{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9002776297708853,
  "eval_steps": 2378,
  "global_step": 10701,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001682604672032306,
      "grad_norm": 0.890625,
      "learning_rate": 1.260504201680672e-06,
      "loss": 2.4691038131713867,
      "step": 2
    },
    {
      "epoch": 0.0003365209344064612,
      "grad_norm": 0.8203125,
      "learning_rate": 3.7815126050420167e-06,
      "loss": 2.366840362548828,
      "step": 4
    },
    {
      "epoch": 0.0005047814016096918,
      "grad_norm": 0.88671875,
      "learning_rate": 6.3025210084033615e-06,
      "loss": 2.3581132888793945,
      "step": 6
    },
    {
      "epoch": 0.0006730418688129224,
      "grad_norm": 0.87109375,
      "learning_rate": 8.823529411764705e-06,
      "loss": 2.3826358318328857,
      "step": 8
    },
    {
      "epoch": 0.000841302336016153,
      "grad_norm": 0.87109375,
      "learning_rate": 1.1344537815126049e-05,
      "loss": 2.362799644470215,
      "step": 10
    },
    {
      "epoch": 0.0010095628032193837,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3865546218487394e-05,
      "loss": 2.4329168796539307,
      "step": 12
    },
    {
      "epoch": 0.0011778232704226141,
      "grad_norm": 0.9140625,
      "learning_rate": 1.6386554621848738e-05,
      "loss": 2.4057555198669434,
      "step": 14
    },
    {
      "epoch": 0.0013460837376258448,
      "grad_norm": 0.84375,
      "learning_rate": 1.8907563025210083e-05,
      "loss": 2.37949275970459,
      "step": 16
    },
    {
      "epoch": 0.0015143442048290755,
      "grad_norm": 0.81640625,
      "learning_rate": 2.1428571428571425e-05,
      "loss": 2.382742166519165,
      "step": 18
    },
    {
      "epoch": 0.001682604672032306,
      "grad_norm": 0.7421875,
      "learning_rate": 2.394957983193277e-05,
      "loss": 2.415689468383789,
      "step": 20
    },
    {
      "epoch": 0.0018508651392355367,
      "grad_norm": 0.8046875,
      "learning_rate": 2.647058823529412e-05,
      "loss": 2.3525829315185547,
      "step": 22
    },
    {
      "epoch": 0.0020191256064387673,
      "grad_norm": 0.765625,
      "learning_rate": 2.899159663865546e-05,
      "loss": 2.3600056171417236,
      "step": 24
    },
    {
      "epoch": 0.002187386073641998,
      "grad_norm": 0.69921875,
      "learning_rate": 3.1512605042016806e-05,
      "loss": 2.449960947036743,
      "step": 26
    },
    {
      "epoch": 0.0023556465408452283,
      "grad_norm": 0.69921875,
      "learning_rate": 3.403361344537815e-05,
      "loss": 2.3475403785705566,
      "step": 28
    },
    {
      "epoch": 0.002523907008048459,
      "grad_norm": 0.734375,
      "learning_rate": 3.655462184873949e-05,
      "loss": 2.3883109092712402,
      "step": 30
    },
    {
      "epoch": 0.0026921674752516896,
      "grad_norm": 0.7421875,
      "learning_rate": 3.907563025210084e-05,
      "loss": 2.4003872871398926,
      "step": 32
    },
    {
      "epoch": 0.00286042794245492,
      "grad_norm": 0.59765625,
      "learning_rate": 4.159663865546218e-05,
      "loss": 2.34065842628479,
      "step": 34
    },
    {
      "epoch": 0.003028688409658151,
      "grad_norm": 0.609375,
      "learning_rate": 4.4117647058823526e-05,
      "loss": 2.342637538909912,
      "step": 36
    },
    {
      "epoch": 0.0031969488768613815,
      "grad_norm": 0.55078125,
      "learning_rate": 4.663865546218487e-05,
      "loss": 2.3691353797912598,
      "step": 38
    },
    {
      "epoch": 0.003365209344064612,
      "grad_norm": 0.4921875,
      "learning_rate": 4.9159663865546216e-05,
      "loss": 2.306152820587158,
      "step": 40
    },
    {
      "epoch": 0.0035334698112678424,
      "grad_norm": 0.44140625,
      "learning_rate": 5.168067226890756e-05,
      "loss": 2.3288981914520264,
      "step": 42
    },
    {
      "epoch": 0.0037017302784710733,
      "grad_norm": 0.412109375,
      "learning_rate": 5.42016806722689e-05,
      "loss": 2.3376402854919434,
      "step": 44
    },
    {
      "epoch": 0.003869990745674304,
      "grad_norm": 0.451171875,
      "learning_rate": 5.6722689075630245e-05,
      "loss": 2.2842459678649902,
      "step": 46
    },
    {
      "epoch": 0.004038251212877535,
      "grad_norm": 0.38671875,
      "learning_rate": 5.924369747899159e-05,
      "loss": 2.2941184043884277,
      "step": 48
    },
    {
      "epoch": 0.004206511680080765,
      "grad_norm": 0.5,
      "learning_rate": 6.176470588235294e-05,
      "loss": 2.2915849685668945,
      "step": 50
    },
    {
      "epoch": 0.004374772147283996,
      "grad_norm": 0.345703125,
      "learning_rate": 6.428571428571427e-05,
      "loss": 2.247030258178711,
      "step": 52
    },
    {
      "epoch": 0.0045430326144872265,
      "grad_norm": 0.365234375,
      "learning_rate": 6.680672268907563e-05,
      "loss": 2.208951234817505,
      "step": 54
    },
    {
      "epoch": 0.0047112930816904566,
      "grad_norm": 0.302734375,
      "learning_rate": 6.932773109243698e-05,
      "loss": 2.2868590354919434,
      "step": 56
    },
    {
      "epoch": 0.0048795535488936875,
      "grad_norm": 0.275390625,
      "learning_rate": 7.184873949579832e-05,
      "loss": 2.240825653076172,
      "step": 58
    },
    {
      "epoch": 0.005047814016096918,
      "grad_norm": 0.275390625,
      "learning_rate": 7.436974789915966e-05,
      "loss": 2.274174213409424,
      "step": 60
    },
    {
      "epoch": 0.005216074483300148,
      "grad_norm": 0.25390625,
      "learning_rate": 7.6890756302521e-05,
      "loss": 2.26889967918396,
      "step": 62
    },
    {
      "epoch": 0.005384334950503379,
      "grad_norm": 0.265625,
      "learning_rate": 7.941176470588235e-05,
      "loss": 2.218982696533203,
      "step": 64
    },
    {
      "epoch": 0.00555259541770661,
      "grad_norm": 0.259765625,
      "learning_rate": 8.193277310924368e-05,
      "loss": 2.244980812072754,
      "step": 66
    },
    {
      "epoch": 0.00572085588490984,
      "grad_norm": 0.275390625,
      "learning_rate": 8.445378151260504e-05,
      "loss": 2.1840691566467285,
      "step": 68
    },
    {
      "epoch": 0.005889116352113071,
      "grad_norm": 0.28515625,
      "learning_rate": 8.697478991596639e-05,
      "loss": 2.2637429237365723,
      "step": 70
    },
    {
      "epoch": 0.006057376819316302,
      "grad_norm": 0.26953125,
      "learning_rate": 8.949579831932773e-05,
      "loss": 2.186166763305664,
      "step": 72
    },
    {
      "epoch": 0.006225637286519532,
      "grad_norm": 0.265625,
      "learning_rate": 9.201680672268908e-05,
      "loss": 2.197683572769165,
      "step": 74
    },
    {
      "epoch": 0.006393897753722763,
      "grad_norm": 0.2470703125,
      "learning_rate": 9.45378151260504e-05,
      "loss": 2.14741849899292,
      "step": 76
    },
    {
      "epoch": 0.006562158220925993,
      "grad_norm": 0.24609375,
      "learning_rate": 9.705882352941176e-05,
      "loss": 2.1839256286621094,
      "step": 78
    },
    {
      "epoch": 0.006730418688129224,
      "grad_norm": 0.2158203125,
      "learning_rate": 9.95798319327731e-05,
      "loss": 2.1234772205352783,
      "step": 80
    },
    {
      "epoch": 0.006898679155332455,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00010210084033613445,
      "loss": 2.176161289215088,
      "step": 82
    },
    {
      "epoch": 0.007066939622535685,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.00010462184873949579,
      "loss": 2.1807897090911865,
      "step": 84
    },
    {
      "epoch": 0.007235200089738916,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00010714285714285714,
      "loss": 2.0876541137695312,
      "step": 86
    },
    {
      "epoch": 0.007403460556942147,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00010966386554621848,
      "loss": 2.1456685066223145,
      "step": 88
    },
    {
      "epoch": 0.007571721024145377,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00011218487394957983,
      "loss": 2.155117988586426,
      "step": 90
    },
    {
      "epoch": 0.007739981491348608,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00011470588235294115,
      "loss": 2.0569896697998047,
      "step": 92
    },
    {
      "epoch": 0.007908241958551838,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0001172268907563025,
      "loss": 2.0766124725341797,
      "step": 94
    },
    {
      "epoch": 0.00807650242575507,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00011974789915966386,
      "loss": 2.1183228492736816,
      "step": 96
    },
    {
      "epoch": 0.0082447628929583,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0001222689075630252,
      "loss": 2.1125264167785645,
      "step": 98
    },
    {
      "epoch": 0.00841302336016153,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00012478991596638655,
      "loss": 2.1278624534606934,
      "step": 100
    },
    {
      "epoch": 0.00858128382736476,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00012731092436974787,
      "loss": 2.1354124546051025,
      "step": 102
    },
    {
      "epoch": 0.008749544294567991,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00012983193277310922,
      "loss": 2.1045663356781006,
      "step": 104
    },
    {
      "epoch": 0.008917804761771222,
      "grad_norm": 0.169921875,
      "learning_rate": 0.00013235294117647058,
      "loss": 2.1220180988311768,
      "step": 106
    },
    {
      "epoch": 0.009086065228974453,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.00013487394957983193,
      "loss": 2.0815682411193848,
      "step": 108
    },
    {
      "epoch": 0.009254325696177682,
      "grad_norm": 0.1875,
      "learning_rate": 0.00013739495798319325,
      "loss": 2.1093480587005615,
      "step": 110
    },
    {
      "epoch": 0.009422586163380913,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0001399159663865546,
      "loss": 2.052865505218506,
      "step": 112
    },
    {
      "epoch": 0.009590846630584144,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00014243697478991596,
      "loss": 1.9954745769500732,
      "step": 114
    },
    {
      "epoch": 0.009759107097787375,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0001449579831932773,
      "loss": 2.0809431076049805,
      "step": 116
    },
    {
      "epoch": 0.009927367564990606,
      "grad_norm": 0.1640625,
      "learning_rate": 0.00014747899159663863,
      "loss": 2.1390209197998047,
      "step": 118
    },
    {
      "epoch": 0.010095628032193837,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00015,
      "loss": 2.0581886768341064,
      "step": 120
    },
    {
      "epoch": 0.010263888499397066,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0001525210084033613,
      "loss": 2.063462257385254,
      "step": 122
    },
    {
      "epoch": 0.010432148966600297,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0001550420168067227,
      "loss": 2.127774477005005,
      "step": 124
    },
    {
      "epoch": 0.010600409433803528,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00015756302521008402,
      "loss": 2.071662425994873,
      "step": 126
    },
    {
      "epoch": 0.010768669901006759,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00016008403361344537,
      "loss": 2.0585618019104004,
      "step": 128
    },
    {
      "epoch": 0.01093693036820999,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0001626050420168067,
      "loss": 2.017852306365967,
      "step": 130
    },
    {
      "epoch": 0.01110519083541322,
      "grad_norm": 0.15625,
      "learning_rate": 0.00016512605042016807,
      "loss": 2.0723910331726074,
      "step": 132
    },
    {
      "epoch": 0.01127345130261645,
      "grad_norm": 0.171875,
      "learning_rate": 0.0001676470588235294,
      "loss": 2.0319981575012207,
      "step": 134
    },
    {
      "epoch": 0.01144171176981968,
      "grad_norm": 0.1640625,
      "learning_rate": 0.00017016806722689075,
      "loss": 2.1004624366760254,
      "step": 136
    },
    {
      "epoch": 0.011609972237022911,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00017268907563025207,
      "loss": 2.084381103515625,
      "step": 138
    },
    {
      "epoch": 0.011778232704226142,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00017521008403361345,
      "loss": 2.02555513381958,
      "step": 140
    },
    {
      "epoch": 0.011946493171429373,
      "grad_norm": 0.166015625,
      "learning_rate": 0.00017773109243697478,
      "loss": 2.1459860801696777,
      "step": 142
    },
    {
      "epoch": 0.012114753638632604,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0001802521008403361,
      "loss": 2.1069960594177246,
      "step": 144
    },
    {
      "epoch": 0.012283014105835833,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00018277310924369746,
      "loss": 2.1008729934692383,
      "step": 146
    },
    {
      "epoch": 0.012451274573039064,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0001852941176470588,
      "loss": 2.064427137374878,
      "step": 148
    },
    {
      "epoch": 0.012619535040242295,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.00018781512605042016,
      "loss": 1.9980311393737793,
      "step": 150
    },
    {
      "epoch": 0.012787795507445526,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00019033613445378148,
      "loss": 2.062361717224121,
      "step": 152
    },
    {
      "epoch": 0.012956055974648757,
      "grad_norm": 0.15625,
      "learning_rate": 0.00019285714285714286,
      "loss": 2.053767681121826,
      "step": 154
    },
    {
      "epoch": 0.013124316441851986,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0001953781512605042,
      "loss": 2.0441789627075195,
      "step": 156
    },
    {
      "epoch": 0.013292576909055217,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0001978991596638655,
      "loss": 2.049647092819214,
      "step": 158
    },
    {
      "epoch": 0.013460837376258448,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.00020042016806722687,
      "loss": 2.0260157585144043,
      "step": 160
    },
    {
      "epoch": 0.013629097843461679,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00020294117647058822,
      "loss": 2.084946870803833,
      "step": 162
    },
    {
      "epoch": 0.01379735831066491,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00020546218487394957,
      "loss": 2.0785281658172607,
      "step": 164
    },
    {
      "epoch": 0.01396561877786814,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002079831932773109,
      "loss": 2.1569368839263916,
      "step": 166
    },
    {
      "epoch": 0.01413387924507137,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00021050420168067225,
      "loss": 2.0413806438446045,
      "step": 168
    },
    {
      "epoch": 0.0143021397122746,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002130252100840336,
      "loss": 2.119694471359253,
      "step": 170
    },
    {
      "epoch": 0.014470400179477831,
      "grad_norm": 0.2421875,
      "learning_rate": 0.00021554621848739495,
      "loss": 2.0799708366394043,
      "step": 172
    },
    {
      "epoch": 0.014638660646681062,
      "grad_norm": 0.15625,
      "learning_rate": 0.00021806722689075628,
      "loss": 2.0739758014678955,
      "step": 174
    },
    {
      "epoch": 0.014806921113884293,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00022058823529411765,
      "loss": 2.0421411991119385,
      "step": 176
    },
    {
      "epoch": 0.014975181581087524,
      "grad_norm": 0.158203125,
      "learning_rate": 0.00022310924369747898,
      "loss": 2.087824821472168,
      "step": 178
    },
    {
      "epoch": 0.015143442048290753,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002256302521008403,
      "loss": 1.9886949062347412,
      "step": 180
    },
    {
      "epoch": 0.015311702515493984,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00022815126050420166,
      "loss": 1.989719033241272,
      "step": 182
    },
    {
      "epoch": 0.015479962982697215,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.000230672268907563,
      "loss": 2.085930347442627,
      "step": 184
    },
    {
      "epoch": 0.015648223449900446,
      "grad_norm": 0.169921875,
      "learning_rate": 0.00023319327731092436,
      "loss": 2.091665029525757,
      "step": 186
    },
    {
      "epoch": 0.015816483917103677,
      "grad_norm": 0.15625,
      "learning_rate": 0.00023571428571428569,
      "loss": 2.031763792037964,
      "step": 188
    },
    {
      "epoch": 0.015984744384306908,
      "grad_norm": 0.1796875,
      "learning_rate": 0.000238235294117647,
      "loss": 2.0749149322509766,
      "step": 190
    },
    {
      "epoch": 0.01615300485151014,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002407563025210084,
      "loss": 1.9726346731185913,
      "step": 192
    },
    {
      "epoch": 0.01632126531871337,
      "grad_norm": 0.15625,
      "learning_rate": 0.00024327731092436971,
      "loss": 2.0352015495300293,
      "step": 194
    },
    {
      "epoch": 0.0164895257859166,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00024579831932773107,
      "loss": 2.0122499465942383,
      "step": 196
    },
    {
      "epoch": 0.016657786253119828,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002483193277310924,
      "loss": 2.087919235229492,
      "step": 198
    },
    {
      "epoch": 0.01682604672032306,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00025084033613445377,
      "loss": 1.9917337894439697,
      "step": 200
    },
    {
      "epoch": 0.01699430718752629,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002533613445378151,
      "loss": 1.9840965270996094,
      "step": 202
    },
    {
      "epoch": 0.01716256765472952,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002558823529411764,
      "loss": 1.984767198562622,
      "step": 204
    },
    {
      "epoch": 0.01733082812193275,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00025840336134453783,
      "loss": 2.0480666160583496,
      "step": 206
    },
    {
      "epoch": 0.017499088589135983,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002609243697478991,
      "loss": 1.989534854888916,
      "step": 208
    },
    {
      "epoch": 0.017667349056339213,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002634453781512605,
      "loss": 2.0356667041778564,
      "step": 210
    },
    {
      "epoch": 0.017835609523542444,
      "grad_norm": 0.248046875,
      "learning_rate": 0.00026596638655462183,
      "loss": 2.069643259048462,
      "step": 212
    },
    {
      "epoch": 0.018003869990745675,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002684873949579832,
      "loss": 2.0190212726593018,
      "step": 214
    },
    {
      "epoch": 0.018172130457948906,
      "grad_norm": 0.259765625,
      "learning_rate": 0.00027100840336134453,
      "loss": 2.0430734157562256,
      "step": 216
    },
    {
      "epoch": 0.018340390925152137,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00027352941176470583,
      "loss": 2.0029714107513428,
      "step": 218
    },
    {
      "epoch": 0.018508651392355364,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002760504201680672,
      "loss": 1.9820024967193604,
      "step": 220
    },
    {
      "epoch": 0.018676911859558595,
      "grad_norm": 0.203125,
      "learning_rate": 0.00027857142857142854,
      "loss": 2.091647148132324,
      "step": 222
    },
    {
      "epoch": 0.018845172326761826,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002810924369747899,
      "loss": 2.0615999698638916,
      "step": 224
    },
    {
      "epoch": 0.019013432793965057,
      "grad_norm": 0.1640625,
      "learning_rate": 0.00028361344537815124,
      "loss": 2.096843957901001,
      "step": 226
    },
    {
      "epoch": 0.019181693261168288,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002861344537815126,
      "loss": 1.99099862575531,
      "step": 228
    },
    {
      "epoch": 0.01934995372837152,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00028865546218487394,
      "loss": 2.0636544227600098,
      "step": 230
    },
    {
      "epoch": 0.01951821419557475,
      "grad_norm": 0.16796875,
      "learning_rate": 0.00029117647058823524,
      "loss": 2.0021512508392334,
      "step": 232
    },
    {
      "epoch": 0.01968647466277798,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002936974789915966,
      "loss": 1.985487461090088,
      "step": 234
    },
    {
      "epoch": 0.01985473512998121,
      "grad_norm": 0.16796875,
      "learning_rate": 0.00029621848739495795,
      "loss": 2.0403635501861572,
      "step": 236
    },
    {
      "epoch": 0.020022995597184443,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002987394957983193,
      "loss": 2.060939311981201,
      "step": 238
    },
    {
      "epoch": 0.020191256064387673,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 2.04160213470459,
      "step": 240
    },
    {
      "epoch": 0.020359516531590904,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 2.073460102081299,
      "step": 242
    },
    {
      "epoch": 0.020527776998794132,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 2.1027708053588867,
      "step": 244
    },
    {
      "epoch": 0.020696037465997363,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 2.0462005138397217,
      "step": 246
    },
    {
      "epoch": 0.020864297933200594,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 2.0052802562713623,
      "step": 248
    },
    {
      "epoch": 0.021032558400403824,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 2.0267109870910645,
      "step": 250
    },
    {
      "epoch": 0.021200818867607055,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 2.0172860622406006,
      "step": 252
    },
    {
      "epoch": 0.021369079334810286,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9662808179855347,
      "step": 254
    },
    {
      "epoch": 0.021537339802013517,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 2.0605382919311523,
      "step": 256
    },
    {
      "epoch": 0.021705600269216748,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 2.0562660694122314,
      "step": 258
    },
    {
      "epoch": 0.02187386073641998,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 2.0552806854248047,
      "step": 260
    },
    {
      "epoch": 0.02204212120362321,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 2.0619161128997803,
      "step": 262
    },
    {
      "epoch": 0.02221038167082644,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9910318851470947,
      "step": 264
    },
    {
      "epoch": 0.022378642138029668,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 2.032259941101074,
      "step": 266
    },
    {
      "epoch": 0.0225469026052329,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 2.003819227218628,
      "step": 268
    },
    {
      "epoch": 0.02271516307243613,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9919540882110596,
      "step": 270
    },
    {
      "epoch": 0.02288342353963936,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 2.036320209503174,
      "step": 272
    },
    {
      "epoch": 0.023051684006842592,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 2.006593704223633,
      "step": 274
    },
    {
      "epoch": 0.023219944474045823,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.9652570486068726,
      "step": 276
    },
    {
      "epoch": 0.023388204941249054,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 2.0443458557128906,
      "step": 278
    },
    {
      "epoch": 0.023556465408452285,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 2.015209436416626,
      "step": 280
    },
    {
      "epoch": 0.023724725875655515,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9911295175552368,
      "step": 282
    },
    {
      "epoch": 0.023892986342858746,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 2.0124270915985107,
      "step": 284
    },
    {
      "epoch": 0.024061246810061977,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9906249046325684,
      "step": 286
    },
    {
      "epoch": 0.024229507277265208,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 2.053417205810547,
      "step": 288
    },
    {
      "epoch": 0.024397767744468436,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 2.058159351348877,
      "step": 290
    },
    {
      "epoch": 0.024566028211671666,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 2.011868953704834,
      "step": 292
    },
    {
      "epoch": 0.024734288678874897,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9976524114608765,
      "step": 294
    },
    {
      "epoch": 0.02490254914607813,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 2.03645396232605,
      "step": 296
    },
    {
      "epoch": 0.02507080961328136,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.9487731456756592,
      "step": 298
    },
    {
      "epoch": 0.02523907008048459,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9961670637130737,
      "step": 300
    },
    {
      "epoch": 0.02540733054768782,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 2.0677599906921387,
      "step": 302
    },
    {
      "epoch": 0.025575591014891052,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 2.029195785522461,
      "step": 304
    },
    {
      "epoch": 0.025743851482094283,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 2.0025458335876465,
      "step": 306
    },
    {
      "epoch": 0.025912111949297514,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 2.0746097564697266,
      "step": 308
    },
    {
      "epoch": 0.026080372416500745,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 2.036525011062622,
      "step": 310
    },
    {
      "epoch": 0.026248632883703972,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.9544272422790527,
      "step": 312
    },
    {
      "epoch": 0.026416893350907203,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 2.0170743465423584,
      "step": 314
    },
    {
      "epoch": 0.026585153818110434,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0003,
      "loss": 1.922873616218567,
      "step": 316
    },
    {
      "epoch": 0.026753414285313665,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9666386842727661,
      "step": 318
    },
    {
      "epoch": 0.026921674752516896,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 2.0089683532714844,
      "step": 320
    },
    {
      "epoch": 0.027089935219720127,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9875130653381348,
      "step": 322
    },
    {
      "epoch": 0.027258195686923357,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003,
      "loss": 1.9853742122650146,
      "step": 324
    },
    {
      "epoch": 0.02742645615412659,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9890427589416504,
      "step": 326
    },
    {
      "epoch": 0.02759471662132982,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 2.1050667762756348,
      "step": 328
    },
    {
      "epoch": 0.02776297708853305,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 2.0180253982543945,
      "step": 330
    },
    {
      "epoch": 0.02793123755573628,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 2.032142162322998,
      "step": 332
    },
    {
      "epoch": 0.028099498022939512,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 2.0697362422943115,
      "step": 334
    },
    {
      "epoch": 0.02826775849014274,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 2.0211009979248047,
      "step": 336
    },
    {
      "epoch": 0.02843601895734597,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 2.0108823776245117,
      "step": 338
    },
    {
      "epoch": 0.0286042794245492,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 2.02127742767334,
      "step": 340
    },
    {
      "epoch": 0.028772539891752432,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9652338027954102,
      "step": 342
    },
    {
      "epoch": 0.028940800358955663,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 2.0384597778320312,
      "step": 344
    },
    {
      "epoch": 0.029109060826158894,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 2.017561674118042,
      "step": 346
    },
    {
      "epoch": 0.029277321293362125,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 2.062777519226074,
      "step": 348
    },
    {
      "epoch": 0.029445581760565356,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.988010048866272,
      "step": 350
    },
    {
      "epoch": 0.029613842227768587,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 2.0020999908447266,
      "step": 352
    },
    {
      "epoch": 0.029782102694971817,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 2.039154052734375,
      "step": 354
    },
    {
      "epoch": 0.02995036316217505,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 2.024914264678955,
      "step": 356
    },
    {
      "epoch": 0.03011862362937828,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.979623556137085,
      "step": 358
    },
    {
      "epoch": 0.030286884096581507,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 2.0230166912078857,
      "step": 360
    },
    {
      "epoch": 0.030455144563784738,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9747154712677002,
      "step": 362
    },
    {
      "epoch": 0.03062340503098797,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.988269329071045,
      "step": 364
    },
    {
      "epoch": 0.0307916654981912,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 1.9855906963348389,
      "step": 366
    },
    {
      "epoch": 0.03095992596539443,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0003,
      "loss": 1.953933596611023,
      "step": 368
    },
    {
      "epoch": 0.03112818643259766,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.9796221256256104,
      "step": 370
    },
    {
      "epoch": 0.03129644689980089,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9828559160232544,
      "step": 372
    },
    {
      "epoch": 0.03146470736700412,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 2.0373551845550537,
      "step": 374
    },
    {
      "epoch": 0.031632967834207354,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.9725497961044312,
      "step": 376
    },
    {
      "epoch": 0.031801228301410585,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003,
      "loss": 1.9752633571624756,
      "step": 378
    },
    {
      "epoch": 0.031969488768613816,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0003,
      "loss": 1.9848291873931885,
      "step": 380
    },
    {
      "epoch": 0.03213774923581705,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 1.9936060905456543,
      "step": 382
    },
    {
      "epoch": 0.03230600970302028,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 2.0503222942352295,
      "step": 384
    },
    {
      "epoch": 0.03247427017022351,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.9907240867614746,
      "step": 386
    },
    {
      "epoch": 0.03264253063742674,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.9190280437469482,
      "step": 388
    },
    {
      "epoch": 0.03281079110462997,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003,
      "loss": 2.014657974243164,
      "step": 390
    },
    {
      "epoch": 0.0329790515718332,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9439013004302979,
      "step": 392
    },
    {
      "epoch": 0.033147312039036425,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9209346771240234,
      "step": 394
    },
    {
      "epoch": 0.033315572506239656,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.978372573852539,
      "step": 396
    },
    {
      "epoch": 0.03348383297344289,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 2.024970054626465,
      "step": 398
    },
    {
      "epoch": 0.03365209344064612,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9825183153152466,
      "step": 400
    },
    {
      "epoch": 0.03382035390784935,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.942143201828003,
      "step": 402
    },
    {
      "epoch": 0.03398861437505258,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 2.0022389888763428,
      "step": 404
    },
    {
      "epoch": 0.03415687484225581,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 2.0376572608947754,
      "step": 406
    },
    {
      "epoch": 0.03432513530945904,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 2.027191400527954,
      "step": 408
    },
    {
      "epoch": 0.03449339577666227,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.922903060913086,
      "step": 410
    },
    {
      "epoch": 0.0346616562438655,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.984115719795227,
      "step": 412
    },
    {
      "epoch": 0.034829916711068734,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9339840412139893,
      "step": 414
    },
    {
      "epoch": 0.034998177178271965,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.930540680885315,
      "step": 416
    },
    {
      "epoch": 0.035166437645475196,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0003,
      "loss": 1.995635986328125,
      "step": 418
    },
    {
      "epoch": 0.03533469811267843,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 2.0234413146972656,
      "step": 420
    },
    {
      "epoch": 0.03550295857988166,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0003,
      "loss": 1.9845763444900513,
      "step": 422
    },
    {
      "epoch": 0.03567121904708489,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0003,
      "loss": 2.017443895339966,
      "step": 424
    },
    {
      "epoch": 0.03583947951428812,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0003,
      "loss": 1.9358736276626587,
      "step": 426
    },
    {
      "epoch": 0.03600773998149135,
      "grad_norm": 0.46875,
      "learning_rate": 0.0003,
      "loss": 1.993273138999939,
      "step": 428
    },
    {
      "epoch": 0.03617600044869458,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0003,
      "loss": 2.001911163330078,
      "step": 430
    },
    {
      "epoch": 0.03634426091589781,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9373407363891602,
      "step": 432
    },
    {
      "epoch": 0.03651252138310104,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0003,
      "loss": 1.9039721488952637,
      "step": 434
    },
    {
      "epoch": 0.036680781850304274,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0003,
      "loss": 2.065171957015991,
      "step": 436
    },
    {
      "epoch": 0.036849042317507505,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0003,
      "loss": 2.0433545112609863,
      "step": 438
    },
    {
      "epoch": 0.03701730278471073,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0003,
      "loss": 2.0285892486572266,
      "step": 440
    },
    {
      "epoch": 0.03718556325191396,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9522912502288818,
      "step": 442
    },
    {
      "epoch": 0.03735382371911719,
      "grad_norm": 0.203125,
      "learning_rate": 0.0003,
      "loss": 1.972991704940796,
      "step": 444
    },
    {
      "epoch": 0.03752208418632042,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 2.0275721549987793,
      "step": 446
    },
    {
      "epoch": 0.03769034465352365,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9839547872543335,
      "step": 448
    },
    {
      "epoch": 0.03785860512072688,
      "grad_norm": 0.25,
      "learning_rate": 0.0003,
      "loss": 2.001451015472412,
      "step": 450
    },
    {
      "epoch": 0.038026865587930114,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0003,
      "loss": 2.01112699508667,
      "step": 452
    },
    {
      "epoch": 0.038195126055133345,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003,
      "loss": 1.9828572273254395,
      "step": 454
    },
    {
      "epoch": 0.038363386522336576,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9841010570526123,
      "step": 456
    },
    {
      "epoch": 0.03853164698953981,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0003,
      "loss": 1.9982948303222656,
      "step": 458
    },
    {
      "epoch": 0.03869990745674304,
      "grad_norm": 0.203125,
      "learning_rate": 0.0003,
      "loss": 1.9435060024261475,
      "step": 460
    },
    {
      "epoch": 0.03886816792394627,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0003,
      "loss": 2.0545706748962402,
      "step": 462
    },
    {
      "epoch": 0.0390364283911495,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.9742114543914795,
      "step": 464
    },
    {
      "epoch": 0.03920468885835273,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.9246653318405151,
      "step": 466
    },
    {
      "epoch": 0.03937294932555596,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9806394577026367,
      "step": 468
    },
    {
      "epoch": 0.03954120979275919,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9919495582580566,
      "step": 470
    },
    {
      "epoch": 0.03970947025996242,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9926968812942505,
      "step": 472
    },
    {
      "epoch": 0.039877730727165654,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 2.0499534606933594,
      "step": 474
    },
    {
      "epoch": 0.040045991194368885,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9395323991775513,
      "step": 476
    },
    {
      "epoch": 0.040214251661572116,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9915237426757812,
      "step": 478
    },
    {
      "epoch": 0.04038251212877535,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9922442436218262,
      "step": 480
    },
    {
      "epoch": 0.04055077259597858,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.981299638748169,
      "step": 482
    },
    {
      "epoch": 0.04071903306318181,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.9370598793029785,
      "step": 484
    },
    {
      "epoch": 0.04088729353038503,
      "grad_norm": 0.203125,
      "learning_rate": 0.0003,
      "loss": 1.962736964225769,
      "step": 486
    },
    {
      "epoch": 0.041055553997588264,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.992995023727417,
      "step": 488
    },
    {
      "epoch": 0.041223814464791494,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9571824073791504,
      "step": 490
    },
    {
      "epoch": 0.041392074931994725,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9102977514266968,
      "step": 492
    },
    {
      "epoch": 0.041560335399197956,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9919326305389404,
      "step": 494
    },
    {
      "epoch": 0.04172859586640119,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9433780908584595,
      "step": 496
    },
    {
      "epoch": 0.04189685633360442,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.9753245115280151,
      "step": 498
    },
    {
      "epoch": 0.04206511680080765,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 2.0142593383789062,
      "step": 500
    },
    {
      "epoch": 0.04223337726801088,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003,
      "loss": 1.9915118217468262,
      "step": 502
    },
    {
      "epoch": 0.04240163773521411,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0003,
      "loss": 1.968039631843567,
      "step": 504
    },
    {
      "epoch": 0.04256989820241734,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0003,
      "loss": 1.9798378944396973,
      "step": 506
    },
    {
      "epoch": 0.04273815866962057,
      "grad_norm": 0.296875,
      "learning_rate": 0.0003,
      "loss": 1.9489612579345703,
      "step": 508
    },
    {
      "epoch": 0.042906419136823803,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0003,
      "loss": 1.9632441997528076,
      "step": 510
    },
    {
      "epoch": 0.043074679604027034,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0003,
      "loss": 1.9661905765533447,
      "step": 512
    },
    {
      "epoch": 0.043242940071230265,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0003,
      "loss": 1.9851961135864258,
      "step": 514
    },
    {
      "epoch": 0.043411200538433496,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0003,
      "loss": 1.989220142364502,
      "step": 516
    },
    {
      "epoch": 0.04357946100563673,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0003,
      "loss": 1.958688735961914,
      "step": 518
    },
    {
      "epoch": 0.04374772147283996,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0003,
      "loss": 1.9549750089645386,
      "step": 520
    },
    {
      "epoch": 0.04391598194004319,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0003,
      "loss": 1.942992925643921,
      "step": 522
    },
    {
      "epoch": 0.04408424240724642,
      "grad_norm": 0.296875,
      "learning_rate": 0.0003,
      "loss": 1.9817299842834473,
      "step": 524
    },
    {
      "epoch": 0.04425250287444965,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0003,
      "loss": 2.0074384212493896,
      "step": 526
    },
    {
      "epoch": 0.04442076334165288,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003,
      "loss": 1.988731861114502,
      "step": 528
    },
    {
      "epoch": 0.04458902380885611,
      "grad_norm": 0.25,
      "learning_rate": 0.0003,
      "loss": 1.9524304866790771,
      "step": 530
    },
    {
      "epoch": 0.044757284276059336,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9519264698028564,
      "step": 532
    },
    {
      "epoch": 0.04492554474326257,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0003,
      "loss": 1.9676859378814697,
      "step": 534
    },
    {
      "epoch": 0.0450938052104658,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0003,
      "loss": 1.9795098304748535,
      "step": 536
    },
    {
      "epoch": 0.04526206567766903,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0003,
      "loss": 1.99741792678833,
      "step": 538
    },
    {
      "epoch": 0.04543032614487226,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0003,
      "loss": 1.9753245115280151,
      "step": 540
    },
    {
      "epoch": 0.04559858661207549,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9693810939788818,
      "step": 542
    },
    {
      "epoch": 0.04576684707927872,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0003,
      "loss": 1.9900774955749512,
      "step": 544
    },
    {
      "epoch": 0.04593510754648195,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0003,
      "loss": 1.9629302024841309,
      "step": 546
    },
    {
      "epoch": 0.046103368013685184,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.9653240442276,
      "step": 548
    },
    {
      "epoch": 0.046271628480888415,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9371516704559326,
      "step": 550
    },
    {
      "epoch": 0.046439888948091645,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.975390076637268,
      "step": 552
    },
    {
      "epoch": 0.046608149415294876,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9020578861236572,
      "step": 554
    },
    {
      "epoch": 0.04677640988249811,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 2.0180180072784424,
      "step": 556
    },
    {
      "epoch": 0.04694467034970134,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9519697427749634,
      "step": 558
    },
    {
      "epoch": 0.04711293081690457,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9566893577575684,
      "step": 560
    },
    {
      "epoch": 0.0472811912841078,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.991062879562378,
      "step": 562
    },
    {
      "epoch": 0.04744945175131103,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9476513862609863,
      "step": 564
    },
    {
      "epoch": 0.04761771221851426,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 2.012618064880371,
      "step": 566
    },
    {
      "epoch": 0.04778597268571749,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003,
      "loss": 1.9834308624267578,
      "step": 568
    },
    {
      "epoch": 0.047954233152920724,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003,
      "loss": 1.981619119644165,
      "step": 570
    },
    {
      "epoch": 0.048122493620123954,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 2.0213985443115234,
      "step": 572
    },
    {
      "epoch": 0.048290754087327185,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 2.021845817565918,
      "step": 574
    },
    {
      "epoch": 0.048459014554530416,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.979712963104248,
      "step": 576
    },
    {
      "epoch": 0.04862727502173364,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0003,
      "loss": 1.9884014129638672,
      "step": 578
    },
    {
      "epoch": 0.04879553548893687,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.957204818725586,
      "step": 580
    },
    {
      "epoch": 0.0489637959561401,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.9626176357269287,
      "step": 582
    },
    {
      "epoch": 0.04913205642334333,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0003,
      "loss": 1.9916534423828125,
      "step": 584
    },
    {
      "epoch": 0.049300316890546564,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0003,
      "loss": 1.942765235900879,
      "step": 586
    },
    {
      "epoch": 0.049468577357749795,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0003,
      "loss": 1.9592390060424805,
      "step": 588
    },
    {
      "epoch": 0.049636837824953026,
      "grad_norm": 0.203125,
      "learning_rate": 0.0003,
      "loss": 1.9957294464111328,
      "step": 590
    },
    {
      "epoch": 0.04980509829215626,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.9474029541015625,
      "step": 592
    },
    {
      "epoch": 0.04997335875935949,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0003,
      "loss": 1.9673320055007935,
      "step": 594
    },
    {
      "epoch": 0.05014161922656272,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0003,
      "loss": 2.009812355041504,
      "step": 596
    },
    {
      "epoch": 0.05030987969376595,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 2.0004100799560547,
      "step": 598
    },
    {
      "epoch": 0.05047814016096918,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9751455783843994,
      "step": 600
    },
    {
      "epoch": 0.05064640062817241,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.951069951057434,
      "step": 602
    },
    {
      "epoch": 0.05081466109537564,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.9662647247314453,
      "step": 604
    },
    {
      "epoch": 0.05098292156257887,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9945075511932373,
      "step": 606
    },
    {
      "epoch": 0.051151182029782104,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9382736682891846,
      "step": 608
    },
    {
      "epoch": 0.051319442496985335,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9552886486053467,
      "step": 610
    },
    {
      "epoch": 0.051487702964188566,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9338929653167725,
      "step": 612
    },
    {
      "epoch": 0.051655963431391796,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.937036156654358,
      "step": 614
    },
    {
      "epoch": 0.05182422389859503,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.981328010559082,
      "step": 616
    },
    {
      "epoch": 0.05199248436579826,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9123508930206299,
      "step": 618
    },
    {
      "epoch": 0.05216074483300149,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.96173894405365,
      "step": 620
    },
    {
      "epoch": 0.05232900530020472,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9740513563156128,
      "step": 622
    },
    {
      "epoch": 0.052497265767407944,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9665355682373047,
      "step": 624
    },
    {
      "epoch": 0.052665526234611175,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9571629762649536,
      "step": 626
    },
    {
      "epoch": 0.052833786701814406,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9562804698944092,
      "step": 628
    },
    {
      "epoch": 0.05300204716901764,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.950750708580017,
      "step": 630
    },
    {
      "epoch": 0.05317030763622087,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8895421028137207,
      "step": 632
    },
    {
      "epoch": 0.0533385681034241,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 2.0113649368286133,
      "step": 634
    },
    {
      "epoch": 0.05350682857062733,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9583513736724854,
      "step": 636
    },
    {
      "epoch": 0.05367508903783056,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003,
      "loss": 2.011899948120117,
      "step": 638
    },
    {
      "epoch": 0.05384334950503379,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.9226679801940918,
      "step": 640
    },
    {
      "epoch": 0.05401160997223702,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9380148649215698,
      "step": 642
    },
    {
      "epoch": 0.05417987043944025,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9530277252197266,
      "step": 644
    },
    {
      "epoch": 0.054348130906643484,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9722737073898315,
      "step": 646
    },
    {
      "epoch": 0.054516391373846715,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9479560852050781,
      "step": 648
    },
    {
      "epoch": 0.054684651841049946,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9458119869232178,
      "step": 650
    },
    {
      "epoch": 0.05485291230825318,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.9580166339874268,
      "step": 652
    },
    {
      "epoch": 0.05502117277545641,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.9468915462493896,
      "step": 654
    },
    {
      "epoch": 0.05518943324265964,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9379453659057617,
      "step": 656
    },
    {
      "epoch": 0.05535769370986287,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8600000143051147,
      "step": 658
    },
    {
      "epoch": 0.0555259541770661,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9992973804473877,
      "step": 660
    },
    {
      "epoch": 0.05569421464426933,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8817965984344482,
      "step": 662
    },
    {
      "epoch": 0.05586247511147256,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9229753017425537,
      "step": 664
    },
    {
      "epoch": 0.05603073557867579,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9363641738891602,
      "step": 666
    },
    {
      "epoch": 0.056198996045879024,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9280869960784912,
      "step": 668
    },
    {
      "epoch": 0.056367256513082255,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 2.0148813724517822,
      "step": 670
    },
    {
      "epoch": 0.05653551698028548,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 2.00032114982605,
      "step": 672
    },
    {
      "epoch": 0.05670377744748871,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9379140138626099,
      "step": 674
    },
    {
      "epoch": 0.05687203791469194,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 2.0075132846832275,
      "step": 676
    },
    {
      "epoch": 0.05704029838189517,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003,
      "loss": 1.9237275123596191,
      "step": 678
    },
    {
      "epoch": 0.0572085588490984,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.9404950141906738,
      "step": 680
    },
    {
      "epoch": 0.05737681931630163,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9969398975372314,
      "step": 682
    },
    {
      "epoch": 0.057545079783504864,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9430654048919678,
      "step": 684
    },
    {
      "epoch": 0.057713340250708095,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.965470790863037,
      "step": 686
    },
    {
      "epoch": 0.057881600717911326,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.9066267013549805,
      "step": 688
    },
    {
      "epoch": 0.05804986118511456,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8755056858062744,
      "step": 690
    },
    {
      "epoch": 0.05821812165231779,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 2.011030435562134,
      "step": 692
    },
    {
      "epoch": 0.05838638211952102,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9623510837554932,
      "step": 694
    },
    {
      "epoch": 0.05855464258672425,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9944987297058105,
      "step": 696
    },
    {
      "epoch": 0.05872290305392748,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9605417251586914,
      "step": 698
    },
    {
      "epoch": 0.05889116352113071,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0003,
      "loss": 1.9891598224639893,
      "step": 700
    },
    {
      "epoch": 0.05905942398833394,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0003,
      "loss": 1.9630203247070312,
      "step": 702
    },
    {
      "epoch": 0.05922768445553717,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0003,
      "loss": 1.9561936855316162,
      "step": 704
    },
    {
      "epoch": 0.059395944922740404,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9263983964920044,
      "step": 706
    },
    {
      "epoch": 0.059564205389943635,
      "grad_norm": 0.328125,
      "learning_rate": 0.0003,
      "loss": 1.9304323196411133,
      "step": 708
    },
    {
      "epoch": 0.059732465857146866,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.9655709266662598,
      "step": 710
    },
    {
      "epoch": 0.0599007263243501,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9245038032531738,
      "step": 712
    },
    {
      "epoch": 0.06006898679155333,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.920980453491211,
      "step": 714
    },
    {
      "epoch": 0.06023724725875656,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9185936450958252,
      "step": 716
    },
    {
      "epoch": 0.06040550772595978,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9415130615234375,
      "step": 718
    },
    {
      "epoch": 0.06057376819316301,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9352434873580933,
      "step": 720
    },
    {
      "epoch": 0.060742028660366244,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 2.0316762924194336,
      "step": 722
    },
    {
      "epoch": 0.060910289127569475,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8884985446929932,
      "step": 724
    },
    {
      "epoch": 0.061078549594772706,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9162423610687256,
      "step": 726
    },
    {
      "epoch": 0.06124681006197594,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9901643991470337,
      "step": 728
    },
    {
      "epoch": 0.06141507052917917,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9207351207733154,
      "step": 730
    },
    {
      "epoch": 0.0615833309963824,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9822008609771729,
      "step": 732
    },
    {
      "epoch": 0.06175159146358563,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9567785263061523,
      "step": 734
    },
    {
      "epoch": 0.06191985193078886,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9941750764846802,
      "step": 736
    },
    {
      "epoch": 0.06208811239799209,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9577667713165283,
      "step": 738
    },
    {
      "epoch": 0.06225637286519532,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9010050296783447,
      "step": 740
    },
    {
      "epoch": 0.06242463333239855,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9712854623794556,
      "step": 742
    },
    {
      "epoch": 0.06259289379960178,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9276484251022339,
      "step": 744
    },
    {
      "epoch": 0.06276115426680501,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 2.0309596061706543,
      "step": 746
    },
    {
      "epoch": 0.06292941473400825,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9334442615509033,
      "step": 748
    },
    {
      "epoch": 0.06309767520121147,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9913946390151978,
      "step": 750
    },
    {
      "epoch": 0.06326593566841471,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9219679832458496,
      "step": 752
    },
    {
      "epoch": 0.06343419613561793,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.986389398574829,
      "step": 754
    },
    {
      "epoch": 0.06360245660282117,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.9281837940216064,
      "step": 756
    },
    {
      "epoch": 0.0637707170700244,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.965665340423584,
      "step": 758
    },
    {
      "epoch": 0.06393897753722763,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9223332405090332,
      "step": 760
    },
    {
      "epoch": 0.06410723800443086,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.959839105606079,
      "step": 762
    },
    {
      "epoch": 0.0642754984716341,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9622299671173096,
      "step": 764
    },
    {
      "epoch": 0.06444375893883732,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9364138841629028,
      "step": 766
    },
    {
      "epoch": 0.06461201940604056,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.9019924402236938,
      "step": 768
    },
    {
      "epoch": 0.06478027987324378,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9938486814498901,
      "step": 770
    },
    {
      "epoch": 0.06494854034044702,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8975040912628174,
      "step": 772
    },
    {
      "epoch": 0.06511680080765024,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9074137210845947,
      "step": 774
    },
    {
      "epoch": 0.06528506127485348,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.9663798809051514,
      "step": 776
    },
    {
      "epoch": 0.0654533217420567,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.9786689281463623,
      "step": 778
    },
    {
      "epoch": 0.06562158220925994,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.912980318069458,
      "step": 780
    },
    {
      "epoch": 0.06578984267646316,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9791762828826904,
      "step": 782
    },
    {
      "epoch": 0.0659581031436664,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.913940191268921,
      "step": 784
    },
    {
      "epoch": 0.06612636361086963,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0003,
      "loss": 1.9037126302719116,
      "step": 786
    },
    {
      "epoch": 0.06629462407807285,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.91959810256958,
      "step": 788
    },
    {
      "epoch": 0.06646288454527609,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.9602323770523071,
      "step": 790
    },
    {
      "epoch": 0.06663114501247931,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0003,
      "loss": 1.9592417478561401,
      "step": 792
    },
    {
      "epoch": 0.06679940547968255,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.950226068496704,
      "step": 794
    },
    {
      "epoch": 0.06696766594688577,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0003,
      "loss": 1.9913499355316162,
      "step": 796
    },
    {
      "epoch": 0.06713592641408901,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0003,
      "loss": 1.931213140487671,
      "step": 798
    },
    {
      "epoch": 0.06730418688129224,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0003,
      "loss": 1.962982416152954,
      "step": 800
    },
    {
      "epoch": 0.06747244734849547,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0003,
      "loss": 1.9713337421417236,
      "step": 802
    },
    {
      "epoch": 0.0676407078156987,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0003,
      "loss": 1.9458580017089844,
      "step": 804
    },
    {
      "epoch": 0.06780896828290194,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0003,
      "loss": 1.9213030338287354,
      "step": 806
    },
    {
      "epoch": 0.06797722875010516,
      "grad_norm": 0.46875,
      "learning_rate": 0.0003,
      "loss": 1.9649664163589478,
      "step": 808
    },
    {
      "epoch": 0.0681454892173084,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0003,
      "loss": 1.9210338592529297,
      "step": 810
    },
    {
      "epoch": 0.06831374968451162,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0003,
      "loss": 1.9470151662826538,
      "step": 812
    },
    {
      "epoch": 0.06848201015171486,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003,
      "loss": 1.9581470489501953,
      "step": 814
    },
    {
      "epoch": 0.06865027061891808,
      "grad_norm": 0.53125,
      "learning_rate": 0.0003,
      "loss": 1.9838721752166748,
      "step": 816
    },
    {
      "epoch": 0.06881853108612132,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.951630711555481,
      "step": 818
    },
    {
      "epoch": 0.06898679155332454,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0003,
      "loss": 1.9434454441070557,
      "step": 820
    },
    {
      "epoch": 0.06915505202052778,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9458017349243164,
      "step": 822
    },
    {
      "epoch": 0.069323312487731,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.9569238424301147,
      "step": 824
    },
    {
      "epoch": 0.06949157295493424,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9502720832824707,
      "step": 826
    },
    {
      "epoch": 0.06965983342213747,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.9144351482391357,
      "step": 828
    },
    {
      "epoch": 0.0698280938893407,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9170506000518799,
      "step": 830
    },
    {
      "epoch": 0.06999635435654393,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 1.9477040767669678,
      "step": 832
    },
    {
      "epoch": 0.07016461482374715,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.999624490737915,
      "step": 834
    },
    {
      "epoch": 0.07033287529095039,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.9658105373382568,
      "step": 836
    },
    {
      "epoch": 0.07050113575815362,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9730370044708252,
      "step": 838
    },
    {
      "epoch": 0.07066939622535685,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.950524926185608,
      "step": 840
    },
    {
      "epoch": 0.07083765669256008,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8793505430221558,
      "step": 842
    },
    {
      "epoch": 0.07100591715976332,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.906143069267273,
      "step": 844
    },
    {
      "epoch": 0.07117417762696654,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9077270030975342,
      "step": 846
    },
    {
      "epoch": 0.07134243809416978,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9340252876281738,
      "step": 848
    },
    {
      "epoch": 0.071510698561373,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8510985374450684,
      "step": 850
    },
    {
      "epoch": 0.07167895902857624,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9735691547393799,
      "step": 852
    },
    {
      "epoch": 0.07184721949577946,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9639033079147339,
      "step": 854
    },
    {
      "epoch": 0.0720154799629827,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.931457281112671,
      "step": 856
    },
    {
      "epoch": 0.07218374043018592,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9449543952941895,
      "step": 858
    },
    {
      "epoch": 0.07235200089738916,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.973243236541748,
      "step": 860
    },
    {
      "epoch": 0.07252026136459239,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.9677700996398926,
      "step": 862
    },
    {
      "epoch": 0.07268852183179562,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.996877908706665,
      "step": 864
    },
    {
      "epoch": 0.07285678229899885,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.87092125415802,
      "step": 866
    },
    {
      "epoch": 0.07302504276620209,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.93142831325531,
      "step": 868
    },
    {
      "epoch": 0.07319330323340531,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9785701036453247,
      "step": 870
    },
    {
      "epoch": 0.07336156370060855,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9680613279342651,
      "step": 872
    },
    {
      "epoch": 0.07352982416781177,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.874957799911499,
      "step": 874
    },
    {
      "epoch": 0.07369808463501501,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8946905136108398,
      "step": 876
    },
    {
      "epoch": 0.07386634510221823,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9605982303619385,
      "step": 878
    },
    {
      "epoch": 0.07403460556942146,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9753882884979248,
      "step": 880
    },
    {
      "epoch": 0.0742028660366247,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9043469429016113,
      "step": 882
    },
    {
      "epoch": 0.07437112650382792,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8884528875350952,
      "step": 884
    },
    {
      "epoch": 0.07453938697103116,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9437971115112305,
      "step": 886
    },
    {
      "epoch": 0.07470764743823438,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8856186866760254,
      "step": 888
    },
    {
      "epoch": 0.07487590790543762,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8841743469238281,
      "step": 890
    },
    {
      "epoch": 0.07504416837264084,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.952164649963379,
      "step": 892
    },
    {
      "epoch": 0.07521242883984408,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9038926362991333,
      "step": 894
    },
    {
      "epoch": 0.0753806893070473,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0003,
      "loss": 1.8948378562927246,
      "step": 896
    },
    {
      "epoch": 0.07554894977425054,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0003,
      "loss": 1.9292104244232178,
      "step": 898
    },
    {
      "epoch": 0.07571721024145377,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9576606750488281,
      "step": 900
    },
    {
      "epoch": 0.075885470708657,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.9422367811203003,
      "step": 902
    },
    {
      "epoch": 0.07605373117586023,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.912757158279419,
      "step": 904
    },
    {
      "epoch": 0.07622199164306347,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.962846040725708,
      "step": 906
    },
    {
      "epoch": 0.07639025211026669,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003,
      "loss": 1.8968214988708496,
      "step": 908
    },
    {
      "epoch": 0.07655851257746993,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.947690725326538,
      "step": 910
    },
    {
      "epoch": 0.07672677304467315,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9889655113220215,
      "step": 912
    },
    {
      "epoch": 0.07689503351187639,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.927793025970459,
      "step": 914
    },
    {
      "epoch": 0.07706329397907961,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9505354166030884,
      "step": 916
    },
    {
      "epoch": 0.07723155444628285,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8930877447128296,
      "step": 918
    },
    {
      "epoch": 0.07739981491348608,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.9238629341125488,
      "step": 920
    },
    {
      "epoch": 0.07756807538068931,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8969670534133911,
      "step": 922
    },
    {
      "epoch": 0.07773633584789254,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.9216035604476929,
      "step": 924
    },
    {
      "epoch": 0.07790459631509576,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9372799396514893,
      "step": 926
    },
    {
      "epoch": 0.078072856782299,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9650444984436035,
      "step": 928
    },
    {
      "epoch": 0.07824111724950222,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9307043552398682,
      "step": 930
    },
    {
      "epoch": 0.07840937771670546,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9133226871490479,
      "step": 932
    },
    {
      "epoch": 0.07857763818390869,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9613797664642334,
      "step": 934
    },
    {
      "epoch": 0.07874589865111192,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9733467102050781,
      "step": 936
    },
    {
      "epoch": 0.07891415911831515,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.968252420425415,
      "step": 938
    },
    {
      "epoch": 0.07908241958551838,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9077494144439697,
      "step": 940
    },
    {
      "epoch": 0.07925068005272161,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9294328689575195,
      "step": 942
    },
    {
      "epoch": 0.07941894051992485,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.9335801601409912,
      "step": 944
    },
    {
      "epoch": 0.07958720098712807,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003,
      "loss": 1.9594451189041138,
      "step": 946
    },
    {
      "epoch": 0.07975546145433131,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.861429214477539,
      "step": 948
    },
    {
      "epoch": 0.07992372192153453,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9489378929138184,
      "step": 950
    },
    {
      "epoch": 0.08009198238873777,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8906891345977783,
      "step": 952
    },
    {
      "epoch": 0.080260242855941,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.921875,
      "step": 954
    },
    {
      "epoch": 0.08042850332314423,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.940981149673462,
      "step": 956
    },
    {
      "epoch": 0.08059676379034746,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8859500885009766,
      "step": 958
    },
    {
      "epoch": 0.0807650242575507,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9174455404281616,
      "step": 960
    },
    {
      "epoch": 0.08093328472475392,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 1.9604482650756836,
      "step": 962
    },
    {
      "epoch": 0.08110154519195716,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.914562702178955,
      "step": 964
    },
    {
      "epoch": 0.08126980565916038,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0003,
      "loss": 1.9399354457855225,
      "step": 966
    },
    {
      "epoch": 0.08143806612636362,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9440795183181763,
      "step": 968
    },
    {
      "epoch": 0.08160632659356684,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.9647586345672607,
      "step": 970
    },
    {
      "epoch": 0.08177458706077007,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8957921266555786,
      "step": 972
    },
    {
      "epoch": 0.0819428475279733,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8923792839050293,
      "step": 974
    },
    {
      "epoch": 0.08211110799517653,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9192314147949219,
      "step": 976
    },
    {
      "epoch": 0.08227936846237976,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9343748092651367,
      "step": 978
    },
    {
      "epoch": 0.08244762892958299,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9516870975494385,
      "step": 980
    },
    {
      "epoch": 0.08261588939678623,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8918962478637695,
      "step": 982
    },
    {
      "epoch": 0.08278414986398945,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9630496501922607,
      "step": 984
    },
    {
      "epoch": 0.08295241033119269,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8496665954589844,
      "step": 986
    },
    {
      "epoch": 0.08312067079839591,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8639805316925049,
      "step": 988
    },
    {
      "epoch": 0.08328893126559915,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8814165592193604,
      "step": 990
    },
    {
      "epoch": 0.08345719173280237,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.8797595500946045,
      "step": 992
    },
    {
      "epoch": 0.08362545220000561,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.9375497102737427,
      "step": 994
    },
    {
      "epoch": 0.08379371266720884,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.8633697032928467,
      "step": 996
    },
    {
      "epoch": 0.08396197313441207,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.8858206272125244,
      "step": 998
    },
    {
      "epoch": 0.0841302336016153,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0003,
      "loss": 1.9618487358093262,
      "step": 1000
    },
    {
      "epoch": 0.08429849406881854,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.9462389945983887,
      "step": 1002
    },
    {
      "epoch": 0.08446675453602176,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8670237064361572,
      "step": 1004
    },
    {
      "epoch": 0.084635015003225,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003,
      "loss": 1.878306269645691,
      "step": 1006
    },
    {
      "epoch": 0.08480327547042822,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.9075534343719482,
      "step": 1008
    },
    {
      "epoch": 0.08497153593763146,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9642163515090942,
      "step": 1010
    },
    {
      "epoch": 0.08513979640483468,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9249271154403687,
      "step": 1012
    },
    {
      "epoch": 0.08530805687203792,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 2.00308895111084,
      "step": 1014
    },
    {
      "epoch": 0.08547631733924115,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0003,
      "loss": 1.8873727321624756,
      "step": 1016
    },
    {
      "epoch": 0.08564457780644437,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.946439266204834,
      "step": 1018
    },
    {
      "epoch": 0.08581283827364761,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9052157402038574,
      "step": 1020
    },
    {
      "epoch": 0.08598109874085083,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.9466506242752075,
      "step": 1022
    },
    {
      "epoch": 0.08614935920805407,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.9530543088912964,
      "step": 1024
    },
    {
      "epoch": 0.08631761967525729,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9564145803451538,
      "step": 1026
    },
    {
      "epoch": 0.08648588014246053,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9122086763381958,
      "step": 1028
    },
    {
      "epoch": 0.08665414060966375,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9366737604141235,
      "step": 1030
    },
    {
      "epoch": 0.08682240107686699,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 1.9558972120285034,
      "step": 1032
    },
    {
      "epoch": 0.08699066154407022,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.8891115188598633,
      "step": 1034
    },
    {
      "epoch": 0.08715892201127345,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0003,
      "loss": 1.945936918258667,
      "step": 1036
    },
    {
      "epoch": 0.08732718247847668,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.8997255563735962,
      "step": 1038
    },
    {
      "epoch": 0.08749544294567992,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9343082904815674,
      "step": 1040
    },
    {
      "epoch": 0.08766370341288314,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9334485530853271,
      "step": 1042
    },
    {
      "epoch": 0.08783196388008638,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9299594163894653,
      "step": 1044
    },
    {
      "epoch": 0.0880002243472896,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9151664972305298,
      "step": 1046
    },
    {
      "epoch": 0.08816848481449284,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9414360523223877,
      "step": 1048
    },
    {
      "epoch": 0.08833674528169606,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8991100788116455,
      "step": 1050
    },
    {
      "epoch": 0.0885050057488993,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9260191917419434,
      "step": 1052
    },
    {
      "epoch": 0.08867326621610253,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.941760540008545,
      "step": 1054
    },
    {
      "epoch": 0.08884152668330576,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.950076699256897,
      "step": 1056
    },
    {
      "epoch": 0.08900978715050899,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8965582847595215,
      "step": 1058
    },
    {
      "epoch": 0.08917804761771222,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.8692132234573364,
      "step": 1060
    },
    {
      "epoch": 0.08934630808491545,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8913989067077637,
      "step": 1062
    },
    {
      "epoch": 0.08951456855211867,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9316949844360352,
      "step": 1064
    },
    {
      "epoch": 0.08968282901932191,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9793076515197754,
      "step": 1066
    },
    {
      "epoch": 0.08985108948652513,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8617980480194092,
      "step": 1068
    },
    {
      "epoch": 0.09001934995372837,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9390721321105957,
      "step": 1070
    },
    {
      "epoch": 0.0901876104209316,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9011821746826172,
      "step": 1072
    },
    {
      "epoch": 0.09035587088813483,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.895948052406311,
      "step": 1074
    },
    {
      "epoch": 0.09052413135533806,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9059522151947021,
      "step": 1076
    },
    {
      "epoch": 0.0906923918225413,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 2.015043258666992,
      "step": 1078
    },
    {
      "epoch": 0.09086065228974452,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.854677438735962,
      "step": 1080
    },
    {
      "epoch": 0.09102891275694776,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9428353309631348,
      "step": 1082
    },
    {
      "epoch": 0.09119717322415098,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9167896509170532,
      "step": 1084
    },
    {
      "epoch": 0.09136543369135422,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8753914833068848,
      "step": 1086
    },
    {
      "epoch": 0.09153369415855744,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.967937707901001,
      "step": 1088
    },
    {
      "epoch": 0.09170195462576068,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9067573547363281,
      "step": 1090
    },
    {
      "epoch": 0.0918702150929639,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9694361686706543,
      "step": 1092
    },
    {
      "epoch": 0.09203847556016714,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 1.972168207168579,
      "step": 1094
    },
    {
      "epoch": 0.09220673602737037,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003,
      "loss": 1.9397616386413574,
      "step": 1096
    },
    {
      "epoch": 0.0923749964945736,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0003,
      "loss": 1.903311014175415,
      "step": 1098
    },
    {
      "epoch": 0.09254325696177683,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.9424320459365845,
      "step": 1100
    },
    {
      "epoch": 0.09271151742898007,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9118008613586426,
      "step": 1102
    },
    {
      "epoch": 0.09287977789618329,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.916631817817688,
      "step": 1104
    },
    {
      "epoch": 0.09304803836338653,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9163919687271118,
      "step": 1106
    },
    {
      "epoch": 0.09321629883058975,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9482258558273315,
      "step": 1108
    },
    {
      "epoch": 0.09338455929779298,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9391429424285889,
      "step": 1110
    },
    {
      "epoch": 0.09355281976499621,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8863152265548706,
      "step": 1112
    },
    {
      "epoch": 0.09372108023219944,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9268739223480225,
      "step": 1114
    },
    {
      "epoch": 0.09388934069940268,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.923417568206787,
      "step": 1116
    },
    {
      "epoch": 0.0940576011666059,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.961132287979126,
      "step": 1118
    },
    {
      "epoch": 0.09422586163380914,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9666460752487183,
      "step": 1120
    },
    {
      "epoch": 0.09439412210101236,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9208102226257324,
      "step": 1122
    },
    {
      "epoch": 0.0945623825682156,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9534294605255127,
      "step": 1124
    },
    {
      "epoch": 0.09473064303541882,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.9365112781524658,
      "step": 1126
    },
    {
      "epoch": 0.09489890350262206,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 2.0162062644958496,
      "step": 1128
    },
    {
      "epoch": 0.09506716396982529,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.9455296993255615,
      "step": 1130
    },
    {
      "epoch": 0.09523542443702852,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9499146938323975,
      "step": 1132
    },
    {
      "epoch": 0.09540368490423175,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.9180772304534912,
      "step": 1134
    },
    {
      "epoch": 0.09557194537143499,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8387937545776367,
      "step": 1136
    },
    {
      "epoch": 0.09574020583863821,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9204694032669067,
      "step": 1138
    },
    {
      "epoch": 0.09590846630584145,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8883326053619385,
      "step": 1140
    },
    {
      "epoch": 0.09607672677304467,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9481935501098633,
      "step": 1142
    },
    {
      "epoch": 0.09624498724024791,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9488816261291504,
      "step": 1144
    },
    {
      "epoch": 0.09641324770745113,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9549610614776611,
      "step": 1146
    },
    {
      "epoch": 0.09658150817465437,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8948073387145996,
      "step": 1148
    },
    {
      "epoch": 0.0967497686418576,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.9329922199249268,
      "step": 1150
    },
    {
      "epoch": 0.09691802910906083,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003,
      "loss": 1.9619866609573364,
      "step": 1152
    },
    {
      "epoch": 0.09708628957626406,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.914672613143921,
      "step": 1154
    },
    {
      "epoch": 0.09725455004346728,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 1.9683159589767456,
      "step": 1156
    },
    {
      "epoch": 0.09742281051067052,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 2.020977020263672,
      "step": 1158
    },
    {
      "epoch": 0.09759107097787374,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8901516199111938,
      "step": 1160
    },
    {
      "epoch": 0.09775933144507698,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9614262580871582,
      "step": 1162
    },
    {
      "epoch": 0.0979275919122802,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9173681735992432,
      "step": 1164
    },
    {
      "epoch": 0.09809585237948344,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9562668800354004,
      "step": 1166
    },
    {
      "epoch": 0.09826411284668667,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.940651297569275,
      "step": 1168
    },
    {
      "epoch": 0.0984323733138899,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.930161952972412,
      "step": 1170
    },
    {
      "epoch": 0.09860063378109313,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9337252378463745,
      "step": 1172
    },
    {
      "epoch": 0.09876889424829637,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8356858491897583,
      "step": 1174
    },
    {
      "epoch": 0.09893715471549959,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.970253348350525,
      "step": 1176
    },
    {
      "epoch": 0.09910541518270283,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003,
      "loss": 1.9399564266204834,
      "step": 1178
    },
    {
      "epoch": 0.09927367564990605,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9923701286315918,
      "step": 1180
    },
    {
      "epoch": 0.09944193611710929,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.8998527526855469,
      "step": 1182
    },
    {
      "epoch": 0.09961019658431251,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9313559532165527,
      "step": 1184
    },
    {
      "epoch": 0.09977845705151575,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9721612930297852,
      "step": 1186
    },
    {
      "epoch": 0.09994671751871897,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 2.0018858909606934,
      "step": 1188
    },
    {
      "epoch": 0.10011497798592221,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.9653280973434448,
      "step": 1190
    },
    {
      "epoch": 0.10028323845312544,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9805734157562256,
      "step": 1192
    },
    {
      "epoch": 0.10045149892032867,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.949955701828003,
      "step": 1194
    },
    {
      "epoch": 0.1006197593875319,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.967233657836914,
      "step": 1196
    },
    {
      "epoch": 0.10078801985473514,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.8856146335601807,
      "step": 1198
    },
    {
      "epoch": 0.10095628032193836,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.900649070739746,
      "step": 1200
    },
    {
      "epoch": 0.10112454078914158,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.886762022972107,
      "step": 1202
    },
    {
      "epoch": 0.10129280125634482,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8775185346603394,
      "step": 1204
    },
    {
      "epoch": 0.10146106172354805,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.958686113357544,
      "step": 1206
    },
    {
      "epoch": 0.10162932219075128,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.9657886028289795,
      "step": 1208
    },
    {
      "epoch": 0.10179758265795451,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0003,
      "loss": 1.9787405729293823,
      "step": 1210
    },
    {
      "epoch": 0.10196584312515775,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.9246686697006226,
      "step": 1212
    },
    {
      "epoch": 0.10213410359236097,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.938382863998413,
      "step": 1214
    },
    {
      "epoch": 0.10230236405956421,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9479703903198242,
      "step": 1216
    },
    {
      "epoch": 0.10247062452676743,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9330463409423828,
      "step": 1218
    },
    {
      "epoch": 0.10263888499397067,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0003,
      "loss": 1.9330073595046997,
      "step": 1220
    },
    {
      "epoch": 0.1028071454611739,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0003,
      "loss": 1.888183355331421,
      "step": 1222
    },
    {
      "epoch": 0.10297540592837713,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0003,
      "loss": 1.8636600971221924,
      "step": 1224
    },
    {
      "epoch": 0.10314366639558036,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0003,
      "loss": 1.9760087728500366,
      "step": 1226
    },
    {
      "epoch": 0.10331192686278359,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0003,
      "loss": 1.9119813442230225,
      "step": 1228
    },
    {
      "epoch": 0.10348018732998682,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0003,
      "loss": 1.936758279800415,
      "step": 1230
    },
    {
      "epoch": 0.10364844779719005,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0003,
      "loss": 1.940406322479248,
      "step": 1232
    },
    {
      "epoch": 0.10381670826439328,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.908825159072876,
      "step": 1234
    },
    {
      "epoch": 0.10398496873159652,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.922850251197815,
      "step": 1236
    },
    {
      "epoch": 0.10415322919879974,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003,
      "loss": 1.9346346855163574,
      "step": 1238
    },
    {
      "epoch": 0.10432148966600298,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8852641582489014,
      "step": 1240
    },
    {
      "epoch": 0.1044897501332062,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8675076961517334,
      "step": 1242
    },
    {
      "epoch": 0.10465801060040944,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003,
      "loss": 1.9002351760864258,
      "step": 1244
    },
    {
      "epoch": 0.10482627106761266,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9177359342575073,
      "step": 1246
    },
    {
      "epoch": 0.10499453153481589,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9179494380950928,
      "step": 1248
    },
    {
      "epoch": 0.10516279200201913,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.9197450876235962,
      "step": 1250
    },
    {
      "epoch": 0.10533105246922235,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9222612380981445,
      "step": 1252
    },
    {
      "epoch": 0.10549931293642559,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.955383539199829,
      "step": 1254
    },
    {
      "epoch": 0.10566757340362881,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.956671118736267,
      "step": 1256
    },
    {
      "epoch": 0.10583583387083205,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9368703365325928,
      "step": 1258
    },
    {
      "epoch": 0.10600409433803527,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8774750232696533,
      "step": 1260
    },
    {
      "epoch": 0.10617235480523851,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.873946189880371,
      "step": 1262
    },
    {
      "epoch": 0.10634061527244174,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.890907883644104,
      "step": 1264
    },
    {
      "epoch": 0.10650887573964497,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9571466445922852,
      "step": 1266
    },
    {
      "epoch": 0.1066771362068482,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0003,
      "loss": 2.008236885070801,
      "step": 1268
    },
    {
      "epoch": 0.10684539667405143,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0003,
      "loss": 1.919787883758545,
      "step": 1270
    },
    {
      "epoch": 0.10701365714125466,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0003,
      "loss": 1.9549667835235596,
      "step": 1272
    },
    {
      "epoch": 0.1071819176084579,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0003,
      "loss": 1.9301097393035889,
      "step": 1274
    },
    {
      "epoch": 0.10735017807566112,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0003,
      "loss": 1.889240026473999,
      "step": 1276
    },
    {
      "epoch": 0.10751843854286436,
      "grad_norm": 7.0625,
      "learning_rate": 0.0003,
      "loss": 2.015836477279663,
      "step": 1278
    },
    {
      "epoch": 0.10768669901006758,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0003,
      "loss": 1.935014009475708,
      "step": 1280
    },
    {
      "epoch": 0.10785495947727082,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0003,
      "loss": 1.926422119140625,
      "step": 1282
    },
    {
      "epoch": 0.10802321994447404,
      "grad_norm": 0.421875,
      "learning_rate": 0.0003,
      "loss": 2.055593490600586,
      "step": 1284
    },
    {
      "epoch": 0.10819148041167728,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0003,
      "loss": 1.9950774908065796,
      "step": 1286
    },
    {
      "epoch": 0.1083597408788805,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0003,
      "loss": 1.9511891603469849,
      "step": 1288
    },
    {
      "epoch": 0.10852800134608374,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0003,
      "loss": 1.9354677200317383,
      "step": 1290
    },
    {
      "epoch": 0.10869626181328697,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0003,
      "loss": 1.9176979064941406,
      "step": 1292
    },
    {
      "epoch": 0.1088645222804902,
      "grad_norm": 0.203125,
      "learning_rate": 0.0003,
      "loss": 1.9347909688949585,
      "step": 1294
    },
    {
      "epoch": 0.10903278274769343,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.93163001537323,
      "step": 1296
    },
    {
      "epoch": 0.10920104321489665,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.9148836135864258,
      "step": 1298
    },
    {
      "epoch": 0.10936930368209989,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 1.9672925472259521,
      "step": 1300
    },
    {
      "epoch": 0.10953756414930312,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9662559032440186,
      "step": 1302
    },
    {
      "epoch": 0.10970582461650635,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8804277181625366,
      "step": 1304
    },
    {
      "epoch": 0.10987408508370958,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9431190490722656,
      "step": 1306
    },
    {
      "epoch": 0.11004234555091282,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.91512131690979,
      "step": 1308
    },
    {
      "epoch": 0.11021060601811604,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9714438915252686,
      "step": 1310
    },
    {
      "epoch": 0.11037886648531928,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9106714725494385,
      "step": 1312
    },
    {
      "epoch": 0.1105471269525225,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8971325159072876,
      "step": 1314
    },
    {
      "epoch": 0.11071538741972574,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9331936836242676,
      "step": 1316
    },
    {
      "epoch": 0.11088364788692896,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9694864749908447,
      "step": 1318
    },
    {
      "epoch": 0.1110519083541322,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.9109864234924316,
      "step": 1320
    },
    {
      "epoch": 0.11122016882133542,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9355192184448242,
      "step": 1322
    },
    {
      "epoch": 0.11138842928853866,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.855737328529358,
      "step": 1324
    },
    {
      "epoch": 0.11155668975574189,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9472017288208008,
      "step": 1326
    },
    {
      "epoch": 0.11172495022294512,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9380812644958496,
      "step": 1328
    },
    {
      "epoch": 0.11189321069014835,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.928450584411621,
      "step": 1330
    },
    {
      "epoch": 0.11206147115735159,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8914761543273926,
      "step": 1332
    },
    {
      "epoch": 0.11222973162455481,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8948400020599365,
      "step": 1334
