{
  "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": 0.9316700100898743,
      "learning_rate": 1.9607843137254904e-07,
      "loss": 2.093052625656128,
      "step": 2
    },
    {
      "epoch": 0.01593625498007968,
      "grad_norm": 0.6678388118743896,
      "learning_rate": 5.882352941176471e-07,
      "loss": 2.01019287109375,
      "step": 4
    },
    {
      "epoch": 0.02390438247011952,
      "grad_norm": 2.5262320041656494,
      "learning_rate": 9.80392156862745e-07,
      "loss": 2.179050922393799,
      "step": 6
    },
    {
      "epoch": 0.03187250996015936,
      "grad_norm": 1.2122697830200195,
      "learning_rate": 1.3725490196078434e-06,
      "loss": 1.8842616081237793,
      "step": 8
    },
    {
      "epoch": 0.0398406374501992,
      "grad_norm": 1.7647368907928467,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 1.463491678237915,
      "step": 10
    },
    {
      "epoch": 0.04780876494023904,
      "grad_norm": 2.4031553268432617,
      "learning_rate": 2.1568627450980393e-06,
      "loss": 1.692115068435669,
      "step": 12
    },
    {
      "epoch": 0.055776892430278883,
      "grad_norm": 0.7320537567138672,
      "learning_rate": 2.549019607843137e-06,
      "loss": 2.267002582550049,
      "step": 14
    },
    {
      "epoch": 0.06374501992031872,
      "grad_norm": 1.8810019493103027,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 2.115994453430176,
      "step": 16
    },
    {
      "epoch": 0.07171314741035857,
      "grad_norm": 2.6569278240203857,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.6094053983688354,
      "step": 18
    },
    {
      "epoch": 0.0796812749003984,
      "grad_norm": 0.39756685495376587,
      "learning_rate": 3.7254901960784316e-06,
      "loss": 1.538004994392395,
      "step": 20
    },
    {
      "epoch": 0.08764940239043825,
      "grad_norm": 1.1743954420089722,
      "learning_rate": 4.11764705882353e-06,
      "loss": 1.1438497304916382,
      "step": 22
    },
    {
      "epoch": 0.09561752988047809,
      "grad_norm": 0.7506561875343323,
      "learning_rate": 4.509803921568628e-06,
      "loss": 1.3503583669662476,
      "step": 24
    },
    {
      "epoch": 0.10358565737051793,
      "grad_norm": 0.6740713715553284,
      "learning_rate": 4.901960784313726e-06,
      "loss": 1.43703293800354,
      "step": 26
    },
    {
      "epoch": 0.11155378486055777,
      "grad_norm": 1.1073075532913208,
      "learning_rate": 5.294117647058824e-06,
      "loss": 1.6244215965270996,
      "step": 28
    },
    {
      "epoch": 0.11952191235059761,
      "grad_norm": 0.9537054896354675,
      "learning_rate": 5.686274509803922e-06,
      "loss": 1.3440543413162231,
      "step": 30
    },
    {
      "epoch": 0.12749003984063745,
      "grad_norm": 0.6916010975837708,
      "learning_rate": 6.07843137254902e-06,
      "loss": 1.1549923419952393,
      "step": 32
    },
    {
      "epoch": 0.13545816733067728,
      "grad_norm": 2.0239624977111816,
      "learning_rate": 6.470588235294119e-06,
      "loss": 1.251583218574524,
      "step": 34
    },
    {
      "epoch": 0.14342629482071714,
      "grad_norm": 0.5752863883972168,
      "learning_rate": 6.862745098039216e-06,
      "loss": 1.4084151983261108,
      "step": 36
    },
    {
      "epoch": 0.15139442231075698,
      "grad_norm": 1.2080317735671997,
      "learning_rate": 7.2549019607843145e-06,
      "loss": 1.1945853233337402,
      "step": 38
    },
    {
      "epoch": 0.1593625498007968,
      "grad_norm": 0.36280813813209534,
      "learning_rate": 7.647058823529411e-06,
      "loss": 1.3136399984359741,
      "step": 40
    },
    {
      "epoch": 0.16733067729083664,
      "grad_norm": 2.675469160079956,
      "learning_rate": 8.03921568627451e-06,
      "loss": 0.7877538204193115,
      "step": 42
    },
    {
      "epoch": 0.1752988047808765,
      "grad_norm": 0.6141120195388794,
      "learning_rate": 8.43137254901961e-06,
      "loss": 0.9838396906852722,
      "step": 44
    },
    {
      "epoch": 0.18326693227091634,
      "grad_norm": 1.0343557596206665,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.9531049132347107,
      "step": 46
    },
    {
      "epoch": 0.19123505976095617,
      "grad_norm": 0.8948782086372375,
      "learning_rate": 9.215686274509804e-06,
      "loss": 1.1812816858291626,
      "step": 48
    },
    {
      "epoch": 0.199203187250996,
      "grad_norm": 0.5856284499168396,
      "learning_rate": 9.607843137254903e-06,
      "loss": 1.3345295190811157,
      "step": 50
    },
    {
      "epoch": 0.20717131474103587,
      "grad_norm": 0.5155065655708313,
      "learning_rate": 1e-05,
      "loss": 1.2863095998764038,
      "step": 52
    },
    {
      "epoch": 0.2151394422310757,
      "grad_norm": 3.8382720947265625,
      "learning_rate": 9.999902196391087e-06,
      "loss": 1.0157941579818726,
      "step": 54
    },
    {
      "epoch": 0.22310756972111553,
      "grad_norm": 0.6327527165412903,
      "learning_rate": 9.999608789815699e-06,
      "loss": 1.527570128440857,
      "step": 56
    },
    {
      "epoch": 0.23107569721115537,
      "grad_norm": 0.42920202016830444,
      "learning_rate": 9.999119793027713e-06,
      "loss": 1.3420095443725586,
      "step": 58
    },
    {
      "epoch": 0.23904382470119523,
      "grad_norm": 0.4539412260055542,
      "learning_rate": 9.998435227282975e-06,
      "loss": 0.7585218548774719,
      "step": 60
    },
    {
      "epoch": 0.24701195219123506,
      "grad_norm": 0.33086708188056946,
      "learning_rate": 9.997555122338372e-06,
      "loss": 1.3439724445343018,
      "step": 62
    },
    {
      "epoch": 0.2549800796812749,
      "grad_norm": 0.5372070670127869,
      "learning_rate": 9.996479516450544e-06,
      "loss": 1.1140345335006714,
      "step": 64
    },
    {
      "epoch": 0.26294820717131473,
      "grad_norm": 0.43935078382492065,
      "learning_rate": 9.99520845637422e-06,
      "loss": 1.2779158353805542,
      "step": 66
    },
    {
      "epoch": 0.27091633466135456,
      "grad_norm": 0.5353237986564636,
      "learning_rate": 9.993741997360185e-06,
      "loss": 0.8180304169654846,
      "step": 68
    },
    {
      "epoch": 0.2788844621513944,
      "grad_norm": 0.6237613558769226,
      "learning_rate": 9.992080203152872e-06,
      "loss": 1.565225601196289,
      "step": 70
    },
    {
      "epoch": 0.2868525896414343,
      "grad_norm": 0.7996062636375427,
      "learning_rate": 9.990223145987608e-06,
      "loss": 0.8221546411514282,
      "step": 72
    },
    {
      "epoch": 0.2948207171314741,
      "grad_norm": 2.544907331466675,
      "learning_rate": 9.98817090658745e-06,
      "loss": 1.1499135494232178,
      "step": 74
    },
    {
      "epoch": 0.30278884462151395,
      "grad_norm": 0.35762548446655273,
      "learning_rate": 9.985923574159696e-06,
      "loss": 1.4618713855743408,
      "step": 76
    },
    {
      "epoch": 0.3107569721115538,
      "grad_norm": 1.1268619298934937,
      "learning_rate": 9.983481246392003e-06,
      "loss": 0.9892376065254211,
      "step": 78
    },
    {
      "epoch": 0.3187250996015936,
      "grad_norm": 0.38517534732818604,
      "learning_rate": 9.980844029448135e-06,
      "loss": 0.9308754205703735,
      "step": 80
    },
    {
      "epoch": 0.32669322709163345,
      "grad_norm": 0.2714245915412903,
      "learning_rate": 9.97801203796335e-06,
      "loss": 1.3360328674316406,
      "step": 82
    },
    {
      "epoch": 0.3346613545816733,
      "grad_norm": 1.8015745878219604,
      "learning_rate": 9.974985395039423e-06,
      "loss": 0.9634675979614258,
      "step": 84
    },
    {
      "epoch": 0.3426294820717131,
      "grad_norm": 0.6544349193572998,
      "learning_rate": 9.971764232239287e-06,
      "loss": 1.2982206344604492,
      "step": 86
    },
    {
      "epoch": 0.350597609561753,
      "grad_norm": 0.6585408449172974,
      "learning_rate": 9.968348689581315e-06,
      "loss": 0.9192724227905273,
      "step": 88
    },
    {
      "epoch": 0.35856573705179284,
      "grad_norm": 0.4878370463848114,
      "learning_rate": 9.964738915533242e-06,
      "loss": 1.3063905239105225,
      "step": 90
    },
    {
      "epoch": 0.3665338645418327,
      "grad_norm": 0.6690500974655151,
      "learning_rate": 9.960935067005706e-06,
      "loss": 1.1153125762939453,
      "step": 92
    },
    {
      "epoch": 0.3745019920318725,
      "grad_norm": 0.4258851706981659,
      "learning_rate": 9.956937309345418e-06,
      "loss": 1.2281380891799927,
      "step": 94
    },
    {
      "epoch": 0.38247011952191234,
      "grad_norm": 0.8225271701812744,
      "learning_rate": 9.952745816327998e-06,
      "loss": 1.0963788032531738,
      "step": 96
    },
    {
      "epoch": 0.3904382470119522,
      "grad_norm": 0.5437995791435242,
      "learning_rate": 9.94836077015039e-06,
      "loss": 1.2456581592559814,
      "step": 98
    },
    {
      "epoch": 0.398406374501992,
      "grad_norm": 0.33754414319992065,
      "learning_rate": 9.943782361422975e-06,
      "loss": 0.867386519908905,
      "step": 100
    },
    {
      "epoch": 0.4063745019920319,
      "grad_norm": 0.5295922756195068,
      "learning_rate": 9.939010789161259e-06,
      "loss": 1.257737398147583,
      "step": 102
    },
    {
      "epoch": 0.41434262948207173,
      "grad_norm": 0.4011881351470947,
      "learning_rate": 9.934046260777239e-06,
      "loss": 0.7846571207046509,
      "step": 104
    },
    {
      "epoch": 0.42231075697211157,
      "grad_norm": 2.200401544570923,
      "learning_rate": 9.928888992070373e-06,
      "loss": 0.7390787601470947,
      "step": 106
    },
    {
      "epoch": 0.4302788844621514,
      "grad_norm": 0.9516831636428833,
      "learning_rate": 9.923539207218219e-06,
      "loss": 0.9987615346908569,
      "step": 108
    },
    {
      "epoch": 0.43824701195219123,
      "grad_norm": 0.3815825581550598,
      "learning_rate": 9.91799713876667e-06,
      "loss": 1.2863091230392456,
      "step": 110
    },
    {
      "epoch": 0.44621513944223107,
      "grad_norm": 0.5752545595169067,
      "learning_rate": 9.912263027619856e-06,
      "loss": 0.8385196328163147,
      "step": 112
    },
    {
      "epoch": 0.4541832669322709,
      "grad_norm": 0.730549156665802,
      "learning_rate": 9.906337123029675e-06,
      "loss": 1.1393396854400635,
      "step": 114
    },
    {
      "epoch": 0.46215139442231074,
      "grad_norm": 0.39685970544815063,
      "learning_rate": 9.90021968258495e-06,
      "loss": 1.2790663242340088,
      "step": 116
    },
    {
      "epoch": 0.4701195219123506,
      "grad_norm": 0.6363944411277771,
      "learning_rate": 9.893910972200238e-06,
      "loss": 1.339137077331543,
      "step": 118
    },
    {
      "epoch": 0.47808764940239046,
      "grad_norm": 0.7425268292427063,
      "learning_rate": 9.887411266104268e-06,
      "loss": 1.2219758033752441,
      "step": 120
    },
    {
      "epoch": 0.4860557768924303,
      "grad_norm": 0.7261989712715149,
      "learning_rate": 9.880720846828026e-06,
      "loss": 0.9617432951927185,
      "step": 122
    },
    {
      "epoch": 0.4940239043824701,
      "grad_norm": 0.5349416732788086,
      "learning_rate": 9.873840005192466e-06,
      "loss": 1.0352855920791626,
      "step": 124
    },
    {
      "epoch": 0.50199203187251,
      "grad_norm": 0.7772267460823059,
      "learning_rate": 9.866769040295875e-06,
      "loss": 0.9337604641914368,
      "step": 126
    },
    {
      "epoch": 0.5099601593625498,
      "grad_norm": 0.41817694902420044,
