{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 1004,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00796812749003984,
      "grad_norm": 0.364986777305603,
      "learning_rate": 1.9607843137254904e-07,
      "loss": 2.1478986740112305,
      "step": 2
    },
    {
      "epoch": 0.01593625498007968,
      "grad_norm": 0.28405916690826416,
      "learning_rate": 5.882352941176471e-07,
      "loss": 2.097151756286621,
      "step": 4
    },
    {
      "epoch": 0.02390438247011952,
      "grad_norm": 1.210844874382019,
      "learning_rate": 9.80392156862745e-07,
      "loss": 2.281658172607422,
      "step": 6
    },
    {
      "epoch": 0.03187250996015936,
      "grad_norm": 0.330355167388916,
      "learning_rate": 1.3725490196078434e-06,
      "loss": 1.901322364807129,
      "step": 8
    },
    {
      "epoch": 0.0398406374501992,
      "grad_norm": 0.7469574213027954,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 1.479136347770691,
      "step": 10
    },
    {
      "epoch": 0.04780876494023904,
      "grad_norm": 0.16489814221858978,
      "learning_rate": 2.1568627450980393e-06,
      "loss": 1.7289620637893677,
      "step": 12
    },
    {
      "epoch": 0.055776892430278883,
      "grad_norm": 0.19304245710372925,
      "learning_rate": 2.549019607843137e-06,
      "loss": 2.3416497707366943,
      "step": 14
    },
    {
      "epoch": 0.06374501992031872,
      "grad_norm": 0.9900257587432861,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 2.545485019683838,
      "step": 16
    },
    {
      "epoch": 0.07171314741035857,
      "grad_norm": 0.3923874497413635,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.6745673418045044,
      "step": 18
    },
    {
      "epoch": 0.0796812749003984,
      "grad_norm": 0.190057635307312,
      "learning_rate": 3.7254901960784316e-06,
      "loss": 1.6109886169433594,
      "step": 20
    },
    {
      "epoch": 0.08764940239043825,
      "grad_norm": 0.38564035296440125,
      "learning_rate": 4.11764705882353e-06,
      "loss": 1.2376759052276611,
      "step": 22
    },
    {
      "epoch": 0.09561752988047809,
      "grad_norm": 0.7834727764129639,
      "learning_rate": 4.509803921568628e-06,
      "loss": 1.3991103172302246,
      "step": 24
    },
    {
      "epoch": 0.10358565737051793,
      "grad_norm": 0.1316996067762375,
      "learning_rate": 4.901960784313726e-06,
      "loss": 1.5003830194473267,
      "step": 26
    },
    {
      "epoch": 0.11155378486055777,
      "grad_norm": 0.29018765687942505,
      "learning_rate": 5.294117647058824e-06,
      "loss": 1.6468145847320557,
      "step": 28
    },
    {
      "epoch": 0.11952191235059761,
      "grad_norm": 0.3622463643550873,
      "learning_rate": 5.686274509803922e-06,
      "loss": 1.3556547164916992,
      "step": 30
    },
    {
      "epoch": 0.12749003984063745,
      "grad_norm": 0.17528486251831055,
      "learning_rate": 6.07843137254902e-06,
      "loss": 1.175823450088501,
      "step": 32
    },
    {
      "epoch": 0.13545816733067728,
      "grad_norm": 0.39535778760910034,
      "learning_rate": 6.470588235294119e-06,
      "loss": 1.3078317642211914,
      "step": 34
    },
    {
      "epoch": 0.14342629482071714,
      "grad_norm": 0.18487055599689484,
      "learning_rate": 6.862745098039216e-06,
      "loss": 1.4364827871322632,
      "step": 36
    },
    {
      "epoch": 0.15139442231075698,
      "grad_norm": 0.3245979845523834,
      "learning_rate": 7.2549019607843145e-06,
      "loss": 1.264662265777588,
      "step": 38
    },
    {
      "epoch": 0.1593625498007968,
      "grad_norm": 0.16470961272716522,
      "learning_rate": 7.647058823529411e-06,
      "loss": 1.3487387895584106,
      "step": 40
    },
    {
      "epoch": 0.16733067729083664,
      "grad_norm": 0.25827449560165405,
      "learning_rate": 8.03921568627451e-06,
      "loss": 0.7240933179855347,
      "step": 42
    },
    {
      "epoch": 0.1752988047808765,
      "grad_norm": 0.13965268433094025,
      "learning_rate": 8.43137254901961e-06,
      "loss": 0.9613214731216431,
      "step": 44
    },
    {
      "epoch": 0.18326693227091634,
      "grad_norm": 0.22857411205768585,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.9437090754508972,
      "step": 46
    },
    {
      "epoch": 0.19123505976095617,
      "grad_norm": 0.37201637029647827,
      "learning_rate": 9.215686274509804e-06,
      "loss": 1.1860681772232056,
      "step": 48
    },
    {
      "epoch": 0.199203187250996,
      "grad_norm": 0.21413810551166534,
      "learning_rate": 9.607843137254903e-06,
      "loss": 1.3915151357650757,
      "step": 50
    },
    {
      "epoch": 0.20717131474103587,
      "grad_norm": 0.3054081201553345,
      "learning_rate": 1e-05,
      "loss": 1.338213562965393,
      "step": 52
    },
    {
      "epoch": 0.2151394422310757,
      "grad_norm": 0.577069878578186,
      "learning_rate": 9.999902196391087e-06,
      "loss": 0.9658082127571106,
      "step": 54
    },
    {
      "epoch": 0.22310756972111553,
      "grad_norm": 0.24089600145816803,
      "learning_rate": 9.999608789815699e-06,
      "loss": 1.5122920274734497,
      "step": 56
    },
    {
      "epoch": 0.23107569721115537,
      "grad_norm": 0.17221805453300476,
      "learning_rate": 9.999119793027713e-06,
      "loss": 1.3596100807189941,
      "step": 58
    },
    {
      "epoch": 0.23904382470119523,
      "grad_norm": 0.29021480679512024,
      "learning_rate": 9.998435227282975e-06,
      "loss": 0.7721037864685059,
      "step": 60
    },
    {
      "epoch": 0.24701195219123506,
      "grad_norm": 0.1621028184890747,
      "learning_rate": 9.997555122338372e-06,
      "loss": 1.3582539558410645,
      "step": 62
    },
    {
      "epoch": 0.2549800796812749,
      "grad_norm": 0.22715942561626434,
      "learning_rate": 9.996479516450544e-06,
      "loss": 1.0979129076004028,
      "step": 64
    },
    {
      "epoch": 0.26294820717131473,
      "grad_norm": 0.11652333289384842,
      "learning_rate": 9.99520845637422e-06,
      "loss": 1.285919189453125,
      "step": 66
    },
    {
      "epoch": 0.27091633466135456,
      "grad_norm": 0.2143312394618988,
      "learning_rate": 9.993741997360185e-06,
      "loss": 0.760143518447876,
      "step": 68
    },
    {
      "epoch": 0.2788844621513944,
      "grad_norm": 0.2805745005607605,
      "learning_rate": 9.992080203152872e-06,
      "loss": 1.6005887985229492,
      "step": 70
    },
    {
      "epoch": 0.2868525896414343,
      "grad_norm": 0.23261545598506927,
      "learning_rate": 9.990223145987608e-06,
      "loss": 0.7550407648086548,
      "step": 72
    },
    {
      "epoch": 0.2948207171314741,
      "grad_norm": 0.3619435131549835,
      "learning_rate": 9.98817090658745e-06,
      "loss": 1.1772500276565552,
      "step": 74
    },
    {
      "epoch": 0.30278884462151395,
      "grad_norm": 0.1638466864824295,
      "learning_rate": 9.985923574159696e-06,
      "loss": 1.4381229877471924,
      "step": 76
    },
    {
      "epoch": 0.3107569721115538,
      "grad_norm": 1.124913215637207,
      "learning_rate": 9.983481246392003e-06,
      "loss": 0.9343405961990356,
      "step": 78
    },
    {
      "epoch": 0.3187250996015936,
      "grad_norm": 0.14637482166290283,
      "learning_rate": 9.980844029448135e-06,
      "loss": 0.8086846470832825,
      "step": 80
    },
    {
      "epoch": 0.32669322709163345,
      "grad_norm": 0.26540300250053406,
      "learning_rate": 9.97801203796335e-06,
      "loss": 1.2997255325317383,
      "step": 82
    },
    {
      "epoch": 0.3346613545816733,
      "grad_norm": 0.21794630587100983,
      "learning_rate": 9.974985395039423e-06,
      "loss": 0.8241325616836548,
      "step": 84
    },
    {
      "epoch": 0.3426294820717131,
      "grad_norm": 0.23181264102458954,
      "learning_rate": 9.971764232239287e-06,
      "loss": 1.3146854639053345,
      "step": 86
    },
    {
      "epoch": 0.350597609561753,
      "grad_norm": 0.42058876156806946,
      "learning_rate": 9.968348689581315e-06,
      "loss": 0.8865960836410522,
      "step": 88
    },
    {
      "epoch": 0.35856573705179284,
      "grad_norm": 0.16636328399181366,
      "learning_rate": 9.964738915533242e-06,
      "loss": 1.329420566558838,
      "step": 90
    },
    {
      "epoch": 0.3665338645418327,
      "grad_norm": 0.21945226192474365,
      "learning_rate": 9.960935067005706e-06,
      "loss": 1.1200248003005981,
      "step": 92
    },
    {
      "epoch": 0.3745019920318725,
      "grad_norm": 0.14254772663116455,
      "learning_rate": 9.956937309345418e-06,
      "loss": 1.2416845560073853,
      "step": 94
    },
    {
      "epoch": 0.38247011952191234,
      "grad_norm": 0.2565023601055145,
      "learning_rate": 9.952745816327998e-06,
      "loss": 1.1250306367874146,
      "step": 96
    },
    {
      "epoch": 0.3904382470119522,
      "grad_norm": 0.12139222770929337,
      "learning_rate": 9.94836077015039e-06,
      "loss": 1.2636781930923462,
      "step": 98
    },
    {
      "epoch": 0.398406374501992,
      "grad_norm": 0.1594734638929367,
      "learning_rate": 9.943782361422975e-06,
      "loss": 0.8420242071151733,
      "step": 100
    },
    {
      "epoch": 0.4063745019920319,
      "grad_norm": 0.141156867146492,
      "learning_rate": 9.939010789161259e-06,
      "loss": 1.2728060483932495,
      "step": 102
    },
    {
      "epoch": 0.41434262948207173,
      "grad_norm": 0.138778418302536,
      "learning_rate": 9.934046260777239e-06,
      "loss": 0.7574232220649719,
      "step": 104
    },
    {
      "epoch": 0.42231075697211157,
      "grad_norm": 0.4497663080692291,
      "learning_rate": 9.928888992070373e-06,
      "loss": 0.7535025477409363,
      "step": 106
    },
    {
      "epoch": 0.4302788844621514,
      "grad_norm": 0.15345805883407593,
      "learning_rate": 9.923539207218219e-06,
      "loss": 1.0066508054733276,
      "step": 108
    },
    {
      "epoch": 0.43824701195219123,
      "grad_norm": 0.10952593386173248,
      "learning_rate": 9.91799713876667e-06,
      "loss": 1.2843388319015503,
      "step": 110
    },
    {
      "epoch": 0.44621513944223107,
      "grad_norm": 0.4354335069656372,
      "learning_rate": 9.912263027619856e-06,
      "loss": 0.8039376139640808,
      "step": 112
    },
    {
      "epoch": 0.4541832669322709,
      "grad_norm": 1.0137566328048706,
      "learning_rate": 9.906337123029675e-06,
      "loss": 1.0992188453674316,
      "step": 114
    },
    {
      "epoch": 0.46215139442231074,
      "grad_norm": 0.16643138229846954,
      "learning_rate": 9.90021968258495e-06,
      "loss": 1.2867761850357056,
      "step": 116
    },
    {
      "epoch": 0.4701195219123506,
      "grad_norm": 0.11957172304391861,
      "learning_rate": 9.893910972200238e-06,
      "loss": 1.355262041091919,
      "step": 118
    },
    {
      "epoch": 0.47808764940239046,
      "grad_norm": 0.13334529101848602,
      "learning_rate": 9.887411266104268e-06,
      "loss": 1.2227801084518433,
      "step": 120
    },
    {
      "epoch": 0.4860557768924303,
      "grad_norm": 0.1560998260974884,
      "learning_rate": 9.880720846828026e-06,
      "loss": 0.9436216950416565,
      "step": 122
    },
    {
      "epoch": 0.4940239043824701,
      "grad_norm": 0.6945304870605469,
      "learning_rate": 9.873840005192466e-06,
      "loss": 1.0742814540863037,
      "step": 124
    },
    {
      "epoch": 0.50199203187251,
      "grad_norm": 0.20763267576694489,
      "learning_rate": 9.866769040295875e-06,
      "loss": 0.9257575869560242,
      "step": 126
    },
    {
