408,
+      "learning_rate": 0.004920425192969151,
+      "loss": 2.4347,
+      "step": 139300
+    },
+    {
+      "epoch": 1.873706282426947,
+      "grad_norm": 0.7894236445426941,
+      "learning_rate": 0.0049202924493743794,
+      "loss": 2.435,
+      "step": 139400
+    },
+    {
+      "epoch": 1.8750504045807683,
+      "grad_norm": 0.37953993678092957,
+      "learning_rate": 0.00492015959695027,
+      "loss": 2.4363,
+      "step": 139500
+    },
+    {
+      "epoch": 1.8763945267345896,
+      "grad_norm": 0.32298392057418823,
+      "learning_rate": 0.004920026635702793,
+      "loss": 2.4356,
+      "step": 139600
+    },
+    {
+      "epoch": 1.877738648888411,
+      "grad_norm": 1.1138973236083984,
+      "learning_rate": 0.004919893565637941,
+      "loss": 2.4246,
+      "step": 139700
+    },
+    {
+      "epoch": 1.8790827710422322,
+      "grad_norm": 0.4021407663822174,
+      "learning_rate": 0.00491976038676171,
+      "loss": 2.4395,
+      "step": 139800
+    },
+    {
+      "epoch": 1.8804268931960535,
+      "grad_norm": 0.37863534688949585,
+      "learning_rate": 0.004919627099080102,
+      "loss": 2.4341,
+      "step": 139900
+    },
+    {
+      "epoch": 1.881771015349875,
+      "grad_norm": 1.6591562032699585,
+      "learning_rate": 0.004919493702599126,
+      "loss": 2.4329,
+      "step": 140000
+    },
+    {
+      "epoch": 1.881771015349875,
+      "eval_MaskedAccuracy": 0.4456517448251654,
+      "eval_loss": 2.7490234375,
+      "eval_runtime": 144.1126,
+      "eval_samples_per_second": 440.461,
+      "eval_steps_per_second": 1.721,
+      "step": 140000
+    },
+    {
+      "epoch": 1.8831151375036963,
+      "grad_norm": 0.8508136868476868,
+      "learning_rate": 0.004919360197324791,
+      "loss": 2.4389,
+      "step": 140100
+    },
+    {
+      "epoch": 1.8844592596575178,
+      "grad_norm": 0.9984667301177979,
+      "learning_rate": 0.004919226583263112,
+      "loss": 2.4296,
+      "step": 140200
+    },
+    {
+      "epoch": 1.885803381811339,
+      "grad_norm": 1.3535406589508057,
+      "learning_rate": 0.004919092860420102,
+      "loss": 2.4339,
+      "step": 140300
+    },
+    {
+      "epoch": 1.8871475039651604,
+      "grad_norm": 0.6479485630989075,
+      "learning_rate": 0.004918959028801795,
+      "loss": 2.4319,
+      "step": 140400
+    },
+    {
+      "epoch": 1.8884916261189817,
+      "grad_norm": 0.6749982237815857,
+      "learning_rate": 0.004918825088414224,
+      "loss": 2.433,
+      "step": 140500
+    },
+    {
+      "epoch": 1.889835748272803,
+      "grad_norm": 0.31086307764053345,
+      "learning_rate": 0.0049186910392634235,
+      "loss": 2.435,
+      "step": 140600
+    },
+    {
+      "epoch": 1.8911798704266243,
+      "grad_norm": 0.5695422291755676,
+      "learning_rate": 0.0049185568813554185,
+      "loss": 2.432,
+      "step": 140700
+    },
+    {
+      "epoch": 1.8925239925804456,
+      "grad_norm": 0.7300012707710266,
+      "learning_rate": 0.004918422614696274,
+      "loss": 2.4368,
+      "step": 140800
+    },
+    {
+      "epoch": 1.893868114734267,
+      "grad_norm": 0.415073424577713,
+      "learning_rate": 0.004918288239292025,
+      "loss": 2.443,
+      "step": 140900
+    },
+    {
+      "epoch": 1.8952122368880884,
+      "grad_norm": 0.9157136678695679,
+      "learning_rate": 0.0049181537551487315,
+      "loss": 2.4297,
+      "step": 141000
+    },
+    {
+      "epoch": 1.8952122368880884,
+      "eval_MaskedAccuracy": 0.4457131985564527,
+      "eval_loss": 2.7454721927642822,
+      "eval_runtime": 153.2602,
+      "eval_samples_per_second": 414.172,
+      "eval_steps_per_second": 1.618,
+      "step": 141000
+    },
+    {
+      "epoch": 1.89655635904191,
+      "grad_norm": 0.43132588267326355,
+      "learning_rate": 0.00491801916227246,
+      "loss": 2.4381,
+      "step": 141100
+    },
+    {
+      "epoch": 1.8979004811957312,
+      "grad_norm": 1.6390697956085205,
+      "learning_rate": 0.004917884460669262,
+      "loss": 2.4192,
+      "step": 141200
+    },
+    {
+      "epoch": 1.8992446033495525,
+      "grad_norm": 0.5319965481758118,
+      "learning_rate": 0.004917749650345217,
+      "loss": 2.4424,
+      "step": 141300
+    },
+    {
+      "epoch": 1.9005887255033738,
+      "grad_norm": 0.43463683128356934,
+      "learning_rate": 0.004917614731306398,
+      "loss": 2.4381,
+      "step": 141400
+    },
+    {
+      "epoch": 1.901932847657195,
+      "grad_norm": 1.1953731775283813,
+      "learning_rate": 0.004917479703558886,
+      "loss": 2.4382,
+      "step": 141500
+    },
+    {
+      "epoch": 1.9032769698110164,
+      "grad_norm": 3.3171629905700684,
+      "learning_rate": 0.004917344567108767,
+      "loss": 2.4285,
+      "step": 141600
+    },
+    {
+      "epoch": 1.9046210919648376,
+      "grad_norm": 1.354646921157837,
+      "learning_rate": 0.004917209321962115,
+      "loss": 2.4329,
+      "step": 141700
+    },
+    {
+      "epoch": 1.9059652141186592,
+      "grad_norm": 0.46209368109703064,
+      "learning_rate": 0.004917073968125036,
+      "loss": 2.4287,
+      "step": 141800
+    },
+    {
+      "epoch": 1.9073093362724804,
+      "grad_norm": 0.2556077241897583,
+      "learning_rate": 0.004916938505603623,
+      "loss": 2.4351,
+      "step": 141900
+    },
+    {
+      "epoch": 1.9086534584263017,
+      "grad_norm": 0.9437853097915649,
+      "learning_rate": 0.004916802934403986,
+      "loss": 2.4264,
+      "step": 142000
+    },
+    {
+      "epoch": 1.9086534584263017,
+      "eval_MaskedAccuracy": 0.44617851609047027,
+      "eval_loss": 2.7451670169830322,
+      "eval_runtime": 156.2618,
+      "eval_samples_per_second": 406.216,
+      "eval_steps_per_second": 1.587,
+      "step": 142000
+    },
+    {
+      "epoch": 1.9099975805801233,
+      "grad_norm": 0.6614341139793396,
+      "learning_rate": 0.004916667254532234,
+      "loss": 2.4258,
+      "step": 142100
+    },
+    {
+      "epoch": 1.9113417027339445,
+      "grad_norm": 2.363129138946533,
+      "learning_rate": 0.004916531465994474,
+      "loss": 2.4318,
+      "step": 142200
+    },
+    {
+      "epoch": 1.9126858248877658,
+      "grad_norm": 0.27388179302215576,
+      "learning_rate": 0.004916395568796826,
+      "loss": 2.4305,
+      "step": 142300
+    },
+    {
+      "epoch": 1.9140299470415871,
+      "grad_norm": 1.5506696701049805,
+      "learning_rate": 0.004916259562945416,
+      "loss": 2.4361,
+      "step": 142400
+    },
+    {
+      "epoch": 1.9153740691954084,
+      "grad_norm": 0.8063324093818665,
+      "learning_rate": 0.004916123448446375,
+      "loss": 2.4362,
+      "step": 142500
+    },
+    {
+      "epoch": 1.9167181913492297,
+      "grad_norm": 0.9321148991584778,
+      "learning_rate": 0.004915987225305831,
+      "loss": 2.4293,
+      "step": 142600
+    },
+    {
+      "epoch": 1.918062313503051,
+      "grad_norm": 0.9221600294113159,
+      "learning_rate": 0.004915850893529926,
+      "loss": 2.4282,
+      "step": 142700
+    },
+    {
+      "epoch": 1.9194064356568725,
+      "grad_norm": 0.31020745635032654,
+      "learning_rate": 0.004915714453124789,
+      "loss": 2.4326,
+      "step": 142800
+    },
+    {
+      "epoch": 1.9207505578106938,
+      "grad_norm": 0.9556496143341064,
+      "learning_rate": 0.004915577904096585,
+      "loss": 2.4317,
+      "step": 142900
+    },
+    {
+      "epoch": 1.9220946799645153,
+      "grad_norm": 0.9458506107330322,
+      "learning_rate": 0.004915441246451457,
+      "loss": 2.4328,
+      "step": 143000
+    },
+    {
+      "epoch": 1.9220946799645153,
+      "eval_MaskedAccuracy": 0.44678351331325233,
+      "eval_loss": 2.7399027347564697,
+      "eval_runtime": 154.7107,
+      "eval_samples_per_second": 410.288,
+      "eval_steps_per_second": 1.603,
+      "step": 143000
+    },
+    {
+      "epoch": 1.9234388021183366,
+      "grad_norm": 0.26708322763442993,
+      "learning_rate": 0.004915304480195574,
+      "loss": 2.4305,
+      "step": 143100
+    },
+    {
+      "epoch": 1.924782924272158,
+      "grad_norm": 0.42214831709861755,
+      "learning_rate": 0.004915167605335087,
+      "loss": 2.4342,
+      "step": 143200
+    },
+    {
+      "epoch": 1.9261270464259792,
+      "grad_norm": 1.0620460510253906,
+      "learning_rate": 0.00491503062187617,
+      "loss": 2.4318,
+      "step": 143300
+    },
+    {
+      "epoch": 1.9274711685798005,
+      "grad_norm": 0.45594775676727295,
+      "learning_rate": 0.004914893529824991,
+      "loss": 2.4298,
+      "step": 143400
+    },
+    {
+      "epoch": 1.9288152907336218,
+      "grad_norm": 0.8724369406700134,
+      "learning_rate": 0.00491475632918773,
+      "loss": 2.4303,
+      "step": 143500
+    },
+    {
+      "epoch": 1.930159412887443,
+      "grad_norm": 1.169735074043274,
+      "learning_rate": 0.004914619019970569,
+      "loss": 2.4335,
+      "step": 143600
+    },
+    {
+      "epoch": 1.9315035350412646,
+      "grad_norm": 0.28445443511009216,
+      "learning_rate": 0.00491448160217969,
+      "loss": 2.4301,
+      "step": 143700
+    },
+    {
+      "epoch": 1.9328476571950859,
+      "grad_norm": 1.0240504741668701,
+      "learning_rate": 0.004914344075821288,
+      "loss": 2.4395,
+      "step": 143800
+    },
+    {
+      "epoch": 1.9341917793489072,
+      "grad_norm": 0.2876899242401123,
+      "learning_rate": 0.004914206440901561,
+      "loss": 2.4273,
+      "step": 143900
+    },
+    {
+      "epoch": 1.9355359015027287,
+      "grad_norm": 1.793506383895874,
+      "learning_rate": 0.004914068697426708,
+      "loss": 2.4316,
+      "step": 144000
+    },
+    {
+      "epoch": 1.9355359015027287,
+      "eval_MaskedAccuracy": 0.4459802489403792,
+      "eval_loss": 2.7441201210021973,
+      "eval_runtime": 159.5024,
+      "eval_samples_per_second": 397.963,
+      "eval_steps_per_second": 1.555,
+      "step": 144000
+    },
+    {
+      "epoch": 1.93688002365655,
+      "grad_norm": 0.4511992931365967,
+      "learning_rate": 0.004913930845402937,
+      "loss": 2.4312,
+      "step": 144100
+    },
+    {
+      "epoch": 1.9382241458103713,
+      "grad_norm": 1.3304485082626343,
+      "learning_rate": 0.004913792884836461,
+      "loss": 2.4234,
+      "step": 144200
+    },
+    {
+      "epoch": 1.9395682679641926,
+      "grad_norm": 0.7679903507232666,
+      "learning_rate": 0.004913654815733498,
+      "loss": 2.4288,
+      "step": 144300
+    },
+    {
+      "epoch": 1.9409123901180139,
+      "grad_norm": 0.2566910982131958,
+      "learning_rate": 0.004913516638100267,
+      "loss": 2.4307,
+      "step": 144400
+    },
+    {
+      "epoch": 1.9422565122718352,
+      "grad_norm": 0.3400803804397583,
+      "learning_rate": 0.004913378351942991,
+      "loss": 2.4325,
+      "step": 144500
+    },
+    {
+      "epoch": 1.9436006344256564,
+      "grad_norm": 0.5881883502006531,
+      "learning_rate": 0.0049132399572679045,
+      "loss": 2.4283,
+      "step": 144600
+    },
+    {
+      "epoch": 1.944944756579478,
+      "grad_norm": 2.5710055828094482,
+      "learning_rate": 0.004913101454081242,
+      "loss": 2.4259,
+      "step": 144700
+    },
+    {
+      "epoch": 1.9462888787332993,
+      "grad_norm": 0.6169648766517639,
+      "learning_rate": 0.0049129628423892424,
+      "loss": 2.4406,
+      "step": 144800
+    },
+    {
+      "epoch": 1.9476330008871208,
+      "grad_norm": 0.541467010974884,
+      "learning_rate": 0.004912824122198156,
+      "loss": 2.4308,
+      "step": 144900
+    },
+    {
+      "epoch": 1.948977123040942,
+      "grad_norm": 1.3558932542800903,
+      "learning_rate": 0.004912685293514229,
+      "loss": 2.426,
+      "step": 145000
+    },
+    {
+      "epoch": 1.948977123040942,
+      "eval_MaskedAccuracy": 0.4466767450773424,
+      "eval_loss": 2.741316556930542,
+      "eval_runtime": 153.255,
+      "eval_samples_per_second": 414.186,
+      "eval_steps_per_second": 1.618,
+      "step": 145000
+    },
+    {
+      "epoch": 1.9503212451947634,
+      "grad_norm": 0.6421293616294861,
+      "learning_rate": 0.00491254635634372,
+      "loss": 2.4282,
+      "step": 145100
+    },
+    {
+      "epoch": 1.9516653673485846,
+      "grad_norm": 0.5589990615844727,
+      "learning_rate": 0.004912407310692889,
+      "loss": 2.4237,
+      "step": 145200
+    },
+    {
+      "epoch": 1.953009489502406,
+      "grad_norm": 0.45161235332489014,
+      "learning_rate": 0.004912268156567995,
+      "loss": 2.4336,
+      "step": 145300
+    },
+    {
+      "epoch": 1.9543536116562272,
+      "grad_norm": 0.6716010570526123,
+      "learning_rate": 0.004912128893975318,
+      "loss": 2.4239,
+      "step": 145400
+    },
+    {
+      "epoch": 1.9556977338100485,
+      "grad_norm": 1.577767014503479,
+      "learning_rate": 0.004911989522921124,
+      "loss": 2.4261,
+      "step": 145500
+    },
+    {
+      "epoch": 1.95704185596387,
+      "grad_norm": 1.3624939918518066,
+      "learning_rate": 0.004911850043411697,
+      "loss": 2.4406,
+      "step": 145600
+    },
+    {
+      "epoch": 1.9583859781176913,
+      "grad_norm": 0.23402433097362518,
+      "learning_rate": 0.004911710455453321,
+      "loss": 2.4291,
+      "step": 145700
+    },
+    {
+      "epoch": 1.9597301002715126,
+      "grad_norm": 0.6562730669975281,
+      "learning_rate": 0.004911570759052287,
+      "loss": 2.4325,
+      "step": 145800
+    },
+    {
+      "epoch": 1.9610742224253341,
+      "grad_norm": 0.2946057915687561,
+      "learning_rate": 0.004911430954214888,
+      "loss": 2.4268,
+      "step": 145900
+    },
+    {
+      "epoch": 1.9624183445791554,
+      "grad_norm": 2.216837167739868,
+      "learning_rate": 0.0049112910409474225,
+      "loss": 2.4272,
+      "step": 146000
+    },
+    {
+      "epoch": 1.9624183445791554,
+      "eval_MaskedAccuracy": 0.44681526480055744,
+      "eval_loss": 2.739447593688965,
+      "eval_runtime": 244.5176,
+      "eval_samples_per_second": 259.597,
+      "eval_steps_per_second": 1.014,
+      "step": 146000
+    },
+    {
+      "epoch": 1.9637624667329767,
+      "grad_norm": 0.9857944250106812,
+      "learning_rate": 0.004911151019256194,
+      "loss": 2.4263,
+      "step": 146100
+    },
+    {
+      "epoch": 1.965106588886798,
+      "grad_norm": 0.9716507792472839,
+      "learning_rate": 0.004911010889147516,
+      "loss": 2.4287,
+      "step": 146200
+    },
+    {
+      "epoch": 1.9664507110406193,
+      "grad_norm": 0.6002909541130066,
+      "learning_rate": 0.0049108706506277005,
+      "loss": 2.4169,
+      "step": 146300
+    },
+    {
+      "epoch": 1.9677948331944406,
+      "grad_norm": 0.33834025263786316,
+      "learning_rate": 0.004910730303703066,
+      "loss": 2.425,
+      "step": 146400
+    },
+    {
+      "epoch": 1.9691389553482619,
+      "grad_norm": 0.3823884129524231,
+      "learning_rate": 0.004910589848379935,
+      "loss": 2.4274,
+      "step": 146500
+    },
+    {
+      "epoch": 1.9704830775020834,
+      "grad_norm": 0.7107839584350586,
+      "learning_rate": 0.004910449284664644,
+      "loss": 2.4339,
+      "step": 146600
+    },
+    {
+      "epoch": 1.9718271996559047,
+      "grad_norm": 1.5292255878448486,
+      "learning_rate": 0.0049103086125635185,
+      "loss": 2.4211,
+      "step": 146700
+    },
+    {
+      "epoch": 1.9731713218097262,
+      "grad_norm": 0.4462113678455353,
+      "learning_rate": 0.0049101678320829035,
+      "loss": 2.4163,
+      "step": 146800
+    },
+    {
+      "epoch": 1.9745154439635475,
+      "grad_norm": 0.4614112377166748,
+      "learning_rate": 0.004910026943229135,
+      "loss": 2.4249,
+      "step": 146900
+    },
+    {
+      "epoch": 1.9758595661173688,
+      "grad_norm": 1.6407921314239502,