      "learning_rate": 9.859508259500869e-06,
      "loss": 1.2733639478683472,
      "step": 128
    },
    {
      "epoch": 0.5179282868525896,
      "grad_norm": 0.5852633714675903,
      "learning_rate": 9.852057978421033e-06,
      "loss": 0.9402267336845398,
      "step": 130
    },
    {
      "epoch": 0.5258964143426295,
      "grad_norm": 0.44234931468963623,
      "learning_rate": 9.844418520907198e-06,
      "loss": 0.7868728637695312,
      "step": 132
    },
    {
      "epoch": 0.5338645418326693,
      "grad_norm": 0.36357221007347107,
      "learning_rate": 9.836590219033375e-06,
      "loss": 0.9318094253540039,
      "step": 134
    },
    {
      "epoch": 0.5418326693227091,
      "grad_norm": 0.7532193064689636,
      "learning_rate": 9.828573413082306e-06,
      "loss": 1.2890671491622925,
      "step": 136
    },
    {
      "epoch": 0.549800796812749,
      "grad_norm": 0.5671216249465942,
      "learning_rate": 9.820368451530681e-06,
      "loss": 0.9606557488441467,
      "step": 138
    },
    {
      "epoch": 0.5577689243027888,
      "grad_norm": 0.6483688354492188,
      "learning_rate": 9.811975691033992e-06,
      "loss": 1.0854955911636353,
      "step": 140
    },
    {
      "epoch": 0.5657370517928287,
      "grad_norm": 0.4566214680671692,
      "learning_rate": 9.80339549641102e-06,
      "loss": 1.0139864683151245,
      "step": 142
    },
    {
      "epoch": 0.5737051792828686,
      "grad_norm": 0.5953824520111084,
      "learning_rate": 9.794628240627992e-06,
      "loss": 1.2217941284179688,
      "step": 144
    },
    {
      "epoch": 0.5816733067729084,
      "grad_norm": 0.6513963341712952,
      "learning_rate": 9.785674304782349e-06,
      "loss": 1.3207577466964722,
      "step": 146
    },
    {
      "epoch": 0.5896414342629482,
      "grad_norm": 0.5430207252502441,
      "learning_rate": 9.776534078086202e-06,
      "loss": 1.0061651468276978,
      "step": 148
    },
    {
      "epoch": 0.5976095617529881,
      "grad_norm": 0.5469725131988525,
      "learning_rate": 9.767207957849396e-06,
      "loss": 0.9545348286628723,
      "step": 150
    },
    {
      "epoch": 0.6055776892430279,
      "grad_norm": 0.790772020816803,
      "learning_rate": 9.757696349462252e-06,
      "loss": 0.9549607038497925,
      "step": 152
    },
    {
      "epoch": 0.6135458167330677,
      "grad_norm": 0.8317003846168518,
      "learning_rate": 9.747999666377934e-06,
      "loss": 1.1262625455856323,
      "step": 154
    },
    {
      "epoch": 0.6215139442231076,
      "grad_norm": 4.543304920196533,
      "learning_rate": 9.738118330094491e-06,
      "loss": 0.9728249311447144,
      "step": 156
    },
    {
      "epoch": 0.6294820717131474,
      "grad_norm": 1.140038013458252,
      "learning_rate": 9.728052770136518e-06,
      "loss": 1.1600499153137207,
      "step": 158
    },
    {
      "epoch": 0.6374501992031872,
      "grad_norm": 0.3960021436214447,
      "learning_rate": 9.717803424036501e-06,
      "loss": 1.0158053636550903,
      "step": 160
    },
    {
      "epoch": 0.6454183266932271,
      "grad_norm": 1.2506061792373657,
      "learning_rate": 9.707370737315793e-06,
      "loss": 0.7808167934417725,
      "step": 162
    },
    {
      "epoch": 0.6533864541832669,
      "grad_norm": 0.32601407170295715,
      "learning_rate": 9.69675516346524e-06,
      "loss": 1.1591135263442993,
      "step": 164
    },
    {
      "epoch": 0.6613545816733067,
      "grad_norm": 0.4265863299369812,
      "learning_rate": 9.685957163925479e-06,
      "loss": 0.8330628871917725,
      "step": 166
    },
    {
      "epoch": 0.6693227091633466,
      "grad_norm": 0.2724871337413788,
      "learning_rate": 9.67497720806688e-06,
      "loss": 1.2323929071426392,
      "step": 168
    },
    {
      "epoch": 0.6772908366533864,
      "grad_norm": 1.6088201999664307,
      "learning_rate": 9.663815773169131e-06,
      "loss": 0.9288667440414429,
      "step": 170
    },
    {
      "epoch": 0.6852589641434262,
      "grad_norm": 0.4231555759906769,
      "learning_rate": 9.652473344400507e-06,
      "loss": 0.8601210117340088,
      "step": 172
    },
    {
      "epoch": 0.6932270916334662,
      "grad_norm": 0.4612506628036499,
      "learning_rate": 9.640950414796772e-06,
      "loss": 1.225210428237915,
      "step": 174
    },
    {
      "epoch": 0.701195219123506,
      "grad_norm": 1.6827380657196045,
      "learning_rate": 9.629247485239747e-06,
      "loss": 0.7526090741157532,
      "step": 176
    },
    {
      "epoch": 0.7091633466135459,
      "grad_norm": 0.7703327536582947,
      "learning_rate": 9.617365064435544e-06,
      "loss": 0.9219900369644165,
      "step": 178
    },
    {
      "epoch": 0.7171314741035857,
      "grad_norm": 0.7876484394073486,
      "learning_rate": 9.605303668892443e-06,
      "loss": 1.0643880367279053,
      "step": 180
    },
    {
      "epoch": 0.7250996015936255,
      "grad_norm": 0.25483438372612,
      "learning_rate": 9.593063822898451e-06,
      "loss": 0.505620539188385,
      "step": 182
    },
    {
      "epoch": 0.7330677290836654,
      "grad_norm": 0.3652317225933075,
      "learning_rate": 9.580646058498505e-06,
      "loss": 1.296199083328247,
      "step": 184
    },
    {
      "epoch": 0.7410358565737052,
      "grad_norm": 0.27735066413879395,
      "learning_rate": 9.56805091547135e-06,
      "loss": 1.251572608947754,
      "step": 186
    },
    {
      "epoch": 0.749003984063745,
      "grad_norm": 0.3912353515625,
      "learning_rate": 9.55527894130607e-06,
      "loss": 1.1036205291748047,
      "step": 188
    },
    {
      "epoch": 0.7569721115537849,
      "grad_norm": 0.37258675694465637,
      "learning_rate": 9.542330691178296e-06,
      "loss": 0.8744907975196838,
      "step": 190
    },
    {
      "epoch": 0.7649402390438247,
      "grad_norm": 0.4719068706035614,
      "learning_rate": 9.529206727926071e-06,
      "loss": 1.2135030031204224,
      "step": 192
    },
    {
      "epoch": 0.7729083665338645,
      "grad_norm": 0.3896011412143707,
      "learning_rate": 9.515907622025376e-06,
      "loss": 0.8844252824783325,
      "step": 194
    },
    {
      "epoch": 0.7808764940239044,
      "grad_norm": 0.8427772521972656,
      "learning_rate": 9.502433951565347e-06,
      "loss": 0.6156318187713623,
      "step": 196
    },
    {
      "epoch": 0.7888446215139442,
      "grad_norm": 1.4091371297836304,
      "learning_rate": 9.48878630222314e-06,
      "loss": 0.772961437702179,
      "step": 198
    },
    {
      "epoch": 0.796812749003984,
      "grad_norm": 0.5533599853515625,
      "learning_rate": 9.474965267238469e-06,
      "loss": 1.0491026639938354,
      "step": 200
    },
    {
      "epoch": 0.8047808764940239,
      "grad_norm": 0.4289070665836334,
      "learning_rate": 9.460971447387824e-06,
      "loss": 0.9913599491119385,
      "step": 202
    },
    {
      "epoch": 0.8127490039840638,
      "grad_norm": 1.0124455690383911,
      "learning_rate": 9.446805450958353e-06,
      "loss": 1.5288549661636353,
      "step": 204
    },
    {
      "epoch": 0.8207171314741036,
      "grad_norm": 0.2913559377193451,
      "learning_rate": 9.43246789372142e-06,
      "loss": 1.2301074266433716,
      "step": 206
    },
    {
      "epoch": 0.8286852589641435,
      "grad_norm": 0.5000865459442139,
      "learning_rate": 9.417959398905846e-06,
      "loss": 0.6086893677711487,
      "step": 208
    },
    {
      "epoch": 0.8366533864541833,
      "grad_norm": 0.33062100410461426,
      "learning_rate": 9.40328059717081e-06,
      "loss": 1.2701423168182373,
      "step": 210
    },
    {
      "epoch": 0.8446215139442231,
      "grad_norm": 0.6745465397834778,
      "learning_rate": 9.388432126578438e-06,
      "loss": 1.213331937789917,
      "step": 212
    },
    {
      "epoch": 0.852589641434263,
      "grad_norm": 0.3171221911907196,
      "learning_rate": 9.373414632566066e-06,
      "loss": 0.9477609395980835,
      "step": 214
    },
    {
      "epoch": 0.8605577689243028,
      "grad_norm": 0.8724876046180725,
      "learning_rate": 9.35822876791819e-06,
      "loss": 1.430282711982727,
      "step": 216
    },
    {
      "epoch": 0.8685258964143426,
      "grad_norm": 1.122143030166626,
      "learning_rate": 9.342875192738083e-06,
      "loss": 0.8672413229942322,
      "step": 218
    },
    {
      "epoch": 0.8764940239043825,
      "grad_norm": 0.3571845293045044,
      "learning_rate": 9.327354574419109e-06,
      "loss": 1.2346898317337036,
      "step": 220
    },
    {
      "epoch": 0.8844621513944223,
      "grad_norm": 0.6231569051742554,
      "learning_rate": 9.311667587615701e-06,
      "loss": 1.4827804565429688,
      "step": 222
    },
    {
      "epoch": 0.8924302788844621,
      "grad_norm": 1.1705824136734009,
      "learning_rate": 9.29581491421405e-06,
      "loss": 0.7086166143417358,
      "step": 224
    },
    {
      "epoch": 0.900398406374502,
      "grad_norm": 0.3164781928062439,
      "learning_rate": 9.279797243302453e-06,
      "loss": 1.005565881729126,
      "step": 226
    },
    {
      "epoch": 0.9083665338645418,
      "grad_norm": 1.2484079599380493,
      "learning_rate": 9.263615271141362e-06,
      "loss": 1.093484878540039,
      "step": 228
    },
    {
      "epoch": 0.9163346613545816,
      "grad_norm": 0.5352246165275574,
      "learning_rate": 9.247269701133125e-06,
      "loss": 1.1812286376953125,
      "step": 230
    },
    {
      "epoch": 0.9243027888446215,
      "grad_norm": 0.6244534850120544,
      "learning_rate": 9.230761243791401e-06,
      "loss": 1.3227640390396118,
      "step": 232
    },
    {
      "epoch": 0.9322709163346613,
      "grad_norm": 0.3455016016960144,
      "learning_rate": 9.214090616710282e-06,
      "loss": 0.8153970837593079,
      "step": 234
    },
    {
      "epoch": 0.9402390438247012,
      "grad_norm": 0.2861965596675873,
      "learning_rate": 9.197258544533098e-06,
      "loss": 0.8405250310897827,
      "step": 236
    },
    {
      "epoch": 0.9482071713147411,
      "grad_norm": 3.556896209716797,
      "learning_rate": 9.180265758920916e-06,
      "loss": 0.4659785032272339,
      "step": 238
    },
    {
      "epoch": 0.9561752988047809,
      "grad_norm": 0.9116905331611633,
      "learning_rate": 9.163112998520741e-06,
      "loss": 0.8530660271644592,
      "step": 240
    },
    {
      "epoch": 0.9641434262948207,
      "grad_norm": 0.5339516997337341,
      "learning_rate": 9.145801008933403e-06,
      "loss": 1.121193528175354,
      "step": 242
    },
    {
      "epoch": 0.9721115537848606,
      "grad_norm": 0.2892410159111023,
      "learning_rate": 9.128330542681154e-06,
      "loss": 1.2122989892959595,
      "step": 244
    },
    {
      "epoch": 0.9800796812749004,
      "grad_norm": 1.7912423610687256,
      "learning_rate": 9.110702359174944e-06,
      "loss": 0.9652339220046997,
      "step": 246
    },
    {
      "epoch": 0.9880478087649402,
      "grad_norm": 0.7738572955131531,
      "learning_rate": 9.092917224681427e-06,
      "loss": 1.0111494064331055,
      "step": 248
    },
    {
      "epoch": 0.9960159362549801,
      "grad_norm": 0.570904016494751,
      "learning_rate": 9.074975912289641e-06,
      "loss": 0.9990465641021729,
      "step": 250
    },
    {
      "epoch": 1.00398406374502,
      "grad_norm": 0.571216344833374,
      "learning_rate": 9.056879201877412e-06,
      "loss": 0.9137588143348694,
      "step": 252
    },
    {
      "epoch": 1.0119521912350598,
      "grad_norm": 0.28236615657806396,
      "learning_rate": 9.038627880077441e-06,
