diff --git "a/checkpoint-770/trainer_state.json" "b/checkpoint-770/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-770/trainer_state.json"
@@ -0,0 +1,5432 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.541516245487364,
+  "eval_steps": 770,
+  "global_step": 770,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.007220216606498195,
+      "grad_norm": 2.68963885307312,
+      "learning_rate": 0.0,
+      "loss": 1.880106806755066,
+      "step": 1
+    },
+    {
+      "epoch": 0.01444043321299639,
+      "grad_norm": 2.3691370487213135,
+      "learning_rate": 8e-05,
+      "loss": 1.882678508758545,
+      "step": 2
+    },
+    {
+      "epoch": 0.021660649819494584,
+      "grad_norm": 3.7024030685424805,
+      "learning_rate": 0.00016,
+      "loss": 1.9426246881484985,
+      "step": 3
+    },
+    {
+      "epoch": 0.02888086642599278,
+      "grad_norm": 0.7646510601043701,
+      "learning_rate": 0.00024,
+      "loss": 1.6261038780212402,
+      "step": 4
+    },
+    {
+      "epoch": 0.036101083032490974,
+      "grad_norm": 0.6628475189208984,
+      "learning_rate": 0.00032,
+      "loss": 1.478842854499817,
+      "step": 5
+    },
+    {
+      "epoch": 0.04332129963898917,
+      "grad_norm": 0.7141556143760681,
+      "learning_rate": 0.0004,
+      "loss": 1.1376413106918335,
+      "step": 6
+    },
+    {
+      "epoch": 0.05054151624548736,
+      "grad_norm": 0.7212001085281372,
+      "learning_rate": 0.0003999983135391464,
+      "loss": 1.0410869121551514,
+      "step": 7
+    },
+    {
+      "epoch": 0.05776173285198556,
+      "grad_norm": 0.7515780925750732,
+      "learning_rate": 0.0003999932541850271,
+      "loss": 0.8246882557868958,
+      "step": 8
+    },
+    {
+      "epoch": 0.06498194945848375,
+      "grad_norm": 2.474442481994629,
+      "learning_rate": 0.0003999848220229661,
+      "loss": 0.7787771821022034,
+      "step": 9
+    },
+    {
+      "epoch": 0.07220216606498195,
+      "grad_norm": 1.2009592056274414,
+      "learning_rate": 0.0003999730171951686,
+      "loss": 0.6830324530601501,
+      "step": 10
+    },
+    {
+      "epoch": 0.07942238267148015,
+      "grad_norm": 0.7545379996299744,
+      "learning_rate": 0.00039995783990071825,
+      "loss": 0.7438161969184875,
+      "step": 11
+    },
+    {
+      "epoch": 0.08664259927797834,
+      "grad_norm": 0.9728625416755676,
+      "learning_rate": 0.0003999392903955744,
+      "loss": 0.6851149797439575,
+      "step": 12
+    },
+    {
+      "epoch": 0.09386281588447654,
+      "grad_norm": 0.5509404540061951,
+      "learning_rate": 0.00039991736899256693,
+      "loss": 0.5853980183601379,
+      "step": 13
+    },
+    {
+      "epoch": 0.10108303249097472,
+      "grad_norm": 0.715829074382782,
+      "learning_rate": 0.0003998920760613919,
+      "loss": 0.7855823040008545,
+      "step": 14
+    },
+    {
+      "epoch": 0.10830324909747292,
+      "grad_norm": 0.6777167320251465,
+      "learning_rate": 0.0003998634120286046,
+      "loss": 0.5202292799949646,
+      "step": 15
+    },
+    {
+      "epoch": 0.11552346570397112,
+      "grad_norm": 0.5348321199417114,
+      "learning_rate": 0.0003998313773776128,
+      "loss": 0.6527559757232666,
+      "step": 16
+    },
+    {
+      "epoch": 0.12274368231046931,
+      "grad_norm": 0.4939132332801819,
+      "learning_rate": 0.0003997959726486683,
+      "loss": 0.584251880645752,
+      "step": 17
+    },
+    {
+      "epoch": 0.1299638989169675,
+      "grad_norm": 0.47802454233169556,
+      "learning_rate": 0.00039975719843885805,
+      "loss": 0.5235528349876404,
+      "step": 18
+    },
+    {
+      "epoch": 0.1371841155234657,
+      "grad_norm": 0.44072362780570984,
+      "learning_rate": 0.0003997150554020938,
+      "loss": 0.5184933543205261,
+      "step": 19
+    },
+    {
+      "epoch": 0.1444043321299639,
+      "grad_norm": 0.4251253306865692,
+      "learning_rate": 0.00039966954424910147,
+      "loss": 0.6167617440223694,
+      "step": 20
+    },
+    {
+      "epoch": 0.15162454873646208,
+      "grad_norm": 0.5198166370391846,
+      "learning_rate": 0.00039962066574740886,
+      "loss": 0.7412333488464355,
+      "step": 21
+    },
+    {
+      "epoch": 0.1588447653429603,
+      "grad_norm": 0.44667476415634155,
+      "learning_rate": 0.00039956842072133266,
+      "loss": 0.6361726522445679,
+      "step": 22
+    },
+    {
+      "epoch": 0.16606498194945848,
+      "grad_norm": 0.40787550806999207,
+      "learning_rate": 0.0003995128100519649,
+      "loss": 0.49531102180480957,
+      "step": 23
+    },
+    {
+      "epoch": 0.17328519855595667,
+      "grad_norm": 0.4681910276412964,
+      "learning_rate": 0.00039945383467715766,
+      "loss": 0.584551990032196,
+      "step": 24
+    },
+    {
+      "epoch": 0.18050541516245489,
+      "grad_norm": 0.41188451647758484,
+      "learning_rate": 0.00039939149559150753,
+      "loss": 0.6003429889678955,
+      "step": 25
+    },
+    {
+      "epoch": 0.18772563176895307,
+      "grad_norm": 0.43262341618537903,
+      "learning_rate": 0.0003993257938463389,
+      "loss": 0.5725842118263245,
+      "step": 26
+    },
+    {
+      "epoch": 0.19494584837545126,
+      "grad_norm": 0.45091742277145386,
+      "learning_rate": 0.0003992567305496859,
+      "loss": 0.6373569965362549,
+      "step": 27
+    },
+    {
+      "epoch": 0.20216606498194944,
+      "grad_norm": 0.3687536120414734,
+      "learning_rate": 0.00039918430686627406,
+      "loss": 0.4958152174949646,
+      "step": 28
+    },
+    {
+      "epoch": 0.20938628158844766,
+      "grad_norm": 0.3994904160499573,
+      "learning_rate": 0.00039910852401750036,
+      "loss": 0.6146813631057739,
+      "step": 29
+    },
+    {
+      "epoch": 0.21660649819494585,
+      "grad_norm": 0.41312268376350403,
+      "learning_rate": 0.0003990293832814129,
+      "loss": 0.5313689708709717,
+      "step": 30
+    },
+    {
+      "epoch": 0.22382671480144403,
+      "grad_norm": 0.37877994775772095,
+      "learning_rate": 0.0003989468859926893,
+      "loss": 0.5694329738616943,
+      "step": 31
+    },
+    {
+      "epoch": 0.23104693140794225,
+      "grad_norm": 0.32251644134521484,
+      "learning_rate": 0.0003988610335426139,
+      "loss": 0.4476504921913147,
+      "step": 32
+    },
+    {
+      "epoch": 0.23826714801444043,
+      "grad_norm": 0.40205422043800354,
+      "learning_rate": 0.0003987718273790548,
+      "loss": 0.4696302115917206,
+      "step": 33
+    },
+    {
+      "epoch": 0.24548736462093862,
+      "grad_norm": 0.3664834201335907,
+      "learning_rate": 0.0003986792690064389,
+      "loss": 0.5304725170135498,
+      "step": 34
+    },
+    {
+      "epoch": 0.2527075812274368,
+      "grad_norm": 0.5150560736656189,
+      "learning_rate": 0.000398583359985727,
+      "loss": 0.5886795520782471,
+      "step": 35
+    },
+    {
+      "epoch": 0.259927797833935,
+      "grad_norm": 0.37544649839401245,
+      "learning_rate": 0.0003984841019343872,
+      "loss": 0.6374161839485168,
+      "step": 36
+    },
+    {
+      "epoch": 0.26714801444043323,
+      "grad_norm": 0.3426123857498169,
+      "learning_rate": 0.0003983814965263676,
+      "loss": 0.507249116897583,
+      "step": 37
+    },
+    {
+      "epoch": 0.2743682310469314,
+      "grad_norm": 0.36909183859825134,
+      "learning_rate": 0.00039827554549206836,
+      "loss": 0.5216564536094666,
+      "step": 38
+    },
+    {
+      "epoch": 0.2815884476534296,
+      "grad_norm": 0.3864560127258301,
+      "learning_rate": 0.00039816625061831206,
+      "loss": 0.4607892632484436,
+      "step": 39
+    },
+    {
+      "epoch": 0.2888086642599278,
+      "grad_norm": 0.3553301692008972,
+      "learning_rate": 0.0003980536137483141,
+      "loss": 0.6021184921264648,
+      "step": 40
+    },
+    {
+      "epoch": 0.296028880866426,
+      "grad_norm": 0.3300604224205017,
+      "learning_rate": 0.0003979376367816511,
+      "loss": 0.49985307455062866,
+      "step": 41
+    },
+    {
+      "epoch": 0.30324909747292417,
+      "grad_norm": 0.3894491493701935,
+      "learning_rate": 0.0003978183216742292,
+      "loss": 0.5403987169265747,
+      "step": 42
+    },
+    {
+      "epoch": 0.3104693140794224,
+      "grad_norm": 0.32339245080947876,
+      "learning_rate": 0.0003976956704382511,
+      "loss": 0.5262795090675354,
+      "step": 43
+    },
+    {
+      "epoch": 0.3176895306859206,
+      "grad_norm": 0.3121689260005951,
+      "learning_rate": 0.00039756968514218166,
+      "loss": 0.4988410174846649,
+      "step": 44
+    },
+    {
+      "epoch": 0.3249097472924188,
+      "grad_norm": 0.37562957406044006,
+      "learning_rate": 0.0003974403679107138,
+      "loss": 0.5539376735687256,
+      "step": 45
+    },
+    {
+      "epoch": 0.33212996389891697,
+      "grad_norm": 0.3560684323310852,
+      "learning_rate": 0.0003973077209247319,
+      "loss": 0.4829791784286499,
+      "step": 46
+    },
+    {
+      "epoch": 0.33935018050541516,
+      "grad_norm": 0.3543923497200012,
+      "learning_rate": 0.0003971717464212755,
+      "loss": 0.4733068346977234,
+      "step": 47
+    },
+    {
+      "epoch": 0.34657039711191334,
+      "grad_norm": 0.3253995180130005,
+      "learning_rate": 0.0003970324466935013,
+      "loss": 0.4318773150444031,
+      "step": 48
+    },
+    {
+      "epoch": 0.35379061371841153,
+      "grad_norm": 0.3587929904460907,
+      "learning_rate": 0.0003968898240906447,
+      "loss": 0.6348429322242737,
+      "step": 49
+    },
+    {
+      "epoch": 0.36101083032490977,
+      "grad_norm": 0.3772837519645691,
+      "learning_rate": 0.0003967438810179802,
+      "loss": 0.5192855596542358,
+      "step": 50
+    },
+    {
+      "epoch": 0.36823104693140796,
+      "grad_norm": 0.3878256380558014,
+      "learning_rate": 0.0003965946199367804,
+      "loss": 0.4653000831604004,
+      "step": 51
+    },
+    {
+      "epoch": 0.37545126353790614,
+      "grad_norm": 0.306704044342041,
+      "learning_rate": 0.00039644204336427506,
+      "loss": 0.46054646372795105,
+      "step": 52
+    },
+    {
+      "epoch": 0.38267148014440433,
+      "grad_norm": 0.3290722072124481,
+      "learning_rate": 0.0003962861538736084,
+      "loss": 0.4637136459350586,
+      "step": 53
+    },
+    {
+      "epoch": 0.3898916967509025,
+      "grad_norm": 0.3877376914024353,
+      "learning_rate": 0.00039612695409379557,
+      "loss": 0.6008591651916504,
+      "step": 54
+    },
+    {
+      "epoch": 0.3971119133574007,
+      "grad_norm": 0.3738860785961151,
+      "learning_rate": 0.00039596444670967864,
+      "loss": 0.49255216121673584,
+      "step": 55
+    },
+    {
+      "epoch": 0.4043321299638989,
+      "grad_norm": 0.43176719546318054,
+      "learning_rate": 0.0003957986344618809,
+      "loss": 0.6328399777412415,
+      "step": 56
+    },
+    {
+      "epoch": 0.41155234657039713,
+      "grad_norm": 0.34965041279792786,
+      "learning_rate": 0.00039562952014676116,
+      "loss": 0.5274545550346375,
+      "step": 57
+    },
+    {
+      "epoch": 0.4187725631768953,
+      "grad_norm": 0.3525114059448242,
+      "learning_rate": 0.000395457106616366,
+      "loss": 0.5174035429954529,
+      "step": 58
+    },
+    {
+      "epoch": 0.4259927797833935,
+      "grad_norm": 0.36805179715156555,
+      "learning_rate": 0.0003952813967783822,
+      "loss": 0.5132086277008057,
+      "step": 59
+    },
+    {
+      "epoch": 0.4332129963898917,
+      "grad_norm": 0.3320341110229492,
+      "learning_rate": 0.00039510239359608735,
+      "loss": 0.47466331720352173,
+      "step": 60
+    },
+    {
+      "epoch": 0.4404332129963899,
+      "grad_norm": 0.36727970838546753,
+      "learning_rate": 0.0003949201000883001,
+      "loss": 0.5845056772232056,
+      "step": 61
+    },
+    {
+      "epoch": 0.44765342960288806,
+      "grad_norm": 0.31494706869125366,
+      "learning_rate": 0.000394734519329329,
+      "loss": 0.5123739242553711,
+      "step": 62
+    },
+    {
+      "epoch": 0.4548736462093863,
+      "grad_norm": 0.31959858536720276,
+      "learning_rate": 0.00039454565444892094,
+      "loss": 0.3888605237007141,
+      "step": 63
+    },
+    {
+      "epoch": 0.4620938628158845,
+      "grad_norm": 0.4212755262851715,
+      "learning_rate": 0.0003943535086322083,
+      "loss": 0.6065781712532043,
+      "step": 64
+    },
+    {
+      "epoch": 0.4693140794223827,
+      "grad_norm": 0.2938922643661499,
+      "learning_rate": 0.0003941580851196549,
+      "loss": 0.47488927841186523,
+      "step": 65
+    },
+    {
+      "epoch": 0.47653429602888087,
+      "grad_norm": 0.27896228432655334,
+      "learning_rate": 0.00039395938720700196,
+      "loss": 0.36559349298477173,
+      "step": 66
+    },
+    {
+      "epoch": 0.48375451263537905,
+      "grad_norm": 0.34284013509750366,
+      "learning_rate": 0.0003937574182452118,
+      "loss": 0.49944937229156494,
+      "step": 67
+    },
+    {
+      "epoch": 0.49097472924187724,
+      "grad_norm": 0.3411049544811249,
+      "learning_rate": 0.00039355218164041194,
+      "loss": 0.5018436312675476,
+      "step": 68
+    },
+    {
+      "epoch": 0.4981949458483754,
+      "grad_norm": 0.35431182384490967,
+      "learning_rate": 0.0003933436808538375,
+      "loss": 0.42643067240715027,
+      "step": 69
+    },
+    {
+      "epoch": 0.5054151624548736,
+      "grad_norm": 0.3351145386695862,
+      "learning_rate": 0.00039313191940177256,
+      "loss": 0.5355833768844604,
+      "step": 70
+    },
+    {
+      "epoch": 0.5126353790613718,
+      "grad_norm": 0.3065779507160187,
+      "learning_rate": 0.00039291690085549105,
+      "loss": 0.4329970180988312,
+      "step": 71
+    },
+    {
+      "epoch": 0.51985559566787,
+      "grad_norm": 0.5208595395088196,
+      "learning_rate": 0.00039269862884119664,
+      "loss": 0.5850666165351868,
+      "step": 72
+    },
+    {
+      "epoch": 0.5270758122743683,
+      "grad_norm": 0.3183553218841553,
+      "learning_rate": 0.00039247710703996144,
+      "loss": 0.5500473380088806,
+      "step": 73
+    },
+    {
+      "epoch": 0.5342960288808665,
+      "grad_norm": 0.32788676023483276,
+      "learning_rate": 0.0003922523391876638,
+      "loss": 0.5200982093811035,
+      "step": 74
+    },
+    {
+      "epoch": 0.5415162454873647,
+      "grad_norm": 0.2959986925125122,
+      "learning_rate": 0.00039202432907492566,
+      "loss": 0.4983074963092804,
+      "step": 75
+    },
+    {
+      "epoch": 0.5487364620938628,
+      "grad_norm": 0.32182440161705017,
+      "learning_rate": 0.0003917930805470483,
+      "loss": 0.44362854957580566,
+      "step": 76
+    },
+    {
+      "epoch": 0.555956678700361,
+      "grad_norm": 0.2816471755504608,
+      "learning_rate": 0.00039155859750394754,
+      "loss": 0.44515836238861084,
+      "step": 77
+    },
+    {
+      "epoch": 0.5631768953068592,
+      "grad_norm": 0.290783554315567,
+      "learning_rate": 0.0003913208839000882,
+      "loss": 0.46542418003082275,
+      "step": 78
+    },
+    {
+      "epoch": 0.5703971119133574,
+      "grad_norm": 0.366773396730423,
+      "learning_rate": 0.0003910799437444171,
+      "loss": 0.5370551943778992,
+      "step": 79
+    },
+    {
+      "epoch": 0.5776173285198556,
+      "grad_norm": 0.32066911458969116,
+      "learning_rate": 0.0003908357811002957,
+      "loss": 0.5146384239196777,
+      "step": 80
+    },
+    {
+      "epoch": 0.5848375451263538,
+      "grad_norm": 0.3420974016189575,
+      "learning_rate": 0.00039058840008543135,
+      "loss": 0.6218519806861877,
+      "step": 81
+    },
+    {
+      "epoch": 0.592057761732852,
+      "grad_norm": 0.3156997263431549,
+      "learning_rate": 0.000390337804871808,
+      "loss": 0.549630343914032,
+      "step": 82
+    },
+    {
+      "epoch": 0.5992779783393501,
+      "grad_norm": 0.3494798541069031,
+      "learning_rate": 0.00039008399968561597,
+      "loss": 0.5499352216720581,
+      "step": 83
+    },
+    {
+      "epoch": 0.6064981949458483,
+      "grad_norm": 0.3652164936065674,
+      "learning_rate": 0.0003898269888071803,
+      "loss": 0.5562577843666077,
+      "step": 84
+    },
+    {
+      "epoch": 0.6137184115523465,
+      "grad_norm": 0.34751319885253906,
+      "learning_rate": 0.0003895667765708887,
+      "loss": 0.5954532027244568,
+      "step": 85
+    },
+    {
+      "epoch": 0.6209386281588448,
+      "grad_norm": 0.28338274359703064,
+      "learning_rate": 0.0003893033673651189,
+      "loss": 0.4894581735134125,
+      "step": 86
+    },
+    {
+      "epoch": 0.628158844765343,
+      "grad_norm": 0.29019951820373535,
+      "learning_rate": 0.00038903676563216394,
+      "loss": 0.4635826349258423,
+      "step": 87
+    },
+    {
+      "epoch": 0.6353790613718412,
+      "grad_norm": 0.2782176434993744,
+      "learning_rate": 0.00038876697586815763,
+      "loss": 0.4121658205986023,
+      "step": 88
+    },
+    {
+      "epoch": 0.6425992779783394,
+      "grad_norm": 0.27983078360557556,
+      "learning_rate": 0.0003884940026229988,
+      "loss": 0.40188780426979065,
+      "step": 89
+    },
+    {
+      "epoch": 0.6498194945848376,
+      "grad_norm": 0.2786664366722107,
+      "learning_rate": 0.0003882178505002744,
+      "loss": 0.44132208824157715,
+      "step": 90
+    },
+    {
+      "epoch": 0.6570397111913358,
+      "grad_norm": 0.3355483412742615,
+      "learning_rate": 0.0003879385241571817,
+      "loss": 0.512315034866333,
+      "step": 91
+    },
+    {
+      "epoch": 0.6642599277978339,
+      "grad_norm": 0.29554808139801025,
+      "learning_rate": 0.0003876560283044503,
+      "loss": 0.4540112018585205,
+      "step": 92
+    },
+    {
+      "epoch": 0.6714801444043321,
+      "grad_norm": 0.3431720435619354,
+      "learning_rate": 0.00038737036770626214,
+      "loss": 0.45164141058921814,
+      "step": 93
+    },
+    {
+      "epoch": 0.6787003610108303,
+      "grad_norm": 0.37090009450912476,
+      "learning_rate": 0.0003870815471801714,
+      "loss": 0.4878736734390259,
+      "step": 94
+    },
+    {
+      "epoch": 0.6859205776173285,
+      "grad_norm": 0.35553082823753357,
+      "learning_rate": 0.0003867895715970231,
+      "loss": 0.5573399662971497,
+      "step": 95
+    },
+    {
+      "epoch": 0.6931407942238267,
+      "grad_norm": 0.30914071202278137,
+      "learning_rate": 0.00038649444588087115,
+      "loss": 0.5417919754981995,