+      "learning_rate": 0.004909885946008562,
+      "loss": 2.4243,
+      "step": 147000
+    },
+    {
+      "epoch": 1.9758595661173688,
+      "eval_MaskedAccuracy": 0.44691000163355754,
+      "eval_loss": 2.740039348602295,
+      "eval_runtime": 160.8268,
+      "eval_samples_per_second": 394.686,
+      "eval_steps_per_second": 1.542,
+      "step": 147000
+    },
+    {
+      "epoch": 1.97720368827119,
+      "grad_norm": 0.3130664527416229,
+      "learning_rate": 0.004909744840427542,
+      "loss": 2.4207,
+      "step": 147100
+    },
+    {
+      "epoch": 1.9785478104250114,
+      "grad_norm": 0.3282364010810852,
+      "learning_rate": 0.004909603626492431,
+      "loss": 2.421,
+      "step": 147200
+    },
+    {
+      "epoch": 1.9798919325788327,
+      "grad_norm": 0.5056743621826172,
+      "learning_rate": 0.004909462304209587,
+      "loss": 2.4254,
+      "step": 147300
+    },
+    {
+      "epoch": 1.981236054732654,
+      "grad_norm": 1.1724507808685303,
+      "learning_rate": 0.0049093208735853855,
+      "loss": 2.428,
+      "step": 147400
+    },
+    {
+      "epoch": 1.9825801768864755,
+      "grad_norm": 0.2965724766254425,
+      "learning_rate": 0.004909179334626195,
+      "loss": 2.4272,
+      "step": 147500
+    },
+    {
+      "epoch": 1.9839242990402968,
+      "grad_norm": 0.37335798144340515,
+      "learning_rate": 0.004909037687338392,
+      "loss": 2.4197,
+      "step": 147600
+    },
+    {
+      "epoch": 1.9852684211941183,
+      "grad_norm": 0.5587185621261597,
+      "learning_rate": 0.0049088959317283645,
+      "loss": 2.4206,
+      "step": 147700
+    },
+    {
+      "epoch": 1.9866125433479396,
+      "grad_norm": 0.5161391496658325,
+      "learning_rate": 0.004908754067802488,
+      "loss": 2.4308,
+      "step": 147800
+    },
+    {
+      "epoch": 1.9879566655017609,
+      "grad_norm": 0.2679111361503601,
+      "learning_rate": 0.004908612095567173,
+      "loss": 2.4215,
+      "step": 147900
+    },
+    {
+      "epoch": 1.9893007876555822,
+      "grad_norm": 0.6613872051239014,
+      "learning_rate": 0.004908470015028803,
+      "loss": 2.4228,
+      "step": 148000
+    },
+    {
+      "epoch": 1.9893007876555822,
+      "eval_MaskedAccuracy": 0.4469897602234249,
+      "eval_loss": 2.7378077507019043,
+      "eval_runtime": 157.5436,
+      "eval_samples_per_second": 402.911,
+      "eval_steps_per_second": 1.574,
+      "step": 148000
+    },
+    {
+      "epoch": 1.9906449098094035,
+      "grad_norm": 0.6632961630821228,
+      "learning_rate": 0.0049083278261937814,
+      "loss": 2.4307,
+      "step": 148100
+    },
+    {
+      "epoch": 1.9919890319632247,
+      "grad_norm": 0.31573984026908875,
+      "learning_rate": 0.0049081855290685145,
+      "loss": 2.4226,
+      "step": 148200
+    },
+    {
+      "epoch": 1.993333154117046,
+      "grad_norm": 2.4147422313690186,
+      "learning_rate": 0.00490804312365942,
+      "loss": 2.4199,
+      "step": 148300
+    },
+    {
+      "epoch": 1.9946772762708675,
+      "grad_norm": 2.1030828952789307,
+      "learning_rate": 0.004907900609972911,
+      "loss": 2.4279,
+      "step": 148400
+    },
+    {
+      "epoch": 1.9960213984246888,
+      "grad_norm": 0.395871639251709,
+      "learning_rate": 0.004907757988015409,
+      "loss": 2.425,
+      "step": 148500
+    },
+    {
+      "epoch": 1.9973655205785101,
+      "grad_norm": 0.538977324962616,
+      "learning_rate": 0.004907615257793332,
+      "loss": 2.4217,
+      "step": 148600
+    },
+    {
+      "epoch": 1.9987096427323316,
+      "grad_norm": 1.8294909000396729,
+      "learning_rate": 0.004907472419313125,
+      "loss": 2.4302,
+      "step": 148700
+    },
+    {
+      "epoch": 2.000053764886153,
+      "grad_norm": 0.25931796431541443,
+      "learning_rate": 0.004907329472581221,
+      "loss": 2.4247,
+      "step": 148800
+    },
+    {
+      "epoch": 2.0013978870399742,
+      "grad_norm": 0.34021487832069397,
+      "learning_rate": 0.0049071864176040555,
+      "loss": 2.4319,
+      "step": 148900
+    },
+    {
+      "epoch": 2.0027420091937955,
+      "grad_norm": 0.7299911379814148,
+      "learning_rate": 0.004907043254388077,
+      "loss": 2.4311,
+      "step": 149000
+    },
+    {
+      "epoch": 2.0027420091937955,
+      "eval_MaskedAccuracy": 0.4465608726612502,
+      "eval_loss": 2.7390899658203125,
+      "eval_runtime": 595.6963,
+      "eval_samples_per_second": 106.558,
+      "eval_steps_per_second": 0.416,
+      "step": 149000
+    },
+    {
+      "epoch": 2.004086131347617,
+      "grad_norm": 0.7680445313453674,
+      "learning_rate": 0.004906899982939738,
+      "loss": 2.4315,
+      "step": 149100
+    },
+    {
+      "epoch": 2.005430253501438,
+      "grad_norm": 0.603999137878418,
+      "learning_rate": 0.004906756603265495,
+      "loss": 2.4207,
+      "step": 149200
+    },
+    {
+      "epoch": 2.0067743756552594,
+      "grad_norm": 0.995240330696106,
+      "learning_rate": 0.004906613115371805,
+      "loss": 2.4293,
+      "step": 149300
+    },
+    {
+      "epoch": 2.0081184978090807,
+      "grad_norm": 0.858191967010498,
+      "learning_rate": 0.0049064695192651345,
+      "loss": 2.4275,
+      "step": 149400
+    },
+    {
+      "epoch": 2.0094626199629024,
+      "grad_norm": 0.5625921487808228,
+      "learning_rate": 0.004906325814951952,
+      "loss": 2.4333,
+      "step": 149500
+    },
+    {
+      "epoch": 2.0108067421167237,
+      "grad_norm": 0.3029104769229889,
+      "learning_rate": 0.004906182002438735,
+      "loss": 2.4288,
+      "step": 149600
+    },
+    {
+      "epoch": 2.012150864270545,
+      "grad_norm": 0.6609681248664856,
+      "learning_rate": 0.004906038081731963,
+      "loss": 2.4128,
+      "step": 149700
+    },
+    {
+      "epoch": 2.0134949864243663,
+      "grad_norm": 0.48055407404899597,
+      "learning_rate": 0.004905894052838118,
+      "loss": 2.4281,
+      "step": 149800
+    },
+    {
+      "epoch": 2.0148391085781876,
+      "grad_norm": 0.2819332778453827,
+      "learning_rate": 0.00490574991576369,
+      "loss": 2.4247,
+      "step": 149900
+    },
+    {
+      "epoch": 2.016183230732009,
+      "grad_norm": 0.32838475704193115,
+      "learning_rate": 0.004905605670515174,
+      "loss": 2.4259,
+      "step": 150000
+    },
+    {
+      "epoch": 2.016183230732009,
+      "eval_MaskedAccuracy": 0.44722750040117315,
+      "eval_loss": 2.738093376159668,
+      "eval_runtime": 142.6909,
+      "eval_samples_per_second": 444.85,
+      "eval_steps_per_second": 1.738,
+      "step": 150000
+    },
+    {
+      "epoch": 2.01752735288583,
+      "grad_norm": 0.35216397047042847,
+      "learning_rate": 0.004905461317099078,
+      "loss": 2.432,
+      "step": 150100
+    },
+    {
+      "epoch": 2.0188714750396515,
+      "grad_norm": 0.7930958271026611,
+      "learning_rate": 0.004905316855521893,
+      "loss": 2.43,
+      "step": 150200
+    },
+    {
+      "epoch": 2.0202155971934728,
+      "grad_norm": 0.6561233401298523,
+      "learning_rate": 0.004905172285790133,
+      "loss": 2.432,
+      "step": 150300
+    },
+    {
+      "epoch": 2.0215597193472945,
+      "grad_norm": 0.8430966138839722,
+      "learning_rate": 0.0049050276079103195,
+      "loss": 2.4273,
+      "step": 150400
+    },
+    {
+      "epoch": 2.022903841501116,
+      "grad_norm": 0.4468148946762085,
+      "learning_rate": 0.004904882821888969,
+      "loss": 2.4255,
+      "step": 150500
+    },
+    {
+      "epoch": 2.024247963654937,
+      "grad_norm": 0.26562803983688354,
+      "learning_rate": 0.004904737927732603,
+      "loss": 2.428,
+      "step": 150600
+    },
+    {
+      "epoch": 2.0255920858087584,
+      "grad_norm": 0.35955631732940674,
+      "learning_rate": 0.004904592925447746,
+      "loss": 2.4227,
+      "step": 150700
+    },
+    {
+      "epoch": 2.0269362079625797,
+      "grad_norm": 1.039691686630249,
+      "learning_rate": 0.0049044478150409385,
+      "loss": 2.4229,
+      "step": 150800
+    },
+    {
+      "epoch": 2.028280330116401,
+      "grad_norm": 2.361480474472046,
+      "learning_rate": 0.004904302596518712,
+      "loss": 2.4246,
+      "step": 150900
+    },
+    {
+      "epoch": 2.0296244522702223,
+      "grad_norm": 0.4659264087677002,
+      "learning_rate": 0.004904157269887615,
+      "loss": 2.429,
+      "step": 151000
+    },
+    {
+      "epoch": 2.0296244522702223,
+      "eval_MaskedAccuracy": 0.44815486516670616,
+      "eval_loss": 2.7315616607666016,
+      "eval_runtime": 142.8587,
+      "eval_samples_per_second": 444.327,
+      "eval_steps_per_second": 1.736,
+      "step": 151000
+    },
+    {
+      "epoch": 2.0309685744240435,
+      "grad_norm": 0.3324325680732727,
+      "learning_rate": 0.004904011835154192,
+      "loss": 2.429,
+      "step": 151100
+    },
+    {
+      "epoch": 2.032312696577865,
+      "grad_norm": 0.4291480481624603,
+      "learning_rate": 0.0049038662923250025,
+      "loss": 2.4253,
+      "step": 151200
+    },
+    {
+      "epoch": 2.033656818731686,
+      "grad_norm": 0.5924742817878723,
+      "learning_rate": 0.004903720641406596,
+      "loss": 2.4181,
+      "step": 151300
+    },
+    {
+      "epoch": 2.035000940885508,
+      "grad_norm": 1.0112535953521729,
+      "learning_rate": 0.004903574882405539,
+      "loss": 2.422,
+      "step": 151400
+    },
+    {
+      "epoch": 2.036345063039329,
+      "grad_norm": 0.6629267334938049,
+      "learning_rate": 0.004903429015328402,
+      "loss": 2.4134,
+      "step": 151500
+    },
+    {
+      "epoch": 2.0376891851931505,
+      "grad_norm": 0.5437670350074768,
+      "learning_rate": 0.004903283040181754,
+      "loss": 2.4211,
+      "step": 151600
+    },
+    {
+      "epoch": 2.0390333073469717,
+      "grad_norm": 0.8838233947753906,
+      "learning_rate": 0.004903136956972172,
+      "loss": 2.4261,
+      "step": 151700
+    },
+    {
+      "epoch": 2.040377429500793,
+      "grad_norm": 0.493783563375473,
+      "learning_rate": 0.00490299076570624,
+      "loss": 2.4228,
+      "step": 151800
+    },
+    {
+      "epoch": 2.0417215516546143,
+      "grad_norm": 0.5075348019599915,
+      "learning_rate": 0.004902844466390544,
+      "loss": 2.4249,
+      "step": 151900
+    },
+    {
+      "epoch": 2.0430656738084356,
+      "grad_norm": 0.31101787090301514,
+      "learning_rate": 0.004902698059031681,
+      "loss": 2.4234,
+      "step": 152000
+    },
+    {
+      "epoch": 2.0430656738084356,
+      "eval_MaskedAccuracy": 0.4476862797973037,
+      "eval_loss": 2.733727216720581,
+      "eval_runtime": 142.8049,
+      "eval_samples_per_second": 444.495,
+      "eval_steps_per_second": 1.737,
+      "step": 152000
+    },
+    {
+      "epoch": 2.044409795962257,
+      "grad_norm": 0.3274334967136383,
+      "learning_rate": 0.004902551543636243,
+      "loss": 2.4265,
+      "step": 152100
+    },
+    {
+      "epoch": 2.045753918116078,
+      "grad_norm": 0.4391040503978729,
+      "learning_rate": 0.004902404920210837,
+      "loss": 2.4252,
+      "step": 152200
+    },
+    {
+      "epoch": 2.0470980402699,
+      "grad_norm": 0.36864349246025085,
+      "learning_rate": 0.004902258188762062,
+      "loss": 2.4191,
+      "step": 152300
+    },
+    {
+      "epoch": 2.0484421624237212,
+      "grad_norm": 0.32997363805770874,
+      "learning_rate": 0.004902111349296531,
+      "loss": 2.4301,
+      "step": 152400
+    },
+    {
+      "epoch": 2.0497862845775425,
+      "grad_norm": 0.649544358253479,
+      "learning_rate": 0.004901964401820856,
+      "loss": 2.4262,
+      "step": 152500
+    },
+    {
+      "epoch": 2.051130406731364,
+      "grad_norm": 0.5093933939933777,
+      "learning_rate": 0.004901817346341668,
+      "loss": 2.4184,
+      "step": 152600
+    },
+    {
+      "epoch": 2.052474528885185,
+      "grad_norm": 0.5928727984428406,
+      "learning_rate": 0.004901670182865587,
+      "loss": 2.4273,
+      "step": 152700
+    },
+    {
+      "epoch": 2.0538186510390064,
+      "grad_norm": 0.8361156582832336,
+      "learning_rate": 0.004901522911399248,
+      "loss": 2.4176,
+      "step": 152800
+    },
+    {
+      "epoch": 2.0551627731928277,
+      "grad_norm": 1.1973950862884521,
+      "learning_rate": 0.004901375531949283,
+      "loss": 2.4253,
+      "step": 152900
+    },
+    {
+      "epoch": 2.056506895346649,
+      "grad_norm": 0.7281692028045654,
+      "learning_rate": 0.004901228044522336,
+      "loss": 2.4213,
+      "step": 153000
+    },
+    {
+      "epoch": 2.056506895346649,
+      "eval_MaskedAccuracy": 0.4476455974274515,
+      "eval_loss": 2.7339048385620117,
+      "eval_runtime": 146.6823,
+      "eval_samples_per_second": 432.745,
+      "eval_steps_per_second": 1.691,
+      "step": 153000
+    },
+    {
+      "epoch": 2.0578510175004703,
+      "grad_norm": 0.908572793006897,
+      "learning_rate": 0.004901080449125051,
+      "loss": 2.4233,
+      "step": 153100
+    },
+    {
+      "epoch": 2.0591951396542916,
+      "grad_norm": 0.3152801990509033,
+      "learning_rate": 0.004900932745764079,
+      "loss": 2.4217,
+      "step": 153200
+    },
+    {
+      "epoch": 2.0605392618081133,
+      "grad_norm": 0.34237778186798096,
+      "learning_rate": 0.004900784934446075,
+      "loss": 2.4311,
+      "step": 153300
+    },
+    {
+      "epoch": 2.0618833839619346,
+      "grad_norm": 0.9181241393089294,
+      "learning_rate": 0.004900637015177697,
+      "loss": 2.4211,
+      "step": 153400
+    },
+    {
+      "epoch": 2.063227506115756,
+      "grad_norm": 1.0392390489578247,
+      "learning_rate": 0.004900488987965611,
+      "loss": 2.4163,
+      "step": 153500
+    },
+    {
+      "epoch": 2.064571628269577,
+      "grad_norm": 1.2790428400039673,
+      "learning_rate": 0.0049003408528164915,
+      "loss": 2.4157,
+      "step": 153600
+    },
+    {
+      "epoch": 2.0659157504233985,
+      "grad_norm": 0.2699611186981201,
+      "learning_rate": 0.004900192609737011,
+      "loss": 2.4267,
+      "step": 153700
+    },
+    {
+      "epoch": 2.0672598725772198,
+      "grad_norm": 1.472973346710205,
+      "learning_rate": 0.004900044258733849,
+      "loss": 2.4306,
+      "step": 153800
+    },
+    {
+      "epoch": 2.068603994731041,
+      "grad_norm": 0.27637550234794617,
+      "learning_rate": 0.004899895799813685,
+      "loss": 2.4221,
+      "step": 153900
+    },
+    {
+      "epoch": 2.0699481168848624,
+      "grad_norm": 1.1613603830337524,
+      "learning_rate": 0.004899747232983215,
+      "loss": 2.4191,
+      "step": 154000
+    },
+    {
+      "epoch": 2.0699481168848624,
+      "eval_MaskedAccuracy": 0.4478018154999935,
+      "eval_loss": 2.7331624031066895,
+      "eval_runtime": 142.1229,
+      "eval_samples_per_second": 446.628,
+      "eval_steps_per_second": 1.745,
+      "step": 154000
+    },
+    {
+      "epoch": 2.0712922390386836,
+      "grad_norm": 1.339164137840271,
+      "learning_rate": 0.004899598558249125,
+      "loss": 2.4229,
+      "step": 154100
+    },
+    {
+      "epoch": 2.0726363611925054,
+      "grad_norm": 0.558399498462677,
+      "learning_rate": 0.004899449775618128,
+      "loss": 2.4168,
+      "step": 154200
+    },
+    {
+      "epoch": 2.0739804833463267,
+      "grad_norm": 0.6973606944084167,
+      "learning_rate": 0.00489930088509692,
+      "loss": 2.4206,
+      "step": 154300
+    },
+    {
+      "epoch": 2.075324605500148,
+      "grad_norm": 0.7255196571350098,
+      "learning_rate": 0.004899151886692205,
+      "loss": 2.4187,
+      "step": 154400
+    },
+    {
+      "epoch": 2.0766687276539693,
+      "grad_norm": 1.0322102308273315,
+      "learning_rate": 0.004899002780410702,
+      "loss": 2.4304,
+      "step": 154500