      "loss": 0.6581600308418274,
      "step": 254
    },
    {
      "epoch": 1.0199203187250996,
      "grad_norm": 0.29759418964385986,
      "learning_rate": 9.020222740243127e-06,
      "loss": 1.0936863422393799,
      "step": 256
    },
    {
      "epoch": 1.0278884462151394,
      "grad_norm": 1.3199347257614136,
      "learning_rate": 9.001664582414066e-06,
      "loss": 1.0800862312316895,
      "step": 258
    },
    {
      "epoch": 1.0358565737051793,
      "grad_norm": 1.234781265258789,
      "learning_rate": 8.982954213281291e-06,
      "loss": 1.0242048501968384,
      "step": 260
    },
    {
      "epoch": 1.043824701195219,
      "grad_norm": 0.8170076012611389,
      "learning_rate": 8.964092446152183e-06,
      "loss": 0.9958657026290894,
      "step": 262
    },
    {
      "epoch": 1.051792828685259,
      "grad_norm": 0.5700747966766357,
      "learning_rate": 8.945080100915145e-06,
      "loss": 1.0254192352294922,
      "step": 264
    },
    {
      "epoch": 1.0597609561752988,
      "grad_norm": 2.9084181785583496,
      "learning_rate": 8.925918004003941e-06,
      "loss": 0.5778894424438477,
      "step": 266
    },
    {
      "epoch": 1.0677290836653386,
      "grad_norm": 0.24685457348823547,
      "learning_rate": 8.90660698836179e-06,
      "loss": 0.8626284599304199,
      "step": 268
    },
    {
      "epoch": 1.0756972111553784,
      "grad_norm": 1.2787214517593384,
      "learning_rate": 8.887147893405145e-06,
      "loss": 0.5841003656387329,
      "step": 270
    },
    {
      "epoch": 1.0836653386454183,
      "grad_norm": 0.296922892332077,
      "learning_rate": 8.867541564987212e-06,
      "loss": 1.196135401725769,
      "step": 272
    },
    {
      "epoch": 1.091633466135458,
      "grad_norm": 0.6439626216888428,
      "learning_rate": 8.84778885536118e-06,
      "loss": 0.5580145120620728,
      "step": 274
    },
    {
      "epoch": 1.099601593625498,
      "grad_norm": 0.4150227904319763,
      "learning_rate": 8.827890623143172e-06,
      "loss": 0.8041074872016907,
      "step": 276
    },
    {
      "epoch": 1.1075697211155378,
      "grad_norm": 0.3217131793498993,
      "learning_rate": 8.80784773327494e-06,
      "loss": 0.7971775531768799,
      "step": 278
    },
    {
      "epoch": 1.1155378486055776,
      "grad_norm": 0.5422036647796631,
      "learning_rate": 8.787661056986236e-06,
      "loss": 0.9509108662605286,
      "step": 280
    },
    {
      "epoch": 1.1235059760956174,
      "grad_norm": 0.29051142930984497,
      "learning_rate": 8.767331471756975e-06,
      "loss": 1.1285297870635986,
      "step": 282
    },
    {
      "epoch": 1.1314741035856573,
      "grad_norm": 0.505642831325531,
      "learning_rate": 8.746859861279068e-06,
      "loss": 1.13683021068573,
      "step": 284
    },
    {
      "epoch": 1.139442231075697,
      "grad_norm": 1.167439579963684,
      "learning_rate": 8.726247115418017e-06,
      "loss": 0.28178226947784424,
      "step": 286
    },
    {
      "epoch": 1.1474103585657371,
      "grad_norm": 0.31547048687934875,
      "learning_rate": 8.70549413017424e-06,
      "loss": 0.6513753533363342,
      "step": 288
    },
    {
      "epoch": 1.155378486055777,
      "grad_norm": 0.5137228965759277,
      "learning_rate": 8.684601807644114e-06,
      "loss": 0.936997652053833,
      "step": 290
    },
    {
      "epoch": 1.1633466135458168,
      "grad_norm": 1.013256549835205,
      "learning_rate": 8.663571055980769e-06,
      "loss": 0.5543504357337952,
      "step": 292
    },
    {
      "epoch": 1.1713147410358566,
      "grad_norm": 0.24896587431430817,
      "learning_rate": 8.642402789354609e-06,
      "loss": 1.123894214630127,
      "step": 294
    },
    {
      "epoch": 1.1792828685258965,
      "grad_norm": 0.31399160623550415,
      "learning_rate": 8.62109792791358e-06,
      "loss": 1.1999843120574951,
      "step": 296
    },
    {
      "epoch": 1.1872509960159363,
      "grad_norm": 0.30823755264282227,
      "learning_rate": 8.599657397743163e-06,
      "loss": 1.2604495286941528,
      "step": 298
    },
    {
      "epoch": 1.1952191235059761,
      "grad_norm": 0.42640066146850586,
      "learning_rate": 8.578082130826123e-06,
      "loss": 0.5553680658340454,
      "step": 300
    },
    {
      "epoch": 1.203187250996016,
      "grad_norm": 0.5658683776855469,
      "learning_rate": 8.556373065002006e-06,
      "loss": 0.902449369430542,
      "step": 302
    },
    {
      "epoch": 1.2111553784860558,
      "grad_norm": 0.43325480818748474,
      "learning_rate": 8.534531143926359e-06,
      "loss": 0.538908839225769,
      "step": 304
    },
    {
      "epoch": 1.2191235059760956,
      "grad_norm": 0.5187049508094788,
      "learning_rate": 8.512557317029715e-06,
      "loss": 1.11819326877594,
      "step": 306
    },
    {
      "epoch": 1.2270916334661355,
      "grad_norm": 1.5206022262573242,
      "learning_rate": 8.49045253947633e-06,
      "loss": 0.8958917260169983,
      "step": 308
    },
    {
      "epoch": 1.2350597609561753,
      "grad_norm": 0.6700809001922607,
      "learning_rate": 8.46821777212266e-06,
      "loss": 1.1043747663497925,
      "step": 310
    },
    {
      "epoch": 1.2430278884462151,
      "grad_norm": 0.5006802678108215,
      "learning_rate": 8.445853981475587e-06,
      "loss": 1.0075000524520874,
      "step": 312
    },
    {
      "epoch": 1.250996015936255,
      "grad_norm": 0.4754709005355835,
      "learning_rate": 8.423362139650414e-06,
      "loss": 0.8615016341209412,
      "step": 314
    },
    {
      "epoch": 1.2589641434262948,
      "grad_norm": 0.9323794841766357,
      "learning_rate": 8.40074322432861e-06,
      "loss": 0.7096800208091736,
      "step": 316
    },
    {
      "epoch": 1.2669322709163346,
      "grad_norm": 0.5477792620658875,
      "learning_rate": 8.377998218715311e-06,
      "loss": 0.6627480387687683,
      "step": 318
    },
    {
      "epoch": 1.2749003984063745,
      "grad_norm": 1.4046838283538818,
      "learning_rate": 8.355128111496569e-06,
      "loss": 0.5815978050231934,
      "step": 320
    },
    {
      "epoch": 1.2828685258964143,
      "grad_norm": 0.443997323513031,
      "learning_rate": 8.332133896796399e-06,
      "loss": 0.9199466109275818,
      "step": 322
    },
    {
      "epoch": 1.2908366533864541,
      "grad_norm": 0.4384806752204895,
      "learning_rate": 8.309016574133545e-06,
      "loss": 0.8846327066421509,
      "step": 324
    },
    {
      "epoch": 1.298804780876494,
      "grad_norm": 0.642951250076294,
      "learning_rate": 8.28577714837805e-06,
      "loss": 1.1947497129440308,
      "step": 326
    },
    {
      "epoch": 1.3067729083665338,
      "grad_norm": 0.2723095417022705,
      "learning_rate": 8.262416629707555e-06,
      "loss": 0.6702521443367004,
      "step": 328
    },
    {
      "epoch": 1.3147410358565736,
      "grad_norm": 0.296138197183609,
      "learning_rate": 8.238936033563415e-06,
      "loss": 0.7461103796958923,
      "step": 330
    },
    {
      "epoch": 1.3227091633466135,
      "grad_norm": 0.9188587069511414,
      "learning_rate": 8.215336380606533e-06,
      "loss": 0.9019520282745361,
      "step": 332
    },
    {
      "epoch": 1.3306772908366533,
      "grad_norm": 1.0994230508804321,
      "learning_rate": 8.191618696673012e-06,
      "loss": 0.7043916583061218,
      "step": 334
    },
    {
      "epoch": 1.3386454183266931,
      "grad_norm": 0.5795750617980957,
      "learning_rate": 8.167784012729555e-06,
      "loss": 0.8661832213401794,
      "step": 336
    },
    {
      "epoch": 1.3466135458167332,
      "grad_norm": 0.5290566682815552,
      "learning_rate": 8.143833364828654e-06,
      "loss": 1.0530855655670166,
      "step": 338
    },
    {
      "epoch": 1.354581673306773,
      "grad_norm": 0.4656870663166046,
      "learning_rate": 8.119767794063553e-06,
      "loss": 1.0912537574768066,
      "step": 340
    },
    {
      "epoch": 1.3625498007968129,
      "grad_norm": 0.32077422738075256,
      "learning_rate": 8.095588346523e-06,
      "loss": 1.1272807121276855,
      "step": 342
    },
    {
      "epoch": 1.3705179282868527,
      "grad_norm": 0.331540584564209,
      "learning_rate": 8.071296073245759e-06,
      "loss": 1.0714155435562134,
      "step": 344
    },
    {
      "epoch": 1.3784860557768925,
      "grad_norm": 0.8204175233840942,
      "learning_rate": 8.046892030174938e-06,
      "loss": 0.7155398726463318,
      "step": 346
    },
    {
      "epoch": 1.3864541832669324,
      "grad_norm": 0.3193747103214264,
      "learning_rate": 8.022377278112089e-06,
      "loss": 1.1310451030731201,
      "step": 348
    },
    {
      "epoch": 1.3944223107569722,
      "grad_norm": 0.42103439569473267,
      "learning_rate": 7.997752882671088e-06,
      "loss": 0.6394126415252686,
      "step": 350
    },
    {
      "epoch": 1.402390438247012,
      "grad_norm": 0.26277822256088257,
      "learning_rate": 7.97301991423182e-06,
      "loss": 1.140794038772583,
      "step": 352
    },
    {
      "epoch": 1.4103585657370519,
      "grad_norm": 1.0092138051986694,
      "learning_rate": 7.948179447893652e-06,
      "loss": 1.0688328742980957,
      "step": 354
    },
    {
      "epoch": 1.4183266932270917,
      "grad_norm": 1.8953406810760498,
      "learning_rate": 7.923232563428693e-06,
      "loss": 0.9266577959060669,
      "step": 356
    },
    {
      "epoch": 1.4262948207171315,
      "grad_norm": 0.32846713066101074,
      "learning_rate": 7.898180345234875e-06,
      "loss": 1.1128524541854858,
      "step": 358
    },
    {
      "epoch": 1.4342629482071714,
      "grad_norm": 0.30604761838912964,
      "learning_rate": 7.873023882288793e-06,
      "loss": 0.5628020167350769,
      "step": 360
    },
    {
      "epoch": 1.4422310756972112,
      "grad_norm": 0.29621729254722595,
      "learning_rate": 7.847764268098387e-06,
      "loss": 0.6502870321273804,
      "step": 362
    },
    {
      "epoch": 1.450199203187251,
      "grad_norm": 0.4790603816509247,
      "learning_rate": 7.822402600655405e-06,
      "loss": 1.0038015842437744,
      "step": 364
    },
    {
      "epoch": 1.4581673306772909,
      "grad_norm": 0.2318412810564041,
      "learning_rate": 7.79693998238767e-06,
      "loss": 1.0806806087493896,
      "step": 366
    },
    {
      "epoch": 1.4661354581673307,
      "grad_norm": 0.3429558277130127,
      "learning_rate": 7.771377520111163e-06,
      "loss": 0.5924015045166016,
      "step": 368
    },
    {
      "epoch": 1.4741035856573705,
      "grad_norm": 1.1847248077392578,
      "learning_rate": 7.745716324981912e-06,
      "loss": 0.7455736398696899,
      "step": 370
    },
    {
      "epoch": 1.4820717131474104,
      "grad_norm": 0.6212412714958191,
      "learning_rate": 7.71995751244769e-06,
      "loss": 1.252964973449707,
      "step": 372
    },
    {
      "epoch": 1.4900398406374502,
      "grad_norm": 0.2830260396003723,
      "learning_rate": 7.694102202199536e-06,
      "loss": 1.065024495124817,
      "step": 374
    },
    {
      "epoch": 1.49800796812749,
      "grad_norm": 0.20884928107261658,
      "learning_rate": 7.66815151812307e-06,
      "loss": 1.0614570379257202,
      "step": 376
    },
    {
      "epoch": 1.5059760956175299,
      "grad_norm": 1.9683042764663696,
      "learning_rate": 7.642106588249651e-06,
      "loss": 0.9400176405906677,
      "step": 378
    },
    {
      "epoch": 1.5139442231075697,
      "grad_norm": 0.27558979392051697,
      "learning_rate": 7.6159685447073376e-06,