+      "step": 96
+    },
+    {
+      "epoch": 0.7003610108303249,
+      "grad_norm": 0.27384090423583984,
+      "learning_rate": 0.0003861961750088954,
+      "loss": 0.4475242793560028,
+      "step": 97
+    },
+    {
+      "epoch": 0.7075812274368231,
+      "grad_norm": 0.2919006049633026,
+      "learning_rate": 0.0003858947640113171,
+      "loss": 0.44341611862182617,
+      "step": 98
+    },
+    {
+      "epoch": 0.7148014440433214,
+      "grad_norm": 0.33408623933792114,
+      "learning_rate": 0.000385590217971315,
+      "loss": 0.5056685209274292,
+      "step": 99
+    },
+    {
+      "epoch": 0.7220216606498195,
+      "grad_norm": 0.3498390018939972,
+      "learning_rate": 0.0003852825420249386,
+      "loss": 0.45975226163864136,
+      "step": 100
+    },
+    {
+      "epoch": 0.7292418772563177,
+      "grad_norm": 0.32381942868232727,
+      "learning_rate": 0.0003849717413610224,
+      "loss": 0.43437302112579346,
+      "step": 101
+    },
+    {
+      "epoch": 0.7364620938628159,
+      "grad_norm": 0.30022308230400085,
+      "learning_rate": 0.0003846578212210979,
+      "loss": 0.41799411177635193,
+      "step": 102
+    },
+    {
+      "epoch": 0.7436823104693141,
+      "grad_norm": 0.31061094999313354,
+      "learning_rate": 0.0003843407868993054,
+      "loss": 0.5682204365730286,
+      "step": 103
+    },
+    {
+      "epoch": 0.7509025270758123,
+      "grad_norm": 0.31062883138656616,
+      "learning_rate": 0.0003840206437423047,
+      "loss": 0.3520132303237915,
+      "step": 104
+    },
+    {
+      "epoch": 0.7581227436823105,
+      "grad_norm": 0.36826908588409424,
+      "learning_rate": 0.00038369739714918464,
+      "loss": 0.546819269657135,
+      "step": 105
+    },
+    {
+      "epoch": 0.7653429602888087,
+      "grad_norm": 0.35304203629493713,
+      "learning_rate": 0.00038337105257137263,
+      "loss": 0.5633726716041565,
+      "step": 106
+    },
+    {
+      "epoch": 0.7725631768953068,
+      "grad_norm": 0.2879215478897095,
+      "learning_rate": 0.00038304161551254207,
+      "loss": 0.42212507128715515,
+      "step": 107
+    },
+    {
+      "epoch": 0.779783393501805,
+      "grad_norm": 0.32802048325538635,
+      "learning_rate": 0.0003827090915285202,
+      "loss": 0.4973101019859314,
+      "step": 108
+    },
+    {
+      "epoch": 0.7870036101083032,
+      "grad_norm": 0.29462599754333496,
+      "learning_rate": 0.0003823734862271937,
+      "loss": 0.4543662667274475,
+      "step": 109
+    },
+    {
+      "epoch": 0.7942238267148014,
+      "grad_norm": 0.34288832545280457,
+      "learning_rate": 0.0003820348052684146,
+      "loss": 0.5599743127822876,
+      "step": 110
+    },
+    {
+      "epoch": 0.8014440433212996,
+      "grad_norm": 0.2871699333190918,
+      "learning_rate": 0.00038169305436390477,
+      "loss": 0.5132560133934021,
+      "step": 111
+    },
+    {
+      "epoch": 0.8086642599277978,
+      "grad_norm": 0.3307580053806305,
+      "learning_rate": 0.00038134823927715933,
+      "loss": 0.5319223403930664,
+      "step": 112
+    },
+    {
+      "epoch": 0.8158844765342961,
+      "grad_norm": 0.3292873799800873,
+      "learning_rate": 0.0003810003658233498,
+      "loss": 0.5933513045310974,
+      "step": 113
+    },
+    {
+      "epoch": 0.8231046931407943,
+      "grad_norm": 0.34675127267837524,
+      "learning_rate": 0.0003806494398692258,
+      "loss": 0.6370081901550293,
+      "step": 114
+    },
+    {
+      "epoch": 0.8303249097472925,
+      "grad_norm": 0.31837624311447144,
+      "learning_rate": 0.00038029546733301615,
+      "loss": 0.44530773162841797,
+      "step": 115
+    },
+    {
+      "epoch": 0.8375451263537906,
+      "grad_norm": 0.3014152944087982,
+      "learning_rate": 0.0003799384541843291,
+      "loss": 0.5310161113739014,
+      "step": 116
+    },
+    {
+      "epoch": 0.8447653429602888,
+      "grad_norm": 0.30693644285202026,
+      "learning_rate": 0.00037957840644405164,
+      "loss": 0.5011177659034729,
+      "step": 117
+    },
+    {
+      "epoch": 0.851985559566787,
+      "grad_norm": 0.2840467095375061,
+      "learning_rate": 0.000379215330184248,
+      "loss": 0.37618446350097656,
+      "step": 118
+    },
+    {
+      "epoch": 0.8592057761732852,
+      "grad_norm": 0.3380180597305298,
+      "learning_rate": 0.00037884923152805715,
+      "loss": 0.5536396503448486,
+      "step": 119
+    },
+    {
+      "epoch": 0.8664259927797834,
+      "grad_norm": 0.30026137828826904,
+      "learning_rate": 0.00037848011664958956,
+      "loss": 0.5487097501754761,
+      "step": 120
+    },
+    {
+      "epoch": 0.8736462093862816,
+      "grad_norm": 0.3990669250488281,
+      "learning_rate": 0.00037810799177382325,
+      "loss": 0.34712734818458557,
+      "step": 121
+    },
+    {
+      "epoch": 0.8808664259927798,
+      "grad_norm": 0.3168427646160126,
+      "learning_rate": 0.0003777328631764986,
+      "loss": 0.44405245780944824,
+      "step": 122
+    },
+    {
+      "epoch": 0.8880866425992779,
+      "grad_norm": 0.2761262059211731,
+      "learning_rate": 0.0003773547371840124,
+      "loss": 0.4112606942653656,
+      "step": 123
+    },
+    {
+      "epoch": 0.8953068592057761,
+      "grad_norm": 0.27056002616882324,
+      "learning_rate": 0.00037697362017331164,
+      "loss": 0.45218032598495483,
+      "step": 124
+    },
+    {
+      "epoch": 0.9025270758122743,
+      "grad_norm": 0.2833767831325531,
+      "learning_rate": 0.00037658951857178543,
+      "loss": 0.4833817481994629,
+      "step": 125
+    },
+    {
+      "epoch": 0.9097472924187726,
+      "grad_norm": 0.30818742513656616,
+      "learning_rate": 0.0003762024388571569,
+      "loss": 0.4847618341445923,
+      "step": 126
+    },
+    {
+      "epoch": 0.9169675090252708,
+      "grad_norm": 0.32447490096092224,
+      "learning_rate": 0.00037581238755737405,
+      "loss": 0.5068320035934448,
+      "step": 127
+    },
+    {
+      "epoch": 0.924187725631769,
+      "grad_norm": 0.307781845331192,
+      "learning_rate": 0.0003754193712504992,
+      "loss": 0.455619752407074,
+      "step": 128
+    },
+    {
+      "epoch": 0.9314079422382672,
+      "grad_norm": 0.29909971356391907,
+      "learning_rate": 0.00037502339656459853,
+      "loss": 0.5618020296096802,
+      "step": 129
+    },
+    {
+      "epoch": 0.9386281588447654,
+      "grad_norm": 0.33699020743370056,
+      "learning_rate": 0.0003746244701776302,
+      "loss": 0.50435870885849,
+      "step": 130
+    },
+    {
+      "epoch": 0.9458483754512635,
+      "grad_norm": 0.2932925820350647,
+      "learning_rate": 0.0003742225988173315,
+      "loss": 0.503471314907074,
+      "step": 131
+    },
+    {
+      "epoch": 0.9530685920577617,
+      "grad_norm": 0.27647075057029724,
+      "learning_rate": 0.00037381778926110567,
+      "loss": 0.4979417324066162,
+      "step": 132
+    },
+    {
+      "epoch": 0.9602888086642599,
+      "grad_norm": 0.23902934789657593,
+      "learning_rate": 0.00037341004833590734,
+      "loss": 0.4127359390258789,
+      "step": 133
+    },
+    {
+      "epoch": 0.9675090252707581,
+      "grad_norm": 0.3308359980583191,
+      "learning_rate": 0.0003729993829181276,
+      "loss": 0.6199836730957031,
+      "step": 134
+    },
+    {
+      "epoch": 0.9747292418772563,
+      "grad_norm": 0.30997487902641296,
+      "learning_rate": 0.00037258579993347795,
+      "loss": 0.5043972134590149,
+      "step": 135
+    },
+    {
+      "epoch": 0.9819494584837545,
+      "grad_norm": 0.5225651264190674,
+      "learning_rate": 0.00037216930635687357,
+      "loss": 0.5078832507133484,
+      "step": 136
+    },
+    {
+      "epoch": 0.9891696750902527,
+      "grad_norm": 0.2910468280315399,
+      "learning_rate": 0.00037174990921231557,
+      "loss": 0.5478032827377319,
+      "step": 137
+    },
+    {
+      "epoch": 0.9963898916967509,
+      "grad_norm": 0.3448226749897003,
+      "learning_rate": 0.00037132761557277257,
+      "loss": 0.5461761355400085,
+      "step": 138
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5731388926506042,
+      "learning_rate": 0.00037090243256006155,
+      "loss": 0.39478886127471924,
+      "step": 139
+    },
+    {
+      "epoch": 1.0072202166064983,
+      "grad_norm": 0.3017331659793854,
+      "learning_rate": 0.0003704743673447276,
+      "loss": 0.3729703426361084,
+      "step": 140
+    },
+    {
+      "epoch": 1.0144404332129964,
+      "grad_norm": 0.29272574186325073,
+      "learning_rate": 0.0003700434271459229,
+      "loss": 0.45498228073120117,
+      "step": 141
+    },
+    {
+      "epoch": 1.0216606498194947,
+      "grad_norm": 0.2818097770214081,
+      "learning_rate": 0.0003696096192312852,
+      "loss": 0.37411680817604065,
+      "step": 142
+    },
+    {
+      "epoch": 1.0288808664259927,
+      "grad_norm": 0.31325557827949524,
+      "learning_rate": 0.00036917295091681526,
+      "loss": 0.4299224019050598,
+      "step": 143
+    },
+    {
+      "epoch": 1.036101083032491,
+      "grad_norm": 0.2971343994140625,
+      "learning_rate": 0.0003687334295667533,
+      "loss": 0.4103966951370239,
+      "step": 144
+    },
+    {
+      "epoch": 1.0433212996389891,
+      "grad_norm": 0.32004889845848083,
+      "learning_rate": 0.00036829106259345465,
+      "loss": 0.49494248628616333,
+      "step": 145
+    },
+    {
+      "epoch": 1.0505415162454874,
+      "grad_norm": 0.29688122868537903,
+      "learning_rate": 0.00036784585745726535,
+      "loss": 0.4550933241844177,
+      "step": 146
+    },
+    {
+      "epoch": 1.0577617328519855,
+      "grad_norm": 0.3322184681892395,
+      "learning_rate": 0.00036739782166639556,
+      "loss": 0.46367114782333374,
+      "step": 147
+    },
+    {
+      "epoch": 1.0649819494584838,
+      "grad_norm": 0.267844021320343,
+      "learning_rate": 0.00036694696277679367,
+      "loss": 0.33760106563568115,
+      "step": 148
+    },
+    {
+      "epoch": 1.0722021660649819,
+      "grad_norm": 0.3240787386894226,
+      "learning_rate": 0.0003664932883920182,
+      "loss": 0.43709027767181396,
+      "step": 149
+    },
+    {
+      "epoch": 1.0794223826714802,
+      "grad_norm": 0.29441356658935547,
+      "learning_rate": 0.00036603680616311015,
+      "loss": 0.4205693304538727,
+      "step": 150
+    },
+    {
+      "epoch": 1.0866425992779782,
+      "grad_norm": 0.33195292949676514,
+      "learning_rate": 0.00036557752378846357,
+      "loss": 0.44472751021385193,
+      "step": 151
+    },
+    {
+      "epoch": 1.0938628158844765,
+      "grad_norm": 0.3113086223602295,
+      "learning_rate": 0.00036511544901369597,
+      "loss": 0.3894999921321869,
+      "step": 152
+    },
+    {
+      "epoch": 1.1010830324909748,
+      "grad_norm": 0.30926311016082764,
+      "learning_rate": 0.00036465058963151744,
+      "loss": 0.33893927931785583,
+      "step": 153
+    },
+    {
+      "epoch": 1.108303249097473,
+      "grad_norm": 0.31725767254829407,
+      "learning_rate": 0.00036418295348159967,
+      "loss": 0.4384872019290924,
+      "step": 154
+    },
+    {
+      "epoch": 1.1155234657039712,
+      "grad_norm": 0.3411625027656555,
+      "learning_rate": 0.00036371254845044297,
+      "loss": 0.4130849838256836,
+      "step": 155
+    },
+    {
+      "epoch": 1.1227436823104693,
+      "grad_norm": 0.3470022976398468,
+      "learning_rate": 0.0003632393824712444,
+      "loss": 0.4436883330345154,
+      "step": 156
+    },
+    {
+      "epoch": 1.1299638989169676,
+      "grad_norm": 0.31315702199935913,
+      "learning_rate": 0.00036276346352376266,
+      "loss": 0.39922523498535156,
+      "step": 157
+    },
+    {
+      "epoch": 1.1371841155234657,
+      "grad_norm": 0.36140701174736023,
+      "learning_rate": 0.00036228479963418465,
+      "loss": 0.44266432523727417,
+      "step": 158
+    },
+    {
+      "epoch": 1.144404332129964,
+      "grad_norm": 0.37621912360191345,
+      "learning_rate": 0.0003618033988749895,
+      "loss": 0.3703776001930237,
+      "step": 159
+    },
+    {
+      "epoch": 1.151624548736462,
+      "grad_norm": 0.40457242727279663,
+      "learning_rate": 0.0003613192693648125,
+      "loss": 0.48410341143608093,
+      "step": 160
+    },
+    {
+      "epoch": 1.1588447653429603,
+      "grad_norm": 0.35907864570617676,
+      "learning_rate": 0.00036083241926830833,
+      "loss": 0.3249608874320984,
+      "step": 161
+    },
+    {
+      "epoch": 1.1660649819494584,
+      "grad_norm": 0.3328985869884491,
+      "learning_rate": 0.00036034285679601336,
+      "loss": 0.3642440140247345,
+      "step": 162
+    },
+    {
+      "epoch": 1.1732851985559567,
+      "grad_norm": 0.3639651834964752,
+      "learning_rate": 0.00035985059020420695,
+      "loss": 0.4125223457813263,
+      "step": 163
+    },
+    {
+      "epoch": 1.1805054151624548,
+      "grad_norm": 0.3122009038925171,
+      "learning_rate": 0.0003593556277947725,
+      "loss": 0.26333218812942505,
+      "step": 164
+    },
+    {
+      "epoch": 1.187725631768953,
+      "grad_norm": 0.3917737901210785,
+      "learning_rate": 0.0003588579779150572,
+      "loss": 0.41683149337768555,
+      "step": 165
+    },
+    {
+      "epoch": 1.1949458483754514,
+      "grad_norm": 0.3892008066177368,
+      "learning_rate": 0.00035835764895773164,
+      "loss": 0.34965166449546814,
+      "step": 166
+    },
+    {
+      "epoch": 1.2021660649819494,
+      "grad_norm": 0.32206663489341736,
+      "learning_rate": 0.0003578546493606477,
+      "loss": 0.3289134204387665,
+      "step": 167
+    },
+    {
+      "epoch": 1.2093862815884477,
+      "grad_norm": 0.373098760843277,
+      "learning_rate": 0.0003573489876066967,
+      "loss": 0.37044239044189453,
+      "step": 168
+    },
+    {
+      "epoch": 1.2166064981949458,
+      "grad_norm": 0.4301343262195587,
+      "learning_rate": 0.00035684067222366614,
+      "loss": 0.46339157223701477,
+      "step": 169
+    },
+    {
+      "epoch": 1.2238267148014441,
+      "grad_norm": 0.3458724915981293,
+      "learning_rate": 0.00035632971178409603,
+      "loss": 0.41736456751823425,
+      "step": 170
+    },
+    {
+      "epoch": 1.2310469314079422,
+      "grad_norm": 0.45238223671913147,
+      "learning_rate": 0.0003558161149051341,
+      "loss": 0.4182761311531067,
+      "step": 171
+    },
+    {
+      "epoch": 1.2382671480144405,
+      "grad_norm": 0.32411810755729675,
+      "learning_rate": 0.0003552998902483907,
+      "loss": 0.30536049604415894,
+      "step": 172
+    },
+    {
+      "epoch": 1.2454873646209386,
+      "grad_norm": 0.40648043155670166,
+      "learning_rate": 0.0003547810465197926,
+      "loss": 0.45940107107162476,
+      "step": 173
+    },
+    {
+      "epoch": 1.2527075812274369,
+      "grad_norm": 0.34474775195121765,
+      "learning_rate": 0.00035425959246943614,
+      "loss": 0.4267578423023224,
+      "step": 174
+    },
+    {
+      "epoch": 1.259927797833935,
+      "grad_norm": 0.33473461866378784,
+      "learning_rate": 0.00035373553689143977,
+      "loss": 0.31750980019569397,
+      "step": 175
+    },
+    {
+      "epoch": 1.2671480144404332,
+      "grad_norm": 0.29968079924583435,
+      "learning_rate": 0.0003532088886237956,
+      "loss": 0.3412264883518219,
+      "step": 176
+    },
+    {
+      "epoch": 1.2743682310469313,
+      "grad_norm": 0.3362952172756195,
+      "learning_rate": 0.00035267965654822063,
+      "loss": 0.347223162651062,
+      "step": 177
+    },
+    {
+      "epoch": 1.2815884476534296,
+      "grad_norm": 0.38441580533981323,
+      "learning_rate": 0.0003521478495900065,
+      "loss": 0.4132072925567627,
+      "step": 178
+    },
+    {
+      "epoch": 1.288808664259928,
+      "grad_norm": 0.3259715437889099,
+      "learning_rate": 0.00035161347671786947,
+      "loss": 0.3259287476539612,
+      "step": 179
+    },
+    {
+      "epoch": 1.296028880866426,
+      "grad_norm": 0.30564218759536743,
+      "learning_rate": 0.00035107654694379876,
+      "loss": 0.33284130692481995,
+      "step": 180
+    },
+    {
+      "epoch": 1.303249097472924,
+      "grad_norm": 0.3464779853820801,
+      "learning_rate": 0.0003505370693229049,
+      "loss": 0.3613452613353729,
+      "step": 181
+    },
+    {
+      "epoch": 1.3104693140794224,
+      "grad_norm": 0.4113422632217407,
+      "learning_rate": 0.0003499950529532668,
+      "loss": 0.49952858686447144,
+      "step": 182
+    },
+    {
+      "epoch": 1.3176895306859207,
+      "grad_norm": 0.3503414988517761,
+      "learning_rate": 0.0003494505069757782,
+      "loss": 0.43928611278533936,
+      "step": 183
+    },
+    {
+      "epoch": 1.3249097472924187,
+      "grad_norm": 0.3231402337551117,
+      "learning_rate": 0.000348903440573994,
+      "loss": 0.35302072763442993,
+      "step": 184
+    },
+    {
+      "epoch": 1.332129963898917,
+      "grad_norm": 0.3849830627441406,
+      "learning_rate": 0.00034835386297397486,
+      "loss": 0.42382097244262695,
+      "step": 185
+    },
+    {
+      "epoch": 1.339350180505415,
+      "grad_norm": 0.35116714239120483,
+      "learning_rate": 0.00034780178344413185,
+      "loss": 0.3274853229522705,
+      "step": 186
+    },
+    {
+      "epoch": 1.3465703971119134,
+      "grad_norm": 0.3941578269004822,
+      "learning_rate": 0.0003472472112950701,
+      "loss": 0.34617680311203003,
+      "step": 187
+    },
+    {
+      "epoch": 1.3537906137184115,
+      "grad_norm": 0.32104700803756714,
+      "learning_rate": 0.00034669015587943184,
+      "loss": 0.39410364627838135,
+      "step": 188
+    },
+    {
+      "epoch": 1.3610108303249098,
+      "grad_norm": 0.42431211471557617,
+      "learning_rate": 0.00034613062659173867,
+      "loss": 0.49896761775016785,
+      "step": 189
+    },
+    {
+      "epoch": 1.3682310469314078,
+      "grad_norm": 0.4387979209423065,
+      "learning_rate": 0.00034556863286823284,
+      "loss": 0.4435005784034729,
+      "step": 190
+    },
+    {
+      "epoch": 1.3754512635379061,
+      "grad_norm": 0.3983216881752014,
+      "learning_rate": 0.00034500418418671855,
+      "loss": 0.44415876269340515,
+      "step": 191
+    },
+    {
+      "epoch": 1.3826714801444044,
+      "grad_norm": 0.30740320682525635,
+      "learning_rate": 0.00034443729006640184,
+      "loss": 0.36664414405822754,
+      "step": 192
+    },
+    {
+      "epoch": 1.3898916967509025,
+      "grad_norm": 0.35397860407829285,