+    },
+    {
+      "epoch": 2.0780128498077906,
+      "grad_norm": 0.8069867491722107,
+      "learning_rate": 0.0048988535662591245,
+      "loss": 2.4222,
+      "step": 154600
+    },
+    {
+      "epoch": 2.079356971961612,
+      "grad_norm": 1.4786266088485718,
+      "learning_rate": 0.004898704244244206,
+      "loss": 2.4193,
+      "step": 154700
+    },
+    {
+      "epoch": 2.080701094115433,
+      "grad_norm": 0.7500434517860413,
+      "learning_rate": 0.004898554814372665,
+      "loss": 2.4213,
+      "step": 154800
+    },
+    {
+      "epoch": 2.0820452162692544,
+      "grad_norm": 1.6421656608581543,
+      "learning_rate": 0.004898405276651237,
+      "loss": 2.4169,
+      "step": 154900
+    },
+    {
+      "epoch": 2.0833893384230757,
+      "grad_norm": 1.4270743131637573,
+      "learning_rate": 0.0048982556310866576,
+      "loss": 2.4242,
+      "step": 155000
+    },
+    {
+      "epoch": 2.0833893384230757,
+      "eval_MaskedAccuracy": 0.44795841587608753,
+      "eval_loss": 2.7314157485961914,
+      "eval_runtime": 142.2388,
+      "eval_samples_per_second": 446.264,
+      "eval_steps_per_second": 1.744,
+      "step": 155000
+    },
+    {
+      "epoch": 2.0847334605768975,
+      "grad_norm": 1.0751755237579346,
+      "learning_rate": 0.004898105877685676,
+      "loss": 2.4239,
+      "step": 155100
+    },
+    {
+      "epoch": 2.0860775827307187,
+      "grad_norm": 0.43115946650505066,
+      "learning_rate": 0.004897956016455043,
+      "loss": 2.4202,
+      "step": 155200
+    },
+    {
+      "epoch": 2.08742170488454,
+      "grad_norm": 0.6999830007553101,
+      "learning_rate": 0.004897806047401499,
+      "loss": 2.4164,
+      "step": 155300
+    },
+    {
+      "epoch": 2.0887658270383613,
+      "grad_norm": 0.6027560830116272,
+      "learning_rate": 0.004897655970531812,
+      "loss": 2.4118,
+      "step": 155400
+    },
+    {
+      "epoch": 2.0901099491921826,
+      "grad_norm": 2.5767462253570557,
+      "learning_rate": 0.004897505785852734,
+      "loss": 2.4254,
+      "step": 155500
+    },
+    {
+      "epoch": 2.091454071346004,
+      "grad_norm": 0.6744506359100342,
+      "learning_rate": 0.004897355493371032,
+      "loss": 2.4192,
+      "step": 155600
+    },
+    {
+      "epoch": 2.092798193499825,
+      "grad_norm": 0.6279339790344238,
+      "learning_rate": 0.004897205093093492,
+      "loss": 2.4212,
+      "step": 155700
+    },
+    {
+      "epoch": 2.0941423156536465,
+      "grad_norm": 0.317688912153244,
+      "learning_rate": 0.00489705458502688,
+      "loss": 2.4179,
+      "step": 155800
+    },
+    {
+      "epoch": 2.095486437807468,
+      "grad_norm": 0.5679840445518494,
+      "learning_rate": 0.004896903969177977,
+      "loss": 2.4157,
+      "step": 155900
+    },
+    {
+      "epoch": 2.096830559961289,
+      "grad_norm": 0.6120138764381409,
+      "learning_rate": 0.0048967532455535725,
+      "loss": 2.4172,
+      "step": 156000
+    },
+    {
+      "epoch": 2.096830559961289,
+      "eval_MaskedAccuracy": 0.44820394067035313,
+      "eval_loss": 2.730539321899414,
+      "eval_runtime": 143.8277,
+      "eval_samples_per_second": 441.334,
+      "eval_steps_per_second": 1.724,
+      "step": 156000
+    },
+    {
+      "epoch": 2.098174682115111,
+      "grad_norm": 1.3471684455871582,
+      "learning_rate": 0.004896602414160458,
+      "loss": 2.412,
+      "step": 156100
+    },
+    {
+      "epoch": 2.099518804268932,
+      "grad_norm": 2.606961250305176,
+      "learning_rate": 0.0048964514750054365,
+      "loss": 2.4199,
+      "step": 156200
+    },
+    {
+      "epoch": 2.1008629264227534,
+      "grad_norm": 0.44379231333732605,
+      "learning_rate": 0.004896300428095295,
+      "loss": 2.4295,
+      "step": 156300
+    },
+    {
+      "epoch": 2.1022070485765747,
+      "grad_norm": 0.4089781641960144,
+      "learning_rate": 0.004896149273436849,
+      "loss": 2.4195,
+      "step": 156400
+    },
+    {
+      "epoch": 2.103551170730396,
+      "grad_norm": 0.39541006088256836,
+      "learning_rate": 0.004895998011036907,
+      "loss": 2.4226,
+      "step": 156500
+    },
+    {
+      "epoch": 2.1048952928842173,
+      "grad_norm": 0.7441678047180176,
+      "learning_rate": 0.004895846640902289,
+      "loss": 2.4146,
+      "step": 156600
+    },
+    {
+      "epoch": 2.1062394150380386,
+      "grad_norm": 0.2654988765716553,
+      "learning_rate": 0.004895695163039808,
+      "loss": 2.4115,
+      "step": 156700
+    },
+    {
+      "epoch": 2.10758353719186,
+      "grad_norm": 0.23396815359592438,
+      "learning_rate": 0.004895543577456294,
+      "loss": 2.422,
+      "step": 156800
+    },
+    {
+      "epoch": 2.108927659345681,
+      "grad_norm": 0.3832063674926758,
+      "learning_rate": 0.004895391884158574,
+      "loss": 2.4195,
+      "step": 156900
+    },
+    {
+      "epoch": 2.1102717814995025,
+      "grad_norm": 0.6434290409088135,
+      "learning_rate": 0.0048952400831534845,
+      "loss": 2.4227,
+      "step": 157000
+    },
+    {
+      "epoch": 2.1102717814995025,
+      "eval_MaskedAccuracy": 0.448189960789518,
+      "eval_loss": 2.7293646335601807,
+      "eval_runtime": 149.7286,
+      "eval_samples_per_second": 423.94,
+      "eval_steps_per_second": 1.656,
+      "step": 157000
+    },
+    {
+      "epoch": 2.111615903653324,
+      "grad_norm": 0.5817483067512512,
+      "learning_rate": 0.004895088174447862,
+      "loss": 2.4267,
+      "step": 157100
+    },
+    {
+      "epoch": 2.1129600258071455,
+      "grad_norm": 0.36821791529655457,
+      "learning_rate": 0.004894936158048559,
+      "loss": 2.4268,
+      "step": 157200
+    },
+    {
+      "epoch": 2.1143041479609668,
+      "grad_norm": 0.5711908340454102,
+      "learning_rate": 0.004894784033962416,
+      "loss": 2.4233,
+      "step": 157300
+    },
+    {
+      "epoch": 2.115648270114788,
+      "grad_norm": 0.4792524576187134,
+      "learning_rate": 0.004894631802196298,
+      "loss": 2.4199,
+      "step": 157400
+    },
+    {
+      "epoch": 2.1169923922686094,
+      "grad_norm": 0.9362654089927673,
+      "learning_rate": 0.0048944794627570525,
+      "loss": 2.425,
+      "step": 157500
+    },
+    {
+      "epoch": 2.1183365144224306,
+      "grad_norm": 0.8550450801849365,
+      "learning_rate": 0.004894327015651554,
+      "loss": 2.4171,
+      "step": 157600
+    },
+    {
+      "epoch": 2.119680636576252,
+      "grad_norm": 0.2275136560201645,
+      "learning_rate": 0.004894174460886669,
+      "loss": 2.4112,
+      "step": 157700
+    },
+    {
+      "epoch": 2.1210247587300732,
+      "grad_norm": 0.8724859356880188,
+      "learning_rate": 0.004894021798469268,
+      "loss": 2.4187,
+      "step": 157800
+    },
+    {
+      "epoch": 2.1223688808838945,
+      "grad_norm": 0.3414950966835022,
+      "learning_rate": 0.004893869028406233,
+      "loss": 2.4113,
+      "step": 157900
+    },
+    {
+      "epoch": 2.1237130030377163,
+      "grad_norm": 0.5261027216911316,
+      "learning_rate": 0.004893716150704447,
+      "loss": 2.4249,
+      "step": 158000
+    },
+    {
+      "epoch": 2.1237130030377163,
+      "eval_MaskedAccuracy": 0.44844991045030136,
+      "eval_loss": 2.7277510166168213,
+      "eval_runtime": 152.1905,
+      "eval_samples_per_second": 417.083,
+      "eval_steps_per_second": 1.63,
+      "step": 158000
+    },
+    {
+      "epoch": 2.1250571251915376,
+      "grad_norm": 0.24052013456821442,
+      "learning_rate": 0.0048935631653708,
+      "loss": 2.4195,
+      "step": 158100
+    },
+    {
+      "epoch": 2.126401247345359,
+      "grad_norm": 0.9782668352127075,
+      "learning_rate": 0.004893410072412176,
+      "loss": 2.4313,
+      "step": 158200
+    },
+    {
+      "epoch": 2.12774536949918,
+      "grad_norm": 1.6755679845809937,
+      "learning_rate": 0.004893256871835483,
+      "loss": 2.4201,
+      "step": 158300
+    },
+    {
+      "epoch": 2.1290894916530014,
+      "grad_norm": 0.42762598395347595,
+      "learning_rate": 0.00489310356364762,
+      "loss": 2.4158,
+      "step": 158400
+    },
+    {
+      "epoch": 2.1304336138068227,
+      "grad_norm": 0.2700982093811035,
+      "learning_rate": 0.0048929501478555005,
+      "loss": 2.4236,
+      "step": 158500
+    },
+    {
+      "epoch": 2.131777735960644,
+      "grad_norm": 0.9225497245788574,
+      "learning_rate": 0.004892796624466027,
+      "loss": 2.4191,
+      "step": 158600
+    },
+    {
+      "epoch": 2.1331218581144653,
+      "grad_norm": 0.9907844066619873,
+      "learning_rate": 0.004892642993486124,
+      "loss": 2.4111,
+      "step": 158700
+    },
+    {
+      "epoch": 2.1344659802682866,
+      "grad_norm": 0.24535821378231049,
+      "learning_rate": 0.004892489254922714,
+      "loss": 2.4277,
+      "step": 158800
+    },
+    {
+      "epoch": 2.1358101024221083,
+      "grad_norm": 1.3044700622558594,
+      "learning_rate": 0.004892335408782724,
+      "loss": 2.4191,
+      "step": 158900
+    },
+    {
+      "epoch": 2.1371542245759296,
+      "grad_norm": 1.1276298761367798,
+      "learning_rate": 0.0048921814550730814,
+      "loss": 2.4142,
+      "step": 159000
+    },
+    {
+      "epoch": 2.1371542245759296,
+      "eval_MaskedAccuracy": 0.44877584775466844,
+      "eval_loss": 2.726320505142212,
+      "eval_runtime": 144.1741,
+      "eval_samples_per_second": 440.273,
+      "eval_steps_per_second": 1.72,
+      "step": 159000
+    },
+    {
+      "epoch": 2.138498346729751,
+      "grad_norm": 0.3968954086303711,
+      "learning_rate": 0.004892027393800723,
+      "loss": 2.4149,
+      "step": 159100
+    },
+    {
+      "epoch": 2.139842468883572,
+      "grad_norm": 0.3040700852870941,
+      "learning_rate": 0.004891873224972598,
+      "loss": 2.4198,
+      "step": 159200
+    },
+    {
+      "epoch": 2.1411865910373935,
+      "grad_norm": 0.24721510708332062,
+      "learning_rate": 0.00489171894859565,
+      "loss": 2.4209,
+      "step": 159300
+    },
+    {
+      "epoch": 2.142530713191215,
+      "grad_norm": 0.3159930408000946,
+      "learning_rate": 0.004891564564676828,
+      "loss": 2.4214,
+      "step": 159400
+    },
+    {
+      "epoch": 2.143874835345036,
+      "grad_norm": 0.31632721424102783,
+      "learning_rate": 0.004891410073223092,
+      "loss": 2.4242,
+      "step": 159500
+    },
+    {
+      "epoch": 2.1452189574988574,
+      "grad_norm": 0.422014981508255,
+      "learning_rate": 0.0048912554742414,
+      "loss": 2.43,
+      "step": 159600
+    },
+    {
+      "epoch": 2.1465630796526787,
+      "grad_norm": 0.2504478394985199,
+      "learning_rate": 0.004891100767738721,
+      "loss": 2.4201,
+      "step": 159700
+    },
+    {
+      "epoch": 2.1479072018065004,
+      "grad_norm": 0.9411721229553223,
+      "learning_rate": 0.0048909459537220265,
+      "loss": 2.4305,
+      "step": 159800
+    },
+    {
+      "epoch": 2.1492513239603217,
+      "grad_norm": 0.7317768931388855,
+      "learning_rate": 0.004890791032198286,
+      "loss": 2.4167,
+      "step": 159900
+    },
+    {
+      "epoch": 2.150595446114143,
+      "grad_norm": 0.462750643491745,
+      "learning_rate": 0.004890636003174482,
+      "loss": 2.4022,
+      "step": 160000
+    },
+    {
+      "epoch": 2.150595446114143,
+      "eval_MaskedAccuracy": 0.44730266116401834,
+      "eval_loss": 2.733778953552246,
+      "eval_runtime": 157.466,
+      "eval_samples_per_second": 403.109,
+      "eval_steps_per_second": 1.575,
+      "step": 160000
+    },
+    {
+      "epoch": 2.1519395682679643,
+      "grad_norm": 1.2427386045455933,
+      "learning_rate": 0.004890480866657605,
+      "loss": 2.4187,
+      "step": 160100
+    },
+    {
+      "epoch": 2.1532836904217856,
+      "grad_norm": 1.1014420986175537,
+      "learning_rate": 0.004890325622654636,
+      "loss": 2.4167,
+      "step": 160200
+    },
+    {
+      "epoch": 2.154627812575607,
+      "grad_norm": 1.7939949035644531,
+      "learning_rate": 0.004890170271172577,
+      "loss": 2.413,
+      "step": 160300
+    },
+    {
+      "epoch": 2.155971934729428,
+      "grad_norm": 0.28519657254219055,
+      "learning_rate": 0.004890014812218431,
+      "loss": 2.4248,
+      "step": 160400
+    },
+    {
+      "epoch": 2.1573160568832495,
+      "grad_norm": 0.3315064609050751,
+      "learning_rate": 0.0048898592457991994,
+      "loss": 2.4238,
+      "step": 160500
+    },
+    {
+      "epoch": 2.1586601790370707,
+      "grad_norm": 0.5888499617576599,
+      "learning_rate": 0.004889703571921885,
+      "loss": 2.4134,
+      "step": 160600
+    },
+    {
+      "epoch": 2.160004301190892,
+      "grad_norm": 0.3281387686729431,
+      "learning_rate": 0.004889547790593514,
+      "loss": 2.4204,
+      "step": 160700
+    },
+    {
+      "epoch": 2.1613484233447133,
+      "grad_norm": 0.3036564588546753,
+      "learning_rate": 0.004889391901821095,
+      "loss": 2.4143,
+      "step": 160800
+    },
+    {
+      "epoch": 2.162692545498535,
+      "grad_norm": 0.5783784985542297,
+      "learning_rate": 0.004889235905611661,
+      "loss": 2.4287,
+      "step": 160900
+    },
+    {
+      "epoch": 2.1640366676523564,
+      "grad_norm": 0.7875732183456421,
+      "learning_rate": 0.004889079801972228,
+      "loss": 2.4151,
+      "step": 161000
+    },
+    {
+      "epoch": 2.1640366676523564,
+      "eval_MaskedAccuracy": 0.4483840000513704,
+      "eval_loss": 2.7266931533813477,
+      "eval_runtime": 157.0844,
+      "eval_samples_per_second": 404.089,
+      "eval_steps_per_second": 1.579,
+      "step": 161000
+    },
+    {
+      "epoch": 2.1653807898061777,
+      "grad_norm": 1.2114744186401367,
+      "learning_rate": 0.0048889235909098374,
+      "loss": 2.4198,
+      "step": 161100
+    },
+    {
+      "epoch": 2.166724911959999,
+      "grad_norm": 3.989194631576538,
+      "learning_rate": 0.0048887672724315335,
+      "loss": 2.4213,
+      "step": 161200
+    },
+    {
+      "epoch": 2.1680690341138202,
+      "grad_norm": 1.195610761642456,
+      "learning_rate": 0.0048886108465443535,
+      "loss": 2.4187,
+      "step": 161300
+    },
+    {
+      "epoch": 2.1694131562676415,
+      "grad_norm": 1.6807376146316528,
+      "learning_rate": 0.004888454313255354,
+      "loss": 2.4067,
+      "step": 161400
+    },
+    {
+      "epoch": 2.170757278421463,
+      "grad_norm": 0.3829794228076935,
+      "learning_rate": 0.0048882976725715815,
+      "loss": 2.41,
+      "step": 161500
+    },
+    {
+      "epoch": 2.172101400575284,
+      "grad_norm": 0.2655125558376312,
+      "learning_rate": 0.00488814092450009,
+      "loss": 2.416,
+      "step": 161600
+    },
+    {
+      "epoch": 2.1734455227291054,
+      "grad_norm": 0.6671358942985535,
+      "learning_rate": 0.004887984069047942,
+      "loss": 2.4155,
+      "step": 161700
+    },
+    {
+      "epoch": 2.174789644882927,
+      "grad_norm": 0.49830421805381775,
+      "learning_rate": 0.004887827106222209,
+      "loss": 2.4146,
+      "step": 161800
+    },
+    {
+      "epoch": 2.1761337670367484,
+      "grad_norm": 0.43950071930885315,
+      "learning_rate": 0.004887670036029963,
+      "loss": 2.4123,
+      "step": 161900
+    },
+    {
+      "epoch": 2.1774778891905697,
+      "grad_norm": 0.9079190492630005,
+      "learning_rate": 0.004887512858478287,
+      "loss": 2.4123,
+      "step": 162000
+    },
+    {
+      "epoch": 2.1774778891905697,
+      "eval_MaskedAccuracy": 0.4486591905523086,
+      "eval_loss": 2.7250816822052,
+      "eval_runtime": 154.8776,
+      "eval_samples_per_second": 409.846,
+      "eval_steps_per_second": 1.601,
+      "step": 162000
+    },
+    {
+      "epoch": 2.178822011344391,
+      "grad_norm": 1.3062362670898438,
