diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.5503329847542517,
+  "epoch": 0.6003632560955474,
   "eval_steps": 500,
-  "global_step": 9999,
+  "global_step": 10908,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -70000,6 +70000,6369 @@
       "learning_rate": 8.269088862876066e-06,
       "loss": 0.6735,
       "step": 9999
+    },
+    {
+      "epoch": 0.5503880235566074,
+      "grad_norm": 0.8945016264915466,
+      "learning_rate": 8.268760867188586e-06,
+      "loss": 0.7575,
+      "step": 10000
+    },
+    {
+      "epoch": 0.5504430623589631,
+      "grad_norm": 0.7708195447921753,
+      "learning_rate": 8.268432846933974e-06,
+      "loss": 0.6988,
+      "step": 10001
+    },
+    {
+      "epoch": 0.5504981011613187,
+      "grad_norm": 0.7884799838066101,
+      "learning_rate": 8.268104802114696e-06,
+      "loss": 0.8085,
+      "step": 10002
+    },
+    {
+      "epoch": 0.5505531399636744,
+      "grad_norm": 0.7801569104194641,
+      "learning_rate": 8.267776732733217e-06,
+      "loss": 0.886,
+      "step": 10003
+    },
+    {
+      "epoch": 0.5506081787660301,
+      "grad_norm": 0.714645504951477,
+      "learning_rate": 8.267448638792004e-06,
+      "loss": 0.7151,
+      "step": 10004
+    },
+    {
+      "epoch": 0.5506632175683858,
+      "grad_norm": 0.653136134147644,
+      "learning_rate": 8.267120520293519e-06,
+      "loss": 0.6347,
+      "step": 10005
+    },
+    {
+      "epoch": 0.5507182563707413,
+      "grad_norm": 0.8821585774421692,
+      "learning_rate": 8.266792377240233e-06,
+      "loss": 0.6457,
+      "step": 10006
+    },
+    {
+      "epoch": 0.550773295173097,
+      "grad_norm": 0.7056930661201477,
+      "learning_rate": 8.266464209634608e-06,
+      "loss": 0.8709,
+      "step": 10007
+    },
+    {
+      "epoch": 0.5508283339754527,
+      "grad_norm": 0.6505821347236633,
+      "learning_rate": 8.266136017479113e-06,
+      "loss": 0.7674,
+      "step": 10008
+    },
+    {
+      "epoch": 0.5508833727778083,
+      "grad_norm": 0.7947389483451843,
+      "learning_rate": 8.265807800776216e-06,
+      "loss": 0.7882,
+      "step": 10009
+    },
+    {
+      "epoch": 0.550938411580164,
+      "grad_norm": 0.7466071844100952,
+      "learning_rate": 8.265479559528379e-06,
+      "loss": 0.7673,
+      "step": 10010
+    },
+    {
+      "epoch": 0.5509934503825197,
+      "grad_norm": 0.706430971622467,
+      "learning_rate": 8.265151293738074e-06,
+      "loss": 0.7796,
+      "step": 10011
+    },
+    {
+      "epoch": 0.5510484891848754,
+      "grad_norm": 0.7701015472412109,
+      "learning_rate": 8.264823003407765e-06,
+      "loss": 0.7631,
+      "step": 10012
+    },
+    {
+      "epoch": 0.551103527987231,
+      "grad_norm": 0.6923625469207764,
+      "learning_rate": 8.264494688539922e-06,
+      "loss": 0.7659,
+      "step": 10013
+    },
+    {
+      "epoch": 0.5511585667895866,
+      "grad_norm": 0.6585322618484497,
+      "learning_rate": 8.264166349137008e-06,
+      "loss": 0.7248,
+      "step": 10014
+    },
+    {
+      "epoch": 0.5512136055919423,
+      "grad_norm": 0.698451578617096,
+      "learning_rate": 8.263837985201493e-06,
+      "loss": 0.7768,
+      "step": 10015
+    },
+    {
+      "epoch": 0.551268644394298,
+      "grad_norm": 0.7585058808326721,
+      "learning_rate": 8.263509596735847e-06,
+      "loss": 0.8535,
+      "step": 10016
+    },
+    {
+      "epoch": 0.5513236831966536,
+      "grad_norm": 0.6973930597305298,
+      "learning_rate": 8.263181183742536e-06,
+      "loss": 0.8253,
+      "step": 10017
+    },
+    {
+      "epoch": 0.5513787219990093,
+      "grad_norm": 0.6752467751502991,
+      "learning_rate": 8.26285274622403e-06,
+      "loss": 0.7402,
+      "step": 10018
+    },
+    {
+      "epoch": 0.551433760801365,
+      "grad_norm": 0.717555820941925,
+      "learning_rate": 8.262524284182794e-06,
+      "loss": 0.8057,
+      "step": 10019
+    },
+    {
+      "epoch": 0.5514887996037207,
+      "grad_norm": 0.6975438594818115,
+      "learning_rate": 8.2621957976213e-06,
+      "loss": 0.803,
+      "step": 10020
+    },
+    {
+      "epoch": 0.5515438384060762,
+      "grad_norm": 0.667797327041626,
+      "learning_rate": 8.261867286542016e-06,
+      "loss": 0.7387,
+      "step": 10021
+    },
+    {
+      "epoch": 0.5515988772084319,
+      "grad_norm": 0.7330532670021057,
+      "learning_rate": 8.261538750947411e-06,
+      "loss": 0.8143,
+      "step": 10022
+    },
+    {
+      "epoch": 0.5516539160107876,
+      "grad_norm": 0.7034017443656921,
+      "learning_rate": 8.261210190839952e-06,
+      "loss": 0.739,
+      "step": 10023
+    },
+    {
+      "epoch": 0.5517089548131433,
+      "grad_norm": 0.709284245967865,
+      "learning_rate": 8.260881606222113e-06,
+      "loss": 0.8021,
+      "step": 10024
+    },
+    {
+      "epoch": 0.5517639936154989,
+      "grad_norm": 0.7587909698486328,
+      "learning_rate": 8.260552997096359e-06,
+      "loss": 0.8346,
+      "step": 10025
+    },
+    {
+      "epoch": 0.5518190324178546,
+      "grad_norm": 0.7413986325263977,
+      "learning_rate": 8.26022436346516e-06,
+      "loss": 0.6777,
+      "step": 10026
+    },
+    {
+      "epoch": 0.5518740712202103,
+      "grad_norm": 0.7112768292427063,
+      "learning_rate": 8.25989570533099e-06,
+      "loss": 0.7017,
+      "step": 10027
+    },
+    {
+      "epoch": 0.551929110022566,
+      "grad_norm": 0.7097088098526001,
+      "learning_rate": 8.259567022696315e-06,
+      "loss": 0.7315,
+      "step": 10028
+    },
+    {
+      "epoch": 0.5519841488249215,
+      "grad_norm": 0.6544226408004761,
+      "learning_rate": 8.259238315563606e-06,
+      "loss": 0.7729,
+      "step": 10029
+    },
+    {
+      "epoch": 0.5520391876272772,
+      "grad_norm": 0.6892885565757751,
+      "learning_rate": 8.258909583935335e-06,
+      "loss": 0.7919,
+      "step": 10030
+    },
+    {
+      "epoch": 0.5520942264296329,
+      "grad_norm": 0.697424054145813,
+      "learning_rate": 8.258580827813972e-06,
+      "loss": 0.7514,
+      "step": 10031
+    },
+    {
+      "epoch": 0.5521492652319886,
+      "grad_norm": 0.7021437883377075,
+      "learning_rate": 8.258252047201989e-06,
+      "loss": 0.747,
+      "step": 10032
+    },
+    {
+      "epoch": 0.5522043040343442,
+      "grad_norm": 0.6974816918373108,
+      "learning_rate": 8.257923242101854e-06,
+      "loss": 0.7245,
+      "step": 10033
+    },
+    {
+      "epoch": 0.5522593428366999,
+      "grad_norm": 0.6645311117172241,
+      "learning_rate": 8.25759441251604e-06,
+      "loss": 0.649,
+      "step": 10034
+    },
+    {
+      "epoch": 0.5523143816390556,
+      "grad_norm": 0.7223736643791199,
+      "learning_rate": 8.25726555844702e-06,
+      "loss": 0.7792,
+      "step": 10035
+    },
+    {
+      "epoch": 0.5523694204414112,
+      "grad_norm": 0.7253531813621521,
+      "learning_rate": 8.256936679897262e-06,
+      "loss": 0.7636,
+      "step": 10036
+    },
+    {
+      "epoch": 0.5524244592437668,
+      "grad_norm": 0.6979514956474304,
+      "learning_rate": 8.256607776869241e-06,
+      "loss": 0.7929,
+      "step": 10037
+    },
+    {
+      "epoch": 0.5524794980461225,
+      "grad_norm": 0.7442019581794739,
+      "learning_rate": 8.25627884936543e-06,
+      "loss": 0.6984,
+      "step": 10038
+    },
+    {
+      "epoch": 0.5525345368484782,
+      "grad_norm": 0.7519513964653015,
+      "learning_rate": 8.255949897388294e-06,
+      "loss": 0.7228,
+      "step": 10039
+    },
+    {
+      "epoch": 0.5525895756508339,
+      "grad_norm": 0.7302790880203247,
+      "learning_rate": 8.255620920940313e-06,
+      "loss": 0.7555,
+      "step": 10040
+    },
+    {
+      "epoch": 0.5526446144531895,
+      "grad_norm": 0.6521434187889099,
+      "learning_rate": 8.255291920023956e-06,
+      "loss": 0.7825,
+      "step": 10041
+    },
+    {
+      "epoch": 0.5526996532555452,
+      "grad_norm": 0.8270126581192017,
+      "learning_rate": 8.254962894641695e-06,
+      "loss": 0.7939,
+      "step": 10042
+    },
+    {
+      "epoch": 0.5527546920579008,
+      "grad_norm": 0.7209310531616211,
+      "learning_rate": 8.254633844796007e-06,
+      "loss": 0.8286,
+      "step": 10043
+    },
+    {
+      "epoch": 0.5528097308602565,
+      "grad_norm": 0.6506814360618591,
+      "learning_rate": 8.25430477048936e-06,
+      "loss": 0.7209,
+      "step": 10044
+    },
+    {
+      "epoch": 0.5528647696626121,
+      "grad_norm": 0.6914637684822083,
+      "learning_rate": 8.25397567172423e-06,
+      "loss": 0.705,
+      "step": 10045
+    },
+    {
+      "epoch": 0.5529198084649678,
+      "grad_norm": 0.8369725942611694,
+      "learning_rate": 8.253646548503091e-06,
+      "loss": 0.8254,
+      "step": 10046
+    },
+    {
+      "epoch": 0.5529748472673235,
+      "grad_norm": 0.7809324860572815,
+      "learning_rate": 8.253317400828414e-06,
+      "loss": 0.8117,
+      "step": 10047
+    },
+    {
+      "epoch": 0.5530298860696792,
+      "grad_norm": 0.7184550762176514,
+      "learning_rate": 8.252988228702676e-06,
+      "loss": 0.738,
+      "step": 10048
+    },
+    {
+      "epoch": 0.5530849248720348,
+      "grad_norm": 0.7111478447914124,
+      "learning_rate": 8.252659032128347e-06,
+      "loss": 0.7143,
+      "step": 10049
+    },
+    {
+      "epoch": 0.5531399636743904,
+      "grad_norm": 0.7506794333457947,
+      "learning_rate": 8.252329811107905e-06,
+      "loss": 0.7721,
+      "step": 10050
+    },
+    {
+      "epoch": 0.5531950024767461,
+      "grad_norm": 0.7700625658035278,
+      "learning_rate": 8.252000565643823e-06,
+      "loss": 0.7993,
+      "step": 10051
+    },
+    {
+      "epoch": 0.5532500412791017,
+      "grad_norm": 0.6985816955566406,
+      "learning_rate": 8.251671295738575e-06,
+      "loss": 0.7461,
+      "step": 10052
+    },
+    {
+      "epoch": 0.5533050800814574,
+      "grad_norm": 0.6932175755500793,
+      "learning_rate": 8.251342001394635e-06,
+      "loss": 0.6804,
+      "step": 10053
+    },
+    {
+      "epoch": 0.5533601188838131,
+      "grad_norm": 0.8060765266418457,
+      "learning_rate": 8.25101268261448e-06,
+      "loss": 0.7137,
+      "step": 10054
+    },
+    {
+      "epoch": 0.5534151576861688,
+      "grad_norm": 0.6853482127189636,
+      "learning_rate": 8.250683339400582e-06,
+      "loss": 0.7229,
+      "step": 10055
+    },
+    {
+      "epoch": 0.5534701964885244,
+      "grad_norm": 0.7581862211227417,
+      "learning_rate": 8.25035397175542e-06,
+      "loss": 0.8091,
+      "step": 10056
+    },
+    {
+      "epoch": 0.55352523529088,
+      "grad_norm": 0.7375245094299316,
+      "learning_rate": 8.250024579681466e-06,
+      "loss": 0.7234,
+      "step": 10057
+    },
+    {
+      "epoch": 0.5535802740932357,
+      "grad_norm": 0.7904585599899292,
+      "learning_rate": 8.249695163181198e-06,
+      "loss": 0.7295,
+      "step": 10058
+    },
+    {
+      "epoch": 0.5536353128955914,
+      "grad_norm": 0.6593602895736694,
+      "learning_rate": 8.249365722257092e-06,
+      "loss": 0.7492,
+      "step": 10059
+    },
+    {
+      "epoch": 0.553690351697947,
+      "grad_norm": 0.7226922512054443,
+      "learning_rate": 8.249036256911622e-06,
+      "loss": 0.8177,
+      "step": 10060
+    },
+    {
+      "epoch": 0.5537453905003027,
+      "grad_norm": 0.7268722653388977,
+      "learning_rate": 8.248706767147265e-06,
+      "loss": 0.8059,
+      "step": 10061
+    },
+    {
+      "epoch": 0.5538004293026584,
+      "grad_norm": 0.7797269225120544,
+      "learning_rate": 8.248377252966499e-06,
+      "loss": 0.8122,
+      "step": 10062
+    },
+    {
+      "epoch": 0.5538554681050141,
+      "grad_norm": 0.7199145555496216,
+      "learning_rate": 8.248047714371797e-06,
+      "loss": 0.7312,
+      "step": 10063
+    },
+    {
+      "epoch": 0.5539105069073696,
+      "grad_norm": 0.6950703263282776,
+      "learning_rate": 8.24771815136564e-06,
+      "loss": 0.757,
+      "step": 10064
+    },
+    {
+      "epoch": 0.5539655457097253,
+      "grad_norm": 0.6413441896438599,
+      "learning_rate": 8.247388563950502e-06,
+      "loss": 0.6955,
+      "step": 10065
+    },
+    {
+      "epoch": 0.554020584512081,
+      "grad_norm": 0.7650758624076843,
+      "learning_rate": 8.24705895212886e-06,
+      "loss": 0.8355,
+      "step": 10066
+    },
+    {
+      "epoch": 0.5540756233144367,
+      "grad_norm": 0.7067090272903442,
+      "learning_rate": 8.246729315903192e-06,
+      "loss": 0.7409,
+      "step": 10067
+    },
+    {
+      "epoch": 0.5541306621167923,
+      "grad_norm": 0.7763532996177673,
+      "learning_rate": 8.246399655275976e-06,
+      "loss": 0.8097,
+      "step": 10068
+    },
+    {
+      "epoch": 0.554185700919148,
+      "grad_norm": 0.6865057945251465,
+      "learning_rate": 8.246069970249689e-06,
+      "loss": 0.7597,
+      "step": 10069
+    },
+    {
+      "epoch": 0.5542407397215037,
+      "grad_norm": 0.7643107771873474,
+      "learning_rate": 8.24574026082681e-06,
+      "loss": 0.7403,
+      "step": 10070
+    },
+    {
+      "epoch": 0.5542957785238594,
+      "grad_norm": 0.7354087829589844,
+      "learning_rate": 8.245410527009815e-06,
+      "loss": 0.8896,
+      "step": 10071
+    },
+    {
+      "epoch": 0.5543508173262149,
+      "grad_norm": 0.7971135973930359,
+      "learning_rate": 8.245080768801183e-06,
+      "loss": 0.7738,
+      "step": 10072
+    },
+    {
+      "epoch": 0.5544058561285706,
+      "grad_norm": 1.0506731271743774,
+      "learning_rate": 8.244750986203394e-06,
+      "loss": 0.7888,
+      "step": 10073
+    },
+    {
+      "epoch": 0.5544608949309263,
+      "grad_norm": 0.8305885195732117,
+      "learning_rate": 8.244421179218925e-06,
+      "loss": 0.8186,
+      "step": 10074
+    },
+    {
+      "epoch": 0.554515933733282,
+      "grad_norm": 0.9507874250411987,
+      "learning_rate": 8.244091347850253e-06,
+      "loss": 0.7975,
+      "step": 10075
+    },
+    {
+      "epoch": 0.5545709725356376,
+      "grad_norm": 0.7146797776222229,
+      "learning_rate": 8.243761492099861e-06,
+      "loss": 0.6895,
+      "step": 10076
+    },
+    {
+      "epoch": 0.5546260113379933,
+      "grad_norm": 0.734990656375885,
+      "learning_rate": 8.243431611970225e-06,
+      "loss": 0.8087,
+      "step": 10077
+    },
+    {
+      "epoch": 0.554681050140349,
+      "grad_norm": 0.6807795166969299,
+      "learning_rate": 8.243101707463825e-06,
+      "loss": 0.7861,
+      "step": 10078
+    },
+    {
+      "epoch": 0.5547360889427047,
+      "grad_norm": 0.7412874698638916,
+      "learning_rate": 8.242771778583142e-06,
+      "loss": 0.7864,
+      "step": 10079
+    },
+    {
+      "epoch": 0.5547911277450602,
+      "grad_norm": 0.6655074954032898,
+      "learning_rate": 8.242441825330652e-06,
+      "loss": 0.6554,
+      "step": 10080
+    },
+    {
+      "epoch": 0.5548461665474159,
+      "grad_norm": 0.7549700140953064,
+      "learning_rate": 8.242111847708838e-06,
+      "loss": 0.8031,
+      "step": 10081
+    },
+    {
+      "epoch": 0.5549012053497716,
+      "grad_norm": 0.8907766342163086,
+      "learning_rate": 8.241781845720181e-06,
+      "loss": 0.8068,
+      "step": 10082
+    },
+    {
+      "epoch": 0.5549562441521273,
+      "grad_norm": 0.7347774505615234,
+      "learning_rate": 8.241451819367157e-06,
+      "loss": 0.7453,
+      "step": 10083
+    },
+    {
+      "epoch": 0.5550112829544829,
+      "grad_norm": 0.6856632828712463,
+      "learning_rate": 8.24112176865225e-06,
+      "loss": 0.6235,
+      "step": 10084
+    },
+    {
+      "epoch": 0.5550663217568386,
+      "grad_norm": 0.7134507298469543,
+      "learning_rate": 8.24079169357794e-06,
+      "loss": 0.7991,
+      "step": 10085
+    },
+    {
+      "epoch": 0.5551213605591943,
+      "grad_norm": 0.7814854383468628,
+      "learning_rate": 8.240461594146704e-06,
+      "loss": 0.7681,
+      "step": 10086
+    },
+    {
+      "epoch": 0.5551763993615499,
+      "grad_norm": 0.6893261671066284,
+      "learning_rate": 8.240131470361028e-06,
+      "loss": 0.7746,
+      "step": 10087
+    },
+    {
+      "epoch": 0.5552314381639055,
+      "grad_norm": 0.925003170967102,
+      "learning_rate": 8.239801322223393e-06,
+      "loss": 0.7621,
+      "step": 10088
+    },
+    {
+      "epoch": 0.5552864769662612,
+      "grad_norm": 0.6261017918586731,
+      "learning_rate": 8.239471149736277e-06,
+      "loss": 0.7673,
+      "step": 10089
+    },
+    {
+      "epoch": 0.5553415157686169,
+      "grad_norm": 0.7268226146697998,
+      "learning_rate": 8.239140952902162e-06,
+      "loss": 0.7375,
+      "step": 10090
+    },
+    {
+      "epoch": 0.5553965545709726,
+      "grad_norm": 0.8062194585800171,
+      "learning_rate": 8.238810731723532e-06,
+      "loss": 0.8002,
+      "step": 10091
+    },
+    {
+      "epoch": 0.5554515933733282,
+      "grad_norm": 0.892842173576355,
+      "learning_rate": 8.238480486202867e-06,
+      "loss": 0.7959,
+      "step": 10092
+    },
+    {
+      "epoch": 0.5555066321756839,
+      "grad_norm": 0.7530377507209778,
+      "learning_rate": 8.23815021634265e-06,
+      "loss": 0.8137,
+      "step": 10093
+    },
+    {
+      "epoch": 0.5555616709780395,
+      "grad_norm": 0.6994850635528564,
+      "learning_rate": 8.237819922145364e-06,
+      "loss": 0.7966,
+      "step": 10094
+    },
+    {
+      "epoch": 0.5556167097803951,
+      "grad_norm": 0.8502941727638245,
+      "learning_rate": 8.237489603613488e-06,
+      "loss": 0.7668,
+      "step": 10095
+    },
+    {
+      "epoch": 0.5556717485827508,
+      "grad_norm": 0.6583576798439026,
+      "learning_rate": 8.237159260749507e-06,
+      "loss": 0.7379,
+      "step": 10096
+    },
+    {
+      "epoch": 0.5557267873851065,
+      "grad_norm": 0.9539539217948914,
+      "learning_rate": 8.236828893555904e-06,
+      "loss": 0.7563,
+      "step": 10097
+    },
+    {
+      "epoch": 0.5557818261874622,
+      "grad_norm": 0.7446413040161133,
+      "learning_rate": 8.236498502035162e-06,
+      "loss": 0.7329,
+      "step": 10098
+    },
+    {
+      "epoch": 0.5558368649898178,
+      "grad_norm": 0.8950835466384888,
+      "learning_rate": 8.236168086189761e-06,
+      "loss": 0.8144,
+      "step": 10099
+    },
+    {
+      "epoch": 0.5558919037921735,
+      "grad_norm": 0.7255009412765503,
+      "learning_rate": 8.235837646022191e-06,
+      "loss": 0.6946,
+      "step": 10100
+    },
+    {
+      "epoch": 0.5559469425945291,
+      "grad_norm": 0.6983402967453003,
+      "learning_rate": 8.235507181534929e-06,
+      "loss": 0.7371,
+      "step": 10101
+    },
+    {
+      "epoch": 0.5560019813968848,
+      "grad_norm": 1.043593168258667,
+      "learning_rate": 8.235176692730463e-06,
+      "loss": 0.6763,
+      "step": 10102
+    },
+    {
+      "epoch": 0.5560570201992404,
+      "grad_norm": 0.7452800869941711,
+      "learning_rate": 8.234846179611272e-06,
+      "loss": 0.8945,
+      "step": 10103
+    },
+    {
+      "epoch": 0.5561120590015961,
+      "grad_norm": 0.6367164254188538,
+      "learning_rate": 8.234515642179845e-06,
+      "loss": 0.6542,
+      "step": 10104
+    },
+    {
+      "epoch": 0.5561670978039518,
+      "grad_norm": 0.8377598524093628,
+      "learning_rate": 8.234185080438664e-06,
+      "loss": 0.787,
+      "step": 10105
+    },
+    {
+      "epoch": 0.5562221366063075,
+      "grad_norm": 0.7353680729866028,
+      "learning_rate": 8.233854494390214e-06,
+      "loss": 0.6391,
+      "step": 10106
+    },
+    {
+      "epoch": 0.5562771754086631,
+      "grad_norm": 0.7431599497795105,
+      "learning_rate": 8.233523884036977e-06,
+      "loss": 0.8221,
+      "step": 10107
+    },
+    {
+      "epoch": 0.5563322142110187,
+      "grad_norm": 0.7292743921279907,
+      "learning_rate": 8.233193249381442e-06,
+      "loss": 0.7791,
+      "step": 10108
+    },
+    {
+      "epoch": 0.5563872530133744,
+      "grad_norm": 0.7251895666122437,
+      "learning_rate": 8.232862590426091e-06,
+      "loss": 0.7993,
+      "step": 10109
+    },
+    {
+      "epoch": 0.5564422918157301,
+      "grad_norm": 0.7373167276382446,
+      "learning_rate": 8.23253190717341e-06,
+      "loss": 0.861,
+      "step": 10110
+    },
+    {
+      "epoch": 0.5564973306180857,
+      "grad_norm": 0.6689401268959045,
+      "learning_rate": 8.232201199625887e-06,
+      "loss": 0.7002,
+      "step": 10111
+    },
+    {