      "epoch": 0.5099601593625498,
      "grad_norm": 0.14976738393306732,
      "learning_rate": 9.859508259500869e-06,
      "loss": 1.3002715110778809,
      "step": 128
    },
    {
      "epoch": 0.5179282868525896,
      "grad_norm": 0.12004569917917252,
      "learning_rate": 9.852057978421033e-06,
      "loss": 0.9138576984405518,
      "step": 130
    },
    {
      "epoch": 0.5258964143426295,
      "grad_norm": 0.24855820834636688,
      "learning_rate": 9.844418520907198e-06,
      "loss": 0.8074920773506165,
      "step": 132
    },
    {
      "epoch": 0.5338645418326693,
      "grad_norm": 0.126007542014122,
      "learning_rate": 9.836590219033375e-06,
      "loss": 0.8707377910614014,
      "step": 134
    },
    {
      "epoch": 0.5418326693227091,
      "grad_norm": 0.16694070398807526,
      "learning_rate": 9.828573413082306e-06,
      "loss": 1.2868237495422363,
      "step": 136
    },
    {
      "epoch": 0.549800796812749,
      "grad_norm": 0.12794046103954315,
      "learning_rate": 9.820368451530681e-06,
      "loss": 0.9321343898773193,
      "step": 138
    },
    {
      "epoch": 0.5577689243027888,
      "grad_norm": 0.30766522884368896,
      "learning_rate": 9.811975691033992e-06,
      "loss": 1.0828709602355957,
      "step": 140
    },
    {
      "epoch": 0.5657370517928287,
      "grad_norm": 0.15038612484931946,
      "learning_rate": 9.80339549641102e-06,
      "loss": 0.9955046772956848,
      "step": 142
    },
    {
      "epoch": 0.5737051792828686,
      "grad_norm": 0.11382584273815155,
      "learning_rate": 9.794628240627992e-06,
      "loss": 1.2232928276062012,
      "step": 144
    },
    {
      "epoch": 0.5816733067729084,
      "grad_norm": 0.2687312364578247,
      "learning_rate": 9.785674304782349e-06,
      "loss": 1.2767001390457153,
      "step": 146
    },
    {
      "epoch": 0.5896414342629482,
      "grad_norm": 0.2042328417301178,
      "learning_rate": 9.776534078086202e-06,
      "loss": 0.9674968719482422,
      "step": 148
    },
    {
      "epoch": 0.5976095617529881,
      "grad_norm": 0.24133750796318054,
      "learning_rate": 9.767207957849396e-06,
      "loss": 0.9319218397140503,
      "step": 150
    },
    {
      "epoch": 0.6055776892430279,
      "grad_norm": 1.62368905544281,
      "learning_rate": 9.757696349462252e-06,
      "loss": 0.9243829846382141,
      "step": 152
    },
    {
      "epoch": 0.6135458167330677,
      "grad_norm": 0.366428405046463,
      "learning_rate": 9.747999666377934e-06,
      "loss": 1.111814022064209,
      "step": 154
    },
    {
      "epoch": 0.6215139442231076,
      "grad_norm": 0.40344610810279846,
      "learning_rate": 9.738118330094491e-06,
      "loss": 0.9363324046134949,
      "step": 156
    },
    {
      "epoch": 0.6294820717131474,
      "grad_norm": 0.22840072214603424,
      "learning_rate": 9.728052770136518e-06,
      "loss": 1.1430773735046387,
      "step": 158
    },
    {
      "epoch": 0.6374501992031872,
      "grad_norm": 0.15481047332286835,
      "learning_rate": 9.717803424036501e-06,
      "loss": 1.0062944889068604,
      "step": 160
    },
    {
      "epoch": 0.6454183266932271,
      "grad_norm": 0.08594097942113876,
      "learning_rate": 9.707370737315793e-06,
      "loss": 0.7411895990371704,
      "step": 162
    },
    {
      "epoch": 0.6533864541832669,
      "grad_norm": 0.2677382528781891,
      "learning_rate": 9.69675516346524e-06,
      "loss": 1.17550528049469,
      "step": 164
    },
    {
      "epoch": 0.6613545816733067,
      "grad_norm": 0.565375030040741,
      "learning_rate": 9.685957163925479e-06,
      "loss": 0.8197770118713379,
      "step": 166
    },
    {
      "epoch": 0.6693227091633466,
      "grad_norm": 0.10801292955875397,
      "learning_rate": 9.67497720806688e-06,
      "loss": 1.2339354753494263,
      "step": 168
    },
    {
      "epoch": 0.6772908366533864,
      "grad_norm": 0.5754889249801636,
      "learning_rate": 9.663815773169131e-06,
      "loss": 0.8984463810920715,
      "step": 170
    },
    {
      "epoch": 0.6852589641434262,
      "grad_norm": 0.1492973268032074,
      "learning_rate": 9.652473344400507e-06,
      "loss": 0.8594431281089783,
      "step": 172
    },
    {
      "epoch": 0.6932270916334662,
      "grad_norm": 0.12211178243160248,
      "learning_rate": 9.640950414796772e-06,
      "loss": 1.2287001609802246,
      "step": 174
    },
    {
      "epoch": 0.701195219123506,
      "grad_norm": 0.2570830285549164,
      "learning_rate": 9.629247485239747e-06,
      "loss": 0.713386595249176,
      "step": 176
    },
    {
      "epoch": 0.7091633466135459,
      "grad_norm": 0.32329094409942627,
      "learning_rate": 9.617365064435544e-06,
      "loss": 0.9358075857162476,
      "step": 178
    },
    {
      "epoch": 0.7171314741035857,
      "grad_norm": 0.2823147475719452,
      "learning_rate": 9.605303668892443e-06,
      "loss": 1.0932412147521973,
      "step": 180
    },
    {
      "epoch": 0.7250996015936255,
      "grad_norm": 0.04298524931073189,
      "learning_rate": 9.593063822898451e-06,
      "loss": 0.4358002841472626,
      "step": 182
    },
    {
      "epoch": 0.7330677290836654,
      "grad_norm": 0.1834041327238083,
      "learning_rate": 9.580646058498505e-06,
      "loss": 1.3084431886672974,
      "step": 184
    },
    {
      "epoch": 0.7410358565737052,
      "grad_norm": 0.22123558819293976,
      "learning_rate": 9.56805091547135e-06,
      "loss": 1.2738910913467407,
      "step": 186
    },
    {
      "epoch": 0.749003984063745,
      "grad_norm": 0.15455125272274017,
      "learning_rate": 9.55527894130607e-06,
      "loss": 1.0830506086349487,
      "step": 188
    },
    {
      "epoch": 0.7569721115537849,
      "grad_norm": 0.2986137866973877,
      "learning_rate": 9.542330691178296e-06,
      "loss": 0.8603120446205139,
      "step": 190
    },
    {
      "epoch": 0.7649402390438247,
      "grad_norm": 0.40803152322769165,
      "learning_rate": 9.529206727926071e-06,
      "loss": 1.2306050062179565,
      "step": 192
    },
    {
      "epoch": 0.7729083665338645,
      "grad_norm": 0.17157235741615295,
      "learning_rate": 9.515907622025376e-06,
      "loss": 0.8791062831878662,
      "step": 194
    },
    {
      "epoch": 0.7808764940239044,
      "grad_norm": 0.27570194005966187,
      "learning_rate": 9.502433951565347e-06,
      "loss": 0.5943114161491394,
      "step": 196
    },
    {
      "epoch": 0.7888446215139442,
      "grad_norm": 0.22980396449565887,
      "learning_rate": 9.48878630222314e-06,
      "loss": 0.796610414981842,
      "step": 198
    },
    {
      "epoch": 0.796812749003984,
      "grad_norm": 0.26152628660202026,
      "learning_rate": 9.474965267238469e-06,
      "loss": 1.005728006362915,
      "step": 200
    },
    {
      "epoch": 0.8047808764940239,
      "grad_norm": 0.09973812848329544,
      "learning_rate": 9.460971447387824e-06,
      "loss": 0.9312565922737122,
      "step": 202
    },
    {
      "epoch": 0.8127490039840638,
      "grad_norm": 0.497194766998291,
      "learning_rate": 9.446805450958353e-06,
      "loss": 1.5533764362335205,
      "step": 204
    },
    {
      "epoch": 0.8207171314741036,
      "grad_norm": 0.16945381462574005,
      "learning_rate": 9.43246789372142e-06,
      "loss": 1.2373756170272827,
      "step": 206
    },
    {
      "epoch": 0.8286852589641435,
      "grad_norm": 0.1157103106379509,
      "learning_rate": 9.417959398905846e-06,
      "loss": 0.5758540630340576,
      "step": 208
    },
    {
      "epoch": 0.8366533864541833,
      "grad_norm": 0.14730894565582275,
      "learning_rate": 9.40328059717081e-06,
      "loss": 1.2672688961029053,
      "step": 210
    },
    {
      "epoch": 0.8446215139442231,
      "grad_norm": 0.12053020298480988,
      "learning_rate": 9.388432126578438e-06,
      "loss": 1.2181390523910522,
      "step": 212
    },
    {
      "epoch": 0.852589641434263,
      "grad_norm": 0.15825557708740234,
      "learning_rate": 9.373414632566066e-06,
      "loss": 0.9476991891860962,
      "step": 214
    },
    {
      "epoch": 0.8605577689243028,
      "grad_norm": 0.31758689880371094,
      "learning_rate": 9.35822876791819e-06,
      "loss": 1.424242377281189,
      "step": 216
    },
    {
      "epoch": 0.8685258964143426,
      "grad_norm": 0.17400968074798584,
      "learning_rate": 9.342875192738083e-06,
      "loss": 0.8652182817459106,
      "step": 218
    },
    {
      "epoch": 0.8764940239043825,
      "grad_norm": 0.1382993459701538,
      "learning_rate": 9.327354574419109e-06,
      "loss": 1.2354328632354736,
      "step": 220
    },
    {
      "epoch": 0.8844621513944223,
      "grad_norm": 0.21812206506729126,
      "learning_rate": 9.311667587615701e-06,
      "loss": 1.4844191074371338,
      "step": 222
    },
    {
      "epoch": 0.8924302788844621,
      "grad_norm": 0.48735082149505615,
      "learning_rate": 9.29581491421405e-06,
      "loss": 0.708640456199646,
      "step": 224
    },
    {
      "epoch": 0.900398406374502,
      "grad_norm": 0.1356334537267685,
      "learning_rate": 9.279797243302453e-06,
      "loss": 0.9836586713790894,
      "step": 226
    },
    {
      "epoch": 0.9083665338645418,
      "grad_norm": 0.18977560102939606,
      "learning_rate": 9.263615271141362e-06,
      "loss": 1.0754379034042358,
      "step": 228
    },
    {
      "epoch": 0.9163346613545816,
      "grad_norm": 0.18321223556995392,
      "learning_rate": 9.247269701133125e-06,
      "loss": 1.1610815525054932,
      "step": 230
    },
    {
      "epoch": 0.9243027888446215,
      "grad_norm": 0.18827883899211884,
      "learning_rate": 9.230761243791401e-06,
      "loss": 1.3425302505493164,
      "step": 232
    },
    {
      "epoch": 0.9322709163346613,
      "grad_norm": 0.15381808578968048,
      "learning_rate": 9.214090616710282e-06,
      "loss": 0.7759059071540833,
      "step": 234
    },
    {
      "epoch": 0.9402390438247012,
      "grad_norm": 0.13686630129814148,
      "learning_rate": 9.197258544533098e-06,
      "loss": 0.7925941944122314,
      "step": 236
    },
    {
      "epoch": 0.9482071713147411,
      "grad_norm": 0.20447702705860138,
      "learning_rate": 9.180265758920916e-06,
      "loss": 0.4479016363620758,
      "step": 238
    },
    {
      "epoch": 0.9561752988047809,
      "grad_norm": 0.4583527147769928,
      "learning_rate": 9.163112998520741e-06,
      "loss": 0.8426061272621155,
      "step": 240
    },
    {
      "epoch": 0.9641434262948207,
      "grad_norm": 0.20723827183246613,
      "learning_rate": 9.145801008933403e-06,
      "loss": 1.0996769666671753,
      "step": 242
    },
    {
      "epoch": 0.9721115537848606,
      "grad_norm": 0.21747617423534393,
      "learning_rate": 9.128330542681154e-06,
      "loss": 1.2090086936950684,
      "step": 244
    },
    {
      "epoch": 0.9800796812749004,
      "grad_norm": 0.17013338208198547,
      "learning_rate": 9.110702359174944e-06,
      "loss": 0.9386983513832092,
      "step": 246
    },
    {
      "epoch": 0.9880478087649402,
      "grad_norm": 0.1476457267999649,
      "learning_rate": 9.092917224681427e-06,
      "loss": 0.9683389067649841,
      "step": 248
    },
    {
      "epoch": 0.9960159362549801,
      "grad_norm": 0.1514909267425537,
      "learning_rate": 9.074975912289641e-06,
      "loss": 0.9698129892349243,
      "step": 250
    },
    {
      "epoch": 1.00398406374502,
      "grad_norm": 0.29872578382492065,
      "learning_rate": 9.056879201877412e-06,
      "loss": 0.89365553855896,
      "step": 252
    },
    {
      "epoch": 1.0119521912350598,