+      "learning_rate": 0.004887355573574258,
+      "loss": 2.4201,
+      "step": 162100
+    },
+    {
+      "epoch": 2.1801661334982123,
+      "grad_norm": 0.9617155194282532,
+      "learning_rate": 0.004887198181324969,
+      "loss": 2.4178,
+      "step": 162200
+    },
+    {
+      "epoch": 2.1815102556520336,
+      "grad_norm": 0.23437173664569855,
+      "learning_rate": 0.004887040681737504,
+      "loss": 2.4276,
+      "step": 162300
+    },
+    {
+      "epoch": 2.182854377805855,
+      "grad_norm": 0.7280550599098206,
+      "learning_rate": 0.004886883074818961,
+      "loss": 2.4122,
+      "step": 162400
+    },
+    {
+      "epoch": 2.184198499959676,
+      "grad_norm": 0.26790088415145874,
+      "learning_rate": 0.004886725360576443,
+      "loss": 2.4176,
+      "step": 162500
+    },
+    {
+      "epoch": 2.1855426221134975,
+      "grad_norm": 0.24585595726966858,
+      "learning_rate": 0.004886567539017049,
+      "loss": 2.4256,
+      "step": 162600
+    },
+    {
+      "epoch": 2.186886744267319,
+      "grad_norm": 0.2857859134674072,
+      "learning_rate": 0.0048864096101479005,
+      "loss": 2.4147,
+      "step": 162700
+    },
+    {
+      "epoch": 2.1882308664211405,
+      "grad_norm": 0.4526360332965851,
+      "learning_rate": 0.00488625157397611,
+      "loss": 2.4197,
+      "step": 162800
+    },
+    {
+      "epoch": 2.189574988574962,
+      "grad_norm": 0.2927253246307373,
+      "learning_rate": 0.004886093430508798,
+      "loss": 2.4114,
+      "step": 162900
+    },
+    {
+      "epoch": 2.190919110728783,
+      "grad_norm": 0.4610482156276703,
+      "learning_rate": 0.004885935179753095,
+      "loss": 2.4147,
+      "step": 163000
+    },
+    {
+      "epoch": 2.190919110728783,
+      "eval_MaskedAccuracy": 0.44903692639620885,
+      "eval_loss": 2.7242319583892822,
+      "eval_runtime": 155.5888,
+      "eval_samples_per_second": 407.973,
+      "eval_steps_per_second": 1.594,
+      "step": 163000
+    },
+    {
+      "epoch": 2.1922632328826044,
+      "grad_norm": 0.6389359831809998,
+      "learning_rate": 0.004885776821716123,
+      "loss": 2.4158,
+      "step": 163100
+    },
+    {
+      "epoch": 2.1936073550364257,
+      "grad_norm": 1.6146184206008911,
+      "learning_rate": 0.004885618356405023,
+      "loss": 2.4194,
+      "step": 163200
+    },
+    {
+      "epoch": 2.194951477190247,
+      "grad_norm": 1.686237096786499,
+      "learning_rate": 0.004885459783826932,
+      "loss": 2.4109,
+      "step": 163300
+    },
+    {
+      "epoch": 2.1962955993440683,
+      "grad_norm": 1.0167760848999023,
+      "learning_rate": 0.004885301103989002,
+      "loss": 2.4146,
+      "step": 163400
+    },
+    {
+      "epoch": 2.1976397214978896,
+      "grad_norm": 0.20749007165431976,
+      "learning_rate": 0.004885142316898374,
+      "loss": 2.4179,
+      "step": 163500
+    },
+    {
+      "epoch": 2.1989838436517113,
+      "grad_norm": 0.3694406747817993,
+      "learning_rate": 0.0048849834225622064,
+      "loss": 2.411,
+      "step": 163600
+    },
+    {
+      "epoch": 2.2003279658055326,
+      "grad_norm": 0.6222627758979797,
+      "learning_rate": 0.004884824420987657,
+      "loss": 2.4116,
+      "step": 163700
+    },
+    {
+      "epoch": 2.201672087959354,
+      "grad_norm": 0.637772798538208,
+      "learning_rate": 0.004884665312181901,
+      "loss": 2.4217,
+      "step": 163800
+    },
+    {
+      "epoch": 2.203016210113175,
+      "grad_norm": 1.5972890853881836,
+      "learning_rate": 0.004884506096152096,
+      "loss": 2.4098,
+      "step": 163900
+    },
+    {
+      "epoch": 2.2043603322669965,
+      "grad_norm": 0.2235487997531891,
+      "learning_rate": 0.004884346772905415,
+      "loss": 2.4088,
+      "step": 164000
+    },
+    {
+      "epoch": 2.2043603322669965,
+      "eval_MaskedAccuracy": 0.44855764145601995,
+      "eval_loss": 2.725599527359009,
+      "eval_runtime": 153.7593,
+      "eval_samples_per_second": 412.827,
+      "eval_steps_per_second": 1.613,
+      "step": 164000
+    },
+    {
+      "epoch": 2.2057044544208178,
+      "grad_norm": 0.5953556895256042,
+      "learning_rate": 0.004884187342449037,
+      "loss": 2.4125,
+      "step": 164100
+    },
+    {
+      "epoch": 2.207048576574639,
+      "grad_norm": 0.6413670778274536,
+      "learning_rate": 0.004884027804790152,
+      "loss": 2.4191,
+      "step": 164200
+    },
+    {
+      "epoch": 2.2083926987284603,
+      "grad_norm": 0.5590651035308838,
+      "learning_rate": 0.004883868159935944,
+      "loss": 2.4153,
+      "step": 164300
+    },
+    {
+      "epoch": 2.2097368208822816,
+      "grad_norm": 0.6301490664482117,
+      "learning_rate": 0.0048837084078936135,
+      "loss": 2.4125,
+      "step": 164400
+    },
+    {
+      "epoch": 2.211080943036103,
+      "grad_norm": 1.1809839010238647,
+      "learning_rate": 0.004883548548670345,
+      "loss": 2.4097,
+      "step": 164500
+    },
+    {
+      "epoch": 2.2124250651899247,
+      "grad_norm": 0.8796495199203491,
+      "learning_rate": 0.004883388582273348,
+      "loss": 2.4151,
+      "step": 164600
+    },
+    {
+      "epoch": 2.213769187343746,
+      "grad_norm": 0.5694831609725952,
+      "learning_rate": 0.004883228508709839,
+      "loss": 2.4123,
+      "step": 164700
+    },
+    {
+      "epoch": 2.2151133094975672,
+      "grad_norm": 1.4738579988479614,
+      "learning_rate": 0.0048830683279870194,
+      "loss": 2.3999,
+      "step": 164800
+    },
+    {
+      "epoch": 2.2164574316513885,
+      "grad_norm": 0.6466926336288452,
+      "learning_rate": 0.004882908040112116,
+      "loss": 2.4129,
+      "step": 164900
+    },
+    {
+      "epoch": 2.21780155380521,
+      "grad_norm": 0.2502327263355255,
+      "learning_rate": 0.004882747645092345,
+      "loss": 2.4139,
+      "step": 165000
+    },
+    {
+      "epoch": 2.21780155380521,
+      "eval_MaskedAccuracy": 0.44872007353941323,
+      "eval_loss": 2.723762273788452,
+      "eval_runtime": 158.1038,
+      "eval_samples_per_second": 401.483,
+      "eval_steps_per_second": 1.569,
+      "step": 165000
+    },
+    {
+      "epoch": 2.219145675959031,
+      "grad_norm": 0.5904813408851624,
+      "learning_rate": 0.004882587142934935,
+      "loss": 2.4146,
+      "step": 165100
+    },
+    {
+      "epoch": 2.2204897981128524,
+      "grad_norm": 0.41806700825691223,
+      "learning_rate": 0.004882426533647115,
+      "loss": 2.4111,
+      "step": 165200
+    },
+    {
+      "epoch": 2.2218339202666737,
+      "grad_norm": 0.30772215127944946,
+      "learning_rate": 0.00488226581723612,
+      "loss": 2.4097,
+      "step": 165300
+    },
+    {
+      "epoch": 2.223178042420495,
+      "grad_norm": 0.6158583760261536,
+      "learning_rate": 0.0048821049937092026,
+      "loss": 2.4132,
+      "step": 165400
+    },
+    {
+      "epoch": 2.2245221645743163,
+      "grad_norm": 0.5486952066421509,
+      "learning_rate": 0.004881944063073602,
+      "loss": 2.4082,
+      "step": 165500
+    },
+    {
+      "epoch": 2.225866286728138,
+      "grad_norm": 0.44802579283714294,
+      "learning_rate": 0.00488178302533657,
+      "loss": 2.4052,
+      "step": 165600
+    },
+    {
+      "epoch": 2.2272104088819593,
+      "grad_norm": 0.6078771948814392,
+      "learning_rate": 0.004881621880505358,
+      "loss": 2.4174,
+      "step": 165700
+    },
+    {
+      "epoch": 2.2285545310357806,
+      "grad_norm": 1.0394670963287354,
+      "learning_rate": 0.004881460628587237,
+      "loss": 2.4053,
+      "step": 165800
+    },
+    {
+      "epoch": 2.229898653189602,
+      "grad_norm": 0.7442373037338257,
+      "learning_rate": 0.0048812992695894675,
+      "loss": 2.4099,
+      "step": 165900
+    },
+    {
+      "epoch": 2.231242775343423,
+      "grad_norm": 0.39984771609306335,
+      "learning_rate": 0.004881137803519318,
+      "loss": 2.4191,
+      "step": 166000
+    },
+    {
+      "epoch": 2.231242775343423,
+      "eval_MaskedAccuracy": 0.44866405937837617,
+      "eval_loss": 2.7227699756622314,
+      "eval_runtime": 176.0029,
+      "eval_samples_per_second": 360.653,
+      "eval_steps_per_second": 1.409,
+      "step": 166000
+    },
+    {
+      "epoch": 2.2325868974972445,
+      "grad_norm": 0.40001779794692993,
+      "learning_rate": 0.004880976230384064,
+      "loss": 2.4152,
+      "step": 166100
+    },
+    {
+      "epoch": 2.2339310196510658,
+      "grad_norm": 0.8670006394386292,
+      "learning_rate": 0.00488081455019099,
+      "loss": 2.4158,
+      "step": 166200
+    },
+    {
+      "epoch": 2.235275141804887,
+      "grad_norm": 0.5148455500602722,
+      "learning_rate": 0.004880652762947384,
+      "loss": 2.4162,
+      "step": 166300
+    },
+    {
+      "epoch": 2.2366192639587084,
+      "grad_norm": 1.6046350002288818,
+      "learning_rate": 0.004880490868660532,
+      "loss": 2.4131,
+      "step": 166400
+    },
+    {
+      "epoch": 2.23796338611253,
+      "grad_norm": 0.896712601184845,
+      "learning_rate": 0.004880328867337728,
+      "loss": 2.4037,
+      "step": 166500
+    },
+    {
+      "epoch": 2.2393075082663514,
+      "grad_norm": 0.881832480430603,
+      "learning_rate": 0.0048801667589862706,
+      "loss": 2.4167,
+      "step": 166600
+    },
+    {
+      "epoch": 2.2406516304201727,
+      "grad_norm": 3.5371510982513428,
+      "learning_rate": 0.00488000454361346,
+      "loss": 2.411,
+      "step": 166700
+    },
+    {
+      "epoch": 2.241995752573994,
+      "grad_norm": 0.2387372851371765,
+      "learning_rate": 0.004879842221226608,
+      "loss": 2.4134,
+      "step": 166800
+    },
+    {
+      "epoch": 2.2433398747278153,
+      "grad_norm": 0.2676979899406433,
+      "learning_rate": 0.00487967979183303,
+      "loss": 2.4213,
+      "step": 166900
+    },
+    {
+      "epoch": 2.2446839968816366,
+      "grad_norm": 0.9509516954421997,
+      "learning_rate": 0.0048795172554400475,
+      "loss": 2.4127,
+      "step": 167000
+    },
+    {
+      "epoch": 2.2446839968816366,
+      "eval_MaskedAccuracy": 0.44885926012655697,
+      "eval_loss": 2.7260682582855225,
+      "eval_runtime": 166.3086,
+      "eval_samples_per_second": 381.676,
+      "eval_steps_per_second": 1.491,
+      "step": 167000
+    },
+    {
+      "epoch": 2.246028119035458,
+      "grad_norm": 0.3096467852592468,
+      "learning_rate": 0.004879354612054984,
+      "loss": 2.4122,
+      "step": 167100
+    },
+    {
+      "epoch": 2.247372241189279,
+      "grad_norm": 0.6709736585617065,
+      "learning_rate": 0.004879191861685164,
+      "loss": 2.4082,
+      "step": 167200
+    },
+    {
+      "epoch": 2.2487163633431004,
+      "grad_norm": 0.5945542454719543,
+      "learning_rate": 0.004879029004337925,
+      "loss": 2.4125,
+      "step": 167300
+    },
+    {
+      "epoch": 2.250060485496922,
+      "grad_norm": 1.153786063194275,
+      "learning_rate": 0.004878866040020597,
+      "loss": 2.4236,
+      "step": 167400
+    },
+    {
+      "epoch": 2.2514046076507435,
+      "grad_norm": 0.9520905613899231,
+      "learning_rate": 0.004878702968740533,
+      "loss": 2.4141,
+      "step": 167500
+    },
+    {
+      "epoch": 2.2527487298045648,
+      "grad_norm": 0.8178815841674805,
+      "learning_rate": 0.004878539790505079,
+      "loss": 2.4127,
+      "step": 167600
+    },
+    {
+      "epoch": 2.254092851958386,
+      "grad_norm": 0.29091590642929077,
+      "learning_rate": 0.004878376505321578,
+      "loss": 2.4077,
+      "step": 167700
+    },
+    {
+      "epoch": 2.2554369741122073,
+      "grad_norm": 1.5120651721954346,
+      "learning_rate": 0.004878213113197398,
+      "loss": 2.4125,
+      "step": 167800
+    },
+    {
+      "epoch": 2.2567810962660286,
+      "grad_norm": 1.6276648044586182,
+      "learning_rate": 0.004878049614139895,
+      "loss": 2.4122,
+      "step": 167900
+    },
+    {
+      "epoch": 2.25812521841985,
+      "grad_norm": 0.786770761013031,
+      "learning_rate": 0.004877886008156443,
+      "loss": 2.405,
+      "step": 168000
+    },
+    {
+      "epoch": 2.25812521841985,
+      "eval_MaskedAccuracy": 0.4498786265923044,
+      "eval_loss": 2.719548225402832,
+      "eval_runtime": 155.0214,
+      "eval_samples_per_second": 409.466,
+      "eval_steps_per_second": 1.6,
+      "step": 168000
+    },
+    {
+      "epoch": 2.259469340573671,
+      "grad_norm": 0.46489110589027405,
+      "learning_rate": 0.004877722295254405,
+      "loss": 2.4138,
+      "step": 168100
+    },
+    {
+      "epoch": 2.2608134627274925,
+      "grad_norm": 1.4252897500991821,
+      "learning_rate": 0.004877558475441166,
+      "loss": 2.411,
+      "step": 168200
+    },
+    {
+      "epoch": 2.2621575848813142,
+      "grad_norm": 0.2540140450000763,
+      "learning_rate": 0.004877394548724101,
+      "loss": 2.4093,
+      "step": 168300
+    },
+    {
+      "epoch": 2.2635017070351355,
+      "grad_norm": 0.3754124045372009,
+      "learning_rate": 0.004877230515110599,
+      "loss": 2.4146,
+      "step": 168400
+    },
+    {
+      "epoch": 2.264845829188957,
+      "grad_norm": 0.6744637489318848,
+      "learning_rate": 0.0048770663746080505,
+      "loss": 2.4203,
+      "step": 168500
+    },
+    {
+      "epoch": 2.266189951342778,
+      "grad_norm": 0.21679449081420898,
+      "learning_rate": 0.004876902127223849,
+      "loss": 2.4013,
+      "step": 168600
+    },
+    {
+      "epoch": 2.2675340734965994,
+      "grad_norm": 0.3662445843219757,
+      "learning_rate": 0.004876737772965405,
+      "loss": 2.4108,
+      "step": 168700
+    },
+    {
+      "epoch": 2.2688781956504207,
+      "grad_norm": 1.2381483316421509,
+      "learning_rate": 0.004876573311840115,
+      "loss": 2.4117,
+      "step": 168800
+    },
+    {
+      "epoch": 2.270222317804242,
+      "grad_norm": 1.7564358711242676,
+      "learning_rate": 0.0048764087438553875,
+      "loss": 2.4082,
+      "step": 168900
+    },
+    {
+      "epoch": 2.2715664399580633,
+      "grad_norm": 0.6631215810775757,
+      "learning_rate": 0.004876244069018646,
+      "loss": 2.4133,
+      "step": 169000
+    },
+    {
+      "epoch": 2.2715664399580633,
+      "eval_MaskedAccuracy": 0.45023057912506814,
+      "eval_loss": 2.716172695159912,
+      "eval_runtime": 156.7093,
+      "eval_samples_per_second": 405.056,
+      "eval_steps_per_second": 1.583,
+      "step": 169000
+    },
+    {
+      "epoch": 2.2729105621118846,
+      "grad_norm": 0.8016698360443115,
+      "learning_rate": 0.004876079287337305,
+      "loss": 2.4051,
+      "step": 169100
+    },
+    {
+      "epoch": 2.2742546842657063,
+      "grad_norm": 0.519679844379425,
+      "learning_rate": 0.0048759143988187925,
+      "loss": 2.4115,
+      "step": 169200
+    },
+    {
+      "epoch": 2.275598806419527,
+      "grad_norm": 0.7286385297775269,
+      "learning_rate": 0.004875749403470536,
+      "loss": 2.4202,
+      "step": 169300
+    },
+    {
+      "epoch": 2.276942928573349,
+      "grad_norm": 0.24820998311042786,
+      "learning_rate": 0.004875584301299965,
+      "loss": 2.405,
+      "step": 169400
+    },
+    {
+      "epoch": 2.27828705072717,
+      "grad_norm": 0.25017213821411133,
+      "learning_rate": 0.004875419092314521,
+      "loss": 2.4129,
+      "step": 169500
+    },
+    {
+      "epoch": 2.2796311728809915,
+      "grad_norm": 0.8957297801971436,
+      "learning_rate": 0.004875253776521664,
+      "loss": 2.4071,
+      "step": 169600
+    },
+    {
+      "epoch": 2.2809752950348128,
+      "grad_norm": 0.5003229975700378,
+      "learning_rate": 0.004875088353928819,
+      "loss": 2.4099,
+      "step": 169700
+    },
+    {
+      "epoch": 2.282319417188634,
+      "grad_norm": 0.6370465159416199,