+      "epoch": 0.5565523694204414,
+      "grad_norm": 0.7405139207839966,
+      "learning_rate": 8.231870467786003e-06,
+      "loss": 0.8041,
+      "step": 10112
+    },
+    {
+      "epoch": 0.5566074082227971,
+      "grad_norm": 0.7561736702919006,
+      "learning_rate": 8.231539711656246e-06,
+      "loss": 0.7687,
+      "step": 10113
+    },
+    {
+      "epoch": 0.5566624470251528,
+      "grad_norm": 0.6857489943504333,
+      "learning_rate": 8.231208931239103e-06,
+      "loss": 0.7175,
+      "step": 10114
+    },
+    {
+      "epoch": 0.5567174858275084,
+      "grad_norm": 0.7410408854484558,
+      "learning_rate": 8.230878126537057e-06,
+      "loss": 0.7337,
+      "step": 10115
+    },
+    {
+      "epoch": 0.556772524629864,
+      "grad_norm": 0.7533249258995056,
+      "learning_rate": 8.230547297552595e-06,
+      "loss": 0.7226,
+      "step": 10116
+    },
+    {
+      "epoch": 0.5568275634322197,
+      "grad_norm": 0.6227561235427856,
+      "learning_rate": 8.230216444288207e-06,
+      "loss": 0.711,
+      "step": 10117
+    },
+    {
+      "epoch": 0.5568826022345754,
+      "grad_norm": 0.6790871024131775,
+      "learning_rate": 8.229885566746373e-06,
+      "loss": 0.728,
+      "step": 10118
+    },
+    {
+      "epoch": 0.556937641036931,
+      "grad_norm": 1.0007857084274292,
+      "learning_rate": 8.229554664929587e-06,
+      "loss": 0.9193,
+      "step": 10119
+    },
+    {
+      "epoch": 0.5569926798392867,
+      "grad_norm": 0.7167220711708069,
+      "learning_rate": 8.229223738840331e-06,
+      "loss": 0.8288,
+      "step": 10120
+    },
+    {
+      "epoch": 0.5570477186416424,
+      "grad_norm": 0.8037107586860657,
+      "learning_rate": 8.228892788481095e-06,
+      "loss": 0.8462,
+      "step": 10121
+    },
+    {
+      "epoch": 0.5571027574439981,
+      "grad_norm": 0.7355597615242004,
+      "learning_rate": 8.228561813854363e-06,
+      "loss": 0.7998,
+      "step": 10122
+    },
+    {
+      "epoch": 0.5571577962463536,
+      "grad_norm": 0.7384124994277954,
+      "learning_rate": 8.228230814962625e-06,
+      "loss": 0.7861,
+      "step": 10123
+    },
+    {
+      "epoch": 0.5572128350487093,
+      "grad_norm": 0.8170364499092102,
+      "learning_rate": 8.227899791808371e-06,
+      "loss": 0.8005,
+      "step": 10124
+    },
+    {
+      "epoch": 0.557267873851065,
+      "grad_norm": 0.678702175617218,
+      "learning_rate": 8.227568744394084e-06,
+      "loss": 0.7408,
+      "step": 10125
+    },
+    {
+      "epoch": 0.5573229126534207,
+      "grad_norm": 0.7212443947792053,
+      "learning_rate": 8.227237672722255e-06,
+      "loss": 0.7127,
+      "step": 10126
+    },
+    {
+      "epoch": 0.5573779514557763,
+      "grad_norm": 0.7035290002822876,
+      "learning_rate": 8.22690657679537e-06,
+      "loss": 0.8263,
+      "step": 10127
+    },
+    {
+      "epoch": 0.557432990258132,
+      "grad_norm": 0.6535285115242004,
+      "learning_rate": 8.226575456615921e-06,
+      "loss": 0.6979,
+      "step": 10128
+    },
+    {
+      "epoch": 0.5574880290604877,
+      "grad_norm": 0.7353794574737549,
+      "learning_rate": 8.226244312186396e-06,
+      "loss": 0.6838,
+      "step": 10129
+    },
+    {
+      "epoch": 0.5575430678628434,
+      "grad_norm": 0.5839618444442749,
+      "learning_rate": 8.225913143509278e-06,
+      "loss": 0.5925,
+      "step": 10130
+    },
+    {
+      "epoch": 0.5575981066651989,
+      "grad_norm": 0.6922228336334229,
+      "learning_rate": 8.225581950587063e-06,
+      "loss": 0.6808,
+      "step": 10131
+    },
+    {
+      "epoch": 0.5576531454675546,
+      "grad_norm": 0.753989040851593,
+      "learning_rate": 8.225250733422236e-06,
+      "loss": 0.6567,
+      "step": 10132
+    },
+    {
+      "epoch": 0.5577081842699103,
+      "grad_norm": 0.7327600717544556,
+      "learning_rate": 8.22491949201729e-06,
+      "loss": 0.8311,
+      "step": 10133
+    },
+    {
+      "epoch": 0.557763223072266,
+      "grad_norm": 0.6435133218765259,
+      "learning_rate": 8.224588226374712e-06,
+      "loss": 0.6684,
+      "step": 10134
+    },
+    {
+      "epoch": 0.5578182618746216,
+      "grad_norm": 0.6402057409286499,
+      "learning_rate": 8.22425693649699e-06,
+      "loss": 0.7569,
+      "step": 10135
+    },
+    {
+      "epoch": 0.5578733006769773,
+      "grad_norm": 0.7454472780227661,
+      "learning_rate": 8.223925622386617e-06,
+      "loss": 0.7908,
+      "step": 10136
+    },
+    {
+      "epoch": 0.557928339479333,
+      "grad_norm": 0.7373154759407043,
+      "learning_rate": 8.223594284046084e-06,
+      "loss": 0.8232,
+      "step": 10137
+    },
+    {
+      "epoch": 0.5579833782816885,
+      "grad_norm": 0.6478374004364014,
+      "learning_rate": 8.223262921477878e-06,
+      "loss": 0.7353,
+      "step": 10138
+    },
+    {
+      "epoch": 0.5580384170840442,
+      "grad_norm": 0.715212881565094,
+      "learning_rate": 8.222931534684488e-06,
+      "loss": 0.729,
+      "step": 10139
+    },
+    {
+      "epoch": 0.5580934558863999,
+      "grad_norm": 0.9226915240287781,
+      "learning_rate": 8.22260012366841e-06,
+      "loss": 0.7846,
+      "step": 10140
+    },
+    {
+      "epoch": 0.5581484946887556,
+      "grad_norm": 0.6481993198394775,
+      "learning_rate": 8.222268688432132e-06,
+      "loss": 0.6955,
+      "step": 10141
+    },
+    {
+      "epoch": 0.5582035334911112,
+      "grad_norm": 0.7240349054336548,
+      "learning_rate": 8.221937228978145e-06,
+      "loss": 0.7956,
+      "step": 10142
+    },
+    {
+      "epoch": 0.5582585722934669,
+      "grad_norm": 0.7089122533798218,
+      "learning_rate": 8.221605745308939e-06,
+      "loss": 0.7481,
+      "step": 10143
+    },
+    {
+      "epoch": 0.5583136110958226,
+      "grad_norm": 0.7292537093162537,
+      "learning_rate": 8.221274237427009e-06,
+      "loss": 0.7797,
+      "step": 10144
+    },
+    {
+      "epoch": 0.5583686498981782,
+      "grad_norm": 0.7104652523994446,
+      "learning_rate": 8.220942705334841e-06,
+      "loss": 0.7966,
+      "step": 10145
+    },
+    {
+      "epoch": 0.5584236887005338,
+      "grad_norm": 0.7656546831130981,
+      "learning_rate": 8.220611149034931e-06,
+      "loss": 0.7541,
+      "step": 10146
+    },
+    {
+      "epoch": 0.5584787275028895,
+      "grad_norm": 0.7618892788887024,
+      "learning_rate": 8.22027956852977e-06,
+      "loss": 0.6994,
+      "step": 10147
+    },
+    {
+      "epoch": 0.5585337663052452,
+      "grad_norm": 0.6445756554603577,
+      "learning_rate": 8.219947963821851e-06,
+      "loss": 0.7303,
+      "step": 10148
+    },
+    {
+      "epoch": 0.5585888051076009,
+      "grad_norm": 0.6529820561408997,
+      "learning_rate": 8.219616334913663e-06,
+      "loss": 0.7008,
+      "step": 10149
+    },
+    {
+      "epoch": 0.5586438439099565,
+      "grad_norm": 0.6890642046928406,
+      "learning_rate": 8.219284681807703e-06,
+      "loss": 0.8124,
+      "step": 10150
+    },
+    {
+      "epoch": 0.5586988827123122,
+      "grad_norm": 0.7273370027542114,
+      "learning_rate": 8.218953004506458e-06,
+      "loss": 0.7507,
+      "step": 10151
+    },
+    {
+      "epoch": 0.5587539215146679,
+      "grad_norm": 0.7239277362823486,
+      "learning_rate": 8.218621303012425e-06,
+      "loss": 0.7929,
+      "step": 10152
+    },
+    {
+      "epoch": 0.5588089603170235,
+      "grad_norm": 0.660275399684906,
+      "learning_rate": 8.218289577328096e-06,
+      "loss": 0.7418,
+      "step": 10153
+    },
+    {
+      "epoch": 0.5588639991193791,
+      "grad_norm": 0.7406648993492126,
+      "learning_rate": 8.217957827455965e-06,
+      "loss": 0.8072,
+      "step": 10154
+    },
+    {
+      "epoch": 0.5589190379217348,
+      "grad_norm": 0.7051703333854675,
+      "learning_rate": 8.217626053398522e-06,
+      "loss": 0.6562,
+      "step": 10155
+    },
+    {
+      "epoch": 0.5589740767240905,
+      "grad_norm": 0.93423992395401,
+      "learning_rate": 8.217294255158266e-06,
+      "loss": 0.738,
+      "step": 10156
+    },
+    {
+      "epoch": 0.5590291155264462,
+      "grad_norm": 0.8362720608711243,
+      "learning_rate": 8.216962432737685e-06,
+      "loss": 0.8585,
+      "step": 10157
+    },
+    {
+      "epoch": 0.5590841543288018,
+      "grad_norm": 0.9195587038993835,
+      "learning_rate": 8.216630586139277e-06,
+      "loss": 0.8778,
+      "step": 10158
+    },
+    {
+      "epoch": 0.5591391931311575,
+      "grad_norm": 0.7181550860404968,
+      "learning_rate": 8.216298715365534e-06,
+      "loss": 0.702,
+      "step": 10159
+    },
+    {
+      "epoch": 0.5591942319335131,
+      "grad_norm": 0.6900259852409363,
+      "learning_rate": 8.21596682041895e-06,
+      "loss": 0.7652,
+      "step": 10160
+    },
+    {
+      "epoch": 0.5592492707358688,
+      "grad_norm": 0.7523833513259888,
+      "learning_rate": 8.215634901302022e-06,
+      "loss": 0.7881,
+      "step": 10161
+    },
+    {
+      "epoch": 0.5593043095382244,
+      "grad_norm": 0.6659645438194275,
+      "learning_rate": 8.215302958017241e-06,
+      "loss": 0.694,
+      "step": 10162
+    },
+    {
+      "epoch": 0.5593593483405801,
+      "grad_norm": 0.8898606300354004,
+      "learning_rate": 8.214970990567105e-06,
+      "loss": 0.8534,
+      "step": 10163
+    },
+    {
+      "epoch": 0.5594143871429358,
+      "grad_norm": 0.6759241819381714,
+      "learning_rate": 8.214638998954108e-06,
+      "loss": 0.8241,
+      "step": 10164
+    },
+    {
+      "epoch": 0.5594694259452915,
+      "grad_norm": 0.7136911749839783,
+      "learning_rate": 8.214306983180744e-06,
+      "loss": 0.7846,
+      "step": 10165
+    },
+    {
+      "epoch": 0.559524464747647,
+      "grad_norm": 0.6781616806983948,
+      "learning_rate": 8.213974943249509e-06,
+      "loss": 0.7116,
+      "step": 10166
+    },
+    {
+      "epoch": 0.5595795035500027,
+      "grad_norm": 0.7134156227111816,
+      "learning_rate": 8.213642879162898e-06,
+      "loss": 0.7537,
+      "step": 10167
+    },
+    {
+      "epoch": 0.5596345423523584,
+      "grad_norm": 1.306710124015808,
+      "learning_rate": 8.213310790923408e-06,
+      "loss": 0.8506,
+      "step": 10168
+    },
+    {
+      "epoch": 0.5596895811547141,
+      "grad_norm": 0.725304901599884,
+      "learning_rate": 8.212978678533534e-06,
+      "loss": 0.8115,
+      "step": 10169
+    },
+    {
+      "epoch": 0.5597446199570697,
+      "grad_norm": 0.7833520174026489,
+      "learning_rate": 8.212646541995772e-06,
+      "loss": 0.919,
+      "step": 10170
+    },
+    {
+      "epoch": 0.5597996587594254,
+      "grad_norm": 0.6938104033470154,
+      "learning_rate": 8.212314381312621e-06,
+      "loss": 0.7303,
+      "step": 10171
+    },
+    {
+      "epoch": 0.5598546975617811,
+      "grad_norm": 0.6860232949256897,
+      "learning_rate": 8.211982196486573e-06,
+      "loss": 0.7709,
+      "step": 10172
+    },
+    {
+      "epoch": 0.5599097363641368,
+      "grad_norm": 0.6611567139625549,
+      "learning_rate": 8.211649987520126e-06,
+      "loss": 0.7711,
+      "step": 10173
+    },
+    {
+      "epoch": 0.5599647751664923,
+      "grad_norm": 0.8603463172912598,
+      "learning_rate": 8.211317754415778e-06,
+      "loss": 0.8527,
+      "step": 10174
+    },
+    {
+      "epoch": 0.560019813968848,
+      "grad_norm": 0.7350558638572693,
+      "learning_rate": 8.210985497176025e-06,
+      "loss": 0.8148,
+      "step": 10175
+    },
+    {
+      "epoch": 0.5600748527712037,
+      "grad_norm": 0.6881470084190369,
+      "learning_rate": 8.210653215803365e-06,
+      "loss": 0.7526,
+      "step": 10176
+    },
+    {
+      "epoch": 0.5601298915735594,
+      "grad_norm": 0.6879626512527466,
+      "learning_rate": 8.210320910300296e-06,
+      "loss": 0.7649,
+      "step": 10177
+    },
+    {
+      "epoch": 0.560184930375915,
+      "grad_norm": 0.6843587160110474,
+      "learning_rate": 8.209988580669312e-06,
+      "loss": 0.8131,
+      "step": 10178
+    },
+    {
+      "epoch": 0.5602399691782707,
+      "grad_norm": 0.6684302687644958,
+      "learning_rate": 8.209656226912915e-06,
+      "loss": 0.7256,
+      "step": 10179
+    },
+    {
+      "epoch": 0.5602950079806264,
+      "grad_norm": 0.7973861694335938,
+      "learning_rate": 8.209323849033601e-06,
+      "loss": 0.7924,
+      "step": 10180
+    },
+    {
+      "epoch": 0.560350046782982,
+      "grad_norm": 0.6850616931915283,
+      "learning_rate": 8.208991447033867e-06,
+      "loss": 0.7423,
+      "step": 10181
+    },
+    {
+      "epoch": 0.5604050855853376,
+      "grad_norm": 0.8284440636634827,
+      "learning_rate": 8.208659020916213e-06,
+      "loss": 0.7637,
+      "step": 10182
+    },
+    {
+      "epoch": 0.5604601243876933,
+      "grad_norm": 0.7671821713447571,
+      "learning_rate": 8.208326570683136e-06,
+      "loss": 0.7688,
+      "step": 10183
+    },
+    {
+      "epoch": 0.560515163190049,
+      "grad_norm": 0.8359144330024719,
+      "learning_rate": 8.207994096337135e-06,
+      "loss": 0.8179,
+      "step": 10184
+    },
+    {
+      "epoch": 0.5605702019924046,
+      "grad_norm": 0.6389699578285217,
+      "learning_rate": 8.207661597880709e-06,
+      "loss": 0.6987,
+      "step": 10185
+    },
+    {
+      "epoch": 0.5606252407947603,
+      "grad_norm": 0.6472755074501038,
+      "learning_rate": 8.20732907531636e-06,
+      "loss": 0.6984,
+      "step": 10186
+    },
+    {
+      "epoch": 0.560680279597116,
+      "grad_norm": 0.8231903314590454,
+      "learning_rate": 8.20699652864658e-06,
+      "loss": 0.8212,
+      "step": 10187
+    },
+    {
+      "epoch": 0.5607353183994717,
+      "grad_norm": 0.7550386190414429,
+      "learning_rate": 8.206663957873876e-06,
+      "loss": 0.7446,
+      "step": 10188
+    },
+    {
+      "epoch": 0.5607903572018272,
+      "grad_norm": 0.6704659461975098,
+      "learning_rate": 8.206331363000743e-06,
+      "loss": 0.7035,
+      "step": 10189
+    },
+    {
+      "epoch": 0.5608453960041829,
+      "grad_norm": 0.7258654236793518,
+      "learning_rate": 8.20599874402968e-06,
+      "loss": 0.7032,
+      "step": 10190
+    },
+    {
+      "epoch": 0.5609004348065386,
+      "grad_norm": 0.674609363079071,
+      "learning_rate": 8.20566610096319e-06,
+      "loss": 0.7545,
+      "step": 10191
+    },
+    {
+      "epoch": 0.5609554736088943,
+      "grad_norm": 0.6978347301483154,
+      "learning_rate": 8.205333433803773e-06,
+      "loss": 0.8198,
+      "step": 10192
+    },
+    {
+      "epoch": 0.5610105124112499,
+      "grad_norm": 0.6252121329307556,
+      "learning_rate": 8.205000742553925e-06,
+      "loss": 0.6639,
+      "step": 10193
+    },
+    {
+      "epoch": 0.5610655512136056,
+      "grad_norm": 0.7288224101066589,
+      "learning_rate": 8.204668027216152e-06,
+      "loss": 0.8035,
+      "step": 10194
+    },
+    {
+      "epoch": 0.5611205900159613,
+      "grad_norm": 0.6591556072235107,
+      "learning_rate": 8.20433528779295e-06,
+      "loss": 0.7552,
+      "step": 10195
+    },
+    {
+      "epoch": 0.561175628818317,
+      "grad_norm": 0.769827127456665,
+      "learning_rate": 8.204002524286823e-06,
+      "loss": 0.7279,
+      "step": 10196
+    },
+    {
+      "epoch": 0.5612306676206725,
+      "grad_norm": 0.74398273229599,
+      "learning_rate": 8.203669736700271e-06,
+      "loss": 0.7638,
+      "step": 10197
+    },
+    {
+      "epoch": 0.5612857064230282,
+      "grad_norm": 0.9343454241752625,
+      "learning_rate": 8.203336925035795e-06,
+      "loss": 0.7513,
+      "step": 10198
+    },
+    {
+      "epoch": 0.5613407452253839,
+      "grad_norm": 0.6667190194129944,
+      "learning_rate": 8.203004089295894e-06,
+      "loss": 0.77,
+      "step": 10199
+    },
+    {
+      "epoch": 0.5613957840277396,
+      "grad_norm": 0.7684557437896729,
+      "learning_rate": 8.202671229483073e-06,
+      "loss": 0.803,
+      "step": 10200
+    },
+    {
+      "epoch": 0.5614508228300952,
+      "grad_norm": 0.6551374793052673,
+      "learning_rate": 8.202338345599832e-06,
+      "loss": 0.6914,
+      "step": 10201
+    },
+    {
+      "epoch": 0.5615058616324509,
+      "grad_norm": 0.717464029788971,
+      "learning_rate": 8.202005437648674e-06,
+      "loss": 0.6797,
+      "step": 10202
+    },
+    {
+      "epoch": 0.5615609004348066,
+      "grad_norm": 0.7053301334381104,
+      "learning_rate": 8.2016725056321e-06,
+      "loss": 0.7857,
+      "step": 10203
+    },
+    {
+      "epoch": 0.5616159392371622,
+      "grad_norm": 0.8392077684402466,
+      "learning_rate": 8.20133954955261e-06,
+      "loss": 0.8321,
+      "step": 10204
+    },
+    {
+      "epoch": 0.5616709780395178,
+      "grad_norm": 0.6630520820617676,
+      "learning_rate": 8.201006569412711e-06,
+      "loss": 0.7093,
+      "step": 10205
+    },
+    {
+      "epoch": 0.5617260168418735,
+      "grad_norm": 0.6835867762565613,
+      "learning_rate": 8.200673565214905e-06,
+      "loss": 0.6623,
+      "step": 10206
+    },
+    {
+      "epoch": 0.5617810556442292,
+      "grad_norm": 0.7635336518287659,
+      "learning_rate": 8.200340536961691e-06,
+      "loss": 0.8378,
+      "step": 10207
+    },
+    {
+      "epoch": 0.5618360944465849,
+      "grad_norm": 0.6500052213668823,
+      "learning_rate": 8.200007484655575e-06,
+      "loss": 0.6836,
+      "step": 10208
+    },
+    {
+      "epoch": 0.5618911332489405,
+      "grad_norm": 0.6549860835075378,
+      "learning_rate": 8.199674408299058e-06,
+      "loss": 0.6868,
+      "step": 10209
+    },
+    {
+      "epoch": 0.5619461720512962,
+      "grad_norm": 0.7995957732200623,
+      "learning_rate": 8.199341307894647e-06,
+      "loss": 0.7719,
+      "step": 10210
+    },
+    {
+      "epoch": 0.5620012108536518,
+      "grad_norm": 0.6869412064552307,
+      "learning_rate": 8.199008183444843e-06,
+      "loss": 0.7921,
+      "step": 10211
+    },
+    {
+      "epoch": 0.5620562496560075,
+      "grad_norm": 0.9125131964683533,
+      "learning_rate": 8.198675034952149e-06,
+      "loss": 0.9015,
+      "step": 10212
+    },
+    {
+      "epoch": 0.5621112884583631,
+      "grad_norm": 0.6851146221160889,
+      "learning_rate": 8.198341862419068e-06,
+      "loss": 0.7773,
+      "step": 10213
+    },
+    {
+      "epoch": 0.5621663272607188,
+      "grad_norm": 0.6808778047561646,
+      "learning_rate": 8.198008665848108e-06,
+      "loss": 0.7375,
+      "step": 10214
+    },
+    {
+      "epoch": 0.5622213660630745,
+      "grad_norm": 0.6419697999954224,
+      "learning_rate": 8.19767544524177e-06,
+      "loss": 0.7496,
+      "step": 10215
+    },
+    {
+      "epoch": 0.5622764048654302,
+      "grad_norm": 0.7325716614723206,
+      "learning_rate": 8.197342200602559e-06,
+      "loss": 0.7424,
+      "step": 10216
+    },
+    {
+      "epoch": 0.5623314436677858,
+      "grad_norm": 0.6165832281112671,
+      "learning_rate": 8.19700893193298e-06,
+      "loss": 0.6364,
+      "step": 10217
+    },
+    {
+      "epoch": 0.5623864824701414,
+      "grad_norm": 0.7632125020027161,
+      "learning_rate": 8.196675639235539e-06,
+      "loss": 0.7175,
+      "step": 10218
+    },
+    {
+      "epoch": 0.5624415212724971,
+      "grad_norm": 0.6789713501930237,
+      "learning_rate": 8.196342322512738e-06,
+      "loss": 0.7122,
+      "step": 10219
+    },
+    {
+      "epoch": 0.5624965600748528,
+      "grad_norm": 0.7341050505638123,
+      "learning_rate": 8.196008981767084e-06,
+      "loss": 0.7598,
+      "step": 10220
+    },
+    {
+      "epoch": 0.5625515988772084,
+      "grad_norm": 0.7318429350852966,
+      "learning_rate": 8.195675617001083e-06,
+      "loss": 0.7723,
+      "step": 10221
+    },
+    {
+      "epoch": 0.5626066376795641,
+      "grad_norm": 0.6940313577651978,
+      "learning_rate": 8.195342228217238e-06,
+      "loss": 0.7885,
+      "step": 10222
+    },
+    {
+      "epoch": 0.5626616764819198,
+      "grad_norm": 0.8792300820350647,
+      "learning_rate": 8.195008815418058e-06,
+      "loss": 0.7657,
+      "step": 10223
+    },
+    {
+      "epoch": 0.5627167152842754,
+      "grad_norm": 0.7234559655189514,
+      "learning_rate": 8.194675378606044e-06,
+      "loss": 0.7988,
+      "step": 10224
+    },
+    {
+      "epoch": 0.562771754086631,
+      "grad_norm": 0.6698254942893982,
+      "learning_rate": 8.194341917783708e-06,
+      "loss": 0.6378,
+      "step": 10225
+    },
+    {