    },
    {
      "epoch": 0.11239799209175805,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9259252548217773,
      "step": 1336
    },
    {
      "epoch": 0.11256625255896127,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8807357549667358,
      "step": 1338
    },
    {
      "epoch": 0.11273451302616451,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 1.8884329795837402,
      "step": 1340
    },
    {
      "epoch": 0.11290277349336773,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9450271129608154,
      "step": 1342
    },
    {
      "epoch": 0.11307103396057096,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 2.007864475250244,
      "step": 1344
    },
    {
      "epoch": 0.1132392944277742,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.9787633419036865,
      "step": 1346
    },
    {
      "epoch": 0.11340755489497742,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.91082763671875,
      "step": 1348
    },
    {
      "epoch": 0.11357581536218066,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0003,
      "loss": 1.9385350942611694,
      "step": 1350
    },
    {
      "epoch": 0.11374407582938388,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003,
      "loss": 1.917553424835205,
      "step": 1352
    },
    {
      "epoch": 0.11391233629658712,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.90462327003479,
      "step": 1354
    },
    {
      "epoch": 0.11408059676379034,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9053934812545776,
      "step": 1356
    },
    {
      "epoch": 0.11424885723099358,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9211571216583252,
      "step": 1358
    },
    {
      "epoch": 0.1144171176981968,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.929023027420044,
      "step": 1360
    },
    {
      "epoch": 0.11458537816540004,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8640564680099487,
      "step": 1362
    },
    {
      "epoch": 0.11475363863260327,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8834364414215088,
      "step": 1364
    },
    {
      "epoch": 0.1149218990998065,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.9007751941680908,
      "step": 1366
    },
    {
      "epoch": 0.11509015956700973,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0003,
      "loss": 1.8911672830581665,
      "step": 1368
    },
    {
      "epoch": 0.11525842003421297,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003,
      "loss": 1.9256126880645752,
      "step": 1370
    },
    {
      "epoch": 0.11542668050141619,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.9157308340072632,
      "step": 1372
    },
    {
      "epoch": 0.11559494096861943,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.9555965662002563,
      "step": 1374
    },
    {
      "epoch": 0.11576320143582265,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.912801742553711,
      "step": 1376
    },
    {
      "epoch": 0.11593146190302589,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.9241381883621216,
      "step": 1378
    },
    {
      "epoch": 0.11609972237022911,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8898686170578003,
      "step": 1380
    },
    {
      "epoch": 0.11626798283743235,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9342970848083496,
      "step": 1382
    },
    {
      "epoch": 0.11643624330463558,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9026743173599243,
      "step": 1384
    },
    {
      "epoch": 0.11660450377183881,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.92124342918396,
      "step": 1386
    },
    {
      "epoch": 0.11677276423904204,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.9418835639953613,
      "step": 1388
    },
    {
      "epoch": 0.11694102470624526,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8985450267791748,
      "step": 1390
    },
    {
      "epoch": 0.1171092851734485,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.938198447227478,
      "step": 1392
    },
    {
      "epoch": 0.11727754564065172,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8539841175079346,
      "step": 1394
    },
    {
      "epoch": 0.11744580610785496,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9347398281097412,
      "step": 1396
    },
    {
      "epoch": 0.11761406657505818,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9070472717285156,
      "step": 1398
    },
    {
      "epoch": 0.11778232704226142,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9620471000671387,
      "step": 1400
    },
    {
      "epoch": 0.11795058750946465,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9475021362304688,
      "step": 1402
    },
    {
      "epoch": 0.11811884797666788,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9082636833190918,
      "step": 1404
    },
    {
      "epoch": 0.11828710844387111,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9146697521209717,
      "step": 1406
    },
    {
      "epoch": 0.11845536891107435,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.9484641551971436,
      "step": 1408
    },
    {
      "epoch": 0.11862362937827757,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.934142827987671,
      "step": 1410
    },
    {
      "epoch": 0.11879188984548081,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.910097599029541,
      "step": 1412
    },
    {
      "epoch": 0.11896015031268403,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0003,
      "loss": 1.9728598594665527,
      "step": 1414
    },
    {
      "epoch": 0.11912841077988727,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003,
      "loss": 1.8408620357513428,
      "step": 1416
    },
    {
      "epoch": 0.1192966712470905,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9596705436706543,
      "step": 1418
    },
    {
      "epoch": 0.11946493171429373,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 1.9008151292800903,
      "step": 1420
    },
    {
      "epoch": 0.11963319218149696,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.9676182270050049,
      "step": 1422
    },
    {
      "epoch": 0.1198014526487002,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8787163496017456,
      "step": 1424
    },
    {
      "epoch": 0.11996971311590342,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8992860317230225,
      "step": 1426
    },
    {
      "epoch": 0.12013797358310666,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9389331340789795,
      "step": 1428
    },
    {
      "epoch": 0.12030623405030988,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 1.9002516269683838,
      "step": 1430
    },
    {
      "epoch": 0.12047449451751312,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9220097064971924,
      "step": 1432
    },
    {
      "epoch": 0.12064275498471634,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.9098525047302246,
      "step": 1434
    },
    {
      "epoch": 0.12081101545191957,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003,
      "loss": 1.922823190689087,
      "step": 1436
    },
    {
      "epoch": 0.1209792759191228,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8835954666137695,
      "step": 1438
    },
    {
      "epoch": 0.12114753638632603,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9585754871368408,
      "step": 1440
    },
    {
      "epoch": 0.12131579685352926,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.9392826557159424,
      "step": 1442
    },
    {
      "epoch": 0.12148405732073249,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9020988941192627,
      "step": 1444
    },
    {
      "epoch": 0.12165231778793573,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.912530541419983,
      "step": 1446
    },
    {
      "epoch": 0.12182057825513895,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.925055980682373,
      "step": 1448
    },
    {
      "epoch": 0.12198883872234219,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9380073547363281,
      "step": 1450
    },
    {
      "epoch": 0.12215709918954541,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9606401920318604,
      "step": 1452
    },
    {
      "epoch": 0.12232535965674865,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9039701223373413,
      "step": 1454
    },
    {
      "epoch": 0.12249362012395187,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9113785028457642,
      "step": 1456
    },
    {
      "epoch": 0.12266188059115511,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9013999700546265,
      "step": 1458
    },
    {
      "epoch": 0.12283014105835834,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0003,
      "loss": 1.9261503219604492,
      "step": 1460
    },
    {
      "epoch": 0.12299840152556157,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8820581436157227,
      "step": 1462
    },
    {
      "epoch": 0.1231666619927648,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9189097881317139,
      "step": 1464
    },
    {
      "epoch": 0.12333492245996804,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.918060302734375,
      "step": 1466
    },
    {
      "epoch": 0.12350318292717126,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9691660404205322,
      "step": 1468
    },
    {
      "epoch": 0.1236714433943745,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 1.8908185958862305,
      "step": 1470
    },
    {
      "epoch": 0.12383970386157772,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9254097938537598,
      "step": 1472
    },
    {
      "epoch": 0.12400796432878096,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8695099353790283,
      "step": 1474
    },
    {
      "epoch": 0.12417622479598418,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9370555877685547,
      "step": 1476
    },
    {
      "epoch": 0.12434448526318742,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9035735130310059,
      "step": 1478
    },
    {
      "epoch": 0.12451274573039064,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9203996658325195,
      "step": 1480
    },
    {
      "epoch": 0.12468100619759387,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9661009311676025,
      "step": 1482
    },
    {
      "epoch": 0.1248492666647971,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9500954151153564,
      "step": 1484
    },
    {
      "epoch": 0.12501752713200034,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9874556064605713,
      "step": 1486
    },
    {
      "epoch": 0.12518578759920357,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.9230115413665771,
      "step": 1488
    },
    {
      "epoch": 0.1253540480664068,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.926455020904541,
      "step": 1490
    },
    {
      "epoch": 0.12552230853361002,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9307589530944824,
      "step": 1492
    },
    {
      "epoch": 0.12569056900081327,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 1.9020943641662598,
      "step": 1494
    },
    {
      "epoch": 0.1258588294680165,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0003,
      "loss": 1.943596363067627,
      "step": 1496
    },
    {
      "epoch": 0.12602708993521972,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.8687384128570557,
      "step": 1498
    },
    {
      "epoch": 0.12619535040242294,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0003,
      "loss": 1.8728657960891724,
      "step": 1500
    },
    {
      "epoch": 0.1263636108696262,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9269788265228271,
      "step": 1502
    },
    {
      "epoch": 0.12653187133682942,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003,
      "loss": 1.9444890022277832,
      "step": 1504
    },
    {
      "epoch": 0.12670013180403264,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.886991024017334,
      "step": 1506
    },
    {
      "epoch": 0.12686839227123586,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9223649501800537,
      "step": 1508
    },
    {
      "epoch": 0.12703665273843912,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8746082782745361,
      "step": 1510
    },
    {
      "epoch": 0.12720491320564234,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.8975021839141846,
      "step": 1512
    },
    {
      "epoch": 0.12737317367284556,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9312758445739746,
      "step": 1514
    },
    {
      "epoch": 0.1275414341400488,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9282803535461426,
      "step": 1516
    },
    {
      "epoch": 0.12770969460725204,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9538241624832153,
      "step": 1518
    },
    {
      "epoch": 0.12787795507445526,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.918857216835022,
      "step": 1520
    },
    {
      "epoch": 0.1280462155416585,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0003,
      "loss": 1.9400659799575806,
      "step": 1522
    },
    {
      "epoch": 0.1282144760088617,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9128485918045044,
      "step": 1524
    },
    {
      "epoch": 0.12838273647606493,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9793527126312256,
      "step": 1526
    },
    {
      "epoch": 0.1285509969432682,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0003,
      "loss": 1.9110498428344727,
      "step": 1528
    },
    {
      "epoch": 0.1287192574104714,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0003,
      "loss": 1.9394385814666748,
      "step": 1530
    },
    {
      "epoch": 0.12888751787767463,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0003,
      "loss": 1.8879005908966064,
      "step": 1532
    },
    {
      "epoch": 0.12905577834487786,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0003,
      "loss": 1.8270690441131592,
      "step": 1534
    },
    {
      "epoch": 0.1292240388120811,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0003,
      "loss": 1.9067952632904053,
      "step": 1536
    },
    {
      "epoch": 0.12939229927928433,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9525420665740967,
      "step": 1538
    },
    {
      "epoch": 0.12956055974648756,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0003,
      "loss": 1.959310531616211,
      "step": 1540
    },
    {
      "epoch": 0.12972882021369078,
      "grad_norm": 0.28125,
      "learning_rate": 0.0003,
      "loss": 1.9132158756256104,
      "step": 1542
    },
    {
      "epoch": 0.12989708068089403,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0003,
      "loss": 1.917759895324707,
      "step": 1544
    },
    {
      "epoch": 0.13006534114809726,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9803400039672852,
      "step": 1546
    },
    {
      "epoch": 0.13023360161530048,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.8750720024108887,
      "step": 1548
    },
    {
      "epoch": 0.1304018620825037,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.9353982210159302,
      "step": 1550
    },
    {
      "epoch": 0.13057012254970696,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 1.9718754291534424,
      "step": 1552
    },
    {
      "epoch": 0.13073838301691018,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.9102880954742432,
      "step": 1554
    },
    {
      "epoch": 0.1309066434841134,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9387660026550293,
      "step": 1556
    },
    {
      "epoch": 0.13107490395131663,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8836321830749512,
      "step": 1558
    },
    {
      "epoch": 0.13124316441851988,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003,
      "loss": 1.9512689113616943,
      "step": 1560
    },
    {
      "epoch": 0.1314114248857231,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9147279262542725,
      "step": 1562
    },
    {
      "epoch": 0.13157968535292633,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.8989226818084717,
      "step": 1564
    },
    {
      "epoch": 0.13174794582012955,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8916362524032593,
      "step": 1566
    },
    {
      "epoch": 0.1319162062873328,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9530980587005615,
      "step": 1568
    },
    {
      "epoch": 0.13208446675453603,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9556283950805664,
      "step": 1570
    },
    {
      "epoch": 0.13225272722173925,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9469244480133057,
      "step": 1572
    },
    {
      "epoch": 0.13242098768894248,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8763182163238525,
      "step": 1574
    },
    {
      "epoch": 0.1325892481561457,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.876267910003662,
      "step": 1576
    },
    {
      "epoch": 0.13275750862334895,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 1.9887369871139526,
      "step": 1578
    },
    {
      "epoch": 0.13292576909055218,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9403808116912842,
      "step": 1580
    },
    {
      "epoch": 0.1330940295577554,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9282402992248535,
      "step": 1582
    },
    {
      "epoch": 0.13326229002495862,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9058828353881836,
      "step": 1584
    },
    {
      "epoch": 0.13343055049216188,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9461846351623535,
      "step": 1586
    },
    {
      "epoch": 0.1335988109593651,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.8889780044555664,
      "step": 1588
    },
    {
      "epoch": 0.13376707142656832,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003,
      "loss": 1.8459279537200928,
      "step": 1590
    },
    {
      "epoch": 0.13393533189377155,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9330594539642334,
      "step": 1592
    },
    {
      "epoch": 0.1341035923609748,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.902472734451294,
      "step": 1594
    },
    {
      "epoch": 0.13427185282817802,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9289157390594482,
      "step": 1596
    },
    {
      "epoch": 0.13444011329538125,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8994752168655396,
      "step": 1598
    },
    {
      "epoch": 0.13460837376258447,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9086780548095703,
      "step": 1600
    },
    {
      "epoch": 0.13477663422978772,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.88742995262146,
      "step": 1602
    },
    {
      "epoch": 0.13494489469699095,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9064100980758667,
      "step": 1604
    },
    {
      "epoch": 0.13511315516419417,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0003,
      "loss": 1.909019112586975,
      "step": 1606
    },
    {
      "epoch": 0.1352814156313974,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.9131169319152832,
      "step": 1608
    },
    {
      "epoch": 0.13544967609860065,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9193834066390991,
      "step": 1610
    },
    {
      "epoch": 0.13561793656580387,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 1.8415677547454834,
      "step": 1612
    },
    {
      "epoch": 0.1357861970330071,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.941443920135498,
      "step": 1614
    },
    {
      "epoch": 0.13595445750021032,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8901164531707764,
      "step": 1616
    },
    {
      "epoch": 0.13612271796741354,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9697635173797607,
      "step": 1618
    },
    {
      "epoch": 0.1362909784346168,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9256943464279175,
      "step": 1620
    },
    {
      "epoch": 0.13645923890182002,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9321902990341187,
      "step": 1622
    },
    {
      "epoch": 0.13662749936902324,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9907758235931396,
      "step": 1624
    },
    {
      "epoch": 0.13679575983622647,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8780033588409424,
      "step": 1626
    },
    {
      "epoch": 0.13696402030342972,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.894285798072815,
      "step": 1628
    },
    {
      "epoch": 0.13713228077063294,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8752095699310303,
      "step": 1630
    },
    {
      "epoch": 0.13730054123783617,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.878859281539917,
      "step": 1632
    },
    {
      "epoch": 0.1374688017050394,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9353609085083008,
      "step": 1634
    },
    {
      "epoch": 0.13763706217224264,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9139010906219482,
      "step": 1636
    },
    {
      "epoch": 0.13780532263944587,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0003,
      "loss": 1.9122400283813477,
      "step": 1638
    },
    {
      "epoch": 0.1379735831066491,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8992652893066406,
      "step": 1640
    },
    {
      "epoch": 0.1381418435738523,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.891580581665039,
      "step": 1642
    },
    {
      "epoch": 0.13831010404105556,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.926457166671753,
      "step": 1644
    },
    {
      "epoch": 0.1384783645082588,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8813509941101074,
      "step": 1646
    },
    {
      "epoch": 0.138646624975462,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9348347187042236,
      "step": 1648
    },
    {
      "epoch": 0.13881488544266524,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0003,
      "loss": 1.944808006286621,
      "step": 1650
    },
    {
      "epoch": 0.1389831459098685,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.942755937576294,
      "step": 1652
    },
    {
      "epoch": 0.1391514063770717,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0003,
      "loss": 1.8784301280975342,
      "step": 1654
    },
    {
      "epoch": 0.13931966684427494,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9113514423370361,
      "step": 1656
    },
    {
      "epoch": 0.13948792731147816,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003,
      "loss": 1.9650113582611084,
      "step": 1658
    },
    {
      "epoch": 0.1396561877786814,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 1.9518228769302368,
      "step": 1660
    },
    {
      "epoch": 0.13982444824588464,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.942530632019043,
      "step": 1662
    },
    {
      "epoch": 0.13999270871308786,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.891254186630249,
      "step": 1664
    },
    {
      "epoch": 0.14016096918029108,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.913703441619873,
      "step": 1666
    },
    {
      "epoch": 0.1403292296474943,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 1.9339816570281982,
      "step": 1668
    },
    {
      "epoch": 0.14049749011469756,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.906704068183899,
      "step": 1670
    },
    {
      "epoch": 0.14066575058190078,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8969719409942627,
      "step": 1672
    },
    {
      "epoch": 0.140834011049104,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8693054914474487,
      "step": 1674
    },
    {
      "epoch": 0.14100227151630723,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8892066478729248,
      "step": 1676
    },
    {
      "epoch": 0.14117053198351048,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9135198593139648,
      "step": 1678
    },
    {
      "epoch": 0.1413387924507137,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8984663486480713,
      "step": 1680
    },
    {
      "epoch": 0.14150705291791693,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 1.9017069339752197,
      "step": 1682
    },
    {
      "epoch": 0.14167531338512016,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0003,
      "loss": 1.9224605560302734,
      "step": 1684
    },
    {
      "epoch": 0.1418435738523234,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.9038444757461548,
      "step": 1686
    },
    {
      "epoch": 0.14201183431952663,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.874229073524475,
      "step": 1688
    },
    {
      "epoch": 0.14218009478672985,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8504348993301392,
      "step": 1690
    },
    {
      "epoch": 0.14234835525393308,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.851838231086731,
      "step": 1692
    },
    {
      "epoch": 0.14251661572113633,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.8895413875579834,
      "step": 1694
    },
    {
      "epoch": 0.14268487618833955,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9130281209945679,
      "step": 1696
    },
    {
      "epoch": 0.14285313665554278,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 1.8889652490615845,
      "step": 1698
    },
    {
      "epoch": 0.143021397122746,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9354965686798096,
      "step": 1700
    },
    {
      "epoch": 0.14318965758994925,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8560874462127686,
      "step": 1702
    },
    {
      "epoch": 0.14335791805715248,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8989894390106201,
      "step": 1704
    },
    {
      "epoch": 0.1435261785243557,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.879460334777832,
      "step": 1706
    },
    {
      "epoch": 0.14369443899155893,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8895301818847656,
      "step": 1708
    },
    {
      "epoch": 0.14386269945876218,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9694263935089111,
      "step": 1710
    },
    {
      "epoch": 0.1440309599259654,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8473427295684814,
      "step": 1712
    },
    {
      "epoch": 0.14419922039316863,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.892897605895996,
      "step": 1714
    },
    {
      "epoch": 0.14436748086037185,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9652364253997803,
      "step": 1716
    },
    {
      "epoch": 0.14453574132757507,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8382337093353271,
      "step": 1718
    },
    {
      "epoch": 0.14470400179477833,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9097328186035156,
      "step": 1720
    },
    {
      "epoch": 0.14487226226198155,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0003,
      "loss": 1.8483141660690308,
      "step": 1722
    },
    {
      "epoch": 0.14504052272918477,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8157551288604736,
      "step": 1724
    },
    {
      "epoch": 0.145208783196388,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8763340711593628,
      "step": 1726
    },
    {
      "epoch": 0.14537704366359125,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8998136520385742,
      "step": 1728
    },
    {
      "epoch": 0.14554530413079447,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003,
      "loss": 1.9248456954956055,
      "step": 1730
    },
    {
      "epoch": 0.1457135645979977,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9096980094909668,
      "step": 1732
    },
    {
      "epoch": 0.14588182506520092,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.857722282409668,
      "step": 1734
    },
    {
      "epoch": 0.14605008553240417,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003,
      "loss": 1.8995428085327148,
      "step": 1736
    },
    {
      "epoch": 0.1462183459996074,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.943412184715271,
      "step": 1738
    },
    {
      "epoch": 0.14638660646681062,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.898258924484253,
      "step": 1740
    },
    {
      "epoch": 0.14655486693401384,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8995907306671143,
      "step": 1742
    },
    {
      "epoch": 0.1467231274012171,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9514412879943848,
      "step": 1744
    },
    {
      "epoch": 0.14689138786842032,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.9101895093917847,
      "step": 1746
    },
    {
      "epoch": 0.14705964833562354,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9517016410827637,
      "step": 1748
    },
    {
      "epoch": 0.14722790880282677,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9457144737243652,
      "step": 1750
    },
    {
      "epoch": 0.14739616927003002,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9343342781066895,
      "step": 1752
    },
    {
      "epoch": 0.14756442973723324,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8750689029693604,
      "step": 1754
    },
    {
      "epoch": 0.14773269020443647,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.960473895072937,
      "step": 1756
    },
    {
      "epoch": 0.1479009506716397,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9108078479766846,
      "step": 1758
    },
    {
      "epoch": 0.14806921113884292,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9007946252822876,
      "step": 1760
    },
    {
      "epoch": 0.14823747160604617,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.897582769393921,
      "step": 1762
    },
    {
      "epoch": 0.1484057320732494,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8524744510650635,
      "step": 1764
    },
    {
      "epoch": 0.14857399254045262,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9460142850875854,
      "step": 1766
    },
    {
      "epoch": 0.14874225300765584,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.892939805984497,
      "step": 1768
    },
    {
      "epoch": 0.1489105134748591,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9314773082733154,
      "step": 1770
    },
    {
      "epoch": 0.14907877394206231,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.890549659729004,
      "step": 1772
    },
    {
      "epoch": 0.14924703440926554,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8880906105041504,
      "step": 1774
    },
    {
      "epoch": 0.14941529487646876,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9373466968536377,
      "step": 1776
    },
    {
      "epoch": 0.14958355534367201,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.902622938156128,
      "step": 1778
    },
    {
      "epoch": 0.14975181581087524,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9048649072647095,
      "step": 1780
    },
    {
      "epoch": 0.14992007627807846,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9274513721466064,
      "step": 1782
    },
    {
      "epoch": 0.1500883367452817,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0003,
      "loss": 1.9271514415740967,
      "step": 1784
    },
    {
      "epoch": 0.15025659721248494,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8906430006027222,
      "step": 1786
    },
    {
      "epoch": 0.15042485767968816,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.933445930480957,
      "step": 1788
    },
    {
      "epoch": 0.15059311814689139,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8756215572357178,
      "step": 1790
    },
    {
      "epoch": 0.1507613786140946,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9951579570770264,
      "step": 1792
    },
    {
      "epoch": 0.15092963908129786,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9419078826904297,
      "step": 1794
    },
    {
      "epoch": 0.15109789954850109,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.907381534576416,
      "step": 1796
    },
    {
      "epoch": 0.1512661600157043,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9230918884277344,
      "step": 1798
    },
    {
      "epoch": 0.15143442048290753,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9106059074401855,
      "step": 1800
    },
    {
      "epoch": 0.15160268095011079,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9121595621109009,
      "step": 1802
    },
    {
      "epoch": 0.151770941417314,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.929678201675415,
      "step": 1804
    },
    {
      "epoch": 0.15193920188451723,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.930748462677002,
      "step": 1806
    },
    {
      "epoch": 0.15210746235172046,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9208400249481201,
      "step": 1808
    },
    {
      "epoch": 0.15227572281892368,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8854045867919922,
      "step": 1810
    },
    {
      "epoch": 0.15244398328612693,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8867055177688599,
      "step": 1812
    },
    {
      "epoch": 0.15261224375333016,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9168126583099365,
      "step": 1814
    },
    {
      "epoch": 0.15278050422053338,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8986386060714722,
      "step": 1816
    },
    {
      "epoch": 0.1529487646877366,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9546396732330322,
      "step": 1818
    },
    {
      "epoch": 0.15311702515493986,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.919301152229309,
      "step": 1820
    },
    {
      "epoch": 0.15328528562214308,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.922861099243164,
      "step": 1822
    },
    {
      "epoch": 0.1534535460893463,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9093223810195923,
      "step": 1824
    },
    {
      "epoch": 0.15362180655654953,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9522987604141235,
      "step": 1826
    },
    {
      "epoch": 0.15379006702375278,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8948850631713867,
      "step": 1828
    },
    {
      "epoch": 0.153958327490956,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8965816497802734,
      "step": 1830
    },
    {
      "epoch": 0.15412658795815923,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9258787631988525,
      "step": 1832
    },
    {
      "epoch": 0.15429484842536245,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8690595626831055,
      "step": 1834
    },
    {
      "epoch": 0.1544631088925657,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 1.9410321712493896,
      "step": 1836
    },
    {
      "epoch": 0.15463136935976893,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8994879722595215,
      "step": 1838
    },
    {
      "epoch": 0.15479962982697215,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9002888202667236,
      "step": 1840
    },
    {
      "epoch": 0.15496789029417538,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.898413062095642,
      "step": 1842
    },
    {
      "epoch": 0.15513615076137863,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003,
      "loss": 1.9042813777923584,
      "step": 1844
    },
    {
      "epoch": 0.15530441122858185,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.909853219985962,
      "step": 1846
    },
    {
      "epoch": 0.15547267169578508,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.908280372619629,
      "step": 1848
    },
    {
      "epoch": 0.1556409321629883,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.86564302444458,
      "step": 1850
    },
    {
      "epoch": 0.15580919263019152,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8647704124450684,
      "step": 1852
    },
    {
      "epoch": 0.15597745309739477,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9200609922409058,
      "step": 1854
    },
    {
      "epoch": 0.156145713564598,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9245845079421997,
      "step": 1856
    },
    {
      "epoch": 0.15631397403180122,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8852876424789429,
      "step": 1858
    },
    {
      "epoch": 0.15648223449900445,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.911393165588379,
      "step": 1860
    },
    {
      "epoch": 0.1566504949662077,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8676481246948242,
      "step": 1862
    },
    {
      "epoch": 0.15681875543341092,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8621420860290527,
      "step": 1864
    },
    {
      "epoch": 0.15698701590061415,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9409252405166626,
      "step": 1866
    },
    {
      "epoch": 0.15715527636781737,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8547532558441162,
      "step": 1868
    },
    {
      "epoch": 0.15732353683502062,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8879642486572266,
      "step": 1870
    },
    {
      "epoch": 0.15749179730222385,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.869751214981079,
      "step": 1872
    },
    {
      "epoch": 0.15766005776942707,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9576621055603027,
      "step": 1874
    },
    {
      "epoch": 0.1578283182366303,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9053308963775635,
      "step": 1876
    },
    {
      "epoch": 0.15799657870383355,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8636329174041748,
      "step": 1878
    },
    {
      "epoch": 0.15816483917103677,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9255702495574951,
      "step": 1880
    },
    {
      "epoch": 0.15833309963824,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9268097877502441,
      "step": 1882
    },
    {
      "epoch": 0.15850136010544322,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.918994665145874,
      "step": 1884
    },
    {
      "epoch": 0.15866962057264647,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9072155952453613,
      "step": 1886
    },
    {
      "epoch": 0.1588378810398497,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.9028345346450806,
      "step": 1888
    },
    {
      "epoch": 0.15900614150705292,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.83545982837677,
      "step": 1890
    },
    {
      "epoch": 0.15917440197425614,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9076398611068726,
      "step": 1892
    },
    {
      "epoch": 0.1593426624414594,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9333021640777588,
      "step": 1894
    },
    {
      "epoch": 0.15951092290866262,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.888129472732544,
      "step": 1896
    },
    {
      "epoch": 0.15967918337586584,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8531389236450195,
      "step": 1898
    },
    {
      "epoch": 0.15984744384306906,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.878471851348877,
      "step": 1900
    },
    {
      "epoch": 0.1600157043102723,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.9046697616577148,
      "step": 1902
    },
    {
      "epoch": 0.16018396477747554,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9137377738952637,
      "step": 1904
    },
    {
      "epoch": 0.16035222524467876,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8737066984176636,
      "step": 1906
    },
    {
      "epoch": 0.160520485711882,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8786468505859375,
      "step": 1908
    },
    {
      "epoch": 0.1606887461790852,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8830469846725464,
      "step": 1910
    },
    {
      "epoch": 0.16085700664628846,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8983227014541626,
      "step": 1912
    },
    {
      "epoch": 0.1610252671134917,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9175633192062378,
      "step": 1914
    },
    {
      "epoch": 0.1611935275806949,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8765082359313965,
      "step": 1916
    },
    {
      "epoch": 0.16136178804789814,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9081376791000366,
      "step": 1918
    },
    {
      "epoch": 0.1615300485151014,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9461287260055542,
      "step": 1920
    },
    {
      "epoch": 0.1616983089823046,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8732094764709473,
      "step": 1922
    },
    {
      "epoch": 0.16186656944950784,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8930509090423584,
      "step": 1924
    },
    {
      "epoch": 0.16203482991671106,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8589091300964355,
      "step": 1926
    },
    {
      "epoch": 0.1622030903839143,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9120795726776123,
      "step": 1928
    },
    {
      "epoch": 0.16237135085111754,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8983142375946045,
      "step": 1930
    },
    {
      "epoch": 0.16253961131832076,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.878803014755249,
      "step": 1932
    },
    {
      "epoch": 0.16270787178552398,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.899629831314087,
      "step": 1934
    },
    {
      "epoch": 0.16287613225272723,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9440147876739502,
      "step": 1936
    },
    {
      "epoch": 0.16304439271993046,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.868260383605957,
      "step": 1938
    },
    {
      "epoch": 0.16321265318713368,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9430912733078003,
      "step": 1940
    },
    {
      "epoch": 0.1633809136543369,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.9207162857055664,
      "step": 1942
    },
    {
      "epoch": 0.16354917412154013,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.92635178565979,
      "step": 1944
    },
    {
      "epoch": 0.16371743458874338,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9298070669174194,
      "step": 1946
    },
    {
      "epoch": 0.1638856950559466,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.82794189453125,
      "step": 1948
    },
    {
      "epoch": 0.16405395552314983,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8791770935058594,
      "step": 1950
    },
    {
      "epoch": 0.16422221599035305,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9234704971313477,
      "step": 1952
    },
    {
      "epoch": 0.1643904764575563,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9317119121551514,
      "step": 1954
    },
    {
      "epoch": 0.16455873692475953,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.9528237581253052,
      "step": 1956
    },
    {
      "epoch": 0.16472699739196275,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.952071189880371,
      "step": 1958
    },
    {
      "epoch": 0.16489525785916598,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9093620777130127,
      "step": 1960
    },
    {
      "epoch": 0.16506351832636923,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.903261661529541,
      "step": 1962
    },
    {
      "epoch": 0.16523177879357245,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.937160611152649,
      "step": 1964
    },
    {
      "epoch": 0.16540003926077568,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9098243713378906,
      "step": 1966
    },
    {
      "epoch": 0.1655682997279789,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8944246768951416,
      "step": 1968
    },
    {
      "epoch": 0.16573656019518215,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9223178625106812,
      "step": 1970
    },
    {
      "epoch": 0.16590482066238538,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8958390951156616,
      "step": 1972
    },
    {
      "epoch": 0.1660730811295886,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8870856761932373,
      "step": 1974
    },
    {
      "epoch": 0.16624134159679183,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.910272240638733,
      "step": 1976
    },
    {
      "epoch": 0.16640960206399508,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.9190630912780762,
      "step": 1978
    },
    {
      "epoch": 0.1665778625311983,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8840372562408447,
      "step": 1980
    },
    {
      "epoch": 0.16674612299840152,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9015569686889648,
      "step": 1982
    },
    {
      "epoch": 0.16691438346560475,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9520008563995361,
      "step": 1984
    },
    {
      "epoch": 0.167082643932808,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8736324310302734,
      "step": 1986
    },
    {
      "epoch": 0.16725090440001122,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.932429552078247,
      "step": 1988
    },
    {
      "epoch": 0.16741916486721445,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 1.9470398426055908,
      "step": 1990
    },
    {
      "epoch": 0.16758742533441767,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.884209394454956,
      "step": 1992
    },
    {
      "epoch": 0.1677556858016209,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9018237590789795,
      "step": 1994
    },
    {
      "epoch": 0.16792394626882415,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8821579217910767,
      "step": 1996
    },
    {
      "epoch": 0.16809220673602737,
      "grad_norm": 2.25,
      "learning_rate": 0.0003,
      "loss": 1.873974323272705,
      "step": 1998
    },
    {
      "epoch": 0.1682604672032306,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0003,
      "loss": 1.8591142892837524,
      "step": 2000
    },
    {
      "epoch": 0.16842872767043382,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0003,
      "loss": 1.9452348947525024,
      "step": 2002
    },
    {
      "epoch": 0.16859698813763707,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9141290187835693,
      "step": 2004
    },
    {
      "epoch": 0.1687652486048403,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 1.837856650352478,
      "step": 2006
    },
    {
      "epoch": 0.16893350907204352,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0003,
      "loss": 1.9166812896728516,
      "step": 2008
    },
    {
      "epoch": 0.16910176953924674,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9178876876831055,
      "step": 2010
    },
    {
      "epoch": 0.16927003000645,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0003,
      "loss": 1.9326244592666626,
      "step": 2012
    },
    {
      "epoch": 0.16943829047365322,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0003,
      "loss": 1.8681262731552124,
      "step": 2014
    },
    {
      "epoch": 0.16960655094085644,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0003,
      "loss": 1.9190623760223389,
      "step": 2016
    },
    {
      "epoch": 0.16977481140805967,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0003,
      "loss": 1.8956677913665771,
      "step": 2018
    },
    {
      "epoch": 0.16994307187526292,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8758394718170166,
      "step": 2020
    },
    {
      "epoch": 0.17011133234246614,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0003,
      "loss": 1.9177134037017822,
      "step": 2022
    },
    {
      "epoch": 0.17027959280966937,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.9071934223175049,
      "step": 2024
    },
    {
      "epoch": 0.1704478532768726,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9139493703842163,
      "step": 2026
    },
    {
      "epoch": 0.17061611374407584,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.8913884162902832,
      "step": 2028
    },
    {
      "epoch": 0.17078437421127907,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9215452671051025,
      "step": 2030
    },
    {
      "epoch": 0.1709526346784823,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.891826868057251,
      "step": 2032
    },
    {
      "epoch": 0.17112089514568551,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9349538087844849,
      "step": 2034
    },
    {
      "epoch": 0.17128915561288874,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9072521924972534,
      "step": 2036
    },
    {
      "epoch": 0.171457416080092,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8940359354019165,
      "step": 2038
    },
    {
      "epoch": 0.17162567654729521,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.9159648418426514,
      "step": 2040
    },
    {
      "epoch": 0.17179393701449844,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8832592964172363,
      "step": 2042
    },
    {
      "epoch": 0.17196219748170166,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8573832511901855,
      "step": 2044
    },
    {
      "epoch": 0.1721304579489049,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 1.8690813779830933,
      "step": 2046
    },
    {
      "epoch": 0.17229871841610814,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9088480472564697,
      "step": 2048
    },
    {
      "epoch": 0.17246697888331136,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8870704174041748,
      "step": 2050
    },
    {
      "epoch": 0.17263523935051459,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.881723165512085,
      "step": 2052
    },
    {
      "epoch": 0.17280349981771784,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8444240093231201,
      "step": 2054
    },
    {
      "epoch": 0.17297176028492106,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.901038646697998,
      "step": 2056
    },
    {
      "epoch": 0.17314002075212429,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8534339666366577,
      "step": 2058
    },
    {
      "epoch": 0.1733082812193275,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.842780351638794,
      "step": 2060
    },
    {
      "epoch": 0.17347654168653076,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9291367530822754,
      "step": 2062
    },
    {
      "epoch": 0.17364480215373398,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003,
      "loss": 1.8362650871276855,
      "step": 2064
    },
    {
      "epoch": 0.1738130626209372,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003,
      "loss": 1.8636300563812256,
      "step": 2066
    },
    {
      "epoch": 0.17398132308814043,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8909094333648682,
      "step": 2068
    },
    {
      "epoch": 0.17414958355534368,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8569002151489258,
      "step": 2070
    },
    {
      "epoch": 0.1743178440225469,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003,
      "loss": 1.921522617340088,
      "step": 2072
    },
    {
      "epoch": 0.17448610448975013,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.9088025093078613,
      "step": 2074
    },
    {
      "epoch": 0.17465436495695336,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9105111360549927,
      "step": 2076
    },
    {
      "epoch": 0.1748226254241566,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9110347032546997,
      "step": 2078
    },
    {
      "epoch": 0.17499088589135983,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8619871139526367,
      "step": 2080
    },
    {
      "epoch": 0.17515914635856306,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9244492053985596,
      "step": 2082
    },
    {
      "epoch": 0.17532740682576628,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9392669200897217,
      "step": 2084
    },
    {
      "epoch": 0.1754956672929695,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.87531316280365,
      "step": 2086
    },
    {
      "epoch": 0.17566392776017276,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8729488849639893,
      "step": 2088
    },
    {
      "epoch": 0.17583218822737598,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.945138692855835,
      "step": 2090
    },
    {
      "epoch": 0.1760004486945792,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.8613954782485962,
      "step": 2092
    },
    {
      "epoch": 0.17616870916178243,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8956997394561768,
      "step": 2094
    },
    {
      "epoch": 0.17633696962898568,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.931321144104004,
      "step": 2096
    },
    {
      "epoch": 0.1765052300961889,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8895444869995117,
      "step": 2098
    },
    {
      "epoch": 0.17667349056339213,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9359303712844849,
      "step": 2100
    },
    {
      "epoch": 0.17684175103059535,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8560831546783447,
      "step": 2102
    },
    {
      "epoch": 0.1770100114977986,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8994574546813965,
      "step": 2104
    },
    {
      "epoch": 0.17717827196500183,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9093024730682373,
      "step": 2106
    },
    {
      "epoch": 0.17734653243220505,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.8999031782150269,
      "step": 2108
    },
    {
      "epoch": 0.17751479289940827,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003,
      "loss": 1.830657720565796,
      "step": 2110
    },
    {
      "epoch": 0.17768305336661153,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9131743907928467,
      "step": 2112
    },
    {
      "epoch": 0.17785131383381475,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.9264907836914062,
      "step": 2114
    },
    {
      "epoch": 0.17801957430101797,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8963432312011719,
      "step": 2116
    },
    {
      "epoch": 0.1781878347682212,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.889513373374939,
      "step": 2118
    },
    {
      "epoch": 0.17835609523542445,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8996819257736206,
      "step": 2120
    },
    {
      "epoch": 0.17852435570262767,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.894258975982666,
      "step": 2122
    },
    {
      "epoch": 0.1786926161698309,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8646239042282104,
      "step": 2124
    },
    {
      "epoch": 0.17886087663703412,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9088584184646606,
      "step": 2126
    },
    {
      "epoch": 0.17902913710423735,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8768305778503418,
      "step": 2128
    },
    {
      "epoch": 0.1791973975714406,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9053537845611572,
      "step": 2130
    },
    {
      "epoch": 0.17936565803864382,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.895362138748169,
      "step": 2132
    },
    {
      "epoch": 0.17953391850584705,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9364423751831055,
      "step": 2134
    },
    {
      "epoch": 0.17970217897305027,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.900244951248169,
      "step": 2136
    },
    {
      "epoch": 0.17987043944025352,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.891688346862793,
      "step": 2138
    },
    {
      "epoch": 0.18003869990745675,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8543533086776733,
      "step": 2140
    },
    {
      "epoch": 0.18020696037465997,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.911575436592102,
      "step": 2142
    },
    {
      "epoch": 0.1803752208418632,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.966556191444397,
      "step": 2144
    },
    {
      "epoch": 0.18054348130906644,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8940355777740479,
      "step": 2146
    },
    {
      "epoch": 0.18071174177626967,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8871724605560303,
      "step": 2148
    },
    {
      "epoch": 0.1808800022434729,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8933637142181396,
      "step": 2150
    },
    {
      "epoch": 0.18104826271067612,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9045274257659912,
      "step": 2152
    },
    {
      "epoch": 0.18121652317787937,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.903721570968628,
      "step": 2154
    },
    {
      "epoch": 0.1813847836450826,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8764883279800415,
      "step": 2156
    },
    {
      "epoch": 0.18155304411228582,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.873652458190918,
      "step": 2158
    },
    {
      "epoch": 0.18172130457948904,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9222973585128784,
      "step": 2160
    },
    {
      "epoch": 0.1818895650466923,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9169238805770874,
      "step": 2162
    },
    {
      "epoch": 0.18205782551389552,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8299540281295776,
      "step": 2164
    },
    {
      "epoch": 0.18222608598109874,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8805582523345947,
      "step": 2166
    },
    {
      "epoch": 0.18239434644830196,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8616507053375244,
      "step": 2168
    },
    {
      "epoch": 0.18256260691550522,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8890275955200195,
      "step": 2170
    },
    {
      "epoch": 0.18273086738270844,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9325604438781738,
      "step": 2172
    },
    {
      "epoch": 0.18289912784991166,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8744213581085205,
      "step": 2174
    },
    {
      "epoch": 0.1830673883171149,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.9184887409210205,
      "step": 2176
    },
    {
      "epoch": 0.1832356487843181,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9057884216308594,
      "step": 2178
    },
    {
      "epoch": 0.18340390925152136,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003,
      "loss": 1.8536261320114136,
      "step": 2180
    },
    {
      "epoch": 0.1835721697187246,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8988028764724731,
      "step": 2182
    },
    {
      "epoch": 0.1837404301859278,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003,
      "loss": 1.8009405136108398,
      "step": 2184
    },
    {
      "epoch": 0.18390869065313104,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8986436128616333,
      "step": 2186
    },
    {
      "epoch": 0.1840769511203343,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8954215049743652,
      "step": 2188
    },
    {
      "epoch": 0.1842452115875375,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.873329520225525,
      "step": 2190
    },
    {
      "epoch": 0.18441347205474073,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8822336196899414,
      "step": 2192
    },
    {
      "epoch": 0.18458173252194396,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003,
      "loss": 1.9465553760528564,
      "step": 2194
    },
    {
      "epoch": 0.1847499929891472,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9513742923736572,
      "step": 2196
    },
    {
      "epoch": 0.18491825345635043,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8866721391677856,
      "step": 2198
    },
    {
      "epoch": 0.18508651392355366,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8958959579467773,
      "step": 2200
    },
    {
      "epoch": 0.18525477439075688,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.893910527229309,
      "step": 2202
    },
    {
      "epoch": 0.18542303485796013,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8532644510269165,
      "step": 2204
    },
    {
      "epoch": 0.18559129532516336,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9097349643707275,
      "step": 2206
    },
    {
      "epoch": 0.18575955579236658,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.869520902633667,
      "step": 2208
    },
    {
      "epoch": 0.1859278162595698,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8858039379119873,
      "step": 2210
    },
    {
      "epoch": 0.18609607672677306,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8910448551177979,
      "step": 2212
    },
    {
      "epoch": 0.18626433719397628,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8475911617279053,
      "step": 2214
    },
    {
      "epoch": 0.1864325976611795,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8649992942810059,
      "step": 2216
    },
    {
      "epoch": 0.18660085812838273,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8680737018585205,
      "step": 2218
    },
    {
      "epoch": 0.18676911859558595,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.912632703781128,
      "step": 2220
    },
    {
      "epoch": 0.1869373790627892,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8501811027526855,
      "step": 2222
    },
    {
      "epoch": 0.18710563952999243,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9060676097869873,
      "step": 2224
    },
    {
      "epoch": 0.18727389999719565,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 1.8460618257522583,
      "step": 2226
    },
    {
      "epoch": 0.18744216046439888,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.920224905014038,
      "step": 2228
    },
    {
      "epoch": 0.18761042093160213,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.849515438079834,
      "step": 2230
    },
    {
      "epoch": 0.18777868139880535,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.8414572477340698,
      "step": 2232
    },
    {
      "epoch": 0.18794694186600858,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8758145570755005,
      "step": 2234
    },
    {
      "epoch": 0.1881152023332118,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003,
      "loss": 1.8957586288452148,
      "step": 2236
    },
    {
      "epoch": 0.18828346280041505,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.8633513450622559,
      "step": 2238
    },
    {
      "epoch": 0.18845172326761828,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.9372608661651611,
      "step": 2240
    },
    {
      "epoch": 0.1886199837348215,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003,
      "loss": 1.8389325141906738,
      "step": 2242
    },
    {
      "epoch": 0.18878824420202472,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8891093730926514,
      "step": 2244
    },
    {
      "epoch": 0.18895650466922798,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003,
      "loss": 1.9154226779937744,
      "step": 2246
    },
    {
      "epoch": 0.1891247651364312,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.864914059638977,
      "step": 2248
    },
    {
      "epoch": 0.18929302560363442,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9351489543914795,
      "step": 2250
    },
    {
      "epoch": 0.18946128607083765,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8818093538284302,
      "step": 2252
    },
    {
      "epoch": 0.1896295465380409,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9165527820587158,
      "step": 2254
    },
    {
      "epoch": 0.18979780700524412,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8950060606002808,
      "step": 2256
    },
    {
      "epoch": 0.18996606747244735,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9644579887390137,
      "step": 2258
    },
    {
      "epoch": 0.19013432793965057,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8388442993164062,
      "step": 2260
    },
    {
      "epoch": 0.19030258840685382,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.881636142730713,
      "step": 2262
    },
    {
      "epoch": 0.19047084887405705,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003,
      "loss": 1.9005322456359863,
      "step": 2264
    },
    {
      "epoch": 0.19063910934126027,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.9179902076721191,
      "step": 2266
    },
    {
      "epoch": 0.1908073698084635,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.878709077835083,
      "step": 2268
    },
    {
      "epoch": 0.19097563027566672,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.891688346862793,
      "step": 2270
    },
    {
      "epoch": 0.19114389074286997,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8465840816497803,
      "step": 2272
    },
    {
      "epoch": 0.1913121512100732,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8850433826446533,
      "step": 2274
    },
    {
      "epoch": 0.19148041167727642,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9318976402282715,
      "step": 2276
    },
    {
      "epoch": 0.19164867214447964,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8486225605010986,
      "step": 2278
    },
    {
      "epoch": 0.1918169326116829,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8399088382720947,
      "step": 2280
    },
    {
      "epoch": 0.19198519307888612,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9374675750732422,
      "step": 2282
    },
    {
      "epoch": 0.19215345354608934,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9223980903625488,
      "step": 2284
    },
    {
      "epoch": 0.19232171401329257,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8563040494918823,
      "step": 2286
    },
    {
      "epoch": 0.19248997448049582,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8222931623458862,
      "step": 2288
    },
    {
      "epoch": 0.19265823494769904,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8707482814788818,
      "step": 2290
    },
    {
      "epoch": 0.19282649541490227,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.865958571434021,
      "step": 2292
    },
    {
      "epoch": 0.1929947558821055,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8938682079315186,
      "step": 2294
    },
    {
      "epoch": 0.19316301634930874,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8308284282684326,
      "step": 2296
    },
    {
      "epoch": 0.19333127681651197,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8941811323165894,
      "step": 2298
    },
    {
      "epoch": 0.1934995372837152,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8708326816558838,
      "step": 2300
    },
    {
      "epoch": 0.1936677977509184,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.9350665807724,
      "step": 2302
    },
    {
      "epoch": 0.19383605821812167,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8526084423065186,
      "step": 2304
    },
    {
      "epoch": 0.1940043186853249,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.902608871459961,
      "step": 2306
    },
    {
      "epoch": 0.1941725791525281,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9083927869796753,
      "step": 2308
    },
    {
      "epoch": 0.19434083961973134,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.903369665145874,
      "step": 2310
    },
    {
      "epoch": 0.19450910008693456,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.941399335861206,
      "step": 2312
    },
    {
      "epoch": 0.1946773605541378,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8552172183990479,
      "step": 2314
    },
    {
      "epoch": 0.19484562102134104,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9012655019760132,
      "step": 2316
    },
    {
      "epoch": 0.19501388148854426,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.9537087678909302,
      "step": 2318
    },
    {
      "epoch": 0.19518214195574748,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8863122463226318,
      "step": 2320
    },
    {
      "epoch": 0.19535040242295074,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0003,
      "loss": 1.8727446794509888,
      "step": 2322
    },
    {
      "epoch": 0.19551866289015396,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8987562656402588,
      "step": 2324
    },
    {
      "epoch": 0.19568692335735718,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8671696186065674,
      "step": 2326
    },
    {
      "epoch": 0.1958551838245604,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8628203868865967,
      "step": 2328
    },
    {
      "epoch": 0.19602344429176366,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8989986181259155,
      "step": 2330
    },
    {
      "epoch": 0.19619170475896688,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8386404514312744,
      "step": 2332
    },
    {
      "epoch": 0.1963599652261701,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9309161901474,
      "step": 2334
    },
    {
      "epoch": 0.19652822569337333,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.863363265991211,
      "step": 2336
    },
    {
      "epoch": 0.19669648616057658,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003,
      "loss": 1.8830854892730713,
      "step": 2338
    },
    {
      "epoch": 0.1968647466277798,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9014075994491577,
      "step": 2340
    },
    {
      "epoch": 0.19703300709498303,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.941175103187561,
      "step": 2342
    },
    {
      "epoch": 0.19720126756218626,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9079504013061523,
      "step": 2344
    },
    {
      "epoch": 0.1973695280293895,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 1.8784403800964355,
      "step": 2346
    },
    {
      "epoch": 0.19753778849659273,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.870534896850586,
      "step": 2348
    },
    {
      "epoch": 0.19770604896379596,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8589330911636353,
      "step": 2350
    },
    {
      "epoch": 0.19787430943099918,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8615875244140625,
      "step": 2352
    },
    {
      "epoch": 0.19804256989820243,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.897202968597412,
      "step": 2354
    },
    {
      "epoch": 0.19821083036540565,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.877577543258667,
      "step": 2356
    },
    {
      "epoch": 0.19837909083260888,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.7930324077606201,
      "step": 2358
    },
    {
      "epoch": 0.1985473512998121,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8821184635162354,
      "step": 2360
    },
    {
      "epoch": 0.19871561176701533,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9251675605773926,
      "step": 2362
    },
    {
      "epoch": 0.19888387223421858,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8628315925598145,
      "step": 2364
    },
    {
      "epoch": 0.1990521327014218,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.829897403717041,
      "step": 2366
    },
    {
      "epoch": 0.19922039316862503,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8638097047805786,
      "step": 2368
    },
    {
      "epoch": 0.19938865363582825,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9052437543869019,
      "step": 2370
    },
    {
      "epoch": 0.1995569141030315,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.973907709121704,
      "step": 2372
    },
    {
      "epoch": 0.19972517457023473,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.882057785987854,
      "step": 2374
    },
    {
      "epoch": 0.19989343503743795,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9004693031311035,
      "step": 2376
    },
    {
      "epoch": 0.20006169550464117,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8788132667541504,
      "step": 2378
    },
    {
      "epoch": 0.20006169550464117,
      "eval_loss": 1.8531941175460815,
      "eval_runtime": 528.1678,
      "eval_samples_per_second": 105.389,
      "eval_steps_per_second": 13.174,
      "step": 2378
    },
    {
      "epoch": 0.20022995597184443,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8918235301971436,
      "step": 2380
    },
    {
      "epoch": 0.20039821643904765,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8499177694320679,
      "step": 2382
    },
    {
      "epoch": 0.20056647690625087,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8917226791381836,
      "step": 2384
    },
    {
      "epoch": 0.2007347373734541,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8606752157211304,
      "step": 2386
    },
    {
      "epoch": 0.20090299784065735,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8381330966949463,
      "step": 2388
    },
    {
      "epoch": 0.20107125830786057,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8698296546936035,
      "step": 2390
    },
    {
      "epoch": 0.2012395187750638,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9187374114990234,
      "step": 2392
    },
    {
      "epoch": 0.20140777924226702,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 1.8404881954193115,
      "step": 2394
    },
    {
      "epoch": 0.20157603970947027,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8539648056030273,
      "step": 2396
    },
    {
      "epoch": 0.2017443001766735,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.904233455657959,
      "step": 2398
    },
    {
      "epoch": 0.20191256064387672,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9030877351760864,
      "step": 2400
    },
    {
      "epoch": 0.20208082111107994,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9202420711517334,
      "step": 2402
    },
    {
      "epoch": 0.20224908157828317,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8934664726257324,
      "step": 2404
    },
    {
      "epoch": 0.20241734204548642,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8463613986968994,
      "step": 2406
    },
    {
      "epoch": 0.20258560251268964,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8378918170928955,
      "step": 2408
    },
    {
      "epoch": 0.20275386297989287,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8759517669677734,
      "step": 2410
    },
    {
      "epoch": 0.2029221234470961,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.864206075668335,
      "step": 2412
    },
    {
      "epoch": 0.20309038391429934,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.917436122894287,
      "step": 2414
    },
    {
      "epoch": 0.20325864438150257,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9038970470428467,
      "step": 2416
    },
    {
      "epoch": 0.2034269048487058,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.910412311553955,
      "step": 2418
    },
    {
      "epoch": 0.20359516531590902,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9099743366241455,
      "step": 2420
    },
    {
      "epoch": 0.20376342578311227,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8752193450927734,
      "step": 2422
    },
    {
      "epoch": 0.2039316862503155,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.897463083267212,
      "step": 2424
    },
    {
      "epoch": 0.20409994671751872,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8666105270385742,
      "step": 2426
    },
    {
      "epoch": 0.20426820718472194,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.884295105934143,
      "step": 2428
    },
    {
      "epoch": 0.2044364676519252,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8791651725769043,
      "step": 2430
    },
    {
      "epoch": 0.20460472811912842,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8768031597137451,
      "step": 2432
    },
    {
      "epoch": 0.20477298858633164,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8736836910247803,
      "step": 2434
    },
    {
      "epoch": 0.20494124905353486,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8971792459487915,
      "step": 2436
    },
    {
      "epoch": 0.20510950952073811,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9268066883087158,
      "step": 2438
    },
    {
      "epoch": 0.20527776998794134,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8837859630584717,
      "step": 2440
    },
    {
      "epoch": 0.20544603045514456,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9074170589447021,
      "step": 2442
    },
    {
      "epoch": 0.2056142909223478,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8796582221984863,
      "step": 2444
    },
    {
      "epoch": 0.20578255138955104,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8575907945632935,
      "step": 2446
    },
    {
      "epoch": 0.20595081185675426,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8895928859710693,
      "step": 2448
    },
    {
      "epoch": 0.2061190723239575,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8465087413787842,
      "step": 2450
    },
    {
      "epoch": 0.2062873327911607,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9146661758422852,
      "step": 2452
    },
    {
      "epoch": 0.20645559325836393,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9340906143188477,
      "step": 2454
    },
    {
      "epoch": 0.20662385372556719,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9148595333099365,
      "step": 2456
    },
    {
      "epoch": 0.2067921141927704,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9054845571517944,
      "step": 2458
    },
    {
      "epoch": 0.20696037465997363,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.851936936378479,
      "step": 2460
    },
    {
      "epoch": 0.20712863512717686,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.860978126525879,
      "step": 2462
    },
    {
      "epoch": 0.2072968955943801,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8887393474578857,
      "step": 2464
    },
    {
      "epoch": 0.20746515606158333,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8480037450790405,
      "step": 2466
    },
    {
      "epoch": 0.20763341652878656,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8305096626281738,
      "step": 2468
    },
    {
      "epoch": 0.20780167699598978,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8986009359359741,
      "step": 2470
    },
    {
      "epoch": 0.20796993746319303,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8833378553390503,
      "step": 2472
    },
    {
      "epoch": 0.20813819793039626,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9136226177215576,
      "step": 2474
    },
    {
      "epoch": 0.20830645839759948,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.886906623840332,
      "step": 2476
    },
    {
      "epoch": 0.2084747188648027,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8353731632232666,
      "step": 2478
    },
    {
      "epoch": 0.20864297933200596,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.872154951095581,
      "step": 2480
    },
    {
      "epoch": 0.20881123979920918,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8802354335784912,
      "step": 2482
    },
    {
      "epoch": 0.2089795002664124,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8644089698791504,
      "step": 2484
    },
    {
      "epoch": 0.20914776073361563,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.944056749343872,
      "step": 2486
    },
    {
      "epoch": 0.20931602120081888,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8672547340393066,
      "step": 2488
    },
    {
      "epoch": 0.2094842816680221,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8306796550750732,
      "step": 2490
    },
    {
      "epoch": 0.20965254213522533,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9031376838684082,
      "step": 2492
    },
    {
      "epoch": 0.20982080260242855,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8229275941848755,
      "step": 2494
    },
    {
      "epoch": 0.20998906306963178,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8886736631393433,
      "step": 2496
    },
    {
      "epoch": 0.21015732353683503,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8214788436889648,
      "step": 2498
    },
    {
      "epoch": 0.21032558400403825,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8972009420394897,
      "step": 2500
    },
    {
      "epoch": 0.21049384447124148,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8813490867614746,
      "step": 2502
    },
    {
      "epoch": 0.2106621049384447,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.9082911014556885,
      "step": 2504
    },
    {
      "epoch": 0.21083036540564795,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8787038326263428,
      "step": 2506
    },
    {
      "epoch": 0.21099862587285118,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8613176345825195,
      "step": 2508
    },
    {
      "epoch": 0.2111668863400544,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8973994255065918,
      "step": 2510
    },
    {
      "epoch": 0.21133514680725762,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8962352275848389,
      "step": 2512
    },
    {
      "epoch": 0.21150340727446088,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8961138725280762,
      "step": 2514
    },
    {
      "epoch": 0.2116716677416641,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.819783091545105,
      "step": 2516
    },
    {
      "epoch": 0.21183992820886732,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9153187274932861,
      "step": 2518
    },
    {
      "epoch": 0.21200818867607055,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8984448909759521,
      "step": 2520
    },
    {
      "epoch": 0.2121764491432738,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9200396537780762,
      "step": 2522
    },
    {
      "epoch": 0.21234470961047702,
      "grad_norm": 0.25,
      "learning_rate": 0.0003,
      "loss": 1.9178752899169922,
      "step": 2524
    },
    {
      "epoch": 0.21251297007768025,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.9199483394622803,
      "step": 2526
    },
    {
      "epoch": 0.21268123054488347,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.8363702297210693,
      "step": 2528
    },
    {
      "epoch": 0.21284949101208672,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.8431811332702637,
      "step": 2530
    },
    {
      "epoch": 0.21301775147928995,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8965916633605957,
      "step": 2532
    },
    {
      "epoch": 0.21318601194649317,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8883694410324097,
      "step": 2534
    },
    {
      "epoch": 0.2133542724136964,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8859868049621582,
      "step": 2536
    },
    {
      "epoch": 0.21352253288089965,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8759138584136963,
      "step": 2538
    },
    {
      "epoch": 0.21369079334810287,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.880584716796875,
      "step": 2540
    },
    {
      "epoch": 0.2138590538153061,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8553612232208252,
      "step": 2542
    },
    {
      "epoch": 0.21402731428250932,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.884256362915039,
      "step": 2544
    },
    {
      "epoch": 0.21419557474971254,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8671263456344604,
      "step": 2546
    },
    {
      "epoch": 0.2143638352169158,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8226943016052246,
      "step": 2548
    },
    {
      "epoch": 0.21453209568411902,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8670274019241333,
      "step": 2550
    },
    {
      "epoch": 0.21470035615132224,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9221431016921997,
      "step": 2552
    },
    {
      "epoch": 0.21486861661852547,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8924552202224731,
      "step": 2554
    },
    {
      "epoch": 0.21503687708572872,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003,
      "loss": 1.875789999961853,
      "step": 2556
    },
    {
      "epoch": 0.21520513755293194,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8916780948638916,
      "step": 2558
    },
    {
      "epoch": 0.21537339802013516,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.917212963104248,
      "step": 2560
    },
    {
      "epoch": 0.2155416584873384,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.900407314300537,
      "step": 2562
    },
    {
      "epoch": 0.21570991895454164,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9050205945968628,
      "step": 2564
    },
    {
      "epoch": 0.21587817942174486,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8673715591430664,
      "step": 2566
    },
    {
      "epoch": 0.2160464398889481,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8719139099121094,
      "step": 2568
    },
    {
      "epoch": 0.2162147003561513,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9558584690093994,
      "step": 2570
    },
    {
      "epoch": 0.21638296082335456,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8693523406982422,
      "step": 2572
    },
    {
      "epoch": 0.2165512212905578,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8622217178344727,
      "step": 2574
    },
    {
      "epoch": 0.216719481757761,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9724653959274292,
      "step": 2576
    },
    {
      "epoch": 0.21688774222496424,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9517152309417725,
      "step": 2578
    },
    {
      "epoch": 0.2170560026921675,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8758957386016846,
      "step": 2580
    },
    {
      "epoch": 0.2172242631593707,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8867919445037842,
      "step": 2582
    },
    {
      "epoch": 0.21739252362657394,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.903228759765625,
      "step": 2584
    },
    {
      "epoch": 0.21756078409377716,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8501629829406738,
      "step": 2586
    },
    {
      "epoch": 0.2177290445609804,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8242254257202148,
      "step": 2588
    },
    {
      "epoch": 0.21789730502818364,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8688526153564453,
      "step": 2590
    },
    {
      "epoch": 0.21806556549538686,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8711903095245361,
      "step": 2592
    },
    {
      "epoch": 0.21823382596259008,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9110751152038574,
      "step": 2594
    },
    {
      "epoch": 0.2184020864297933,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9284318685531616,
      "step": 2596
    },
    {
      "epoch": 0.21857034689699656,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8443536758422852,
      "step": 2598
    },
    {
      "epoch": 0.21873860736419978,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8873469829559326,
      "step": 2600
    },
    {
      "epoch": 0.218906867831403,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8445956707000732,
      "step": 2602
    },
    {
      "epoch": 0.21907512829860623,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9095325469970703,
      "step": 2604
    },
    {
      "epoch": 0.21924338876580948,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8635461330413818,
      "step": 2606
    },
    {
      "epoch": 0.2194116492330127,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.854446291923523,
      "step": 2608
    },
    {
      "epoch": 0.21957990970021593,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8150538206100464,
      "step": 2610
    },
    {
      "epoch": 0.21974817016741915,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.909688949584961,
      "step": 2612
    },
    {
      "epoch": 0.2199164306346224,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9274461269378662,
      "step": 2614
    },
    {
      "epoch": 0.22008469110182563,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8744330406188965,
      "step": 2616
    },
    {
      "epoch": 0.22025295156902885,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.898446798324585,
      "step": 2618
    },
    {
      "epoch": 0.22042121203623208,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8568707704544067,
      "step": 2620
    },
    {
      "epoch": 0.22058947250343533,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8874421119689941,
      "step": 2622
    },
    {
      "epoch": 0.22075773297063855,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8698123693466187,
      "step": 2624
    },
    {
      "epoch": 0.22092599343784178,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8629450798034668,
      "step": 2626
    },
    {
      "epoch": 0.221094253905045,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8815085887908936,
      "step": 2628
    },
    {
      "epoch": 0.22126251437224825,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.885282039642334,
      "step": 2630
    },
    {
      "epoch": 0.22143077483945148,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9244436025619507,
      "step": 2632
    },
    {
      "epoch": 0.2215990353066547,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8511152267456055,
      "step": 2634
    },
    {
      "epoch": 0.22176729577385793,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8901498317718506,
      "step": 2636
    },
    {
      "epoch": 0.22193555624106115,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8747613430023193,
      "step": 2638
    },
    {
      "epoch": 0.2221038167082644,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0003,
      "loss": 1.8894016742706299,
      "step": 2640
    },
    {
      "epoch": 0.22227207717546762,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0003,
      "loss": 1.9096343517303467,
      "step": 2642
    },
    {
      "epoch": 0.22244033764267085,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9244468212127686,
      "step": 2644
    },
    {
      "epoch": 0.22260859810987407,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.854464054107666,
      "step": 2646
    },
    {
      "epoch": 0.22277685857707732,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9055533409118652,
      "step": 2648
    },
    {
      "epoch": 0.22294511904428055,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8844306468963623,
      "step": 2650
    },
    {
      "epoch": 0.22311337951148377,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8463428020477295,
      "step": 2652
    },
    {
      "epoch": 0.223281639978687,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9283151626586914,
      "step": 2654
    },
    {
      "epoch": 0.22344990044589025,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8822001218795776,
      "step": 2656
    },
    {
      "epoch": 0.22361816091309347,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8863521814346313,
      "step": 2658
    },
    {
      "epoch": 0.2237864213802967,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9059209823608398,
      "step": 2660
    },
    {
      "epoch": 0.22395468184749992,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8609366416931152,
      "step": 2662
    },
    {
      "epoch": 0.22412294231470317,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.922555685043335,
      "step": 2664
    },
    {
      "epoch": 0.2242912027819064,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.890061855316162,
      "step": 2666
    },
    {
      "epoch": 0.22445946324910962,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8420352935791016,
      "step": 2668
    },
    {
      "epoch": 0.22462772371631284,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8868992328643799,
      "step": 2670
    },
    {
      "epoch": 0.2247959841835161,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.868120551109314,
      "step": 2672
    },
    {
      "epoch": 0.22496424465071932,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9200103282928467,
      "step": 2674
    },
    {
      "epoch": 0.22513250511792254,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8398784399032593,
      "step": 2676
    },
    {
      "epoch": 0.22530076558512577,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8783336877822876,
      "step": 2678
    },
    {
      "epoch": 0.22546902605232902,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.929810643196106,
      "step": 2680
    },
    {
      "epoch": 0.22563728651953224,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.920559287071228,
      "step": 2682
    },
    {
      "epoch": 0.22580554698673547,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8799467086791992,
      "step": 2684
    },
    {
      "epoch": 0.2259738074539387,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.867621898651123,
      "step": 2686
    },
    {
      "epoch": 0.22614206792114191,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 1.9163930416107178,
      "step": 2688
    },
    {
      "epoch": 0.22631032838834517,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8321728706359863,
      "step": 2690
    },
    {
      "epoch": 0.2264785888555484,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8534255027770996,
      "step": 2692
    },
    {
      "epoch": 0.22664684932275161,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9043371677398682,
      "step": 2694
    },
    {
      "epoch": 0.22681510978995484,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8924858570098877,
      "step": 2696
    },
    {
      "epoch": 0.2269833702571581,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9009597301483154,
      "step": 2698
    },
    {
      "epoch": 0.22715163072436131,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8623430728912354,
      "step": 2700
    },
    {
      "epoch": 0.22731989119156454,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.908335566520691,
      "step": 2702
    },
    {
      "epoch": 0.22748815165876776,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8354105949401855,
      "step": 2704
    },
    {
      "epoch": 0.227656412125971,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9163963794708252,
      "step": 2706
    },
    {
      "epoch": 0.22782467259317424,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8979425430297852,
      "step": 2708
    },
    {
      "epoch": 0.22799293306037746,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.855201244354248,
      "step": 2710
    },
    {
      "epoch": 0.22816119352758069,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8985729217529297,
      "step": 2712
    },
    {
      "epoch": 0.22832945399478394,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.903731346130371,
      "step": 2714
    },
    {
      "epoch": 0.22849771446198716,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8999871015548706,
      "step": 2716
    },
    {
      "epoch": 0.22866597492919039,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8915799856185913,
      "step": 2718
    },
    {
      "epoch": 0.2288342353963936,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8475620746612549,
      "step": 2720
    },
    {
      "epoch": 0.22900249586359686,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.869617223739624,
      "step": 2722
    },
    {
      "epoch": 0.22917075633080009,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.859818935394287,
      "step": 2724
    },
    {
      "epoch": 0.2293390167980033,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8834078311920166,
      "step": 2726
    },
    {
      "epoch": 0.22950727726520653,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.884641170501709,
      "step": 2728
    },
    {
      "epoch": 0.22967553773240976,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8673315048217773,
      "step": 2730
    },
    {
      "epoch": 0.229843798199613,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8946144580841064,
      "step": 2732
    },
    {
      "epoch": 0.23001205866681623,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8261878490447998,
      "step": 2734
    },
    {
      "epoch": 0.23018031913401946,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.921569585800171,
      "step": 2736
    },
    {
      "epoch": 0.23034857960122268,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.862008810043335,
      "step": 2738
    },
    {
      "epoch": 0.23051684006842593,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8787444829940796,
      "step": 2740
    },
    {
      "epoch": 0.23068510053562916,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8387501239776611,
      "step": 2742
    },
    {
      "epoch": 0.23085336100283238,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.7958323955535889,
      "step": 2744
    },
    {
      "epoch": 0.2310216214700356,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9352949857711792,
      "step": 2746
    },
    {
      "epoch": 0.23118988193723886,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8495659828186035,
      "step": 2748
    },
    {
      "epoch": 0.23135814240444208,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8554437160491943,