+      "learning_rate": 0.0003438679600677302,
+      "loss": 0.36680418252944946,
+      "step": 193
+    },
+    {
+      "epoch": 1.3971119133574006,
+      "grad_norm": 0.3522591292858124,
+      "learning_rate": 0.0003432962037922312,
+      "loss": 0.3908054232597351,
+      "step": 194
+    },
+    {
+      "epoch": 1.404332129963899,
+      "grad_norm": 0.36350661516189575,
+      "learning_rate": 0.0003427220308823505,
+      "loss": 0.34363672137260437,
+      "step": 195
+    },
+    {
+      "epoch": 1.4115523465703972,
+      "grad_norm": 0.3392064869403839,
+      "learning_rate": 0.00034214545102128946,
+      "loss": 0.40172988176345825,
+      "step": 196
+    },
+    {
+      "epoch": 1.4187725631768953,
+      "grad_norm": 0.3826808035373688,
+      "learning_rate": 0.0003415664739328418,
+      "loss": 0.4292600452899933,
+      "step": 197
+    },
+    {
+      "epoch": 1.4259927797833936,
+      "grad_norm": 0.4102179706096649,
+      "learning_rate": 0.00034098510938122947,
+      "loss": 0.40238481760025024,
+      "step": 198
+    },
+    {
+      "epoch": 1.4332129963898916,
+      "grad_norm": 0.38712596893310547,
+      "learning_rate": 0.00034040136717093796,
+      "loss": 0.3980088233947754,
+      "step": 199
+    },
+    {
+      "epoch": 1.44043321299639,
+      "grad_norm": 0.35182803869247437,
+      "learning_rate": 0.0003398152571465512,
+      "loss": 0.3569096028804779,
+      "step": 200
+    },
+    {
+      "epoch": 1.447653429602888,
+      "grad_norm": 0.4129750728607178,
+      "learning_rate": 0.0003392267891925854,
+      "loss": 0.4729324281215668,
+      "step": 201
+    },
+    {
+      "epoch": 1.4548736462093863,
+      "grad_norm": 0.3427615463733673,
+      "learning_rate": 0.000338635973233322,
+      "loss": 0.38466876745224,
+      "step": 202
+    },
+    {
+      "epoch": 1.4620938628158844,
+      "grad_norm": 0.29937732219696045,
+      "learning_rate": 0.0003380428192326411,
+      "loss": 0.24431845545768738,
+      "step": 203
+    },
+    {
+      "epoch": 1.4693140794223827,
+      "grad_norm": 0.35563215613365173,
+      "learning_rate": 0.00033744733719385253,
+      "loss": 0.34825772047042847,
+      "step": 204
+    },
+    {
+      "epoch": 1.476534296028881,
+      "grad_norm": 0.3395077586174011,
+      "learning_rate": 0.0003368495371595279,
+      "loss": 0.37704023718833923,
+      "step": 205
+    },
+    {
+      "epoch": 1.483754512635379,
+      "grad_norm": 0.32358789443969727,
+      "learning_rate": 0.0003362494292113308,
+      "loss": 0.29866379499435425,
+      "step": 206
+    },
+    {
+      "epoch": 1.4909747292418771,
+      "grad_norm": 0.44219517707824707,
+      "learning_rate": 0.00033564702346984684,
+      "loss": 0.5036117434501648,
+      "step": 207
+    },
+    {
+      "epoch": 1.4981949458483754,
+      "grad_norm": 0.358549565076828,
+      "learning_rate": 0.00033504233009441287,
+      "loss": 0.42227745056152344,
+      "step": 208
+    },
+    {
+      "epoch": 1.5054151624548737,
+      "grad_norm": 0.3784799575805664,
+      "learning_rate": 0.0003344353592829461,
+      "loss": 0.4267178773880005,
+      "step": 209
+    },
+    {
+      "epoch": 1.5126353790613718,
+      "grad_norm": 0.3619599938392639,
+      "learning_rate": 0.00033382612127177166,
+      "loss": 0.4075395166873932,
+      "step": 210
+    },
+    {
+      "epoch": 1.5198555956678699,
+      "grad_norm": 0.44484809041023254,
+      "learning_rate": 0.0003332146263354501,
+      "loss": 0.5016494989395142,
+      "step": 211
+    },
+    {
+      "epoch": 1.5270758122743682,
+      "grad_norm": 0.34379443526268005,
+      "learning_rate": 0.00033260088478660407,
+      "loss": 0.317508339881897,
+      "step": 212
+    },
+    {
+      "epoch": 1.5342960288808665,
+      "grad_norm": 0.4008065164089203,
+      "learning_rate": 0.0003319849069757446,
+      "loss": 0.3974913954734802,
+      "step": 213
+    },
+    {
+      "epoch": 1.5415162454873648,
+      "grad_norm": 0.4006946384906769,
+      "learning_rate": 0.00033136670329109624,
+      "loss": 0.5162532925605774,
+      "step": 214
+    },
+    {
+      "epoch": 1.5487364620938628,
+      "grad_norm": 0.32255908846855164,
+      "learning_rate": 0.0003307462841584223,
+      "loss": 0.37669965624809265,
+      "step": 215
+    },
+    {
+      "epoch": 1.555956678700361,
+      "grad_norm": 0.7169090509414673,
+      "learning_rate": 0.0003301236600408484,
+      "loss": 0.4477896988391876,
+      "step": 216
+    },
+    {
+      "epoch": 1.5631768953068592,
+      "grad_norm": 0.3694959580898285,
+      "learning_rate": 0.00032949884143868675,
+      "loss": 0.45744967460632324,
+      "step": 217
+    },
+    {
+      "epoch": 1.5703971119133575,
+      "grad_norm": 0.3709769546985626,
+      "learning_rate": 0.0003288718388892583,
+      "loss": 0.4952532947063446,
+      "step": 218
+    },
+    {
+      "epoch": 1.5776173285198556,
+      "grad_norm": 0.29159069061279297,
+      "learning_rate": 0.00032824266296671567,
+      "loss": 0.28994375467300415,
+      "step": 219
+    },
+    {
+      "epoch": 1.5848375451263537,
+      "grad_norm": 0.3600374460220337,
+      "learning_rate": 0.0003276113242818645,
+      "loss": 0.43825557827949524,
+      "step": 220
+    },
+    {
+      "epoch": 1.592057761732852,
+      "grad_norm": 0.2843645215034485,
+      "learning_rate": 0.0003269778334819846,
+      "loss": 0.32124435901641846,
+      "step": 221
+    },
+    {
+      "epoch": 1.5992779783393503,
+      "grad_norm": 0.3688318431377411,
+      "learning_rate": 0.0003263422012506502,
+      "loss": 0.4742942750453949,
+      "step": 222
+    },
+    {
+      "epoch": 1.6064981949458483,
+      "grad_norm": 0.3213002383708954,
+      "learning_rate": 0.00032570443830755015,
+      "loss": 0.42100387811660767,
+      "step": 223
+    },
+    {
+      "epoch": 1.6137184115523464,
+      "grad_norm": 0.3240319490432739,
+      "learning_rate": 0.0003250645554083068,
+      "loss": 0.29403799772262573,
+      "step": 224
+    },
+    {
+      "epoch": 1.6209386281588447,
+      "grad_norm": 0.3569975197315216,
+      "learning_rate": 0.0003244225633442948,
+      "loss": 0.4555712938308716,
+      "step": 225
+    },
+    {
+      "epoch": 1.628158844765343,
+      "grad_norm": 0.43396008014678955,
+      "learning_rate": 0.00032377847294245895,
+      "loss": 0.4877527356147766,
+      "step": 226
+    },
+    {
+      "epoch": 1.6353790613718413,
+      "grad_norm": 0.3694853186607361,
+      "learning_rate": 0.0003231322950651316,
+      "loss": 0.41299888491630554,
+      "step": 227
+    },
+    {
+      "epoch": 1.6425992779783394,
+      "grad_norm": 0.3358525335788727,
+      "learning_rate": 0.00032248404060985,
+      "loss": 0.4551587998867035,
+      "step": 228
+    },
+    {
+      "epoch": 1.6498194945848375,
+      "grad_norm": 0.38492903113365173,
+      "learning_rate": 0.0003218337205091715,
+      "loss": 0.4316423535346985,
+      "step": 229
+    },
+    {
+      "epoch": 1.6570397111913358,
+      "grad_norm": 0.34664952754974365,
+      "learning_rate": 0.0003211813457304902,
+      "loss": 0.39314523339271545,
+      "step": 230
+    },
+    {
+      "epoch": 1.664259927797834,
+      "grad_norm": 0.3254722058773041,
+      "learning_rate": 0.00032052692727585133,
+      "loss": 0.3327226936817169,
+      "step": 231
+    },
+    {
+      "epoch": 1.6714801444043321,
+      "grad_norm": 0.4386709928512573,
+      "learning_rate": 0.0003198704761817658,
+      "loss": 0.4558185935020447,
+      "step": 232
+    },
+    {
+      "epoch": 1.6787003610108302,
+      "grad_norm": 0.5447010397911072,
+      "learning_rate": 0.00031921200351902446,
+      "loss": 0.48867422342300415,
+      "step": 233
+    },
+    {
+      "epoch": 1.6859205776173285,
+      "grad_norm": 0.2976516783237457,
+      "learning_rate": 0.000318551520392511,
+      "loss": 0.3376544117927551,
+      "step": 234
+    },
+    {
+      "epoch": 1.6931407942238268,
+      "grad_norm": 0.3854595124721527,
+      "learning_rate": 0.00031788903794101477,
+      "loss": 0.45083147287368774,
+      "step": 235
+    },
+    {
+      "epoch": 1.7003610108303249,
+      "grad_norm": 0.37051668763160706,
+      "learning_rate": 0.00031722456733704297,
+      "loss": 0.4483460485935211,
+      "step": 236
+    },
+    {
+      "epoch": 1.707581227436823,
+      "grad_norm": 0.42628687620162964,
+      "learning_rate": 0.0003165581197866322,
+      "loss": 0.5173429250717163,
+      "step": 237
+    },
+    {
+      "epoch": 1.7148014440433212,
+      "grad_norm": 0.44184717535972595,
+      "learning_rate": 0.0003158897065291596,
+      "loss": 0.45995786786079407,
+      "step": 238
+    },
+    {
+      "epoch": 1.7220216606498195,
+      "grad_norm": 0.3687286376953125,
+      "learning_rate": 0.00031521933883715293,
+      "loss": 0.33321040868759155,
+      "step": 239
+    },
+    {
+      "epoch": 1.7292418772563178,
+      "grad_norm": 0.3514050841331482,
+      "learning_rate": 0.00031454702801610103,
+      "loss": 0.3461155891418457,
+      "step": 240
+    },
+    {
+      "epoch": 1.736462093862816,
+      "grad_norm": 0.4148181676864624,
+      "learning_rate": 0.0003138727854042627,
+      "loss": 0.4323638677597046,
+      "step": 241
+    },
+    {
+      "epoch": 1.743682310469314,
+      "grad_norm": 0.44285446405410767,
+      "learning_rate": 0.0003131966223724756,
+      "loss": 0.4780181646347046,
+      "step": 242
+    },
+    {
+      "epoch": 1.7509025270758123,
+      "grad_norm": 0.3196166157722473,
+      "learning_rate": 0.00031251855032396464,
+      "loss": 0.3380797505378723,
+      "step": 243
+    },
+    {
+      "epoch": 1.7581227436823106,
+      "grad_norm": 0.48847395181655884,
+      "learning_rate": 0.0003118385806941494,
+      "loss": 0.4878506064414978,
+      "step": 244
+    },
+    {
+      "epoch": 1.7653429602888087,
+      "grad_norm": 0.398876816034317,
+      "learning_rate": 0.0003111567249504515,
+      "loss": 0.4532281458377838,
+      "step": 245
+    },
+    {
+      "epoch": 1.7725631768953067,
+      "grad_norm": 0.34218019247055054,
+      "learning_rate": 0.0003104729945921012,
+      "loss": 0.3738666772842407,
+      "step": 246
+    },
+    {
+      "epoch": 1.779783393501805,
+      "grad_norm": 0.4071042835712433,
+      "learning_rate": 0.0003097874011499433,
+      "loss": 0.4437788724899292,
+      "step": 247
+    },
+    {
+      "epoch": 1.7870036101083033,
+      "grad_norm": 0.35893604159355164,
+      "learning_rate": 0.0003090999561862428,
+      "loss": 0.35174691677093506,
+      "step": 248
+    },
+    {
+      "epoch": 1.7942238267148014,
+      "grad_norm": 0.34495967626571655,
+      "learning_rate": 0.0003084106712944899,
+      "loss": 0.34072697162628174,
+      "step": 249
+    },
+    {
+      "epoch": 1.8014440433212995,
+      "grad_norm": 0.4653945565223694,
+      "learning_rate": 0.0003077195580992045,
+      "loss": 0.45023709535598755,
+      "step": 250
+    },
+    {
+      "epoch": 1.8086642599277978,
+      "grad_norm": 0.42124322056770325,
+      "learning_rate": 0.0003070266282557401,
+      "loss": 0.4846741557121277,
+      "step": 251
+    },
+    {
+      "epoch": 1.815884476534296,
+      "grad_norm": 0.38208404183387756,
+      "learning_rate": 0.0003063318934500872,
+      "loss": 0.3900390565395355,
+      "step": 252
+    },
+    {
+      "epoch": 1.8231046931407944,
+      "grad_norm": 0.3911641538143158,
+      "learning_rate": 0.0003056353653986764,
+      "loss": 0.3795917332172394,
+      "step": 253
+    },
+    {
+      "epoch": 1.8303249097472925,
+      "grad_norm": 0.34185168147087097,
+      "learning_rate": 0.00030493705584818065,
+      "loss": 0.39876672625541687,
+      "step": 254
+    },
+    {
+      "epoch": 1.8375451263537905,
+      "grad_norm": 0.4229717552661896,
+      "learning_rate": 0.000304236976575317,
+      "loss": 0.357430100440979,
+      "step": 255
+    },
+    {
+      "epoch": 1.8447653429602888,
+      "grad_norm": 0.35156992077827454,
+      "learning_rate": 0.0003035351393866485,
+      "loss": 0.37869274616241455,
+      "step": 256
+    },
+    {
+      "epoch": 1.8519855595667871,
+      "grad_norm": 0.3945099711418152,
+      "learning_rate": 0.0003028315561183845,
+      "loss": 0.45211949944496155,
+      "step": 257
+    },
+    {
+      "epoch": 1.8592057761732852,
+      "grad_norm": 0.31473544239997864,
+      "learning_rate": 0.0003021262386361814,
+      "loss": 0.32624444365501404,
+      "step": 258
+    },
+    {
+      "epoch": 1.8664259927797833,
+      "grad_norm": 0.4487520158290863,
+      "learning_rate": 0.00030141919883494247,
+      "loss": 0.48553428053855896,
+      "step": 259
+    },
+    {
+      "epoch": 1.8736462093862816,
+      "grad_norm": 0.3847423791885376,
+      "learning_rate": 0.00030071044863861713,
+      "loss": 0.457511842250824,
+      "step": 260
+    },
+    {
+      "epoch": 1.8808664259927799,
+      "grad_norm": 0.3822494447231293,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 0.38700124621391296,
+      "step": 261
+    },
+    {
+      "epoch": 1.888086642599278,
+      "grad_norm": 0.37934961915016174,
+      "learning_rate": 0.0002992878649005293,
+      "loss": 0.4150720238685608,
+      "step": 262
+    },
+    {
+      "epoch": 1.895306859205776,
+      "grad_norm": 0.37126392126083374,
+      "learning_rate": 0.00029857405535008467,
+      "loss": 0.3389851450920105,
+      "step": 263
+    },
+    {
+      "epoch": 1.9025270758122743,
+      "grad_norm": 0.3264472484588623,
+      "learning_rate": 0.0002978585833867847,
+      "loss": 0.327554315328598,
+      "step": 264
+    },
+    {
+      "epoch": 1.9097472924187726,
+      "grad_norm": 0.3889877200126648,
+      "learning_rate": 0.00029714146107678413,
+      "loss": 0.4393918812274933,
+      "step": 265
+    },
+    {
+      "epoch": 1.916967509025271,
+      "grad_norm": 0.4926806092262268,
+      "learning_rate": 0.0002964227005140698,
+      "loss": 0.5468560457229614,
+      "step": 266
+    },
+    {
+      "epoch": 1.924187725631769,
+      "grad_norm": 0.4259951114654541,
+      "learning_rate": 0.0002957023138202573,
+      "loss": 0.39216533303260803,
+      "step": 267
+    },
+    {
+      "epoch": 1.931407942238267,
+      "grad_norm": 0.39791861176490784,
+      "learning_rate": 0.00029498031314438626,
+      "loss": 0.37104716897010803,
+      "step": 268
+    },
+    {
+      "epoch": 1.9386281588447654,
+      "grad_norm": 0.3682475686073303,
+      "learning_rate": 0.0002942567106627155,
+      "loss": 0.41986221075057983,
+      "step": 269
+    },
+    {
+      "epoch": 1.9458483754512637,
+      "grad_norm": 0.3230769634246826,
+      "learning_rate": 0.00029353151857851735,
+      "loss": 0.35548532009124756,
+      "step": 270
+    },
+    {
+      "epoch": 1.9530685920577617,
+      "grad_norm": 0.4607864320278168,
+      "learning_rate": 0.0002928047491218727,
+      "loss": 0.4246275722980499,
+      "step": 271
+    },
+    {
+      "epoch": 1.9602888086642598,
+      "grad_norm": 0.4302678108215332,
+      "learning_rate": 0.0002920764145494638,
+      "loss": 0.42147913575172424,
+      "step": 272
+    },
+    {
+      "epoch": 1.967509025270758,
+      "grad_norm": 0.36212071776390076,
+      "learning_rate": 0.00029134652714436814,
+      "loss": 0.4659682512283325,
+      "step": 273
+    },
+    {
+      "epoch": 1.9747292418772564,
+      "grad_norm": 0.31535375118255615,
+      "learning_rate": 0.00029061509921585107,
+      "loss": 0.38229790329933167,
+      "step": 274
+    },
+    {
+      "epoch": 1.9819494584837545,
+      "grad_norm": 0.3685518801212311,
+      "learning_rate": 0.0002898821430991582,
+      "loss": 0.46467846632003784,
+      "step": 275
+    },
+    {
+      "epoch": 1.9891696750902526,
+      "grad_norm": 0.3998563885688782,
+      "learning_rate": 0.0002891476711553077,
+      "loss": 0.42970407009124756,
+      "step": 276
+    },
+    {
+      "epoch": 1.9963898916967509,
+      "grad_norm": 0.35800614953041077,
+      "learning_rate": 0.00028841169577088115,
+      "loss": 0.4408795237541199,
+      "step": 277
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5006654858589172,
+      "learning_rate": 0.0002876742293578155,
+      "loss": 0.38447684049606323,
+      "step": 278
+    },
+    {
+      "epoch": 2.0072202166064983,
+      "grad_norm": 0.31330448389053345,
+      "learning_rate": 0.00028693528435319305,
+      "loss": 0.30373793840408325,
+      "step": 279
+    },
+    {
+      "epoch": 2.0144404332129966,
+      "grad_norm": 0.29946643114089966,
+      "learning_rate": 0.0002861948732190319,
+      "loss": 0.26699960231781006,
+      "step": 280
+    },
+    {
+      "epoch": 2.0216606498194944,
+      "grad_norm": 0.26591944694519043,
+      "learning_rate": 0.0002854530084420762,
+      "loss": 0.23807109892368317,
+      "step": 281
+    },
+    {
+      "epoch": 2.0288808664259927,
+      "grad_norm": 0.34681931138038635,
+      "learning_rate": 0.00028470970253358487,
+      "loss": 0.32200539112091064,
+      "step": 282
+    },
+    {
+      "epoch": 2.036101083032491,
+      "grad_norm": 0.3542478382587433,
+      "learning_rate": 0.0002839649680291211,
+      "loss": 0.3349723815917969,
+      "step": 283
+    },
+    {
+      "epoch": 2.0433212996389893,
+      "grad_norm": 0.3007175624370575,
+      "learning_rate": 0.0002832188174883408,
+      "loss": 0.2172810435295105,
+      "step": 284
+    },
+    {
+      "epoch": 2.050541516245487,
+      "grad_norm": 0.381719708442688,
+      "learning_rate": 0.00028247126349478075,
+      "loss": 0.3058876693248749,
+      "step": 285
+    },
+    {
+      "epoch": 2.0577617328519855,
+      "grad_norm": 0.4056151211261749,
+      "learning_rate": 0.0002817223186556463,
+      "loss": 0.25402823090553284,
+      "step": 286
+    },
+    {
+      "epoch": 2.064981949458484,
+      "grad_norm": 0.5069922804832458,
+      "learning_rate": 0.00028097199560159913,
+      "loss": 0.2375117838382721,
+      "step": 287
+    },
+    {
+      "epoch": 2.072202166064982,
+      "grad_norm": 0.574663519859314,
+      "learning_rate": 0.00028022030698654374,
+      "loss": 0.3134710490703583,
+      "step": 288
+    },
+    {
+      "epoch": 2.07942238267148,
+      "grad_norm": 0.5239531993865967,
+      "learning_rate": 0.00027946726548741443,
+      "loss": 0.20423674583435059,
+      "step": 289
+    },
+    {