+      "learning_rate": 0.004874922824543451,
+      "loss": 2.4257,
+      "step": 169800
+    },
+    {
+      "epoch": 2.2836635393424554,
+      "grad_norm": 0.30033034086227417,
+      "learning_rate": 0.004874757188373019,
+      "loss": 2.4151,
+      "step": 169900
+    },
+    {
+      "epoch": 2.2850076614962767,
+      "grad_norm": 1.7922015190124512,
+      "learning_rate": 0.004874591445424987,
+      "loss": 2.4176,
+      "step": 170000
+    },
+    {
+      "epoch": 2.2850076614962767,
+      "eval_MaskedAccuracy": 0.4488507191873566,
+      "eval_loss": 2.722308874130249,
+      "eval_runtime": 156.2994,
+      "eval_samples_per_second": 406.118,
+      "eval_steps_per_second": 1.587,
+      "step": 170000
+    },
+    {
+      "epoch": 2.286351783650098,
+      "grad_norm": 1.080620288848877,
+      "learning_rate": 0.004874425595706819,
+      "loss": 2.4126,
+      "step": 170100
+    },
+    {
+      "epoch": 2.2876959058039192,
+      "grad_norm": 0.4531891644001007,
+      "learning_rate": 0.004874259639225995,
+      "loss": 2.4058,
+      "step": 170200
+    },
+    {
+      "epoch": 2.289040027957741,
+      "grad_norm": 0.2731966972351074,
+      "learning_rate": 0.0048740935759899835,
+      "loss": 2.4072,
+      "step": 170300
+    },
+    {
+      "epoch": 2.2903841501115623,
+      "grad_norm": 0.5683355927467346,
+      "learning_rate": 0.004873927406006273,
+      "loss": 2.4048,
+      "step": 170400
+    },
+    {
+      "epoch": 2.2917282722653836,
+      "grad_norm": 1.147813081741333,
+      "learning_rate": 0.004873761129282351,
+      "loss": 2.403,
+      "step": 170500
+    },
+    {
+      "epoch": 2.293072394419205,
+      "grad_norm": 0.30079033970832825,
+      "learning_rate": 0.0048735947458257095,
+      "loss": 2.4076,
+      "step": 170600
+    },
+    {
+      "epoch": 2.294416516573026,
+      "grad_norm": 0.29191407561302185,
+      "learning_rate": 0.00487342825564384,
+      "loss": 2.4128,
+      "step": 170700
+    },
+    {
+      "epoch": 2.2957606387268474,
+      "grad_norm": 0.7784181833267212,
+      "learning_rate": 0.004873261658744253,
+      "loss": 2.409,
+      "step": 170800
+    },
+    {
+      "epoch": 2.2971047608806687,
+      "grad_norm": 0.7186245322227478,
+      "learning_rate": 0.004873094955134448,
+      "loss": 2.4138,
+      "step": 170900
+    },
+    {
+      "epoch": 2.29844888303449,
+      "grad_norm": 0.9298619031906128,
+      "learning_rate": 0.004872928144821944,
+      "loss": 2.413,
+      "step": 171000
+    },
+    {
+      "epoch": 2.29844888303449,
+      "eval_MaskedAccuracy": 0.4487523079732683,
+      "eval_loss": 2.7223293781280518,
+      "eval_runtime": 154.8012,
+      "eval_samples_per_second": 410.049,
+      "eval_steps_per_second": 1.602,
+      "step": 171000
+    },
+    {
+      "epoch": 2.2997930051883113,
+      "grad_norm": 0.9932366013526917,
+      "learning_rate": 0.00487276122781425,
+      "loss": 2.417,
+      "step": 171100
+    },
+    {
+      "epoch": 2.301137127342133,
+      "grad_norm": 1.6287879943847656,
+      "learning_rate": 0.0048725942041188885,
+      "loss": 2.41,
+      "step": 171200
+    },
+    {
+      "epoch": 2.3024812494959543,
+      "grad_norm": 0.38423383235931396,
+      "learning_rate": 0.00487242707374338,
+      "loss": 2.4128,
+      "step": 171300
+    },
+    {
+      "epoch": 2.3038253716497756,
+      "grad_norm": 1.0105299949645996,
+      "learning_rate": 0.0048722598366952715,
+      "loss": 2.4199,
+      "step": 171400
+    },
+    {
+      "epoch": 2.305169493803597,
+      "grad_norm": 1.1391443014144897,
+      "learning_rate": 0.004872092492982082,
+      "loss": 2.4101,
+      "step": 171500
+    },
+    {
+      "epoch": 2.306513615957418,
+      "grad_norm": 0.2690454125404358,
+      "learning_rate": 0.004871925042611355,
+      "loss": 2.4109,
+      "step": 171600
+    },
+    {
+      "epoch": 2.3078577381112395,
+      "grad_norm": 0.3632848262786865,
+      "learning_rate": 0.004871757485590642,
+      "loss": 2.4156,
+      "step": 171700
+    },
+    {
+      "epoch": 2.309201860265061,
+      "grad_norm": 1.6692334413528442,
+      "learning_rate": 0.004871589821927492,
+      "loss": 2.415,
+      "step": 171800
+    },
+    {
+      "epoch": 2.310545982418882,
+      "grad_norm": 0.8343886137008667,
+      "learning_rate": 0.004871422051629453,
+      "loss": 2.4143,
+      "step": 171900
+    },
+    {
+      "epoch": 2.3118901045727034,
+      "grad_norm": 0.42459598183631897,
+      "learning_rate": 0.004871254174704084,
+      "loss": 2.4116,
+      "step": 172000
+    },
+    {
+      "epoch": 2.3118901045727034,
+      "eval_MaskedAccuracy": 0.44924426931396766,
+      "eval_loss": 2.720871925354004,
+      "eval_runtime": 154.4054,
+      "eval_samples_per_second": 411.099,
+      "eval_steps_per_second": 1.606,
+      "step": 172000
+    },
+    {
+      "epoch": 2.313234226726525,
+      "grad_norm": 0.23972781002521515,
+      "learning_rate": 0.004871086191158964,
+      "loss": 2.4127,
+      "step": 172100
+    },
+    {
+      "epoch": 2.3145783488803464,
+      "grad_norm": 0.5230980515480042,
+      "learning_rate": 0.004870918101001645,
+      "loss": 2.4101,
+      "step": 172200
+    },
+    {
+      "epoch": 2.3159224710341677,
+      "grad_norm": 1.2739980220794678,
+      "learning_rate": 0.004870749904239708,
+      "loss": 2.4125,
+      "step": 172300
+    },
+    {
+      "epoch": 2.317266593187989,
+      "grad_norm": 0.2384161502122879,
+      "learning_rate": 0.004870581600880735,
+      "loss": 2.4088,
+      "step": 172400
+    },
+    {
+      "epoch": 2.3186107153418103,
+      "grad_norm": 0.5129702091217041,
+      "learning_rate": 0.004870413190932297,
+      "loss": 2.4088,
+      "step": 172500
+    },
+    {
+      "epoch": 2.3199548374956316,
+      "grad_norm": 0.2832634449005127,
+      "learning_rate": 0.004870244674402003,
+      "loss": 2.4016,
+      "step": 172600
+    },
+    {
+      "epoch": 2.321298959649453,
+      "grad_norm": 0.287016361951828,
+      "learning_rate": 0.004870076051297425,
+      "loss": 2.4126,
+      "step": 172700
+    },
+    {
+      "epoch": 2.322643081803274,
+      "grad_norm": 0.2521541118621826,
+      "learning_rate": 0.0048699073216261725,
+      "loss": 2.4122,
+      "step": 172800
+    },
+    {
+      "epoch": 2.3239872039570955,
+      "grad_norm": 0.29343220591545105,
+      "learning_rate": 0.004869738485395845,
+      "loss": 2.3996,
+      "step": 172900
+    },
+    {
+      "epoch": 2.325331326110917,
+      "grad_norm": 0.2879350185394287,
+      "learning_rate": 0.004869569542614052,
+      "loss": 2.4084,
+      "step": 173000
+    },
+    {
+      "epoch": 2.325331326110917,
+      "eval_MaskedAccuracy": 0.4492615010440258,
+      "eval_loss": 2.7211360931396484,
+      "eval_runtime": 156.2801,
+      "eval_samples_per_second": 406.168,
+      "eval_steps_per_second": 1.587,
+      "step": 173000
+    },
+    {
+      "epoch": 2.3266754482647385,
+      "grad_norm": 0.30386149883270264,
+      "learning_rate": 0.004869400493288407,
+      "loss": 2.4125,
+      "step": 173100
+    },
+    {
+      "epoch": 2.32801957041856,
+      "grad_norm": 0.30869200825691223,
+      "learning_rate": 0.004869231337426524,
+      "loss": 2.413,
+      "step": 173200
+    },
+    {
+      "epoch": 2.329363692572381,
+      "grad_norm": 0.6456079483032227,
+      "learning_rate": 0.0048690620750360255,
+      "loss": 2.4144,
+      "step": 173300
+    },
+    {
+      "epoch": 2.3307078147262024,
+      "grad_norm": 0.6145516633987427,
+      "learning_rate": 0.004868892706124542,
+      "loss": 2.4062,
+      "step": 173400
+    },
+    {
+      "epoch": 2.3320519368800237,
+      "grad_norm": 0.5485299229621887,
+      "learning_rate": 0.004868723230699693,
+      "loss": 2.401,
+      "step": 173500
+    },
+    {
+      "epoch": 2.333396059033845,
+      "grad_norm": 1.082629680633545,
+      "learning_rate": 0.004868553648769127,
+      "loss": 2.413,
+      "step": 173600
+    },
+    {
+      "epoch": 2.3347401811876662,
+      "grad_norm": 0.5388498306274414,
+      "learning_rate": 0.004868383960340481,
+      "loss": 2.4097,
+      "step": 173700
+    },
+    {
+      "epoch": 2.3360843033414875,
+      "grad_norm": 0.23575937747955322,
+      "learning_rate": 0.004868214165421399,
+      "loss": 2.4054,
+      "step": 173800
+    },
+    {
+      "epoch": 2.337428425495309,
+      "grad_norm": 0.7409723997116089,
+      "learning_rate": 0.004868044264019538,
+      "loss": 2.4085,
+      "step": 173900
+    },
+    {
+      "epoch": 2.33877254764913,
+      "grad_norm": 0.2735578715801239,
+      "learning_rate": 0.004867874256142546,
+      "loss": 2.4156,
+      "step": 174000
+    },
+    {
+      "epoch": 2.33877254764913,
+      "eval_MaskedAccuracy": 0.4495013367228055,
+      "eval_loss": 2.7185449600219727,
+      "eval_runtime": 154.8489,
+      "eval_samples_per_second": 409.922,
+      "eval_steps_per_second": 1.602,
+      "step": 174000
+    },
+    {
+      "epoch": 2.340116669802952,
+      "grad_norm": 0.26877549290657043,
+      "learning_rate": 0.004867704141798087,
+      "loss": 2.4045,
+      "step": 174100
+    },
+    {
+      "epoch": 2.341460791956773,
+      "grad_norm": 0.6106776595115662,
+      "learning_rate": 0.004867533920993822,
+      "loss": 2.4046,
+      "step": 174200
+    },
+    {
+      "epoch": 2.3428049141105944,
+      "grad_norm": 0.2555699348449707,
+      "learning_rate": 0.00486736359373743,
+      "loss": 2.4151,
+      "step": 174300
+    },
+    {
+      "epoch": 2.3441490362644157,
+      "grad_norm": 0.3675636649131775,
+      "learning_rate": 0.004867193160036579,
+      "loss": 2.4072,
+      "step": 174400
+    },
+    {
+      "epoch": 2.345493158418237,
+      "grad_norm": 0.7142601609230042,
+      "learning_rate": 0.004867022619898949,
+      "loss": 2.4049,
+      "step": 174500
+    },
+    {
+      "epoch": 2.3468372805720583,
+      "grad_norm": 1.72738778591156,
+      "learning_rate": 0.004866851973332225,
+      "loss": 2.412,
+      "step": 174600
+    },
+    {
+      "epoch": 2.3481814027258796,
+      "grad_norm": 0.45770561695098877,
+      "learning_rate": 0.004866681220344099,
+      "loss": 2.416,
+      "step": 174700
+    },
+    {
+      "epoch": 2.349525524879701,
+      "grad_norm": 0.788829505443573,
+      "learning_rate": 0.004866510360942256,
+      "loss": 2.4055,
+      "step": 174800
+    },
+    {
+      "epoch": 2.350869647033522,
+      "grad_norm": 0.2863014042377472,
+      "learning_rate": 0.004866339395134404,
+      "loss": 2.4019,
+      "step": 174900
+    },
+    {
+      "epoch": 2.352213769187344,
+      "grad_norm": 0.27914437651634216,
+      "learning_rate": 0.004866168322928247,
+      "loss": 2.4036,
+      "step": 175000
+    },
+    {
+      "epoch": 2.352213769187344,
+      "eval_MaskedAccuracy": 0.4499790247285866,
+      "eval_loss": 2.7163689136505127,
+      "eval_runtime": 155.1875,
+      "eval_samples_per_second": 409.028,
+      "eval_steps_per_second": 1.598,
+      "step": 175000
+    },
+    {
+      "epoch": 2.353557891341165,
+      "grad_norm": 0.49107837677001953,
+      "learning_rate": 0.004865997144331489,
+      "loss": 2.4051,
+      "step": 175100
+    },
+    {
+      "epoch": 2.3549020134949865,
+      "grad_norm": 1.704633355140686,
+      "learning_rate": 0.004865825859351838,
+      "loss": 2.4091,
+      "step": 175200
+    },
+    {
+      "epoch": 2.356246135648808,
+      "grad_norm": 0.2347813844680786,
+      "learning_rate": 0.004865654467997019,
+      "loss": 2.4098,
+      "step": 175300
+    },
+    {
+      "epoch": 2.357590257802629,
+      "grad_norm": 0.9905913472175598,
+      "learning_rate": 0.0048654829702747555,
+      "loss": 2.4093,
+      "step": 175400
+    },
+    {
+      "epoch": 2.3589343799564504,
+      "grad_norm": 0.3170483410358429,
+      "learning_rate": 0.0048653113661927695,
+      "loss": 2.4048,
+      "step": 175500
+    },
+    {
+      "epoch": 2.3602785021102717,
+      "grad_norm": 0.8622209429740906,
+      "learning_rate": 0.004865139655758797,
+      "loss": 2.4073,
+      "step": 175600
+    },
+    {
+      "epoch": 2.361622624264093,
+      "grad_norm": 0.3935220241546631,
+      "learning_rate": 0.004864967838980571,
+      "loss": 2.4155,
+      "step": 175700
+    },
+    {
+      "epoch": 2.3629667464179143,
+      "grad_norm": 1.1118823289871216,
+      "learning_rate": 0.004864795915865833,
+      "loss": 2.4112,
+      "step": 175800
+    },
+    {
+      "epoch": 2.364310868571736,
+      "grad_norm": 0.7513082027435303,
+      "learning_rate": 0.00486462388642233,
+      "loss": 2.4152,
+      "step": 175900
+    },
+    {
+      "epoch": 2.3656549907255573,
+      "grad_norm": 1.073836326599121,
+      "learning_rate": 0.0048644517506578255,
+      "loss": 2.4023,
+      "step": 176000
+    },
+    {
+      "epoch": 2.3656549907255573,
+      "eval_MaskedAccuracy": 0.44973931526656014,
+      "eval_loss": 2.7163290977478027,
+      "eval_runtime": 156.8073,
+      "eval_samples_per_second": 404.803,
+      "eval_steps_per_second": 1.582,
+      "step": 176000
+    },
+    {
+      "epoch": 2.3669991128793786,
+      "grad_norm": 0.6424139142036438,
+      "learning_rate": 0.004864279508580062,
+      "loss": 2.3978,
+      "step": 176100
+    },
+    {
+      "epoch": 2.3683432350332,
+      "grad_norm": 0.9152630567550659,
+      "learning_rate": 0.004864107160196811,
+      "loss": 2.405,
+      "step": 176200
+    },
+    {
+      "epoch": 2.369687357187021,
+      "grad_norm": 0.44839566946029663,
+      "learning_rate": 0.004863934705515825,
+      "loss": 2.4032,
+      "step": 176300
+    },
+    {
+      "epoch": 2.3710314793408425,
+      "grad_norm": 1.236681342124939,
+      "learning_rate": 0.0048637621445448845,
+      "loss": 2.4041,
+      "step": 176400
+    },
+    {
+      "epoch": 2.3723756014946638,
+      "grad_norm": 0.62148517370224,
+      "learning_rate": 0.004863589477291759,
+      "loss": 2.4113,
+      "step": 176500
+    },
+    {
+      "epoch": 2.373719723648485,
+      "grad_norm": 0.2154432088136673,
+      "learning_rate": 0.004863416703764239,
+      "loss": 2.4113,
+      "step": 176600
+    },
+    {
+      "epoch": 2.3750638458023063,
+      "grad_norm": 0.5083281993865967,
+      "learning_rate": 0.004863243823970095,
+      "loss": 2.4058,
+      "step": 176700
+    },
+    {
+      "epoch": 2.376407967956128,
+      "grad_norm": 0.33949902653694153,
+      "learning_rate": 0.004863070837917129,
+      "loss": 2.4015,
+      "step": 176800
+    },
+    {
+      "epoch": 2.3777520901099494,
+      "grad_norm": 1.1010327339172363,
+      "learning_rate": 0.004862897745613131,
+      "loss": 2.4008,
+      "step": 176900
+    },
+    {
+      "epoch": 2.3790962122637707,
+      "grad_norm": 1.2555139064788818,
+      "learning_rate": 0.004862724547065901,
+      "loss": 2.4059,
+      "step": 177000
+    },
+    {
+      "epoch": 2.3790962122637707,
+      "eval_MaskedAccuracy": 0.4503048499742003,
+      "eval_loss": 2.716306447982788,
+      "eval_runtime": 155.5308,
+      "eval_samples_per_second": 408.125,
+      "eval_steps_per_second": 1.595,
+      "step": 177000
+    },
+    {
+      "epoch": 2.380440334417592,
+      "grad_norm": 0.5980218648910522,
+      "learning_rate": 0.0048625512422832366,
+      "loss": 2.4036,
+      "step": 177100
+    },
+    {
+      "epoch": 2.3817844565714132,
+      "grad_norm": 0.252000629901886,
+      "learning_rate": 0.004862377831272951,
+      "loss": 2.4125,
+      "step": 177200
+    },
+    {
+      "epoch": 2.3831285787252345,
+      "grad_norm": 0.6052896976470947,
+      "learning_rate": 0.004862204314042862,
+      "loss": 2.415,
+      "step": 177300
+    },
+    {
+      "epoch": 2.384472700879056,