      "step": 2750
    },
    {
      "epoch": 0.2315264028716453,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8357136249542236,
      "step": 2752
    },
    {
      "epoch": 0.23169466333884853,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8778045177459717,
      "step": 2754
    },
    {
      "epoch": 0.23186292380605178,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8484947681427002,
      "step": 2756
    },
    {
      "epoch": 0.232031184273255,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9095567464828491,
      "step": 2758
    },
    {
      "epoch": 0.23219944474045823,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8742036819458008,
      "step": 2760
    },
    {
      "epoch": 0.23236770520766145,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.9174658060073853,
      "step": 2762
    },
    {
      "epoch": 0.2325359656748647,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8438241481781006,
      "step": 2764
    },
    {
      "epoch": 0.23270422614206793,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8112657070159912,
      "step": 2766
    },
    {
      "epoch": 0.23287248660927115,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.92279052734375,
      "step": 2768
    },
    {
      "epoch": 0.23304074707647437,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8618665933609009,
      "step": 2770
    },
    {
      "epoch": 0.23320900754367763,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8351373672485352,
      "step": 2772
    },
    {
      "epoch": 0.23337726801088085,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.875737190246582,
      "step": 2774
    },
    {
      "epoch": 0.23354552847808407,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8807313442230225,
      "step": 2776
    },
    {
      "epoch": 0.2337137889452873,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.868659257888794,
      "step": 2778
    },
    {
      "epoch": 0.23388204941249052,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8463995456695557,
      "step": 2780
    },
    {
      "epoch": 0.23405030987969377,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8386752605438232,
      "step": 2782
    },
    {
      "epoch": 0.234218570346897,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8733923435211182,
      "step": 2784
    },
    {
      "epoch": 0.23438683081410022,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9510694742202759,
      "step": 2786
    },
    {
      "epoch": 0.23455509128130345,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.873216152191162,
      "step": 2788
    },
    {
      "epoch": 0.2347233517485067,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.862964153289795,
      "step": 2790
    },
    {
      "epoch": 0.23489161221570992,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8892133235931396,
      "step": 2792
    },
    {
      "epoch": 0.23505987268291315,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8484368324279785,
      "step": 2794
    },
    {
      "epoch": 0.23522813315011637,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8900601863861084,
      "step": 2796
    },
    {
      "epoch": 0.23539639361731962,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8515071868896484,
      "step": 2798
    },
    {
      "epoch": 0.23556465408452285,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8898144960403442,
      "step": 2800
    },
    {
      "epoch": 0.23573291455172607,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8621717691421509,
      "step": 2802
    },
    {
      "epoch": 0.2359011750189293,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8765792846679688,
      "step": 2804
    },
    {
      "epoch": 0.23606943548613255,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8890440464019775,
      "step": 2806
    },
    {
      "epoch": 0.23623769595333577,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.887814998626709,
      "step": 2808
    },
    {
      "epoch": 0.236405956420539,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.9142069816589355,
      "step": 2810
    },
    {
      "epoch": 0.23657421688774222,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8868740797042847,
      "step": 2812
    },
    {
      "epoch": 0.23674247735494547,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.869403600692749,
      "step": 2814
    },
    {
      "epoch": 0.2369107378221487,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.874135971069336,
      "step": 2816
    },
    {
      "epoch": 0.23707899828935192,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9072186946868896,
      "step": 2818
    },
    {
      "epoch": 0.23724725875655514,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.897245168685913,
      "step": 2820
    },
    {
      "epoch": 0.23741551922375836,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8887101411819458,
      "step": 2822
    },
    {
      "epoch": 0.23758377969096162,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8877075910568237,
      "step": 2824
    },
    {
      "epoch": 0.23775204015816484,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8908970355987549,
      "step": 2826
    },
    {
      "epoch": 0.23792030062536806,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8831470012664795,
      "step": 2828
    },
    {
      "epoch": 0.2380885610925713,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8647334575653076,
      "step": 2830
    },
    {
      "epoch": 0.23825682155977454,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8922103643417358,
      "step": 2832
    },
    {
      "epoch": 0.23842508202697776,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.877692461013794,
      "step": 2834
    },
    {
      "epoch": 0.238593342494181,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9526077508926392,
      "step": 2836
    },
    {
      "epoch": 0.2387616029613842,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9217331409454346,
      "step": 2838
    },
    {
      "epoch": 0.23892986342858746,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9033727645874023,
      "step": 2840
    },
    {
      "epoch": 0.2390981238957907,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8973650932312012,
      "step": 2842
    },
    {
      "epoch": 0.2392663843629939,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8908278942108154,
      "step": 2844
    },
    {
      "epoch": 0.23943464483019714,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8962745666503906,
      "step": 2846
    },
    {
      "epoch": 0.2396029052974004,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.86026930809021,
      "step": 2848
    },
    {
      "epoch": 0.2397711657646036,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8230162858963013,
      "step": 2850
    },
    {
      "epoch": 0.23993942623180683,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8318099975585938,
      "step": 2852
    },
    {
      "epoch": 0.24010768669901006,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8450207710266113,
      "step": 2854
    },
    {
      "epoch": 0.2402759471662133,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9057466983795166,
      "step": 2856
    },
    {
      "epoch": 0.24044420763341653,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8587586879730225,
      "step": 2858
    },
    {
      "epoch": 0.24061246810061976,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.887007474899292,
      "step": 2860
    },
    {
      "epoch": 0.24078072856782298,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9042458534240723,
      "step": 2862
    },
    {
      "epoch": 0.24094898903502623,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8696728944778442,
      "step": 2864
    },
    {
      "epoch": 0.24111724950222946,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8663671016693115,
      "step": 2866
    },
    {
      "epoch": 0.24128550996943268,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9468204975128174,
      "step": 2868
    },
    {
      "epoch": 0.2414537704366359,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8843989372253418,
      "step": 2870
    },
    {
      "epoch": 0.24162203090383913,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9151825904846191,
      "step": 2872
    },
    {
      "epoch": 0.24179029137104238,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8819860219955444,
      "step": 2874
    },
    {
      "epoch": 0.2419585518382456,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.890951156616211,
      "step": 2876
    },
    {
      "epoch": 0.24212681230544883,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8578972816467285,
      "step": 2878
    },
    {
      "epoch": 0.24229507277265205,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.863383412361145,
      "step": 2880
    },
    {
      "epoch": 0.2424633332398553,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003,
      "loss": 1.882206678390503,
      "step": 2882
    },
    {
      "epoch": 0.24263159370705853,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9058516025543213,
      "step": 2884
    },
    {
      "epoch": 0.24279985417426175,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9052767753601074,
      "step": 2886
    },
    {
      "epoch": 0.24296811464146498,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9215350151062012,
      "step": 2888
    },
    {
      "epoch": 0.24313637510866823,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9140534400939941,
      "step": 2890
    },
    {
      "epoch": 0.24330463557587145,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8708794116973877,
      "step": 2892
    },
    {
      "epoch": 0.24347289604307468,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8733689785003662,
      "step": 2894
    },
    {
      "epoch": 0.2436411565102779,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8773137331008911,
      "step": 2896
    },
    {
      "epoch": 0.24380941697748115,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8779692649841309,
      "step": 2898
    },
    {
      "epoch": 0.24397767744468438,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9115407466888428,
      "step": 2900
    },
    {
      "epoch": 0.2441459379118876,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8559805154800415,
      "step": 2902
    },
    {
      "epoch": 0.24431419837909082,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.951850414276123,
      "step": 2904
    },
    {
      "epoch": 0.24448245884629408,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.846613883972168,
      "step": 2906
    },
    {
      "epoch": 0.2446507193134973,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.852430820465088,
      "step": 2908
    },
    {
      "epoch": 0.24481897978070052,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9384982585906982,
      "step": 2910
    },
    {
      "epoch": 0.24498724024790375,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8514857292175293,
      "step": 2912
    },
    {
      "epoch": 0.24515550071510697,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8159314393997192,
      "step": 2914
    },
    {
      "epoch": 0.24532376118231022,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.901989221572876,
      "step": 2916
    },
    {
      "epoch": 0.24549202164951345,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8628402948379517,
      "step": 2918
    },
    {
      "epoch": 0.24566028211671667,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.910447597503662,
      "step": 2920
    },
    {
      "epoch": 0.2458285425839199,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.87015700340271,
      "step": 2922
    },
    {
      "epoch": 0.24599680305112315,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8233139514923096,
      "step": 2924
    },
    {
      "epoch": 0.24616506351832637,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.899294137954712,
      "step": 2926
    },
    {
      "epoch": 0.2463333239855296,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8736231327056885,
      "step": 2928
    },
    {
      "epoch": 0.24650158445273282,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9222290515899658,
      "step": 2930
    },
    {
      "epoch": 0.24666984491993607,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9276052713394165,
      "step": 2932
    },
    {
      "epoch": 0.2468381053871393,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8974111080169678,
      "step": 2934
    },
    {
      "epoch": 0.24700636585434252,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.848832368850708,
      "step": 2936
    },
    {
      "epoch": 0.24717462632154574,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8821473121643066,
      "step": 2938
    },
    {
      "epoch": 0.247342886788749,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.871675968170166,
      "step": 2940
    },
    {
      "epoch": 0.24751114725595222,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8890318870544434,
      "step": 2942
    },
    {
      "epoch": 0.24767940772315544,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9442709684371948,
      "step": 2944
    },
    {
      "epoch": 0.24784766819035867,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8822420835494995,
      "step": 2946
    },
    {
      "epoch": 0.24801592865756192,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8950763940811157,
      "step": 2948
    },
    {
      "epoch": 0.24818418912476514,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8755054473876953,
      "step": 2950
    },
    {
      "epoch": 0.24835244959196837,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8640451431274414,
      "step": 2952
    },
    {
      "epoch": 0.2485207100591716,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8239399194717407,
      "step": 2954
    },
    {
      "epoch": 0.24868897052637484,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9086270332336426,
      "step": 2956
    },
    {
      "epoch": 0.24885723099357807,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.869236946105957,
      "step": 2958
    },
    {
      "epoch": 0.2490254914607813,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8828363418579102,
      "step": 2960
    },
    {
      "epoch": 0.2491937519279845,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9156357049942017,
      "step": 2962
    },
    {
      "epoch": 0.24936201239518774,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8568023443222046,
      "step": 2964
    },
    {
      "epoch": 0.249530272862391,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8872981071472168,
      "step": 2966
    },
    {
      "epoch": 0.2496985333295942,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8789199590682983,
      "step": 2968
    },
    {
      "epoch": 0.24986679379679744,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8681443929672241,
      "step": 2970
    },
    {
      "epoch": 0.2500350542640007,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9170730113983154,
      "step": 2972
    },
    {
      "epoch": 0.2502033147312039,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8939733505249023,
      "step": 2974
    },
    {
      "epoch": 0.25037157519840714,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8847920894622803,
      "step": 2976
    },
    {
      "epoch": 0.2505398356656104,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8382043838500977,
      "step": 2978
    },
    {
      "epoch": 0.2507080961328136,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8245179653167725,
      "step": 2980
    },
    {
      "epoch": 0.25087635660001684,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9202015399932861,
      "step": 2982
    },
    {
      "epoch": 0.25104461706722003,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9102747440338135,
      "step": 2984
    },
    {
      "epoch": 0.2512128775344233,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9101377725601196,
      "step": 2986
    },
    {
      "epoch": 0.25138113800162654,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9300272464752197,
      "step": 2988
    },
    {
      "epoch": 0.25154939846882973,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.881932258605957,
      "step": 2990
    },
    {
      "epoch": 0.251717658936033,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.887772798538208,
      "step": 2992
    },
    {
      "epoch": 0.25188591940323624,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8845584392547607,
      "step": 2994
    },
    {
      "epoch": 0.25205417987043943,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8950356245040894,
      "step": 2996
    },
    {
      "epoch": 0.2522224403376427,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8592448234558105,
      "step": 2998
    },
    {
      "epoch": 0.2523907008048459,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8811382055282593,
      "step": 3000
    },
    {
      "epoch": 0.25255896127204913,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8729259967803955,
      "step": 3002
    },
    {
      "epoch": 0.2527272217392524,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8638328313827515,
      "step": 3004
    },
    {
      "epoch": 0.2528954822064556,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8488091230392456,
      "step": 3006
    },
    {
      "epoch": 0.25306374267365883,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8992512226104736,
      "step": 3008
    },
    {
      "epoch": 0.253232003140862,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8789143562316895,
      "step": 3010
    },
    {
      "epoch": 0.2534002636080653,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8924612998962402,
      "step": 3012
    },
    {
      "epoch": 0.25356852407526853,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8977869749069214,
      "step": 3014
    },
    {
      "epoch": 0.2537367845424717,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8581494092941284,
      "step": 3016
    },
    {
      "epoch": 0.253905045009675,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8332231044769287,
      "step": 3018
    },
    {
      "epoch": 0.25407330547687823,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8407278060913086,
      "step": 3020
    },
    {
      "epoch": 0.2542415659440814,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8560960292816162,
      "step": 3022
    },
    {
      "epoch": 0.2544098264112847,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8650498390197754,
      "step": 3024
    },
    {
      "epoch": 0.2545780868784879,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8348109722137451,
      "step": 3026
    },
    {
      "epoch": 0.2547463473456911,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.895792841911316,
      "step": 3028
    },
    {
      "epoch": 0.2549146078128944,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8987611532211304,
      "step": 3030
    },
    {
      "epoch": 0.2550828682800976,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8227717876434326,
      "step": 3032
    },
    {
      "epoch": 0.2552511287473008,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8419108390808105,
      "step": 3034
    },
    {
      "epoch": 0.2554193892145041,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8465135097503662,
      "step": 3036
    },
    {
      "epoch": 0.2555876496817073,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9202295541763306,
      "step": 3038
    },
    {
      "epoch": 0.2557559101489105,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.8657450675964355,
      "step": 3040
    },
    {
      "epoch": 0.2559241706161137,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8644499778747559,
      "step": 3042
    },
    {
      "epoch": 0.256092431083317,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9136632680892944,
      "step": 3044
    },
    {
      "epoch": 0.2562606915505202,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8712952136993408,
      "step": 3046
    },
    {
      "epoch": 0.2564289520177234,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8553709983825684,
      "step": 3048
    },
    {
      "epoch": 0.2565972124849267,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8290854692459106,
      "step": 3050
    },
    {
      "epoch": 0.25676547295212987,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8972852230072021,
      "step": 3052
    },
    {
      "epoch": 0.2569337334193331,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.835843563079834,
      "step": 3054
    },
    {
      "epoch": 0.2571019938865364,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9165441989898682,
      "step": 3056
    },
    {
      "epoch": 0.25727025435373957,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8561158180236816,
      "step": 3058
    },
    {
      "epoch": 0.2574385148209428,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9061245918273926,
      "step": 3060
    },
    {
      "epoch": 0.2576067752881461,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8196547031402588,
      "step": 3062
    },
    {
      "epoch": 0.25777503575534927,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.921690821647644,
      "step": 3064
    },
    {
      "epoch": 0.2579432962225525,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8270201683044434,
      "step": 3066
    },
    {
      "epoch": 0.2581115566897557,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8520450592041016,
      "step": 3068
    },
    {
      "epoch": 0.25827981715695897,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8905645608901978,
      "step": 3070
    },
    {
      "epoch": 0.2584480776241622,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.829707145690918,
      "step": 3072
    },
    {
      "epoch": 0.2586163380913654,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8955073356628418,
      "step": 3074
    },
    {
      "epoch": 0.25878459855856867,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8317666053771973,
      "step": 3076
    },
    {
      "epoch": 0.2589528590257719,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8672267198562622,
      "step": 3078
    },
    {
      "epoch": 0.2591211194929751,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9324471950531006,
      "step": 3080
    },
    {
      "epoch": 0.25928937996017837,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8641631603240967,
      "step": 3082
    },
    {
      "epoch": 0.25945764042738156,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.844289779663086,
      "step": 3084
    },
    {
      "epoch": 0.2596259008945848,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8540849685668945,
      "step": 3086
    },
    {
      "epoch": 0.25979416136178807,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9056406021118164,
      "step": 3088
    },
    {
      "epoch": 0.25996242182899126,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.861724615097046,
      "step": 3090
    },
    {
      "epoch": 0.2601306822961945,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8412243127822876,
      "step": 3092
    },
    {
      "epoch": 0.26029894276339777,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.90041184425354,
      "step": 3094
    },
    {
      "epoch": 0.26046720323060096,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8837145566940308,
      "step": 3096
    },
    {
      "epoch": 0.2606354636978042,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.875992774963379,
      "step": 3098
    },
    {
      "epoch": 0.2608037241650074,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9054152965545654,
      "step": 3100
    },
    {
      "epoch": 0.26097198463221066,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8423762321472168,
      "step": 3102
    },
    {
      "epoch": 0.2611402450994139,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8750646114349365,
      "step": 3104
    },
    {
      "epoch": 0.2613085055666171,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8785638809204102,
      "step": 3106
    },
    {
      "epoch": 0.26147676603382036,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8777384757995605,
      "step": 3108
    },
    {
      "epoch": 0.26164502650102356,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8671660423278809,
      "step": 3110
    },
    {
      "epoch": 0.2618132869682268,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8972692489624023,
      "step": 3112
    },
    {
      "epoch": 0.26198154743543006,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8981778621673584,
      "step": 3114
    },
    {
      "epoch": 0.26214980790263326,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0003,
      "loss": 1.9187886714935303,
      "step": 3116
    },
    {
      "epoch": 0.2623180683698365,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8370201587677002,
      "step": 3118
    },
    {
      "epoch": 0.26248632883703976,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.8727036714553833,
      "step": 3120
    },
    {
      "epoch": 0.26265458930424296,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.895568609237671,
      "step": 3122
    },
    {
      "epoch": 0.2628228497714462,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8322652578353882,
      "step": 3124
    },
    {
      "epoch": 0.2629911102386494,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.860079050064087,
      "step": 3126
    },
    {
      "epoch": 0.26315937070585266,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8549771308898926,
      "step": 3128
    },
    {
      "epoch": 0.2633276311730559,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8789892196655273,
      "step": 3130
    },
    {
      "epoch": 0.2634958916402591,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.857250452041626,
      "step": 3132
    },
    {
      "epoch": 0.26366415210746236,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8859878778457642,
      "step": 3134
    },
    {
      "epoch": 0.2638324125746656,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.893284797668457,
      "step": 3136
    },
    {
      "epoch": 0.2640006730418688,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8644254207611084,
      "step": 3138
    },
    {
      "epoch": 0.26416893350907206,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8888039588928223,
      "step": 3140
    },
    {
      "epoch": 0.26433719397627525,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8689630031585693,
      "step": 3142
    },
    {
      "epoch": 0.2645054544434785,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8317666053771973,
      "step": 3144
    },
    {
      "epoch": 0.26467371491068176,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8729615211486816,
      "step": 3146
    },
    {
      "epoch": 0.26484197537788495,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.918393611907959,
      "step": 3148
    },
    {
      "epoch": 0.2650102358450882,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.881615400314331,
      "step": 3150
    },
    {
      "epoch": 0.2651784963122914,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8179441690444946,
      "step": 3152
    },
    {
      "epoch": 0.26534675677949465,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8298217058181763,
      "step": 3154
    },
    {
      "epoch": 0.2655150172466979,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8943259716033936,
      "step": 3156
    },
    {
      "epoch": 0.2656832777139011,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8354926109313965,
      "step": 3158
    },
    {
      "epoch": 0.26585153818110435,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9133589267730713,
      "step": 3160
    },
    {
      "epoch": 0.2660197986483076,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8794219493865967,
      "step": 3162
    },
    {
      "epoch": 0.2661880591155108,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.860735297203064,
      "step": 3164
    },
    {
      "epoch": 0.26635631958271405,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.876822829246521,
      "step": 3166
    },
    {
      "epoch": 0.26652458004991725,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8857851028442383,
      "step": 3168
    },
    {
      "epoch": 0.2666928405171205,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.90088951587677,
      "step": 3170
    },
    {
      "epoch": 0.26686110098432375,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8580515384674072,
      "step": 3172
    },
    {
      "epoch": 0.26702936145152695,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.851397156715393,
      "step": 3174
    },
    {
      "epoch": 0.2671976219187302,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8405373096466064,
      "step": 3176
    },
    {
      "epoch": 0.26736588238593345,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8357053995132446,
      "step": 3178
    },
    {
      "epoch": 0.26753414285313665,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8990567922592163,
      "step": 3180
    },
    {
      "epoch": 0.2677024033203399,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.844382643699646,
      "step": 3182
    },
    {
      "epoch": 0.2678706637875431,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8173645734786987,
      "step": 3184
    },
    {
      "epoch": 0.26803892425474635,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8650732040405273,
      "step": 3186
    },
    {
      "epoch": 0.2682071847219496,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8596446514129639,
      "step": 3188
    },
    {
      "epoch": 0.2683754451891528,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8669238090515137,
      "step": 3190
    },
    {
      "epoch": 0.26854370565635605,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.900693655014038,
      "step": 3192
    },
    {
      "epoch": 0.26871196612355924,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9254920482635498,
      "step": 3194
    },
    {
      "epoch": 0.2688802265907625,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.843451738357544,
      "step": 3196
    },
    {
      "epoch": 0.26904848705796575,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8634321689605713,
      "step": 3198
    },
    {
      "epoch": 0.26921674752516894,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.832934021949768,
      "step": 3200
    },
    {
      "epoch": 0.2693850079923722,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8933981657028198,
      "step": 3202
    },
    {
      "epoch": 0.26955326845957545,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8872950077056885,
      "step": 3204
    },
    {
      "epoch": 0.26972152892677864,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8306140899658203,
      "step": 3206
    },
    {
      "epoch": 0.2698897893939819,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.7875301837921143,
      "step": 3208
    },
    {
      "epoch": 0.2700580498611851,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.9035704135894775,
      "step": 3210
    },
    {
      "epoch": 0.27022631032838834,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.837846279144287,
      "step": 3212
    },
    {
      "epoch": 0.2703945707955916,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8785748481750488,
      "step": 3214
    },
    {
      "epoch": 0.2705628312627948,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8619394302368164,
      "step": 3216
    },
    {
      "epoch": 0.27073109172999804,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8550726175308228,
      "step": 3218
    },
    {
      "epoch": 0.2708993521972013,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.891996145248413,
      "step": 3220
    },
    {
      "epoch": 0.2710676126644045,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.906307578086853,
      "step": 3222
    },
    {
      "epoch": 0.27123587313160774,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.897355556488037,
      "step": 3224
    },
    {
      "epoch": 0.27140413359881094,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8757216930389404,
      "step": 3226
    },
    {
      "epoch": 0.2715723940660142,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8683801889419556,
      "step": 3228
    },
    {
      "epoch": 0.27174065453321744,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8493150472640991,
      "step": 3230
    },
    {
      "epoch": 0.27190891500042064,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8689993619918823,
      "step": 3232
    },
    {
      "epoch": 0.2720771754676239,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8789758682250977,
      "step": 3234
    },
    {
      "epoch": 0.2722454359348271,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.850790023803711,
      "step": 3236
    },
    {
      "epoch": 0.27241369640203034,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8922373056411743,
      "step": 3238
    },
    {
      "epoch": 0.2725819568692336,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8787682056427002,
      "step": 3240
    },
    {
      "epoch": 0.2727502173364368,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8718161582946777,
      "step": 3242
    },
    {
      "epoch": 0.27291847780364004,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8688337802886963,
      "step": 3244
    },
    {
      "epoch": 0.2730867382708433,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8463727235794067,
      "step": 3246
    },
    {
      "epoch": 0.2732549987380465,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8892571926116943,
      "step": 3248
    },
    {
      "epoch": 0.27342325920524974,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8122692108154297,
      "step": 3250
    },
    {
      "epoch": 0.27359151967245293,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8240017890930176,
      "step": 3252
    },
    {
      "epoch": 0.2737597801396562,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8752942085266113,
      "step": 3254
    },
    {
      "epoch": 0.27392804060685944,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8433451652526855,
      "step": 3256
    },
    {
      "epoch": 0.27409630107406263,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8626364469528198,
      "step": 3258
    },
    {
      "epoch": 0.2742645615412659,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8147199153900146,
      "step": 3260
    },
    {
      "epoch": 0.27443282200846913,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9179339408874512,
      "step": 3262
    },
    {
      "epoch": 0.27460108247567233,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.876093864440918,
      "step": 3264
    },
    {
      "epoch": 0.2747693429428756,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.8985507488250732,
      "step": 3266
    },
    {
      "epoch": 0.2749376034100788,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0003,
      "loss": 1.8344101905822754,
      "step": 3268
    },
    {
      "epoch": 0.27510586387728203,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0003,
      "loss": 1.8771886825561523,
      "step": 3270
    },
    {
      "epoch": 0.2752741243444853,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003,
      "loss": 1.926322340965271,
      "step": 3272
    },
    {
      "epoch": 0.2754423848116885,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.8937790393829346,
      "step": 3274
    },
    {
      "epoch": 0.27561064527889173,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.913184642791748,
      "step": 3276
    },
    {
      "epoch": 0.275778905746095,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.906518578529358,
      "step": 3278
    },
    {
      "epoch": 0.2759471662132982,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8961617946624756,
      "step": 3280
    },
    {
      "epoch": 0.27611542668050143,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.904283046722412,
      "step": 3282
    },
    {
      "epoch": 0.2762836871477046,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8780951499938965,
      "step": 3284
    },
    {
      "epoch": 0.2764519476149079,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8134222030639648,
      "step": 3286
    },
    {
      "epoch": 0.27662020808211113,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8607853651046753,
      "step": 3288
    },
    {
      "epoch": 0.2767884685493143,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8843684196472168,
      "step": 3290
    },
    {
      "epoch": 0.2769567290165176,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8532389402389526,
      "step": 3292
    },
    {
      "epoch": 0.2771249894837208,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8140994310379028,
      "step": 3294
    },
    {
      "epoch": 0.277293249950924,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8465523719787598,
      "step": 3296
    },
    {
      "epoch": 0.2774615104181273,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8625171184539795,
      "step": 3298
    },
    {
      "epoch": 0.2776297708853305,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.82668137550354,
      "step": 3300
    },
    {
      "epoch": 0.2777980313525337,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8364675045013428,
      "step": 3302
    },
    {
      "epoch": 0.277966291819737,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.9137320518493652,
      "step": 3304
    },
    {
      "epoch": 0.2781345522869402,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.808650255203247,
      "step": 3306
    },
    {
      "epoch": 0.2783028127541434,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.867314100265503,
      "step": 3308
    },
    {
      "epoch": 0.2784710732213466,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8249554634094238,
      "step": 3310
    },
    {
      "epoch": 0.2786393336885499,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.790658950805664,
      "step": 3312
    },
    {
      "epoch": 0.2788075941557531,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.913550615310669,
      "step": 3314
    },
    {
      "epoch": 0.2789758546229563,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8534421920776367,
      "step": 3316
    },
    {
      "epoch": 0.2791441150901596,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8863084316253662,
      "step": 3318
    },
    {
      "epoch": 0.2793123755573628,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8995580673217773,
      "step": 3320
    },
    {
      "epoch": 0.279480636024566,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8419835567474365,
      "step": 3322
    },
    {
      "epoch": 0.27964889649176927,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.832056999206543,
      "step": 3324
    },
    {
      "epoch": 0.27981715695897247,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8926503658294678,
      "step": 3326
    },
    {
      "epoch": 0.2799854174261757,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8881720304489136,
      "step": 3328
    },
    {
      "epoch": 0.28015367789337897,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8902665376663208,
      "step": 3330
    },
    {
      "epoch": 0.28032193836058217,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.880469799041748,
      "step": 3332
    },
    {
      "epoch": 0.2804901988277854,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8646562099456787,
      "step": 3334
    },
    {
      "epoch": 0.2806584592949886,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8994884490966797,
      "step": 3336
    },
    {
      "epoch": 0.28082671976219187,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8559415340423584,
      "step": 3338
    },
    {
      "epoch": 0.2809949802293951,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8474715948104858,
      "step": 3340
    },
    {
      "epoch": 0.2811632406965983,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8643250465393066,
      "step": 3342
    },
    {
      "epoch": 0.28133150116380157,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.852321982383728,
      "step": 3344
    },
    {
      "epoch": 0.2814997616310048,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8344388008117676,
      "step": 3346
    },
    {
      "epoch": 0.281668022098208,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.898447871208191,
      "step": 3348
    },
    {
      "epoch": 0.28183628256541127,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9108023643493652,
      "step": 3350
    },
    {
      "epoch": 0.28200454303261446,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9083607196807861,
      "step": 3352
    },
    {
      "epoch": 0.2821728034998177,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.865666389465332,
      "step": 3354
    },
    {
      "epoch": 0.28234106396702097,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8762094974517822,
      "step": 3356
    },
    {
      "epoch": 0.28250932443422416,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.878679871559143,
      "step": 3358
    },
    {
      "epoch": 0.2826775849014274,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8622525930404663,
      "step": 3360
    },
    {
      "epoch": 0.28284584536863067,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9196500778198242,
      "step": 3362
    },
    {
      "epoch": 0.28301410583583386,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8941612243652344,
      "step": 3364
    },
    {
      "epoch": 0.2831823663030371,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.7953133583068848,
      "step": 3366
    },
    {
      "epoch": 0.2833506267702403,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.854511022567749,
      "step": 3368
    },
    {
      "epoch": 0.28351888723744356,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9137613773345947,
      "step": 3370
    },
    {
      "epoch": 0.2836871477046468,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8739721775054932,
      "step": 3372
    },
    {
      "epoch": 0.28385540817185,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8667056560516357,
      "step": 3374
    },
    {
      "epoch": 0.28402366863905326,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8361566066741943,
      "step": 3376
    },
    {
      "epoch": 0.28419192910625646,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9219236373901367,
      "step": 3378
    },
    {
      "epoch": 0.2843601895734597,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8535802364349365,
      "step": 3380
    },
    {
      "epoch": 0.28452845004066296,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.869412899017334,
      "step": 3382
    },
    {
      "epoch": 0.28469671050786616,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8627632856369019,
      "step": 3384
    },
    {
      "epoch": 0.2848649709750694,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9168916940689087,
      "step": 3386
    },
    {
      "epoch": 0.28503323144227266,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8747608661651611,
      "step": 3388
    },
    {
      "epoch": 0.28520149190947586,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.8689842224121094,
      "step": 3390
    },
    {
      "epoch": 0.2853697523766791,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8774120807647705,
      "step": 3392
    },
    {
      "epoch": 0.2855380128438823,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.877009391784668,
      "step": 3394
    },
    {
      "epoch": 0.28570627331108556,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.887885570526123,
      "step": 3396
    },
    {
      "epoch": 0.2858745337782888,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.895578384399414,
      "step": 3398
    },
    {
      "epoch": 0.286042794245492,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9170576333999634,
      "step": 3400
    },
    {
      "epoch": 0.28621105471269526,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8974868059158325,
      "step": 3402
    },
    {
      "epoch": 0.2863793151798985,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8262208700180054,
      "step": 3404
    },
    {
      "epoch": 0.2865475756471017,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8496284484863281,
      "step": 3406
    },
    {
      "epoch": 0.28671583611430496,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9032862186431885,
      "step": 3408
    },
    {
      "epoch": 0.28688409658150815,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.87919282913208,
      "step": 3410
    },
    {
      "epoch": 0.2870523570487114,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8344480991363525,
      "step": 3412
    },
    {
      "epoch": 0.28722061751591466,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.870591163635254,
      "step": 3414
    },
    {
      "epoch": 0.28738887798311785,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8891026973724365,
      "step": 3416
    },
    {
      "epoch": 0.2875571384503211,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.86173677444458,
      "step": 3418
    },
    {
      "epoch": 0.28772539891752436,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8951568603515625,
      "step": 3420
    },
    {
      "epoch": 0.28789365938472755,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8699982166290283,
      "step": 3422
    },
    {
      "epoch": 0.2880619198519308,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.860522747039795,
      "step": 3424
    },
    {
      "epoch": 0.288230180319134,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003,
      "loss": 1.8473031520843506,
      "step": 3426
    },
    {
      "epoch": 0.28839844078633725,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8931434154510498,
      "step": 3428
    },
    {
      "epoch": 0.2885667012535405,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.853391170501709,
      "step": 3430
    },
    {
      "epoch": 0.2887349617207437,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8910959959030151,
      "step": 3432
    },
    {
      "epoch": 0.28890322218794695,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8486597537994385,
      "step": 3434
    },
    {
      "epoch": 0.28907148265515015,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8814618587493896,
      "step": 3436
    },
    {
      "epoch": 0.2892397431223534,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8941593170166016,
      "step": 3438
    },
    {
      "epoch": 0.28940800358955665,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.863787055015564,
      "step": 3440
    },
    {
      "epoch": 0.28957626405675985,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9135799407958984,
      "step": 3442
    },
    {
      "epoch": 0.2897445245239631,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8287431001663208,
      "step": 3444
    },
    {
      "epoch": 0.28991278499116635,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8727765083312988,
      "step": 3446
    },
    {
      "epoch": 0.29008104545836955,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8580970764160156,
      "step": 3448
    },
    {
      "epoch": 0.2902493059255728,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.867619514465332,
      "step": 3450
    },
    {
      "epoch": 0.290417566392776,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8662925958633423,
      "step": 3452
    },
    {
      "epoch": 0.29058582685997925,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.87797212600708,
      "step": 3454
    },
    {
      "epoch": 0.2907540873271825,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8538389205932617,
      "step": 3456
    },
    {
      "epoch": 0.2909223477943857,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9146101474761963,
      "step": 3458
    },
    {
      "epoch": 0.29109060826158895,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9076781272888184,
      "step": 3460
    },
    {
      "epoch": 0.2912588687287922,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8445765972137451,
      "step": 3462
    },
    {
      "epoch": 0.2914271291959954,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.873124599456787,
      "step": 3464
    },
    {
      "epoch": 0.29159538966319865,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8695573806762695,
      "step": 3466
    },
    {
      "epoch": 0.29176365013040184,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.861147403717041,
      "step": 3468
    },
    {
      "epoch": 0.2919319105976051,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.894752025604248,
      "step": 3470
    },
    {
      "epoch": 0.29210017106480834,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8843766450881958,
      "step": 3472
    },
    {
      "epoch": 0.29226843153201154,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8641018867492676,
      "step": 3474
    },
    {
      "epoch": 0.2924366919992148,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8791038990020752,
      "step": 3476
    },
    {
      "epoch": 0.292604952466418,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8540399074554443,
      "step": 3478
    },
    {
      "epoch": 0.29277321293362124,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8794680833816528,
      "step": 3480
    },
    {
      "epoch": 0.2929414734008245,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8642399311065674,
      "step": 3482
    },
    {
      "epoch": 0.2931097338680277,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9025368690490723,
      "step": 3484
    },
    {
      "epoch": 0.29327799433523094,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9088737964630127,
      "step": 3486
    },
    {
      "epoch": 0.2934462548024342,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8538222312927246,
      "step": 3488
    },
    {
      "epoch": 0.2936145152696374,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.9040887355804443,
      "step": 3490
    },
    {
      "epoch": 0.29378277573684064,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8559389114379883,
      "step": 3492
    },
    {
      "epoch": 0.29395103620404384,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8348047733306885,
      "step": 3494
    },
    {
      "epoch": 0.2941192966712471,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.854455590248108,
      "step": 3496
    },
    {
      "epoch": 0.29428755713845034,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8741997480392456,
      "step": 3498
    },
    {
      "epoch": 0.29445581760565354,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8406360149383545,
      "step": 3500
    },
    {
      "epoch": 0.2946240780728568,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9639798402786255,
      "step": 3502
    },
    {
      "epoch": 0.29479233854006004,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8607097864151,
      "step": 3504
    },
    {
      "epoch": 0.29496059900726324,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8991074562072754,
      "step": 3506
    },
    {
      "epoch": 0.2951288594744665,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8867425918579102,
      "step": 3508
    },
    {
      "epoch": 0.2952971199416697,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8978204727172852,
      "step": 3510
    },
    {
      "epoch": 0.29546538040887294,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9087927341461182,
      "step": 3512
    },
    {
      "epoch": 0.2956336408760762,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8832217454910278,
      "step": 3514
    },
    {
      "epoch": 0.2958019013432794,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003,
      "loss": 1.8715660572052002,
      "step": 3516
    },
    {
      "epoch": 0.29597016181048263,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8473100662231445,
      "step": 3518
    },
    {
      "epoch": 0.29613842227768583,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8622734546661377,
      "step": 3520
    },
    {
      "epoch": 0.2963066827448891,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8666963577270508,
      "step": 3522
    },
    {
      "epoch": 0.29647494321209233,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8607895374298096,
      "step": 3524
    },
    {
      "epoch": 0.29664320367929553,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.843174934387207,
      "step": 3526
    },
    {
      "epoch": 0.2968114641464988,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8476256132125854,
      "step": 3528
    },
    {
      "epoch": 0.29697972461370203,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8535393476486206,
      "step": 3530
    },
    {
      "epoch": 0.29714798508090523,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8670868873596191,
      "step": 3532
    },
    {
      "epoch": 0.2973162455481085,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9105762243270874,
      "step": 3534
    },
    {
      "epoch": 0.2974845060153117,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8914451599121094,
      "step": 3536
    },
    {
      "epoch": 0.29765276648251493,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8613178730010986,
      "step": 3538
    },
    {
      "epoch": 0.2978210269497182,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8899122476577759,
      "step": 3540
    },
    {
      "epoch": 0.2979892874169214,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8745774030685425,
      "step": 3542
    },
    {
      "epoch": 0.29815754788412463,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8806946277618408,
      "step": 3544
    },
    {
      "epoch": 0.2983258083513279,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9007518291473389,
      "step": 3546
    },
    {
      "epoch": 0.2984940688185311,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9156569242477417,
      "step": 3548
    },
    {
      "epoch": 0.29866232928573433,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8634800910949707,
      "step": 3550
    },
    {
      "epoch": 0.2988305897529375,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8830186128616333,
      "step": 3552
    },
    {
      "epoch": 0.2989988502201408,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8644263744354248,
      "step": 3554
    },
    {
      "epoch": 0.29916711068734403,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8633571863174438,
      "step": 3556
    },
    {
      "epoch": 0.2993353711545472,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8935626745224,
      "step": 3558
    },
    {
      "epoch": 0.2995036316217505,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8801356554031372,
      "step": 3560
    },
    {
      "epoch": 0.2996718920889537,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8661222457885742,
      "step": 3562
    },
    {
      "epoch": 0.2998401525561569,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8208799362182617,
      "step": 3564
    },
    {
      "epoch": 0.3000084130233602,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8886229991912842,
      "step": 3566
    },
    {
      "epoch": 0.3001766734905634,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8791732788085938,
      "step": 3568
    },
    {
      "epoch": 0.3003449339577666,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.85847806930542,
      "step": 3570
    },
    {
      "epoch": 0.3005131944249699,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8934385776519775,
      "step": 3572
    },
    {
      "epoch": 0.3006814548921731,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9081249237060547,
      "step": 3574
    },
    {
      "epoch": 0.3008497153593763,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8954684734344482,
      "step": 3576
    },
    {
      "epoch": 0.3010179758265795,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8690669536590576,
      "step": 3578
    },
    {
      "epoch": 0.30118623629378277,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8758726119995117,
      "step": 3580
    },
    {
      "epoch": 0.301354496760986,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.869187355041504,
      "step": 3582
    },
    {
      "epoch": 0.3015227572281892,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8464605808258057,
      "step": 3584
    },
    {
      "epoch": 0.30169101769539247,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8511884212493896,
      "step": 3586
    },
    {
      "epoch": 0.3018592781625957,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8904540538787842,
      "step": 3588
    },
    {
      "epoch": 0.3020275386297989,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.8502109050750732,
      "step": 3590
    },
    {
      "epoch": 0.30219579909700217,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.865281581878662,
      "step": 3592
    },
    {
      "epoch": 0.30236405956420537,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.7882630825042725,
      "step": 3594
    },
    {
      "epoch": 0.3025323200314086,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.925553798675537,
      "step": 3596
    },
    {
      "epoch": 0.30270058049861187,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8359683752059937,
      "step": 3598
    },
    {
      "epoch": 0.30286884096581507,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.853502869606018,
      "step": 3600
    },
    {
      "epoch": 0.3030371014330183,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8571507930755615,
      "step": 3602
    },
    {
      "epoch": 0.30320536190022157,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8611356019973755,
      "step": 3604
    },
    {
      "epoch": 0.30337362236742477,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.865299940109253,
      "step": 3606
    },
    {
      "epoch": 0.303541882834628,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9234116077423096,
      "step": 3608
    },
    {
      "epoch": 0.3037101433018312,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9533205032348633,
      "step": 3610
    },
    {
      "epoch": 0.30387840376903447,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8358526229858398,
      "step": 3612
    },
    {
      "epoch": 0.3040466642362377,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.7999510765075684,
      "step": 3614
    },
    {
      "epoch": 0.3042149247034409,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8740417957305908,
      "step": 3616
    },
    {
      "epoch": 0.30438318517064417,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.851014494895935,
      "step": 3618
    },
    {
      "epoch": 0.30455144563784736,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.781557321548462,
      "step": 3620
    },
    {
      "epoch": 0.3047197061050506,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9052770137786865,
      "step": 3622
    },
    {
      "epoch": 0.30488796657225387,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8342921733856201,
      "step": 3624
    },
    {
      "epoch": 0.30505622703945706,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8293700218200684,
      "step": 3626
    },
    {
      "epoch": 0.3052244875066603,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8566792011260986,
      "step": 3628
    },
    {
      "epoch": 0.30539274797386357,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9174964427947998,
      "step": 3630
    },
    {
      "epoch": 0.30556100844106676,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.815117359161377,
      "step": 3632
    },
    {
      "epoch": 0.30572926890827,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8271584510803223,
      "step": 3634
    },
    {
      "epoch": 0.3058975293754732,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8668638467788696,
      "step": 3636
    },
    {
      "epoch": 0.30606578984267646,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8688249588012695,
      "step": 3638
    },
    {
      "epoch": 0.3062340503098797,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9120137691497803,
      "step": 3640
    },
    {
      "epoch": 0.3064023107770829,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0003,
      "loss": 1.8708055019378662,
      "step": 3642
    },
    {
      "epoch": 0.30657057124428616,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003,
      "loss": 1.8885983228683472,
      "step": 3644
    },
    {
      "epoch": 0.3067388317114894,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.848218560218811,
      "step": 3646
    },
    {
      "epoch": 0.3069070921786926,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8439409732818604,
      "step": 3648
    },
    {
      "epoch": 0.30707535264589586,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8663172721862793,
      "step": 3650
    },
    {
      "epoch": 0.30724361311309906,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.955939769744873,
      "step": 3652
    },
    {
      "epoch": 0.3074118735803023,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.871226191520691,
      "step": 3654
    },
    {
      "epoch": 0.30758013404750556,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.836936593055725,
      "step": 3656
    },
    {
      "epoch": 0.30774839451470876,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8596620559692383,
      "step": 3658
    },
    {
      "epoch": 0.307916654981912,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8819119930267334,
      "step": 3660
    },
    {
      "epoch": 0.3080849154491152,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9014511108398438,
      "step": 3662
    },
    {
      "epoch": 0.30825317591631846,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8783725500106812,
      "step": 3664
    },
    {
      "epoch": 0.3084214363835217,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8670105934143066,
      "step": 3666
    },
    {
      "epoch": 0.3085896968507249,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8725711107254028,
      "step": 3668
    },
    {
      "epoch": 0.30875795731792816,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8577990531921387,
      "step": 3670
    },
    {
      "epoch": 0.3089262177851314,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.896641492843628,
      "step": 3672
    },
    {
      "epoch": 0.3090944782523346,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8387861251831055,
      "step": 3674
    },
    {
      "epoch": 0.30926273871953786,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8609778881072998,
      "step": 3676
    },
    {
      "epoch": 0.30943099918674105,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8766117095947266,
      "step": 3678
    },
    {
      "epoch": 0.3095992596539443,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8612436056137085,
      "step": 3680
    },
    {
      "epoch": 0.30976752012114755,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.879903793334961,
      "step": 3682
    },
    {
      "epoch": 0.30993578058835075,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8374532461166382,
      "step": 3684
    },
    {
      "epoch": 0.310104041055554,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.7740446329116821,
      "step": 3686
    },
    {
      "epoch": 0.31027230152275725,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.7790193557739258,
      "step": 3688
    },
    {
      "epoch": 0.31044056198996045,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8381619453430176,
      "step": 3690
    },
    {
      "epoch": 0.3106088224571637,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8676321506500244,
      "step": 3692
    },
    {
      "epoch": 0.3107770829243669,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.849055528640747,
      "step": 3694
    },
    {
      "epoch": 0.31094534339157015,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8608002662658691,
      "step": 3696
    },
    {
      "epoch": 0.3111136038587734,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9172919988632202,
      "step": 3698
    },
    {
      "epoch": 0.3112818643259766,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8847105503082275,
      "step": 3700
    },
    {
      "epoch": 0.31145012479317985,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9292223453521729,
      "step": 3702
    },
    {
      "epoch": 0.31161838526038305,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9048868417739868,
      "step": 3704
    },
    {
      "epoch": 0.3117866457275863,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.843309998512268,
      "step": 3706
    },
    {
      "epoch": 0.31195490619478955,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8788464069366455,
      "step": 3708
    },
    {
      "epoch": 0.31212316666199275,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8821687698364258,
      "step": 3710
    },
    {
      "epoch": 0.312291427129196,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.888781189918518,
      "step": 3712
    },
    {
      "epoch": 0.31245968759639925,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8492558002471924,
      "step": 3714
    },
    {
      "epoch": 0.31262794806360245,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8836572170257568,
      "step": 3716
    },
    {
      "epoch": 0.3127962085308057,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8774948120117188,
      "step": 3718
    },
    {
      "epoch": 0.3129644689980089,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8469822406768799,
      "step": 3720
    },
    {
      "epoch": 0.31313272946521215,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8612725734710693,
      "step": 3722
    },
    {
      "epoch": 0.3133009899324154,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9077210426330566,
      "step": 3724
    },
    {
      "epoch": 0.3134692503996186,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8379788398742676,
      "step": 3726
    },
    {
      "epoch": 0.31363751086682184,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8925726413726807,
      "step": 3728
    },
    {
      "epoch": 0.3138057713340251,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.9144246578216553,
      "step": 3730
    },
    {
      "epoch": 0.3139740318012283,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8567466735839844,
      "step": 3732
    },
    {
      "epoch": 0.31414229226843154,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8199453353881836,
      "step": 3734
    },
    {
      "epoch": 0.31431055273563474,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8588147163391113,
      "step": 3736
    },
    {
      "epoch": 0.314478813202838,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.7948260307312012,
      "step": 3738
    },
    {
      "epoch": 0.31464707367004124,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8405332565307617,
      "step": 3740
    },
    {
      "epoch": 0.31481533413724444,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8540245294570923,
      "step": 3742
    },
    {
      "epoch": 0.3149835946044477,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8421465158462524,
      "step": 3744
    },
    {
      "epoch": 0.3151518550716509,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.887857437133789,
      "step": 3746
    },
    {
      "epoch": 0.31532011553885414,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.881984829902649,
      "step": 3748
    },
    {
      "epoch": 0.3154883760060574,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.824082612991333,
      "step": 3750
    },
    {
      "epoch": 0.3156566364732606,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8149815797805786,
      "step": 3752
    },
    {
      "epoch": 0.31582489694046384,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8328945636749268,
      "step": 3754
    },
    {
      "epoch": 0.3159931574076671,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8917242288589478,
      "step": 3756
    },
    {
      "epoch": 0.3161614178748703,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8336012363433838,
      "step": 3758
    },
    {
      "epoch": 0.31632967834207354,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8584048748016357,
      "step": 3760
    },
    {
      "epoch": 0.31649793880927674,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.9122538566589355,
      "step": 3762
    },
    {
      "epoch": 0.31666619927648,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8859353065490723,
      "step": 3764
    },
    {
      "epoch": 0.31683445974368324,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8861830234527588,
      "step": 3766
    },
    {
      "epoch": 0.31700272021088644,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8504984378814697,
      "step": 3768
    },
    {
      "epoch": 0.3171709806780897,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8950271606445312,
      "step": 3770
    },
    {
      "epoch": 0.31733924114529294,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8434422016143799,
      "step": 3772
    },
    {
      "epoch": 0.31750750161249613,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.823119878768921,
      "step": 3774
    },
    {
      "epoch": 0.3176757620796994,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.7941339015960693,
      "step": 3776
    },
    {
      "epoch": 0.3178440225469026,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.809309482574463,
      "step": 3778
    },
    {
      "epoch": 0.31801228301410583,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8305456638336182,
      "step": 3780
    },
    {
      "epoch": 0.3181805434813091,
      "grad_norm": 0.21875,
      "learning_rate": 0.0003,
      "loss": 1.8734121322631836,
      "step": 3782
    },
    {
      "epoch": 0.3183488039485123,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.862046480178833,
      "step": 3784
    },
    {
      "epoch": 0.31851706441571553,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8620274066925049,
      "step": 3786
    },
    {
      "epoch": 0.3186853248829188,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8918719291687012,
      "step": 3788
    },
    {
      "epoch": 0.318853585350122,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8448455333709717,
      "step": 3790
    },
    {
      "epoch": 0.31902184581732523,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.879143238067627,
      "step": 3792
    },
    {
      "epoch": 0.31919010628452843,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.8448827266693115,
      "step": 3794
    },
    {
      "epoch": 0.3193583667517317,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8693859577178955,
      "step": 3796
    },
    {
      "epoch": 0.31952662721893493,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9114189147949219,
      "step": 3798
    },
    {
      "epoch": 0.31969488768613813,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8946382999420166,
      "step": 3800
    },
    {
      "epoch": 0.3198631481533414,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8792357444763184,
      "step": 3802
    },
    {
      "epoch": 0.3200314086205446,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.851743221282959,
      "step": 3804
    },
    {
      "epoch": 0.32019966908774783,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8510115146636963,
      "step": 3806
    },
    {
      "epoch": 0.3203679295549511,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8973456621170044,
      "step": 3808
    },
    {
      "epoch": 0.3205361900221543,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.79490327835083,
      "step": 3810
    },
    {
      "epoch": 0.32070445048935753,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8858728408813477,
      "step": 3812
    },
    {
      "epoch": 0.3208727109565608,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.9002485275268555,
      "step": 3814
    },
    {
      "epoch": 0.321040971423764,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8303015232086182,
      "step": 3816
    },
    {
      "epoch": 0.32120923189096723,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8684226274490356,
      "step": 3818
    },
    {
      "epoch": 0.3213774923581704,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8956971168518066,
      "step": 3820
    },
    {
      "epoch": 0.3215457528253737,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8989932537078857,
      "step": 3822
    },
    {
      "epoch": 0.32171401329257693,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.838505506515503,
      "step": 3824
    },
    {
      "epoch": 0.3218822737597801,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8919272422790527,
      "step": 3826
    },
    {
      "epoch": 0.3220505342269834,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8320837020874023,
      "step": 3828
    },
    {
      "epoch": 0.3222187946941866,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8338614702224731,
      "step": 3830
    },
    {
      "epoch": 0.3223870551613898,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8390660285949707,
      "step": 3832
    },
    {
      "epoch": 0.3225553156285931,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8995847702026367,
      "step": 3834
    },
    {
      "epoch": 0.32272357609579627,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8751614093780518,
      "step": 3836
    },
    {
      "epoch": 0.3228918365629995,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.9016811847686768,
      "step": 3838
    },
    {
      "epoch": 0.3230600970302028,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.850046157836914,
      "step": 3840
    },
    {
      "epoch": 0.32322835749740597,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8559223413467407,
      "step": 3842
    },
    {
      "epoch": 0.3233966179646092,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.870532512664795,
      "step": 3844
    },
    {
      "epoch": 0.3235648784318124,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8562122583389282,
      "step": 3846
    },
    {
      "epoch": 0.32373313889901567,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8243489265441895,
      "step": 3848
    },
    {
      "epoch": 0.3239013993662189,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.820356845855713,
      "step": 3850
    },
    {
      "epoch": 0.3240696598334221,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.871504306793213,
      "step": 3852
    },
    {
      "epoch": 0.32423792030062537,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.865014910697937,
      "step": 3854
    },
    {
      "epoch": 0.3244061807678286,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8796753883361816,
      "step": 3856
    },
    {
      "epoch": 0.3245744412350318,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.920353651046753,
      "step": 3858
    },
    {
      "epoch": 0.32474270170223507,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8169820308685303,
      "step": 3860
    },
    {
      "epoch": 0.32491096216943827,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8427238464355469,
      "step": 3862
    },
    {
      "epoch": 0.3250792226366415,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.904712438583374,
      "step": 3864
    },
    {
      "epoch": 0.32524748310384477,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8915772438049316,
      "step": 3866
    },
    {
      "epoch": 0.32541574357104797,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9368259906768799,
      "step": 3868
    },
    {
      "epoch": 0.3255840040382512,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8592915534973145,
      "step": 3870
    },
    {
      "epoch": 0.32575226450545447,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.86061692237854,
      "step": 3872
    },
    {
      "epoch": 0.32592052497265767,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8890175819396973,
      "step": 3874
    },
    {
      "epoch": 0.3260887854398609,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8447871208190918,
      "step": 3876
    },
    {
      "epoch": 0.3262570459070641,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.856191635131836,
      "step": 3878
    },
    {
      "epoch": 0.32642530637426737,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8729684352874756,
      "step": 3880
    },
    {
      "epoch": 0.3265935668414706,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8386845588684082,
      "step": 3882
    },
    {
      "epoch": 0.3267618273086738,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8728570938110352,
      "step": 3884
    },
    {
      "epoch": 0.32693008777587707,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.866088628768921,
      "step": 3886
    },
    {
      "epoch": 0.32709834824308026,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8823490142822266,
      "step": 3888
    },
    {
      "epoch": 0.3272666087102835,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8277677297592163,
      "step": 3890
    },
    {
      "epoch": 0.32743486917748676,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8401596546173096,
      "step": 3892
    },
    {
      "epoch": 0.32760312964468996,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8584022521972656,
      "step": 3894
    },
    {
      "epoch": 0.3277713901118932,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8188765048980713,
      "step": 3896
    },
    {
      "epoch": 0.32793965057909646,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.875680923461914,
      "step": 3898
    },
    {
      "epoch": 0.32810791104629966,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8383197784423828,
      "step": 3900
    },
    {
      "epoch": 0.3282761715135029,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8785862922668457,
      "step": 3902
    },
    {
      "epoch": 0.3284444319807061,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8587710857391357,
      "step": 3904
    },
    {
      "epoch": 0.32861269244790936,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8723313808441162,
      "step": 3906
    },
    {
      "epoch": 0.3287809529151126,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8828476667404175,
      "step": 3908
    },
    {
      "epoch": 0.3289492133823158,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8386127948760986,
      "step": 3910
    },
    {
      "epoch": 0.32911747384951906,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8685197830200195,
      "step": 3912
    },
    {
      "epoch": 0.3292857343167223,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9180203676223755,
      "step": 3914
    },
    {
      "epoch": 0.3294539947839255,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8418514728546143,
      "step": 3916
    },
    {
      "epoch": 0.32962225525112876,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.896508812904358,
      "step": 3918
    },
    {
      "epoch": 0.32979051571833196,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9026496410369873,
      "step": 3920
    },
    {
      "epoch": 0.3299587761855352,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.863297939300537,
      "step": 3922
    },
    {
      "epoch": 0.33012703665273846,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9035916328430176,
      "step": 3924
    },
    {
      "epoch": 0.33029529711994166,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.863051414489746,
      "step": 3926
    },
    {
      "epoch": 0.3304635575871449,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.858727216720581,
      "step": 3928
    },
    {
      "epoch": 0.3306318180543481,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8483350276947021,
      "step": 3930
    },
    {
      "epoch": 0.33080007852155136,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8101766109466553,
      "step": 3932
    },
    {
      "epoch": 0.3309683389887546,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9169988632202148,
      "step": 3934
    },
    {
      "epoch": 0.3311365994559578,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8523852825164795,
      "step": 3936
    },
    {
      "epoch": 0.33130485992316105,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9023165702819824,
      "step": 3938
    },
    {
      "epoch": 0.3314731203903643,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.859055519104004,
      "step": 3940
    },
    {
      "epoch": 0.3316413808575675,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.839821696281433,
      "step": 3942
    },
    {
      "epoch": 0.33180964132477075,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0003,
      "loss": 1.8347830772399902,
      "step": 3944
    },
    {
      "epoch": 0.33197790179197395,
      "grad_norm": 1.65625,
      "learning_rate": 0.0003,
      "loss": 1.9241106510162354,
      "step": 3946
    },
    {
      "epoch": 0.3321461622591772,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0003,
      "loss": 1.9086036682128906,
      "step": 3948
    },
    {
      "epoch": 0.33231442272638045,
      "grad_norm": 24.875,
      "learning_rate": 0.0003,
      "loss": 2.586730480194092,
      "step": 3950
    },
    {
      "epoch": 0.33248268319358365,
      "grad_norm": 0.75,
      "learning_rate": 0.0003,
      "loss": 2.0153965950012207,
      "step": 3952
    },
    {
      "epoch": 0.3326509436607869,
      "grad_norm": 0.3125,
      "learning_rate": 0.0003,
      "loss": 1.982718825340271,
      "step": 3954
    },
    {
      "epoch": 0.33281920412799015,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0003,
      "loss": 1.848696231842041,
      "step": 3956
    },
    {
      "epoch": 0.33298746459519335,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0003,
      "loss": 1.9203574657440186,
      "step": 3958
    },
    {
      "epoch": 0.3331557250623966,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0003,
      "loss": 1.884813904762268,
      "step": 3960
    },
    {
      "epoch": 0.3333239855295998,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0003,
      "loss": 1.914201259613037,
      "step": 3962
    },
    {
      "epoch": 0.33349224599680305,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003,
      "loss": 1.8636889457702637,
      "step": 3964
    },
    {
      "epoch": 0.3336605064640063,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0003,
      "loss": 1.91408109664917,
      "step": 3966
    },
    {
      "epoch": 0.3338287669312095,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003,
      "loss": 1.911180019378662,
      "step": 3968
    },
    {
      "epoch": 0.33399702739841275,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8750474452972412,
      "step": 3970
    },
    {
      "epoch": 0.334165287865616,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003,
      "loss": 1.9138474464416504,
      "step": 3972
    },
    {
      "epoch": 0.3343335483328192,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8643827438354492,
      "step": 3974
    },
    {
      "epoch": 0.33450180880002245,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8882591724395752,
      "step": 3976
    },
    {
      "epoch": 0.33467006926722564,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0003,
      "loss": 1.8955390453338623,
      "step": 3978
    },
    {
      "epoch": 0.3348383297344289,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.858117699623108,
      "step": 3980
    },
    {
      "epoch": 0.33500659020163215,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.928159475326538,
      "step": 3982
    },
    {
      "epoch": 0.33517485066883534,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8956348896026611,
      "step": 3984
    },
    {
      "epoch": 0.3353431111360386,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8543643951416016,
      "step": 3986
    },
    {
      "epoch": 0.3355113716032418,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8436110019683838,
      "step": 3988
    },
    {
      "epoch": 0.33567963207044504,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.924584984779358,
      "step": 3990
    },
    {
      "epoch": 0.3358478925376483,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8808845281600952,
      "step": 3992
    },
    {
      "epoch": 0.3360161530048515,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9404008388519287,
      "step": 3994
    },
    {
      "epoch": 0.33618441347205474,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8268177509307861,
      "step": 3996
    },
    {
      "epoch": 0.336352673939258,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8831335306167603,
      "step": 3998
    },
    {
      "epoch": 0.3365209344064612,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8302100896835327,
      "step": 4000
    },
    {
      "epoch": 0.33668919487366444,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8537179231643677,
      "step": 4002
    },
    {
      "epoch": 0.33685745534086764,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.875118374824524,
      "step": 4004
    },
    {
      "epoch": 0.3370257158080709,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8614304065704346,
      "step": 4006
    },
    {
      "epoch": 0.33719397627527414,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8663849830627441,
      "step": 4008
    },
    {
      "epoch": 0.33736223674247734,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8526201248168945,
      "step": 4010
    },
    {
      "epoch": 0.3375304972096806,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8697057962417603,
      "step": 4012
    },
    {
      "epoch": 0.33769875767688384,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.810081958770752,
      "step": 4014
    },
    {
      "epoch": 0.33786701814408704,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9480533599853516,
      "step": 4016
    },
    {
      "epoch": 0.3380352786112903,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.9049072265625,
      "step": 4018
    },
    {
      "epoch": 0.3382035390784935,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9020788669586182,
      "step": 4020
    },
    {
      "epoch": 0.33837179954569674,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8666861057281494,
      "step": 4022
    },
    {
      "epoch": 0.3385400600129,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9119019508361816,
      "step": 4024
    },
    {
      "epoch": 0.3387083204801032,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8384666442871094,
      "step": 4026
    },
    {
      "epoch": 0.33887658094730644,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8974761962890625,
      "step": 4028
    },
    {
      "epoch": 0.33904484141450963,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8816516399383545,
      "step": 4030
    },
    {
      "epoch": 0.3392131018817129,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.908820390701294,
      "step": 4032
    },
    {
      "epoch": 0.33938136234891614,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.9065861701965332,
      "step": 4034
    },
    {
      "epoch": 0.33954962281611933,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8656210899353027,
      "step": 4036
    },
    {
      "epoch": 0.3397178832833226,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8544164896011353,
      "step": 4038
    },
    {
      "epoch": 0.33988614375052584,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8324735164642334,
      "step": 4040
    },
    {
      "epoch": 0.34005440421772903,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.841951847076416,
      "step": 4042
    },
    {
      "epoch": 0.3402226646849323,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.9143900871276855,
      "step": 4044
    },
    {
      "epoch": 0.3403909251521355,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8422844409942627,
      "step": 4046
    },
    {
      "epoch": 0.34055918561933873,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8144325017929077,
      "step": 4048
    },
    {
      "epoch": 0.340727446086542,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8615750074386597,
      "step": 4050
    },
    {
      "epoch": 0.3408957065537452,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8646533489227295,
      "step": 4052
    },
    {
      "epoch": 0.34106396702094843,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8293901681900024,
      "step": 4054
    },
    {
      "epoch": 0.3412322274881517,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8809282779693604,
      "step": 4056
    },
    {
      "epoch": 0.3414004879553549,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.873234510421753,
      "step": 4058
    },
    {
      "epoch": 0.34156874842255813,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8918458223342896,
      "step": 4060
    },
    {
      "epoch": 0.34173700888976133,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 1.8683303594589233,
      "step": 4062
    },
    {
      "epoch": 0.3419052693569646,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8012259006500244,
      "step": 4064
    },
    {
      "epoch": 0.34207352982416783,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8581525087356567,
      "step": 4066
    },
    {
      "epoch": 0.34224179029137103,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.899673581123352,
      "step": 4068
    },
    {
      "epoch": 0.3424100507585743,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8437117338180542,
      "step": 4070
    },
    {
      "epoch": 0.3425783112257775,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8295531272888184,
      "step": 4072
    },
    {
      "epoch": 0.34274657169298073,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8007183074951172,
      "step": 4074
    },
    {
      "epoch": 0.342914832160184,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8211514949798584,
      "step": 4076
    },
    {
      "epoch": 0.3430830926273872,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.846879243850708,
      "step": 4078
    },
    {
      "epoch": 0.34325135309459043,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8556374311447144,
      "step": 4080
    },
    {
      "epoch": 0.3434196135617937,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.849456787109375,
      "step": 4082
    },
    {
      "epoch": 0.3435878740289969,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.9221388101577759,
      "step": 4084
    },
    {
      "epoch": 0.3437561344962001,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8587887287139893,
      "step": 4086
    },
    {
      "epoch": 0.3439243949634033,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003,
      "loss": 1.8546226024627686,
      "step": 4088
    },
    {
      "epoch": 0.3440926554306066,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.8572349548339844,
      "step": 4090
    },
    {
      "epoch": 0.3442609158978098,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0003,
      "loss": 1.8384013175964355,
      "step": 4092
    },
    {
      "epoch": 0.344429176365013,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0003,
      "loss": 1.8738865852355957,
      "step": 4094
    },
    {
      "epoch": 0.3445974368322163,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0003,
      "loss": 1.8584131002426147,
      "step": 4096
    },
    {
      "epoch": 0.3447656972994195,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0003,
      "loss": 1.8879488706588745,
      "step": 4098
    },
    {
      "epoch": 0.3449339577666227,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.9069781303405762,
      "step": 4100
    },
    {
      "epoch": 0.345102218233826,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 1.8844830989837646,
      "step": 4102
    },
    {
      "epoch": 0.34527047870102917,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0003,
      "loss": 1.8600679636001587,
      "step": 4104
    },
    {
      "epoch": 0.3454387391682324,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0003,
      "loss": 1.889657974243164,
      "step": 4106
    },
    {
      "epoch": 0.3456069996354357,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8428858518600464,
      "step": 4108
    },
    {
      "epoch": 0.34577526010263887,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003,
      "loss": 1.8077696561813354,
      "step": 4110
    },
    {
      "epoch": 0.3459435205698421,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0003,
      "loss": 1.884803056716919,
      "step": 4112
    },
    {
      "epoch": 0.3461117810370453,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0003,
      "loss": 1.8845510482788086,
      "step": 4114
    },
    {
      "epoch": 0.34628004150424857,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9281530380249023,
      "step": 4116
    },
    {
      "epoch": 0.3464483019714518,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8643503189086914,
      "step": 4118
    },
    {
      "epoch": 0.346616562438655,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.861614465713501,
      "step": 4120
    },
    {
      "epoch": 0.34678482290585827,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.877102017402649,
      "step": 4122
    },
    {
      "epoch": 0.3469530833730615,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8114712238311768,
      "step": 4124
    },
    {
      "epoch": 0.3471213438402647,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8660168647766113,
      "step": 4126
    },
    {
      "epoch": 0.34728960430746797,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.9023497104644775,
      "step": 4128
    },
    {
      "epoch": 0.34745786477467117,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8590810298919678,
      "step": 4130
    },
    {
      "epoch": 0.3476261252418744,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8751860857009888,
      "step": 4132
    },
    {
      "epoch": 0.34779438570907767,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.863155722618103,
      "step": 4134
    },
    {
      "epoch": 0.34796264617628087,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.9208976030349731,
      "step": 4136
    },
    {
      "epoch": 0.3481309066434841,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.806704044342041,
      "step": 4138
    },
    {
      "epoch": 0.34829916711068737,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9564613103866577,
      "step": 4140
    },
    {
      "epoch": 0.34846742757789056,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8453530073165894,
      "step": 4142
    },
    {
      "epoch": 0.3486356880450938,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8754546642303467,
      "step": 4144
    },
    {
      "epoch": 0.348803948512297,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.896795630455017,
      "step": 4146
    },
    {
      "epoch": 0.34897220897950026,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8569164276123047,
      "step": 4148
    },
    {
      "epoch": 0.3491404694467035,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8424506187438965,
      "step": 4150
    },
    {
      "epoch": 0.3493087299139067,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8189446926116943,
      "step": 4152
    },
    {
      "epoch": 0.34947699038110996,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.9150117635726929,
      "step": 4154
    },
    {
      "epoch": 0.3496452508483132,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8320398330688477,
      "step": 4156
    },
    {
      "epoch": 0.3498135113155164,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8553013801574707,
      "step": 4158
    },
    {
      "epoch": 0.34998177178271966,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8649288415908813,
      "step": 4160
    },
    {
      "epoch": 0.35015003224992286,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8052525520324707,
      "step": 4162
    },
    {
      "epoch": 0.3503182927171261,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8695592880249023,
      "step": 4164
    },
    {
      "epoch": 0.35048655318432936,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.7996783256530762,
      "step": 4166
    },
    {
      "epoch": 0.35065481365153256,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.895764946937561,
      "step": 4168
    },
    {
      "epoch": 0.3508230741187358,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.868869662284851,
      "step": 4170
    },
    {
      "epoch": 0.350991334585939,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8374518156051636,
      "step": 4172
    },
    {
      "epoch": 0.35115959505314226,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8608019351959229,
      "step": 4174
    },
    {
      "epoch": 0.3513278555203455,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0003,
      "loss": 1.8410956859588623,
      "step": 4176
    },
    {
      "epoch": 0.3514961159875487,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8924307823181152,
      "step": 4178
    },
    {