      "grad_norm": 0.2790847718715668,
      "learning_rate": 9.038627880077441e-06,
      "loss": 0.6684755086898804,
      "step": 254
    },
    {
      "epoch": 1.0199203187250996,
      "grad_norm": 0.1847427636384964,
      "learning_rate": 9.020222740243127e-06,
      "loss": 1.1455042362213135,
      "step": 256
    },
    {
      "epoch": 1.0278884462151394,
      "grad_norm": 0.14955571293830872,
      "learning_rate": 9.001664582414066e-06,
      "loss": 1.1182340383529663,
      "step": 258
    },
    {
      "epoch": 1.0358565737051793,
      "grad_norm": 0.19441288709640503,
      "learning_rate": 8.982954213281291e-06,
      "loss": 1.0341322422027588,
      "step": 260
    },
    {
      "epoch": 1.043824701195219,
      "grad_norm": 0.362791508436203,
      "learning_rate": 8.964092446152183e-06,
      "loss": 0.9678187966346741,
      "step": 262
    },
    {
      "epoch": 1.051792828685259,
      "grad_norm": 0.2039051651954651,
      "learning_rate": 8.945080100915145e-06,
      "loss": 1.048351526260376,
      "step": 264
    },
    {
      "epoch": 1.0597609561752988,
      "grad_norm": 0.4163295030593872,
      "learning_rate": 8.925918004003941e-06,
      "loss": 0.5999967455863953,
      "step": 266
    },
    {
      "epoch": 1.0677290836653386,
      "grad_norm": 0.11615937203168869,
      "learning_rate": 8.90660698836179e-06,
      "loss": 0.8929613828659058,
      "step": 268
    },
    {
      "epoch": 1.0756972111553784,
      "grad_norm": 0.11563047021627426,
      "learning_rate": 8.887147893405145e-06,
      "loss": 0.5231808423995972,
      "step": 270
    },
    {
      "epoch": 1.0836653386454183,
      "grad_norm": 0.15155117213726044,
      "learning_rate": 8.867541564987212e-06,
      "loss": 1.2267755270004272,
      "step": 272
    },
    {
      "epoch": 1.091633466135458,
      "grad_norm": 0.21607372164726257,
      "learning_rate": 8.84778885536118e-06,
      "loss": 0.5920886397361755,
      "step": 274
    },
    {
      "epoch": 1.099601593625498,
      "grad_norm": 0.4365248382091522,
      "learning_rate": 8.827890623143172e-06,
      "loss": 0.8858764171600342,
      "step": 276
    },
    {
      "epoch": 1.1075697211155378,
      "grad_norm": 0.15347373485565186,
      "learning_rate": 8.80784773327494e-06,
      "loss": 0.7853067517280579,
      "step": 278
    },
    {
      "epoch": 1.1155378486055776,
      "grad_norm": 0.19446605443954468,
      "learning_rate": 8.787661056986236e-06,
      "loss": 0.9884380102157593,
      "step": 280
    },
    {
      "epoch": 1.1235059760956174,
      "grad_norm": 0.3765554130077362,
      "learning_rate": 8.767331471756975e-06,
      "loss": 1.1270825862884521,
      "step": 282
    },
    {
      "epoch": 1.1314741035856573,
      "grad_norm": 0.38874736428260803,
      "learning_rate": 8.746859861279068e-06,
      "loss": 1.1766220331192017,
      "step": 284
    },
    {
      "epoch": 1.139442231075697,
      "grad_norm": 0.6246842741966248,
      "learning_rate": 8.726247115418017e-06,
      "loss": 0.23498383164405823,
      "step": 286
    },
    {
      "epoch": 1.1474103585657371,
      "grad_norm": 0.09193812310695648,
      "learning_rate": 8.70549413017424e-06,
      "loss": 0.6870949864387512,
      "step": 288
    },
    {
      "epoch": 1.155378486055777,
      "grad_norm": 0.22841566801071167,
      "learning_rate": 8.684601807644114e-06,
      "loss": 0.8961982131004333,
      "step": 290
    },
    {
      "epoch": 1.1633466135458168,
      "grad_norm": 0.17654278874397278,
      "learning_rate": 8.663571055980769e-06,
      "loss": 0.4762253165245056,
      "step": 292
    },
    {
      "epoch": 1.1713147410358566,
      "grad_norm": 0.11051709949970245,
      "learning_rate": 8.642402789354609e-06,
      "loss": 1.1876903772354126,
      "step": 294
    },
    {
      "epoch": 1.1792828685258965,
      "grad_norm": 0.11673513799905777,
      "learning_rate": 8.62109792791358e-06,
      "loss": 1.2314369678497314,
      "step": 296
    },
    {
      "epoch": 1.1872509960159363,
      "grad_norm": 0.15819886326789856,
      "learning_rate": 8.599657397743163e-06,
      "loss": 1.3261650800704956,
      "step": 298
    },
    {
      "epoch": 1.1952191235059761,
      "grad_norm": 0.3119302988052368,
      "learning_rate": 8.578082130826123e-06,
      "loss": 0.5580907464027405,
      "step": 300
    },
    {
      "epoch": 1.203187250996016,
      "grad_norm": 0.6036977767944336,
      "learning_rate": 8.556373065002006e-06,
      "loss": 0.8679631948471069,
      "step": 302
    },
    {
      "epoch": 1.2111553784860558,
      "grad_norm": 0.14448155462741852,
      "learning_rate": 8.534531143926359e-06,
      "loss": 0.48680248856544495,
      "step": 304
    },
    {
      "epoch": 1.2191235059760956,
      "grad_norm": 0.30012473464012146,
      "learning_rate": 8.512557317029715e-06,
      "loss": 1.1665215492248535,
      "step": 306
    },
    {
      "epoch": 1.2270916334661355,
      "grad_norm": 0.8571121096611023,
      "learning_rate": 8.49045253947633e-06,
      "loss": 0.9532862305641174,
      "step": 308
    },
    {
      "epoch": 1.2350597609561753,
      "grad_norm": 0.2318251132965088,
      "learning_rate": 8.46821777212266e-06,
      "loss": 1.1333225965499878,
      "step": 310
    },
    {
      "epoch": 1.2430278884462151,
      "grad_norm": 0.5253759622573853,
      "learning_rate": 8.445853981475587e-06,
      "loss": 1.035222053527832,
      "step": 312
    },
    {
      "epoch": 1.250996015936255,
      "grad_norm": 0.13817377388477325,
      "learning_rate": 8.423362139650414e-06,
      "loss": 0.8419394493103027,
      "step": 314
    },
    {
      "epoch": 1.2589641434262948,
      "grad_norm": 0.4624038338661194,
      "learning_rate": 8.40074322432861e-06,
      "loss": 0.7096641659736633,
      "step": 316
    },
    {
      "epoch": 1.2669322709163346,
      "grad_norm": 0.25667160749435425,
      "learning_rate": 8.377998218715311e-06,
      "loss": 0.5867374539375305,
      "step": 318
    },
    {
      "epoch": 1.2749003984063745,
      "grad_norm": 0.10003433376550674,
      "learning_rate": 8.355128111496569e-06,
      "loss": 0.4398966431617737,
      "step": 320
    },
    {
      "epoch": 1.2828685258964143,
      "grad_norm": 0.24401697516441345,
      "learning_rate": 8.332133896796399e-06,
      "loss": 0.9452098608016968,
      "step": 322
    },
    {
      "epoch": 1.2908366533864541,
      "grad_norm": 0.5646545886993408,
      "learning_rate": 8.309016574133545e-06,
      "loss": 0.8878511190414429,
      "step": 324
    },
    {
      "epoch": 1.298804780876494,
      "grad_norm": 0.3865255117416382,
      "learning_rate": 8.28577714837805e-06,
      "loss": 1.2442195415496826,
      "step": 326
    },
    {
      "epoch": 1.3067729083665338,
      "grad_norm": 0.15615998208522797,
      "learning_rate": 8.262416629707555e-06,
      "loss": 0.6620932221412659,
      "step": 328
    },
    {
      "epoch": 1.3147410358565736,
      "grad_norm": 0.22632110118865967,
      "learning_rate": 8.238936033563415e-06,
      "loss": 0.8339309096336365,
      "step": 330
    },
    {
      "epoch": 1.3227091633466135,
      "grad_norm": 0.2634675204753876,
      "learning_rate": 8.215336380606533e-06,
      "loss": 0.8840557932853699,
      "step": 332
    },
    {
      "epoch": 1.3306772908366533,
      "grad_norm": 0.1798751801252365,
      "learning_rate": 8.191618696673012e-06,
      "loss": 0.5947738289833069,
      "step": 334
    },
    {
      "epoch": 1.3386454183266931,
      "grad_norm": 0.5607253909111023,
      "learning_rate": 8.167784012729555e-06,
      "loss": 0.8520745635032654,
      "step": 336
    },
    {
      "epoch": 1.3466135458167332,
      "grad_norm": 0.24325330555438995,
      "learning_rate": 8.143833364828654e-06,
      "loss": 1.1230289936065674,
      "step": 338
    },
    {
      "epoch": 1.354581673306773,
      "grad_norm": 0.172824889421463,
      "learning_rate": 8.119767794063553e-06,
      "loss": 1.1187266111373901,
      "step": 340
    },
    {
      "epoch": 1.3625498007968129,
      "grad_norm": 0.09730469435453415,
      "learning_rate": 8.095588346523e-06,
      "loss": 1.1664950847625732,
      "step": 342
    },
    {
      "epoch": 1.3705179282868527,
      "grad_norm": 0.14186157286167145,
      "learning_rate": 8.071296073245759e-06,
      "loss": 1.1263258457183838,
      "step": 344
    },
    {
      "epoch": 1.3784860557768925,
      "grad_norm": 0.14493584632873535,
      "learning_rate": 8.046892030174938e-06,
      "loss": 0.6797537803649902,
      "step": 346
    },
    {
      "epoch": 1.3864541832669324,
      "grad_norm": 0.16807496547698975,
      "learning_rate": 8.022377278112089e-06,
      "loss": 1.1494190692901611,
      "step": 348
    },
    {
      "epoch": 1.3944223107569722,
      "grad_norm": 0.27777084708213806,
      "learning_rate": 7.997752882671088e-06,
      "loss": 0.6691007018089294,
      "step": 350
    },
    {
      "epoch": 1.402390438247012,
      "grad_norm": 0.1443774253129959,
      "learning_rate": 7.97301991423182e-06,
      "loss": 1.1747478246688843,
      "step": 352
    },
    {
      "epoch": 1.4103585657370519,
      "grad_norm": 0.1637537181377411,
      "learning_rate": 7.948179447893652e-06,
      "loss": 1.1217589378356934,
      "step": 354
    },
    {
      "epoch": 1.4183266932270917,
      "grad_norm": 0.7767637968063354,
      "learning_rate": 7.923232563428693e-06,
      "loss": 0.8744209408760071,
      "step": 356
    },
    {
      "epoch": 1.4262948207171315,
      "grad_norm": 0.439284086227417,
      "learning_rate": 7.898180345234875e-06,
      "loss": 1.142197847366333,
      "step": 358
    },
    {
      "epoch": 1.4342629482071714,
      "grad_norm": 0.2838995158672333,
      "learning_rate": 7.873023882288793e-06,
      "loss": 0.5824839472770691,
      "step": 360
    },
    {
      "epoch": 1.4422310756972112,
      "grad_norm": 0.12531805038452148,
      "learning_rate": 7.847764268098387e-06,
      "loss": 0.6964688301086426,
      "step": 362
    },
    {
      "epoch": 1.450199203187251,
      "grad_norm": 0.1316349059343338,
      "learning_rate": 7.822402600655405e-06,
      "loss": 1.0379220247268677,
      "step": 364
    },
    {
      "epoch": 1.4581673306772909,
      "grad_norm": 0.10242468863725662,
      "learning_rate": 7.79693998238767e-06,
      "loss": 1.10782790184021,
      "step": 366
    },
    {
      "epoch": 1.4661354581673307,
      "grad_norm": 0.16310665011405945,
      "learning_rate": 7.771377520111163e-06,
      "loss": 0.5999088287353516,
      "step": 368
    },
    {
      "epoch": 1.4741035856573705,
      "grad_norm": 0.15483590960502625,
      "learning_rate": 7.745716324981912e-06,
      "loss": 0.6717012524604797,
      "step": 370
    },
    {
      "epoch": 1.4820717131474104,
      "grad_norm": 0.2711755633354187,
      "learning_rate": 7.71995751244769e-06,
      "loss": 1.223280668258667,
      "step": 372
    },
    {
      "epoch": 1.4900398406374502,
      "grad_norm": 0.12904302775859833,
      "learning_rate": 7.694102202199536e-06,
      "loss": 1.114804744720459,
      "step": 374
    },
    {
      "epoch": 1.49800796812749,
      "grad_norm": 0.13068151473999023,
      "learning_rate": 7.66815151812307e-06,
      "loss": 1.1189380884170532,
      "step": 376
    },
    {
      "epoch": 1.5059760956175299,
      "grad_norm": 0.20732992887496948,
      "learning_rate": 7.642106588249651e-06,
      "loss": 0.8951546549797058,
      "step": 378
    },
    {
      "epoch": 1.5139442231075697,
      "grad_norm": 0.13041703402996063,