+      "grad_norm": 0.37761661410331726,
+      "learning_rate": 0.004862030690600784,
+      "loss": 2.4009,
+      "step": 177400
+    },
+    {
+      "epoch": 2.385816823032877,
+      "grad_norm": 0.3634903132915497,
+      "learning_rate": 0.004861856960954538,
+      "loss": 2.4027,
+      "step": 177500
+    },
+    {
+      "epoch": 2.3871609451866984,
+      "grad_norm": 0.24861650168895721,
+      "learning_rate": 0.004861683125111956,
+      "loss": 2.4014,
+      "step": 177600
+    },
+    {
+      "epoch": 2.38850506734052,
+      "grad_norm": 0.893480122089386,
+      "learning_rate": 0.004861509183080868,
+      "loss": 2.4062,
+      "step": 177700
+    },
+    {
+      "epoch": 2.389849189494341,
+      "grad_norm": 0.22390282154083252,
+      "learning_rate": 0.004861335134869118,
+      "loss": 2.3922,
+      "step": 177800
+    },
+    {
+      "epoch": 2.3911933116481627,
+      "grad_norm": 0.6363741159439087,
+      "learning_rate": 0.004861160980484538,
+      "loss": 2.402,
+      "step": 177900
+    },
+    {
+      "epoch": 2.392537433801984,
+      "grad_norm": 0.918346107006073,
+      "learning_rate": 0.004860986719934979,
+      "loss": 2.3986,
+      "step": 178000
+    },
+    {
+      "epoch": 2.392537433801984,
+      "eval_MaskedAccuracy": 0.44992926294273605,
+      "eval_loss": 2.714897394180298,
+      "eval_runtime": 156.6614,
+      "eval_samples_per_second": 405.18,
+      "eval_steps_per_second": 1.583,
+      "step": 178000
+    },
+    {
+      "epoch": 2.3938815559558053,
+      "grad_norm": 0.3443443477153778,
+      "learning_rate": 0.0048608123532282955,
+      "loss": 2.4041,
+      "step": 178100
+    },
+    {
+      "epoch": 2.3952256781096266,
+      "grad_norm": 1.1353808641433716,
+      "learning_rate": 0.004860637880372347,
+      "loss": 2.4011,
+      "step": 178200
+    },
+    {
+      "epoch": 2.396569800263448,
+      "grad_norm": 0.813539445400238,
+      "learning_rate": 0.004860463301374987,
+      "loss": 2.4029,
+      "step": 178300
+    },
+    {
+      "epoch": 2.397913922417269,
+      "grad_norm": 1.0734734535217285,
+      "learning_rate": 0.004860288616244085,
+      "loss": 2.4092,
+      "step": 178400
+    },
+    {
+      "epoch": 2.3992580445710905,
+      "grad_norm": 0.3121657073497772,
+      "learning_rate": 0.004860113824987515,
+      "loss": 2.3985,
+      "step": 178500
+    },
+    {
+      "epoch": 2.400602166724912,
+      "grad_norm": 0.5276472568511963,
+      "learning_rate": 0.004859938927613148,
+      "loss": 2.4032,
+      "step": 178600
+    },
+    {
+      "epoch": 2.401946288878733,
+      "grad_norm": 0.21578922867774963,
+      "learning_rate": 0.004859763924128867,
+      "loss": 2.4002,
+      "step": 178700
+    },
+    {
+      "epoch": 2.403290411032555,
+      "grad_norm": 0.36338990926742554,
+      "learning_rate": 0.004859588814542557,
+      "loss": 2.4103,
+      "step": 178800
+    },
+    {
+      "epoch": 2.404634533186376,
+      "grad_norm": 0.7060322165489197,
+      "learning_rate": 0.004859413598862113,
+      "loss": 2.4039,
+      "step": 178900
+    },
+    {
+      "epoch": 2.4059786553401974,
+      "grad_norm": 0.3517087399959564,
+      "learning_rate": 0.004859238277095424,
+      "loss": 2.4107,
+      "step": 179000
+    },
+    {
+      "epoch": 2.4059786553401974,
+      "eval_MaskedAccuracy": 0.4501261935846993,
+      "eval_loss": 2.7151551246643066,
+      "eval_runtime": 158.7753,
+      "eval_samples_per_second": 399.785,
+      "eval_steps_per_second": 1.562,
+      "step": 179000
+    },
+    {
+      "epoch": 2.4073227774940187,
+      "grad_norm": 0.850573718547821,
+      "learning_rate": 0.004859062849250387,
+      "loss": 2.4033,
+      "step": 179100
+    },
+    {
+      "epoch": 2.40866689964784,
+      "grad_norm": 0.25686702132225037,
+      "learning_rate": 0.004858887315334917,
+      "loss": 2.4033,
+      "step": 179200
+    },
+    {
+      "epoch": 2.4100110218016613,
+      "grad_norm": 0.580507755279541,
+      "learning_rate": 0.0048587116753569136,
+      "loss": 2.4075,
+      "step": 179300
+    },
+    {
+      "epoch": 2.4113551439554826,
+      "grad_norm": 0.21325333416461945,
+      "learning_rate": 0.004858535929324301,
+      "loss": 2.4066,
+      "step": 179400
+    },
+    {
+      "epoch": 2.412699266109304,
+      "grad_norm": 1.397865891456604,
+      "learning_rate": 0.004858360077244987,
+      "loss": 2.4069,
+      "step": 179500
+    },
+    {
+      "epoch": 2.414043388263125,
+      "grad_norm": 0.7146439552307129,
+      "learning_rate": 0.004858184119126909,
+      "loss": 2.4055,
+      "step": 179600
+    },
+    {
+      "epoch": 2.415387510416947,
+      "grad_norm": 0.6387777328491211,
+      "learning_rate": 0.004858008054977973,
+      "loss": 2.3984,
+      "step": 179700
+    },
+    {
+      "epoch": 2.416731632570768,
+      "grad_norm": 0.6029757261276245,
+      "learning_rate": 0.004857831884806135,
+      "loss": 2.4017,
+      "step": 179800
+    },
+    {
+      "epoch": 2.4180757547245895,
+      "grad_norm": 0.3883086144924164,
+      "learning_rate": 0.004857655608619315,
+      "loss": 2.4009,
+      "step": 179900
+    },
+    {
+      "epoch": 2.4194198768784108,
+      "grad_norm": 0.24268724024295807,
+      "learning_rate": 0.004857479226425462,
+      "loss": 2.4127,
+      "step": 180000
+    },
+    {
+      "epoch": 2.4194198768784108,
+      "eval_MaskedAccuracy": 0.4500598610681297,
+      "eval_loss": 2.7149128913879395,
+      "eval_runtime": 160.6745,
+      "eval_samples_per_second": 395.06,
+      "eval_steps_per_second": 1.543,
+      "step": 180000
+    },
+    {
+      "epoch": 2.420763999032232,
+      "grad_norm": 0.2926810085773468,
+      "learning_rate": 0.004857302738232527,
+      "loss": 2.4048,
+      "step": 180100
+    },
+    {
+      "epoch": 2.4221081211860533,
+      "grad_norm": 0.5268754363059998,
+      "learning_rate": 0.004857126144048464,
+      "loss": 2.3981,
+      "step": 180200
+    },
+    {
+      "epoch": 2.4234522433398746,
+      "grad_norm": 0.27951580286026,
+      "learning_rate": 0.004856949443881227,
+      "loss": 2.3983,
+      "step": 180300
+    },
+    {
+      "epoch": 2.424796365493696,
+      "grad_norm": 0.5589247345924377,
+      "learning_rate": 0.004856772637738778,
+      "loss": 2.405,
+      "step": 180400
+    },
+    {
+      "epoch": 2.426140487647517,
+      "grad_norm": 0.2754684090614319,
+      "learning_rate": 0.004856595725629084,
+      "loss": 2.4081,
+      "step": 180500
+    },
+    {
+      "epoch": 2.427484609801339,
+      "grad_norm": 1.014289379119873,
+      "learning_rate": 0.004856418707560116,
+      "loss": 2.4028,
+      "step": 180600
+    },
+    {
+      "epoch": 2.4288287319551602,
+      "grad_norm": 1.224565029144287,
+      "learning_rate": 0.004856241583539849,
+      "loss": 2.4041,
+      "step": 180700
+    },
+    {
+      "epoch": 2.4301728541089815,
+      "grad_norm": 0.2553938329219818,
+      "learning_rate": 0.004856064353576271,
+      "loss": 2.4039,
+      "step": 180800
+    },
+    {
+      "epoch": 2.431516976262803,
+      "grad_norm": 0.36857032775878906,
+      "learning_rate": 0.004855887017677354,
+      "loss": 2.4021,
+      "step": 180900
+    },
+    {
+      "epoch": 2.432861098416624,
+      "grad_norm": 0.3493588864803314,
+      "learning_rate": 0.004855709575851092,
+      "loss": 2.4009,
+      "step": 181000
+    },
+    {
+      "epoch": 2.432861098416624,
+      "eval_MaskedAccuracy": 0.4509061473369637,
+      "eval_loss": 2.7095935344696045,
+      "eval_runtime": 159.2973,
+      "eval_samples_per_second": 398.475,
+      "eval_steps_per_second": 1.557,
+      "step": 181000
+    },
+    {
+      "epoch": 2.4342052205704454,
+      "grad_norm": 0.6775467395782471,
+      "learning_rate": 0.004855532028105485,
+      "loss": 2.393,
+      "step": 181100
+    },
+    {
+      "epoch": 2.4355493427242667,
+      "grad_norm": 1.7895002365112305,
+      "learning_rate": 0.004855354374448533,
+      "loss": 2.3995,
+      "step": 181200
+    },
+    {
+      "epoch": 2.436893464878088,
+      "grad_norm": 0.2109525501728058,
+      "learning_rate": 0.004855176614888239,
+      "loss": 2.4031,
+      "step": 181300
+    },
+    {
+      "epoch": 2.4382375870319093,
+      "grad_norm": 0.24894700944423676,
+      "learning_rate": 0.004854998749432621,
+      "loss": 2.4049,
+      "step": 181400
+    },
+    {
+      "epoch": 2.439581709185731,
+      "grad_norm": 0.3863070011138916,
+      "learning_rate": 0.004854820778089685,
+      "loss": 2.3983,
+      "step": 181500
+    },
+    {
+      "epoch": 2.4409258313395523,
+      "grad_norm": 0.4976949095726013,
+      "learning_rate": 0.0048546427008674484,
+      "loss": 2.4076,
+      "step": 181600
+    },
+    {
+      "epoch": 2.4422699534933736,
+      "grad_norm": 1.53170907497406,
+      "learning_rate": 0.00485446451777394,
+      "loss": 2.4009,
+      "step": 181700
+    },
+    {
+      "epoch": 2.443614075647195,
+      "grad_norm": 0.8086992502212524,
+      "learning_rate": 0.0048542862288171885,
+      "loss": 2.4014,
+      "step": 181800
+    },
+    {
+      "epoch": 2.444958197801016,
+      "grad_norm": 0.8978860378265381,
+      "learning_rate": 0.00485410783400522,
+      "loss": 2.4002,
+      "step": 181900
+    },
+    {
+      "epoch": 2.4463023199548375,
+      "grad_norm": 0.40652504563331604,
+      "learning_rate": 0.004853929333346081,
+      "loss": 2.3978,
+      "step": 182000
+    },
+    {
+      "epoch": 2.4463023199548375,
+      "eval_MaskedAccuracy": 0.4505315380239766,
+      "eval_loss": 2.7120492458343506,
+      "eval_runtime": 158.1071,
+      "eval_samples_per_second": 401.475,
+      "eval_steps_per_second": 1.569,
+      "step": 182000
+    },
+    {
+      "epoch": 2.447646442108659,
+      "grad_norm": 0.544252336025238,
+      "learning_rate": 0.004853750726847811,
+      "loss": 2.3905,
+      "step": 182100
+    },
+    {
+      "epoch": 2.44899056426248,
+      "grad_norm": 1.0650444030761719,
+      "learning_rate": 0.0048535720145184615,
+      "loss": 2.4042,
+      "step": 182200
+    },
+    {
+      "epoch": 2.4503346864163014,
+      "grad_norm": 0.23677361011505127,
+      "learning_rate": 0.004853393196366076,
+      "loss": 2.3987,
+      "step": 182300
+    },
+    {
+      "epoch": 2.4516788085701227,
+      "grad_norm": 0.5979803204536438,
+      "learning_rate": 0.004853214272398713,
+      "loss": 2.4045,
+      "step": 182400
+    },
+    {
+      "epoch": 2.453022930723944,
+      "grad_norm": 0.37009915709495544,
+      "learning_rate": 0.0048530352426244434,
+      "loss": 2.4025,
+      "step": 182500
+    },
+    {
+      "epoch": 2.4543670528777657,
+      "grad_norm": 0.9107745289802551,
+      "learning_rate": 0.00485285610705133,
+      "loss": 2.4047,
+      "step": 182600
+    },
+    {
+      "epoch": 2.455711175031587,
+      "grad_norm": 0.6059080958366394,
+      "learning_rate": 0.004852676865687442,
+      "loss": 2.4036,
+      "step": 182700
+    },
+    {
+      "epoch": 2.4570552971854083,
+      "grad_norm": 1.2172014713287354,
+      "learning_rate": 0.004852497518540862,
+      "loss": 2.4039,
+      "step": 182800
+    },
+    {
+      "epoch": 2.4583994193392296,
+      "grad_norm": 0.27083879709243774,
+      "learning_rate": 0.004852318065619665,
+      "loss": 2.4028,
+      "step": 182900
+    },
+    {
+      "epoch": 2.459743541493051,
+      "grad_norm": 0.2236872911453247,
+      "learning_rate": 0.0048521385069319346,
+      "loss": 2.3937,
+      "step": 183000
+    },
+    {
+      "epoch": 2.459743541493051,
+      "eval_MaskedAccuracy": 0.4513215347110807,
+      "eval_loss": 2.7078845500946045,
+      "eval_runtime": 156.7319,
+      "eval_samples_per_second": 404.997,
+      "eval_steps_per_second": 1.582,
+      "step": 183000
+    },
+    {
+      "epoch": 2.461087663646872,
+      "grad_norm": 0.7614181637763977,
+      "learning_rate": 0.004851958842485767,
+      "loss": 2.3939,
+      "step": 183100
+    },
+    {
+      "epoch": 2.4624317858006934,
+      "grad_norm": 0.20438380539417267,
+      "learning_rate": 0.004851779072289257,
+      "loss": 2.4072,
+      "step": 183200
+    },
+    {
+      "epoch": 2.4637759079545147,
+      "grad_norm": 0.995496392250061,
+      "learning_rate": 0.004851599196350496,
+      "loss": 2.3985,
+      "step": 183300
+    },
+    {
+      "epoch": 2.465120030108336,
+      "grad_norm": 0.5554972290992737,
+      "learning_rate": 0.004851419214677599,
+      "loss": 2.3987,
+      "step": 183400
+    },
+    {
+      "epoch": 2.4664641522621578,
+      "grad_norm": 0.7563191056251526,
+      "learning_rate": 0.004851239127278679,
+      "loss": 2.4124,
+      "step": 183500
+    },
+    {
+      "epoch": 2.467808274415979,
+      "grad_norm": 0.9702069163322449,
+      "learning_rate": 0.004851058934161847,
+      "loss": 2.4062,
+      "step": 183600
+    },
+    {
+      "epoch": 2.4691523965698003,
+      "grad_norm": 0.3375505208969116,
+      "learning_rate": 0.004850878635335215,
+      "loss": 2.3993,
+      "step": 183700
+    },
+    {
+      "epoch": 2.4704965187236216,
+      "grad_norm": 0.3639039993286133,
+      "learning_rate": 0.004850698230806917,
+      "loss": 2.3888,
+      "step": 183800
+    },
+    {
+      "epoch": 2.471840640877443,
+      "grad_norm": 1.2747180461883545,
+      "learning_rate": 0.004850517720585071,
+      "loss": 2.3952,
+      "step": 183900
+    },
+    {
+      "epoch": 2.473184763031264,
+      "grad_norm": 0.2078840732574463,
+      "learning_rate": 0.004850337104677819,
+      "loss": 2.3996,
+      "step": 184000
+    },
+    {
+      "epoch": 2.473184763031264,
+      "eval_MaskedAccuracy": 0.4516848487899179,
+      "eval_loss": 2.7041962146759033,
+      "eval_runtime": 156.0607,
+      "eval_samples_per_second": 406.739,
+      "eval_steps_per_second": 1.589,
+      "step": 184000
+    },
+    {
+      "epoch": 2.4745288851850855,
+      "grad_norm": 0.7602828145027161,
+      "learning_rate": 0.004850156383093299,
+      "loss": 2.3861,
+      "step": 184100
+    },
+    {
+      "epoch": 2.475873007338907,
+      "grad_norm": 0.3689172863960266,
+      "learning_rate": 0.004849975555839646,
+      "loss": 2.3889,
+      "step": 184200
+    },
+    {
+      "epoch": 2.477217129492728,
+      "grad_norm": 0.26970091462135315,
+      "learning_rate": 0.004849794622925014,
+      "loss": 2.4053,
+      "step": 184300
+    },
+    {
+      "epoch": 2.47856125164655,
+      "grad_norm": 0.237064927816391,
+      "learning_rate": 0.004849613584357556,
+      "loss": 2.4008,
+      "step": 184400
+    },
+    {
+      "epoch": 2.479905373800371,
+      "grad_norm": 1.7367196083068848,
+      "learning_rate": 0.004849432440145427,
+      "loss": 2.4008,
+      "step": 184500
+    },
+    {
+      "epoch": 2.4812494959541924,
+      "grad_norm": 0.3666498064994812,
+      "learning_rate": 0.004849251190296789,
+      "loss": 2.4011,
+      "step": 184600
+    },
+    {
+      "epoch": 2.4825936181080137,
+      "grad_norm": 0.4193936288356781,
+      "learning_rate": 0.004849069834819819,
+      "loss": 2.4085,
+      "step": 184700
+    },
+    {
+      "epoch": 2.483937740261835,
+      "grad_norm": 0.3407999277114868,
+      "learning_rate": 0.004848888373722671,
+      "loss": 2.4006,
+      "step": 184800
+    },
+    {
+      "epoch": 2.4852818624156563,
+      "grad_norm": 0.3108844459056854,
+      "learning_rate": 0.0048487068070135345,
+      "loss": 2.4081,
+      "step": 184900
+    },
+    {
+      "epoch": 2.4866259845694776,
+      "grad_norm": 0.3092864453792572,
+      "learning_rate": 0.004848525134700581,
+      "loss": 2.4082,
+      "step": 185000
+    },
+    {
+      "epoch": 2.4866259845694776,