+      "epoch": 2.0866425992779782,
+      "grad_norm": 0.5845417976379395,
+      "learning_rate": 0.0002787128838039612,
+      "loss": 0.3297584354877472,
+      "step": 290
+    },
+    {
+      "epoch": 2.0938628158844765,
+      "grad_norm": 0.47120949625968933,
+      "learning_rate": 0.00027795717465853585,
+      "loss": 0.32672739028930664,
+      "step": 291
+    },
+    {
+      "epoch": 2.101083032490975,
+      "grad_norm": 0.40136632323265076,
+      "learning_rate": 0.00027720015079587743,
+      "loss": 0.23871661722660065,
+      "step": 292
+    },
+    {
+      "epoch": 2.108303249097473,
+      "grad_norm": 0.5156406164169312,
+      "learning_rate": 0.00027644182498289677,
+      "loss": 0.23360338807106018,
+      "step": 293
+    },
+    {
+      "epoch": 2.115523465703971,
+      "grad_norm": 0.4302510917186737,
+      "learning_rate": 0.0002756822100084621,
+      "loss": 0.31100887060165405,
+      "step": 294
+    },
+    {
+      "epoch": 2.1227436823104693,
+      "grad_norm": 0.3673456609249115,
+      "learning_rate": 0.00027492131868318246,
+      "loss": 0.21802300214767456,
+      "step": 295
+    },
+    {
+      "epoch": 2.1299638989169676,
+      "grad_norm": 0.32138535380363464,
+      "learning_rate": 0.00027415916383919216,
+      "loss": 0.19008119404315948,
+      "step": 296
+    },
+    {
+      "epoch": 2.137184115523466,
+      "grad_norm": 0.3203730285167694,
+      "learning_rate": 0.00027339575832993444,
+      "loss": 0.1889929622411728,
+      "step": 297
+    },
+    {
+      "epoch": 2.1444043321299637,
+      "grad_norm": 0.4612288773059845,
+      "learning_rate": 0.0002726311150299443,
+      "loss": 0.22323200106620789,
+      "step": 298
+    },
+    {
+      "epoch": 2.151624548736462,
+      "grad_norm": 0.36709311604499817,
+      "learning_rate": 0.00027186524683463155,
+      "loss": 0.20466932654380798,
+      "step": 299
+    },
+    {
+      "epoch": 2.1588447653429603,
+      "grad_norm": 0.4253864586353302,
+      "learning_rate": 0.0002710981666600636,
+      "loss": 0.2431459128856659,
+      "step": 300
+    },
+    {
+      "epoch": 2.1660649819494586,
+      "grad_norm": 0.4018993675708771,
+      "learning_rate": 0.00027032988744274736,
+      "loss": 0.2751820385456085,
+      "step": 301
+    },
+    {
+      "epoch": 2.1732851985559565,
+      "grad_norm": 0.4428861737251282,
+      "learning_rate": 0.000269560422139411,
+      "loss": 0.269814133644104,
+      "step": 302
+    },
+    {
+      "epoch": 2.1805054151624548,
+      "grad_norm": 0.47122061252593994,
+      "learning_rate": 0.00026878978372678564,
+      "loss": 0.22829411923885345,
+      "step": 303
+    },
+    {
+      "epoch": 2.187725631768953,
+      "grad_norm": 0.4426339268684387,
+      "learning_rate": 0.00026801798520138646,
+      "loss": 0.2637779414653778,
+      "step": 304
+    },
+    {
+      "epoch": 2.1949458483754514,
+      "grad_norm": 0.4073690176010132,
+      "learning_rate": 0.0002672450395792935,
+      "loss": 0.1904142051935196,
+      "step": 305
+    },
+    {
+      "epoch": 2.2021660649819497,
+      "grad_norm": 0.4896828234195709,
+      "learning_rate": 0.00026647095989593193,
+      "loss": 0.34650635719299316,
+      "step": 306
+    },
+    {
+      "epoch": 2.2093862815884475,
+      "grad_norm": 0.48900970816612244,
+      "learning_rate": 0.0002656957592058528,
+      "loss": 0.2542106807231903,
+      "step": 307
+    },
+    {
+      "epoch": 2.216606498194946,
+      "grad_norm": 0.4600171446800232,
+      "learning_rate": 0.0002649194505825121,
+      "loss": 0.21008341014385223,
+      "step": 308
+    },
+    {
+      "epoch": 2.223826714801444,
+      "grad_norm": 0.43380653858184814,
+      "learning_rate": 0.00026414204711805103,
+      "loss": 0.2683791518211365,
+      "step": 309
+    },
+    {
+      "epoch": 2.2310469314079424,
+      "grad_norm": 0.5792773365974426,
+      "learning_rate": 0.0002633635619230746,
+      "loss": 0.33160915970802307,
+      "step": 310
+    },
+    {
+      "epoch": 2.2382671480144403,
+      "grad_norm": 0.5626965165138245,
+      "learning_rate": 0.0002625840081264309,
+      "loss": 0.26700398325920105,
+      "step": 311
+    },
+    {
+      "epoch": 2.2454873646209386,
+      "grad_norm": 0.4952887296676636,
+      "learning_rate": 0.00026180339887498953,
+      "loss": 0.2689879536628723,
+      "step": 312
+    },
+    {
+      "epoch": 2.252707581227437,
+      "grad_norm": 0.4501405656337738,
+      "learning_rate": 0.0002610217473334199,
+      "loss": 0.26021096110343933,
+      "step": 313
+    },
+    {
+      "epoch": 2.259927797833935,
+      "grad_norm": 0.36552777886390686,
+      "learning_rate": 0.00026023906668396933,
+      "loss": 0.26156651973724365,
+      "step": 314
+    },
+    {
+      "epoch": 2.2671480144404335,
+      "grad_norm": 0.3733731806278229,
+      "learning_rate": 0.00025945537012624054,
+      "loss": 0.21406950056552887,
+      "step": 315
+    },
+    {
+      "epoch": 2.2743682310469313,
+      "grad_norm": 0.4240867495536804,
+      "learning_rate": 0.00025867067087696923,
+      "loss": 0.32105109095573425,
+      "step": 316
+    },
+    {
+      "epoch": 2.2815884476534296,
+      "grad_norm": 0.4198870360851288,
+      "learning_rate": 0.000257884982169801,
+      "loss": 0.3047807216644287,
+      "step": 317
+    },
+    {
+      "epoch": 2.288808664259928,
+      "grad_norm": 0.4444144666194916,
+      "learning_rate": 0.00025709831725506845,
+      "loss": 0.23268437385559082,
+      "step": 318
+    },
+    {
+      "epoch": 2.2960288808664258,
+      "grad_norm": 0.41258376836776733,
+      "learning_rate": 0.00025631068939956726,
+      "loss": 0.26838576793670654,
+      "step": 319
+    },
+    {
+      "epoch": 2.303249097472924,
+      "grad_norm": 0.5179259777069092,
+      "learning_rate": 0.00025552211188633293,
+      "loss": 0.2882261574268341,
+      "step": 320
+    },
+    {
+      "epoch": 2.3104693140794224,
+      "grad_norm": 0.4838998019695282,
+      "learning_rate": 0.0002547325980144166,
+      "loss": 0.30366814136505127,
+      "step": 321
+    },
+    {
+      "epoch": 2.3176895306859207,
+      "grad_norm": 0.4862940311431885,
+      "learning_rate": 0.0002539421610986605,
+      "loss": 0.29283270239830017,
+      "step": 322
+    },
+    {
+      "epoch": 2.324909747292419,
+      "grad_norm": 0.441685289144516,
+      "learning_rate": 0.000253150814469474,
+      "loss": 0.22393643856048584,
+      "step": 323
+    },
+    {
+      "epoch": 2.332129963898917,
+      "grad_norm": 0.39844486117362976,
+      "learning_rate": 0.0002523585714726081,
+      "loss": 0.17744283378124237,
+      "step": 324
+    },
+    {
+      "epoch": 2.339350180505415,
+      "grad_norm": 0.4977899491786957,
+      "learning_rate": 0.0002515654454689307,
+      "loss": 0.24069495499134064,
+      "step": 325
+    },
+    {
+      "epoch": 2.3465703971119134,
+      "grad_norm": 0.4101859927177429,
+      "learning_rate": 0.0002507714498342016,
+      "loss": 0.25044453144073486,
+      "step": 326
+    },
+    {
+      "epoch": 2.3537906137184117,
+      "grad_norm": 0.3521929085254669,
+      "learning_rate": 0.0002499765979588462,
+      "loss": 0.20651675760746002,
+      "step": 327
+    },
+    {
+      "epoch": 2.3610108303249095,
+      "grad_norm": 0.5269883871078491,
+      "learning_rate": 0.00024918090324773024,
+      "loss": 0.301647424697876,
+      "step": 328
+    },
+    {
+      "epoch": 2.368231046931408,
+      "grad_norm": 0.47672051191329956,
+      "learning_rate": 0.0002483843791199335,
+      "loss": 0.23506474494934082,
+      "step": 329
+    },
+    {
+      "epoch": 2.375451263537906,
+      "grad_norm": 0.5546545386314392,
+      "learning_rate": 0.00024758703900852375,
+      "loss": 0.3066970705986023,
+      "step": 330
+    },
+    {
+      "epoch": 2.3826714801444044,
+      "grad_norm": 0.3792511820793152,
+      "learning_rate": 0.0002467888963603298,
+      "loss": 0.20687991380691528,
+      "step": 331
+    },
+    {
+      "epoch": 2.3898916967509027,
+      "grad_norm": 0.4949963092803955,
+      "learning_rate": 0.00024598996463571474,
+      "loss": 0.2865826189517975,
+      "step": 332
+    },
+    {
+      "epoch": 2.3971119133574006,
+      "grad_norm": 0.4250241219997406,
+      "learning_rate": 0.00024519025730834954,
+      "loss": 0.25499409437179565,
+      "step": 333
+    },
+    {
+      "epoch": 2.404332129963899,
+      "grad_norm": 0.30932316184043884,
+      "learning_rate": 0.00024438978786498524,
+      "loss": 0.18888786435127258,
+      "step": 334
+    },
+    {
+      "epoch": 2.411552346570397,
+      "grad_norm": 0.5183025002479553,
+      "learning_rate": 0.00024358856980522556,
+      "loss": 0.2950358986854553,
+      "step": 335
+    },
+    {
+      "epoch": 2.4187725631768955,
+      "grad_norm": 0.46398693323135376,
+      "learning_rate": 0.00024278661664129948,
+      "loss": 0.28177422285079956,
+      "step": 336
+    },
+    {
+      "epoch": 2.4259927797833933,
+      "grad_norm": 0.4938473701477051,
+      "learning_rate": 0.00024198394189783317,
+      "loss": 0.31616371870040894,
+      "step": 337
+    },
+    {
+      "epoch": 2.4332129963898916,
+      "grad_norm": 0.4439496099948883,
+      "learning_rate": 0.00024118055911162198,
+      "loss": 0.2702808380126953,
+      "step": 338
+    },
+    {
+      "epoch": 2.44043321299639,
+      "grad_norm": 0.45705899596214294,
+      "learning_rate": 0.00024037648183140205,
+      "loss": 0.33506399393081665,
+      "step": 339
+    },
+    {
+      "epoch": 2.4476534296028882,
+      "grad_norm": 0.4123782217502594,
+      "learning_rate": 0.00023957172361762199,
+      "loss": 0.2569669783115387,
+      "step": 340
+    },
+    {
+      "epoch": 2.4548736462093865,
+      "grad_norm": 0.5178057551383972,
+      "learning_rate": 0.00023876629804221402,
+      "loss": 0.2357858568429947,
+      "step": 341
+    },
+    {
+      "epoch": 2.4620938628158844,
+      "grad_norm": 0.5206875205039978,
+      "learning_rate": 0.0002379602186883652,
+      "loss": 0.30847087502479553,
+      "step": 342
+    },
+    {
+      "epoch": 2.4693140794223827,
+      "grad_norm": 0.4264805018901825,
+      "learning_rate": 0.00023715349915028823,
+      "loss": 0.27529531717300415,
+      "step": 343
+    },
+    {
+      "epoch": 2.476534296028881,
+      "grad_norm": 0.5162002444267273,
+      "learning_rate": 0.00023634615303299233,
+      "loss": 0.3068305253982544,
+      "step": 344
+    },
+    {
+      "epoch": 2.483754512635379,
+      "grad_norm": 0.45099541544914246,
+      "learning_rate": 0.00023553819395205378,
+      "loss": 0.2681335508823395,
+      "step": 345
+    },
+    {
+      "epoch": 2.490974729241877,
+      "grad_norm": 0.49566006660461426,
+      "learning_rate": 0.00023472963553338613,
+      "loss": 0.2626950144767761,
+      "step": 346
+    },
+    {
+      "epoch": 2.4981949458483754,
+      "grad_norm": 0.4181882441043854,
+      "learning_rate": 0.00023392049141301055,
+      "loss": 0.2935839891433716,
+      "step": 347
+    },
+    {
+      "epoch": 2.5054151624548737,
+      "grad_norm": 0.4298352897167206,
+      "learning_rate": 0.000233110775236826,
+      "loss": 0.2565048038959503,
+      "step": 348
+    },
+    {
+      "epoch": 2.512635379061372,
+      "grad_norm": 0.46044519543647766,
+      "learning_rate": 0.00023230050066037872,
+      "loss": 0.2740374207496643,
+      "step": 349
+    },
+    {
+      "epoch": 2.51985559566787,
+      "grad_norm": 0.3774242401123047,
+      "learning_rate": 0.00023148968134863233,
+      "loss": 0.23688335716724396,
+      "step": 350
+    },
+    {
+      "epoch": 2.527075812274368,
+      "grad_norm": 0.4179275631904602,
+      "learning_rate": 0.00023067833097573714,
+      "loss": 0.2116960883140564,
+      "step": 351
+    },
+    {
+      "epoch": 2.5342960288808665,
+      "grad_norm": 0.5179049968719482,
+      "learning_rate": 0.0002298664632247994,
+      "loss": 0.2446313500404358,
+      "step": 352
+    },
+    {
+      "epoch": 2.5415162454873648,
+      "grad_norm": 0.44997915625572205,
+      "learning_rate": 0.000229054091787651,
+      "loss": 0.3158394694328308,
+      "step": 353
+    },
+    {
+      "epoch": 2.5487364620938626,
+      "grad_norm": 0.41355574131011963,
+      "learning_rate": 0.0002282412303646183,
+      "loss": 0.2720823884010315,
+      "step": 354
+    },
+    {
+      "epoch": 2.555956678700361,
+      "grad_norm": 0.36182165145874023,
+      "learning_rate": 0.00022742789266429095,
+      "loss": 0.22511911392211914,
+      "step": 355
+    },
+    {
+      "epoch": 2.563176895306859,
+      "grad_norm": 0.43841102719306946,
+      "learning_rate": 0.00022661409240329076,
+      "loss": 0.21212677657604218,
+      "step": 356
+    },
+    {
+      "epoch": 2.5703971119133575,
+      "grad_norm": 0.4933500289916992,
+      "learning_rate": 0.0002257998433060407,
+      "loss": 0.3581770956516266,
+      "step": 357
+    },
+    {
+      "epoch": 2.577617328519856,
+      "grad_norm": 0.4325965344905853,
+      "learning_rate": 0.00022498515910453296,
+      "loss": 0.3252311646938324,
+      "step": 358
+    },
+    {
+      "epoch": 2.5848375451263537,
+      "grad_norm": 0.45968931913375854,
+      "learning_rate": 0.00022417005353809772,
+      "loss": 0.3097744286060333,
+      "step": 359
+    },
+    {
+      "epoch": 2.592057761732852,
+      "grad_norm": 0.43083667755126953,
+      "learning_rate": 0.00022335454035317124,
+      "loss": 0.25429341197013855,
+      "step": 360
+    },
+    {
+      "epoch": 2.5992779783393503,
+      "grad_norm": 0.4504421353340149,
+      "learning_rate": 0.00022253863330306425,
+      "loss": 0.315784215927124,
+      "step": 361
+    },
+    {
+      "epoch": 2.606498194945848,
+      "grad_norm": 0.44943565130233765,
+      "learning_rate": 0.0002217223461477296,
+      "loss": 0.22878926992416382,
+      "step": 362
+    },
+    {
+      "epoch": 2.6137184115523464,
+      "grad_norm": 0.4279639720916748,
+      "learning_rate": 0.00022090569265353072,
+      "loss": 0.2623154819011688,
+      "step": 363
+    },
+    {
+      "epoch": 2.6209386281588447,
+      "grad_norm": 0.4907088279724121,
+      "learning_rate": 0.00022008868659300905,
+      "loss": 0.317205548286438,
+      "step": 364
+    },
+    {
+      "epoch": 2.628158844765343,
+      "grad_norm": 0.49302950501441956,
+      "learning_rate": 0.000219271341744652,
+      "loss": 0.30752891302108765,
+      "step": 365
+    },
+    {
+      "epoch": 2.6353790613718413,
+      "grad_norm": 0.38831761479377747,
+      "learning_rate": 0.00021845367189266042,
+      "loss": 0.27115532755851746,
+      "step": 366
+    },
+    {
+      "epoch": 2.6425992779783396,
+      "grad_norm": 0.47725188732147217,
+      "learning_rate": 0.00021763569082671635,
+      "loss": 0.20296484231948853,
+      "step": 367
+    },
+    {
+      "epoch": 2.6498194945848375,
+      "grad_norm": 0.4198381006717682,
+      "learning_rate": 0.00021681741234175027,
+      "loss": 0.22836127877235413,
+      "step": 368
+    },
+    {
+      "epoch": 2.6570397111913358,
+      "grad_norm": 0.5218875408172607,
+      "learning_rate": 0.00021599885023770835,
+      "loss": 0.29816269874572754,
+      "step": 369
+    },
+    {
+      "epoch": 2.664259927797834,
+      "grad_norm": 0.4321244955062866,
+      "learning_rate": 0.00021518001831932022,
+      "loss": 0.21631377935409546,
+      "step": 370
+    },
+    {
+      "epoch": 2.671480144404332,
+      "grad_norm": 0.4036301374435425,
+      "learning_rate": 0.00021436093039586552,
+      "loss": 0.19903570413589478,
+      "step": 371
+    },
+    {
+      "epoch": 2.67870036101083,
+      "grad_norm": 0.543828010559082,
+      "learning_rate": 0.00021354160028094153,
+      "loss": 0.2922933101654053,
+      "step": 372
+    },
+    {
+      "epoch": 2.6859205776173285,
+      "grad_norm": 0.5146738290786743,
+      "learning_rate": 0.00021272204179222982,
+      "loss": 0.2749359607696533,
+      "step": 373
+    },
+    {
+      "epoch": 2.693140794223827,
+      "grad_norm": 0.47690504789352417,
+      "learning_rate": 0.00021190226875126352,
+      "loss": 0.296896755695343,
+      "step": 374
+    },
+    {
+      "epoch": 2.700361010830325,
+      "grad_norm": 0.545520007610321,
+      "learning_rate": 0.000211082294983194,
+      "loss": 0.27578210830688477,
+      "step": 375
+    },
+    {
+      "epoch": 2.707581227436823,
+      "grad_norm": 0.42757925391197205,
+      "learning_rate": 0.00021026213431655792,
+      "loss": 0.2315436154603958,
+      "step": 376
+    },
+    {
+      "epoch": 2.7148014440433212,
+      "grad_norm": 0.5044121146202087,
+      "learning_rate": 0.0002094418005830439,
+      "loss": 0.34600791335105896,
+      "step": 377
+    },
+    {
+      "epoch": 2.7220216606498195,
+      "grad_norm": 0.5155430436134338,
+      "learning_rate": 0.00020862130761725917,
+      "loss": 0.3247469663619995,
+      "step": 378
+    },
+    {
+      "epoch": 2.729241877256318,
+      "grad_norm": 0.5196051001548767,
+      "learning_rate": 0.00020780066925649634,
+      "loss": 0.3277815878391266,
+      "step": 379
+    },
+    {
+      "epoch": 2.7364620938628157,
+      "grad_norm": 0.44720444083213806,
+      "learning_rate": 0.00020697989934050025,
+      "loss": 0.25156158208847046,
+      "step": 380
+    },
+    {
+      "epoch": 2.743682310469314,
+      "grad_norm": 0.43984708189964294,
+      "learning_rate": 0.00020615901171123412,
+      "loss": 0.28022634983062744,
+      "step": 381
+    },
+    {
+      "epoch": 2.7509025270758123,
+      "grad_norm": 0.47703853249549866,
+      "learning_rate": 0.00020533802021264648,
+      "loss": 0.2780344486236572,
+      "step": 382
+    },
+    {
+      "epoch": 2.7581227436823106,
+      "grad_norm": 0.43746253848075867,
+      "learning_rate": 0.00020451693869043763,
+      "loss": 0.23613542318344116,
+      "step": 383
+    },
+    {
+      "epoch": 2.765342960288809,
+      "grad_norm": 0.41468992829322815,
+      "learning_rate": 0.00020369578099182597,
+      "loss": 0.3185104429721832,
+      "step": 384
+    },
+    {
+      "epoch": 2.7725631768953067,
+      "grad_norm": 0.5149108171463013,
+      "learning_rate": 0.0002028745609653147,
+      "loss": 0.2768135070800781,
+      "step": 385
+    },
+    {
+      "epoch": 2.779783393501805,
+      "grad_norm": 0.47348129749298096,
+      "learning_rate": 0.00020205329246045798,