      "learning_rate": 7.6159685447073376e-06,
      "loss": 1.1092724800109863,
      "step": 380
    },
    {
      "epoch": 1.5219123505976095,
      "grad_norm": 0.08554384112358093,
      "learning_rate": 7.589738523671685e-06,
      "loss": 0.717506468296051,
      "step": 382
    },
    {
      "epoch": 1.5298804780876494,
      "grad_norm": 0.26063498854637146,
      "learning_rate": 7.563417665316338e-06,
      "loss": 0.7131892442703247,
      "step": 384
    },
    {
      "epoch": 1.5378486055776892,
      "grad_norm": 0.37084245681762695,
      "learning_rate": 7.5370071137634984e-06,
      "loss": 1.0123889446258545,
      "step": 386
    },
    {
      "epoch": 1.545816733067729,
      "grad_norm": 0.16996140778064728,
      "learning_rate": 7.510508017034165e-06,
      "loss": 1.1446300745010376,
      "step": 388
    },
    {
      "epoch": 1.5537848605577689,
      "grad_norm": 0.21291318535804749,
      "learning_rate": 7.483921526998247e-06,
      "loss": 0.5220435857772827,
      "step": 390
    },
    {
      "epoch": 1.5617529880478087,
      "grad_norm": 0.16681550443172455,
      "learning_rate": 7.4572487993244865e-06,
      "loss": 0.5643850564956665,
      "step": 392
    },
    {
      "epoch": 1.5697211155378485,
      "grad_norm": 0.22322708368301392,
      "learning_rate": 7.430490993430232e-06,
      "loss": 1.0830570459365845,
      "step": 394
    },
    {
      "epoch": 1.5776892430278884,
      "grad_norm": 0.41750508546829224,
      "learning_rate": 7.403649272431027e-06,
      "loss": 1.0145124197006226,
      "step": 396
    },
    {
      "epoch": 1.5856573705179282,
      "grad_norm": 0.09574401378631592,
      "learning_rate": 7.376724803090069e-06,
      "loss": 1.130380630493164,
      "step": 398
    },
    {
      "epoch": 1.593625498007968,
      "grad_norm": 0.41027480363845825,
      "learning_rate": 7.349718755767473e-06,
      "loss": 0.567794680595398,
      "step": 400
    },
    {
      "epoch": 1.6015936254980079,
      "grad_norm": 0.1381840854883194,
      "learning_rate": 7.322632304369416e-06,
      "loss": 1.1123874187469482,
      "step": 402
    },
    {
      "epoch": 1.6095617529880477,
      "grad_norm": 0.12441328912973404,
      "learning_rate": 7.2954666262970965e-06,
      "loss": 1.2539772987365723,
      "step": 404
    },
    {
      "epoch": 1.6175298804780875,
      "grad_norm": 0.21142208576202393,
      "learning_rate": 7.268222902395562e-06,
      "loss": 0.7022032141685486,
      "step": 406
    },
    {
      "epoch": 1.6254980079681274,
      "grad_norm": 0.08305402845144272,
      "learning_rate": 7.240902316902373e-06,
      "loss": 0.5497481226921082,
      "step": 408
    },
    {
      "epoch": 1.6334661354581672,
      "grad_norm": 0.17785273492336273,
      "learning_rate": 7.213506057396134e-06,
      "loss": 0.7389368414878845,
      "step": 410
    },
    {
      "epoch": 1.641434262948207,
      "grad_norm": 0.6749900579452515,
      "learning_rate": 7.186035314744872e-06,
      "loss": 0.4744718074798584,
      "step": 412
    },
    {
      "epoch": 1.6494023904382469,
      "grad_norm": 0.12851163744926453,
      "learning_rate": 7.1584912830542575e-06,
      "loss": 0.8076127171516418,
      "step": 414
    },
    {
      "epoch": 1.6573705179282867,
      "grad_norm": 0.14811666309833527,
      "learning_rate": 7.130875159615716e-06,
      "loss": 0.7206007838249207,
      "step": 416
    },
    {
      "epoch": 1.6653386454183265,
      "grad_norm": 0.20830723643302917,
      "learning_rate": 7.103188144854376e-06,
      "loss": 0.9853091835975647,
      "step": 418
    },
    {
      "epoch": 1.6733067729083664,
      "grad_norm": 0.0935424342751503,
      "learning_rate": 7.075431442276888e-06,
      "loss": 1.1377902030944824,
      "step": 420
    },
    {
      "epoch": 1.6812749003984062,
      "grad_norm": 0.20101168751716614,
      "learning_rate": 7.047606258419108e-06,
      "loss": 1.0096081495285034,
      "step": 422
    },
    {
      "epoch": 1.6892430278884463,
      "grad_norm": 0.22718675434589386,
      "learning_rate": 7.019713802793663e-06,
      "loss": 0.5819125771522522,
      "step": 424
    },
    {
      "epoch": 1.697211155378486,
      "grad_norm": 0.1758713275194168,
      "learning_rate": 6.99175528783736e-06,
      "loss": 1.1902439594268799,
      "step": 426
    },
    {
      "epoch": 1.705179282868526,
      "grad_norm": 0.1724633425474167,
      "learning_rate": 6.963731928858495e-06,
      "loss": 1.127792477607727,
      "step": 428
    },
    {
      "epoch": 1.7131474103585658,
      "grad_norm": 0.18502382934093475,
      "learning_rate": 6.935644943984017e-06,
      "loss": 0.7112741470336914,
      "step": 430
    },
    {
      "epoch": 1.7211155378486056,
      "grad_norm": 0.16653676331043243,
      "learning_rate": 6.9074955541065865e-06,
      "loss": 0.8753823637962341,
      "step": 432
    },
    {
      "epoch": 1.7290836653386454,
      "grad_norm": 0.4369586706161499,
      "learning_rate": 6.8792849828315054e-06,
      "loss": 0.6039544939994812,
      "step": 434
    },
    {
      "epoch": 1.7370517928286853,
      "grad_norm": 0.8986008167266846,
      "learning_rate": 6.851014456423515e-06,
      "loss": 0.3703668415546417,
      "step": 436
    },
    {
      "epoch": 1.745019920318725,
      "grad_norm": 0.2696269750595093,
      "learning_rate": 6.822685203753508e-06,
      "loss": 0.7240628004074097,
      "step": 438
    },
    {
      "epoch": 1.752988047808765,
      "grad_norm": 0.7449815273284912,
      "learning_rate": 6.794298456245111e-06,
      "loss": 0.6291514039039612,
      "step": 440
    },
    {
      "epoch": 1.7609561752988048,
      "grad_norm": 0.2776779234409332,
      "learning_rate": 6.765855447821144e-06,
      "loss": 1.4720019102096558,
      "step": 442
    },
    {
      "epoch": 1.7689243027888446,
      "grad_norm": 0.15569284558296204,
      "learning_rate": 6.737357414849993e-06,
      "loss": 0.7313789129257202,
      "step": 444
    },
    {
      "epoch": 1.7768924302788844,
      "grad_norm": 0.12689454853534698,
      "learning_rate": 6.70880559609187e-06,
      "loss": 1.0231692790985107,
      "step": 446
    },
    {
      "epoch": 1.7848605577689243,
      "grad_norm": 0.2637998163700104,
      "learning_rate": 6.680201232644958e-06,
      "loss": 1.0926713943481445,
      "step": 448
    },
    {
      "epoch": 1.792828685258964,
      "grad_norm": 0.14777080714702606,
      "learning_rate": 6.651545567891471e-06,
      "loss": 0.8266294598579407,
      "step": 450
    },
    {
      "epoch": 1.800796812749004,
      "grad_norm": 0.0680837631225586,
      "learning_rate": 6.622839847443596e-06,
      "loss": 0.4956511855125427,
      "step": 452
    },
    {
      "epoch": 1.8087649402390438,
      "grad_norm": 0.3037063181400299,
      "learning_rate": 6.59408531908936e-06,
      "loss": 0.961536705493927,
      "step": 454
    },
    {
      "epoch": 1.8167330677290838,
      "grad_norm": 1.0879807472229004,
      "learning_rate": 6.565283232738386e-06,
      "loss": 0.6515597701072693,
      "step": 456
    },
    {
      "epoch": 1.8247011952191237,
      "grad_norm": 0.22389857470989227,
      "learning_rate": 6.536434840367555e-06,
      "loss": 1.0253115892410278,
      "step": 458
    },
    {
      "epoch": 1.8326693227091635,
      "grad_norm": 0.15206734836101532,
      "learning_rate": 6.507541395966591e-06,
      "loss": 0.7219138145446777,
      "step": 460
    },
    {
      "epoch": 1.8406374501992033,
      "grad_norm": 0.13305911421775818,
      "learning_rate": 6.47860415548356e-06,
      "loss": 0.8085894584655762,
      "step": 462
    },
    {
      "epoch": 1.8486055776892432,
      "grad_norm": 0.08373041450977325,
      "learning_rate": 6.449624376770259e-06,
      "loss": 0.1179397702217102,
      "step": 464
    },
    {
      "epoch": 1.856573705179283,
      "grad_norm": 0.13342876732349396,
      "learning_rate": 6.420603319527554e-06,
      "loss": 1.153374195098877,
      "step": 466
    },
    {
      "epoch": 1.8645418326693228,
      "grad_norm": 0.22394229471683502,
      "learning_rate": 6.3915422452506146e-06,
      "loss": 0.5254758596420288,
      "step": 468
    },
    {
      "epoch": 1.8725099601593627,
      "grad_norm": 0.2631344795227051,
      "learning_rate": 6.362442417174082e-06,
      "loss": 1.1494216918945312,
      "step": 470
    },
    {
      "epoch": 1.8804780876494025,
      "grad_norm": 0.19804665446281433,
      "learning_rate": 6.33330510021716e-06,
      "loss": 1.0185717344284058,
      "step": 472
    },
    {
      "epoch": 1.8884462151394423,
      "grad_norm": 0.24636146426200867,
      "learning_rate": 6.304131560928622e-06,
      "loss": 1.1287875175476074,
      "step": 474
    },
    {
      "epoch": 1.8964143426294822,
      "grad_norm": 0.1704067885875702,
      "learning_rate": 6.274923067431775e-06,
      "loss": 0.9416290521621704,
      "step": 476
    },
    {
      "epoch": 1.904382470119522,
      "grad_norm": 0.1688898205757141,
      "learning_rate": 6.2456808893693155e-06,
      "loss": 0.8116242289543152,
      "step": 478
    },
    {
      "epoch": 1.9123505976095618,
      "grad_norm": 0.25278931856155396,
      "learning_rate": 6.216406297848155e-06,
      "loss": 0.5594226121902466,
      "step": 480
    },
    {
      "epoch": 1.9203187250996017,
      "grad_norm": 0.24208518862724304,
      "learning_rate": 6.187100565384162e-06,
      "loss": 0.7302346229553223,
      "step": 482
    },
    {
      "epoch": 1.9282868525896415,
      "grad_norm": 0.140226811170578,
      "learning_rate": 6.157764965846841e-06,
      "loss": 1.1030246019363403,
      "step": 484
    },
    {
      "epoch": 1.9362549800796813,
      "grad_norm": 0.2927384674549103,
      "learning_rate": 6.128400774403974e-06,
      "loss": 1.0868550539016724,
      "step": 486
    },
    {
      "epoch": 1.9442231075697212,
      "grad_norm": 0.1566203534603119,
      "learning_rate": 6.0990092674661834e-06,
      "loss": 1.0964586734771729,
      "step": 488
    },
    {
      "epoch": 1.952191235059761,
      "grad_norm": 0.10223723202943802,
      "learning_rate": 6.069591722631446e-06,
      "loss": 1.103720784187317,
      "step": 490
    },
    {
      "epoch": 1.9601593625498008,
      "grad_norm": 0.14172522723674774,
      "learning_rate": 6.040149418629562e-06,
      "loss": 0.764342725276947,
      "step": 492
    },
    {
      "epoch": 1.9681274900398407,
      "grad_norm": 3.1486051082611084,
      "learning_rate": 6.010683635266569e-06,
      "loss": 0.6474576592445374,
      "step": 494
    },
    {
      "epoch": 1.9760956175298805,
      "grad_norm": 0.1419641375541687,
      "learning_rate": 5.981195653369113e-06,
      "loss": 1.0781517028808594,
      "step": 496
    },
    {
      "epoch": 1.9840637450199203,
      "grad_norm": 0.36362552642822266,
      "learning_rate": 5.95168675472877e-06,
      "loss": 0.8261182904243469,
      "step": 498
    },
    {
      "epoch": 1.9920318725099602,
      "grad_norm": 0.15697282552719116,
      "learning_rate": 5.922158222046334e-06,
      "loss": 1.1619532108306885,
      "step": 500
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.3233239948749542,
      "learning_rate": 5.892611338876053e-06,
      "loss": 0.623477578163147,
      "step": 502
    },
    {
      "epoch": 2.00796812749004,
      "grad_norm": 0.11791647225618362,
      "learning_rate": 5.8630473895698426e-06,
      "loss": 0.5960298776626587,
      "step": 504
    },
    {
      "epoch": 2.0159362549800797,
      "grad_norm": 0.1991225928068161,
      "learning_rate": 5.833467659221452e-06,