      "loss": 1.0889798402786255,
      "step": 380
    },
    {
      "epoch": 1.5219123505976095,
      "grad_norm": 0.29121148586273193,
      "learning_rate": 7.589738523671685e-06,
      "loss": 0.6911762952804565,
      "step": 382
    },
    {
      "epoch": 1.5298804780876494,
      "grad_norm": 0.23157551884651184,
      "learning_rate": 7.563417665316338e-06,
      "loss": 0.7454714179039001,
      "step": 384
    },
    {
      "epoch": 1.5378486055776892,
      "grad_norm": 0.6436242461204529,
      "learning_rate": 7.5370071137634984e-06,
      "loss": 0.960307776927948,
      "step": 386
    },
    {
      "epoch": 1.545816733067729,
      "grad_norm": 0.48214268684387207,
      "learning_rate": 7.510508017034165e-06,
      "loss": 1.103023648262024,
      "step": 388
    },
    {
      "epoch": 1.5537848605577689,
      "grad_norm": 0.5468902587890625,
      "learning_rate": 7.483921526998247e-06,
      "loss": 0.5839257836341858,
      "step": 390
    },
    {
      "epoch": 1.5617529880478087,
      "grad_norm": 0.3464953601360321,
      "learning_rate": 7.4572487993244865e-06,
      "loss": 0.555505633354187,
      "step": 392
    },
    {
      "epoch": 1.5697211155378485,
      "grad_norm": 2.1821701526641846,
      "learning_rate": 7.430490993430232e-06,
      "loss": 1.0839892625808716,
      "step": 394
    },
    {
      "epoch": 1.5776892430278884,
      "grad_norm": 0.4379599988460541,
      "learning_rate": 7.403649272431027e-06,
      "loss": 0.9683752059936523,
      "step": 396
    },
    {
      "epoch": 1.5856573705179282,
      "grad_norm": 0.3230619728565216,
      "learning_rate": 7.376724803090069e-06,
      "loss": 1.0813394784927368,
      "step": 398
    },
    {
      "epoch": 1.593625498007968,
      "grad_norm": 2.186976671218872,
      "learning_rate": 7.349718755767473e-06,
      "loss": 0.5913651585578918,
      "step": 400
    },
    {
      "epoch": 1.6015936254980079,
      "grad_norm": 0.6617003083229065,
      "learning_rate": 7.322632304369416e-06,
      "loss": 1.0874972343444824,
      "step": 402
    },
    {
      "epoch": 1.6095617529880477,
      "grad_norm": 0.22927327454090118,
      "learning_rate": 7.2954666262970965e-06,
      "loss": 1.1802277565002441,
      "step": 404
    },
    {
      "epoch": 1.6175298804780875,
      "grad_norm": 1.8990100622177124,
      "learning_rate": 7.268222902395562e-06,
      "loss": 0.7137308716773987,
      "step": 406
    },
    {
      "epoch": 1.6254980079681274,
      "grad_norm": 0.29405704140663147,
      "learning_rate": 7.240902316902373e-06,
      "loss": 0.5633601546287537,
      "step": 408
    },
    {
      "epoch": 1.6334661354581672,
      "grad_norm": 0.3894561529159546,
      "learning_rate": 7.213506057396134e-06,
      "loss": 0.6938993334770203,
      "step": 410
    },
    {
      "epoch": 1.641434262948207,
      "grad_norm": 1.3900845050811768,
      "learning_rate": 7.186035314744872e-06,
      "loss": 0.511903703212738,
      "step": 412
    },
    {
      "epoch": 1.6494023904382469,
      "grad_norm": 0.3045184314250946,
      "learning_rate": 7.1584912830542575e-06,
      "loss": 0.7494630813598633,
      "step": 414
    },
    {
      "epoch": 1.6573705179282867,
      "grad_norm": 0.3465200960636139,
      "learning_rate": 7.130875159615716e-06,
      "loss": 0.7218171954154968,
      "step": 416
    },
    {
      "epoch": 1.6653386454183265,
      "grad_norm": 0.4777250587940216,
      "learning_rate": 7.103188144854376e-06,
      "loss": 0.8778831958770752,
      "step": 418
    },
    {
      "epoch": 1.6733067729083664,
      "grad_norm": 0.27671298384666443,
      "learning_rate": 7.075431442276888e-06,
      "loss": 1.1158913373947144,
      "step": 420
    },
    {
      "epoch": 1.6812749003984062,
      "grad_norm": 0.35358166694641113,
      "learning_rate": 7.047606258419108e-06,
      "loss": 1.0098072290420532,
      "step": 422
    },
    {
      "epoch": 1.6892430278884463,
      "grad_norm": 0.43482786417007446,
      "learning_rate": 7.019713802793663e-06,
      "loss": 0.536350429058075,
      "step": 424
    },
    {
      "epoch": 1.697211155378486,
      "grad_norm": 0.32529133558273315,
      "learning_rate": 6.99175528783736e-06,
      "loss": 1.1366305351257324,
      "step": 426
    },
    {
      "epoch": 1.705179282868526,
      "grad_norm": 0.28394225239753723,
      "learning_rate": 6.963731928858495e-06,
      "loss": 1.070862054824829,
      "step": 428
    },
    {
      "epoch": 1.7131474103585658,
      "grad_norm": 0.4192523658275604,
      "learning_rate": 6.935644943984017e-06,
      "loss": 0.7123104929924011,
      "step": 430
    },
    {
      "epoch": 1.7211155378486056,
      "grad_norm": 0.4438948929309845,
      "learning_rate": 6.9074955541065865e-06,
      "loss": 0.8589892983436584,
      "step": 432
    },
    {
      "epoch": 1.7290836653386454,
      "grad_norm": 0.9661609530448914,
      "learning_rate": 6.8792849828315054e-06,
      "loss": 0.6671584844589233,
      "step": 434
    },
    {
      "epoch": 1.7370517928286853,
      "grad_norm": 0.4608452618122101,
      "learning_rate": 6.851014456423515e-06,
      "loss": 0.39532050490379333,
      "step": 436
    },
    {
      "epoch": 1.745019920318725,
      "grad_norm": 0.4160957634449005,
      "learning_rate": 6.822685203753508e-06,
      "loss": 0.8038726449012756,
      "step": 438
    },
    {
      "epoch": 1.752988047808765,
      "grad_norm": 1.6464133262634277,
      "learning_rate": 6.794298456245111e-06,
      "loss": 0.7063311338424683,
      "step": 440
    },
    {
      "epoch": 1.7609561752988048,
      "grad_norm": 0.8938610553741455,
      "learning_rate": 6.765855447821144e-06,
      "loss": 1.439618706703186,
      "step": 442
    },
    {
      "epoch": 1.7689243027888446,
      "grad_norm": 0.3867045044898987,
      "learning_rate": 6.737357414849993e-06,
      "loss": 0.6978985667228699,
      "step": 444
    },
    {
      "epoch": 1.7768924302788844,
      "grad_norm": 0.32330453395843506,
      "learning_rate": 6.70880559609187e-06,
      "loss": 1.029992699623108,
      "step": 446
    },
    {
      "epoch": 1.7848605577689243,
      "grad_norm": 1.1258419752120972,
      "learning_rate": 6.680201232644958e-06,
      "loss": 1.0915586948394775,
      "step": 448
    },
    {
      "epoch": 1.792828685258964,
      "grad_norm": 0.23717984557151794,
      "learning_rate": 6.651545567891471e-06,
      "loss": 0.8761530518531799,
      "step": 450
    },
    {
      "epoch": 1.800796812749004,
      "grad_norm": 0.33776843547821045,
      "learning_rate": 6.622839847443596e-06,
      "loss": 0.40660956501960754,
      "step": 452
    },
    {
      "epoch": 1.8087649402390438,
      "grad_norm": 2.475672483444214,
      "learning_rate": 6.59408531908936e-06,
      "loss": 0.950668454170227,
      "step": 454
    },
    {
      "epoch": 1.8167330677290838,
      "grad_norm": 0.7502440214157104,
      "learning_rate": 6.565283232738386e-06,
      "loss": 0.6837414503097534,
      "step": 456
    },
    {
      "epoch": 1.8247011952191237,
      "grad_norm": 0.3184780180454254,
      "learning_rate": 6.536434840367555e-06,
      "loss": 0.9988897442817688,
      "step": 458
    },
    {
      "epoch": 1.8326693227091635,
      "grad_norm": 0.6095825433731079,
      "learning_rate": 6.507541395966591e-06,
      "loss": 0.7473199367523193,
      "step": 460
    },
    {
      "epoch": 1.8406374501992033,
      "grad_norm": 0.3309530019760132,
      "learning_rate": 6.47860415548356e-06,
      "loss": 0.8035681247711182,
      "step": 462
    },
    {
      "epoch": 1.8486055776892432,
      "grad_norm": 0.22135062515735626,
      "learning_rate": 6.449624376770259e-06,
      "loss": 0.12004426121711731,
      "step": 464
    },
    {
      "epoch": 1.856573705179283,
      "grad_norm": 0.2587142586708069,
      "learning_rate": 6.420603319527554e-06,
      "loss": 1.1019535064697266,
      "step": 466
    },
    {
      "epoch": 1.8645418326693228,
      "grad_norm": 0.8834278583526611,
      "learning_rate": 6.3915422452506146e-06,
      "loss": 0.6125680208206177,
      "step": 468
    },
    {
      "epoch": 1.8725099601593627,
      "grad_norm": 0.3534308969974518,
      "learning_rate": 6.362442417174082e-06,
      "loss": 1.1139684915542603,
      "step": 470
    },
    {
      "epoch": 1.8804780876494025,
      "grad_norm": 0.582276463508606,
      "learning_rate": 6.33330510021716e-06,
      "loss": 1.0157603025436401,
      "step": 472
    },
    {
      "epoch": 1.8884462151394423,
      "grad_norm": 0.39019274711608887,
      "learning_rate": 6.304131560928622e-06,
      "loss": 1.0498619079589844,
      "step": 474
    },
    {
      "epoch": 1.8964143426294822,
      "grad_norm": 0.23048311471939087,
      "learning_rate": 6.274923067431775e-06,
      "loss": 0.9724476337432861,
      "step": 476
    },
    {
      "epoch": 1.904382470119522,
      "grad_norm": 0.19976598024368286,
      "learning_rate": 6.2456808893693155e-06,
      "loss": 0.8315828442573547,
      "step": 478
    },
    {
      "epoch": 1.9123505976095618,
      "grad_norm": 0.6196638941764832,
      "learning_rate": 6.216406297848155e-06,
      "loss": 0.6385701894760132,
      "step": 480
    },
    {
      "epoch": 1.9203187250996017,
      "grad_norm": 0.4662269353866577,
      "learning_rate": 6.187100565384162e-06,
      "loss": 0.7646976709365845,
      "step": 482
    },
    {
      "epoch": 1.9282868525896415,
      "grad_norm": 0.22306957840919495,
      "learning_rate": 6.157764965846841e-06,
      "loss": 1.0547288656234741,
      "step": 484
    },
    {
      "epoch": 1.9362549800796813,
      "grad_norm": 0.27468961477279663,
      "learning_rate": 6.128400774403974e-06,
      "loss": 1.006617784500122,
      "step": 486
    },
    {
      "epoch": 1.9442231075697212,
      "grad_norm": 0.21906159818172455,
      "learning_rate": 6.0990092674661834e-06,
      "loss": 1.0428978204727173,
      "step": 488
    },
    {
      "epoch": 1.952191235059761,
      "grad_norm": 0.39873772859573364,
      "learning_rate": 6.069591722631446e-06,
      "loss": 1.0501173734664917,
      "step": 490
    },
    {
      "epoch": 1.9601593625498008,
      "grad_norm": 0.2867768108844757,
      "learning_rate": 6.040149418629562e-06,
      "loss": 0.7425338625907898,
      "step": 492
    },
    {
      "epoch": 1.9681274900398407,
      "grad_norm": 1.727679967880249,
      "learning_rate": 6.010683635266569e-06,
      "loss": 0.7619685530662537,
      "step": 494
    },
    {
      "epoch": 1.9760956175298805,
      "grad_norm": 0.3420916497707367,
      "learning_rate": 5.981195653369113e-06,
      "loss": 1.0616827011108398,
      "step": 496
    },
    {
      "epoch": 1.9840637450199203,
      "grad_norm": 2.505432605743408,
      "learning_rate": 5.95168675472877e-06,
      "loss": 0.8652823567390442,
      "step": 498
    },
    {
      "epoch": 1.9920318725099602,
      "grad_norm": 0.24245813488960266,
      "learning_rate": 5.922158222046334e-06,
      "loss": 1.1408816576004028,
      "step": 500
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.27242010831832886,
      "learning_rate": 5.892611338876053e-06,
      "loss": 0.5883684754371643,
      "step": 502
    },
    {
      "epoch": 2.00796812749004,
      "grad_norm": 0.19679775834083557,
      "learning_rate": 5.8630473895698426e-06,
      "loss": 0.5334652662277222,
      "step": 504
    },
    {
      "epoch": 2.0159362549800797,
      "grad_norm": 0.4099384546279907,
      "learning_rate": 5.833467659221452e-06,
      "loss": 0.5121970772743225,