+      "loss": 0.31962212920188904,
+      "step": 386
+    },
+    {
+      "epoch": 2.7870036101083033,
+      "grad_norm": 0.4738944172859192,
+      "learning_rate": 0.00020123198932762778,
+      "loss": 0.23498407006263733,
+      "step": 387
+    },
+    {
+      "epoch": 2.794223826714801,
+      "grad_norm": 0.48899075388908386,
+      "learning_rate": 0.0002004106654177798,
+      "loss": 0.23857903480529785,
+      "step": 388
+    },
+    {
+      "epoch": 2.8014440433212995,
+      "grad_norm": 0.45867517590522766,
+      "learning_rate": 0.00019958933458222026,
+      "loss": 0.3246588706970215,
+      "step": 389
+    },
+    {
+      "epoch": 2.808664259927798,
+      "grad_norm": 0.5840234160423279,
+      "learning_rate": 0.0001987680106723723,
+      "loss": 0.2935522198677063,
+      "step": 390
+    },
+    {
+      "epoch": 2.815884476534296,
+      "grad_norm": 0.3764658272266388,
+      "learning_rate": 0.00019794670753954204,
+      "loss": 0.2096547782421112,
+      "step": 391
+    },
+    {
+      "epoch": 2.8231046931407944,
+      "grad_norm": 0.44675493240356445,
+      "learning_rate": 0.00019712543903468535,
+      "loss": 0.2988222539424896,
+      "step": 392
+    },
+    {
+      "epoch": 2.8303249097472927,
+      "grad_norm": 0.43063509464263916,
+      "learning_rate": 0.00019630421900817407,
+      "loss": 0.2222481369972229,
+      "step": 393
+    },
+    {
+      "epoch": 2.8375451263537905,
+      "grad_norm": 0.47969475388526917,
+      "learning_rate": 0.00019548306130956242,
+      "loss": 0.26167353987693787,
+      "step": 394
+    },
+    {
+      "epoch": 2.844765342960289,
+      "grad_norm": 0.47235366702079773,
+      "learning_rate": 0.00019466197978735354,
+      "loss": 0.30541300773620605,
+      "step": 395
+    },
+    {
+      "epoch": 2.851985559566787,
+      "grad_norm": 0.47744160890579224,
+      "learning_rate": 0.00019384098828876598,
+      "loss": 0.23645427823066711,
+      "step": 396
+    },
+    {
+      "epoch": 2.859205776173285,
+      "grad_norm": 0.5037774443626404,
+      "learning_rate": 0.00019302010065949988,
+      "loss": 0.26774662733078003,
+      "step": 397
+    },
+    {
+      "epoch": 2.8664259927797833,
+      "grad_norm": 0.4689895510673523,
+      "learning_rate": 0.0001921993307435037,
+      "loss": 0.3405228853225708,
+      "step": 398
+    },
+    {
+      "epoch": 2.8736462093862816,
+      "grad_norm": 0.44282686710357666,
+      "learning_rate": 0.00019137869238274093,
+      "loss": 0.26063668727874756,
+      "step": 399
+    },
+    {
+      "epoch": 2.88086642599278,
+      "grad_norm": 0.4379267990589142,
+      "learning_rate": 0.0001905581994169562,
+      "loss": 0.26478758454322815,
+      "step": 400
+    },
+    {
+      "epoch": 2.888086642599278,
+      "grad_norm": 0.519749641418457,
+      "learning_rate": 0.00018973786568344205,
+      "loss": 0.31215131282806396,
+      "step": 401
+    },
+    {
+      "epoch": 2.895306859205776,
+      "grad_norm": 0.4515770375728607,
+      "learning_rate": 0.000188917705016806,
+      "loss": 0.2881377339363098,
+      "step": 402
+    },
+    {
+      "epoch": 2.9025270758122743,
+      "grad_norm": 0.4650344252586365,
+      "learning_rate": 0.0001880977312487365,
+      "loss": 0.2557958960533142,
+      "step": 403
+    },
+    {
+      "epoch": 2.9097472924187726,
+      "grad_norm": 0.4621173143386841,
+      "learning_rate": 0.0001872779582077702,
+      "loss": 0.23401126265525818,
+      "step": 404
+    },
+    {
+      "epoch": 2.916967509025271,
+      "grad_norm": 0.4094926118850708,
+      "learning_rate": 0.00018645839971905852,
+      "loss": 0.24556361138820648,
+      "step": 405
+    },
+    {
+      "epoch": 2.9241877256317688,
+      "grad_norm": 0.48592710494995117,
+      "learning_rate": 0.0001856390696041345,
+      "loss": 0.27918654680252075,
+      "step": 406
+    },
+    {
+      "epoch": 2.931407942238267,
+      "grad_norm": 0.4039171040058136,
+      "learning_rate": 0.00018481998168067983,
+      "loss": 0.22918914258480072,
+      "step": 407
+    },
+    {
+      "epoch": 2.9386281588447654,
+      "grad_norm": 0.5521522760391235,
+      "learning_rate": 0.0001840011497622917,
+      "loss": 0.35596632957458496,
+      "step": 408
+    },
+    {
+      "epoch": 2.9458483754512637,
+      "grad_norm": 0.5527123212814331,
+      "learning_rate": 0.0001831825876582498,
+      "loss": 0.25526192784309387,
+      "step": 409
+    },
+    {
+      "epoch": 2.953068592057762,
+      "grad_norm": 0.4386238157749176,
+      "learning_rate": 0.0001823643091732837,
+      "loss": 0.2627878189086914,
+      "step": 410
+    },
+    {
+      "epoch": 2.96028880866426,
+      "grad_norm": 0.48318973183631897,
+      "learning_rate": 0.00018154632810733962,
+      "loss": 0.29408901929855347,
+      "step": 411
+    },
+    {
+      "epoch": 2.967509025270758,
+      "grad_norm": 0.5193997025489807,
+      "learning_rate": 0.00018072865825534805,
+      "loss": 0.2762555480003357,
+      "step": 412
+    },
+    {
+      "epoch": 2.9747292418772564,
+      "grad_norm": 0.488490492105484,
+      "learning_rate": 0.000179911313406991,
+      "loss": 0.3119064271450043,
+      "step": 413
+    },
+    {
+      "epoch": 2.9819494584837543,
+      "grad_norm": 0.4461340010166168,
+      "learning_rate": 0.00017909430734646935,
+      "loss": 0.2738839089870453,
+      "step": 414
+    },
+    {
+      "epoch": 2.9891696750902526,
+      "grad_norm": 0.4488433003425598,
+      "learning_rate": 0.0001782776538522704,
+      "loss": 0.25788843631744385,
+      "step": 415
+    },
+    {
+      "epoch": 2.996389891696751,
+      "grad_norm": 0.5179741382598877,
+      "learning_rate": 0.0001774613666969358,
+      "loss": 0.2556946575641632,
+      "step": 416
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.6138741374015808,
+      "learning_rate": 0.00017664545964682878,
+      "loss": 0.23274531960487366,
+      "step": 417
+    },
+    {
+      "epoch": 3.0072202166064983,
+      "grad_norm": 0.27745357155799866,
+      "learning_rate": 0.00017582994646190233,
+      "loss": 0.14754480123519897,
+      "step": 418
+    },
+    {
+      "epoch": 3.0144404332129966,
+      "grad_norm": 0.40759217739105225,
+      "learning_rate": 0.00017501484089546708,
+      "loss": 0.21522817015647888,
+      "step": 419
+    },
+    {
+      "epoch": 3.0216606498194944,
+      "grad_norm": 0.3123788833618164,
+      "learning_rate": 0.00017420015669395938,
+      "loss": 0.15608534216880798,
+      "step": 420
+    },
+    {
+      "epoch": 3.0288808664259927,
+      "grad_norm": 0.4599035978317261,
+      "learning_rate": 0.00017338590759670934,
+      "loss": 0.1309528797864914,
+      "step": 421
+    },
+    {
+      "epoch": 3.036101083032491,
+      "grad_norm": 0.38352325558662415,
+      "learning_rate": 0.00017257210733570916,
+      "loss": 0.1724625676870346,
+      "step": 422
+    },
+    {
+      "epoch": 3.0433212996389893,
+      "grad_norm": 0.44771838188171387,
+      "learning_rate": 0.00017175876963538178,
+      "loss": 0.1661359965801239,
+      "step": 423
+    },
+    {
+      "epoch": 3.050541516245487,
+      "grad_norm": 0.5998123288154602,
+      "learning_rate": 0.00017094590821234906,
+      "loss": 0.1318468153476715,
+      "step": 424
+    },
+    {
+      "epoch": 3.0577617328519855,
+      "grad_norm": 0.47632187604904175,
+      "learning_rate": 0.00017013353677520064,
+      "loss": 0.10464094579219818,
+      "step": 425
+    },
+    {
+      "epoch": 3.064981949458484,
+      "grad_norm": 0.3745962083339691,
+      "learning_rate": 0.00016932166902426288,
+      "loss": 0.10677687078714371,
+      "step": 426
+    },
+    {
+      "epoch": 3.072202166064982,
+      "grad_norm": 0.6340925693511963,
+      "learning_rate": 0.00016851031865136763,
+      "loss": 0.1600916087627411,
+      "step": 427
+    },
+    {
+      "epoch": 3.07942238267148,
+      "grad_norm": 0.8183390498161316,
+      "learning_rate": 0.00016769949933962127,
+      "loss": 0.13513822853565216,
+      "step": 428
+    },
+    {
+      "epoch": 3.0866425992779782,
+      "grad_norm": 0.8068543076515198,
+      "learning_rate": 0.00016688922476317398,
+      "loss": 0.14319300651550293,
+      "step": 429
+    },
+    {
+      "epoch": 3.0938628158844765,
+      "grad_norm": 0.6381204128265381,
+      "learning_rate": 0.00016607950858698946,
+      "loss": 0.1376371830701828,
+      "step": 430
+    },
+    {
+      "epoch": 3.101083032490975,
+      "grad_norm": 0.5576755404472351,
+      "learning_rate": 0.00016527036446661395,
+      "loss": 0.14407199621200562,
+      "step": 431
+    },
+    {
+      "epoch": 3.108303249097473,
+      "grad_norm": 0.6089445948600769,
+      "learning_rate": 0.00016446180604794627,
+      "loss": 0.1575869917869568,
+      "step": 432
+    },
+    {
+      "epoch": 3.115523465703971,
+      "grad_norm": 0.6250684857368469,
+      "learning_rate": 0.0001636538469670077,
+      "loss": 0.17905279994010925,
+      "step": 433
+    },
+    {
+      "epoch": 3.1227436823104693,
+      "grad_norm": 0.4243159592151642,
+      "learning_rate": 0.00016284650084971185,
+      "loss": 0.12401723116636276,
+      "step": 434
+    },
+    {
+      "epoch": 3.1299638989169676,
+      "grad_norm": 0.4369412362575531,
+      "learning_rate": 0.00016203978131163485,
+      "loss": 0.11108125746250153,
+      "step": 435
+    },
+    {
+      "epoch": 3.137184115523466,
+      "grad_norm": 0.45113661885261536,
+      "learning_rate": 0.000161233701957786,
+      "loss": 0.1364113986492157,
+      "step": 436
+    },
+    {
+      "epoch": 3.1444043321299637,
+      "grad_norm": 0.33899277448654175,
+      "learning_rate": 0.00016042827638237806,
+      "loss": 0.1253184676170349,
+      "step": 437
+    },
+    {
+      "epoch": 3.151624548736462,
+      "grad_norm": 0.48023247718811035,
+      "learning_rate": 0.000159623518168598,
+      "loss": 0.1729699671268463,
+      "step": 438
+    },
+    {
+      "epoch": 3.1588447653429603,
+      "grad_norm": 0.4260300397872925,
+      "learning_rate": 0.00015881944088837804,
+      "loss": 0.1630483865737915,
+      "step": 439
+    },
+    {
+      "epoch": 3.1660649819494586,
+      "grad_norm": 0.34896689653396606,
+      "learning_rate": 0.00015801605810216685,
+      "loss": 0.14477582275867462,
+      "step": 440
+    },
+    {
+      "epoch": 3.1732851985559565,
+      "grad_norm": 0.46189841628074646,
+      "learning_rate": 0.00015721338335870057,
+      "loss": 0.16415920853614807,
+      "step": 441
+    },
+    {
+      "epoch": 3.1805054151624548,
+      "grad_norm": 0.40856343507766724,
+      "learning_rate": 0.00015641143019477446,
+      "loss": 0.12028312683105469,
+      "step": 442
+    },
+    {
+      "epoch": 3.187725631768953,
+      "grad_norm": 0.36561018228530884,
+      "learning_rate": 0.00015561021213501483,
+      "loss": 0.12599362432956696,
+      "step": 443
+    },
+    {
+      "epoch": 3.1949458483754514,
+      "grad_norm": 0.4341471493244171,
+      "learning_rate": 0.0001548097426916505,
+      "loss": 0.14446969330310822,
+      "step": 444
+    },
+    {
+      "epoch": 3.2021660649819497,
+      "grad_norm": 0.3864421844482422,
+      "learning_rate": 0.00015401003536428534,
+      "loss": 0.11642799526453018,
+      "step": 445
+    },
+    {
+      "epoch": 3.2093862815884475,
+      "grad_norm": 0.5154263973236084,
+      "learning_rate": 0.00015321110363967035,
+      "loss": 0.17028403282165527,
+      "step": 446
+    },
+    {
+      "epoch": 3.216606498194946,
+      "grad_norm": 0.4077145755290985,
+      "learning_rate": 0.0001524129609914763,
+      "loss": 0.11544163525104523,
+      "step": 447
+    },
+    {
+      "epoch": 3.223826714801444,
+      "grad_norm": 0.41609570384025574,
+      "learning_rate": 0.00015161562088006648,
+      "loss": 0.1203891783952713,
+      "step": 448
+    },
+    {
+      "epoch": 3.2310469314079424,
+      "grad_norm": 0.430624783039093,
+      "learning_rate": 0.00015081909675226975,
+      "loss": 0.09192588925361633,
+      "step": 449
+    },
+    {
+      "epoch": 3.2382671480144403,
+      "grad_norm": 0.5538221001625061,
+      "learning_rate": 0.0001500234020411538,
+      "loss": 0.16532525420188904,
+      "step": 450
+    },
+    {
+      "epoch": 3.2454873646209386,
+      "grad_norm": 0.6360041499137878,
+      "learning_rate": 0.00014922855016579845,
+      "loss": 0.13534685969352722,
+      "step": 451
+    },
+    {
+      "epoch": 3.252707581227437,
+      "grad_norm": 0.556538462638855,
+      "learning_rate": 0.00014843455453106925,
+      "loss": 0.16164641082286835,
+      "step": 452
+    },
+    {
+      "epoch": 3.259927797833935,
+      "grad_norm": 0.8166404366493225,
+      "learning_rate": 0.00014764142852739193,
+      "loss": 0.17865729331970215,
+      "step": 453
+    },
+    {
+      "epoch": 3.2671480144404335,
+      "grad_norm": 0.6231580972671509,
+      "learning_rate": 0.00014684918553052603,
+      "loss": 0.11281149089336395,
+      "step": 454
+    },
+    {
+      "epoch": 3.2743682310469313,
+      "grad_norm": 0.5884938836097717,
+      "learning_rate": 0.00014605783890133948,
+      "loss": 0.15221013128757477,
+      "step": 455
+    },
+    {
+      "epoch": 3.2815884476534296,
+      "grad_norm": 0.40336933732032776,
+      "learning_rate": 0.00014526740198558344,
+      "loss": 0.08725619316101074,
+      "step": 456
+    },
+    {
+      "epoch": 3.288808664259928,
+      "grad_norm": 0.7082973718643188,
+      "learning_rate": 0.0001444778881136671,
+      "loss": 0.1795981079339981,
+      "step": 457
+    },
+    {
+      "epoch": 3.2960288808664258,
+      "grad_norm": 0.44312670826911926,
+      "learning_rate": 0.0001436893106004328,
+      "loss": 0.11532506346702576,
+      "step": 458
+    },
+    {
+      "epoch": 3.303249097472924,
+      "grad_norm": 0.4892312288284302,
+      "learning_rate": 0.0001429016827449316,
+      "loss": 0.15218153595924377,
+      "step": 459
+    },
+    {
+      "epoch": 3.3104693140794224,
+      "grad_norm": 0.36000511050224304,
+      "learning_rate": 0.000142115017830199,
+      "loss": 0.1174839586019516,
+      "step": 460
+    },
+    {
+      "epoch": 3.3176895306859207,
+      "grad_norm": 0.42502933740615845,
+      "learning_rate": 0.00014132932912303082,
+      "loss": 0.1354883909225464,
+      "step": 461
+    },
+    {
+      "epoch": 3.324909747292419,
+      "grad_norm": 0.5004318356513977,
+      "learning_rate": 0.00014054462987375947,
+      "loss": 0.14879979193210602,
+      "step": 462
+    },
+    {
+      "epoch": 3.332129963898917,
+      "grad_norm": 0.5428838729858398,
+      "learning_rate": 0.00013976093331603072,
+      "loss": 0.1601594090461731,
+      "step": 463
+    },
+    {
+      "epoch": 3.339350180505415,
+      "grad_norm": 0.5523360967636108,
+      "learning_rate": 0.00013897825266658015,
+      "loss": 0.1349392980337143,
+      "step": 464
+    },
+    {
+      "epoch": 3.3465703971119134,
+      "grad_norm": 0.4672771394252777,
+      "learning_rate": 0.00013819660112501054,
+      "loss": 0.1563996523618698,
+      "step": 465
+    },
+    {
+      "epoch": 3.3537906137184117,
+      "grad_norm": 0.4747507870197296,
+      "learning_rate": 0.00013741599187356917,
+      "loss": 0.12792187929153442,
+      "step": 466
+    },
+    {
+      "epoch": 3.3610108303249095,
+      "grad_norm": 0.4293171167373657,
+      "learning_rate": 0.00013663643807692547,
+      "loss": 0.13704359531402588,
+      "step": 467
+    },
+    {
+      "epoch": 3.368231046931408,
+      "grad_norm": 0.406646728515625,
+      "learning_rate": 0.00013585795288194902,
+      "loss": 0.14688344299793243,
+      "step": 468
+    },
+    {
+      "epoch": 3.375451263537906,
+      "grad_norm": 0.44870486855506897,
+      "learning_rate": 0.00013508054941748792,
+      "loss": 0.11994664371013641,
+      "step": 469
+    },
+    {
+      "epoch": 3.3826714801444044,
+      "grad_norm": 0.46383994817733765,
+      "learning_rate": 0.00013430424079414728,
+      "loss": 0.13228918612003326,
+      "step": 470
+    },
+    {
+      "epoch": 3.3898916967509027,
+      "grad_norm": 0.873016893863678,
+      "learning_rate": 0.0001335290401040681,
+      "loss": 0.14495904743671417,
+      "step": 471
+    },
+    {
+      "epoch": 3.3971119133574006,
+      "grad_norm": 0.5575613975524902,
+      "learning_rate": 0.0001327549604207066,
+      "loss": 0.17171433568000793,
+      "step": 472
+    },
+    {
+      "epoch": 3.404332129963899,
+      "grad_norm": 0.541860818862915,
+      "learning_rate": 0.00013198201479861353,
+      "loss": 0.14465925097465515,
+      "step": 473
+    },
+    {
+      "epoch": 3.411552346570397,
+      "grad_norm": 0.5210021734237671,
+      "learning_rate": 0.00013121021627321438,
+      "loss": 0.15102122724056244,
+      "step": 474
+    },
+    {
+      "epoch": 3.4187725631768955,
+      "grad_norm": 0.46373456716537476,
+      "learning_rate": 0.00013043957786058904,
+      "loss": 0.15518702566623688,
+      "step": 475
+    },
+    {
+      "epoch": 3.4259927797833933,
+      "grad_norm": 0.48513785004615784,
+      "learning_rate": 0.00012967011255725263,
+      "loss": 0.14110815525054932,
+      "step": 476
+    },
+    {
+      "epoch": 3.4332129963898916,
+      "grad_norm": 0.5956123471260071,
+      "learning_rate": 0.0001289018333399364,
+      "loss": 0.14913466572761536,
+      "step": 477
+    },
+    {
+      "epoch": 3.44043321299639,
+      "grad_norm": 0.5387746095657349,
+      "learning_rate": 0.0001281347531653685,
+      "loss": 0.14882197976112366,
+      "step": 478
+    },
+    {
+      "epoch": 3.4476534296028882,
+      "grad_norm": 0.43160125613212585,
+      "learning_rate": 0.00012736888497005578,
+      "loss": 0.13204768300056458,
+      "step": 479
+    },
+    {
+      "epoch": 3.4548736462093865,
+      "grad_norm": 0.41130170226097107,
+      "learning_rate": 0.0001266042416700656,
+      "loss": 0.13427188992500305,
+      "step": 480
+    },
+    {
+      "epoch": 3.4620938628158844,
+      "grad_norm": 0.5126029253005981,
+      "learning_rate": 0.00012584083616080786,
+      "loss": 0.12719954550266266,
+      "step": 481
+    },
+    {
+      "epoch": 3.4693140794223827,
+      "grad_norm": 0.4994872212409973,
+      "learning_rate": 0.0001250786813168176,
+      "loss": 0.15062202513217926,
+      "step": 482
+    },
+    {