+      "epoch": 0.5628267928889867,
+      "grad_norm": 0.6546483635902405,
+      "learning_rate": 8.194008432953552e-06,
+      "loss": 0.7113,
+      "step": 10226
+    },
+    {
+      "epoch": 0.5628818316913424,
+      "grad_norm": 0.6532583832740784,
+      "learning_rate": 8.193674924118085e-06,
+      "loss": 0.6782,
+      "step": 10227
+    },
+    {
+      "epoch": 0.562936870493698,
+      "grad_norm": 0.770578920841217,
+      "learning_rate": 8.19334139127981e-06,
+      "loss": 0.8519,
+      "step": 10228
+    },
+    {
+      "epoch": 0.5629919092960537,
+      "grad_norm": 0.7255409359931946,
+      "learning_rate": 8.193007834441235e-06,
+      "loss": 0.6555,
+      "step": 10229
+    },
+    {
+      "epoch": 0.5630469480984094,
+      "grad_norm": 0.6659883856773376,
+      "learning_rate": 8.19267425360487e-06,
+      "loss": 0.7836,
+      "step": 10230
+    },
+    {
+      "epoch": 0.5631019869007651,
+      "grad_norm": 0.6596028208732605,
+      "learning_rate": 8.192340648773221e-06,
+      "loss": 0.6199,
+      "step": 10231
+    },
+    {
+      "epoch": 0.5631570257031207,
+      "grad_norm": 0.8226001858711243,
+      "learning_rate": 8.192007019948793e-06,
+      "loss": 0.8101,
+      "step": 10232
+    },
+    {
+      "epoch": 0.5632120645054763,
+      "grad_norm": 0.7465038895606995,
+      "learning_rate": 8.191673367134094e-06,
+      "loss": 0.8437,
+      "step": 10233
+    },
+    {
+      "epoch": 0.563267103307832,
+      "grad_norm": 1.0008004903793335,
+      "learning_rate": 8.191339690331632e-06,
+      "loss": 0.8626,
+      "step": 10234
+    },
+    {
+      "epoch": 0.5633221421101877,
+      "grad_norm": 0.7538222670555115,
+      "learning_rate": 8.191005989543917e-06,
+      "loss": 0.7222,
+      "step": 10235
+    },
+    {
+      "epoch": 0.5633771809125433,
+      "grad_norm": 0.6252872943878174,
+      "learning_rate": 8.190672264773454e-06,
+      "loss": 0.8038,
+      "step": 10236
+    },
+    {
+      "epoch": 0.563432219714899,
+      "grad_norm": 0.7083514928817749,
+      "learning_rate": 8.190338516022752e-06,
+      "loss": 0.7863,
+      "step": 10237
+    },
+    {
+      "epoch": 0.5634872585172547,
+      "grad_norm": 0.6887454390525818,
+      "learning_rate": 8.19000474329432e-06,
+      "loss": 0.7034,
+      "step": 10238
+    },
+    {
+      "epoch": 0.5635422973196104,
+      "grad_norm": 0.7487072348594666,
+      "learning_rate": 8.189670946590666e-06,
+      "loss": 0.8618,
+      "step": 10239
+    },
+    {
+      "epoch": 0.5635973361219659,
+      "grad_norm": 0.6999371647834778,
+      "learning_rate": 8.189337125914298e-06,
+      "loss": 0.7613,
+      "step": 10240
+    },
+    {
+      "epoch": 0.5636523749243216,
+      "grad_norm": 0.8265380263328552,
+      "learning_rate": 8.18900328126773e-06,
+      "loss": 0.7576,
+      "step": 10241
+    },
+    {
+      "epoch": 0.5637074137266773,
+      "grad_norm": 0.6688962578773499,
+      "learning_rate": 8.188669412653463e-06,
+      "loss": 0.712,
+      "step": 10242
+    },
+    {
+      "epoch": 0.563762452529033,
+      "grad_norm": 0.6343923211097717,
+      "learning_rate": 8.188335520074011e-06,
+      "loss": 0.7239,
+      "step": 10243
+    },
+    {
+      "epoch": 0.5638174913313886,
+      "grad_norm": 0.7122388482093811,
+      "learning_rate": 8.188001603531883e-06,
+      "loss": 0.7892,
+      "step": 10244
+    },
+    {
+      "epoch": 0.5638725301337443,
+      "grad_norm": 0.6646286845207214,
+      "learning_rate": 8.187667663029587e-06,
+      "loss": 0.7805,
+      "step": 10245
+    },
+    {
+      "epoch": 0.5639275689361,
+      "grad_norm": 0.742938220500946,
+      "learning_rate": 8.187333698569638e-06,
+      "loss": 0.8444,
+      "step": 10246
+    },
+    {
+      "epoch": 0.5639826077384557,
+      "grad_norm": 0.7260885238647461,
+      "learning_rate": 8.18699971015454e-06,
+      "loss": 0.8621,
+      "step": 10247
+    },
+    {
+      "epoch": 0.5640376465408112,
+      "grad_norm": 0.7920067310333252,
+      "learning_rate": 8.186665697786804e-06,
+      "loss": 0.7391,
+      "step": 10248
+    },
+    {
+      "epoch": 0.5640926853431669,
+      "grad_norm": 0.7472825646400452,
+      "learning_rate": 8.186331661468943e-06,
+      "loss": 0.7249,
+      "step": 10249
+    },
+    {
+      "epoch": 0.5641477241455226,
+      "grad_norm": 0.692643940448761,
+      "learning_rate": 8.185997601203465e-06,
+      "loss": 0.7884,
+      "step": 10250
+    },
+    {
+      "epoch": 0.5642027629478783,
+      "grad_norm": 0.715455174446106,
+      "learning_rate": 8.185663516992884e-06,
+      "loss": 0.7369,
+      "step": 10251
+    },
+    {
+      "epoch": 0.5642578017502339,
+      "grad_norm": 0.7566105723381042,
+      "learning_rate": 8.185329408839705e-06,
+      "loss": 0.7378,
+      "step": 10252
+    },
+    {
+      "epoch": 0.5643128405525896,
+      "grad_norm": 0.8163520693778992,
+      "learning_rate": 8.184995276746445e-06,
+      "loss": 0.7326,
+      "step": 10253
+    },
+    {
+      "epoch": 0.5643678793549453,
+      "grad_norm": 0.6280468106269836,
+      "learning_rate": 8.184661120715615e-06,
+      "loss": 0.6858,
+      "step": 10254
+    },
+    {
+      "epoch": 0.564422918157301,
+      "grad_norm": 0.7246795892715454,
+      "learning_rate": 8.184326940749723e-06,
+      "loss": 0.8111,
+      "step": 10255
+    },
+    {
+      "epoch": 0.5644779569596565,
+      "grad_norm": 0.7429527640342712,
+      "learning_rate": 8.18399273685128e-06,
+      "loss": 0.7642,
+      "step": 10256
+    },
+    {
+      "epoch": 0.5645329957620122,
+      "grad_norm": 0.7308861017227173,
+      "learning_rate": 8.183658509022802e-06,
+      "loss": 0.7844,
+      "step": 10257
+    },
+    {
+      "epoch": 0.5645880345643679,
+      "grad_norm": 0.7549033164978027,
+      "learning_rate": 8.1833242572668e-06,
+      "loss": 0.8585,
+      "step": 10258
+    },
+    {
+      "epoch": 0.5646430733667236,
+      "grad_norm": 0.6779888868331909,
+      "learning_rate": 8.182989981585782e-06,
+      "loss": 0.6808,
+      "step": 10259
+    },
+    {
+      "epoch": 0.5646981121690792,
+      "grad_norm": 0.887113630771637,
+      "learning_rate": 8.182655681982266e-06,
+      "loss": 0.8229,
+      "step": 10260
+    },
+    {
+      "epoch": 0.5647531509714349,
+      "grad_norm": 0.6405711770057678,
+      "learning_rate": 8.18232135845876e-06,
+      "loss": 0.6901,
+      "step": 10261
+    },
+    {
+      "epoch": 0.5648081897737905,
+      "grad_norm": 0.7302486300468445,
+      "learning_rate": 8.18198701101778e-06,
+      "loss": 0.6853,
+      "step": 10262
+    },
+    {
+      "epoch": 0.5648632285761462,
+      "grad_norm": 0.6374662518501282,
+      "learning_rate": 8.181652639661837e-06,
+      "loss": 0.7177,
+      "step": 10263
+    },
+    {
+      "epoch": 0.5649182673785018,
+      "grad_norm": 0.9267570972442627,
+      "learning_rate": 8.181318244393444e-06,
+      "loss": 0.7926,
+      "step": 10264
+    },
+    {
+      "epoch": 0.5649733061808575,
+      "grad_norm": 0.8196623921394348,
+      "learning_rate": 8.180983825215114e-06,
+      "loss": 0.7127,
+      "step": 10265
+    },
+    {
+      "epoch": 0.5650283449832132,
+      "grad_norm": 0.7004575133323669,
+      "learning_rate": 8.180649382129361e-06,
+      "loss": 0.7858,
+      "step": 10266
+    },
+    {
+      "epoch": 0.5650833837855688,
+      "grad_norm": 0.7667824625968933,
+      "learning_rate": 8.180314915138701e-06,
+      "loss": 0.7742,
+      "step": 10267
+    },
+    {
+      "epoch": 0.5651384225879245,
+      "grad_norm": 0.7372623682022095,
+      "learning_rate": 8.179980424245644e-06,
+      "loss": 0.7949,
+      "step": 10268
+    },
+    {
+      "epoch": 0.5651934613902801,
+      "grad_norm": 0.6417940258979797,
+      "learning_rate": 8.179645909452704e-06,
+      "loss": 0.6683,
+      "step": 10269
+    },
+    {
+      "epoch": 0.5652485001926358,
+      "grad_norm": 0.6736140251159668,
+      "learning_rate": 8.179311370762398e-06,
+      "loss": 0.6564,
+      "step": 10270
+    },
+    {
+      "epoch": 0.5653035389949914,
+      "grad_norm": 0.6727200746536255,
+      "learning_rate": 8.178976808177239e-06,
+      "loss": 0.8065,
+      "step": 10271
+    },
+    {
+      "epoch": 0.5653585777973471,
+      "grad_norm": 0.7565415501594543,
+      "learning_rate": 8.17864222169974e-06,
+      "loss": 0.9055,
+      "step": 10272
+    },
+    {
+      "epoch": 0.5654136165997028,
+      "grad_norm": 0.8938627243041992,
+      "learning_rate": 8.178307611332418e-06,
+      "loss": 0.8009,
+      "step": 10273
+    },
+    {
+      "epoch": 0.5654686554020585,
+      "grad_norm": 0.7439131140708923,
+      "learning_rate": 8.177972977077786e-06,
+      "loss": 0.7807,
+      "step": 10274
+    },
+    {
+      "epoch": 0.5655236942044141,
+      "grad_norm": 0.7603998184204102,
+      "learning_rate": 8.17763831893836e-06,
+      "loss": 0.818,
+      "step": 10275
+    },
+    {
+      "epoch": 0.5655787330067698,
+      "grad_norm": 0.7088946104049683,
+      "learning_rate": 8.177303636916655e-06,
+      "loss": 0.7741,
+      "step": 10276
+    },
+    {
+      "epoch": 0.5656337718091254,
+      "grad_norm": 0.6801518201828003,
+      "learning_rate": 8.176968931015187e-06,
+      "loss": 0.7633,
+      "step": 10277
+    },
+    {
+      "epoch": 0.5656888106114811,
+      "grad_norm": 0.6739299297332764,
+      "learning_rate": 8.17663420123647e-06,
+      "loss": 0.7772,
+      "step": 10278
+    },
+    {
+      "epoch": 0.5657438494138367,
+      "grad_norm": 0.7432494759559631,
+      "learning_rate": 8.176299447583021e-06,
+      "loss": 0.7368,
+      "step": 10279
+    },
+    {
+      "epoch": 0.5657988882161924,
+      "grad_norm": 0.7847158908843994,
+      "learning_rate": 8.175964670057357e-06,
+      "loss": 0.7824,
+      "step": 10280
+    },
+    {
+      "epoch": 0.5658539270185481,
+      "grad_norm": 0.8732449412345886,
+      "learning_rate": 8.17562986866199e-06,
+      "loss": 0.8035,
+      "step": 10281
+    },
+    {
+      "epoch": 0.5659089658209038,
+      "grad_norm": 0.7988447546958923,
+      "learning_rate": 8.17529504339944e-06,
+      "loss": 0.828,
+      "step": 10282
+    },
+    {
+      "epoch": 0.5659640046232594,
+      "grad_norm": 0.7063263058662415,
+      "learning_rate": 8.174960194272224e-06,
+      "loss": 0.7723,
+      "step": 10283
+    },
+    {
+      "epoch": 0.566019043425615,
+      "grad_norm": 0.7635022401809692,
+      "learning_rate": 8.174625321282856e-06,
+      "loss": 0.7156,
+      "step": 10284
+    },
+    {
+      "epoch": 0.5660740822279707,
+      "grad_norm": 0.6505927443504333,
+      "learning_rate": 8.174290424433853e-06,
+      "loss": 0.7409,
+      "step": 10285
+    },
+    {
+      "epoch": 0.5661291210303264,
+      "grad_norm": 0.6919816136360168,
+      "learning_rate": 8.173955503727734e-06,
+      "loss": 0.7829,
+      "step": 10286
+    },
+    {
+      "epoch": 0.566184159832682,
+      "grad_norm": 0.7024216651916504,
+      "learning_rate": 8.173620559167015e-06,
+      "loss": 0.7378,
+      "step": 10287
+    },
+    {
+      "epoch": 0.5662391986350377,
+      "grad_norm": 0.7134365439414978,
+      "learning_rate": 8.173285590754212e-06,
+      "loss": 0.7737,
+      "step": 10288
+    },
+    {
+      "epoch": 0.5662942374373934,
+      "grad_norm": 0.6867973804473877,
+      "learning_rate": 8.172950598491845e-06,
+      "loss": 0.7169,
+      "step": 10289
+    },
+    {
+      "epoch": 0.5663492762397491,
+      "grad_norm": 0.6900742650032043,
+      "learning_rate": 8.172615582382432e-06,
+      "loss": 0.7888,
+      "step": 10290
+    },
+    {
+      "epoch": 0.5664043150421046,
+      "grad_norm": 0.7026718854904175,
+      "learning_rate": 8.172280542428488e-06,
+      "loss": 0.8179,
+      "step": 10291
+    },
+    {
+      "epoch": 0.5664593538444603,
+      "grad_norm": 0.6940855979919434,
+      "learning_rate": 8.171945478632533e-06,
+      "loss": 0.7686,
+      "step": 10292
+    },
+    {
+      "epoch": 0.566514392646816,
+      "grad_norm": 0.6717686653137207,
+      "learning_rate": 8.171610390997085e-06,
+      "loss": 0.7865,
+      "step": 10293
+    },
+    {
+      "epoch": 0.5665694314491717,
+      "grad_norm": 0.6947711110115051,
+      "learning_rate": 8.171275279524661e-06,
+      "loss": 0.7811,
+      "step": 10294
+    },
+    {
+      "epoch": 0.5666244702515273,
+      "grad_norm": 0.6907814741134644,
+      "learning_rate": 8.170940144217782e-06,
+      "loss": 0.7095,
+      "step": 10295
+    },
+    {
+      "epoch": 0.566679509053883,
+      "grad_norm": 0.723952054977417,
+      "learning_rate": 8.170604985078965e-06,
+      "loss": 0.7814,
+      "step": 10296
+    },
+    {
+      "epoch": 0.5667345478562387,
+      "grad_norm": 0.7775490880012512,
+      "learning_rate": 8.17026980211073e-06,
+      "loss": 0.797,
+      "step": 10297
+    },
+    {
+      "epoch": 0.5667895866585944,
+      "grad_norm": 0.7557885646820068,
+      "learning_rate": 8.169934595315597e-06,
+      "loss": 0.8423,
+      "step": 10298
+    },
+    {
+      "epoch": 0.5668446254609499,
+      "grad_norm": 0.7838338017463684,
+      "learning_rate": 8.169599364696083e-06,
+      "loss": 0.7114,
+      "step": 10299
+    },
+    {
+      "epoch": 0.5668996642633056,
+      "grad_norm": 0.6632605791091919,
+      "learning_rate": 8.169264110254707e-06,
+      "loss": 0.6723,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5669547030656613,
+      "grad_norm": 0.735756516456604,
+      "learning_rate": 8.168928831993991e-06,
+      "loss": 0.7533,
+      "step": 10301
+    },
+    {
+      "epoch": 0.567009741868017,
+      "grad_norm": 0.6981016993522644,
+      "learning_rate": 8.168593529916457e-06,
+      "loss": 0.7882,
+      "step": 10302
+    },
+    {
+      "epoch": 0.5670647806703726,
+      "grad_norm": 0.6413942575454712,
+      "learning_rate": 8.168258204024619e-06,
+      "loss": 0.6593,
+      "step": 10303
+    },
+    {
+      "epoch": 0.5671198194727283,
+      "grad_norm": 0.7040891051292419,
+      "learning_rate": 8.167922854321002e-06,
+      "loss": 0.7295,
+      "step": 10304
+    },
+    {
+      "epoch": 0.567174858275084,
+      "grad_norm": 0.7132521867752075,
+      "learning_rate": 8.167587480808126e-06,
+      "loss": 0.7128,
+      "step": 10305
+    },
+    {
+      "epoch": 0.5672298970774396,
+      "grad_norm": 0.756529688835144,
+      "learning_rate": 8.167252083488508e-06,
+      "loss": 0.7044,
+      "step": 10306
+    },
+    {
+      "epoch": 0.5672849358797952,
+      "grad_norm": 0.8456888198852539,
+      "learning_rate": 8.166916662364672e-06,
+      "loss": 0.8304,
+      "step": 10307
+    },
+    {
+      "epoch": 0.5673399746821509,
+      "grad_norm": 0.7758522629737854,
+      "learning_rate": 8.166581217439138e-06,
+      "loss": 0.7192,
+      "step": 10308
+    },
+    {
+      "epoch": 0.5673950134845066,
+      "grad_norm": 0.8110343217849731,
+      "learning_rate": 8.166245748714428e-06,
+      "loss": 0.8794,
+      "step": 10309
+    },
+    {
+      "epoch": 0.5674500522868622,
+      "grad_norm": 0.6803586483001709,
+      "learning_rate": 8.165910256193062e-06,
+      "loss": 0.7402,
+      "step": 10310
+    },
+    {
+      "epoch": 0.5675050910892179,
+      "grad_norm": 0.7294176816940308,
+      "learning_rate": 8.165574739877563e-06,
+      "loss": 0.7325,
+      "step": 10311
+    },
+    {
+      "epoch": 0.5675601298915736,
+      "grad_norm": 0.835488498210907,
+      "learning_rate": 8.165239199770448e-06,
+      "loss": 0.8317,
+      "step": 10312
+    },
+    {
+      "epoch": 0.5676151686939293,
+      "grad_norm": 0.6497608423233032,
+      "learning_rate": 8.164903635874246e-06,
+      "loss": 0.6902,
+      "step": 10313
+    },
+    {
+      "epoch": 0.5676702074962848,
+      "grad_norm": 0.6782082915306091,
+      "learning_rate": 8.164568048191474e-06,
+      "loss": 0.7941,
+      "step": 10314
+    },
+    {
+      "epoch": 0.5677252462986405,
+      "grad_norm": 0.6974388957023621,
+      "learning_rate": 8.164232436724656e-06,
+      "loss": 0.7899,
+      "step": 10315
+    },
+    {
+      "epoch": 0.5677802851009962,
+      "grad_norm": 0.7222558259963989,
+      "learning_rate": 8.163896801476314e-06,
+      "loss": 0.8034,
+      "step": 10316
+    },
+    {
+      "epoch": 0.5678353239033519,
+      "grad_norm": 0.6562586426734924,
+      "learning_rate": 8.16356114244897e-06,
+      "loss": 0.7864,
+      "step": 10317
+    },
+    {
+      "epoch": 0.5678903627057075,
+      "grad_norm": 0.6888270378112793,
+      "learning_rate": 8.16322545964515e-06,
+      "loss": 0.8455,
+      "step": 10318
+    },
+    {
+      "epoch": 0.5679454015080632,
+      "grad_norm": 0.642084002494812,
+      "learning_rate": 8.162889753067372e-06,
+      "loss": 0.7478,
+      "step": 10319
+    },
+    {
+      "epoch": 0.5680004403104189,
+      "grad_norm": 0.7077270746231079,
+      "learning_rate": 8.16255402271816e-06,
+      "loss": 0.7281,
+      "step": 10320
+    },
+    {
+      "epoch": 0.5680554791127745,
+      "grad_norm": 0.7202198505401611,
+      "learning_rate": 8.16221826860004e-06,
+      "loss": 0.7893,
+      "step": 10321
+    },
+    {
+      "epoch": 0.5681105179151301,
+      "grad_norm": 0.8950369954109192,
+      "learning_rate": 8.161882490715534e-06,
+      "loss": 0.772,
+      "step": 10322
+    },
+    {
+      "epoch": 0.5681655567174858,
+      "grad_norm": 0.6986666917800903,
+      "learning_rate": 8.161546689067166e-06,
+      "loss": 0.7712,
+      "step": 10323
+    },
+    {
+      "epoch": 0.5682205955198415,
+      "grad_norm": 0.7095959782600403,
+      "learning_rate": 8.161210863657458e-06,
+      "loss": 0.8373,
+      "step": 10324
+    },
+    {
+      "epoch": 0.5682756343221972,
+      "grad_norm": 0.7510485649108887,
+      "learning_rate": 8.160875014488936e-06,
+      "loss": 0.9106,
+      "step": 10325
+    },
+    {
+      "epoch": 0.5683306731245528,
+      "grad_norm": 0.7558283805847168,
+      "learning_rate": 8.160539141564123e-06,
+      "loss": 0.8192,
+      "step": 10326
+    },
+    {
+      "epoch": 0.5683857119269085,
+      "grad_norm": 0.7523400187492371,
+      "learning_rate": 8.160203244885545e-06,
+      "loss": 0.8276,
+      "step": 10327
+    },
+    {
+      "epoch": 0.5684407507292641,
+      "grad_norm": 0.6911195516586304,
+      "learning_rate": 8.159867324455724e-06,
+      "loss": 0.6286,
+      "step": 10328
+    },
+    {
+      "epoch": 0.5684957895316198,
+      "grad_norm": 0.6456325054168701,
+      "learning_rate": 8.159531380277188e-06,
+      "loss": 0.7419,
+      "step": 10329
+    },
+    {
+      "epoch": 0.5685508283339754,
+      "grad_norm": 0.9318492412567139,
+      "learning_rate": 8.159195412352458e-06,
+      "loss": 0.8131,
+      "step": 10330
+    },
+    {
+      "epoch": 0.5686058671363311,
+      "grad_norm": 0.7012938857078552,
+      "learning_rate": 8.158859420684062e-06,
+      "loss": 0.7074,
+      "step": 10331
+    },
+    {
+      "epoch": 0.5686609059386868,
+      "grad_norm": 0.7152053117752075,
+      "learning_rate": 8.158523405274523e-06,
+      "loss": 0.7186,
+      "step": 10332
+    },
+    {
+      "epoch": 0.5687159447410425,
+      "grad_norm": 0.7074982523918152,
+      "learning_rate": 8.158187366126368e-06,
+      "loss": 0.8021,
+      "step": 10333
+    },
+    {
+      "epoch": 0.5687709835433981,
+      "grad_norm": 0.689536452293396,
+      "learning_rate": 8.157851303242123e-06,
+      "loss": 0.7493,
+      "step": 10334
+    },
+    {
+      "epoch": 0.5688260223457537,
+      "grad_norm": 0.7411753535270691,
+      "learning_rate": 8.157515216624313e-06,
+      "loss": 0.8012,
+      "step": 10335
+    },
+    {
+      "epoch": 0.5688810611481094,
+      "grad_norm": 0.6831420063972473,
+      "learning_rate": 8.157179106275463e-06,
+      "loss": 0.7114,
+      "step": 10336
+    },
+    {
+      "epoch": 0.5689360999504651,
+      "grad_norm": 0.6786901950836182,
+      "learning_rate": 8.1568429721981e-06,
+      "loss": 0.7638,
+      "step": 10337
+    },
+    {
+      "epoch": 0.5689911387528207,
+      "grad_norm": 0.7546970844268799,
+      "learning_rate": 8.15650681439475e-06,
+      "loss": 0.7711,
+      "step": 10338
+    },
+    {
+      "epoch": 0.5690461775551764,
+      "grad_norm": 0.8071785569190979,
+      "learning_rate": 8.156170632867942e-06,
+      "loss": 0.8105,
+      "step": 10339
+    },
+    {