      "epoch": 0.35166437645475196,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.84513521194458,
      "step": 4180
    },
    {
      "epoch": 0.3518326369219552,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8388750553131104,
      "step": 4182
    },
    {
      "epoch": 0.3520008973891584,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8793985843658447,
      "step": 4184
    },
    {
      "epoch": 0.35216915785636166,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.9256815910339355,
      "step": 4186
    },
    {
      "epoch": 0.35233741832356485,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.876656174659729,
      "step": 4188
    },
    {
      "epoch": 0.3525056787907681,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003,
      "loss": 1.9215724468231201,
      "step": 4190
    },
    {
      "epoch": 0.35267393925797136,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8181390762329102,
      "step": 4192
    },
    {
      "epoch": 0.35284219972517455,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.9076271057128906,
      "step": 4194
    },
    {
      "epoch": 0.3530104601923778,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8911964893341064,
      "step": 4196
    },
    {
      "epoch": 0.35317872065958106,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8886250257492065,
      "step": 4198
    },
    {
      "epoch": 0.35334698112678425,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.885007619857788,
      "step": 4200
    },
    {
      "epoch": 0.3535152415939875,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.860021948814392,
      "step": 4202
    },
    {
      "epoch": 0.3536835020611907,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.871260166168213,
      "step": 4204
    },
    {
      "epoch": 0.35385176252839395,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8346993923187256,
      "step": 4206
    },
    {
      "epoch": 0.3540200229955972,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8705899715423584,
      "step": 4208
    },
    {
      "epoch": 0.3541882834628004,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.870624303817749,
      "step": 4210
    },
    {
      "epoch": 0.35435654393000365,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0003,
      "loss": 1.886182188987732,
      "step": 4212
    },
    {
      "epoch": 0.35452480439720685,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003,
      "loss": 1.817286491394043,
      "step": 4214
    },
    {
      "epoch": 0.3546930648644101,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003,
      "loss": 1.843151330947876,
      "step": 4216
    },
    {
      "epoch": 0.35486132533161335,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8760435581207275,
      "step": 4218
    },
    {
      "epoch": 0.35502958579881655,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.8958981037139893,
      "step": 4220
    },
    {
      "epoch": 0.3551978462660198,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8763055801391602,
      "step": 4222
    },
    {
      "epoch": 0.35536610673322305,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.820326805114746,
      "step": 4224
    },
    {
      "epoch": 0.35553436720042625,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8241221904754639,
      "step": 4226
    },
    {
      "epoch": 0.3557026276676295,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8463923931121826,
      "step": 4228
    },
    {
      "epoch": 0.3558708881348327,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8759849071502686,
      "step": 4230
    },
    {
      "epoch": 0.35603914860203595,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8613405227661133,
      "step": 4232
    },
    {
      "epoch": 0.3562074090692392,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.9118133783340454,
      "step": 4234
    },
    {
      "epoch": 0.3563756695364424,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8698644638061523,
      "step": 4236
    },
    {
      "epoch": 0.35654393000364565,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8885126113891602,
      "step": 4238
    },
    {
      "epoch": 0.3567121904708489,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8697750568389893,
      "step": 4240
    },
    {
      "epoch": 0.3568804509380521,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8194037675857544,
      "step": 4242
    },
    {
      "epoch": 0.35704871140525535,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8804447650909424,
      "step": 4244
    },
    {
      "epoch": 0.35721697187245854,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9110827445983887,
      "step": 4246
    },
    {
      "epoch": 0.3573852323396618,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.832970142364502,
      "step": 4248
    },
    {
      "epoch": 0.35755349280686505,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.878289818763733,
      "step": 4250
    },
    {
      "epoch": 0.35772175327406824,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8703241348266602,
      "step": 4252
    },
    {
      "epoch": 0.3578900137412715,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8497341871261597,
      "step": 4254
    },
    {
      "epoch": 0.3580582742084747,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003,
      "loss": 1.830588936805725,
      "step": 4256
    },
    {
      "epoch": 0.35822653467567794,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.857637643814087,
      "step": 4258
    },
    {
      "epoch": 0.3583947951428812,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.9041574001312256,
      "step": 4260
    },
    {
      "epoch": 0.3585630556100844,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.7744847536087036,
      "step": 4262
    },
    {
      "epoch": 0.35873131607728764,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8236842155456543,
      "step": 4264
    },
    {
      "epoch": 0.3588995765444909,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.829077959060669,
      "step": 4266
    },
    {
      "epoch": 0.3590678370116941,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8866846561431885,
      "step": 4268
    },
    {
      "epoch": 0.35923609747889734,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8571105003356934,
      "step": 4270
    },
    {
      "epoch": 0.35940435794610054,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8782892227172852,
      "step": 4272
    },
    {
      "epoch": 0.3595726184133038,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.798835039138794,
      "step": 4274
    },
    {
      "epoch": 0.35974087888050704,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8431508541107178,
      "step": 4276
    },
    {
      "epoch": 0.35990913934771024,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8610095977783203,
      "step": 4278
    },
    {
      "epoch": 0.3600773998149135,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8431506156921387,
      "step": 4280
    },
    {
      "epoch": 0.36024566028211674,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.864717721939087,
      "step": 4282
    },
    {
      "epoch": 0.36041392074931994,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8658771514892578,
      "step": 4284
    },
    {
      "epoch": 0.3605821812165232,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8568891286849976,
      "step": 4286
    },
    {
      "epoch": 0.3607504416837264,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8448036909103394,
      "step": 4288
    },
    {
      "epoch": 0.36091870215092964,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8930323123931885,
      "step": 4290
    },
    {
      "epoch": 0.3610869626181329,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.865108847618103,
      "step": 4292
    },
    {
      "epoch": 0.3612552230853361,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.9002172946929932,
      "step": 4294
    },
    {
      "epoch": 0.36142348355253934,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8452471494674683,
      "step": 4296
    },
    {
      "epoch": 0.3615917440197426,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8971211910247803,
      "step": 4298
    },
    {
      "epoch": 0.3617600044869458,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8451032638549805,
      "step": 4300
    },
    {
      "epoch": 0.36192826495414904,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.860536813735962,
      "step": 4302
    },
    {
      "epoch": 0.36209652542135223,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8372702598571777,
      "step": 4304
    },
    {
      "epoch": 0.3622647858885555,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8120756149291992,
      "step": 4306
    },
    {
      "epoch": 0.36243304635575874,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8306198120117188,
      "step": 4308
    },
    {
      "epoch": 0.36260130682296193,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8404529094696045,
      "step": 4310
    },
    {
      "epoch": 0.3627695672901652,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8514331579208374,
      "step": 4312
    },
    {
      "epoch": 0.3629378277573684,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8299592733383179,
      "step": 4314
    },
    {
      "epoch": 0.36310608822457163,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8316278457641602,
      "step": 4316
    },
    {
      "epoch": 0.3632743486917749,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8346469402313232,
      "step": 4318
    },
    {
      "epoch": 0.3634426091589781,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8541990518569946,
      "step": 4320
    },
    {
      "epoch": 0.36361086962618133,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8548333644866943,
      "step": 4322
    },
    {
      "epoch": 0.3637791300933846,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.856055498123169,
      "step": 4324
    },
    {
      "epoch": 0.3639473905605878,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8420028686523438,
      "step": 4326
    },
    {
      "epoch": 0.36411565102779103,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.789348840713501,
      "step": 4328
    },
    {
      "epoch": 0.36428391149499423,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9047610759735107,
      "step": 4330
    },
    {
      "epoch": 0.3644521719621975,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8302688598632812,
      "step": 4332
    },
    {
      "epoch": 0.36462043242940073,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8713937997817993,
      "step": 4334
    },
    {
      "epoch": 0.3647886928966039,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8275055885314941,
      "step": 4336
    },
    {
      "epoch": 0.3649569533638072,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.830660343170166,
      "step": 4338
    },
    {
      "epoch": 0.36512521383101043,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9004476070404053,
      "step": 4340
    },
    {
      "epoch": 0.3652934742982136,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8527209758758545,
      "step": 4342
    },
    {
      "epoch": 0.3654617347654169,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8637635707855225,
      "step": 4344
    },
    {
      "epoch": 0.3656299952326201,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8736116886138916,
      "step": 4346
    },
    {
      "epoch": 0.3657982556998233,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.835370421409607,
      "step": 4348
    },
    {
      "epoch": 0.3659665161670266,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8609960079193115,
      "step": 4350
    },
    {
      "epoch": 0.3661347766342298,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8754910230636597,
      "step": 4352
    },
    {
      "epoch": 0.366303037101433,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.848397970199585,
      "step": 4354
    },
    {
      "epoch": 0.3664712975686362,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.872434377670288,
      "step": 4356
    },
    {
      "epoch": 0.3666395580358395,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8631129264831543,
      "step": 4358
    },
    {
      "epoch": 0.3668078185030427,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.862055778503418,
      "step": 4360
    },
    {
      "epoch": 0.3669760789702459,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9129970073699951,
      "step": 4362
    },
    {
      "epoch": 0.3671443394374492,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8514277935028076,
      "step": 4364
    },
    {
      "epoch": 0.3673125999046524,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.9031832218170166,
      "step": 4366
    },
    {
      "epoch": 0.3674808603718556,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.848862648010254,
      "step": 4368
    },
    {
      "epoch": 0.3676491208390589,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8735995292663574,
      "step": 4370
    },
    {
      "epoch": 0.36781738130626207,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.844201683998108,
      "step": 4372
    },
    {
      "epoch": 0.3679856417734653,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8694851398468018,
      "step": 4374
    },
    {
      "epoch": 0.3681539022406686,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8224499225616455,
      "step": 4376
    },
    {
      "epoch": 0.36832216270787177,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8128178119659424,
      "step": 4378
    },
    {
      "epoch": 0.368490423175075,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8011059761047363,
      "step": 4380
    },
    {
      "epoch": 0.3686586836422783,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9612146615982056,
      "step": 4382
    },
    {
      "epoch": 0.36882694410948147,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8327138423919678,
      "step": 4384
    },
    {
      "epoch": 0.3689952045766847,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8678789138793945,
      "step": 4386
    },
    {
      "epoch": 0.3691634650438879,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9200565814971924,
      "step": 4388
    },
    {
      "epoch": 0.36933172551109117,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8437559604644775,
      "step": 4390
    },
    {
      "epoch": 0.3694999859782944,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8826746940612793,
      "step": 4392
    },
    {
      "epoch": 0.3696682464454976,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9213488101959229,
      "step": 4394
    },
    {
      "epoch": 0.36983650691270087,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8561475276947021,
      "step": 4396
    },
    {
      "epoch": 0.37000476737990406,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.866892695426941,
      "step": 4398
    },
    {
      "epoch": 0.3701730278471073,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8548835515975952,
      "step": 4400
    },
    {
      "epoch": 0.37034128831431057,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.7662588357925415,
      "step": 4402
    },
    {
      "epoch": 0.37050954878151376,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8592026233673096,
      "step": 4404
    },
    {
      "epoch": 0.370677809248717,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.890699028968811,
      "step": 4406
    },
    {
      "epoch": 0.37084606971592027,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.846623182296753,
      "step": 4408
    },
    {
      "epoch": 0.37101433018312346,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8920495510101318,
      "step": 4410
    },
    {
      "epoch": 0.3711825906503267,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8361213207244873,
      "step": 4412
    },
    {
      "epoch": 0.3713508511175299,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.846116065979004,
      "step": 4414
    },
    {
      "epoch": 0.37151911158473316,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8252332210540771,
      "step": 4416
    },
    {
      "epoch": 0.3716873720519364,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9050908088684082,
      "step": 4418
    },
    {
      "epoch": 0.3718556325191396,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.823420524597168,
      "step": 4420
    },
    {
      "epoch": 0.37202389298634286,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8413670063018799,
      "step": 4422
    },
    {
      "epoch": 0.3721921534535461,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 1.8193632364273071,
      "step": 4424
    },
    {
      "epoch": 0.3723604139207493,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8389079570770264,
      "step": 4426
    },
    {
      "epoch": 0.37252867438795256,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8357930183410645,
      "step": 4428
    },
    {
      "epoch": 0.37269693485515576,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8615214824676514,
      "step": 4430
    },
    {
      "epoch": 0.372865195322359,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8593577146530151,
      "step": 4432
    },
    {
      "epoch": 0.37303345578956226,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8403033018112183,
      "step": 4434
    },
    {
      "epoch": 0.37320171625676546,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8928072452545166,
      "step": 4436
    },
    {
      "epoch": 0.3733699767239687,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8315999507904053,
      "step": 4438
    },
    {
      "epoch": 0.3735382371911719,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003,
      "loss": 1.8425942659378052,
      "step": 4440
    },
    {
      "epoch": 0.37370649765837516,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8644912242889404,
      "step": 4442
    },
    {
      "epoch": 0.3738747581255784,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.8560198545455933,
      "step": 4444
    },
    {
      "epoch": 0.3740430185927816,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8572437763214111,
      "step": 4446
    },
    {
      "epoch": 0.37421127905998486,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8357101678848267,
      "step": 4448
    },
    {
      "epoch": 0.3743795395271881,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.868126392364502,
      "step": 4450
    },
    {
      "epoch": 0.3745477999943913,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8136754035949707,
      "step": 4452
    },
    {
      "epoch": 0.37471606046159456,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8570661544799805,
      "step": 4454
    },
    {
      "epoch": 0.37488432092879775,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.886802077293396,
      "step": 4456
    },
    {
      "epoch": 0.375052581396001,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8707962036132812,
      "step": 4458
    },
    {
      "epoch": 0.37522084186320426,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.815312147140503,
      "step": 4460
    },
    {
      "epoch": 0.37538910233040745,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8640072345733643,
      "step": 4462
    },
    {
      "epoch": 0.3755573627976107,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9307156801223755,
      "step": 4464
    },
    {
      "epoch": 0.37572562326481396,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8569077253341675,
      "step": 4466
    },
    {
      "epoch": 0.37589388373201715,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.815842866897583,
      "step": 4468
    },
    {
      "epoch": 0.3760621441992204,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.898988962173462,
      "step": 4470
    },
    {
      "epoch": 0.3762304046664236,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9069769382476807,
      "step": 4472
    },
    {
      "epoch": 0.37639866513362685,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8389627933502197,
      "step": 4474
    },
    {
      "epoch": 0.3765669256008301,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8581401109695435,
      "step": 4476
    },
    {
      "epoch": 0.3767351860680333,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8521840572357178,
      "step": 4478
    },
    {
      "epoch": 0.37690344653523655,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8483788967132568,
      "step": 4480
    },
    {
      "epoch": 0.3770717070024398,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8686535358428955,
      "step": 4482
    },
    {
      "epoch": 0.377239967469643,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.836313247680664,
      "step": 4484
    },
    {
      "epoch": 0.37740822793684625,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8637442588806152,
      "step": 4486
    },
    {
      "epoch": 0.37757648840404945,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8761377334594727,
      "step": 4488
    },
    {
      "epoch": 0.3777447488712527,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8924980163574219,
      "step": 4490
    },
    {
      "epoch": 0.37791300933845595,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8605420589447021,
      "step": 4492
    },
    {
      "epoch": 0.37808126980565915,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.86045241355896,
      "step": 4494
    },
    {
      "epoch": 0.3782495302728624,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8641526699066162,
      "step": 4496
    },
    {
      "epoch": 0.3784177907400656,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8831942081451416,
      "step": 4498
    },
    {
      "epoch": 0.37858605120726885,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.855360507965088,
      "step": 4500
    },
    {
      "epoch": 0.3787543116744721,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8675832748413086,
      "step": 4502
    },
    {
      "epoch": 0.3789225721416753,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8975629806518555,
      "step": 4504
    },
    {
      "epoch": 0.37909083260887855,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8475128412246704,
      "step": 4506
    },
    {
      "epoch": 0.3792590930760818,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8539950847625732,
      "step": 4508
    },
    {
      "epoch": 0.379427353543285,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8739655017852783,
      "step": 4510
    },
    {
      "epoch": 0.37959561401048825,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.912283182144165,
      "step": 4512
    },
    {
      "epoch": 0.37976387447769144,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9028751850128174,
      "step": 4514
    },
    {
      "epoch": 0.3799321349448947,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8576847314834595,
      "step": 4516
    },
    {
      "epoch": 0.38010039541209795,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8255009651184082,
      "step": 4518
    },
    {
      "epoch": 0.38026865587930114,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.811992883682251,
      "step": 4520
    },
    {
      "epoch": 0.3804369163465044,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8497283458709717,
      "step": 4522
    },
    {
      "epoch": 0.38060517681370765,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.9177377223968506,
      "step": 4524
    },
    {
      "epoch": 0.38077343728091084,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8679866790771484,
      "step": 4526
    },
    {
      "epoch": 0.3809416977481141,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8835303783416748,
      "step": 4528
    },
    {
      "epoch": 0.3811099582153173,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8791511058807373,
      "step": 4530
    },
    {
      "epoch": 0.38127821868252054,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.919022798538208,
      "step": 4532
    },
    {
      "epoch": 0.3814464791497238,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8440027236938477,
      "step": 4534
    },
    {
      "epoch": 0.381614739616927,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0003,
      "loss": 1.8808114528656006,
      "step": 4536
    },
    {
      "epoch": 0.38178300008413024,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.851625680923462,
      "step": 4538
    },
    {
      "epoch": 0.38195126055133344,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.856220006942749,
      "step": 4540
    },
    {
      "epoch": 0.3821195210185367,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8564107418060303,
      "step": 4542
    },
    {
      "epoch": 0.38228778148573994,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.9167286157608032,
      "step": 4544
    },
    {
      "epoch": 0.38245604195294314,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.830963134765625,
      "step": 4546
    },
    {
      "epoch": 0.3826243024201464,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8850429058074951,
      "step": 4548
    },
    {
      "epoch": 0.38279256288734964,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8752870559692383,
      "step": 4550
    },
    {
      "epoch": 0.38296082335455284,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8368842601776123,
      "step": 4552
    },
    {
      "epoch": 0.3831290838217561,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.856083869934082,
      "step": 4554
    },
    {
      "epoch": 0.3832973442889593,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8794653415679932,
      "step": 4556
    },
    {
      "epoch": 0.38346560475616254,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.841750144958496,
      "step": 4558
    },
    {
      "epoch": 0.3836338652233658,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9026367664337158,
      "step": 4560
    },
    {
      "epoch": 0.383802125690569,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8421781063079834,
      "step": 4562
    },
    {
      "epoch": 0.38397038615777224,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8425276279449463,
      "step": 4564
    },
    {
      "epoch": 0.3841386466249755,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8457183837890625,
      "step": 4566
    },
    {
      "epoch": 0.3843069070921787,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8797197341918945,
      "step": 4568
    },
    {
      "epoch": 0.38447516755938194,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8145244121551514,
      "step": 4570
    },
    {
      "epoch": 0.38464342802658513,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.7860229015350342,
      "step": 4572
    },
    {
      "epoch": 0.3848116884937884,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8531218767166138,
      "step": 4574
    },
    {
      "epoch": 0.38497994896099164,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8404762744903564,
      "step": 4576
    },
    {
      "epoch": 0.38514820942819483,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8330953121185303,
      "step": 4578
    },
    {
      "epoch": 0.3853164698953981,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8996546268463135,
      "step": 4580
    },
    {
      "epoch": 0.3854847303626013,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.830596685409546,
      "step": 4582
    },
    {
      "epoch": 0.38565299082980453,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.893639326095581,
      "step": 4584
    },
    {
      "epoch": 0.3858212512970078,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8650248050689697,
      "step": 4586
    },
    {
      "epoch": 0.385989511764211,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9030554294586182,
      "step": 4588
    },
    {
      "epoch": 0.38615777223141423,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8899855613708496,
      "step": 4590
    },
    {
      "epoch": 0.3863260326986175,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.7952806949615479,
      "step": 4592
    },
    {
      "epoch": 0.3864942931658207,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8598946332931519,
      "step": 4594
    },
    {
      "epoch": 0.38666255363302393,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.893235206604004,
      "step": 4596
    },
    {
      "epoch": 0.3868308141002271,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.7838258743286133,
      "step": 4598
    },
    {
      "epoch": 0.3869990745674304,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.836708903312683,
      "step": 4600
    },
    {
      "epoch": 0.38716733503463363,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8134894371032715,
      "step": 4602
    },
    {
      "epoch": 0.3873355955018368,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.86518394947052,
      "step": 4604
    },
    {
      "epoch": 0.3875038559690401,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8280094861984253,
      "step": 4606
    },
    {
      "epoch": 0.38767211643624333,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.907750129699707,
      "step": 4608
    },
    {
      "epoch": 0.3878403769034465,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8739879131317139,
      "step": 4610
    },
    {
      "epoch": 0.3880086373706498,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8808284997940063,
      "step": 4612
    },
    {
      "epoch": 0.388176897837853,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.852165937423706,
      "step": 4614
    },
    {
      "epoch": 0.3883451583050562,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.881048321723938,
      "step": 4616
    },
    {
      "epoch": 0.3885134187722595,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8551963567733765,
      "step": 4618
    },
    {
      "epoch": 0.3886816792394627,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8690121173858643,
      "step": 4620
    },
    {
      "epoch": 0.3888499397066659,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8595457077026367,
      "step": 4622
    },
    {
      "epoch": 0.3890182001738691,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8525559902191162,
      "step": 4624
    },
    {
      "epoch": 0.3891864606410724,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8493109941482544,
      "step": 4626
    },
    {
      "epoch": 0.3893547211082756,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.818713903427124,
      "step": 4628
    },
    {
      "epoch": 0.3895229815754788,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8869866132736206,
      "step": 4630
    },
    {
      "epoch": 0.3896912420426821,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.856992244720459,
      "step": 4632
    },
    {
      "epoch": 0.3898595025098853,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8784983158111572,
      "step": 4634
    },
    {
      "epoch": 0.3900277629770885,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0003,
      "loss": 1.819836139678955,
      "step": 4636
    },
    {
      "epoch": 0.3901960234442918,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8386175632476807,
      "step": 4638
    },
    {
      "epoch": 0.39036428391149497,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.886093020439148,
      "step": 4640
    },
    {
      "epoch": 0.3905325443786982,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8293447494506836,
      "step": 4642
    },
    {
      "epoch": 0.3907008048459015,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8541921377182007,
      "step": 4644
    },
    {
      "epoch": 0.39086906531310467,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8652796745300293,
      "step": 4646
    },
    {
      "epoch": 0.3910373257803079,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8483929634094238,
      "step": 4648
    },
    {
      "epoch": 0.3912055862475112,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.854693055152893,
      "step": 4650
    },
    {
      "epoch": 0.39137384671471437,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.873854160308838,
      "step": 4652
    },
    {
      "epoch": 0.3915421071819176,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8768669366836548,
      "step": 4654
    },
    {
      "epoch": 0.3917103676491208,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8217474222183228,
      "step": 4656
    },
    {
      "epoch": 0.39187862811632407,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8607592582702637,
      "step": 4658
    },
    {
      "epoch": 0.3920468885835273,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.931011438369751,
      "step": 4660
    },
    {
      "epoch": 0.3922151490507305,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8376190662384033,
      "step": 4662
    },
    {
      "epoch": 0.39238340951793377,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8500524759292603,
      "step": 4664
    },
    {
      "epoch": 0.392551669985137,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.849021077156067,
      "step": 4666
    },
    {
      "epoch": 0.3927199304523402,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8586902618408203,
      "step": 4668
    },
    {
      "epoch": 0.39288819091954347,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8521685600280762,
      "step": 4670
    },
    {
      "epoch": 0.39305645138674666,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.819950819015503,
      "step": 4672
    },
    {
      "epoch": 0.3932247118539499,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.868251919746399,
      "step": 4674
    },
    {
      "epoch": 0.39339297232115317,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9228920936584473,
      "step": 4676
    },
    {
      "epoch": 0.39356123278835636,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8226845264434814,
      "step": 4678
    },
    {
      "epoch": 0.3937294932555596,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.831139326095581,
      "step": 4680
    },
    {
      "epoch": 0.3938977537227628,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8721537590026855,
      "step": 4682
    },
    {
      "epoch": 0.39406601418996606,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8529844284057617,
      "step": 4684
    },
    {
      "epoch": 0.3942342746571693,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003,
      "loss": 1.8452403545379639,
      "step": 4686
    },
    {
      "epoch": 0.3944025351243725,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.83066725730896,
      "step": 4688
    },
    {
      "epoch": 0.39457079559157576,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8282550573349,
      "step": 4690
    },
    {
      "epoch": 0.394739056058779,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.848116159439087,
      "step": 4692
    },
    {
      "epoch": 0.3949073165259822,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8791431188583374,
      "step": 4694
    },
    {
      "epoch": 0.39507557699318546,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.882152795791626,
      "step": 4696
    },
    {
      "epoch": 0.39524383746038866,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003,
      "loss": 1.8694283962249756,
      "step": 4698
    },
    {
      "epoch": 0.3954120979275919,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8666706085205078,
      "step": 4700
    },
    {
      "epoch": 0.39558035839479516,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8148703575134277,
      "step": 4702
    },
    {
      "epoch": 0.39574861886199836,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8247408866882324,
      "step": 4704
    },
    {
      "epoch": 0.3959168793292016,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8786565065383911,
      "step": 4706
    },
    {
      "epoch": 0.39608513979640486,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.853364109992981,
      "step": 4708
    },
    {
      "epoch": 0.39625340026360806,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8746209144592285,
      "step": 4710
    },
    {
      "epoch": 0.3964216607308113,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8242783546447754,
      "step": 4712
    },
    {
      "epoch": 0.3965899211980145,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8371549844741821,
      "step": 4714
    },
    {
      "epoch": 0.39675818166521776,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.7918696403503418,
      "step": 4716
    },
    {
      "epoch": 0.396926442132421,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8196983337402344,
      "step": 4718
    },
    {
      "epoch": 0.3970947025996242,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8171378374099731,
      "step": 4720
    },
    {
      "epoch": 0.39726296306682746,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8597605228424072,
      "step": 4722
    },
    {
      "epoch": 0.39743122353403065,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8424537181854248,
      "step": 4724
    },
    {
      "epoch": 0.3975994840012339,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8094136714935303,
      "step": 4726
    },
    {
      "epoch": 0.39776774446843716,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.850404977798462,
      "step": 4728
    },
    {
      "epoch": 0.39793600493564035,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8381047248840332,
      "step": 4730
    },
    {
      "epoch": 0.3981042654028436,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.8592967987060547,
      "step": 4732
    },
    {
      "epoch": 0.39827252587004686,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8533565998077393,
      "step": 4734
    },
    {
      "epoch": 0.39844078633725005,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8584849834442139,
      "step": 4736
    },
    {
      "epoch": 0.3986090468044533,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8420213460922241,
      "step": 4738
    },
    {
      "epoch": 0.3987773072716565,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8453192710876465,
      "step": 4740
    },
    {
      "epoch": 0.39894556773885975,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8347465991973877,
      "step": 4742
    },
    {
      "epoch": 0.399113828206063,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9128386974334717,
      "step": 4744
    },
    {
      "epoch": 0.3992820886732662,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8935587406158447,
      "step": 4746
    },
    {
      "epoch": 0.39945034914046945,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.7936766147613525,
      "step": 4748
    },
    {
      "epoch": 0.3996186096076727,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8881841897964478,
      "step": 4750
    },
    {
      "epoch": 0.3997868700748759,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8318591117858887,
      "step": 4752
    },
    {
      "epoch": 0.39995513054207915,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.887589931488037,
      "step": 4754
    },
    {
      "epoch": 0.40012339100928235,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.84372079372406,
      "step": 4756
    },
    {
      "epoch": 0.40012339100928235,
      "eval_loss": 1.8230888843536377,
      "eval_runtime": 539.9826,
      "eval_samples_per_second": 103.083,
      "eval_steps_per_second": 12.886,
      "step": 4756
    },
    {
      "epoch": 0.4002916514764856,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8204362392425537,
      "step": 4758
    },
    {
      "epoch": 0.40045991194368885,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.846121072769165,
      "step": 4760
    },
    {
      "epoch": 0.40062817241089205,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8437823057174683,
      "step": 4762
    },
    {
      "epoch": 0.4007964328780953,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8411822319030762,
      "step": 4764
    },
    {
      "epoch": 0.4009646933452985,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8601104021072388,
      "step": 4766
    },
    {
      "epoch": 0.40113295381250175,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.858335256576538,
      "step": 4768
    },
    {
      "epoch": 0.401301214279705,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8325574398040771,
      "step": 4770
    },
    {
      "epoch": 0.4014694747469082,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.80202054977417,
      "step": 4772
    },
    {
      "epoch": 0.40163773521411145,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8657782077789307,
      "step": 4774
    },
    {
      "epoch": 0.4018059956813147,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8360810279846191,
      "step": 4776
    },
    {
      "epoch": 0.4019742561485179,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.845475196838379,
      "step": 4778
    },
    {
      "epoch": 0.40214251661572115,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8819303512573242,
      "step": 4780
    },
    {
      "epoch": 0.40231077708292434,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.880976915359497,
      "step": 4782
    },
    {
      "epoch": 0.4024790375501276,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.856764554977417,
      "step": 4784
    },
    {
      "epoch": 0.40264729801733085,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8782942295074463,
      "step": 4786
    },
    {
      "epoch": 0.40281555848453404,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8895702362060547,
      "step": 4788
    },
    {
      "epoch": 0.4029838189517373,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003,
      "loss": 1.8697278499603271,
      "step": 4790
    },
    {
      "epoch": 0.40315207941894055,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8844778537750244,
      "step": 4792
    },
    {
      "epoch": 0.40332033988614374,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8470101356506348,
      "step": 4794
    },
    {
      "epoch": 0.403488600353347,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8538079261779785,
      "step": 4796
    },
    {
      "epoch": 0.4036568608205502,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8900970220565796,
      "step": 4798
    },
    {
      "epoch": 0.40382512128775344,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.817952275276184,
      "step": 4800
    },
    {
      "epoch": 0.4039933817549567,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.872812271118164,
      "step": 4802
    },
    {
      "epoch": 0.4041616422221599,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8414208889007568,
      "step": 4804
    },
    {
      "epoch": 0.40432990268936314,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003,
      "loss": 1.8481721878051758,
      "step": 4806
    },
    {
      "epoch": 0.40449816315656634,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8112316131591797,
      "step": 4808
    },
    {
      "epoch": 0.4046664236237696,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8322917222976685,
      "step": 4810
    },
    {
      "epoch": 0.40483468409097284,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8688421249389648,
      "step": 4812
    },
    {
      "epoch": 0.40500294455817604,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8335949182510376,
      "step": 4814
    },
    {
      "epoch": 0.4051712050253793,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9111578464508057,
      "step": 4816
    },
    {
      "epoch": 0.40533946549258254,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.832777500152588,
      "step": 4818
    },
    {
      "epoch": 0.40550772595978574,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8160617351531982,
      "step": 4820
    },
    {
      "epoch": 0.405675986426989,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8185396194458008,
      "step": 4822
    },
    {
      "epoch": 0.4058442468941922,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8617687225341797,
      "step": 4824
    },
    {
      "epoch": 0.40601250736139544,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8558969497680664,
      "step": 4826
    },
    {
      "epoch": 0.4061807678285987,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8508987426757812,
      "step": 4828
    },
    {
      "epoch": 0.4063490282958019,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8264577388763428,
      "step": 4830
    },
    {
      "epoch": 0.40651728876300514,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8642231225967407,
      "step": 4832
    },
    {
      "epoch": 0.4066855492302084,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8364858627319336,
      "step": 4834
    },
    {
      "epoch": 0.4068538096974116,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003,
      "loss": 1.8646694421768188,
      "step": 4836
    },
    {
      "epoch": 0.40702207016461484,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.7808382511138916,
      "step": 4838
    },
    {
      "epoch": 0.40719033063181803,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8825949430465698,
      "step": 4840
    },
    {
      "epoch": 0.4073585910990213,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8663816452026367,
      "step": 4842
    },
    {
      "epoch": 0.40752685156622453,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8326139450073242,
      "step": 4844
    },
    {
      "epoch": 0.40769511203342773,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8989505767822266,
      "step": 4846
    },
    {
      "epoch": 0.407863372500631,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9054362773895264,
      "step": 4848
    },
    {
      "epoch": 0.40803163296783423,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8205817937850952,
      "step": 4850
    },
    {
      "epoch": 0.40819989343503743,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8383492231369019,
      "step": 4852
    },
    {
      "epoch": 0.4083681539022407,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.857367753982544,
      "step": 4854
    },
    {
      "epoch": 0.4085364143694439,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8395779132843018,
      "step": 4856
    },
    {
      "epoch": 0.40870467483664713,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8510242700576782,
      "step": 4858
    },
    {
      "epoch": 0.4088729353038504,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8753166198730469,
      "step": 4860
    },
    {
      "epoch": 0.4090411957710536,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.867672324180603,
      "step": 4862
    },
    {
      "epoch": 0.40920945623825683,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8283164501190186,
      "step": 4864
    },
    {
      "epoch": 0.40937771670546,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8423609733581543,
      "step": 4866
    },
    {
      "epoch": 0.4095459771726633,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8585090637207031,
      "step": 4868
    },
    {
      "epoch": 0.40971423763986653,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.875443458557129,
      "step": 4870
    },
    {
      "epoch": 0.4098824981070697,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.867616891860962,
      "step": 4872
    },
    {
      "epoch": 0.410050758574273,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8418222665786743,
      "step": 4874
    },
    {
      "epoch": 0.41021901904147623,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.840241551399231,
      "step": 4876
    },
    {
      "epoch": 0.4103872795086794,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8789030313491821,
      "step": 4878
    },
    {
      "epoch": 0.4105555399758827,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8290181159973145,
      "step": 4880
    },
    {
      "epoch": 0.4107238004430859,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.845411777496338,
      "step": 4882
    },
    {
      "epoch": 0.4108920609102891,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8055942058563232,
      "step": 4884
    },
    {
      "epoch": 0.4110603213774924,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.875232219696045,
      "step": 4886
    },
    {
      "epoch": 0.4112285818446956,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8679254055023193,
      "step": 4888
    },
    {
      "epoch": 0.4113968423118988,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003,
      "loss": 1.8536744117736816,
      "step": 4890
    },
    {
      "epoch": 0.4115651027791021,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8352406024932861,
      "step": 4892
    },
    {
      "epoch": 0.4117333632463053,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8426165580749512,
      "step": 4894
    },
    {
      "epoch": 0.4119016237135085,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8410687446594238,
      "step": 4896
    },
    {
      "epoch": 0.4120698841807117,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8151347637176514,
      "step": 4898
    },
    {
      "epoch": 0.412238144647915,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8536725044250488,
      "step": 4900
    },
    {
      "epoch": 0.4124064051151182,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.901092529296875,
      "step": 4902
    },
    {
      "epoch": 0.4125746655823214,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8517358303070068,
      "step": 4904
    },
    {
      "epoch": 0.41274292604952467,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.869875192642212,
      "step": 4906
    },
    {
      "epoch": 0.41291118651672787,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8387603759765625,
      "step": 4908
    },
    {
      "epoch": 0.4130794469839311,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8322443962097168,
      "step": 4910
    },
    {
      "epoch": 0.41324770745113437,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8559932708740234,
      "step": 4912
    },
    {
      "epoch": 0.41341596791833757,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8649612665176392,
      "step": 4914
    },
    {
      "epoch": 0.4135842283855408,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.837500810623169,
      "step": 4916
    },
    {
      "epoch": 0.41375248885274407,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8196276426315308,
      "step": 4918
    },
    {
      "epoch": 0.41392074931994727,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8636773824691772,
      "step": 4920
    },
    {
      "epoch": 0.4140890097871505,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8311206102371216,
      "step": 4922
    },
    {
      "epoch": 0.4142572702543537,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8478431701660156,
      "step": 4924
    },
    {
      "epoch": 0.41442553072155697,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8167450428009033,
      "step": 4926
    },
    {
      "epoch": 0.4145937911887602,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8699976205825806,
      "step": 4928
    },
    {
      "epoch": 0.4147620516559634,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8707785606384277,
      "step": 4930
    },
    {
      "epoch": 0.41493031212316667,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.866210699081421,
      "step": 4932
    },
    {
      "epoch": 0.4150985725903699,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8654046058654785,
      "step": 4934
    },
    {
      "epoch": 0.4152668330575731,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8272833824157715,
      "step": 4936
    },
    {
      "epoch": 0.41543509352477637,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8218203783035278,
      "step": 4938
    },
    {
      "epoch": 0.41560335399197956,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8182874917984009,
      "step": 4940
    },
    {
      "epoch": 0.4157716144591828,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.864674687385559,
      "step": 4942
    },
    {
      "epoch": 0.41593987492638607,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8923275470733643,
      "step": 4944
    },
    {
      "epoch": 0.41610813539358926,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8524607419967651,
      "step": 4946
    },
    {
      "epoch": 0.4162763958607925,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8177255392074585,
      "step": 4948
    },
    {
      "epoch": 0.4164446563279957,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8578803539276123,
      "step": 4950
    },
    {
      "epoch": 0.41661291679519896,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8308703899383545,
      "step": 4952
    },
    {
      "epoch": 0.4167811772624022,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8820980787277222,
      "step": 4954
    },
    {
      "epoch": 0.4169494377296054,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9149575233459473,
      "step": 4956
    },
    {
      "epoch": 0.41711769819680866,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8804335594177246,
      "step": 4958
    },
    {
      "epoch": 0.4172859586640119,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8903619050979614,
      "step": 4960
    },
    {
      "epoch": 0.4174542191312151,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8898673057556152,
      "step": 4962
    },
    {
      "epoch": 0.41762247959841836,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.867480993270874,
      "step": 4964
    },
    {
      "epoch": 0.41779074006562156,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.809093952178955,
      "step": 4966
    },
    {
      "epoch": 0.4179590005328248,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8365998268127441,
      "step": 4968
    },
    {
      "epoch": 0.41812726100002806,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8458998203277588,
      "step": 4970
    },
    {
      "epoch": 0.41829552146723126,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8719065189361572,
      "step": 4972
    },
    {
      "epoch": 0.4184637819344345,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8163866996765137,
      "step": 4974
    },
    {
      "epoch": 0.41863204240163776,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8576040267944336,
      "step": 4976
    },
    {
      "epoch": 0.41880030286884096,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8511459827423096,
      "step": 4978
    },
    {
      "epoch": 0.4189685633360442,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.869091510772705,
      "step": 4980
    },
    {
      "epoch": 0.4191368238032474,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8328659534454346,
      "step": 4982
    },
    {
      "epoch": 0.41930508427045066,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8554950952529907,
      "step": 4984
    },
    {
      "epoch": 0.4194733447376539,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.866563320159912,
      "step": 4986
    },
    {
      "epoch": 0.4196416052048571,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.834499478340149,
      "step": 4988
    },
    {
      "epoch": 0.41980986567206036,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8757071495056152,
      "step": 4990
    },
    {
      "epoch": 0.41997812613926355,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8289196491241455,
      "step": 4992
    },
    {
      "epoch": 0.4201463866064668,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8158631324768066,
      "step": 4994
    },
    {
      "epoch": 0.42031464707367006,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.854619026184082,
      "step": 4996
    },
    {
      "epoch": 0.42048290754087325,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.847519874572754,
      "step": 4998
    },
    {
      "epoch": 0.4206511680080765,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8518385887145996,
      "step": 5000
    },
    {
      "epoch": 0.42081942847527976,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.844959020614624,
      "step": 5002
    },
    {
      "epoch": 0.42098768894248295,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8425681591033936,
      "step": 5004
    },
    {
      "epoch": 0.4211559494096862,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.861499547958374,
      "step": 5006
    },
    {
      "epoch": 0.4213242098768894,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003,
      "loss": 1.878270149230957,
      "step": 5008
    },
    {
      "epoch": 0.42149247034409265,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8242779970169067,
      "step": 5010
    },
    {
      "epoch": 0.4216607308112959,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.894126296043396,
      "step": 5012
    },
    {
      "epoch": 0.4218289912784991,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.9282348155975342,
      "step": 5014
    },
    {
      "epoch": 0.42199725174570235,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8782057762145996,
      "step": 5016
    },
    {
      "epoch": 0.4221655122129056,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8652466535568237,
      "step": 5018
    },
    {
      "epoch": 0.4223337726801088,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.7806001901626587,
      "step": 5020
    },
    {
      "epoch": 0.42250203314731205,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8154668807983398,
      "step": 5022
    },
    {
      "epoch": 0.42267029361451525,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8470356464385986,
      "step": 5024
    },
    {
      "epoch": 0.4228385540817185,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8761954307556152,
      "step": 5026
    },
    {
      "epoch": 0.42300681454892175,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8938063383102417,
      "step": 5028
    },
    {
      "epoch": 0.42317507501612495,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8245296478271484,
      "step": 5030
    },
    {
      "epoch": 0.4233433354833282,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8385615348815918,
      "step": 5032
    },
    {
      "epoch": 0.42351159595053145,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.857254981994629,
      "step": 5034
    },
    {
      "epoch": 0.42367985641773465,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8885836601257324,
      "step": 5036
    },
    {
      "epoch": 0.4238481168849379,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.786860704421997,
      "step": 5038
    },
    {
      "epoch": 0.4240163773521411,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.7976672649383545,
      "step": 5040
    },
    {
      "epoch": 0.42418463781934435,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8312358856201172,
      "step": 5042
    },
    {
      "epoch": 0.4243528982865476,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8637996912002563,
      "step": 5044
    },
    {
      "epoch": 0.4245211587537508,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8086175918579102,
      "step": 5046
    },
    {
      "epoch": 0.42468941922095405,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8522164821624756,
      "step": 5048
    },
    {
      "epoch": 0.42485767968815724,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.813535451889038,
      "step": 5050
    },
    {
      "epoch": 0.4250259401553605,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.879728078842163,
      "step": 5052
    },
    {
      "epoch": 0.42519420062256374,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8417651653289795,
      "step": 5054
    },
    {
      "epoch": 0.42536246108976694,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8877267837524414,
      "step": 5056
    },
    {
      "epoch": 0.4255307215569702,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8777426481246948,
      "step": 5058
    },
    {
      "epoch": 0.42569898202417344,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8308894634246826,
      "step": 5060
    },
    {
      "epoch": 0.42586724249137664,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9045298099517822,
      "step": 5062
    },
    {
      "epoch": 0.4260355029585799,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.9297128915786743,
      "step": 5064
    },
    {
      "epoch": 0.4262037634257831,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.856966257095337,
      "step": 5066
    },
    {
      "epoch": 0.42637202389298634,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8534338474273682,
      "step": 5068
    },
    {
      "epoch": 0.4265402843601896,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8340699672698975,
      "step": 5070
    },
    {
      "epoch": 0.4267085448273928,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.829815149307251,
      "step": 5072
    },
    {
      "epoch": 0.42687680529459604,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8797483444213867,
      "step": 5074
    },
    {
      "epoch": 0.4270450657617993,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.890449047088623,
      "step": 5076
    },
    {
      "epoch": 0.4272133262290025,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8720083236694336,
      "step": 5078
    },
    {
      "epoch": 0.42738158669620574,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8542966842651367,
      "step": 5080
    },
    {
      "epoch": 0.42754984716340894,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8529342412948608,
      "step": 5082
    },
    {
      "epoch": 0.4277181076306122,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8670660257339478,
      "step": 5084
    },
    {
      "epoch": 0.42788636809781544,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8118938207626343,
      "step": 5086
    },
    {
      "epoch": 0.42805462856501864,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.9051192998886108,
      "step": 5088
    },
    {
      "epoch": 0.4282228890322219,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8169578313827515,
      "step": 5090
    },
    {
      "epoch": 0.4283911494994251,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.851027488708496,
      "step": 5092
    },
    {
      "epoch": 0.42855940996662834,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8524409532546997,
      "step": 5094
    },
    {
      "epoch": 0.4287276704338316,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9234716892242432,
      "step": 5096
    },
    {
      "epoch": 0.4288959309010348,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8409911394119263,
      "step": 5098
    },
    {
      "epoch": 0.42906419136823803,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8734869956970215,
      "step": 5100
    },
    {
      "epoch": 0.4292324518354413,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8349225521087646,
      "step": 5102
    },
    {
      "epoch": 0.4294007123026445,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8290014266967773,
      "step": 5104
    },
    {
      "epoch": 0.42956897276984773,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8859753608703613,
      "step": 5106
    },
    {
      "epoch": 0.42973723323705093,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0003,
      "loss": 1.8500820398330688,
      "step": 5108
    },
    {
      "epoch": 0.4299054937042542,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003,
      "loss": 1.8487718105316162,
      "step": 5110
    },
    {
      "epoch": 0.43007375417145743,
      "grad_norm": 1.59375,
      "learning_rate": 0.0003,
      "loss": 1.9827172756195068,
      "step": 5112
    },
    {
      "epoch": 0.43024201463866063,
      "grad_norm": 2.734375,
      "learning_rate": 0.0003,
      "loss": 2.0592446327209473,
      "step": 5114
    },
    {
      "epoch": 0.4304102751058639,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0003,
      "loss": 2.0257461071014404,
      "step": 5116
    },
    {
      "epoch": 0.43057853557306713,
      "grad_norm": 0.71875,
      "learning_rate": 0.0003,
      "loss": 2.045689582824707,
      "step": 5118
    },
    {
      "epoch": 0.43074679604027033,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0003,
      "loss": 1.9712104797363281,
      "step": 5120
    },
    {
      "epoch": 0.4309150565074736,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0003,
      "loss": 1.994288682937622,
      "step": 5122
    },
    {
      "epoch": 0.4310833169746768,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0003,
      "loss": 1.9515581130981445,
      "step": 5124
    },
    {
      "epoch": 0.43125157744188003,
      "grad_norm": 0.234375,
      "learning_rate": 0.0003,
      "loss": 1.9778897762298584,
      "step": 5126
    },
    {
      "epoch": 0.4314198379090833,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003,
      "loss": 1.978132724761963,
      "step": 5128
    },
    {
      "epoch": 0.4315880983762865,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003,
      "loss": 1.9485716819763184,
      "step": 5130
    },
    {
      "epoch": 0.43175635884348973,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.9412591457366943,
      "step": 5132
    },
    {
      "epoch": 0.4319246193106929,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.9120941162109375,
      "step": 5134
    },
    {
      "epoch": 0.4320928797778962,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8900258541107178,
      "step": 5136
    },
    {
      "epoch": 0.43226114024509943,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.89882230758667,
      "step": 5138
    },
    {
      "epoch": 0.4324294007123026,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.940398097038269,
      "step": 5140
    },
    {
      "epoch": 0.4325976611795059,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.9075276851654053,
      "step": 5142
    },
    {
      "epoch": 0.43276592164670913,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8847041130065918,
      "step": 5144
    },
    {
      "epoch": 0.4329341821139123,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.909827709197998,
      "step": 5146
    },
    {
      "epoch": 0.4331024425811156,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9140040874481201,
      "step": 5148
    },
    {
      "epoch": 0.4332707030483188,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.865070104598999,
      "step": 5150
    },
    {
      "epoch": 0.433438963515522,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8763169050216675,
      "step": 5152
    },
    {
      "epoch": 0.4336072239827253,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8946975469589233,
      "step": 5154
    },
    {
      "epoch": 0.4337754844499285,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8523186445236206,
      "step": 5156
    },
    {
      "epoch": 0.4339437449171317,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9016883373260498,
      "step": 5158
    },
    {
      "epoch": 0.434112005384335,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9254387617111206,
      "step": 5160
    },
    {
      "epoch": 0.43428026585153817,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9053566455841064,
      "step": 5162
    },
    {
      "epoch": 0.4344485263187414,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8829138278961182,
      "step": 5164
    },
    {
      "epoch": 0.4346167867859446,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.867133378982544,
      "step": 5166
    },
    {
      "epoch": 0.43478504725314787,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8555493354797363,
      "step": 5168
    },
    {
      "epoch": 0.4349533077203511,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003,
      "loss": 1.8949813842773438,
      "step": 5170
    },
    {
      "epoch": 0.4351215681875543,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.8651165962219238,
      "step": 5172
    },
    {
      "epoch": 0.43528982865475757,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8275350332260132,
      "step": 5174
    },
    {
      "epoch": 0.4354580891219608,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8872151374816895,
      "step": 5176
    },
    {
      "epoch": 0.435626349589164,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.7930710315704346,
      "step": 5178
    },
    {
      "epoch": 0.43579461005636727,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8489298820495605,
      "step": 5180
    },
    {
      "epoch": 0.43596287052357047,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.9133497476577759,
      "step": 5182
    },
    {
      "epoch": 0.4361311309907737,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8998597860336304,
      "step": 5184
    },
    {
      "epoch": 0.43629939145797697,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8871028423309326,
      "step": 5186
    },
    {
      "epoch": 0.43646765192518017,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8783812522888184,
      "step": 5188
    },
    {
      "epoch": 0.4366359123923834,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8929307460784912,
      "step": 5190
    },
    {
      "epoch": 0.4368041728595866,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8280582427978516,
      "step": 5192
    },
    {
      "epoch": 0.43697243332678987,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8683323860168457,
      "step": 5194
    },
    {
      "epoch": 0.4371406937939931,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8150275945663452,
      "step": 5196
    },
    {
      "epoch": 0.4373089542611963,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.889363408088684,
      "step": 5198
    },
    {
      "epoch": 0.43747721472839957,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.838592529296875,
      "step": 5200
    },
    {
      "epoch": 0.4376454751956028,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.942219853401184,
      "step": 5202
    },
    {
      "epoch": 0.437813735662806,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8942644596099854,
      "step": 5204
    },
    {
      "epoch": 0.43798199613000927,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.9254858493804932,
      "step": 5206
    },
    {
      "epoch": 0.43815025659721246,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8604316711425781,
      "step": 5208
    },
    {
      "epoch": 0.4383185170644157,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8187631368637085,
      "step": 5210
    },
    {
      "epoch": 0.43848677753161897,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.870947003364563,
      "step": 5212
    },
    {
      "epoch": 0.43865503799882216,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8725275993347168,
      "step": 5214
    },
    {
      "epoch": 0.4388232984660254,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8604071140289307,
      "step": 5216
    },
    {
      "epoch": 0.43899155893322866,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.9041380882263184,
      "step": 5218
    },
    {
      "epoch": 0.43915981940043186,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8613276481628418,
      "step": 5220
    },
    {
      "epoch": 0.4393280798676351,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.846703290939331,
      "step": 5222
    },
    {
      "epoch": 0.4394963403348383,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8695021867752075,
      "step": 5224
    },
    {
      "epoch": 0.43966460080204156,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0003,
      "loss": 1.8777716159820557,
      "step": 5226
    },
    {
      "epoch": 0.4398328612692448,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.883468747138977,
      "step": 5228
    },
    {
      "epoch": 0.440001121736448,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8789043426513672,
      "step": 5230
    },
    {
      "epoch": 0.44016938220365126,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8369626998901367,
      "step": 5232
    },
    {
      "epoch": 0.44033764267085446,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8328208923339844,
      "step": 5234
    },
    {
      "epoch": 0.4405059031380577,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.912573218345642,
      "step": 5236
    },
    {
      "epoch": 0.44067416360526096,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8574092388153076,
      "step": 5238
    },
    {
      "epoch": 0.44084242407246416,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.7845542430877686,
      "step": 5240
    },
    {
      "epoch": 0.4410106845396674,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8542823791503906,
      "step": 5242
    },
    {
      "epoch": 0.44117894500687066,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8798987865447998,
      "step": 5244
    },
    {
      "epoch": 0.44134720547407386,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8249592781066895,
      "step": 5246
    },
    {
      "epoch": 0.4415154659412771,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8868303298950195,
      "step": 5248
    },
    {
      "epoch": 0.4416837264084803,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8406586647033691,
      "step": 5250
    },
    {
      "epoch": 0.44185198687568356,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.927297830581665,
      "step": 5252
    },
    {
      "epoch": 0.4420202473428868,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.908623218536377,
      "step": 5254
    },
    {
      "epoch": 0.44218850781009,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8577653169631958,
      "step": 5256
    },
    {
      "epoch": 0.44235676827729326,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.867362380027771,
      "step": 5258
    },
    {
      "epoch": 0.4425250287444965,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.854485034942627,
      "step": 5260
    },
    {
      "epoch": 0.4426932892116997,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.910072922706604,
      "step": 5262
    },
    {
      "epoch": 0.44286154967890295,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.892256259918213,
      "step": 5264
    },
    {
      "epoch": 0.44302981014610615,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8976072072982788,
      "step": 5266
    },
    {
      "epoch": 0.4431980706133094,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8686752319335938,
      "step": 5268
    },
    {
      "epoch": 0.44336633108051265,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8326730728149414,
      "step": 5270
    },
    {
      "epoch": 0.44353459154771585,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8354787826538086,
      "step": 5272
    },
    {
      "epoch": 0.4437028520149191,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0003,
      "loss": 1.8357237577438354,
      "step": 5274
    },
    {
      "epoch": 0.4438711124821223,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.9084913730621338,
      "step": 5276
    },
    {
      "epoch": 0.44403937294932555,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8632546663284302,
      "step": 5278
    },
    {
      "epoch": 0.4442076334165288,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8828327655792236,
      "step": 5280
    },
    {
      "epoch": 0.444375893883732,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8218305110931396,
      "step": 5282
    },
    {
      "epoch": 0.44454415435093525,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8766192197799683,
      "step": 5284
    },
    {
      "epoch": 0.4447124148181385,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8943684101104736,
      "step": 5286
    },
    {
      "epoch": 0.4448806752853417,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8638417720794678,
      "step": 5288
    },
    {
      "epoch": 0.44504893575254495,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8109736442565918,
      "step": 5290
    },
    {
      "epoch": 0.44521719621974815,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8365658521652222,
      "step": 5292
    },
    {
      "epoch": 0.4453854566869514,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.82277512550354,
      "step": 5294
    },
    {
      "epoch": 0.44555371715415465,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.849085807800293,
      "step": 5296
    },
    {
      "epoch": 0.44572197762135785,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8984382152557373,
      "step": 5298
    },
    {
      "epoch": 0.4458902380885611,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.808199405670166,
      "step": 5300
    },
    {
      "epoch": 0.44605849855576435,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.808314323425293,
      "step": 5302
    },
    {
      "epoch": 0.44622675902296755,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0003,
      "loss": 1.8006086349487305,
      "step": 5304
    },
    {
      "epoch": 0.4463950194901708,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8930131196975708,
      "step": 5306
    },
    {
      "epoch": 0.446563279957374,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.841903805732727,
      "step": 5308
    },
    {
      "epoch": 0.44673154042457724,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.841183066368103,
      "step": 5310
    },
    {
      "epoch": 0.4468998008917805,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8476521968841553,
      "step": 5312
    },
    {
      "epoch": 0.4470680613589837,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.843845248222351,
      "step": 5314
    },
    {
      "epoch": 0.44723632182618694,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8571088314056396,
      "step": 5316
    },
    {
      "epoch": 0.44740458229339014,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8497490882873535,
      "step": 5318
    },
    {
      "epoch": 0.4475728427605934,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.851987600326538,
      "step": 5320
    },
    {
      "epoch": 0.44774110322779664,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8916161060333252,
      "step": 5322
    },
    {
      "epoch": 0.44790936369499984,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8544466495513916,
      "step": 5324
    },
    {
      "epoch": 0.4480776241622031,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8566031455993652,
      "step": 5326
    },
    {
      "epoch": 0.44824588462940634,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8512012958526611,
      "step": 5328
    },
    {
      "epoch": 0.44841414509660954,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8151590824127197,
      "step": 5330
    },
    {
      "epoch": 0.4485824055638128,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8652503490447998,
      "step": 5332
    },
    {
      "epoch": 0.448750666031016,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.9295122623443604,
      "step": 5334
    },
    {
      "epoch": 0.44891892649821924,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8500659465789795,
      "step": 5336
    },
    {
      "epoch": 0.4490871869654225,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8964040279388428,
      "step": 5338
    },
    {
      "epoch": 0.4492554474326257,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8929883241653442,
      "step": 5340
    },
    {
      "epoch": 0.44942370789982894,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.821066975593567,
      "step": 5342
    },
    {
      "epoch": 0.4495919683670322,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.883821964263916,
      "step": 5344
    },
    {
      "epoch": 0.4497602288342354,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8910892009735107,
      "step": 5346
    },
    {
      "epoch": 0.44992848930143864,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8954401016235352,
      "step": 5348
    },
    {
      "epoch": 0.45009674976864184,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8348100185394287,
      "step": 5350
    },
    {
      "epoch": 0.4502650102358451,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0003,
      "loss": 1.8322563171386719,
      "step": 5352
    },
    {
      "epoch": 0.45043327070304834,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.9253042936325073,
      "step": 5354
    },
    {
      "epoch": 0.45060153117025153,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8728573322296143,
      "step": 5356
    },
    {
      "epoch": 0.4507697916374548,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.886345624923706,
      "step": 5358
    },
    {
      "epoch": 0.45093805210465804,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8891772031784058,
      "step": 5360
    },
    {
      "epoch": 0.45110631257186123,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8464785814285278,
      "step": 5362
    },
    {
      "epoch": 0.4512745730390645,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8374913930892944,
      "step": 5364
    },
    {
      "epoch": 0.4514428335062677,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8767975568771362,
      "step": 5366
    },
    {
      "epoch": 0.45161109397347093,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8869285583496094,
      "step": 5368
    },
    {
      "epoch": 0.4517793544406742,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8910973072052002,
      "step": 5370
    },
    {
      "epoch": 0.4519476149078774,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8354072570800781,
      "step": 5372
    },
    {
      "epoch": 0.45211587537508063,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8639553785324097,
      "step": 5374
    },
    {
      "epoch": 0.45228413584228383,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8325058221817017,
      "step": 5376
    },
    {
      "epoch": 0.4524523963094871,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8477115631103516,
      "step": 5378
    },
    {
      "epoch": 0.45262065677669033,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8281694650650024,
      "step": 5380
    },
    {
      "epoch": 0.45278891724389353,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8478338718414307,
      "step": 5382
    },
    {
      "epoch": 0.4529571777110968,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.879786491394043,
      "step": 5384
    },
    {
      "epoch": 0.45312543817830003,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.816864252090454,
      "step": 5386
    },
    {
      "epoch": 0.45329369864550323,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8762390613555908,
      "step": 5388
    },
    {
      "epoch": 0.4534619591127065,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8783068656921387,
      "step": 5390
    },
    {
      "epoch": 0.4536302195799097,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8628743886947632,
      "step": 5392
    },
    {
      "epoch": 0.45379848004711293,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.835709810256958,
      "step": 5394
    },
    {
      "epoch": 0.4539667405143162,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8660805225372314,
      "step": 5396
    },
    {
      "epoch": 0.4541350009815194,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8374173641204834,
      "step": 5398
    },
    {
      "epoch": 0.45430326144872263,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8869471549987793,
      "step": 5400
    },
    {
      "epoch": 0.4544715219159259,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8876157999038696,
      "step": 5402
    },
    {
      "epoch": 0.4546397823831291,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8535182476043701,
      "step": 5404
    },
    {
      "epoch": 0.45480804285033233,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.854591727256775,
      "step": 5406
    },
    {
      "epoch": 0.4549763033175355,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.7866885662078857,
      "step": 5408
    },
    {
      "epoch": 0.4551445637847388,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8622195720672607,
      "step": 5410
    },
    {
      "epoch": 0.455312824251942,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8847346305847168,
      "step": 5412
    },
    {
      "epoch": 0.4554810847191452,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0003,
      "loss": 1.867648720741272,
      "step": 5414
    },
    {
      "epoch": 0.4556493451863485,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8111932277679443,
      "step": 5416
    },
    {
      "epoch": 0.45581760565355167,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8750970363616943,
      "step": 5418
    },
    {
      "epoch": 0.4559858661207549,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.87458074092865,
      "step": 5420
    },
    {
      "epoch": 0.4561541265879582,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8572351932525635,
      "step": 5422
    },
    {
      "epoch": 0.45632238705516137,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8309375047683716,
      "step": 5424
    },
    {
      "epoch": 0.4564906475223646,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8660783767700195,
      "step": 5426
    },
    {
      "epoch": 0.4566589079895679,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8558696508407593,
      "step": 5428
    },
    {
      "epoch": 0.45682716845677107,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8706345558166504,
      "step": 5430
    },
    {
      "epoch": 0.4569954289239743,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8299366235733032,
      "step": 5432
    },
    {
      "epoch": 0.4571636893911775,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0003,
      "loss": 1.8203890323638916,
      "step": 5434
    },
    {
      "epoch": 0.45733194985838077,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8871244192123413,
      "step": 5436
    },
    {
      "epoch": 0.457500210325584,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8730669021606445,
      "step": 5438
    },
    {
      "epoch": 0.4576684707927872,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.9056720733642578,
      "step": 5440
    },
    {
      "epoch": 0.45783673125999047,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.893676519393921,
      "step": 5442
    },
    {
      "epoch": 0.4580049917271937,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8050551414489746,
      "step": 5444
    },
    {
      "epoch": 0.4581732521943969,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8398643732070923,
      "step": 5446
    },
    {
      "epoch": 0.45834151266160017,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8958936929702759,
      "step": 5448
    },
    {
      "epoch": 0.45850977312880337,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8647483587265015,
      "step": 5450
    },
    {
      "epoch": 0.4586780335960066,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8325027227401733,
      "step": 5452
    },
    {
      "epoch": 0.45884629406320987,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8365510702133179,
      "step": 5454
    },
    {
      "epoch": 0.45901455453041307,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8245388269424438,
      "step": 5456
    },
    {
      "epoch": 0.4591828149976163,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.914254903793335,
      "step": 5458
    },
    {
      "epoch": 0.4593510754648195,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.838131308555603,
      "step": 5460
    },
    {
      "epoch": 0.45951933593202277,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8815791606903076,
      "step": 5462
    },
    {
      "epoch": 0.459687596399226,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8461024761199951,
      "step": 5464
    },
    {
      "epoch": 0.4598558568664292,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8493854999542236,
      "step": 5466
    },
    {
      "epoch": 0.46002411733363247,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.9045684337615967,
      "step": 5468
    },
    {
      "epoch": 0.4601923778008357,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8968528509140015,
      "step": 5470
    },
    {
      "epoch": 0.4603606382680389,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8336641788482666,
      "step": 5472
    },
    {
      "epoch": 0.46052889873524216,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8683714866638184,
      "step": 5474
    },
    {
      "epoch": 0.46069715920244536,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8367663621902466,
      "step": 5476
    },
    {
      "epoch": 0.4608654196696486,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003,
      "loss": 1.8581829071044922,
      "step": 5478
    },
    {
      "epoch": 0.46103368013685186,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.827835202217102,
      "step": 5480
    },
    {
      "epoch": 0.46120194060405506,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8320121765136719,
      "step": 5482
    },
    {
      "epoch": 0.4613702010712583,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.859140157699585,
      "step": 5484
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8830540180206299,
      "step": 5486
    },
    {
      "epoch": 0.46170672200566476,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8665298223495483,
      "step": 5488
    },
    {
      "epoch": 0.461874982472868,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8618652820587158,
      "step": 5490
    },
    {
      "epoch": 0.4620432429400712,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8440864086151123,
      "step": 5492
    },
    {
      "epoch": 0.46221150340727446,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8725669384002686,
      "step": 5494
    },
    {
      "epoch": 0.4623797638744777,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8717167377471924,
      "step": 5496
    },
    {
      "epoch": 0.4625480243416809,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8240416049957275,
      "step": 5498
    },
    {
      "epoch": 0.46271628480888416,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8555972576141357,
      "step": 5500
    },
    {
      "epoch": 0.46288454527608736,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8353780508041382,
      "step": 5502
    },
    {
      "epoch": 0.4630528057432906,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8919432163238525,
      "step": 5504
    },
    {
      "epoch": 0.46322106621049386,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8880773782730103,
      "step": 5506
    },
    {
      "epoch": 0.46338932667769706,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.85103440284729,
      "step": 5508
    },
    {
      "epoch": 0.4635575871449003,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8618991374969482,
      "step": 5510
    },
    {
      "epoch": 0.46372584761210356,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8568588495254517,
      "step": 5512
    },
    {
      "epoch": 0.46389410807930676,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9132959842681885,
      "step": 5514
    },
    {
      "epoch": 0.46406236854651,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8426105976104736,
      "step": 5516
    },
    {
      "epoch": 0.4642306290137132,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.825427770614624,
      "step": 5518
    },
    {
      "epoch": 0.46439888948091645,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.819854974746704,
      "step": 5520
    },
    {
      "epoch": 0.4645671499481197,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.829544186592102,
      "step": 5522
    },
    {
      "epoch": 0.4647354104153229,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.7694684267044067,
      "step": 5524
    },
    {
      "epoch": 0.46490367088252615,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8583238124847412,
      "step": 5526
    },
    {
      "epoch": 0.4650719313497294,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8406904935836792,
      "step": 5528
    },
    {
      "epoch": 0.4652401918169326,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.879626989364624,
      "step": 5530
    },
    {
      "epoch": 0.46540845228413585,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0003,
      "loss": 1.829959750175476,
      "step": 5532
    },
    {
      "epoch": 0.46557671275133905,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8407893180847168,
      "step": 5534
    },
    {
      "epoch": 0.4657449732185423,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8583735227584839,
      "step": 5536
    },
    {
      "epoch": 0.46591323368574555,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8474990129470825,
      "step": 5538
    },
    {
      "epoch": 0.46608149415294875,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8565003871917725,
      "step": 5540
    },
    {
      "epoch": 0.466249754620152,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.850085735321045,
      "step": 5542
    },
    {
      "epoch": 0.46641801508735525,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8244609832763672,
      "step": 5544
    },
    {
      "epoch": 0.46658627555455845,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8510874509811401,
      "step": 5546
    },
    {
      "epoch": 0.4667545360217617,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8366775512695312,
      "step": 5548
    },
    {
      "epoch": 0.4669227964889649,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.833579182624817,
      "step": 5550
    },
    {
      "epoch": 0.46709105695616815,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8925817012786865,
      "step": 5552
    },
    {
      "epoch": 0.4672593174233714,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8280715942382812,
      "step": 5554
    },
    {
      "epoch": 0.4674275778905746,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8275939226150513,
      "step": 5556
    },
    {
      "epoch": 0.46759583835777785,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8864471912384033,
      "step": 5558
    },
    {
      "epoch": 0.46776409882498104,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9023559093475342,
      "step": 5560
    },
    {
      "epoch": 0.4679323592921843,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8453950881958008,
      "step": 5562
    },
    {
      "epoch": 0.46810061975938755,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8342480659484863,
      "step": 5564
    },
    {
      "epoch": 0.46826888022659074,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.88969087600708,
      "step": 5566
    },
    {
      "epoch": 0.468437140693794,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8701565265655518,
      "step": 5568
    },
    {
      "epoch": 0.46860540116099725,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8376394510269165,
      "step": 5570
    },
    {
      "epoch": 0.46877366162820044,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.891002893447876,
      "step": 5572
    },
    {
      "epoch": 0.4689419220954037,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8623106479644775,
      "step": 5574
    },
    {
      "epoch": 0.4691101825626069,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.857226848602295,
      "step": 5576
    },
    {
      "epoch": 0.46927844302981014,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8863677978515625,
      "step": 5578
    },
    {
      "epoch": 0.4694467034970134,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8292392492294312,
      "step": 5580
    },
    {
      "epoch": 0.4696149639642166,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.7746028900146484,
      "step": 5582
    },
    {
      "epoch": 0.46978322443141984,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8695952892303467,
      "step": 5584
    },
    {
      "epoch": 0.4699514848986231,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.9011144638061523,
      "step": 5586
    },
    {
      "epoch": 0.4701197453658263,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8679029941558838,
      "step": 5588
    },
    {
      "epoch": 0.47028800583302954,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.884771466255188,
      "step": 5590
    },
    {
      "epoch": 0.47045626630023274,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.872206449508667,
      "step": 5592
    },
    {
      "epoch": 0.470624526767436,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8835575580596924,
      "step": 5594
    },
    {
      "epoch": 0.47079278723463924,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8582497835159302,
      "step": 5596
    },
    {
      "epoch": 0.47096104770184244,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8160433769226074,
      "step": 5598
    },
    {
      "epoch": 0.4711293081690457,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8928561210632324,
      "step": 5600
    },
    {
      "epoch": 0.4712975686362489,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8467472791671753,
      "step": 5602
    },
    {
      "epoch": 0.47146582910345214,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8584730625152588,
      "step": 5604
    },
    {
      "epoch": 0.4716340895706554,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8476479053497314,