+      "epoch": 3.476534296028881,
+      "grad_norm": 0.4697980284690857,
+      "learning_rate": 0.00012431778999153796,
+      "loss": 0.14715459942817688,
+      "step": 483
+    },
+    {
+      "epoch": 3.483754512635379,
+      "grad_norm": 0.5356115102767944,
+      "learning_rate": 0.00012355817501710328,
+      "loss": 0.12667298316955566,
+      "step": 484
+    },
+    {
+      "epoch": 3.490974729241877,
+      "grad_norm": 0.48629310727119446,
+      "learning_rate": 0.00012279984920412264,
+      "loss": 0.12396648526191711,
+      "step": 485
+    },
+    {
+      "epoch": 3.4981949458483754,
+      "grad_norm": 0.6295841336250305,
+      "learning_rate": 0.00012204282534146414,
+      "loss": 0.1579296588897705,
+      "step": 486
+    },
+    {
+      "epoch": 3.5054151624548737,
+      "grad_norm": 0.524736225605011,
+      "learning_rate": 0.00012128711619603885,
+      "loss": 0.14176616072654724,
+      "step": 487
+    },
+    {
+      "epoch": 3.512635379061372,
+      "grad_norm": 0.4429115355014801,
+      "learning_rate": 0.0001205327345125856,
+      "loss": 0.16075468063354492,
+      "step": 488
+    },
+    {
+      "epoch": 3.51985559566787,
+      "grad_norm": 0.5202926993370056,
+      "learning_rate": 0.00011977969301345627,
+      "loss": 0.155876025557518,
+      "step": 489
+    },
+    {
+      "epoch": 3.527075812274368,
+      "grad_norm": 0.5090999603271484,
+      "learning_rate": 0.00011902800439840091,
+      "loss": 0.1420796513557434,
+      "step": 490
+    },
+    {
+      "epoch": 3.5342960288808665,
+      "grad_norm": 0.5266745686531067,
+      "learning_rate": 0.0001182776813443537,
+      "loss": 0.1911281943321228,
+      "step": 491
+    },
+    {
+      "epoch": 3.5415162454873648,
+      "grad_norm": 0.4100717306137085,
+      "learning_rate": 0.00011752873650521934,
+      "loss": 0.12646690011024475,
+      "step": 492
+    },
+    {
+      "epoch": 3.5487364620938626,
+      "grad_norm": 0.4897909164428711,
+      "learning_rate": 0.00011678118251165929,
+      "loss": 0.14893396198749542,
+      "step": 493
+    },
+    {
+      "epoch": 3.555956678700361,
+      "grad_norm": 0.48633724451065063,
+      "learning_rate": 0.00011603503197087893,
+      "loss": 0.11430861800909042,
+      "step": 494
+    },
+    {
+      "epoch": 3.563176895306859,
+      "grad_norm": 0.3797968327999115,
+      "learning_rate": 0.00011529029746641522,
+      "loss": 0.10336866229772568,
+      "step": 495
+    },
+    {
+      "epoch": 3.5703971119133575,
+      "grad_norm": 0.4714502692222595,
+      "learning_rate": 0.00011454699155792387,
+      "loss": 0.12609894573688507,
+      "step": 496
+    },
+    {
+      "epoch": 3.577617328519856,
+      "grad_norm": 0.5311459898948669,
+      "learning_rate": 0.00011380512678096805,
+      "loss": 0.14762037992477417,
+      "step": 497
+    },
+    {
+      "epoch": 3.5848375451263537,
+      "grad_norm": 0.44842103123664856,
+      "learning_rate": 0.00011306471564680703,
+      "loss": 0.1137317568063736,
+      "step": 498
+    },
+    {
+      "epoch": 3.592057761732852,
+      "grad_norm": 0.6083047986030579,
+      "learning_rate": 0.0001123257706421845,
+      "loss": 0.1538015455007553,
+      "step": 499
+    },
+    {
+      "epoch": 3.5992779783393503,
+      "grad_norm": 0.454277902841568,
+      "learning_rate": 0.00011158830422911882,
+      "loss": 0.13900214433670044,
+      "step": 500
+    },
+    {
+      "epoch": 3.606498194945848,
+      "grad_norm": 0.46477752923965454,
+      "learning_rate": 0.00011085232884469236,
+      "loss": 0.1317400485277176,
+      "step": 501
+    },
+    {
+      "epoch": 3.6137184115523464,
+      "grad_norm": 0.5491266250610352,
+      "learning_rate": 0.00011011785690084178,
+      "loss": 0.20672887563705444,
+      "step": 502
+    },
+    {
+      "epoch": 3.6209386281588447,
+      "grad_norm": 0.48666200041770935,
+      "learning_rate": 0.00010938490078414902,
+      "loss": 0.15307360887527466,
+      "step": 503
+    },
+    {
+      "epoch": 3.628158844765343,
+      "grad_norm": 0.557000458240509,
+      "learning_rate": 0.00010865347285563189,
+      "loss": 0.18239088356494904,
+      "step": 504
+    },
+    {
+      "epoch": 3.6353790613718413,
+      "grad_norm": 0.45608651638031006,
+      "learning_rate": 0.0001079235854505362,
+      "loss": 0.15936806797981262,
+      "step": 505
+    },
+    {
+      "epoch": 3.6425992779783396,
+      "grad_norm": 0.455575555562973,
+      "learning_rate": 0.00010719525087812736,
+      "loss": 0.12316213548183441,
+      "step": 506
+    },
+    {
+      "epoch": 3.6498194945848375,
+      "grad_norm": 0.48413556814193726,
+      "learning_rate": 0.00010646848142148267,
+      "loss": 0.13552163541316986,
+      "step": 507
+    },
+    {
+      "epoch": 3.6570397111913358,
+      "grad_norm": 0.31646862626075745,
+      "learning_rate": 0.0001057432893372846,
+      "loss": 0.09000004827976227,
+      "step": 508
+    },
+    {
+      "epoch": 3.664259927797834,
+      "grad_norm": 0.43280190229415894,
+      "learning_rate": 0.00010501968685561379,
+      "loss": 0.13823340833187103,
+      "step": 509
+    },
+    {
+      "epoch": 3.671480144404332,
+      "grad_norm": 0.48566341400146484,
+      "learning_rate": 0.00010429768617974271,
+      "loss": 0.1494002640247345,
+      "step": 510
+    },
+    {
+      "epoch": 3.67870036101083,
+      "grad_norm": 0.5894930362701416,
+      "learning_rate": 0.00010357729948593022,
+      "loss": 0.16500279307365417,
+      "step": 511
+    },
+    {
+      "epoch": 3.6859205776173285,
+      "grad_norm": 0.4345530867576599,
+      "learning_rate": 0.00010285853892321597,
+      "loss": 0.15392449498176575,
+      "step": 512
+    },
+    {
+      "epoch": 3.693140794223827,
+      "grad_norm": 0.38582298159599304,
+      "learning_rate": 0.0001021414166132153,
+      "loss": 0.11830861121416092,
+      "step": 513
+    },
+    {
+      "epoch": 3.700361010830325,
+      "grad_norm": 0.5571300983428955,
+      "learning_rate": 0.00010142594464991538,
+      "loss": 0.17028063535690308,
+      "step": 514
+    },
+    {
+      "epoch": 3.707581227436823,
+      "grad_norm": 0.442558616399765,
+      "learning_rate": 0.00010071213509947078,
+      "loss": 0.13233450055122375,
+      "step": 515
+    },
+    {
+      "epoch": 3.7148014440433212,
+      "grad_norm": 0.38991135358810425,
+      "learning_rate": 0.00010000000000000005,
+      "loss": 0.11185239255428314,
+      "step": 516
+    },
+    {
+      "epoch": 3.7220216606498195,
+      "grad_norm": 0.46868595480918884,
+      "learning_rate": 9.928955136138292e-05,
+      "loss": 0.13171209394931793,
+      "step": 517
+    },
+    {
+      "epoch": 3.729241877256318,
+      "grad_norm": 0.3938763439655304,
+      "learning_rate": 9.858080116505763e-05,
+      "loss": 0.12150520086288452,
+      "step": 518
+    },
+    {
+      "epoch": 3.7364620938628157,
+      "grad_norm": 0.5428814888000488,
+      "learning_rate": 9.787376136381865e-05,
+      "loss": 0.17324000597000122,
+      "step": 519
+    },
+    {
+      "epoch": 3.743682310469314,
+      "grad_norm": 0.5528351664543152,
+      "learning_rate": 9.716844388161555e-05,
+      "loss": 0.13830417394638062,
+      "step": 520
+    },
+    {
+      "epoch": 3.7509025270758123,
+      "grad_norm": 0.43436095118522644,
+      "learning_rate": 9.646486061335151e-05,
+      "loss": 0.12715619802474976,
+      "step": 521
+    },
+    {
+      "epoch": 3.7581227436823106,
+      "grad_norm": 0.5142949223518372,
+      "learning_rate": 9.576302342468297e-05,
+      "loss": 0.13084562122821808,
+      "step": 522
+    },
+    {
+      "epoch": 3.765342960288809,
+      "grad_norm": 0.6056551933288574,
+      "learning_rate": 9.506294415181942e-05,
+      "loss": 0.14757800102233887,
+      "step": 523
+    },
+    {
+      "epoch": 3.7725631768953067,
+      "grad_norm": 0.7759309411048889,
+      "learning_rate": 9.436463460132359e-05,
+      "loss": 0.14157943427562714,
+      "step": 524
+    },
+    {
+      "epoch": 3.779783393501805,
+      "grad_norm": 0.5732694268226624,
+      "learning_rate": 9.366810654991276e-05,
+      "loss": 0.12228554487228394,
+      "step": 525
+    },
+    {
+      "epoch": 3.7870036101083033,
+      "grad_norm": 0.5100772976875305,
+      "learning_rate": 9.297337174425995e-05,
+      "loss": 0.13228444755077362,
+      "step": 526
+    },
+    {
+      "epoch": 3.794223826714801,
+      "grad_norm": 0.3461199104785919,
+      "learning_rate": 9.228044190079553e-05,
+      "loss": 0.11099657416343689,
+      "step": 527
+    },
+    {
+      "epoch": 3.8014440433212995,
+      "grad_norm": 0.5344772338867188,
+      "learning_rate": 9.158932870551011e-05,
+      "loss": 0.1298564374446869,
+      "step": 528
+    },
+    {
+      "epoch": 3.808664259927798,
+      "grad_norm": 0.5982236266136169,
+      "learning_rate": 9.09000438137573e-05,
+      "loss": 0.13814345002174377,
+      "step": 529
+    },
+    {
+      "epoch": 3.815884476534296,
+      "grad_norm": 0.540401816368103,
+      "learning_rate": 9.021259885005678e-05,
+      "loss": 0.15752826631069183,
+      "step": 530
+    },
+    {
+      "epoch": 3.8231046931407944,
+      "grad_norm": 0.5778837203979492,
+      "learning_rate": 8.952700540789885e-05,
+      "loss": 0.11452727019786835,
+      "step": 531
+    },
+    {
+      "epoch": 3.8303249097472927,
+      "grad_norm": 0.5168759226799011,
+      "learning_rate": 8.884327504954857e-05,
+      "loss": 0.18627649545669556,
+      "step": 532
+    },
+    {
+      "epoch": 3.8375451263537905,
+      "grad_norm": 0.5951539278030396,
+      "learning_rate": 8.816141930585067e-05,
+      "loss": 0.17812731862068176,
+      "step": 533
+    },
+    {
+      "epoch": 3.844765342960289,
+      "grad_norm": 0.46762675046920776,
+      "learning_rate": 8.748144967603538e-05,
+      "loss": 0.1432511955499649,
+      "step": 534
+    },
+    {
+      "epoch": 3.851985559566787,
+      "grad_norm": 0.4484635889530182,
+      "learning_rate": 8.680337762752444e-05,
+      "loss": 0.13658685982227325,
+      "step": 535
+    },
+    {
+      "epoch": 3.859205776173285,
+      "grad_norm": 0.518476128578186,
+      "learning_rate": 8.612721459573734e-05,
+      "loss": 0.14582251012325287,
+      "step": 536
+    },
+    {
+      "epoch": 3.8664259927797833,
+      "grad_norm": 0.5416932106018066,
+      "learning_rate": 8.545297198389896e-05,
+      "loss": 0.15608924627304077,
+      "step": 537
+    },
+    {
+      "epoch": 3.8736462093862816,
+      "grad_norm": 0.502341628074646,
+      "learning_rate": 8.478066116284713e-05,
+      "loss": 0.14149951934814453,
+      "step": 538
+    },
+    {
+      "epoch": 3.88086642599278,
+      "grad_norm": 0.3645715117454529,
+      "learning_rate": 8.411029347084049e-05,
+      "loss": 0.1299845278263092,
+      "step": 539
+    },
+    {
+      "epoch": 3.888086642599278,
+      "grad_norm": 0.4532802700996399,
+      "learning_rate": 8.344188021336783e-05,
+      "loss": 0.11957640945911407,
+      "step": 540
+    },
+    {
+      "epoch": 3.895306859205776,
+      "grad_norm": 0.39594075083732605,
+      "learning_rate": 8.277543266295713e-05,
+      "loss": 0.09765089303255081,
+      "step": 541
+    },
+    {
+      "epoch": 3.9025270758122743,
+      "grad_norm": 0.4911992847919464,
+      "learning_rate": 8.211096205898529e-05,
+      "loss": 0.13661810755729675,
+      "step": 542
+    },
+    {
+      "epoch": 3.9097472924187726,
+      "grad_norm": 0.41524603962898254,
+      "learning_rate": 8.144847960748904e-05,
+      "loss": 0.11461715400218964,
+      "step": 543
+    },
+    {
+      "epoch": 3.916967509025271,
+      "grad_norm": 0.6111493110656738,
+      "learning_rate": 8.078799648097555e-05,
+      "loss": 0.14272424578666687,
+      "step": 544
+    },
+    {
+      "epoch": 3.9241877256317688,
+      "grad_norm": 0.6367743611335754,
+      "learning_rate": 8.01295238182342e-05,
+      "loss": 0.15256257355213165,
+      "step": 545
+    },
+    {
+      "epoch": 3.931407942238267,
+      "grad_norm": 0.5055350065231323,
+      "learning_rate": 7.947307272414874e-05,
+      "loss": 0.12612725794315338,
+      "step": 546
+    },
+    {
+      "epoch": 3.9386281588447654,
+      "grad_norm": 0.6354690790176392,
+      "learning_rate": 7.881865426950979e-05,
+      "loss": 0.18444940447807312,
+      "step": 547
+    },
+    {
+      "epoch": 3.9458483754512637,
+      "grad_norm": 0.48749735951423645,
+      "learning_rate": 7.816627949082844e-05,
+      "loss": 0.14657062292099,
+      "step": 548
+    },
+    {
+      "epoch": 3.953068592057762,
+      "grad_norm": 0.5144006013870239,
+      "learning_rate": 7.751595939015004e-05,
+      "loss": 0.13369342684745789,
+      "step": 549
+    },
+    {
+      "epoch": 3.96028880866426,
+      "grad_norm": 0.6086586117744446,
+      "learning_rate": 7.686770493486834e-05,
+      "loss": 0.12459664046764374,
+      "step": 550
+    },
+    {
+      "epoch": 3.967509025270758,
+      "grad_norm": 0.3899252712726593,
+      "learning_rate": 7.622152705754109e-05,
+      "loss": 0.09838660806417465,
+      "step": 551
+    },
+    {
+      "epoch": 3.9747292418772564,
+      "grad_norm": 0.5426426529884338,
+      "learning_rate": 7.557743665570524e-05,
+      "loss": 0.12174687534570694,
+      "step": 552
+    },
+    {
+      "epoch": 3.9819494584837543,
+      "grad_norm": 0.4566539227962494,
+      "learning_rate": 7.49354445916932e-05,
+      "loss": 0.12463852018117905,
+      "step": 553
+    },
+    {
+      "epoch": 3.9891696750902526,
+      "grad_norm": 0.4546673595905304,
+      "learning_rate": 7.429556169244984e-05,
+      "loss": 0.145570307970047,
+      "step": 554
+    },
+    {
+      "epoch": 3.996389891696751,
+      "grad_norm": 0.4223948121070862,
+      "learning_rate": 7.365779874934985e-05,
+      "loss": 0.1327395737171173,
+      "step": 555
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.9028692245483398,
+      "learning_rate": 7.302216651801546e-05,
+      "loss": 0.1679631471633911,
+      "step": 556
+    },
+    {
+      "epoch": 4.007220216606498,
+      "grad_norm": 0.2356356829404831,
+      "learning_rate": 7.23886757181355e-05,
+      "loss": 0.07719021290540695,
+      "step": 557
+    },
+    {
+      "epoch": 4.014440433212997,
+      "grad_norm": 0.2547387480735779,
+      "learning_rate": 7.175733703328438e-05,
+      "loss": 0.08970507979393005,
+      "step": 558
+    },
+    {
+      "epoch": 4.021660649819495,
+      "grad_norm": 0.2796124815940857,
+      "learning_rate": 7.112816111074175e-05,
+      "loss": 0.08122967183589935,
+      "step": 559
+    },
+    {
+      "epoch": 4.028880866425993,
+      "grad_norm": 0.3060125410556793,
+      "learning_rate": 7.050115856131327e-05,
+      "loss": 0.10071869194507599,
+      "step": 560
+    },
+    {
+      "epoch": 4.036101083032491,
+      "grad_norm": 0.23872718214988708,
+      "learning_rate": 6.987633995915164e-05,
+      "loss": 0.07773898541927338,
+      "step": 561
+    },
+    {
+      "epoch": 4.043321299638989,
+      "grad_norm": 0.2474278211593628,
+      "learning_rate": 6.925371584157776e-05,
+      "loss": 0.08390785753726959,
+      "step": 562
+    },
+    {
+      "epoch": 4.050541516245487,
+      "grad_norm": 0.2561244070529938,
+      "learning_rate": 6.863329670890379e-05,
+      "loss": 0.07564981281757355,
+      "step": 563
+    },
+    {
+      "epoch": 4.0577617328519855,
+      "grad_norm": 0.23667839169502258,
+      "learning_rate": 6.801509302425553e-05,
+      "loss": 0.06735668331384659,
+      "step": 564
+    },
+    {
+      "epoch": 4.064981949458484,
+      "grad_norm": 0.30244675278663635,
+      "learning_rate": 6.739911521339603e-05,
+      "loss": 0.08246147632598877,
+      "step": 565
+    },
+    {
+      "epoch": 4.072202166064982,
+      "grad_norm": 0.322549968957901,
+      "learning_rate": 6.678537366454998e-05,
+      "loss": 0.0768466591835022,
+      "step": 566
+    },
+    {
+      "epoch": 4.07942238267148,
+      "grad_norm": 0.3082004487514496,
+      "learning_rate": 6.617387872822842e-05,
+      "loss": 0.07058844715356827,
+      "step": 567
+    },
+    {
+      "epoch": 4.086642599277979,
+      "grad_norm": 0.4027414917945862,
+      "learning_rate": 6.556464071705386e-05,
+      "loss": 0.06483431905508041,
+      "step": 568
+    },
+    {
+      "epoch": 4.093862815884476,
+      "grad_norm": 0.32241883873939514,
+      "learning_rate": 6.495766990558716e-05,
+      "loss": 0.08380893617868423,
+      "step": 569
+    },
+    {
+      "epoch": 4.101083032490974,
+      "grad_norm": 0.6206040978431702,
+      "learning_rate": 6.435297653015319e-05,
+      "loss": 0.08805538713932037,
+      "step": 570
+    },
+    {
+      "epoch": 4.108303249097473,
+      "grad_norm": 0.37588536739349365,
+      "learning_rate": 6.375057078866916e-05,
+      "loss": 0.06972259283065796,
+      "step": 571
+    },
+    {
+      "epoch": 4.115523465703971,
+      "grad_norm": 0.7320848703384399,
+      "learning_rate": 6.315046284047207e-05,
+      "loss": 0.07265684008598328,
+      "step": 572
+    },
+    {
+      "epoch": 4.122743682310469,
+      "grad_norm": 0.6254473328590393,
+      "learning_rate": 6.255266280614747e-05,
+      "loss": 0.06212320178747177,
+      "step": 573
+    },
+    {
+      "epoch": 4.129963898916968,
+      "grad_norm": 0.5760471820831299,
+      "learning_rate": 6.195718076735894e-05,
+      "loss": 0.08673934638500214,
+      "step": 574
+    },
+    {
+      "epoch": 4.137184115523466,
+      "grad_norm": 0.49967148900032043,
+      "learning_rate": 6.136402676667806e-05,
+      "loss": 0.06496398150920868,
+      "step": 575
+    },
+    {
+      "epoch": 4.144404332129964,
+      "grad_norm": 0.43019023537635803,
+      "learning_rate": 6.077321080741469e-05,
+      "loss": 0.06604783982038498,
+      "step": 576
+    },
+    {
+      "epoch": 4.1516245487364625,
+      "grad_norm": 0.4466577172279358,
+      "learning_rate": 6.0184742853448796e-05,
+      "loss": 0.06710691004991531,
+      "step": 577
+    },
+    {
+      "epoch": 4.15884476534296,
+      "grad_norm": 0.5732907056808472,
+      "learning_rate": 5.959863282906208e-05,
+      "loss": 0.09353934228420258,
+      "step": 578
+    },
+    {
+      "epoch": 4.166064981949458,
+      "grad_norm": 0.3326954245567322,
+      "learning_rate": 5.901489061877059e-05,
+      "loss": 0.06768821179866791,