+      "epoch": 0.5691012163575321,
+      "grad_norm": 0.7872087359428406,
+      "learning_rate": 8.155834427620198e-06,
+      "loss": 0.7657,
+      "step": 10340
+    },
+    {
+      "epoch": 0.5691562551598878,
+      "grad_norm": 0.724328875541687,
+      "learning_rate": 8.155498198654047e-06,
+      "loss": 0.7978,
+      "step": 10341
+    },
+    {
+      "epoch": 0.5692112939622433,
+      "grad_norm": 0.8559905886650085,
+      "learning_rate": 8.155161945972016e-06,
+      "loss": 0.7766,
+      "step": 10342
+    },
+    {
+      "epoch": 0.569266332764599,
+      "grad_norm": 0.607418417930603,
+      "learning_rate": 8.154825669576635e-06,
+      "loss": 0.642,
+      "step": 10343
+    },
+    {
+      "epoch": 0.5693213715669547,
+      "grad_norm": 0.7403624653816223,
+      "learning_rate": 8.154489369470426e-06,
+      "loss": 0.7301,
+      "step": 10344
+    },
+    {
+      "epoch": 0.5693764103693104,
+      "grad_norm": 0.7388540506362915,
+      "learning_rate": 8.154153045655922e-06,
+      "loss": 0.7895,
+      "step": 10345
+    },
+    {
+      "epoch": 0.569431449171666,
+      "grad_norm": 0.8327579498291016,
+      "learning_rate": 8.153816698135646e-06,
+      "loss": 0.7589,
+      "step": 10346
+    },
+    {
+      "epoch": 0.5694864879740217,
+      "grad_norm": 0.7738710641860962,
+      "learning_rate": 8.153480326912128e-06,
+      "loss": 0.7828,
+      "step": 10347
+    },
+    {
+      "epoch": 0.5695415267763774,
+      "grad_norm": 0.8280724287033081,
+      "learning_rate": 8.153143931987896e-06,
+      "loss": 0.8194,
+      "step": 10348
+    },
+    {
+      "epoch": 0.5695965655787331,
+      "grad_norm": 0.8290724754333496,
+      "learning_rate": 8.152807513365478e-06,
+      "loss": 0.5941,
+      "step": 10349
+    },
+    {
+      "epoch": 0.5696516043810886,
+      "grad_norm": 0.7514322400093079,
+      "learning_rate": 8.152471071047403e-06,
+      "loss": 0.676,
+      "step": 10350
+    },
+    {
+      "epoch": 0.5697066431834443,
+      "grad_norm": 0.6990258693695068,
+      "learning_rate": 8.1521346050362e-06,
+      "loss": 0.804,
+      "step": 10351
+    },
+    {
+      "epoch": 0.5697616819858,
+      "grad_norm": 0.6781288981437683,
+      "learning_rate": 8.151798115334396e-06,
+      "loss": 0.7372,
+      "step": 10352
+    },
+    {
+      "epoch": 0.5698167207881556,
+      "grad_norm": 0.764301061630249,
+      "learning_rate": 8.151461601944523e-06,
+      "loss": 0.8242,
+      "step": 10353
+    },
+    {
+      "epoch": 0.5698717595905113,
+      "grad_norm": 0.7577376961708069,
+      "learning_rate": 8.151125064869106e-06,
+      "loss": 0.7354,
+      "step": 10354
+    },
+    {
+      "epoch": 0.569926798392867,
+      "grad_norm": 0.767764687538147,
+      "learning_rate": 8.150788504110678e-06,
+      "loss": 0.7262,
+      "step": 10355
+    },
+    {
+      "epoch": 0.5699818371952227,
+      "grad_norm": 0.6634765267372131,
+      "learning_rate": 8.150451919671767e-06,
+      "loss": 0.7527,
+      "step": 10356
+    },
+    {
+      "epoch": 0.5700368759975782,
+      "grad_norm": 0.8803308010101318,
+      "learning_rate": 8.150115311554901e-06,
+      "loss": 0.8172,
+      "step": 10357
+    },
+    {
+      "epoch": 0.5700919147999339,
+      "grad_norm": 0.695791482925415,
+      "learning_rate": 8.149778679762611e-06,
+      "loss": 0.7538,
+      "step": 10358
+    },
+    {
+      "epoch": 0.5701469536022896,
+      "grad_norm": 0.7047555446624756,
+      "learning_rate": 8.149442024297432e-06,
+      "loss": 0.7533,
+      "step": 10359
+    },
+    {
+      "epoch": 0.5702019924046453,
+      "grad_norm": 0.7148274183273315,
+      "learning_rate": 8.149105345161886e-06,
+      "loss": 0.6736,
+      "step": 10360
+    },
+    {
+      "epoch": 0.5702570312070009,
+      "grad_norm": 0.673204243183136,
+      "learning_rate": 8.148768642358508e-06,
+      "loss": 0.7713,
+      "step": 10361
+    },
+    {
+      "epoch": 0.5703120700093566,
+      "grad_norm": 0.6258989572525024,
+      "learning_rate": 8.148431915889827e-06,
+      "loss": 0.6578,
+      "step": 10362
+    },
+    {
+      "epoch": 0.5703671088117123,
+      "grad_norm": 0.8411956429481506,
+      "learning_rate": 8.148095165758377e-06,
+      "loss": 0.8387,
+      "step": 10363
+    },
+    {
+      "epoch": 0.570422147614068,
+      "grad_norm": 0.7802130579948425,
+      "learning_rate": 8.147758391966685e-06,
+      "loss": 0.8564,
+      "step": 10364
+    },
+    {
+      "epoch": 0.5704771864164235,
+      "grad_norm": 0.6665176153182983,
+      "learning_rate": 8.147421594517282e-06,
+      "loss": 0.688,
+      "step": 10365
+    },
+    {
+      "epoch": 0.5705322252187792,
+      "grad_norm": 0.7166683673858643,
+      "learning_rate": 8.147084773412702e-06,
+      "loss": 0.6704,
+      "step": 10366
+    },
+    {
+      "epoch": 0.5705872640211349,
+      "grad_norm": 0.6948957443237305,
+      "learning_rate": 8.146747928655476e-06,
+      "loss": 0.7116,
+      "step": 10367
+    },
+    {
+      "epoch": 0.5706423028234906,
+      "grad_norm": 0.588965892791748,
+      "learning_rate": 8.146411060248134e-06,
+      "loss": 0.5644,
+      "step": 10368
+    },
+    {
+      "epoch": 0.5706973416258462,
+      "grad_norm": 0.8020890355110168,
+      "learning_rate": 8.14607416819321e-06,
+      "loss": 0.6978,
+      "step": 10369
+    },
+    {
+      "epoch": 0.5707523804282019,
+      "grad_norm": 0.9900732040405273,
+      "learning_rate": 8.145737252493234e-06,
+      "loss": 0.7295,
+      "step": 10370
+    },
+    {
+      "epoch": 0.5708074192305576,
+      "grad_norm": 0.7236563563346863,
+      "learning_rate": 8.145400313150737e-06,
+      "loss": 0.7555,
+      "step": 10371
+    },
+    {
+      "epoch": 0.5708624580329132,
+      "grad_norm": 0.6784152984619141,
+      "learning_rate": 8.145063350168257e-06,
+      "loss": 0.7283,
+      "step": 10372
+    },
+    {
+      "epoch": 0.5709174968352688,
+      "grad_norm": 0.6255244612693787,
+      "learning_rate": 8.14472636354832e-06,
+      "loss": 0.6722,
+      "step": 10373
+    },
+    {
+      "epoch": 0.5709725356376245,
+      "grad_norm": 0.8250948786735535,
+      "learning_rate": 8.14438935329346e-06,
+      "loss": 0.8406,
+      "step": 10374
+    },
+    {
+      "epoch": 0.5710275744399802,
+      "grad_norm": 0.7308233380317688,
+      "learning_rate": 8.144052319406215e-06,
+      "loss": 0.8084,
+      "step": 10375
+    },
+    {
+      "epoch": 0.5710826132423359,
+      "grad_norm": 0.7850058674812317,
+      "learning_rate": 8.143715261889112e-06,
+      "loss": 0.7892,
+      "step": 10376
+    },
+    {
+      "epoch": 0.5711376520446915,
+      "grad_norm": 0.81241774559021,
+      "learning_rate": 8.143378180744687e-06,
+      "loss": 0.7819,
+      "step": 10377
+    },
+    {
+      "epoch": 0.5711926908470472,
+      "grad_norm": 0.7174570560455322,
+      "learning_rate": 8.143041075975473e-06,
+      "loss": 0.7104,
+      "step": 10378
+    },
+    {
+      "epoch": 0.5712477296494028,
+      "grad_norm": 0.6954129934310913,
+      "learning_rate": 8.142703947584004e-06,
+      "loss": 0.7821,
+      "step": 10379
+    },
+    {
+      "epoch": 0.5713027684517585,
+      "grad_norm": 0.6895242929458618,
+      "learning_rate": 8.142366795572813e-06,
+      "loss": 0.7687,
+      "step": 10380
+    },
+    {
+      "epoch": 0.5713578072541141,
+      "grad_norm": 0.6543757319450378,
+      "learning_rate": 8.142029619944434e-06,
+      "loss": 0.7042,
+      "step": 10381
+    },
+    {
+      "epoch": 0.5714128460564698,
+      "grad_norm": 0.6712427139282227,
+      "learning_rate": 8.141692420701404e-06,
+      "loss": 0.6861,
+      "step": 10382
+    },
+    {
+      "epoch": 0.5714678848588255,
+      "grad_norm": 1.6716055870056152,
+      "learning_rate": 8.141355197846253e-06,
+      "loss": 0.8209,
+      "step": 10383
+    },
+    {
+      "epoch": 0.5715229236611812,
+      "grad_norm": 0.7509854435920715,
+      "learning_rate": 8.141017951381516e-06,
+      "loss": 0.8246,
+      "step": 10384
+    },
+    {
+      "epoch": 0.5715779624635368,
+      "grad_norm": 0.7161786556243896,
+      "learning_rate": 8.14068068130973e-06,
+      "loss": 0.835,
+      "step": 10385
+    },
+    {
+      "epoch": 0.5716330012658924,
+      "grad_norm": 0.7423714995384216,
+      "learning_rate": 8.140343387633427e-06,
+      "loss": 0.8004,
+      "step": 10386
+    },
+    {
+      "epoch": 0.5716880400682481,
+      "grad_norm": 0.6955768465995789,
+      "learning_rate": 8.140006070355146e-06,
+      "loss": 0.7299,
+      "step": 10387
+    },
+    {
+      "epoch": 0.5717430788706038,
+      "grad_norm": 0.6742254495620728,
+      "learning_rate": 8.13966872947742e-06,
+      "loss": 0.6549,
+      "step": 10388
+    },
+    {
+      "epoch": 0.5717981176729594,
+      "grad_norm": 0.7332299947738647,
+      "learning_rate": 8.139331365002782e-06,
+      "loss": 0.7945,
+      "step": 10389
+    },
+    {
+      "epoch": 0.5718531564753151,
+      "grad_norm": 0.6552133560180664,
+      "learning_rate": 8.138993976933771e-06,
+      "loss": 0.7193,
+      "step": 10390
+    },
+    {
+      "epoch": 0.5719081952776708,
+      "grad_norm": 0.6708530187606812,
+      "learning_rate": 8.138656565272923e-06,
+      "loss": 0.8053,
+      "step": 10391
+    },
+    {
+      "epoch": 0.5719632340800265,
+      "grad_norm": 0.7837093472480774,
+      "learning_rate": 8.138319130022771e-06,
+      "loss": 0.7752,
+      "step": 10392
+    },
+    {
+      "epoch": 0.572018272882382,
+      "grad_norm": 0.6910337805747986,
+      "learning_rate": 8.137981671185853e-06,
+      "loss": 0.7573,
+      "step": 10393
+    },
+    {
+      "epoch": 0.5720733116847377,
+      "grad_norm": 0.6758334636688232,
+      "learning_rate": 8.137644188764704e-06,
+      "loss": 0.8251,
+      "step": 10394
+    },
+    {
+      "epoch": 0.5721283504870934,
+      "grad_norm": 0.7513287663459778,
+      "learning_rate": 8.137306682761862e-06,
+      "loss": 0.6491,
+      "step": 10395
+    },
+    {
+      "epoch": 0.572183389289449,
+      "grad_norm": 0.678210973739624,
+      "learning_rate": 8.136969153179863e-06,
+      "loss": 0.7761,
+      "step": 10396
+    },
+    {
+      "epoch": 0.5722384280918047,
+      "grad_norm": 0.8256083726882935,
+      "learning_rate": 8.13663160002124e-06,
+      "loss": 0.7813,
+      "step": 10397
+    },
+    {
+      "epoch": 0.5722934668941604,
+      "grad_norm": 0.8383314609527588,
+      "learning_rate": 8.136294023288538e-06,
+      "loss": 0.7669,
+      "step": 10398
+    },
+    {
+      "epoch": 0.5723485056965161,
+      "grad_norm": 0.7150036692619324,
+      "learning_rate": 8.135956422984287e-06,
+      "loss": 0.8322,
+      "step": 10399
+    },
+    {
+      "epoch": 0.5724035444988717,
+      "grad_norm": 1.3011385202407837,
+      "learning_rate": 8.13561879911103e-06,
+      "loss": 0.8044,
+      "step": 10400
+    },
+    {
+      "epoch": 0.5724585833012273,
+      "grad_norm": 0.6749194860458374,
+      "learning_rate": 8.135281151671298e-06,
+      "loss": 0.6426,
+      "step": 10401
+    },
+    {
+      "epoch": 0.572513622103583,
+      "grad_norm": 0.7370286583900452,
+      "learning_rate": 8.134943480667635e-06,
+      "loss": 0.8051,
+      "step": 10402
+    },
+    {
+      "epoch": 0.5725686609059387,
+      "grad_norm": 0.6827631592750549,
+      "learning_rate": 8.134605786102574e-06,
+      "loss": 0.6961,
+      "step": 10403
+    },
+    {
+      "epoch": 0.5726236997082943,
+      "grad_norm": 0.7593247294425964,
+      "learning_rate": 8.134268067978655e-06,
+      "loss": 0.7514,
+      "step": 10404
+    },
+    {
+      "epoch": 0.57267873851065,
+      "grad_norm": 0.7229800224304199,
+      "learning_rate": 8.133930326298417e-06,
+      "loss": 0.8105,
+      "step": 10405
+    },
+    {
+      "epoch": 0.5727337773130057,
+      "grad_norm": 0.720973551273346,
+      "learning_rate": 8.133592561064396e-06,
+      "loss": 0.6866,
+      "step": 10406
+    },
+    {
+      "epoch": 0.5727888161153614,
+      "grad_norm": 0.7530742883682251,
+      "learning_rate": 8.133254772279135e-06,
+      "loss": 0.773,
+      "step": 10407
+    },
+    {
+      "epoch": 0.5728438549177169,
+      "grad_norm": 0.6897457838058472,
+      "learning_rate": 8.132916959945167e-06,
+      "loss": 0.8107,
+      "step": 10408
+    },
+    {
+      "epoch": 0.5728988937200726,
+      "grad_norm": 0.6659066081047058,
+      "learning_rate": 8.132579124065034e-06,
+      "loss": 0.8036,
+      "step": 10409
+    },
+    {
+      "epoch": 0.5729539325224283,
+      "grad_norm": 0.6925005316734314,
+      "learning_rate": 8.132241264641276e-06,
+      "loss": 0.7869,
+      "step": 10410
+    },
+    {
+      "epoch": 0.573008971324784,
+      "grad_norm": 0.8681634068489075,
+      "learning_rate": 8.131903381676433e-06,
+      "loss": 0.7411,
+      "step": 10411
+    },
+    {
+      "epoch": 0.5730640101271396,
+      "grad_norm": 0.669561505317688,
+      "learning_rate": 8.13156547517304e-06,
+      "loss": 0.7398,
+      "step": 10412
+    },
+    {
+      "epoch": 0.5731190489294953,
+      "grad_norm": 0.6737409234046936,
+      "learning_rate": 8.131227545133639e-06,
+      "loss": 0.7319,
+      "step": 10413
+    },
+    {
+      "epoch": 0.573174087731851,
+      "grad_norm": 0.7111513614654541,
+      "learning_rate": 8.130889591560772e-06,
+      "loss": 0.7192,
+      "step": 10414
+    },
+    {
+      "epoch": 0.5732291265342067,
+      "grad_norm": 0.6618744134902954,
+      "learning_rate": 8.130551614456974e-06,
+      "loss": 0.6636,
+      "step": 10415
+    },
+    {
+      "epoch": 0.5732841653365622,
+      "grad_norm": 0.8150144815444946,
+      "learning_rate": 8.13021361382479e-06,
+      "loss": 0.7168,
+      "step": 10416
+    },
+    {
+      "epoch": 0.5733392041389179,
+      "grad_norm": 0.744898796081543,
+      "learning_rate": 8.129875589666758e-06,
+      "loss": 0.8562,
+      "step": 10417
+    },
+    {
+      "epoch": 0.5733942429412736,
+      "grad_norm": 0.7831705212593079,
+      "learning_rate": 8.129537541985419e-06,
+      "loss": 0.8491,
+      "step": 10418
+    },
+    {
+      "epoch": 0.5734492817436293,
+      "grad_norm": 0.8097667098045349,
+      "learning_rate": 8.129199470783313e-06,
+      "loss": 0.7623,
+      "step": 10419
+    },
+    {
+      "epoch": 0.5735043205459849,
+      "grad_norm": 0.7951840758323669,
+      "learning_rate": 8.128861376062982e-06,
+      "loss": 0.8195,
+      "step": 10420
+    },
+    {
+      "epoch": 0.5735593593483406,
+      "grad_norm": 0.5902833938598633,
+      "learning_rate": 8.128523257826966e-06,
+      "loss": 0.6244,
+      "step": 10421
+    },
+    {
+      "epoch": 0.5736143981506963,
+      "grad_norm": 1.113287329673767,
+      "learning_rate": 8.128185116077805e-06,
+      "loss": 0.8382,
+      "step": 10422
+    },
+    {
+      "epoch": 0.573669436953052,
+      "grad_norm": 0.6899390816688538,
+      "learning_rate": 8.127846950818046e-06,
+      "loss": 0.7632,
+      "step": 10423
+    },
+    {
+      "epoch": 0.5737244757554075,
+      "grad_norm": 0.6905965805053711,
+      "learning_rate": 8.127508762050225e-06,
+      "loss": 0.7429,
+      "step": 10424
+    },
+    {
+      "epoch": 0.5737795145577632,
+      "grad_norm": 0.7036122679710388,
+      "learning_rate": 8.127170549776882e-06,
+      "loss": 0.7699,
+      "step": 10425
+    },
+    {
+      "epoch": 0.5738345533601189,
+      "grad_norm": 0.6599798202514648,
+      "learning_rate": 8.126832314000566e-06,
+      "loss": 0.7169,
+      "step": 10426
+    },
+    {
+      "epoch": 0.5738895921624746,
+      "grad_norm": 0.8682155609130859,
+      "learning_rate": 8.126494054723815e-06,
+      "loss": 0.851,
+      "step": 10427
+    },
+    {
+      "epoch": 0.5739446309648302,
+      "grad_norm": 0.6661516427993774,
+      "learning_rate": 8.12615577194917e-06,
+      "loss": 0.7287,
+      "step": 10428
+    },
+    {
+      "epoch": 0.5739996697671859,
+      "grad_norm": 0.6805256009101868,
+      "learning_rate": 8.125817465679176e-06,
+      "loss": 0.7033,
+      "step": 10429
+    },
+    {
+      "epoch": 0.5740547085695415,
+      "grad_norm": 0.7088646292686462,
+      "learning_rate": 8.125479135916375e-06,
+      "loss": 0.7295,
+      "step": 10430
+    },
+    {
+      "epoch": 0.5741097473718972,
+      "grad_norm": 0.6854971647262573,
+      "learning_rate": 8.12514078266331e-06,
+      "loss": 0.8102,
+      "step": 10431
+    },
+    {
+      "epoch": 0.5741647861742528,
+      "grad_norm": 0.7481474876403809,
+      "learning_rate": 8.124802405922521e-06,
+      "loss": 0.7463,
+      "step": 10432
+    },
+    {
+      "epoch": 0.5742198249766085,
+      "grad_norm": 0.8280898928642273,
+      "learning_rate": 8.124464005696556e-06,
+      "loss": 0.8067,
+      "step": 10433
+    },
+    {
+      "epoch": 0.5742748637789642,
+      "grad_norm": 0.696812629699707,
+      "learning_rate": 8.124125581987953e-06,
+      "loss": 0.7041,
+      "step": 10434
+    },
+    {
+      "epoch": 0.5743299025813199,
+      "grad_norm": 0.791084349155426,
+      "learning_rate": 8.123787134799262e-06,
+      "loss": 0.8244,
+      "step": 10435
+    },
+    {
+      "epoch": 0.5743849413836755,
+      "grad_norm": 0.7422665953636169,
+      "learning_rate": 8.123448664133022e-06,
+      "loss": 0.7792,
+      "step": 10436
+    },
+    {
+      "epoch": 0.5744399801860312,
+      "grad_norm": 0.7302834987640381,
+      "learning_rate": 8.123110169991777e-06,
+      "loss": 0.7617,
+      "step": 10437
+    },
+    {
+      "epoch": 0.5744950189883868,
+      "grad_norm": 0.6640440821647644,
+      "learning_rate": 8.122771652378071e-06,
+      "loss": 0.7965,
+      "step": 10438
+    },
+    {
+      "epoch": 0.5745500577907424,
+      "grad_norm": 0.7704516649246216,
+      "learning_rate": 8.12243311129445e-06,
+      "loss": 0.7814,
+      "step": 10439
+    },
+    {
+      "epoch": 0.5746050965930981,
+      "grad_norm": 0.673254668712616,
+      "learning_rate": 8.122094546743459e-06,
+      "loss": 0.7364,
+      "step": 10440
+    },
+    {
+      "epoch": 0.5746601353954538,
+      "grad_norm": 0.7648451924324036,
+      "learning_rate": 8.121755958727639e-06,
+      "loss": 0.8585,
+      "step": 10441
+    },
+    {
+      "epoch": 0.5747151741978095,
+      "grad_norm": 0.6660173535346985,
+      "learning_rate": 8.121417347249539e-06,
+      "loss": 0.6989,
+      "step": 10442
+    },
+    {
+      "epoch": 0.5747702130001651,
+      "grad_norm": 0.7128653526306152,
+      "learning_rate": 8.1210787123117e-06,
+      "loss": 0.8317,
+      "step": 10443
+    },
+    {
+      "epoch": 0.5748252518025208,
+      "grad_norm": 0.6404966115951538,
+      "learning_rate": 8.12074005391667e-06,
+      "loss": 0.6957,
+      "step": 10444
+    },
+    {
+      "epoch": 0.5748802906048764,
+      "grad_norm": 0.9597657918930054,
+      "learning_rate": 8.120401372066993e-06,
+      "loss": 0.9266,
+      "step": 10445
+    },
+    {
+      "epoch": 0.5749353294072321,
+      "grad_norm": 0.7735045552253723,
+      "learning_rate": 8.120062666765213e-06,
+      "loss": 0.8159,
+      "step": 10446
+    },
+    {
+      "epoch": 0.5749903682095877,
+      "grad_norm": 0.8031814098358154,
+      "learning_rate": 8.11972393801388e-06,
+      "loss": 0.7741,
+      "step": 10447
+    },
+    {
+      "epoch": 0.5750454070119434,
+      "grad_norm": 0.7008558511734009,
+      "learning_rate": 8.119385185815535e-06,
+      "loss": 0.6558,
+      "step": 10448
+    },
+    {
+      "epoch": 0.5751004458142991,
+      "grad_norm": 0.8162875175476074,
+      "learning_rate": 8.119046410172725e-06,
+      "loss": 0.7196,
+      "step": 10449
+    },
+    {
+      "epoch": 0.5751554846166548,
+      "grad_norm": 0.8142701983451843,
+      "learning_rate": 8.118707611088e-06,
+      "loss": 0.7709,
+      "step": 10450
+    },
+    {
+      "epoch": 0.5752105234190104,
+      "grad_norm": 0.7671986818313599,
+      "learning_rate": 8.118368788563902e-06,
+      "loss": 0.8725,
+      "step": 10451
+    },
+    {
+      "epoch": 0.575265562221366,
+      "grad_norm": 0.6604374051094055,
+      "learning_rate": 8.118029942602979e-06,
+      "loss": 0.7119,
+      "step": 10452
+    },
+    {
+      "epoch": 0.5753206010237217,
+      "grad_norm": 0.7119179368019104,
+      "learning_rate": 8.117691073207776e-06,
+      "loss": 0.7445,
+      "step": 10453
+    },
+    {