      "step": 506
    },
    {
      "epoch": 2.0239043824701195,
      "grad_norm": 0.1672699898481369,
      "learning_rate": 5.803873433610604e-06,
      "loss": 0.4877749979496002,
      "step": 508
    },
    {
      "epoch": 2.0318725099601593,
      "grad_norm": 0.22908855974674225,
      "learning_rate": 5.774265999147108e-06,
      "loss": 0.7408496141433716,
      "step": 510
    },
    {
      "epoch": 2.039840637450199,
      "grad_norm": 0.32839062809944153,
      "learning_rate": 5.744646642814935e-06,
      "loss": 0.6727060675621033,
      "step": 512
    },
    {
      "epoch": 2.047808764940239,
      "grad_norm": 4.0945634841918945,
      "learning_rate": 5.715016652116284e-06,
      "loss": 0.34946340322494507,
      "step": 514
    },
    {
      "epoch": 2.055776892430279,
      "grad_norm": 0.2665126621723175,
      "learning_rate": 5.68537731501561e-06,
      "loss": 0.8418608903884888,
      "step": 516
    },
    {
      "epoch": 2.0637450199203187,
      "grad_norm": 0.87298583984375,
      "learning_rate": 5.655729919883639e-06,
      "loss": 0.4820442795753479,
      "step": 518
    },
    {
      "epoch": 2.0717131474103585,
      "grad_norm": 0.4436075687408447,
      "learning_rate": 5.626075755441366e-06,
      "loss": 0.7624883055686951,
      "step": 520
    },
    {
      "epoch": 2.0796812749003983,
      "grad_norm": 0.5095271468162537,
      "learning_rate": 5.596416110704034e-06,
      "loss": 0.8474641442298889,
      "step": 522
    },
    {
      "epoch": 2.087649402390438,
      "grad_norm": 1.8589236736297607,
      "learning_rate": 5.566752274925109e-06,
      "loss": 0.8073615431785583,
      "step": 524
    },
    {
      "epoch": 2.095617529880478,
      "grad_norm": 0.8471553325653076,
      "learning_rate": 5.5370855375402345e-06,
      "loss": 0.4542064368724823,
      "step": 526
    },
    {
      "epoch": 2.103585657370518,
      "grad_norm": 0.23006175458431244,
      "learning_rate": 5.507417188111176e-06,
      "loss": 0.6293087601661682,
      "step": 528
    },
    {
      "epoch": 2.1115537848605577,
      "grad_norm": 0.21818548440933228,
      "learning_rate": 5.477748516269779e-06,
      "loss": 0.5145719051361084,
      "step": 530
    },
    {
      "epoch": 2.1195219123505975,
      "grad_norm": 0.25169360637664795,
      "learning_rate": 5.448080811661897e-06,
      "loss": 0.20230402052402496,
      "step": 532
    },
    {
      "epoch": 2.1274900398406373,
      "grad_norm": 0.5802799463272095,
      "learning_rate": 5.418415363891346e-06,
      "loss": 0.8121124505996704,
      "step": 534
    },
    {
      "epoch": 2.135458167330677,
      "grad_norm": 2.6178159713745117,
      "learning_rate": 5.3887534624638335e-06,
      "loss": 0.6049786806106567,
      "step": 536
    },
    {
      "epoch": 2.143426294820717,
      "grad_norm": 0.7545949220657349,
      "learning_rate": 5.359096396730924e-06,
      "loss": 0.6464651823043823,
      "step": 538
    },
    {
      "epoch": 2.151394422310757,
      "grad_norm": 0.2867981791496277,
      "learning_rate": 5.329445455833975e-06,
      "loss": 0.8182328939437866,
      "step": 540
    },
    {
      "epoch": 2.1593625498007967,
      "grad_norm": 0.5017597079277039,
      "learning_rate": 5.299801928648105e-06,
      "loss": 0.8306395411491394,
      "step": 542
    },
    {
      "epoch": 2.1673306772908365,
      "grad_norm": 0.30368703603744507,
      "learning_rate": 5.270167103726181e-06,
      "loss": 0.5657422542572021,
      "step": 544
    },
    {
      "epoch": 2.1752988047808763,
      "grad_norm": 1.346498727798462,
      "learning_rate": 5.240542269242794e-06,
      "loss": 0.5811312794685364,
      "step": 546
    },
    {
      "epoch": 2.183266932270916,
      "grad_norm": 0.27454039454460144,
      "learning_rate": 5.210928712938262e-06,
      "loss": 0.8369240164756775,
      "step": 548
    },
    {
      "epoch": 2.191235059760956,
      "grad_norm": 0.25529855489730835,
      "learning_rate": 5.181327722062667e-06,
      "loss": 0.8771988749504089,
      "step": 550
    },
    {
      "epoch": 2.199203187250996,
      "grad_norm": 2.9861326217651367,
      "learning_rate": 5.151740583319892e-06,
      "loss": 0.5763852596282959,
      "step": 552
    },
    {
      "epoch": 2.2071713147410357,
      "grad_norm": 0.5112147331237793,
      "learning_rate": 5.122168582811689e-06,
      "loss": 0.46077486872673035,
      "step": 554
    },
    {
      "epoch": 2.2151394422310755,
      "grad_norm": 0.23867323994636536,
      "learning_rate": 5.092613005981781e-06,
      "loss": 0.607995867729187,
      "step": 556
    },
    {
      "epoch": 2.2231075697211153,
      "grad_norm": 1.360227346420288,
      "learning_rate": 5.063075137559979e-06,
      "loss": 0.7866989374160767,
      "step": 558
    },
    {
      "epoch": 2.231075697211155,
      "grad_norm": 0.2265808880329132,
      "learning_rate": 5.033556261506343e-06,
      "loss": 0.8641742467880249,
      "step": 560
    },
    {
      "epoch": 2.239043824701195,
      "grad_norm": 0.314727246761322,
      "learning_rate": 5.0040576609553645e-06,
      "loss": 0.9733244180679321,
      "step": 562
    },
    {
      "epoch": 2.247011952191235,
      "grad_norm": 6.417182445526123,
      "learning_rate": 4.974580618160194e-06,
      "loss": 0.8019009828567505,
      "step": 564
    },
    {
      "epoch": 2.2549800796812747,
      "grad_norm": 0.22401520609855652,
      "learning_rate": 4.945126414436908e-06,
      "loss": 0.5445197820663452,
      "step": 566
    },
    {
      "epoch": 2.2629482071713145,
      "grad_norm": 0.1901124268770218,
      "learning_rate": 4.915696330108804e-06,
      "loss": 0.6923782825469971,
      "step": 568
    },
    {
      "epoch": 2.2709163346613543,
      "grad_norm": 0.17700311541557312,
      "learning_rate": 4.886291644450751e-06,
      "loss": 0.8518681526184082,
      "step": 570
    },
    {
      "epoch": 2.278884462151394,
      "grad_norm": 0.39328062534332275,
      "learning_rate": 4.8569136356335845e-06,
      "loss": 0.6248937249183655,
      "step": 572
    },
    {
      "epoch": 2.2868525896414345,
      "grad_norm": 0.43977269530296326,
      "learning_rate": 4.827563580668544e-06,
      "loss": 0.6675489544868469,
      "step": 574
    },
    {
      "epoch": 2.2948207171314743,
      "grad_norm": 0.17548246681690216,
      "learning_rate": 4.798242755351755e-06,
      "loss": 0.2537887692451477,
      "step": 576
    },
    {
      "epoch": 2.302788844621514,
      "grad_norm": 0.31994521617889404,
      "learning_rate": 4.768952434208795e-06,
      "loss": 0.5556949973106384,
      "step": 578
    },
    {
      "epoch": 2.310756972111554,
      "grad_norm": 0.2803627848625183,
      "learning_rate": 4.7396938904392656e-06,
      "loss": 0.5626878142356873,
      "step": 580
    },
    {
      "epoch": 2.318725099601594,
      "grad_norm": 0.8239148855209351,
      "learning_rate": 4.71046839586147e-06,
      "loss": 0.5838683843612671,
      "step": 582
    },
    {
      "epoch": 2.3266932270916336,
      "grad_norm": 0.22264306247234344,
      "learning_rate": 4.681277220857108e-06,
      "loss": 0.734087347984314,
      "step": 584
    },
    {
      "epoch": 2.3346613545816735,
      "grad_norm": 0.27492600679397583,
      "learning_rate": 4.6521216343160824e-06,
      "loss": 0.6127607226371765,
      "step": 586
    },
    {
      "epoch": 2.3426294820717133,
      "grad_norm": 0.47190821170806885,
      "learning_rate": 4.623002903581316e-06,
      "loss": 0.7112446427345276,
      "step": 588
    },
    {
      "epoch": 2.350597609561753,
      "grad_norm": 0.23274028301239014,
      "learning_rate": 4.593922294393677e-06,
      "loss": 0.8977214097976685,
      "step": 590
    },
    {
      "epoch": 2.358565737051793,
      "grad_norm": 0.21238867938518524,
      "learning_rate": 4.564881070836955e-06,
      "loss": 0.8718763589859009,
      "step": 592
    },
    {
      "epoch": 2.366533864541833,
      "grad_norm": 0.34060487151145935,
      "learning_rate": 4.535880495282916e-06,
      "loss": 0.6093429327011108,
      "step": 594
    },
    {
      "epoch": 2.3745019920318726,
      "grad_norm": 0.22852496802806854,
      "learning_rate": 4.506921828336427e-06,
      "loss": 0.8165010809898376,
      "step": 596
    },
    {
      "epoch": 2.3824701195219125,
      "grad_norm": 0.1634281426668167,
      "learning_rate": 4.478006328780657e-06,
      "loss": 0.6037092208862305,
      "step": 598
    },
    {
      "epoch": 2.3904382470119523,
      "grad_norm": 1.3593310117721558,
      "learning_rate": 4.44913525352237e-06,
      "loss": 0.7917259931564331,
      "step": 600
    },
    {
      "epoch": 2.398406374501992,
      "grad_norm": 0.22936953604221344,
      "learning_rate": 4.420309857537277e-06,
      "loss": 0.8677392601966858,
      "step": 602
    },
    {
      "epoch": 2.406374501992032,
      "grad_norm": 0.43788209557533264,
      "learning_rate": 4.39153139381549e-06,
      "loss": 0.5103025436401367,
      "step": 604
    },
    {
      "epoch": 2.414342629482072,
      "grad_norm": 1.2928889989852905,
      "learning_rate": 4.3628011133070605e-06,
      "loss": 0.2453230768442154,
      "step": 606
    },
    {
      "epoch": 2.4223107569721116,
      "grad_norm": 0.13447590172290802,
      "learning_rate": 4.334120264867596e-06,
      "loss": 0.45397987961769104,
      "step": 608
    },
    {
      "epoch": 2.4302788844621515,
      "grad_norm": 0.14010317623615265,
      "learning_rate": 4.305490095203979e-06,
      "loss": 0.708511471748352,
      "step": 610
    },
    {
      "epoch": 2.4382470119521913,
      "grad_norm": 0.2558565139770508,
      "learning_rate": 4.27691184882017e-06,
      "loss": 0.7274853587150574,
      "step": 612
    },
    {
      "epoch": 2.446215139442231,
      "grad_norm": 0.2692249119281769,
      "learning_rate": 4.248386767963121e-06,
      "loss": 0.5029411911964417,
      "step": 614
    },
    {
      "epoch": 2.454183266932271,
      "grad_norm": 0.23596222698688507,
      "learning_rate": 4.219916092568764e-06,
      "loss": 0.8314698338508606,
      "step": 616
    },
    {
      "epoch": 2.462151394422311,
      "grad_norm": 0.3547062873840332,
      "learning_rate": 4.191501060208121e-06,
      "loss": 0.6102959513664246,
      "step": 618
    },
    {
      "epoch": 2.4701195219123506,
      "grad_norm": 0.5084841251373291,
      "learning_rate": 4.163142906033512e-06,
      "loss": 0.670396625995636,
      "step": 620
    },
    {
      "epoch": 2.4780876494023905,
      "grad_norm": 0.24352110922336578,
      "learning_rate": 4.134842862724854e-06,
      "loss": 0.756343424320221,
      "step": 622
    },
    {
      "epoch": 2.4860557768924303,
      "grad_norm": 0.483839750289917,
      "learning_rate": 4.106602160436088e-06,
      "loss": 0.6942517757415771,
      "step": 624
    },
    {
      "epoch": 2.49402390438247,
      "grad_norm": 0.2871508300304413,
      "learning_rate": 4.078422026741706e-06,
      "loss": 0.5420860052108765,
      "step": 626
    },
    {
      "epoch": 2.50199203187251,
      "grad_norm": 0.30408766865730286,
      "learning_rate": 4.050303686583386e-06,
      "loss": 0.8625388741493225,
      "step": 628
    },
    {
      "epoch": 2.50996015936255,
      "grad_norm": 0.3474479019641876,
      "learning_rate": 4.0222483622167465e-06,
      "loss": 0.37638863921165466,
      "step": 630
    },
    {
      "epoch": 2.5179282868525896,
      "grad_norm": 2.7704296112060547,
      "learning_rate": 3.9942572731582185e-06,
      "loss": 0.43901708722114563,