+      "step": 579
+    },
+    {
+      "epoch": 4.1732851985559565,
+      "grad_norm": 0.5838833451271057,
+      "learning_rate": 5.843352606715819e-05,
+      "loss": 0.09657922387123108,
+      "step": 580
+    },
+    {
+      "epoch": 4.180505415162455,
+      "grad_norm": 0.6080812215805054,
+      "learning_rate": 5.7854548978710586e-05,
+      "loss": 0.07862652093172073,
+      "step": 581
+    },
+    {
+      "epoch": 4.187725631768953,
+      "grad_norm": 0.35978272557258606,
+      "learning_rate": 5.7277969117649554e-05,
+      "loss": 0.06750591099262238,
+      "step": 582
+    },
+    {
+      "epoch": 4.194945848375451,
+      "grad_norm": 0.5055731534957886,
+      "learning_rate": 5.670379620776882e-05,
+      "loss": 0.07750953733921051,
+      "step": 583
+    },
+    {
+      "epoch": 4.20216606498195,
+      "grad_norm": 0.3300122618675232,
+      "learning_rate": 5.613203993226981e-05,
+      "loss": 0.05905657261610031,
+      "step": 584
+    },
+    {
+      "epoch": 4.209386281588448,
+      "grad_norm": 0.39663630723953247,
+      "learning_rate": 5.5562709933598156e-05,
+      "loss": 0.06651993840932846,
+      "step": 585
+    },
+    {
+      "epoch": 4.216606498194946,
+      "grad_norm": 0.2632470726966858,
+      "learning_rate": 5.4995815813281483e-05,
+      "loss": 0.04662226885557175,
+      "step": 586
+    },
+    {
+      "epoch": 4.223826714801444,
+      "grad_norm": 0.3943081796169281,
+      "learning_rate": 5.443136713176724e-05,
+      "loss": 0.055592283606529236,
+      "step": 587
+    },
+    {
+      "epoch": 4.231046931407942,
+      "grad_norm": 0.3067891001701355,
+      "learning_rate": 5.386937340826139e-05,
+      "loss": 0.06188460439443588,
+      "step": 588
+    },
+    {
+      "epoch": 4.23826714801444,
+      "grad_norm": 0.5765032768249512,
+      "learning_rate": 5.330984412056814e-05,
+      "loss": 0.07393093407154083,
+      "step": 589
+    },
+    {
+      "epoch": 4.245487364620939,
+      "grad_norm": 0.3043580949306488,
+      "learning_rate": 5.2752788704929944e-05,
+      "loss": 0.04691813141107559,
+      "step": 590
+    },
+    {
+      "epoch": 4.252707581227437,
+      "grad_norm": 0.41853103041648865,
+      "learning_rate": 5.2198216555868206e-05,
+      "loss": 0.08564123511314392,
+      "step": 591
+    },
+    {
+      "epoch": 4.259927797833935,
+      "grad_norm": 0.366338849067688,
+      "learning_rate": 5.164613702602519e-05,
+      "loss": 0.06465509533882141,
+      "step": 592
+    },
+    {
+      "epoch": 4.2671480144404335,
+      "grad_norm": 0.4903438985347748,
+      "learning_rate": 5.1096559426005994e-05,
+      "loss": 0.0653722956776619,
+      "step": 593
+    },
+    {
+      "epoch": 4.274368231046932,
+      "grad_norm": 0.3799297511577606,
+      "learning_rate": 5.054949302422178e-05,
+      "loss": 0.06573019176721573,
+      "step": 594
+    },
+    {
+      "epoch": 4.28158844765343,
+      "grad_norm": 0.3962422013282776,
+      "learning_rate": 5.000494704673328e-05,
+      "loss": 0.06330323964357376,
+      "step": 595
+    },
+    {
+      "epoch": 4.2888086642599275,
+      "grad_norm": 0.40771204233169556,
+      "learning_rate": 4.946293067709511e-05,
+      "loss": 0.0744054913520813,
+      "step": 596
+    },
+    {
+      "epoch": 4.296028880866426,
+      "grad_norm": 0.358303040266037,
+      "learning_rate": 4.892345305620123e-05,
+      "loss": 0.06824557483196259,
+      "step": 597
+    },
+    {
+      "epoch": 4.303249097472924,
+      "grad_norm": 0.5681015253067017,
+      "learning_rate": 4.83865232821306e-05,
+      "loss": 0.07922443747520447,
+      "step": 598
+    },
+    {
+      "epoch": 4.310469314079422,
+      "grad_norm": 0.40545016527175903,
+      "learning_rate": 4.785215040999354e-05,
+      "loss": 0.06584658473730087,
+      "step": 599
+    },
+    {
+      "epoch": 4.317689530685921,
+      "grad_norm": 0.4532371759414673,
+      "learning_rate": 4.73203434517794e-05,
+      "loss": 0.06798739731311798,
+      "step": 600
+    },
+    {
+      "epoch": 4.324909747292419,
+      "grad_norm": 0.34823986887931824,
+      "learning_rate": 4.679111137620442e-05,
+      "loss": 0.060036975890398026,
+      "step": 601
+    },
+    {
+      "epoch": 4.332129963898917,
+      "grad_norm": 0.7183476686477661,
+      "learning_rate": 4.626446310856027e-05,
+      "loss": 0.09737586230039597,
+      "step": 602
+    },
+    {
+      "epoch": 4.3393501805054155,
+      "grad_norm": 0.3646744191646576,
+      "learning_rate": 4.574040753056385e-05,
+      "loss": 0.07068562507629395,
+      "step": 603
+    },
+    {
+      "epoch": 4.346570397111913,
+      "grad_norm": 0.7281261682510376,
+      "learning_rate": 4.521895348020744e-05,
+      "loss": 0.0830075815320015,
+      "step": 604
+    },
+    {
+      "epoch": 4.353790613718411,
+      "grad_norm": 0.3349117338657379,
+      "learning_rate": 4.470010975160932e-05,
+      "loss": 0.057703591883182526,
+      "step": 605
+    },
+    {
+      "epoch": 4.3610108303249095,
+      "grad_norm": 0.4553733766078949,
+      "learning_rate": 4.418388509486591e-05,
+      "loss": 0.07166286557912827,
+      "step": 606
+    },
+    {
+      "epoch": 4.368231046931408,
+      "grad_norm": 0.44004717469215393,
+      "learning_rate": 4.3670288215904044e-05,
+      "loss": 0.06402762979269028,
+      "step": 607
+    },
+    {
+      "epoch": 4.375451263537906,
+      "grad_norm": 0.4734828472137451,
+      "learning_rate": 4.315932777633391e-05,
+      "loss": 0.06359273195266724,
+      "step": 608
+    },
+    {
+      "epoch": 4.382671480144404,
+      "grad_norm": 0.42865970730781555,
+      "learning_rate": 4.265101239330336e-05,
+      "loss": 0.05663272365927696,
+      "step": 609
+    },
+    {
+      "epoch": 4.389891696750903,
+      "grad_norm": 0.3049170970916748,
+      "learning_rate": 4.214535063935236e-05,
+      "loss": 0.05514144152402878,
+      "step": 610
+    },
+    {
+      "epoch": 4.397111913357401,
+      "grad_norm": 0.5843676328659058,
+      "learning_rate": 4.1642351042268413e-05,
+      "loss": 0.0814075767993927,
+      "step": 611
+    },
+    {
+      "epoch": 4.404332129963899,
+      "grad_norm": 0.45971769094467163,
+      "learning_rate": 4.11420220849428e-05,
+      "loss": 0.08407925814390182,
+      "step": 612
+    },
+    {
+      "epoch": 4.411552346570397,
+      "grad_norm": 0.3730124235153198,
+      "learning_rate": 4.064437220522759e-05,
+      "loss": 0.08016012609004974,
+      "step": 613
+    },
+    {
+      "epoch": 4.418772563176895,
+      "grad_norm": 0.47459709644317627,
+      "learning_rate": 4.0149409795793116e-05,
+      "loss": 0.06777369976043701,
+      "step": 614
+    },
+    {
+      "epoch": 4.425992779783393,
+      "grad_norm": 0.43131333589553833,
+      "learning_rate": 3.965714320398668e-05,
+      "loss": 0.06275366246700287,
+      "step": 615
+    },
+    {
+      "epoch": 4.433212996389892,
+      "grad_norm": 0.5383766293525696,
+      "learning_rate": 3.9167580731691644e-05,
+      "loss": 0.07567012310028076,
+      "step": 616
+    },
+    {
+      "epoch": 4.44043321299639,
+      "grad_norm": 0.5049149990081787,
+      "learning_rate": 3.868073063518749e-05,
+      "loss": 0.07030000537633896,
+      "step": 617
+    },
+    {
+      "epoch": 4.447653429602888,
+      "grad_norm": 0.4588138461112976,
+      "learning_rate": 3.819660112501053e-05,
+      "loss": 0.06413187086582184,
+      "step": 618
+    },
+    {
+      "epoch": 4.4548736462093865,
+      "grad_norm": 0.5270079374313354,
+      "learning_rate": 3.771520036581535e-05,
+      "loss": 0.08776699006557465,
+      "step": 619
+    },
+    {
+      "epoch": 4.462093862815885,
+      "grad_norm": 0.4434908330440521,
+      "learning_rate": 3.723653647623735e-05,
+      "loss": 0.06680673360824585,
+      "step": 620
+    },
+    {
+      "epoch": 4.469314079422382,
+      "grad_norm": 0.40275996923446655,
+      "learning_rate": 3.6760617528755684e-05,
+      "loss": 0.06380727887153625,
+      "step": 621
+    },
+    {
+      "epoch": 4.4765342960288805,
+      "grad_norm": 0.33545786142349243,
+      "learning_rate": 3.6287451549557016e-05,
+      "loss": 0.06220482289791107,
+      "step": 622
+    },
+    {
+      "epoch": 4.483754512635379,
+      "grad_norm": 0.358530193567276,
+      "learning_rate": 3.5817046518400385e-05,
+      "loss": 0.07663938403129578,
+      "step": 623
+    },
+    {
+      "epoch": 4.490974729241877,
+      "grad_norm": 0.4760512411594391,
+      "learning_rate": 3.534941036848258e-05,
+      "loss": 0.07023464143276215,
+      "step": 624
+    },
+    {
+      "epoch": 4.498194945848375,
+      "grad_norm": 0.5976642966270447,
+      "learning_rate": 3.4884550986304074e-05,
+      "loss": 0.07387922704219818,
+      "step": 625
+    },
+    {
+      "epoch": 4.505415162454874,
+      "grad_norm": 0.4269956052303314,
+      "learning_rate": 3.442247621153643e-05,
+      "loss": 0.06992574781179428,
+      "step": 626
+    },
+    {
+      "epoch": 4.512635379061372,
+      "grad_norm": 0.5576417446136475,
+      "learning_rate": 3.3963193836889905e-05,
+      "loss": 0.06586994230747223,
+      "step": 627
+    },
+    {
+      "epoch": 4.51985559566787,
+      "grad_norm": 0.27365460991859436,
+      "learning_rate": 3.3506711607981824e-05,
+      "loss": 0.05236521363258362,
+      "step": 628
+    },
+    {
+      "epoch": 4.527075812274369,
+      "grad_norm": 0.32343900203704834,
+      "learning_rate": 3.305303722320636e-05,
+      "loss": 0.0833154246211052,
+      "step": 629
+    },
+    {
+      "epoch": 4.534296028880867,
+      "grad_norm": 0.6516956686973572,
+      "learning_rate": 3.260217833360446e-05,
+      "loss": 0.08919483423233032,
+      "step": 630
+    },
+    {
+      "epoch": 4.541516245487364,
+      "grad_norm": 0.3491802513599396,
+      "learning_rate": 3.215414254273468e-05,
+      "loss": 0.06530842185020447,
+      "step": 631
+    },
+    {
+      "epoch": 4.548736462093863,
+      "grad_norm": 0.4860411286354065,
+      "learning_rate": 3.170893740654533e-05,
+      "loss": 0.08303852379322052,
+      "step": 632
+    },
+    {
+      "epoch": 4.555956678700361,
+      "grad_norm": 0.4222336709499359,
+      "learning_rate": 3.126657043324677e-05,
+      "loss": 0.0689733549952507,
+      "step": 633
+    },
+    {
+      "epoch": 4.563176895306859,
+      "grad_norm": 0.6741357445716858,
+      "learning_rate": 3.082704908318474e-05,
+      "loss": 0.07350336015224457,
+      "step": 634
+    },
+    {
+      "epoch": 4.5703971119133575,
+      "grad_norm": 0.3562377691268921,
+      "learning_rate": 3.039038076871481e-05,
+      "loss": 0.06916918605566025,
+      "step": 635
+    },
+    {
+      "epoch": 4.577617328519856,
+      "grad_norm": 0.5075894594192505,
+      "learning_rate": 2.9956572854077202e-05,
+      "loss": 0.05884088948369026,
+      "step": 636
+    },
+    {
+      "epoch": 4.584837545126354,
+      "grad_norm": 0.49467241764068604,
+      "learning_rate": 2.9525632655272482e-05,
+      "loss": 0.05398339778184891,
+      "step": 637
+    },
+    {
+      "epoch": 4.5920577617328515,
+      "grad_norm": 0.30880579352378845,
+      "learning_rate": 2.9097567439938478e-05,
+      "loss": 0.06401652097702026,
+      "step": 638
+    },
+    {
+      "epoch": 4.59927797833935,
+      "grad_norm": 0.3337753713130951,
+      "learning_rate": 2.8672384427227484e-05,
+      "loss": 0.06975904852151871,
+      "step": 639
+    },
+    {
+      "epoch": 4.606498194945848,
+      "grad_norm": 1.321722149848938,
+      "learning_rate": 2.8250090787684437e-05,
+      "loss": 0.07469654083251953,
+      "step": 640
+    },
+    {
+      "epoch": 4.613718411552346,
+      "grad_norm": 0.3456924855709076,
+      "learning_rate": 2.783069364312647e-05,
+      "loss": 0.061028316617012024,
+      "step": 641
+    },
+    {
+      "epoch": 4.620938628158845,
+      "grad_norm": 0.9587940573692322,
+      "learning_rate": 2.7414200066522067e-05,
+      "loss": 0.08222727477550507,
+      "step": 642
+    },
+    {
+      "epoch": 4.628158844765343,
+      "grad_norm": 0.34773939847946167,
+      "learning_rate": 2.7000617081872402e-05,
+      "loss": 0.06201966106891632,
+      "step": 643
+    },
+    {
+      "epoch": 4.635379061371841,
+      "grad_norm": 0.4925590455532074,
+      "learning_rate": 2.6589951664092706e-05,
+      "loss": 0.06247459352016449,
+      "step": 644
+    },
+    {
+      "epoch": 4.64259927797834,
+      "grad_norm": 0.34186214208602905,
+      "learning_rate": 2.618221073889433e-05,
+      "loss": 0.054603368043899536,
+      "step": 645
+    },
+    {
+      "epoch": 4.649819494584838,
+      "grad_norm": 0.3584718406200409,
+      "learning_rate": 2.5777401182668446e-05,
+      "loss": 0.05636471137404442,
+      "step": 646
+    },
+    {
+      "epoch": 4.657039711191336,
+      "grad_norm": 0.33370015025138855,
+      "learning_rate": 2.5375529822369815e-05,
+      "loss": 0.06488262116909027,
+      "step": 647
+    },
+    {
+      "epoch": 4.664259927797834,
+      "grad_norm": 0.428996205329895,
+      "learning_rate": 2.497660343540147e-05,
+      "loss": 0.08126433193683624,
+      "step": 648
+    },
+    {
+      "epoch": 4.671480144404332,
+      "grad_norm": 0.306087464094162,
+      "learning_rate": 2.4580628749500844e-05,
+      "loss": 0.056748297065496445,
+      "step": 649
+    },
+    {
+      "epoch": 4.67870036101083,
+      "grad_norm": 0.45556285977363586,
+      "learning_rate": 2.4187612442626016e-05,
+      "loss": 0.08052687346935272,
+      "step": 650
+    },
+    {
+      "epoch": 4.6859205776173285,
+      "grad_norm": 0.24860779941082,
+      "learning_rate": 2.3797561142843105e-05,
+      "loss": 0.06610046327114105,
+      "step": 651
+    },
+    {
+      "epoch": 4.693140794223827,
+      "grad_norm": 0.452534556388855,
+      "learning_rate": 2.3410481428214603e-05,
+      "loss": 0.056214869022369385,
+      "step": 652
+    },
+    {
+      "epoch": 4.700361010830325,
+      "grad_norm": 0.38668516278266907,
+      "learning_rate": 2.302637982668843e-05,
+      "loss": 0.06206169351935387,
+      "step": 653
+    },
+    {
+      "epoch": 4.707581227436823,
+      "grad_norm": 0.2503337562084198,
+      "learning_rate": 2.2645262815987622e-05,
+      "loss": 0.06331133097410202,
+      "step": 654
+    },
+    {
+      "epoch": 4.714801444043322,
+      "grad_norm": 0.40086954832077026,
+      "learning_rate": 2.2267136823501455e-05,
+      "loss": 0.06614114344120026,
+      "step": 655
+    },
+    {
+      "epoch": 4.722021660649819,
+      "grad_norm": 0.3914971351623535,
+      "learning_rate": 2.1892008226176762e-05,
+      "loss": 0.06879283487796783,
+      "step": 656
+    },
+    {
+      "epoch": 4.729241877256317,
+      "grad_norm": 0.3335961699485779,
+      "learning_rate": 2.1519883350410443e-05,
+      "loss": 0.07069812715053558,
+      "step": 657
+    },
+    {
+      "epoch": 4.736462093862816,
+      "grad_norm": 0.45657894015312195,
+      "learning_rate": 2.115076847194293e-05,
+      "loss": 0.07669930905103683,
+      "step": 658
+    },
+    {
+      "epoch": 4.743682310469314,
+      "grad_norm": 0.28149735927581787,
+      "learning_rate": 2.0784669815752044e-05,
+      "loss": 0.049429114907979965,
+      "step": 659
+    },
+    {
+      "epoch": 4.750902527075812,
+      "grad_norm": 0.3293699026107788,
+      "learning_rate": 2.0421593555948393e-05,
+      "loss": 0.05612622946500778,
+      "step": 660
+    },
+    {
+      "epoch": 4.758122743682311,
+      "grad_norm": 0.4966195821762085,
+      "learning_rate": 2.0061545815670967e-05,
+      "loss": 0.0798168033361435,
+      "step": 661
+    },
+    {
+      "epoch": 4.765342960288809,
+      "grad_norm": 0.40849727392196655,
+      "learning_rate": 1.970453266698391e-05,
+      "loss": 0.06668490171432495,
+      "step": 662
+    },
+    {
+      "epoch": 4.772563176895307,
+      "grad_norm": 0.37348175048828125,
+      "learning_rate": 1.935056013077423e-05,
+      "loss": 0.06494098156690598,
+      "step": 663
+    },
+    {
+      "epoch": 4.7797833935018055,
+      "grad_norm": 0.29984328150749207,
+      "learning_rate": 1.8999634176650205e-05,
+      "loss": 0.051809899508953094,
+      "step": 664
+    },
+    {
+      "epoch": 4.787003610108303,
+      "grad_norm": 0.366937518119812,
+      "learning_rate": 1.8651760722840672e-05,
+      "loss": 0.06859136372804642,
+      "step": 665
+    },
+    {
+      "epoch": 4.794223826714801,
+      "grad_norm": 0.46067899465560913,
+      "learning_rate": 1.830694563609525e-05,
+      "loss": 0.06466175615787506,
+      "step": 666
+    },
+    {
+      "epoch": 4.8014440433212995,
+      "grad_norm": 0.45101597905158997,
+      "learning_rate": 1.7965194731585422e-05,
+      "loss": 0.062113627791404724,
+      "step": 667
+    },
+    {
+      "epoch": 4.808664259927798,
+      "grad_norm": 0.39610734581947327,
+      "learning_rate": 1.762651377280633e-05,
+      "loss": 0.06937667727470398,
+      "step": 668
+    },
+    {
+      "epoch": 4.815884476534296,
+      "grad_norm": 0.406488835811615,
+      "learning_rate": 1.7290908471479805e-05,
+      "loss": 0.06851080060005188,
+      "step": 669
+    },
+    {
+      "epoch": 4.823104693140794,
+      "grad_norm": 0.4525218904018402,
+      "learning_rate": 1.6958384487457923e-05,
+      "loss": 0.05941899120807648,
+      "step": 670
+    },
+    {
+      "epoch": 4.830324909747293,
+      "grad_norm": 0.3088747560977936,
+      "learning_rate": 1.6628947428627438e-05,
+      "loss": 0.05612848699092865,
+      "step": 671
+    },
+    {
+      "epoch": 4.837545126353791,
+      "grad_norm": 0.330507755279541,
+      "learning_rate": 1.6302602850815397e-05,
+      "loss": 0.06643390655517578,
+      "step": 672
+    },
+    {
+      "epoch": 4.844765342960288,
+      "grad_norm": 0.3919360041618347,
+      "learning_rate": 1.59793562576954e-05,
+      "loss": 0.06949331611394882,
+      "step": 673
+    },
+    {
+      "epoch": 4.851985559566787,
+      "grad_norm": 0.43942415714263916,
+      "learning_rate": 1.5659213100694626e-05,
+      "loss": 0.06495010107755661,
+      "step": 674
+    },
+    {
+      "epoch": 4.859205776173285,
+      "grad_norm": 0.3314034640789032,
+      "learning_rate": 1.5342178778902116e-05,
+      "loss": 0.0641111359000206,
+      "step": 675
+    },
+    {
+      "epoch": 4.866425992779783,
+      "grad_norm": 0.6920318007469177,