+      "epoch": 0.5753756398260774,
+      "grad_norm": 0.7572842240333557,
+      "learning_rate": 8.117352180380843e-06,
+      "loss": 0.7672,
+      "step": 10454
+    },
+    {
+      "epoch": 0.575430678628433,
+      "grad_norm": 0.688667356967926,
+      "learning_rate": 8.117013264124725e-06,
+      "loss": 0.7733,
+      "step": 10455
+    },
+    {
+      "epoch": 0.5754857174307887,
+      "grad_norm": 0.6683163046836853,
+      "learning_rate": 8.116674324441971e-06,
+      "loss": 0.6381,
+      "step": 10456
+    },
+    {
+      "epoch": 0.5755407562331444,
+      "grad_norm": 0.7792099714279175,
+      "learning_rate": 8.116335361335126e-06,
+      "loss": 0.7781,
+      "step": 10457
+    },
+    {
+      "epoch": 0.5755957950355001,
+      "grad_norm": 0.702132523059845,
+      "learning_rate": 8.115996374806738e-06,
+      "loss": 0.7442,
+      "step": 10458
+    },
+    {
+      "epoch": 0.5756508338378556,
+      "grad_norm": 0.7021365761756897,
+      "learning_rate": 8.115657364859356e-06,
+      "loss": 0.7215,
+      "step": 10459
+    },
+    {
+      "epoch": 0.5757058726402113,
+      "grad_norm": 0.7032247185707092,
+      "learning_rate": 8.115318331495527e-06,
+      "loss": 0.7069,
+      "step": 10460
+    },
+    {
+      "epoch": 0.575760911442567,
+      "grad_norm": 0.8301237225532532,
+      "learning_rate": 8.1149792747178e-06,
+      "loss": 0.789,
+      "step": 10461
+    },
+    {
+      "epoch": 0.5758159502449227,
+      "grad_norm": 0.7051018476486206,
+      "learning_rate": 8.11464019452872e-06,
+      "loss": 0.7511,
+      "step": 10462
+    },
+    {
+      "epoch": 0.5758709890472783,
+      "grad_norm": 0.8422626256942749,
+      "learning_rate": 8.114301090930843e-06,
+      "loss": 0.6507,
+      "step": 10463
+    },
+    {
+      "epoch": 0.575926027849634,
+      "grad_norm": 0.7751632332801819,
+      "learning_rate": 8.113961963926708e-06,
+      "loss": 0.7357,
+      "step": 10464
+    },
+    {
+      "epoch": 0.5759810666519897,
+      "grad_norm": 0.7158333659172058,
+      "learning_rate": 8.11362281351887e-06,
+      "loss": 0.8382,
+      "step": 10465
+    },
+    {
+      "epoch": 0.5760361054543454,
+      "grad_norm": 0.6926481127738953,
+      "learning_rate": 8.113283639709878e-06,
+      "loss": 0.7078,
+      "step": 10466
+    },
+    {
+      "epoch": 0.5760911442567009,
+      "grad_norm": 0.7091588973999023,
+      "learning_rate": 8.112944442502277e-06,
+      "loss": 0.7932,
+      "step": 10467
+    },
+    {
+      "epoch": 0.5761461830590566,
+      "grad_norm": 0.6979780197143555,
+      "learning_rate": 8.11260522189862e-06,
+      "loss": 0.6812,
+      "step": 10468
+    },
+    {
+      "epoch": 0.5762012218614123,
+      "grad_norm": 0.6735736131668091,
+      "learning_rate": 8.112265977901455e-06,
+      "loss": 0.7499,
+      "step": 10469
+    },
+    {
+      "epoch": 0.576256260663768,
+      "grad_norm": 0.6995692849159241,
+      "learning_rate": 8.111926710513334e-06,
+      "loss": 0.7123,
+      "step": 10470
+    },
+    {
+      "epoch": 0.5763112994661236,
+      "grad_norm": 0.7162681818008423,
+      "learning_rate": 8.111587419736802e-06,
+      "loss": 0.7586,
+      "step": 10471
+    },
+    {
+      "epoch": 0.5763663382684793,
+      "grad_norm": 0.945935070514679,
+      "learning_rate": 8.111248105574414e-06,
+      "loss": 0.8474,
+      "step": 10472
+    },
+    {
+      "epoch": 0.576421377070835,
+      "grad_norm": 0.608730673789978,
+      "learning_rate": 8.110908768028716e-06,
+      "loss": 0.6433,
+      "step": 10473
+    },
+    {
+      "epoch": 0.5764764158731907,
+      "grad_norm": 0.6777853965759277,
+      "learning_rate": 8.110569407102263e-06,
+      "loss": 0.7913,
+      "step": 10474
+    },
+    {
+      "epoch": 0.5765314546755462,
+      "grad_norm": 0.6310930848121643,
+      "learning_rate": 8.1102300227976e-06,
+      "loss": 0.719,
+      "step": 10475
+    },
+    {
+      "epoch": 0.5765864934779019,
+      "grad_norm": 0.7048485279083252,
+      "learning_rate": 8.109890615117282e-06,
+      "loss": 0.7341,
+      "step": 10476
+    },
+    {
+      "epoch": 0.5766415322802576,
+      "grad_norm": 0.672987163066864,
+      "learning_rate": 8.10955118406386e-06,
+      "loss": 0.7637,
+      "step": 10477
+    },
+    {
+      "epoch": 0.5766965710826133,
+      "grad_norm": 0.7018216252326965,
+      "learning_rate": 8.109211729639882e-06,
+      "loss": 0.6924,
+      "step": 10478
+    },
+    {
+      "epoch": 0.5767516098849689,
+      "grad_norm": 0.7183761596679688,
+      "learning_rate": 8.108872251847901e-06,
+      "loss": 0.7945,
+      "step": 10479
+    },
+    {
+      "epoch": 0.5768066486873246,
+      "grad_norm": 0.7332683801651001,
+      "learning_rate": 8.108532750690469e-06,
+      "loss": 0.7686,
+      "step": 10480
+    },
+    {
+      "epoch": 0.5768616874896803,
+      "grad_norm": 0.7118290066719055,
+      "learning_rate": 8.108193226170139e-06,
+      "loss": 0.6917,
+      "step": 10481
+    },
+    {
+      "epoch": 0.5769167262920358,
+      "grad_norm": 0.8242507576942444,
+      "learning_rate": 8.107853678289456e-06,
+      "loss": 0.9119,
+      "step": 10482
+    },
+    {
+      "epoch": 0.5769717650943915,
+      "grad_norm": 0.7138590216636658,
+      "learning_rate": 8.10751410705098e-06,
+      "loss": 0.7095,
+      "step": 10483
+    },
+    {
+      "epoch": 0.5770268038967472,
+      "grad_norm": 0.7541199326515198,
+      "learning_rate": 8.107174512457259e-06,
+      "loss": 0.8042,
+      "step": 10484
+    },
+    {
+      "epoch": 0.5770818426991029,
+      "grad_norm": 0.7776939868927002,
+      "learning_rate": 8.106834894510846e-06,
+      "loss": 0.8075,
+      "step": 10485
+    },
+    {
+      "epoch": 0.5771368815014585,
+      "grad_norm": 0.6466917395591736,
+      "learning_rate": 8.106495253214293e-06,
+      "loss": 0.707,
+      "step": 10486
+    },
+    {
+      "epoch": 0.5771919203038142,
+      "grad_norm": 0.687101423740387,
+      "learning_rate": 8.106155588570153e-06,
+      "loss": 0.6945,
+      "step": 10487
+    },
+    {
+      "epoch": 0.5772469591061699,
+      "grad_norm": 0.8338418006896973,
+      "learning_rate": 8.10581590058098e-06,
+      "loss": 0.8044,
+      "step": 10488
+    },
+    {
+      "epoch": 0.5773019979085255,
+      "grad_norm": 0.7052263617515564,
+      "learning_rate": 8.105476189249325e-06,
+      "loss": 0.8216,
+      "step": 10489
+    },
+    {
+      "epoch": 0.5773570367108811,
+      "grad_norm": 0.7205906510353088,
+      "learning_rate": 8.105136454577744e-06,
+      "loss": 0.8853,
+      "step": 10490
+    },
+    {
+      "epoch": 0.5774120755132368,
+      "grad_norm": 0.7875076532363892,
+      "learning_rate": 8.10479669656879e-06,
+      "loss": 0.822,
+      "step": 10491
+    },
+    {
+      "epoch": 0.5774671143155925,
+      "grad_norm": 0.6858797669410706,
+      "learning_rate": 8.104456915225012e-06,
+      "loss": 0.7924,
+      "step": 10492
+    },
+    {
+      "epoch": 0.5775221531179482,
+      "grad_norm": 0.6991322636604309,
+      "learning_rate": 8.104117110548968e-06,
+      "loss": 0.8144,
+      "step": 10493
+    },
+    {
+      "epoch": 0.5775771919203038,
+      "grad_norm": 0.7768846750259399,
+      "learning_rate": 8.103777282543209e-06,
+      "loss": 0.7793,
+      "step": 10494
+    },
+    {
+      "epoch": 0.5776322307226595,
+      "grad_norm": 0.7055716514587402,
+      "learning_rate": 8.103437431210293e-06,
+      "loss": 0.7653,
+      "step": 10495
+    },
+    {
+      "epoch": 0.5776872695250151,
+      "grad_norm": 1.009839653968811,
+      "learning_rate": 8.10309755655277e-06,
+      "loss": 0.7646,
+      "step": 10496
+    },
+    {
+      "epoch": 0.5777423083273708,
+      "grad_norm": 0.699435293674469,
+      "learning_rate": 8.102757658573197e-06,
+      "loss": 0.7806,
+      "step": 10497
+    },
+    {
+      "epoch": 0.5777973471297264,
+      "grad_norm": 0.8566381931304932,
+      "learning_rate": 8.102417737274129e-06,
+      "loss": 0.8302,
+      "step": 10498
+    },
+    {
+      "epoch": 0.5778523859320821,
+      "grad_norm": 0.745801568031311,
+      "learning_rate": 8.10207779265812e-06,
+      "loss": 0.91,
+      "step": 10499
+    },
+    {
+      "epoch": 0.5779074247344378,
+      "grad_norm": 0.6867349743843079,
+      "learning_rate": 8.101737824727724e-06,
+      "loss": 0.771,
+      "step": 10500
+    },
+    {
+      "epoch": 0.5779624635367935,
+      "grad_norm": 0.6693048477172852,
+      "learning_rate": 8.101397833485496e-06,
+      "loss": 0.7967,
+      "step": 10501
+    },
+    {
+      "epoch": 0.5780175023391491,
+      "grad_norm": 0.7485450506210327,
+      "learning_rate": 8.101057818933993e-06,
+      "loss": 0.7132,
+      "step": 10502
+    },
+    {
+      "epoch": 0.5780725411415047,
+      "grad_norm": 0.7619839906692505,
+      "learning_rate": 8.100717781075769e-06,
+      "loss": 0.7379,
+      "step": 10503
+    },
+    {
+      "epoch": 0.5781275799438604,
+      "grad_norm": 0.7651955485343933,
+      "learning_rate": 8.100377719913382e-06,
+      "loss": 0.8437,
+      "step": 10504
+    },
+    {
+      "epoch": 0.5781826187462161,
+      "grad_norm": 0.692385196685791,
+      "learning_rate": 8.100037635449384e-06,
+      "loss": 0.7666,
+      "step": 10505
+    },
+    {
+      "epoch": 0.5782376575485717,
+      "grad_norm": 0.7332374453544617,
+      "learning_rate": 8.099697527686334e-06,
+      "loss": 0.7476,
+      "step": 10506
+    },
+    {
+      "epoch": 0.5782926963509274,
+      "grad_norm": 0.6934877634048462,
+      "learning_rate": 8.099357396626786e-06,
+      "loss": 0.8054,
+      "step": 10507
+    },
+    {
+      "epoch": 0.5783477351532831,
+      "grad_norm": 0.8393011689186096,
+      "learning_rate": 8.099017242273298e-06,
+      "loss": 0.8655,
+      "step": 10508
+    },
+    {
+      "epoch": 0.5784027739556388,
+      "grad_norm": 0.6850646734237671,
+      "learning_rate": 8.098677064628425e-06,
+      "loss": 0.7424,
+      "step": 10509
+    },
+    {
+      "epoch": 0.5784578127579943,
+      "grad_norm": 0.7302095293998718,
+      "learning_rate": 8.098336863694728e-06,
+      "loss": 0.903,
+      "step": 10510
+    },
+    {
+      "epoch": 0.57851285156035,
+      "grad_norm": 0.7474033236503601,
+      "learning_rate": 8.097996639474757e-06,
+      "loss": 0.7509,
+      "step": 10511
+    },
+    {
+      "epoch": 0.5785678903627057,
+      "grad_norm": 0.6525655388832092,
+      "learning_rate": 8.097656391971074e-06,
+      "loss": 0.7097,
+      "step": 10512
+    },
+    {
+      "epoch": 0.5786229291650614,
+      "grad_norm": 0.8197451829910278,
+      "learning_rate": 8.097316121186234e-06,
+      "loss": 0.7401,
+      "step": 10513
+    },
+    {
+      "epoch": 0.578677967967417,
+      "grad_norm": 0.7048231959342957,
+      "learning_rate": 8.096975827122795e-06,
+      "loss": 0.7964,
+      "step": 10514
+    },
+    {
+      "epoch": 0.5787330067697727,
+      "grad_norm": 0.8417022228240967,
+      "learning_rate": 8.096635509783315e-06,
+      "loss": 0.7703,
+      "step": 10515
+    },
+    {
+      "epoch": 0.5787880455721284,
+      "grad_norm": 0.7313926815986633,
+      "learning_rate": 8.096295169170352e-06,
+      "loss": 0.7565,
+      "step": 10516
+    },
+    {
+      "epoch": 0.5788430843744841,
+      "grad_norm": 0.7156692147254944,
+      "learning_rate": 8.095954805286464e-06,
+      "loss": 0.7456,
+      "step": 10517
+    },
+    {
+      "epoch": 0.5788981231768396,
+      "grad_norm": 0.7366768717765808,
+      "learning_rate": 8.095614418134205e-06,
+      "loss": 0.72,
+      "step": 10518
+    },
+    {
+      "epoch": 0.5789531619791953,
+      "grad_norm": 0.7011533379554749,
+      "learning_rate": 8.09527400771614e-06,
+      "loss": 0.7683,
+      "step": 10519
+    },
+    {
+      "epoch": 0.579008200781551,
+      "grad_norm": 0.6849086284637451,
+      "learning_rate": 8.094933574034823e-06,
+      "loss": 0.6938,
+      "step": 10520
+    },
+    {
+      "epoch": 0.5790632395839067,
+      "grad_norm": 0.7351469397544861,
+      "learning_rate": 8.094593117092814e-06,
+      "loss": 0.7364,
+      "step": 10521
+    },
+    {
+      "epoch": 0.5791182783862623,
+      "grad_norm": 0.7133724689483643,
+      "learning_rate": 8.09425263689267e-06,
+      "loss": 0.7328,
+      "step": 10522
+    },
+    {
+      "epoch": 0.579173317188618,
+      "grad_norm": 0.6713461875915527,
+      "learning_rate": 8.093912133436954e-06,
+      "loss": 0.7296,
+      "step": 10523
+    },
+    {
+      "epoch": 0.5792283559909737,
+      "grad_norm": 0.7057825922966003,
+      "learning_rate": 8.093571606728222e-06,
+      "loss": 0.7732,
+      "step": 10524
+    },
+    {
+      "epoch": 0.5792833947933292,
+      "grad_norm": 0.7378783226013184,
+      "learning_rate": 8.093231056769033e-06,
+      "loss": 0.7907,
+      "step": 10525
+    },
+    {
+      "epoch": 0.5793384335956849,
+      "grad_norm": 0.8796947598457336,
+      "learning_rate": 8.092890483561947e-06,
+      "loss": 0.7325,
+      "step": 10526
+    },
+    {
+      "epoch": 0.5793934723980406,
+      "grad_norm": 0.7326352000236511,
+      "learning_rate": 8.092549887109525e-06,
+      "loss": 0.7948,
+      "step": 10527
+    },
+    {
+      "epoch": 0.5794485112003963,
+      "grad_norm": 0.7131063342094421,
+      "learning_rate": 8.092209267414325e-06,
+      "loss": 0.7595,
+      "step": 10528
+    },
+    {
+      "epoch": 0.5795035500027519,
+      "grad_norm": 0.6993252635002136,
+      "learning_rate": 8.091868624478908e-06,
+      "loss": 0.782,
+      "step": 10529
+    },
+    {
+      "epoch": 0.5795585888051076,
+      "grad_norm": 0.6945857405662537,
+      "learning_rate": 8.091527958305835e-06,
+      "loss": 0.7283,
+      "step": 10530
+    },
+    {
+      "epoch": 0.5796136276074633,
+      "grad_norm": 0.8203904032707214,
+      "learning_rate": 8.091187268897667e-06,
+      "loss": 0.7787,
+      "step": 10531
+    },
+    {
+      "epoch": 0.579668666409819,
+      "grad_norm": 0.6450221538543701,
+      "learning_rate": 8.09084655625696e-06,
+      "loss": 0.7092,
+      "step": 10532
+    },
+    {
+      "epoch": 0.5797237052121745,
+      "grad_norm": 0.6852096915245056,
+      "learning_rate": 8.090505820386279e-06,
+      "loss": 0.7916,
+      "step": 10533
+    },
+    {
+      "epoch": 0.5797787440145302,
+      "grad_norm": 1.0816445350646973,
+      "learning_rate": 8.090165061288182e-06,
+      "loss": 0.7545,
+      "step": 10534
+    },
+    {
+      "epoch": 0.5798337828168859,
+      "grad_norm": 0.7312847375869751,
+      "learning_rate": 8.089824278965233e-06,
+      "loss": 0.7395,
+      "step": 10535
+    },
+    {
+      "epoch": 0.5798888216192416,
+      "grad_norm": 0.7281426191329956,
+      "learning_rate": 8.089483473419992e-06,
+      "loss": 0.7677,
+      "step": 10536
+    },
+    {
+      "epoch": 0.5799438604215972,
+      "grad_norm": 0.7392409443855286,
+      "learning_rate": 8.08914264465502e-06,
+      "loss": 0.7674,
+      "step": 10537
+    },
+    {
+      "epoch": 0.5799988992239529,
+      "grad_norm": 0.7041863799095154,
+      "learning_rate": 8.088801792672877e-06,
+      "loss": 0.6156,
+      "step": 10538
+    },
+    {
+      "epoch": 0.5800539380263086,
+      "grad_norm": 0.7113755345344543,
+      "learning_rate": 8.088460917476128e-06,
+      "loss": 0.7677,
+      "step": 10539
+    },
+    {
+      "epoch": 0.5801089768286642,
+      "grad_norm": 0.673966646194458,
+      "learning_rate": 8.088120019067334e-06,
+      "loss": 0.7557,
+      "step": 10540
+    },
+    {
+      "epoch": 0.5801640156310198,
+      "grad_norm": 0.8165854215621948,
+      "learning_rate": 8.087779097449055e-06,
+      "loss": 0.8102,
+      "step": 10541
+    },
+    {
+      "epoch": 0.5802190544333755,
+      "grad_norm": 0.7010880708694458,
+      "learning_rate": 8.087438152623857e-06,
+      "loss": 0.7816,
+      "step": 10542
+    },
+    {
+      "epoch": 0.5802740932357312,
+      "grad_norm": 0.726177990436554,
+      "learning_rate": 8.0870971845943e-06,
+      "loss": 0.7671,
+      "step": 10543
+    },
+    {
+      "epoch": 0.5803291320380869,
+      "grad_norm": 0.7403919696807861,
+      "learning_rate": 8.086756193362946e-06,
+      "loss": 0.8449,
+      "step": 10544
+    },
+    {
+      "epoch": 0.5803841708404425,
+      "grad_norm": 0.6897104382514954,
+      "learning_rate": 8.086415178932358e-06,
+      "loss": 0.7563,
+      "step": 10545
+    },
+    {
+      "epoch": 0.5804392096427982,
+      "grad_norm": 0.7682604193687439,
+      "learning_rate": 8.0860741413051e-06,
+      "loss": 0.8019,
+      "step": 10546
+    },
+    {
+      "epoch": 0.5804942484451538,
+      "grad_norm": 0.7317522168159485,
+      "learning_rate": 8.085733080483736e-06,
+      "loss": 0.7446,
+      "step": 10547
+    },
+    {
+      "epoch": 0.5805492872475095,
+      "grad_norm": 0.8503430485725403,
+      "learning_rate": 8.085391996470826e-06,
+      "loss": 0.7343,
+      "step": 10548
+    },
+    {
+      "epoch": 0.5806043260498651,
+      "grad_norm": 0.8550657629966736,
+      "learning_rate": 8.085050889268937e-06,
+      "loss": 0.9267,
+      "step": 10549
+    },
+    {
+      "epoch": 0.5806593648522208,
+      "grad_norm": 0.7751224637031555,
+      "learning_rate": 8.084709758880633e-06,
+      "loss": 0.7404,
+      "step": 10550
+    },
+    {
+      "epoch": 0.5807144036545765,
+      "grad_norm": 0.6346186399459839,
+      "learning_rate": 8.084368605308475e-06,
+      "loss": 0.66,
+      "step": 10551
+    },
+    {
+      "epoch": 0.5807694424569322,
+      "grad_norm": 0.7295717597007751,
+      "learning_rate": 8.084027428555027e-06,
+      "loss": 0.8313,
+      "step": 10552
+    },
+    {
+      "epoch": 0.5808244812592878,
+      "grad_norm": 0.6962289810180664,
+      "learning_rate": 8.083686228622856e-06,
+      "loss": 0.7871,
+      "step": 10553
+    },
+    {
+      "epoch": 0.5808795200616435,
+      "grad_norm": 0.6968896389007568,
+      "learning_rate": 8.083345005514522e-06,
+      "loss": 0.7261,
+      "step": 10554
+    },
+    {
+      "epoch": 0.5809345588639991,
+      "grad_norm": 0.8374869227409363,
+      "learning_rate": 8.083003759232595e-06,
+      "loss": 0.797,
+      "step": 10555
+    },
+    {
+      "epoch": 0.5809895976663548,
+      "grad_norm": 0.6511034369468689,
+      "learning_rate": 8.082662489779637e-06,
+      "loss": 0.7237,
+      "step": 10556
+    },
+    {
+      "epoch": 0.5810446364687104,
+      "grad_norm": 0.6644287705421448,
+      "learning_rate": 8.082321197158212e-06,
+      "loss": 0.6969,
+      "step": 10557
+    },
+    {
+      "epoch": 0.5810996752710661,
+      "grad_norm": 0.7681102752685547,
+      "learning_rate": 8.081979881370884e-06,
+      "loss": 0.7193,
+      "step": 10558
+    },
+    {
+      "epoch": 0.5811547140734218,
+      "grad_norm": 0.7930792570114136,
+      "learning_rate": 8.081638542420224e-06,
+      "loss": 0.7198,
+      "step": 10559
+    },
+    {
+      "epoch": 0.5812097528757775,
+      "grad_norm": 0.7227992415428162,
+      "learning_rate": 8.081297180308791e-06,
+      "loss": 0.7533,
+      "step": 10560
+    },
+    {
+      "epoch": 0.581264791678133,
+      "grad_norm": 0.7293071150779724,
+      "learning_rate": 8.080955795039156e-06,
+      "loss": 0.6228,
+      "step": 10561
+    },
+    {
+      "epoch": 0.5813198304804887,
+      "grad_norm": 0.7356483936309814,
+      "learning_rate": 8.080614386613879e-06,
+      "loss": 0.7299,
+      "step": 10562
+    },
+    {
+      "epoch": 0.5813748692828444,
+      "grad_norm": 0.8181473016738892,
+      "learning_rate": 8.080272955035531e-06,
+      "loss": 0.6576,
+      "step": 10563
+    },
+    {
+      "epoch": 0.5814299080852001,
+      "grad_norm": 0.7066958546638489,
+      "learning_rate": 8.079931500306675e-06,
+      "loss": 0.7372,
+      "step": 10564
+    },
+    {
+      "epoch": 0.5814849468875557,
+      "grad_norm": 0.6821097135543823,
+      "learning_rate": 8.079590022429877e-06,
+      "loss": 0.7516,
+      "step": 10565
+    },
+    {
+      "epoch": 0.5815399856899114,
+      "grad_norm": 0.6879069209098816,
+      "learning_rate": 8.079248521407707e-06,
+      "loss": 0.7525,
+      "step": 10566
+    },
+    {
+      "epoch": 0.5815950244922671,
+      "grad_norm": 0.956345796585083,
+      "learning_rate": 8.078906997242729e-06,
+      "loss": 0.8175,
+      "step": 10567
+    },
+    {
+      "epoch": 0.5816500632946227,