      "step": 632
    },
    {
      "epoch": 2.5258964143426295,
      "grad_norm": 0.20278728008270264,
      "learning_rate": 3.966331636132039e-06,
      "loss": 0.8589592576026917,
      "step": 634
    },
    {
      "epoch": 2.5338645418326693,
      "grad_norm": 0.6090526580810547,
      "learning_rate": 3.938472665017355e-06,
      "loss": 0.852121114730835,
      "step": 636
    },
    {
      "epoch": 2.541832669322709,
      "grad_norm": 0.38922619819641113,
      "learning_rate": 3.91068157079546e-06,
      "loss": 0.8877055048942566,
      "step": 638
    },
    {
      "epoch": 2.549800796812749,
      "grad_norm": 0.3310116231441498,
      "learning_rate": 3.882959561497159e-06,
      "loss": 0.5493086576461792,
      "step": 640
    },
    {
      "epoch": 2.557768924302789,
      "grad_norm": 1.107908010482788,
      "learning_rate": 3.855307842150259e-06,
      "loss": 0.8867993950843811,
      "step": 642
    },
    {
      "epoch": 2.5657370517928286,
      "grad_norm": 0.23238185048103333,
      "learning_rate": 3.827727614727177e-06,
      "loss": 0.9654583930969238,
      "step": 644
    },
    {
      "epoch": 2.5737051792828685,
      "grad_norm": 0.26392483711242676,
      "learning_rate": 3.8002200780927e-06,
      "loss": 0.5441569685935974,
      "step": 646
    },
    {
      "epoch": 2.5816733067729083,
      "grad_norm": 0.31749483942985535,
      "learning_rate": 3.772786427951878e-06,
      "loss": 0.8563807606697083,
      "step": 648
    },
    {
      "epoch": 2.589641434262948,
      "grad_norm": 0.17193931341171265,
      "learning_rate": 3.7454278567980406e-06,
      "loss": 0.6253618597984314,
      "step": 650
    },
    {
      "epoch": 2.597609561752988,
      "grad_norm": 3.2469472885131836,
      "learning_rate": 3.7181455538609577e-06,
      "loss": 0.6294826865196228,
      "step": 652
    },
    {
      "epoch": 2.605577689243028,
      "grad_norm": 0.2697989344596863,
      "learning_rate": 3.690940705055163e-06,
      "loss": 0.5965220928192139,
      "step": 654
    },
    {
      "epoch": 2.6135458167330676,
      "grad_norm": 0.50716632604599,
      "learning_rate": 3.6638144929283824e-06,
      "loss": 0.30099672079086304,
      "step": 656
    },
    {
      "epoch": 2.6215139442231075,
      "grad_norm": 0.323240727186203,
      "learning_rate": 3.6367680966101483e-06,
      "loss": 0.4847179055213928,
      "step": 658
    },
    {
      "epoch": 2.6294820717131473,
      "grad_norm": 0.44548723101615906,
      "learning_rate": 3.6098026917605334e-06,
      "loss": 0.8643190860748291,
      "step": 660
    },
    {
      "epoch": 2.637450199203187,
      "grad_norm": 1.7422468662261963,
      "learning_rate": 3.5829194505190534e-06,
      "loss": 0.7982045412063599,
      "step": 662
    },
    {
      "epoch": 2.645418326693227,
      "grad_norm": 0.4717836081981659,
      "learning_rate": 3.5561195414537164e-06,
      "loss": 0.5058800578117371,
      "step": 664
    },
    {
      "epoch": 2.653386454183267,
      "grad_norm": 0.44365137815475464,
      "learning_rate": 3.529404129510221e-06,
      "loss": 0.8421945571899414,
      "step": 666
    },
    {
      "epoch": 2.6613545816733066,
      "grad_norm": 0.08579611778259277,
      "learning_rate": 3.502774375961321e-06,
      "loss": 0.5676506161689758,
      "step": 668
    },
    {
      "epoch": 2.6693227091633465,
      "grad_norm": 0.3980202078819275,
      "learning_rate": 3.4762314383563533e-06,
      "loss": 0.6362449526786804,
      "step": 670
    },
    {
      "epoch": 2.6772908366533863,
      "grad_norm": 0.1160992830991745,
      "learning_rate": 3.4497764704709143e-06,
      "loss": 0.3021758794784546,
      "step": 672
    },
    {
      "epoch": 2.685258964143426,
      "grad_norm": 0.17738822102546692,
      "learning_rate": 3.4234106222567055e-06,
      "loss": 0.8663972020149231,
      "step": 674
    },
    {
      "epoch": 2.6932270916334664,
      "grad_norm": 0.8303455114364624,
      "learning_rate": 3.3971350397915504e-06,
      "loss": 0.5851129293441772,
      "step": 676
    },
    {
      "epoch": 2.7011952191235062,
      "grad_norm": 0.45925116539001465,
      "learning_rate": 3.370950865229583e-06,
      "loss": 0.5390315055847168,
      "step": 678
    },
    {
      "epoch": 2.709163346613546,
      "grad_norm": 0.2804052531719208,
      "learning_rate": 3.344859236751585e-06,
      "loss": 0.8887301683425903,
      "step": 680
    },
    {
      "epoch": 2.717131474103586,
      "grad_norm": 0.22559693455696106,
      "learning_rate": 3.3188612885155257e-06,
      "loss": 0.8851104974746704,
      "step": 682
    },
    {
      "epoch": 2.7250996015936257,
      "grad_norm": 0.24337342381477356,
      "learning_rate": 3.2929581506072596e-06,
      "loss": 0.619876503944397,
      "step": 684
    },
    {
      "epoch": 2.7330677290836656,
      "grad_norm": 0.5463287234306335,
      "learning_rate": 3.267150948991387e-06,
      "loss": 0.4967731833457947,
      "step": 686
    },
    {
      "epoch": 2.7410358565737054,
      "grad_norm": 0.19549770653247833,
      "learning_rate": 3.241440805462337e-06,
      "loss": 0.8568810820579529,
      "step": 688
    },
    {
      "epoch": 2.7490039840637452,
      "grad_norm": 0.2904706597328186,
      "learning_rate": 3.215828837595585e-06,
      "loss": 0.4567462205886841,
      "step": 690
    },
    {
      "epoch": 2.756972111553785,
      "grad_norm": 0.25626641511917114,
      "learning_rate": 3.190316158699084e-06,
      "loss": 0.9870220422744751,
      "step": 692
    },
    {
      "epoch": 2.764940239043825,
      "grad_norm": 0.6251752376556396,
      "learning_rate": 3.1649038777648622e-06,
      "loss": 0.7637002468109131,
      "step": 694
    },
    {
      "epoch": 2.7729083665338647,
      "grad_norm": 0.4932820498943329,
      "learning_rate": 3.1395930994208234e-06,
      "loss": 0.44007396697998047,
      "step": 696
    },
    {
      "epoch": 2.7808764940239046,
      "grad_norm": 0.2649643123149872,
      "learning_rate": 3.114384923882734e-06,
      "loss": 0.7740649580955505,
      "step": 698
    },
    {
      "epoch": 2.7888446215139444,
      "grad_norm": 0.21873942017555237,
      "learning_rate": 3.089280446906387e-06,
      "loss": 0.896877646446228,
      "step": 700
    },
    {
      "epoch": 2.7968127490039842,
      "grad_norm": 0.5701000690460205,
      "learning_rate": 3.0642807597399847e-06,
      "loss": 0.5173239707946777,
      "step": 702
    },
    {
      "epoch": 2.804780876494024,
      "grad_norm": 0.5561186671257019,
      "learning_rate": 3.0393869490766913e-06,
      "loss": 1.088994026184082,
      "step": 704
    },
    {
      "epoch": 2.812749003984064,
      "grad_norm": 0.20816925168037415,
      "learning_rate": 3.0146000970074087e-06,
      "loss": 1.0059046745300293,
      "step": 706
    },
    {
      "epoch": 2.8207171314741037,
      "grad_norm": 0.9060568809509277,
      "learning_rate": 2.9899212809737275e-06,
      "loss": 0.9540885090827942,
      "step": 708
    },
    {
      "epoch": 2.8286852589641436,
      "grad_norm": 0.2360645979642868,
      "learning_rate": 2.9653515737211037e-06,
      "loss": 0.45551544427871704,
      "step": 710
    },
    {
      "epoch": 2.8366533864541834,
      "grad_norm": 0.19540555775165558,
      "learning_rate": 2.940892043252225e-06,
      "loss": 0.9112032651901245,
      "step": 712
    },
    {
      "epoch": 2.8446215139442232,
      "grad_norm": 0.20937809348106384,
      "learning_rate": 2.916543752780574e-06,
      "loss": 0.559850811958313,
      "step": 714
    },
    {
      "epoch": 2.852589641434263,
      "grad_norm": 0.24084621667861938,
      "learning_rate": 2.892307760684234e-06,
      "loss": 0.8589080572128296,
      "step": 716
    },
    {
      "epoch": 2.860557768924303,
      "grad_norm": 0.3589399755001068,
      "learning_rate": 2.8681851204598675e-06,
      "loss": 0.6580166220664978,
      "step": 718
    },
    {
      "epoch": 2.8685258964143427,
      "grad_norm": 0.257345050573349,
      "learning_rate": 2.844176880676931e-06,
      "loss": 0.5470048785209656,
      "step": 720
    },
    {
      "epoch": 2.8764940239043826,
      "grad_norm": 0.6320711970329285,
      "learning_rate": 2.8202840849320867e-06,
      "loss": 0.5753337740898132,
      "step": 722
    },
    {
      "epoch": 2.8844621513944224,
      "grad_norm": 0.22646014392375946,
      "learning_rate": 2.7965077718038437e-06,
      "loss": 0.5759199261665344,
      "step": 724
    },
    {
      "epoch": 2.8924302788844622,
      "grad_norm": 0.2723312973976135,
      "learning_rate": 2.7728489748074193e-06,
      "loss": 0.8322911262512207,
      "step": 726
    },
    {
      "epoch": 2.900398406374502,
      "grad_norm": 0.21613740921020508,
      "learning_rate": 2.7493087223498015e-06,
      "loss": 0.49863553047180176,
      "step": 728
    },
    {
      "epoch": 2.908366533864542,
      "grad_norm": 0.25844162702560425,
      "learning_rate": 2.725888037685057e-06,
      "loss": 0.8254385590553284,
      "step": 730
    },
    {
      "epoch": 2.9163346613545817,
      "grad_norm": 0.4716702997684479,
      "learning_rate": 2.702587938869842e-06,
      "loss": 0.34096500277519226,
      "step": 732
    },
    {
      "epoch": 2.9243027888446216,
      "grad_norm": 0.34133321046829224,
      "learning_rate": 2.6794094387191617e-06,
      "loss": 0.890186071395874,
      "step": 734
    },
    {
      "epoch": 2.9322709163346614,
      "grad_norm": 1.1006333827972412,
      "learning_rate": 2.6563535447623302e-06,
      "loss": 0.4161735773086548,
      "step": 736
    },
    {
      "epoch": 2.9402390438247012,
      "grad_norm": 1.3671048879623413,
      "learning_rate": 2.6334212591991847e-06,
      "loss": 0.7910690307617188,
      "step": 738
    },
    {
      "epoch": 2.948207171314741,
      "grad_norm": 0.22590197622776031,
      "learning_rate": 2.6106135788565248e-06,
      "loss": 0.9301161170005798,
      "step": 740
    },
    {
      "epoch": 2.956175298804781,
      "grad_norm": 2.009145975112915,
      "learning_rate": 2.5879314951447688e-06,
      "loss": 0.3573679029941559,
      "step": 742
    },
    {
      "epoch": 2.9641434262948207,
      "grad_norm": 0.23535369336605072,
      "learning_rate": 2.5653759940148692e-06,
      "loss": 0.7875599265098572,
      "step": 744
    },
    {
      "epoch": 2.9721115537848606,
      "grad_norm": 0.35985681414604187,
      "learning_rate": 2.542948055915454e-06,
      "loss": 0.8422313928604126,
      "step": 746
    },
    {
      "epoch": 2.9800796812749004,
      "grad_norm": 0.21578483283519745,
      "learning_rate": 2.5206486557502087e-06,
      "loss": 0.45459961891174316,
      "step": 748
    },
    {
      "epoch": 2.9880478087649402,
      "grad_norm": 0.23628054559230804,
      "learning_rate": 2.498478762835492e-06,
      "loss": 0.13406522572040558,
      "step": 750
    },
    {
      "epoch": 2.99601593625498,
      "grad_norm": 0.23027797043323517,
      "learning_rate": 2.476439340858208e-06,
      "loss": 0.8818100690841675,
      "step": 752
    },
    {
      "epoch": 3.00398406374502,
      "grad_norm": 0.2023279219865799,
      "learning_rate": 2.4545313478339178e-06,
      "loss": 0.43777215480804443,
      "step": 754
    },
    {
      "epoch": 3.0119521912350598,
      "grad_norm": 0.26390212774276733,
      "learning_rate": 2.4327557360651895e-06,
      "loss": 0.47713276743888855,
      "step": 756
    },
    {
      "epoch": 3.0199203187250996,
      "grad_norm": 0.2905995845794678,
      "learning_rate": 2.411113452100211e-06,