      "step": 5606
    },
    {
      "epoch": 0.4718023500378586,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8302178382873535,
      "step": 5608
    },
    {
      "epoch": 0.47197061050506184,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8277881145477295,
      "step": 5610
    },
    {
      "epoch": 0.4721388709722651,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8456183671951294,
      "step": 5612
    },
    {
      "epoch": 0.4723071314394683,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8541700839996338,
      "step": 5614
    },
    {
      "epoch": 0.47247539190667154,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8398160934448242,
      "step": 5616
    },
    {
      "epoch": 0.47264365237387473,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0003,
      "loss": 1.8133342266082764,
      "step": 5618
    },
    {
      "epoch": 0.472811912841078,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.860898733139038,
      "step": 5620
    },
    {
      "epoch": 0.47298017330828124,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.802849292755127,
      "step": 5622
    },
    {
      "epoch": 0.47314843377548443,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.9002327919006348,
      "step": 5624
    },
    {
      "epoch": 0.4733166942426877,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.86818528175354,
      "step": 5626
    },
    {
      "epoch": 0.47348495470989094,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8620264530181885,
      "step": 5628
    },
    {
      "epoch": 0.47365321517709413,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.834169864654541,
      "step": 5630
    },
    {
      "epoch": 0.4738214756442974,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8645024299621582,
      "step": 5632
    },
    {
      "epoch": 0.4739897361115006,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.834524154663086,
      "step": 5634
    },
    {
      "epoch": 0.47415799657870383,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8522653579711914,
      "step": 5636
    },
    {
      "epoch": 0.4743262570459071,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8749196529388428,
      "step": 5638
    },
    {
      "epoch": 0.4744945175131103,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8450909852981567,
      "step": 5640
    },
    {
      "epoch": 0.47466277798031353,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.840910792350769,
      "step": 5642
    },
    {
      "epoch": 0.47483103844751673,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8347790241241455,
      "step": 5644
    },
    {
      "epoch": 0.47499929891472,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.86039137840271,
      "step": 5646
    },
    {
      "epoch": 0.47516755938192323,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8909393548965454,
      "step": 5648
    },
    {
      "epoch": 0.47533581984912643,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8372044563293457,
      "step": 5650
    },
    {
      "epoch": 0.4755040803163297,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8271564245224,
      "step": 5652
    },
    {
      "epoch": 0.47567234078353293,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.854422926902771,
      "step": 5654
    },
    {
      "epoch": 0.47584060125073613,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8710482120513916,
      "step": 5656
    },
    {
      "epoch": 0.4760088617179394,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8534765243530273,
      "step": 5658
    },
    {
      "epoch": 0.4761771221851426,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8364768028259277,
      "step": 5660
    },
    {
      "epoch": 0.47634538265234583,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8845243453979492,
      "step": 5662
    },
    {
      "epoch": 0.4765136431195491,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0003,
      "loss": 1.8159297704696655,
      "step": 5664
    },
    {
      "epoch": 0.4766819035867523,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8614778518676758,
      "step": 5666
    },
    {
      "epoch": 0.4768501640539555,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8420078754425049,
      "step": 5668
    },
    {
      "epoch": 0.4770184245211588,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8595819473266602,
      "step": 5670
    },
    {
      "epoch": 0.477186684988362,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8386173248291016,
      "step": 5672
    },
    {
      "epoch": 0.4773549454555652,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.7993652820587158,
      "step": 5674
    },
    {
      "epoch": 0.4775232059227684,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0003,
      "loss": 1.8501596450805664,
      "step": 5676
    },
    {
      "epoch": 0.4776914663899717,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.9040745496749878,
      "step": 5678
    },
    {
      "epoch": 0.4778597268571749,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8567783832550049,
      "step": 5680
    },
    {
      "epoch": 0.4780279873243781,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8400416374206543,
      "step": 5682
    },
    {
      "epoch": 0.4781962477915814,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8234291076660156,
      "step": 5684
    },
    {
      "epoch": 0.47836450825878457,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8380521535873413,
      "step": 5686
    },
    {
      "epoch": 0.4785327687259878,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.814704179763794,
      "step": 5688
    },
    {
      "epoch": 0.4787010291931911,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.877873420715332,
      "step": 5690
    },
    {
      "epoch": 0.47886928966039427,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8395812511444092,
      "step": 5692
    },
    {
      "epoch": 0.4790375501275975,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8681962490081787,
      "step": 5694
    },
    {
      "epoch": 0.4792058105948008,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.84226655960083,
      "step": 5696
    },
    {
      "epoch": 0.47937407106200397,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8289748430252075,
      "step": 5698
    },
    {
      "epoch": 0.4795423315292072,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8431401252746582,
      "step": 5700
    },
    {
      "epoch": 0.4797105919964104,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8721810579299927,
      "step": 5702
    },
    {
      "epoch": 0.47987885246361367,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8515934944152832,
      "step": 5704
    },
    {
      "epoch": 0.4800471129308169,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8184764385223389,
      "step": 5706
    },
    {
      "epoch": 0.4802153733980201,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0003,
      "loss": 1.8939648866653442,
      "step": 5708
    },
    {
      "epoch": 0.48038363386522337,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9085718393325806,
      "step": 5710
    },
    {
      "epoch": 0.4805518943324266,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8339669704437256,
      "step": 5712
    },
    {
      "epoch": 0.4807201547996298,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.9185373783111572,
      "step": 5714
    },
    {
      "epoch": 0.48088841526683307,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8517284393310547,
      "step": 5716
    },
    {
      "epoch": 0.48105667573403627,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8024563789367676,
      "step": 5718
    },
    {
      "epoch": 0.4812249362012395,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8862398862838745,
      "step": 5720
    },
    {
      "epoch": 0.48139319666844277,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.877204179763794,
      "step": 5722
    },
    {
      "epoch": 0.48156145713564596,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8264895677566528,
      "step": 5724
    },
    {
      "epoch": 0.4817297176028492,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.850816011428833,
      "step": 5726
    },
    {
      "epoch": 0.48189797807005247,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8538262844085693,
      "step": 5728
    },
    {
      "epoch": 0.48206623853725566,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.87843656539917,
      "step": 5730
    },
    {
      "epoch": 0.4822344990044589,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8753070831298828,
      "step": 5732
    },
    {
      "epoch": 0.4824027594716621,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8342270851135254,
      "step": 5734
    },
    {
      "epoch": 0.48257101993886536,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8503787517547607,
      "step": 5736
    },
    {
      "epoch": 0.4827392804060686,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8319487571716309,
      "step": 5738
    },
    {
      "epoch": 0.4829075408732718,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.88318932056427,
      "step": 5740
    },
    {
      "epoch": 0.48307580134047506,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8558893203735352,
      "step": 5742
    },
    {
      "epoch": 0.48324406180767826,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8673282861709595,
      "step": 5744
    },
    {
      "epoch": 0.4834123222748815,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8713464736938477,
      "step": 5746
    },
    {
      "epoch": 0.48358058274208476,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8134338855743408,
      "step": 5748
    },
    {
      "epoch": 0.48374884320928796,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.886000394821167,
      "step": 5750
    },
    {
      "epoch": 0.4839171036764912,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8789587020874023,
      "step": 5752
    },
    {
      "epoch": 0.48408536414369446,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8519771099090576,
      "step": 5754
    },
    {
      "epoch": 0.48425362461089766,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8951332569122314,
      "step": 5756
    },
    {
      "epoch": 0.4844218850781009,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.874114751815796,
      "step": 5758
    },
    {
      "epoch": 0.4845901455453041,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8479435443878174,
      "step": 5760
    },
    {
      "epoch": 0.48475840601250736,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8735382556915283,
      "step": 5762
    },
    {
      "epoch": 0.4849266664797106,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.831144094467163,
      "step": 5764
    },
    {
      "epoch": 0.4850949269469138,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8643746376037598,
      "step": 5766
    },
    {
      "epoch": 0.48526318741411706,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8252935409545898,
      "step": 5768
    },
    {
      "epoch": 0.4854314478813203,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8689696788787842,
      "step": 5770
    },
    {
      "epoch": 0.4855997083485235,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.9023597240447998,
      "step": 5772
    },
    {
      "epoch": 0.48576796881572676,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.9361011981964111,
      "step": 5774
    },
    {
      "epoch": 0.48593622928292995,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8680464029312134,
      "step": 5776
    },
    {
      "epoch": 0.4861044897501332,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8628551959991455,
      "step": 5778
    },
    {
      "epoch": 0.48627275021733646,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8901357650756836,
      "step": 5780
    },
    {
      "epoch": 0.48644101068453965,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8653843402862549,
      "step": 5782
    },
    {
      "epoch": 0.4866092711517429,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.849057674407959,
      "step": 5784
    },
    {
      "epoch": 0.4867775316189461,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8717219829559326,
      "step": 5786
    },
    {
      "epoch": 0.48694579208614935,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8757715225219727,
      "step": 5788
    },
    {
      "epoch": 0.4871140525533526,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8626258373260498,
      "step": 5790
    },
    {
      "epoch": 0.4872823130205558,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.853032112121582,
      "step": 5792
    },
    {
      "epoch": 0.48745057348775905,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8313617706298828,
      "step": 5794
    },
    {
      "epoch": 0.4876188339549623,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8821523189544678,
      "step": 5796
    },
    {
      "epoch": 0.4877870944221655,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.815430998802185,
      "step": 5798
    },
    {
      "epoch": 0.48795535488936875,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8326473236083984,
      "step": 5800
    },
    {
      "epoch": 0.48812361535657195,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8851494789123535,
      "step": 5802
    },
    {
      "epoch": 0.4882918758237752,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8203041553497314,
      "step": 5804
    },
    {
      "epoch": 0.48846013629097845,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8072082996368408,
      "step": 5806
    },
    {
      "epoch": 0.48862839675818165,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.7703933715820312,
      "step": 5808
    },
    {
      "epoch": 0.4887966572253849,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8899893760681152,
      "step": 5810
    },
    {
      "epoch": 0.48896491769258815,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8366014957427979,
      "step": 5812
    },
    {
      "epoch": 0.48913317815979135,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8425061702728271,
      "step": 5814
    },
    {
      "epoch": 0.4893014386269946,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8172311782836914,
      "step": 5816
    },
    {
      "epoch": 0.4894696990941978,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8537228107452393,
      "step": 5818
    },
    {
      "epoch": 0.48963795956140105,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8179311752319336,
      "step": 5820
    },
    {
      "epoch": 0.4898062200286043,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8391342163085938,
      "step": 5822
    },
    {
      "epoch": 0.4899744804958075,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.832735538482666,
      "step": 5824
    },
    {
      "epoch": 0.49014274096301075,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8255679607391357,
      "step": 5826
    },
    {
      "epoch": 0.49031100143021394,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.786961317062378,
      "step": 5828
    },
    {
      "epoch": 0.4904792618974172,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8640729188919067,
      "step": 5830
    },
    {
      "epoch": 0.49064752236462045,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8705401420593262,
      "step": 5832
    },
    {
      "epoch": 0.49081578283182364,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8516933917999268,
      "step": 5834
    },
    {
      "epoch": 0.4909840432990269,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8122504949569702,
      "step": 5836
    },
    {
      "epoch": 0.49115230376623015,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8975062370300293,
      "step": 5838
    },
    {
      "epoch": 0.49132056423343334,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8317869901657104,
      "step": 5840
    },
    {
      "epoch": 0.4914888247006366,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9248077869415283,
      "step": 5842
    },
    {
      "epoch": 0.4916570851678398,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8262948989868164,
      "step": 5844
    },
    {
      "epoch": 0.49182534563504304,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8054454326629639,
      "step": 5846
    },
    {
      "epoch": 0.4919936061022463,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.908886194229126,
      "step": 5848
    },
    {
      "epoch": 0.4921618665694495,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8104603290557861,
      "step": 5850
    },
    {
      "epoch": 0.49233012703665274,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8106988668441772,
      "step": 5852
    },
    {
      "epoch": 0.492498387503856,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.862583875656128,
      "step": 5854
    },
    {
      "epoch": 0.4926666479710592,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9091814756393433,
      "step": 5856
    },
    {
      "epoch": 0.49283490843826244,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8456982374191284,
      "step": 5858
    },
    {
      "epoch": 0.49300316890546564,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.860119342803955,
      "step": 5860
    },
    {
      "epoch": 0.4931714293726689,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.7965960502624512,
      "step": 5862
    },
    {
      "epoch": 0.49333968983987214,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8338327407836914,
      "step": 5864
    },
    {
      "epoch": 0.49350795030707534,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8305420875549316,
      "step": 5866
    },
    {
      "epoch": 0.4936762107742786,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8464574813842773,
      "step": 5868
    },
    {
      "epoch": 0.4938444712414818,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8370134830474854,
      "step": 5870
    },
    {
      "epoch": 0.49401273170868504,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8443732261657715,
      "step": 5872
    },
    {
      "epoch": 0.4941809921758883,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.860398292541504,
      "step": 5874
    },
    {
      "epoch": 0.4943492526430915,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8717961311340332,
      "step": 5876
    },
    {
      "epoch": 0.49451751311029474,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.829319715499878,
      "step": 5878
    },
    {
      "epoch": 0.494685773577498,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.842812180519104,
      "step": 5880
    },
    {
      "epoch": 0.4948540340447012,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8268134593963623,
      "step": 5882
    },
    {
      "epoch": 0.49502229451190444,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8349741697311401,
      "step": 5884
    },
    {
      "epoch": 0.49519055497910763,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.818479061126709,
      "step": 5886
    },
    {
      "epoch": 0.4953588154463109,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8321267366409302,
      "step": 5888
    },
    {
      "epoch": 0.49552707591351414,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.831193447113037,
      "step": 5890
    },
    {
      "epoch": 0.49569533638071733,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.865218162536621,
      "step": 5892
    },
    {
      "epoch": 0.4958635968479206,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.864062786102295,
      "step": 5894
    },
    {
      "epoch": 0.49603185731512384,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8668606281280518,
      "step": 5896
    },
    {
      "epoch": 0.49620011778232703,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8481688499450684,
      "step": 5898
    },
    {
      "epoch": 0.4963683782495303,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8357467651367188,
      "step": 5900
    },
    {
      "epoch": 0.4965366387167335,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8224523067474365,
      "step": 5902
    },
    {
      "epoch": 0.49670489918393673,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.856367588043213,
      "step": 5904
    },
    {
      "epoch": 0.49687315965114,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8236010074615479,
      "step": 5906
    },
    {
      "epoch": 0.4970414201183432,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.876981496810913,
      "step": 5908
    },
    {
      "epoch": 0.49720968058554643,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.872553825378418,
      "step": 5910
    },
    {
      "epoch": 0.4973779410527497,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8919224739074707,
      "step": 5912
    },
    {
      "epoch": 0.4975462015199529,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.839388132095337,
      "step": 5914
    },
    {
      "epoch": 0.49771446198715613,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.7895276546478271,
      "step": 5916
    },
    {
      "epoch": 0.4978827224543593,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8755145072937012,
      "step": 5918
    },
    {
      "epoch": 0.4980509829215626,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.886352777481079,
      "step": 5920
    },
    {
      "epoch": 0.49821924338876583,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8424843549728394,
      "step": 5922
    },
    {
      "epoch": 0.498387503855969,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8833050727844238,
      "step": 5924
    },
    {
      "epoch": 0.4985557643231723,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8046488761901855,
      "step": 5926
    },
    {
      "epoch": 0.4987240247903755,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8874001502990723,
      "step": 5928
    },
    {
      "epoch": 0.4988922852575787,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8102126121520996,
      "step": 5930
    },
    {
      "epoch": 0.499060545724782,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8670287132263184,
      "step": 5932
    },
    {
      "epoch": 0.4992288061919852,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8272700309753418,
      "step": 5934
    },
    {
      "epoch": 0.4993970666591884,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.830449104309082,
      "step": 5936
    },
    {
      "epoch": 0.4995653271263917,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8690992593765259,
      "step": 5938
    },
    {
      "epoch": 0.4997335875935949,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8558512926101685,
      "step": 5940
    },
    {
      "epoch": 0.4999018480607981,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8200737237930298,
      "step": 5942
    },
    {
      "epoch": 0.5000701085280014,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8385487794876099,
      "step": 5944
    },
    {
      "epoch": 0.5002383689952046,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8305362462997437,
      "step": 5946
    },
    {
      "epoch": 0.5004066294624078,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.857137680053711,
      "step": 5948
    },
    {
      "epoch": 0.5005748899296111,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8515459299087524,
      "step": 5950
    },
    {
      "epoch": 0.5007431503968143,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8290153741836548,
      "step": 5952
    },
    {
      "epoch": 0.5009114108640175,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.885138750076294,
      "step": 5954
    },
    {
      "epoch": 0.5010796713312208,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.836418628692627,
      "step": 5956
    },
    {
      "epoch": 0.501247931798424,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8904597759246826,
      "step": 5958
    },
    {
      "epoch": 0.5014161922656272,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8763914108276367,
      "step": 5960
    },
    {
      "epoch": 0.5015844527328305,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8308939933776855,
      "step": 5962
    },
    {
      "epoch": 0.5017527132000337,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8333802223205566,
      "step": 5964
    },
    {
      "epoch": 0.5019209736672369,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.850446105003357,
      "step": 5966
    },
    {
      "epoch": 0.5020892341344401,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9122726917266846,
      "step": 5968
    },
    {
      "epoch": 0.5022574946016434,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8519742488861084,
      "step": 5970
    },
    {
      "epoch": 0.5024257550688466,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8170044422149658,
      "step": 5972
    },
    {
      "epoch": 0.5025940155360498,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8271305561065674,
      "step": 5974
    },
    {
      "epoch": 0.5027622760032531,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.810429573059082,
      "step": 5976
    },
    {
      "epoch": 0.5029305364704563,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8355791568756104,
      "step": 5978
    },
    {
      "epoch": 0.5030987969376595,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8288023471832275,
      "step": 5980
    },
    {
      "epoch": 0.5032670574048628,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8788337707519531,
      "step": 5982
    },
    {
      "epoch": 0.503435317872066,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.866550326347351,
      "step": 5984
    },
    {
      "epoch": 0.5036035783392692,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8104677200317383,
      "step": 5986
    },
    {
      "epoch": 0.5037718388064725,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8537516593933105,
      "step": 5988
    },
    {
      "epoch": 0.5039400992736757,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8985137939453125,
      "step": 5990
    },
    {
      "epoch": 0.5041083597408789,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.852118730545044,
      "step": 5992
    },
    {
      "epoch": 0.5042766202080821,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8288078308105469,
      "step": 5994
    },
    {
      "epoch": 0.5044448806752854,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.868248462677002,
      "step": 5996
    },
    {
      "epoch": 0.5046131411424886,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8513069152832031,
      "step": 5998
    },
    {
      "epoch": 0.5047814016096918,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8239179849624634,
      "step": 6000
    },
    {
      "epoch": 0.5049496620768951,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8321843147277832,
      "step": 6002
    },
    {
      "epoch": 0.5051179225440983,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8213047981262207,
      "step": 6004
    },
    {
      "epoch": 0.5052861830113015,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.866120457649231,
      "step": 6006
    },
    {
      "epoch": 0.5054544434785048,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.846738338470459,
      "step": 6008
    },
    {
      "epoch": 0.505622703945708,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.869840383529663,
      "step": 6010
    },
    {
      "epoch": 0.5057909644129112,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8670490980148315,
      "step": 6012
    },
    {
      "epoch": 0.5059592248801145,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8860530853271484,
      "step": 6014
    },
    {
      "epoch": 0.5061274853473177,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.876488447189331,
      "step": 6016
    },
    {
      "epoch": 0.5062957458145209,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8373572826385498,
      "step": 6018
    },
    {
      "epoch": 0.506464006281724,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.844501256942749,
      "step": 6020
    },
    {
      "epoch": 0.5066322667489274,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.871741533279419,
      "step": 6022
    },
    {
      "epoch": 0.5068005272161306,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8442409038543701,
      "step": 6024
    },
    {
      "epoch": 0.5069687876833338,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8588489294052124,
      "step": 6026
    },
    {
      "epoch": 0.5071370481505371,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0003,
      "loss": 1.8185261487960815,
      "step": 6028
    },
    {
      "epoch": 0.5073053086177403,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8873560428619385,
      "step": 6030
    },
    {
      "epoch": 0.5074735690849435,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.849708080291748,
      "step": 6032
    },
    {
      "epoch": 0.5076418295521468,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8344666957855225,
      "step": 6034
    },
    {
      "epoch": 0.50781009001935,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8331562280654907,
      "step": 6036
    },
    {
      "epoch": 0.5079783504865532,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.907973289489746,
      "step": 6038
    },
    {
      "epoch": 0.5081466109537565,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8468761444091797,
      "step": 6040
    },
    {
      "epoch": 0.5083148714209597,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.785849928855896,
      "step": 6042
    },
    {
      "epoch": 0.5084831318881629,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8632025718688965,
      "step": 6044
    },
    {
      "epoch": 0.5086513923553662,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8391520977020264,
      "step": 6046
    },
    {
      "epoch": 0.5088196528225694,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8950114250183105,
      "step": 6048
    },
    {
      "epoch": 0.5089879132897726,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8304716348648071,
      "step": 6050
    },
    {
      "epoch": 0.5091561737569757,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.872921109199524,
      "step": 6052
    },
    {
      "epoch": 0.5093244342241791,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.7945663928985596,
      "step": 6054
    },
    {
      "epoch": 0.5094926946913823,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8865597248077393,
      "step": 6056
    },
    {
      "epoch": 0.5096609551585854,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0003,
      "loss": 1.7749130725860596,
      "step": 6058
    },
    {
      "epoch": 0.5098292156257888,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0003,
      "loss": 1.855605125427246,
      "step": 6060
    },
    {
      "epoch": 0.509997476092992,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.8478004932403564,
      "step": 6062
    },
    {
      "epoch": 0.5101657365601951,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8940538167953491,
      "step": 6064
    },
    {
      "epoch": 0.5103339970273985,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8264565467834473,
      "step": 6066
    },
    {
      "epoch": 0.5105022574946017,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8291655778884888,
      "step": 6068
    },
    {
      "epoch": 0.5106705179618048,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8491830825805664,
      "step": 6070
    },
    {
      "epoch": 0.5108387784290082,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8584810495376587,
      "step": 6072
    },
    {
      "epoch": 0.5110070388962114,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.879284381866455,
      "step": 6074
    },
    {
      "epoch": 0.5111752993634145,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.867747187614441,
      "step": 6076
    },
    {
      "epoch": 0.5113435598306177,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8187614679336548,
      "step": 6078
    },
    {
      "epoch": 0.511511820297821,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.7985014915466309,
      "step": 6080
    },
    {
      "epoch": 0.5116800807650242,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8606899976730347,
      "step": 6082
    },
    {
      "epoch": 0.5118483412322274,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.826566219329834,
      "step": 6084
    },
    {
      "epoch": 0.5120166016994308,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8564798831939697,
      "step": 6086
    },
    {
      "epoch": 0.512184862166634,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8153717517852783,
      "step": 6088
    },
    {
      "epoch": 0.5123531226338371,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8394551277160645,
      "step": 6090
    },
    {
      "epoch": 0.5125213831010405,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.7760918140411377,
      "step": 6092
    },
    {
      "epoch": 0.5126896435682436,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8688507080078125,
      "step": 6094
    },
    {
      "epoch": 0.5128579040354468,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8752548694610596,
      "step": 6096
    },
    {
      "epoch": 0.5130261645026502,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.868476152420044,
      "step": 6098
    },
    {
      "epoch": 0.5131944249698533,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9121216535568237,
      "step": 6100
    },
    {
      "epoch": 0.5133626854370565,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8755061626434326,
      "step": 6102
    },
    {
      "epoch": 0.5135309459042597,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8359277248382568,
      "step": 6104
    },
    {
      "epoch": 0.513699206371463,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8868389129638672,
      "step": 6106
    },
    {
      "epoch": 0.5138674668386662,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8778643608093262,
      "step": 6108
    },
    {
      "epoch": 0.5140357273058694,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.7912776470184326,
      "step": 6110
    },
    {
      "epoch": 0.5142039877730727,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8259689807891846,
      "step": 6112
    },
    {
      "epoch": 0.5143722482402759,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9051284790039062,
      "step": 6114
    },
    {
      "epoch": 0.5145405087074791,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8470370769500732,
      "step": 6116
    },
    {
      "epoch": 0.5147087691746824,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8705755472183228,
      "step": 6118
    },
    {
      "epoch": 0.5148770296418856,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8625798225402832,
      "step": 6120
    },
    {
      "epoch": 0.5150452901090888,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8614660501480103,
      "step": 6122
    },
    {
      "epoch": 0.5152135505762921,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8614931106567383,
      "step": 6124
    },
    {
      "epoch": 0.5153818110434953,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8276216983795166,
      "step": 6126
    },
    {
      "epoch": 0.5155500715106985,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8351970911026,
      "step": 6128
    },
    {
      "epoch": 0.5157183319779018,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8367124795913696,
      "step": 6130
    },
    {
      "epoch": 0.515886592445105,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8368951082229614,
      "step": 6132
    },
    {
      "epoch": 0.5160548529123082,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8354151248931885,
      "step": 6134
    },
    {
      "epoch": 0.5162231133795114,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8100955486297607,
      "step": 6136
    },
    {
      "epoch": 0.5163913738467147,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8532748222351074,
      "step": 6138
    },
    {
      "epoch": 0.5165596343139179,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8972771167755127,
      "step": 6140
    },
    {
      "epoch": 0.5167278947811211,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8716564178466797,
      "step": 6142
    },
    {
      "epoch": 0.5168961552483244,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.7811713218688965,
      "step": 6144
    },
    {
      "epoch": 0.5170644157155276,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8297640085220337,
      "step": 6146
    },
    {
      "epoch": 0.5172326761827308,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8307249546051025,
      "step": 6148
    },
    {
      "epoch": 0.5174009366499341,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9313689470291138,
      "step": 6150
    },
    {
      "epoch": 0.5175691971171373,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8891340494155884,
      "step": 6152
    },
    {
      "epoch": 0.5177374575843405,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8629354238510132,
      "step": 6154
    },
    {
      "epoch": 0.5179057180515438,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8446818590164185,
      "step": 6156
    },
    {
      "epoch": 0.518073978518747,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8949127197265625,
      "step": 6158
    },
    {
      "epoch": 0.5182422389859502,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8213222026824951,
      "step": 6160
    },
    {
      "epoch": 0.5184104994531534,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.87493896484375,
      "step": 6162
    },
    {
      "epoch": 0.5185787599203567,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.776595950126648,
      "step": 6164
    },
    {
      "epoch": 0.5187470203875599,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8248701095581055,
      "step": 6166
    },
    {
      "epoch": 0.5189152808547631,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.839904546737671,
      "step": 6168
    },
    {
      "epoch": 0.5190835413219664,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8166368007659912,
      "step": 6170
    },
    {
      "epoch": 0.5192518017891696,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.851348876953125,
      "step": 6172
    },
    {
      "epoch": 0.5194200622563728,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.857954502105713,
      "step": 6174
    },
    {
      "epoch": 0.5195883227235761,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8058016300201416,
      "step": 6176
    },
    {
      "epoch": 0.5197565831907793,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8300421237945557,
      "step": 6178
    },
    {
      "epoch": 0.5199248436579825,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8437671661376953,
      "step": 6180
    },
    {
      "epoch": 0.5200931041251858,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.827370524406433,
      "step": 6182
    },
    {
      "epoch": 0.520261364592389,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8562226295471191,
      "step": 6184
    },
    {
      "epoch": 0.5204296250595922,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.842428207397461,
      "step": 6186
    },
    {
      "epoch": 0.5205978855267955,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003,
      "loss": 1.8091812133789062,
      "step": 6188
    },
    {
      "epoch": 0.5207661459939987,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8474454879760742,
      "step": 6190
    },
    {
      "epoch": 0.5209344064612019,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8174148797988892,
      "step": 6192
    },
    {
      "epoch": 0.5211026669284051,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.9029275178909302,
      "step": 6194
    },
    {
      "epoch": 0.5212709273956084,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0003,
      "loss": 1.8422166109085083,
      "step": 6196
    },
    {
      "epoch": 0.5214391878628116,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8858871459960938,
      "step": 6198
    },
    {
      "epoch": 0.5216074483300148,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8348329067230225,
      "step": 6200
    },
    {
      "epoch": 0.5217757087972181,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.790744423866272,
      "step": 6202
    },
    {
      "epoch": 0.5219439692644213,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8944575786590576,
      "step": 6204
    },
    {
      "epoch": 0.5221122297316245,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8398821353912354,
      "step": 6206
    },
    {
      "epoch": 0.5222804901988278,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8137458562850952,
      "step": 6208
    },
    {
      "epoch": 0.522448750666031,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.797667145729065,
      "step": 6210
    },
    {
      "epoch": 0.5226170111332342,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.871333360671997,
      "step": 6212
    },
    {
      "epoch": 0.5227852716004375,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8749456405639648,
      "step": 6214
    },
    {
      "epoch": 0.5229535320676407,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8651180267333984,
      "step": 6216
    },
    {
      "epoch": 0.5231217925348439,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8592619895935059,
      "step": 6218
    },
    {
      "epoch": 0.5232900530020471,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.893657922744751,
      "step": 6220
    },
    {
      "epoch": 0.5234583134692504,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8648474216461182,
      "step": 6222
    },
    {
      "epoch": 0.5236265739364536,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.837424635887146,
      "step": 6224
    },
    {
      "epoch": 0.5237948344036568,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8561962842941284,
      "step": 6226
    },
    {
      "epoch": 0.5239630948708601,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8723602294921875,
      "step": 6228
    },
    {
      "epoch": 0.5241313553380633,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8407106399536133,
      "step": 6230
    },
    {
      "epoch": 0.5242996158052665,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8112614154815674,
      "step": 6232
    },
    {
      "epoch": 0.5244678762724698,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8223941326141357,
      "step": 6234
    },
    {
      "epoch": 0.524636136739673,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.796363115310669,
      "step": 6236
    },
    {
      "epoch": 0.5248043972068762,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8530263900756836,
      "step": 6238
    },
    {
      "epoch": 0.5249726576740795,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8077607154846191,
      "step": 6240
    },
    {
      "epoch": 0.5251409181412827,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.899458646774292,
      "step": 6242
    },
    {
      "epoch": 0.5253091786084859,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8295061588287354,
      "step": 6244
    },
    {
      "epoch": 0.5254774390756891,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8637800216674805,
      "step": 6246
    },
    {
      "epoch": 0.5256456995428924,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 1.8603895902633667,
      "step": 6248
    },
    {
      "epoch": 0.5258139600100956,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8288397789001465,
      "step": 6250
    },
    {
      "epoch": 0.5259822204772988,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8393745422363281,
      "step": 6252
    },
    {
      "epoch": 0.5261504809445021,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8086326122283936,
      "step": 6254
    },
    {
      "epoch": 0.5263187414117053,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.852476716041565,
      "step": 6256
    },
    {
      "epoch": 0.5264870018789085,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8859260082244873,
      "step": 6258
    },
    {
      "epoch": 0.5266552623461118,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8462042808532715,
      "step": 6260
    },
    {
      "epoch": 0.526823522813315,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8520373106002808,
      "step": 6262
    },
    {
      "epoch": 0.5269917832805182,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8395129442214966,
      "step": 6264
    },
    {
      "epoch": 0.5271600437477215,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.851799726486206,
      "step": 6266
    },
    {
      "epoch": 0.5273283042149247,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8359230756759644,
      "step": 6268
    },
    {
      "epoch": 0.5274965646821279,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8672409057617188,
      "step": 6270
    },
    {
      "epoch": 0.5276648251493312,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8544018268585205,
      "step": 6272
    },
    {
      "epoch": 0.5278330856165344,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8583135604858398,
      "step": 6274
    },
    {
      "epoch": 0.5280013460837376,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8333417177200317,
      "step": 6276
    },
    {
      "epoch": 0.5281696065509408,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8192830085754395,
      "step": 6278
    },
    {
      "epoch": 0.5283378670181441,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.87782621383667,
      "step": 6280
    },
    {
      "epoch": 0.5285061274853473,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8487603664398193,
      "step": 6282
    },
    {
      "epoch": 0.5286743879525505,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8016149997711182,
      "step": 6284
    },
    {
      "epoch": 0.5288426484197538,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8453525304794312,
      "step": 6286
    },
    {
      "epoch": 0.529010908886957,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.858016014099121,
      "step": 6288
    },
    {
      "epoch": 0.5291791693541602,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.838106393814087,
      "step": 6290
    },
    {
      "epoch": 0.5293474298213635,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8126566410064697,
      "step": 6292
    },
    {
      "epoch": 0.5295156902885667,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8603483438491821,
      "step": 6294
    },
    {
      "epoch": 0.5296839507557699,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8235323429107666,
      "step": 6296
    },
    {
      "epoch": 0.5298522112229732,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8375970125198364,
      "step": 6298
    },
    {
      "epoch": 0.5300204716901764,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8868989944458008,
      "step": 6300
    },
    {
      "epoch": 0.5301887321573796,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8329651355743408,
      "step": 6302
    },
    {
      "epoch": 0.5303569926245828,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003,
      "loss": 1.8135600090026855,
      "step": 6304
    },
    {
      "epoch": 0.5305252530917861,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.815781831741333,
      "step": 6306
    },
    {
      "epoch": 0.5306935135589893,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8156261444091797,
      "step": 6308
    },
    {
      "epoch": 0.5308617740261925,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8529224395751953,
      "step": 6310
    },
    {
      "epoch": 0.5310300344933958,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.7790745496749878,
      "step": 6312
    },
    {
      "epoch": 0.531198294960599,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8447566032409668,
      "step": 6314
    },
    {
      "epoch": 0.5313665554278022,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8322652578353882,
      "step": 6316
    },
    {
      "epoch": 0.5315348158950055,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8830182552337646,
      "step": 6318
    },
    {
      "epoch": 0.5317030763622087,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8755083084106445,
      "step": 6320
    },
    {
      "epoch": 0.5318713368294119,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.9081590175628662,
      "step": 6322
    },
    {
      "epoch": 0.5320395972966152,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003,
      "loss": 1.8260178565979004,
      "step": 6324
    },
    {
      "epoch": 0.5322078577638184,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8776381015777588,
      "step": 6326
    },
    {
      "epoch": 0.5323761182310216,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8260422945022583,
      "step": 6328
    },
    {
      "epoch": 0.5325443786982249,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.830700159072876,
      "step": 6330
    },
    {
      "epoch": 0.5327126391654281,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0003,
      "loss": 1.8533059358596802,
      "step": 6332
    },
    {
      "epoch": 0.5328808996326313,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.841245174407959,
      "step": 6334
    },
    {
      "epoch": 0.5330491600998345,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8691117763519287,
      "step": 6336
    },
    {
      "epoch": 0.5332174205670378,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8164451122283936,
      "step": 6338
    },
    {
      "epoch": 0.533385681034241,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8295009136199951,
      "step": 6340
    },
    {
      "epoch": 0.5335539415014442,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8157498836517334,
      "step": 6342
    },
    {
      "epoch": 0.5337222019686475,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8355228900909424,
      "step": 6344
    },
    {
      "epoch": 0.5338904624358507,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8921916484832764,
      "step": 6346
    },
    {
      "epoch": 0.5340587229030539,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.90574312210083,
      "step": 6348
    },
    {
      "epoch": 0.5342269833702572,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.7977995872497559,
      "step": 6350
    },
    {
      "epoch": 0.5343952438374604,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9068644046783447,
      "step": 6352
    },
    {
      "epoch": 0.5345635043046636,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8686838150024414,
      "step": 6354
    },
    {
      "epoch": 0.5347317647718669,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8206048011779785,
      "step": 6356
    },
    {
      "epoch": 0.5349000252390701,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8419082164764404,
      "step": 6358
    },
    {
      "epoch": 0.5350682857062733,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.827216625213623,
      "step": 6360
    },
    {
      "epoch": 0.5352365461734765,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8424545526504517,
      "step": 6362
    },
    {
      "epoch": 0.5354048066406798,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8337948322296143,
      "step": 6364
    },
    {
      "epoch": 0.535573067107883,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8472094535827637,
      "step": 6366
    },
    {
      "epoch": 0.5357413275750862,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0003,
      "loss": 1.8404481410980225,
      "step": 6368
    },
    {
      "epoch": 0.5359095880422895,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8437896966934204,
      "step": 6370
    },
    {
      "epoch": 0.5360778485094927,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8637559413909912,
      "step": 6372
    },
    {
      "epoch": 0.5362461089766959,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8560147285461426,
      "step": 6374
    },
    {
      "epoch": 0.5364143694438992,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003,
      "loss": 1.8628015518188477,
      "step": 6376
    },
    {
      "epoch": 0.5365826299111024,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8314448595046997,
      "step": 6378
    },
    {
      "epoch": 0.5367508903783056,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8007421493530273,
      "step": 6380
    },
    {
      "epoch": 0.5369191508455089,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.8576924800872803,
      "step": 6382
    },
    {
      "epoch": 0.5370874113127121,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8761768341064453,
      "step": 6384
    },
    {
      "epoch": 0.5372556717799153,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8876712322235107,
      "step": 6386
    },
    {
      "epoch": 0.5374239322471185,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8293898105621338,
      "step": 6388
    },
    {
      "epoch": 0.5375921927143218,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8594799041748047,
      "step": 6390
    },
    {
      "epoch": 0.537760453181525,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8457677364349365,
      "step": 6392
    },
    {
      "epoch": 0.5379287136487282,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8798189163208008,
      "step": 6394
    },
    {
      "epoch": 0.5380969741159315,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8360882997512817,
      "step": 6396
    },
    {
      "epoch": 0.5382652345831347,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8288729190826416,
      "step": 6398
    },
    {
      "epoch": 0.5384334950503379,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8557460308074951,
      "step": 6400
    },
    {
      "epoch": 0.5386017555175412,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8126862049102783,
      "step": 6402
    },
    {
      "epoch": 0.5387700159847444,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.828134298324585,
      "step": 6404
    },
    {
      "epoch": 0.5389382764519476,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8406801223754883,
      "step": 6406
    },
    {
      "epoch": 0.5391065369191509,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8408546447753906,
      "step": 6408
    },
    {
      "epoch": 0.5392747973863541,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8443419933319092,
      "step": 6410
    },
    {
      "epoch": 0.5394430578535573,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8721303939819336,
      "step": 6412
    },
    {
      "epoch": 0.5396113183207606,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8907132148742676,
      "step": 6414
    },
    {
      "epoch": 0.5397795787879638,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8448576927185059,
      "step": 6416
    },
    {
      "epoch": 0.539947839255167,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8338711261749268,
      "step": 6418
    },
    {
      "epoch": 0.5401160997223702,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8537280559539795,
      "step": 6420
    },
    {
      "epoch": 0.5402843601895735,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8213281631469727,
      "step": 6422
    },
    {
      "epoch": 0.5404526206567767,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.81974196434021,
      "step": 6424
    },
    {
      "epoch": 0.5406208811239799,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8074665069580078,
      "step": 6426
    },
    {
      "epoch": 0.5407891415911832,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.803879737854004,
      "step": 6428
    },
    {
      "epoch": 0.5409574020583864,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8357245922088623,
      "step": 6430
    },
    {
      "epoch": 0.5411256625255896,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8094369173049927,
      "step": 6432
    },
    {
      "epoch": 0.5412939229927929,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8192569017410278,
      "step": 6434
    },
    {
      "epoch": 0.5414621834599961,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8519504070281982,
      "step": 6436
    },
    {
      "epoch": 0.5416304439271993,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.902604341506958,
      "step": 6438
    },
    {
      "epoch": 0.5417987043944026,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8143868446350098,
      "step": 6440
    },
    {
      "epoch": 0.5419669648616058,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.883967638015747,
      "step": 6442
    },
    {
      "epoch": 0.542135225328809,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8414936065673828,
      "step": 6444
    },
    {
      "epoch": 0.5423034857960122,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8416873216629028,
      "step": 6446
    },
    {
      "epoch": 0.5424717462632155,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8416261672973633,
      "step": 6448
    },
    {
      "epoch": 0.5426400067304187,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8400394916534424,
      "step": 6450
    },
    {
      "epoch": 0.5428082671976219,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.846447467803955,
      "step": 6452
    },
    {
      "epoch": 0.5429765276648252,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.840559482574463,
      "step": 6454
    },
    {
      "epoch": 0.5431447881320284,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.7959558963775635,
      "step": 6456
    },
    {
      "epoch": 0.5433130485992316,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8034279346466064,
      "step": 6458
    },
    {
      "epoch": 0.5434813090664349,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.891671895980835,
      "step": 6460
    },
    {
      "epoch": 0.5436495695336381,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.838822603225708,
      "step": 6462
    },
    {
      "epoch": 0.5438178300008413,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8162569999694824,
      "step": 6464
    },
    {
      "epoch": 0.5439860904680446,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8326315879821777,
      "step": 6466
    },
    {
      "epoch": 0.5441543509352478,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8384312391281128,
      "step": 6468
    },
    {
      "epoch": 0.544322611402451,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8501248359680176,
      "step": 6470
    },
    {
      "epoch": 0.5444908718696542,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.845119595527649,
      "step": 6472
    },
    {
      "epoch": 0.5446591323368575,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.798396110534668,
      "step": 6474
    },
    {
      "epoch": 0.5448273928040607,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.807959794998169,
      "step": 6476
    },
    {
      "epoch": 0.5449956532712639,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.7973582744598389,
      "step": 6478
    },
    {
      "epoch": 0.5451639137384672,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8302091360092163,
      "step": 6480
    },
    {
      "epoch": 0.5453321742056704,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.812163233757019,
      "step": 6482
    },
    {
      "epoch": 0.5455004346728736,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8452218770980835,
      "step": 6484
    },
    {
      "epoch": 0.5456686951400769,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.7919321060180664,
      "step": 6486
    },
    {
      "epoch": 0.5458369556072801,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.865593433380127,
      "step": 6488
    },
    {
      "epoch": 0.5460052160744833,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.7910236120224,
      "step": 6490
    },
    {
      "epoch": 0.5461734765416866,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8250446319580078,
      "step": 6492
    },
    {
      "epoch": 0.5463417370088898,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.854727864265442,
      "step": 6494
    },
    {
      "epoch": 0.546509997476093,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8625752925872803,
      "step": 6496
    },
    {
      "epoch": 0.5466782579432963,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8765251636505127,
      "step": 6498
    },
    {
      "epoch": 0.5468465184104995,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.847043514251709,
      "step": 6500
    },
    {
      "epoch": 0.5470147788777027,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8366634845733643,
      "step": 6502
    },
    {
      "epoch": 0.5471830393449059,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.841134786605835,
      "step": 6504
    },
    {
      "epoch": 0.5473512998121092,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8090081214904785,
      "step": 6506
    },
    {
      "epoch": 0.5475195602793124,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8832683563232422,
      "step": 6508
    },
    {
      "epoch": 0.5476878207465156,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8833577632904053,
      "step": 6510
    },
    {
      "epoch": 0.5478560812137189,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.7984977960586548,
      "step": 6512
    },
    {
      "epoch": 0.5480243416809221,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8482873439788818,
      "step": 6514
    },
    {
      "epoch": 0.5481926021481253,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8201210498809814,
      "step": 6516
    },
    {
      "epoch": 0.5483608626153286,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8794063329696655,
      "step": 6518
    },
    {
      "epoch": 0.5485291230825318,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.7960814237594604,
      "step": 6520
    },
    {
      "epoch": 0.548697383549735,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8313608169555664,
      "step": 6522
    },
    {
      "epoch": 0.5488656440169383,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8884515762329102,
      "step": 6524
    },
    {
      "epoch": 0.5490339044841415,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8627564907073975,
      "step": 6526
    },
    {
      "epoch": 0.5492021649513447,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.7970798015594482,
      "step": 6528
    },
    {
      "epoch": 0.5493704254185479,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8435524702072144,
      "step": 6530
    },
    {
      "epoch": 0.5495386858857512,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8458539247512817,
      "step": 6532
    },
    {
      "epoch": 0.5497069463529544,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8366572856903076,
      "step": 6534
    },
    {
      "epoch": 0.5498752068201576,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8774176836013794,
      "step": 6536
    },
    {
      "epoch": 0.5500434672873609,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.856332778930664,
      "step": 6538
    },
    {
      "epoch": 0.5502117277545641,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8084311485290527,
      "step": 6540
    },
    {
      "epoch": 0.5503799882217673,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8561151027679443,
      "step": 6542
    },
    {
      "epoch": 0.5505482486889706,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.816391110420227,
      "step": 6544
    },
    {
      "epoch": 0.5507165091561738,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8412225246429443,
      "step": 6546
    },
    {
      "epoch": 0.550884769623377,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.842756986618042,
      "step": 6548
    },
    {
      "epoch": 0.5510530300905803,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8362126350402832,
      "step": 6550
    },
    {
      "epoch": 0.5512212905577835,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8236725330352783,
      "step": 6552
    },
    {
      "epoch": 0.5513895510249867,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8479514122009277,
      "step": 6554
    },
    {
      "epoch": 0.55155781149219,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.856986403465271,
      "step": 6556
    },
    {
      "epoch": 0.5517260719593932,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.9040968418121338,
      "step": 6558
    },
    {
      "epoch": 0.5518943324265964,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8304696083068848,
      "step": 6560
    },
    {
      "epoch": 0.5520625928937996,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8503310680389404,
      "step": 6562
    },
    {
      "epoch": 0.5522308533610029,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8475193977355957,
      "step": 6564
    },
    {
      "epoch": 0.5523991138282061,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8536431789398193,
      "step": 6566
    },
    {
      "epoch": 0.5525673742954093,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8589022159576416,
      "step": 6568
    },
    {
      "epoch": 0.5527356347626126,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.790926456451416,
      "step": 6570
    },
    {
      "epoch": 0.5529038952298158,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.881044626235962,
      "step": 6572
    },
    {
      "epoch": 0.553072155697019,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8425933122634888,
      "step": 6574
    },
    {
      "epoch": 0.5532404161642223,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8357281684875488,
      "step": 6576
    },
    {
      "epoch": 0.5534086766314255,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8995380401611328,
      "step": 6578
    },
    {
      "epoch": 0.5535769370986287,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.897728681564331,
      "step": 6580
    },
    {
      "epoch": 0.553745197565832,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8146684169769287,
      "step": 6582
    },
    {
      "epoch": 0.5539134580330352,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.7764084339141846,
      "step": 6584
    },
    {
      "epoch": 0.5540817185002384,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8111395835876465,
      "step": 6586
    },
    {
      "epoch": 0.5542499789674415,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8208839893341064,
      "step": 6588
    },
    {
      "epoch": 0.5544182394346449,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8081486225128174,
      "step": 6590
    },
    {
      "epoch": 0.554586499901848,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.880208969116211,
      "step": 6592
    },
    {
      "epoch": 0.5547547603690512,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8281974792480469,
      "step": 6594
    },
    {
      "epoch": 0.5549230208362546,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8058321475982666,
      "step": 6596
    },
    {
      "epoch": 0.5550912813034578,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.860772967338562,
      "step": 6598
    },
    {
      "epoch": 0.555259541770661,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8301217555999756,
      "step": 6600
    },
    {
      "epoch": 0.5554278022378643,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8479695320129395,
      "step": 6602
    },
    {
      "epoch": 0.5555960627050675,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8905130624771118,
      "step": 6604
    },
    {
      "epoch": 0.5557643231722706,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8623120784759521,
      "step": 6606
    },
    {
      "epoch": 0.555932583639474,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8337385654449463,
      "step": 6608
    },
    {
      "epoch": 0.5561008441066772,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.7913267612457275,
      "step": 6610
    },
    {
      "epoch": 0.5562691045738803,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8194339275360107,
      "step": 6612
    },
    {
      "epoch": 0.5564373650410835,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.850247859954834,
      "step": 6614
    },
    {
      "epoch": 0.5566056255082868,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8278110027313232,
      "step": 6616
    },
    {
      "epoch": 0.55677388597549,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8276722431182861,
      "step": 6618
    },
    {
      "epoch": 0.5569421464426932,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.844024419784546,
      "step": 6620
    },
    {
      "epoch": 0.5571104069098965,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8633663654327393,
      "step": 6622
    },
    {
      "epoch": 0.5572786673770997,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8026745319366455,
      "step": 6624
    },
    {
      "epoch": 0.5574469278443029,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.862415075302124,
      "step": 6626
    },
    {
      "epoch": 0.5576151883115062,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8089735507965088,
      "step": 6628
    },
    {
      "epoch": 0.5577834487787094,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.816622257232666,
      "step": 6630
    },
    {
      "epoch": 0.5579517092459126,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8523929119110107,
      "step": 6632
    },
    {
      "epoch": 0.558119969713116,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8433189392089844,
      "step": 6634
    },
    {
      "epoch": 0.5582882301803191,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8172228336334229,
      "step": 6636
    },
    {
      "epoch": 0.5584564906475223,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8132102489471436,
      "step": 6638
    },
    {
      "epoch": 0.5586247511147256,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8897631168365479,
      "step": 6640
    },
    {
      "epoch": 0.5587930115819288,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8341152667999268,
      "step": 6642
    },
    {
      "epoch": 0.558961272049132,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8190363645553589,
      "step": 6644
    },
    {
      "epoch": 0.5591295325163352,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8362700939178467,
      "step": 6646
    },
    {
      "epoch": 0.5592977929835385,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8013601303100586,
      "step": 6648
    },
    {
      "epoch": 0.5594660534507417,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.9021742343902588,
      "step": 6650
    },
    {
      "epoch": 0.5596343139179449,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8277878761291504,
      "step": 6652
    },
    {
      "epoch": 0.5598025743851482,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8151803016662598,
      "step": 6654
    },
    {
      "epoch": 0.5599708348523514,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8292887210845947,
      "step": 6656
    },
    {
      "epoch": 0.5601390953195546,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8488177061080933,
      "step": 6658
    },
    {
      "epoch": 0.5603073557867579,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8644568920135498,
      "step": 6660
    },
    {
      "epoch": 0.5604756162539611,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8588274717330933,
      "step": 6662
    },
    {
      "epoch": 0.5606438767211643,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8188376426696777,
      "step": 6664
    },
    {
      "epoch": 0.5608121371883676,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8717349767684937,
      "step": 6666
    },
    {
      "epoch": 0.5609803976555708,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8196494579315186,
      "step": 6668
    },
    {
      "epoch": 0.561148658122774,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003,
      "loss": 1.8878250122070312,
      "step": 6670
    },
    {
      "epoch": 0.5613169185899772,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.808474063873291,
      "step": 6672
    },
    {
      "epoch": 0.5614851790571805,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8770866394042969,
      "step": 6674
    },
    {
      "epoch": 0.5616534395243837,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.7917777299880981,
      "step": 6676
    },
    {
      "epoch": 0.5618216999915869,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003,
      "loss": 1.8418302536010742,
      "step": 6678
    },
    {
      "epoch": 0.5619899604587902,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8643522262573242,
      "step": 6680
    },
    {
      "epoch": 0.5621582209259934,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8463222980499268,
      "step": 6682
    },
    {
      "epoch": 0.5623264813931966,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8320109844207764,
      "step": 6684
    },
    {
      "epoch": 0.5624947418603999,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8377130031585693,
      "step": 6686
    },
    {
      "epoch": 0.5626630023276031,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.833949089050293,
      "step": 6688
    },
    {
      "epoch": 0.5628312627948063,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8930296897888184,
      "step": 6690
    },
    {
      "epoch": 0.5629995232620096,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8597642183303833,
      "step": 6692
    },
    {
      "epoch": 0.5631677837292128,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.834914207458496,
      "step": 6694
    },
    {
      "epoch": 0.563336044196416,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8476731777191162,
      "step": 6696
    },
    {
      "epoch": 0.5635043046636193,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8457434177398682,
      "step": 6698
    },
    {
      "epoch": 0.5636725651308225,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8024015426635742,
      "step": 6700
    },
    {
      "epoch": 0.5638408255980257,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8455498218536377,
      "step": 6702
    },
    {
      "epoch": 0.5640090860652289,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8307862281799316,
      "step": 6704
    },
    {
      "epoch": 0.5641773465324322,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.7950801849365234,
      "step": 6706
    },
    {
      "epoch": 0.5643456069996354,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8228659629821777,
      "step": 6708
    },
    {
      "epoch": 0.5645138674668386,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8848693370819092,
      "step": 6710
    },
    {
      "epoch": 0.5646821279340419,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8508062362670898,
      "step": 6712
    },
    {
      "epoch": 0.5648503884012451,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.7869386672973633,
      "step": 6714
    },
    {
      "epoch": 0.5650186488684483,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8975082635879517,
      "step": 6716
    },
    {
      "epoch": 0.5651869093356516,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.860522985458374,
      "step": 6718
    },
    {
      "epoch": 0.5653551698028548,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8362889289855957,
      "step": 6720
    },
    {
      "epoch": 0.565523430270058,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8772480487823486,
      "step": 6722
    },
    {
      "epoch": 0.5656916907372613,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.85263991355896,
      "step": 6724
    },
    {
      "epoch": 0.5658599512044645,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.823664903640747,
      "step": 6726
    },
    {
      "epoch": 0.5660282116716677,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8792427778244019,
      "step": 6728
    },
    {
      "epoch": 0.5661964721388709,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8708817958831787,
      "step": 6730
    },
    {
      "epoch": 0.5663647326060742,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8451626300811768,
      "step": 6732
    },
    {
      "epoch": 0.5665329930732774,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.852536678314209,
      "step": 6734
    },
    {
      "epoch": 0.5667012535404806,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.873767614364624,
      "step": 6736
    },
    {
      "epoch": 0.5668695140076839,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8085048198699951,
      "step": 6738
    },
    {
      "epoch": 0.5670377744748871,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8510501384735107,
      "step": 6740
    },
    {
      "epoch": 0.5672060349420903,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8410899639129639,
      "step": 6742
    },
    {
      "epoch": 0.5673742954092936,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8513734340667725,
      "step": 6744
    },
    {
      "epoch": 0.5675425558764968,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8216161727905273,
      "step": 6746
    },
    {
      "epoch": 0.5677108163437,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8285350799560547,
      "step": 6748
    },
    {
      "epoch": 0.5678790768109033,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8120535612106323,
      "step": 6750
    },
    {
      "epoch": 0.5680473372781065,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.816211223602295,
      "step": 6752
    },
    {
      "epoch": 0.5682155977453097,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.871464729309082,
      "step": 6754
    },
    {
      "epoch": 0.5683838582125129,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8803085088729858,
      "step": 6756
    },
    {
      "epoch": 0.5685521186797162,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.7898900508880615,
      "step": 6758
    },
    {
      "epoch": 0.5687203791469194,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8451340198516846,
      "step": 6760
    },
    {
      "epoch": 0.5688886396141226,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8715415000915527,
      "step": 6762
    },
    {
      "epoch": 0.5690569000813259,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8725786209106445,
      "step": 6764
    },
    {
      "epoch": 0.5692251605485291,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8250426054000854,
      "step": 6766
    },
    {
      "epoch": 0.5693934210157323,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8347986936569214,
      "step": 6768
    },
    {
      "epoch": 0.5695616814829356,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8515220880508423,
      "step": 6770
    },
    {
      "epoch": 0.5697299419501388,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.7789446115493774,
      "step": 6772
    },
    {
      "epoch": 0.569898202417342,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.836829662322998,
      "step": 6774
    },
    {
      "epoch": 0.5700664628845453,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8436015844345093,
      "step": 6776
    },
    {
      "epoch": 0.5702347233517485,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003,
      "loss": 1.7950525283813477,
      "step": 6778
    },
    {
      "epoch": 0.5704029838189517,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.867929220199585,
      "step": 6780
    },
    {
      "epoch": 0.570571244286155,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8364677429199219,
      "step": 6782
    },
    {
      "epoch": 0.5707395047533582,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.803072214126587,
      "step": 6784
    },
    {
      "epoch": 0.5709077652205614,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8302298784255981,
      "step": 6786
    },
    {
      "epoch": 0.5710760256877646,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8632253408432007,
      "step": 6788
    },
    {
      "epoch": 0.5712442861549679,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.844343662261963,
      "step": 6790
    },
    {
      "epoch": 0.5714125466221711,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.85696280002594,
      "step": 6792
    },
    {
      "epoch": 0.5715808070893743,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.838792085647583,
      "step": 6794
    },
    {
      "epoch": 0.5717490675565776,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.870007872581482,
      "step": 6796
    },
    {
      "epoch": 0.5719173280237808,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8782925605773926,
      "step": 6798
    },
    {
      "epoch": 0.572085588490984,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8370327949523926,
      "step": 6800
    },
    {
      "epoch": 0.5722538489581873,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8312740325927734,
      "step": 6802
    },
    {
      "epoch": 0.5724221094253905,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8425631523132324,
      "step": 6804
    },
    {
      "epoch": 0.5725903698925937,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.863903284072876,
      "step": 6806
    },
    {
      "epoch": 0.572758630359797,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.7772746086120605,
      "step": 6808
    },
    {
      "epoch": 0.5729268908270002,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8439817428588867,
      "step": 6810
    },
    {
      "epoch": 0.5730951512942034,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8247575759887695,
      "step": 6812
    },
    {
      "epoch": 0.5732634117614066,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8306171894073486,
      "step": 6814
    },
    {
      "epoch": 0.5734316722286099,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003,
      "loss": 1.8430547714233398,
      "step": 6816
    },
    {
      "epoch": 0.5735999326958131,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.822359561920166,
      "step": 6818
    },
    {
      "epoch": 0.5737681931630163,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8450584411621094,
      "step": 6820
    },
    {
      "epoch": 0.5739364536302196,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003,
      "loss": 1.8647267818450928,
      "step": 6822
    },
    {
      "epoch": 0.5741047140974228,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8713912963867188,
      "step": 6824
    },
    {
      "epoch": 0.574272974564626,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.7810461521148682,
      "step": 6826
    },
    {
      "epoch": 0.5744412350318293,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.821700096130371,
      "step": 6828
    },
    {
      "epoch": 0.5746094954990325,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.7614305019378662,
      "step": 6830
    },
    {
      "epoch": 0.5747777559662357,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.852152943611145,
      "step": 6832
    },
    {
      "epoch": 0.574946016433439,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8088759183883667,
      "step": 6834
    },
    {
      "epoch": 0.5751142769006422,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8151906728744507,
      "step": 6836
    },
    {
      "epoch": 0.5752825373678454,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.851491928100586,
      "step": 6838
    },
    {
      "epoch": 0.5754507978350487,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.8196749687194824,
      "step": 6840
    },
    {
      "epoch": 0.5756190583022519,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8929126262664795,
      "step": 6842
    },
    {
      "epoch": 0.5757873187694551,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8434326648712158,
      "step": 6844
    },
    {
      "epoch": 0.5759555792366583,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0003,
      "loss": 1.7853505611419678,
      "step": 6846
    },
    {
      "epoch": 0.5761238397038616,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8773813247680664,
      "step": 6848
    },
    {
      "epoch": 0.5762921001710648,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.826221227645874,
      "step": 6850
    },
    {
      "epoch": 0.576460360638268,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8875575065612793,
      "step": 6852
    },
    {
      "epoch": 0.5766286211054713,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.852051854133606,
      "step": 6854
    },
    {
      "epoch": 0.5767968815726745,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.792067050933838,
      "step": 6856
    },
    {
      "epoch": 0.5769651420398777,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8596413135528564,
      "step": 6858
    },
    {
      "epoch": 0.577133402507081,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8158833980560303,
      "step": 6860
    },
    {
      "epoch": 0.5773016629742842,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8621058464050293,
      "step": 6862
    },
    {
      "epoch": 0.5774699234414874,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8882360458374023,
      "step": 6864
    },
    {
      "epoch": 0.5776381839086907,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8149725198745728,
      "step": 6866
    },
    {
      "epoch": 0.5778064443758939,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.814427137374878,
      "step": 6868
    },
    {
      "epoch": 0.5779747048430971,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.7935571670532227,
      "step": 6870
    },
    {
      "epoch": 0.5781429653103003,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.9026113748550415,
      "step": 6872
    },
    {
      "epoch": 0.5783112257775036,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8761922121047974,
      "step": 6874
    },
    {
      "epoch": 0.5784794862447068,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8824820518493652,
      "step": 6876
    },
    {
      "epoch": 0.57864774671191,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8097028732299805,
      "step": 6878
    },
    {
      "epoch": 0.5788160071791133,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8311055898666382,
      "step": 6880
    },
    {
      "epoch": 0.5789842676463165,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.840061902999878,
      "step": 6882
    },
    {
      "epoch": 0.5791525281135197,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8301706314086914,
      "step": 6884
    },
    {
      "epoch": 0.579320788580723,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8737449645996094,
      "step": 6886
    },
    {
      "epoch": 0.5794890490479262,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8695361614227295,
      "step": 6888
    },
    {
      "epoch": 0.5796573095151294,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.835585355758667,
      "step": 6890
    },
    {
      "epoch": 0.5798255699823327,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8055400848388672,
      "step": 6892
    },
    {
      "epoch": 0.5799938304495359,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8510273694992065,
      "step": 6894
    },
    {
      "epoch": 0.5801620909167391,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8069655895233154,
      "step": 6896
    },
    {
      "epoch": 0.5803303513839423,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.837843656539917,
      "step": 6898
    },
    {
      "epoch": 0.5804986118511456,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8080826997756958,
      "step": 6900
    },
    {
      "epoch": 0.5806668723183488,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8406956195831299,
      "step": 6902
    },
    {
      "epoch": 0.580835132785552,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8329715728759766,
      "step": 6904
    },
    {
      "epoch": 0.5810033932527553,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8724116086959839,
      "step": 6906
    },
    {
      "epoch": 0.5811716537199585,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8362478017807007,
      "step": 6908
    },
    {
      "epoch": 0.5813399141871617,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8197100162506104,
      "step": 6910
    },
    {
      "epoch": 0.581508174654365,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8666319847106934,
      "step": 6912
    },
    {
      "epoch": 0.5816764351215682,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8102080821990967,
      "step": 6914
    },
    {
      "epoch": 0.5818446955887714,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.829654335975647,
      "step": 6916
    },
    {
      "epoch": 0.5820129560559747,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8190845251083374,
      "step": 6918
    },
    {
      "epoch": 0.5821812165231779,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8657829761505127,
      "step": 6920
    },
    {
      "epoch": 0.5823494769903811,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8807408809661865,
      "step": 6922
    },
    {
      "epoch": 0.5825177374575844,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8528108596801758,
      "step": 6924
    },
    {
      "epoch": 0.5826859979247876,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8160814046859741,
      "step": 6926
    },
    {
      "epoch": 0.5828542583919908,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8494644165039062,
      "step": 6928
    },
    {
      "epoch": 0.583022518859194,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8098838329315186,
      "step": 6930
    },
    {
      "epoch": 0.5831907793263973,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.775921106338501,
      "step": 6932
    },
    {
      "epoch": 0.5833590397936005,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8500875234603882,
      "step": 6934
    },
    {
      "epoch": 0.5835273002608037,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8609752655029297,
      "step": 6936
    },
    {
      "epoch": 0.583695560728007,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8497142791748047,
      "step": 6938
    },
    {
      "epoch": 0.5838638211952102,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8613553047180176,
      "step": 6940
    },
    {
      "epoch": 0.5840320816624134,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8424057960510254,
      "step": 6942
    },
    {
      "epoch": 0.5842003421296167,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8549919128417969,
      "step": 6944
    },
    {
      "epoch": 0.5843686025968199,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8342978954315186,
      "step": 6946
    },
    {
      "epoch": 0.5845368630640231,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.844310998916626,
      "step": 6948
    },
    {
      "epoch": 0.5847051235312264,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8797192573547363,
      "step": 6950
    },
    {
      "epoch": 0.5848733839984296,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8600749969482422,
      "step": 6952
    },
    {
      "epoch": 0.5850416444656328,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8462133407592773,
      "step": 6954
    },
    {
      "epoch": 0.585209904932836,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8865866661071777,
      "step": 6956
    },
    {
      "epoch": 0.5853781654000393,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8146040439605713,
      "step": 6958
    },
    {
      "epoch": 0.5855464258672425,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8409061431884766,
      "step": 6960
    },
    {
      "epoch": 0.5857146863344457,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9069311618804932,
      "step": 6962
    },
    {
      "epoch": 0.585882946801649,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.7907757759094238,
      "step": 6964
    },
    {
      "epoch": 0.5860512072688522,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.865976333618164,
      "step": 6966
    },
    {
      "epoch": 0.5862194677360554,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.862554669380188,
      "step": 6968
    },
    {
      "epoch": 0.5863877282032587,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9122257232666016,
      "step": 6970
    },
    {
      "epoch": 0.5865559886704619,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8475182056427002,
      "step": 6972
    },
    {
      "epoch": 0.5867242491376651,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8627724647521973,
      "step": 6974
    },
    {
      "epoch": 0.5868925096048684,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.852548360824585,
      "step": 6976
    },
    {
      "epoch": 0.5870607700720716,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8725972175598145,
      "step": 6978
    },
    {
      "epoch": 0.5872290305392748,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8209478855133057,
      "step": 6980
    },
    {
      "epoch": 0.587397291006478,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.884181261062622,
      "step": 6982
    },
    {
      "epoch": 0.5875655514736813,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8204631805419922,
      "step": 6984
    },
    {
      "epoch": 0.5877338119408845,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8164830207824707,
      "step": 6986
    },
    {
      "epoch": 0.5879020724080877,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8502044677734375,
      "step": 6988
    },
    {
      "epoch": 0.588070332875291,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8470127582550049,
      "step": 6990
    },
    {
      "epoch": 0.5882385933424942,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.818113923072815,
      "step": 6992
    },
    {
      "epoch": 0.5884068538096974,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8448923826217651,
      "step": 6994
    },
    {
      "epoch": 0.5885751142769007,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8457825183868408,
      "step": 6996
    },
    {
      "epoch": 0.5887433747441039,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.841991901397705,
      "step": 6998
    },
    {
      "epoch": 0.5889116352113071,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8630890846252441,
      "step": 7000
    },
    {
      "epoch": 0.5890798956785104,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8296699523925781,
      "step": 7002
    },
    {
      "epoch": 0.5892481561457136,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8358324766159058,
      "step": 7004
    },
    {
      "epoch": 0.5894164166129168,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.837693691253662,
      "step": 7006
    },
    {
      "epoch": 0.5895846770801201,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8254293203353882,
      "step": 7008
    },
    {
      "epoch": 0.5897529375473233,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.839503288269043,
      "step": 7010
    },
    {
      "epoch": 0.5899211980145265,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.817673921585083,
      "step": 7012
    },
    {
      "epoch": 0.5900894584817297,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8321360349655151,
      "step": 7014
    },
    {
      "epoch": 0.590257718948933,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.858731985092163,
      "step": 7016
    },
    {
      "epoch": 0.5904259794161362,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8210113048553467,
      "step": 7018
    },
    {
      "epoch": 0.5905942398833394,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8203914165496826,
      "step": 7020
    },
    {
      "epoch": 0.5907625003505427,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.846574306488037,
      "step": 7022
    },
    {
      "epoch": 0.5909307608177459,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8140504360198975,
      "step": 7024
    },
    {
      "epoch": 0.5910990212849491,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8206865787506104,
      "step": 7026
    },
    {
      "epoch": 0.5912672817521524,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.825526237487793,
      "step": 7028
    },
    {
      "epoch": 0.5914355422193556,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8551273345947266,
      "step": 7030
    },
    {
      "epoch": 0.5916038026865588,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8681013584136963,
      "step": 7032
    },
    {
      "epoch": 0.5917720631537621,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.7949274778366089,
      "step": 7034
    },
    {
      "epoch": 0.5919403236209653,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8755393028259277,
      "step": 7036
    },
    {
      "epoch": 0.5921085840881685,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0003,
      "loss": 1.8606845140457153,
      "step": 7038
    },
    {
      "epoch": 0.5922768445553717,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.7984182834625244,
      "step": 7040
    },
    {
      "epoch": 0.592445105022575,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8141541481018066,
      "step": 7042
    },
    {
      "epoch": 0.5926133654897782,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.7877318859100342,
      "step": 7044
    },