+      "grad_norm": 0.6942328214645386,
+      "learning_rate": 8.078565449937508e-06,
+      "loss": 0.6264,
+      "step": 10568
+    },
+    {
+      "epoch": 0.5817051020969783,
+      "grad_norm": 0.7073766589164734,
+      "learning_rate": 8.078223879494615e-06,
+      "loss": 0.766,
+      "step": 10569
+    },
+    {
+      "epoch": 0.581760140899334,
+      "grad_norm": 0.7649571895599365,
+      "learning_rate": 8.077882285916614e-06,
+      "loss": 0.8767,
+      "step": 10570
+    },
+    {
+      "epoch": 0.5818151797016897,
+      "grad_norm": 0.6384355425834656,
+      "learning_rate": 8.077540669206076e-06,
+      "loss": 0.7444,
+      "step": 10571
+    },
+    {
+      "epoch": 0.5818702185040453,
+      "grad_norm": 0.7173928022384644,
+      "learning_rate": 8.077199029365565e-06,
+      "loss": 0.8277,
+      "step": 10572
+    },
+    {
+      "epoch": 0.581925257306401,
+      "grad_norm": 0.7310757637023926,
+      "learning_rate": 8.076857366397648e-06,
+      "loss": 0.8425,
+      "step": 10573
+    },
+    {
+      "epoch": 0.5819802961087567,
+      "grad_norm": 0.6888872385025024,
+      "learning_rate": 8.076515680304897e-06,
+      "loss": 0.6961,
+      "step": 10574
+    },
+    {
+      "epoch": 0.5820353349111124,
+      "grad_norm": 0.7290124297142029,
+      "learning_rate": 8.076173971089877e-06,
+      "loss": 0.7865,
+      "step": 10575
+    },
+    {
+      "epoch": 0.582090373713468,
+      "grad_norm": 0.7402634024620056,
+      "learning_rate": 8.075832238755156e-06,
+      "loss": 0.7196,
+      "step": 10576
+    },
+    {
+      "epoch": 0.5821454125158236,
+      "grad_norm": 0.74916672706604,
+      "learning_rate": 8.075490483303305e-06,
+      "loss": 0.8361,
+      "step": 10577
+    },
+    {
+      "epoch": 0.5822004513181793,
+      "grad_norm": 0.8146494626998901,
+      "learning_rate": 8.07514870473689e-06,
+      "loss": 0.7398,
+      "step": 10578
+    },
+    {
+      "epoch": 0.582255490120535,
+      "grad_norm": 0.6632487177848816,
+      "learning_rate": 8.07480690305848e-06,
+      "loss": 0.7239,
+      "step": 10579
+    },
+    {
+      "epoch": 0.5823105289228906,
+      "grad_norm": 0.6912766695022583,
+      "learning_rate": 8.074465078270645e-06,
+      "loss": 0.7488,
+      "step": 10580
+    },
+    {
+      "epoch": 0.5823655677252463,
+      "grad_norm": 0.7410522699356079,
+      "learning_rate": 8.074123230375952e-06,
+      "loss": 0.7413,
+      "step": 10581
+    },
+    {
+      "epoch": 0.582420606527602,
+      "grad_norm": 0.7932689189910889,
+      "learning_rate": 8.073781359376972e-06,
+      "loss": 0.7894,
+      "step": 10582
+    },
+    {
+      "epoch": 0.5824756453299577,
+      "grad_norm": 0.6710309982299805,
+      "learning_rate": 8.073439465276277e-06,
+      "loss": 0.6727,
+      "step": 10583
+    },
+    {
+      "epoch": 0.5825306841323132,
+      "grad_norm": 0.7457143068313599,
+      "learning_rate": 8.07309754807643e-06,
+      "loss": 0.6719,
+      "step": 10584
+    },
+    {
+      "epoch": 0.5825857229346689,
+      "grad_norm": 0.7340453863143921,
+      "learning_rate": 8.072755607780008e-06,
+      "loss": 0.7397,
+      "step": 10585
+    },
+    {
+      "epoch": 0.5826407617370246,
+      "grad_norm": 0.7532176971435547,
+      "learning_rate": 8.072413644389574e-06,
+      "loss": 0.7368,
+      "step": 10586
+    },
+    {
+      "epoch": 0.5826958005393803,
+      "grad_norm": 0.9317812919616699,
+      "learning_rate": 8.072071657907703e-06,
+      "loss": 0.9113,
+      "step": 10587
+    },
+    {
+      "epoch": 0.5827508393417359,
+      "grad_norm": 0.8535491228103638,
+      "learning_rate": 8.071729648336963e-06,
+      "loss": 0.7708,
+      "step": 10588
+    },
+    {
+      "epoch": 0.5828058781440916,
+      "grad_norm": 0.6720348000526428,
+      "learning_rate": 8.071387615679926e-06,
+      "loss": 0.7521,
+      "step": 10589
+    },
+    {
+      "epoch": 0.5828609169464473,
+      "grad_norm": 0.7113864421844482,
+      "learning_rate": 8.071045559939162e-06,
+      "loss": 0.8713,
+      "step": 10590
+    },
+    {
+      "epoch": 0.582915955748803,
+      "grad_norm": 0.7760024070739746,
+      "learning_rate": 8.070703481117242e-06,
+      "loss": 0.7567,
+      "step": 10591
+    },
+    {
+      "epoch": 0.5829709945511585,
+      "grad_norm": 0.9548617005348206,
+      "learning_rate": 8.070361379216735e-06,
+      "loss": 0.7937,
+      "step": 10592
+    },
+    {
+      "epoch": 0.5830260333535142,
+      "grad_norm": 0.7796840667724609,
+      "learning_rate": 8.070019254240216e-06,
+      "loss": 0.7485,
+      "step": 10593
+    },
+    {
+      "epoch": 0.5830810721558699,
+      "grad_norm": 0.7006514668464661,
+      "learning_rate": 8.069677106190253e-06,
+      "loss": 0.7813,
+      "step": 10594
+    },
+    {
+      "epoch": 0.5831361109582256,
+      "grad_norm": 0.646396279335022,
+      "learning_rate": 8.069334935069417e-06,
+      "loss": 0.7437,
+      "step": 10595
+    },
+    {
+      "epoch": 0.5831911497605812,
+      "grad_norm": 0.8257368206977844,
+      "learning_rate": 8.068992740880283e-06,
+      "loss": 0.7351,
+      "step": 10596
+    },
+    {
+      "epoch": 0.5832461885629369,
+      "grad_norm": 0.6646208763122559,
+      "learning_rate": 8.068650523625422e-06,
+      "loss": 0.6554,
+      "step": 10597
+    },
+    {
+      "epoch": 0.5833012273652926,
+      "grad_norm": 0.8495579957962036,
+      "learning_rate": 8.068308283307402e-06,
+      "loss": 0.791,
+      "step": 10598
+    },
+    {
+      "epoch": 0.5833562661676482,
+      "grad_norm": 0.7283076047897339,
+      "learning_rate": 8.0679660199288e-06,
+      "loss": 0.7327,
+      "step": 10599
+    },
+    {
+      "epoch": 0.5834113049700038,
+      "grad_norm": 0.704572856426239,
+      "learning_rate": 8.067623733492187e-06,
+      "loss": 0.6094,
+      "step": 10600
+    },
+    {
+      "epoch": 0.5834663437723595,
+      "grad_norm": 0.6435144543647766,
+      "learning_rate": 8.067281424000136e-06,
+      "loss": 0.6974,
+      "step": 10601
+    },
+    {
+      "epoch": 0.5835213825747152,
+      "grad_norm": 0.9628346562385559,
+      "learning_rate": 8.066939091455215e-06,
+      "loss": 0.8933,
+      "step": 10602
+    },
+    {
+      "epoch": 0.5835764213770709,
+      "grad_norm": 0.6856930255889893,
+      "learning_rate": 8.066596735860004e-06,
+      "loss": 0.7414,
+      "step": 10603
+    },
+    {
+      "epoch": 0.5836314601794265,
+      "grad_norm": 0.7341175675392151,
+      "learning_rate": 8.066254357217072e-06,
+      "loss": 0.7553,
+      "step": 10604
+    },
+    {
+      "epoch": 0.5836864989817822,
+      "grad_norm": 0.7124871611595154,
+      "learning_rate": 8.065911955528995e-06,
+      "loss": 0.663,
+      "step": 10605
+    },
+    {
+      "epoch": 0.5837415377841378,
+      "grad_norm": 0.816028892993927,
+      "learning_rate": 8.065569530798341e-06,
+      "loss": 0.8778,
+      "step": 10606
+    },
+    {
+      "epoch": 0.5837965765864935,
+      "grad_norm": 0.8735721111297607,
+      "learning_rate": 8.06522708302769e-06,
+      "loss": 0.7866,
+      "step": 10607
+    },
+    {
+      "epoch": 0.5838516153888491,
+      "grad_norm": 0.6780036687850952,
+      "learning_rate": 8.06488461221961e-06,
+      "loss": 0.7329,
+      "step": 10608
+    },
+    {
+      "epoch": 0.5839066541912048,
+      "grad_norm": 0.7624822854995728,
+      "learning_rate": 8.06454211837668e-06,
+      "loss": 0.8095,
+      "step": 10609
+    },
+    {
+      "epoch": 0.5839616929935605,
+      "grad_norm": 0.8269234895706177,
+      "learning_rate": 8.06419960150147e-06,
+      "loss": 0.7194,
+      "step": 10610
+    },
+    {
+      "epoch": 0.5840167317959161,
+      "grad_norm": 0.6748649477958679,
+      "learning_rate": 8.063857061596558e-06,
+      "loss": 0.702,
+      "step": 10611
+    },
+    {
+      "epoch": 0.5840717705982718,
+      "grad_norm": 0.9700273275375366,
+      "learning_rate": 8.063514498664515e-06,
+      "loss": 0.7917,
+      "step": 10612
+    },
+    {
+      "epoch": 0.5841268094006274,
+      "grad_norm": 0.7798827290534973,
+      "learning_rate": 8.063171912707916e-06,
+      "loss": 0.798,
+      "step": 10613
+    },
+    {
+      "epoch": 0.5841818482029831,
+      "grad_norm": 0.6613249778747559,
+      "learning_rate": 8.06282930372934e-06,
+      "loss": 0.7216,
+      "step": 10614
+    },
+    {
+      "epoch": 0.5842368870053387,
+      "grad_norm": 0.727116048336029,
+      "learning_rate": 8.062486671731357e-06,
+      "loss": 0.8054,
+      "step": 10615
+    },
+    {
+      "epoch": 0.5842919258076944,
+      "grad_norm": 0.6704444289207458,
+      "learning_rate": 8.062144016716543e-06,
+      "loss": 0.7503,
+      "step": 10616
+    },
+    {
+      "epoch": 0.5843469646100501,
+      "grad_norm": 0.6867938041687012,
+      "learning_rate": 8.061801338687477e-06,
+      "loss": 0.8005,
+      "step": 10617
+    },
+    {
+      "epoch": 0.5844020034124058,
+      "grad_norm": 0.7097555994987488,
+      "learning_rate": 8.061458637646729e-06,
+      "loss": 0.8515,
+      "step": 10618
+    },
+    {
+      "epoch": 0.5844570422147614,
+      "grad_norm": 0.6624881625175476,
+      "learning_rate": 8.061115913596878e-06,
+      "loss": 0.7735,
+      "step": 10619
+    },
+    {
+      "epoch": 0.584512081017117,
+      "grad_norm": 0.6649004220962524,
+      "learning_rate": 8.060773166540498e-06,
+      "loss": 0.7837,
+      "step": 10620
+    },
+    {
+      "epoch": 0.5845671198194727,
+      "grad_norm": 0.6732968091964722,
+      "learning_rate": 8.06043039648017e-06,
+      "loss": 0.7846,
+      "step": 10621
+    },
+    {
+      "epoch": 0.5846221586218284,
+      "grad_norm": 0.7551947236061096,
+      "learning_rate": 8.060087603418464e-06,
+      "loss": 0.6868,
+      "step": 10622
+    },
+    {
+      "epoch": 0.584677197424184,
+      "grad_norm": 0.7781728506088257,
+      "learning_rate": 8.059744787357959e-06,
+      "loss": 0.8088,
+      "step": 10623
+    },
+    {
+      "epoch": 0.5847322362265397,
+      "grad_norm": 0.6362790465354919,
+      "learning_rate": 8.05940194830123e-06,
+      "loss": 0.664,
+      "step": 10624
+    },
+    {
+      "epoch": 0.5847872750288954,
+      "grad_norm": 0.670386791229248,
+      "learning_rate": 8.059059086250856e-06,
+      "loss": 0.6839,
+      "step": 10625
+    },
+    {
+      "epoch": 0.5848423138312511,
+      "grad_norm": 0.7030045986175537,
+      "learning_rate": 8.058716201209414e-06,
+      "loss": 0.7243,
+      "step": 10626
+    },
+    {
+      "epoch": 0.5848973526336066,
+      "grad_norm": 0.7881805896759033,
+      "learning_rate": 8.058373293179477e-06,
+      "loss": 0.7994,
+      "step": 10627
+    },
+    {
+      "epoch": 0.5849523914359623,
+      "grad_norm": 0.7077344059944153,
+      "learning_rate": 8.058030362163628e-06,
+      "loss": 0.822,
+      "step": 10628
+    },
+    {
+      "epoch": 0.585007430238318,
+      "grad_norm": 0.6787039637565613,
+      "learning_rate": 8.057687408164439e-06,
+      "loss": 0.7619,
+      "step": 10629
+    },
+    {
+      "epoch": 0.5850624690406737,
+      "grad_norm": 1.1377217769622803,
+      "learning_rate": 8.05734443118449e-06,
+      "loss": 0.8632,
+      "step": 10630
+    },
+    {
+      "epoch": 0.5851175078430293,
+      "grad_norm": 0.7002600431442261,
+      "learning_rate": 8.05700143122636e-06,
+      "loss": 0.8184,
+      "step": 10631
+    },
+    {
+      "epoch": 0.585172546645385,
+      "grad_norm": 0.7016324400901794,
+      "learning_rate": 8.056658408292626e-06,
+      "loss": 0.658,
+      "step": 10632
+    },
+    {
+      "epoch": 0.5852275854477407,
+      "grad_norm": 0.6674843430519104,
+      "learning_rate": 8.056315362385864e-06,
+      "loss": 0.7281,
+      "step": 10633
+    },
+    {
+      "epoch": 0.5852826242500964,
+      "grad_norm": 0.6789288520812988,
+      "learning_rate": 8.055972293508653e-06,
+      "loss": 0.8192,
+      "step": 10634
+    },
+    {
+      "epoch": 0.5853376630524519,
+      "grad_norm": 0.6740062236785889,
+      "learning_rate": 8.055629201663575e-06,
+      "loss": 0.7343,
+      "step": 10635
+    },
+    {
+      "epoch": 0.5853927018548076,
+      "grad_norm": 0.7417730689048767,
+      "learning_rate": 8.055286086853204e-06,
+      "loss": 0.8161,
+      "step": 10636
+    },
+    {
+      "epoch": 0.5854477406571633,
+      "grad_norm": 0.6680465340614319,
+      "learning_rate": 8.054942949080122e-06,
+      "loss": 0.7589,
+      "step": 10637
+    },
+    {
+      "epoch": 0.585502779459519,
+      "grad_norm": 0.7205108404159546,
+      "learning_rate": 8.054599788346904e-06,
+      "loss": 0.6837,
+      "step": 10638
+    },
+    {
+      "epoch": 0.5855578182618746,
+      "grad_norm": 0.8694404363632202,
+      "learning_rate": 8.054256604656134e-06,
+      "loss": 0.8033,
+      "step": 10639
+    },
+    {
+      "epoch": 0.5856128570642303,
+      "grad_norm": 0.685471773147583,
+      "learning_rate": 8.053913398010389e-06,
+      "loss": 0.7654,
+      "step": 10640
+    },
+    {
+      "epoch": 0.585667895866586,
+      "grad_norm": 1.3463424444198608,
+      "learning_rate": 8.053570168412249e-06,
+      "loss": 0.7743,
+      "step": 10641
+    },
+    {
+      "epoch": 0.5857229346689417,
+      "grad_norm": 0.9380106329917908,
+      "learning_rate": 8.05322691586429e-06,
+      "loss": 0.8984,
+      "step": 10642
+    },
+    {
+      "epoch": 0.5857779734712972,
+      "grad_norm": 0.7408519387245178,
+      "learning_rate": 8.052883640369096e-06,
+      "loss": 0.7716,
+      "step": 10643
+    },
+    {
+      "epoch": 0.5858330122736529,
+      "grad_norm": 0.7712904214859009,
+      "learning_rate": 8.052540341929248e-06,
+      "loss": 0.7767,
+      "step": 10644
+    },
+    {
+      "epoch": 0.5858880510760086,
+      "grad_norm": 0.8464158177375793,
+      "learning_rate": 8.052197020547321e-06,
+      "loss": 0.8333,
+      "step": 10645
+    },
+    {
+      "epoch": 0.5859430898783643,
+      "grad_norm": 0.6970158219337463,
+      "learning_rate": 8.0518536762259e-06,
+      "loss": 0.7354,
+      "step": 10646
+    },
+    {
+      "epoch": 0.5859981286807199,
+      "grad_norm": 0.7048965096473694,
+      "learning_rate": 8.051510308967563e-06,
+      "loss": 0.8333,
+      "step": 10647
+    },
+    {
+      "epoch": 0.5860531674830756,
+      "grad_norm": 0.6443868279457092,
+      "learning_rate": 8.05116691877489e-06,
+      "loss": 0.7386,
+      "step": 10648
+    },
+    {
+      "epoch": 0.5861082062854313,
+      "grad_norm": 0.6653542518615723,
+      "learning_rate": 8.050823505650465e-06,
+      "loss": 0.8116,
+      "step": 10649
+    },
+    {
+      "epoch": 0.5861632450877869,
+      "grad_norm": 0.7293158769607544,
+      "learning_rate": 8.050480069596868e-06,
+      "loss": 0.8231,
+      "step": 10650
+    },
+    {
+      "epoch": 0.5862182838901425,
+      "grad_norm": 0.6876117587089539,
+      "learning_rate": 8.050136610616676e-06,
+      "loss": 0.7856,
+      "step": 10651
+    },
+    {
+      "epoch": 0.5862733226924982,
+      "grad_norm": 0.6811665296554565,
+      "learning_rate": 8.049793128712477e-06,
+      "loss": 0.7667,
+      "step": 10652
+    },
+    {
+      "epoch": 0.5863283614948539,
+      "grad_norm": 0.701034426689148,
+      "learning_rate": 8.049449623886849e-06,
+      "loss": 0.7812,
+      "step": 10653
+    },
+    {
+      "epoch": 0.5863834002972095,
+      "grad_norm": 0.6872833967208862,
+      "learning_rate": 8.049106096142372e-06,
+      "loss": 0.755,
+      "step": 10654
+    },
+    {
+      "epoch": 0.5864384390995652,
+      "grad_norm": 0.6643580198287964,
+      "learning_rate": 8.04876254548163e-06,
+      "loss": 0.7692,
+      "step": 10655
+    },
+    {
+      "epoch": 0.5864934779019209,
+      "grad_norm": 0.6672106981277466,
+      "learning_rate": 8.048418971907206e-06,
+      "loss": 0.7424,
+      "step": 10656
+    },
+    {
+      "epoch": 0.5865485167042765,
+      "grad_norm": 0.8030515313148499,
+      "learning_rate": 8.04807537542168e-06,
+      "loss": 0.8074,
+      "step": 10657
+    },
+    {
+      "epoch": 0.5866035555066321,
+      "grad_norm": 0.713417112827301,
+      "learning_rate": 8.047731756027637e-06,
+      "loss": 0.6974,
+      "step": 10658
+    },
+    {
+      "epoch": 0.5866585943089878,
+      "grad_norm": 0.7715572118759155,
+      "learning_rate": 8.047388113727657e-06,
+      "loss": 0.7353,
+      "step": 10659
+    },
+    {
+      "epoch": 0.5867136331113435,
+      "grad_norm": 0.7009812593460083,
+      "learning_rate": 8.047044448524323e-06,
+      "loss": 0.7992,
+      "step": 10660
+    },
+    {
+      "epoch": 0.5867686719136992,
+      "grad_norm": 0.6425079107284546,
+      "learning_rate": 8.046700760420219e-06,
+      "loss": 0.7394,
+      "step": 10661
+    },
+    {
+      "epoch": 0.5868237107160548,
+      "grad_norm": 0.7713460922241211,
+      "learning_rate": 8.046357049417927e-06,
+      "loss": 0.7759,
+      "step": 10662
+    },
+    {
+      "epoch": 0.5868787495184105,
+      "grad_norm": 0.7310347557067871,
+      "learning_rate": 8.046013315520033e-06,
+      "loss": 0.7278,
+      "step": 10663
+    },
+    {
+      "epoch": 0.5869337883207661,
+      "grad_norm": 0.7493315935134888,
+      "learning_rate": 8.045669558729117e-06,
+      "loss": 0.7808,
+      "step": 10664
+    },
+    {
+      "epoch": 0.5869888271231218,
+      "grad_norm": 0.7547439336776733,
+      "learning_rate": 8.045325779047763e-06,
+      "loss": 0.8245,
+      "step": 10665
+    },
+    {
+      "epoch": 0.5870438659254774,
+      "grad_norm": 0.7556985020637512,
+      "learning_rate": 8.044981976478557e-06,
+      "loss": 0.8,
+      "step": 10666
+    },
+    {
+      "epoch": 0.5870989047278331,
+      "grad_norm": 0.8330736756324768,
+      "learning_rate": 8.04463815102408e-06,
+      "loss": 0.8177,
+      "step": 10667
+    },
+    {
+      "epoch": 0.5871539435301888,
+      "grad_norm": 0.7823941111564636,
+      "learning_rate": 8.04429430268692e-06,
+      "loss": 0.8306,
+      "step": 10668
+    },
+    {
+      "epoch": 0.5872089823325445,
+      "grad_norm": 0.9141719937324524,
+      "learning_rate": 8.043950431469657e-06,
+      "loss": 0.9137,
+      "step": 10669
+    },
+    {
+      "epoch": 0.5872640211349001,
+      "grad_norm": 0.6967095732688904,
+      "learning_rate": 8.043606537374878e-06,
+      "loss": 0.7262,
+      "step": 10670
+    },
+    {
+      "epoch": 0.5873190599372557,
+      "grad_norm": 0.7909649014472961,
+      "learning_rate": 8.043262620405166e-06,
+      "loss": 0.8332,
+      "step": 10671
+    },
+    {
+      "epoch": 0.5873740987396114,
+      "grad_norm": 0.7967168092727661,
+      "learning_rate": 8.042918680563107e-06,
+      "loss": 0.7966,
+      "step": 10672
+    },
+    {
+      "epoch": 0.5874291375419671,
+      "grad_norm": 0.7637625336647034,
+      "learning_rate": 8.042574717851287e-06,
+      "loss": 0.8322,
+      "step": 10673
+    },
+    {
+      "epoch": 0.5874841763443227,
+      "grad_norm": 0.6968004107475281,
+      "learning_rate": 8.04223073227229e-06,
+      "loss": 0.8061,
+      "step": 10674
+    },
+    {
+      "epoch": 0.5875392151466784,
+      "grad_norm": 0.7325586080551147,
+      "learning_rate": 8.0418867238287e-06,
+      "loss": 0.7922,
+      "step": 10675
+    },
+    {
+      "epoch": 0.5875942539490341,
+      "grad_norm": 0.6784406304359436,
+      "learning_rate": 8.041542692523103e-06,
+      "loss": 0.7327,
+      "step": 10676
+    },
+    {
+      "epoch": 0.5876492927513898,
+      "grad_norm": 0.8297861218452454,
+      "learning_rate": 8.041198638358088e-06,
+      "loss": 0.9347,
+      "step": 10677
+    },
+    {
+      "epoch": 0.5877043315537454,
+      "grad_norm": 0.6227413415908813,
+      "learning_rate": 8.040854561336236e-06,
+      "loss": 0.655,
+      "step": 10678
+    },
+    {
+      "epoch": 0.587759370356101,
+      "grad_norm": 0.752098023891449,
+      "learning_rate": 8.040510461460134e-06,
+      "loss": 0.7608,
+      "step": 10679
+    },
+    {
+      "epoch": 0.5878144091584567,
+      "grad_norm": 0.7008342146873474,
+      "learning_rate": 8.040166338732372e-06,
+      "loss": 0.7385,
+      "step": 10680
+    },
+    {
+      "epoch": 0.5878694479608124,
+      "grad_norm": 0.6768027544021606,
+      "learning_rate": 8.039822193155532e-06,
+      "loss": 0.6812,
+      "step": 10681
+    },
+    {
+      "epoch": 0.587924486763168,