      "loss": 0.23930948972702026,
      "step": 758
    },
    {
      "epoch": 3.0278884462151394,
      "grad_norm": 0.7211907505989075,
      "learning_rate": 2.3896054366916365e-06,
      "loss": 0.5089601874351501,
      "step": 760
    },
    {
      "epoch": 3.0358565737051793,
      "grad_norm": 0.8466331958770752,
      "learning_rate": 2.3682326247557e-06,
      "loss": 0.3438795208930969,
      "step": 762
    },
    {
      "epoch": 3.043824701195219,
      "grad_norm": 1.432413101196289,
      "learning_rate": 2.346995945331576e-06,
      "loss": 0.662820041179657,
      "step": 764
    },
    {
      "epoch": 3.051792828685259,
      "grad_norm": 0.7132479548454285,
      "learning_rate": 2.325896321540993e-06,
      "loss": 0.17177388072013855,
      "step": 766
    },
    {
      "epoch": 3.0597609561752988,
      "grad_norm": 0.29185473918914795,
      "learning_rate": 2.3049346705481086e-06,
      "loss": 0.4204632639884949,
      "step": 768
    },
    {
      "epoch": 3.0677290836653386,
      "grad_norm": 0.3408202528953552,
      "learning_rate": 2.284111903519641e-06,
      "loss": 0.5492228269577026,
      "step": 770
    },
    {
      "epoch": 3.0756972111553784,
      "grad_norm": 0.32708925008773804,
      "learning_rate": 2.26342892558526e-06,
      "loss": 0.6939098834991455,
      "step": 772
    },
    {
      "epoch": 3.0836653386454183,
      "grad_norm": 0.8320415019989014,
      "learning_rate": 2.2428866357982515e-06,
      "loss": 0.621051013469696,
      "step": 774
    },
    {
      "epoch": 3.091633466135458,
      "grad_norm": 0.03806205466389656,
      "learning_rate": 2.222485927096424e-06,
      "loss": 0.38304874300956726,
      "step": 776
    },
    {
      "epoch": 3.099601593625498,
      "grad_norm": 0.25648078322410583,
      "learning_rate": 2.2022276862633075e-06,
      "loss": 0.4233095049858093,
      "step": 778
    },
    {
      "epoch": 3.1075697211155378,
      "grad_norm": 0.2527969479560852,
      "learning_rate": 2.1821127938895938e-06,
      "loss": 0.6554626226425171,
      "step": 780
    },
    {
      "epoch": 3.1155378486055776,
      "grad_norm": 0.11690159142017365,
      "learning_rate": 2.162142124334871e-06,
      "loss": 0.26096802949905396,
      "step": 782
    },
    {
      "epoch": 3.1235059760956174,
      "grad_norm": 0.27716323733329773,
      "learning_rate": 2.1423165456896068e-06,
      "loss": 0.7666482329368591,
      "step": 784
    },
    {
      "epoch": 3.1314741035856573,
      "grad_norm": 0.233504980802536,
      "learning_rate": 2.1226369197374176e-06,
      "loss": 0.5519140958786011,
      "step": 786
    },
    {
      "epoch": 3.139442231075697,
      "grad_norm": 0.8129276633262634,
      "learning_rate": 2.103104101917615e-06,
      "loss": 0.4260145425796509,
      "step": 788
    },
    {
      "epoch": 3.147410358565737,
      "grad_norm": 2.6786768436431885,
      "learning_rate": 2.0837189412880044e-06,
      "loss": 0.35502249002456665,
      "step": 790
    },
    {
      "epoch": 3.1553784860557768,
      "grad_norm": 1.0312765836715698,
      "learning_rate": 2.0644822804879964e-06,
      "loss": 0.44495531916618347,
      "step": 792
    },
    {
      "epoch": 3.1633466135458166,
      "grad_norm": 0.06341253966093063,
      "learning_rate": 2.045394955701969e-06,
      "loss": 0.32705411314964294,
      "step": 794
    },
    {
      "epoch": 3.1713147410358564,
      "grad_norm": 0.6740862727165222,
      "learning_rate": 2.0264577966229222e-06,
      "loss": 0.484709233045578,
      "step": 796
    },
    {
      "epoch": 3.1792828685258963,
      "grad_norm": 1.0063650608062744,
      "learning_rate": 2.007671626416412e-06,
      "loss": 0.457725465297699,
      "step": 798
    },
    {
      "epoch": 3.187250996015936,
      "grad_norm": 0.41252169013023376,
      "learning_rate": 1.9890372616847663e-06,
      "loss": 0.6346026659011841,
      "step": 800
    },
    {
      "epoch": 3.195219123505976,
      "grad_norm": 0.18677018582820892,
      "learning_rate": 1.9705555124315965e-06,
      "loss": 0.4674780070781708,
      "step": 802
    },
    {
      "epoch": 3.2031872509960158,
      "grad_norm": 1.4164493083953857,
      "learning_rate": 1.95222718202658e-06,
      "loss": 0.5149998664855957,
      "step": 804
    },
    {
      "epoch": 3.2111553784860556,
      "grad_norm": 0.6007558107376099,
      "learning_rate": 1.934053067170545e-06,
      "loss": 0.6689565777778625,
      "step": 806
    },
    {
      "epoch": 3.2191235059760954,
      "grad_norm": 0.6316778659820557,
      "learning_rate": 1.9160339578608305e-06,
      "loss": 0.1798117458820343,
      "step": 808
    },
    {
      "epoch": 3.2270916334661353,
      "grad_norm": 0.24809357523918152,
      "learning_rate": 1.8981706373569602e-06,
      "loss": 0.47069182991981506,
      "step": 810
    },
    {
      "epoch": 3.235059760956175,
      "grad_norm": 0.7230861186981201,
      "learning_rate": 1.8804638821465821e-06,
      "loss": 0.3359948396682739,
      "step": 812
    },
    {
      "epoch": 3.243027888446215,
      "grad_norm": 0.24000729620456696,
      "learning_rate": 1.8629144619117234e-06,
      "loss": 0.5788341760635376,
      "step": 814
    },
    {
      "epoch": 3.2509960159362548,
      "grad_norm": 1.1049745082855225,
      "learning_rate": 1.8455231394953348e-06,
      "loss": 0.2038959264755249,
      "step": 816
    },
    {
      "epoch": 3.2589641434262946,
      "grad_norm": 0.23211677372455597,
      "learning_rate": 1.8282906708681238e-06,
      "loss": 0.6457157731056213,
      "step": 818
    },
    {
      "epoch": 3.2669322709163344,
      "grad_norm": 1.8020168542861938,
      "learning_rate": 1.811217805095699e-06,
      "loss": 0.3817474842071533,
      "step": 820
    },
    {
      "epoch": 3.2749003984063743,
      "grad_norm": 1.051099419593811,
      "learning_rate": 1.7943052843060133e-06,
      "loss": 0.46903231739997864,
      "step": 822
    },
    {
      "epoch": 3.2828685258964145,
      "grad_norm": 5.814231872558594,
      "learning_rate": 1.7775538436570937e-06,
      "loss": 0.4421203136444092,
      "step": 824
    },
    {
      "epoch": 3.2908366533864544,
      "grad_norm": 0.26077303290367126,
      "learning_rate": 1.7609642113050996e-06,
      "loss": 0.6923336386680603,
      "step": 826
    },
    {
      "epoch": 3.298804780876494,
      "grad_norm": 0.28432515263557434,
      "learning_rate": 1.7445371083726567e-06,
      "loss": 0.6867501139640808,
      "step": 828
    },
    {
      "epoch": 3.306772908366534,
      "grad_norm": 0.6227379441261292,
      "learning_rate": 1.7282732489175224e-06,
      "loss": 0.5801105499267578,
      "step": 830
    },
    {
      "epoch": 3.314741035856574,
      "grad_norm": 1.1650967597961426,
      "learning_rate": 1.7121733399015406e-06,
      "loss": 0.41004452109336853,
      "step": 832
    },
    {
      "epoch": 3.3227091633466137,
      "grad_norm": 1.9622670412063599,
      "learning_rate": 1.6962380811599138e-06,
      "loss": 0.2994910180568695,
      "step": 834
    },
    {
      "epoch": 3.3306772908366535,
      "grad_norm": 0.934207022190094,
      "learning_rate": 1.6804681653707816e-06,
      "loss": 0.41912150382995605,
      "step": 836
    },
    {
      "epoch": 3.3386454183266934,
      "grad_norm": 0.33417809009552,
      "learning_rate": 1.6648642780251096e-06,
      "loss": 0.637283980846405,
      "step": 838
    },
    {
      "epoch": 3.346613545816733,
      "grad_norm": 1.4292770624160767,
      "learning_rate": 1.6494270973968974e-06,
      "loss": 0.19428566098213196,
      "step": 840
    },
    {
      "epoch": 3.354581673306773,
      "grad_norm": 0.46277549862861633,
      "learning_rate": 1.6341572945136902e-06,
      "loss": 0.43366187810897827,
      "step": 842
    },
    {
      "epoch": 3.362549800796813,
      "grad_norm": 0.28821876645088196,
      "learning_rate": 1.6190555331274122e-06,
      "loss": 0.4826190769672394,
      "step": 844
    },
    {
      "epoch": 3.3705179282868527,
      "grad_norm": 0.3622353672981262,
      "learning_rate": 1.6041224696855145e-06,
      "loss": 0.711381733417511,
      "step": 846
    },
    {
      "epoch": 3.3784860557768925,
      "grad_norm": 0.2941267192363739,
      "learning_rate": 1.589358753302439e-06,
      "loss": 0.6465040445327759,
      "step": 848
    },
    {
      "epoch": 3.3864541832669324,
      "grad_norm": 0.719002902507782,
      "learning_rate": 1.5747650257314068e-06,
      "loss": 0.743245542049408,
      "step": 850
    },
    {
      "epoch": 3.394422310756972,
      "grad_norm": 0.29168501496315,
      "learning_rate": 1.5603419213365146e-06,
      "loss": 0.6720424890518188,
      "step": 852
    },
    {
      "epoch": 3.402390438247012,
      "grad_norm": 0.39615362882614136,
      "learning_rate": 1.5460900670651705e-06,
      "loss": 0.38796430826187134,
      "step": 854
    },
    {
      "epoch": 3.410358565737052,
      "grad_norm": 0.3184586465358734,
      "learning_rate": 1.5320100824208306e-06,
      "loss": 0.5875858664512634,
      "step": 856
    },
    {
      "epoch": 3.4183266932270917,
      "grad_norm": 0.2626410126686096,
      "learning_rate": 1.51810257943608e-06,
      "loss": 0.650122880935669,
      "step": 858
    },
    {
      "epoch": 3.4262948207171315,
      "grad_norm": 0.3557261824607849,
      "learning_rate": 1.5043681626460196e-06,
      "loss": 0.8494365811347961,
      "step": 860
    },
    {
      "epoch": 3.4342629482071714,
      "grad_norm": 0.24497739970684052,
      "learning_rate": 1.4908074290619972e-06,
      "loss": 0.431637167930603,
      "step": 862
    },
    {
      "epoch": 3.442231075697211,
      "grad_norm": 1.1705156564712524,
      "learning_rate": 1.477420968145652e-06,
      "loss": 0.5145531296730042,
      "step": 864
    },
    {
      "epoch": 3.450199203187251,
      "grad_norm": 0.369912326335907,
      "learning_rate": 1.4642093617832853e-06,
      "loss": 0.6736663579940796,
      "step": 866
    },
    {
      "epoch": 3.458167330677291,
      "grad_norm": 0.2663828730583191,
      "learning_rate": 1.4511731842605805e-06,
      "loss": 0.5267163515090942,
      "step": 868
    },
    {
      "epoch": 3.4661354581673307,
      "grad_norm": 0.2748737633228302,
      "learning_rate": 1.4383130022376298e-06,
      "loss": 0.38511818647384644,
      "step": 870
    },
    {
      "epoch": 3.4741035856573705,
      "grad_norm": 0.08580473065376282,
      "learning_rate": 1.4256293747243065e-06,
      "loss": 0.40568140149116516,
      "step": 872
    },
    {
      "epoch": 3.4820717131474104,
      "grad_norm": 0.11655905842781067,
      "learning_rate": 1.4131228530559637e-06,
      "loss": 0.4467930793762207,
      "step": 874
    },
    {
      "epoch": 3.49003984063745,
      "grad_norm": 0.24579553306102753,
      "learning_rate": 1.400793980869468e-06,
      "loss": 0.33361196517944336,
      "step": 876
    },
    {
      "epoch": 3.49800796812749,
      "grad_norm": 0.4674082100391388,
      "learning_rate": 1.3886432940795744e-06,
      "loss": 0.4383583068847656,
      "step": 878
    },
    {
      "epoch": 3.50597609561753,
      "grad_norm": 0.11904944479465485,
      "learning_rate": 1.3766713208556232e-06,
      "loss": 0.34720107913017273,
      "step": 880
    },
    {
      "epoch": 3.5139442231075697,
      "grad_norm": 0.22730985283851624,
      "learning_rate": 1.3648785815985876e-06,
      "loss": 0.278349906206131,
      "step": 882
    },
    {
      "epoch": 3.5219123505976095,
      "grad_norm": 6.253781795501709,