+      "learning_rate": 1.5028258638977677e-05,
+      "loss": 0.08164434880018234,
+      "step": 676
+    },
+    {
+      "epoch": 4.873646209386282,
+      "grad_norm": 0.4740530550479889,
+      "learning_rate": 1.4717457975061455e-05,
+      "loss": 0.07530295848846436,
+      "step": 677
+    },
+    {
+      "epoch": 4.88086642599278,
+      "grad_norm": 0.4190375506877899,
+      "learning_rate": 1.4409782028685104e-05,
+      "loss": 0.07053963840007782,
+      "step": 678
+    },
+    {
+      "epoch": 4.888086642599278,
+      "grad_norm": 0.3520694375038147,
+      "learning_rate": 1.4105235988682896e-05,
+      "loss": 0.06393534690141678,
+      "step": 679
+    },
+    {
+      "epoch": 4.8953068592057765,
+      "grad_norm": 0.40102067589759827,
+      "learning_rate": 1.3803824991104663e-05,
+      "loss": 0.07699807733297348,
+      "step": 680
+    },
+    {
+      "epoch": 4.902527075812275,
+      "grad_norm": 0.4686407148838043,
+      "learning_rate": 1.350555411912886e-05,
+      "loss": 0.06640764325857162,
+      "step": 681
+    },
+    {
+      "epoch": 4.909747292418773,
+      "grad_norm": 0.31581881642341614,
+      "learning_rate": 1.3210428402976971e-05,
+      "loss": 0.06866060942411423,
+      "step": 682
+    },
+    {
+      "epoch": 4.9169675090252705,
+      "grad_norm": 0.39863917231559753,
+      "learning_rate": 1.2918452819828663e-05,
+      "loss": 0.05927126109600067,
+      "step": 683
+    },
+    {
+      "epoch": 4.924187725631769,
+      "grad_norm": 0.7777624130249023,
+      "learning_rate": 1.2629632293737903e-05,
+      "loss": 0.07142484188079834,
+      "step": 684
+    },
+    {
+      "epoch": 4.931407942238267,
+      "grad_norm": 0.3229425251483917,
+      "learning_rate": 1.2343971695549727e-05,
+      "loss": 0.06009211018681526,
+      "step": 685
+    },
+    {
+      "epoch": 4.938628158844765,
+      "grad_norm": 0.5577942132949829,
+      "learning_rate": 1.2061475842818338e-05,
+      "loss": 0.08745917677879333,
+      "step": 686
+    },
+    {
+      "epoch": 4.945848375451264,
+      "grad_norm": 0.4703651964664459,
+      "learning_rate": 1.1782149499725714e-05,
+      "loss": 0.059495482593774796,
+      "step": 687
+    },
+    {
+      "epoch": 4.953068592057762,
+      "grad_norm": 0.4242652356624603,
+      "learning_rate": 1.150599737700122e-05,
+      "loss": 0.05950907990336418,
+      "step": 688
+    },
+    {
+      "epoch": 4.96028880866426,
+      "grad_norm": 0.454228013753891,
+      "learning_rate": 1.1233024131842374e-05,
+      "loss": 0.08122330904006958,
+      "step": 689
+    },
+    {
+      "epoch": 4.967509025270758,
+      "grad_norm": 0.28975188732147217,
+      "learning_rate": 1.0963234367836106e-05,
+      "loss": 0.05605383217334747,
+      "step": 690
+    },
+    {
+      "epoch": 4.974729241877256,
+      "grad_norm": 0.38855794072151184,
+      "learning_rate": 1.0696632634881099e-05,
+      "loss": 0.059195615351200104,
+      "step": 691
+    },
+    {
+      "epoch": 4.981949458483754,
+      "grad_norm": 0.46655526757240295,
+      "learning_rate": 1.043322342911126e-05,
+      "loss": 0.07140585780143738,
+      "step": 692
+    },
+    {
+      "epoch": 4.9891696750902526,
+      "grad_norm": 0.4152084290981293,
+      "learning_rate": 1.017301119281977e-05,
+      "loss": 0.07906543463468552,
+      "step": 693
+    },
+    {
+      "epoch": 4.996389891696751,
+      "grad_norm": 0.39406299591064453,
+      "learning_rate": 9.916000314384044e-06,
+      "loss": 0.0740804374217987,
+      "step": 694
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.4732819199562073,
+      "learning_rate": 9.662195128191997e-06,
+      "loss": 0.06949488818645477,
+      "step": 695
+    },
+    {
+      "epoch": 5.007220216606498,
+      "grad_norm": 0.2503197491168976,
+      "learning_rate": 9.411599914568702e-06,
+      "loss": 0.048096612095832825,
+      "step": 696
+    },
+    {
+      "epoch": 5.014440433212997,
+      "grad_norm": 0.20594151318073273,
+      "learning_rate": 9.164218899704336e-06,
+      "loss": 0.044573940336704254,
+      "step": 697
+    },
+    {
+      "epoch": 5.021660649819495,
+      "grad_norm": 0.3241271674633026,
+      "learning_rate": 8.920056255582943e-06,
+      "loss": 0.05117207020521164,
+      "step": 698
+    },
+    {
+      "epoch": 5.028880866425993,
+      "grad_norm": 0.18174150586128235,
+      "learning_rate": 8.679116099911854e-06,
+      "loss": 0.04179506003856659,
+      "step": 699
+    },
+    {
+      "epoch": 5.036101083032491,
+      "grad_norm": 0.1984250396490097,
+      "learning_rate": 8.441402496052496e-06,
+      "loss": 0.053596131503582,
+      "step": 700
+    },
+    {
+      "epoch": 5.043321299638989,
+      "grad_norm": 0.21945425868034363,
+      "learning_rate": 8.206919452951778e-06,
+      "loss": 0.042779602110385895,
+      "step": 701
+    },
+    {
+      "epoch": 5.050541516245487,
+      "grad_norm": 0.260366827249527,
+      "learning_rate": 7.975670925074386e-06,
+      "loss": 0.06215063855051994,
+      "step": 702
+    },
+    {
+      "epoch": 5.0577617328519855,
+      "grad_norm": 0.14613628387451172,
+      "learning_rate": 7.747660812336222e-06,
+      "loss": 0.04405611380934715,
+      "step": 703
+    },
+    {
+      "epoch": 5.064981949458484,
+      "grad_norm": 0.14823344349861145,
+      "learning_rate": 7.52289296003863e-06,
+      "loss": 0.040258534252643585,
+      "step": 704
+    },
+    {
+      "epoch": 5.072202166064982,
+      "grad_norm": 0.11375842988491058,
+      "learning_rate": 7.301371158803383e-06,
+      "loss": 0.037275541573762894,
+      "step": 705
+    },
+    {
+      "epoch": 5.07942238267148,
+      "grad_norm": 0.18186606466770172,
+      "learning_rate": 7.083099144508976e-06,
+      "loss": 0.04733014106750488,
+      "step": 706
+    },
+    {
+      "epoch": 5.086642599277979,
+      "grad_norm": 0.21977707743644714,
+      "learning_rate": 6.8680805982275e-06,
+      "loss": 0.04784165322780609,
+      "step": 707
+    },
+    {
+      "epoch": 5.093862815884476,
+      "grad_norm": 0.2574014663696289,
+      "learning_rate": 6.656319146162515e-06,
+      "loss": 0.05287627875804901,
+      "step": 708
+    },
+    {
+      "epoch": 5.101083032490974,
+      "grad_norm": 0.24309682846069336,
+      "learning_rate": 6.447818359588054e-06,
+      "loss": 0.05101453512907028,
+      "step": 709
+    },
+    {
+      "epoch": 5.108303249097473,
+      "grad_norm": 0.18539798259735107,
+      "learning_rate": 6.242581754788268e-06,
+      "loss": 0.04852500185370445,
+      "step": 710
+    },
+    {
+      "epoch": 5.115523465703971,
+      "grad_norm": 0.2783971130847931,
+      "learning_rate": 6.040612792998123e-06,
+      "loss": 0.04606611281633377,
+      "step": 711
+    },
+    {
+      "epoch": 5.122743682310469,
+      "grad_norm": 0.2432968020439148,
+      "learning_rate": 5.841914880345111e-06,
+      "loss": 0.05015704780817032,
+      "step": 712
+    },
+    {
+      "epoch": 5.129963898916968,
+      "grad_norm": 0.21053971350193024,
+      "learning_rate": 5.64649136779174e-06,
+      "loss": 0.04552324116230011,
+      "step": 713
+    },
+    {
+      "epoch": 5.137184115523466,
+      "grad_norm": 0.2822020947933197,
+      "learning_rate": 5.454345551079043e-06,
+      "loss": 0.04811366647481918,
+      "step": 714
+    },
+    {
+      "epoch": 5.144404332129964,
+      "grad_norm": 0.3009110689163208,
+      "learning_rate": 5.265480670671053e-06,
+      "loss": 0.057499293237924576,
+      "step": 715
+    },
+    {
+      "epoch": 5.1516245487364625,
+      "grad_norm": 0.2482829988002777,
+      "learning_rate": 5.07989991169997e-06,
+      "loss": 0.04851806163787842,
+      "step": 716
+    },
+    {
+      "epoch": 5.15884476534296,
+      "grad_norm": 0.4322669506072998,
+      "learning_rate": 4.8976064039126805e-06,
+      "loss": 0.040396105498075485,
+      "step": 717
+    },
+    {
+      "epoch": 5.166064981949458,
+      "grad_norm": 0.22342568635940552,
+      "learning_rate": 4.718603221617834e-06,
+      "loss": 0.047426290810108185,
+      "step": 718
+    },
+    {
+      "epoch": 5.1732851985559565,
+      "grad_norm": 0.3421519994735718,
+      "learning_rate": 4.542893383634028e-06,
+      "loss": 0.050219837576150894,
+      "step": 719
+    },
+    {
+      "epoch": 5.180505415162455,
+      "grad_norm": 0.2810814678668976,
+      "learning_rate": 4.370479853238863e-06,
+      "loss": 0.039352793246507645,
+      "step": 720
+    },
+    {
+      "epoch": 5.187725631768953,
+      "grad_norm": 0.16700661182403564,
+      "learning_rate": 4.201365538119096e-06,
+      "loss": 0.03363039717078209,
+      "step": 721
+    },
+    {
+      "epoch": 5.194945848375451,
+      "grad_norm": 0.21759076416492462,
+      "learning_rate": 4.035553290321414e-06,
+      "loss": 0.04147165268659592,
+      "step": 722
+    },
+    {
+      "epoch": 5.20216606498195,
+      "grad_norm": 0.13985103368759155,
+      "learning_rate": 3.873045906204453e-06,
+      "loss": 0.043792638927698135,
+      "step": 723
+    },
+    {
+      "epoch": 5.209386281588448,
+      "grad_norm": 0.2625335156917572,
+      "learning_rate": 3.7138461263916513e-06,
+      "loss": 0.05631346255540848,
+      "step": 724
+    },
+    {
+      "epoch": 5.216606498194946,
+      "grad_norm": 0.16157718002796173,
+      "learning_rate": 3.5579566357249793e-06,
+      "loss": 0.03788129985332489,
+      "step": 725
+    },
+    {
+      "epoch": 5.223826714801444,
+      "grad_norm": 0.19516326487064362,
+      "learning_rate": 3.405380063219643e-06,
+      "loss": 0.044141985476017,
+      "step": 726
+    },
+    {
+      "epoch": 5.231046931407942,
+      "grad_norm": 0.15679222345352173,
+      "learning_rate": 3.256118982019851e-06,
+      "loss": 0.04181442782282829,
+      "step": 727
+    },
+    {
+      "epoch": 5.23826714801444,
+      "grad_norm": 0.17116223275661469,
+      "learning_rate": 3.1101759093552728e-06,
+      "loss": 0.036379870027303696,
+      "step": 728
+    },
+    {
+      "epoch": 5.245487364620939,
+      "grad_norm": 0.13156861066818237,
+      "learning_rate": 2.9675533064986936e-06,
+      "loss": 0.035947442054748535,
+      "step": 729
+    },
+    {
+      "epoch": 5.252707581227437,
+      "grad_norm": 0.31538212299346924,
+      "learning_rate": 2.828253578724538e-06,
+      "loss": 0.04313802346587181,
+      "step": 730
+    },
+    {
+      "epoch": 5.259927797833935,
+      "grad_norm": 0.16643384099006653,
+      "learning_rate": 2.6922790752681004e-06,
+      "loss": 0.03961348533630371,
+      "step": 731
+    },
+    {
+      "epoch": 5.2671480144404335,
+      "grad_norm": 0.2200719565153122,
+      "learning_rate": 2.5596320892862013e-06,
+      "loss": 0.051561444997787476,
+      "step": 732
+    },
+    {
+      "epoch": 5.274368231046932,
+      "grad_norm": 0.3769620954990387,
+      "learning_rate": 2.4303148578183497e-06,
+      "loss": 0.07501974701881409,
+      "step": 733
+    },
+    {
+      "epoch": 5.28158844765343,
+      "grad_norm": 0.20830948650836945,
+      "learning_rate": 2.3043295617489746e-06,
+      "loss": 0.04220421612262726,
+      "step": 734
+    },
+    {
+      "epoch": 5.2888086642599275,
+      "grad_norm": 0.19785495102405548,
+      "learning_rate": 2.1816783257708083e-06,
+      "loss": 0.03773656487464905,
+      "step": 735
+    },
+    {
+      "epoch": 5.296028880866426,
+      "grad_norm": 0.2532854974269867,
+      "learning_rate": 2.0623632183489396e-06,
+      "loss": 0.04975757375359535,
+      "step": 736
+    },
+    {
+      "epoch": 5.303249097472924,
+      "grad_norm": 0.21113696694374084,
+      "learning_rate": 1.9463862516859276e-06,
+      "loss": 0.04151391610503197,
+      "step": 737
+    },
+    {
+      "epoch": 5.310469314079422,
+      "grad_norm": 0.41521739959716797,
+      "learning_rate": 1.8337493816879436e-06,
+      "loss": 0.04442136362195015,
+      "step": 738
+    },
+    {
+      "epoch": 5.317689530685921,
+      "grad_norm": 0.21844181418418884,
+      "learning_rate": 1.724454507931683e-06,
+      "loss": 0.04176602512598038,
+      "step": 739
+    },
+    {
+      "epoch": 5.324909747292419,
+      "grad_norm": 0.17051415145397186,
+      "learning_rate": 1.6185034736324156e-06,
+      "loss": 0.03697787970304489,
+      "step": 740
+    },
+    {
+      "epoch": 5.332129963898917,
+      "grad_norm": 0.18966346979141235,
+      "learning_rate": 1.515898065612853e-06,
+      "loss": 0.038028594106435776,
+      "step": 741
+    },
+    {
+      "epoch": 5.3393501805054155,
+      "grad_norm": 0.17065943777561188,
+      "learning_rate": 1.4166400142730407e-06,
+      "loss": 0.035803914070129395,
+      "step": 742
+    },
+    {
+      "epoch": 5.346570397111913,
+      "grad_norm": 0.31374311447143555,
+      "learning_rate": 1.3207309935611367e-06,
+      "loss": 0.056527022272348404,
+      "step": 743
+    },
+    {
+      "epoch": 5.353790613718411,
+      "grad_norm": 0.1854705810546875,
+      "learning_rate": 1.2281726209452781e-06,
+      "loss": 0.043832339346408844,
+      "step": 744
+    },
+    {
+      "epoch": 5.3610108303249095,
+      "grad_norm": 0.14780643582344055,
+      "learning_rate": 1.138966457386137e-06,
+      "loss": 0.03484298288822174,
+      "step": 745
+    },
+    {
+      "epoch": 5.368231046931408,
+      "grad_norm": 0.20580680668354034,
+      "learning_rate": 1.0531140073107626e-06,
+      "loss": 0.042070530354976654,
+      "step": 746
+    },
+    {
+      "epoch": 5.375451263537906,
+      "grad_norm": 0.18943031132221222,
+      "learning_rate": 9.70616718587114e-07,
+      "loss": 0.04158391058444977,
+      "step": 747
+    },
+    {
+      "epoch": 5.382671480144404,
+      "grad_norm": 0.1829371601343155,
+      "learning_rate": 8.914759824996788e-07,
+      "loss": 0.03915674239397049,
+      "step": 748
+    },
+    {
+      "epoch": 5.389891696750903,
+      "grad_norm": 0.23177319765090942,
+      "learning_rate": 8.15693133725981e-07,
+      "loss": 0.03685947507619858,
+      "step": 749
+    },
+    {
+      "epoch": 5.397111913357401,
+      "grad_norm": 0.2238905131816864,
+      "learning_rate": 7.432694503141102e-07,
+      "loss": 0.04124440625309944,
+      "step": 750
+    },
+    {
+      "epoch": 5.404332129963899,
+      "grad_norm": 0.19328288733959198,
+      "learning_rate": 6.742061536611166e-07,
+      "loss": 0.04230578616261482,
+      "step": 751
+    },
+    {
+      "epoch": 5.411552346570397,
+      "grad_norm": 0.2329723834991455,
+      "learning_rate": 6.085044084924718e-07,
+      "loss": 0.04548173397779465,
+      "step": 752
+    },
+    {
+      "epoch": 5.418772563176895,
+      "grad_norm": 0.17935019731521606,
+      "learning_rate": 5.461653228423957e-07,
+      "loss": 0.04122517257928848,
+      "step": 753
+    },
+    {
+      "epoch": 5.425992779783393,
+      "grad_norm": 0.19964119791984558,
+      "learning_rate": 4.871899480351605e-07,
+      "loss": 0.03701728954911232,
+      "step": 754
+    },
+    {
+      "epoch": 5.433212996389892,
+      "grad_norm": 0.17978627979755402,
+      "learning_rate": 4.315792786673489e-07,
+      "loss": 0.04114942252635956,
+      "step": 755
+    },
+    {
+      "epoch": 5.44043321299639,
+      "grad_norm": 0.3241960406303406,
+      "learning_rate": 3.7933425259117914e-07,
+      "loss": 0.0526733361184597,
+      "step": 756
+    },
+    {
+      "epoch": 5.447653429602888,
+      "grad_norm": 0.22060273587703705,
+      "learning_rate": 3.3045575089853954e-07,
+      "loss": 0.04859931021928787,
+      "step": 757
+    },
+    {
+      "epoch": 5.4548736462093865,
+      "grad_norm": 0.22234350442886353,
+      "learning_rate": 2.849445979062226e-07,
+      "loss": 0.0446445494890213,
+      "step": 758
+    },
+    {
+      "epoch": 5.462093862815885,
+      "grad_norm": 0.3016712963581085,
+      "learning_rate": 2.428015611420253e-07,
+      "loss": 0.0628553107380867,
+      "step": 759
+    },
+    {
+      "epoch": 5.469314079422382,
+      "grad_norm": 0.21905440092086792,
+      "learning_rate": 2.040273513317148e-07,
+      "loss": 0.04386765509843826,
+      "step": 760
+    },
+    {
+      "epoch": 5.4765342960288805,
+      "grad_norm": 0.24936382472515106,
+      "learning_rate": 1.686226223872378e-07,
+      "loss": 0.0364193394780159,
+      "step": 761
+    },
+    {
+      "epoch": 5.483754512635379,
+      "grad_norm": 0.21109315752983093,
+      "learning_rate": 1.3658797139541878e-07,
+      "loss": 0.036623530089855194,
+      "step": 762
+    },
+    {
+      "epoch": 5.490974729241877,
+      "grad_norm": 0.3846798241138458,
+      "learning_rate": 1.0792393860814543e-07,
+      "loss": 0.039832159876823425,
+      "step": 763
+    },
+    {
+      "epoch": 5.498194945848375,
+      "grad_norm": 0.3015185296535492,
+      "learning_rate": 8.263100743310937e-08,
+      "loss": 0.054986074566841125,
+      "step": 764
+    },
+    {
+      "epoch": 5.505415162454874,
+      "grad_norm": 0.22440600395202637,
+      "learning_rate": 6.070960442567941e-08,
+      "loss": 0.04008050635457039,
+      "step": 765
+    },
+    {
+      "epoch": 5.512635379061372,
+      "grad_norm": 0.2297099381685257,
+      "learning_rate": 4.216009928172948e-08,
+      "loss": 0.05160742253065109,
+      "step": 766
+    },
+    {
+      "epoch": 5.51985559566787,
+      "grad_norm": 0.2735305726528168,
+      "learning_rate": 2.698280483142135e-08,
+      "loss": 0.047364577651023865,
+      "step": 767
+    },
+    {
+      "epoch": 5.527075812274369,
+      "grad_norm": 0.18135972321033478,
+      "learning_rate": 1.517797703387558e-08,
+      "loss": 0.03480882942676544,
+      "step": 768
+    },
+    {
+      "epoch": 5.534296028880867,
+      "grad_norm": 0.28884103894233704,
+      "learning_rate": 6.745814972908271e-09,
+      "loss": 0.049272023141384125,
+      "step": 769
+    },
+    {
+      "epoch": 5.541516245487364,
+      "grad_norm": 0.22770103812217712,
+      "learning_rate": 1.6864608536115711e-09,
+      "loss": 0.04704172536730766,
+      "step": 770
+    },
+    {
+      "epoch": 5.541516245487364,
+      "eval_loss": 0.7697137594223022,
+      "eval_runtime": 48.9364,
+      "eval_samples_per_second": 2.003,
+      "eval_steps_per_second": 0.511,
+      "step": 770
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 770,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.51023270933007e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}