+      "grad_norm": 0.7728545069694519,
+      "learning_rate": 8.039478024732203e-06,
+      "loss": 0.7696,
+      "step": 10682
+    },
+    {
+      "epoch": 0.5879795255655237,
+      "grad_norm": 0.7257505655288696,
+      "learning_rate": 8.03913383346497e-06,
+      "loss": 0.6686,
+      "step": 10683
+    },
+    {
+      "epoch": 0.5880345643678794,
+      "grad_norm": 0.7755837440490723,
+      "learning_rate": 8.03878961935642e-06,
+      "loss": 0.8469,
+      "step": 10684
+    },
+    {
+      "epoch": 0.5880896031702351,
+      "grad_norm": 0.7187668085098267,
+      "learning_rate": 8.038445382409142e-06,
+      "loss": 0.8249,
+      "step": 10685
+    },
+    {
+      "epoch": 0.5881446419725906,
+      "grad_norm": 0.638053834438324,
+      "learning_rate": 8.038101122625722e-06,
+      "loss": 0.6876,
+      "step": 10686
+    },
+    {
+      "epoch": 0.5881996807749463,
+      "grad_norm": 0.7323756217956543,
+      "learning_rate": 8.037756840008746e-06,
+      "loss": 0.7489,
+      "step": 10687
+    },
+    {
+      "epoch": 0.588254719577302,
+      "grad_norm": 0.6795439720153809,
+      "learning_rate": 8.037412534560804e-06,
+      "loss": 0.7246,
+      "step": 10688
+    },
+    {
+      "epoch": 0.5883097583796577,
+      "grad_norm": 0.8136376142501831,
+      "learning_rate": 8.037068206284482e-06,
+      "loss": 0.8518,
+      "step": 10689
+    },
+    {
+      "epoch": 0.5883647971820133,
+      "grad_norm": 0.6484195590019226,
+      "learning_rate": 8.036723855182367e-06,
+      "loss": 0.7018,
+      "step": 10690
+    },
+    {
+      "epoch": 0.588419835984369,
+      "grad_norm": 0.7465028166770935,
+      "learning_rate": 8.036379481257048e-06,
+      "loss": 0.8276,
+      "step": 10691
+    },
+    {
+      "epoch": 0.5884748747867247,
+      "grad_norm": 0.7761173844337463,
+      "learning_rate": 8.036035084511116e-06,
+      "loss": 0.6371,
+      "step": 10692
+    },
+    {
+      "epoch": 0.5885299135890804,
+      "grad_norm": 0.830008864402771,
+      "learning_rate": 8.035690664947156e-06,
+      "loss": 0.8199,
+      "step": 10693
+    },
+    {
+      "epoch": 0.5885849523914359,
+      "grad_norm": 0.6614254117012024,
+      "learning_rate": 8.03534622256776e-06,
+      "loss": 0.656,
+      "step": 10694
+    },
+    {
+      "epoch": 0.5886399911937916,
+      "grad_norm": 0.7229047417640686,
+      "learning_rate": 8.035001757375509e-06,
+      "loss": 0.7622,
+      "step": 10695
+    },
+    {
+      "epoch": 0.5886950299961473,
+      "grad_norm": 0.7044325470924377,
+      "learning_rate": 8.034657269373001e-06,
+      "loss": 0.7678,
+      "step": 10696
+    },
+    {
+      "epoch": 0.5887500687985029,
+      "grad_norm": 0.7109018564224243,
+      "learning_rate": 8.03431275856282e-06,
+      "loss": 0.7976,
+      "step": 10697
+    },
+    {
+      "epoch": 0.5888051076008586,
+      "grad_norm": 0.7812879085540771,
+      "learning_rate": 8.033968224947557e-06,
+      "loss": 0.7163,
+      "step": 10698
+    },
+    {
+      "epoch": 0.5888601464032143,
+      "grad_norm": 0.7408469915390015,
+      "learning_rate": 8.033623668529802e-06,
+      "loss": 0.6895,
+      "step": 10699
+    },
+    {
+      "epoch": 0.58891518520557,
+      "grad_norm": 0.7654302716255188,
+      "learning_rate": 8.033279089312142e-06,
+      "loss": 0.8126,
+      "step": 10700
+    },
+    {
+      "epoch": 0.5889702240079255,
+      "grad_norm": 0.7307846546173096,
+      "learning_rate": 8.032934487297169e-06,
+      "loss": 0.7958,
+      "step": 10701
+    },
+    {
+      "epoch": 0.5890252628102812,
+      "grad_norm": 0.6658591032028198,
+      "learning_rate": 8.032589862487472e-06,
+      "loss": 0.717,
+      "step": 10702
+    },
+    {
+      "epoch": 0.5890803016126369,
+      "grad_norm": 1.4167139530181885,
+      "learning_rate": 8.03224521488564e-06,
+      "loss": 0.8599,
+      "step": 10703
+    },
+    {
+      "epoch": 0.5891353404149926,
+      "grad_norm": 0.6723609566688538,
+      "learning_rate": 8.031900544494266e-06,
+      "loss": 0.8167,
+      "step": 10704
+    },
+    {
+      "epoch": 0.5891903792173482,
+      "grad_norm": 0.6420501470565796,
+      "learning_rate": 8.03155585131594e-06,
+      "loss": 0.692,
+      "step": 10705
+    },
+    {
+      "epoch": 0.5892454180197039,
+      "grad_norm": 0.6973454356193542,
+      "learning_rate": 8.031211135353251e-06,
+      "loss": 0.7709,
+      "step": 10706
+    },
+    {
+      "epoch": 0.5893004568220596,
+      "grad_norm": 0.7752252221107483,
+      "learning_rate": 8.03086639660879e-06,
+      "loss": 0.7795,
+      "step": 10707
+    },
+    {
+      "epoch": 0.5893554956244152,
+      "grad_norm": 0.8193135857582092,
+      "learning_rate": 8.030521635085149e-06,
+      "loss": 0.812,
+      "step": 10708
+    },
+    {
+      "epoch": 0.5894105344267708,
+      "grad_norm": 0.7976878881454468,
+      "learning_rate": 8.03017685078492e-06,
+      "loss": 0.8039,
+      "step": 10709
+    },
+    {
+      "epoch": 0.5894655732291265,
+      "grad_norm": 0.7545839548110962,
+      "learning_rate": 8.02983204371069e-06,
+      "loss": 0.8238,
+      "step": 10710
+    },
+    {
+      "epoch": 0.5895206120314822,
+      "grad_norm": 0.6544732451438904,
+      "learning_rate": 8.029487213865054e-06,
+      "loss": 0.7471,
+      "step": 10711
+    },
+    {
+      "epoch": 0.5895756508338379,
+      "grad_norm": 0.7054508924484253,
+      "learning_rate": 8.029142361250603e-06,
+      "loss": 0.8283,
+      "step": 10712
+    },
+    {
+      "epoch": 0.5896306896361935,
+      "grad_norm": 0.7425236105918884,
+      "learning_rate": 8.02879748586993e-06,
+      "loss": 0.8031,
+      "step": 10713
+    },
+    {
+      "epoch": 0.5896857284385492,
+      "grad_norm": 0.8390052318572998,
+      "learning_rate": 8.028452587725626e-06,
+      "loss": 0.7218,
+      "step": 10714
+    },
+    {
+      "epoch": 0.5897407672409049,
+      "grad_norm": 0.8116903901100159,
+      "learning_rate": 8.028107666820282e-06,
+      "loss": 0.8057,
+      "step": 10715
+    },
+    {
+      "epoch": 0.5897958060432605,
+      "grad_norm": 0.602308452129364,
+      "learning_rate": 8.027762723156492e-06,
+      "loss": 0.6428,
+      "step": 10716
+    },
+    {
+      "epoch": 0.5898508448456161,
+      "grad_norm": 0.7480159401893616,
+      "learning_rate": 8.027417756736848e-06,
+      "loss": 0.7566,
+      "step": 10717
+    },
+    {
+      "epoch": 0.5899058836479718,
+      "grad_norm": 0.6823177933692932,
+      "learning_rate": 8.027072767563943e-06,
+      "loss": 0.8337,
+      "step": 10718
+    },
+    {
+      "epoch": 0.5899609224503275,
+      "grad_norm": 0.6841796040534973,
+      "learning_rate": 8.026727755640367e-06,
+      "loss": 0.751,
+      "step": 10719
+    },
+    {
+      "epoch": 0.5900159612526832,
+      "grad_norm": 0.7257139086723328,
+      "learning_rate": 8.026382720968718e-06,
+      "loss": 0.7373,
+      "step": 10720
+    },
+    {
+      "epoch": 0.5900710000550388,
+      "grad_norm": 0.6318400502204895,
+      "learning_rate": 8.026037663551584e-06,
+      "loss": 0.7205,
+      "step": 10721
+    },
+    {
+      "epoch": 0.5901260388573945,
+      "grad_norm": 0.6612908840179443,
+      "learning_rate": 8.025692583391564e-06,
+      "loss": 0.7613,
+      "step": 10722
+    },
+    {
+      "epoch": 0.5901810776597501,
+      "grad_norm": 0.7555351853370667,
+      "learning_rate": 8.025347480491246e-06,
+      "loss": 0.718,
+      "step": 10723
+    },
+    {
+      "epoch": 0.5902361164621058,
+      "grad_norm": 0.6944366097450256,
+      "learning_rate": 8.025002354853227e-06,
+      "loss": 0.7775,
+      "step": 10724
+    },
+    {
+      "epoch": 0.5902911552644614,
+      "grad_norm": 0.6968230605125427,
+      "learning_rate": 8.0246572064801e-06,
+      "loss": 0.7316,
+      "step": 10725
+    },
+    {
+      "epoch": 0.5903461940668171,
+      "grad_norm": 0.7083567380905151,
+      "learning_rate": 8.024312035374459e-06,
+      "loss": 0.7844,
+      "step": 10726
+    },
+    {
+      "epoch": 0.5904012328691728,
+      "grad_norm": 0.7183080315589905,
+      "learning_rate": 8.0239668415389e-06,
+      "loss": 0.8308,
+      "step": 10727
+    },
+    {
+      "epoch": 0.5904562716715285,
+      "grad_norm": 0.8350495100021362,
+      "learning_rate": 8.023621624976014e-06,
+      "loss": 0.9077,
+      "step": 10728
+    },
+    {
+      "epoch": 0.590511310473884,
+      "grad_norm": 0.6876987218856812,
+      "learning_rate": 8.023276385688396e-06,
+      "loss": 0.7483,
+      "step": 10729
+    },
+    {
+      "epoch": 0.5905663492762397,
+      "grad_norm": 0.8617128133773804,
+      "learning_rate": 8.022931123678646e-06,
+      "loss": 0.7058,
+      "step": 10730
+    },
+    {
+      "epoch": 0.5906213880785954,
+      "grad_norm": 0.6921959519386292,
+      "learning_rate": 8.02258583894935e-06,
+      "loss": 0.7542,
+      "step": 10731
+    },
+    {
+      "epoch": 0.5906764268809511,
+      "grad_norm": 0.7394077181816101,
+      "learning_rate": 8.02224053150311e-06,
+      "loss": 0.7761,
+      "step": 10732
+    },
+    {
+      "epoch": 0.5907314656833067,
+      "grad_norm": 0.6672187447547913,
+      "learning_rate": 8.02189520134252e-06,
+      "loss": 0.6904,
+      "step": 10733
+    },
+    {
+      "epoch": 0.5907865044856624,
+      "grad_norm": 0.7498076558113098,
+      "learning_rate": 8.021549848470174e-06,
+      "loss": 0.7994,
+      "step": 10734
+    },
+    {
+      "epoch": 0.5908415432880181,
+      "grad_norm": 0.699832558631897,
+      "learning_rate": 8.021204472888669e-06,
+      "loss": 0.7413,
+      "step": 10735
+    },
+    {
+      "epoch": 0.5908965820903738,
+      "grad_norm": 0.7628722190856934,
+      "learning_rate": 8.020859074600598e-06,
+      "loss": 0.8202,
+      "step": 10736
+    },
+    {
+      "epoch": 0.5909516208927293,
+      "grad_norm": 0.8023744225502014,
+      "learning_rate": 8.020513653608558e-06,
+      "loss": 0.8225,
+      "step": 10737
+    },
+    {
+      "epoch": 0.591006659695085,
+      "grad_norm": 0.7283689379692078,
+      "learning_rate": 8.02016820991515e-06,
+      "loss": 0.6706,
+      "step": 10738
+    },
+    {
+      "epoch": 0.5910616984974407,
+      "grad_norm": 0.7199996113777161,
+      "learning_rate": 8.019822743522962e-06,
+      "loss": 0.8258,
+      "step": 10739
+    },
+    {
+      "epoch": 0.5911167372997963,
+      "grad_norm": 0.623249888420105,
+      "learning_rate": 8.019477254434598e-06,
+      "loss": 0.6188,
+      "step": 10740
+    },
+    {
+      "epoch": 0.591171776102152,
+      "grad_norm": 0.7331949472427368,
+      "learning_rate": 8.01913174265265e-06,
+      "loss": 0.8013,
+      "step": 10741
+    },
+    {
+      "epoch": 0.5912268149045077,
+      "grad_norm": 0.7003010511398315,
+      "learning_rate": 8.018786208179716e-06,
+      "loss": 0.8305,
+      "step": 10742
+    },
+    {
+      "epoch": 0.5912818537068634,
+      "grad_norm": 0.6879638433456421,
+      "learning_rate": 8.01844065101839e-06,
+      "loss": 0.7622,
+      "step": 10743
+    },
+    {
+      "epoch": 0.591336892509219,
+      "grad_norm": 0.6597324013710022,
+      "learning_rate": 8.018095071171276e-06,
+      "loss": 0.7362,
+      "step": 10744
+    },
+    {
+      "epoch": 0.5913919313115746,
+      "grad_norm": 0.664905846118927,
+      "learning_rate": 8.017749468640967e-06,
+      "loss": 0.7629,
+      "step": 10745
+    },
+    {
+      "epoch": 0.5914469701139303,
+      "grad_norm": 0.7358053922653198,
+      "learning_rate": 8.017403843430059e-06,
+      "loss": 0.7798,
+      "step": 10746
+    },
+    {
+      "epoch": 0.591502008916286,
+      "grad_norm": 0.699603259563446,
+      "learning_rate": 8.017058195541152e-06,
+      "loss": 0.6249,
+      "step": 10747
+    },
+    {
+      "epoch": 0.5915570477186416,
+      "grad_norm": 0.6736140847206116,
+      "learning_rate": 8.016712524976843e-06,
+      "loss": 0.6904,
+      "step": 10748
+    },
+    {
+      "epoch": 0.5916120865209973,
+      "grad_norm": 0.6803401112556458,
+      "learning_rate": 8.016366831739732e-06,
+      "loss": 0.6868,
+      "step": 10749
+    },
+    {
+      "epoch": 0.591667125323353,
+      "grad_norm": 0.7152959704399109,
+      "learning_rate": 8.016021115832413e-06,
+      "loss": 0.7747,
+      "step": 10750
+    },
+    {
+      "epoch": 0.5917221641257087,
+      "grad_norm": 0.6469255685806274,
+      "learning_rate": 8.015675377257489e-06,
+      "loss": 0.7309,
+      "step": 10751
+    },
+    {
+      "epoch": 0.5917772029280642,
+      "grad_norm": 0.7902734875679016,
+      "learning_rate": 8.015329616017554e-06,
+      "loss": 0.7575,
+      "step": 10752
+    },
+    {
+      "epoch": 0.5918322417304199,
+      "grad_norm": 0.7447189688682556,
+      "learning_rate": 8.014983832115208e-06,
+      "loss": 0.7759,
+      "step": 10753
+    },
+    {
+      "epoch": 0.5918872805327756,
+      "grad_norm": 0.6135374903678894,
+      "learning_rate": 8.014638025553053e-06,
+      "loss": 0.6681,
+      "step": 10754
+    },
+    {
+      "epoch": 0.5919423193351313,
+      "grad_norm": 0.8614835739135742,
+      "learning_rate": 8.014292196333684e-06,
+      "loss": 0.7203,
+      "step": 10755
+    },
+    {
+      "epoch": 0.5919973581374869,
+      "grad_norm": 0.7649008631706238,
+      "learning_rate": 8.013946344459703e-06,
+      "loss": 0.7966,
+      "step": 10756
+    },
+    {
+      "epoch": 0.5920523969398426,
+      "grad_norm": 1.0862764120101929,
+      "learning_rate": 8.013600469933707e-06,
+      "loss": 0.866,
+      "step": 10757
+    },
+    {
+      "epoch": 0.5921074357421983,
+      "grad_norm": 0.7304185628890991,
+      "learning_rate": 8.013254572758296e-06,
+      "loss": 0.7599,
+      "step": 10758
+    },
+    {
+      "epoch": 0.592162474544554,
+      "grad_norm": 0.6329634785652161,
+      "learning_rate": 8.012908652936072e-06,
+      "loss": 0.6855,
+      "step": 10759
+    },
+    {
+      "epoch": 0.5922175133469095,
+      "grad_norm": 0.6692202687263489,
+      "learning_rate": 8.012562710469631e-06,
+      "loss": 0.817,
+      "step": 10760
+    },
+    {
+      "epoch": 0.5922725521492652,
+      "grad_norm": 0.6577631235122681,
+      "learning_rate": 8.012216745361577e-06,
+      "loss": 0.7813,
+      "step": 10761
+    },
+    {
+      "epoch": 0.5923275909516209,
+      "grad_norm": 0.6877861022949219,
+      "learning_rate": 8.011870757614506e-06,
+      "loss": 0.7142,
+      "step": 10762
+    },
+    {
+      "epoch": 0.5923826297539766,
+      "grad_norm": 0.7132022380828857,
+      "learning_rate": 8.011524747231023e-06,
+      "loss": 0.747,
+      "step": 10763
+    },
+    {
+      "epoch": 0.5924376685563322,
+      "grad_norm": 0.7841360569000244,
+      "learning_rate": 8.011178714213726e-06,
+      "loss": 0.7511,
+      "step": 10764
+    },
+    {
+      "epoch": 0.5924927073586879,
+      "grad_norm": 0.8572794198989868,
+      "learning_rate": 8.010832658565215e-06,
+      "loss": 0.8704,
+      "step": 10765
+    },
+    {
+      "epoch": 0.5925477461610436,
+      "grad_norm": 0.6825506687164307,
+      "learning_rate": 8.010486580288092e-06,
+      "loss": 0.7472,
+      "step": 10766
+    },
+    {
+      "epoch": 0.5926027849633992,
+      "grad_norm": 0.7484591603279114,
+      "learning_rate": 8.010140479384957e-06,
+      "loss": 0.7679,
+      "step": 10767
+    },
+    {
+      "epoch": 0.5926578237657548,
+      "grad_norm": 0.712602436542511,
+      "learning_rate": 8.009794355858412e-06,
+      "loss": 0.7706,
+      "step": 10768
+    },
+    {
+      "epoch": 0.5927128625681105,
+      "grad_norm": 0.8911493420600891,
+      "learning_rate": 8.00944820971106e-06,
+      "loss": 0.8396,
+      "step": 10769
+    },
+    {
+      "epoch": 0.5927679013704662,
+      "grad_norm": 0.7300251126289368,
+      "learning_rate": 8.009102040945498e-06,
+      "loss": 0.7611,
+      "step": 10770
+    },
+    {
+      "epoch": 0.5928229401728219,
+      "grad_norm": 0.727343738079071,
+      "learning_rate": 8.008755849564333e-06,
+      "loss": 0.6785,
+      "step": 10771
+    },
+    {
+      "epoch": 0.5928779789751775,
+      "grad_norm": 0.8323808908462524,
+      "learning_rate": 8.008409635570163e-06,
+      "loss": 0.7429,
+      "step": 10772
+    },
+    {
+      "epoch": 0.5929330177775332,
+      "grad_norm": 0.6651942133903503,
+      "learning_rate": 8.00806339896559e-06,
+      "loss": 0.7683,
+      "step": 10773
+    },
+    {
+      "epoch": 0.5929880565798888,
+      "grad_norm": 0.7164554595947266,
+      "learning_rate": 8.007717139753222e-06,
+      "loss": 0.7742,
+      "step": 10774
+    },
+    {
+      "epoch": 0.5930430953822445,
+      "grad_norm": 0.6906408667564392,
+      "learning_rate": 8.007370857935654e-06,
+      "loss": 0.7322,
+      "step": 10775
+    },
+    {
+      "epoch": 0.5930981341846001,
+      "grad_norm": 0.6384999752044678,
+      "learning_rate": 8.007024553515493e-06,
+      "loss": 0.7011,
+      "step": 10776
+    },
+    {
+      "epoch": 0.5931531729869558,
+      "grad_norm": 0.6997355222702026,
+      "learning_rate": 8.006678226495338e-06,
+      "loss": 0.7303,
+      "step": 10777
+    },
+    {
+      "epoch": 0.5932082117893115,
+      "grad_norm": 0.6730707287788391,
+      "learning_rate": 8.006331876877797e-06,
+      "loss": 0.7461,
+      "step": 10778
+    },
+    {
+      "epoch": 0.5932632505916672,
+      "grad_norm": 0.7529115080833435,
+      "learning_rate": 8.00598550466547e-06,
+      "loss": 0.7487,
+      "step": 10779
+    },
+    {
+      "epoch": 0.5933182893940228,
+      "grad_norm": 0.7186329960823059,
+      "learning_rate": 8.00563910986096e-06,
+      "loss": 0.8025,
+      "step": 10780
+    },
+    {
+      "epoch": 0.5933733281963784,
+      "grad_norm": 0.7523752450942993,
+      "learning_rate": 8.005292692466869e-06,
+      "loss": 0.8291,
+      "step": 10781
+    },
+    {
+      "epoch": 0.5934283669987341,
+      "grad_norm": 1.182645559310913,
+      "learning_rate": 8.004946252485806e-06,
+      "loss": 0.8037,
+      "step": 10782
+    },
+    {
+      "epoch": 0.5934834058010897,
+      "grad_norm": 0.736570417881012,
+      "learning_rate": 8.004599789920369e-06,
+      "loss": 0.8259,
+      "step": 10783
+    },
+    {
+      "epoch": 0.5935384446034454,
+      "grad_norm": 0.757665753364563,
+      "learning_rate": 8.004253304773165e-06,
+      "loss": 0.7773,
+      "step": 10784
+    },
+    {
+      "epoch": 0.5935934834058011,
+      "grad_norm": 0.6988566517829895,
+      "learning_rate": 8.003906797046798e-06,
+      "loss": 0.7895,
+      "step": 10785
+    },
+    {
+      "epoch": 0.5936485222081568,
+      "grad_norm": 0.6921454071998596,
+      "learning_rate": 8.00356026674387e-06,
+      "loss": 0.8068,
+      "step": 10786
+    },
+    {
+      "epoch": 0.5937035610105124,
+      "grad_norm": 0.7053877115249634,
+      "learning_rate": 8.003213713866988e-06,
+      "loss": 0.7632,
+      "step": 10787
+    },
+    {
+      "epoch": 0.593758599812868,
+      "grad_norm": 0.8193650245666504,
+      "learning_rate": 8.002867138418757e-06,
+      "loss": 0.759,
+      "step": 10788
+    },
+    {
+      "epoch": 0.5938136386152237,
+      "grad_norm": 0.6089804768562317,
+      "learning_rate": 8.002520540401779e-06,
+      "loss": 0.7117,
+      "step": 10789
+    },
+    {
+      "epoch": 0.5938686774175794,
+      "grad_norm": 0.6869456768035889,
+      "learning_rate": 8.002173919818662e-06,
+      "loss": 0.7724,
+      "step": 10790
+    },
+    {
+      "epoch": 0.593923716219935,
+      "grad_norm": 0.7279118895530701,
+      "learning_rate": 8.001827276672007e-06,
+      "loss": 0.7578,
+      "step": 10791
+    },
+    {
+      "epoch": 0.5939787550222907,
+      "grad_norm": 0.6960133910179138,
+      "learning_rate": 8.00148061096442e-06,
+      "loss": 0.7887,
+      "step": 10792
+    },
+    {
+      "epoch": 0.5940337938246464,
+      "grad_norm": 0.6774740815162659,
+      "learning_rate": 8.001133922698511e-06,
+      "loss": 0.7146,
+      "step": 10793
+    },
+    {
+      "epoch": 0.5940888326270021,
+      "grad_norm": 0.6696349382400513,
+      "learning_rate": 8.000787211876883e-06,
+      "loss": 0.7829,
+      "step": 10794
+    },
+    {
+      "epoch": 0.5941438714293577,
+      "grad_norm": 1.5037024021148682,
+      "learning_rate": 8.000440478502142e-06,
+      "loss": 0.8198,
+      "step": 10795
+    },
+    {
+      "epoch": 0.5941989102317133,