    {
      "epoch": 0.5927816259569814,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.9187488555908203,
      "step": 7046
    },
    {
      "epoch": 0.5929498864241847,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.7859327793121338,
      "step": 7048
    },
    {
      "epoch": 0.5931181468913879,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.85759437084198,
      "step": 7050
    },
    {
      "epoch": 0.5932864073585911,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8284142017364502,
      "step": 7052
    },
    {
      "epoch": 0.5934546678257944,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.8008017539978027,
      "step": 7054
    },
    {
      "epoch": 0.5936229282929976,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.819831371307373,
      "step": 7056
    },
    {
      "epoch": 0.5937911887602008,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.850856065750122,
      "step": 7058
    },
    {
      "epoch": 0.5939594492274041,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8645236492156982,
      "step": 7060
    },
    {
      "epoch": 0.5941277096946073,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8321034908294678,
      "step": 7062
    },
    {
      "epoch": 0.5942959701618105,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8636021614074707,
      "step": 7064
    },
    {
      "epoch": 0.5944642306290138,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.873770833015442,
      "step": 7066
    },
    {
      "epoch": 0.594632491096217,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8662960529327393,
      "step": 7068
    },
    {
      "epoch": 0.5948007515634202,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.803368091583252,
      "step": 7070
    },
    {
      "epoch": 0.5949690120306234,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8535311222076416,
      "step": 7072
    },
    {
      "epoch": 0.5951372724978267,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8350067138671875,
      "step": 7074
    },
    {
      "epoch": 0.5953055329650299,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.819990873336792,
      "step": 7076
    },
    {
      "epoch": 0.5954737934322331,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8298194408416748,
      "step": 7078
    },
    {
      "epoch": 0.5956420538994364,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.839776873588562,
      "step": 7080
    },
    {
      "epoch": 0.5958103143666396,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8073556423187256,
      "step": 7082
    },
    {
      "epoch": 0.5959785748338428,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8380156755447388,
      "step": 7084
    },
    {
      "epoch": 0.5961468353010461,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8331087827682495,
      "step": 7086
    },
    {
      "epoch": 0.5963150957682493,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8563323020935059,
      "step": 7088
    },
    {
      "epoch": 0.5964833562354525,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.830461859703064,
      "step": 7090
    },
    {
      "epoch": 0.5966516167026558,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8484270572662354,
      "step": 7092
    },
    {
      "epoch": 0.596819877169859,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8470641374588013,
      "step": 7094
    },
    {
      "epoch": 0.5969881376370622,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8535054922103882,
      "step": 7096
    },
    {
      "epoch": 0.5971563981042654,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.805802583694458,
      "step": 7098
    },
    {
      "epoch": 0.5973246585714687,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8068773746490479,
      "step": 7100
    },
    {
      "epoch": 0.5974929190386719,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8206642866134644,
      "step": 7102
    },
    {
      "epoch": 0.597661179505875,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8616039752960205,
      "step": 7104
    },
    {
      "epoch": 0.5978294399730784,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.8543572425842285,
      "step": 7106
    },
    {
      "epoch": 0.5979977004402816,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8932126760482788,
      "step": 7108
    },
    {
      "epoch": 0.5981659609074848,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8491090536117554,
      "step": 7110
    },
    {
      "epoch": 0.5983342213746881,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8011622428894043,
      "step": 7112
    },
    {
      "epoch": 0.5985024818418913,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8411930799484253,
      "step": 7114
    },
    {
      "epoch": 0.5986707423090945,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.7771309614181519,
      "step": 7116
    },
    {
      "epoch": 0.5988390027762978,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8803081512451172,
      "step": 7118
    },
    {
      "epoch": 0.599007263243501,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.847373127937317,
      "step": 7120
    },
    {
      "epoch": 0.5991755237107041,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8501886129379272,
      "step": 7122
    },
    {
      "epoch": 0.5993437841779073,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.7981771230697632,
      "step": 7124
    },
    {
      "epoch": 0.5995120446451107,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.829690933227539,
      "step": 7126
    },
    {
      "epoch": 0.5996803051123138,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.85075044631958,
      "step": 7128
    },
    {
      "epoch": 0.599848565579517,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8165779113769531,
      "step": 7130
    },
    {
      "epoch": 0.6000168260467204,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8434803485870361,
      "step": 7132
    },
    {
      "epoch": 0.6001850865139235,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.9078083038330078,
      "step": 7134
    },
    {
      "epoch": 0.6001850865139235,
      "eval_loss": 1.8079233169555664,
      "eval_runtime": 537.3903,
      "eval_samples_per_second": 103.58,
      "eval_steps_per_second": 12.948,
      "step": 7134
    },
    {
      "epoch": 0.6003533469811267,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8006616830825806,
      "step": 7136
    },
    {
      "epoch": 0.60052160744833,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8311787843704224,
      "step": 7138
    },
    {
      "epoch": 0.6006898679155332,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.795119047164917,
      "step": 7140
    },
    {
      "epoch": 0.6008581283827364,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8403925895690918,
      "step": 7142
    },
    {
      "epoch": 0.6010263888499398,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8435275554656982,
      "step": 7144
    },
    {
      "epoch": 0.601194649317143,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8684611320495605,
      "step": 7146
    },
    {
      "epoch": 0.6013629097843461,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8373883962631226,
      "step": 7148
    },
    {
      "epoch": 0.6015311702515495,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.812728762626648,
      "step": 7150
    },
    {
      "epoch": 0.6016994307187526,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8421393632888794,
      "step": 7152
    },
    {
      "epoch": 0.6018676911859558,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8664357662200928,
      "step": 7154
    },
    {
      "epoch": 0.602035951653159,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8379042148590088,
      "step": 7156
    },
    {
      "epoch": 0.6022042121203623,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8673436641693115,
      "step": 7158
    },
    {
      "epoch": 0.6023724725875655,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8376669883728027,
      "step": 7160
    },
    {
      "epoch": 0.6025407330547687,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.853177547454834,
      "step": 7162
    },
    {
      "epoch": 0.602708993521972,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.851130723953247,
      "step": 7164
    },
    {
      "epoch": 0.6028772539891752,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.829164981842041,
      "step": 7166
    },
    {
      "epoch": 0.6030455144563784,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8436002731323242,
      "step": 7168
    },
    {
      "epoch": 0.6032137749235817,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8047187328338623,
      "step": 7170
    },
    {
      "epoch": 0.6033820353907849,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8403449058532715,
      "step": 7172
    },
    {
      "epoch": 0.6035502958579881,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8935136795043945,
      "step": 7174
    },
    {
      "epoch": 0.6037185563251914,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8577351570129395,
      "step": 7176
    },
    {
      "epoch": 0.6038868167923946,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8490910530090332,
      "step": 7178
    },
    {
      "epoch": 0.6040550772595978,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8372259140014648,
      "step": 7180
    },
    {
      "epoch": 0.604223337726801,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8409993648529053,
      "step": 7182
    },
    {
      "epoch": 0.6043915981940043,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8592514991760254,
      "step": 7184
    },
    {
      "epoch": 0.6045598586612075,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8755689859390259,
      "step": 7186
    },
    {
      "epoch": 0.6047281191284107,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8552335500717163,
      "step": 7188
    },
    {
      "epoch": 0.604896379595614,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.7974483966827393,
      "step": 7190
    },
    {
      "epoch": 0.6050646400628172,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.824864149093628,
      "step": 7192
    },
    {
      "epoch": 0.6052329005300204,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8019510507583618,
      "step": 7194
    },
    {
      "epoch": 0.6054011609972237,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8648247718811035,
      "step": 7196
    },
    {
      "epoch": 0.6055694214644269,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.9010679721832275,
      "step": 7198
    },
    {
      "epoch": 0.6057376819316301,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8484792709350586,
      "step": 7200
    },
    {
      "epoch": 0.6059059423988334,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.890458106994629,
      "step": 7202
    },
    {
      "epoch": 0.6060742028660366,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.870463490486145,
      "step": 7204
    },
    {
      "epoch": 0.6062424633332398,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.7985284328460693,
      "step": 7206
    },
    {
      "epoch": 0.6064107238004431,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8738787174224854,
      "step": 7208
    },
    {
      "epoch": 0.6065789842676463,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8584096431732178,
      "step": 7210
    },
    {
      "epoch": 0.6067472447348495,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.84344482421875,
      "step": 7212
    },
    {
      "epoch": 0.6069155052020527,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.869478464126587,
      "step": 7214
    },
    {
      "epoch": 0.607083765669256,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8611383438110352,
      "step": 7216
    },
    {
      "epoch": 0.6072520261364592,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8714232444763184,
      "step": 7218
    },
    {
      "epoch": 0.6074202866036624,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.814186453819275,
      "step": 7220
    },
    {
      "epoch": 0.6075885470708657,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8358800411224365,
      "step": 7222
    },
    {
      "epoch": 0.6077568075380689,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003,
      "loss": 1.8675596714019775,
      "step": 7224
    },
    {
      "epoch": 0.6079250680052721,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003,
      "loss": 1.8068556785583496,
      "step": 7226
    },
    {
      "epoch": 0.6080933284724754,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8528554439544678,
      "step": 7228
    },
    {
      "epoch": 0.6082615889396786,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8384730815887451,
      "step": 7230
    },
    {
      "epoch": 0.6084298494068818,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8236913681030273,
      "step": 7232
    },
    {
      "epoch": 0.6085981098740851,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8701786994934082,
      "step": 7234
    },
    {
      "epoch": 0.6087663703412883,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8510304689407349,
      "step": 7236
    },
    {
      "epoch": 0.6089346308084915,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8104500770568848,
      "step": 7238
    },
    {
      "epoch": 0.6091028912756947,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8227343559265137,
      "step": 7240
    },
    {
      "epoch": 0.609271151742898,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8404319286346436,
      "step": 7242
    },
    {
      "epoch": 0.6094394122101012,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8228273391723633,
      "step": 7244
    },
    {
      "epoch": 0.6096076726773044,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.7559633255004883,
      "step": 7246
    },
    {
      "epoch": 0.6097759331445077,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8063523769378662,
      "step": 7248
    },
    {
      "epoch": 0.6099441936117109,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8240647315979004,
      "step": 7250
    },
    {
      "epoch": 0.6101124540789141,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.876920223236084,
      "step": 7252
    },
    {
      "epoch": 0.6102807145461174,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8840537071228027,
      "step": 7254
    },
    {
      "epoch": 0.6104489750133206,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8480528593063354,
      "step": 7256
    },
    {
      "epoch": 0.6106172354805238,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.875593662261963,
      "step": 7258
    },
    {
      "epoch": 0.6107854959477271,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8421201705932617,
      "step": 7260
    },
    {
      "epoch": 0.6109537564149303,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8531306982040405,
      "step": 7262
    },
    {
      "epoch": 0.6111220168821335,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8298646211624146,
      "step": 7264
    },
    {
      "epoch": 0.6112902773493367,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8424063920974731,
      "step": 7266
    },
    {
      "epoch": 0.61145853781654,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8393957614898682,
      "step": 7268
    },
    {
      "epoch": 0.6116267982837432,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8596782684326172,
      "step": 7270
    },
    {
      "epoch": 0.6117950587509464,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8670711517333984,
      "step": 7272
    },
    {
      "epoch": 0.6119633192181497,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8613067865371704,
      "step": 7274
    },
    {
      "epoch": 0.6121315796853529,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8522870540618896,
      "step": 7276
    },
    {
      "epoch": 0.6122998401525561,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8522813320159912,
      "step": 7278
    },
    {
      "epoch": 0.6124681006197594,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8097273111343384,
      "step": 7280
    },
    {
      "epoch": 0.6126363610869626,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8508557081222534,
      "step": 7282
    },
    {
      "epoch": 0.6128046215541658,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8792988061904907,
      "step": 7284
    },
    {
      "epoch": 0.6129728820213691,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8376301527023315,
      "step": 7286
    },
    {
      "epoch": 0.6131411424885723,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8346641063690186,
      "step": 7288
    },
    {
      "epoch": 0.6133094029557755,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8114427328109741,
      "step": 7290
    },
    {
      "epoch": 0.6134776634229788,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8374536037445068,
      "step": 7292
    },
    {
      "epoch": 0.613645923890182,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.823652744293213,
      "step": 7294
    },
    {
      "epoch": 0.6138141843573852,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.837003469467163,
      "step": 7296
    },
    {
      "epoch": 0.6139824448245884,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.849332332611084,
      "step": 7298
    },
    {
      "epoch": 0.6141507052917917,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8478364944458008,
      "step": 7300
    },
    {
      "epoch": 0.6143189657589949,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.7693346738815308,
      "step": 7302
    },
    {
      "epoch": 0.6144872262261981,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.836620807647705,
      "step": 7304
    },
    {
      "epoch": 0.6146554866934014,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.7797150611877441,
      "step": 7306
    },
    {
      "epoch": 0.6148237471606046,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8410507440567017,
      "step": 7308
    },
    {
      "epoch": 0.6149920076278078,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.838696002960205,
      "step": 7310
    },
    {
      "epoch": 0.6151602680950111,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8582481145858765,
      "step": 7312
    },
    {
      "epoch": 0.6153285285622143,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8394365310668945,
      "step": 7314
    },
    {
      "epoch": 0.6154967890294175,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8773356676101685,
      "step": 7316
    },
    {
      "epoch": 0.6156650494966208,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8445539474487305,
      "step": 7318
    },
    {
      "epoch": 0.615833309963824,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8738865852355957,
      "step": 7320
    },
    {
      "epoch": 0.6160015704310272,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8241450786590576,
      "step": 7322
    },
    {
      "epoch": 0.6161698308982304,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8049685955047607,
      "step": 7324
    },
    {
      "epoch": 0.6163380913654337,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.810915470123291,
      "step": 7326
    },
    {
      "epoch": 0.6165063518326369,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.786655068397522,
      "step": 7328
    },
    {
      "epoch": 0.6166746122998401,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.919859528541565,
      "step": 7330
    },
    {
      "epoch": 0.6168428727670434,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8191640377044678,
      "step": 7332
    },
    {
      "epoch": 0.6170111332342466,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8545387983322144,
      "step": 7334
    },
    {
      "epoch": 0.6171793937014498,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.814344882965088,
      "step": 7336
    },
    {
      "epoch": 0.6173476541686531,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8488895893096924,
      "step": 7338
    },
    {
      "epoch": 0.6175159146358563,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.804473638534546,
      "step": 7340
    },
    {
      "epoch": 0.6176841751030595,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8466086387634277,
      "step": 7342
    },
    {
      "epoch": 0.6178524355702628,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8143377304077148,
      "step": 7344
    },
    {
      "epoch": 0.618020696037466,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8480877876281738,
      "step": 7346
    },
    {
      "epoch": 0.6181889565046692,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.805263876914978,
      "step": 7348
    },
    {
      "epoch": 0.6183572169718724,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003,
      "loss": 1.8135814666748047,
      "step": 7350
    },
    {
      "epoch": 0.6185254774390757,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.838120460510254,
      "step": 7352
    },
    {
      "epoch": 0.6186937379062789,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8135740756988525,
      "step": 7354
    },
    {
      "epoch": 0.6188619983734821,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8474795818328857,
      "step": 7356
    },
    {
      "epoch": 0.6190302588406854,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8547377586364746,
      "step": 7358
    },
    {
      "epoch": 0.6191985193078886,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8394358158111572,
      "step": 7360
    },
    {
      "epoch": 0.6193667797750918,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8583683967590332,
      "step": 7362
    },
    {
      "epoch": 0.6195350402422951,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8698458671569824,
      "step": 7364
    },
    {
      "epoch": 0.6197033007094983,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8561804294586182,
      "step": 7366
    },
    {
      "epoch": 0.6198715611767015,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8730757236480713,
      "step": 7368
    },
    {
      "epoch": 0.6200398216439048,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.7862540483474731,
      "step": 7370
    },
    {
      "epoch": 0.620208082111108,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8267453908920288,
      "step": 7372
    },
    {
      "epoch": 0.6203763425783112,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8234890699386597,
      "step": 7374
    },
    {
      "epoch": 0.6205446030455145,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.852694034576416,
      "step": 7376
    },
    {
      "epoch": 0.6207128635127177,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8218882083892822,
      "step": 7378
    },
    {
      "epoch": 0.6208811239799209,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003,
      "loss": 1.8092528581619263,
      "step": 7380
    },
    {
      "epoch": 0.6210493844471241,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.802565574645996,
      "step": 7382
    },
    {
      "epoch": 0.6212176449143274,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8381379842758179,
      "step": 7384
    },
    {
      "epoch": 0.6213859053815306,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8755059242248535,
      "step": 7386
    },
    {
      "epoch": 0.6215541658487338,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.883028268814087,
      "step": 7388
    },
    {
      "epoch": 0.6217224263159371,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8489065170288086,
      "step": 7390
    },
    {
      "epoch": 0.6218906867831403,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8697960376739502,
      "step": 7392
    },
    {
      "epoch": 0.6220589472503435,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.7862565517425537,
      "step": 7394
    },
    {
      "epoch": 0.6222272077175468,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8512916564941406,
      "step": 7396
    },
    {
      "epoch": 0.62239546818475,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8046594858169556,
      "step": 7398
    },
    {
      "epoch": 0.6225637286519532,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.840925931930542,
      "step": 7400
    },
    {
      "epoch": 0.6227319891191565,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8809720277786255,
      "step": 7402
    },
    {
      "epoch": 0.6229002495863597,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8082035779953003,
      "step": 7404
    },
    {
      "epoch": 0.6230685100535629,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8288304805755615,
      "step": 7406
    },
    {
      "epoch": 0.6232367705207661,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8375147581100464,
      "step": 7408
    },
    {
      "epoch": 0.6234050309879694,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8410577774047852,
      "step": 7410
    },
    {
      "epoch": 0.6235732914551726,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8715112209320068,
      "step": 7412
    },
    {
      "epoch": 0.6237415519223758,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8646862506866455,
      "step": 7414
    },
    {
      "epoch": 0.6239098123895791,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8449149131774902,
      "step": 7416
    },
    {
      "epoch": 0.6240780728567823,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8144779205322266,
      "step": 7418
    },
    {
      "epoch": 0.6242463333239855,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8378117084503174,
      "step": 7420
    },
    {
      "epoch": 0.6244145937911888,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.840905785560608,
      "step": 7422
    },
    {
      "epoch": 0.624582854258392,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8194403648376465,
      "step": 7424
    },
    {
      "epoch": 0.6247511147255952,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.847280502319336,
      "step": 7426
    },
    {
      "epoch": 0.6249193751927985,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8290613889694214,
      "step": 7428
    },
    {
      "epoch": 0.6250876356600017,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8178998231887817,
      "step": 7430
    },
    {
      "epoch": 0.6252558961272049,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8032876253128052,
      "step": 7432
    },
    {
      "epoch": 0.6254241565944082,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8285694122314453,
      "step": 7434
    },
    {
      "epoch": 0.6255924170616114,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8750134706497192,
      "step": 7436
    },
    {
      "epoch": 0.6257606775288146,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8646817207336426,
      "step": 7438
    },
    {
      "epoch": 0.6259289379960178,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8798801898956299,
      "step": 7440
    },
    {
      "epoch": 0.6260971984632211,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8203665018081665,
      "step": 7442
    },
    {
      "epoch": 0.6262654589304243,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8305810689926147,
      "step": 7444
    },
    {
      "epoch": 0.6264337193976275,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8780293464660645,
      "step": 7446
    },
    {
      "epoch": 0.6266019798648308,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8510701656341553,
      "step": 7448
    },
    {
      "epoch": 0.626770240332034,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8268095254898071,
      "step": 7450
    },
    {
      "epoch": 0.6269385007992372,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8642488718032837,
      "step": 7452
    },
    {
      "epoch": 0.6271067612664405,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8681223392486572,
      "step": 7454
    },
    {
      "epoch": 0.6272750217336437,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8760343790054321,
      "step": 7456
    },
    {
      "epoch": 0.6274432822008469,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.868269681930542,
      "step": 7458
    },
    {
      "epoch": 0.6276115426680502,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.754111647605896,
      "step": 7460
    },
    {
      "epoch": 0.6277798031352534,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.837515950202942,
      "step": 7462
    },
    {
      "epoch": 0.6279480636024566,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8083451986312866,
      "step": 7464
    },
    {
      "epoch": 0.6281163240696598,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.840416669845581,
      "step": 7466
    },
    {
      "epoch": 0.6282845845368631,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.7958297729492188,
      "step": 7468
    },
    {
      "epoch": 0.6284528450040663,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.880020022392273,
      "step": 7470
    },
    {
      "epoch": 0.6286211054712695,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8482978343963623,
      "step": 7472
    },
    {
      "epoch": 0.6287893659384728,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8321609497070312,
      "step": 7474
    },
    {
      "epoch": 0.628957626405676,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8757479190826416,
      "step": 7476
    },
    {
      "epoch": 0.6291258868728792,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.805016040802002,
      "step": 7478
    },
    {
      "epoch": 0.6292941473400825,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8362681865692139,
      "step": 7480
    },
    {
      "epoch": 0.6294624078072857,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8344452381134033,
      "step": 7482
    },
    {
      "epoch": 0.6296306682744889,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.778493881225586,
      "step": 7484
    },
    {
      "epoch": 0.6297989287416922,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8217321634292603,
      "step": 7486
    },
    {
      "epoch": 0.6299671892088954,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8139679431915283,
      "step": 7488
    },
    {
      "epoch": 0.6301354496760986,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8770191669464111,
      "step": 7490
    },
    {
      "epoch": 0.6303037101433018,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8605916500091553,
      "step": 7492
    },
    {
      "epoch": 0.6304719706105051,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.892150640487671,
      "step": 7494
    },
    {
      "epoch": 0.6306402310777083,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8697452545166016,
      "step": 7496
    },
    {
      "epoch": 0.6308084915449115,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8027575016021729,
      "step": 7498
    },
    {
      "epoch": 0.6309767520121148,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8396501541137695,
      "step": 7500
    },
    {
      "epoch": 0.631145012479318,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8278063535690308,
      "step": 7502
    },
    {
      "epoch": 0.6313132729465212,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.828863501548767,
      "step": 7504
    },
    {
      "epoch": 0.6314815334137245,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.846962332725525,
      "step": 7506
    },
    {
      "epoch": 0.6316497938809277,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8716888427734375,
      "step": 7508
    },
    {
      "epoch": 0.6318180543481309,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.850477933883667,
      "step": 7510
    },
    {
      "epoch": 0.6319863148153342,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8465169668197632,
      "step": 7512
    },
    {
      "epoch": 0.6321545752825374,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8162665367126465,
      "step": 7514
    },
    {
      "epoch": 0.6323228357497406,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8221734762191772,
      "step": 7516
    },
    {
      "epoch": 0.6324910962169439,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8390438556671143,
      "step": 7518
    },
    {
      "epoch": 0.6326593566841471,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8168342113494873,
      "step": 7520
    },
    {
      "epoch": 0.6328276171513503,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8373358249664307,
      "step": 7522
    },
    {
      "epoch": 0.6329958776185535,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8441320657730103,
      "step": 7524
    },
    {
      "epoch": 0.6331641380857568,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.816483974456787,
      "step": 7526
    },
    {
      "epoch": 0.63333239855296,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.837378740310669,
      "step": 7528
    },
    {
      "epoch": 0.6335006590201632,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8286129236221313,
      "step": 7530
    },
    {
      "epoch": 0.6336689194873665,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8451404571533203,
      "step": 7532
    },
    {
      "epoch": 0.6338371799545697,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.9010499715805054,
      "step": 7534
    },
    {
      "epoch": 0.6340054404217729,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8395307064056396,
      "step": 7536
    },
    {
      "epoch": 0.6341737008889762,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8123000860214233,
      "step": 7538
    },
    {
      "epoch": 0.6343419613561794,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8803362846374512,
      "step": 7540
    },
    {
      "epoch": 0.6345102218233826,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8039050102233887,
      "step": 7542
    },
    {
      "epoch": 0.6346784822905859,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8336094617843628,
      "step": 7544
    },
    {
      "epoch": 0.6348467427577891,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8155336380004883,
      "step": 7546
    },
    {
      "epoch": 0.6350150032249923,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8172308206558228,
      "step": 7548
    },
    {
      "epoch": 0.6351832636921955,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8453216552734375,
      "step": 7550
    },
    {
      "epoch": 0.6353515241593988,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.870932936668396,
      "step": 7552
    },
    {
      "epoch": 0.635519784626602,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.806220531463623,
      "step": 7554
    },
    {
      "epoch": 0.6356880450938052,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0003,
      "loss": 1.812788724899292,
      "step": 7556
    },
    {
      "epoch": 0.6358563055610085,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003,
      "loss": 1.8183109760284424,
      "step": 7558
    },
    {
      "epoch": 0.6360245660282117,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.7966952323913574,
      "step": 7560
    },
    {
      "epoch": 0.6361928264954149,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8688563108444214,
      "step": 7562
    },
    {
      "epoch": 0.6363610869626182,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8293383121490479,
      "step": 7564
    },
    {
      "epoch": 0.6365293474298214,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8514833450317383,
      "step": 7566
    },
    {
      "epoch": 0.6366976078970246,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8434877395629883,
      "step": 7568
    },
    {
      "epoch": 0.6368658683642279,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8560571670532227,
      "step": 7570
    },
    {
      "epoch": 0.6370341288314311,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.7986218929290771,
      "step": 7572
    },
    {
      "epoch": 0.6372023892986343,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.84461510181427,
      "step": 7574
    },
    {
      "epoch": 0.6373706497658376,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8661489486694336,
      "step": 7576
    },
    {
      "epoch": 0.6375389102330408,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.7932710647583008,
      "step": 7578
    },
    {
      "epoch": 0.637707170700244,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8979198932647705,
      "step": 7580
    },
    {
      "epoch": 0.6378754311674472,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8734074831008911,
      "step": 7582
    },
    {
      "epoch": 0.6380436916346505,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8010334968566895,
      "step": 7584
    },
    {
      "epoch": 0.6382119521018537,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8439815044403076,
      "step": 7586
    },
    {
      "epoch": 0.6383802125690569,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.852477788925171,
      "step": 7588
    },
    {
      "epoch": 0.6385484730362602,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8370468616485596,
      "step": 7590
    },
    {
      "epoch": 0.6387167335034634,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8304860591888428,
      "step": 7592
    },
    {
      "epoch": 0.6388849939706666,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8330168724060059,
      "step": 7594
    },
    {
      "epoch": 0.6390532544378699,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8161886930465698,
      "step": 7596
    },
    {
      "epoch": 0.6392215149050731,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8605055809020996,
      "step": 7598
    },
    {
      "epoch": 0.6393897753722763,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8236520290374756,
      "step": 7600
    },
    {
      "epoch": 0.6395580358394796,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8553149700164795,
      "step": 7602
    },
    {
      "epoch": 0.6397262963066828,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8023219108581543,
      "step": 7604
    },
    {
      "epoch": 0.639894556773886,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8358407020568848,
      "step": 7606
    },
    {
      "epoch": 0.6400628172410892,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.832641839981079,
      "step": 7608
    },
    {
      "epoch": 0.6402310777082925,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8047584295272827,
      "step": 7610
    },
    {
      "epoch": 0.6403993381754957,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.827697992324829,
      "step": 7612
    },
    {
      "epoch": 0.6405675986426989,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0003,
      "loss": 1.844022512435913,
      "step": 7614
    },
    {
      "epoch": 0.6407358591099022,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8990771770477295,
      "step": 7616
    },
    {
      "epoch": 0.6409041195771054,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8493306636810303,
      "step": 7618
    },
    {
      "epoch": 0.6410723800443086,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.844544529914856,
      "step": 7620
    },
    {
      "epoch": 0.6412406405115119,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8243260383605957,
      "step": 7622
    },
    {
      "epoch": 0.6414089009787151,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8475500345230103,
      "step": 7624
    },
    {
      "epoch": 0.6415771614459183,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8108158111572266,
      "step": 7626
    },
    {
      "epoch": 0.6417454219131216,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0003,
      "loss": 1.8454533815383911,
      "step": 7628
    },
    {
      "epoch": 0.6419136823803248,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8326324224472046,
      "step": 7630
    },
    {
      "epoch": 0.642081942847528,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0003,
      "loss": 1.8367586135864258,
      "step": 7632
    },
    {
      "epoch": 0.6422502033147311,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8200644254684448,
      "step": 7634
    },
    {
      "epoch": 0.6424184637819345,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8224341869354248,
      "step": 7636
    },
    {
      "epoch": 0.6425867242491377,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.7996107339859009,
      "step": 7638
    },
    {
      "epoch": 0.6427549847163408,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8572609424591064,
      "step": 7640
    },
    {
      "epoch": 0.6429232451835442,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.819139003753662,
      "step": 7642
    },
    {
      "epoch": 0.6430915056507474,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.825951099395752,
      "step": 7644
    },
    {
      "epoch": 0.6432597661179505,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8577468395233154,
      "step": 7646
    },
    {
      "epoch": 0.6434280265851539,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8057923316955566,
      "step": 7648
    },
    {
      "epoch": 0.643596287052357,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003,
      "loss": 1.8056672811508179,
      "step": 7650
    },
    {
      "epoch": 0.6437645475195602,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8487857580184937,
      "step": 7652
    },
    {
      "epoch": 0.6439328079867636,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8495349884033203,
      "step": 7654
    },
    {
      "epoch": 0.6441010684539668,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8411569595336914,
      "step": 7656
    },
    {
      "epoch": 0.64426932892117,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8189877271652222,
      "step": 7658
    },
    {
      "epoch": 0.6444375893883733,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8011200428009033,
      "step": 7660
    },
    {
      "epoch": 0.6446058498555765,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8675847053527832,
      "step": 7662
    },
    {
      "epoch": 0.6447741103227796,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.825608253479004,
      "step": 7664
    },
    {
      "epoch": 0.6449423707899828,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.836174488067627,
      "step": 7666
    },
    {
      "epoch": 0.6451106312571862,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8687796592712402,
      "step": 7668
    },
    {
      "epoch": 0.6452788917243893,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.884685754776001,
      "step": 7670
    },
    {
      "epoch": 0.6454471521915925,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0003,
      "loss": 1.8256573677062988,
      "step": 7672
    },
    {
      "epoch": 0.6456154126587959,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.841443419456482,
      "step": 7674
    },
    {
      "epoch": 0.645783673125999,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.884012222290039,
      "step": 7676
    },
    {
      "epoch": 0.6459519335932022,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.81431245803833,
      "step": 7678
    },
    {
      "epoch": 0.6461201940604056,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8237735033035278,
      "step": 7680
    },
    {
      "epoch": 0.6462884545276087,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8415145874023438,
      "step": 7682
    },
    {
      "epoch": 0.6464567149948119,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0003,
      "loss": 1.8626141548156738,
      "step": 7684
    },
    {
      "epoch": 0.6466249754620153,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.846382737159729,
      "step": 7686
    },
    {
      "epoch": 0.6467932359292184,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8715050220489502,
      "step": 7688
    },
    {
      "epoch": 0.6469614963964216,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.810410737991333,
      "step": 7690
    },
    {
      "epoch": 0.6471297568636248,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.828620433807373,
      "step": 7692
    },
    {
      "epoch": 0.6472980173308281,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8799771070480347,
      "step": 7694
    },
    {
      "epoch": 0.6474662777980313,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8336249589920044,
      "step": 7696
    },
    {
      "epoch": 0.6476345382652345,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8276855945587158,
      "step": 7698
    },
    {
      "epoch": 0.6478027987324378,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.853872537612915,
      "step": 7700
    },
    {
      "epoch": 0.647971059199641,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.866526484489441,
      "step": 7702
    },
    {
      "epoch": 0.6481393196668442,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8658195734024048,
      "step": 7704
    },
    {
      "epoch": 0.6483075801340475,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8081176280975342,
      "step": 7706
    },
    {
      "epoch": 0.6484758406012507,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8778924942016602,
      "step": 7708
    },
    {
      "epoch": 0.6486441010684539,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0003,
      "loss": 1.7858366966247559,
      "step": 7710
    },
    {
      "epoch": 0.6488123615356572,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.7742292881011963,
      "step": 7712
    },
    {
      "epoch": 0.6489806220028604,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8689510822296143,
      "step": 7714
    },
    {
      "epoch": 0.6491488824700636,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.8420354127883911,
      "step": 7716
    },
    {
      "epoch": 0.649317142937267,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.846379041671753,
      "step": 7718
    },
    {
      "epoch": 0.6494854034044701,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.835815668106079,
      "step": 7720
    },
    {
      "epoch": 0.6496536638716733,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8655577898025513,
      "step": 7722
    },
    {
      "epoch": 0.6498219243388765,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.78276526927948,
      "step": 7724
    },
    {
      "epoch": 0.6499901848060798,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8323553800582886,
      "step": 7726
    },
    {
      "epoch": 0.650158445273283,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8378878831863403,
      "step": 7728
    },
    {
      "epoch": 0.6503267057404862,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8648738861083984,
      "step": 7730
    },
    {
      "epoch": 0.6504949662076895,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8701491355895996,
      "step": 7732
    },
    {
      "epoch": 0.6506632266748927,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.827115774154663,
      "step": 7734
    },
    {
      "epoch": 0.6508314871420959,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.7939460277557373,
      "step": 7736
    },
    {
      "epoch": 0.6509997476092992,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.7971889972686768,
      "step": 7738
    },
    {
      "epoch": 0.6511680080765024,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8160996437072754,
      "step": 7740
    },
    {
      "epoch": 0.6513362685437056,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8761483430862427,
      "step": 7742
    },
    {
      "epoch": 0.6515045290109089,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8679711818695068,
      "step": 7744
    },
    {
      "epoch": 0.6516727894781121,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8440258502960205,
      "step": 7746
    },
    {
      "epoch": 0.6518410499453153,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8767352104187012,
      "step": 7748
    },
    {
      "epoch": 0.6520093104125185,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8271950483322144,
      "step": 7750
    },
    {
      "epoch": 0.6521775708797218,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8249471187591553,
      "step": 7752
    },
    {
      "epoch": 0.652345831346925,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.827711582183838,
      "step": 7754
    },
    {
      "epoch": 0.6525140918141282,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.826134443283081,
      "step": 7756
    },
    {
      "epoch": 0.6526823522813315,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8292031288146973,
      "step": 7758
    },
    {
      "epoch": 0.6528506127485347,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.80049729347229,
      "step": 7760
    },
    {
      "epoch": 0.6530188732157379,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8573752641677856,
      "step": 7762
    },
    {
      "epoch": 0.6531871336829412,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8707997798919678,
      "step": 7764
    },
    {
      "epoch": 0.6533553941501444,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8922923803329468,
      "step": 7766
    },
    {
      "epoch": 0.6535236546173476,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8655683994293213,
      "step": 7768
    },
    {
      "epoch": 0.6536919150845509,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.813483476638794,
      "step": 7770
    },
    {
      "epoch": 0.6538601755517541,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.7872185707092285,
      "step": 7772
    },
    {
      "epoch": 0.6540284360189573,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8326165676116943,
      "step": 7774
    },
    {
      "epoch": 0.6541966964861605,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8245398998260498,
      "step": 7776
    },
    {
      "epoch": 0.6543649569533638,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8160680532455444,
      "step": 7778
    },
    {
      "epoch": 0.654533217420567,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8036054372787476,
      "step": 7780
    },
    {
      "epoch": 0.6547014778877702,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8430172204971313,
      "step": 7782
    },
    {
      "epoch": 0.6548697383549735,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8384058475494385,
      "step": 7784
    },
    {
      "epoch": 0.6550379988221767,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.7982380390167236,
      "step": 7786
    },
    {
      "epoch": 0.6552062592893799,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8440253734588623,
      "step": 7788
    },
    {
      "epoch": 0.6553745197565832,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.897500991821289,
      "step": 7790
    },
    {
      "epoch": 0.6555427802237864,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.820894718170166,
      "step": 7792
    },
    {
      "epoch": 0.6557110406909896,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8465557098388672,
      "step": 7794
    },
    {
      "epoch": 0.6558793011581929,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8623625040054321,
      "step": 7796
    },
    {
      "epoch": 0.6560475616253961,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8430241346359253,
      "step": 7798
    },
    {
      "epoch": 0.6562158220925993,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8382103443145752,
      "step": 7800
    },
    {
      "epoch": 0.6563840825598026,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8901517391204834,
      "step": 7802
    },
    {
      "epoch": 0.6565523430270058,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8516309261322021,
      "step": 7804
    },
    {
      "epoch": 0.656720603494209,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8166896104812622,
      "step": 7806
    },
    {
      "epoch": 0.6568888639614122,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8410581350326538,
      "step": 7808
    },
    {
      "epoch": 0.6570571244286155,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.842523217201233,
      "step": 7810
    },
    {
      "epoch": 0.6572253848958187,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.799959421157837,
      "step": 7812
    },
    {
      "epoch": 0.6573936453630219,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8440673351287842,
      "step": 7814
    },
    {
      "epoch": 0.6575619058302252,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8416199684143066,
      "step": 7816
    },
    {
      "epoch": 0.6577301662974284,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8625376224517822,
      "step": 7818
    },
    {
      "epoch": 0.6578984267646316,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8708124160766602,
      "step": 7820
    },
    {
      "epoch": 0.6580666872318349,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8009223937988281,
      "step": 7822
    },
    {
      "epoch": 0.6582349476990381,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8321616649627686,
      "step": 7824
    },
    {
      "epoch": 0.6584032081662413,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8212242126464844,
      "step": 7826
    },
    {
      "epoch": 0.6585714686334446,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8521573543548584,
      "step": 7828
    },
    {
      "epoch": 0.6587397291006478,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.842756748199463,
      "step": 7830
    },
    {
      "epoch": 0.658907989567851,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8645305633544922,
      "step": 7832
    },
    {
      "epoch": 0.6590762500350542,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.812554121017456,
      "step": 7834
    },
    {
      "epoch": 0.6592445105022575,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.791438341140747,
      "step": 7836
    },
    {
      "epoch": 0.6594127709694607,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.7908563613891602,
      "step": 7838
    },
    {
      "epoch": 0.6595810314366639,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8117101192474365,
      "step": 7840
    },
    {
      "epoch": 0.6597492919038672,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8686367273330688,
      "step": 7842
    },
    {
      "epoch": 0.6599175523710704,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003,
      "loss": 1.8450703620910645,
      "step": 7844
    },
    {
      "epoch": 0.6600858128382736,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8365986347198486,
      "step": 7846
    },
    {
      "epoch": 0.6602540733054769,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8732472658157349,
      "step": 7848
    },
    {
      "epoch": 0.6604223337726801,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.814185619354248,
      "step": 7850
    },
    {
      "epoch": 0.6605905942398833,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8573853969573975,
      "step": 7852
    },
    {
      "epoch": 0.6607588547070866,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8345537185668945,
      "step": 7854
    },
    {
      "epoch": 0.6609271151742898,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8068974018096924,
      "step": 7856
    },
    {
      "epoch": 0.661095375641493,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.825638771057129,
      "step": 7858
    },
    {
      "epoch": 0.6612636361086962,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8861727714538574,
      "step": 7860
    },
    {
      "epoch": 0.6614318965758995,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.7997950315475464,
      "step": 7862
    },
    {
      "epoch": 0.6616001570431027,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.806443214416504,
      "step": 7864
    },
    {
      "epoch": 0.6617684175103059,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8705604076385498,
      "step": 7866
    },
    {
      "epoch": 0.6619366779775092,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8069649934768677,
      "step": 7868
    },
    {
      "epoch": 0.6621049384447124,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8063209056854248,
      "step": 7870
    },
    {
      "epoch": 0.6622731989119156,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8146018981933594,
      "step": 7872
    },
    {
      "epoch": 0.6624414593791189,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8021290302276611,
      "step": 7874
    },
    {
      "epoch": 0.6626097198463221,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.804781198501587,
      "step": 7876
    },
    {
      "epoch": 0.6627779803135253,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8086330890655518,
      "step": 7878
    },
    {
      "epoch": 0.6629462407807286,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.851076602935791,
      "step": 7880
    },
    {
      "epoch": 0.6631145012479318,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.837763786315918,
      "step": 7882
    },
    {
      "epoch": 0.663282761715135,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8487157821655273,
      "step": 7884
    },
    {
      "epoch": 0.6634510221823383,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8502088785171509,
      "step": 7886
    },
    {
      "epoch": 0.6636192826495415,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.7587324380874634,
      "step": 7888
    },
    {
      "epoch": 0.6637875431167447,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.846090316772461,
      "step": 7890
    },
    {
      "epoch": 0.6639558035839479,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8272314071655273,
      "step": 7892
    },
    {
      "epoch": 0.6641240640511512,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8476096391677856,
      "step": 7894
    },
    {
      "epoch": 0.6642923245183544,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.879351019859314,
      "step": 7896
    },
    {
      "epoch": 0.6644605849855576,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8029296398162842,
      "step": 7898
    },
    {
      "epoch": 0.6646288454527609,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8769456148147583,
      "step": 7900
    },
    {
      "epoch": 0.6647971059199641,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8628023862838745,
      "step": 7902
    },
    {
      "epoch": 0.6649653663871673,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8533155918121338,
      "step": 7904
    },
    {
      "epoch": 0.6651336268543706,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8596165180206299,
      "step": 7906
    },
    {
      "epoch": 0.6653018873215738,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8787471055984497,
      "step": 7908
    },
    {
      "epoch": 0.665470147788777,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8705534934997559,
      "step": 7910
    },
    {
      "epoch": 0.6656384082559803,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8295555114746094,
      "step": 7912
    },
    {
      "epoch": 0.6658066687231835,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9017407894134521,
      "step": 7914
    },
    {
      "epoch": 0.6659749291903867,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8544294834136963,
      "step": 7916
    },
    {
      "epoch": 0.6661431896575899,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8510642051696777,
      "step": 7918
    },
    {
      "epoch": 0.6663114501247932,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8693902492523193,
      "step": 7920
    },
    {
      "epoch": 0.6664797105919964,
      "grad_norm": 0.15625,
      "learning_rate": 0.0003,
      "loss": 1.8409603834152222,
      "step": 7922
    },
    {
      "epoch": 0.6666479710591996,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8341906070709229,
      "step": 7924
    },
    {
      "epoch": 0.6668162315264029,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8651752471923828,
      "step": 7926
    },
    {
      "epoch": 0.6669844919936061,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8370082378387451,
      "step": 7928
    },
    {
      "epoch": 0.6671527524608093,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8407988548278809,
      "step": 7930
    },
    {
      "epoch": 0.6673210129280126,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8259804248809814,
      "step": 7932
    },
    {
      "epoch": 0.6674892733952158,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8288366794586182,
      "step": 7934
    },
    {
      "epoch": 0.667657533862419,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8153266906738281,
      "step": 7936
    },
    {
      "epoch": 0.6678257943296223,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8837811946868896,
      "step": 7938
    },
    {
      "epoch": 0.6679940547968255,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003,
      "loss": 1.8488285541534424,
      "step": 7940
    },
    {
      "epoch": 0.6681623152640287,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.7971932888031006,
      "step": 7942
    },
    {
      "epoch": 0.668330575731232,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8091130256652832,
      "step": 7944
    },
    {
      "epoch": 0.6684988361984352,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8071393966674805,
      "step": 7946
    },
    {
      "epoch": 0.6686670966656384,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003,
      "loss": 1.8037028312683105,
      "step": 7948
    },
    {
      "epoch": 0.6688353571328416,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8687729835510254,
      "step": 7950
    },
    {
      "epoch": 0.6690036176000449,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8438912630081177,
      "step": 7952
    },
    {
      "epoch": 0.6691718780672481,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.7963910102844238,
      "step": 7954
    },
    {
      "epoch": 0.6693401385344513,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8686320781707764,
      "step": 7956
    },
    {
      "epoch": 0.6695083990016546,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.806961178779602,
      "step": 7958
    },
    {
      "epoch": 0.6696766594688578,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.843037486076355,
      "step": 7960
    },
    {
      "epoch": 0.669844919936061,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8633898496627808,
      "step": 7962
    },
    {
      "epoch": 0.6700131804032643,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8050997257232666,
      "step": 7964
    },
    {
      "epoch": 0.6701814408704675,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.7870556116104126,
      "step": 7966
    },
    {
      "epoch": 0.6703497013376707,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8212292194366455,
      "step": 7968
    },
    {
      "epoch": 0.670517961804874,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8459593057632446,
      "step": 7970
    },
    {
      "epoch": 0.6706862222720772,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8606582880020142,
      "step": 7972
    },
    {
      "epoch": 0.6708544827392804,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8411192893981934,
      "step": 7974
    },
    {
      "epoch": 0.6710227432064836,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8148839473724365,
      "step": 7976
    },
    {
      "epoch": 0.6711910036736869,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.7922487258911133,
      "step": 7978
    },
    {
      "epoch": 0.6713592641408901,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.805362582206726,
      "step": 7980
    },
    {
      "epoch": 0.6715275246080933,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.805192232131958,
      "step": 7982
    },
    {
      "epoch": 0.6716957850752966,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8398686647415161,
      "step": 7984
    },
    {
      "epoch": 0.6718640455424998,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8576407432556152,
      "step": 7986
    },
    {
      "epoch": 0.672032306009703,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8371872901916504,
      "step": 7988
    },
    {
      "epoch": 0.6722005664769063,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8291082382202148,
      "step": 7990
    },
    {
      "epoch": 0.6723688269441095,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.895066261291504,
      "step": 7992
    },
    {
      "epoch": 0.6725370874113127,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8153672218322754,
      "step": 7994
    },
    {
      "epoch": 0.672705347878516,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.9166113138198853,
      "step": 7996
    },
    {
      "epoch": 0.6728736083457192,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.771510362625122,
      "step": 7998
    },
    {
      "epoch": 0.6730418688129224,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.847531795501709,
      "step": 8000
    },
    {
      "epoch": 0.6732101292801256,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.824177861213684,
      "step": 8002
    },
    {
      "epoch": 0.6733783897473289,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8408516645431519,
      "step": 8004
    },
    {
      "epoch": 0.6735466502145321,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8377275466918945,
      "step": 8006
    },
    {
      "epoch": 0.6737149106817353,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8211705684661865,
      "step": 8008
    },
    {
      "epoch": 0.6738831711489386,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0003,
      "loss": 1.8417319059371948,
      "step": 8010
    },
    {
      "epoch": 0.6740514316161418,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.805934190750122,
      "step": 8012
    },
    {
      "epoch": 0.674219692083345,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8728175163269043,
      "step": 8014
    },
    {
      "epoch": 0.6743879525505483,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8628644943237305,
      "step": 8016
    },
    {
      "epoch": 0.6745562130177515,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.848183035850525,
      "step": 8018
    },
    {
      "epoch": 0.6747244734849547,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8376799821853638,
      "step": 8020
    },
    {
      "epoch": 0.674892733952158,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.821729302406311,
      "step": 8022
    },
    {
      "epoch": 0.6750609944193612,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8987177610397339,
      "step": 8024
    },
    {
      "epoch": 0.6752292548865644,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0003,
      "loss": 1.8578459024429321,
      "step": 8026
    },
    {
      "epoch": 0.6753975153537677,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.776274561882019,
      "step": 8028
    },
    {
      "epoch": 0.6755657758209709,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8143327236175537,
      "step": 8030
    },
    {
      "epoch": 0.6757340362881741,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.841071367263794,
      "step": 8032
    },
    {
      "epoch": 0.6759022967553773,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0003,
      "loss": 1.7949402332305908,
      "step": 8034
    },
    {
      "epoch": 0.6760705572225806,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8528283834457397,
      "step": 8036
    },
    {
      "epoch": 0.6762388176897838,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8359286785125732,
      "step": 8038
    },
    {
      "epoch": 0.676407078156987,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.776014804840088,
      "step": 8040
    },
    {
      "epoch": 0.6765753386241903,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.820932388305664,
      "step": 8042
    },
    {
      "epoch": 0.6767435990913935,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8387861251831055,
      "step": 8044
    },
    {
      "epoch": 0.6769118595585967,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8666863441467285,
      "step": 8046
    },
    {
      "epoch": 0.6770801200258,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.859168291091919,
      "step": 8048
    },
    {
      "epoch": 0.6772483804930032,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8452932834625244,
      "step": 8050
    },
    {
      "epoch": 0.6774166409602064,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8199565410614014,
      "step": 8052
    },
    {
      "epoch": 0.6775849014274097,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8284969329833984,
      "step": 8054
    },
    {
      "epoch": 0.6777531618946129,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8433201313018799,
      "step": 8056
    },
    {
      "epoch": 0.6779214223618161,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.9011263847351074,
      "step": 8058
    },
    {
      "epoch": 0.6780896828290193,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.882885456085205,
      "step": 8060
    },
    {
      "epoch": 0.6782579432962226,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8078563213348389,
      "step": 8062
    },
    {
      "epoch": 0.6784262037634258,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.7862471342086792,
      "step": 8064
    },
    {
      "epoch": 0.678594464230629,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.7996118068695068,
      "step": 8066
    },
    {
      "epoch": 0.6787627246978323,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8620712757110596,
      "step": 8068
    },
    {
      "epoch": 0.6789309851650355,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8217202425003052,
      "step": 8070
    },
    {
      "epoch": 0.6790992456322387,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8349674940109253,
      "step": 8072
    },
    {
      "epoch": 0.679267506099442,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8095343112945557,
      "step": 8074
    },
    {
      "epoch": 0.6794357665666452,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8031532764434814,
      "step": 8076
    },
    {
      "epoch": 0.6796040270338484,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8794901371002197,
      "step": 8078
    },
    {
      "epoch": 0.6797722875010517,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8501553535461426,
      "step": 8080
    },
    {
      "epoch": 0.6799405479682549,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8423537015914917,
      "step": 8082
    },
    {
      "epoch": 0.6801088084354581,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8088983297348022,
      "step": 8084
    },
    {
      "epoch": 0.6802770689026614,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0003,
      "loss": 1.8314764499664307,
      "step": 8086
    },
    {
      "epoch": 0.6804453293698646,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.8409247398376465,
      "step": 8088
    },
    {
      "epoch": 0.6806135898370678,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.7680301666259766,
      "step": 8090
    },
    {
      "epoch": 0.680781850304271,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8173261880874634,
      "step": 8092
    },
    {
      "epoch": 0.6809501107714743,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8244731426239014,
      "step": 8094
    },
    {
      "epoch": 0.6811183712386775,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0003,
      "loss": 1.7952790260314941,
      "step": 8096
    },
    {
      "epoch": 0.6812866317058807,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.8209967613220215,
      "step": 8098
    },
    {
      "epoch": 0.681454892173084,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.83610200881958,
      "step": 8100
    },
    {
      "epoch": 0.6816231526402872,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8444850444793701,
      "step": 8102
    },
    {
      "epoch": 0.6817914131074904,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.84622323513031,
      "step": 8104
    },
    {
      "epoch": 0.6819596735746937,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0003,
      "loss": 1.8554999828338623,
      "step": 8106
    },
    {
      "epoch": 0.6821279340418969,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8529130220413208,
      "step": 8108
    },
    {
      "epoch": 0.6822961945091001,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8413969278335571,
      "step": 8110
    },
    {
      "epoch": 0.6824644549763034,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0003,
      "loss": 1.8135454654693604,
      "step": 8112
    },
    {
      "epoch": 0.6826327154435066,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8316936492919922,
      "step": 8114
    },
    {
      "epoch": 0.6828009759107098,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8915092945098877,
      "step": 8116
    },
    {
      "epoch": 0.682969236377913,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8152942657470703,
      "step": 8118
    },
    {
      "epoch": 0.6831374968451163,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8072352409362793,
      "step": 8120
    },
    {
      "epoch": 0.6833057573123195,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.8028002977371216,
      "step": 8122
    },
    {
      "epoch": 0.6834740177795227,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8085328340530396,
      "step": 8124
    },
    {
      "epoch": 0.683642278246726,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8300905227661133,
      "step": 8126
    },
    {
      "epoch": 0.6838105387139292,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8851637840270996,
      "step": 8128
    },
    {
      "epoch": 0.6839787991811324,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8300634622573853,
      "step": 8130
    },
    {
      "epoch": 0.6841470596483357,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8650834560394287,
      "step": 8132
    },
    {
      "epoch": 0.6843153201155389,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.874964952468872,
      "step": 8134
    },
    {
      "epoch": 0.6844835805827421,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8679791688919067,
      "step": 8136
    },
    {
      "epoch": 0.6846518410499454,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8493129014968872,
      "step": 8138
    },
    {
      "epoch": 0.6848201015171486,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8014229536056519,
      "step": 8140
    },
    {
      "epoch": 0.6849883619843518,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8249170780181885,
      "step": 8142
    },
    {
      "epoch": 0.685156622451555,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8464672565460205,
      "step": 8144
    },
    {
      "epoch": 0.6853248829187583,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.785599946975708,
      "step": 8146
    },
    {
      "epoch": 0.6854931433859615,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8387510776519775,
      "step": 8148
    },
    {
      "epoch": 0.6856614038531647,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8752472400665283,
      "step": 8150
    },
    {
      "epoch": 0.685829664320368,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.8631885051727295,
      "step": 8152
    },
    {
      "epoch": 0.6859979247875712,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8499914407730103,
      "step": 8154
    },
    {
      "epoch": 0.6861661852547744,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8520512580871582,
      "step": 8156
    },
    {
      "epoch": 0.6863344457219777,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8072006702423096,
      "step": 8158
    },
    {
      "epoch": 0.6865027061891809,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.854610562324524,
      "step": 8160
    },
    {
      "epoch": 0.686670966656384,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.855637788772583,
      "step": 8162
    },
    {
      "epoch": 0.6868392271235874,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8632055521011353,
      "step": 8164
    },
    {
      "epoch": 0.6870074875907906,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.8314008712768555,
      "step": 8166
    },
    {
      "epoch": 0.6871757480579938,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8198764324188232,
      "step": 8168
    },
    {
      "epoch": 0.6873440085251971,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8245679140090942,
      "step": 8170
    },
    {
      "epoch": 0.6875122689924003,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8810532093048096,
      "step": 8172
    },
    {
      "epoch": 0.6876805294596035,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8394807577133179,
      "step": 8174
    },
    {
      "epoch": 0.6878487899268066,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0003,
      "loss": 1.851392149925232,
      "step": 8176
    },
    {
      "epoch": 0.68801705039401,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8280854225158691,
      "step": 8178
    },
    {
      "epoch": 0.6881853108612132,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8086460828781128,
      "step": 8180
    },
    {
      "epoch": 0.6883535713284163,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8267700672149658,
      "step": 8182
    },
    {
      "epoch": 0.6885218317956197,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.7772901058197021,
      "step": 8184
    },
    {
      "epoch": 0.6886900922628229,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8179035186767578,
      "step": 8186
    },
    {
      "epoch": 0.688858352730026,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.8237122297286987,
      "step": 8188
    },
    {
      "epoch": 0.6890266131972294,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8195276260375977,
      "step": 8190
    },
    {
      "epoch": 0.6891948736644326,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0003,
      "loss": 1.854102611541748,
      "step": 8192
    },
    {
      "epoch": 0.6893631341316357,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.9027035236358643,
      "step": 8194
    },
    {
      "epoch": 0.689531394598839,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8106598854064941,
      "step": 8196
    },
    {
      "epoch": 0.6896996550660422,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0003,
      "loss": 1.8517906665802002,
      "step": 8198
    },
    {
      "epoch": 0.6898679155332454,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8280346393585205,
      "step": 8200
    },
    {
      "epoch": 0.6900361760004486,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8363893032073975,
      "step": 8202
    },
    {
      "epoch": 0.690204436467652,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.7867846488952637,
      "step": 8204
    },
    {
      "epoch": 0.6903726969348551,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0003,
      "loss": 1.818001389503479,
      "step": 8206
    },
    {
      "epoch": 0.6905409574020583,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8494415283203125,
      "step": 8208
    },
    {
      "epoch": 0.6907092178692616,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8204445838928223,
      "step": 8210
    },
    {
      "epoch": 0.6908774783364648,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.7869510650634766,
      "step": 8212
    },
    {
      "epoch": 0.691045738803668,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0003,
      "loss": 1.7932045459747314,
      "step": 8214
    },
    {
      "epoch": 0.6912139992708713,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8098678588867188,
      "step": 8216
    },
    {
      "epoch": 0.6913822597380745,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8456652164459229,
      "step": 8218
    },
    {
      "epoch": 0.6915505202052777,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8094148635864258,
      "step": 8220
    },
    {
      "epoch": 0.691718780672481,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8394297361373901,
      "step": 8222
    },
    {
      "epoch": 0.6918870411396842,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0003,
      "loss": 1.8297810554504395,
      "step": 8224
    },
    {
      "epoch": 0.6920553016068874,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0003,
      "loss": 1.8427808284759521,
      "step": 8226
    },
    {
      "epoch": 0.6922235620740906,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0003,
      "loss": 1.749032735824585,
      "step": 8228
    },
    {
      "epoch": 0.6923918225412939,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0003,
      "loss": 1.8308714628219604,
      "step": 8230
    },
    {
      "epoch": 0.6925600830084971,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0003,
      "loss": 1.8629499673843384,
      "step": 8232
    },
    {
      "epoch": 0.6927283434757003,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8528633117675781,
      "step": 8234
    },
    {
      "epoch": 0.6928966039429036,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0003,
      "loss": 1.8481330871582031,
      "step": 8236
    },
    {
      "epoch": 0.6930648644101068,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0003,
      "loss": 1.8658093214035034,
      "step": 8238
    },
    {
      "epoch": 0.69323312487731,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8302016258239746,
      "step": 8240
    },
    {
      "epoch": 0.6934013853445133,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0003,
      "loss": 1.8194986581802368,
      "step": 8242
    },
    {
      "epoch": 0.6935696458117165,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.841973066329956,
      "step": 8244
    },
    {
      "epoch": 0.6937379062789197,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.830629587173462,
      "step": 8246
    },
    {
      "epoch": 0.693906166746123,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0003,
      "loss": 1.8813130855560303,
      "step": 8248
    },
    {
      "epoch": 0.6940744272133262,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003,
      "loss": 1.7970116138458252,
      "step": 8250
    },
    {
      "epoch": 0.6942426876805294,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.856227159500122,
      "step": 8252
    },
    {
      "epoch": 0.6944109481477327,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0003,
      "loss": 1.8441555500030518,
      "step": 8254
    },
    {
      "epoch": 0.6945792086149359,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8511348962783813,
      "step": 8256
    },
    {
      "epoch": 0.6947474690821391,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8314862251281738,
      "step": 8258
    },
    {
      "epoch": 0.6949157295493423,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.810760259628296,
      "step": 8260
    },
    {
      "epoch": 0.6950839900165456,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.7952719926834106,
      "step": 8262
    },
    {
      "epoch": 0.6952522504837488,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8841428756713867,
      "step": 8264
    },
    {
      "epoch": 0.695420510950952,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8040802478790283,
      "step": 8266
    },
    {
      "epoch": 0.6955887714181553,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0003,
      "loss": 1.7851977348327637,
      "step": 8268
    },
    {
      "epoch": 0.6957570318853585,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.897027611732483,
      "step": 8270
    },
    {
      "epoch": 0.6959252923525617,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8740520477294922,
      "step": 8272
    },
    {
      "epoch": 0.696093552819765,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0003,
      "loss": 1.8518829345703125,
      "step": 8274
    },
    {
      "epoch": 0.6962618132869682,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8380463123321533,
      "step": 8276
    },
    {
      "epoch": 0.6964300737541714,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8118088245391846,
      "step": 8278
    },
    {
      "epoch": 0.6965983342213747,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0003,
      "loss": 1.8084231615066528,
      "step": 8280
    },
    {
      "epoch": 0.6967665946885779,
      "grad_norm": 0.140625,
      "learning_rate": 0.0003,
      "loss": 1.8307759761810303,
      "step": 8282
    },
    {
      "epoch": 0.6969348551557811,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0003,
      "loss": 1.843420147895813,
      "step": 8284
    },
    {
      "epoch": 0.6971031156229843,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.7971837520599365,
      "step": 8286
    },
    {
      "epoch": 0.6972713760901876,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8395297527313232,
      "step": 8288
    },
    {
      "epoch": 0.6974396365573908,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0003,
      "loss": 1.855322003364563,
      "step": 8290
    },
    {
      "epoch": 0.697607897024594,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8819184303283691,
      "step": 8292
    },
    {
      "epoch": 0.6977761574917973,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8137657642364502,
      "step": 8294
    },
    {
      "epoch": 0.6979444179590005,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8113319873809814,
      "step": 8296
    },
    {
      "epoch": 0.6981126784262037,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.8755836486816406,
      "step": 8298
    },
    {
      "epoch": 0.698280938893407,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8612651824951172,
      "step": 8300
    },
    {
      "epoch": 0.6984491993606102,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0003,
      "loss": 1.8287386894226074,
      "step": 8302
    },
    {
      "epoch": 0.6986174598278134,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.8156287670135498,
      "step": 8304
    },
    {
      "epoch": 0.6987857202950167,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0003,
      "loss": 1.8211485147476196,
      "step": 8306
    },
    {
      "epoch": 0.6989539807622199,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0003,
      "loss": 1.8356273174285889,
      "step": 8308
    },
    {
      "epoch": 0.6991222412294231,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0003,
      "loss": 1.8296780586242676,
      "step": 8310
    },
    {
      "epoch": 0.6992905016966264,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0003,
      "loss": 1.8251934051513672,
      "step": 8312
    },
    {
      "epoch": 0.6994587621638296,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0003,
      "loss": 1.8311152458190918,
      "step": 8314
    },
    {
      "epoch": 0.6996270226310328,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0003,
      "loss": 1.840162754058838,
      "step": 8316
    },
    {
      "epoch": 0.699795283098236,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0003,
      "loss": 1.8406784534454346,
      "step": 8318
    },
    {
      "epoch": 0.6999635435654393,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003,
      "loss": 1.8368821144104004,
      "step": 8320
    },
    {
      "epoch": 0.7001318040326425,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002999999418225392,
      "loss": 1.7895938158035278,
      "step": 8322
    },
    {
      "epoch": 0.7003000644998457,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002999994764031238,
      "loss": 1.8990931510925293,
      "step": 8324
    },
    {
      "epoch": 0.700468324967049,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000299998545565737,
      "loss": 1.8468278646469116,
      "step": 8326
    },
    {
      "epoch": 0.7006365854342522,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002999971493132671,
      "loss": 1.8570469617843628,
      "step": 8328
    },
    {
      "epoch": 0.7008048459014554,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002999952876500463,
      "loss": 1.7982532978057861,
      "step": 8330
    },
    {
      "epoch": 0.7009731063686587,
      "grad_norm": 0.16015625,
      "learning_rate": 0.00029999296058185105,
      "loss": 1.8126469850540161,
      "step": 8332
    },
    {
      "epoch": 0.7011413668358619,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00029999016811590177,
      "loss": 1.8634814023971558,
      "step": 8334
    },
    {
      "epoch": 0.7013096273030651,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00029998691026086286,
      "loss": 1.8306618928909302,
      "step": 8336
    },
    {
      "epoch": 0.7014778877702684,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002999831870268427,
      "loss": 1.8590298891067505,
      "step": 8338
    },
    {
      "epoch": 0.7016461482374716,
      "grad_norm": 0.140625,
      "learning_rate": 0.000299978998425394,
      "loss": 1.840123176574707,
      "step": 8340
    },
    {
      "epoch": 0.7018144087046748,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00029997434446951286,
      "loss": 1.860687255859375,
      "step": 8342
    },
    {
      "epoch": 0.701982669171878,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00029996922517363974,
      "loss": 1.8419744968414307,
      "step": 8344
    },
    {
      "epoch": 0.7021509296390813,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002999636405536587,
      "loss": 1.8380126953125,
      "step": 8346
    },
    {
      "epoch": 0.7023191901062845,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002999575906268977,
      "loss": 1.802882194519043,
      "step": 8348
    },
    {
      "epoch": 0.7024874505734877,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00029995107541212843,
      "loss": 1.8407187461853027,
      "step": 8350
    },
    {
      "epoch": 0.702655711040691,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002999440949295663,
      "loss": 1.8008718490600586,
      "step": 8352
    },
    {
      "epoch": 0.7028239715078942,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002999366492008703,
      "loss": 1.8852235078811646,
      "step": 8354
    },
    {
      "epoch": 0.7029922319750974,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00029992873824914306,
      "loss": 1.8709607124328613,
      "step": 8356
    },
    {
      "epoch": 0.7031604924423007,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002999203620989306,
      "loss": 1.8724839687347412,
      "step": 8358
    },
    {
      "epoch": 0.7033287529095039,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00029991152077622253,
      "loss": 1.832047939300537,
      "step": 8360
    },
    {
      "epoch": 0.7034970133767071,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002999022143084515,
      "loss": 1.78397798538208,
      "step": 8362
    },
    {
      "epoch": 0.7036652738439104,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00029989244272449374,
      "loss": 1.8401751518249512,
      "step": 8364
    },
    {
      "epoch": 0.7038335343111136,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002998822060546685,
      "loss": 1.7950260639190674,
      "step": 8366
    },
    {
      "epoch": 0.7040017947783168,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000299871504330738,
      "loss": 1.8555718660354614,
      "step": 8368
    },
    {
      "epoch": 0.70417005524552,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00029986033758590757,
      "loss": 1.8219223022460938,
      "step": 8370
    },
    {
      "epoch": 0.7043383157127233,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00029984870585482535,
      "loss": 1.7947317361831665,
      "step": 8372
    },
    {
      "epoch": 0.7045065761799265,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00029983660917358223,
      "loss": 1.8598883152008057,
      "step": 8374
    },
    {
      "epoch": 0.7046748366471297,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002998240475797118,
      "loss": 1.820469617843628,
      "step": 8376
    },
    {
      "epoch": 0.704843097114333,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002998110211121901,
      "loss": 1.8476920127868652,
      "step": 8378
    },
    {
      "epoch": 0.7050113575815362,
      "grad_norm": 0.140625,
      "learning_rate": 0.00029979752981143564,
      "loss": 1.7881208658218384,
      "step": 8380
    },
    {
      "epoch": 0.7051796180487394,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002997835737193091,
      "loss": 1.81507408618927,
      "step": 8382
    },
    {
      "epoch": 0.7053478785159427,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00029976915287911363,
      "loss": 1.879138469696045,
      "step": 8384
    },
    {
      "epoch": 0.7055161389831459,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00029975426733559393,
      "loss": 1.8123488426208496,
      "step": 8386
    },
    {
      "epoch": 0.7056843994503491,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000299738917134937,
      "loss": 1.8499782085418701,
      "step": 8388
    },
    {
      "epoch": 0.7058526599175524,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029972310232477127,
      "loss": 1.8668426275253296,
      "step": 8390
    },
    {
      "epoch": 0.7060209203847556,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000299706822954167,
      "loss": 1.8783745765686035,
      "step": 8392
    },
    {
      "epoch": 0.7061891808519588,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00029969007907363574,
      "loss": 1.8697123527526855,
      "step": 8394
    },
    {
      "epoch": 0.7063574413191621,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002996728707351303,
      "loss": 1.8130815029144287,
      "step": 8396
    },
    {
      "epoch": 0.7065257017863653,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00029965519799204465,
      "loss": 1.8494970798492432,
      "step": 8398
    },
    {
      "epoch": 0.7066939622535685,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00029963706089921384,
      "loss": 1.8107943534851074,
      "step": 8400
    },
    {
      "epoch": 0.7068622227207717,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002996184595129136,
      "loss": 1.8532679080963135,
      "step": 8402
    },
    {
      "epoch": 0.707030483187975,
      "grad_norm": 0.16015625,
      "learning_rate": 0.00029959939389086014,
      "loss": 1.8140795230865479,
      "step": 8404
    },
    {
      "epoch": 0.7071987436551782,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002995798640922103,
      "loss": 1.8145718574523926,
      "step": 8406
    },
    {
      "epoch": 0.7073670041223814,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000299559870177561,
      "loss": 1.8492653369903564,
      "step": 8408
    },
    {
      "epoch": 0.7075352645895847,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002995394122089495,
      "loss": 1.813413143157959,
      "step": 8410
    },
    {
      "epoch": 0.7077035250567879,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002995184902498525,
      "loss": 1.8011122941970825,
      "step": 8412
    },
    {
      "epoch": 0.7078717855239911,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00029949710436518676,
      "loss": 1.8464747667312622,
      "step": 8414
    },
    {
      "epoch": 0.7080400459911944,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002994752546213083,
      "loss": 1.823185682296753,
      "step": 8416
    },
    {
      "epoch": 0.7082083064583976,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002994529410860124,
      "loss": 1.8128830194473267,
      "step": 8418
    },
    {
      "epoch": 0.7083765669256008,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029943016382853353,
      "loss": 1.8150964975357056,
      "step": 8420
    },
    {
      "epoch": 0.7085448273928041,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002994069229195448,
      "loss": 1.8603503704071045,
      "step": 8422
    },
    {
      "epoch": 0.7087130878600073,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002993832184311581,
      "loss": 1.8123068809509277,
      "step": 8424
    },
    {
      "epoch": 0.7088813483272105,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002993590504369236,
      "loss": 1.8259176015853882,
      "step": 8426
    },
    {
      "epoch": 0.7090496087944137,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00029933441901182974,
      "loss": 1.7873989343643188,
      "step": 8428
    },
    {
      "epoch": 0.709217869261617,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00029930932423230283,
      "loss": 1.8348369598388672,
      "step": 8430
    },
    {
      "epoch": 0.7093861297288202,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002992837661762069,
      "loss": 1.8651955127716064,
      "step": 8432
    },
    {
      "epoch": 0.7095543901960234,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002992577449228434,
      "loss": 1.7865402698516846,
      "step": 8434
    },
    {
      "epoch": 0.7097226506632267,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00029923126055295097,
      "loss": 1.8393077850341797,
      "step": 8436
    },
    {
      "epoch": 0.7098909111304299,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002992043131487053,
      "loss": 1.8159847259521484,
      "step": 8438