      "learning_rate": 1.353265588918448e-06,
      "loss": 0.16171355545520782,
      "step": 884
    },
    {
      "epoch": 3.5298804780876494,
      "grad_norm": 0.45735225081443787,
      "learning_rate": 1.3418328476119136e-06,
      "loss": 0.5551536679267883,
      "step": 886
    },
    {
      "epoch": 3.537848605577689,
      "grad_norm": 0.35474294424057007,
      "learning_rate": 1.330580854640477e-06,
      "loss": 0.6177930235862732,
      "step": 888
    },
    {
      "epoch": 3.545816733067729,
      "grad_norm": 0.43323010206222534,
      "learning_rate": 1.3195100991088142e-06,
      "loss": 0.3514676094055176,
      "step": 890
    },
    {
      "epoch": 3.553784860557769,
      "grad_norm": 0.7189896702766418,
      "learning_rate": 1.308621062243524e-06,
      "loss": 0.5006885528564453,
      "step": 892
    },
    {
      "epoch": 3.5617529880478087,
      "grad_norm": 0.259891152381897,
      "learning_rate": 1.2979142173722054e-06,
      "loss": 0.8925083875656128,
      "step": 894
    },
    {
      "epoch": 3.5697211155378485,
      "grad_norm": 0.12638863921165466,
      "learning_rate": 1.2873900299028896e-06,
      "loss": 0.13373291492462158,
      "step": 896
    },
    {
      "epoch": 3.5776892430278884,
      "grad_norm": 0.2114347517490387,
      "learning_rate": 1.2770489573038073e-06,
      "loss": 0.37734314799308777,
      "step": 898
    },
    {
      "epoch": 3.585657370517928,
      "grad_norm": 0.7182096838951111,
      "learning_rate": 1.2668914490834983e-06,
      "loss": 0.2670367956161499,
      "step": 900
    },
    {
      "epoch": 3.593625498007968,
      "grad_norm": 0.31399330496788025,
      "learning_rate": 1.2569179467712805e-06,
      "loss": 0.09790647029876709,
      "step": 902
    },
    {
      "epoch": 3.601593625498008,
      "grad_norm": 0.27743661403656006,
      "learning_rate": 1.2471288838980505e-06,
      "loss": 0.4643532335758209,
      "step": 904
    },
    {
      "epoch": 3.6095617529880477,
      "grad_norm": 0.6836845874786377,
      "learning_rate": 1.237524685977444e-06,
      "loss": 0.3613964319229126,
      "step": 906
    },
    {
      "epoch": 3.6175298804780875,
      "grad_norm": 0.2841336727142334,
      "learning_rate": 1.2281057704873331e-06,
      "loss": 0.6562983989715576,
      "step": 908
    },
    {
      "epoch": 3.6254980079681274,
      "grad_norm": 0.3593156635761261,
      "learning_rate": 1.2188725468516876e-06,
      "loss": 0.6140581369400024,
      "step": 910
    },
    {
      "epoch": 3.633466135458167,
      "grad_norm": 0.0681595429778099,
      "learning_rate": 1.2098254164227706e-06,
      "loss": 0.021015794947743416,
      "step": 912
    },
    {
      "epoch": 3.641434262948207,
      "grad_norm": 0.29087039828300476,
      "learning_rate": 1.200964772463695e-06,
      "loss": 0.4797722101211548,
      "step": 914
    },
    {
      "epoch": 3.649402390438247,
      "grad_norm": 0.5029170513153076,
      "learning_rate": 1.1922910001313312e-06,
      "loss": 0.5871038436889648,
      "step": 916
    },
    {
      "epoch": 3.6573705179282867,
      "grad_norm": 0.8182309865951538,
      "learning_rate": 1.1838044764595617e-06,
      "loss": 0.12434842437505722,
      "step": 918
    },
    {
      "epoch": 3.6653386454183265,
      "grad_norm": 0.25871777534484863,
      "learning_rate": 1.1755055703428958e-06,
      "loss": 0.6588802337646484,
      "step": 920
    },
    {
      "epoch": 3.6733067729083664,
      "grad_norm": 0.17757800221443176,
      "learning_rate": 1.1673946425204288e-06,
      "loss": 0.07635926455259323,
      "step": 922
    },
    {
      "epoch": 3.681274900398406,
      "grad_norm": 0.7696473598480225,
      "learning_rate": 1.1594720455601664e-06,
      "loss": 0.4940272271633148,
      "step": 924
    },
    {
      "epoch": 3.6892430278884465,
      "grad_norm": 1.6124153137207031,
      "learning_rate": 1.1517381238436986e-06,
      "loss": 0.2697877287864685,
      "step": 926
    },
    {
      "epoch": 3.6972111553784863,
      "grad_norm": 0.24673812091350555,
      "learning_rate": 1.1441932135512262e-06,
      "loss": 0.37748289108276367,
      "step": 928
    },
    {
      "epoch": 3.705179282868526,
      "grad_norm": 0.16552996635437012,
      "learning_rate": 1.1368376426469526e-06,
      "loss": 0.5062107443809509,
      "step": 930
    },
    {
      "epoch": 3.713147410358566,
      "grad_norm": 0.25886663794517517,
      "learning_rate": 1.1296717308648237e-06,
      "loss": 0.6300727128982544,
      "step": 932
    },
    {
      "epoch": 3.721115537848606,
      "grad_norm": 0.3514450788497925,
      "learning_rate": 1.1226957896946332e-06,
      "loss": 0.34530383348464966,
      "step": 934
    },
    {
      "epoch": 3.7290836653386457,
      "grad_norm": 0.28320619463920593,
      "learning_rate": 1.115910122368479e-06,
      "loss": 0.605874240398407,
      "step": 936
    },
    {
      "epoch": 3.7370517928286855,
      "grad_norm": 0.6926787495613098,
      "learning_rate": 1.1093150238475847e-06,
      "loss": 0.3987686038017273,
      "step": 938
    },
    {
      "epoch": 3.7450199203187253,
      "grad_norm": 0.3855150640010834,
      "learning_rate": 1.1029107808094789e-06,
      "loss": 0.7403500080108643,
      "step": 940
    },
    {
      "epoch": 3.752988047808765,
      "grad_norm": 0.9000049829483032,
      "learning_rate": 1.0966976716355294e-06,
      "loss": 0.512790322303772,
      "step": 942
    },
    {
      "epoch": 3.760956175298805,
      "grad_norm": 0.2785480320453644,
      "learning_rate": 1.0906759663988475e-06,
      "loss": 0.3739131987094879,
      "step": 944
    },
    {
      "epoch": 3.768924302788845,
      "grad_norm": 0.28096991777420044,
      "learning_rate": 1.084845926852547e-06,
      "loss": 0.7322132587432861,
      "step": 946
    },
    {
      "epoch": 3.7768924302788847,
      "grad_norm": 0.28296729922294617,
      "learning_rate": 1.0792078064183652e-06,
      "loss": 0.4156455099582672,
      "step": 948
    },
    {
      "epoch": 3.7848605577689245,
      "grad_norm": 0.2130855917930603,
      "learning_rate": 1.0737618501756455e-06,
      "loss": 0.6804937720298767,
      "step": 950
    },
    {
      "epoch": 3.7928286852589643,
      "grad_norm": 0.29680532217025757,
      "learning_rate": 1.0685082948506875e-06,
      "loss": 0.6481207013130188,
      "step": 952
    },
    {
      "epoch": 3.800796812749004,
      "grad_norm": 0.21896500885486603,
      "learning_rate": 1.063447368806457e-06,
      "loss": 0.38140422105789185,
      "step": 954
    },
    {
      "epoch": 3.808764940239044,
      "grad_norm": 1.2831289768218994,
      "learning_rate": 1.0585792920326565e-06,
      "loss": 0.5196990966796875,
      "step": 956
    },
    {
      "epoch": 3.816733067729084,
      "grad_norm": 1.1605606079101562,
      "learning_rate": 1.0539042761361646e-06,
      "loss": 0.40277406573295593,
      "step": 958
    },
    {
      "epoch": 3.8247011952191237,
      "grad_norm": 0.544163703918457,
      "learning_rate": 1.0494225243318373e-06,
      "loss": 0.37141847610473633,
      "step": 960
    },
    {
      "epoch": 3.8326693227091635,
      "grad_norm": 0.12706421315670013,
      "learning_rate": 1.0451342314336755e-06,
      "loss": 0.4112641215324402,
      "step": 962
    },
    {
      "epoch": 3.8406374501992033,
      "grad_norm": 0.3214135766029358,
      "learning_rate": 1.0410395838463547e-06,
      "loss": 0.7260571122169495,
      "step": 964
    },
    {
      "epoch": 3.848605577689243,
      "grad_norm": 0.6689302921295166,
      "learning_rate": 1.0371387595571248e-06,
      "loss": 0.22925657033920288,
      "step": 966
    },
    {
      "epoch": 3.856573705179283,
      "grad_norm": 0.23933514952659607,
      "learning_rate": 1.0334319281280715e-06,
      "loss": 0.4337913990020752,
      "step": 968
    },
    {
      "epoch": 3.864541832669323,
      "grad_norm": 10.142091751098633,
      "learning_rate": 1.0299192506887467e-06,
      "loss": 0.5167869329452515,
      "step": 970
    },
    {
      "epoch": 3.8725099601593627,
      "grad_norm": 0.09362570941448212,
      "learning_rate": 1.0266008799291635e-06,
      "loss": 0.44974231719970703,
      "step": 972
    },
    {
      "epoch": 3.8804780876494025,
      "grad_norm": 0.24123288691043854,
      "learning_rate": 1.0234769600931612e-06,
      "loss": 0.37262213230133057,
      "step": 974
    },
    {
      "epoch": 3.8884462151394423,
      "grad_norm": 0.27388399839401245,
      "learning_rate": 1.0205476269721312e-06,
      "loss": 0.6645011901855469,
      "step": 976
    },
    {
      "epoch": 3.896414342629482,
      "grad_norm": 0.4157608151435852,
      "learning_rate": 1.0178130078991196e-06,
      "loss": 0.37185800075531006,
      "step": 978
    },
    {
      "epoch": 3.904382470119522,
      "grad_norm": 0.14850561320781708,
      "learning_rate": 1.0152732217432875e-06,
      "loss": 0.3079684376716614,
      "step": 980
    },
    {
      "epoch": 3.912350597609562,
      "grad_norm": 0.48838549852371216,
      "learning_rate": 1.0129283789047473e-06,
      "loss": 0.29368099570274353,
      "step": 982
    },
    {
      "epoch": 3.9203187250996017,
      "grad_norm": 0.4035380184650421,
      "learning_rate": 1.0107785813097621e-06,
      "loss": 0.759061872959137,
      "step": 984
    },
    {
      "epoch": 3.9282868525896415,
      "grad_norm": 0.5321217179298401,
      "learning_rate": 1.0088239224063153e-06,
      "loss": 0.16407278180122375,
      "step": 986
    },
    {
      "epoch": 3.9362549800796813,
      "grad_norm": 0.9683710336685181,
      "learning_rate": 1.0070644871600495e-06,
      "loss": 0.48237884044647217,
      "step": 988
    },
    {
      "epoch": 3.944223107569721,
      "grad_norm": 0.09458450227975845,
      "learning_rate": 1.0055003520505714e-06,
      "loss": 0.3799082338809967,
      "step": 990
    },
    {
      "epoch": 3.952191235059761,
      "grad_norm": 0.2854856848716736,
      "learning_rate": 1.00413158506813e-06,
      "loss": 0.6594796776771545,
      "step": 992
    },
    {
      "epoch": 3.960159362549801,
      "grad_norm": 0.8761612176895142,
      "learning_rate": 1.0029582457106588e-06,
      "loss": 0.1987125426530838,
      "step": 994
    },
    {
      "epoch": 3.9681274900398407,
      "grad_norm": 0.4088381826877594,
      "learning_rate": 1.00198038498119e-06,
      "loss": 0.5873836874961853,
      "step": 996
    },
    {
      "epoch": 3.9760956175298805,
      "grad_norm": 0.13188602030277252,
      "learning_rate": 1.0011980453856391e-06,
      "loss": 0.40052562952041626,
      "step": 998
    },
    {
      "epoch": 3.9840637450199203,
      "grad_norm": 0.33241450786590576,
      "learning_rate": 1.0006112609309548e-06,
      "loss": 0.3138795793056488,
      "step": 1000
    },
    {
      "epoch": 3.99203187250996,
      "grad_norm": 0.2598963677883148,
      "learning_rate": 1.0002200571236429e-06,
      "loss": 0.6573696732521057,
      "step": 1002
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.18074186146259308,
      "learning_rate": 1.0000244509686564e-06,
      "loss": 0.1689499020576477,
      "step": 1004
    },
    {
      "epoch": 4.0,
      "step": 1004,
      "total_flos": 4.038502240003031e+18,
      "train_loss": 0.7884150154599215,
      "train_runtime": 10532.0612,
      "train_samples_per_second": 5.72,
      "train_steps_per_second": 0.095
    }
  ],
  "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": 4.038502240003031e+18,
  "train_batch_size": 3,
  "trial_name": null,
  "trial_params": null
}