+      "eval_MaskedAccuracy": 0.4504291166073478,
+      "eval_loss": 2.712066411972046,
+      "eval_runtime": 156.1824,
+      "eval_samples_per_second": 406.422,
+      "eval_steps_per_second": 1.588,
+      "step": 185000
+    },
+    {
+      "epoch": 2.487970106723299,
+      "grad_norm": 0.25049683451652527,
+      "learning_rate": 0.004848343356792004,
+      "loss": 2.4032,
+      "step": 185100
+    },
+    {
+      "epoch": 2.48931422887712,
+      "grad_norm": 0.9619961977005005,
+      "learning_rate": 0.004848161473295994,
+      "loss": 2.4099,
+      "step": 185200
+    },
+    {
+      "epoch": 2.490658351030942,
+      "grad_norm": 1.3118278980255127,
+      "learning_rate": 0.004847979484220742,
+      "loss": 2.3951,
+      "step": 185300
+    },
+    {
+      "epoch": 2.492002473184763,
+      "grad_norm": 0.890388011932373,
+      "learning_rate": 0.004847797389574453,
+      "loss": 2.404,
+      "step": 185400
+    },
+    {
+      "epoch": 2.4933465953385845,
+      "grad_norm": 0.9836694002151489,
+      "learning_rate": 0.004847615189365324,
+      "loss": 2.4056,
+      "step": 185500
+    },
+    {
+      "epoch": 2.494690717492406,
+      "grad_norm": 1.4075289964675903,
+      "learning_rate": 0.004847432883601571,
+      "loss": 2.3974,
+      "step": 185600
+    },
+    {
+      "epoch": 2.496034839646227,
+      "grad_norm": 1.5059231519699097,
+      "learning_rate": 0.0048472504722914134,
+      "loss": 2.399,
+      "step": 185700
+    },
+    {
+      "epoch": 2.4973789618000484,
+      "grad_norm": 0.42734503746032715,
+      "learning_rate": 0.004847067955443057,
+      "loss": 2.4063,
+      "step": 185800
+    },
+    {
+      "epoch": 2.4987230839538697,
+      "grad_norm": 1.4821892976760864,
+      "learning_rate": 0.004846885333064743,
+      "loss": 2.3914,
+      "step": 185900
+    },
+    {
+      "epoch": 2.500067206107691,
+      "grad_norm": 1.791476845741272,
+      "learning_rate": 0.004846702605164673,
+      "loss": 2.3974,
+      "step": 186000
+    },
+    {
+      "epoch": 2.500067206107691,
+      "eval_MaskedAccuracy": 0.4501642596224747,
+      "eval_loss": 2.7116963863372803,
+      "eval_runtime": 159.5858,
+      "eval_samples_per_second": 397.755,
+      "eval_steps_per_second": 1.554,
+      "step": 186000
+    },
+    {
+      "epoch": 2.5014113282615122,
+      "grad_norm": 0.5543153285980225,
+      "learning_rate": 0.004846519771751113,
+      "loss": 2.4015,
+      "step": 186100
+    },
+    {
+      "epoch": 2.502755450415334,
+      "grad_norm": 1.6506179571151733,
+      "learning_rate": 0.004846336832832279,
+      "loss": 2.3985,
+      "step": 186200
+    },
+    {
+      "epoch": 2.504099572569155,
+      "grad_norm": 0.9173561334609985,
+      "learning_rate": 0.004846153788416425,
+      "loss": 2.3929,
+      "step": 186300
+    },
+    {
+      "epoch": 2.5054436947229766,
+      "grad_norm": 0.9084792733192444,
+      "learning_rate": 0.004845970638511788,
+      "loss": 2.3981,
+      "step": 186400
+    },
+    {
+      "epoch": 2.506787816876798,
+      "grad_norm": 0.4403022825717926,
+      "learning_rate": 0.004845787383126632,
+      "loss": 2.3893,
+      "step": 186500
+    },
+    {
+      "epoch": 2.508131939030619,
+      "grad_norm": 0.3516016900539398,
+      "learning_rate": 0.0048456040222692055,
+      "loss": 2.3998,
+      "step": 186600
+    },
+    {
+      "epoch": 2.5094760611844404,
+      "grad_norm": 0.5586411952972412,
+      "learning_rate": 0.0048454205559477735,
+      "loss": 2.4068,
+      "step": 186700
+    },
+    {
+      "epoch": 2.5108201833382617,
+      "grad_norm": 0.2955363094806671,
+      "learning_rate": 0.004845236984170605,
+      "loss": 2.4005,
+      "step": 186800
+    },
+    {
+      "epoch": 2.512164305492083,
+      "grad_norm": 0.22038711607456207,
+      "learning_rate": 0.004845053306945971,
+      "loss": 2.4009,
+      "step": 186900
+    },
+    {
+      "epoch": 2.5135084276459043,
+      "grad_norm": 0.7231718301773071,
+      "learning_rate": 0.004844869524282139,
+      "loss": 2.3994,
+      "step": 187000
+    },
+    {
+      "epoch": 2.5135084276459043,
+      "eval_MaskedAccuracy": 0.45075855288065936,
+      "eval_loss": 2.709667682647705,
+      "eval_runtime": 159.519,
+      "eval_samples_per_second": 397.921,
+      "eval_steps_per_second": 1.555,
+      "step": 187000
+    },
+    {
+      "epoch": 2.514852549799726,
+      "grad_norm": 1.2093353271484375,
+      "learning_rate": 0.004844685636187404,
+      "loss": 2.4022,
+      "step": 187100
+    },
+    {
+      "epoch": 2.516196671953547,
+      "grad_norm": 0.46012699604034424,
+      "learning_rate": 0.004844501642670045,
+      "loss": 2.4025,
+      "step": 187200
+    },
+    {
+      "epoch": 2.5175407941073686,
+      "grad_norm": 0.5770087838172913,
+      "learning_rate": 0.00484431754373835,
+      "loss": 2.3926,
+      "step": 187300
+    },
+    {
+      "epoch": 2.51888491626119,
+      "grad_norm": 0.8018302321434021,
+      "learning_rate": 0.004844133339400616,
+      "loss": 2.403,
+      "step": 187400
+    },
+    {
+      "epoch": 2.5202290384150112,
+      "grad_norm": 1.2442452907562256,
+      "learning_rate": 0.004843949029665142,
+      "loss": 2.402,
+      "step": 187500
+    },
+    {
+      "epoch": 2.5215731605688325,
+      "grad_norm": 0.22802333533763885,
+      "learning_rate": 0.0048437646145402385,
+      "loss": 2.4039,
+      "step": 187600
+    },
+    {
+      "epoch": 2.522917282722654,
+      "grad_norm": 0.5682822465896606,
+      "learning_rate": 0.004843580094034207,
+      "loss": 2.3969,
+      "step": 187700
+    },
+    {
+      "epoch": 2.524261404876475,
+      "grad_norm": 0.9464827179908752,
+      "learning_rate": 0.004843395468155368,
+      "loss": 2.3924,
+      "step": 187800
+    },
+    {
+      "epoch": 2.5256055270302964,
+      "grad_norm": 0.5253119468688965,
+      "learning_rate": 0.00484321073691204,
+      "loss": 2.3929,
+      "step": 187900
+    },
+    {
+      "epoch": 2.526949649184118,
+      "grad_norm": 0.6649153828620911,
+      "learning_rate": 0.004843025900312542,
+      "loss": 2.3994,
+      "step": 188000
+    },
+    {
+      "epoch": 2.526949649184118,
+      "eval_MaskedAccuracy": 0.45129980447121243,
+      "eval_loss": 2.707331657409668,
+      "eval_runtime": 162.5712,
+      "eval_samples_per_second": 390.45,
+      "eval_steps_per_second": 1.525,
+      "step": 188000
+    },
+    {
+      "epoch": 2.528293771337939,
+      "grad_norm": 0.21869996190071106,
+      "learning_rate": 0.004842840958365209,
+      "loss": 2.3981,
+      "step": 188100
+    },
+    {
+      "epoch": 2.5296378934917607,
+      "grad_norm": 0.394054651260376,
+      "learning_rate": 0.004842655911078364,
+      "loss": 2.3985,
+      "step": 188200
+    },
+    {
+      "epoch": 2.530982015645582,
+      "grad_norm": 0.5307244658470154,
+      "learning_rate": 0.004842470758460362,
+      "loss": 2.3977,
+      "step": 188300
+    },
+    {
+      "epoch": 2.5323261377994033,
+      "grad_norm": 0.19963762164115906,
+      "learning_rate": 0.004842285500519532,
+      "loss": 2.4024,
+      "step": 188400
+    },
+    {
+      "epoch": 2.5336702599532246,
+      "grad_norm": 0.7292469143867493,
+      "learning_rate": 0.004842100137264226,
+      "loss": 2.3958,
+      "step": 188500
+    },
+    {
+      "epoch": 2.535014382107046,
+      "grad_norm": 2.722954034805298,
+      "learning_rate": 0.004841914668702794,
+      "loss": 2.3943,
+      "step": 188600
+    },
+    {
+      "epoch": 2.536358504260867,
+      "grad_norm": 0.30424684286117554,
+      "learning_rate": 0.004841729094843593,
+      "loss": 2.3951,
+      "step": 188700
+    },
+    {
+      "epoch": 2.5377026264146885,
+      "grad_norm": 0.6298586130142212,
+      "learning_rate": 0.004841543415694982,
+      "loss": 2.3964,
+      "step": 188800
+    },
+    {
+      "epoch": 2.5390467485685098,
+      "grad_norm": 0.30196303129196167,
+      "learning_rate": 0.004841357631265328,
+      "loss": 2.3894,
+      "step": 188900
+    },
+    {
+      "epoch": 2.540390870722331,
+      "grad_norm": 0.2881263792514801,
+      "learning_rate": 0.004841171741563008,
+      "loss": 2.4015,
+      "step": 189000
+    },
+    {
+      "epoch": 2.540390870722331,
+      "eval_MaskedAccuracy": 0.4518744871539106,
+      "eval_loss": 2.7033305168151855,
+      "eval_runtime": 153.1061,
+      "eval_samples_per_second": 414.588,
+      "eval_steps_per_second": 1.62,
+      "step": 189000
+    },
+    {
+      "epoch": 2.541734992876153,
+      "grad_norm": 0.23784345388412476,
+      "learning_rate": 0.004840985746596394,
+      "loss": 2.3971,
+      "step": 189100
+    },
+    {
+      "epoch": 2.543079115029974,
+      "grad_norm": 0.2152082920074463,
+      "learning_rate": 0.004840799646373867,
+      "loss": 2.3981,
+      "step": 189200
+    },
+    {
+      "epoch": 2.5444232371837954,
+      "grad_norm": 0.2263517528772354,
+      "learning_rate": 0.004840613440903815,
+      "loss": 2.3943,
+      "step": 189300
+    },
+    {
+      "epoch": 2.5457673593376167,
+      "grad_norm": 0.6147149801254272,
+      "learning_rate": 0.00484042713019462,
+      "loss": 2.3895,
+      "step": 189400
+    },
+    {
+      "epoch": 2.547111481491438,
+      "grad_norm": 0.3678668439388275,
+      "learning_rate": 0.0048402407142546875,
+      "loss": 2.3991,
+      "step": 189500
+    },
+    {
+      "epoch": 2.5484556036452592,
+      "grad_norm": 0.323637455701828,
+      "learning_rate": 0.0048400541930924135,
+      "loss": 2.3957,
+      "step": 189600
+    },
+    {
+      "epoch": 2.5497997257990805,
+      "grad_norm": 1.3563669919967651,
+      "learning_rate": 0.004839867566716193,
+      "loss": 2.3984,
+      "step": 189700
+    },
+    {
+      "epoch": 2.551143847952902,
+      "grad_norm": 1.2862578630447388,
+      "learning_rate": 0.004839680835134451,
+      "loss": 2.4005,
+      "step": 189800
+    },
+    {
+      "epoch": 2.552487970106723,
+      "grad_norm": 0.30557116866111755,
+      "learning_rate": 0.00483949399835559,
+      "loss": 2.4025,
+      "step": 189900
+    },
+    {
+      "epoch": 2.553832092260545,
+      "grad_norm": 0.46706828474998474,
+      "learning_rate": 0.004839307056388037,
+      "loss": 2.3974,
+      "step": 190000
+    },
+    {
+      "epoch": 2.553832092260545,
+      "eval_MaskedAccuracy": 0.4516189117545165,
+      "eval_loss": 2.704475164413452,
+      "eval_runtime": 155.082,
+      "eval_samples_per_second": 409.306,
+      "eval_steps_per_second": 1.599,
+      "step": 190000
+    },
+    {
+      "epoch": 2.5551762144143657,
+      "grad_norm": 0.8386965990066528,
+      "learning_rate": 0.0048391200092402055,
+      "loss": 2.3969,
+      "step": 190100
+    },
+    {
+      "epoch": 2.5565203365681874,
+      "grad_norm": 0.35076892375946045,
+      "learning_rate": 0.004838932856920533,
+      "loss": 2.4024,
+      "step": 190200
+    },
+    {
+      "epoch": 2.5578644587220087,
+      "grad_norm": 0.2803960144519806,
+      "learning_rate": 0.0048387455994374435,
+      "loss": 2.389,
+      "step": 190300
+    },
+    {
+      "epoch": 2.55920858087583,
+      "grad_norm": 0.4562201499938965,
+      "learning_rate": 0.004838558236799381,
+      "loss": 2.3991,
+      "step": 190400
+    },
+    {
+      "epoch": 2.5605527030296513,
+      "grad_norm": 1.0036048889160156,
+      "learning_rate": 0.004838370769014779,
+      "loss": 2.3928,
+      "step": 190500
+    },
+    {
+      "epoch": 2.5618968251834726,
+      "grad_norm": 0.2660607099533081,
+      "learning_rate": 0.004838183196092097,
+      "loss": 2.4001,
+      "step": 190600
+    },
+    {
+      "epoch": 2.563240947337294,
+      "grad_norm": 0.6841756701469421,
+      "learning_rate": 0.004837995518039784,
+      "loss": 2.4028,
+      "step": 190700
+    },
+    {
+      "epoch": 2.564585069491115,
+      "grad_norm": 1.2373573780059814,
+      "learning_rate": 0.0048378077348662845,
+      "loss": 2.4022,
+      "step": 190800
+    },
+    {
+      "epoch": 2.565929191644937,
+      "grad_norm": 0.6526246070861816,
+      "learning_rate": 0.004837619846580077,
+      "loss": 2.3894,
+      "step": 190900
+    },
+    {
+      "epoch": 2.567273313798758,
+      "grad_norm": 0.2829650938510895,
+      "learning_rate": 0.004837431853189622,
+      "loss": 2.4013,
+      "step": 191000
+    },
+    {
+      "epoch": 2.567273313798758,
+      "eval_MaskedAccuracy": 0.4523377197669412,
+      "eval_loss": 2.701768636703491,
+      "eval_runtime": 154.0538,
+      "eval_samples_per_second": 412.038,
+      "eval_steps_per_second": 1.61,
+      "step": 191000
+    },
+    {
+      "epoch": 2.5686174359525795,
+      "grad_norm": 0.32170721888542175,
+      "learning_rate": 0.004837243754703377,
+      "loss": 2.3919,
+      "step": 191100
+    },
+    {
+      "epoch": 2.569961558106401,
+      "grad_norm": 0.49978765845298767,
+      "learning_rate": 0.004837055551129837,
+      "loss": 2.398,
+      "step": 191200
+    },
+    {
+      "epoch": 2.571305680260222,
+      "grad_norm": 0.7216925024986267,
+      "learning_rate": 0.004836867242477465,
+      "loss": 2.389,
+      "step": 191300
+    },
+    {
+      "epoch": 2.5726498024140434,
+      "grad_norm": 0.381653368473053,
+      "learning_rate": 0.004836678828754757,
+      "loss": 2.4022,
+      "step": 191400
+    },
+    {
+      "epoch": 2.5739939245678647,
+      "grad_norm": 0.5341126322746277,
+      "learning_rate": 0.004836490309970204,
+      "loss": 2.4023,
+      "step": 191500
+    },
+    {
+      "epoch": 2.575338046721686,
+      "grad_norm": 1.2294304370880127,
+      "learning_rate": 0.00483630168613229,
+      "loss": 2.3937,
+      "step": 191600
+    },
+    {
+      "epoch": 2.5766821688755073,
+      "grad_norm": 0.9563962817192078,
+      "learning_rate": 0.004836112957249523,
+      "loss": 2.3978,
+      "step": 191700
+    },
+    {
+      "epoch": 2.578026291029329,
+      "grad_norm": 0.6789734959602356,
+      "learning_rate": 0.0048359241233304024,
+      "loss": 2.3914,
+      "step": 191800
+    },
+    {
+      "epoch": 2.57937041318315,
+      "grad_norm": 2.394522190093994,
+      "learning_rate": 0.004835735184383436,
+      "loss": 2.397,
+      "step": 191900
+    },
+    {
+      "epoch": 2.5807145353369716,
+      "grad_norm": 0.9763964414596558,
+      "learning_rate": 0.0048355461404171445,
+      "loss": 2.3978,
+      "step": 192000
+    },
+    {
+      "epoch": 2.5807145353369716,
+      "eval_MaskedAccuracy": 0.45172072887484477,
+      "eval_loss": 2.704576253890991,
+      "eval_runtime": 152.9163,
+      "eval_samples_per_second": 415.103,
+      "eval_steps_per_second": 1.622,
+      "step": 192000
+    },
+    {
+      "epoch": 2.582058657490793,
+      "grad_norm": 1.0108835697174072,
+      "learning_rate": 0.004835356991440036,
+      "loss": 2.3952,
+      "step": 192100
+    },
+    {
+      "epoch": 2.583402779644614,
+      "grad_norm": 0.4426333010196686,
+      "learning_rate": 0.004835167737460643,
+      "loss": 2.3953,
+      "step": 192200
+    },
+    {
+      "epoch": 2.5847469017984355,
+      "grad_norm": 0.35795265436172485,
+      "learning_rate": 0.004834978378487477,
+      "loss": 2.3937,
+      "step": 192300
+    },
+    {
+      "epoch": 2.5860910239522568,
+      "grad_norm": 0.5657187104225159,
+      "learning_rate": 0.004834788914529085,
+      "loss": 2.3972,
+      "step": 192400
+    },
+    {
+      "epoch": 2.587435146106078,
+      "grad_norm": 1.4484549760818481,
+      "learning_rate": 0.004834599345594001,
+      "loss": 2.4053,
+      "step": 192500
+    },
+    {
+      "epoch": 2.5887792682598993,
+      "grad_norm": 0.5882700085639954,
+      "learning_rate": 0.004834409671690765,
+      "loss": 2.3941,