      "loss": 0.5541172027587891,
      "step": 506
    },
    {
      "epoch": 2.0239043824701195,
      "grad_norm": 0.028176333755254745,
      "learning_rate": 5.803873433610604e-06,
      "loss": 0.5265404582023621,
      "step": 508
    },
    {
      "epoch": 2.0318725099601593,
      "grad_norm": 0.14614051580429077,
      "learning_rate": 5.774265999147108e-06,
      "loss": 0.819342851638794,
      "step": 510
    },
    {
      "epoch": 2.039840637450199,
      "grad_norm": 0.12275773286819458,
      "learning_rate": 5.744646642814935e-06,
      "loss": 0.6939613819122314,
      "step": 512
    },
    {
      "epoch": 2.047808764940239,
      "grad_norm": 1.7416125535964966,
      "learning_rate": 5.715016652116284e-06,
      "loss": 0.2898235619068146,
      "step": 514
    },
    {
      "epoch": 2.055776892430279,
      "grad_norm": 0.2686458230018616,
      "learning_rate": 5.68537731501561e-06,
      "loss": 0.8659465312957764,
      "step": 516
    },
    {
      "epoch": 2.0637450199203187,
      "grad_norm": 0.2778455913066864,
      "learning_rate": 5.655729919883639e-06,
      "loss": 0.49256813526153564,
      "step": 518
    },
    {
      "epoch": 2.0717131474103585,
      "grad_norm": 0.3662952184677124,
      "learning_rate": 5.626075755441366e-06,
      "loss": 0.7744277715682983,
      "step": 520
    },
    {
      "epoch": 2.0796812749003983,
      "grad_norm": 0.17675438523292542,
      "learning_rate": 5.596416110704034e-06,
      "loss": 0.9661864042282104,
      "step": 522
    },
    {
      "epoch": 2.087649402390438,
      "grad_norm": 0.4687350392341614,
      "learning_rate": 5.566752274925109e-06,
      "loss": 0.8630577325820923,
      "step": 524
    },
    {
      "epoch": 2.095617529880478,
      "grad_norm": 0.21669980883598328,
      "learning_rate": 5.5370855375402345e-06,
      "loss": 0.39600545167922974,
      "step": 526
    },
    {
      "epoch": 2.103585657370518,
      "grad_norm": 0.14760233461856842,
      "learning_rate": 5.507417188111176e-06,
      "loss": 0.7052125334739685,
      "step": 528
    },
    {
      "epoch": 2.1115537848605577,
      "grad_norm": 0.13765133917331696,
      "learning_rate": 5.477748516269779e-06,
      "loss": 0.6329566836357117,
      "step": 530
    },
    {
      "epoch": 2.1195219123505975,
      "grad_norm": 0.2127397209405899,
      "learning_rate": 5.448080811661897e-06,
      "loss": 0.1936207264661789,
      "step": 532
    },
    {
      "epoch": 2.1274900398406373,
      "grad_norm": 0.250801146030426,
      "learning_rate": 5.418415363891346e-06,
      "loss": 0.7265528440475464,
      "step": 534
    },
    {
      "epoch": 2.135458167330677,
      "grad_norm": 2.4114866256713867,
      "learning_rate": 5.3887534624638335e-06,
      "loss": 0.6055819988250732,
      "step": 536
    },
    {
      "epoch": 2.143426294820717,
      "grad_norm": 0.4504620432853699,
      "learning_rate": 5.359096396730924e-06,
      "loss": 0.6730338931083679,
      "step": 538
    },
    {
      "epoch": 2.151394422310757,
      "grad_norm": 0.1654483675956726,
      "learning_rate": 5.329445455833975e-06,
      "loss": 0.8896601796150208,
      "step": 540
    },
    {
      "epoch": 2.1593625498007967,
      "grad_norm": 1.2194499969482422,
      "learning_rate": 5.299801928648105e-06,
      "loss": 0.8691336512565613,
      "step": 542
    },
    {
      "epoch": 2.1673306772908365,
      "grad_norm": 0.3409273028373718,
      "learning_rate": 5.270167103726181e-06,
      "loss": 0.7230013012886047,
      "step": 544
    },
    {
      "epoch": 2.1752988047808763,
      "grad_norm": 0.24869979918003082,
      "learning_rate": 5.240542269242794e-06,
      "loss": 0.6018685102462769,
      "step": 546
    },
    {
      "epoch": 2.183266932270916,
      "grad_norm": 0.34317705035209656,
      "learning_rate": 5.210928712938262e-06,
      "loss": 0.9386495351791382,
      "step": 548
    },
    {
      "epoch": 2.191235059760956,
      "grad_norm": 0.12627029418945312,
      "learning_rate": 5.181327722062667e-06,
      "loss": 1.0006942749023438,
      "step": 550
    },
    {
      "epoch": 2.199203187250996,
      "grad_norm": 0.19774127006530762,
      "learning_rate": 5.151740583319892e-06,
      "loss": 0.6490278840065002,
      "step": 552
    },
    {
      "epoch": 2.2071713147410357,
      "grad_norm": 0.557928204536438,
      "learning_rate": 5.122168582811689e-06,
      "loss": 0.42054125666618347,
      "step": 554
    },
    {
      "epoch": 2.2151394422310755,
      "grad_norm": 0.14655070006847382,
      "learning_rate": 5.092613005981781e-06,
      "loss": 0.6261390447616577,
      "step": 556
    },
    {
      "epoch": 2.2231075697211153,
      "grad_norm": 0.45267534255981445,
      "learning_rate": 5.063075137559979e-06,
      "loss": 0.8401848673820496,
      "step": 558
    },
    {
      "epoch": 2.231075697211155,
      "grad_norm": 0.12097624689340591,
      "learning_rate": 5.033556261506343e-06,
      "loss": 0.8791764974594116,
      "step": 560
    },
    {
      "epoch": 2.239043824701195,
      "grad_norm": 0.26983964443206787,
      "learning_rate": 5.0040576609553645e-06,
      "loss": 0.9999547004699707,
      "step": 562
    },
    {
      "epoch": 2.247011952191235,
      "grad_norm": 1.0746606588363647,
      "learning_rate": 4.974580618160194e-06,
      "loss": 0.7313042283058167,
      "step": 564
    },
    {
      "epoch": 2.2549800796812747,
      "grad_norm": 0.13656942546367645,
      "learning_rate": 4.945126414436908e-06,
      "loss": 0.6323325037956238,
      "step": 566
    },
    {
      "epoch": 2.2629482071713145,
      "grad_norm": 0.15358102321624756,
      "learning_rate": 4.915696330108804e-06,
      "loss": 0.7094585299491882,
      "step": 568
    },
    {
      "epoch": 2.2709163346613543,
      "grad_norm": 0.15575262904167175,
      "learning_rate": 4.886291644450751e-06,
      "loss": 0.9807476997375488,
      "step": 570
    },
    {
      "epoch": 2.278884462151394,
      "grad_norm": 0.1870071440935135,
      "learning_rate": 4.8569136356335845e-06,
      "loss": 0.5671411156654358,
      "step": 572
    },
    {
      "epoch": 2.2868525896414345,
      "grad_norm": 0.14295098185539246,
      "learning_rate": 4.827563580668544e-06,
      "loss": 0.6624703407287598,
      "step": 574
    },
    {
      "epoch": 2.2948207171314743,
      "grad_norm": 0.09388113021850586,
      "learning_rate": 4.798242755351755e-06,
      "loss": 0.20577380061149597,
      "step": 576
    },
    {
      "epoch": 2.302788844621514,
      "grad_norm": 0.20584431290626526,
      "learning_rate": 4.768952434208795e-06,
      "loss": 0.6171508431434631,
      "step": 578
    },
    {
      "epoch": 2.310756972111554,
      "grad_norm": 0.1313192993402481,
      "learning_rate": 4.7396938904392656e-06,
      "loss": 0.6086855530738831,
      "step": 580
    },
    {
      "epoch": 2.318725099601594,
      "grad_norm": 0.23046337068080902,
      "learning_rate": 4.71046839586147e-06,
      "loss": 0.5453673601150513,
      "step": 582
    },
    {
      "epoch": 2.3266932270916336,
      "grad_norm": 0.12405805289745331,
      "learning_rate": 4.681277220857108e-06,
      "loss": 0.7182409763336182,
      "step": 584
    },
    {
      "epoch": 2.3346613545816735,
      "grad_norm": 0.2038748413324356,
      "learning_rate": 4.6521216343160824e-06,
      "loss": 0.6064735651016235,
      "step": 586
    },
    {
      "epoch": 2.3426294820717133,
      "grad_norm": 0.4655267596244812,
      "learning_rate": 4.623002903581316e-06,
      "loss": 0.8138838410377502,
      "step": 588
    },
    {
      "epoch": 2.350597609561753,
      "grad_norm": 0.11591798067092896,
      "learning_rate": 4.593922294393677e-06,
      "loss": 0.9658578038215637,
      "step": 590
    },
    {
      "epoch": 2.358565737051793,
      "grad_norm": 0.11815351247787476,
      "learning_rate": 4.564881070836955e-06,
      "loss": 1.010229229927063,
      "step": 592
    },
    {
      "epoch": 2.366533864541833,
      "grad_norm": 0.294574111700058,
      "learning_rate": 4.535880495282916e-06,
      "loss": 0.6586866974830627,
      "step": 594
    },
    {
      "epoch": 2.3745019920318726,
      "grad_norm": 0.10035061836242676,
      "learning_rate": 4.506921828336427e-06,
      "loss": 0.9030760526657104,
      "step": 596
    },
    {
      "epoch": 2.3824701195219125,
      "grad_norm": 0.14462533593177795,
      "learning_rate": 4.478006328780657e-06,
      "loss": 0.6834123730659485,
      "step": 598
    },
    {
      "epoch": 2.3904382470119523,
      "grad_norm": 0.389182984828949,
      "learning_rate": 4.44913525352237e-06,
      "loss": 0.8373348712921143,
      "step": 600
    },
    {
      "epoch": 2.398406374501992,
      "grad_norm": 0.18416228890419006,
      "learning_rate": 4.420309857537277e-06,
      "loss": 0.9915989637374878,
      "step": 602
    },
    {
      "epoch": 2.406374501992032,
      "grad_norm": 0.2437923103570938,
      "learning_rate": 4.39153139381549e-06,
      "loss": 0.5638095736503601,
      "step": 604
    },
    {
      "epoch": 2.414342629482072,
      "grad_norm": 0.49585264921188354,
      "learning_rate": 4.3628011133070605e-06,
      "loss": 0.2088540494441986,
      "step": 606
    },
    {
      "epoch": 2.4223107569721116,
      "grad_norm": 0.04721204936504364,
      "learning_rate": 4.334120264867596e-06,
      "loss": 0.4597742557525635,
      "step": 608
    },
    {
      "epoch": 2.4302788844621515,
      "grad_norm": 0.09569624811410904,
      "learning_rate": 4.305490095203979e-06,
      "loss": 0.8684208393096924,
      "step": 610
    },
    {
      "epoch": 2.4382470119521913,
      "grad_norm": 0.1270139515399933,
      "learning_rate": 4.27691184882017e-06,
      "loss": 0.6606026291847229,
      "step": 612
    },
    {
      "epoch": 2.446215139442231,
      "grad_norm": 0.1697314977645874,
      "learning_rate": 4.248386767963121e-06,
      "loss": 0.5479908585548401,
      "step": 614
    },
    {
      "epoch": 2.454183266932271,
      "grad_norm": 0.13808858394622803,
      "learning_rate": 4.219916092568764e-06,
      "loss": 0.8740329742431641,
      "step": 616
    },
    {
      "epoch": 2.462151394422311,
      "grad_norm": 0.22472700476646423,
      "learning_rate": 4.191501060208121e-06,
      "loss": 0.6489477753639221,
      "step": 618
    },
    {
      "epoch": 2.4701195219123506,
      "grad_norm": 0.31426283717155457,
      "learning_rate": 4.163142906033512e-06,
      "loss": 0.7165242433547974,
      "step": 620
    },
    {
      "epoch": 2.4780876494023905,
      "grad_norm": 0.195907860994339,
      "learning_rate": 4.134842862724854e-06,
      "loss": 0.7869659066200256,
      "step": 622
    },
    {
      "epoch": 2.4860557768924303,
      "grad_norm": 0.21041953563690186,
      "learning_rate": 4.106602160436088e-06,
      "loss": 0.7412861585617065,
      "step": 624
    },
    {
      "epoch": 2.49402390438247,
      "grad_norm": 0.18993598222732544,
      "learning_rate": 4.078422026741706e-06,
      "loss": 0.5796581506729126,
      "step": 626
    },
    {
      "epoch": 2.50199203187251,
      "grad_norm": 0.20352329313755035,
      "learning_rate": 4.050303686583386e-06,
      "loss": 0.9587329030036926,
      "step": 628
    },
    {
      "epoch": 2.50996015936255,
      "grad_norm": 0.17588691413402557,
      "learning_rate": 4.0222483622167465e-06,
      "loss": 0.4242645800113678,
      "step": 630
    },
    {
      "epoch": 2.5179282868525896,
      "grad_norm": 0.26455914974212646,
      "learning_rate": 3.9942572731582185e-06,