    },
    {
      "epoch": 0.7100591715976331,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002991769027937187,
      "loss": 1.8901045322418213,
      "step": 8440
    },
    {
      "epoch": 0.7102274320648364,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002991490295730399,
      "loss": 1.8446431159973145,
      "step": 8442
    },
    {
      "epoch": 0.7103956925320396,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00029912069357315393,
      "loss": 1.840991735458374,
      "step": 8444
    },
    {
      "epoch": 0.7105639529992428,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002990918948819816,
      "loss": 1.8138597011566162,
      "step": 8446
    },
    {
      "epoch": 0.7107322134664461,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00029906263358887945,
      "loss": 1.8340080976486206,
      "step": 8448
    },
    {
      "epoch": 0.7109004739336493,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002990329097846393,
      "loss": 1.906317114830017,
      "step": 8450
    },
    {
      "epoch": 0.7110687344008525,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00029900272356148816,
      "loss": 1.8040978908538818,
      "step": 8452
    },
    {
      "epoch": 0.7112369948680558,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00029897207501308764,
      "loss": 1.774714708328247,
      "step": 8454
    },
    {
      "epoch": 0.711405255335259,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002989409642345341,
      "loss": 1.8616214990615845,
      "step": 8456
    },
    {
      "epoch": 0.7115735158024622,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00029890939132235796,
      "loss": 1.9072380065917969,
      "step": 8458
    },
    {
      "epoch": 0.7117417762696654,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00029887735637452344,
      "loss": 1.7945291996002197,
      "step": 8460
    },
    {
      "epoch": 0.7119100367368687,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002988448594904287,
      "loss": 1.8131639957427979,
      "step": 8462
    },
    {
      "epoch": 0.7120782972040719,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00029881190077090476,
      "loss": 1.8208391666412354,
      "step": 8464
    },
    {
      "epoch": 0.7122465576712751,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000298778480318216,
      "loss": 1.8290441036224365,
      "step": 8466
    },
    {
      "epoch": 0.7124148181384784,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002987445982360593,
      "loss": 1.847928524017334,
      "step": 8468
    },
    {
      "epoch": 0.7125830786056816,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002987102546295638,
      "loss": 1.8702664375305176,
      "step": 8470
    },
    {
      "epoch": 0.7127513390728848,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00029867544960529086,
      "loss": 1.8162468671798706,
      "step": 8472
    },
    {
      "epoch": 0.7129195995400881,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002986401832712334,
      "loss": 1.7918552160263062,
      "step": 8474
    },
    {
      "epoch": 0.7130878600072913,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002986044557368156,
      "loss": 1.811090350151062,
      "step": 8476
    },
    {
      "epoch": 0.7132561204744945,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002985682671128928,
      "loss": 1.864563226699829,
      "step": 8478
    },
    {
      "epoch": 0.7134243809416978,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002985316175117511,
      "loss": 1.8263646364212036,
      "step": 8480
    },
    {
      "epoch": 0.713592641408901,
      "grad_norm": 0.15625,
      "learning_rate": 0.00029849450704710656,
      "loss": 1.8516467809677124,
      "step": 8482
    },
    {
      "epoch": 0.7137609018761042,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002984569358341056,
      "loss": 1.8400070667266846,
      "step": 8484
    },
    {
      "epoch": 0.7139291623433074,
      "grad_norm": 0.140625,
      "learning_rate": 0.00029841890398932405,
      "loss": 1.8301100730895996,
      "step": 8486
    },
    {
      "epoch": 0.7140974228105107,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00029838041163076696,
      "loss": 1.8381800651550293,
      "step": 8488
    },
    {
      "epoch": 0.7142656832777139,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00029834145887786836,
      "loss": 1.8052544593811035,
      "step": 8490
    },
    {
      "epoch": 0.7144339437449171,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002983020458514907,
      "loss": 1.8245983123779297,
      "step": 8492
    },
    {
      "epoch": 0.7146022042121204,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002982621726739247,
      "loss": 1.829383373260498,
      "step": 8494
    },
    {
      "epoch": 0.7147704646793236,
      "grad_norm": 0.16015625,
      "learning_rate": 0.00029822183946888865,
      "loss": 1.8224300146102905,
      "step": 8496
    },
    {
      "epoch": 0.7149387251465268,
      "grad_norm": 0.16015625,
      "learning_rate": 0.00029818104636152834,
      "loss": 1.7657434940338135,
      "step": 8498
    },
    {
      "epoch": 0.7151069856137301,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002981397934784166,
      "loss": 1.8928970098495483,
      "step": 8500
    },
    {
      "epoch": 0.7152752460809333,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00029809808094755265,
      "loss": 1.796283483505249,
      "step": 8502
    },
    {
      "epoch": 0.7154435065481365,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00029805590889836204,
      "loss": 1.8259165287017822,
      "step": 8504
    },
    {
      "epoch": 0.7156117670153398,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002980132774616961,
      "loss": 1.8491019010543823,
      "step": 8506
    },
    {
      "epoch": 0.715780027482543,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002979701867698316,
      "loss": 1.842362403869629,
      "step": 8508
    },
    {
      "epoch": 0.7159482879497462,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00029792663695647016,
      "loss": 1.8153791427612305,
      "step": 8510
    },
    {
      "epoch": 0.7161165484169494,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029788262815673807,
      "loss": 1.8616528511047363,
      "step": 8512
    },
    {
      "epoch": 0.7162848088841527,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002978381605071856,
      "loss": 1.8350731134414673,
      "step": 8514
    },
    {
      "epoch": 0.7164530693513559,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000297793234145787,
      "loss": 1.7891151905059814,
      "step": 8516
    },
    {
      "epoch": 0.7166213298185591,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002977478492119395,
      "loss": 1.813618540763855,
      "step": 8518
    },
    {
      "epoch": 0.7167895902857624,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002977020058464635,
      "loss": 1.8115320205688477,
      "step": 8520
    },
    {
      "epoch": 0.7169578507529656,
      "grad_norm": 0.16015625,
      "learning_rate": 0.00029765570419160165,
      "loss": 1.8912737369537354,
      "step": 8522
    },
    {
      "epoch": 0.7171261112201688,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00029760894439101855,
      "loss": 1.8339840173721313,
      "step": 8524
    },
    {
      "epoch": 0.7172943716873721,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002975617265898004,
      "loss": 1.8320006132125854,
      "step": 8526
    },
    {
      "epoch": 0.7174626321545753,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002975140509344545,
      "loss": 1.865631341934204,
      "step": 8528
    },
    {
      "epoch": 0.7176308926217785,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002974659175729086,
      "loss": 1.8177330493927002,
      "step": 8530
    },
    {
      "epoch": 0.7177991530889818,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029741732665451096,
      "loss": 1.867095947265625,
      "step": 8532
    },
    {
      "epoch": 0.717967413556185,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002973682783300293,
      "loss": 1.8560519218444824,
      "step": 8534
    },
    {
      "epoch": 0.7181356740233882,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002973187727516505,
      "loss": 1.8126368522644043,
      "step": 8536
    },
    {
      "epoch": 0.7183039344905915,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002972688100729805,
      "loss": 1.8145861625671387,
      "step": 8538
    },
    {
      "epoch": 0.7184721949577947,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002972183904490433,
      "loss": 1.8034918308258057,
      "step": 8540
    },
    {
      "epoch": 0.7186404554249979,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00029716751403628075,
      "loss": 1.8924615383148193,
      "step": 8542
    },
    {
      "epoch": 0.7188087158922011,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029711618099255205,
      "loss": 1.8011823892593384,
      "step": 8544
    },
    {
      "epoch": 0.7189769763594044,
      "grad_norm": 0.140625,
      "learning_rate": 0.00029706439147713325,
      "loss": 1.8225728273391724,
      "step": 8546
    },
    {
      "epoch": 0.7191452368266076,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00029701214565071673,
      "loss": 1.8105340003967285,
      "step": 8548
    },
    {
      "epoch": 0.7193134972938108,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002969594436754107,
      "loss": 1.8535929918289185,
      "step": 8550
    },
    {
      "epoch": 0.7194817577610141,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00029690628571473864,
      "loss": 1.8279025554656982,
      "step": 8552
    },
    {
      "epoch": 0.7196500182282173,
      "grad_norm": 0.15625,
      "learning_rate": 0.00029685267193363906,
      "loss": 1.8574192523956299,
      "step": 8554
    },
    {
      "epoch": 0.7198182786954205,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002967986024984645,
      "loss": 1.8263403177261353,
      "step": 8556
    },
    {
      "epoch": 0.7199865391626238,
      "grad_norm": 0.16015625,
      "learning_rate": 0.00029674407757698156,
      "loss": 1.7647042274475098,
      "step": 8558
    },
    {
      "epoch": 0.720154799629827,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00029668909733836994,
      "loss": 1.806577444076538,
      "step": 8560
    },
    {
      "epoch": 0.7203230600970302,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00029663366195322236,
      "loss": 1.860095500946045,
      "step": 8562
    },
    {
      "epoch": 0.7204913205642335,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002965777715935433,
      "loss": 1.818493127822876,
      "step": 8564
    },
    {
      "epoch": 0.7206595810314367,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00029652142643274947,
      "loss": 1.850123643875122,
      "step": 8566
    },
    {
      "epoch": 0.7208278414986399,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002964646266456683,
      "loss": 1.8694918155670166,
      "step": 8568
    },
    {
      "epoch": 0.7209961019658431,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002964073724085382,
      "loss": 1.7968498468399048,
      "step": 8570
    },
    {
      "epoch": 0.7211643624330464,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002963496638990073,
      "loss": 1.8908193111419678,
      "step": 8572
    },
    {
      "epoch": 0.7213326229002496,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00029629150129613347,
      "loss": 1.7941346168518066,
      "step": 8574
    },
    {
      "epoch": 0.7215008833674528,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002962328847803835,
      "loss": 1.8390626907348633,
      "step": 8576
    },
    {
      "epoch": 0.7216691438346561,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002961738145336325,
      "loss": 1.8539061546325684,
      "step": 8578
    },
    {
      "epoch": 0.7218374043018593,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00029611429073916353,
      "loss": 1.8663713932037354,
      "step": 8580
    },
    {
      "epoch": 0.7220056647690625,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00029605431358166684,
      "loss": 1.823725700378418,
      "step": 8582
    },
    {
      "epoch": 0.7221739252362658,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002959938832472394,
      "loss": 1.8074278831481934,
      "step": 8584
    },
    {
      "epoch": 0.722342185703469,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002959329999233842,
      "loss": 1.801011085510254,
      "step": 8586
    },
    {
      "epoch": 0.7225104461706722,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00029587166379901,
      "loss": 1.8581907749176025,
      "step": 8588
    },
    {
      "epoch": 0.7226787066378755,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00029580987506443027,
      "loss": 1.811882734298706,
      "step": 8590
    },
    {
      "epoch": 0.7228469671050787,
      "grad_norm": 0.15625,
      "learning_rate": 0.000295747633911363,
      "loss": 1.850489854812622,
      "step": 8592
    },
    {
      "epoch": 0.7230152275722819,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00029568494053292975,
      "loss": 1.8540327548980713,
      "step": 8594
    },
    {
      "epoch": 0.7231834880394852,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002956217951236555,
      "loss": 1.8148231506347656,
      "step": 8596
    },
    {
      "epoch": 0.7233517485066884,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00029555819787946755,
      "loss": 1.8378381729125977,
      "step": 8598
    },
    {
      "epoch": 0.7235200089738916,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002954941489976953,
      "loss": 1.8469455242156982,
      "step": 8600
    },
    {
      "epoch": 0.7236882694410948,
      "grad_norm": 0.15625,
      "learning_rate": 0.00029542964867706946,
      "loss": 1.8631083965301514,
      "step": 8602
    },
    {
      "epoch": 0.7238565299082981,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029536469711772146,
      "loss": 1.8576703071594238,
      "step": 8604
    },
    {
      "epoch": 0.7240247903755013,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002952992945211828,
      "loss": 1.805189847946167,
      "step": 8606
    },
    {
      "epoch": 0.7241930508427045,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002952334410903845,
      "loss": 1.825810194015503,
      "step": 8608
    },
    {
      "epoch": 0.7243613113099078,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00029516713702965633,
      "loss": 1.8477048873901367,
      "step": 8610
    },
    {
      "epoch": 0.724529571777111,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002951003825447265,
      "loss": 1.794377088546753,
      "step": 8612
    },
    {
      "epoch": 0.7246978322443142,
      "grad_norm": 0.158203125,
      "learning_rate": 0.00029503317784272047,
      "loss": 1.802333950996399,
      "step": 8614
    },
    {
      "epoch": 0.7248660927115175,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00029496552313216097,
      "loss": 1.8569632768630981,
      "step": 8616
    },
    {
      "epoch": 0.7250343531787207,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002948974186229667,
      "loss": 1.7957661151885986,
      "step": 8618
    },
    {
      "epoch": 0.7252026136459239,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00029482886452645234,
      "loss": 1.791773796081543,
      "step": 8620
    },
    {
      "epoch": 0.7253708741131272,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002947598610553272,
      "loss": 1.7736616134643555,
      "step": 8622
    },
    {
      "epoch": 0.7255391345803304,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002946904084236951,
      "loss": 1.8701753616333008,
      "step": 8624
    },
    {
      "epoch": 0.7257073950475336,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00029462050684705347,
      "loss": 1.8782552480697632,
      "step": 8626
    },
    {
      "epoch": 0.7258756555147368,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002945501565422928,
      "loss": 1.8375273942947388,
      "step": 8628
    },
    {
      "epoch": 0.7260439159819401,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002944793577276957,
      "loss": 1.8828777074813843,
      "step": 8630
    },
    {
      "epoch": 0.7262121764491433,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002944081106229367,
      "loss": 1.8394558429718018,
      "step": 8632
    },
    {
      "epoch": 0.7263804369163465,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00029433641544908103,
      "loss": 1.8731348514556885,
      "step": 8634
    },
    {
      "epoch": 0.7265486973835498,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002942642724285843,
      "loss": 1.8701361417770386,
      "step": 8636
    },
    {
      "epoch": 0.726716957850753,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002941916817852916,
      "loss": 1.870474100112915,
      "step": 8638
    },
    {
      "epoch": 0.7268852183179562,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00029411864374443716,
      "loss": 1.8296356201171875,
      "step": 8640
    },
    {
      "epoch": 0.7270534787851595,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00029404515853264306,
      "loss": 1.8364388942718506,
      "step": 8642
    },
    {
      "epoch": 0.7272217392523627,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029397122637791904,
      "loss": 1.8552289009094238,
      "step": 8644
    },
    {
      "epoch": 0.7273899997195659,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00029389684750966166,
      "loss": 1.8046045303344727,
      "step": 8646
    },
    {
      "epoch": 0.7275582601867692,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002938220221586535,
      "loss": 1.8008126020431519,
      "step": 8648
    },
    {
      "epoch": 0.7277265206539724,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002937467505570623,
      "loss": 1.910240650177002,
      "step": 8650
    },
    {
      "epoch": 0.7278947811211756,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002936710329384407,
      "loss": 1.7932566404342651,
      "step": 8652
    },
    {
      "epoch": 0.7280630415883788,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00029359486953772513,
      "loss": 1.791978359222412,
      "step": 8654
    },
    {
      "epoch": 0.7282313020555821,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00029351826059123516,
      "loss": 1.8508353233337402,
      "step": 8656
    },
    {
      "epoch": 0.7283995625227853,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002934412063366727,
      "loss": 1.8280079364776611,
      "step": 8658
    },
    {
      "epoch": 0.7285678229899885,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002933637070131217,
      "loss": 1.859595537185669,
      "step": 8660
    },
    {
      "epoch": 0.7287360834571918,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00029328576286104675,
      "loss": 1.8328527212142944,
      "step": 8662
    },
    {
      "epoch": 0.728904343924395,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00029320737412229274,
      "loss": 1.8150734901428223,
      "step": 8664
    },
    {
      "epoch": 0.7290726043915982,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002931285410400841,
      "loss": 1.8702658414840698,
      "step": 8666
    },
    {
      "epoch": 0.7292408648588015,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00029304926385902375,
      "loss": 1.7847737073898315,
      "step": 8668
    },
    {
      "epoch": 0.7294091253260047,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002929695428250929,
      "loss": 1.7777025699615479,
      "step": 8670
    },
    {
      "epoch": 0.7295773857932079,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002928893781856497,
      "loss": 1.835672378540039,
      "step": 8672
    },
    {
      "epoch": 0.7297456462604112,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002928087701894287,
      "loss": 1.779684066772461,
      "step": 8674
    },
    {
      "epoch": 0.7299139067276144,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029272771908654026,
      "loss": 1.8175454139709473,
      "step": 8676
    },
    {
      "epoch": 0.7300821671948176,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002926462251284695,
      "loss": 1.8610165119171143,
      "step": 8678
    },
    {
      "epoch": 0.7302504276620209,
      "grad_norm": 0.140625,
      "learning_rate": 0.00029256428856807567,
      "loss": 1.7958910465240479,
      "step": 8680
    },
    {
      "epoch": 0.7304186881292241,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002924819096595913,
      "loss": 1.8771566152572632,
      "step": 8682
    },
    {
      "epoch": 0.7305869485964273,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002923990886586214,
      "loss": 1.832718014717102,
      "step": 8684
    },
    {
      "epoch": 0.7307552090636305,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002923158258221428,
      "loss": 1.8312938213348389,
      "step": 8686
    },
    {
      "epoch": 0.7309234695308338,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029223212140850317,
      "loss": 1.764393925666809,
      "step": 8688
    },
    {
      "epoch": 0.731091729998037,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.00029214797567742035,
      "loss": 1.7975504398345947,
      "step": 8690
    },
    {
      "epoch": 0.7312599904652402,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002920633888899815,
      "loss": 1.8220856189727783,
      "step": 8692
    },
    {
      "epoch": 0.7314282509324435,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00029197836130864227,
      "loss": 1.828742265701294,
      "step": 8694
    },
    {
      "epoch": 0.7315965113996467,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000291892893197226,
      "loss": 1.8719005584716797,
      "step": 8696
    },
    {
      "epoch": 0.7317647718668499,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000291806984820923,
      "loss": 1.8440322875976562,
      "step": 8698
    },
    {
      "epoch": 0.7319330323340532,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002917206364462894,
      "loss": 1.8092658519744873,
      "step": 8700
    },
    {
      "epoch": 0.7321012928012564,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002916338483412468,
      "loss": 1.8666075468063354,
      "step": 8702
    },
    {
      "epoch": 0.7322695532684595,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002915466207750811,
      "loss": 1.8518052101135254,
      "step": 8704
    },
    {
      "epoch": 0.7324378137356629,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002914589540184417,
      "loss": 1.8538782596588135,
      "step": 8706
    },
    {
      "epoch": 0.732606074202866,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.00029137084834334083,
      "loss": 1.831737756729126,
      "step": 8708
    },
    {
      "epoch": 0.7327743346700692,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00029128230402315254,
      "loss": 1.8135111331939697,
      "step": 8710
    },
    {
      "epoch": 0.7329425951372724,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029119332133261186,
      "loss": 1.7830228805541992,
      "step": 8712
    },
    {
      "epoch": 0.7331108556044758,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00029110390054781413,
      "loss": 1.8119139671325684,
      "step": 8714
    },
    {
      "epoch": 0.733279116071679,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00029101404194621384,
      "loss": 1.8360059261322021,
      "step": 8716
    },
    {
      "epoch": 0.7334473765388821,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00029092374580662404,
      "loss": 1.8133928775787354,
      "step": 8718
    },
    {
      "epoch": 0.7336156370060855,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002908330124092153,
      "loss": 1.837253451347351,
      "step": 8720
    },
    {
      "epoch": 0.7337838974732886,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00029074184203551506,
      "loss": 1.815619707107544,
      "step": 8722
    },
    {
      "epoch": 0.7339521579404918,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002906502349684065,
      "loss": 1.8207966089248657,
      "step": 8724
    },
    {
      "epoch": 0.7341204184076952,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002905581914921277,
      "loss": 1.8732283115386963,
      "step": 8726
    },
    {
      "epoch": 0.7342886788748983,
      "grad_norm": 0.140625,
      "learning_rate": 0.00029046571189227094,
      "loss": 1.8094518184661865,
      "step": 8728
    },
    {
      "epoch": 0.7344569393421015,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002903727964557817,
      "loss": 1.831796646118164,
      "step": 8730
    },
    {
      "epoch": 0.7346251998093049,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002902794454709577,
      "loss": 1.8585642576217651,
      "step": 8732
    },
    {
      "epoch": 0.734793460276508,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00029018565922744817,
      "loss": 1.8143378496170044,
      "step": 8734
    },
    {
      "epoch": 0.7349617207437112,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.00029009143801625285,
      "loss": 1.835749864578247,
      "step": 8736
    },
    {
      "epoch": 0.7351299812109144,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00028999678212972095,
      "loss": 1.8283390998840332,
      "step": 8738
    },
    {
      "epoch": 0.7352982416781177,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002899016918615506,
      "loss": 1.8292875289916992,
      "step": 8740
    },
    {
      "epoch": 0.7354665021453209,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00028980616750678756,
      "loss": 1.8367784023284912,
      "step": 8742
    },
    {
      "epoch": 0.7356347626125241,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002897102093618246,
      "loss": 1.8622620105743408,
      "step": 8744
    },
    {
      "epoch": 0.7358030230797274,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002896138177244003,
      "loss": 1.883021354675293,
      "step": 8746
    },
    {
      "epoch": 0.7359712835469306,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00028951699289359853,
      "loss": 1.8490495681762695,
      "step": 8748
    },
    {
      "epoch": 0.7361395440141338,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000289419735169847,
      "loss": 1.851457953453064,
      "step": 8750
    },
    {
      "epoch": 0.7363078044813371,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00028932204485491666,
      "loss": 1.7773983478546143,
      "step": 8752
    },
    {
      "epoch": 0.7364760649485403,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002892239222519209,
      "loss": 1.8262646198272705,
      "step": 8754
    },
    {
      "epoch": 0.7366443254157435,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002891253676653142,
      "loss": 1.7875444889068604,
      "step": 8756
    },
    {
      "epoch": 0.7368125858829468,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00028902638140089145,
      "loss": 1.8982475996017456,
      "step": 8758
    },
    {
      "epoch": 0.73698084635015,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00028892696376578696,
      "loss": 1.8104829788208008,
      "step": 8760
    },
    {
      "epoch": 0.7371491068173532,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002888271150684735,
      "loss": 1.8551642894744873,
      "step": 8762
    },
    {
      "epoch": 0.7373173672845565,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00028872683561876136,
      "loss": 1.84297513961792,
      "step": 8764
    },
    {
      "epoch": 0.7374856277517597,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00028862612572779737,
      "loss": 1.8613276481628418,
      "step": 8766
    },
    {
      "epoch": 0.7376538882189629,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00028852498570806386,
      "loss": 1.8579189777374268,
      "step": 8768
    },
    {
      "epoch": 0.7378221486861661,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002884234158733779,
      "loss": 1.7744345664978027,
      "step": 8770
    },
    {
      "epoch": 0.7379904091533694,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.00028832141653888993,
      "loss": 1.8513822555541992,
      "step": 8772
    },
    {
      "epoch": 0.7381586696205726,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002882189880210834,
      "loss": 1.8474924564361572,
      "step": 8774
    },
    {
      "epoch": 0.7383269300877758,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002881161306377731,
      "loss": 1.8539667129516602,
      "step": 8776
    },
    {
      "epoch": 0.7384951905549791,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002880128447081047,
      "loss": 1.791276454925537,
      "step": 8778
    },
    {
      "epoch": 0.7386634510221823,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002879091305525535,
      "loss": 1.7812480926513672,
      "step": 8780
    },
    {
      "epoch": 0.7388317114893855,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00028780498849292354,
      "loss": 1.8177769184112549,
      "step": 8782
    },
    {
      "epoch": 0.7389999719565888,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002877004188523464,
      "loss": 1.8328818082809448,
      "step": 8784
    },
    {
      "epoch": 0.739168232423792,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00028759542195528064,
      "loss": 1.842647671699524,
      "step": 8786
    },
    {
      "epoch": 0.7393364928909952,
      "grad_norm": 0.140625,
      "learning_rate": 0.00028748999812751026,
      "loss": 1.8094760179519653,
      "step": 8788
    },
    {
      "epoch": 0.7395047533581985,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000287384147696144,
      "loss": 1.8826112747192383,
      "step": 8790
    },
    {
      "epoch": 0.7396730138254017,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00028727787098961437,
      "loss": 1.832634687423706,
      "step": 8792
    },
    {
      "epoch": 0.7398412742926049,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002871711683376764,
      "loss": 1.9018802642822266,
      "step": 8794
    },
    {
      "epoch": 0.7400095347598081,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00028706404007140677,
      "loss": 1.8099900484085083,
      "step": 8796
    },
    {
      "epoch": 0.7401777952270114,
      "grad_norm": 0.162109375,
      "learning_rate": 0.00028695648652320285,
      "loss": 1.8037991523742676,
      "step": 8798
    },
    {
      "epoch": 0.7403460556942146,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002868485080267815,
      "loss": 1.8273701667785645,
      "step": 8800
    },
    {
      "epoch": 0.7405143161614178,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000286740104917178,
      "loss": 1.8388582468032837,
      "step": 8802
    },
    {
      "epoch": 0.7406825766286211,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002866312775307453,
      "loss": 1.8047676086425781,
      "step": 8804
    },
    {
      "epoch": 0.7408508370958243,
      "grad_norm": 0.166015625,
      "learning_rate": 0.00028652202620515273,
      "loss": 1.8514435291290283,
      "step": 8806
    },
    {
      "epoch": 0.7410190975630275,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00028641235127938497,
      "loss": 1.850799322128296,
      "step": 8808
    },
    {
      "epoch": 0.7411873580302308,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002863022530937411,
      "loss": 1.8349828720092773,
      "step": 8810
    },
    {
      "epoch": 0.741355618497434,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00028619173198983353,
      "loss": 1.865175724029541,
      "step": 8812
    },
    {
      "epoch": 0.7415238789646372,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002860807883105868,
      "loss": 1.797144889831543,
      "step": 8814
    },
    {
      "epoch": 0.7416921394318405,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002859694224002366,
      "loss": 1.8356027603149414,
      "step": 8816
    },
    {
      "epoch": 0.7418603998990437,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.00028585763460432887,
      "loss": 1.8082737922668457,
      "step": 8818
    },
    {
      "epoch": 0.7420286603662469,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00028574542526971843,
      "loss": 1.892000436782837,
      "step": 8820
    },
    {
      "epoch": 0.7421969208334502,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002856327947445681,
      "loss": 1.848557472229004,
      "step": 8822
    },
    {
      "epoch": 0.7423651813006534,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00028551974337834757,
      "loss": 1.8018662929534912,
      "step": 8824
    },
    {
      "epoch": 0.7425334417678566,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002854062715218324,
      "loss": 1.896080732345581,
      "step": 8826
    },
    {
      "epoch": 0.7427017022350598,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002852923795271026,
      "loss": 1.8765701055526733,
      "step": 8828
    },
    {
      "epoch": 0.7428699627022631,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002851780677475421,
      "loss": 1.813946008682251,
      "step": 8830
    },
    {
      "epoch": 0.7430382231694663,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002850633365378371,
      "loss": 1.8553149700164795,
      "step": 8832
    },
    {
      "epoch": 0.7432064836366695,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002849481862539753,
      "loss": 1.8383551836013794,
      "step": 8834
    },
    {
      "epoch": 0.7433747441038728,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002848326172532447,
      "loss": 1.8448152542114258,
      "step": 8836
    },
    {
      "epoch": 0.743543004571076,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002847166298942325,
      "loss": 1.84800124168396,
      "step": 8838
    },
    {
      "epoch": 0.7437112650382792,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002846002245368239,
      "loss": 1.9041719436645508,
      "step": 8840
    },
    {
      "epoch": 0.7438795255054825,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00028448340154220116,
      "loss": 1.8533892631530762,
      "step": 8842
    },
    {
      "epoch": 0.7440477859726857,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002843661612728424,
      "loss": 1.8208653926849365,
      "step": 8844
    },
    {
      "epoch": 0.7442160464398889,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00028424850409252035,
      "loss": 1.8480854034423828,
      "step": 8846
    },
    {
      "epoch": 0.7443843069070922,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00028413043036630147,
      "loss": 1.8180055618286133,
      "step": 8848
    },
    {
      "epoch": 0.7445525673742954,
      "grad_norm": 0.1640625,
      "learning_rate": 0.00028401194046054454,
      "loss": 1.8111225366592407,
      "step": 8850
    },
    {
      "epoch": 0.7447208278414986,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00028389303474289965,
      "loss": 1.8055859804153442,
      "step": 8852
    },
    {
      "epoch": 0.7448890883087018,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002837737135823073,
      "loss": 1.8445401191711426,
      "step": 8854
    },
    {
      "epoch": 0.7450573487759051,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002836539773489968,
      "loss": 1.8321324586868286,
      "step": 8856
    },
    {
      "epoch": 0.7452256092431083,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00028353382641448533,
      "loss": 1.8611403703689575,
      "step": 8858
    },
    {
      "epoch": 0.7453938697103115,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000283413261151577,
      "loss": 1.8002145290374756,
      "step": 8860
    },
    {
      "epoch": 0.7455621301775148,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00028329228193436137,
      "loss": 1.8472979068756104,
      "step": 8862
    },
    {
      "epoch": 0.745730390644718,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00028317088913821234,
      "loss": 1.8546886444091797,
      "step": 8864
    },
    {
      "epoch": 0.7458986511119212,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002830490831397872,
      "loss": 1.84401273727417,
      "step": 8866
    },
    {
      "epoch": 0.7460669115791245,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002829268643170253,
      "loss": 1.81426203250885,
      "step": 8868
    },
    {
      "epoch": 0.7462351720463277,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002828042330491469,
      "loss": 1.8220261335372925,
      "step": 8870
    },
    {
      "epoch": 0.7464034325135309,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00028268118971665185,
      "loss": 1.8639609813690186,
      "step": 8872
    },
    {
      "epoch": 0.7465716929807342,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002825577347013188,
      "loss": 1.8724309206008911,
      "step": 8874
    },
    {
      "epoch": 0.7467399534479374,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00028243386838620356,
      "loss": 1.7851307392120361,
      "step": 8876
    },
    {
      "epoch": 0.7469082139151406,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00028230959115563823,
      "loss": 1.8268041610717773,
      "step": 8878
    },
    {
      "epoch": 0.7470764743823438,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002821849033952299,
      "loss": 1.8306913375854492,
      "step": 8880
    },
    {
      "epoch": 0.7472447348495471,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002820598054918593,
      "loss": 1.819580316543579,
      "step": 8882
    },
    {
      "epoch": 0.7474129953167503,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00028193429783368,
      "loss": 1.8445980548858643,
      "step": 8884
    },
    {
      "epoch": 0.7475812557839535,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002818083808101167,
      "loss": 1.8462677001953125,
      "step": 8886
    },
    {
      "epoch": 0.7477495162511568,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002816820548118645,
      "loss": 1.8660573959350586,
      "step": 8888
    },
    {
      "epoch": 0.74791777671836,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00028155532023088734,
      "loss": 1.843395471572876,
      "step": 8890
    },
    {
      "epoch": 0.7480860371855632,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000281428177460417,
      "loss": 1.8103333711624146,
      "step": 8892
    },
    {
      "epoch": 0.7482542976527665,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00028130062689495154,
      "loss": 1.831254482269287,
      "step": 8894
    },
    {
      "epoch": 0.7484225581199697,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.00028117266893025466,
      "loss": 1.8147289752960205,
      "step": 8896
    },
    {
      "epoch": 0.7485908185871729,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000281044303963354,
      "loss": 1.8368070125579834,
      "step": 8898
    },
    {
      "epoch": 0.7487590790543762,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00028091553239253996,
      "loss": 1.869166612625122,
      "step": 8900
    },
    {
      "epoch": 0.7489273395215794,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00028078635461736465,
      "loss": 1.850893497467041,
      "step": 8902
    },
    {
      "epoch": 0.7490955999887826,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00028065677103864055,
      "loss": 1.7864880561828613,
      "step": 8904
    },
    {
      "epoch": 0.7492638604559859,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002805267820584392,
      "loss": 1.826582908630371,
      "step": 8906
    },
    {
      "epoch": 0.7494321209231891,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00028039638808009007,
      "loss": 1.842336654663086,
      "step": 8908
    },
    {
      "epoch": 0.7496003813903923,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002802655895081793,
      "loss": 1.8316229581832886,
      "step": 8910
    },
    {
      "epoch": 0.7497686418575955,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002801343867485482,
      "loss": 1.8138864040374756,
      "step": 8912
    },
    {
      "epoch": 0.7499369023247988,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002800027802082925,
      "loss": 1.8045542240142822,
      "step": 8914
    },
    {
      "epoch": 0.750105162792002,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002798707702957605,
      "loss": 1.810992956161499,
      "step": 8916
    },
    {
      "epoch": 0.7502734232592052,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00027973835742055227,
      "loss": 1.8411390781402588,
      "step": 8918
    },
    {
      "epoch": 0.7504416837264085,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00027960554199351814,
      "loss": 1.8480608463287354,
      "step": 8920
    },
    {
      "epoch": 0.7506099441936117,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002794723244267574,
      "loss": 1.8271546363830566,
      "step": 8922
    },
    {
      "epoch": 0.7507782046608149,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002793387051336172,
      "loss": 1.840609073638916,
      "step": 8924
    },
    {
      "epoch": 0.7509464651280182,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002792046845286911,
      "loss": 1.8304417133331299,
      "step": 8926
    },
    {
      "epoch": 0.7511147255952214,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002790702630278178,
      "loss": 1.8766663074493408,
      "step": 8928
    },
    {
      "epoch": 0.7512829860624246,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002789354410480802,
      "loss": 1.8685028553009033,
      "step": 8930
    },
    {
      "epoch": 0.7514512465296279,
      "grad_norm": 0.2421875,
      "learning_rate": 0.00027880021900780327,
      "loss": 1.9094637632369995,
      "step": 8932
    },
    {
      "epoch": 0.7516195069968311,
      "grad_norm": 0.21875,
      "learning_rate": 0.00027866459732655386,
      "loss": 1.903400182723999,
      "step": 8934
    },
    {
      "epoch": 0.7517877674640343,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.00027852857642513836,
      "loss": 1.8650267124176025,
      "step": 8936
    },
    {
      "epoch": 0.7519560279312375,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002783921567256022,
      "loss": 1.8494975566864014,
      "step": 8938
    },
    {
      "epoch": 0.7521242883984408,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002782553386512279,
      "loss": 1.8838376998901367,
      "step": 8940
    },
    {
      "epoch": 0.752292548865644,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002781181226265344,
      "loss": 1.8667669296264648,
      "step": 8942
    },
    {
      "epoch": 0.7524608093328472,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002779805090772751,
      "loss": 1.8888264894485474,
      "step": 8944
    },
    {
      "epoch": 0.7526290698000505,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.000277842498430437,
      "loss": 1.8632793426513672,
      "step": 8946
    },
    {
      "epoch": 0.7527973302672537,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002777040911142391,
      "loss": 1.8603894710540771,
      "step": 8948
    },
    {
      "epoch": 0.7529655907344569,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00027756528755813124,
      "loss": 1.8611618280410767,
      "step": 8950
    },
    {
      "epoch": 0.7531338512016602,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.00027742608819279285,
      "loss": 1.8905339241027832,
      "step": 8952
    },
    {
      "epoch": 0.7533021116688634,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002772864934501312,
      "loss": 1.8567075729370117,
      "step": 8954
    },
    {
      "epoch": 0.7534703721360666,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002771465037632806,
      "loss": 1.8539183139801025,
      "step": 8956
    },
    {
      "epoch": 0.7536386326032699,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002770061195666005,
      "loss": 1.8756003379821777,
      "step": 8958
    },
    {
      "epoch": 0.7538068930704731,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00027686534129567477,
      "loss": 1.8304998874664307,
      "step": 8960
    },
    {
      "epoch": 0.7539751535376763,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002767241693873098,
      "loss": 1.8155819177627563,
      "step": 8962
    },
    {
      "epoch": 0.7541434140048796,
      "grad_norm": 0.140625,
      "learning_rate": 0.00027658260427953346,
      "loss": 1.8016221523284912,
      "step": 8964
    },
    {
      "epoch": 0.7543116744720828,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00027644064641159344,
      "loss": 1.7527015209197998,
      "step": 8966
    },
    {
      "epoch": 0.754479934939286,
      "grad_norm": 0.140625,
      "learning_rate": 0.00027629829622395635,
      "loss": 1.880126953125,
      "step": 8968
    },
    {
      "epoch": 0.7546481954064892,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000276155554158306,
      "loss": 1.8578946590423584,
      "step": 8970
    },
    {
      "epoch": 0.7548164558736925,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00027601242065754196,
      "loss": 1.814528465270996,
      "step": 8972
    },
    {
      "epoch": 0.7549847163408957,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00027586889616577854,
      "loss": 1.8270596265792847,
      "step": 8974
    },
    {
      "epoch": 0.7551529768080989,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002757249811283432,
      "loss": 1.8101258277893066,
      "step": 8976
    },
    {
      "epoch": 0.7553212372753022,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000275580675991775,
      "loss": 1.8528211116790771,
      "step": 8978
    },
    {
      "epoch": 0.7554894977425054,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00027543598120382363,
      "loss": 1.8657304048538208,
      "step": 8980
    },
    {
      "epoch": 0.7556577582097086,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002752908972134477,
      "loss": 1.7975690364837646,
      "step": 8982
    },
    {
      "epoch": 0.7558260186769119,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00027514542447081334,
      "loss": 1.8102877140045166,
      "step": 8984
    },
    {
      "epoch": 0.7559942791441151,
      "grad_norm": 0.140625,
      "learning_rate": 0.00027499956342729305,
      "loss": 1.8467531204223633,
      "step": 8986
    },
    {
      "epoch": 0.7561625396113183,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00027485331453546406,
      "loss": 1.774156928062439,
      "step": 8988
    },
    {
      "epoch": 0.7563308000785216,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000274706678249107,
      "loss": 1.8520431518554688,
      "step": 8990
    },
    {
      "epoch": 0.7564990605457248,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00027455965502320463,
      "loss": 1.880547046661377,
      "step": 8992
    },
    {
      "epoch": 0.756667321012928,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002744122453139402,
      "loss": 1.822178840637207,
      "step": 8994
    },
    {
      "epoch": 0.7568355814801312,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00027426444957869613,
      "loss": 1.8449288606643677,
      "step": 8996
    },
    {
      "epoch": 0.7570038419473345,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00027411626827605255,
      "loss": 1.874418020248413,
      "step": 8998
    },
    {
      "epoch": 0.7571721024145377,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002739677018657862,
      "loss": 1.8197630643844604,
      "step": 9000
    },
    {
      "epoch": 0.7573403628817409,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002738187508088683,
      "loss": 1.8327217102050781,
      "step": 9002
    },
    {
      "epoch": 0.7575086233489442,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00027366941556746405,
      "loss": 1.8009089231491089,
      "step": 9004
    },
    {
      "epoch": 0.7576768838161474,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002735196966049302,
      "loss": 1.822040319442749,
      "step": 9006
    },
    {
      "epoch": 0.7578451442833506,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00027336959438581447,
      "loss": 1.8438992500305176,
      "step": 9008
    },
    {
      "epoch": 0.7580134047505539,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00027321910937585364,
      "loss": 1.8030643463134766,
      "step": 9010
    },
    {
      "epoch": 0.7581816652177571,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002730682420419721,
      "loss": 1.8062326908111572,
      "step": 9012
    },
    {
      "epoch": 0.7583499256849603,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00027291699285228064,
      "loss": 1.830003023147583,
      "step": 9014
    },
    {
      "epoch": 0.7585181861521636,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00027276536227607487,
      "loss": 1.7890081405639648,
      "step": 9016
    },
    {
      "epoch": 0.7586864466193668,
      "grad_norm": 0.140625,
      "learning_rate": 0.00027261335078383373,
      "loss": 1.800846815109253,
      "step": 9018
    },
    {
      "epoch": 0.75885470708657,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00027246095884721807,
      "loss": 1.818969488143921,
      "step": 9020
    },
    {
      "epoch": 0.7590229675537732,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002723081869390691,
      "loss": 1.8287935256958008,
      "step": 9022
    },
    {
      "epoch": 0.7591912280209765,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002721550355334072,
      "loss": 1.8085482120513916,
      "step": 9024
    },
    {
      "epoch": 0.7593594884881797,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00027200150510543015,
      "loss": 1.8311831951141357,
      "step": 9026
    },
    {
      "epoch": 0.7595277489553829,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002718475961315116,
      "loss": 1.823134422302246,
      "step": 9028
    },
    {
      "epoch": 0.7596960094225862,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002716933090892,
      "loss": 1.8215885162353516,
      "step": 9030
    },
    {
      "epoch": 0.7598642698897894,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002715386444572168,
      "loss": 1.8360953330993652,
      "step": 9032
    },
    {
      "epoch": 0.7600325303569926,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00027138360271545495,
      "loss": 1.8396610021591187,
      "step": 9034
    },
    {
      "epoch": 0.7602007908241959,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00027122818434497753,
      "loss": 1.8506104946136475,
      "step": 9036
    },
    {
      "epoch": 0.7603690512913991,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002710723898280163,
      "loss": 1.8471980094909668,
      "step": 9038
    },
    {
      "epoch": 0.7605373117586023,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00027091621964796996,
      "loss": 1.7951210737228394,
      "step": 9040
    },
    {
      "epoch": 0.7607055722258056,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002707596742894031,
      "loss": 1.8556747436523438,
      "step": 9042
    },
    {
      "epoch": 0.7608738326930088,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00027060275423804414,
      "loss": 1.8462584018707275,
      "step": 9044
    },
    {
      "epoch": 0.761042093160212,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002704454599807841,
      "loss": 1.829227089881897,
      "step": 9046
    },
    {
      "epoch": 0.7612103536274153,
      "grad_norm": 0.158203125,
      "learning_rate": 0.00027028779200567537,
      "loss": 1.8444693088531494,
      "step": 9048
    },
    {
      "epoch": 0.7613786140946185,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002701297508019295,
      "loss": 1.814220666885376,
      "step": 9050
    },
    {
      "epoch": 0.7615468745618217,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002699713368599165,
      "loss": 1.8565798997879028,
      "step": 9052
    },
    {
      "epoch": 0.7617151350290249,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00026981255067116267,
      "loss": 1.8087549209594727,
      "step": 9054
    },
    {
      "epoch": 0.7618833954962282,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00026965339272834937,
      "loss": 1.8458645343780518,
      "step": 9056
    },
    {
      "epoch": 0.7620516559634314,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00026949386352531136,
      "loss": 1.7924787998199463,
      "step": 9058
    },
    {
      "epoch": 0.7622199164306346,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002693339635570355,
      "loss": 1.8507633209228516,
      "step": 9060
    },
    {
      "epoch": 0.7623881768978379,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00026917369331965895,
      "loss": 1.8137531280517578,
      "step": 9062
    },
    {
      "epoch": 0.7625564373650411,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00026901305331046777,
      "loss": 1.8491499423980713,
      "step": 9064
    },
    {
      "epoch": 0.7627246978322443,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002688520440278954,
      "loss": 1.83095383644104,
      "step": 9066
    },
    {
      "epoch": 0.7628929582994476,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00026869066597152096,
      "loss": 1.7941243648529053,
      "step": 9068
    },
    {
      "epoch": 0.7630612187666508,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002685289196420678,
      "loss": 1.8680310249328613,
      "step": 9070
    },
    {
      "epoch": 0.763229479233854,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002683668055414021,
      "loss": 1.8119053840637207,
      "step": 9072
    },
    {
      "epoch": 0.7633977397010573,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000268204324172531,
      "loss": 1.8207275867462158,
      "step": 9074
    },
    {
      "epoch": 0.7635660001682605,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00026804147603960124,
      "loss": 1.8486363887786865,
      "step": 9076
    },
    {
      "epoch": 0.7637342606354637,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002678782616478976,
      "loss": 1.7962614297866821,
      "step": 9078
    },
    {
      "epoch": 0.7639025211026669,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00026771468150384127,
      "loss": 1.8605763912200928,
      "step": 9080
    },
    {
      "epoch": 0.7640707815698702,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002675507361149882,
      "loss": 1.843949794769287,
      "step": 9082
    },
    {
      "epoch": 0.7642390420370734,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002673864259900278,
      "loss": 1.859186053276062,
      "step": 9084
    },
    {
      "epoch": 0.7644073025042766,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000267221751638781,
      "loss": 1.872516393661499,
      "step": 9086
    },
    {
      "epoch": 0.7645755629714799,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000267056713572199,
      "loss": 1.8774422407150269,
      "step": 9088
    },
    {
      "epoch": 0.7647438234386831,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002668913123023614,
      "loss": 1.794471263885498,
      "step": 9090
    },
    {
      "epoch": 0.7649120839058863,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002667255483424749,
      "loss": 1.8562190532684326,
      "step": 9092
    },
    {
      "epoch": 0.7650803443730896,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002665594222068714,
      "loss": 1.8736724853515625,
      "step": 9094
    },
    {
      "epoch": 0.7652486048402928,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002663929344110066,
      "loss": 1.8307629823684692,
      "step": 9096
    },
    {
      "epoch": 0.765416865307496,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00026622608547145847,
      "loss": 1.8594938516616821,
      "step": 9098
    },
    {
      "epoch": 0.7655851257746993,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00026605887590592543,
      "loss": 1.789650797843933,
      "step": 9100
    },
    {
      "epoch": 0.7657533862419025,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00026589130623322485,
      "loss": 1.8008372783660889,
      "step": 9102
    },
    {
      "epoch": 0.7659216467091057,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00026572337697329144,
      "loss": 1.8228950500488281,
      "step": 9104
    },
    {
      "epoch": 0.7660899071763089,
      "grad_norm": 0.267578125,
      "learning_rate": 0.00026555508864717565,
      "loss": 1.864142894744873,
      "step": 9106
    },
    {
      "epoch": 0.7662581676435122,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00026538644177704214,
      "loss": 1.8520300388336182,
      "step": 9108
    },
    {
      "epoch": 0.7664264281107154,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00026521743688616786,
      "loss": 1.827566385269165,
      "step": 9110
    },
    {
      "epoch": 0.7665946885779186,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002650480744989407,
      "loss": 1.8295530080795288,
      "step": 9112
    },
    {
      "epoch": 0.7667629490451219,
      "grad_norm": 0.166015625,
      "learning_rate": 0.00026487835514085793,
      "loss": 1.8571224212646484,
      "step": 9114
    },
    {
      "epoch": 0.7669312095123251,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002647082793385243,
      "loss": 1.9172008037567139,
      "step": 9116
    },
    {
      "epoch": 0.7670994699795283,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00026453784761965054,
      "loss": 1.8055365085601807,
      "step": 9118
    },
    {
      "epoch": 0.7672677304467316,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00026436706051305165,
      "loss": 1.8266873359680176,
      "step": 9120
    },
    {
      "epoch": 0.7674359909139348,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002641959185486455,
      "loss": 1.8703672885894775,
      "step": 9122
    },
    {
      "epoch": 0.767604251381138,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002640244222574509,
      "loss": 1.8711165189743042,
      "step": 9124
    },
    {
      "epoch": 0.7677725118483413,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000263852572171586,
      "loss": 1.8267033100128174,
      "step": 9126
    },
    {
      "epoch": 0.7679407723155445,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002636803688242669,
      "loss": 1.8088641166687012,
      "step": 9128
    },
    {
      "epoch": 0.7681090327827477,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002635078127498056,
      "loss": 1.8155295848846436,
      "step": 9130
    },
    {
      "epoch": 0.768277293249951,
      "grad_norm": 0.140625,
      "learning_rate": 0.00026333490448360867,
      "loss": 1.842378854751587,
      "step": 9132
    },
    {
      "epoch": 0.7684455537171542,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002631616445621754,
      "loss": 1.851716160774231,
      "step": 9134
    },
    {
      "epoch": 0.7686138141843574,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00026298803352309616,
      "loss": 1.806494951248169,
      "step": 9136
    },
    {
      "epoch": 0.7687820746515606,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00026281407190505086,
      "loss": 1.8823492527008057,
      "step": 9138
    },
    {
      "epoch": 0.7689503351187639,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002626397602478072,
      "loss": 1.8315966129302979,
      "step": 9140
    },
    {
      "epoch": 0.7691185955859671,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002624650990922188,
      "loss": 1.8659372329711914,
      "step": 9142
    },
    {
      "epoch": 0.7692868560531703,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00026229008898022384,
      "loss": 1.8425354957580566,
      "step": 9144
    },
    {
      "epoch": 0.7694551165203736,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00026211473045484335,
      "loss": 1.7868034839630127,
      "step": 9146
    },
    {
      "epoch": 0.7696233769875768,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00026193902406017913,
      "loss": 1.8368192911148071,
      "step": 9148
    },
    {
      "epoch": 0.76979163745478,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002617629703414126,
      "loss": 1.8749134540557861,
      "step": 9150
    },
    {
      "epoch": 0.7699598979219833,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002615865698448028,
      "loss": 1.8215208053588867,
      "step": 9152
    },
    {
      "epoch": 0.7701281583891865,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002614098231176846,
      "loss": 1.8453458547592163,
      "step": 9154
    },
    {
      "epoch": 0.7702964188563897,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002612327307084674,
      "loss": 1.8618137836456299,
      "step": 9156
    },
    {
      "epoch": 0.770464679323593,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002610552931666329,
      "loss": 1.8337024450302124,
      "step": 9158
    },
    {
      "epoch": 0.7706329397907962,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002608775110427339,
      "loss": 1.802424669265747,
      "step": 9160
    },
    {
      "epoch": 0.7708012002579994,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00026069938488839235,
      "loss": 1.8085839748382568,
      "step": 9162
    },
    {
      "epoch": 0.7709694607252026,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00026052091525629754,
      "loss": 1.831002950668335,
      "step": 9164
    },
    {
      "epoch": 0.7711377211924059,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002603421027002046,
      "loss": 1.826420783996582,
      "step": 9166
    },
    {
      "epoch": 0.7713059816596091,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00026016294777493257,
      "loss": 1.8714808225631714,
      "step": 9168
    },
    {
      "epoch": 0.7714742421268123,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000259983451036363,
      "loss": 1.8171846866607666,
      "step": 9170
    },
    {
      "epoch": 0.7716425025940156,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002598036130414378,
      "loss": 1.8598706722259521,
      "step": 9172
    },
    {
      "epoch": 0.7718107630612188,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00025962343434815776,
      "loss": 1.7799298763275146,
      "step": 9174
    },
    {
      "epoch": 0.771979023528422,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00025944291551558096,
      "loss": 1.8744728565216064,
      "step": 9176
    },
    {
      "epoch": 0.7721472839956253,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00025926205710382065,
      "loss": 1.8586658239364624,
      "step": 9178
    },
    {
      "epoch": 0.7723155444628285,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002590808596740439,
      "loss": 1.8209452629089355,
      "step": 9180
    },
    {
      "epoch": 0.7724838049300317,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002588993237884696,
      "loss": 1.8673639297485352,
      "step": 9182
    },
    {
      "epoch": 0.772652065397235,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002587174500103668,
      "loss": 1.8293766975402832,
      "step": 9184
    },
    {
      "epoch": 0.7728203258644382,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002585352389040529,
      "loss": 1.8138519525527954,
      "step": 9186
    },
    {
      "epoch": 0.7729885863316414,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002583526910348921,
      "loss": 1.8325138092041016,
      "step": 9188
    },
    {
      "epoch": 0.7731568467988447,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00025816980696929346,
      "loss": 1.805180549621582,
      "step": 9190
    },
    {
      "epoch": 0.7733251072660479,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002579865872747091,
      "loss": 1.7990508079528809,
      "step": 9192
    },
    {
      "epoch": 0.7734933677332511,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00025780303251963263,
      "loss": 1.8357905149459839,
      "step": 9194
    },
    {
      "epoch": 0.7736616282004543,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00025761914327359725,
      "loss": 1.781388282775879,
      "step": 9196
    },
    {
      "epoch": 0.7738298886676576,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000257434920107174,
      "loss": 1.8220677375793457,
      "step": 9198
    },
    {
      "epoch": 0.7739981491348608,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00025725036359197,
      "loss": 1.8097245693206787,
      "step": 9200
    },
    {
      "epoch": 0.774166409602064,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002570654743006268,
      "loss": 1.8414753675460815,
      "step": 9202
    },
    {
      "epoch": 0.7743346700692673,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00025688025280681825,
      "loss": 1.764427900314331,
      "step": 9204
    },
    {
      "epoch": 0.7745029305364705,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00025669469968524927,
      "loss": 1.8986413478851318,
      "step": 9206
    },
    {
      "epoch": 0.7746711910036737,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002565088155116535,
      "loss": 1.8721706867218018,
      "step": 9208
    },
    {
      "epoch": 0.774839451470877,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00025632260086279183,
      "loss": 1.7512484788894653,
      "step": 9210
    },
    {
      "epoch": 0.7750077119380802,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002561360563164506,
      "loss": 1.787672758102417,
      "step": 9212
    },
    {
      "epoch": 0.7751759724052834,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00025594918245143975,
      "loss": 1.8259432315826416,
      "step": 9214
    },
    {
      "epoch": 0.7753442328724867,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.00025576197984759094,
      "loss": 1.8924686908721924,
      "step": 9216
    },
    {
      "epoch": 0.7755124933396899,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000255574449085756,
      "loss": 1.8172476291656494,
      "step": 9218
    },
    {
      "epoch": 0.775680753806893,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002553865907478048,
      "loss": 1.7994070053100586,
      "step": 9220
    },
    {
      "epoch": 0.7758490142740962,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00025519840541662366,
      "loss": 1.7733800411224365,
      "step": 9222
    },
    {
      "epoch": 0.7760172747412996,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002550098936761136,
      "loss": 1.769362449645996,
      "step": 9224
    },
    {
      "epoch": 0.7761855352085028,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00025482105611118825,
      "loss": 1.8385376930236816,
      "step": 9226
    },
    {
      "epoch": 0.776353795675706,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002546318933077724,
      "loss": 1.7727479934692383,
      "step": 9228
    },
    {
      "epoch": 0.7765220561429093,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00025444240585279984,
      "loss": 1.8199495077133179,
      "step": 9230
    },
    {
      "epoch": 0.7766903166101125,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00025425259433421173,
      "loss": 1.8416476249694824,
      "step": 9232
    },
    {
      "epoch": 0.7768585770773156,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00025406245934095475,
      "loss": 1.788468360900879,
      "step": 9234
    },
    {
      "epoch": 0.777026837544519,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002538720014629792,
      "loss": 1.8047715425491333,
      "step": 9236
    },
    {
      "epoch": 0.7771950980117222,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002536812212912373,
      "loss": 1.8617525100708008,
      "step": 9238
    },
    {
      "epoch": 0.7773633584789253,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00025349011941768124,
      "loss": 1.8410627841949463,
      "step": 9240
    },
    {
      "epoch": 0.7775316189461287,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002532986964352615,
      "loss": 1.826300859451294,
      "step": 9242
    },
    {
      "epoch": 0.7776998794133319,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002531069529379247,
      "loss": 1.8403160572052002,
      "step": 9244
    },
    {
      "epoch": 0.777868139880535,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000252914889520612,
      "loss": 1.8030017614364624,
      "step": 9246
    },
    {
      "epoch": 0.7780364003477382,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002527225067792573,
      "loss": 1.8145015239715576,
      "step": 9248
    },
    {
      "epoch": 0.7782046608149416,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002525298053107853,
      "loss": 1.8046293258666992,
      "step": 9250
    },
    {
      "epoch": 0.7783729212821447,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00025233678571310957,
      "loss": 1.8513628244400024,
      "step": 9252
    },
    {
      "epoch": 0.7785411817493479,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00025214344858513077,
      "loss": 1.8580496311187744,
      "step": 9254
    },
    {
      "epoch": 0.7787094422165513,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002519497945267349,
      "loss": 1.8693780899047852,
      "step": 9256
    },
    {
      "epoch": 0.7788777026837544,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00025175582413879116,
      "loss": 1.829506278038025,
      "step": 9258
    },
    {
      "epoch": 0.7790459631509576,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002515615380231503,
      "loss": 1.8642957210540771,
      "step": 9260
    },
    {
      "epoch": 0.779214223618161,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002513669367826429,
      "loss": 1.820169448852539,
      "step": 9262
    },
    {
      "epoch": 0.7793824840853641,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00025117202102107705,
      "loss": 1.8393492698669434,
      "step": 9264
    },
    {
      "epoch": 0.7795507445525673,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00025097679134323696,
      "loss": 1.8467947244644165,
      "step": 9266
    },
    {
      "epoch": 0.7797190050197707,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00025078124835488066,
      "loss": 1.8411916494369507,
      "step": 9268
    },
    {
      "epoch": 0.7798872654869738,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002505853926627384,
      "loss": 1.855724573135376,
      "step": 9270
    },
    {
      "epoch": 0.780055525954177,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002503892248745108,
      "loss": 1.8312512636184692,
      "step": 9272
    },
    {
      "epoch": 0.7802237864213803,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002501927455988666,
      "loss": 1.8183212280273438,
      "step": 9274
    },
    {
      "epoch": 0.7803920468885835,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00024999595544544133,
      "loss": 1.8326714038848877,
      "step": 9276
    },
    {
      "epoch": 0.7805603073557867,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00024979885502483476,
      "loss": 1.8117444515228271,
      "step": 9278
    },
    {
      "epoch": 0.7807285678229899,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002496014449486097,
      "loss": 1.8485164642333984,
      "step": 9280
    },
    {
      "epoch": 0.7808968282901932,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00024940372582928956,
      "loss": 1.8508542776107788,
      "step": 9282
    },
    {
      "epoch": 0.7810650887573964,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002492056982803566,
      "loss": 1.8351573944091797,
      "step": 9284
    },
    {
      "epoch": 0.7812333492245996,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002490073629162503,
      "loss": 1.8573716878890991,
      "step": 9286
    },
    {
      "epoch": 0.781401609691803,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00024880872035236505,
      "loss": 1.7899127006530762,
      "step": 9288
    },
    {
      "epoch": 0.7815698701590061,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002486097712050484,
      "loss": 1.806758165359497,
      "step": 9290
    },
    {
      "epoch": 0.7817381306262093,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00024841051609159927,
      "loss": 1.8372116088867188,
      "step": 9292
    },
    {
      "epoch": 0.7819063910934126,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00024821095563026596,
      "loss": 1.7999165058135986,
      "step": 9294
    },
    {
      "epoch": 0.7820746515606158,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.000248011090440244,
      "loss": 1.8378522396087646,
      "step": 9296
    },
    {
      "epoch": 0.782242912027819,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00024781092114167466,
      "loss": 1.865127444267273,
      "step": 9298
    },
    {
      "epoch": 0.7824111724950223,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002476104483556427,
      "loss": 1.8119968175888062,
      "step": 9300
    },
    {
      "epoch": 0.7825794329622255,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00024740967270417446,
      "loss": 1.8702692985534668,
      "step": 9302
    },
    {
      "epoch": 0.7827476934294287,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002472085948102362,
      "loss": 1.8857817649841309,
      "step": 9304
    },
    {
      "epoch": 0.7829159538966319,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002470072152977317,
      "loss": 1.862051010131836,
      "step": 9306
    },
    {
      "epoch": 0.7830842143638352,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00024680553479150097,
      "loss": 1.8528058528900146,
      "step": 9308
    },
    {
      "epoch": 0.7832524748310384,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00024660355391731764,
      "loss": 1.812973976135254,
      "step": 9310
    },
    {
      "epoch": 0.7834207352982416,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002464012733018874,
      "loss": 1.7807812690734863,
      "step": 9312
    },
    {
      "epoch": 0.7835889957654449,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00024619869357284606,
      "loss": 1.859920620918274,
      "step": 9314
    },
    {
      "epoch": 0.7837572562326481,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002459958153587574,
      "loss": 1.798720121383667,
      "step": 9316
    },
    {
      "epoch": 0.7839255166998513,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002457926392891114,
      "loss": 1.8025473356246948,
      "step": 9318
    },
    {
      "epoch": 0.7840937771670546,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00024558916599432234,
      "loss": 1.7979103326797485,
      "step": 9320
    },
    {
      "epoch": 0.7842620376342578,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00024538539610572643,
      "loss": 1.8310050964355469,
      "step": 9322
    },
    {
      "epoch": 0.784430298101461,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00024518133025558045,
      "loss": 1.830194354057312,
      "step": 9324
    },
    {
      "epoch": 0.7845985585686643,
      "grad_norm": 0.140625,
      "learning_rate": 0.00024497696907705937,
      "loss": 1.9013776779174805,
      "step": 9326
    },
    {
      "epoch": 0.7847668190358675,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002447723132042544,
      "loss": 1.7897865772247314,
      "step": 9328
    },
    {
      "epoch": 0.7849350795030707,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00024456736327217136,
      "loss": 1.8770601749420166,
      "step": 9330
    },
    {
      "epoch": 0.785103339970274,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00024436211991672827,
      "loss": 1.8251497745513916,
      "step": 9332
    },
    {
      "epoch": 0.7852716004374772,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002441565837747537,
      "loss": 1.8226170539855957,
      "step": 9334
    },
    {
      "epoch": 0.7854398609046804,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00024395075548398456,
      "loss": 1.8553736209869385,
      "step": 9336
    },
    {
      "epoch": 0.7856081213718836,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00024374463568306443,
      "loss": 1.8114097118377686,
      "step": 9338
    },
    {
      "epoch": 0.7857763818390869,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00024353822501154118,
      "loss": 1.8422126770019531,
      "step": 9340
    },
    {
      "epoch": 0.7859446423062901,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00024333152410986538,
      "loss": 1.8509352207183838,
      "step": 9342
    },
    {
      "epoch": 0.7861129027734933,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00024312453361938796,
      "loss": 1.813276767730713,
      "step": 9344
    },
    {
      "epoch": 0.7862811632406966,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00024291725418235845,
      "loss": 1.8445675373077393,
      "step": 9346
    },
    {
      "epoch": 0.7864494237078998,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002427096864419229,
      "loss": 1.788644552230835,
      "step": 9348
    },
    {
      "epoch": 0.786617684175103,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00024250183104212208,
      "loss": 1.8426542282104492,
      "step": 9350
    },
    {
      "epoch": 0.7867859446423063,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000242293688627889,
      "loss": 1.7935500144958496,
      "step": 9352
    },
    {
      "epoch": 0.7869542051095095,
      "grad_norm": 0.140625,
      "learning_rate": 0.00024208525984504748,
      "loss": 1.8150556087493896,
      "step": 9354
    },
    {
      "epoch": 0.7871224655767127,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00024187654534030977,
      "loss": 1.8558709621429443,
      "step": 9356
    },
    {
      "epoch": 0.787290726043916,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00024166754576127465,
      "loss": 1.7757902145385742,
      "step": 9358
    },
    {
      "epoch": 0.7874589865111192,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00024145826175642549,
      "loss": 1.805711030960083,
      "step": 9360
    },
    {
      "epoch": 0.7876272469783224,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00024124869397512807,
      "loss": 1.8268401622772217,
      "step": 9362
    },
    {
      "epoch": 0.7877955074455256,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00024103884306762876,
      "loss": 1.7717701196670532,
      "step": 9364
    },
    {
      "epoch": 0.7879637679127289,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.00024082870968505248,
      "loss": 1.785775899887085,
      "step": 9366
    },
    {
      "epoch": 0.7881320283799321,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00024061829447940037,
      "loss": 1.8519684076309204,
      "step": 9368
    },
    {
      "epoch": 0.7883002888471353,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002404075981035483,
      "loss": 1.7983005046844482,
      "step": 9370
    },
    {
      "epoch": 0.7884685493143386,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002401966212112443,
      "loss": 1.802543044090271,
      "step": 9372
    },
    {
      "epoch": 0.7886368097815418,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00023998536445710706,
      "loss": 1.8511040210723877,
      "step": 9374
    },
    {
      "epoch": 0.788805070248745,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00023977382849662329,
      "loss": 1.7912533283233643,
      "step": 9376
    },
    {
      "epoch": 0.7889733307159483,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002395620139861463,
      "loss": 1.8514974117279053,
      "step": 9378
    },
    {
      "epoch": 0.7891415911831515,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00023934992158289355,
      "loss": 1.8291139602661133,
      "step": 9380
    },
    {
      "epoch": 0.7893098516503547,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00023913755194494482,
      "loss": 1.8531851768493652,
      "step": 9382
    },
    {
      "epoch": 0.789478112117558,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00023892490573124,
      "loss": 1.8575819730758667,
      "step": 9384
    },
    {
      "epoch": 0.7896463725847612,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00023871198360157723,
      "loss": 1.8253458738327026,
      "step": 9386
    },
    {
      "epoch": 0.7898146330519644,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002384987862166107,
      "loss": 1.8466529846191406,
      "step": 9388
    },
    {
      "epoch": 0.7899828935191676,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00023828531423784866,
      "loss": 1.8443231582641602,
      "step": 9390
    },
    {
      "epoch": 0.7901511539863709,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00023807156832765142,
      "loss": 1.842411994934082,
      "step": 9392
    },
    {
      "epoch": 0.7903194144535741,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002378575491492292,
      "loss": 1.8079794645309448,
      "step": 9394
    },
    {
      "epoch": 0.7904876749207773,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00023764325736664013,
      "loss": 1.8280715942382812,
      "step": 9396
    },
    {
      "epoch": 0.7906559353879806,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002374286936447882,
      "loss": 1.814848780632019,
      "step": 9398
    },
    {
      "epoch": 0.7908241958551838,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00023721385864942116,
      "loss": 1.8342742919921875,
      "step": 9400
    },
    {
      "epoch": 0.790992456322387,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002369987530471284,
      "loss": 1.7963597774505615,
      "step": 9402
    },
    {
      "epoch": 0.7911607167895903,
      "grad_norm": 0.140625,
      "learning_rate": 0.000236783377505339,
      "loss": 1.854562759399414,
      "step": 9404
    },
    {
      "epoch": 0.7913289772567935,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002365677326923197,
      "loss": 1.8477497100830078,
      "step": 9406
    },
    {
      "epoch": 0.7914972377239967,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00023635181927717257,
      "loss": 1.802664875984192,
      "step": 9408
    },
    {
      "epoch": 0.7916654981912,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00023613563792983315,
      "loss": 1.806227207183838,
      "step": 9410
    },
    {
      "epoch": 0.7918337586584032,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002359191893210684,
      "loss": 1.8364441394805908,
      "step": 9412
    },
    {
      "epoch": 0.7920020191256064,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00023570247412247445,
      "loss": 1.8811116218566895,
      "step": 9414
    },
    {
      "epoch": 0.7921702795928097,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00023548549300647466,
      "loss": 1.8071205615997314,
      "step": 9416
    },
    {
      "epoch": 0.7923385400600129,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00023526824664631745,
      "loss": 1.8361084461212158,
      "step": 9418
    },
    {
      "epoch": 0.7925068005272161,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002350507357160742,
      "loss": 1.8491911888122559,
      "step": 9420
    },
    {
      "epoch": 0.7926750609944193,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002348329608906373,
      "loss": 1.8060637712478638,
      "step": 9422
    },
    {
      "epoch": 0.7928433214616226,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00023461492284571783,
      "loss": 1.826298713684082,
      "step": 9424
    },
    {
      "epoch": 0.7930115819288258,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002343966222578437,
      "loss": 1.8418796062469482,
      "step": 9426
    },
    {
      "epoch": 0.793179842396029,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00023417805980435736,
      "loss": 1.8417270183563232,
      "step": 9428
    },
    {
      "epoch": 0.7933481028632323,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00023395923616341378,
      "loss": 1.870889663696289,
      "step": 9430
    },
    {
      "epoch": 0.7935163633304355,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00023374015201397839,
      "loss": 1.8649967908859253,
      "step": 9432
    },
    {
      "epoch": 0.7936846237976387,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00023352080803582494,
      "loss": 1.810030460357666,
      "step": 9434
    },
    {
      "epoch": 0.793852884264842,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002333012049095333,
      "loss": 1.8077688217163086,
      "step": 9436
    },
    {
      "epoch": 0.7940211447320452,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00023308134331648738,
      "loss": 1.8295856714248657,
      "step": 9438
    },
    {
      "epoch": 0.7941894051992484,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00023286122393887323,
      "loss": 1.7921639680862427,
      "step": 9440
    },
    {
      "epoch": 0.7943576656664517,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00023264084745967665,
      "loss": 1.8427931070327759,
      "step": 9442
    },
    {
      "epoch": 0.7945259261336549,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00023242021456268118,
      "loss": 1.8011810779571533,
      "step": 9444
    },
    {
      "epoch": 0.7946941866008581,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00023219932593246598,
      "loss": 1.7359905242919922,
      "step": 9446
    },
    {
      "epoch": 0.7948624470680613,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00023197818225440374,
      "loss": 1.814941167831421,
      "step": 9448
    },
    {
      "epoch": 0.7950307075352646,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.00023175678421465836,
      "loss": 1.7948070764541626,
      "step": 9450
    },
    {
      "epoch": 0.7951989680024678,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00023153513250018316,
      "loss": 1.8530611991882324,
      "step": 9452
    },
    {
      "epoch": 0.795367228469671,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00023131322779871845,
      "loss": 1.8077865839004517,
      "step": 9454
    },
    {
      "epoch": 0.7955354889368743,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00023109107079878955,
      "loss": 1.8585121631622314,
      "step": 9456
    },
    {
      "epoch": 0.7957037494040775,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00023086866218970467,
      "loss": 1.8424898386001587,
      "step": 9458
    },
    {
      "epoch": 0.7958720098712807,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002306460026615525,
      "loss": 1.8407001495361328,
      "step": 9460
    },
    {
      "epoch": 0.796040270338484,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00023042309290520057,
      "loss": 1.8642244338989258,
      "step": 9462
    },
    {
      "epoch": 0.7962085308056872,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002301999336122926,
      "loss": 1.8041281700134277,
      "step": 9464
    },
    {
      "epoch": 0.7963767912728904,
      "grad_norm": 0.140625,
      "learning_rate": 0.00022997652547524658,
      "loss": 1.8479968309402466,
      "step": 9466
    },
    {
      "epoch": 0.7965450517400937,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00022975286918725278,
      "loss": 1.8055171966552734,
      "step": 9468
    },
    {
      "epoch": 0.7967133122072969,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00022952896544227132,
      "loss": 1.8278987407684326,
      "step": 9470
    },
    {
      "epoch": 0.7968815726745001,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00022930481493503013,
      "loss": 1.8017473220825195,
      "step": 9472
    },
    {
      "epoch": 0.7970498331417034,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002290804183610227,
      "loss": 1.8367862701416016,
      "step": 9474
    },
    {
      "epoch": 0.7972180936089066,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00022885577641650623,
      "loss": 1.8114068508148193,
      "step": 9476
    },
    {
      "epoch": 0.7973863540761098,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00022863088979849913,
      "loss": 1.8635579347610474,
      "step": 9478
    },
    {
      "epoch": 0.797554614543313,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00022840575920477895,
      "loss": 1.8039169311523438,
      "step": 9480
    },
    {
      "epoch": 0.7977228750105163,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002281803853338803,
      "loss": 1.8687736988067627,
      "step": 9482
    },
    {
      "epoch": 0.7978911354777195,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00022795476888509269,
      "loss": 1.845482587814331,
      "step": 9484
    },
    {
      "epoch": 0.7980593959449227,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00022772891055845807,
      "loss": 1.8163573741912842,
      "step": 9486
    },
    {
      "epoch": 0.798227656412126,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002275028110547692,
      "loss": 1.787839412689209,
      "step": 9488
    },
    {
      "epoch": 0.7983959168793292,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00022727647107556693,
      "loss": 1.8453009128570557,
      "step": 9490
    },
    {
      "epoch": 0.7985641773465324,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00022704989132313832,
      "loss": 1.831998348236084,
      "step": 9492
    },
    {
      "epoch": 0.7987324378137357,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00022682307250051454,
      "loss": 1.8562735319137573,
      "step": 9494
    },
    {
      "epoch": 0.7989006982809389,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00022659601531146832,
      "loss": 1.838932991027832,
      "step": 9496
    },
    {
      "epoch": 0.7990689587481421,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00022636872046051206,
      "loss": 1.8605695962905884,
      "step": 9498
    },
    {
      "epoch": 0.7992372192153454,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002261411886528957,
      "loss": 1.8359626531600952,
      "step": 9500
    },
    {
      "epoch": 0.7994054796825486,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00022591342059460431,
      "loss": 1.8459100723266602,
      "step": 9502
    },
    {
      "epoch": 0.7995737401497518,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00022568541699235595,
      "loss": 1.8001916408538818,
      "step": 9504
    },
    {
      "epoch": 0.799742000616955,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00022545717855359962,
      "loss": 1.832448959350586,
      "step": 9506
    },
    {
      "epoch": 0.7999102610841583,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000225228705986513,
      "loss": 1.8383009433746338,
      "step": 9508
    },
    {
      "epoch": 0.8000785215513615,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000225,
      "loss": 1.8481522798538208,
      "step": 9510
    },
    {
      "epoch": 0.8002467820185647,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00022477106130368903,
      "loss": 1.7874271869659424,
      "step": 9512
    },
    {
      "epoch": 0.8002467820185647,
      "eval_loss": 1.7952238321304321,
      "eval_runtime": 535.4687,
      "eval_samples_per_second": 103.952,
      "eval_steps_per_second": 12.994,
      "step": 9512
    },
    {
      "epoch": 0.800415042485768,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00022454189060793048,
      "loss": 1.8688840866088867,
      "step": 9514
    },
    {
      "epoch": 0.8005833029529712,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00022431248862379453,
      "loss": 1.8349323272705078,
      "step": 9516
    },
    {
      "epoch": 0.8007515634201744,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00022408285606306905,
      "loss": 1.8606958389282227,
      "step": 9518
    },
    {
      "epoch": 0.8009198238873777,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00022385299363825734,
      "loss": 1.8636901378631592,
      "step": 9520
    },
    {
      "epoch": 0.8010880843545809,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002236229020625759,
      "loss": 1.854252815246582,
      "step": 9522
    },
    {
      "epoch": 0.8012563448217841,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00022339258204995228,
      "loss": 1.821629524230957,
      "step": 9524
    },
    {
      "epoch": 0.8014246052889874,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00022316203431502272,
      "loss": 1.8415162563323975,
      "step": 9526
    },
    {
      "epoch": 0.8015928657561906,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00022293125957313016,
      "loss": 1.8453059196472168,
      "step": 9528
    },
    {
      "epoch": 0.8017611262233938,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00022270025854032187,
      "loss": 1.7915256023406982,
      "step": 9530
    },
    {
      "epoch": 0.801929386690597,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00022246903193334723,
      "loss": 1.8256394863128662,
      "step": 9532
    },
    {
      "epoch": 0.8020976471578003,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00022223758046965552,
      "loss": 1.8617889881134033,
      "step": 9534
    },
    {
      "epoch": 0.8022659076250035,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00022200590486739366,
      "loss": 1.833371639251709,
      "step": 9536
    },
    {
      "epoch": 0.8024341680922067,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00022177400584540415,
      "loss": 1.804419994354248,
      "step": 9538
    },
    {
      "epoch": 0.80260242855941,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00022154188412322267,
      "loss": 1.8172340393066406,
      "step": 9540
    },
    {
      "epoch": 0.8027706890266132,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00022130954042107583,
      "loss": 1.8157756328582764,
      "step": 9542
    },
    {
      "epoch": 0.8029389494938164,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00022107697545987906,
      "loss": 1.8611489534378052,
      "step": 9544
    },
    {
      "epoch": 0.8031072099610197,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00022084418996123423,
      "loss": 1.8388047218322754,
      "step": 9546
    },
    {
      "epoch": 0.8032754704282229,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00022061118464742772,
      "loss": 1.779839277267456,
      "step": 9548
    },
    {
      "epoch": 0.8034437308954261,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002203779602414276,
      "loss": 1.833999514579773,
      "step": 9550
    },
    {
      "epoch": 0.8036119913626294,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00022014451746688206,
      "loss": 1.831554651260376,
      "step": 9552
    },
    {
      "epoch": 0.8037802518298326,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.00021991085704811664,
      "loss": 1.827341079711914,
      "step": 9554
    },
    {
      "epoch": 0.8039485122970358,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002196769797101323,
      "loss": 1.778041124343872,
      "step": 9556
    },
    {
      "epoch": 0.8041167727642391,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000219442886178603,
      "loss": 1.855804204940796,
      "step": 9558
    },
    {
      "epoch": 0.8042850332314423,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00021920857717987356,
      "loss": 1.8330737352371216,
      "step": 9560
    },
    {
      "epoch": 0.8044532936986455,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002189740534409572,
      "loss": 1.8235056400299072,
      "step": 9562
    },
    {
      "epoch": 0.8046215541658487,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00021873931568953364,
      "loss": 1.832338809967041,
      "step": 9564
    },
    {
      "epoch": 0.804789814633052,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00021850436465394653,
      "loss": 1.8374745845794678,
      "step": 9566
    },
    {
      "epoch": 0.8049580751002552,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002182692010632013,
      "loss": 1.8111106157302856,
      "step": 9568
    },
    {
      "epoch": 0.8051263355674584,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002180338256469629,
      "loss": 1.870656967163086,
      "step": 9570
    },
    {
      "epoch": 0.8052945960346617,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00021779823913555357,
      "loss": 1.8382186889648438,
      "step": 9572
    },
    {
      "epoch": 0.8054628565018649,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00021756244225995045,
      "loss": 1.8460191488265991,
      "step": 9574
    },
    {
      "epoch": 0.8056311169690681,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00021732643575178345,