+      "step": 192600
+    },
+    {
+      "epoch": 2.5901233904137206,
+      "grad_norm": 0.2406308501958847,
+      "learning_rate": 0.004834219892827928,
+      "loss": 2.3877,
+      "step": 192700
+    },
+    {
+      "epoch": 2.591467512567542,
+      "grad_norm": 0.6876055002212524,
+      "learning_rate": 0.004834030009014036,
+      "loss": 2.3869,
+      "step": 192800
+    },
+    {
+      "epoch": 2.5928116347213637,
+      "grad_norm": 2.6464781761169434,
+      "learning_rate": 0.0048338400202576435,
+      "loss": 2.3974,
+      "step": 192900
+    },
+    {
+      "epoch": 2.594155756875185,
+      "grad_norm": 0.5188945531845093,
+      "learning_rate": 0.004833649926567314,
+      "loss": 2.3896,
+      "step": 193000
+    },
+    {
+      "epoch": 2.594155756875185,
+      "eval_MaskedAccuracy": 0.4505340990281368,
+      "eval_loss": 2.70825457572937,
+      "eval_runtime": 153.498,
+      "eval_samples_per_second": 413.53,
+      "eval_steps_per_second": 1.616,
+      "step": 193000
+    },
+    {
+      "epoch": 2.5954998790290063,
+      "grad_norm": 0.28811684250831604,
+      "learning_rate": 0.004833459727951611,
+      "loss": 2.3915,
+      "step": 193100
+    },
+    {
+      "epoch": 2.5968440011828275,
+      "grad_norm": 0.47558093070983887,
+      "learning_rate": 0.004833269424419109,
+      "loss": 2.39,
+      "step": 193200
+    },
+    {
+      "epoch": 2.598188123336649,
+      "grad_norm": 0.34942418336868286,
+      "learning_rate": 0.004833079015978379,
+      "loss": 2.3917,
+      "step": 193300
+    },
+    {
+      "epoch": 2.59953224549047,
+      "grad_norm": 0.28590691089630127,
+      "learning_rate": 0.004832888502638003,
+      "loss": 2.3953,
+      "step": 193400
+    },
+    {
+      "epoch": 2.6008763676442914,
+      "grad_norm": 0.4065092206001282,
+      "learning_rate": 0.004832697884406561,
+      "loss": 2.3919,
+      "step": 193500
+    },
+    {
+      "epoch": 2.6022204897981127,
+      "grad_norm": 0.7588686943054199,
+      "learning_rate": 0.004832507161292647,
+      "loss": 2.3976,
+      "step": 193600
+    },
+    {
+      "epoch": 2.603564611951934,
+      "grad_norm": 0.8062926530838013,
+      "learning_rate": 0.004832316333304851,
+      "loss": 2.3993,
+      "step": 193700
+    },
+    {
+      "epoch": 2.6049087341057557,
+      "grad_norm": 0.3884819447994232,
+      "learning_rate": 0.004832125400451772,
+      "loss": 2.3946,
+      "step": 193800
+    },
+    {
+      "epoch": 2.606252856259577,
+      "grad_norm": 0.950999915599823,
+      "learning_rate": 0.004831934362742015,
+      "loss": 2.399,
+      "step": 193900
+    },
+    {
+      "epoch": 2.6075969784133983,
+      "grad_norm": 0.2583158016204834,
+      "learning_rate": 0.004831743220184189,
+      "loss": 2.3895,
+      "step": 194000
+    },
+    {
+      "epoch": 2.6075969784133983,
+      "eval_MaskedAccuracy": 0.45153046662578544,
+      "eval_loss": 2.70489764213562,
+      "eval_runtime": 156.1126,
+      "eval_samples_per_second": 406.604,
+      "eval_steps_per_second": 1.589,
+      "step": 194000
+    },
+    {
+      "epoch": 2.6089411005672196,
+      "grad_norm": 0.20159654319286346,
+      "learning_rate": 0.004831551972786904,
+      "loss": 2.4009,
+      "step": 194100
+    },
+    {
+      "epoch": 2.610285222721041,
+      "grad_norm": 0.3360646963119507,
+      "learning_rate": 0.004831360620558774,
+      "loss": 2.4018,
+      "step": 194200
+    },
+    {
+      "epoch": 2.611629344874862,
+      "grad_norm": 1.5870589017868042,
+      "learning_rate": 0.004831169163508427,
+      "loss": 2.3897,
+      "step": 194300
+    },
+    {
+      "epoch": 2.6129734670286835,
+      "grad_norm": 0.9621824622154236,
+      "learning_rate": 0.004830977601644483,
+      "loss": 2.3896,
+      "step": 194400
+    },
+    {
+      "epoch": 2.614317589182505,
+      "grad_norm": 0.5799399018287659,
+      "learning_rate": 0.004830785934975585,
+      "loss": 2.3932,
+      "step": 194500
+    },
+    {
+      "epoch": 2.615661711336326,
+      "grad_norm": 0.3511755168437958,
+      "learning_rate": 0.004830594163510357,
+      "loss": 2.3863,
+      "step": 194600
+    },
+    {
+      "epoch": 2.617005833490148,
+      "grad_norm": 0.6359513401985168,
+      "learning_rate": 0.004830402287257447,
+      "loss": 2.3967,
+      "step": 194700
+    },
+    {
+      "epoch": 2.6183499556439687,
+      "grad_norm": 0.7922179102897644,
+      "learning_rate": 0.004830210306225503,
+      "loss": 2.397,
+      "step": 194800
+    },
+    {
+      "epoch": 2.6196940777977904,
+      "grad_norm": 0.43679991364479065,
+      "learning_rate": 0.004830018220423168,
+      "loss": 2.3966,
+      "step": 194900
+    },
+    {
+      "epoch": 2.6210381999516117,
+      "grad_norm": 0.5114357471466064,
+      "learning_rate": 0.004829826029859097,
+      "loss": 2.3997,
+      "step": 195000
+    },
+    {
+      "epoch": 2.6210381999516117,
+      "eval_MaskedAccuracy": 0.45087460050173545,
+      "eval_loss": 2.7082619667053223,
+      "eval_runtime": 153.3827,
+      "eval_samples_per_second": 413.841,
+      "eval_steps_per_second": 1.617,
+      "step": 195000
+    },
+    {
+      "epoch": 2.622382322105433,
+      "grad_norm": 0.5121012330055237,
+      "learning_rate": 0.004829633734541953,
+      "loss": 2.3954,
+      "step": 195100
+    },
+    {
+      "epoch": 2.6237264442592543,
+      "grad_norm": 0.693209171295166,
+      "learning_rate": 0.0048294413344804,
+      "loss": 2.3919,
+      "step": 195200
+    },
+    {
+      "epoch": 2.6250705664130756,
+      "grad_norm": 0.23428793251514435,
+      "learning_rate": 0.004829248829683105,
+      "loss": 2.3982,
+      "step": 195300
+    },
+    {
+      "epoch": 2.626414688566897,
+      "grad_norm": 0.37016478180885315,
+      "learning_rate": 0.00482905622015875,
+      "loss": 2.3932,
+      "step": 195400
+    },
+    {
+      "epoch": 2.627758810720718,
+      "grad_norm": 0.6766939759254456,
+      "learning_rate": 0.004828863505916004,
+      "loss": 2.394,
+      "step": 195500
+    },
+    {
+      "epoch": 2.62910293287454,
+      "grad_norm": 0.5270081758499146,
+      "learning_rate": 0.004828670686963563,
+      "loss": 2.3912,
+      "step": 195600
+    },
+    {
+      "epoch": 2.6304470550283607,
+      "grad_norm": 0.31191486120224,
+      "learning_rate": 0.004828477763310103,
+      "loss": 2.4008,
+      "step": 195700
+    },
+    {
+      "epoch": 2.6317911771821825,
+      "grad_norm": 0.21520403027534485,
+      "learning_rate": 0.00482828473496432,
+      "loss": 2.3868,
+      "step": 195800
+    },
+    {
+      "epoch": 2.6331352993360038,
+      "grad_norm": 0.2524843215942383,
+      "learning_rate": 0.00482809160193491,
+      "loss": 2.3957,
+      "step": 195900
+    },
+    {
+      "epoch": 2.634479421489825,
+      "grad_norm": 0.6567283272743225,
+      "learning_rate": 0.004827898364230586,
+      "loss": 2.3915,
+      "step": 196000
+    },
+    {
+      "epoch": 2.634479421489825,
+      "eval_MaskedAccuracy": 0.4512480287660106,
+      "eval_loss": 2.7046916484832764,
+      "eval_runtime": 154.0885,
+      "eval_samples_per_second": 411.945,
+      "eval_steps_per_second": 1.609,
+      "step": 196000
+    },
+    {
+      "epoch": 2.6358235436436463,
+      "grad_norm": 0.542457103729248,
+      "learning_rate": 0.004827705021860038,
+      "loss": 2.3947,
+      "step": 196100
+    },
+    {
+      "epoch": 2.6371676657974676,
+      "grad_norm": 0.44133803248405457,
+      "learning_rate": 0.004827511574831989,
+      "loss": 2.3872,
+      "step": 196200
+    },
+    {
+      "epoch": 2.638511787951289,
+      "grad_norm": 0.9183239936828613,
+      "learning_rate": 0.004827318023155146,
+      "loss": 2.3966,
+      "step": 196300
+    },
+    {
+      "epoch": 2.6398559101051102,
+      "grad_norm": 0.851475715637207,
+      "learning_rate": 0.0048271243668382485,
+      "loss": 2.3931,
+      "step": 196400
+    },
+    {
+      "epoch": 2.641200032258932,
+      "grad_norm": 0.9451562166213989,
+      "learning_rate": 0.004826930605890004,
+      "loss": 2.3922,
+      "step": 196500
+    },
+    {
+      "epoch": 2.642544154412753,
+      "grad_norm": 0.5648373961448669,
+      "learning_rate": 0.004826736740319149,
+      "loss": 2.3882,
+      "step": 196600
+    },
+    {
+      "epoch": 2.6438882765665745,
+      "grad_norm": 0.32394424080848694,
+      "learning_rate": 0.0048265427701344185,
+      "loss": 2.3918,
+      "step": 196700
+    },
+    {
+      "epoch": 2.645232398720396,
+      "grad_norm": 0.5081258416175842,
+      "learning_rate": 0.0048263486953445565,
+      "loss": 2.3863,
+      "step": 196800
+    },
+    {
+      "epoch": 2.646576520874217,
+      "grad_norm": 0.7854281067848206,
+      "learning_rate": 0.004826154515958302,
+      "loss": 2.3833,
+      "step": 196900
+    },
+    {
+      "epoch": 2.6479206430280384,
+      "grad_norm": 0.5658829212188721,
+      "learning_rate": 0.004825960231984411,
+      "loss": 2.3879,
+      "step": 197000
+    },
+    {
+      "epoch": 2.6479206430280384,
+      "eval_MaskedAccuracy": 0.4514419081074835,
+      "eval_loss": 2.7042078971862793,
+      "eval_runtime": 172.0922,
+      "eval_samples_per_second": 368.849,
+      "eval_steps_per_second": 1.441,
+      "step": 197000
+    },
+    {
+      "epoch": 2.6492647651818597,
+      "grad_norm": 0.9744159579277039,
+      "learning_rate": 0.004825765843431632,
+      "loss": 2.3993,
+      "step": 197100
+    },
+    {
+      "epoch": 2.650608887335681,
+      "grad_norm": 1.349397897720337,
+      "learning_rate": 0.004825571350308726,
+      "loss": 2.3985,
+      "step": 197200
+    },
+    {
+      "epoch": 2.6519530094895023,
+      "grad_norm": 0.8859503865242004,
+      "learning_rate": 0.004825376752624455,
+      "loss": 2.3838,
+      "step": 197300
+    },
+    {
+      "epoch": 2.6532971316433236,
+      "grad_norm": 0.27515995502471924,
+      "learning_rate": 0.004825182050387594,
+      "loss": 2.3907,
+      "step": 197400
+    },
+    {
+      "epoch": 2.654641253797145,
+      "grad_norm": 0.2807437777519226,
+      "learning_rate": 0.0048249872436069065,
+      "loss": 2.3969,
+      "step": 197500
+    },
+    {
+      "epoch": 2.6559853759509666,
+      "grad_norm": 0.9948381185531616,
+      "learning_rate": 0.00482479233229117,
+      "loss": 2.392,
+      "step": 197600
+    },
+    {
+      "epoch": 2.657329498104788,
+      "grad_norm": 1.0630486011505127,
+      "learning_rate": 0.004824597316449177,
+      "loss": 2.3895,
+      "step": 197700
+    },
+    {
+      "epoch": 2.658673620258609,
+      "grad_norm": 0.24495822191238403,
+      "learning_rate": 0.004824402196089705,
+      "loss": 2.3984,
+      "step": 197800
+    },
+    {
+      "epoch": 2.6600177424124305,
+      "grad_norm": 1.0143710374832153,
+      "learning_rate": 0.004824206971221553,
+      "loss": 2.3864,
+      "step": 197900
+    },
+    {
+      "epoch": 2.661361864566252,
+      "grad_norm": 0.2758951187133789,
+      "learning_rate": 0.0048240116418535135,
+      "loss": 2.3837,
+      "step": 198000
+    },
+    {
+      "epoch": 2.661361864566252,
+      "eval_MaskedAccuracy": 0.4513660960606149,
+      "eval_loss": 2.70383882522583,
+      "eval_runtime": 159.3018,
+      "eval_samples_per_second": 398.464,
+      "eval_steps_per_second": 1.557,
+      "step": 198000
+    },
+    {
+      "epoch": 2.662705986720073,
+      "grad_norm": 0.34540656208992004,
+      "learning_rate": 0.004823816207994382,
+      "loss": 2.3947,
+      "step": 198100
+    },
+    {
+      "epoch": 2.6640501088738944,
+      "grad_norm": 0.716076672077179,
+      "learning_rate": 0.004823620669652978,
+      "loss": 2.3981,
+      "step": 198200
+    },
+    {
+      "epoch": 2.6653942310277157,
+      "grad_norm": 0.37496283650398254,
+      "learning_rate": 0.004823425026838102,
+      "loss": 2.3902,
+      "step": 198300
+    },
+    {
+      "epoch": 2.666738353181537,
+      "grad_norm": 2.4844629764556885,
+      "learning_rate": 0.004823229279558573,
+      "loss": 2.387,
+      "step": 198400
+    },
+    {
+      "epoch": 2.6680824753353587,
+      "grad_norm": 0.2198270559310913,
+      "learning_rate": 0.004823033427823211,
+      "loss": 2.3895,
+      "step": 198500
+    },
+    {
+      "epoch": 2.6694265974891795,
+      "grad_norm": 0.23235642910003662,
+      "learning_rate": 0.004822837471640838,
+      "loss": 2.3955,
+      "step": 198600
+    },
+    {
+      "epoch": 2.6707707196430013,
+      "grad_norm": 0.5798563957214355,
+      "learning_rate": 0.00482264141102029,
+      "loss": 2.3911,
+      "step": 198700
+    },
+    {
+      "epoch": 2.6721148417968226,
+      "grad_norm": 0.36171451210975647,
+      "learning_rate": 0.004822445245970393,
+      "loss": 2.3919,
+      "step": 198800
+    },
+    {
+      "epoch": 2.673458963950644,
+      "grad_norm": 0.9405841827392578,
+      "learning_rate": 0.00482224897649999,
+      "loss": 2.3965,
+      "step": 198900
+    },
+    {
+      "epoch": 2.674803086104465,
+      "grad_norm": 0.40147000551223755,
+      "learning_rate": 0.004822052602617928,
+      "loss": 2.3928,
+      "step": 199000
+    },
+    {
+      "epoch": 2.674803086104465,
+      "eval_MaskedAccuracy": 0.451330398429542,
+      "eval_loss": 2.7040762901306152,
+      "eval_runtime": 152.7346,
+      "eval_samples_per_second": 415.597,
+      "eval_steps_per_second": 1.624,
+      "step": 199000
+    },
+    {
+      "epoch": 2.6761472082582864,
+      "grad_norm": 0.9061090350151062,
+      "learning_rate": 0.004821856124333049,
+      "loss": 2.3848,
+      "step": 199100
+    },
+    {
+      "epoch": 2.6774913304121077,
+      "grad_norm": 0.8056171536445618,
+      "learning_rate": 0.004821659541654204,
+      "loss": 2.3965,
+      "step": 199200
+    },
+    {
+      "epoch": 2.678835452565929,
+      "grad_norm": 2.8304381370544434,
+      "learning_rate": 0.004821462854590257,
+      "loss": 2.3895,
+      "step": 199300
+    },
+    {
+      "epoch": 2.6801795747197508,
+      "grad_norm": 0.6876057386398315,
+      "learning_rate": 0.004821266063150074,
+      "loss": 2.3914,
+      "step": 199400
+    },
+    {
+      "epoch": 2.6815236968735716,
+      "grad_norm": 1.1147319078445435,
+      "learning_rate": 0.004821069167342508,
+      "loss": 2.3884,
+      "step": 199500
+    },
+    {
+      "epoch": 2.6828678190273934,
+      "grad_norm": 0.563893735408783,
+      "learning_rate": 0.004820872167176442,
+      "loss": 2.3824,
+      "step": 199600
+    },
+    {
+      "epoch": 2.6842119411812146,
+      "grad_norm": 0.9097170829772949,
+      "learning_rate": 0.004820675062660752,
+      "loss": 2.3951,
+      "step": 199700
+    },
+    {
+      "epoch": 2.685556063335036,
+      "grad_norm": 0.5297996997833252,
+      "learning_rate": 0.004820477853804319,
+      "loss": 2.3987,
+      "step": 199800
+    },
+    {
+      "epoch": 2.6869001854888572,
+      "grad_norm": 0.4210087060928345,
+      "learning_rate": 0.0048202805406160265,
+      "loss": 2.3966,
+      "step": 199900
+    },
+    {
+      "epoch": 2.6882443076426785,
+      "grad_norm": 0.9611253142356873,
+      "learning_rate": 0.004820083123104765,
+      "loss": 2.3946,
+      "step": 200000
+    },
+    {
+      "epoch": 2.6882443076426785,
+      "eval_MaskedAccuracy": 0.4521056806971866,
+      "eval_loss": 2.7007529735565186,
+      "eval_runtime": 154.1926,
+      "eval_samples_per_second": 411.667,
+      "eval_steps_per_second": 1.608,
+      "step": 200000
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 1500000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 21,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.56886375582715e+18,
+  "train_batch_size": 384,
+  "trial_name": null,
+  "trial_params": null
+}