      "loss": 0.403324156999588,
      "step": 632
    },
    {
      "epoch": 2.5258964143426295,
      "grad_norm": 0.1372862458229065,
      "learning_rate": 3.966331636132039e-06,
      "loss": 0.9816950559616089,
      "step": 634
    },
    {
      "epoch": 2.5338645418326693,
      "grad_norm": 0.890044629573822,
      "learning_rate": 3.938472665017355e-06,
      "loss": 0.9140403270721436,
      "step": 636
    },
    {
      "epoch": 2.541832669322709,
      "grad_norm": 0.24590884149074554,
      "learning_rate": 3.91068157079546e-06,
      "loss": 0.9193362593650818,
      "step": 638
    },
    {
      "epoch": 2.549800796812749,
      "grad_norm": 0.056746870279312134,
      "learning_rate": 3.882959561497159e-06,
      "loss": 0.5603985786437988,
      "step": 640
    },
    {
      "epoch": 2.557768924302789,
      "grad_norm": 0.19658251106739044,
      "learning_rate": 3.855307842150259e-06,
      "loss": 0.9370026588439941,
      "step": 642
    },
    {
      "epoch": 2.5657370517928286,
      "grad_norm": 0.1134665459394455,
      "learning_rate": 3.827727614727177e-06,
      "loss": 0.9933424592018127,
      "step": 644
    },
    {
      "epoch": 2.5737051792828685,
      "grad_norm": 0.118997722864151,
      "learning_rate": 3.8002200780927e-06,
      "loss": 0.6442151665687561,
      "step": 646
    },
    {
      "epoch": 2.5816733067729083,
      "grad_norm": 0.18749064207077026,
      "learning_rate": 3.772786427951878e-06,
      "loss": 0.9765391945838928,
      "step": 648
    },
    {
      "epoch": 2.589641434262948,
      "grad_norm": 0.20333199203014374,
      "learning_rate": 3.7454278567980406e-06,
      "loss": 0.728888988494873,
      "step": 650
    },
    {
      "epoch": 2.597609561752988,
      "grad_norm": 0.17941781878471375,
      "learning_rate": 3.7181455538609577e-06,
      "loss": 0.7437627911567688,
      "step": 652
    },
    {
      "epoch": 2.605577689243028,
      "grad_norm": 0.13907383382320404,
      "learning_rate": 3.690940705055163e-06,
      "loss": 0.6450785994529724,
      "step": 654
    },
    {
      "epoch": 2.6135458167330676,
      "grad_norm": 0.16438795626163483,
      "learning_rate": 3.6638144929283824e-06,
      "loss": 0.30199572443962097,
      "step": 656
    },
    {
      "epoch": 2.6215139442231075,
      "grad_norm": 0.16593554615974426,
      "learning_rate": 3.6367680966101483e-06,
      "loss": 0.5315187573432922,
      "step": 658
    },
    {
      "epoch": 2.6294820717131473,
      "grad_norm": 0.1273457556962967,
      "learning_rate": 3.6098026917605334e-06,
      "loss": 0.8293731808662415,
      "step": 660
    },
    {
      "epoch": 2.637450199203187,
      "grad_norm": 1.5358424186706543,
      "learning_rate": 3.5829194505190534e-06,
      "loss": 0.8861653804779053,
      "step": 662
    },
    {
      "epoch": 2.645418326693227,
      "grad_norm": 0.18468861281871796,
      "learning_rate": 3.5561195414537164e-06,
      "loss": 0.5043506026268005,
      "step": 664
    },
    {
      "epoch": 2.653386454183267,
      "grad_norm": 0.3537488877773285,
      "learning_rate": 3.529404129510221e-06,
      "loss": 0.8800249695777893,
      "step": 666
    },
    {
      "epoch": 2.6613545816733066,
      "grad_norm": 0.08804573118686676,
      "learning_rate": 3.502774375961321e-06,
      "loss": 0.5786421298980713,
      "step": 668
    },
    {
      "epoch": 2.6693227091633465,
      "grad_norm": 0.269173264503479,
      "learning_rate": 3.4762314383563533e-06,
      "loss": 0.677842378616333,
      "step": 670
    },
    {
      "epoch": 2.6772908366533863,
      "grad_norm": 0.07226338982582092,
      "learning_rate": 3.4497764704709143e-06,
      "loss": 0.32216712832450867,
      "step": 672
    },
    {
      "epoch": 2.685258964143426,
      "grad_norm": 0.18977269530296326,
      "learning_rate": 3.4234106222567055e-06,
      "loss": 0.9900534749031067,
      "step": 674
    },
    {
      "epoch": 2.6932270916334664,
      "grad_norm": 8.442211151123047,
      "learning_rate": 3.3971350397915504e-06,
      "loss": 0.6448943614959717,
      "step": 676
    },
    {
      "epoch": 2.7011952191235062,
      "grad_norm": 0.2724821865558624,
      "learning_rate": 3.370950865229583e-06,
      "loss": 0.5593557357788086,
      "step": 678
    },
    {
      "epoch": 2.709163346613546,
      "grad_norm": 0.13181589543819427,
      "learning_rate": 3.344859236751585e-06,
      "loss": 0.9824644923210144,
      "step": 680
    },
    {
      "epoch": 2.717131474103586,
      "grad_norm": 0.12464668601751328,
      "learning_rate": 3.3188612885155257e-06,
      "loss": 1.0139515399932861,
      "step": 682
    },
    {
      "epoch": 2.7250996015936257,
      "grad_norm": 0.14147010445594788,
      "learning_rate": 3.2929581506072596e-06,
      "loss": 0.6543970108032227,
      "step": 684
    },
    {
      "epoch": 2.7330677290836656,
      "grad_norm": 0.3615143895149231,
      "learning_rate": 3.267150948991387e-06,
      "loss": 0.3849199712276459,
      "step": 686
    },
    {
      "epoch": 2.7410358565737054,
      "grad_norm": 0.19376139342784882,
      "learning_rate": 3.241440805462337e-06,
      "loss": 0.9890676736831665,
      "step": 688
    },
    {
      "epoch": 2.7490039840637452,
      "grad_norm": 0.3360038697719574,
      "learning_rate": 3.215828837595585e-06,
      "loss": 0.47163745760917664,
      "step": 690
    },
    {
      "epoch": 2.756972111553785,
      "grad_norm": 0.1981293261051178,
      "learning_rate": 3.190316158699084e-06,
      "loss": 1.0654531717300415,
      "step": 692
    },
    {
      "epoch": 2.764940239043825,
      "grad_norm": 0.8281208276748657,
      "learning_rate": 3.1649038777648622e-06,
      "loss": 0.8090680837631226,
      "step": 694
    },
    {
      "epoch": 2.7729083665338647,
      "grad_norm": 0.3178573548793793,
      "learning_rate": 3.1395930994208234e-06,
      "loss": 0.6035836338996887,
      "step": 696
    },
    {
      "epoch": 2.7808764940239046,
      "grad_norm": 0.31108981370925903,
      "learning_rate": 3.114384923882734e-06,
      "loss": 0.8315184712409973,
      "step": 698
    },
    {
      "epoch": 2.7888446215139444,
      "grad_norm": 0.11245593428611755,
      "learning_rate": 3.089280446906387e-06,
      "loss": 1.0156062841415405,
      "step": 700
    },
    {
      "epoch": 2.7968127490039842,
      "grad_norm": 0.1584748476743698,
      "learning_rate": 3.0642807597399847e-06,
      "loss": 0.6059562563896179,
      "step": 702
    },
    {
      "epoch": 2.804780876494024,
      "grad_norm": 0.23952287435531616,
      "learning_rate": 3.0393869490766913e-06,
      "loss": 1.144424557685852,
      "step": 704
    },
    {
      "epoch": 2.812749003984064,
      "grad_norm": 0.12983287870883942,
      "learning_rate": 3.0146000970074087e-06,
      "loss": 1.145877718925476,
      "step": 706
    },
    {
      "epoch": 2.8207171314741037,
      "grad_norm": 0.23081286251544952,
      "learning_rate": 2.9899212809737275e-06,
      "loss": 1.0499770641326904,
      "step": 708
    },
    {
      "epoch": 2.8286852589641436,
      "grad_norm": 0.0911397635936737,
      "learning_rate": 2.9653515737211037e-06,
      "loss": 0.506878137588501,
      "step": 710
    },
    {
      "epoch": 2.8366533864541834,
      "grad_norm": 0.1302153617143631,
      "learning_rate": 2.940892043252225e-06,
      "loss": 1.0316249132156372,
      "step": 712
    },
    {
      "epoch": 2.8446215139442232,
      "grad_norm": 0.12898322939872742,
      "learning_rate": 2.916543752780574e-06,
      "loss": 0.5781832337379456,
      "step": 714
    },
    {
      "epoch": 2.852589641434263,
      "grad_norm": 0.1421491950750351,
      "learning_rate": 2.892307760684234e-06,
      "loss": 0.9909021258354187,
      "step": 716
    },
    {
      "epoch": 2.860557768924303,
      "grad_norm": 0.5054803490638733,
      "learning_rate": 2.8681851204598675e-06,
      "loss": 0.5557779669761658,
      "step": 718
    },
    {
      "epoch": 2.8685258964143427,
      "grad_norm": 0.1159234568476677,
      "learning_rate": 2.844176880676931e-06,
      "loss": 0.576461672782898,
      "step": 720
    },
    {
      "epoch": 2.8764940239043826,
      "grad_norm": 0.19920121133327484,
      "learning_rate": 2.8202840849320867e-06,
      "loss": 0.6151230335235596,
      "step": 722
    },
    {
      "epoch": 2.8844621513944224,
      "grad_norm": 0.3590832054615021,
      "learning_rate": 2.7965077718038437e-06,
      "loss": 0.5856032967567444,
      "step": 724
    },
    {
      "epoch": 2.8924302788844622,
      "grad_norm": 0.1798829734325409,
      "learning_rate": 2.7728489748074193e-06,
      "loss": 0.9423098564147949,
      "step": 726
    },
    {
      "epoch": 2.900398406374502,
      "grad_norm": 0.15152226388454437,
      "learning_rate": 2.7493087223498015e-06,
      "loss": 0.5390990972518921,
      "step": 728
    },
    {
      "epoch": 2.908366533864542,
      "grad_norm": 0.16586051881313324,
      "learning_rate": 2.725888037685057e-06,
      "loss": 0.9130787253379822,
      "step": 730
    },
    {
      "epoch": 2.9163346613545817,
      "grad_norm": 0.12153919786214828,
      "learning_rate": 2.702587938869842e-06,
      "loss": 0.30029648542404175,
      "step": 732
    },
    {
      "epoch": 2.9243027888446216,
      "grad_norm": 0.17477543652057648,
      "learning_rate": 2.6794094387191617e-06,
      "loss": 1.0147703886032104,
      "step": 734
    },
    {
      "epoch": 2.9322709163346614,
      "grad_norm": 0.3891335725784302,
      "learning_rate": 2.6563535447623302e-06,
      "loss": 0.3808979094028473,
      "step": 736
    },
    {
      "epoch": 2.9402390438247012,
      "grad_norm": 0.2984495460987091,
      "learning_rate": 2.6334212591991847e-06,
      "loss": 0.7716872096061707,
      "step": 738
    },
    {
      "epoch": 2.948207171314741,
      "grad_norm": 0.1185426265001297,
      "learning_rate": 2.6106135788565248e-06,
      "loss": 1.0928815603256226,
      "step": 740
    },
    {
      "epoch": 2.956175298804781,
      "grad_norm": 0.6638994812965393,
      "learning_rate": 2.5879314951447688e-06,
      "loss": 0.2910023033618927,
      "step": 742
    },
    {
      "epoch": 2.9641434262948207,
      "grad_norm": 0.3299606144428253,
      "learning_rate": 2.5653759940148692e-06,
      "loss": 0.8220379948616028,
      "step": 744
    },
    {
      "epoch": 2.9721115537848606,
      "grad_norm": 0.26891013979911804,
      "learning_rate": 2.542948055915454e-06,
      "loss": 0.8336227536201477,
      "step": 746
    },
    {
      "epoch": 2.9800796812749004,
      "grad_norm": 0.24989786744117737,
      "learning_rate": 2.5206486557502087e-06,
      "loss": 0.525390625,
      "step": 748
    },
    {
      "epoch": 2.9880478087649402,
      "grad_norm": 0.14200733602046967,
      "learning_rate": 2.498478762835492e-06,
      "loss": 0.14072617888450623,
      "step": 750
    },
    {
      "epoch": 2.99601593625498,
      "grad_norm": 0.14495570957660675,
      "learning_rate": 2.476439340858208e-06,
      "loss": 0.9838566780090332,
      "step": 752
    },
    {
      "epoch": 3.00398406374502,
      "grad_norm": 0.17396603524684906,
      "learning_rate": 2.4545313478339178e-06,
      "loss": 0.5045596361160278,
      "step": 754
    },
    {
      "epoch": 3.0119521912350598,
      "grad_norm": 0.14038436114788055,
      "learning_rate": 2.4327557360651895e-06,
      "loss": 0.5435022115707397,
      "step": 756
    },
    {
      "epoch": 3.0199203187250996,
      "grad_norm": 0.08444824069738388,
      "learning_rate": 2.411113452100211e-06,
      "loss": 0.23670323193073273,