+      "grad_norm": 0.7373353838920593,
+      "learning_rate": 8.000093722576893e-06,
+      "loss": 0.7864,
+      "step": 10796
+    },
+    {
+      "epoch": 0.594253949034069,
+      "grad_norm": 0.8120700120925903,
+      "learning_rate": 7.999746944103743e-06,
+      "loss": 0.7918,
+      "step": 10797
+    },
+    {
+      "epoch": 0.5943089878364247,
+      "grad_norm": 0.7669811844825745,
+      "learning_rate": 7.999400143085296e-06,
+      "loss": 0.751,
+      "step": 10798
+    },
+    {
+      "epoch": 0.5943640266387803,
+      "grad_norm": 0.8090860843658447,
+      "learning_rate": 7.999053319524163e-06,
+      "loss": 0.8387,
+      "step": 10799
+    },
+    {
+      "epoch": 0.594419065441136,
+      "grad_norm": 0.6994315385818481,
+      "learning_rate": 7.998706473422945e-06,
+      "loss": 0.7084,
+      "step": 10800
+    },
+    {
+      "epoch": 0.5944741042434917,
+      "grad_norm": 0.7913107872009277,
+      "learning_rate": 7.998359604784254e-06,
+      "loss": 0.7454,
+      "step": 10801
+    },
+    {
+      "epoch": 0.5945291430458474,
+      "grad_norm": 0.6831398010253906,
+      "learning_rate": 7.998012713610696e-06,
+      "loss": 0.7422,
+      "step": 10802
+    },
+    {
+      "epoch": 0.5945841818482029,
+      "grad_norm": 0.7324068546295166,
+      "learning_rate": 7.997665799904875e-06,
+      "loss": 0.7622,
+      "step": 10803
+    },
+    {
+      "epoch": 0.5946392206505586,
+      "grad_norm": 0.8192811012268066,
+      "learning_rate": 7.997318863669399e-06,
+      "loss": 0.7783,
+      "step": 10804
+    },
+    {
+      "epoch": 0.5946942594529143,
+      "grad_norm": 0.8008341789245605,
+      "learning_rate": 7.996971904906879e-06,
+      "loss": 0.7673,
+      "step": 10805
+    },
+    {
+      "epoch": 0.59474929825527,
+      "grad_norm": 0.6899568438529968,
+      "learning_rate": 7.99662492361992e-06,
+      "loss": 0.7477,
+      "step": 10806
+    },
+    {
+      "epoch": 0.5948043370576256,
+      "grad_norm": 0.7322555780410767,
+      "learning_rate": 7.996277919811132e-06,
+      "loss": 0.7673,
+      "step": 10807
+    },
+    {
+      "epoch": 0.5948593758599813,
+      "grad_norm": 1.008300542831421,
+      "learning_rate": 7.995930893483117e-06,
+      "loss": 0.7556,
+      "step": 10808
+    },
+    {
+      "epoch": 0.594914414662337,
+      "grad_norm": 0.7211925387382507,
+      "learning_rate": 7.99558384463849e-06,
+      "loss": 0.761,
+      "step": 10809
+    },
+    {
+      "epoch": 0.5949694534646927,
+      "grad_norm": 0.7143383622169495,
+      "learning_rate": 7.995236773279855e-06,
+      "loss": 0.7972,
+      "step": 10810
+    },
+    {
+      "epoch": 0.5950244922670482,
+      "grad_norm": 0.7682802677154541,
+      "learning_rate": 7.994889679409825e-06,
+      "loss": 0.8538,
+      "step": 10811
+    },
+    {
+      "epoch": 0.5950795310694039,
+      "grad_norm": 0.6304698586463928,
+      "learning_rate": 7.994542563031004e-06,
+      "loss": 0.7343,
+      "step": 10812
+    },
+    {
+      "epoch": 0.5951345698717596,
+      "grad_norm": 0.6704440116882324,
+      "learning_rate": 7.994195424146002e-06,
+      "loss": 0.6921,
+      "step": 10813
+    },
+    {
+      "epoch": 0.5951896086741153,
+      "grad_norm": 0.8626209497451782,
+      "learning_rate": 7.99384826275743e-06,
+      "loss": 0.7049,
+      "step": 10814
+    },
+    {
+      "epoch": 0.5952446474764709,
+      "grad_norm": 0.810922384262085,
+      "learning_rate": 7.993501078867895e-06,
+      "loss": 0.793,
+      "step": 10815
+    },
+    {
+      "epoch": 0.5952996862788266,
+      "grad_norm": 0.8495855927467346,
+      "learning_rate": 7.993153872480009e-06,
+      "loss": 0.8078,
+      "step": 10816
+    },
+    {
+      "epoch": 0.5953547250811823,
+      "grad_norm": 0.7430331707000732,
+      "learning_rate": 7.992806643596378e-06,
+      "loss": 0.7957,
+      "step": 10817
+    },
+    {
+      "epoch": 0.595409763883538,
+      "grad_norm": 0.7188051342964172,
+      "learning_rate": 7.992459392219614e-06,
+      "loss": 0.725,
+      "step": 10818
+    },
+    {
+      "epoch": 0.5954648026858935,
+      "grad_norm": 0.7046926021575928,
+      "learning_rate": 7.992112118352326e-06,
+      "loss": 0.7438,
+      "step": 10819
+    },
+    {
+      "epoch": 0.5955198414882492,
+      "grad_norm": 0.7982804775238037,
+      "learning_rate": 7.991764821997123e-06,
+      "loss": 0.7046,
+      "step": 10820
+    },
+    {
+      "epoch": 0.5955748802906049,
+      "grad_norm": 0.6392245292663574,
+      "learning_rate": 7.991417503156618e-06,
+      "loss": 0.7413,
+      "step": 10821
+    },
+    {
+      "epoch": 0.5956299190929606,
+      "grad_norm": 0.7518960237503052,
+      "learning_rate": 7.99107016183342e-06,
+      "loss": 0.7661,
+      "step": 10822
+    },
+    {
+      "epoch": 0.5956849578953162,
+      "grad_norm": 0.7413721680641174,
+      "learning_rate": 7.99072279803014e-06,
+      "loss": 0.6538,
+      "step": 10823
+    },
+    {
+      "epoch": 0.5957399966976719,
+      "grad_norm": 0.7729454636573792,
+      "learning_rate": 7.990375411749384e-06,
+      "loss": 0.8056,
+      "step": 10824
+    },
+    {
+      "epoch": 0.5957950355000275,
+      "grad_norm": 0.8059296607971191,
+      "learning_rate": 7.99002800299377e-06,
+      "loss": 0.8699,
+      "step": 10825
+    },
+    {
+      "epoch": 0.5958500743023831,
+      "grad_norm": 0.5947105288505554,
+      "learning_rate": 7.989680571765907e-06,
+      "loss": 0.6481,
+      "step": 10826
+    },
+    {
+      "epoch": 0.5959051131047388,
+      "grad_norm": 0.7303743362426758,
+      "learning_rate": 7.989333118068404e-06,
+      "loss": 0.7401,
+      "step": 10827
+    },
+    {
+      "epoch": 0.5959601519070945,
+      "grad_norm": 0.7121400237083435,
+      "learning_rate": 7.988985641903873e-06,
+      "loss": 0.78,
+      "step": 10828
+    },
+    {
+      "epoch": 0.5960151907094502,
+      "grad_norm": 0.6921802163124084,
+      "learning_rate": 7.988638143274926e-06,
+      "loss": 0.7234,
+      "step": 10829
+    },
+    {
+      "epoch": 0.5960702295118058,
+      "grad_norm": 0.6715331673622131,
+      "learning_rate": 7.988290622184174e-06,
+      "loss": 0.7606,
+      "step": 10830
+    },
+    {
+      "epoch": 0.5961252683141615,
+      "grad_norm": 0.6315215229988098,
+      "learning_rate": 7.98794307863423e-06,
+      "loss": 0.6902,
+      "step": 10831
+    },
+    {
+      "epoch": 0.5961803071165171,
+      "grad_norm": 0.6884782314300537,
+      "learning_rate": 7.987595512627707e-06,
+      "loss": 0.7808,
+      "step": 10832
+    },
+    {
+      "epoch": 0.5962353459188728,
+      "grad_norm": 0.7050700783729553,
+      "learning_rate": 7.987247924167215e-06,
+      "loss": 0.7248,
+      "step": 10833
+    },
+    {
+      "epoch": 0.5962903847212284,
+      "grad_norm": 0.7232446074485779,
+      "learning_rate": 7.986900313255367e-06,
+      "loss": 0.8686,
+      "step": 10834
+    },
+    {
+      "epoch": 0.5963454235235841,
+      "grad_norm": 0.693631649017334,
+      "learning_rate": 7.986552679894778e-06,
+      "loss": 0.7567,
+      "step": 10835
+    },
+    {
+      "epoch": 0.5964004623259398,
+      "grad_norm": 0.6462356448173523,
+      "learning_rate": 7.986205024088054e-06,
+      "loss": 0.7091,
+      "step": 10836
+    },
+    {
+      "epoch": 0.5964555011282955,
+      "grad_norm": 0.7465559840202332,
+      "learning_rate": 7.985857345837814e-06,
+      "loss": 0.8965,
+      "step": 10837
+    },
+    {
+      "epoch": 0.5965105399306511,
+      "grad_norm": 0.6803271770477295,
+      "learning_rate": 7.985509645146672e-06,
+      "loss": 0.7602,
+      "step": 10838
+    },
+    {
+      "epoch": 0.5965655787330068,
+      "grad_norm": 1.1414798498153687,
+      "learning_rate": 7.985161922017238e-06,
+      "loss": 0.7806,
+      "step": 10839
+    },
+    {
+      "epoch": 0.5966206175353624,
+      "grad_norm": 0.6583230495452881,
+      "learning_rate": 7.984814176452123e-06,
+      "loss": 0.6727,
+      "step": 10840
+    },
+    {
+      "epoch": 0.5966756563377181,
+      "grad_norm": 0.6582550406455994,
+      "learning_rate": 7.984466408453946e-06,
+      "loss": 0.6794,
+      "step": 10841
+    },
+    {
+      "epoch": 0.5967306951400737,
+      "grad_norm": 0.8680793642997742,
+      "learning_rate": 7.984118618025318e-06,
+      "loss": 0.7999,
+      "step": 10842
+    },
+    {
+      "epoch": 0.5967857339424294,
+      "grad_norm": 0.772777795791626,
+      "learning_rate": 7.983770805168853e-06,
+      "loss": 0.6278,
+      "step": 10843
+    },
+    {
+      "epoch": 0.5968407727447851,
+      "grad_norm": 0.8099700808525085,
+      "learning_rate": 7.983422969887167e-06,
+      "loss": 0.7631,
+      "step": 10844
+    },
+    {
+      "epoch": 0.5968958115471408,
+      "grad_norm": 0.660271406173706,
+      "learning_rate": 7.983075112182871e-06,
+      "loss": 0.7557,
+      "step": 10845
+    },
+    {
+      "epoch": 0.5969508503494964,
+      "grad_norm": 0.7205530405044556,
+      "learning_rate": 7.982727232058582e-06,
+      "loss": 0.8258,
+      "step": 10846
+    },
+    {
+      "epoch": 0.597005889151852,
+      "grad_norm": 0.7925810813903809,
+      "learning_rate": 7.982379329516912e-06,
+      "loss": 0.7534,
+      "step": 10847
+    },
+    {
+      "epoch": 0.5970609279542077,
+      "grad_norm": 0.7255545854568481,
+      "learning_rate": 7.982031404560477e-06,
+      "loss": 0.8394,
+      "step": 10848
+    },
+    {
+      "epoch": 0.5971159667565634,
+      "grad_norm": 0.835394561290741,
+      "learning_rate": 7.981683457191893e-06,
+      "loss": 0.8384,
+      "step": 10849
+    },
+    {
+      "epoch": 0.597171005558919,
+      "grad_norm": 0.6781747937202454,
+      "learning_rate": 7.981335487413775e-06,
+      "loss": 0.8173,
+      "step": 10850
+    },
+    {
+      "epoch": 0.5972260443612747,
+      "grad_norm": 0.8602943420410156,
+      "learning_rate": 7.980987495228737e-06,
+      "loss": 0.8257,
+      "step": 10851
+    },
+    {
+      "epoch": 0.5972810831636304,
+      "grad_norm": 0.7157264947891235,
+      "learning_rate": 7.980639480639394e-06,
+      "loss": 0.7267,
+      "step": 10852
+    },
+    {
+      "epoch": 0.5973361219659861,
+      "grad_norm": 0.7695063352584839,
+      "learning_rate": 7.980291443648364e-06,
+      "loss": 0.7794,
+      "step": 10853
+    },
+    {
+      "epoch": 0.5973911607683416,
+      "grad_norm": 0.723971426486969,
+      "learning_rate": 7.979943384258262e-06,
+      "loss": 0.7761,
+      "step": 10854
+    },
+    {
+      "epoch": 0.5974461995706973,
+      "grad_norm": 0.691722571849823,
+      "learning_rate": 7.979595302471702e-06,
+      "loss": 0.7276,
+      "step": 10855
+    },
+    {
+      "epoch": 0.597501238373053,
+      "grad_norm": 0.7019701600074768,
+      "learning_rate": 7.9792471982913e-06,
+      "loss": 0.7965,
+      "step": 10856
+    },
+    {
+      "epoch": 0.5975562771754087,
+      "grad_norm": 0.6626996994018555,
+      "learning_rate": 7.978899071719675e-06,
+      "loss": 0.7124,
+      "step": 10857
+    },
+    {
+      "epoch": 0.5976113159777643,
+      "grad_norm": 0.6871625781059265,
+      "learning_rate": 7.978550922759443e-06,
+      "loss": 0.7742,
+      "step": 10858
+    },
+    {
+      "epoch": 0.59766635478012,
+      "grad_norm": 0.7153579592704773,
+      "learning_rate": 7.978202751413217e-06,
+      "loss": 0.7852,
+      "step": 10859
+    },
+    {
+      "epoch": 0.5977213935824757,
+      "grad_norm": 0.6891841292381287,
+      "learning_rate": 7.977854557683619e-06,
+      "loss": 0.7873,
+      "step": 10860
+    },
+    {
+      "epoch": 0.5977764323848314,
+      "grad_norm": 0.6864004731178284,
+      "learning_rate": 7.977506341573262e-06,
+      "loss": 0.7223,
+      "step": 10861
+    },
+    {
+      "epoch": 0.5978314711871869,
+      "grad_norm": 0.7163059115409851,
+      "learning_rate": 7.977158103084764e-06,
+      "loss": 0.679,
+      "step": 10862
+    },
+    {
+      "epoch": 0.5978865099895426,
+      "grad_norm": 0.6727336049079895,
+      "learning_rate": 7.976809842220742e-06,
+      "loss": 0.7148,
+      "step": 10863
+    },
+    {
+      "epoch": 0.5979415487918983,
+      "grad_norm": 0.672960638999939,
+      "learning_rate": 7.976461558983814e-06,
+      "loss": 0.7263,
+      "step": 10864
+    },
+    {
+      "epoch": 0.597996587594254,
+      "grad_norm": 0.9124444127082825,
+      "learning_rate": 7.976113253376601e-06,
+      "loss": 0.6876,
+      "step": 10865
+    },
+    {
+      "epoch": 0.5980516263966096,
+      "grad_norm": 0.6415041089057922,
+      "learning_rate": 7.975764925401715e-06,
+      "loss": 0.6655,
+      "step": 10866
+    },
+    {
+      "epoch": 0.5981066651989653,
+      "grad_norm": 0.7342595458030701,
+      "learning_rate": 7.975416575061776e-06,
+      "loss": 0.7753,
+      "step": 10867
+    },
+    {
+      "epoch": 0.598161704001321,
+      "grad_norm": 0.7161775231361389,
+      "learning_rate": 7.975068202359402e-06,
+      "loss": 0.7525,
+      "step": 10868
+    },
+    {
+      "epoch": 0.5982167428036765,
+      "grad_norm": 0.7087578773498535,
+      "learning_rate": 7.974719807297212e-06,
+      "loss": 0.7196,
+      "step": 10869
+    },
+    {
+      "epoch": 0.5982717816060322,
+      "grad_norm": 0.6472536325454712,
+      "learning_rate": 7.974371389877826e-06,
+      "loss": 0.6837,
+      "step": 10870
+    },
+    {
+      "epoch": 0.5983268204083879,
+      "grad_norm": 0.6625581383705139,
+      "learning_rate": 7.97402295010386e-06,
+      "loss": 0.6379,
+      "step": 10871
+    },
+    {
+      "epoch": 0.5983818592107436,
+      "grad_norm": 0.7621071934700012,
+      "learning_rate": 7.973674487977934e-06,
+      "loss": 0.8291,
+      "step": 10872
+    },
+    {
+      "epoch": 0.5984368980130992,
+      "grad_norm": 0.693394660949707,
+      "learning_rate": 7.973326003502666e-06,
+      "loss": 0.7677,
+      "step": 10873
+    },
+    {
+      "epoch": 0.5984919368154549,
+      "grad_norm": 0.6393985152244568,
+      "learning_rate": 7.972977496680674e-06,
+      "loss": 0.7058,
+      "step": 10874
+    },
+    {
+      "epoch": 0.5985469756178106,
+      "grad_norm": 0.7101462483406067,
+      "learning_rate": 7.972628967514582e-06,
+      "loss": 0.7396,
+      "step": 10875
+    },
+    {
+      "epoch": 0.5986020144201663,
+      "grad_norm": 0.8131522536277771,
+      "learning_rate": 7.972280416007003e-06,
+      "loss": 0.8461,
+      "step": 10876
+    },
+    {
+      "epoch": 0.5986570532225218,
+      "grad_norm": 0.7186655402183533,
+      "learning_rate": 7.971931842160564e-06,
+      "loss": 0.7721,
+      "step": 10877
+    },
+    {
+      "epoch": 0.5987120920248775,
+      "grad_norm": 0.7520855069160461,
+      "learning_rate": 7.971583245977877e-06,
+      "loss": 0.7733,
+      "step": 10878
+    },
+    {
+      "epoch": 0.5987671308272332,
+      "grad_norm": 0.6548848748207092,
+      "learning_rate": 7.971234627461569e-06,
+      "loss": 0.6555,
+      "step": 10879
+    },
+    {
+      "epoch": 0.5988221696295889,
+      "grad_norm": 0.7341775894165039,
+      "learning_rate": 7.970885986614254e-06,
+      "loss": 0.8292,
+      "step": 10880
+    },
+    {
+      "epoch": 0.5988772084319445,
+      "grad_norm": 0.7126352190971375,
+      "learning_rate": 7.970537323438556e-06,
+      "loss": 0.7704,
+      "step": 10881
+    },
+    {
+      "epoch": 0.5989322472343002,
+      "grad_norm": 0.7291527390480042,
+      "learning_rate": 7.970188637937097e-06,
+      "loss": 0.8175,
+      "step": 10882
+    },
+    {
+      "epoch": 0.5989872860366559,
+      "grad_norm": 0.682767927646637,
+      "learning_rate": 7.969839930112493e-06,
+      "loss": 0.8187,
+      "step": 10883
+    },
+    {
+      "epoch": 0.5990423248390115,
+      "grad_norm": 0.7820014953613281,
+      "learning_rate": 7.969491199967368e-06,
+      "loss": 0.7949,
+      "step": 10884
+    },
+    {
+      "epoch": 0.5990973636413671,
+      "grad_norm": 0.7257336974143982,
+      "learning_rate": 7.969142447504341e-06,
+      "loss": 0.8461,
+      "step": 10885
+    },
+    {
+      "epoch": 0.5991524024437228,
+      "grad_norm": 0.6813532114028931,
+      "learning_rate": 7.968793672726033e-06,
+      "loss": 0.7889,
+      "step": 10886
+    },
+    {
+      "epoch": 0.5992074412460785,
+      "grad_norm": 0.6868439316749573,
+      "learning_rate": 7.96844487563507e-06,
+      "loss": 0.7268,
+      "step": 10887
+    },
+    {
+      "epoch": 0.5992624800484342,
+      "grad_norm": 0.6547278761863708,
+      "learning_rate": 7.968096056234067e-06,
+      "loss": 0.7026,
+      "step": 10888
+    },
+    {
+      "epoch": 0.5993175188507898,
+      "grad_norm": 0.6704558730125427,
+      "learning_rate": 7.96774721452565e-06,
+      "loss": 0.6994,
+      "step": 10889
+    },
+    {
+      "epoch": 0.5993725576531455,
+      "grad_norm": 0.7134065628051758,
+      "learning_rate": 7.967398350512439e-06,
+      "loss": 0.7728,
+      "step": 10890
+    },
+    {
+      "epoch": 0.5994275964555011,
+      "grad_norm": 0.751265823841095,
+      "learning_rate": 7.967049464197056e-06,
+      "loss": 0.8421,
+      "step": 10891
+    },
+    {
+      "epoch": 0.5994826352578568,
+      "grad_norm": 0.8558571934700012,
+      "learning_rate": 7.966700555582125e-06,
+      "loss": 0.9144,
+      "step": 10892
+    },
+    {
+      "epoch": 0.5995376740602124,
+      "grad_norm": 0.8338084816932678,
+      "learning_rate": 7.966351624670263e-06,
+      "loss": 0.7502,
+      "step": 10893
+    },
+    {
+      "epoch": 0.5995927128625681,
+      "grad_norm": 0.7017131447792053,
+      "learning_rate": 7.9660026714641e-06,
+      "loss": 0.7778,
+      "step": 10894
+    },
+    {
+      "epoch": 0.5996477516649238,
+      "grad_norm": 0.7176111340522766,
+      "learning_rate": 7.965653695966253e-06,
+      "loss": 0.8478,
+      "step": 10895
+    },
+    {
+      "epoch": 0.5997027904672795,
+      "grad_norm": 0.7026060819625854,
+      "learning_rate": 7.965304698179349e-06,
+      "loss": 0.7111,
+      "step": 10896
+    },
+    {
+      "epoch": 0.5997578292696351,
+      "grad_norm": 0.6383810639381409,
+      "learning_rate": 7.964955678106005e-06,
+      "loss": 0.6429,
+      "step": 10897
+    },
+    {
+      "epoch": 0.5998128680719907,
+      "grad_norm": 0.8024059534072876,
+      "learning_rate": 7.96460663574885e-06,
+      "loss": 0.7308,
+      "step": 10898
+    },
+    {
+      "epoch": 0.5998679068743464,
+      "grad_norm": 0.7378466725349426,
+      "learning_rate": 7.964257571110504e-06,
+      "loss": 0.7593,
+      "step": 10899
+    },
+    {
+      "epoch": 0.5999229456767021,
+      "grad_norm": 0.7089043855667114,
+      "learning_rate": 7.963908484193593e-06,
+      "loss": 0.6862,
+      "step": 10900
+    },
+    {
+      "epoch": 0.5999779844790577,
+      "grad_norm": 0.765295684337616,
+      "learning_rate": 7.963559375000738e-06,
+      "loss": 0.6759,
+      "step": 10901
+    },
+    {
+      "epoch": 0.6000330232814134,
+      "grad_norm": 0.7040783166885376,
+      "learning_rate": 7.963210243534565e-06,
+      "loss": 0.7754,
+      "step": 10902
+    },
+    {
+      "epoch": 0.6000880620837691,
+      "grad_norm": 0.8593736886978149,
+      "learning_rate": 7.962861089797698e-06,
+      "loss": 0.8765,
+      "step": 10903
+    },
+    {
+      "epoch": 0.6001431008861248,
+      "grad_norm": 0.6613926291465759,
+      "learning_rate": 7.962511913792758e-06,
+      "loss": 0.6697,
+      "step": 10904
+    },
+    {
+      "epoch": 0.6001981396884803,
+      "grad_norm": 0.6369597911834717,
+      "learning_rate": 7.962162715522372e-06,
+      "loss": 0.7145,
+      "step": 10905
+    },
+    {
+      "epoch": 0.600253178490836,
+      "grad_norm": 1.1790162324905396,
+      "learning_rate": 7.961813494989164e-06,
+      "loss": 0.8067,
+      "step": 10906
+    },
+    {
+      "epoch": 0.6003082172931917,
+      "grad_norm": 0.7548268437385559,
+      "learning_rate": 7.961464252195759e-06,
+      "loss": 0.7936,
+      "step": 10907
+    },
+    {
+      "epoch": 0.6003632560955474,
+      "grad_norm": 0.6204384565353394,
+      "learning_rate": 7.961114987144781e-06,
+      "loss": 0.6374,
+      "step": 10908
     }
   ],
   "logging_steps": 1,
@@ -70019,7 +76382,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2.9507661855341937e+19,
+  "total_flos": 3.219017656946393e+19,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null