      "loss": 1.8570075035095215,
      "step": 9576
    },
    {
      "epoch": 0.8057993774362714,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00021709022034333296,
      "loss": 1.7946466207504272,
      "step": 9578
    },
    {
      "epoch": 0.8059676379034746,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00021685379676752753,
      "loss": 1.8569450378417969,
      "step": 9580
    },
    {
      "epoch": 0.8061358983706778,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00021661716575794156,
      "loss": 1.8245850801467896,
      "step": 9582
    },
    {
      "epoch": 0.8063041588378811,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00021638032804879314,
      "loss": 1.8434021472930908,
      "step": 9584
    },
    {
      "epoch": 0.8064724193050843,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00021614328437494164,
      "loss": 1.82460355758667,
      "step": 9586
    },
    {
      "epoch": 0.8066406797722875,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00021590603547188552,
      "loss": 1.8108410835266113,
      "step": 9588
    },
    {
      "epoch": 0.8068089402394907,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00021566858207576012,
      "loss": 1.861106038093567,
      "step": 9590
    },
    {
      "epoch": 0.806977200706694,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002154309249233351,
      "loss": 1.7847239971160889,
      "step": 9592
    },
    {
      "epoch": 0.8071454611738972,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00021519306475201248,
      "loss": 1.8275220394134521,
      "step": 9594
    },
    {
      "epoch": 0.8073137216411004,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00021495500229982414,
      "loss": 1.794785976409912,
      "step": 9596
    },
    {
      "epoch": 0.8074819821083037,
      "grad_norm": 0.140625,
      "learning_rate": 0.00021471673830542967,
      "loss": 1.769310474395752,
      "step": 9598
    },
    {
      "epoch": 0.8076502425755069,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002144782735081139,
      "loss": 1.8450950384140015,
      "step": 9600
    },
    {
      "epoch": 0.8078185030427101,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002142396086477848,
      "loss": 1.810652256011963,
      "step": 9602
    },
    {
      "epoch": 0.8079867635099134,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000214000744464971,
      "loss": 1.7861034870147705,
      "step": 9604
    },
    {
      "epoch": 0.8081550239771166,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00021376168170081966,
      "loss": 1.8463783264160156,
      "step": 9606
    },
    {
      "epoch": 0.8083232844443198,
      "grad_norm": 0.140625,
      "learning_rate": 0.00021352242109709408,
      "loss": 1.8028678894042969,
      "step": 9608
    },
    {
      "epoch": 0.8084915449115231,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00021328296339617145,
      "loss": 1.832820177078247,
      "step": 9610
    },
    {
      "epoch": 0.8086598053787263,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00021304330934104047,
      "loss": 1.801715612411499,
      "step": 9612
    },
    {
      "epoch": 0.8088280658459295,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00021280345967529903,
      "loss": 1.8346668481826782,
      "step": 9614
    },
    {
      "epoch": 0.8089963263131327,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00021256341514315212,
      "loss": 1.785402774810791,
      "step": 9616
    },
    {
      "epoch": 0.809164586780336,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002123231764894092,
      "loss": 1.812907338142395,
      "step": 9618
    },
    {
      "epoch": 0.8093328472475392,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00021208274445948217,
      "loss": 1.8283079862594604,
      "step": 9620
    },
    {
      "epoch": 0.8095011077147424,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00021184211979938288,
      "loss": 1.8164337873458862,
      "step": 9622
    },
    {
      "epoch": 0.8096693681819457,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00021160130325572084,
      "loss": 1.8328700065612793,
      "step": 9624
    },
    {
      "epoch": 0.8098376286491489,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000211360295575701,
      "loss": 1.7783238887786865,
      "step": 9626
    },
    {
      "epoch": 0.8100058891163521,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00021111909750712128,
      "loss": 1.8571009635925293,
      "step": 9628
    },
    {
      "epoch": 0.8101741495835554,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002108777097983705,
      "loss": 1.784805178642273,
      "step": 9630
    },
    {
      "epoch": 0.8103424100507586,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002106361331984257,
      "loss": 1.8270992040634155,
      "step": 9632
    },
    {
      "epoch": 0.8105106705179618,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002103943684568502,
      "loss": 1.83774733543396,
      "step": 9634
    },
    {
      "epoch": 0.8106789309851651,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00021015241632379086,
      "loss": 1.7745342254638672,
      "step": 9636
    },
    {
      "epoch": 0.8108471914523683,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00020991027754997625,
      "loss": 1.8441569805145264,
      "step": 9638
    },
    {
      "epoch": 0.8110154519195715,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00020966795288671379,
      "loss": 1.8385577201843262,
      "step": 9640
    },
    {
      "epoch": 0.8111837123867748,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00020942544308588784,
      "loss": 1.7957602739334106,
      "step": 9642
    },
    {
      "epoch": 0.811351972853978,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002091827488999572,
      "loss": 1.8043549060821533,
      "step": 9644
    },
    {
      "epoch": 0.8115202333211812,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00020893987108195264,
      "loss": 1.8298993110656738,
      "step": 9646
    },
    {
      "epoch": 0.8116884937883844,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00020869681038547487,
      "loss": 1.8189992904663086,
      "step": 9648
    },
    {
      "epoch": 0.8118567542555877,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00020845356756469196,
      "loss": 1.8188159465789795,
      "step": 9650
    },
    {
      "epoch": 0.8120250147227909,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00020821014337433705,
      "loss": 1.8385086059570312,
      "step": 9652
    },
    {
      "epoch": 0.8121932751899941,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00020796653856970605,
      "loss": 1.8022279739379883,
      "step": 9654
    },
    {
      "epoch": 0.8123615356571974,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00020772275390665534,
      "loss": 1.8288723230361938,
      "step": 9656
    },
    {
      "epoch": 0.8125297961244006,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00020747879014159932,
      "loss": 1.8405683040618896,
      "step": 9658
    },
    {
      "epoch": 0.8126980565916038,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00020723464803150805,
      "loss": 1.8042080402374268,
      "step": 9660
    },
    {
      "epoch": 0.8128663170588071,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00020699032833390504,
      "loss": 1.8384959697723389,
      "step": 9662
    },
    {
      "epoch": 0.8130345775260103,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00020674583180686477,
      "loss": 1.8248705863952637,
      "step": 9664
    },
    {
      "epoch": 0.8132028379932135,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00020650115920901047,
      "loss": 1.8130180835723877,
      "step": 9666
    },
    {
      "epoch": 0.8133710984604168,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00020625631129951154,
      "loss": 1.7949481010437012,
      "step": 9668
    },
    {
      "epoch": 0.81353935892762,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00020601128883808147,
      "loss": 1.8503198623657227,
      "step": 9670
    },
    {
      "epoch": 0.8137076193948232,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00020576609258497522,
      "loss": 1.8416669368743896,
      "step": 9672
    },
    {
      "epoch": 0.8138758798620264,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00020552072330098716,
      "loss": 1.7716346979141235,
      "step": 9674
    },
    {
      "epoch": 0.8140441403292297,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00020527518174744838,
      "loss": 1.8935742378234863,
      "step": 9676
    },
    {
      "epoch": 0.8142124007964329,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00020502946868622453,
      "loss": 1.7939109802246094,
      "step": 9678
    },
    {
      "epoch": 0.8143806612636361,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002047835848797135,
      "loss": 1.8466964960098267,
      "step": 9680
    },
    {
      "epoch": 0.8145489217308394,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00020453753109084285,
      "loss": 1.7988097667694092,
      "step": 9682
    },
    {
      "epoch": 0.8147171821980426,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00020429130808306766,
      "loss": 1.8085572719573975,
      "step": 9684
    },
    {
      "epoch": 0.8148854426652458,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00020404491662036803,
      "loss": 1.8182331323623657,
      "step": 9686
    },
    {
      "epoch": 0.8150537031324491,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00020379835746724663,
      "loss": 1.8667889833450317,
      "step": 9688
    },
    {
      "epoch": 0.8152219635996523,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00020355163138872662,
      "loss": 1.8544211387634277,
      "step": 9690
    },
    {
      "epoch": 0.8153902240668555,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00020330473915034896,
      "loss": 1.8314061164855957,
      "step": 9692
    },
    {
      "epoch": 0.8155584845340588,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00020305768151817027,
      "loss": 1.8049817085266113,
      "step": 9694
    },
    {
      "epoch": 0.815726745001262,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00020281045925876027,
      "loss": 1.8466815948486328,
      "step": 9696
    },
    {
      "epoch": 0.8158950054684652,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00020256307313919956,
      "loss": 1.8569203615188599,
      "step": 9698
    },
    {
      "epoch": 0.8160632659356685,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002023155239270771,
      "loss": 1.8233028650283813,
      "step": 9700
    },
    {
      "epoch": 0.8162315264028717,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00020206781239048792,
      "loss": 1.7950053215026855,
      "step": 9702
    },
    {
      "epoch": 0.8163997868700749,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00020181993929803069,
      "loss": 1.795240879058838,
      "step": 9704
    },
    {
      "epoch": 0.8165680473372781,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00020157190541880536,
      "loss": 1.800694227218628,
      "step": 9706
    },
    {
      "epoch": 0.8167363078044814,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00020132371152241084,
      "loss": 1.844083309173584,
      "step": 9708
    },
    {
      "epoch": 0.8169045682716846,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002010753583789425,
      "loss": 1.800480842590332,
      "step": 9710
    },
    {
      "epoch": 0.8170728287388878,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002008268467589897,
      "loss": 1.7907887697219849,
      "step": 9712
    },
    {
      "epoch": 0.8172410892060911,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002005781774336337,
      "loss": 1.8152122497558594,
      "step": 9714
    },
    {
      "epoch": 0.8174093496732943,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00020032935117444503,
      "loss": 1.8605833053588867,
      "step": 9716
    },
    {
      "epoch": 0.8175776101404975,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002000803687534811,
      "loss": 1.813916802406311,
      "step": 9718
    },
    {
      "epoch": 0.8177458706077008,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00019983123094328387,
      "loss": 1.833884596824646,
      "step": 9720
    },
    {
      "epoch": 0.817914131074904,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00019958193851687747,
      "loss": 1.808490514755249,
      "step": 9722
    },
    {
      "epoch": 0.8180823915421072,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00019933249224776589,
      "loss": 1.8299058675765991,
      "step": 9724
    },
    {
      "epoch": 0.8182506520093105,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0001990828929099302,
      "loss": 1.8044819831848145,
      "step": 9726
    },
    {
      "epoch": 0.8184189124765137,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00019883314127782658,
      "loss": 1.8158924579620361,
      "step": 9728
    },
    {
      "epoch": 0.8185871729437169,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001985832381263838,
      "loss": 1.8293439149856567,
      "step": 9730
    },
    {
      "epoch": 0.81875543341092,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00019833318423100057,
      "loss": 1.8348362445831299,
      "step": 9732
    },
    {
      "epoch": 0.8189236938781234,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0001980829803675435,
      "loss": 1.7998802661895752,
      "step": 9734
    },
    {
      "epoch": 0.8190919543453266,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00019783262731234454,
      "loss": 1.8606064319610596,
      "step": 9736
    },
    {
      "epoch": 0.8192602148125298,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001975821258421983,
      "loss": 1.8017258644104004,
      "step": 9738
    },
    {
      "epoch": 0.8194284752797331,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00019733147673436024,
      "loss": 1.755021572113037,
      "step": 9740
    },
    {
      "epoch": 0.8195967357469363,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0001970806807665436,
      "loss": 1.7907869815826416,
      "step": 9742
    },
    {
      "epoch": 0.8197649962141395,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00019682973871691756,
      "loss": 1.9137850999832153,
      "step": 9744
    },
    {
      "epoch": 0.8199332566813428,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00019657865136410432,
      "loss": 1.7790266275405884,
      "step": 9746
    },
    {
      "epoch": 0.820101517148546,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0001963274194871771,
      "loss": 1.7774637937545776,
      "step": 9748
    },
    {
      "epoch": 0.8202697776157492,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00019607604386565752,
      "loss": 1.8057196140289307,
      "step": 9750
    },
    {
      "epoch": 0.8204380380829525,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00019582452527951315,
      "loss": 1.8161063194274902,
      "step": 9752
    },
    {
      "epoch": 0.8206062985501557,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00019557286450915513,
      "loss": 1.8285460472106934,
      "step": 9754
    },
    {
      "epoch": 0.8207745590173589,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00019532106233543587,
      "loss": 1.829406976699829,
      "step": 9756
    },
    {
      "epoch": 0.820942819484562,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00019506911953964652,
      "loss": 1.8197603225708008,
      "step": 9758
    },
    {
      "epoch": 0.8211110799517654,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00019481703690351444,
      "loss": 1.8534083366394043,
      "step": 9760
    },
    {
      "epoch": 0.8212793404189686,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00019456481520920097,
      "loss": 1.8247323036193848,
      "step": 9762
    },
    {
      "epoch": 0.8214476008861717,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00019431245523929893,
      "loss": 1.8237218856811523,
      "step": 9764
    },
    {
      "epoch": 0.821615861353375,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00019405995777683013,
      "loss": 1.8219997882843018,
      "step": 9766
    },
    {
      "epoch": 0.8217841218205783,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000193807323605243,
      "loss": 1.7673691511154175,
      "step": 9768
    },
    {
      "epoch": 0.8219523822877814,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00019355455350841022,
      "loss": 1.8303041458129883,
      "step": 9770
    },
    {
      "epoch": 0.8221206427549848,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0001933016482706261,
      "loss": 1.805048942565918,
      "step": 9772
    },
    {
      "epoch": 0.822288903222188,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001930486086766044,
      "loss": 1.8213343620300293,
      "step": 9774
    },
    {
      "epoch": 0.8224571636893911,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00019279543551147565,
      "loss": 1.8180993795394897,
      "step": 9776
    },
    {
      "epoch": 0.8226254241565945,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00019254212956078487,
      "loss": 1.8169360160827637,
      "step": 9778
    },
    {
      "epoch": 0.8227936846237976,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00019228869161048905,
      "loss": 1.8736075162887573,
      "step": 9780
    },
    {
      "epoch": 0.8229619450910008,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00019203512244695485,
      "loss": 1.7944486141204834,
      "step": 9782
    },
    {
      "epoch": 0.8231302055582042,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00019178142285695594,
      "loss": 1.8395283222198486,
      "step": 9784
    },
    {
      "epoch": 0.8232984660254073,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00019152759362767074,
      "loss": 1.81486177444458,
      "step": 9786
    },
    {
      "epoch": 0.8234667264926105,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00019127363554667987,
      "loss": 1.8192803859710693,
      "step": 9788
    },
    {
      "epoch": 0.8236349869598137,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00019101954940196385,
      "loss": 1.8480277061462402,
      "step": 9790
    },
    {
      "epoch": 0.823803247427017,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00019076533598190042,
      "loss": 1.7847678661346436,
      "step": 9792
    },
    {
      "epoch": 0.8239715078942202,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0001905109960752623,
      "loss": 1.8228245973587036,
      "step": 9794
    },
    {
      "epoch": 0.8241397683614234,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00019025653047121467,
      "loss": 1.7634872198104858,
      "step": 9796
    },
    {
      "epoch": 0.8243080288286267,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00019000193995931272,
      "loss": 1.8313343524932861,
      "step": 9798
    },
    {
      "epoch": 0.82447628929583,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00018974722532949927,
      "loss": 1.80926513671875,
      "step": 9800
    },
    {
      "epoch": 0.8246445497630331,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0001894923873721021,
      "loss": 1.785710334777832,
      "step": 9802
    },
    {
      "epoch": 0.8248128102302364,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00018923742687783175,
      "loss": 1.819293737411499,
      "step": 9804
    },
    {
      "epoch": 0.8249810706974396,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00018898234463777898,
      "loss": 1.784306287765503,
      "step": 9806
    },
    {
      "epoch": 0.8251493311646428,
      "grad_norm": 0.140625,
      "learning_rate": 0.00018872714144341226,
      "loss": 1.835662603378296,
      "step": 9808
    },
    {
      "epoch": 0.8253175916318461,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00018847181808657543,
      "loss": 1.8453035354614258,
      "step": 9810
    },
    {
      "epoch": 0.8254858520990493,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00018821637535948511,
      "loss": 1.8444191217422485,
      "step": 9812
    },
    {
      "epoch": 0.8256541125662525,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001879608140547282,
      "loss": 1.7698026895523071,
      "step": 9814
    },
    {
      "epoch": 0.8258223730334557,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001877051349652597,
      "loss": 1.8504194021224976,
      "step": 9816
    },
    {
      "epoch": 0.825990633500659,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0001874493388844,
      "loss": 1.8140826225280762,
      "step": 9818
    },
    {
      "epoch": 0.8261588939678622,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00018719342660583247,
      "loss": 1.8262683153152466,
      "step": 9820
    },
    {
      "epoch": 0.8263271544350654,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00018693739892360105,
      "loss": 1.8286852836608887,
      "step": 9822
    },
    {
      "epoch": 0.8264954149022687,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00018668125663210777,
      "loss": 1.828436017036438,
      "step": 9824
    },
    {
      "epoch": 0.8266636753694719,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00018642500052611023,
      "loss": 1.830739974975586,
      "step": 9826
    },
    {
      "epoch": 0.8268319358366751,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00018616863140071913,
      "loss": 1.8160178661346436,
      "step": 9828
    },
    {
      "epoch": 0.8270001963038784,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00018591215005139597,
      "loss": 1.8435693979263306,
      "step": 9830
    },
    {
      "epoch": 0.8271684567710816,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00018565555727395038,
      "loss": 1.8098199367523193,
      "step": 9832
    },
    {
      "epoch": 0.8273367172382848,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00018539885386453767,
      "loss": 1.8426299095153809,
      "step": 9834
    },
    {
      "epoch": 0.8275049777054881,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0001851420406196566,
      "loss": 1.8765281438827515,
      "step": 9836
    },
    {
      "epoch": 0.8276732381726913,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00018488511833614655,
      "loss": 1.8453720808029175,
      "step": 9838
    },
    {
      "epoch": 0.8278414986398945,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00018462808781118527,
      "loss": 1.8176957368850708,
      "step": 9840
    },
    {
      "epoch": 0.8280097591070978,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00018437094984228642,
      "loss": 1.812211513519287,
      "step": 9842
    },
    {
      "epoch": 0.828178019574301,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00018411370522729696,
      "loss": 1.8342702388763428,
      "step": 9844
    },
    {
      "epoch": 0.8283462800415042,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00018385635476439485,
      "loss": 1.8568968772888184,
      "step": 9846
    },
    {
      "epoch": 0.8285145405087074,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00018359889925208636,
      "loss": 1.8624756336212158,
      "step": 9848
    },
    {
      "epoch": 0.8286828009759107,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001833413394892038,
      "loss": 1.8017340898513794,
      "step": 9850
    },
    {
      "epoch": 0.8288510614431139,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00018308367627490284,
      "loss": 1.8327910900115967,
      "step": 9852
    },
    {
      "epoch": 0.8290193219103171,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00018282591040866026,
      "loss": 1.8423579931259155,
      "step": 9854
    },
    {
      "epoch": 0.8291875823775204,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001825680426902713,
      "loss": 1.816881775856018,
      "step": 9856
    },
    {
      "epoch": 0.8293558428447236,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001823100739198472,
      "loss": 1.8186333179473877,
      "step": 9858
    },
    {
      "epoch": 0.8295241033119268,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00018205200489781274,
      "loss": 1.8017714023590088,
      "step": 9860
    },
    {
      "epoch": 0.8296923637791301,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0001817938364249039,
      "loss": 1.9030544757843018,
      "step": 9862
    },
    {
      "epoch": 0.8298606242463333,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00018153556930216498,
      "loss": 1.8460490703582764,
      "step": 9864
    },
    {
      "epoch": 0.8300288847135365,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001812772043309466,
      "loss": 1.8005867004394531,
      "step": 9866
    },
    {
      "epoch": 0.8301971451807398,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00018101874231290284,
      "loss": 1.828664779663086,
      "step": 9868
    },
    {
      "epoch": 0.830365405647943,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000180760184049989,
      "loss": 1.8408071994781494,
      "step": 9870
    },
    {
      "epoch": 0.8305336661151462,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00018050153034445895,
      "loss": 1.8198368549346924,
      "step": 9872
    },
    {
      "epoch": 0.8307019265823494,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0001802427819988627,
      "loss": 1.8337007761001587,
      "step": 9874
    },
    {
      "epoch": 0.8308701870495527,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00017998393981604396,
      "loss": 1.8642196655273438,
      "step": 9876
    },
    {
      "epoch": 0.8310384475167559,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00017972500459913747,
      "loss": 1.77899169921875,
      "step": 9878
    },
    {
      "epoch": 0.8312067079839591,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00017946597715156682,
      "loss": 1.7923262119293213,
      "step": 9880
    },
    {
      "epoch": 0.8313749684511624,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00017920685827704165,
      "loss": 1.8118852376937866,
      "step": 9882
    },
    {
      "epoch": 0.8315432289183656,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00017894764877955525,
      "loss": 1.8481364250183105,
      "step": 9884
    },
    {
      "epoch": 0.8317114893855688,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00017868834946338218,
      "loss": 1.749047875404358,
      "step": 9886
    },
    {
      "epoch": 0.8318797498527721,
      "grad_norm": 0.162109375,
      "learning_rate": 0.00017842896113307572,
      "loss": 1.8162493705749512,
      "step": 9888
    },
    {
      "epoch": 0.8320480103199753,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00017816948459346523,
      "loss": 1.8029325008392334,
      "step": 9890
    },
    {
      "epoch": 0.8322162707871785,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00017790992064965378,
      "loss": 1.8088722229003906,
      "step": 9892
    },
    {
      "epoch": 0.8323845312543818,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00017765027010701576,
      "loss": 1.7977365255355835,
      "step": 9894
    },
    {
      "epoch": 0.832552791721585,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001773905337711941,
      "loss": 1.8655521869659424,
      "step": 9896
    },
    {
      "epoch": 0.8327210521887882,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00017713071244809806,
      "loss": 1.817348599433899,
      "step": 9898
    },
    {
      "epoch": 0.8328893126559914,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001768708069439005,
      "loss": 1.8319666385650635,
      "step": 9900
    },
    {
      "epoch": 0.8330575731231947,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00017661081806503556,
      "loss": 1.8557934761047363,
      "step": 9902
    },
    {
      "epoch": 0.8332258335903979,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00017635074661819598,
      "loss": 1.8056167364120483,
      "step": 9904
    },
    {
      "epoch": 0.8333940940576011,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00017609059341033078,
      "loss": 1.8258802890777588,
      "step": 9906
    },
    {
      "epoch": 0.8335623545248044,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001758303592486426,
      "loss": 1.7815194129943848,
      "step": 9908
    },
    {
      "epoch": 0.8337306149920076,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00017557004494058535,
      "loss": 1.7952063083648682,
      "step": 9910
    },
    {
      "epoch": 0.8338988754592108,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00017530965129386157,
      "loss": 1.8050448894500732,
      "step": 9912
    },
    {
      "epoch": 0.8340671359264141,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00017504917911641987,
      "loss": 1.8575260639190674,
      "step": 9914
    },
    {
      "epoch": 0.8342353963936173,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00017478862921645273,
      "loss": 1.761277198791504,
      "step": 9916
    },
    {
      "epoch": 0.8344036568608205,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00017452800240239363,
      "loss": 1.8258414268493652,
      "step": 9918
    },
    {
      "epoch": 0.8345719173280238,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00017426729948291474,
      "loss": 1.8244233131408691,
      "step": 9920
    },
    {
      "epoch": 0.834740177795227,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00017400652126692433,
      "loss": 1.8323392868041992,
      "step": 9922
    },
    {
      "epoch": 0.8349084382624302,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00017374566856356445,
      "loss": 1.8412270545959473,
      "step": 9924
    },
    {
      "epoch": 0.8350766987296335,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00017348474218220811,
      "loss": 1.7919371128082275,
      "step": 9926
    },
    {
      "epoch": 0.8352449591968367,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00017322374293245697,
      "loss": 1.8148751258850098,
      "step": 9928
    },
    {
      "epoch": 0.8354132196640399,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0001729626716241388,
      "loss": 1.8414709568023682,
      "step": 9930
    },
    {
      "epoch": 0.8355814801312431,
      "grad_norm": 0.171875,
      "learning_rate": 0.00017270152906730498,
      "loss": 1.8183796405792236,
      "step": 9932
    },
    {
      "epoch": 0.8357497405984464,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0001724403160722279,
      "loss": 1.8250906467437744,
      "step": 9934
    },
    {
      "epoch": 0.8359180010656496,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00017217903344939858,
      "loss": 1.8320837020874023,
      "step": 9936
    },
    {
      "epoch": 0.8360862615328528,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00017191768200952408,
      "loss": 1.8438973426818848,
      "step": 9938
    },
    {
      "epoch": 0.8362545220000561,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00017165626256352478,
      "loss": 1.8457660675048828,
      "step": 9940
    },
    {
      "epoch": 0.8364227824672593,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00017139477592253237,
      "loss": 1.8461740016937256,
      "step": 9942
    },
    {
      "epoch": 0.8365910429344625,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00017113322289788685,
      "loss": 1.8022711277008057,
      "step": 9944
    },
    {
      "epoch": 0.8367593034016658,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00017087160430113426,
      "loss": 1.8272708654403687,
      "step": 9946
    },
    {
      "epoch": 0.836927563868869,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00017060992094402404,
      "loss": 1.8240365982055664,
      "step": 9948
    },
    {
      "epoch": 0.8370958243360722,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001703481736385067,
      "loss": 1.8515727519989014,
      "step": 9950
    },
    {
      "epoch": 0.8372640848032755,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00017008636319673098,
      "loss": 1.7729978561401367,
      "step": 9952
    },
    {
      "epoch": 0.8374323452704787,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00016982449043104157,
      "loss": 1.8099732398986816,
      "step": 9954
    },
    {
      "epoch": 0.8376006057376819,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001695625561539767,
      "loss": 1.8156120777130127,
      "step": 9956
    },
    {
      "epoch": 0.8377688662048851,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00016930056117826528,
      "loss": 1.8039886951446533,
      "step": 9958
    },
    {
      "epoch": 0.8379371266720884,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00016903850631682464,
      "loss": 1.8198652267456055,
      "step": 9960
    },
    {
      "epoch": 0.8381053871392916,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00016877639238275786,
      "loss": 1.83920156955719,
      "step": 9962
    },
    {
      "epoch": 0.8382736476064948,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00016851422018935143,
      "loss": 1.834550142288208,
      "step": 9964
    },
    {
      "epoch": 0.8384419080736981,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00016825199055007243,
      "loss": 1.8234561681747437,
      "step": 9966
    },
    {
      "epoch": 0.8386101685409013,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0001679897042785663,
      "loss": 1.8810362815856934,
      "step": 9968
    },
    {
      "epoch": 0.8387784290081045,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00016772736218865418,
      "loss": 1.7718452215194702,
      "step": 9970
    },
    {
      "epoch": 0.8389466894753078,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00016746496509433044,
      "loss": 1.8378665447235107,
      "step": 9972
    },
    {
      "epoch": 0.839114949942511,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00016720251380976007,
      "loss": 1.8229990005493164,
      "step": 9974
    },
    {
      "epoch": 0.8392832104097142,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00016694000914927614,
      "loss": 1.8581054210662842,
      "step": 9976
    },
    {
      "epoch": 0.8394514708769175,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00016667745192737747,
      "loss": 1.845445156097412,
      "step": 9978
    },
    {
      "epoch": 0.8396197313441207,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0001664148429587259,
      "loss": 1.8803410530090332,
      "step": 9980
    },
    {
      "epoch": 0.8397879918113239,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00016615218305814382,
      "loss": 1.7779786586761475,
      "step": 9982
    },
    {
      "epoch": 0.8399562522785271,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00016588947304061163,
      "loss": 1.8767311573028564,
      "step": 9984
    },
    {
      "epoch": 0.8401245127457304,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0001656267137212653,
      "loss": 1.7913391590118408,
      "step": 9986
    },
    {
      "epoch": 0.8402927732129336,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00016536390591539375,
      "loss": 1.8114224672317505,
      "step": 9988
    },
    {
      "epoch": 0.8404610336801368,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001651010504384363,
      "loss": 1.7569472789764404,
      "step": 9990
    },
    {
      "epoch": 0.8406292941473401,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001648381481059802,
      "loss": 1.798190712928772,
      "step": 9992
    },
    {
      "epoch": 0.8407975546145433,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00016457519973375815,
      "loss": 1.801969051361084,
      "step": 9994
    },
    {
      "epoch": 0.8409658150817465,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00016431220613764562,
      "loss": 1.8724465370178223,
      "step": 9996
    },
    {
      "epoch": 0.8411340755489498,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0001640491681336584,
      "loss": 1.8481249809265137,
      "step": 9998
    },
    {
      "epoch": 0.841302336016153,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00016378608653795017,
      "loss": 1.8089492321014404,
      "step": 10000
    },
    {
      "epoch": 0.8414705964833562,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0001635229621668098,
      "loss": 1.752370834350586,
      "step": 10002
    },
    {
      "epoch": 0.8416388569505595,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001632597958366588,
      "loss": 1.8168597221374512,
      "step": 10004
    },
    {
      "epoch": 0.8418071174177627,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00016299658836404902,
      "loss": 1.8208386898040771,
      "step": 10006
    },
    {
      "epoch": 0.8419753778849659,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00016273334056565985,
      "loss": 1.9121531248092651,
      "step": 10008
    },
    {
      "epoch": 0.8421436383521692,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00016247005325829584,
      "loss": 1.8542203903198242,
      "step": 10010
    },
    {
      "epoch": 0.8423118988193724,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00016220672725888422,
      "loss": 1.837280035018921,
      "step": 10012
    },
    {
      "epoch": 0.8424801592865756,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00016194336338447218,
      "loss": 1.8716154098510742,
      "step": 10014
    },
    {
      "epoch": 0.8426484197537788,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00016167996245222433,
      "loss": 1.796844720840454,
      "step": 10016
    },
    {
      "epoch": 0.8428166802209821,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001614165252794205,
      "loss": 1.8853718042373657,
      "step": 10018
    },
    {
      "epoch": 0.8429849406881853,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00016115305268345274,
      "loss": 1.8414785861968994,
      "step": 10020
    },
    {
      "epoch": 0.8431532011553885,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00016088954548182315,
      "loss": 1.8372554779052734,
      "step": 10022
    },
    {
      "epoch": 0.8433214616225918,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00016062600449214123,
      "loss": 1.8358871936798096,
      "step": 10024
    },
    {
      "epoch": 0.843489722089795,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00016036243053212115,
      "loss": 1.8119571208953857,
      "step": 10026
    },
    {
      "epoch": 0.8436579825569982,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00016009882441957952,
      "loss": 1.8187679052352905,
      "step": 10028
    },
    {
      "epoch": 0.8438262430242015,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001598351869724327,
      "loss": 1.7776987552642822,
      "step": 10030
    },
    {
      "epoch": 0.8439945034914047,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00015957151900869424,
      "loss": 1.8330271244049072,
      "step": 10032
    },
    {
      "epoch": 0.8441627639586079,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0001593078213464724,
      "loss": 1.840736746788025,
      "step": 10034
    },
    {
      "epoch": 0.8443310244258112,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001590440948039675,
      "loss": 1.8498456478118896,
      "step": 10036
    },
    {
      "epoch": 0.8444992848930144,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001587803401994697,
      "loss": 1.8155300617218018,
      "step": 10038
    },
    {
      "epoch": 0.8446675453602176,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00015851655835135586,
      "loss": 1.8591128587722778,
      "step": 10040
    },
    {
      "epoch": 0.8448358058274208,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001582527500780878,
      "loss": 1.852828860282898,
      "step": 10042
    },
    {
      "epoch": 0.8450040662946241,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00015798891619820898,
      "loss": 1.8460612297058105,
      "step": 10044
    },
    {
      "epoch": 0.8451723267618273,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00015772505753034252,
      "loss": 1.8460365533828735,
      "step": 10046
    },
    {
      "epoch": 0.8453405872290305,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00015746117489318838,
      "loss": 1.8287932872772217,
      "step": 10048
    },
    {
      "epoch": 0.8455088476962338,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00015719726910552083,
      "loss": 1.7934272289276123,
      "step": 10050
    },
    {
      "epoch": 0.845677108163437,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00015693334098618617,
      "loss": 1.8148860931396484,
      "step": 10052
    },
    {
      "epoch": 0.8458453686306402,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001566693913540997,
      "loss": 1.8202831745147705,
      "step": 10054
    },
    {
      "epoch": 0.8460136290978435,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0001564054210282437,
      "loss": 1.7939950227737427,
      "step": 10056
    },
    {
      "epoch": 0.8461818895650467,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00015614143082766458,
      "loss": 1.8990797996520996,
      "step": 10058
    },
    {
      "epoch": 0.8463501500322499,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00015587742157147048,
      "loss": 1.7853505611419678,
      "step": 10060
    },
    {
      "epoch": 0.8465184104994532,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00015561339407882852,
      "loss": 1.778873324394226,
      "step": 10062
    },
    {
      "epoch": 0.8466866709666564,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001553493491689626,
      "loss": 1.794269323348999,
      "step": 10064
    },
    {
      "epoch": 0.8468549314338596,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00015508528766115042,
      "loss": 1.7727335691452026,
      "step": 10066
    },
    {
      "epoch": 0.8470231919010629,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00015482121037472141,
      "loss": 1.7970373630523682,
      "step": 10068
    },
    {
      "epoch": 0.8471914523682661,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0001545571181290538,
      "loss": 1.8355185985565186,
      "step": 10070
    },
    {
      "epoch": 0.8473597128354693,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00015429301174357238,
      "loss": 1.81055748462677,
      "step": 10072
    },
    {
      "epoch": 0.8475279733026725,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00015402889203774563,
      "loss": 1.8412175178527832,
      "step": 10074
    },
    {
      "epoch": 0.8476962337698758,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00015376475983108355,
      "loss": 1.7780990600585938,
      "step": 10076
    },
    {
      "epoch": 0.847864494237079,
      "grad_norm": 0.140625,
      "learning_rate": 0.00015350061594313487,
      "loss": 1.8435980081558228,
      "step": 10078
    },
    {
      "epoch": 0.8480327547042822,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0001532364611934844,
      "loss": 1.8082568645477295,
      "step": 10080
    },
    {
      "epoch": 0.8482010151714855,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00015297229640175085,
      "loss": 1.8336408138275146,
      "step": 10082
    },
    {
      "epoch": 0.8483692756386887,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00015270812238758405,
      "loss": 1.8704860210418701,
      "step": 10084
    },
    {
      "epoch": 0.8485375361058919,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0001524439399706624,
      "loss": 1.7883481979370117,
      "step": 10086
    },
    {
      "epoch": 0.8487057965730952,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001521797499706904,
      "loss": 1.8596179485321045,
      "step": 10088
    },
    {
      "epoch": 0.8488740570402984,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00015191555320739607,
      "loss": 1.807480812072754,
      "step": 10090
    },
    {
      "epoch": 0.8490423175075016,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.00015165135050052838,
      "loss": 1.7944951057434082,
      "step": 10092
    },
    {
      "epoch": 0.8492105779747049,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0001513871426698548,
      "loss": 1.8460637331008911,
      "step": 10094
    },
    {
      "epoch": 0.8493788384419081,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0001511229305351587,
      "loss": 1.8596850633621216,
      "step": 10096
    },
    {
      "epoch": 0.8495470989091113,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00015085871491623675,
      "loss": 1.8312137126922607,
      "step": 10098
    },
    {
      "epoch": 0.8497153593763145,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00015059449663289653,
      "loss": 1.853998064994812,
      "step": 10100
    },
    {
      "epoch": 0.8498836198435178,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00015033027650495368,
      "loss": 1.8548657894134521,
      "step": 10102
    },
    {
      "epoch": 0.850051880310721,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00015006605535222982,
      "loss": 1.7862088680267334,
      "step": 10104
    },
    {
      "epoch": 0.8502201407779242,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00014980183399454958,
      "loss": 1.8200011253356934,
      "step": 10106
    },
    {
      "epoch": 0.8503884012451275,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001495376132517383,
      "loss": 1.784240484237671,
      "step": 10108
    },
    {
      "epoch": 0.8505566617123307,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00014927339394361938,
      "loss": 1.8166353702545166,
      "step": 10110
    },
    {
      "epoch": 0.8507249221795339,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00014900917689001175,
      "loss": 1.8510030508041382,
      "step": 10112
    },
    {
      "epoch": 0.8508931826467372,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00014874496291072742,
      "loss": 1.8627190589904785,
      "step": 10114
    },
    {
      "epoch": 0.8510614431139404,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001484807528255688,
      "loss": 1.808750867843628,
      "step": 10116
    },
    {
      "epoch": 0.8512297035811436,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00014821654745432615,
      "loss": 1.7914159297943115,
      "step": 10118
    },
    {
      "epoch": 0.8513979640483469,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001479523476167753,
      "loss": 1.8060412406921387,
      "step": 10120
    },
    {
      "epoch": 0.8515662245155501,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0001476881541326748,
      "loss": 1.770132303237915,
      "step": 10122
    },
    {
      "epoch": 0.8517344849827533,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00014742396782176334,
      "loss": 1.8001489639282227,
      "step": 10124
    },
    {
      "epoch": 0.8519027454499565,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00014715978950375768,
      "loss": 1.842379093170166,
      "step": 10126
    },
    {
      "epoch": 0.8520710059171598,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00014689561999834937,
      "loss": 1.8046324253082275,
      "step": 10128
    },
    {
      "epoch": 0.852239266384363,
      "grad_norm": 0.12109375,
      "learning_rate": 0.00014663146012520306,
      "loss": 1.761889100074768,
      "step": 10130
    },
    {
      "epoch": 0.8524075268515662,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00014636731070395316,
      "loss": 1.8133690357208252,
      "step": 10132
    },
    {
      "epoch": 0.8525757873187695,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00014610317255420185,
      "loss": 1.8176090717315674,
      "step": 10134
    },
    {
      "epoch": 0.8527440477859727,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00014583904649551627,
      "loss": 1.8389244079589844,
      "step": 10136
    },
    {
      "epoch": 0.8529123082531759,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00014557493334742594,
      "loss": 1.8636600971221924,
      "step": 10138
    },
    {
      "epoch": 0.8530805687203792,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00014531083392942058,
      "loss": 1.8451006412506104,
      "step": 10140
    },
    {
      "epoch": 0.8532488291875824,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00014504674906094697,
      "loss": 1.8284004926681519,
      "step": 10142
    },
    {
      "epoch": 0.8534170896547856,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00014478267956140713,
      "loss": 1.808011770248413,
      "step": 10144
    },
    {
      "epoch": 0.8535853501219889,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001445186262501551,
      "loss": 1.9138174057006836,
      "step": 10146
    },
    {
      "epoch": 0.8537536105891921,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001442545899464947,
      "loss": 1.8165696859359741,
      "step": 10148
    },
    {
      "epoch": 0.8539218710563953,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00014399057146967717,
      "loss": 1.7659368515014648,
      "step": 10150
    },
    {
      "epoch": 0.8540901315235986,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00014372657163889827,
      "loss": 1.8110694885253906,
      "step": 10152
    },
    {
      "epoch": 0.8542583919908018,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00014346259127329605,
      "loss": 1.830501675605774,
      "step": 10154
    },
    {
      "epoch": 0.854426652458005,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00014319863119194797,
      "loss": 1.7663686275482178,
      "step": 10156
    },
    {
      "epoch": 0.8545949129252082,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00014293469221386871,
      "loss": 1.7759218215942383,
      "step": 10158
    },
    {
      "epoch": 0.8547631733924115,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00014267077515800746,
      "loss": 1.8503425121307373,
      "step": 10160
    },
    {
      "epoch": 0.8549314338596147,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00014240688084324525,
      "loss": 1.8235937356948853,
      "step": 10162
    },
    {
      "epoch": 0.8550996943268179,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00014214301008839275,
      "loss": 1.7936694622039795,
      "step": 10164
    },
    {
      "epoch": 0.8552679547940212,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00014187916371218736,
      "loss": 1.8227139711380005,
      "step": 10166
    },
    {
      "epoch": 0.8554362152612244,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00014161534253329095,
      "loss": 1.8527570962905884,
      "step": 10168
    },
    {
      "epoch": 0.8556044757284276,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001413515473702871,
      "loss": 1.7769073247909546,
      "step": 10170
    },
    {
      "epoch": 0.8557727361956309,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00014108777904167882,
      "loss": 1.8608719110488892,
      "step": 10172
    },
    {
      "epoch": 0.8559409966628341,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00014082403836588572,
      "loss": 1.7808926105499268,
      "step": 10174
    },
    {
      "epoch": 0.8561092571300373,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00014056032616124165,
      "loss": 1.838006854057312,
      "step": 10176
    },
    {
      "epoch": 0.8562775175972406,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00014029664324599216,
      "loss": 1.7917182445526123,
      "step": 10178
    },
    {
      "epoch": 0.8564457780644438,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0001400329904382918,
      "loss": 1.7873127460479736,
      "step": 10180
    },
    {
      "epoch": 0.856614038531647,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00013976936855620192,
      "loss": 1.8329112529754639,
      "step": 10182
    },
    {
      "epoch": 0.8567822989988502,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00013950577841768776,
      "loss": 1.839892864227295,
      "step": 10184
    },
    {
      "epoch": 0.8569505594660535,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00013924222084061596,
      "loss": 1.7757720947265625,
      "step": 10186
    },
    {
      "epoch": 0.8571188199332567,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00013897869664275243,
      "loss": 1.7786030769348145,
      "step": 10188
    },
    {
      "epoch": 0.8572870804004599,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001387152066417592,
      "loss": 1.8101989030838013,
      "step": 10190
    },
    {
      "epoch": 0.8574553408676632,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00013845175165519248,
      "loss": 1.8224214315414429,
      "step": 10192
    },
    {
      "epoch": 0.8576236013348664,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00013818833250049957,
      "loss": 1.8526585102081299,
      "step": 10194
    },
    {
      "epoch": 0.8577918618020696,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001379249499950168,
      "loss": 1.8708820343017578,
      "step": 10196
    },
    {
      "epoch": 0.8579601222692729,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00013766160495596666,
      "loss": 1.8459336757659912,
      "step": 10198
    },
    {
      "epoch": 0.8581283827364761,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00013739829820045538,
      "loss": 1.853477954864502,
      "step": 10200
    },
    {
      "epoch": 0.8582966432036793,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00013713503054547054,
      "loss": 1.7406511306762695,
      "step": 10202
    },
    {
      "epoch": 0.8584649036708826,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00013687180280787822,
      "loss": 1.830970287322998,
      "step": 10204
    },
    {
      "epoch": 0.8586331641380858,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00013660861580442086,
      "loss": 1.8452842235565186,
      "step": 10206
    },
    {
      "epoch": 0.858801424605289,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001363454703517142,
      "loss": 1.8157684803009033,
      "step": 10208
    },
    {
      "epoch": 0.8589696850724923,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00013608236726624548,
      "loss": 1.8061063289642334,
      "step": 10210
    },
    {
      "epoch": 0.8591379455396955,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0001358193073643701,
      "loss": 1.853405475616455,
      "step": 10212
    },
    {
      "epoch": 0.8593062060068987,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00013555629146230953,
      "loss": 1.7801496982574463,
      "step": 10214
    },
    {
      "epoch": 0.8594744664741019,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000135293320376149,
      "loss": 1.8391424417495728,
      "step": 10216
    },
    {
      "epoch": 0.8596427269413052,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00013503039492183432,
      "loss": 1.8536521196365356,
      "step": 10218
    },
    {
      "epoch": 0.8598109874085084,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00013476751591517,
      "loss": 1.8114778995513916,
      "step": 10220
    },
    {
      "epoch": 0.8599792478757116,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001345046841718163,
      "loss": 1.8196406364440918,
      "step": 10222
    },
    {
      "epoch": 0.8601475083429149,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00013424190050728674,
      "loss": 1.7827916145324707,
      "step": 10224
    },
    {
      "epoch": 0.8603157688101181,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001339791657369459,
      "loss": 1.8373088836669922,
      "step": 10226
    },
    {
      "epoch": 0.8604840292773213,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001337164806760064,
      "loss": 1.8020442724227905,
      "step": 10228
    },
    {
      "epoch": 0.8606522897445246,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00013345384613952684,
      "loss": 1.7810389995574951,
      "step": 10230
    },
    {
      "epoch": 0.8608205502117278,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001331912629424089,
      "loss": 1.7865403890609741,
      "step": 10232
    },
    {
      "epoch": 0.860988810678931,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00013292873189939502,
      "loss": 1.831552267074585,
      "step": 10234
    },
    {
      "epoch": 0.8611570711461343,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00013266625382506582,
      "loss": 1.8454904556274414,
      "step": 10236
    },
    {
      "epoch": 0.8613253316133375,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00013240382953383755,
      "loss": 1.7713679075241089,
      "step": 10238
    },
    {
      "epoch": 0.8614935920805407,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00013214145983995963,
      "loss": 1.8112092018127441,
      "step": 10240
    },
    {
      "epoch": 0.8616618525477439,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00013187914555751198,
      "loss": 1.8285859823226929,
      "step": 10242
    },
    {
      "epoch": 0.8618301130149472,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0001316168875004027,
      "loss": 1.827867031097412,
      "step": 10244
    },
    {
      "epoch": 0.8619983734821504,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00013135468648236534,
      "loss": 1.7966053485870361,
      "step": 10246
    },
    {
      "epoch": 0.8621666339493536,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00013109254331695658,
      "loss": 1.8297573328018188,
      "step": 10248
    },
    {
      "epoch": 0.8623348944165569,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00013083045881755352,
      "loss": 1.77121102809906,
      "step": 10250
    },
    {
      "epoch": 0.8625031548837601,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001305684337973511,
      "loss": 1.843039870262146,
      "step": 10252
    },
    {
      "epoch": 0.8626714153509633,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00013030646906936002,
      "loss": 1.8376723527908325,
      "step": 10254
    },
    {
      "epoch": 0.8628396758181666,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00013004456544640362,
      "loss": 1.8227739334106445,
      "step": 10256
    },
    {
      "epoch": 0.8630079362853698,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00012978272374111586,
      "loss": 1.7825841903686523,
      "step": 10258
    },
    {
      "epoch": 0.863176196752573,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00012952094476593844,
      "loss": 1.8595049381256104,
      "step": 10260
    },
    {
      "epoch": 0.8633444572197763,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00012925922933311834,
      "loss": 1.8516708612442017,
      "step": 10262
    },
    {
      "epoch": 0.8635127176869795,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00012899757825470567,
      "loss": 1.8410699367523193,
      "step": 10264
    },
    {
      "epoch": 0.8636809781541827,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00012873599234255055,
      "loss": 1.822857141494751,
      "step": 10266
    },
    {
      "epoch": 0.8638492386213859,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00012847447240830117,
      "loss": 1.8319164514541626,
      "step": 10268
    },
    {
      "epoch": 0.8640174990885892,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00012821301926340076,
      "loss": 1.8083257675170898,
      "step": 10270
    },
    {
      "epoch": 0.8641857595557924,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00012795163371908557,
      "loss": 1.8507109880447388,
      "step": 10272
    },
    {
      "epoch": 0.8643540200229956,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00012769031658638188,
      "loss": 1.8309581279754639,
      "step": 10274
    },
    {
      "epoch": 0.8645222804901989,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001274290686761037,
      "loss": 1.8360016345977783,
      "step": 10276
    },
    {
      "epoch": 0.864690540957402,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.00012716789079885053,
      "loss": 1.8095251321792603,
      "step": 10278
    },
    {
      "epoch": 0.8648588014246053,
      "grad_norm": 0.140625,
      "learning_rate": 0.0001269067837650042,
      "loss": 1.819960594177246,
      "step": 10280
    },
    {
      "epoch": 0.8650270618918086,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0001266457483847271,
      "loss": 1.8565406799316406,
      "step": 10282
    },
    {
      "epoch": 0.8651953223590118,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00012638478546795893,
      "loss": 1.8702666759490967,
      "step": 10284
    },
    {
      "epoch": 0.865363582826215,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00012612389582441485,
      "loss": 1.7951929569244385,
      "step": 10286
    },
    {
      "epoch": 0.8655318432934183,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001258630802635825,
      "loss": 1.802323341369629,
      "step": 10288
    },
    {
      "epoch": 0.8657001037606215,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00012560233959471964,
      "loss": 1.7732264995574951,
      "step": 10290
    },
    {
      "epoch": 0.8658683642278246,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00012534167462685176,
      "loss": 1.8520487546920776,
      "step": 10292
    },
    {
      "epoch": 0.866036624695028,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00012508108616876947,
      "loss": 1.8278772830963135,
      "step": 10294
    },
    {
      "epoch": 0.8662048851622312,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001248205750290259,
      "loss": 1.8462512493133545,
      "step": 10296
    },
    {
      "epoch": 0.8663731456294343,
      "grad_norm": 0.140625,
      "learning_rate": 0.00012456014201593426,
      "loss": 1.8433963060379028,
      "step": 10298
    },
    {
      "epoch": 0.8665414060966375,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00012429978793756545,
      "loss": 1.839693307876587,
      "step": 10300
    },
    {
      "epoch": 0.8667096665638409,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00012403951360174544,
      "loss": 1.832432508468628,
      "step": 10302
    },
    {
      "epoch": 0.866877927031044,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00012377931981605265,
      "loss": 1.876226544380188,
      "step": 10304
    },
    {
      "epoch": 0.8670461874982472,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00012351920738781576,
      "loss": 1.8313679695129395,
      "step": 10306
    },
    {
      "epoch": 0.8672144479654506,
      "grad_norm": 0.125,
      "learning_rate": 0.00012325917712411078,
      "loss": 1.8396570682525635,
      "step": 10308
    },
    {
      "epoch": 0.8673827084326537,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00012299922983175906,
      "loss": 1.8235797882080078,
      "step": 10310
    },
    {
      "epoch": 0.867550968899857,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00012273936631732428,
      "loss": 1.8437292575836182,
      "step": 10312
    },
    {
      "epoch": 0.8677192293670603,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00012247958738711021,
      "loss": 1.8405916690826416,
      "step": 10314
    },
    {
      "epoch": 0.8678874898342634,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00012221989384715833,
      "loss": 1.8032671213150024,
      "step": 10316
    },
    {
      "epoch": 0.8680557503014666,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00012196028650324492,
      "loss": 1.8314012289047241,
      "step": 10318
    },
    {
      "epoch": 0.86822401076867,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00012170076616087909,
      "loss": 1.7679330110549927,
      "step": 10320
    },
    {
      "epoch": 0.8683922712358731,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00012144133362529971,
      "loss": 1.8172732591629028,
      "step": 10322
    },
    {
      "epoch": 0.8685605317030763,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00012118198970147352,
      "loss": 1.8889851570129395,
      "step": 10324
    },
    {
      "epoch": 0.8687287921702795,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00012092273519409203,
      "loss": 1.8174866437911987,
      "step": 10326
    },
    {
      "epoch": 0.8688970526374828,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0001206635709075694,
      "loss": 1.8562545776367188,
      "step": 10328
    },
    {
      "epoch": 0.869065313104686,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00012040449764604,
      "loss": 1.7608864307403564,
      "step": 10330
    },
    {
      "epoch": 0.8692335735718892,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00012014551621335553,
      "loss": 1.8513953685760498,
      "step": 10332
    },
    {
      "epoch": 0.8694018340390925,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00011988662741308299,
      "loss": 1.824241042137146,
      "step": 10334
    },
    {
      "epoch": 0.8695700945062957,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00011962783204850179,
      "loss": 1.887752890586853,
      "step": 10336
    },
    {
      "epoch": 0.8697383549734989,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00011936913092260146,
      "loss": 1.7893178462982178,
      "step": 10338
    },
    {
      "epoch": 0.8699066154407022,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00011911052483807922,
      "loss": 1.8448505401611328,
      "step": 10340
    },
    {
      "epoch": 0.8700748759079054,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00011885201459733728,
      "loss": 1.8083279132843018,
      "step": 10342
    },
    {
      "epoch": 0.8702431363751086,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00011859360100248059,
      "loss": 1.847838044166565,
      "step": 10344
    },
    {
      "epoch": 0.870411396842312,
      "grad_norm": 0.125,
      "learning_rate": 0.00011833528485531408,
      "loss": 1.8369171619415283,
      "step": 10346
    },
    {
      "epoch": 0.8705796573095151,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00011807706695734043,
      "loss": 1.8379526138305664,
      "step": 10348
    },
    {
      "epoch": 0.8707479177767183,
      "grad_norm": 0.125,
      "learning_rate": 0.00011781894810975745,
      "loss": 1.8497846126556396,
      "step": 10350
    },
    {
      "epoch": 0.8709161782439216,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001175609291134556,
      "loss": 1.8529751300811768,
      "step": 10352
    },
    {
      "epoch": 0.8710844387111248,
      "grad_norm": 0.125,
      "learning_rate": 0.00011730301076901549,
      "loss": 1.8099555969238281,
      "step": 10354
    },
    {
      "epoch": 0.871252699178328,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00011704519387670549,
      "loss": 1.7938547134399414,
      "step": 10356
    },
    {
      "epoch": 0.8714209596455312,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0001167874792364792,
      "loss": 1.8217086791992188,
      "step": 10358
    },
    {
      "epoch": 0.8715892201127345,
      "grad_norm": 0.140625,
      "learning_rate": 0.0001165298676479728,
      "loss": 1.8215312957763672,
      "step": 10360
    },
    {
      "epoch": 0.8717574805799377,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00011627235991050299,
      "loss": 1.8505569696426392,
      "step": 10362
    },
    {
      "epoch": 0.8719257410471409,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00011601495682306394,
      "loss": 1.789318561553955,
      "step": 10364
    },
    {
      "epoch": 0.8720940015143442,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00011575765918432534,
      "loss": 1.7800054550170898,
      "step": 10366
    },
    {
      "epoch": 0.8722622619815474,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00011550046779262961,
      "loss": 1.8199899196624756,
      "step": 10368
    },
    {
      "epoch": 0.8724305224487506,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0001152433834459894,
      "loss": 1.8228750228881836,
      "step": 10370
    },
    {
      "epoch": 0.8725987829159539,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001149864069420855,
      "loss": 1.8115959167480469,
      "step": 10372
    },
    {
      "epoch": 0.8727670433831571,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00011472953907826382,
      "loss": 1.7973060607910156,
      "step": 10374
    },
    {
      "epoch": 0.8729353038503603,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.00011447278065153321,
      "loss": 1.8173240423202515,
      "step": 10376
    },
    {
      "epoch": 0.8731035643175636,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00011421613245856318,
      "loss": 1.7981102466583252,
      "step": 10378
    },
    {
      "epoch": 0.8732718247847668,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00011395959529568087,
      "loss": 1.8515565395355225,
      "step": 10380
    },
    {
      "epoch": 0.87344008525197,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00011370316995886929,
      "loss": 1.8784126043319702,
      "step": 10382
    },
    {
      "epoch": 0.8736083457191732,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00011344685724376412,
      "loss": 1.8183987140655518,
      "step": 10384
    },
    {
      "epoch": 0.8737766061863765,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00011319065794565189,
      "loss": 1.816469669342041,
      "step": 10386
    },
    {
      "epoch": 0.8739448666535797,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00011293457285946703,
      "loss": 1.8269344568252563,
      "step": 10388
    },
    {
      "epoch": 0.8741131271207829,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001126786027797896,
      "loss": 1.8073118925094604,
      "step": 10390
    },
    {
      "epoch": 0.8742813875879862,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00011242274850084301,
      "loss": 1.8218525648117065,
      "step": 10392
    },
    {
      "epoch": 0.8744496480551894,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00011216701081649108,
      "loss": 1.8079239130020142,
      "step": 10394
    },
    {
      "epoch": 0.8746179085223926,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00011191139052023615,
      "loss": 1.840364933013916,
      "step": 10396
    },
    {
      "epoch": 0.8747861689895959,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00011165588840521604,
      "loss": 1.8037405014038086,
      "step": 10398
    },
    {
      "epoch": 0.8749544294567991,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00011140050526420218,
      "loss": 1.8120594024658203,
      "step": 10400
    },
    {
      "epoch": 0.8751226899240023,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00011114524188959659,
      "loss": 1.7770408391952515,
      "step": 10402
    },
    {
      "epoch": 0.8752909503912056,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00011089009907342976,
      "loss": 1.7630571126937866,
      "step": 10404
    },
    {
      "epoch": 0.8754592108584088,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00011063507760735823,
      "loss": 1.8450000286102295,
      "step": 10406
    },
    {
      "epoch": 0.875627471325612,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.00011038017828266184,
      "loss": 1.8053944110870361,
      "step": 10408
    },
    {
      "epoch": 0.8757957317928152,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00011012540189024154,
      "loss": 1.7787978649139404,
      "step": 10410
    },
    {
      "epoch": 0.8759639922600185,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00010987074922061689,
      "loss": 1.8256986141204834,
      "step": 10412
    },
    {
      "epoch": 0.8761322527272217,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00010961622106392344,
      "loss": 1.794630527496338,
      "step": 10414
    },
    {
      "epoch": 0.8763005131944249,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00010936181820991047,
      "loss": 1.7617545127868652,
      "step": 10416
    },
    {
      "epoch": 0.8764687736616282,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00010910754144793851,
      "loss": 1.8371927738189697,
      "step": 10418
    },
    {
      "epoch": 0.8766370341288314,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00010885339156697678,
      "loss": 1.7955780029296875,
      "step": 10420
    },
    {
      "epoch": 0.8768052945960346,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00010859936935560088,
      "loss": 1.7939481735229492,
      "step": 10422
    },
    {
      "epoch": 0.8769735550632379,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00010834547560199025,
      "loss": 1.8699250221252441,
      "step": 10424
    },
    {
      "epoch": 0.8771418155304411,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00010809171109392569,
      "loss": 1.8247511386871338,
      "step": 10426
    },
    {
      "epoch": 0.8773100759976443,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00010783807661878706,
      "loss": 1.825154423713684,
      "step": 10428
    },
    {
      "epoch": 0.8774783364648476,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00010758457296355077,
      "loss": 1.847503900527954,
      "step": 10430
    },
    {
      "epoch": 0.8776465969320508,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00010733120091478715,
      "loss": 1.845982313156128,
      "step": 10432
    },
    {
      "epoch": 0.877814857399254,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00010707796125865848,
      "loss": 1.8467328548431396,
      "step": 10434
    },
    {
      "epoch": 0.8779831178664573,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00010682485478091598,
      "loss": 1.8146226406097412,
      "step": 10436
    },
    {
      "epoch": 0.8781513783336605,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001065718822668977,
      "loss": 1.8215408325195312,
      "step": 10438
    },
    {
      "epoch": 0.8783196388008637,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00010631904450152623,
      "loss": 1.8255783319473267,
      "step": 10440
    },
    {
      "epoch": 0.8784878992680669,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00010606634226930574,
      "loss": 1.8298580646514893,
      "step": 10442
    },
    {
      "epoch": 0.8786561597352702,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00010581377635432018,
      "loss": 1.8196682929992676,
      "step": 10444
    },
    {
      "epoch": 0.8788244202024734,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00010556134754023028,
      "loss": 1.8280870914459229,
      "step": 10446
    },
    {
      "epoch": 0.8789926806696766,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00010530905661027158,
      "loss": 1.8192722797393799,
      "step": 10448
    },
    {
      "epoch": 0.8791609411368799,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00010505690434725163,
      "loss": 1.7947022914886475,
      "step": 10450
    },
    {
      "epoch": 0.8793292016040831,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00010480489153354777,
      "loss": 1.86146879196167,
      "step": 10452
    },
    {
      "epoch": 0.8794974620712863,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00010455301895110478,
      "loss": 1.7622032165527344,
      "step": 10454
    },
    {
      "epoch": 0.8796657225384896,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001043012873814321,
      "loss": 1.825929880142212,
      "step": 10456
    },
    {
      "epoch": 0.8798339830056928,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001040496976056019,
      "loss": 1.8209515810012817,
      "step": 10458
    },
    {
      "epoch": 0.880002243472896,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00010379825040424614,
      "loss": 1.8029673099517822,
      "step": 10460
    },
    {
      "epoch": 0.8801705039400993,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00010354694655755464,
      "loss": 1.8299963474273682,
      "step": 10462
    },
    {
      "epoch": 0.8803387644073025,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00010329578684527221,
      "loss": 1.8509035110473633,
      "step": 10464
    },
    {
      "epoch": 0.8805070248745057,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00010304477204669652,
      "loss": 1.8124629259109497,
      "step": 10466
    },
    {
      "epoch": 0.8806752853417089,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00010279390294067568,
      "loss": 1.8112757205963135,
      "step": 10468
    },
    {
      "epoch": 0.8808435458089122,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001025431803056056,
      "loss": 1.827420949935913,
      "step": 10470
    },
    {
      "epoch": 0.8810118062761154,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.00010229260491942788,
      "loss": 1.8295466899871826,
      "step": 10472
    },
    {
      "epoch": 0.8811800667433186,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00010204217755962711,
      "loss": 1.8114385604858398,
      "step": 10474
    },
    {
      "epoch": 0.8813483272105219,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00010179189900322857,
      "loss": 1.8073551654815674,
      "step": 10476
    },
    {
      "epoch": 0.8815165876777251,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000101541770026796,
      "loss": 1.7833055257797241,
      "step": 10478
    },
    {
      "epoch": 0.8816848481449283,
      "grad_norm": 0.12109375,
      "learning_rate": 0.00010129179140642884,
      "loss": 1.7322614192962646,
      "step": 10480
    },
    {
      "epoch": 0.8818531086121316,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001010419639177601,
      "loss": 1.848292350769043,
      "step": 10482
    },
    {
      "epoch": 0.8820213690793348,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00010079228833595385,
      "loss": 1.850351333618164,
      "step": 10484
    },
    {
      "epoch": 0.882189629546538,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001005427654357028,
      "loss": 1.7868213653564453,
      "step": 10486
    },
    {
      "epoch": 0.8823578900137413,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00010029339599122593,
      "loss": 1.8382256031036377,
      "step": 10488
    },
    {
      "epoch": 0.8825261504809445,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00010004418077626607,
      "loss": 1.8251088857650757,
      "step": 10490
    },
    {
      "epoch": 0.8826944109481477,
      "grad_norm": 0.125,
      "learning_rate": 9.97951205640876e-05,
      "loss": 1.8249109983444214,
      "step": 10492
    },
    {
      "epoch": 0.8828626714153509,
      "grad_norm": 0.12890625,
      "learning_rate": 9.95462161274737e-05,
      "loss": 1.8369827270507812,
      "step": 10494
    },
    {
      "epoch": 0.8830309318825542,
      "grad_norm": 0.126953125,
      "learning_rate": 9.92974682387246e-05,
      "loss": 1.85978364944458,
      "step": 10496
    },
    {
      "epoch": 0.8831991923497574,
      "grad_norm": 0.134765625,
      "learning_rate": 9.904887766965439e-05,
      "loss": 1.837876796722412,
      "step": 10498
    },
    {
      "epoch": 0.8833674528169606,
      "grad_norm": 0.125,
      "learning_rate": 9.880044519158938e-05,
      "loss": 1.814760684967041,
      "step": 10500
    },
    {
      "epoch": 0.8835357132841639,
      "grad_norm": 0.1240234375,
      "learning_rate": 9.85521715753651e-05,
      "loss": 1.8284838199615479,
      "step": 10502
    },
    {
      "epoch": 0.8837039737513671,
      "grad_norm": 0.1318359375,
      "learning_rate": 9.830405759132422e-05,
      "loss": 1.8595483303070068,
      "step": 10504
    },
    {
      "epoch": 0.8838722342185703,
      "grad_norm": 0.134765625,
      "learning_rate": 9.805610400931425e-05,
      "loss": 1.8475183248519897,
      "step": 10506
    },
    {
      "epoch": 0.8840404946857736,
      "grad_norm": 0.1328125,
      "learning_rate": 9.780831159868478e-05,
      "loss": 1.7537281513214111,
      "step": 10508
    },
    {
      "epoch": 0.8842087551529768,
      "grad_norm": 0.1318359375,
      "learning_rate": 9.756068112828555e-05,
      "loss": 1.826585292816162,
      "step": 10510
    },
    {
      "epoch": 0.88437701562018,
      "grad_norm": 0.1279296875,
      "learning_rate": 9.731321336646368e-05,
      "loss": 1.8154162168502808,
      "step": 10512
    },
    {
      "epoch": 0.8845452760873833,
      "grad_norm": 0.130859375,
      "learning_rate": 9.706590908106138e-05,
      "loss": 1.8383324146270752,
      "step": 10514
    },
    {
      "epoch": 0.8847135365545865,
      "grad_norm": 0.1298828125,
      "learning_rate": 9.681876903941386e-05,
      "loss": 1.8386945724487305,
      "step": 10516
    },
    {
      "epoch": 0.8848817970217897,
      "grad_norm": 0.1318359375,
      "learning_rate": 9.657179400834647e-05,
      "loss": 1.84430992603302,
      "step": 10518
    },
    {
      "epoch": 0.885050057488993,
      "grad_norm": 0.1298828125,
      "learning_rate": 9.632498475417275e-05,
      "loss": 1.796187400817871,
      "step": 10520
    },
    {
      "epoch": 0.8852183179561962,
      "grad_norm": 0.126953125,
      "learning_rate": 9.60783420426917e-05,
      "loss": 1.7522003650665283,
      "step": 10522
    },
    {
      "epoch": 0.8853865784233994,
      "grad_norm": 0.1298828125,
      "learning_rate": 9.583186663918579e-05,
      "loss": 1.8228411674499512,
      "step": 10524
    },
    {
      "epoch": 0.8855548388906026,
      "grad_norm": 0.1259765625,
      "learning_rate": 9.558555930841813e-05,
      "loss": 1.8647325038909912,
      "step": 10526
    },
    {
      "epoch": 0.8857230993578059,
      "grad_norm": 0.1279296875,
      "learning_rate": 9.533942081463045e-05,
      "loss": 1.8336642980575562,
      "step": 10528
    },
    {
      "epoch": 0.8858913598250091,
      "grad_norm": 0.1328125,
      "learning_rate": 9.509345192154067e-05,
      "loss": 1.8226854801177979,
      "step": 10530
    },
    {
      "epoch": 0.8860596202922123,
      "grad_norm": 0.1357421875,
      "learning_rate": 9.484765339234032e-05,
      "loss": 1.8022663593292236,
      "step": 10532
    },
    {
      "epoch": 0.8862278807594156,
      "grad_norm": 0.12890625,
      "learning_rate": 9.460202598969251e-05,
      "loss": 1.8154181241989136,
      "step": 10534
    },
    {
      "epoch": 0.8863961412266188,
      "grad_norm": 0.150390625,
      "learning_rate": 9.43565704757292e-05,
      "loss": 1.791346549987793,
      "step": 10536
    },
    {
      "epoch": 0.886564401693822,
      "grad_norm": 0.1220703125,
      "learning_rate": 9.411128761204915e-05,
      "loss": 1.816857099533081,
      "step": 10538
    },
    {
      "epoch": 0.8867326621610253,
      "grad_norm": 0.126953125,
      "learning_rate": 9.386617815971539e-05,
      "loss": 1.83504319190979,
      "step": 10540
    },
    {
      "epoch": 0.8869009226282285,
      "grad_norm": 0.126953125,
      "learning_rate": 9.362124287925279e-05,
      "loss": 1.8385217189788818,
      "step": 10542
    },
    {
      "epoch": 0.8870691830954317,
      "grad_norm": 0.130859375,
      "learning_rate": 9.337648253064591e-05,
      "loss": 1.8373417854309082,
      "step": 10544
    },
    {
      "epoch": 0.887237443562635,
      "grad_norm": 0.1220703125,
      "learning_rate": 9.313189787333651e-05,
      "loss": 1.803870439529419,
      "step": 10546
    },
    {
      "epoch": 0.8874057040298382,
      "grad_norm": 0.1328125,
      "learning_rate": 9.288748966622126e-05,
      "loss": 1.8170305490493774,
      "step": 10548
    },
    {
      "epoch": 0.8875739644970414,
      "grad_norm": 0.1298828125,
      "learning_rate": 9.264325866764916e-05,
      "loss": 1.8131345510482788,
      "step": 10550
    },
    {
      "epoch": 0.8877422249642446,
      "grad_norm": 0.12890625,
      "learning_rate": 9.239920563541962e-05,
      "loss": 1.8015702962875366,
      "step": 10552
    },
    {
      "epoch": 0.8879104854314479,
      "grad_norm": 0.140625,
      "learning_rate": 9.215533132677968e-05,
      "loss": 1.8060026168823242,
      "step": 10554
    },
    {
      "epoch": 0.8880787458986511,
      "grad_norm": 0.13671875,
      "learning_rate": 9.191163649842182e-05,
      "loss": 1.8301334381103516,
      "step": 10556
    },
    {
      "epoch": 0.8882470063658543,
      "grad_norm": 0.12890625,
      "learning_rate": 9.166812190648183e-05,
      "loss": 1.8671655654907227,
      "step": 10558
    },
    {
      "epoch": 0.8884152668330576,
      "grad_norm": 0.126953125,
      "learning_rate": 9.142478830653601e-05,
      "loss": 1.8182122707366943,
      "step": 10560
    },
    {
      "epoch": 0.8885835273002608,
      "grad_norm": 0.1298828125,
      "learning_rate": 9.11816364535993e-05,
      "loss": 1.8378891944885254,
      "step": 10562
    },
    {
      "epoch": 0.888751787767464,
      "grad_norm": 0.1376953125,
      "learning_rate": 9.093866710212257e-05,
      "loss": 1.8901876211166382,
      "step": 10564
    },
    {
      "epoch": 0.8889200482346673,
      "grad_norm": 0.12109375,
      "learning_rate": 9.069588100599041e-05,
      "loss": 1.7951891422271729,
      "step": 10566
    },
    {
      "epoch": 0.8890883087018705,
      "grad_norm": 0.1328125,
      "learning_rate": 9.045327891851897e-05,
      "loss": 1.8456066846847534,
      "step": 10568
    },
    {
      "epoch": 0.8892565691690737,
      "grad_norm": 0.12060546875,
      "learning_rate": 9.021086159245324e-05,
      "loss": 1.8006625175476074,
      "step": 10570
    },
    {
      "epoch": 0.889424829636277,
      "grad_norm": 0.126953125,
      "learning_rate": 8.996862977996519e-05,
      "loss": 1.7647593021392822,
      "step": 10572
    },
    {
      "epoch": 0.8895930901034802,
      "grad_norm": 0.126953125,
      "learning_rate": 8.972658423265088e-05,
      "loss": 1.8095566034317017,
      "step": 10574
    },
    {
      "epoch": 0.8897613505706834,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.948472570152872e-05,
      "loss": 1.7990866899490356,
      "step": 10576
    },
    {
      "epoch": 0.8899296110378867,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.924305493703666e-05,
      "loss": 1.7985013723373413,
      "step": 10578
    },
    {
      "epoch": 0.8900978715050899,
      "grad_norm": 0.125,
      "learning_rate": 8.900157268903e-05,
      "loss": 1.8147401809692383,
      "step": 10580
    },
    {
      "epoch": 0.8902661319722931,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.876027970677938e-05,
      "loss": 1.8018863201141357,
      "step": 10582
    },
    {
      "epoch": 0.8904343924394963,
      "grad_norm": 0.12890625,
      "learning_rate": 8.851917673896785e-05,
      "loss": 1.8015493154525757,
      "step": 10584
    },
    {
      "epoch": 0.8906026529066996,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.827826453368917e-05,
      "loss": 1.835490345954895,
      "step": 10586
    },
    {
      "epoch": 0.8907709133739028,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.803754383844506e-05,
      "loss": 1.8554110527038574,
      "step": 10588
    },
    {
      "epoch": 0.890939173841106,
      "grad_norm": 0.12451171875,
      "learning_rate": 8.779701540014295e-05,
      "loss": 1.8058515787124634,
      "step": 10590
    },
    {
      "epoch": 0.8911074343083093,
      "grad_norm": 0.1328125,
      "learning_rate": 8.755667996509401e-05,
      "loss": 1.8531959056854248,
      "step": 10592
    },
    {
      "epoch": 0.8912756947755125,
      "grad_norm": 0.126953125,
      "learning_rate": 8.731653827901017e-05,
      "loss": 1.811143398284912,
      "step": 10594
    },
    {
      "epoch": 0.8914439552427157,
      "grad_norm": 0.140625,
      "learning_rate": 8.707659108700258e-05,
      "loss": 1.8423328399658203,
      "step": 10596
    },
    {
      "epoch": 0.891612215709919,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.683683913357862e-05,
      "loss": 1.7477567195892334,
      "step": 10598
    },
    {
      "epoch": 0.8917804761771222,
      "grad_norm": 0.126953125,
      "learning_rate": 8.65972831626401e-05,
      "loss": 1.8326442241668701,
      "step": 10600
    },
    {
      "epoch": 0.8919487366443254,
      "grad_norm": 0.162109375,
      "learning_rate": 8.63579239174806e-05,
      "loss": 1.7994303703308105,
      "step": 10602
    },
    {
      "epoch": 0.8921169971115287,
      "grad_norm": 0.130859375,
      "learning_rate": 8.611876214078327e-05,
      "loss": 1.8673512935638428,
      "step": 10604
    },
    {
      "epoch": 0.8922852575787319,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.587979857461877e-05,
      "loss": 1.8465418815612793,
      "step": 10606
    },
    {
      "epoch": 0.8924535180459351,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.564103396044248e-05,
      "loss": 1.8030707836151123,
      "step": 10608
    },
    {
      "epoch": 0.8926217785131383,
      "grad_norm": 0.12890625,
      "learning_rate": 8.540246903909271e-05,
      "loss": 1.7931513786315918,
      "step": 10610
    },
    {
      "epoch": 0.8927900389803416,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.516410455078792e-05,
      "loss": 1.801175594329834,
      "step": 10612
    },
    {
      "epoch": 0.8929582994475448,
      "grad_norm": 0.12890625,
      "learning_rate": 8.492594123512497e-05,
      "loss": 1.8924297094345093,
      "step": 10614
    },
    {
      "epoch": 0.893126559914748,
      "grad_norm": 0.138671875,
      "learning_rate": 8.468797983107629e-05,
      "loss": 1.8298330307006836,
      "step": 10616
    },
    {
      "epoch": 0.8932948203819513,
      "grad_norm": 0.1240234375,
      "learning_rate": 8.445022107698776e-05,
      "loss": 1.7977910041809082,
      "step": 10618
    },
    {
      "epoch": 0.8934630808491545,
      "grad_norm": 0.1337890625,
      "learning_rate": 8.421266571057679e-05,
      "loss": 1.7874908447265625,
      "step": 10620
    },
    {
      "epoch": 0.8936313413163577,
      "grad_norm": 0.14453125,
      "learning_rate": 8.397531446892937e-05,
      "loss": 1.8097655773162842,
      "step": 10622
    },
    {
      "epoch": 0.893799601783561,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.373816808849843e-05,
      "loss": 1.8476923704147339,
      "step": 10624
    },
    {
      "epoch": 0.8939678622507642,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.350122730510106e-05,
      "loss": 1.8247867822647095,
      "step": 10626
    },
    {
      "epoch": 0.8941361227179674,
      "grad_norm": 0.12060546875,
      "learning_rate": 8.326449285391643e-05,
      "loss": 1.8260849714279175,
      "step": 10628
    },
    {
      "epoch": 0.8943043831851707,
      "grad_norm": 0.1240234375,
      "learning_rate": 8.302796546948369e-05,
      "loss": 1.8451807498931885,
      "step": 10630
    },
    {
      "epoch": 0.8944726436523739,
      "grad_norm": 0.1337890625,
      "learning_rate": 8.27916458856992e-05,
      "loss": 1.797168254852295,
      "step": 10632
    },
    {
      "epoch": 0.8946409041195771,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.255553483581489e-05,
      "loss": 1.8159763813018799,
      "step": 10634
    },
    {
      "epoch": 0.8948091645867803,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.231963305243535e-05,
      "loss": 1.779120683670044,
      "step": 10636
    },
    {
      "epoch": 0.8949774250539836,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.208394126751614e-05,
      "loss": 1.8124102354049683,
      "step": 10638
    },
    {
      "epoch": 0.8951456855211868,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.184846021236101e-05,
      "loss": 1.8388490676879883,
      "step": 10640
    },
    {
      "epoch": 0.89531394598839,
      "grad_norm": 0.123046875,
      "learning_rate": 8.161319061761991e-05,
      "loss": 1.8065133094787598,
      "step": 10642
    },
    {
      "epoch": 0.8954822064555933,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.137813321328682e-05,
      "loss": 1.7996330261230469,
      "step": 10644
    },
    {
      "epoch": 0.8956504669227965,
      "grad_norm": 0.1240234375,
      "learning_rate": 8.114328872869708e-05,
      "loss": 1.7628419399261475,
      "step": 10646
    },
    {
      "epoch": 0.8958187273899997,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.090865789252565e-05,
      "loss": 1.8208856582641602,
      "step": 10648
    },
    {
      "epoch": 0.895986987857203,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.067424143278436e-05,
      "loss": 1.8938796520233154,
      "step": 10650
    },
    {
      "epoch": 0.8961552483244062,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.044004007682006e-05,
      "loss": 1.838402271270752,
      "step": 10652
    },
    {
      "epoch": 0.8963235087916094,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.020605455131203e-05,
      "loss": 1.83860445022583,
      "step": 10654
    },
    {
      "epoch": 0.8964917692588127,
      "grad_norm": 0.1396484375,
      "learning_rate": 7.99722855822699e-05,
      "loss": 1.81541907787323,
      "step": 10656
    },
    {
      "epoch": 0.8966600297260159,
      "grad_norm": 0.130859375,
      "learning_rate": 7.973873389503149e-05,
      "loss": 1.8169031143188477,
      "step": 10658
    },
    {
      "epoch": 0.8968282901932191,
      "grad_norm": 0.1318359375,
      "learning_rate": 7.950540021426023e-05,
      "loss": 1.8076441287994385,
      "step": 10660
    },
    {
      "epoch": 0.8969965506604224,
      "grad_norm": 0.12890625,
      "learning_rate": 7.927228526394334e-05,
      "loss": 1.8505065441131592,
      "step": 10662
    },
    {
      "epoch": 0.8971648111276256,
      "grad_norm": 0.1357421875,
      "learning_rate": 7.903938976738924e-05,
      "loss": 1.8464287519454956,
      "step": 10664
    },
    {
      "epoch": 0.8973330715948288,
      "grad_norm": 0.126953125,
      "learning_rate": 7.880671444722536e-05,
      "loss": 1.818138599395752,
      "step": 10666
    },
    {
      "epoch": 0.897501332062032,
      "grad_norm": 0.1259765625,
      "learning_rate": 7.857426002539618e-05,
      "loss": 1.758169174194336,
      "step": 10668
    },
    {
      "epoch": 0.8976695925292353,
      "grad_norm": 0.12890625,
      "learning_rate": 7.834202722316054e-05,
      "loss": 1.7981303930282593,
      "step": 10670
    },
    {
      "epoch": 0.8978378529964385,
      "grad_norm": 0.123046875,
      "learning_rate": 7.811001676108987e-05,
      "loss": 1.759138822555542,
      "step": 10672
    },
    {
      "epoch": 0.8980061134636417,
      "grad_norm": 0.1357421875,
      "learning_rate": 7.787822935906547e-05,
      "loss": 1.7813206911087036,
      "step": 10674
    },
    {
      "epoch": 0.898174373930845,
      "grad_norm": 0.1318359375,
      "learning_rate": 7.764666573627684e-05,
      "loss": 1.7955987453460693,
      "step": 10676
    },
    {
      "epoch": 0.8983426343980482,
      "grad_norm": 0.12890625,
      "learning_rate": 7.741532661121883e-05,
      "loss": 1.840401530265808,
      "step": 10678
    },
    {
      "epoch": 0.8985108948652514,
      "grad_norm": 0.1298828125,
      "learning_rate": 7.718421270168985e-05,
      "loss": 1.8384289741516113,
      "step": 10680
    },
    {
      "epoch": 0.8986791553324547,
      "grad_norm": 0.1337890625,
      "learning_rate": 7.695332472478963e-05,
      "loss": 1.8136204481124878,
      "step": 10682
    },
    {
      "epoch": 0.8988474157996579,
      "grad_norm": 0.12451171875,
      "learning_rate": 7.672266339691661e-05,
      "loss": 1.8239694833755493,
      "step": 10684
    },
    {
      "epoch": 0.8990156762668611,
      "grad_norm": 0.1279296875,
      "learning_rate": 7.64922294337663e-05,
      "loss": 1.8389451503753662,
      "step": 10686
    },
    {
      "epoch": 0.8991839367340644,
      "grad_norm": 0.1279296875,
      "learning_rate": 7.626202355032849e-05,
      "loss": 1.8323720693588257,
      "step": 10688
    },
    {
      "epoch": 0.8993521972012676,
      "grad_norm": 0.1298828125,
      "learning_rate": 7.603204646088533e-05,
      "loss": 1.8228474855422974,
      "step": 10690
    },
    {
      "epoch": 0.8995204576684708,
      "grad_norm": 0.11962890625,
      "learning_rate": 7.58022988790092e-05,
      "loss": 1.8408551216125488,
      "step": 10692
    },
    {
      "epoch": 0.899688718135674,
      "grad_norm": 0.1259765625,
      "learning_rate": 7.557278151756028e-05,
      "loss": 1.7926255464553833,
      "step": 10694
    },
    {
      "epoch": 0.8998569786028773,
      "grad_norm": 0.134765625,
      "learning_rate": 7.534349508868431e-05,
      "loss": 1.8190979957580566,
      "step": 10696
    },
    {
      "epoch": 0.9000252390700805,
      "grad_norm": 0.15625,
      "learning_rate": 7.511444030381076e-05,
      "loss": 1.8609733581542969,
      "step": 10698
    },
    {
      "epoch": 0.9001934995372837,
      "grad_norm": 0.1318359375,
      "learning_rate": 7.488561787365002e-05,
      "loss": 1.815778136253357,
      "step": 10700
    }
  ],
  "logging_steps": 2,
  "max_steps": 11887,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1189,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.208671351568597e+18,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}