      "step": 758
    },
    {
      "epoch": 3.0278884462151394,
      "grad_norm": 0.11717414855957031,
      "learning_rate": 2.3896054366916365e-06,
      "loss": 0.586422324180603,
      "step": 760
    },
    {
      "epoch": 3.0358565737051793,
      "grad_norm": 0.36922261118888855,
      "learning_rate": 2.3682326247557e-06,
      "loss": 0.39174172282218933,
      "step": 762
    },
    {
      "epoch": 3.043824701195219,
      "grad_norm": 0.4215605854988098,
      "learning_rate": 2.346995945331576e-06,
      "loss": 0.8948820233345032,
      "step": 764
    },
    {
      "epoch": 3.051792828685259,
      "grad_norm": 0.14065450429916382,
      "learning_rate": 2.325896321540993e-06,
      "loss": 0.11864655464887619,
      "step": 766
    },
    {
      "epoch": 3.0597609561752988,
      "grad_norm": 0.17616204917430878,
      "learning_rate": 2.3049346705481086e-06,
      "loss": 0.5252296328544617,
      "step": 768
    },
    {
      "epoch": 3.0677290836653386,
      "grad_norm": 0.23492881655693054,
      "learning_rate": 2.284111903519641e-06,
      "loss": 0.7128872871398926,
      "step": 770
    },
    {
      "epoch": 3.0756972111553784,
      "grad_norm": 0.1484983265399933,
      "learning_rate": 2.26342892558526e-06,
      "loss": 0.8851345777511597,
      "step": 772
    },
    {
      "epoch": 3.0836653386454183,
      "grad_norm": 0.40719956159591675,
      "learning_rate": 2.2428866357982515e-06,
      "loss": 0.7606514692306519,
      "step": 774
    },
    {
      "epoch": 3.091633466135458,
      "grad_norm": 0.04664382338523865,
      "learning_rate": 2.222485927096424e-06,
      "loss": 0.5094689726829529,
      "step": 776
    },
    {
      "epoch": 3.099601593625498,
      "grad_norm": 0.26247334480285645,
      "learning_rate": 2.2022276862633075e-06,
      "loss": 0.49606576561927795,
      "step": 778
    },
    {
      "epoch": 3.1075697211155378,
      "grad_norm": 0.12347497045993805,
      "learning_rate": 2.1821127938895938e-06,
      "loss": 0.8695854544639587,
      "step": 780
    },
    {
      "epoch": 3.1155378486055776,
      "grad_norm": 0.08101819455623627,
      "learning_rate": 2.162142124334871e-06,
      "loss": 0.2356264293193817,
      "step": 782
    },
    {
      "epoch": 3.1235059760956174,
      "grad_norm": 0.2343272715806961,
      "learning_rate": 2.1423165456896068e-06,
      "loss": 0.9376544952392578,
      "step": 784
    },
    {
      "epoch": 3.1314741035856573,
      "grad_norm": 0.1727258414030075,
      "learning_rate": 2.1226369197374176e-06,
      "loss": 0.6643160581588745,
      "step": 786
    },
    {
      "epoch": 3.139442231075697,
      "grad_norm": 0.21666266024112701,
      "learning_rate": 2.103104101917615e-06,
      "loss": 0.47477614879608154,
      "step": 788
    },
    {
      "epoch": 3.147410358565737,
      "grad_norm": 0.07991424202919006,
      "learning_rate": 2.0837189412880044e-06,
      "loss": 0.47438961267471313,
      "step": 790
    },
    {
      "epoch": 3.1553784860557768,
      "grad_norm": 0.32614389061927795,
      "learning_rate": 2.0644822804879964e-06,
      "loss": 0.6335086822509766,
      "step": 792
    },
    {
      "epoch": 3.1633466135458166,
      "grad_norm": 0.02057274617254734,
      "learning_rate": 2.045394955701969e-06,
      "loss": 0.15964680910110474,
      "step": 794
    },
    {
      "epoch": 3.1713147410358564,
      "grad_norm": 0.21819208562374115,
      "learning_rate": 2.0264577966229222e-06,
      "loss": 0.5290365815162659,
      "step": 796
    },
    {
      "epoch": 3.1792828685258963,
      "grad_norm": 0.2687428891658783,
      "learning_rate": 2.007671626416412e-06,
      "loss": 0.5017299652099609,
      "step": 798
    },
    {
      "epoch": 3.187250996015936,
      "grad_norm": 0.19458836317062378,
      "learning_rate": 1.9890372616847663e-06,
      "loss": 0.7563250660896301,
      "step": 800
    },
    {
      "epoch": 3.195219123505976,
      "grad_norm": 0.10655785351991653,
      "learning_rate": 1.9705555124315965e-06,
      "loss": 0.6374018788337708,
      "step": 802
    },
    {
      "epoch": 3.2031872509960158,
      "grad_norm": 0.45915767550468445,
      "learning_rate": 1.95222718202658e-06,
      "loss": 0.5263140201568604,
      "step": 804
    },
    {
      "epoch": 3.2111553784860556,
      "grad_norm": 0.4690450131893158,
      "learning_rate": 1.934053067170545e-06,
      "loss": 0.936068594455719,
      "step": 806
    },
    {
      "epoch": 3.2191235059760954,
      "grad_norm": 0.313833087682724,
      "learning_rate": 1.9160339578608305e-06,
      "loss": 0.2009841352701187,
      "step": 808
    },
    {
      "epoch": 3.2270916334661353,
      "grad_norm": 0.17449450492858887,
      "learning_rate": 1.8981706373569602e-06,
      "loss": 0.5904903411865234,
      "step": 810
    },
    {
      "epoch": 3.235059760956175,
      "grad_norm": 0.24412080645561218,
      "learning_rate": 1.8804638821465821e-06,
      "loss": 0.2917124032974243,
      "step": 812
    },
    {
      "epoch": 3.243027888446215,
      "grad_norm": 0.11904601752758026,
      "learning_rate": 1.8629144619117234e-06,
      "loss": 0.7442460060119629,
      "step": 814
    },
    {
      "epoch": 3.2509960159362548,
      "grad_norm": 0.4469807744026184,
      "learning_rate": 1.8455231394953348e-06,
      "loss": 0.20837628841400146,
      "step": 816
    },
    {
      "epoch": 3.2589641434262946,
      "grad_norm": 0.14697785675525665,
      "learning_rate": 1.8282906708681238e-06,
      "loss": 0.8804487586021423,
      "step": 818
    },
    {
      "epoch": 3.2669322709163344,
      "grad_norm": 0.3428151607513428,
      "learning_rate": 1.811217805095699e-06,
      "loss": 0.40904954075813293,
      "step": 820
    },
    {
      "epoch": 3.2749003984063743,
      "grad_norm": 0.30418866872787476,
      "learning_rate": 1.7943052843060133e-06,
      "loss": 0.5646911263465881,
      "step": 822
    },
    {
      "epoch": 3.2828685258964145,
      "grad_norm": 0.559181809425354,
      "learning_rate": 1.7775538436570937e-06,
      "loss": 0.5167331099510193,
      "step": 824
    },
    {
      "epoch": 3.2908366533864544,
      "grad_norm": 0.11576046794652939,
      "learning_rate": 1.7609642113050996e-06,
      "loss": 0.8828203082084656,
      "step": 826
    },
    {
      "epoch": 3.298804780876494,
      "grad_norm": 0.15391217172145844,
      "learning_rate": 1.7445371083726567e-06,
      "loss": 0.8726562261581421,
      "step": 828
    },
    {
      "epoch": 3.306772908366534,
      "grad_norm": 0.308348149061203,
      "learning_rate": 1.7282732489175224e-06,
      "loss": 0.7919400930404663,
      "step": 830
    },
    {
      "epoch": 3.314741035856574,
      "grad_norm": 0.3492012023925781,
      "learning_rate": 1.7121733399015406e-06,
      "loss": 0.358115553855896,
      "step": 832
    },
    {
      "epoch": 3.3227091633466137,
      "grad_norm": 0.9149593114852905,
      "learning_rate": 1.6962380811599138e-06,
      "loss": 0.2333878129720688,
      "step": 834
    },
    {
      "epoch": 3.3306772908366535,
      "grad_norm": 0.4431096315383911,
      "learning_rate": 1.6804681653707816e-06,
      "loss": 0.4343220293521881,
      "step": 836
    },
    {
      "epoch": 3.3386454183266934,
      "grad_norm": 0.22197343409061432,
      "learning_rate": 1.6648642780251096e-06,
      "loss": 0.6893764734268188,
      "step": 838
    },
    {
      "epoch": 3.346613545816733,
      "grad_norm": 0.9335878491401672,
      "learning_rate": 1.6494270973968974e-06,
      "loss": 0.13649165630340576,
      "step": 840
    },
    {
      "epoch": 3.354581673306773,
      "grad_norm": 0.33430683612823486,
      "learning_rate": 1.6341572945136902e-06,
      "loss": 0.5570306181907654,
      "step": 842
    },
    {
      "epoch": 3.362549800796813,
      "grad_norm": 0.17486511170864105,
      "learning_rate": 1.6190555331274122e-06,
      "loss": 0.5539298057556152,
      "step": 844
    },
    {
      "epoch": 3.3705179282868527,
      "grad_norm": 0.14506401121616364,
      "learning_rate": 1.6041224696855145e-06,
      "loss": 0.9352858662605286,
      "step": 846
    },
    {
      "epoch": 3.3784860557768925,
      "grad_norm": 0.14617930352687836,
      "learning_rate": 1.589358753302439e-06,
      "loss": 0.86332106590271,
      "step": 848
    },
    {
      "epoch": 3.3864541832669324,
      "grad_norm": 0.4121112823486328,
      "learning_rate": 1.5747650257314068e-06,
      "loss": 0.8755100965499878,
      "step": 850
    },
    {
      "epoch": 3.394422310756972,
      "grad_norm": 0.22572061419487,
      "learning_rate": 1.5603419213365146e-06,
      "loss": 0.871116578578949,
      "step": 852
    },
    {
      "epoch": 3.402390438247012,
      "grad_norm": 0.18710310757160187,
      "learning_rate": 1.5460900670651705e-06,
      "loss": 0.5007573366165161,
      "step": 854
    },
    {
      "epoch": 3.410358565737052,
      "grad_norm": 0.21413084864616394,
      "learning_rate": 1.5320100824208306e-06,
      "loss": 0.8228604793548584,
      "step": 856
    },
    {
      "epoch": 3.4183266932270917,
      "grad_norm": 0.3382580876350403,
      "learning_rate": 1.51810257943608e-06,
      "loss": 0.9514821171760559,
      "step": 858
    },
    {
      "epoch": 3.4262948207171315,
      "grad_norm": 0.1686478853225708,
      "learning_rate": 1.5043681626460196e-06,
      "loss": 0.9232702851295471,
      "step": 860
    },
    {
      "epoch": 3.4342629482071714,
      "grad_norm": 0.09846412390470505,
      "learning_rate": 1.4908074290619972e-06,
      "loss": 0.5861298441886902,
      "step": 862
    },
    {
      "epoch": 3.442231075697211,
      "grad_norm": 0.33733096718788147,
      "learning_rate": 1.477420968145652e-06,
      "loss": 0.42202669382095337,
      "step": 864
    },
    {
      "epoch": 3.450199203187251,
      "grad_norm": 0.1436883807182312,
      "learning_rate": 1.4642093617832853e-06,
      "loss": 0.7252512574195862,
      "step": 866
    },
    {
      "epoch": 3.458167330677291,
      "grad_norm": 0.13490426540374756,
      "learning_rate": 1.4511731842605805e-06,
      "loss": 0.6275559663772583,
      "step": 868
    },
    {
      "epoch": 3.4661354581673307,
      "grad_norm": 0.1342659443616867,
      "learning_rate": 1.4383130022376298e-06,
      "loss": 0.47992435097694397,
      "step": 870
    },
    {
      "epoch": 3.4741035856573705,
      "grad_norm": 0.03528635576367378,
      "learning_rate": 1.4256293747243065e-06,
      "loss": 0.5249592065811157,
      "step": 872
    },
    {
      "epoch": 3.4820717131474104,
      "grad_norm": 0.06316667050123215,
      "learning_rate": 1.4131228530559637e-06,
      "loss": 0.599078357219696,
      "step": 874
    },
    {
      "epoch": 3.49003984063745,
      "grad_norm": 0.17602014541625977,
      "learning_rate": 1.400793980869468e-06,
      "loss": 0.47573012113571167,
      "step": 876
    },
    {
      "epoch": 3.49800796812749,
      "grad_norm": 0.24554605782032013,
      "learning_rate": 1.3886432940795744e-06,
      "loss": 0.48329198360443115,
      "step": 878
    },
    {
      "epoch": 3.50597609561753,
      "grad_norm": 0.05697697773575783,
      "learning_rate": 1.3766713208556232e-06,
      "loss": 0.3098355233669281,
      "step": 880
    },
    {
      "epoch": 3.5139442231075697,
      "grad_norm": 0.21454374492168427,
      "learning_rate": 1.3648785815985876e-06,
      "loss": 0.5117029547691345,
      "step": 882
    },
    {
      "epoch": 3.5219123505976095,
      "grad_norm": 0.1845294088125229,
      "learning_rate": 1.353265588918448e-06,
      "loss": 0.10415966808795929,
      "step": 884
    },
    {
      "epoch": 3.5298804780876494,
      "grad_norm": 0.2590247690677643,
      "learning_rate": 1.3418328476119136e-06,
      "loss": 0.7016832828521729,
      "step": 886
    },
    {
      "epoch": 3.537848605577689,
      "grad_norm": 0.2117038518190384,
      "learning_rate": 1.330580854640477e-06,
      "loss": 0.8351197838783264,
      "step": 888
    },
    {
      "epoch": 3.545816733067729,
      "grad_norm": 0.11664468050003052,
      "learning_rate": 1.3195100991088142e-06,
      "loss": 0.458038330078125,
      "step": 890
    },
    {
      "epoch": 3.553784860557769,
      "grad_norm": 0.38587722182273865,
      "learning_rate": 1.308621062243524e-06,
      "loss": 0.7673805356025696,
      "step": 892
    },
    {
      "epoch": 3.5617529880478087,
      "grad_norm": 0.12609800696372986,
      "learning_rate": 1.2979142173722054e-06,
      "loss": 0.9808838367462158,
      "step": 894
    },
    {
      "epoch": 3.5697211155378485,
      "grad_norm": 0.31664928793907166,
      "learning_rate": 1.2873900299028896e-06,
      "loss": 0.18043935298919678,
      "step": 896
    },
    {
      "epoch": 3.5776892430278884,
      "grad_norm": 0.16440443694591522,
      "learning_rate": 1.2770489573038073e-06,
      "loss": 0.5510551333427429,
      "step": 898
    },
    {
      "epoch": 3.585657370517928,
      "grad_norm": 0.19209904968738556,
      "learning_rate": 1.2668914490834983e-06,
      "loss": 0.16328421235084534,
      "step": 900
    },
    {
      "epoch": 3.593625498007968,
      "grad_norm": 0.2616488039493561,
      "learning_rate": 1.2569179467712805e-06,
      "loss": 0.07121944427490234,
      "step": 902
    },
    {
      "epoch": 3.601593625498008,
      "grad_norm": 0.17343591153621674,
      "learning_rate": 1.2471288838980505e-06,
      "loss": 0.6597598791122437,
      "step": 904
    },
    {
      "epoch": 3.6095617529880477,
      "grad_norm": 0.13453447818756104,
      "learning_rate": 1.237524685977444e-06,
      "loss": 0.536332905292511,
      "step": 906
    },
    {
      "epoch": 3.6175298804780875,
      "grad_norm": 0.1287977248430252,
      "learning_rate": 1.2281057704873331e-06,
      "loss": 0.8159908056259155,
      "step": 908
    },
    {
      "epoch": 3.6254980079681274,
      "grad_norm": 0.3556680381298065,
      "learning_rate": 1.2188725468516876e-06,
      "loss": 0.7484529614448547,
      "step": 910
    },
    {
      "epoch": 3.633466135458167,
      "grad_norm": 0.06616581231355667,
      "learning_rate": 1.2098254164227706e-06,
      "loss": 0.029973143711686134,
      "step": 912
    },
    {
      "epoch": 3.641434262948207,
      "grad_norm": 0.4066866636276245,
      "learning_rate": 1.200964772463695e-06,
      "loss": 0.6849876642227173,
      "step": 914
    },
    {
      "epoch": 3.649402390438247,
      "grad_norm": 0.27680647373199463,
      "learning_rate": 1.1922910001313312e-06,
      "loss": 0.6810500025749207,
      "step": 916
    },
    {
      "epoch": 3.6573705179282867,
      "grad_norm": 0.645796537399292,
      "learning_rate": 1.1838044764595617e-06,
      "loss": 0.11293944716453552,
      "step": 918
    },
    {
      "epoch": 3.6653386454183265,
      "grad_norm": 0.144248366355896,
      "learning_rate": 1.1755055703428958e-06,
      "loss": 0.8695544600486755,
      "step": 920
    },
    {
      "epoch": 3.6733067729083664,
      "grad_norm": 0.22124017775058746,
      "learning_rate": 1.1673946425204288e-06,
      "loss": 0.1105734258890152,
      "step": 922
    },
    {
      "epoch": 3.681274900398406,
      "grad_norm": 0.5181828737258911,
      "learning_rate": 1.1594720455601664e-06,
      "loss": 0.8183557391166687,
      "step": 924
    },
    {
      "epoch": 3.6892430278884465,
      "grad_norm": 0.25619664788246155,
      "learning_rate": 1.1517381238436986e-06,
      "loss": 0.21246881783008575,
      "step": 926
    },
    {
      "epoch": 3.6972111553784863,
      "grad_norm": 0.13301387429237366,
      "learning_rate": 1.1441932135512262e-06,
      "loss": 0.48757287859916687,
      "step": 928
    },
    {
      "epoch": 3.705179282868526,
      "grad_norm": 0.09774433076381683,
      "learning_rate": 1.1368376426469526e-06,
      "loss": 0.7442837953567505,
      "step": 930
    },
    {
      "epoch": 3.713147410358566,
      "grad_norm": 0.2561400830745697,
      "learning_rate": 1.1296717308648237e-06,
      "loss": 0.8765264749526978,
      "step": 932
    },
    {
      "epoch": 3.721115537848606,
      "grad_norm": 0.564517080783844,
      "learning_rate": 1.1226957896946332e-06,
      "loss": 0.41935089230537415,
      "step": 934
    },
    {
      "epoch": 3.7290836653386457,
      "grad_norm": 0.220582976937294,
      "learning_rate": 1.115910122368479e-06,
      "loss": 0.7852447032928467,
      "step": 936
    },
    {
      "epoch": 3.7370517928286855,
      "grad_norm": 0.2987324297428131,
      "learning_rate": 1.1093150238475847e-06,
      "loss": 0.40063056349754333,
      "step": 938
    },
    {
      "epoch": 3.7450199203187253,
      "grad_norm": 0.11963716149330139,
      "learning_rate": 1.1029107808094789e-06,
      "loss": 0.7047469615936279,
      "step": 940
    },
    {
      "epoch": 3.752988047808765,
      "grad_norm": 0.8242963552474976,
      "learning_rate": 1.0966976716355294e-06,
      "loss": 0.5519139170646667,
      "step": 942
    },
    {
      "epoch": 3.760956175298805,
      "grad_norm": 0.19697557389736176,
      "learning_rate": 1.0906759663988475e-06,
      "loss": 0.48083430528640747,
      "step": 944
    },
    {
      "epoch": 3.768924302788845,
      "grad_norm": 0.1293589323759079,
      "learning_rate": 1.084845926852547e-06,
      "loss": 0.9226465821266174,
      "step": 946
    },
    {
      "epoch": 3.7768924302788847,
      "grad_norm": 0.17823494970798492,
      "learning_rate": 1.0792078064183652e-06,
      "loss": 0.5296029448509216,
      "step": 948
    },
    {
      "epoch": 3.7848605577689245,
      "grad_norm": 0.11703119426965714,
      "learning_rate": 1.0737618501756455e-06,
      "loss": 0.888701319694519,
      "step": 950
    },
    {
      "epoch": 3.7928286852589643,
      "grad_norm": 0.30720436573028564,
      "learning_rate": 1.0685082948506875e-06,
      "loss": 0.8863480687141418,
      "step": 952
    },
    {
      "epoch": 3.800796812749004,
      "grad_norm": 0.14865604043006897,
      "learning_rate": 1.063447368806457e-06,
      "loss": 0.46968895196914673,
      "step": 954
    },
    {
      "epoch": 3.808764940239044,
      "grad_norm": 0.27862486243247986,
      "learning_rate": 1.0585792920326565e-06,
      "loss": 0.6574747562408447,
      "step": 956
    },
    {
      "epoch": 3.816733067729084,
      "grad_norm": 0.8925970792770386,
      "learning_rate": 1.0539042761361646e-06,
      "loss": 0.4635286331176758,
      "step": 958
    },
    {
      "epoch": 3.8247011952191237,
      "grad_norm": 0.14167918264865875,
      "learning_rate": 1.0494225243318373e-06,
      "loss": 0.5599319934844971,
      "step": 960
    },
    {
      "epoch": 3.8326693227091635,
      "grad_norm": 0.06606796383857727,
      "learning_rate": 1.0451342314336755e-06,
      "loss": 0.5078768134117126,
      "step": 962
    },
    {
      "epoch": 3.8406374501992033,
      "grad_norm": 0.21200227737426758,
      "learning_rate": 1.0410395838463547e-06,
      "loss": 0.9009104371070862,
      "step": 964
    },
    {
      "epoch": 3.848605577689243,
      "grad_norm": 8.519966125488281,
      "learning_rate": 1.0371387595571248e-06,
      "loss": 0.29693469405174255,
      "step": 966
    },
    {
      "epoch": 3.856573705179283,
      "grad_norm": 0.34527188539505005,
      "learning_rate": 1.0334319281280715e-06,
      "loss": 0.5275134444236755,
      "step": 968
    },
    {
      "epoch": 3.864541832669323,
      "grad_norm": 0.19093593955039978,
      "learning_rate": 1.0299192506887467e-06,
      "loss": 0.6093609929084778,
      "step": 970
    },
    {
      "epoch": 3.8725099601593627,
      "grad_norm": 0.3833182454109192,
      "learning_rate": 1.0266008799291635e-06,
      "loss": 0.4910091459751129,
      "step": 972
    },
    {
      "epoch": 3.8804780876494025,
      "grad_norm": 0.05908047407865524,
      "learning_rate": 1.0234769600931612e-06,
      "loss": 0.39339911937713623,
      "step": 974
    },
    {
      "epoch": 3.8884462151394423,
      "grad_norm": 0.18856295943260193,
      "learning_rate": 1.0205476269721312e-06,
      "loss": 0.8877154588699341,
      "step": 976
    },
    {
      "epoch": 3.896414342629482,
      "grad_norm": 0.24654258787631989,
      "learning_rate": 1.0178130078991196e-06,
      "loss": 0.4859975576400757,
      "step": 978
    },
    {
      "epoch": 3.904382470119522,
      "grad_norm": 0.07047562301158905,
      "learning_rate": 1.0152732217432875e-06,
      "loss": 0.3432178497314453,
      "step": 980
    },
    {
      "epoch": 3.912350597609562,
      "grad_norm": 0.17790663242340088,
      "learning_rate": 1.0129283789047473e-06,
      "loss": 0.38312843441963196,
      "step": 982
    },
    {
      "epoch": 3.9203187250996017,
      "grad_norm": 0.5323594212532043,
      "learning_rate": 1.0107785813097621e-06,
      "loss": 0.8315017819404602,
      "step": 984
    },
    {
      "epoch": 3.9282868525896415,
      "grad_norm": 0.2661547064781189,
      "learning_rate": 1.0088239224063153e-06,
      "loss": 0.09976834803819656,
      "step": 986
    },
    {
      "epoch": 3.9362549800796813,
      "grad_norm": 0.32426854968070984,
      "learning_rate": 1.0070644871600495e-06,
      "loss": 0.5638818144798279,
      "step": 988
    },
    {
      "epoch": 3.944223107569721,
      "grad_norm": 0.06352943181991577,
      "learning_rate": 1.0055003520505714e-06,
      "loss": 0.5082998275756836,
      "step": 990
    },
    {
      "epoch": 3.952191235059761,
      "grad_norm": 0.27524417638778687,
      "learning_rate": 1.00413158506813e-06,
      "loss": 0.8891046047210693,
      "step": 992
    },
    {
      "epoch": 3.960159362549801,
      "grad_norm": 0.03673136979341507,
      "learning_rate": 1.0029582457106588e-06,
      "loss": 0.18685656785964966,
      "step": 994
    },
    {
      "epoch": 3.9681274900398407,
      "grad_norm": 0.2767471671104431,
      "learning_rate": 1.00198038498119e-06,
      "loss": 0.7628071308135986,
      "step": 996
    },
    {
      "epoch": 3.9760956175298805,
      "grad_norm": 0.07721873372793198,
      "learning_rate": 1.0011980453856391e-06,
      "loss": 0.5093322396278381,
      "step": 998
    },
    {
      "epoch": 3.9840637450199203,
      "grad_norm": 0.20198450982570648,
      "learning_rate": 1.0006112609309548e-06,
      "loss": 0.3758847713470459,
      "step": 1000
    },
    {
      "epoch": 3.99203187250996,
      "grad_norm": 0.25387924909591675,
      "learning_rate": 1.0002200571236429e-06,
      "loss": 0.8914694786071777,
      "step": 1002
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.06263309717178345,
      "learning_rate": 1.0000244509686564e-06,
      "loss": 0.24155621230602264,
      "step": 1004
    },
    {
      "epoch": 4.0,
      "step": 1004,
      "total_flos": 3.954132003103703e+18,
      "train_loss": 0.8280324278013461,
      "train_runtime": 34497.9796,
      "train_samples_per_second": 1.746,
      "train_steps_per_second": 0.029
    }
  ],
  "logging_steps": 2,
  "max_steps": 1004,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 99999,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.954132003103703e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}